Coverage Report

Created: 2026-06-16 07:20

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
0
#define NS_INDEX_EMPTY  INT_MAX
81
0
#define NS_INDEX_XML    (INT_MAX - 1)
82
0
#define URI_HASH_EMPTY  0xD943A04E
83
0
#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
30.9k
#define XML_MAX_ATTRS 100000000 /* 100 million */
94
95
896k
#define XML_SPECIAL_EXTERNAL    (1 << 20)
96
896k
#define XML_SPECIAL_TYPE_MASK   (XML_SPECIAL_EXTERNAL - 1)
97
98
892k
#define XML_ATTVAL_ALLOC        (1 << 0)
99
0
#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
70.3k
#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
76.0k
#define XML_ENT_FIXED_COST 20
170
171
266M
#define XML_PARSER_BIG_BUFFER_SIZE 300
172
1.33M
#define XML_PARSER_BUFFER_SIZE 100
173
32.1k
#define SAX_COMPAT_MODE BAD_CAST "SAX compatibility mode document"
174
175
/**
176
 * XML_PARSER_CHUNK_SIZE
177
 *
178
 * When calling GROW that's the minimal amount of data
179
 * the parser expected to have received. It is not a hard
180
 * limit but an optimization when reading strings like Names
181
 * It is not strictly needed as long as inputs available characters
182
 * are followed by 0, which should be provided by the I/O level
183
 */
184
#define XML_PARSER_CHUNK_SIZE 100
185
186
/**
187
 * Constant string describing the version of the library used at
188
 * run-time.
189
 */
190
const char *const
191
xmlParserVersion = LIBXML_VERSION_STRING LIBXML_VERSION_EXTRA;
192
193
/*
194
 * List of XML prefixed PI allowed by W3C specs
195
 */
196
197
static const char* const xmlW3CPIs[] = {
198
    "xml-stylesheet",
199
    "xml-model",
200
    NULL
201
};
202
203
204
/* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */
205
static xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt,
206
                                              const xmlChar **str);
207
208
static void
209
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent);
210
211
static int
212
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity);
213
214
static void
215
xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl);
216
217
/************************************************************************
218
 *                  *
219
 *    Some factorized error routines        *
220
 *                  *
221
 ************************************************************************/
222
223
static void
224
0
xmlErrMemory(xmlParserCtxtPtr ctxt) {
225
0
    xmlCtxtErrMemory(ctxt);
226
0
}
227
228
/**
229
 * Handle a redefinition of attribute error
230
 *
231
 * @param ctxt  an XML parser context
232
 * @param prefix  the attribute prefix
233
 * @param localname  the attribute localname
234
 */
235
static void
236
xmlErrAttributeDup(xmlParserCtxtPtr ctxt, const xmlChar * prefix,
237
                   const xmlChar * localname)
238
26
{
239
26
    if (prefix == NULL)
240
26
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
241
26
                   XML_ERR_FATAL, localname, NULL, NULL, 0,
242
26
                   "Attribute %s redefined\n", localname);
243
0
    else
244
0
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
245
0
                   XML_ERR_FATAL, prefix, localname, NULL, 0,
246
0
                   "Attribute %s:%s redefined\n", prefix, localname);
247
26
}
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
45.9k
{
260
45.9k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
261
45.9k
               NULL, NULL, NULL, 0, "%s", msg);
262
45.9k
}
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
84.9k
{
277
84.9k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING,
278
84.9k
               str1, str2, NULL, 0, msg, str1, str2);
279
84.9k
}
280
281
#ifdef LIBXML_VALID_ENABLED
282
/**
283
 * Handle a validity error.
284
 *
285
 * @param ctxt  an XML parser context
286
 * @param error  the error number
287
 * @param msg  the error message
288
 * @param str1  extra data
289
 * @param str2  extra data
290
 */
291
static void LIBXML_ATTR_FORMAT(3,0)
292
xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error,
293
              const char *msg, const xmlChar *str1, const xmlChar *str2)
294
0
{
295
0
    ctxt->valid = 0;
296
297
0
    xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR,
298
0
               str1, str2, NULL, 0, msg, str1, str2);
299
0
}
300
#endif
301
302
/**
303
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
304
 *
305
 * @param ctxt  an XML parser context
306
 * @param error  the error number
307
 * @param msg  the error message
308
 * @param val  an integer value
309
 */
310
static void LIBXML_ATTR_FORMAT(3,0)
311
xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error,
312
                  const char *msg, int val)
313
7.08k
{
314
7.08k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
315
7.08k
               NULL, NULL, NULL, val, msg, val);
316
7.08k
}
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
5.51k
{
333
5.51k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
334
5.51k
               str1, str2, NULL, val, msg, str1, val, str2);
335
5.51k
}
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
53.9k
{
349
53.9k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
350
53.9k
               val, NULL, NULL, 0, msg, val);
351
53.9k
}
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
0
{
365
0
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR,
366
0
               val, NULL, NULL, 0, msg, val);
367
0
}
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
452
{
385
452
    ctxt->nsWellFormed = 0;
386
387
452
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR,
388
452
               info1, info2, info3, 0, msg, info1, info2, info3);
389
452
}
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
0
{
407
0
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING,
408
0
               info1, info2, info3, 0, msg, info1, info2, info3);
409
0
}
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
636k
{
436
636k
    unsigned long consumed;
437
636k
    unsigned long *expandedSize;
438
636k
    xmlParserInputPtr input = ctxt->input;
439
636k
    xmlEntityPtr entity = input->entity;
440
441
636k
    if ((entity) && (entity->flags & XML_ENT_CHECKED))
442
565k
        return(0);
443
444
    /*
445
     * Compute total consumed bytes so far, including input streams of
446
     * external entities.
447
     */
448
70.3k
    consumed = input->consumed;
449
70.3k
    xmlSaturatedAddSizeT(&consumed, input->cur - input->base);
450
70.3k
    xmlSaturatedAdd(&consumed, ctxt->sizeentities);
451
452
70.3k
    if (entity)
453
24.1k
        expandedSize = &entity->expandedSize;
454
46.1k
    else
455
46.1k
        expandedSize = &ctxt->sizeentcopy;
456
457
    /*
458
     * Add extra cost and some fixed cost.
459
     */
460
70.3k
    xmlSaturatedAdd(expandedSize, extra);
461
70.3k
    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
70.3k
    if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) &&
469
1.52k
        ((*expandedSize >= ULONG_MAX) ||
470
1.52k
         (*expandedSize / ctxt->maxAmpl > consumed))) {
471
253
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
472
253
                       "Maximum entity amplification factor exceeded, see "
473
253
                       "xmlCtxtSetMaxAmplification.\n");
474
253
        return(1);
475
253
    }
476
477
70.0k
    return(0);
478
70.3k
}
479
480
/************************************************************************
481
 *                  *
482
 *    Library wide options          *
483
 *                  *
484
 ************************************************************************/
485
486
/**
487
 * Examines if the library has been compiled with a given feature.
488
 *
489
 * @param feature  the feature to be examined
490
 * @returns zero (0) if the feature does not exist or an unknown
491
 * feature is requested, non-zero otherwise.
492
 */
493
int
494
xmlHasFeature(xmlFeature feature)
495
0
{
496
0
    switch (feature) {
497
0
  case XML_WITH_THREAD:
498
0
#ifdef LIBXML_THREAD_ENABLED
499
0
      return(1);
500
#else
501
      return(0);
502
#endif
503
0
        case XML_WITH_TREE:
504
0
            return(1);
505
0
        case XML_WITH_OUTPUT:
506
0
#ifdef LIBXML_OUTPUT_ENABLED
507
0
            return(1);
508
#else
509
            return(0);
510
#endif
511
0
        case XML_WITH_PUSH:
512
0
#ifdef LIBXML_PUSH_ENABLED
513
0
            return(1);
514
#else
515
            return(0);
516
#endif
517
0
        case XML_WITH_READER:
518
0
#ifdef LIBXML_READER_ENABLED
519
0
            return(1);
520
#else
521
            return(0);
522
#endif
523
0
        case XML_WITH_PATTERN:
524
0
#ifdef LIBXML_PATTERN_ENABLED
525
0
            return(1);
526
#else
527
            return(0);
528
#endif
529
0
        case XML_WITH_WRITER:
530
0
#ifdef LIBXML_WRITER_ENABLED
531
0
            return(1);
532
#else
533
            return(0);
534
#endif
535
0
        case XML_WITH_SAX1:
536
0
#ifdef LIBXML_SAX1_ENABLED
537
0
            return(1);
538
#else
539
            return(0);
540
#endif
541
0
        case XML_WITH_HTTP:
542
0
            return(0);
543
0
        case XML_WITH_VALID:
544
0
#ifdef LIBXML_VALID_ENABLED
545
0
            return(1);
546
#else
547
            return(0);
548
#endif
549
0
        case XML_WITH_HTML:
550
0
#ifdef LIBXML_HTML_ENABLED
551
0
            return(1);
552
#else
553
            return(0);
554
#endif
555
0
        case XML_WITH_LEGACY:
556
0
            return(0);
557
0
        case XML_WITH_C14N:
558
0
#ifdef LIBXML_C14N_ENABLED
559
0
            return(1);
560
#else
561
            return(0);
562
#endif
563
0
        case XML_WITH_CATALOG:
564
0
#ifdef LIBXML_CATALOG_ENABLED
565
0
            return(1);
566
#else
567
            return(0);
568
#endif
569
0
        case XML_WITH_XPATH:
570
0
#ifdef LIBXML_XPATH_ENABLED
571
0
            return(1);
572
#else
573
            return(0);
574
#endif
575
0
        case XML_WITH_XPTR:
576
0
#ifdef LIBXML_XPTR_ENABLED
577
0
            return(1);
578
#else
579
            return(0);
580
#endif
581
0
        case XML_WITH_XINCLUDE:
582
0
#ifdef LIBXML_XINCLUDE_ENABLED
583
0
            return(1);
584
#else
585
            return(0);
586
#endif
587
0
        case XML_WITH_ICONV:
588
0
#ifdef LIBXML_ICONV_ENABLED
589
0
            return(1);
590
#else
591
            return(0);
592
#endif
593
0
        case XML_WITH_ISO8859X:
594
0
#ifdef LIBXML_ISO8859X_ENABLED
595
0
            return(1);
596
#else
597
            return(0);
598
#endif
599
0
        case XML_WITH_UNICODE:
600
0
            return(0);
601
0
        case XML_WITH_REGEXP:
602
0
#ifdef LIBXML_REGEXP_ENABLED
603
0
            return(1);
604
#else
605
            return(0);
606
#endif
607
0
        case XML_WITH_AUTOMATA:
608
0
#ifdef LIBXML_REGEXP_ENABLED
609
0
            return(1);
610
#else
611
            return(0);
612
#endif
613
0
        case XML_WITH_EXPR:
614
0
            return(0);
615
0
        case XML_WITH_RELAXNG:
616
0
#ifdef LIBXML_RELAXNG_ENABLED
617
0
            return(1);
618
#else
619
            return(0);
620
#endif
621
0
        case XML_WITH_SCHEMAS:
622
0
#ifdef LIBXML_SCHEMAS_ENABLED
623
0
            return(1);
624
#else
625
            return(0);
626
#endif
627
0
        case XML_WITH_SCHEMATRON:
628
#ifdef LIBXML_SCHEMATRON_ENABLED
629
            return(1);
630
#else
631
0
            return(0);
632
0
#endif
633
0
        case XML_WITH_MODULES:
634
0
#ifdef LIBXML_MODULES_ENABLED
635
0
            return(1);
636
#else
637
            return(0);
638
#endif
639
0
        case XML_WITH_DEBUG:
640
#ifdef LIBXML_DEBUG_ENABLED
641
            return(1);
642
#else
643
0
            return(0);
644
0
#endif
645
0
        case XML_WITH_DEBUG_MEM:
646
0
            return(0);
647
0
        case XML_WITH_ZLIB:
648
0
#ifdef LIBXML_ZLIB_ENABLED
649
0
            return(1);
650
#else
651
            return(0);
652
#endif
653
0
        case XML_WITH_LZMA:
654
0
            return(0);
655
0
        case XML_WITH_ICU:
656
#ifdef LIBXML_ICU_ENABLED
657
            return(1);
658
#else
659
0
            return(0);
660
0
#endif
661
0
        default:
662
0
      break;
663
0
     }
664
0
     return(0);
665
0
}
666
667
/************************************************************************
668
 *                  *
669
 *      Simple string buffer        *
670
 *                  *
671
 ************************************************************************/
672
673
typedef struct {
674
    xmlChar *mem;
675
    unsigned size;
676
    unsigned cap; /* size < cap */
677
    unsigned max; /* size <= max */
678
    xmlParserErrors code;
679
} xmlSBuf;
680
681
static void
682
918k
xmlSBufInit(xmlSBuf *buf, unsigned max) {
683
918k
    buf->mem = NULL;
684
918k
    buf->size = 0;
685
918k
    buf->cap = 0;
686
918k
    buf->max = max;
687
918k
    buf->code = XML_ERR_OK;
688
918k
}
689
690
static int
691
902k
xmlSBufGrow(xmlSBuf *buf, unsigned len) {
692
902k
    xmlChar *mem;
693
902k
    unsigned cap;
694
695
902k
    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
902k
    cap = (buf->size + len) * 2;
702
902k
    if (cap < 240)
703
853k
        cap = 240;
704
705
902k
    mem = xmlRealloc(buf->mem, cap);
706
902k
    if (mem == NULL) {
707
0
        buf->code = XML_ERR_NO_MEMORY;
708
0
        return(-1);
709
0
    }
710
711
902k
    buf->mem = mem;
712
902k
    buf->cap = cap;
713
714
902k
    return(0);
715
902k
}
716
717
static void
718
4.51M
xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) {
719
4.51M
    if (buf->max - buf->size < len) {
720
0
        if (buf->code == XML_ERR_OK)
721
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
722
0
        return;
723
0
    }
724
725
4.51M
    if (buf->cap - buf->size <= len) {
726
901k
        if (xmlSBufGrow(buf, len) < 0)
727
0
            return;
728
901k
    }
729
730
4.51M
    if (len > 0)
731
4.51M
        memcpy(buf->mem + buf->size, str, len);
732
4.51M
    buf->size += len;
733
4.51M
}
734
735
static void
736
2.34M
xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) {
737
2.34M
    xmlSBufAddString(buf, (const xmlChar *) str, len);
738
2.34M
}
739
740
static void
741
57.0k
xmlSBufAddChar(xmlSBuf *buf, int c) {
742
57.0k
    xmlChar *end;
743
744
57.0k
    if (buf->max - buf->size < 4) {
745
0
        if (buf->code == XML_ERR_OK)
746
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
747
0
        return;
748
0
    }
749
750
57.0k
    if (buf->cap - buf->size <= 4) {
751
1.74k
        if (xmlSBufGrow(buf, 4) < 0)
752
0
            return;
753
1.74k
    }
754
755
57.0k
    end = buf->mem + buf->size;
756
757
57.0k
    if (c < 0x80) {
758
33.7k
        *end = (xmlChar) c;
759
33.7k
        buf->size += 1;
760
33.7k
    } else {
761
23.2k
        buf->size += xmlCopyCharMultiByte(end, c);
762
23.2k
    }
763
57.0k
}
764
765
static void
766
3.16k
xmlSBufAddReplChar(xmlSBuf *buf) {
767
3.16k
    xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3);
768
3.16k
}
769
770
static void
771
0
xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
772
0
    if (buf->code == XML_ERR_NO_MEMORY)
773
0
        xmlCtxtErrMemory(ctxt);
774
0
    else
775
0
        xmlFatalErr(ctxt, buf->code, errMsg);
776
0
}
777
778
static xmlChar *
779
xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt,
780
913k
              const char *errMsg) {
781
913k
    if (buf->mem == NULL) {
782
33.9k
        buf->mem = xmlMalloc(1);
783
33.9k
        if (buf->mem == NULL) {
784
0
            buf->code = XML_ERR_NO_MEMORY;
785
33.9k
        } else {
786
33.9k
            buf->mem[0] = 0;
787
33.9k
        }
788
879k
    } else {
789
879k
        buf->mem[buf->size] = 0;
790
879k
    }
791
792
913k
    if (buf->code == XML_ERR_OK) {
793
913k
        if (sizeOut != NULL)
794
0
            *sizeOut = buf->size;
795
913k
        return(buf->mem);
796
913k
    }
797
798
0
    xmlSBufReportError(buf, ctxt, errMsg);
799
800
0
    xmlFree(buf->mem);
801
802
0
    if (sizeOut != NULL)
803
0
        *sizeOut = 0;
804
0
    return(NULL);
805
913k
}
806
807
static void
808
4.17k
xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
809
4.17k
    if (buf->code != XML_ERR_OK)
810
0
        xmlSBufReportError(buf, ctxt, errMsg);
811
812
4.17k
    xmlFree(buf->mem);
813
4.17k
}
814
815
static int
816
xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str,
817
135M
                    const char *errMsg) {
818
135M
    int c = str[0];
819
135M
    int c1 = str[1];
820
821
135M
    if ((c1 & 0xC0) != 0x80)
822
504
        goto encoding_error;
823
824
135M
    if (c < 0xE0) {
825
        /* 2-byte sequence */
826
43.7M
        if (c < 0xC2)
827
98
            goto encoding_error;
828
829
43.7M
        return(2);
830
91.9M
    } else {
831
91.9M
        int c2 = str[2];
832
833
91.9M
        if ((c2 & 0xC0) != 0x80)
834
260
            goto encoding_error;
835
836
91.9M
        if (c < 0xF0) {
837
            /* 3-byte sequence */
838
91.9M
            if (c == 0xE0) {
839
                /* overlong */
840
1.34k
                if (c1 < 0xA0)
841
2
                    goto encoding_error;
842
91.9M
            } else if (c == 0xED) {
843
                /* surrogate */
844
530
                if (c1 >= 0xA0)
845
3
                    goto encoding_error;
846
91.9M
            } else if (c == 0xEF) {
847
                /* U+FFFE and U+FFFF are invalid Chars */
848
10.8k
                if ((c1 == 0xBF) && (c2 >= 0xBE))
849
3
                    xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg);
850
10.8k
            }
851
852
91.9M
            return(3);
853
91.9M
        } else {
854
            /* 4-byte sequence */
855
8.56k
            if ((str[3] & 0xC0) != 0x80)
856
9
                goto encoding_error;
857
8.55k
            if (c == 0xF0) {
858
                /* overlong */
859
1.70k
                if (c1 < 0x90)
860
2
                    goto encoding_error;
861
6.85k
            } else if (c >= 0xF4) {
862
                /* greater than 0x10FFFF */
863
1.52k
                if ((c > 0xF4) || (c1 >= 0x90))
864
5
                    goto encoding_error;
865
1.52k
            }
866
867
8.54k
            return(4);
868
8.55k
        }
869
91.9M
    }
870
871
883
encoding_error:
872
    /* Only report the first error */
873
883
    if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) {
874
883
        xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL);
875
883
        ctxt->input->flags |= XML_INPUT_ENCODING_ERROR;
876
883
    }
877
878
883
    return(0);
879
135M
}
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
122k
xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) {
897
122k
    xmlSAXHandlerPtr sax;
898
899
    /* Avoid unused variable warning if features are disabled. */
900
122k
    (void) sax;
901
902
    /*
903
     * Changing the SAX struct directly is still widespread practice
904
     * in internal and external code.
905
     */
906
122k
    if (ctxt == NULL) return;
907
122k
    sax = ctxt->sax;
908
122k
#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
122k
    if (((ctxt->options & XML_PARSE_SAX1) == 0) &&
914
122k
        (sax) &&
915
122k
        (sax->initialized == XML_SAX2_MAGIC) &&
916
0
        ((sax->startElementNs != NULL) ||
917
0
         (sax->endElementNs != NULL) ||
918
0
         ((sax->startElement == NULL) && (sax->endElement == NULL))))
919
0
        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
122k
    ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3);
929
122k
    ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5);
930
122k
    ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36);
931
122k
    if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) ||
932
122k
    (ctxt->str_xml_ns == NULL)) {
933
0
        xmlErrMemory(ctxt);
934
0
    }
935
936
122k
    xmlDictSetLimit(ctxt->dict,
937
122k
                    (ctxt->options & XML_PARSE_HUGE) ?
938
0
                        0 :
939
122k
                        XML_MAX_DICTIONARY_LIMIT);
940
941
122k
#ifdef LIBXML_VALID_ENABLED
942
122k
    if (ctxt->validate)
943
0
        ctxt->vctxt.flags |= XML_VCTXT_VALIDATE;
944
122k
    else
945
122k
        ctxt->vctxt.flags &= ~XML_VCTXT_VALIDATE;
946
122k
#endif /* LIBXML_VALID_ENABLED */
947
122k
}
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
25.6k
{
989
25.6k
    if ((src == NULL) || (dst == NULL))
990
0
        return(NULL);
991
992
32.2k
    while (*src == 0x20) src++;
993
85.1k
    while (*src != 0) {
994
59.5k
  if (*src == 0x20) {
995
25.9k
      while (*src == 0x20) src++;
996
3.95k
      if (*src != 0)
997
2.90k
    *dst++ = 0x20;
998
55.6k
  } else {
999
55.6k
      *dst++ = *src++;
1000
55.6k
  }
1001
59.5k
    }
1002
25.6k
    *dst = 0;
1003
25.6k
    if (dst == src)
1004
23.3k
       return(NULL);
1005
2.26k
    return(dst);
1006
25.6k
}
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
0
               const xmlChar *value) {
1021
0
    xmlDefAttrsPtr defaults;
1022
0
    xmlDefAttr *attr;
1023
0
    int len, expandedSize;
1024
0
    xmlHashedString name;
1025
0
    xmlHashedString prefix;
1026
0
    xmlHashedString hvalue;
1027
0
    const xmlChar *localname;
1028
1029
    /*
1030
     * Allows to detect attribute redefinitions
1031
     */
1032
0
    if (ctxt->attsSpecial != NULL) {
1033
0
        if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL)
1034
0
      return;
1035
0
    }
1036
1037
0
    if (ctxt->attsDefault == NULL) {
1038
0
        ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict);
1039
0
  if (ctxt->attsDefault == NULL)
1040
0
      goto mem_error;
1041
0
    }
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
0
    localname = xmlSplitQName3(fullname, &len);
1048
0
    if (localname == NULL) {
1049
0
        name = xmlDictLookupHashed(ctxt->dict, fullname, -1);
1050
0
  prefix.name = NULL;
1051
0
    } else {
1052
0
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1053
0
  prefix = xmlDictLookupHashed(ctxt->dict, fullname, len);
1054
0
        if (prefix.name == NULL)
1055
0
            goto mem_error;
1056
0
    }
1057
0
    if (name.name == NULL)
1058
0
        goto mem_error;
1059
1060
    /*
1061
     * make sure there is some storage
1062
     */
1063
0
    defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name);
1064
0
    if ((defaults == NULL) ||
1065
0
        (defaults->nbAttrs >= defaults->maxAttrs)) {
1066
0
        xmlDefAttrsPtr temp;
1067
0
        int newSize;
1068
1069
0
        if (defaults == NULL) {
1070
0
            newSize = 4;
1071
0
        } else {
1072
0
            if ((defaults->maxAttrs >= XML_MAX_ATTRS) ||
1073
0
                ((size_t) defaults->maxAttrs >
1074
0
                     SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults)))
1075
0
                goto mem_error;
1076
1077
0
            if (defaults->maxAttrs > XML_MAX_ATTRS / 2)
1078
0
                newSize = XML_MAX_ATTRS;
1079
0
            else
1080
0
                newSize = defaults->maxAttrs * 2;
1081
0
        }
1082
0
        temp = xmlRealloc(defaults,
1083
0
                          sizeof(*defaults) + newSize * sizeof(xmlDefAttr));
1084
0
  if (temp == NULL)
1085
0
      goto mem_error;
1086
0
        if (defaults == NULL)
1087
0
            temp->nbAttrs = 0;
1088
0
  temp->maxAttrs = newSize;
1089
0
        defaults = temp;
1090
0
  if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name,
1091
0
                          defaults, NULL) < 0) {
1092
0
      xmlFree(defaults);
1093
0
      goto mem_error;
1094
0
  }
1095
0
    }
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
0
    localname = xmlSplitQName3(fullattr, &len);
1102
0
    if (localname == NULL) {
1103
0
        name = xmlDictLookupHashed(ctxt->dict, fullattr, -1);
1104
0
  prefix.name = NULL;
1105
0
    } else {
1106
0
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1107
0
  prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len);
1108
0
        if (prefix.name == NULL)
1109
0
            goto mem_error;
1110
0
    }
1111
0
    if (name.name == NULL)
1112
0
        goto mem_error;
1113
1114
    /* intern the string and precompute the end */
1115
0
    len = strlen((const char *) value);
1116
0
    hvalue = xmlDictLookupHashed(ctxt->dict, value, len);
1117
0
    if (hvalue.name == NULL)
1118
0
        goto mem_error;
1119
1120
0
    expandedSize = strlen((const char *) name.name);
1121
0
    if (prefix.name != NULL)
1122
0
        expandedSize += strlen((const char *) prefix.name);
1123
0
    expandedSize += len;
1124
1125
0
    attr = &defaults->attrs[defaults->nbAttrs++];
1126
0
    attr->name = name;
1127
0
    attr->prefix = prefix;
1128
0
    attr->value = hvalue;
1129
0
    attr->valueEnd = hvalue.name + len;
1130
0
    attr->external = PARSER_EXTERNAL(ctxt);
1131
0
    attr->expandedSize = expandedSize;
1132
1133
0
    return;
1134
1135
0
mem_error:
1136
0
    xmlErrMemory(ctxt);
1137
0
}
1138
1139
/**
1140
 * Register this attribute type
1141
 *
1142
 * @param ctxt  an XML parser context
1143
 * @param fullname  the element fullname
1144
 * @param fullattr  the attribute fullname
1145
 * @param type  the attribute type
1146
 */
1147
static void
1148
xmlAddSpecialAttr(xmlParserCtxtPtr ctxt,
1149
      const xmlChar *fullname,
1150
      const xmlChar *fullattr,
1151
      int type)
1152
0
{
1153
0
    if (ctxt->attsSpecial == NULL) {
1154
0
        ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict);
1155
0
  if (ctxt->attsSpecial == NULL)
1156
0
      goto mem_error;
1157
0
    }
1158
1159
0
    if (PARSER_EXTERNAL(ctxt))
1160
0
        type |= XML_SPECIAL_EXTERNAL;
1161
1162
0
    if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr,
1163
0
                    XML_INT_TO_PTR(type)) < 0)
1164
0
        goto mem_error;
1165
0
    return;
1166
1167
0
mem_error:
1168
0
    xmlErrMemory(ctxt);
1169
0
}
1170
1171
/**
1172
 * Removes CDATA attributes from the special attribute table
1173
 */
1174
static void
1175
xmlCleanSpecialAttrCallback(void *payload, void *data,
1176
                            const xmlChar *fullname, const xmlChar *fullattr,
1177
0
                            const xmlChar *unused ATTRIBUTE_UNUSED) {
1178
0
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data;
1179
1180
0
    if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) {
1181
0
        xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL);
1182
0
    }
1183
0
}
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
17.0k
{
1195
17.0k
    if (ctxt->attsSpecial == NULL)
1196
17.0k
        return;
1197
1198
0
    xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt);
1199
1200
0
    if (xmlHashSize(ctxt->attsSpecial) == 0) {
1201
0
        xmlHashFree(ctxt->attsSpecial, NULL);
1202
0
        ctxt->attsSpecial = NULL;
1203
0
    }
1204
0
}
1205
1206
/**
1207
 * Checks that the value conforms to the LanguageID production:
1208
 *
1209
 * @deprecated Internal function, do not use.
1210
 *
1211
 * NOTE: this is somewhat deprecated, those productions were removed from
1212
 * the XML Second edition.
1213
 *
1214
 *     [33] LanguageID ::= Langcode ('-' Subcode)*
1215
 *     [34] Langcode ::= ISO639Code |  IanaCode |  UserCode
1216
 *     [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z])
1217
 *     [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+
1218
 *     [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+
1219
 *     [38] Subcode ::= ([a-z] | [A-Z])+
1220
 *
1221
 * The current REC reference the successors of RFC 1766, currently 5646
1222
 *
1223
 * http://www.rfc-editor.org/rfc/rfc5646.txt
1224
 *
1225
 *     langtag       = language
1226
 *                     ["-" script]
1227
 *                     ["-" region]
1228
 *                     *("-" variant)
1229
 *                     *("-" extension)
1230
 *                     ["-" privateuse]
1231
 *     language      = 2*3ALPHA            ; shortest ISO 639 code
1232
 *                     ["-" extlang]       ; sometimes followed by
1233
 *                                         ; extended language subtags
1234
 *                   / 4ALPHA              ; or reserved for future use
1235
 *                   / 5*8ALPHA            ; or registered language subtag
1236
 *
1237
 *     extlang       = 3ALPHA              ; selected ISO 639 codes
1238
 *                     *2("-" 3ALPHA)      ; permanently reserved
1239
 *
1240
 *     script        = 4ALPHA              ; ISO 15924 code
1241
 *
1242
 *     region        = 2ALPHA              ; ISO 3166-1 code
1243
 *                   / 3DIGIT              ; UN M.49 code
1244
 *
1245
 *     variant       = 5*8alphanum         ; registered variants
1246
 *                   / (DIGIT 3alphanum)
1247
 *
1248
 *     extension     = singleton 1*("-" (2*8alphanum))
1249
 *
1250
 *                                         ; Single alphanumerics
1251
 *                                         ; "x" reserved for private use
1252
 *     singleton     = DIGIT               ; 0 - 9
1253
 *                   / %x41-57             ; A - W
1254
 *                   / %x59-5A             ; Y - Z
1255
 *                   / %x61-77             ; a - w
1256
 *                   / %x79-7A             ; y - z
1257
 *
1258
 * it sounds right to still allow Irregular i-xxx IANA and user codes too
1259
 * The parser below doesn't try to cope with extension or privateuse
1260
 * that could be added but that's not interoperable anyway
1261
 *
1262
 * @param lang  pointer to the string value
1263
 * @returns 1 if correct 0 otherwise
1264
 **/
1265
int
1266
xmlCheckLanguageID(const xmlChar * lang)
1267
0
{
1268
0
    const xmlChar *cur = lang, *nxt;
1269
1270
0
    if (cur == NULL)
1271
0
        return (0);
1272
0
    if (((cur[0] == 'i') && (cur[1] == '-')) ||
1273
0
        ((cur[0] == 'I') && (cur[1] == '-')) ||
1274
0
        ((cur[0] == 'x') && (cur[1] == '-')) ||
1275
0
        ((cur[0] == 'X') && (cur[1] == '-'))) {
1276
        /*
1277
         * Still allow IANA code and user code which were coming
1278
         * from the previous version of the XML-1.0 specification
1279
         * it's deprecated but we should not fail
1280
         */
1281
0
        cur += 2;
1282
0
        while (((cur[0] >= 'A') && (cur[0] <= 'Z')) ||
1283
0
               ((cur[0] >= 'a') && (cur[0] <= 'z')))
1284
0
            cur++;
1285
0
        return(cur[0] == 0);
1286
0
    }
1287
0
    nxt = cur;
1288
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1289
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1290
0
           nxt++;
1291
0
    if (nxt - cur >= 4) {
1292
        /*
1293
         * Reserved
1294
         */
1295
0
        if ((nxt - cur > 8) || (nxt[0] != 0))
1296
0
            return(0);
1297
0
        return(1);
1298
0
    }
1299
0
    if (nxt - cur < 2)
1300
0
        return(0);
1301
    /* we got an ISO 639 code */
1302
0
    if (nxt[0] == 0)
1303
0
        return(1);
1304
0
    if (nxt[0] != '-')
1305
0
        return(0);
1306
1307
0
    nxt++;
1308
0
    cur = nxt;
1309
    /* now we can have extlang or script or region or variant */
1310
0
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1311
0
        goto region_m49;
1312
1313
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1314
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1315
0
           nxt++;
1316
0
    if (nxt - cur == 4)
1317
0
        goto script;
1318
0
    if (nxt - cur == 2)
1319
0
        goto region;
1320
0
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1321
0
        goto variant;
1322
0
    if (nxt - cur != 3)
1323
0
        return(0);
1324
    /* we parsed an extlang */
1325
0
    if (nxt[0] == 0)
1326
0
        return(1);
1327
0
    if (nxt[0] != '-')
1328
0
        return(0);
1329
1330
0
    nxt++;
1331
0
    cur = nxt;
1332
    /* now we can have script or region or variant */
1333
0
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1334
0
        goto region_m49;
1335
1336
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1337
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1338
0
           nxt++;
1339
0
    if (nxt - cur == 2)
1340
0
        goto region;
1341
0
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1342
0
        goto variant;
1343
0
    if (nxt - cur != 4)
1344
0
        return(0);
1345
    /* we parsed a script */
1346
0
script:
1347
0
    if (nxt[0] == 0)
1348
0
        return(1);
1349
0
    if (nxt[0] != '-')
1350
0
        return(0);
1351
1352
0
    nxt++;
1353
0
    cur = nxt;
1354
    /* now we can have region or variant */
1355
0
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1356
0
        goto region_m49;
1357
1358
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1359
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1360
0
           nxt++;
1361
1362
0
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1363
0
        goto variant;
1364
0
    if (nxt - cur != 2)
1365
0
        return(0);
1366
    /* we parsed a region */
1367
0
region:
1368
0
    if (nxt[0] == 0)
1369
0
        return(1);
1370
0
    if (nxt[0] != '-')
1371
0
        return(0);
1372
1373
0
    nxt++;
1374
0
    cur = nxt;
1375
    /* now we can just have a variant */
1376
0
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1377
0
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1378
0
           nxt++;
1379
1380
0
    if ((nxt - cur < 5) || (nxt - cur > 8))
1381
0
        return(0);
1382
1383
    /* we parsed a variant */
1384
0
variant:
1385
0
    if (nxt[0] == 0)
1386
0
        return(1);
1387
0
    if (nxt[0] != '-')
1388
0
        return(0);
1389
    /* extensions and private use subtags not checked */
1390
0
    return (1);
1391
1392
0
region_m49:
1393
0
    if (((nxt[1] >= '0') && (nxt[1] <= '9')) &&
1394
0
        ((nxt[2] >= '0') && (nxt[2] <= '9'))) {
1395
0
        nxt += 3;
1396
0
        goto region;
1397
0
    }
1398
0
    return(0);
1399
0
}
1400
1401
/************************************************************************
1402
 *                  *
1403
 *    Parser stacks related functions and macros    *
1404
 *                  *
1405
 ************************************************************************/
1406
1407
static xmlChar *
1408
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar **str);
1409
1410
/**
1411
 * Create a new namespace database.
1412
 *
1413
 * @returns the new obejct.
1414
 */
1415
xmlParserNsData *
1416
122k
xmlParserNsCreate(void) {
1417
122k
    xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb));
1418
1419
122k
    if (nsdb == NULL)
1420
0
        return(NULL);
1421
122k
    memset(nsdb, 0, sizeof(*nsdb));
1422
122k
    nsdb->defaultNsIndex = INT_MAX;
1423
1424
122k
    return(nsdb);
1425
122k
}
1426
1427
/**
1428
 * Free a namespace database.
1429
 *
1430
 * @param nsdb  namespace database
1431
 */
1432
void
1433
122k
xmlParserNsFree(xmlParserNsData *nsdb) {
1434
122k
    if (nsdb == NULL)
1435
0
        return;
1436
1437
122k
    xmlFree(nsdb->extra);
1438
122k
    xmlFree(nsdb->hash);
1439
122k
    xmlFree(nsdb);
1440
122k
}
1441
1442
/**
1443
 * Reset a namespace database.
1444
 *
1445
 * @param nsdb  namespace database
1446
 */
1447
static void
1448
0
xmlParserNsReset(xmlParserNsData *nsdb) {
1449
0
    if (nsdb == NULL)
1450
0
        return;
1451
1452
0
    nsdb->hashElems = 0;
1453
0
    nsdb->elementId = 0;
1454
0
    nsdb->defaultNsIndex = INT_MAX;
1455
1456
0
    if (nsdb->hash)
1457
0
        memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0]));
1458
0
}
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
0
xmlParserNsStartElement(xmlParserNsData *nsdb) {
1468
0
    if (nsdb->elementId == UINT_MAX)
1469
0
        return(-1);
1470
0
    nsdb->elementId++;
1471
1472
0
    return(0);
1473
0
}
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
0
                  xmlParserNsBucket **bucketPtr) {
1489
0
    xmlParserNsBucket *bucket, *tombstone;
1490
0
    unsigned index, hashValue;
1491
1492
0
    if (prefix->name == NULL)
1493
0
        return(ctxt->nsdb->defaultNsIndex);
1494
1495
0
    if (ctxt->nsdb->hashSize == 0)
1496
0
        return(INT_MAX);
1497
1498
0
    hashValue = prefix->hashValue;
1499
0
    index = hashValue & (ctxt->nsdb->hashSize - 1);
1500
0
    bucket = &ctxt->nsdb->hash[index];
1501
0
    tombstone = NULL;
1502
1503
0
    while (bucket->hashValue) {
1504
0
        if (bucket->index == INT_MAX) {
1505
0
            if (tombstone == NULL)
1506
0
                tombstone = bucket;
1507
0
        } else if (bucket->hashValue == hashValue) {
1508
0
            if (ctxt->nsTab[bucket->index * 2] == prefix->name) {
1509
0
                if (bucketPtr != NULL)
1510
0
                    *bucketPtr = bucket;
1511
0
                return(bucket->index);
1512
0
            }
1513
0
        }
1514
1515
0
        index++;
1516
0
        bucket++;
1517
0
        if (index == ctxt->nsdb->hashSize) {
1518
0
            index = 0;
1519
0
            bucket = ctxt->nsdb->hash;
1520
0
        }
1521
0
    }
1522
1523
0
    if (bucketPtr != NULL)
1524
0
        *bucketPtr = tombstone ? tombstone : bucket;
1525
0
    return(INT_MAX);
1526
0
}
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
0
xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) {
1537
0
    const xmlChar *ret;
1538
0
    int nsIndex;
1539
1540
0
    if (prefix->name == ctxt->str_xml)
1541
0
        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
0
    nsIndex = xmlParserNsLookup(ctxt, prefix, NULL);
1548
0
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1549
0
        return(NULL);
1550
1551
0
    ret = ctxt->nsTab[nsIndex * 2 + 1];
1552
0
    if (ret[0] == 0)
1553
0
        ret = NULL;
1554
0
    return(ret);
1555
0
}
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
0
xmlParserNsLookupSax(xmlParserCtxt *ctxt, const xmlChar *prefix) {
1567
0
    xmlHashedString hprefix;
1568
0
    int nsIndex;
1569
1570
0
    if (prefix == ctxt->str_xml)
1571
0
        return(NULL);
1572
1573
0
    hprefix.name = prefix;
1574
0
    if (prefix != NULL)
1575
0
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1576
0
    else
1577
0
        hprefix.hashValue = 0;
1578
0
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1579
0
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1580
0
        return(NULL);
1581
1582
0
    return(ctxt->nsdb->extra[nsIndex].saxData);
1583
0
}
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
0
                     void *saxData) {
1598
0
    xmlHashedString hprefix;
1599
0
    int nsIndex;
1600
1601
0
    if (prefix == ctxt->str_xml)
1602
0
        return(-1);
1603
1604
0
    hprefix.name = prefix;
1605
0
    if (prefix != NULL)
1606
0
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1607
0
    else
1608
0
        hprefix.hashValue = 0;
1609
0
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1610
0
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1611
0
        return(-1);
1612
1613
0
    ctxt->nsdb->extra[nsIndex].saxData = saxData;
1614
0
    return(0);
1615
0
}
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
0
xmlParserNsGrow(xmlParserCtxtPtr ctxt) {
1625
0
    const xmlChar **table;
1626
0
    xmlParserNsExtra *extra;
1627
0
    int newSize;
1628
1629
0
    newSize = xmlGrowCapacity(ctxt->nsMax,
1630
0
                              sizeof(table[0]) + sizeof(extra[0]),
1631
0
                              16, XML_MAX_ITEMS);
1632
0
    if (newSize < 0)
1633
0
        goto error;
1634
1635
0
    table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0]));
1636
0
    if (table == NULL)
1637
0
        goto error;
1638
0
    ctxt->nsTab = table;
1639
1640
0
    extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0]));
1641
0
    if (extra == NULL)
1642
0
        goto error;
1643
0
    ctxt->nsdb->extra = extra;
1644
1645
0
    ctxt->nsMax = newSize;
1646
0
    return(0);
1647
1648
0
error:
1649
0
    xmlErrMemory(ctxt);
1650
0
    return(-1);
1651
0
}
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
0
                const xmlHashedString *uri, void *saxData, int defAttr) {
1667
0
    xmlParserNsBucket *bucket = NULL;
1668
0
    xmlParserNsExtra *extra;
1669
0
    const xmlChar **ns;
1670
0
    unsigned hashValue, nsIndex, oldIndex;
1671
1672
0
    if ((prefix != NULL) && (prefix->name == ctxt->str_xml))
1673
0
        return(0);
1674
1675
0
    if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) {
1676
0
        xmlErrMemory(ctxt);
1677
0
        return(-1);
1678
0
    }
1679
1680
    /*
1681
     * Default namespace and 'xml' namespace
1682
     */
1683
0
    if ((prefix == NULL) || (prefix->name == NULL)) {
1684
0
        oldIndex = ctxt->nsdb->defaultNsIndex;
1685
1686
0
        if (oldIndex != INT_MAX) {
1687
0
            extra = &ctxt->nsdb->extra[oldIndex];
1688
1689
0
            if (extra->elementId == ctxt->nsdb->elementId) {
1690
0
                if (defAttr == 0)
1691
0
                    xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns");
1692
0
                return(0);
1693
0
            }
1694
1695
0
            if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1696
0
                (uri->name == ctxt->nsTab[oldIndex * 2 + 1]))
1697
0
                return(0);
1698
0
        }
1699
1700
0
        ctxt->nsdb->defaultNsIndex = ctxt->nsNr;
1701
0
        goto populate_entry;
1702
0
    }
1703
1704
    /*
1705
     * Hash table lookup
1706
     */
1707
0
    oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket);
1708
0
    if (oldIndex != INT_MAX) {
1709
0
        extra = &ctxt->nsdb->extra[oldIndex];
1710
1711
        /*
1712
         * Check for duplicate definitions on the same element.
1713
         */
1714
0
        if (extra->elementId == ctxt->nsdb->elementId) {
1715
0
            if (defAttr == 0)
1716
0
                xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name);
1717
0
            return(0);
1718
0
        }
1719
1720
0
        if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1721
0
            (uri->name == ctxt->nsTab[bucket->index * 2 + 1]))
1722
0
            return(0);
1723
1724
0
        bucket->index = ctxt->nsNr;
1725
0
        goto populate_entry;
1726
0
    }
1727
1728
    /*
1729
     * Insert new bucket
1730
     */
1731
1732
0
    hashValue = prefix->hashValue;
1733
1734
    /*
1735
     * Grow hash table, 50% fill factor
1736
     */
1737
0
    if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) {
1738
0
        xmlParserNsBucket *newHash;
1739
0
        unsigned newSize, i, index;
1740
1741
0
        if (ctxt->nsdb->hashSize > UINT_MAX / 2) {
1742
0
            xmlErrMemory(ctxt);
1743
0
            return(-1);
1744
0
        }
1745
0
        newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16;
1746
0
        newHash = xmlMalloc(newSize * sizeof(newHash[0]));
1747
0
        if (newHash == NULL) {
1748
0
            xmlErrMemory(ctxt);
1749
0
            return(-1);
1750
0
        }
1751
0
        memset(newHash, 0, newSize * sizeof(newHash[0]));
1752
1753
0
        for (i = 0; i < ctxt->nsdb->hashSize; i++) {
1754
0
            unsigned hv = ctxt->nsdb->hash[i].hashValue;
1755
0
            unsigned newIndex;
1756
1757
0
            if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX))
1758
0
                continue;
1759
0
            newIndex = hv & (newSize - 1);
1760
1761
0
            while (newHash[newIndex].hashValue != 0) {
1762
0
                newIndex++;
1763
0
                if (newIndex == newSize)
1764
0
                    newIndex = 0;
1765
0
            }
1766
1767
0
            newHash[newIndex] = ctxt->nsdb->hash[i];
1768
0
        }
1769
1770
0
        xmlFree(ctxt->nsdb->hash);
1771
0
        ctxt->nsdb->hash = newHash;
1772
0
        ctxt->nsdb->hashSize = newSize;
1773
1774
        /*
1775
         * Relookup
1776
         */
1777
0
        index = hashValue & (newSize - 1);
1778
1779
0
        while (newHash[index].hashValue != 0) {
1780
0
            index++;
1781
0
            if (index == newSize)
1782
0
                index = 0;
1783
0
        }
1784
1785
0
        bucket = &newHash[index];
1786
0
    }
1787
1788
0
    bucket->hashValue = hashValue;
1789
0
    bucket->index = ctxt->nsNr;
1790
0
    ctxt->nsdb->hashElems++;
1791
0
    oldIndex = INT_MAX;
1792
1793
0
populate_entry:
1794
0
    nsIndex = ctxt->nsNr;
1795
1796
0
    ns = &ctxt->nsTab[nsIndex * 2];
1797
0
    ns[0] = prefix ? prefix->name : NULL;
1798
0
    ns[1] = uri->name;
1799
1800
0
    extra = &ctxt->nsdb->extra[nsIndex];
1801
0
    extra->saxData = saxData;
1802
0
    extra->prefixHashValue = prefix ? prefix->hashValue : 0;
1803
0
    extra->uriHashValue = uri->hashValue;
1804
0
    extra->elementId = ctxt->nsdb->elementId;
1805
0
    extra->oldIndex = oldIndex;
1806
1807
0
    ctxt->nsNr++;
1808
1809
0
    return(1);
1810
0
}
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
0
{
1822
0
    int i;
1823
1824
    /* assert(nr <= ctxt->nsNr); */
1825
1826
0
    for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) {
1827
0
        const xmlChar *prefix = ctxt->nsTab[i * 2];
1828
0
        xmlParserNsExtra *extra = &ctxt->nsdb->extra[i];
1829
1830
0
        if (prefix == NULL) {
1831
0
            ctxt->nsdb->defaultNsIndex = extra->oldIndex;
1832
0
        } else {
1833
0
            xmlHashedString hprefix;
1834
0
            xmlParserNsBucket *bucket = NULL;
1835
1836
0
            hprefix.name = prefix;
1837
0
            hprefix.hashValue = extra->prefixHashValue;
1838
0
            xmlParserNsLookup(ctxt, &hprefix, &bucket);
1839
            /* assert(bucket && bucket->hashValue); */
1840
0
            bucket->index = extra->oldIndex;
1841
0
        }
1842
0
    }
1843
1844
0
    ctxt->nsNr -= nr;
1845
0
    return(nr);
1846
0
}
1847
1848
static int
1849
0
xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) {
1850
0
    const xmlChar **atts;
1851
0
    unsigned *attallocs;
1852
0
    int newSize;
1853
1854
0
    newSize = xmlGrowCapacity(ctxt->maxatts / 5,
1855
0
                              sizeof(atts[0]) * 5 + sizeof(attallocs[0]),
1856
0
                              10, XML_MAX_ATTRS);
1857
0
    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
0
    atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5);
1864
0
    if (atts == NULL)
1865
0
        goto mem_error;
1866
0
    ctxt->atts = atts;
1867
1868
0
    attallocs = xmlRealloc(ctxt->attallocs,
1869
0
                           newSize * sizeof(attallocs[0]));
1870
0
    if (attallocs == NULL)
1871
0
        goto mem_error;
1872
0
    ctxt->attallocs = attallocs;
1873
1874
0
    ctxt->maxatts = newSize * 5;
1875
1876
0
    return(0);
1877
1878
0
mem_error:
1879
0
    xmlErrMemory(ctxt);
1880
0
    return(-1);
1881
0
}
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
746k
{
1893
746k
    char *directory = NULL;
1894
746k
    int maxDepth;
1895
1896
746k
    if ((ctxt == NULL) || (value == NULL))
1897
0
        return(-1);
1898
1899
746k
    maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
1900
1901
746k
    if (ctxt->inputNr >= ctxt->inputMax) {
1902
7.50k
        xmlParserInputPtr *tmp;
1903
7.50k
        int newSize;
1904
1905
7.50k
        newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]),
1906
7.50k
                                  5, maxDepth);
1907
7.50k
        if (newSize < 0) {
1908
3
            xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
1909
3
                           "Maximum entity nesting depth exceeded");
1910
3
            return(-1);
1911
3
        }
1912
7.50k
        tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0]));
1913
7.50k
        if (tmp == NULL) {
1914
0
            xmlErrMemory(ctxt);
1915
0
            return(-1);
1916
0
        }
1917
7.50k
        ctxt->inputTab = tmp;
1918
7.50k
        ctxt->inputMax = newSize;
1919
7.50k
    }
1920
1921
746k
    if ((ctxt->inputNr == 0) && (value->filename != NULL)) {
1922
122k
        directory = xmlParserGetDirectory(value->filename);
1923
122k
        if (directory == NULL) {
1924
0
            xmlErrMemory(ctxt);
1925
0
            return(-1);
1926
0
        }
1927
122k
    }
1928
1929
746k
    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
746k
    ctxt->inputTab[ctxt->inputNr] = value;
1935
746k
    ctxt->input = value;
1936
1937
746k
    if (ctxt->inputNr == 0) {
1938
122k
        xmlFree(ctxt->directory);
1939
122k
        ctxt->directory = directory;
1940
122k
    }
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
746k
    value->id = ctxt->input_id++;
1948
1949
746k
    return(ctxt->inputNr++);
1950
746k
}
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
991k
{
1961
991k
    xmlParserInputPtr ret;
1962
1963
991k
    if (ctxt == NULL)
1964
0
        return(NULL);
1965
991k
    if (ctxt->inputNr <= 0)
1966
245k
        return (NULL);
1967
746k
    ctxt->inputNr--;
1968
746k
    if (ctxt->inputNr > 0)
1969
623k
        ctxt->input = ctxt->inputTab[ctxt->inputNr - 1];
1970
122k
    else
1971
122k
        ctxt->input = NULL;
1972
746k
    ret = ctxt->inputTab[ctxt->inputNr];
1973
746k
    ctxt->inputTab[ctxt->inputNr] = NULL;
1974
746k
    return (ret);
1975
991k
}
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
0
{
1989
0
    if (ctxt == NULL)
1990
0
        return(0);
1991
1992
0
    if (ctxt->nodeNr >= ctxt->nodeMax) {
1993
0
        int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
1994
0
        xmlNodePtr *tmp;
1995
0
        int newSize;
1996
1997
0
        newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]),
1998
0
                                  10, maxDepth);
1999
0
        if (newSize < 0) {
2000
0
            xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
2001
0
                    "Excessive depth in document: %d,"
2002
0
                    " use XML_PARSE_HUGE option\n",
2003
0
                    ctxt->nodeNr);
2004
0
            return(-1);
2005
0
        }
2006
2007
0
  tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0]));
2008
0
        if (tmp == NULL) {
2009
0
            xmlErrMemory(ctxt);
2010
0
            return (-1);
2011
0
        }
2012
0
        ctxt->nodeTab = tmp;
2013
0
  ctxt->nodeMax = newSize;
2014
0
    }
2015
2016
0
    ctxt->nodeTab[ctxt->nodeNr] = value;
2017
0
    ctxt->node = value;
2018
0
    return (ctxt->nodeNr++);
2019
0
}
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
15.6k
{
2032
15.6k
    xmlNodePtr ret;
2033
2034
15.6k
    if (ctxt == NULL) return(NULL);
2035
15.6k
    if (ctxt->nodeNr <= 0)
2036
15.6k
        return (NULL);
2037
0
    ctxt->nodeNr--;
2038
0
    if (ctxt->nodeNr > 0)
2039
0
        ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1];
2040
0
    else
2041
0
        ctxt->node = NULL;
2042
0
    ret = ctxt->nodeTab[ctxt->nodeNr];
2043
0
    ctxt->nodeTab[ctxt->nodeNr] = NULL;
2044
0
    return (ret);
2045
15.6k
}
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
2.26M
{
2062
2.26M
    xmlStartTag *tag;
2063
2064
2.26M
    if (ctxt->nameNr >= ctxt->nameMax) {
2065
105k
        const xmlChar **tmp;
2066
105k
        xmlStartTag *tmp2;
2067
105k
        int newSize;
2068
2069
105k
        newSize = xmlGrowCapacity(ctxt->nameMax,
2070
105k
                                  sizeof(tmp[0]) + sizeof(tmp2[0]),
2071
105k
                                  10, XML_MAX_ITEMS);
2072
105k
        if (newSize < 0)
2073
0
            goto mem_error;
2074
2075
105k
        tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0]));
2076
105k
        if (tmp == NULL)
2077
0
      goto mem_error;
2078
105k
  ctxt->nameTab = tmp;
2079
2080
105k
        tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0]));
2081
105k
        if (tmp2 == NULL)
2082
0
      goto mem_error;
2083
105k
  ctxt->pushTab = tmp2;
2084
2085
105k
        ctxt->nameMax = newSize;
2086
2.16M
    } else if (ctxt->pushTab == NULL) {
2087
46.7k
        ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0]));
2088
46.7k
        if (ctxt->pushTab == NULL)
2089
0
            goto mem_error;
2090
46.7k
    }
2091
2.26M
    ctxt->nameTab[ctxt->nameNr] = value;
2092
2.26M
    ctxt->name = value;
2093
2.26M
    tag = &ctxt->pushTab[ctxt->nameNr];
2094
2.26M
    tag->prefix = prefix;
2095
2.26M
    tag->URI = URI;
2096
2.26M
    tag->line = line;
2097
2.26M
    tag->nsNr = nsNr;
2098
2.26M
    return (ctxt->nameNr++);
2099
0
mem_error:
2100
0
    xmlErrMemory(ctxt);
2101
0
    return (-1);
2102
2.26M
}
2103
#ifdef LIBXML_PUSH_ENABLED
2104
/**
2105
 * Pops the top element/prefix/URI name from the name stack
2106
 *
2107
 * @param ctxt  an XML parser context
2108
 * @returns the name just removed
2109
 */
2110
static const xmlChar *
2111
nameNsPop(xmlParserCtxtPtr ctxt)
2112
0
{
2113
0
    const xmlChar *ret;
2114
2115
0
    if (ctxt->nameNr <= 0)
2116
0
        return (NULL);
2117
0
    ctxt->nameNr--;
2118
0
    if (ctxt->nameNr > 0)
2119
0
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2120
0
    else
2121
0
        ctxt->name = NULL;
2122
0
    ret = ctxt->nameTab[ctxt->nameNr];
2123
0
    ctxt->nameTab[ctxt->nameNr] = NULL;
2124
0
    return (ret);
2125
0
}
2126
#endif /* LIBXML_PUSH_ENABLED */
2127
2128
/**
2129
 * Pops the top element name from the name stack
2130
 *
2131
 * @deprecated Internal function, do not use.
2132
 *
2133
 * @param ctxt  an XML parser context
2134
 * @returns the name just removed
2135
 */
2136
static const xmlChar *
2137
namePop(xmlParserCtxtPtr ctxt)
2138
1.11M
{
2139
1.11M
    const xmlChar *ret;
2140
2141
1.11M
    if ((ctxt == NULL) || (ctxt->nameNr <= 0))
2142
0
        return (NULL);
2143
1.11M
    ctxt->nameNr--;
2144
1.11M
    if (ctxt->nameNr > 0)
2145
1.11M
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2146
2.78k
    else
2147
2.78k
        ctxt->name = NULL;
2148
1.11M
    ret = ctxt->nameTab[ctxt->nameNr];
2149
1.11M
    ctxt->nameTab[ctxt->nameNr] = NULL;
2150
1.11M
    return (ret);
2151
1.11M
}
2152
2153
2.83M
static int spacePush(xmlParserCtxtPtr ctxt, int val) {
2154
2.83M
    if (ctxt->spaceNr >= ctxt->spaceMax) {
2155
159k
        int *tmp;
2156
159k
        int newSize;
2157
2158
159k
        newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]),
2159
159k
                                  10, XML_MAX_ITEMS);
2160
159k
        if (newSize < 0) {
2161
0
      xmlErrMemory(ctxt);
2162
0
      return(-1);
2163
0
        }
2164
2165
159k
        tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0]));
2166
159k
        if (tmp == NULL) {
2167
0
      xmlErrMemory(ctxt);
2168
0
      return(-1);
2169
0
  }
2170
159k
  ctxt->spaceTab = tmp;
2171
2172
159k
        ctxt->spaceMax = newSize;
2173
159k
    }
2174
2.83M
    ctxt->spaceTab[ctxt->spaceNr] = val;
2175
2.83M
    ctxt->space = &ctxt->spaceTab[ctxt->spaceNr];
2176
2.83M
    return(ctxt->spaceNr++);
2177
2.83M
}
2178
2179
1.68M
static int spacePop(xmlParserCtxtPtr ctxt) {
2180
1.68M
    int ret;
2181
1.68M
    if (ctxt->spaceNr <= 0) return(0);
2182
1.68M
    ctxt->spaceNr--;
2183
1.68M
    if (ctxt->spaceNr > 0)
2184
1.68M
  ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1];
2185
0
    else
2186
0
        ctxt->space = &ctxt->spaceTab[0];
2187
1.68M
    ret = ctxt->spaceTab[ctxt->spaceNr];
2188
1.68M
    ctxt->spaceTab[ctxt->spaceNr] = -1;
2189
1.68M
    return(ret);
2190
1.68M
}
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
19.2M
#define RAW (*ctxt->input->cur)
2225
364M
#define CUR (*ctxt->input->cur)
2226
10.7M
#define NXT(val) ctxt->input->cur[(val)]
2227
781M
#define CUR_PTR ctxt->input->cur
2228
0
#define BASE_PTR ctxt->input->base
2229
2230
#define CMP4( s, c1, c2, c3, c4 ) \
2231
5.12M
  ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \
2232
2.62M
    ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 )
2233
#define CMP5( s, c1, c2, c3, c4, c5 ) \
2234
4.90M
  ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 )
2235
#define CMP6( s, c1, c2, c3, c4, c5, c6 ) \
2236
4.35M
  ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 )
2237
#define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \
2238
3.81M
  ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 )
2239
#define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \
2240
3.61M
  ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 )
2241
#define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \
2242
1.78M
  ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \
2243
1.78M
    ((unsigned char *) s)[ 8 ] == c9 )
2244
#define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \
2245
29.7k
  ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \
2246
29.7k
    ((unsigned char *) s)[ 9 ] == c10 )
2247
2248
3.83M
#define SKIP(val) do {             \
2249
3.83M
    ctxt->input->cur += (val),ctxt->input->col+=(val);      \
2250
3.83M
    if (*ctxt->input->cur == 0)           \
2251
3.83M
        xmlParserGrow(ctxt);           \
2252
3.83M
  } 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
5.74M
    if (!PARSER_PROGRESSIVE(ctxt)) \
2268
5.74M
  xmlParserShrink(ctxt);
2269
2270
#define GROW \
2271
18.1M
    if ((!PARSER_PROGRESSIVE(ctxt)) && \
2272
18.1M
        (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \
2273
5.88M
  xmlParserGrow(ctxt);
2274
2275
6.17M
#define SKIP_BLANKS xmlSkipBlankChars(ctxt)
2276
2277
324k
#define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt)
2278
2279
5.83M
#define NEXT xmlNextChar(ctxt)
2280
2281
2.47M
#define NEXT1 {               \
2282
2.47M
  ctxt->input->col++;           \
2283
2.47M
  ctxt->input->cur++;           \
2284
2.47M
  if (*ctxt->input->cur == 0)         \
2285
2.47M
      xmlParserGrow(ctxt);           \
2286
2.47M
    }
2287
2288
675M
#define NEXTL(l) do {             \
2289
675M
    if (*(ctxt->input->cur) == '\n') {         \
2290
2.98M
  ctxt->input->line++; ctxt->input->col = 1;      \
2291
672M
    } else ctxt->input->col++;           \
2292
675M
    ctxt->input->cur += l;        \
2293
675M
  } while (0)
2294
2295
#define COPY_BUF(b, i, v)           \
2296
305M
    if (v < 0x80) b[i++] = v;           \
2297
305M
    else i += xmlCopyCharMultiByte(&b[i],v)
2298
2299
static int
2300
305M
xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) {
2301
305M
    int c = xmlCurrentChar(ctxt, len);
2302
2303
305M
    if (c == XML_INVALID_CHAR)
2304
5.26M
        c = 0xFFFD; /* replacement character */
2305
2306
305M
    return(c);
2307
305M
}
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
6.46M
xmlSkipBlankChars(xmlParserCtxt *ctxt) {
2319
6.46M
    const xmlChar *cur;
2320
6.46M
    int res = 0;
2321
2322
6.46M
    cur = ctxt->input->cur;
2323
6.46M
    while (IS_BLANK_CH(*cur)) {
2324
4.24M
        if (*cur == '\n') {
2325
1.39M
            ctxt->input->line++; ctxt->input->col = 1;
2326
2.85M
        } else {
2327
2.85M
            ctxt->input->col++;
2328
2.85M
        }
2329
4.24M
        cur++;
2330
4.24M
        if (res < INT_MAX)
2331
4.24M
            res++;
2332
4.24M
        if (*cur == 0) {
2333
7.78k
            ctxt->input->cur = cur;
2334
7.78k
            xmlParserGrow(ctxt);
2335
7.78k
            cur = ctxt->input->cur;
2336
7.78k
        }
2337
4.24M
    }
2338
6.46M
    ctxt->input->cur = cur;
2339
2340
6.46M
    if (res > 4)
2341
24.4k
        GROW;
2342
2343
6.46M
    return(res);
2344
6.46M
}
2345
2346
static void
2347
1.82k
xmlPopPE(xmlParserCtxtPtr ctxt) {
2348
1.82k
    unsigned long consumed;
2349
1.82k
    xmlEntityPtr ent;
2350
2351
1.82k
    ent = ctxt->input->entity;
2352
2353
1.82k
    ent->flags &= ~XML_ENT_EXPANDING;
2354
2355
1.82k
    if ((ent->flags & XML_ENT_CHECKED) == 0) {
2356
230
        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
230
        do {
2363
230
            ctxt->input->cur = ctxt->input->end;
2364
230
            xmlParserShrink(ctxt);
2365
230
            result = xmlParserGrow(ctxt);
2366
230
        } while (result > 0);
2367
2368
230
        consumed = ctxt->input->consumed;
2369
230
        xmlSaturatedAddSizeT(&consumed,
2370
230
                             ctxt->input->end - ctxt->input->base);
2371
2372
230
        xmlSaturatedAdd(&ent->expandedSize, consumed);
2373
2374
        /*
2375
         * Add to sizeentities when parsing an external entity
2376
         * for the first time.
2377
         */
2378
230
        if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
2379
0
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
2380
0
        }
2381
2382
230
        ent->flags |= XML_ENT_CHECKED;
2383
230
    }
2384
2385
1.82k
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
2386
2387
1.82k
    xmlParserEntityCheck(ctxt, ent->expandedSize);
2388
2389
1.82k
    GROW;
2390
1.82k
}
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
324k
xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) {
2401
324k
    int res = 0;
2402
324k
    int inParam;
2403
324k
    int expandParam;
2404
2405
324k
    inParam = PARSER_IN_PE(ctxt);
2406
324k
    expandParam = PARSER_EXTERNAL(ctxt);
2407
2408
324k
    if (!inParam && !expandParam)
2409
286k
        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
78.1k
    while (PARSER_STOPPED(ctxt) == 0) {
2416
78.0k
        if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */
2417
39.8k
            NEXT;
2418
39.8k
        } else if (CUR == '%') {
2419
0
            if ((expandParam == 0) ||
2420
0
                (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0))
2421
0
                break;
2422
2423
            /*
2424
             * Expand parameter entity. We continue to consume
2425
             * whitespace at the start of the entity and possible
2426
             * even consume the whole entity and pop it. We might
2427
             * even pop multiple PEs in this loop.
2428
             */
2429
0
            xmlParsePERefInternal(ctxt, 0);
2430
2431
0
            inParam = PARSER_IN_PE(ctxt);
2432
0
            expandParam = PARSER_EXTERNAL(ctxt);
2433
38.2k
        } else if (CUR == 0) {
2434
18
            if (inParam == 0)
2435
0
                break;
2436
2437
            /*
2438
             * Don't pop parameter entities that start a markup
2439
             * declaration to detect Well-formedness constraint:
2440
             * PE Between Declarations.
2441
             */
2442
18
            if (ctxt->input->flags & XML_INPUT_MARKUP_DECL)
2443
18
                break;
2444
2445
0
            xmlPopPE(ctxt);
2446
2447
0
            inParam = PARSER_IN_PE(ctxt);
2448
0
            expandParam = PARSER_EXTERNAL(ctxt);
2449
38.2k
        } else {
2450
38.2k
            break;
2451
38.2k
        }
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
39.8k
        if (res < INT_MAX)
2461
39.8k
            res++;
2462
39.8k
    }
2463
2464
38.3k
    return(res);
2465
324k
}
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
52.6k
xmlParseCharRef(xmlParserCtxt *ctxt) {
2530
52.6k
    int val = 0;
2531
52.6k
    int count = 0;
2532
2533
    /*
2534
     * Using RAW/CUR/NEXT is okay since we are working on ASCII range here
2535
     */
2536
52.6k
    if ((RAW == '&') && (NXT(1) == '#') &&
2537
52.6k
        (NXT(2) == 'x')) {
2538
43.4k
  SKIP(3);
2539
43.4k
  GROW;
2540
161k
  while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) {
2541
118k
      if (count++ > 20) {
2542
1.49k
    count = 0;
2543
1.49k
    GROW;
2544
1.49k
      }
2545
118k
      if ((RAW >= '0') && (RAW <= '9'))
2546
100k
          val = val * 16 + (CUR - '0');
2547
18.2k
      else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20))
2548
11.4k
          val = val * 16 + (CUR - 'a') + 10;
2549
6.78k
      else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20))
2550
6.64k
          val = val * 16 + (CUR - 'A') + 10;
2551
143
      else {
2552
143
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2553
143
    val = 0;
2554
143
    break;
2555
143
      }
2556
118k
      if (val > 0x110000)
2557
8.71k
          val = 0x110000;
2558
2559
118k
      NEXT;
2560
118k
      count++;
2561
118k
  }
2562
43.4k
  if (RAW == ';') {
2563
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2564
43.3k
      ctxt->input->col++;
2565
43.3k
      ctxt->input->cur++;
2566
43.3k
  }
2567
43.4k
    } else if  ((RAW == '&') && (NXT(1) == '#')) {
2568
9.25k
  SKIP(2);
2569
9.25k
  GROW;
2570
58.4k
  while (RAW != ';') { /* loop blocked by count */
2571
50.2k
      if (count++ > 20) {
2572
2.16k
    count = 0;
2573
2.16k
    GROW;
2574
2.16k
      }
2575
50.2k
      if ((RAW >= '0') && (RAW <= '9'))
2576
49.1k
          val = val * 10 + (CUR - '0');
2577
1.03k
      else {
2578
1.03k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2579
1.03k
    val = 0;
2580
1.03k
    break;
2581
1.03k
      }
2582
49.1k
      if (val > 0x110000)
2583
11.2k
          val = 0x110000;
2584
2585
49.1k
      NEXT;
2586
49.1k
      count++;
2587
49.1k
  }
2588
9.25k
  if (RAW == ';') {
2589
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2590
8.21k
      ctxt->input->col++;
2591
8.21k
      ctxt->input->cur++;
2592
8.21k
  }
2593
9.25k
    } 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
52.6k
    if (val >= 0x110000) {
2605
81
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2606
81
                "xmlParseCharRef: character reference out of bounds\n",
2607
81
          val);
2608
81
        val = 0xFFFD;
2609
52.6k
    } else if (!IS_CHAR(val)) {
2610
1.26k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2611
1.26k
                          "xmlParseCharRef: invalid xmlChar value %d\n",
2612
1.26k
                    val);
2613
1.26k
    }
2614
52.6k
    return(val);
2615
52.6k
}
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
48.4k
xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) {
2635
48.4k
    const xmlChar *ptr;
2636
48.4k
    xmlChar cur;
2637
48.4k
    int val = 0;
2638
2639
48.4k
    if ((str == NULL) || (*str == NULL)) return(0);
2640
48.4k
    ptr = *str;
2641
48.4k
    cur = *ptr;
2642
48.4k
    if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) {
2643
27.0k
  ptr += 3;
2644
27.0k
  cur = *ptr;
2645
108k
  while (cur != ';') { /* Non input consuming loop */
2646
82.0k
      if ((cur >= '0') && (cur <= '9'))
2647
64.2k
          val = val * 16 + (cur - '0');
2648
17.8k
      else if ((cur >= 'a') && (cur <= 'f'))
2649
8.90k
          val = val * 16 + (cur - 'a') + 10;
2650
8.91k
      else if ((cur >= 'A') && (cur <= 'F'))
2651
8.81k
          val = val * 16 + (cur - 'A') + 10;
2652
101
      else {
2653
101
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2654
101
    val = 0;
2655
101
    break;
2656
101
      }
2657
81.9k
      if (val > 0x110000)
2658
1.58k
          val = 0x110000;
2659
2660
81.9k
      ptr++;
2661
81.9k
      cur = *ptr;
2662
81.9k
  }
2663
27.0k
  if (cur == ';')
2664
26.9k
      ptr++;
2665
27.0k
    } else if  ((cur == '&') && (ptr[1] == '#')){
2666
21.4k
  ptr += 2;
2667
21.4k
  cur = *ptr;
2668
75.2k
  while (cur != ';') { /* Non input consuming loops */
2669
53.9k
      if ((cur >= '0') && (cur <= '9'))
2670
53.8k
          val = val * 10 + (cur - '0');
2671
96
      else {
2672
96
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2673
96
    val = 0;
2674
96
    break;
2675
96
      }
2676
53.8k
      if (val > 0x110000)
2677
725
          val = 0x110000;
2678
2679
53.8k
      ptr++;
2680
53.8k
      cur = *ptr;
2681
53.8k
  }
2682
21.4k
  if (cur == ';')
2683
21.3k
      ptr++;
2684
21.4k
    } else {
2685
0
  xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2686
0
  return(0);
2687
0
    }
2688
48.4k
    *str = ptr;
2689
2690
    /*
2691
     * [ WFC: Legal Character ]
2692
     * Characters referred to using character references must match the
2693
     * production for Char.
2694
     */
2695
48.4k
    if (val >= 0x110000) {
2696
34
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2697
34
                "xmlParseStringCharRef: character reference out of bounds\n",
2698
34
                val);
2699
48.4k
    } else if (IS_CHAR(val)) {
2700
48.1k
        return(val);
2701
48.1k
    } else {
2702
233
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2703
233
        "xmlParseStringCharRef: invalid xmlChar value %d\n",
2704
233
        val);
2705
233
    }
2706
267
    return(0);
2707
48.4k
}
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
4.30M
                     int blank_chars) {
2816
4.30M
    int i;
2817
4.30M
    xmlNodePtr lastChild;
2818
2819
    /*
2820
     * Check for xml:space value.
2821
     */
2822
4.30M
    if ((ctxt->space == NULL) || (*(ctxt->space) == 1) ||
2823
4.30M
        (*(ctxt->space) == -2))
2824
3.14M
  return(0);
2825
2826
    /*
2827
     * Check that the string is made of blanks
2828
     */
2829
1.16M
    if (blank_chars == 0) {
2830
1.24M
  for (i = 0;i < len;i++)
2831
1.24M
      if (!(IS_BLANK_CH(str[i]))) return(0);
2832
952k
    }
2833
2834
    /*
2835
     * Look if the element is mixed content in the DTD if available
2836
     */
2837
220k
    if (ctxt->node == NULL) return(0);
2838
0
    if (ctxt->myDoc != NULL) {
2839
0
        xmlElementPtr elemDecl = NULL;
2840
0
        xmlDocPtr doc = ctxt->myDoc;
2841
0
        const xmlChar *prefix = NULL;
2842
2843
0
        if (ctxt->node->ns)
2844
0
            prefix = ctxt->node->ns->prefix;
2845
0
        if (doc->intSubset != NULL)
2846
0
            elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name,
2847
0
                                      prefix);
2848
0
        if ((elemDecl == NULL) && (doc->extSubset != NULL))
2849
0
            elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name,
2850
0
                                      prefix);
2851
0
        if (elemDecl != NULL) {
2852
0
            if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)
2853
0
                return(1);
2854
0
            if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) ||
2855
0
                (elemDecl->etype == XML_ELEMENT_TYPE_MIXED))
2856
0
                return(0);
2857
0
        }
2858
0
    }
2859
2860
    /*
2861
     * Otherwise, heuristic :-\
2862
     *
2863
     * When push parsing, we could be at the end of a chunk.
2864
     * This makes the look-ahead and consequently the NOBLANKS
2865
     * option unreliable.
2866
     */
2867
0
    if ((RAW != '<') && (RAW != 0xD)) return(0);
2868
0
    if ((ctxt->node->children == NULL) &&
2869
0
  (RAW == '<') && (NXT(1) == '/')) return(0);
2870
2871
0
    lastChild = xmlGetLastChild(ctxt->node);
2872
0
    if (lastChild == NULL) {
2873
0
        if ((ctxt->node->type != XML_ELEMENT_NODE) &&
2874
0
            (ctxt->node->content != NULL)) return(0);
2875
0
    } else if (xmlNodeIsText(lastChild))
2876
0
        return(0);
2877
0
    else if ((ctxt->node->children != NULL) &&
2878
0
             (xmlNodeIsText(ctxt->node->children)))
2879
0
        return(0);
2880
0
    return(1);
2881
0
}
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
266k
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
266k
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2953
265k
        (((c >= 'a') && (c <= 'z')) ||
2954
64.8k
         ((c >= 'A') && (c <= 'Z')) ||
2955
44.8k
         (c == '_') || (c == ':') ||
2956
43.3k
         ((c >= 0xC0) && (c <= 0xD6)) ||
2957
41.9k
         ((c >= 0xD8) && (c <= 0xF6)) ||
2958
39.1k
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2959
31.6k
         ((c >= 0x370) && (c <= 0x37D)) ||
2960
31.1k
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2961
19.6k
         ((c >= 0x200C) && (c <= 0x200D)) ||
2962
19.0k
         ((c >= 0x2070) && (c <= 0x218F)) ||
2963
18.5k
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2964
17.6k
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2965
15.1k
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2966
14.4k
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2967
12.3k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2968
255k
        return(1);
2969
11.1k
    return(0);
2970
266k
}
2971
2972
static int
2973
8.62M
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
8.62M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2979
8.58M
        (((c >= 'a') && (c <= 'z')) ||
2980
4.44M
         ((c >= 'A') && (c <= 'Z')) ||
2981
4.14M
         ((c >= '0') && (c <= '9')) || /* !start */
2982
3.92M
         (c == '_') || (c == ':') ||
2983
3.87M
         (c == '-') || (c == '.') || (c == 0xB7) || /* !start */
2984
3.83M
         ((c >= 0xC0) && (c <= 0xD6)) ||
2985
3.27M
         ((c >= 0xD8) && (c <= 0xF6)) ||
2986
3.00M
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2987
1.22M
         ((c >= 0x300) && (c <= 0x36F)) || /* !start */
2988
1.21M
         ((c >= 0x370) && (c <= 0x37D)) ||
2989
1.21M
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2990
1.08M
         ((c >= 0x200C) && (c <= 0x200D)) ||
2991
1.08M
         ((c >= 0x203F) && (c <= 0x2040)) || /* !start */
2992
1.08M
         ((c >= 0x2070) && (c <= 0x218F)) ||
2993
1.07M
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2994
1.07M
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2995
791k
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2996
788k
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2997
238k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2998
8.35M
         return(1);
2999
272k
    return(0);
3000
8.62M
}
3001
3002
static int
3003
0
xmlIsNameStartCharOld(int c) {
3004
0
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3005
0
        ((IS_LETTER(c) || (c == '_') || (c == ':'))))
3006
0
        return(1);
3007
0
    return(0);
3008
0
}
3009
3010
static int
3011
0
xmlIsNameCharOld(int c) {
3012
0
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3013
0
        ((IS_LETTER(c)) || (IS_DIGIT(c)) ||
3014
0
         (c == '.') || (c == '-') ||
3015
0
         (c == '_') || (c == ':') ||
3016
0
         (IS_COMBINING(c)) ||
3017
0
         (IS_EXTENDER(c))))
3018
0
        return(1);
3019
0
    return(0);
3020
0
}
3021
3022
static int
3023
266k
xmlIsNameStartChar(int c, int old10) {
3024
266k
    if (!old10)
3025
266k
        return(xmlIsNameStartCharNew(c));
3026
0
    else
3027
0
        return(xmlIsNameStartCharOld(c));
3028
266k
}
3029
3030
static int
3031
8.62M
xmlIsNameChar(int c, int old10) {
3032
8.62M
    if (!old10)
3033
8.62M
        return(xmlIsNameCharNew(c));
3034
0
    else
3035
0
        return(xmlIsNameCharOld(c));
3036
8.62M
}
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
77.4k
xmlScanName(const xmlChar *ptr, size_t maxSize, int flags) {
3052
77.4k
    int stop = flags & XML_SCAN_NC ? ':' : 0;
3053
77.4k
    int old10 = flags & XML_SCAN_OLD10 ? 1 : 0;
3054
3055
417k
    while (1) {
3056
417k
        int c, len;
3057
3058
417k
        c = *ptr;
3059
417k
        if (c < 0x80) {
3060
232k
            if (c == stop)
3061
2
                break;
3062
232k
            len = 1;
3063
232k
        } else {
3064
185k
            len = 4;
3065
185k
            c = xmlGetUTF8Char(ptr, &len);
3066
185k
            if (c < 0)
3067
73
                break;
3068
185k
        }
3069
3070
417k
        if (flags & XML_SCAN_NMTOKEN ?
3071
340k
                !xmlIsNameChar(c, old10) :
3072
417k
                !xmlIsNameStartChar(c, old10))
3073
77.3k
            break;
3074
3075
340k
        if ((size_t) len > maxSize)
3076
7
            return(NULL);
3077
340k
        ptr += len;
3078
340k
        maxSize -= len;
3079
340k
        flags |= XML_SCAN_NMTOKEN;
3080
340k
    }
3081
3082
77.4k
    return(ptr);
3083
77.4k
}
3084
3085
static const xmlChar *
3086
188k
xmlParseNameComplex(xmlParserCtxtPtr ctxt) {
3087
188k
    const xmlChar *ret;
3088
188k
    int len = 0, l;
3089
188k
    int c;
3090
188k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3091
0
                    XML_MAX_TEXT_LENGTH :
3092
188k
                    XML_MAX_NAME_LENGTH;
3093
188k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3094
3095
    /*
3096
     * Handler for more complex cases
3097
     */
3098
188k
    c = xmlCurrentChar(ctxt, &l);
3099
188k
    if (!xmlIsNameStartChar(c, old10))
3100
11.0k
        return(NULL);
3101
177k
    len += l;
3102
177k
    NEXTL(l);
3103
177k
    c = xmlCurrentChar(ctxt, &l);
3104
6.46M
    while (xmlIsNameChar(c, old10)) {
3105
6.28M
        if (len <= INT_MAX - l)
3106
6.28M
            len += l;
3107
6.28M
        NEXTL(l);
3108
6.28M
        c = xmlCurrentChar(ctxt, &l);
3109
6.28M
    }
3110
177k
    if (len > maxLength) {
3111
40
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3112
40
        return(NULL);
3113
40
    }
3114
177k
    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
177k
    if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r'))
3125
2.73k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len);
3126
174k
    else
3127
174k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len);
3128
177k
    if (ret == NULL)
3129
0
        xmlErrMemory(ctxt);
3130
177k
    return(ret);
3131
177k
}
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
4.44M
xmlParseName(xmlParserCtxt *ctxt) {
3151
4.44M
    const xmlChar *in;
3152
4.44M
    const xmlChar *ret;
3153
4.44M
    size_t count = 0;
3154
4.44M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3155
0
                       XML_MAX_TEXT_LENGTH :
3156
4.44M
                       XML_MAX_NAME_LENGTH;
3157
3158
4.44M
    GROW;
3159
3160
    /*
3161
     * Accelerator for simple ASCII names
3162
     */
3163
4.44M
    in = ctxt->input->cur;
3164
4.44M
    if (((*in >= 0x61) && (*in <= 0x7A)) ||
3165
1.16M
  ((*in >= 0x41) && (*in <= 0x5A)) ||
3166
4.40M
  (*in == '_') || (*in == ':')) {
3167
4.40M
  in++;
3168
28.7M
  while (((*in >= 0x61) && (*in <= 0x7A)) ||
3169
8.97M
         ((*in >= 0x41) && (*in <= 0x5A)) ||
3170
6.83M
         ((*in >= 0x30) && (*in <= 0x39)) ||
3171
5.18M
         (*in == '_') || (*in == '-') ||
3172
4.63M
         (*in == ':') || (*in == '.'))
3173
24.3M
      in++;
3174
4.40M
  if ((*in > 0) && (*in < 0x80)) {
3175
4.25M
      count = in - ctxt->input->cur;
3176
4.25M
            if (count > maxLength) {
3177
25
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3178
25
                return(NULL);
3179
25
            }
3180
4.25M
      ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count);
3181
4.25M
      ctxt->input->cur = in;
3182
4.25M
      ctxt->input->col += count;
3183
4.25M
      if (ret == NULL)
3184
0
          xmlErrMemory(ctxt);
3185
4.25M
      return(ret);
3186
4.25M
  }
3187
4.40M
    }
3188
    /* accelerator for special cases */
3189
188k
    return(xmlParseNameComplex(ctxt));
3190
4.44M
}
3191
3192
static xmlHashedString
3193
0
xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) {
3194
0
    xmlHashedString ret;
3195
0
    int len = 0, l;
3196
0
    int c;
3197
0
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3198
0
                    XML_MAX_TEXT_LENGTH :
3199
0
                    XML_MAX_NAME_LENGTH;
3200
0
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3201
0
    size_t startPosition = 0;
3202
3203
0
    ret.name = NULL;
3204
0
    ret.hashValue = 0;
3205
3206
    /*
3207
     * Handler for more complex cases
3208
     */
3209
0
    startPosition = CUR_PTR - BASE_PTR;
3210
0
    c = xmlCurrentChar(ctxt, &l);
3211
0
    if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3212
0
  (!xmlIsNameStartChar(c, old10) || (c == ':'))) {
3213
0
  return(ret);
3214
0
    }
3215
3216
0
    while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */
3217
0
     (xmlIsNameChar(c, old10) && (c != ':'))) {
3218
0
        if (len <= INT_MAX - l)
3219
0
      len += l;
3220
0
  NEXTL(l);
3221
0
  c = xmlCurrentChar(ctxt, &l);
3222
0
    }
3223
0
    if (len > maxLength) {
3224
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3225
0
        return(ret);
3226
0
    }
3227
0
    ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len);
3228
0
    if (ret.name == NULL)
3229
0
        xmlErrMemory(ctxt);
3230
0
    return(ret);
3231
0
}
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
0
xmlParseNCName(xmlParserCtxtPtr ctxt) {
3247
0
    const xmlChar *in, *e;
3248
0
    xmlHashedString ret;
3249
0
    size_t count = 0;
3250
0
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3251
0
                       XML_MAX_TEXT_LENGTH :
3252
0
                       XML_MAX_NAME_LENGTH;
3253
3254
0
    ret.name = NULL;
3255
3256
    /*
3257
     * Accelerator for simple ASCII names
3258
     */
3259
0
    in = ctxt->input->cur;
3260
0
    e = ctxt->input->end;
3261
0
    if ((((*in >= 0x61) && (*in <= 0x7A)) ||
3262
0
   ((*in >= 0x41) && (*in <= 0x5A)) ||
3263
0
   (*in == '_')) && (in < e)) {
3264
0
  in++;
3265
0
  while ((((*in >= 0x61) && (*in <= 0x7A)) ||
3266
0
          ((*in >= 0x41) && (*in <= 0x5A)) ||
3267
0
          ((*in >= 0x30) && (*in <= 0x39)) ||
3268
0
          (*in == '_') || (*in == '-') ||
3269
0
          (*in == '.')) && (in < e))
3270
0
      in++;
3271
0
  if (in >= e)
3272
0
      goto complex;
3273
0
  if ((*in > 0) && (*in < 0x80)) {
3274
0
      count = in - ctxt->input->cur;
3275
0
            if (count > maxLength) {
3276
0
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3277
0
                return(ret);
3278
0
            }
3279
0
      ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count);
3280
0
      ctxt->input->cur = in;
3281
0
      ctxt->input->col += count;
3282
0
      if (ret.name == NULL) {
3283
0
          xmlErrMemory(ctxt);
3284
0
      }
3285
0
      return(ret);
3286
0
  }
3287
0
    }
3288
0
complex:
3289
0
    return(xmlParseNCNameComplex(ctxt));
3290
0
}
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
235k
xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) {
3304
235k
    register const xmlChar *cmp = other;
3305
235k
    register const xmlChar *in;
3306
235k
    const xmlChar *ret;
3307
3308
235k
    GROW;
3309
3310
235k
    in = ctxt->input->cur;
3311
1.26M
    while (*in != 0 && *in == *cmp) {
3312
1.03M
  ++in;
3313
1.03M
  ++cmp;
3314
1.03M
    }
3315
235k
    if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
3316
  /* success */
3317
232k
  ctxt->input->col += in - ctxt->input->cur;
3318
232k
  ctxt->input->cur = in;
3319
232k
  return (const xmlChar*) 1;
3320
232k
    }
3321
    /* failure (or end of input buffer), check with full function */
3322
2.51k
    ret = xmlParseName (ctxt);
3323
    /* strings coming from the dictionary direct compare possible */
3324
2.51k
    if (ret == other) {
3325
39
  return (const xmlChar*) 1;
3326
39
    }
3327
2.47k
    return ret;
3328
2.51k
}
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
77.4k
xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) {
3341
77.4k
    xmlChar *ret;
3342
77.4k
    const xmlChar *cur = *str;
3343
77.4k
    int flags = 0;
3344
77.4k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3345
0
                    XML_MAX_TEXT_LENGTH :
3346
77.4k
                    XML_MAX_NAME_LENGTH;
3347
3348
77.4k
    if (ctxt->options & XML_PARSE_OLD10)
3349
0
        flags |= XML_SCAN_OLD10;
3350
3351
77.4k
    cur = xmlScanName(*str, maxLength, flags);
3352
77.4k
    if (cur == NULL) {
3353
7
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3354
7
        return(NULL);
3355
7
    }
3356
77.4k
    if (cur == *str)
3357
128
        return(NULL);
3358
3359
77.2k
    ret = xmlStrndup(*str, cur - *str);
3360
77.2k
    if (ret == NULL)
3361
0
        xmlErrMemory(ctxt);
3362
77.2k
    *str = cur;
3363
77.2k
    return(ret);
3364
77.4k
}
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
17.6k
xmlParseNmtoken(xmlParserCtxt *ctxt) {
3381
17.6k
    xmlChar buf[XML_MAX_NAMELEN + 5];
3382
17.6k
    xmlChar *ret;
3383
17.6k
    int len = 0, l;
3384
17.6k
    int c;
3385
17.6k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3386
0
                    XML_MAX_TEXT_LENGTH :
3387
17.6k
                    XML_MAX_NAME_LENGTH;
3388
17.6k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3389
3390
17.6k
    c = xmlCurrentChar(ctxt, &l);
3391
3392
228k
    while (xmlIsNameChar(c, old10)) {
3393
212k
  COPY_BUF(buf, len, c);
3394
212k
  NEXTL(l);
3395
212k
  c = xmlCurrentChar(ctxt, &l);
3396
212k
  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
1.85k
      xmlChar *buffer;
3402
1.85k
      int max = len * 2;
3403
3404
1.85k
      buffer = xmlMalloc(max);
3405
1.85k
      if (buffer == NULL) {
3406
0
          xmlErrMemory(ctxt);
3407
0
    return(NULL);
3408
0
      }
3409
1.85k
      memcpy(buffer, buf, len);
3410
1.59M
      while (xmlIsNameChar(c, old10)) {
3411
1.58M
    if (len + 10 > max) {
3412
2.35k
        xmlChar *tmp;
3413
2.35k
                    int newSize;
3414
3415
2.35k
                    newSize = xmlGrowCapacity(max, 1, 1, maxLength);
3416
2.35k
                    if (newSize < 0) {
3417
2
                        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3418
2
                        xmlFree(buffer);
3419
2
                        return(NULL);
3420
2
                    }
3421
2.35k
        tmp = xmlRealloc(buffer, newSize);
3422
2.35k
        if (tmp == NULL) {
3423
0
      xmlErrMemory(ctxt);
3424
0
      xmlFree(buffer);
3425
0
      return(NULL);
3426
0
        }
3427
2.35k
        buffer = tmp;
3428
2.35k
                    max = newSize;
3429
2.35k
    }
3430
1.58M
    COPY_BUF(buffer, len, c);
3431
1.58M
    NEXTL(l);
3432
1.58M
    c = xmlCurrentChar(ctxt, &l);
3433
1.58M
      }
3434
1.85k
      buffer[len] = 0;
3435
1.85k
      return(buffer);
3436
1.85k
  }
3437
212k
    }
3438
15.8k
    if (len == 0)
3439
13
        return(NULL);
3440
15.8k
    if (len > maxLength) {
3441
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3442
0
        return(NULL);
3443
0
    }
3444
15.8k
    ret = xmlStrndup(buf, len);
3445
15.8k
    if (ret == NULL)
3446
0
        xmlErrMemory(ctxt);
3447
15.8k
    return(ret);
3448
15.8k
}
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
21.6k
                          const xmlChar *str, int length, int depth) {
3462
21.6k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3463
21.6k
    const xmlChar *end, *chunk;
3464
21.6k
    int c, l;
3465
3466
21.6k
    if (str == NULL)
3467
0
        return;
3468
3469
21.6k
    depth += 1;
3470
21.6k
    if (depth > maxDepth) {
3471
0
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3472
0
                       "Maximum entity nesting depth exceeded");
3473
0
  return;
3474
0
    }
3475
3476
21.6k
    end = str + length;
3477
21.6k
    chunk = str;
3478
3479
107M
    while ((str < end) && (!PARSER_STOPPED(ctxt))) {
3480
107M
        c = *str;
3481
3482
107M
        if (c >= 0x80) {
3483
78.3M
            l = xmlUTF8MultibyteLen(ctxt, str,
3484
78.3M
                    "invalid character in entity value\n");
3485
78.3M
            if (l == 0) {
3486
127
                if (chunk < str)
3487
112
                    xmlSBufAddString(buf, chunk, str - chunk);
3488
127
                xmlSBufAddReplChar(buf);
3489
127
                str += 1;
3490
127
                chunk = str;
3491
78.3M
            } else {
3492
78.3M
                str += l;
3493
78.3M
            }
3494
78.3M
        } else if (c == '&') {
3495
113k
            if (str[1] == '#') {
3496
46.3k
                if (chunk < str)
3497
44.1k
                    xmlSBufAddString(buf, chunk, str - chunk);
3498
3499
46.3k
                c = xmlParseStringCharRef(ctxt, &str);
3500
46.3k
                if (c == 0)
3501
261
                    return;
3502
3503
46.0k
                xmlSBufAddChar(buf, c);
3504
3505
46.0k
                chunk = str;
3506
66.7k
            } else {
3507
66.7k
                xmlChar *name;
3508
3509
                /*
3510
                 * General entity references are checked for
3511
                 * syntactic validity.
3512
                 */
3513
66.7k
                str++;
3514
66.7k
                name = xmlParseStringName(ctxt, &str);
3515
3516
66.7k
                if ((name == NULL) || (*str++ != ';')) {
3517
213
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR,
3518
213
                            "EntityValue: '&' forbidden except for entities "
3519
213
                            "references\n");
3520
213
                    xmlFree(name);
3521
213
                    return;
3522
213
                }
3523
3524
66.4k
                xmlFree(name);
3525
66.4k
            }
3526
28.6M
        } else if (c == '%') {
3527
2.80k
            xmlEntityPtr ent;
3528
3529
2.80k
            if (chunk < str)
3530
2.66k
                xmlSBufAddString(buf, chunk, str - chunk);
3531
3532
2.80k
            ent = xmlParseStringPEReference(ctxt, &str);
3533
2.80k
            if (ent == NULL)
3534
2.80k
                return;
3535
3536
3
            if (!PARSER_EXTERNAL(ctxt)) {
3537
3
                xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL);
3538
3
                return;
3539
3
            }
3540
3541
0
            if (ent->content == NULL) {
3542
                /*
3543
                 * Note: external parsed entities will not be loaded,
3544
                 * it is not required for a non-validating parser to
3545
                 * complete external PEReferences coming from the
3546
                 * internal subset
3547
                 */
3548
0
                if (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
3549
0
                    ((ctxt->replaceEntities) ||
3550
0
                     (ctxt->validate))) {
3551
0
                    xmlLoadEntityContent(ctxt, ent);
3552
0
                } else {
3553
0
                    xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
3554
0
                                  "not validating will not read content for "
3555
0
                                  "PE entity %s\n", ent->name, NULL);
3556
0
                }
3557
0
            }
3558
3559
            /*
3560
             * TODO: Skip if ent->content is still NULL.
3561
             */
3562
3563
0
            if (xmlParserEntityCheck(ctxt, ent->length))
3564
0
                return;
3565
3566
0
            if (ent->flags & XML_ENT_EXPANDING) {
3567
0
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3568
0
                return;
3569
0
            }
3570
3571
0
            ent->flags |= XML_ENT_EXPANDING;
3572
0
            xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length,
3573
0
                                      depth);
3574
0
            ent->flags &= ~XML_ENT_EXPANDING;
3575
3576
0
            chunk = str;
3577
28.6M
        } else {
3578
            /* Normal ASCII char */
3579
28.6M
            if (!IS_BYTE_CHAR(c)) {
3580
126
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3581
126
                        "invalid character in entity value\n");
3582
126
                if (chunk < str)
3583
115
                    xmlSBufAddString(buf, chunk, str - chunk);
3584
126
                xmlSBufAddReplChar(buf);
3585
126
                str += 1;
3586
126
                chunk = str;
3587
28.6M
            } else {
3588
28.6M
                str += 1;
3589
28.6M
            }
3590
28.6M
        }
3591
107M
    }
3592
3593
18.3k
    if (chunk < str)
3594
17.3k
        xmlSBufAddString(buf, chunk, str - chunk);
3595
18.3k
}
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
21.7k
xmlParseEntityValue(xmlParserCtxt *ctxt, xmlChar **orig) {
3611
21.7k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3612
0
                         XML_MAX_HUGE_LENGTH :
3613
21.7k
                         XML_MAX_TEXT_LENGTH;
3614
21.7k
    xmlSBuf buf;
3615
21.7k
    const xmlChar *start;
3616
21.7k
    int quote, length;
3617
3618
21.7k
    xmlSBufInit(&buf, maxLength);
3619
3620
21.7k
    GROW;
3621
3622
21.7k
    quote = CUR;
3623
21.7k
    if ((quote != '"') && (quote != '\'')) {
3624
0
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
3625
0
  return(NULL);
3626
0
    }
3627
21.7k
    CUR_PTR++;
3628
3629
21.7k
    length = 0;
3630
3631
    /*
3632
     * Copy raw content of the entity into a buffer
3633
     */
3634
253M
    while (1) {
3635
253M
        int c;
3636
3637
253M
        if (PARSER_STOPPED(ctxt))
3638
20
            goto error;
3639
3640
253M
        if (CUR_PTR >= ctxt->input->end) {
3641
27
            xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL);
3642
27
            goto error;
3643
27
        }
3644
3645
253M
        c = CUR;
3646
3647
253M
        if (c == 0) {
3648
38
            xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3649
38
                    "invalid character in entity value\n");
3650
38
            goto error;
3651
38
        }
3652
253M
        if (c == quote)
3653
21.6k
            break;
3654
253M
        NEXTL(1);
3655
253M
        length += 1;
3656
3657
        /*
3658
         * TODO: Check growth threshold
3659
         */
3660
253M
        if (ctxt->input->end - CUR_PTR < 10)
3661
2.33k
            GROW;
3662
253M
    }
3663
3664
21.6k
    start = CUR_PTR - length;
3665
3666
21.6k
    if (orig != NULL) {
3667
21.6k
        *orig = xmlStrndup(start, length);
3668
21.6k
        if (*orig == NULL)
3669
0
            xmlErrMemory(ctxt);
3670
21.6k
    }
3671
3672
21.6k
    xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr);
3673
3674
21.6k
    NEXTL(1);
3675
3676
21.6k
    return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long"));
3677
3678
85
error:
3679
85
    xmlSBufCleanup(&buf, ctxt, "entity length too long");
3680
85
    return(NULL);
3681
21.7k
}
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
1.63k
xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) {
3693
1.63k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3694
1.63k
    const xmlChar *str;
3695
1.63k
    unsigned long expandedSize = pent->length;
3696
1.63k
    int c, flags;
3697
3698
1.63k
    depth += 1;
3699
1.63k
    if (depth > maxDepth) {
3700
2
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3701
2
                       "Maximum entity nesting depth exceeded");
3702
2
  return;
3703
2
    }
3704
3705
1.63k
    if (pent->flags & XML_ENT_EXPANDING) {
3706
26
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3707
26
        return;
3708
26
    }
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
1.60k
    if (ctxt->inSubset == 0)
3716
1.56k
        flags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
3717
38
    else
3718
38
        flags = XML_ENT_VALIDATED;
3719
3720
1.60k
    str = pent->content;
3721
1.60k
    if (str == NULL)
3722
0
        goto done;
3723
3724
    /*
3725
     * Note that entity values are already validated. We only check
3726
     * for illegal less-than signs and compute the expanded size
3727
     * of the entity. No special handling for multi-byte characters
3728
     * is needed.
3729
     */
3730
17.3M
    while (!PARSER_STOPPED(ctxt)) {
3731
17.3M
        c = *str;
3732
3733
17.3M
  if (c != '&') {
3734
17.3M
            if (c == 0)
3735
1.29k
                break;
3736
3737
17.3M
            if (c == '<')
3738
13
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3739
13
                        "'<' in entity '%s' is not allowed in attributes "
3740
13
                        "values\n", pent->name);
3741
3742
17.3M
            str += 1;
3743
17.3M
        } else if (str[1] == '#') {
3744
2.11k
            int val;
3745
3746
2.11k
      val = xmlParseStringCharRef(ctxt, &str);
3747
2.11k
      if (val == 0) {
3748
6
                pent->content[0] = 0;
3749
6
                break;
3750
6
            }
3751
7.92k
  } else {
3752
7.92k
            xmlChar *name;
3753
7.92k
            xmlEntityPtr ent;
3754
3755
7.92k
      name = xmlParseStringEntityRef(ctxt, &str);
3756
7.92k
      if (name == NULL) {
3757
5
                pent->content[0] = 0;
3758
5
                break;
3759
5
            }
3760
3761
7.91k
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3762
7.91k
            xmlFree(name);
3763
3764
7.91k
            if ((ent != NULL) &&
3765
6.96k
                (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) {
3766
5.75k
                if ((ent->flags & flags) != flags) {
3767
971
                    pent->flags |= XML_ENT_EXPANDING;
3768
971
                    xmlCheckEntityInAttValue(ctxt, ent, depth);
3769
971
                    pent->flags &= ~XML_ENT_EXPANDING;
3770
971
                }
3771
3772
5.75k
                xmlSaturatedAdd(&expandedSize, ent->expandedSize);
3773
5.75k
                xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST);
3774
5.75k
            }
3775
7.91k
        }
3776
17.3M
    }
3777
3778
1.60k
done:
3779
1.60k
    if (ctxt->inSubset == 0)
3780
1.56k
        pent->expandedSize = expandedSize;
3781
3782
1.60k
    pent->flags |= flags;
3783
1.60k
}
3784
3785
/**
3786
 * Expand general entity references in an entity or attribute value.
3787
 * Perform attribute value normalization.
3788
 *
3789
 * @param ctxt  parser context
3790
 * @param buf  string buffer
3791
 * @param str  entity or attribute value
3792
 * @param pent  entity for entity value, NULL for attribute values
3793
 * @param normalize  whether to collapse whitespace
3794
 * @param inSpace  whitespace state
3795
 * @param depth  nesting depth
3796
 * @param check  whether to check for amplification
3797
 * @returns  whether there was a normalization change
3798
 */
3799
static int
3800
xmlExpandEntityInAttValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf,
3801
                          const xmlChar *str, xmlEntityPtr pent, int normalize,
3802
0
                          int *inSpace, int depth, int check) {
3803
0
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3804
0
    int c, chunkSize;
3805
0
    int normChange = 0;
3806
3807
0
    if (str == NULL)
3808
0
        return(0);
3809
3810
0
    depth += 1;
3811
0
    if (depth > maxDepth) {
3812
0
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3813
0
                       "Maximum entity nesting depth exceeded");
3814
0
  return(0);
3815
0
    }
3816
3817
0
    if (pent != NULL) {
3818
0
        if (pent->flags & XML_ENT_EXPANDING) {
3819
0
            xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3820
0
            return(0);
3821
0
        }
3822
3823
0
        if (check) {
3824
0
            if (xmlParserEntityCheck(ctxt, pent->length))
3825
0
                return(0);
3826
0
        }
3827
0
    }
3828
3829
0
    chunkSize = 0;
3830
3831
    /*
3832
     * Note that entity values are already validated. No special
3833
     * handling for multi-byte characters is needed.
3834
     */
3835
0
    while (!PARSER_STOPPED(ctxt)) {
3836
0
        c = *str;
3837
3838
0
  if (c != '&') {
3839
0
            if (c == 0)
3840
0
                break;
3841
3842
            /*
3843
             * If this function is called without an entity, it is used to
3844
             * expand entities in an attribute content where less-than was
3845
             * already unscaped and is allowed.
3846
             */
3847
0
            if ((pent != NULL) && (c == '<')) {
3848
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3849
0
                        "'<' in entity '%s' is not allowed in attributes "
3850
0
                        "values\n", pent->name);
3851
0
                break;
3852
0
            }
3853
3854
0
            if (c <= 0x20) {
3855
0
                if ((normalize) && (*inSpace)) {
3856
                    /* Skip char */
3857
0
                    if (chunkSize > 0) {
3858
0
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3859
0
                        chunkSize = 0;
3860
0
                    }
3861
0
                    normChange = 1;
3862
0
                } else if (c < 0x20) {
3863
0
                    if (chunkSize > 0) {
3864
0
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3865
0
                        chunkSize = 0;
3866
0
                    }
3867
3868
0
                    xmlSBufAddCString(buf, " ", 1);
3869
0
                } else {
3870
0
                    chunkSize += 1;
3871
0
                }
3872
3873
0
                *inSpace = 1;
3874
0
            } else {
3875
0
                chunkSize += 1;
3876
0
                *inSpace = 0;
3877
0
            }
3878
3879
0
            str += 1;
3880
0
        } else if (str[1] == '#') {
3881
0
            int val;
3882
3883
0
            if (chunkSize > 0) {
3884
0
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3885
0
                chunkSize = 0;
3886
0
            }
3887
3888
0
      val = xmlParseStringCharRef(ctxt, &str);
3889
0
      if (val == 0) {
3890
0
                if (pent != NULL)
3891
0
                    pent->content[0] = 0;
3892
0
                break;
3893
0
            }
3894
3895
0
            if (val == ' ') {
3896
0
                if ((normalize) && (*inSpace))
3897
0
                    normChange = 1;
3898
0
                else
3899
0
                    xmlSBufAddCString(buf, " ", 1);
3900
0
                *inSpace = 1;
3901
0
            } else {
3902
0
                xmlSBufAddChar(buf, val);
3903
0
                *inSpace = 0;
3904
0
            }
3905
0
  } else {
3906
0
            xmlChar *name;
3907
0
            xmlEntityPtr ent;
3908
3909
0
            if (chunkSize > 0) {
3910
0
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3911
0
                chunkSize = 0;
3912
0
            }
3913
3914
0
      name = xmlParseStringEntityRef(ctxt, &str);
3915
0
            if (name == NULL) {
3916
0
                if (pent != NULL)
3917
0
                    pent->content[0] = 0;
3918
0
                break;
3919
0
            }
3920
3921
0
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3922
0
            xmlFree(name);
3923
3924
0
      if ((ent != NULL) &&
3925
0
    (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
3926
0
    if (ent->content == NULL) {
3927
0
        xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
3928
0
          "predefined entity has no content\n");
3929
0
                    break;
3930
0
                }
3931
3932
0
                xmlSBufAddString(buf, ent->content, ent->length);
3933
3934
0
                *inSpace = 0;
3935
0
      } else if ((ent != NULL) && (ent->content != NULL)) {
3936
0
                if (pent != NULL)
3937
0
                    pent->flags |= XML_ENT_EXPANDING;
3938
0
    normChange |= xmlExpandEntityInAttValue(ctxt, buf,
3939
0
                        ent->content, ent, normalize, inSpace, depth, check);
3940
0
                if (pent != NULL)
3941
0
                    pent->flags &= ~XML_ENT_EXPANDING;
3942
0
      }
3943
0
        }
3944
0
    }
3945
3946
0
    if (chunkSize > 0)
3947
0
        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3948
3949
0
    return(normChange);
3950
0
}
3951
3952
/**
3953
 * Expand general entity references in an entity or attribute value.
3954
 * Perform attribute value normalization.
3955
 *
3956
 * @param ctxt  parser context
3957
 * @param str  entity or attribute value
3958
 * @param normalize  whether to collapse whitespace
3959
 * @returns the expanded attribtue value.
3960
 */
3961
xmlChar *
3962
xmlExpandEntitiesInAttValue(xmlParserCtxt *ctxt, const xmlChar *str,
3963
0
                            int normalize) {
3964
0
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3965
0
                         XML_MAX_HUGE_LENGTH :
3966
0
                         XML_MAX_TEXT_LENGTH;
3967
0
    xmlSBuf buf;
3968
0
    int inSpace = 1;
3969
3970
0
    xmlSBufInit(&buf, maxLength);
3971
3972
0
    xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace,
3973
0
                              ctxt->inputNr, /* check */ 0);
3974
3975
0
    if ((normalize) && (inSpace) && (buf.size > 0))
3976
0
        buf.size--;
3977
3978
0
    return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long"));
3979
0
}
3980
3981
/**
3982
 * Parse a value for an attribute.
3983
 *
3984
 * NOTE: if no normalization is needed, the routine will return pointers
3985
 * directly from the data buffer.
3986
 *
3987
 * 3.3.3 Attribute-Value Normalization:
3988
 *
3989
 * Before the value of an attribute is passed to the application or
3990
 * checked for validity, the XML processor must normalize it as follows:
3991
 *
3992
 * - a character reference is processed by appending the referenced
3993
 *   character to the attribute value
3994
 * - an entity reference is processed by recursively processing the
3995
 *   replacement text of the entity
3996
 * - a whitespace character (\#x20, \#xD, \#xA, \#x9) is processed by
3997
 *   appending \#x20 to the normalized value, except that only a single
3998
 *   \#x20 is appended for a "#xD#xA" sequence that is part of an external
3999
 *   parsed entity or the literal entity value of an internal parsed entity
4000
 * - other characters are processed by appending them to the normalized value
4001
 *
4002
 * If the declared value is not CDATA, then the XML processor must further
4003
 * process the normalized attribute value by discarding any leading and
4004
 * trailing space (\#x20) characters, and by replacing sequences of space
4005
 * (\#x20) characters by a single space (\#x20) character.
4006
 * All attributes for which no declaration has been read should be treated
4007
 * by a non-validating parser as if declared CDATA.
4008
 *
4009
 * @param ctxt  an XML parser context
4010
 * @param attlen  attribute len result
4011
 * @param outFlags  resulting XML_ATTVAL_* flags
4012
 * @param special  value from attsSpecial
4013
 * @param isNamespace  whether this is a namespace declaration
4014
 * @returns the AttValue parsed or NULL. The value has to be freed by the
4015
 *     caller if it was copied, this can be detected by val[*len] == 0.
4016
 */
4017
static xmlChar *
4018
xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *attlen, int *outFlags,
4019
896k
                         int special, int isNamespace) {
4020
896k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4021
0
                         XML_MAX_HUGE_LENGTH :
4022
896k
                         XML_MAX_TEXT_LENGTH;
4023
896k
    xmlSBuf buf;
4024
896k
    xmlChar *ret;
4025
896k
    int c, l, quote, entFlags, chunkSize;
4026
896k
    int inSpace = 1;
4027
896k
    int replaceEntities;
4028
896k
    int normalize = (special & XML_SPECIAL_TYPE_MASK) > XML_ATTRIBUTE_CDATA;
4029
896k
    int attvalFlags = 0;
4030
4031
    /* Always expand namespace URIs */
4032
896k
    replaceEntities = (ctxt->replaceEntities) || (isNamespace);
4033
4034
896k
    xmlSBufInit(&buf, maxLength);
4035
4036
896k
    GROW;
4037
4038
896k
    quote = CUR;
4039
896k
    if ((quote != '"') && (quote != '\'')) {
4040
382
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
4041
382
  return(NULL);
4042
382
    }
4043
896k
    NEXTL(1);
4044
4045
896k
    if (ctxt->inSubset == 0)
4046
870k
        entFlags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
4047
25.8k
    else
4048
25.8k
        entFlags = XML_ENT_VALIDATED;
4049
4050
896k
    inSpace = 1;
4051
896k
    chunkSize = 0;
4052
4053
107M
    while (1) {
4054
107M
        if (PARSER_STOPPED(ctxt))
4055
3.48k
            goto error;
4056
4057
107M
        if (CUR_PTR >= ctxt->input->end) {
4058
421
            xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED,
4059
421
                           "AttValue: ' expected\n");
4060
421
            goto error;
4061
421
        }
4062
4063
        /*
4064
         * TODO: Check growth threshold
4065
         */
4066
107M
        if (ctxt->input->end - CUR_PTR < 10)
4067
129k
            GROW;
4068
4069
107M
        c = CUR;
4070
4071
107M
        if (c >= 0x80) {
4072
57.2M
            l = xmlUTF8MultibyteLen(ctxt, CUR_PTR,
4073
57.2M
                    "invalid character in attribute value\n");
4074
57.2M
            if (l == 0) {
4075
756
                if (chunkSize > 0) {
4076
325
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4077
325
                    chunkSize = 0;
4078
325
                }
4079
756
                xmlSBufAddReplChar(&buf);
4080
756
                NEXTL(1);
4081
57.2M
            } else {
4082
57.2M
                chunkSize += l;
4083
57.2M
                NEXTL(l);
4084
57.2M
            }
4085
4086
57.2M
            inSpace = 0;
4087
57.2M
        } else if (c != '&') {
4088
50.1M
            if (c > 0x20) {
4089
46.5M
                if (c == quote)
4090
892k
                    break;
4091
4092
45.6M
                if (c == '<')
4093
190
                    xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL);
4094
4095
45.6M
                chunkSize += 1;
4096
45.6M
                inSpace = 0;
4097
45.6M
            } else if (!IS_BYTE_CHAR(c)) {
4098
2.15k
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
4099
2.15k
                        "invalid character in attribute value\n");
4100
2.15k
                if (chunkSize > 0) {
4101
1.34k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4102
1.34k
                    chunkSize = 0;
4103
1.34k
                }
4104
2.15k
                xmlSBufAddReplChar(&buf);
4105
2.15k
                inSpace = 0;
4106
3.54M
            } else {
4107
                /* Whitespace */
4108
3.54M
                if ((normalize) && (inSpace)) {
4109
                    /* Skip char */
4110
0
                    if (chunkSize > 0) {
4111
0
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4112
0
                        chunkSize = 0;
4113
0
                    }
4114
0
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4115
3.54M
                } else if (c < 0x20) {
4116
                    /* Convert to space */
4117
2.28M
                    if (chunkSize > 0) {
4118
1.18M
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4119
1.18M
                        chunkSize = 0;
4120
1.18M
                    }
4121
4122
2.28M
                    xmlSBufAddCString(&buf, " ", 1);
4123
2.28M
                } else {
4124
1.26M
                    chunkSize += 1;
4125
1.26M
                }
4126
4127
3.54M
                inSpace = 1;
4128
4129
3.54M
                if ((c == 0xD) && (NXT(1) == 0xA))
4130
25.5k
                    CUR_PTR++;
4131
3.54M
            }
4132
4133
49.2M
            NEXTL(1);
4134
49.2M
        } else if (NXT(1) == '#') {
4135
38.2k
            int val;
4136
4137
38.2k
            if (chunkSize > 0) {
4138
33.2k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4139
33.2k
                chunkSize = 0;
4140
33.2k
            }
4141
4142
38.2k
            val = xmlParseCharRef(ctxt);
4143
38.2k
            if (val == 0)
4144
107
                goto error;
4145
4146
38.1k
            if ((val == '&') && (!replaceEntities)) {
4147
                /*
4148
                 * The reparsing will be done in xmlNodeParseContent()
4149
                 * called from SAX2.c
4150
                 */
4151
25.9k
                xmlSBufAddCString(&buf, "&#38;", 5);
4152
25.9k
                inSpace = 0;
4153
25.9k
            } else if (val == ' ') {
4154
1.31k
                if ((normalize) && (inSpace))
4155
0
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4156
1.31k
                else
4157
1.31k
                    xmlSBufAddCString(&buf, " ", 1);
4158
1.31k
                inSpace = 1;
4159
10.9k
            } else {
4160
10.9k
                xmlSBufAddChar(&buf, val);
4161
10.9k
                inSpace = 0;
4162
10.9k
            }
4163
38.1k
        } else {
4164
32.1k
            const xmlChar *name;
4165
32.1k
            xmlEntityPtr ent;
4166
4167
32.1k
            if (chunkSize > 0) {
4168
20.5k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4169
20.5k
                chunkSize = 0;
4170
20.5k
            }
4171
4172
32.1k
            name = xmlParseEntityRefInternal(ctxt);
4173
32.1k
            if (name == NULL) {
4174
                /*
4175
                 * Probably a literal '&' which wasn't escaped.
4176
                 * TODO: Handle gracefully in recovery mode.
4177
                 */
4178
171
                continue;
4179
171
            }
4180
4181
31.9k
            ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1);
4182
31.9k
            if (ent == NULL)
4183
2.59k
                continue;
4184
4185
29.3k
            if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) {
4186
17.1k
                if ((ent->content[0] == '&') && (!replaceEntities))
4187
8.29k
                    xmlSBufAddCString(&buf, "&#38;", 5);
4188
8.89k
                else
4189
8.89k
                    xmlSBufAddString(&buf, ent->content, ent->length);
4190
17.1k
                inSpace = 0;
4191
17.1k
            } else if (replaceEntities) {
4192
0
                if (xmlExpandEntityInAttValue(ctxt, &buf,
4193
0
                        ent->content, ent, normalize, &inSpace, ctxt->inputNr,
4194
0
                        /* check */ 1) > 0)
4195
0
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4196
12.1k
            } else {
4197
12.1k
                if ((ent->flags & entFlags) != entFlags)
4198
664
                    xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr);
4199
4200
12.1k
                if (xmlParserEntityCheck(ctxt, ent->expandedSize)) {
4201
72
                    ent->content[0] = 0;
4202
72
                    goto error;
4203
72
                }
4204
4205
                /*
4206
                 * Just output the reference
4207
                 */
4208
12.0k
                xmlSBufAddCString(&buf, "&", 1);
4209
12.0k
                xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name));
4210
12.0k
                xmlSBufAddCString(&buf, ";", 1);
4211
4212
12.0k
                inSpace = 0;
4213
12.0k
            }
4214
29.3k
  }
4215
107M
    }
4216
4217
892k
    if ((buf.mem == NULL) && (outFlags != NULL)) {
4218
0
        ret = (xmlChar *) CUR_PTR - chunkSize;
4219
4220
0
        if (attlen != NULL)
4221
0
            *attlen = chunkSize;
4222
0
        if ((normalize) && (inSpace) && (chunkSize > 0)) {
4223
0
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4224
0
            *attlen -= 1;
4225
0
        }
4226
4227
        /* Report potential error */
4228
0
        xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4229
892k
    } else {
4230
892k
        if (chunkSize > 0)
4231
846k
            xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4232
4233
892k
        if ((normalize) && (inSpace) && (buf.size > 0)) {
4234
0
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4235
0
            buf.size--;
4236
0
        }
4237
4238
892k
        ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long");
4239
892k
        attvalFlags |= XML_ATTVAL_ALLOC;
4240
4241
892k
        if (ret != NULL) {
4242
892k
            if (attlen != NULL)
4243
0
                *attlen = buf.size;
4244
892k
        }
4245
892k
    }
4246
4247
892k
    if (outFlags != NULL)
4248
0
        *outFlags = attvalFlags;
4249
4250
892k
    NEXTL(1);
4251
4252
892k
    return(ret);
4253
4254
4.08k
error:
4255
4.08k
    xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4256
4.08k
    return(NULL);
4257
896k
}
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
896k
xmlParseAttValue(xmlParserCtxt *ctxt) {
4297
896k
    if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL);
4298
896k
    return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0));
4299
896k
}
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
10.8k
xmlParseSystemLiteral(xmlParserCtxt *ctxt) {
4314
10.8k
    xmlChar *buf = NULL;
4315
10.8k
    int len = 0;
4316
10.8k
    int size = XML_PARSER_BUFFER_SIZE;
4317
10.8k
    int cur, l;
4318
10.8k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4319
0
                    XML_MAX_TEXT_LENGTH :
4320
10.8k
                    XML_MAX_NAME_LENGTH;
4321
10.8k
    xmlChar stop;
4322
4323
10.8k
    if (RAW == '"') {
4324
7.68k
        NEXT;
4325
7.68k
  stop = '"';
4326
7.68k
    } else if (RAW == '\'') {
4327
3.00k
        NEXT;
4328
3.00k
  stop = '\'';
4329
3.00k
    } else {
4330
209
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4331
209
  return(NULL);
4332
209
    }
4333
4334
10.6k
    buf = xmlMalloc(size);
4335
10.6k
    if (buf == NULL) {
4336
0
        xmlErrMemory(ctxt);
4337
0
  return(NULL);
4338
0
    }
4339
10.6k
    cur = xmlCurrentCharRecover(ctxt, &l);
4340
2.07M
    while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */
4341
2.06M
  if (len + 5 >= size) {
4342
7.54k
      xmlChar *tmp;
4343
7.54k
            int newSize;
4344
4345
7.54k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4346
7.54k
            if (newSize < 0) {
4347
6
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral");
4348
6
                xmlFree(buf);
4349
6
                return(NULL);
4350
6
            }
4351
7.54k
      tmp = xmlRealloc(buf, newSize);
4352
7.54k
      if (tmp == NULL) {
4353
0
          xmlFree(buf);
4354
0
    xmlErrMemory(ctxt);
4355
0
    return(NULL);
4356
0
      }
4357
7.54k
      buf = tmp;
4358
7.54k
            size = newSize;
4359
7.54k
  }
4360
2.06M
  COPY_BUF(buf, len, cur);
4361
2.06M
  NEXTL(l);
4362
2.06M
  cur = xmlCurrentCharRecover(ctxt, &l);
4363
2.06M
    }
4364
10.6k
    buf[len] = 0;
4365
10.6k
    if (!IS_CHAR(cur)) {
4366
117
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4367
10.5k
    } else {
4368
10.5k
  NEXT;
4369
10.5k
    }
4370
10.6k
    return(buf);
4371
10.6k
}
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
10.5k
xmlParsePubidLiteral(xmlParserCtxt *ctxt) {
4386
10.5k
    xmlChar *buf = NULL;
4387
10.5k
    int len = 0;
4388
10.5k
    int size = XML_PARSER_BUFFER_SIZE;
4389
10.5k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4390
0
                    XML_MAX_TEXT_LENGTH :
4391
10.5k
                    XML_MAX_NAME_LENGTH;
4392
10.5k
    xmlChar cur;
4393
10.5k
    xmlChar stop;
4394
4395
10.5k
    if (RAW == '"') {
4396
6.86k
        NEXT;
4397
6.86k
  stop = '"';
4398
6.86k
    } else if (RAW == '\'') {
4399
3.62k
        NEXT;
4400
3.62k
  stop = '\'';
4401
3.62k
    } else {
4402
24
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4403
24
  return(NULL);
4404
24
    }
4405
10.4k
    buf = xmlMalloc(size);
4406
10.4k
    if (buf == NULL) {
4407
0
  xmlErrMemory(ctxt);
4408
0
  return(NULL);
4409
0
    }
4410
10.4k
    cur = CUR;
4411
417k
    while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) &&
4412
407k
           (PARSER_STOPPED(ctxt) == 0)) { /* checked */
4413
407k
  if (len + 1 >= size) {
4414
435
      xmlChar *tmp;
4415
435
            int newSize;
4416
4417
435
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4418
435
            if (newSize < 0) {
4419
3
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID");
4420
3
                xmlFree(buf);
4421
3
                return(NULL);
4422
3
            }
4423
432
      tmp = xmlRealloc(buf, newSize);
4424
432
      if (tmp == NULL) {
4425
0
    xmlErrMemory(ctxt);
4426
0
    xmlFree(buf);
4427
0
    return(NULL);
4428
0
      }
4429
432
      buf = tmp;
4430
432
            size = newSize;
4431
432
  }
4432
407k
  buf[len++] = cur;
4433
407k
  NEXT;
4434
407k
  cur = CUR;
4435
407k
    }
4436
10.4k
    buf[len] = 0;
4437
10.4k
    if (cur != stop) {
4438
125
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4439
10.3k
    } else {
4440
10.3k
  NEXTL(1);
4441
10.3k
    }
4442
10.4k
    return(buf);
4443
10.4k
}
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
4.31M
              int isBlank) {
4488
4.31M
    int checkBlanks;
4489
4490
4.31M
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
4491
10.8k
        return;
4492
4493
4.30M
    checkBlanks = (!ctxt->keepBlanks) ||
4494
4.30M
                  (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
4.30M
    if ((checkBlanks) &&
4502
4.30M
        (areBlanks(ctxt, buf, size, isBlank))) {
4503
0
        if ((ctxt->sax->ignorableWhitespace != NULL) &&
4504
0
            (ctxt->keepBlanks))
4505
0
            ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
4506
4.30M
    } else {
4507
4.30M
        if (ctxt->sax->characters != NULL)
4508
4.30M
            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
4.30M
        if ((checkBlanks) && (*ctxt->space == -1))
4515
1.16M
            *ctxt->space = -2;
4516
4.30M
    }
4517
4.30M
}
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
1.93M
xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) {
4534
1.93M
    const xmlChar *in;
4535
1.93M
    int line = ctxt->input->line;
4536
1.93M
    int col = ctxt->input->col;
4537
1.93M
    int ccol;
4538
1.93M
    int terminate = 0;
4539
4540
1.93M
    GROW;
4541
    /*
4542
     * Accelerated common case where input don't need to be
4543
     * modified before passing it to the handler.
4544
     */
4545
1.93M
    in = ctxt->input->cur;
4546
1.95M
    do {
4547
2.44M
get_more_space:
4548
3.19M
        while (*in == 0x20) { in++; ctxt->input->col++; }
4549
2.44M
        if (*in == 0xA) {
4550
1.54M
            do {
4551
1.54M
                ctxt->input->line++; ctxt->input->col = 1;
4552
1.54M
                in++;
4553
1.54M
            } while (*in == 0xA);
4554
487k
            goto get_more_space;
4555
487k
        }
4556
1.95M
        if (*in == '<') {
4557
683k
            while (in > ctxt->input->cur) {
4558
341k
                const xmlChar *tmp = ctxt->input->cur;
4559
341k
                size_t nbchar = in - tmp;
4560
4561
341k
                if (nbchar > XML_MAX_ITEMS)
4562
0
                    nbchar = XML_MAX_ITEMS;
4563
341k
                ctxt->input->cur += nbchar;
4564
4565
341k
                xmlCharacters(ctxt, tmp, nbchar, 1);
4566
341k
            }
4567
341k
            return;
4568
341k
        }
4569
4570
4.09M
get_more:
4571
4.09M
        ccol = ctxt->input->col;
4572
47.2M
        while (test_char_data[*in]) {
4573
43.1M
            in++;
4574
43.1M
            ccol++;
4575
43.1M
        }
4576
4.09M
        ctxt->input->col = ccol;
4577
4.09M
        if (*in == 0xA) {
4578
1.72M
            do {
4579
1.72M
                ctxt->input->line++; ctxt->input->col = 1;
4580
1.72M
                in++;
4581
1.72M
            } while (*in == 0xA);
4582
287k
            goto get_more;
4583
287k
        }
4584
3.80M
        if (*in == ']') {
4585
2.19M
            size_t avail = ctxt->input->end - in;
4586
4587
2.19M
            if (partial && avail < 2) {
4588
30
                terminate = 1;
4589
30
                goto invoke_callback;
4590
30
            }
4591
2.19M
            if (in[1] == ']') {
4592
1.96M
                if (partial && avail < 3) {
4593
908
                    terminate = 1;
4594
908
                    goto invoke_callback;
4595
908
                }
4596
1.96M
                if (in[2] == '>')
4597
532
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4598
1.96M
            }
4599
4600
2.19M
            in++;
4601
2.19M
            ctxt->input->col++;
4602
2.19M
            goto get_more;
4603
2.19M
        }
4604
4605
1.61M
invoke_callback:
4606
3.17M
        while (in > ctxt->input->cur) {
4607
1.55M
            const xmlChar *tmp = ctxt->input->cur;
4608
1.55M
            size_t nbchar = in - tmp;
4609
4610
1.55M
            if (nbchar > XML_MAX_ITEMS)
4611
0
                nbchar = XML_MAX_ITEMS;
4612
1.55M
            ctxt->input->cur += nbchar;
4613
4614
1.55M
            xmlCharacters(ctxt, tmp, nbchar, 0);
4615
4616
1.55M
            line = ctxt->input->line;
4617
1.55M
            col = ctxt->input->col;
4618
1.55M
        }
4619
1.61M
        ctxt->input->cur = in;
4620
1.61M
        if (*in == 0xD) {
4621
108k
            in++;
4622
108k
            if (*in == 0xA) {
4623
28.1k
                ctxt->input->cur = in;
4624
28.1k
                in++;
4625
28.1k
                ctxt->input->line++; ctxt->input->col = 1;
4626
28.1k
                continue; /* while */
4627
28.1k
            }
4628
80.3k
            in--;
4629
80.3k
        }
4630
1.58M
        if (*in == '<') {
4631
817k
            return;
4632
817k
        }
4633
769k
        if (*in == '&') {
4634
135k
            return;
4635
135k
        }
4636
634k
        if (terminate) {
4637
938
            return;
4638
938
        }
4639
633k
        SHRINK;
4640
633k
        GROW;
4641
633k
        in = ctxt->input->cur;
4642
662k
    } while (((*in >= 0x20) && (*in <= 0x7F)) ||
4643
640k
             (*in == 0x09) || (*in == 0x0a));
4644
635k
    ctxt->input->line = line;
4645
635k
    ctxt->input->col = col;
4646
635k
    xmlParseCharDataComplex(ctxt, partial);
4647
635k
}
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
635k
xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
4661
635k
    xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5];
4662
635k
    int nbchar = 0;
4663
635k
    int cur, l;
4664
4665
635k
    cur = xmlCurrentCharRecover(ctxt, &l);
4666
265M
    while ((cur != '<') && /* checked */
4667
265M
           (cur != '&') &&
4668
265M
     (IS_CHAR(cur))) {
4669
264M
        if (cur == ']') {
4670
2.34M
            size_t avail = ctxt->input->end - ctxt->input->cur;
4671
4672
2.34M
            if (partial && avail < 2)
4673
157
                break;
4674
2.34M
            if (NXT(1) == ']') {
4675
2.10M
                if (partial && avail < 3)
4676
751
                    break;
4677
2.10M
                if (NXT(2) == '>')
4678
813
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4679
2.10M
            }
4680
2.34M
        }
4681
4682
264M
  COPY_BUF(buf, nbchar, cur);
4683
  /* move current position before possible calling of ctxt->sax->characters */
4684
264M
  NEXTL(l);
4685
264M
  if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) {
4686
2.12M
      buf[nbchar] = 0;
4687
4688
2.12M
            xmlCharacters(ctxt, buf, nbchar, 0);
4689
2.12M
      nbchar = 0;
4690
2.12M
            SHRINK;
4691
2.12M
  }
4692
264M
  cur = xmlCurrentCharRecover(ctxt, &l);
4693
264M
    }
4694
635k
    if (nbchar != 0) {
4695
298k
        buf[nbchar] = 0;
4696
4697
298k
        xmlCharacters(ctxt, buf, nbchar, 0);
4698
298k
    }
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
635k
    if (ctxt->input->cur < ctxt->input->end) {
4707
230k
        if ((cur == 0) && (CUR != 0)) {
4708
1.03k
            if (partial == 0) {
4709
272
                xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4710
272
                        "Incomplete UTF-8 sequence starting with %02X\n", CUR);
4711
272
                NEXTL(1);
4712
272
            }
4713
229k
        } else if ((cur != '<') && (cur != '&') && (cur != ']')) {
4714
            /* Generate the error and skip the offending character */
4715
4.94k
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4716
4.94k
                              "PCDATA invalid Char value %d\n", cur);
4717
4.94k
            NEXTL(l);
4718
4.94k
        }
4719
230k
    }
4720
635k
}
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
28.9k
xmlParseExternalID(xmlParserCtxt *ctxt, xmlChar **publicId, int strict) {
4756
28.9k
    xmlChar *URI = NULL;
4757
4758
28.9k
    *publicId = NULL;
4759
28.9k
    if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) {
4760
1.10k
        SKIP(6);
4761
1.10k
  if (SKIP_BLANKS == 0) {
4762
17
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4763
17
                     "Space required after 'SYSTEM'\n");
4764
17
  }
4765
1.10k
  URI = xmlParseSystemLiteral(ctxt);
4766
1.10k
  if (URI == NULL) {
4767
17
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4768
17
        }
4769
27.8k
    } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) {
4770
10.5k
        SKIP(6);
4771
10.5k
  if (SKIP_BLANKS == 0) {
4772
29
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4773
29
        "Space required after 'PUBLIC'\n");
4774
29
  }
4775
10.5k
  *publicId = xmlParsePubidLiteral(ctxt);
4776
10.5k
  if (*publicId == NULL) {
4777
27
      xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL);
4778
27
  }
4779
10.5k
  if (strict) {
4780
      /*
4781
       * We don't handle [83] so "S SystemLiteral" is required.
4782
       */
4783
8.75k
      if (SKIP_BLANKS == 0) {
4784
202
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4785
202
      "Space required after the Public Identifier\n");
4786
202
      }
4787
8.75k
  } else {
4788
      /*
4789
       * We handle [83] so we return immediately, if
4790
       * "S SystemLiteral" is not detected. We skip blanks if no
4791
             * system literal was found, but this is harmless since we must
4792
             * be at the end of a NotationDecl.
4793
       */
4794
1.76k
      if (SKIP_BLANKS == 0) return(NULL);
4795
1.30k
      if ((CUR != '\'') && (CUR != '"')) return(NULL);
4796
1.30k
  }
4797
9.79k
  URI = xmlParseSystemLiteral(ctxt);
4798
9.79k
  if (URI == NULL) {
4799
198
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4800
198
        }
4801
9.79k
    }
4802
28.2k
    return(URI);
4803
28.9k
}
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
76.2k
                       size_t len, size_t size) {
4820
76.2k
    int q, ql;
4821
76.2k
    int r, rl;
4822
76.2k
    int cur, l;
4823
76.2k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4824
0
                    XML_MAX_HUGE_LENGTH :
4825
76.2k
                    XML_MAX_TEXT_LENGTH;
4826
4827
76.2k
    if (buf == NULL) {
4828
15.7k
        len = 0;
4829
15.7k
  size = XML_PARSER_BUFFER_SIZE;
4830
15.7k
  buf = xmlMalloc(size);
4831
15.7k
  if (buf == NULL) {
4832
0
      xmlErrMemory(ctxt);
4833
0
      return;
4834
0
  }
4835
15.7k
    }
4836
76.2k
    q = xmlCurrentCharRecover(ctxt, &ql);
4837
76.2k
    if (q == 0)
4838
208
        goto not_terminated;
4839
76.0k
    if (!IS_CHAR(q)) {
4840
67
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4841
67
                          "xmlParseComment: invalid xmlChar value %d\n",
4842
67
                    q);
4843
67
  xmlFree (buf);
4844
67
  return;
4845
67
    }
4846
76.0k
    NEXTL(ql);
4847
76.0k
    r = xmlCurrentCharRecover(ctxt, &rl);
4848
76.0k
    if (r == 0)
4849
31
        goto not_terminated;
4850
75.9k
    if (!IS_CHAR(r)) {
4851
16
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4852
16
                          "xmlParseComment: invalid xmlChar value %d\n",
4853
16
                    r);
4854
16
  xmlFree (buf);
4855
16
  return;
4856
16
    }
4857
75.9k
    NEXTL(rl);
4858
75.9k
    cur = xmlCurrentCharRecover(ctxt, &l);
4859
75.9k
    if (cur == 0)
4860
11
        goto not_terminated;
4861
10.0M
    while (IS_CHAR(cur) && /* checked */
4862
10.0M
           ((cur != '>') ||
4863
9.99M
      (r != '-') || (q != '-'))) {
4864
9.99M
  if ((r == '-') && (q == '-')) {
4865
1.31k
      xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL);
4866
1.31k
  }
4867
9.99M
  if (len + 5 >= size) {
4868
13.7k
      xmlChar *tmp;
4869
13.7k
            int newSize;
4870
4871
13.7k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4872
13.7k
            if (newSize < 0) {
4873
0
                xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4874
0
                             "Comment too big found", NULL);
4875
0
                xmlFree (buf);
4876
0
                return;
4877
0
            }
4878
13.7k
      tmp = xmlRealloc(buf, newSize);
4879
13.7k
      if (tmp == NULL) {
4880
0
    xmlErrMemory(ctxt);
4881
0
    xmlFree(buf);
4882
0
    return;
4883
0
      }
4884
13.7k
      buf = tmp;
4885
13.7k
            size = newSize;
4886
13.7k
  }
4887
9.99M
  COPY_BUF(buf, len, q);
4888
4889
9.99M
  q = r;
4890
9.99M
  ql = rl;
4891
9.99M
  r = cur;
4892
9.99M
  rl = l;
4893
4894
9.99M
  NEXTL(l);
4895
9.99M
  cur = xmlCurrentCharRecover(ctxt, &l);
4896
4897
9.99M
    }
4898
75.9k
    buf[len] = 0;
4899
75.9k
    if (cur == 0) {
4900
153
  xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4901
153
                       "Comment not terminated \n<!--%.50s\n", buf);
4902
75.8k
    } else if (!IS_CHAR(cur)) {
4903
127
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4904
127
                          "xmlParseComment: invalid xmlChar value %d\n",
4905
127
                    cur);
4906
75.6k
    } else {
4907
75.6k
        NEXT;
4908
75.6k
  if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
4909
75.6k
      (!ctxt->disableSAX))
4910
75.1k
      ctxt->sax->comment(ctxt->userData, buf);
4911
75.6k
    }
4912
75.9k
    xmlFree(buf);
4913
75.9k
    return;
4914
250
not_terminated:
4915
250
    xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4916
250
       "Comment not terminated\n", NULL);
4917
250
    xmlFree(buf);
4918
250
}
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
393k
xmlParseComment(xmlParserCtxt *ctxt) {
4933
393k
    xmlChar *buf = NULL;
4934
393k
    size_t size = XML_PARSER_BUFFER_SIZE;
4935
393k
    size_t len = 0;
4936
393k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4937
0
                       XML_MAX_HUGE_LENGTH :
4938
393k
                       XML_MAX_TEXT_LENGTH;
4939
393k
    const xmlChar *in;
4940
393k
    size_t nbchar = 0;
4941
393k
    int ccol;
4942
4943
    /*
4944
     * Check that there is a comment right here.
4945
     */
4946
393k
    if ((RAW != '<') || (NXT(1) != '!'))
4947
0
        return;
4948
393k
    SKIP(2);
4949
393k
    if ((RAW != '-') || (NXT(1) != '-'))
4950
19
        return;
4951
393k
    SKIP(2);
4952
393k
    GROW;
4953
4954
    /*
4955
     * Accelerated common case where input don't need to be
4956
     * modified before passing it to the handler.
4957
     */
4958
393k
    in = ctxt->input->cur;
4959
393k
    do {
4960
393k
  if (*in == 0xA) {
4961
31.3k
      do {
4962
31.3k
    ctxt->input->line++; ctxt->input->col = 1;
4963
31.3k
    in++;
4964
31.3k
      } while (*in == 0xA);
4965
30.3k
  }
4966
830k
get_more:
4967
830k
        ccol = ctxt->input->col;
4968
5.99M
  while (((*in > '-') && (*in <= 0x7F)) ||
4969
1.28M
         ((*in >= 0x20) && (*in < '-')) ||
4970
5.16M
         (*in == 0x09)) {
4971
5.16M
        in++;
4972
5.16M
        ccol++;
4973
5.16M
  }
4974
830k
  ctxt->input->col = ccol;
4975
830k
  if (*in == 0xA) {
4976
381k
      do {
4977
381k
    ctxt->input->line++; ctxt->input->col = 1;
4978
381k
    in++;
4979
381k
      } while (*in == 0xA);
4980
150k
      goto get_more;
4981
150k
  }
4982
679k
  nbchar = in - ctxt->input->cur;
4983
  /*
4984
   * save current set of data
4985
   */
4986
679k
  if (nbchar > 0) {
4987
498k
            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
498k
            if (buf == NULL) {
4994
237k
                if ((*in == '-') && (in[1] == '-'))
4995
9.21k
                    size = nbchar + 1;
4996
227k
                else
4997
227k
                    size = XML_PARSER_BUFFER_SIZE + nbchar;
4998
237k
                buf = xmlMalloc(size);
4999
237k
                if (buf == NULL) {
5000
0
                    xmlErrMemory(ctxt);
5001
0
                    return;
5002
0
                }
5003
237k
                len = 0;
5004
261k
            } else if (len + nbchar + 1 >= size) {
5005
1.58k
                xmlChar *new_buf;
5006
1.58k
                size += len + nbchar + XML_PARSER_BUFFER_SIZE;
5007
1.58k
                new_buf = xmlRealloc(buf, size);
5008
1.58k
                if (new_buf == NULL) {
5009
0
                    xmlErrMemory(ctxt);
5010
0
                    xmlFree(buf);
5011
0
                    return;
5012
0
                }
5013
1.58k
                buf = new_buf;
5014
1.58k
            }
5015
498k
            memcpy(&buf[len], ctxt->input->cur, nbchar);
5016
498k
            len += nbchar;
5017
498k
            buf[len] = 0;
5018
498k
  }
5019
679k
  ctxt->input->cur = in;
5020
679k
  if (*in == 0xA) {
5021
0
      in++;
5022
0
      ctxt->input->line++; ctxt->input->col = 1;
5023
0
  }
5024
679k
  if (*in == 0xD) {
5025
85.2k
      in++;
5026
85.2k
      if (*in == 0xA) {
5027
20.9k
    ctxt->input->cur = in;
5028
20.9k
    in++;
5029
20.9k
    ctxt->input->line++; ctxt->input->col = 1;
5030
20.9k
    goto get_more;
5031
20.9k
      }
5032
64.2k
      in--;
5033
64.2k
  }
5034
658k
  SHRINK;
5035
658k
  GROW;
5036
658k
  in = ctxt->input->cur;
5037
658k
  if (*in == '-') {
5038
582k
      if (in[1] == '-') {
5039
319k
          if (in[2] == '>') {
5040
317k
        SKIP(3);
5041
317k
        if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
5042
317k
            (!ctxt->disableSAX)) {
5043
317k
      if (buf != NULL)
5044
176k
          ctxt->sax->comment(ctxt->userData, buf);
5045
140k
      else
5046
140k
          ctxt->sax->comment(ctxt->userData, BAD_CAST "");
5047
317k
        }
5048
317k
        if (buf != NULL)
5049
176k
            xmlFree(buf);
5050
317k
        return;
5051
317k
    }
5052
1.85k
    if (buf != NULL) {
5053
1.78k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5054
1.78k
                          "Double hyphen within comment: "
5055
1.78k
                                      "<!--%.50s\n",
5056
1.78k
              buf);
5057
1.78k
    } else
5058
71
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5059
71
                          "Double hyphen within comment\n", NULL);
5060
1.85k
    in++;
5061
1.85k
    ctxt->input->col++;
5062
1.85k
      }
5063
265k
      in++;
5064
265k
      ctxt->input->col++;
5065
265k
      goto get_more;
5066
582k
  }
5067
658k
    } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a));
5068
76.2k
    xmlParseCommentComplex(ctxt, buf, len, size);
5069
76.2k
}
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
391k
xmlParsePITarget(xmlParserCtxt *ctxt) {
5085
391k
    const xmlChar *name;
5086
5087
391k
    name = xmlParseName(ctxt);
5088
391k
    if ((name != NULL) &&
5089
391k
        ((name[0] == 'x') || (name[0] == 'X')) &&
5090
190k
        ((name[1] == 'm') || (name[1] == 'M')) &&
5091
56.3k
        ((name[2] == 'l') || (name[2] == 'L'))) {
5092
53.1k
  int i;
5093
53.1k
  if ((name[0] == 'x') && (name[1] == 'm') &&
5094
14.6k
      (name[2] == 'l') && (name[3] == 0)) {
5095
966
      xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5096
966
     "XML declaration allowed only at the start of the document\n");
5097
966
      return(name);
5098
52.2k
  } else if (name[3] == 0) {
5099
2.57k
      xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL);
5100
2.57k
      return(name);
5101
2.57k
  }
5102
147k
  for (i = 0;;i++) {
5103
147k
      if (xmlW3CPIs[i] == NULL) break;
5104
98.8k
      if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i]))
5105
1.06k
          return(name);
5106
98.8k
  }
5107
48.5k
  xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5108
48.5k
          "xmlParsePITarget: invalid name prefix 'xml'\n",
5109
48.5k
          NULL, NULL);
5110
48.5k
    }
5111
387k
    if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) {
5112
438
  xmlNsErr(ctxt, XML_NS_ERR_COLON,
5113
438
     "colons are forbidden from PI names '%s'\n", name, NULL, NULL);
5114
438
    }
5115
387k
    return(name);
5116
391k
}
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
391k
xmlParsePI(xmlParserCtxt *ctxt) {
5199
391k
    xmlChar *buf = NULL;
5200
391k
    size_t len = 0;
5201
391k
    size_t size = XML_PARSER_BUFFER_SIZE;
5202
391k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5203
0
                       XML_MAX_HUGE_LENGTH :
5204
391k
                       XML_MAX_TEXT_LENGTH;
5205
391k
    int cur, l;
5206
391k
    const xmlChar *target;
5207
5208
391k
    if ((RAW == '<') && (NXT(1) == '?')) {
5209
  /*
5210
   * this is a Processing Instruction.
5211
   */
5212
391k
  SKIP(2);
5213
5214
  /*
5215
   * Parse the target name and check for special support like
5216
   * namespace.
5217
   */
5218
391k
        target = xmlParsePITarget(ctxt);
5219
391k
  if (target != NULL) {
5220
391k
      if ((RAW == '?') && (NXT(1) == '>')) {
5221
224k
    SKIP(2);
5222
5223
    /*
5224
     * SAX: PI detected.
5225
     */
5226
224k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5227
224k
        (ctxt->sax->processingInstruction != NULL))
5228
0
        ctxt->sax->processingInstruction(ctxt->userData,
5229
0
                                         target, NULL);
5230
224k
    return;
5231
224k
      }
5232
167k
      buf = xmlMalloc(size);
5233
167k
      if (buf == NULL) {
5234
0
    xmlErrMemory(ctxt);
5235
0
    return;
5236
0
      }
5237
167k
      if (SKIP_BLANKS == 0) {
5238
10.5k
    xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED,
5239
10.5k
        "ParsePI: PI %s space expected\n", target);
5240
10.5k
      }
5241
167k
      cur = xmlCurrentCharRecover(ctxt, &l);
5242
9.67M
      while (IS_CHAR(cur) && /* checked */
5243
9.65M
       ((cur != '?') || (NXT(1) != '>'))) {
5244
9.50M
    if (len + 5 >= size) {
5245
21.6k
        xmlChar *tmp;
5246
21.6k
                    int newSize;
5247
5248
21.6k
                    newSize = xmlGrowCapacity(size, 1, 1, maxLength);
5249
21.6k
                    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
21.6k
        tmp = xmlRealloc(buf, newSize);
5256
21.6k
        if (tmp == NULL) {
5257
0
      xmlErrMemory(ctxt);
5258
0
      xmlFree(buf);
5259
0
      return;
5260
0
        }
5261
21.6k
        buf = tmp;
5262
21.6k
                    size = newSize;
5263
21.6k
    }
5264
9.50M
    COPY_BUF(buf, len, cur);
5265
9.50M
    NEXTL(l);
5266
9.50M
    cur = xmlCurrentCharRecover(ctxt, &l);
5267
9.50M
      }
5268
167k
      buf[len] = 0;
5269
167k
      if (cur != '?') {
5270
15.5k
    xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5271
15.5k
          "ParsePI: PI %s never end ...\n", target);
5272
151k
      } else {
5273
151k
    SKIP(2);
5274
5275
151k
#ifdef LIBXML_CATALOG_ENABLED
5276
151k
    if ((ctxt->inSubset == 0) &&
5277
144k
        (xmlStrEqual(target, XML_CATALOG_PI))) {
5278
60.8k
        xmlCatalogAllow allow = xmlCatalogGetDefaults();
5279
5280
60.8k
        if ((ctxt->options & XML_PARSE_CATALOG_PI) &&
5281
0
                        ((allow == XML_CATA_ALLOW_DOCUMENT) ||
5282
0
       (allow == XML_CATA_ALLOW_ALL)))
5283
0
      xmlParseCatalogPI(ctxt, buf);
5284
60.8k
    }
5285
151k
#endif
5286
5287
    /*
5288
     * SAX: PI detected.
5289
     */
5290
151k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5291
150k
        (ctxt->sax->processingInstruction != NULL))
5292
0
        ctxt->sax->processingInstruction(ctxt->userData,
5293
0
                                         target, buf);
5294
151k
      }
5295
167k
      xmlFree(buf);
5296
167k
  } else {
5297
83
      xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL);
5298
83
  }
5299
391k
    }
5300
391k
}
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
2.13k
xmlParseNotationDecl(xmlParserCtxt *ctxt) {
5323
2.13k
    const xmlChar *name;
5324
2.13k
    xmlChar *Pubid;
5325
2.13k
    xmlChar *Systemid;
5326
5327
2.13k
    if ((CUR != '<') || (NXT(1) != '!'))
5328
0
        return;
5329
2.13k
    SKIP(2);
5330
5331
2.13k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5332
2.08k
#ifdef LIBXML_VALID_ENABLED
5333
2.08k
  int oldInputNr = ctxt->inputNr;
5334
2.08k
#endif
5335
5336
2.08k
  SKIP(8);
5337
2.08k
  if (SKIP_BLANKS_PE == 0) {
5338
5
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5339
5
         "Space required after '<!NOTATION'\n");
5340
5
      return;
5341
5
  }
5342
5343
2.08k
        name = xmlParseName(ctxt);
5344
2.08k
  if (name == NULL) {
5345
4
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5346
4
      return;
5347
4
  }
5348
2.08k
  if (xmlStrchr(name, ':') != NULL) {
5349
2
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5350
2
         "colons are forbidden from notation names '%s'\n",
5351
2
         name, NULL, NULL);
5352
2
  }
5353
2.08k
  if (SKIP_BLANKS_PE == 0) {
5354
4
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5355
4
         "Space required after the NOTATION name'\n");
5356
4
      return;
5357
4
  }
5358
5359
  /*
5360
   * Parse the IDs.
5361
   */
5362
2.07k
  Systemid = xmlParseExternalID(ctxt, &Pubid, 0);
5363
2.07k
  SKIP_BLANKS_PE;
5364
5365
2.07k
  if (RAW == '>') {
5366
2.02k
#ifdef LIBXML_VALID_ENABLED
5367
2.02k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5368
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5369
0
                           "Notation declaration doesn't start and stop"
5370
0
                                 " in the same entity\n",
5371
0
                                 NULL, NULL);
5372
0
      }
5373
2.02k
#endif
5374
2.02k
      NEXT;
5375
2.02k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5376
2.02k
    (ctxt->sax->notationDecl != NULL))
5377
2.02k
    ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid);
5378
2.02k
  } else {
5379
48
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5380
48
  }
5381
2.07k
  if (Systemid != NULL) xmlFree(Systemid);
5382
2.07k
  if (Pubid != NULL) xmlFree(Pubid);
5383
2.07k
    }
5384
2.13k
}
5385
5386
/**
5387
 * Parse an entity declaration. Always consumes '<!'.
5388
 *
5389
 * @deprecated Internal function, don't use.
5390
 *
5391
 *     [70] EntityDecl ::= GEDecl | PEDecl
5392
 *
5393
 *     [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>'
5394
 *
5395
 *     [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>'
5396
 *
5397
 *     [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?)
5398
 *
5399
 *     [74] PEDef ::= EntityValue | ExternalID
5400
 *
5401
 *     [76] NDataDecl ::= S 'NDATA' S Name
5402
 *
5403
 * [ VC: Notation Declared ]
5404
 * The Name must match the declared name of a notation.
5405
 *
5406
 * @param ctxt  an XML parser context
5407
 */
5408
5409
void
5410
31.2k
xmlParseEntityDecl(xmlParserCtxt *ctxt) {
5411
31.2k
    const xmlChar *name = NULL;
5412
31.2k
    xmlChar *value = NULL;
5413
31.2k
    xmlChar *URI = NULL, *literal = NULL;
5414
31.2k
    const xmlChar *ndata = NULL;
5415
31.2k
    int isParameter = 0;
5416
31.2k
    xmlChar *orig = NULL;
5417
5418
31.2k
    if ((CUR != '<') || (NXT(1) != '!'))
5419
0
        return;
5420
31.2k
    SKIP(2);
5421
5422
    /* GROW; done in the caller */
5423
31.2k
    if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5424
30.9k
#ifdef LIBXML_VALID_ENABLED
5425
30.9k
  int oldInputNr = ctxt->inputNr;
5426
30.9k
#endif
5427
5428
30.9k
  SKIP(6);
5429
30.9k
  if (SKIP_BLANKS_PE == 0) {
5430
58
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5431
58
         "Space required after '<!ENTITY'\n");
5432
58
  }
5433
5434
30.9k
  if (RAW == '%') {
5435
2.84k
      NEXT;
5436
2.84k
      if (SKIP_BLANKS_PE == 0) {
5437
18
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5438
18
             "Space required after '%%'\n");
5439
18
      }
5440
2.84k
      isParameter = 1;
5441
2.84k
  }
5442
5443
30.9k
        name = xmlParseName(ctxt);
5444
30.9k
  if (name == NULL) {
5445
63
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5446
63
                     "xmlParseEntityDecl: no name\n");
5447
63
            return;
5448
63
  }
5449
30.9k
  if (xmlStrchr(name, ':') != NULL) {
5450
12
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5451
12
         "colons are forbidden from entities names '%s'\n",
5452
12
         name, NULL, NULL);
5453
12
  }
5454
30.9k
  if (SKIP_BLANKS_PE == 0) {
5455
95
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5456
95
         "Space required after the entity name\n");
5457
95
  }
5458
5459
  /*
5460
   * handle the various case of definitions...
5461
   */
5462
30.9k
  if (isParameter) {
5463
2.84k
      if ((RAW == '"') || (RAW == '\'')) {
5464
2.68k
          value = xmlParseEntityValue(ctxt, &orig);
5465
2.68k
    if (value) {
5466
2.67k
        if ((ctxt->sax != NULL) &&
5467
2.67k
      (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5468
2.64k
      ctxt->sax->entityDecl(ctxt->userData, name,
5469
2.64k
                        XML_INTERNAL_PARAMETER_ENTITY,
5470
2.64k
            NULL, NULL, value);
5471
2.67k
    }
5472
2.68k
      } else {
5473
155
          URI = xmlParseExternalID(ctxt, &literal, 1);
5474
155
    if ((URI == NULL) && (literal == NULL)) {
5475
13
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5476
13
    }
5477
155
    if (URI) {
5478
139
                    if (xmlStrchr(URI, '#')) {
5479
2
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5480
137
                    } else {
5481
137
                        if ((ctxt->sax != NULL) &&
5482
137
                            (!ctxt->disableSAX) &&
5483
133
                            (ctxt->sax->entityDecl != NULL))
5484
133
                            ctxt->sax->entityDecl(ctxt->userData, name,
5485
133
                                        XML_EXTERNAL_PARAMETER_ENTITY,
5486
133
                                        literal, URI, NULL);
5487
137
                    }
5488
139
    }
5489
155
      }
5490
28.0k
  } else {
5491
28.0k
      if ((RAW == '"') || (RAW == '\'')) {
5492
19.0k
          value = xmlParseEntityValue(ctxt, &orig);
5493
19.0k
    if ((ctxt->sax != NULL) &&
5494
19.0k
        (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5495
18.1k
        ctxt->sax->entityDecl(ctxt->userData, name,
5496
18.1k
        XML_INTERNAL_GENERAL_ENTITY,
5497
18.1k
        NULL, NULL, value);
5498
    /*
5499
     * For expat compatibility in SAX mode.
5500
     */
5501
19.0k
    if ((ctxt->myDoc == NULL) ||
5502
19.0k
        (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) {
5503
0
        if (ctxt->myDoc == NULL) {
5504
0
      ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5505
0
      if (ctxt->myDoc == NULL) {
5506
0
          xmlErrMemory(ctxt);
5507
0
          goto done;
5508
0
      }
5509
0
      ctxt->myDoc->properties = XML_DOC_INTERNAL;
5510
0
        }
5511
0
        if (ctxt->myDoc->intSubset == NULL) {
5512
0
      ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5513
0
              BAD_CAST "fake", NULL, NULL);
5514
0
                        if (ctxt->myDoc->intSubset == NULL) {
5515
0
                            xmlErrMemory(ctxt);
5516
0
                            goto done;
5517
0
                        }
5518
0
                    }
5519
5520
0
        xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY,
5521
0
                    NULL, NULL, value);
5522
0
    }
5523
19.0k
      } else {
5524
9.01k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5525
9.01k
    if ((URI == NULL) && (literal == NULL)) {
5526
194
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5527
194
    }
5528
9.01k
    if (URI) {
5529
8.71k
                    if (xmlStrchr(URI, '#')) {
5530
14
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5531
14
                    }
5532
8.71k
    }
5533
9.01k
    if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) {
5534
378
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5535
378
           "Space required before 'NDATA'\n");
5536
378
    }
5537
9.01k
    if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) {
5538
191
        SKIP(5);
5539
191
        if (SKIP_BLANKS_PE == 0) {
5540
3
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5541
3
               "Space required after 'NDATA'\n");
5542
3
        }
5543
191
        ndata = xmlParseName(ctxt);
5544
191
        if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5545
186
            (ctxt->sax->unparsedEntityDecl != NULL))
5546
186
      ctxt->sax->unparsedEntityDecl(ctxt->userData, name,
5547
186
            literal, URI, ndata);
5548
8.82k
    } else {
5549
8.82k
        if ((ctxt->sax != NULL) &&
5550
8.82k
            (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5551
8.35k
      ctxt->sax->entityDecl(ctxt->userData, name,
5552
8.35k
            XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5553
8.35k
            literal, URI, NULL);
5554
        /*
5555
         * For expat compatibility in SAX mode.
5556
         * assuming the entity replacement was asked for
5557
         */
5558
8.82k
        if ((ctxt->replaceEntities != 0) &&
5559
0
      ((ctxt->myDoc == NULL) ||
5560
0
      (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) {
5561
0
      if (ctxt->myDoc == NULL) {
5562
0
          ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5563
0
          if (ctxt->myDoc == NULL) {
5564
0
              xmlErrMemory(ctxt);
5565
0
        goto done;
5566
0
          }
5567
0
          ctxt->myDoc->properties = XML_DOC_INTERNAL;
5568
0
      }
5569
5570
0
      if (ctxt->myDoc->intSubset == NULL) {
5571
0
          ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5572
0
            BAD_CAST "fake", NULL, NULL);
5573
0
                            if (ctxt->myDoc->intSubset == NULL) {
5574
0
                                xmlErrMemory(ctxt);
5575
0
                                goto done;
5576
0
                            }
5577
0
                        }
5578
0
      xmlSAX2EntityDecl(ctxt, name,
5579
0
                  XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5580
0
                  literal, URI, NULL);
5581
0
        }
5582
8.82k
    }
5583
9.01k
      }
5584
28.0k
  }
5585
30.9k
  SKIP_BLANKS_PE;
5586
30.9k
  if (RAW != '>') {
5587
965
      xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED,
5588
965
              "xmlParseEntityDecl: entity %s not terminated\n", name);
5589
29.9k
  } else {
5590
29.9k
#ifdef LIBXML_VALID_ENABLED
5591
29.9k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5592
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5593
0
                           "Entity declaration doesn't start and stop in"
5594
0
                                 " the same entity\n",
5595
0
                                 NULL, NULL);
5596
0
      }
5597
29.9k
#endif
5598
29.9k
      NEXT;
5599
29.9k
  }
5600
30.9k
  if (orig != NULL) {
5601
      /*
5602
       * Ugly mechanism to save the raw entity value.
5603
       */
5604
21.6k
      xmlEntityPtr cur = NULL;
5605
5606
21.6k
      if (isParameter) {
5607
2.67k
          if ((ctxt->sax != NULL) &&
5608
2.67k
        (ctxt->sax->getParameterEntity != NULL))
5609
2.67k
        cur = ctxt->sax->getParameterEntity(ctxt->userData, name);
5610
18.9k
      } else {
5611
18.9k
          if ((ctxt->sax != NULL) &&
5612
18.9k
        (ctxt->sax->getEntity != NULL))
5613
18.9k
        cur = ctxt->sax->getEntity(ctxt->userData, name);
5614
18.9k
    if ((cur == NULL) && (ctxt->userData==ctxt)) {
5615
765
        cur = xmlSAX2GetEntity(ctxt, name);
5616
765
    }
5617
18.9k
      }
5618
21.6k
            if ((cur != NULL) && (cur->orig == NULL)) {
5619
12.0k
    cur->orig = orig;
5620
12.0k
                orig = NULL;
5621
12.0k
      }
5622
21.6k
  }
5623
5624
30.9k
done:
5625
30.9k
  if (value != NULL) xmlFree(value);
5626
30.9k
  if (URI != NULL) xmlFree(URI);
5627
30.9k
  if (literal != NULL) xmlFree(literal);
5628
30.9k
        if (orig != NULL) xmlFree(orig);
5629
30.9k
    }
5630
31.2k
}
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
40.7k
xmlParseDefaultDecl(xmlParserCtxt *ctxt, xmlChar **value) {
5663
40.7k
    int val;
5664
40.7k
    xmlChar *ret;
5665
5666
40.7k
    *value = NULL;
5667
40.7k
    if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) {
5668
5.71k
  SKIP(9);
5669
5.71k
  return(XML_ATTRIBUTE_REQUIRED);
5670
5.71k
    }
5671
35.0k
    if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) {
5672
8.97k
  SKIP(8);
5673
8.97k
  return(XML_ATTRIBUTE_IMPLIED);
5674
8.97k
    }
5675
26.0k
    val = XML_ATTRIBUTE_NONE;
5676
26.0k
    if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) {
5677
130
  SKIP(6);
5678
130
  val = XML_ATTRIBUTE_FIXED;
5679
130
  if (SKIP_BLANKS_PE == 0) {
5680
10
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5681
10
         "Space required after '#FIXED'\n");
5682
10
  }
5683
130
    }
5684
26.0k
    ret = xmlParseAttValue(ctxt);
5685
26.0k
    if (ret == NULL) {
5686
233
  xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo,
5687
233
           "Attribute default value declaration error\n");
5688
233
    } else
5689
25.8k
        *value = ret;
5690
26.0k
    return(val);
5691
35.0k
}
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
107
xmlParseNotationType(xmlParserCtxt *ctxt) {
5712
107
    const xmlChar *name;
5713
107
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5714
5715
107
    if (RAW != '(') {
5716
5
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5717
5
  return(NULL);
5718
5
    }
5719
529
    do {
5720
529
        NEXT;
5721
529
  SKIP_BLANKS_PE;
5722
529
        name = xmlParseName(ctxt);
5723
529
  if (name == NULL) {
5724
6
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5725
6
         "Name expected in NOTATION declaration\n");
5726
6
            xmlFreeEnumeration(ret);
5727
6
      return(NULL);
5728
6
  }
5729
523
        tmp = NULL;
5730
523
#ifdef LIBXML_VALID_ENABLED
5731
523
        if (ctxt->validate) {
5732
0
            tmp = ret;
5733
0
            while (tmp != NULL) {
5734
0
                if (xmlStrEqual(name, tmp->name)) {
5735
0
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5736
0
              "standalone: attribute notation value token %s duplicated\n",
5737
0
                                     name, NULL);
5738
0
                    if (!xmlDictOwns(ctxt->dict, name))
5739
0
                        xmlFree((xmlChar *) name);
5740
0
                    break;
5741
0
                }
5742
0
                tmp = tmp->next;
5743
0
            }
5744
0
        }
5745
523
#endif /* LIBXML_VALID_ENABLED */
5746
523
  if (tmp == NULL) {
5747
523
      cur = xmlCreateEnumeration(name);
5748
523
      if (cur == NULL) {
5749
0
                xmlErrMemory(ctxt);
5750
0
                xmlFreeEnumeration(ret);
5751
0
                return(NULL);
5752
0
            }
5753
523
      if (last == NULL) ret = last = cur;
5754
423
      else {
5755
423
    last->next = cur;
5756
423
    last = cur;
5757
423
      }
5758
523
  }
5759
523
  SKIP_BLANKS_PE;
5760
523
    } while (RAW == '|');
5761
96
    if (RAW != ')') {
5762
18
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5763
18
        xmlFreeEnumeration(ret);
5764
18
  return(NULL);
5765
18
    }
5766
78
    NEXT;
5767
78
    return(ret);
5768
96
}
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
9.36k
xmlParseEnumerationType(xmlParserCtxt *ctxt) {
5787
9.36k
    xmlChar *name;
5788
9.36k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5789
5790
9.36k
    if (RAW != '(') {
5791
292
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL);
5792
292
  return(NULL);
5793
292
    }
5794
17.6k
    do {
5795
17.6k
        NEXT;
5796
17.6k
  SKIP_BLANKS_PE;
5797
17.6k
        name = xmlParseNmtoken(ctxt);
5798
17.6k
  if (name == NULL) {
5799
15
      xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL);
5800
15
      return(ret);
5801
15
  }
5802
17.6k
        tmp = NULL;
5803
17.6k
#ifdef LIBXML_VALID_ENABLED
5804
17.6k
        if (ctxt->validate) {
5805
0
            tmp = ret;
5806
0
            while (tmp != NULL) {
5807
0
                if (xmlStrEqual(name, tmp->name)) {
5808
0
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5809
0
              "standalone: attribute enumeration value token %s duplicated\n",
5810
0
                                     name, NULL);
5811
0
                    if (!xmlDictOwns(ctxt->dict, name))
5812
0
                        xmlFree(name);
5813
0
                    break;
5814
0
                }
5815
0
                tmp = tmp->next;
5816
0
            }
5817
0
        }
5818
17.6k
#endif /* LIBXML_VALID_ENABLED */
5819
17.6k
  if (tmp == NULL) {
5820
17.6k
      cur = xmlCreateEnumeration(name);
5821
17.6k
      if (!xmlDictOwns(ctxt->dict, name))
5822
17.6k
    xmlFree(name);
5823
17.6k
      if (cur == NULL) {
5824
0
                xmlErrMemory(ctxt);
5825
0
                xmlFreeEnumeration(ret);
5826
0
                return(NULL);
5827
0
            }
5828
17.6k
      if (last == NULL) ret = last = cur;
5829
8.61k
      else {
5830
8.61k
    last->next = cur;
5831
8.61k
    last = cur;
5832
8.61k
      }
5833
17.6k
  }
5834
17.6k
  SKIP_BLANKS_PE;
5835
17.6k
    } while (RAW == '|');
5836
9.05k
    if (RAW != ')') {
5837
197
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL);
5838
197
  return(ret);
5839
197
    }
5840
8.86k
    NEXT;
5841
8.86k
    return(ret);
5842
9.05k
}
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
9.47k
xmlParseEnumeratedType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5860
9.47k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5861
111
  SKIP(8);
5862
111
  if (SKIP_BLANKS_PE == 0) {
5863
4
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5864
4
         "Space required after 'NOTATION'\n");
5865
4
      return(0);
5866
4
  }
5867
107
  *tree = xmlParseNotationType(ctxt);
5868
107
  if (*tree == NULL) return(0);
5869
78
  return(XML_ATTRIBUTE_NOTATION);
5870
107
    }
5871
9.36k
    *tree = xmlParseEnumerationType(ctxt);
5872
9.36k
    if (*tree == NULL) return(0);
5873
9.06k
    return(XML_ATTRIBUTE_ENUMERATION);
5874
9.36k
}
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
41.3k
xmlParseAttributeType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5923
41.3k
    if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) {
5924
792
  SKIP(5);
5925
792
  return(XML_ATTRIBUTE_CDATA);
5926
40.5k
     } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) {
5927
1.87k
  SKIP(6);
5928
1.87k
  return(XML_ATTRIBUTE_IDREFS);
5929
38.7k
     } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) {
5930
5.97k
  SKIP(5);
5931
5.97k
  return(XML_ATTRIBUTE_IDREF);
5932
32.7k
     } else if ((RAW == 'I') && (NXT(1) == 'D')) {
5933
21.8k
        SKIP(2);
5934
21.8k
  return(XML_ATTRIBUTE_ID);
5935
21.8k
     } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5936
71
  SKIP(6);
5937
71
  return(XML_ATTRIBUTE_ENTITY);
5938
10.8k
     } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) {
5939
982
  SKIP(8);
5940
982
  return(XML_ATTRIBUTE_ENTITIES);
5941
9.83k
     } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) {
5942
74
  SKIP(8);
5943
74
  return(XML_ATTRIBUTE_NMTOKENS);
5944
9.76k
     } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) {
5945
284
  SKIP(7);
5946
284
  return(XML_ATTRIBUTE_NMTOKEN);
5947
284
     }
5948
9.47k
     return(xmlParseEnumeratedType(ctxt, tree));
5949
41.3k
}
5950
5951
/**
5952
 * Parse an attribute list declaration for an element. Always consumes '<!'.
5953
 *
5954
 * @deprecated Internal function, don't use.
5955
 *
5956
 *     [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>'
5957
 *
5958
 *     [53] AttDef ::= S Name S AttType S DefaultDecl
5959
 * @param ctxt  an XML parser context
5960
 */
5961
void
5962
12.6k
xmlParseAttributeListDecl(xmlParserCtxt *ctxt) {
5963
12.6k
    const xmlChar *elemName;
5964
12.6k
    const xmlChar *attrName;
5965
12.6k
    xmlEnumerationPtr tree;
5966
5967
12.6k
    if ((CUR != '<') || (NXT(1) != '!'))
5968
0
        return;
5969
12.6k
    SKIP(2);
5970
5971
12.6k
    if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) {
5972
12.6k
#ifdef LIBXML_VALID_ENABLED
5973
12.6k
  int oldInputNr = ctxt->inputNr;
5974
12.6k
#endif
5975
5976
12.6k
  SKIP(7);
5977
12.6k
  if (SKIP_BLANKS_PE == 0) {
5978
15
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5979
15
                     "Space required after '<!ATTLIST'\n");
5980
15
  }
5981
12.6k
        elemName = xmlParseName(ctxt);
5982
12.6k
  if (elemName == NULL) {
5983
8
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5984
8
         "ATTLIST: no name for Element\n");
5985
8
      return;
5986
8
  }
5987
12.5k
  SKIP_BLANKS_PE;
5988
12.5k
  GROW;
5989
53.0k
  while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) {
5990
41.8k
      int type;
5991
41.8k
      int def;
5992
41.8k
      xmlChar *defaultValue = NULL;
5993
5994
41.8k
      GROW;
5995
41.8k
            tree = NULL;
5996
41.8k
      attrName = xmlParseName(ctxt);
5997
41.8k
      if (attrName == NULL) {
5998
123
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5999
123
             "ATTLIST: no name for Attribute\n");
6000
123
    break;
6001
123
      }
6002
41.7k
      GROW;
6003
41.7k
      if (SKIP_BLANKS_PE == 0) {
6004
340
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6005
340
            "Space required after the attribute name\n");
6006
340
    break;
6007
340
      }
6008
6009
41.3k
      type = xmlParseAttributeType(ctxt, &tree);
6010
41.3k
      if (type <= 0) {
6011
331
          break;
6012
331
      }
6013
6014
41.0k
      GROW;
6015
41.0k
      if (SKIP_BLANKS_PE == 0) {
6016
285
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6017
285
             "Space required after the attribute type\n");
6018
285
          if (tree != NULL)
6019
212
        xmlFreeEnumeration(tree);
6020
285
    break;
6021
285
      }
6022
6023
40.7k
      def = xmlParseDefaultDecl(ctxt, &defaultValue);
6024
40.7k
      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
40.7k
      if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL))
6032
25.6k
          xmlAttrNormalizeSpace(defaultValue, defaultValue);
6033
6034
40.7k
      GROW;
6035
40.7k
            if (RAW != '>') {
6036
31.2k
    if (SKIP_BLANKS_PE == 0) {
6037
277
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6038
277
      "Space required after the attribute default value\n");
6039
277
        if (defaultValue != NULL)
6040
56
      xmlFree(defaultValue);
6041
277
        if (tree != NULL)
6042
28
      xmlFreeEnumeration(tree);
6043
277
        break;
6044
277
    }
6045
31.2k
      }
6046
40.4k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6047
40.4k
    (ctxt->sax->attributeDecl != NULL))
6048
40.4k
    ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName,
6049
40.4k
                          type, def, defaultValue, tree);
6050
17
      else if (tree != NULL)
6051
4
    xmlFreeEnumeration(tree);
6052
6053
40.4k
      if ((ctxt->sax2) && (defaultValue != NULL) &&
6054
0
          (def != XML_ATTRIBUTE_IMPLIED) &&
6055
0
    (def != XML_ATTRIBUTE_REQUIRED)) {
6056
0
    xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue);
6057
0
      }
6058
40.4k
      if (ctxt->sax2) {
6059
0
    xmlAddSpecialAttr(ctxt, elemName, attrName, type);
6060
0
      }
6061
40.4k
      if (defaultValue != NULL)
6062
25.7k
          xmlFree(defaultValue);
6063
40.4k
      GROW;
6064
40.4k
  }
6065
12.5k
  if (RAW == '>') {
6066
11.2k
#ifdef LIBXML_VALID_ENABLED
6067
11.2k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6068
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6069
0
                                 "Attribute list declaration doesn't start and"
6070
0
                                 " stop in the same entity\n",
6071
0
                                 NULL, NULL);
6072
0
      }
6073
11.2k
#endif
6074
11.2k
      NEXT;
6075
11.2k
  }
6076
12.5k
    }
6077
12.6k
}
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
298k
xmlSkipBlankCharsPEBalanced(xmlParserCtxt *ctxt, int openInputNr) {
6088
298k
    SKIP_BLANKS;
6089
298k
    GROW;
6090
6091
298k
    (void) openInputNr;
6092
6093
298k
    if (!PARSER_EXTERNAL(ctxt) && !PARSER_IN_PE(ctxt))
6094
295k
        return;
6095
6096
3.75k
    while (!PARSER_STOPPED(ctxt)) {
6097
2.79k
        if (ctxt->input->cur >= ctxt->input->end) {
6098
25
#ifdef LIBXML_VALID_ENABLED
6099
25
            if ((ctxt->validate) && (ctxt->inputNr <= openInputNr)) {
6100
0
                xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6101
0
                                 "Element content declaration doesn't start "
6102
0
                                 "and stop in the same entity\n",
6103
0
                                 NULL, NULL);
6104
0
            }
6105
25
#endif
6106
25
            if (PARSER_IN_PE(ctxt))
6107
25
                xmlPopPE(ctxt);
6108
0
            else
6109
0
                break;
6110
2.76k
        } else if (RAW == '%') {
6111
27
            xmlParsePERefInternal(ctxt, 0);
6112
2.73k
        } else {
6113
2.73k
            break;
6114
2.73k
        }
6115
6116
52
        SKIP_BLANKS;
6117
52
        GROW;
6118
52
    }
6119
3.70k
}
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
4.26k
xmlParseElementMixedContentDecl(xmlParserCtxt *ctxt, int openInputNr) {
6143
4.26k
    xmlElementContentPtr ret = NULL, cur = NULL, n;
6144
4.26k
    const xmlChar *elem = NULL;
6145
6146
4.26k
    GROW;
6147
4.26k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6148
4.26k
  SKIP(7);
6149
4.26k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6150
4.26k
  if (RAW == ')') {
6151
3.08k
#ifdef LIBXML_VALID_ENABLED
6152
3.08k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6153
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6154
0
                                 "Element content declaration doesn't start "
6155
0
                                 "and stop in the same entity\n",
6156
0
                                 NULL, NULL);
6157
0
      }
6158
3.08k
#endif
6159
3.08k
      NEXT;
6160
3.08k
      ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6161
3.08k
      if (ret == NULL)
6162
0
                goto mem_error;
6163
3.08k
      if (RAW == '*') {
6164
245
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6165
245
    NEXT;
6166
245
      }
6167
3.08k
      return(ret);
6168
3.08k
  }
6169
1.18k
  if ((RAW == '(') || (RAW == '|')) {
6170
1.17k
      ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6171
1.17k
      if (ret == NULL)
6172
0
                goto mem_error;
6173
1.17k
  }
6174
2.83k
  while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) {
6175
1.66k
      NEXT;
6176
1.66k
            n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6177
1.66k
            if (n == NULL)
6178
0
                goto mem_error;
6179
1.66k
      if (elem == NULL) {
6180
1.17k
    n->c1 = cur;
6181
1.17k
    if (cur != NULL)
6182
1.17k
        cur->parent = n;
6183
1.17k
    ret = cur = n;
6184
1.17k
      } else {
6185
497
          cur->c2 = n;
6186
497
    n->parent = cur;
6187
497
    n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6188
497
                if (n->c1 == NULL)
6189
0
                    goto mem_error;
6190
497
    n->c1->parent = n;
6191
497
    cur = n;
6192
497
      }
6193
1.66k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6194
1.66k
      elem = xmlParseName(ctxt);
6195
1.66k
      if (elem == NULL) {
6196
15
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6197
15
      "xmlParseElementMixedContentDecl : Name expected\n");
6198
15
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6199
15
    return(NULL);
6200
15
      }
6201
1.65k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6202
1.65k
  }
6203
1.17k
  if ((RAW == ')') && (NXT(1) == '*')) {
6204
1.13k
      if (elem != NULL) {
6205
1.13k
    cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem,
6206
1.13k
                                   XML_ELEMENT_CONTENT_ELEMENT);
6207
1.13k
    if (cur->c2 == NULL)
6208
0
                    goto mem_error;
6209
1.13k
    cur->c2->parent = cur;
6210
1.13k
            }
6211
1.13k
            if (ret != NULL)
6212
1.13k
                ret->ocur = XML_ELEMENT_CONTENT_MULT;
6213
1.13k
#ifdef LIBXML_VALID_ENABLED
6214
1.13k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6215
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6216
0
                                 "Element content declaration doesn't start "
6217
0
                                 "and stop in the same entity\n",
6218
0
                                 NULL, NULL);
6219
0
      }
6220
1.13k
#endif
6221
1.13k
      SKIP(2);
6222
1.13k
  } else {
6223
37
      xmlFreeDocElementContent(ctxt->myDoc, ret);
6224
37
      xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL);
6225
37
      return(NULL);
6226
37
  }
6227
6228
1.17k
    } else {
6229
0
  xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL);
6230
0
    }
6231
1.13k
    return(ret);
6232
6233
0
mem_error:
6234
0
    xmlErrMemory(ctxt);
6235
0
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6236
0
    return(NULL);
6237
4.26k
}
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
37.6k
                                       int depth) {
6272
37.6k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
6273
37.6k
    xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL;
6274
37.6k
    const xmlChar *elem;
6275
37.6k
    xmlChar type = 0;
6276
6277
37.6k
    if (depth > maxDepth) {
6278
4
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
6279
4
                "xmlParseElementChildrenContentDecl : depth %d too deep, "
6280
4
                "use XML_PARSE_HUGE\n", depth);
6281
4
  return(NULL);
6282
4
    }
6283
37.6k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6284
37.6k
    if (RAW == '(') {
6285
19.4k
        int newInputNr = ctxt->inputNr;
6286
6287
        /* Recurse on first child */
6288
19.4k
  NEXT;
6289
19.4k
        cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6290
19.4k
                                                           depth + 1);
6291
19.4k
        if (cur == NULL)
6292
6.56k
            return(NULL);
6293
19.4k
    } else {
6294
18.2k
  elem = xmlParseName(ctxt);
6295
18.2k
  if (elem == NULL) {
6296
141
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6297
141
      return(NULL);
6298
141
  }
6299
18.1k
        cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6300
18.1k
  if (cur == NULL) {
6301
0
      xmlErrMemory(ctxt);
6302
0
      return(NULL);
6303
0
  }
6304
18.1k
  GROW;
6305
18.1k
  if (RAW == '?') {
6306
2.77k
      cur->ocur = XML_ELEMENT_CONTENT_OPT;
6307
2.77k
      NEXT;
6308
15.3k
  } else if (RAW == '*') {
6309
2.35k
      cur->ocur = XML_ELEMENT_CONTENT_MULT;
6310
2.35k
      NEXT;
6311
12.9k
  } else if (RAW == '+') {
6312
939
      cur->ocur = XML_ELEMENT_CONTENT_PLUS;
6313
939
      NEXT;
6314
12.0k
  } else {
6315
12.0k
      cur->ocur = XML_ELEMENT_CONTENT_ONCE;
6316
12.0k
  }
6317
18.1k
  GROW;
6318
18.1k
    }
6319
144k
    while (!PARSER_STOPPED(ctxt)) {
6320
126k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6321
126k
        if (RAW == ')')
6322
10.9k
            break;
6323
        /*
6324
   * Each loop we parse one separator and one element.
6325
   */
6326
115k
        if (RAW == ',') {
6327
57.3k
      if (type == 0) type = CUR;
6328
6329
      /*
6330
       * Detect "Name | Name , Name" error
6331
       */
6332
54.9k
      else if (type != CUR) {
6333
9
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6334
9
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6335
9
                      type);
6336
9
    if ((last != NULL) && (last != ret))
6337
9
        xmlFreeDocElementContent(ctxt->myDoc, last);
6338
9
    if (ret != NULL)
6339
9
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6340
9
    return(NULL);
6341
9
      }
6342
57.3k
      NEXT;
6343
6344
57.3k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ);
6345
57.3k
      if (op == NULL) {
6346
0
                xmlErrMemory(ctxt);
6347
0
    if ((last != NULL) && (last != ret))
6348
0
        xmlFreeDocElementContent(ctxt->myDoc, last);
6349
0
          xmlFreeDocElementContent(ctxt->myDoc, ret);
6350
0
    return(NULL);
6351
0
      }
6352
57.3k
      if (last == NULL) {
6353
2.42k
    op->c1 = ret;
6354
2.42k
    if (ret != NULL)
6355
2.42k
        ret->parent = op;
6356
2.42k
    ret = cur = op;
6357
54.9k
      } else {
6358
54.9k
          cur->c2 = op;
6359
54.9k
    if (op != NULL)
6360
54.9k
        op->parent = cur;
6361
54.9k
    op->c1 = last;
6362
54.9k
    if (last != NULL)
6363
54.9k
        last->parent = op;
6364
54.9k
    cur =op;
6365
54.9k
    last = NULL;
6366
54.9k
      }
6367
58.5k
  } else if (RAW == '|') {
6368
58.3k
      if (type == 0) type = CUR;
6369
6370
      /*
6371
       * Detect "Name , Name | Name" error
6372
       */
6373
46.9k
      else if (type != CUR) {
6374
6
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6375
6
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6376
6
          type);
6377
6
    if ((last != NULL) && (last != ret))
6378
6
        xmlFreeDocElementContent(ctxt->myDoc, last);
6379
6
    if (ret != NULL)
6380
6
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6381
6
    return(NULL);
6382
6
      }
6383
58.3k
      NEXT;
6384
6385
58.3k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6386
58.3k
      if (op == NULL) {
6387
0
                xmlErrMemory(ctxt);
6388
0
    if ((last != NULL) && (last != ret))
6389
0
        xmlFreeDocElementContent(ctxt->myDoc, last);
6390
0
    if (ret != NULL)
6391
0
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6392
0
    return(NULL);
6393
0
      }
6394
58.3k
      if (last == NULL) {
6395
11.3k
    op->c1 = ret;
6396
11.3k
    if (ret != NULL)
6397
11.3k
        ret->parent = op;
6398
11.3k
    ret = cur = op;
6399
46.9k
      } else {
6400
46.9k
          cur->c2 = op;
6401
46.9k
    if (op != NULL)
6402
46.9k
        op->parent = cur;
6403
46.9k
    op->c1 = last;
6404
46.9k
    if (last != NULL)
6405
46.9k
        last->parent = op;
6406
46.9k
    cur =op;
6407
46.9k
    last = NULL;
6408
46.9k
      }
6409
58.3k
  } else {
6410
208
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL);
6411
208
      if ((last != NULL) && (last != ret))
6412
119
          xmlFreeDocElementContent(ctxt->myDoc, last);
6413
208
      if (ret != NULL)
6414
208
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6415
208
      return(NULL);
6416
208
  }
6417
115k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6418
115k
        if (RAW == '(') {
6419
11.3k
            int newInputNr = ctxt->inputNr;
6420
6421
      /* Recurse on second child */
6422
11.3k
      NEXT;
6423
11.3k
      last = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6424
11.3k
                                                          depth + 1);
6425
11.3k
            if (last == NULL) {
6426
2.32k
    if (ret != NULL)
6427
2.32k
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6428
2.32k
    return(NULL);
6429
2.32k
            }
6430
104k
  } else {
6431
104k
      elem = xmlParseName(ctxt);
6432
104k
      if (elem == NULL) {
6433
71
    xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6434
71
    if (ret != NULL)
6435
71
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6436
71
    return(NULL);
6437
71
      }
6438
104k
      last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6439
104k
      if (last == NULL) {
6440
0
                xmlErrMemory(ctxt);
6441
0
    if (ret != NULL)
6442
0
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6443
0
    return(NULL);
6444
0
      }
6445
104k
      if (RAW == '?') {
6446
11.3k
    last->ocur = XML_ELEMENT_CONTENT_OPT;
6447
11.3k
    NEXT;
6448
92.9k
      } else if (RAW == '*') {
6449
2.46k
    last->ocur = XML_ELEMENT_CONTENT_MULT;
6450
2.46k
    NEXT;
6451
90.4k
      } else if (RAW == '+') {
6452
2.12k
    last->ocur = XML_ELEMENT_CONTENT_PLUS;
6453
2.12k
    NEXT;
6454
88.3k
      } else {
6455
88.3k
    last->ocur = XML_ELEMENT_CONTENT_ONCE;
6456
88.3k
      }
6457
104k
  }
6458
115k
    }
6459
28.3k
    if ((cur != NULL) && (last != NULL)) {
6460
11.2k
        cur->c2 = last;
6461
11.2k
  if (last != NULL)
6462
11.2k
      last->parent = cur;
6463
11.2k
    }
6464
28.3k
#ifdef LIBXML_VALID_ENABLED
6465
28.3k
    if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6466
0
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6467
0
                         "Element content declaration doesn't start "
6468
0
                         "and stop in the same entity\n",
6469
0
                         NULL, NULL);
6470
0
    }
6471
28.3k
#endif
6472
28.3k
    NEXT;
6473
28.3k
    if (RAW == '?') {
6474
1.03k
  if (ret != NULL) {
6475
1.03k
      if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) ||
6476
909
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6477
424
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6478
612
      else
6479
612
          ret->ocur = XML_ELEMENT_CONTENT_OPT;
6480
1.03k
  }
6481
1.03k
  NEXT;
6482
27.3k
    } else if (RAW == '*') {
6483
5.22k
  if (ret != NULL) {
6484
5.22k
      ret->ocur = XML_ELEMENT_CONTENT_MULT;
6485
5.22k
      cur = ret;
6486
      /*
6487
       * Some normalization:
6488
       * (a | b* | c?)* == (a | b | c)*
6489
       */
6490
41.7k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6491
36.4k
    if ((cur->c1 != NULL) &&
6492
36.4k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6493
28.8k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT)))
6494
9.26k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6495
36.4k
    if ((cur->c2 != NULL) &&
6496
36.4k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6497
33.2k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT)))
6498
4.62k
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6499
36.4k
    cur = cur->c2;
6500
36.4k
      }
6501
5.22k
  }
6502
5.22k
  NEXT;
6503
22.0k
    } else if (RAW == '+') {
6504
2.05k
  if (ret != NULL) {
6505
2.05k
      int found = 0;
6506
6507
2.05k
      if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) ||
6508
2.03k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6509
347
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6510
1.71k
      else
6511
1.71k
          ret->ocur = XML_ELEMENT_CONTENT_PLUS;
6512
      /*
6513
       * Some normalization:
6514
       * (a | b*)+ == (a | b)*
6515
       * (a | b?)+ == (a | b)*
6516
       */
6517
59.8k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6518
57.7k
    if ((cur->c1 != NULL) &&
6519
57.7k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6520
56.5k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) {
6521
2.36k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6522
2.36k
        found = 1;
6523
2.36k
    }
6524
57.7k
    if ((cur->c2 != NULL) &&
6525
57.7k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6526
57.6k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) {
6527
587
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6528
587
        found = 1;
6529
587
    }
6530
57.7k
    cur = cur->c2;
6531
57.7k
      }
6532
2.05k
      if (found)
6533
618
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6534
2.05k
  }
6535
2.05k
  NEXT;
6536
2.05k
    }
6537
28.3k
    return(ret);
6538
30.9k
}
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
11.1k
                           xmlElementContent **result) {
6593
6594
11.1k
    xmlElementContentPtr tree = NULL;
6595
11.1k
    int openInputNr = ctxt->inputNr;
6596
11.1k
    int res;
6597
6598
11.1k
    *result = NULL;
6599
6600
11.1k
    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
11.1k
    NEXT;
6606
11.1k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6607
11.1k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6608
4.26k
        tree = xmlParseElementMixedContentDecl(ctxt, openInputNr);
6609
4.26k
  res = XML_ELEMENT_TYPE_MIXED;
6610
6.89k
    } else {
6611
6.89k
        tree = xmlParseElementChildrenContentDeclPriv(ctxt, openInputNr, 1);
6612
6.89k
  res = XML_ELEMENT_TYPE_ELEMENT;
6613
6.89k
    }
6614
11.1k
    if (tree == NULL)
6615
491
        return(-1);
6616
10.6k
    SKIP_BLANKS_PE;
6617
10.6k
    *result = tree;
6618
10.6k
    return(res);
6619
11.1k
}
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
12.4k
xmlParseElementDecl(xmlParserCtxt *ctxt) {
6636
12.4k
    const xmlChar *name;
6637
12.4k
    int ret = -1;
6638
12.4k
    xmlElementContentPtr content  = NULL;
6639
6640
12.4k
    if ((CUR != '<') || (NXT(1) != '!'))
6641
0
        return(ret);
6642
12.4k
    SKIP(2);
6643
6644
    /* GROW; done in the caller */
6645
12.4k
    if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) {
6646
11.8k
#ifdef LIBXML_VALID_ENABLED
6647
11.8k
  int oldInputNr = ctxt->inputNr;
6648
11.8k
#endif
6649
6650
11.8k
  SKIP(7);
6651
11.8k
  if (SKIP_BLANKS_PE == 0) {
6652
12
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6653
12
               "Space required after 'ELEMENT'\n");
6654
12
      return(-1);
6655
12
  }
6656
11.8k
        name = xmlParseName(ctxt);
6657
11.8k
  if (name == NULL) {
6658
9
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6659
9
         "xmlParseElementDecl: no name for Element\n");
6660
9
      return(-1);
6661
9
  }
6662
11.7k
  if (SKIP_BLANKS_PE == 0) {
6663
286
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6664
286
         "Space required after the element name\n");
6665
286
  }
6666
11.7k
  if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) {
6667
173
      SKIP(5);
6668
      /*
6669
       * Element must always be empty.
6670
       */
6671
173
      ret = XML_ELEMENT_TYPE_EMPTY;
6672
11.6k
  } else if ((RAW == 'A') && (NXT(1) == 'N') &&
6673
397
             (NXT(2) == 'Y')) {
6674
392
      SKIP(3);
6675
      /*
6676
       * Element is a generic container.
6677
       */
6678
392
      ret = XML_ELEMENT_TYPE_ANY;
6679
11.2k
  } else if (RAW == '(') {
6680
11.1k
      ret = xmlParseElementContentDecl(ctxt, name, &content);
6681
11.1k
            if (ret <= 0)
6682
491
                return(-1);
6683
11.1k
  } else {
6684
      /*
6685
       * [ WFC: PEs in Internal Subset ] error handling.
6686
       */
6687
66
            xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6688
66
                  "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n");
6689
66
      return(-1);
6690
66
  }
6691
6692
11.2k
  SKIP_BLANKS_PE;
6693
6694
11.2k
  if (RAW != '>') {
6695
554
      xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
6696
554
      if (content != NULL) {
6697
546
    xmlFreeDocElementContent(ctxt->myDoc, content);
6698
546
      }
6699
10.6k
  } else {
6700
10.6k
#ifdef LIBXML_VALID_ENABLED
6701
10.6k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6702
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6703
0
                                 "Element declaration doesn't start and stop in"
6704
0
                                 " the same entity\n",
6705
0
                                 NULL, NULL);
6706
0
      }
6707
10.6k
#endif
6708
6709
10.6k
      NEXT;
6710
10.6k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6711
10.6k
    (ctxt->sax->elementDecl != NULL)) {
6712
10.6k
    if (content != NULL)
6713
10.0k
        content->parent = NULL;
6714
10.6k
          ctxt->sax->elementDecl(ctxt->userData, name, ret,
6715
10.6k
                           content);
6716
10.6k
    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
8.59k
        xmlFreeDocElementContent(ctxt->myDoc, content);
6724
8.59k
    }
6725
10.6k
      } else if (content != NULL) {
6726
44
    xmlFreeDocElementContent(ctxt->myDoc, content);
6727
44
      }
6728
10.6k
  }
6729
11.2k
    }
6730
11.8k
    return(ret);
6731
12.4k
}
6732
6733
/**
6734
 * Parse a conditional section. Always consumes '<!['.
6735
 *
6736
 *     [61] conditionalSect ::= includeSect | ignoreSect
6737
 *     [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>'
6738
 *     [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>'
6739
 *     [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>'
6740
 *                                 Ignore)*
6741
 *     [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*)
6742
 * @param ctxt  an XML parser context
6743
 */
6744
6745
static void
6746
0
xmlParseConditionalSections(xmlParserCtxtPtr ctxt) {
6747
0
    size_t depth = 0;
6748
0
    int isFreshPE = 0;
6749
0
    int oldInputNr = ctxt->inputNr;
6750
0
    int declInputNr = ctxt->inputNr;
6751
6752
0
    while (!PARSER_STOPPED(ctxt)) {
6753
0
        if (ctxt->input->cur >= ctxt->input->end) {
6754
0
            if (ctxt->inputNr <= oldInputNr) {
6755
0
                xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6756
0
                return;
6757
0
            }
6758
6759
0
            xmlPopPE(ctxt);
6760
0
            declInputNr = ctxt->inputNr;
6761
0
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6762
0
            SKIP(3);
6763
0
            SKIP_BLANKS_PE;
6764
6765
0
            isFreshPE = 0;
6766
6767
0
            if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) {
6768
0
                SKIP(7);
6769
0
                SKIP_BLANKS_PE;
6770
0
                if (RAW != '[') {
6771
0
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6772
0
                    return;
6773
0
                }
6774
0
#ifdef LIBXML_VALID_ENABLED
6775
0
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6776
0
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6777
0
                                     "All markup of the conditional section is"
6778
0
                                     " not in the same entity\n",
6779
0
                                     NULL, NULL);
6780
0
                }
6781
0
#endif
6782
0
                NEXT;
6783
6784
0
                depth++;
6785
0
            } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) {
6786
0
                size_t ignoreDepth = 0;
6787
6788
0
                SKIP(6);
6789
0
                SKIP_BLANKS_PE;
6790
0
                if (RAW != '[') {
6791
0
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6792
0
                    return;
6793
0
                }
6794
0
#ifdef LIBXML_VALID_ENABLED
6795
0
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6796
0
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6797
0
                                     "All markup of the conditional section is"
6798
0
                                     " not in the same entity\n",
6799
0
                                     NULL, NULL);
6800
0
                }
6801
0
#endif
6802
0
                NEXT;
6803
6804
0
                while (PARSER_STOPPED(ctxt) == 0) {
6805
0
                    if (RAW == 0) {
6806
0
                        xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL);
6807
0
                        return;
6808
0
                    }
6809
0
                    if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6810
0
                        SKIP(3);
6811
0
                        ignoreDepth++;
6812
                        /* Check for integer overflow */
6813
0
                        if (ignoreDepth == 0) {
6814
0
                            xmlErrMemory(ctxt);
6815
0
                            return;
6816
0
                        }
6817
0
                    } else if ((RAW == ']') && (NXT(1) == ']') &&
6818
0
                               (NXT(2) == '>')) {
6819
0
                        SKIP(3);
6820
0
                        if (ignoreDepth == 0)
6821
0
                            break;
6822
0
                        ignoreDepth--;
6823
0
                    } else {
6824
0
                        NEXT;
6825
0
                    }
6826
0
                }
6827
6828
0
#ifdef LIBXML_VALID_ENABLED
6829
0
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6830
0
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6831
0
                                     "All markup of the conditional section is"
6832
0
                                     " not in the same entity\n",
6833
0
                                     NULL, NULL);
6834
0
                }
6835
0
#endif
6836
0
            } else {
6837
0
                xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL);
6838
0
                return;
6839
0
            }
6840
0
        } else if ((depth > 0) &&
6841
0
                   (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) {
6842
0
            if (isFreshPE) {
6843
0
                xmlFatalErrMsg(ctxt, XML_ERR_CONDSEC_INVALID,
6844
0
                               "Parameter entity must match "
6845
0
                               "extSubsetDecl\n");
6846
0
                return;
6847
0
            }
6848
6849
0
            depth--;
6850
0
#ifdef LIBXML_VALID_ENABLED
6851
0
            if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6852
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6853
0
                                 "All markup of the conditional section is not"
6854
0
                                 " in the same entity\n",
6855
0
                                 NULL, NULL);
6856
0
            }
6857
0
#endif
6858
0
            SKIP(3);
6859
0
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
6860
0
            isFreshPE = 0;
6861
0
            xmlParseMarkupDecl(ctxt);
6862
0
        } else if (RAW == '%') {
6863
0
            xmlParsePERefInternal(ctxt, 1);
6864
0
            if (ctxt->inputNr > declInputNr) {
6865
0
                isFreshPE = 1;
6866
0
                declInputNr = ctxt->inputNr;
6867
0
            }
6868
0
        } else {
6869
0
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6870
0
            return;
6871
0
        }
6872
6873
0
        if (depth == 0)
6874
0
            break;
6875
6876
0
        SKIP_BLANKS;
6877
0
        SHRINK;
6878
0
        GROW;
6879
0
    }
6880
0
}
6881
6882
/**
6883
 * Parse markup declarations. Always consumes '<!' or '<?'.
6884
 *
6885
 * @deprecated Internal function, don't use.
6886
 *
6887
 *     [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl |
6888
 *                         NotationDecl | PI | Comment
6889
 *
6890
 * [ VC: Proper Declaration/PE Nesting ]
6891
 * Parameter-entity replacement text must be properly nested with
6892
 * markup declarations. That is to say, if either the first character
6893
 * or the last character of a markup declaration (markupdecl above) is
6894
 * contained in the replacement text for a parameter-entity reference,
6895
 * both must be contained in the same replacement text.
6896
 *
6897
 * [ WFC: PEs in Internal Subset ]
6898
 * In the internal DTD subset, parameter-entity references can occur
6899
 * only where markup declarations can occur, not within markup declarations.
6900
 * (This does not apply to references that occur in external parameter
6901
 * entities or to the external subset.)
6902
 *
6903
 * @param ctxt  an XML parser context
6904
 */
6905
void
6906
67.8k
xmlParseMarkupDecl(xmlParserCtxt *ctxt) {
6907
67.8k
    GROW;
6908
67.8k
    if (CUR == '<') {
6909
67.8k
        if (NXT(1) == '!') {
6910
60.5k
      switch (NXT(2)) {
6911
43.7k
          case 'E':
6912
43.7k
        if (NXT(3) == 'L')
6913
12.4k
      xmlParseElementDecl(ctxt);
6914
31.2k
        else if (NXT(3) == 'N')
6915
31.2k
      xmlParseEntityDecl(ctxt);
6916
78
                    else
6917
78
                        SKIP(2);
6918
43.7k
        break;
6919
12.6k
          case 'A':
6920
12.6k
        xmlParseAttributeListDecl(ctxt);
6921
12.6k
        break;
6922
2.13k
          case 'N':
6923
2.13k
        xmlParseNotationDecl(ctxt);
6924
2.13k
        break;
6925
1.89k
          case '-':
6926
1.89k
        xmlParseComment(ctxt);
6927
1.89k
        break;
6928
108
    default:
6929
108
                    xmlFatalErr(ctxt,
6930
108
                                ctxt->inSubset == 2 ?
6931
0
                                    XML_ERR_EXT_SUBSET_NOT_FINISHED :
6932
108
                                    XML_ERR_INT_SUBSET_NOT_FINISHED,
6933
108
                                NULL);
6934
108
                    SKIP(2);
6935
108
        break;
6936
60.5k
      }
6937
60.5k
  } else if (NXT(1) == '?') {
6938
7.27k
      xmlParsePI(ctxt);
6939
7.27k
  }
6940
67.8k
    }
6941
67.8k
}
6942
6943
/**
6944
 * Parse an XML declaration header for external entities
6945
 *
6946
 * @deprecated Internal function, don't use.
6947
 *
6948
 *     [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>'
6949
 * @param ctxt  an XML parser context
6950
 */
6951
6952
void
6953
0
xmlParseTextDecl(xmlParserCtxt *ctxt) {
6954
0
    xmlChar *version;
6955
6956
    /*
6957
     * We know that '<?xml' is here.
6958
     */
6959
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
6960
0
  SKIP(5);
6961
0
    } else {
6962
0
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL);
6963
0
  return;
6964
0
    }
6965
6966
0
    if (SKIP_BLANKS == 0) {
6967
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6968
0
           "Space needed after '<?xml'\n");
6969
0
    }
6970
6971
    /*
6972
     * We may have the VersionInfo here.
6973
     */
6974
0
    version = xmlParseVersionInfo(ctxt);
6975
0
    if (version == NULL) {
6976
0
  version = xmlCharStrdup(XML_DEFAULT_VERSION);
6977
0
        if (version == NULL) {
6978
0
            xmlErrMemory(ctxt);
6979
0
            return;
6980
0
        }
6981
0
    } else {
6982
0
  if (SKIP_BLANKS == 0) {
6983
0
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6984
0
               "Space needed here\n");
6985
0
  }
6986
0
    }
6987
0
    ctxt->input->version = version;
6988
6989
    /*
6990
     * We must have the encoding declaration
6991
     */
6992
0
    xmlParseEncodingDecl(ctxt);
6993
6994
0
    SKIP_BLANKS;
6995
0
    if ((RAW == '?') && (NXT(1) == '>')) {
6996
0
        SKIP(2);
6997
0
    } else if (RAW == '>') {
6998
        /* Deprecated old WD ... */
6999
0
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7000
0
  NEXT;
7001
0
    } else {
7002
0
        int c;
7003
7004
0
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7005
0
        while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) {
7006
0
            NEXT;
7007
0
            if (c == '>')
7008
0
                break;
7009
0
        }
7010
0
    }
7011
0
}
7012
7013
/**
7014
 * Parse Markup declarations from an external subset
7015
 *
7016
 * @deprecated Internal function, don't use.
7017
 *
7018
 *     [30] extSubset ::= textDecl? extSubsetDecl
7019
 *
7020
 *     [31] extSubsetDecl ::= (markupdecl | conditionalSect |
7021
 *                             PEReference | S) *
7022
 * @param ctxt  an XML parser context
7023
 * @param publicId  the public identifier
7024
 * @param systemId  the system identifier (URL)
7025
 */
7026
void
7027
xmlParseExternalSubset(xmlParserCtxt *ctxt, const xmlChar *publicId,
7028
0
                       const xmlChar *systemId) {
7029
0
    int oldInputNr;
7030
7031
0
    xmlCtxtInitializeLate(ctxt);
7032
7033
0
    xmlDetectEncoding(ctxt);
7034
7035
0
    if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) {
7036
0
  xmlParseTextDecl(ctxt);
7037
0
    }
7038
0
    if (ctxt->myDoc == NULL) {
7039
0
        ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0");
7040
0
  if (ctxt->myDoc == NULL) {
7041
0
      xmlErrMemory(ctxt);
7042
0
      return;
7043
0
  }
7044
0
  ctxt->myDoc->properties = XML_DOC_INTERNAL;
7045
0
    }
7046
0
    if ((ctxt->myDoc->intSubset == NULL) &&
7047
0
        (xmlCreateIntSubset(ctxt->myDoc, NULL, publicId, systemId) == NULL)) {
7048
0
        xmlErrMemory(ctxt);
7049
0
    }
7050
7051
0
    ctxt->inSubset = 2;
7052
0
    oldInputNr = ctxt->inputNr;
7053
7054
0
    SKIP_BLANKS;
7055
0
    while (!PARSER_STOPPED(ctxt)) {
7056
0
        if (ctxt->input->cur >= ctxt->input->end) {
7057
0
            if (ctxt->inputNr <= oldInputNr) {
7058
0
                xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED);
7059
0
                break;
7060
0
            }
7061
7062
0
            xmlPopPE(ctxt);
7063
0
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
7064
0
            xmlParseConditionalSections(ctxt);
7065
0
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
7066
0
            xmlParseMarkupDecl(ctxt);
7067
0
        } else if (RAW == '%') {
7068
0
            xmlParsePERefInternal(ctxt, 1);
7069
0
        } else {
7070
0
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
7071
7072
0
            while (ctxt->inputNr > oldInputNr)
7073
0
                xmlPopPE(ctxt);
7074
0
            break;
7075
0
        }
7076
0
        SKIP_BLANKS;
7077
0
        SHRINK;
7078
0
        GROW;
7079
0
    }
7080
0
}
7081
7082
/**
7083
 * Parse and handle entity references in content, depending on the SAX
7084
 * interface, this may end-up in a call to character() if this is a
7085
 * CharRef, a predefined entity, if there is no reference() callback.
7086
 * or if the parser was asked to switch to that mode.
7087
 *
7088
 * @deprecated Internal function, don't use.
7089
 *
7090
 * Always consumes '&'.
7091
 *
7092
 *     [67] Reference ::= EntityRef | CharRef
7093
 * @param ctxt  an XML parser context
7094
 */
7095
void
7096
695k
xmlParseReference(xmlParserCtxt *ctxt) {
7097
695k
    xmlEntityPtr ent = NULL;
7098
695k
    const xmlChar *name;
7099
695k
    xmlChar *val;
7100
7101
695k
    if (RAW != '&')
7102
0
        return;
7103
7104
    /*
7105
     * Simple case of a CharRef
7106
     */
7107
695k
    if (NXT(1) == '#') {
7108
14.4k
  int i = 0;
7109
14.4k
  xmlChar out[16];
7110
14.4k
  int value = xmlParseCharRef(ctxt);
7111
7112
14.4k
  if (value == 0)
7113
1.08k
      return;
7114
7115
        /*
7116
         * Just encode the value in UTF-8
7117
         */
7118
13.3k
        COPY_BUF(out, i, value);
7119
13.3k
        out[i] = 0;
7120
13.3k
        if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7121
13.3k
            (!ctxt->disableSAX))
7122
13.2k
            ctxt->sax->characters(ctxt->userData, out, i);
7123
13.3k
  return;
7124
14.4k
    }
7125
7126
    /*
7127
     * We are seeing an entity reference
7128
     */
7129
681k
    name = xmlParseEntityRefInternal(ctxt);
7130
681k
    if (name == NULL)
7131
853
        return;
7132
680k
    ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0);
7133
680k
    if (ent == NULL) {
7134
        /*
7135
         * Create a reference for undeclared entities.
7136
         */
7137
16.8k
        if ((ctxt->replaceEntities == 0) &&
7138
16.8k
            (ctxt->sax != NULL) &&
7139
16.8k
            (ctxt->disableSAX == 0) &&
7140
16.4k
            (ctxt->sax->reference != NULL)) {
7141
0
            ctxt->sax->reference(ctxt->userData, name);
7142
0
        }
7143
16.8k
        return;
7144
16.8k
    }
7145
663k
    if (!ctxt->wellFormed)
7146
0
  return;
7147
7148
    /* special case of predefined entities */
7149
663k
    if ((ent->name == NULL) ||
7150
663k
        (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
7151
41.6k
  val = ent->content;
7152
41.6k
  if (val == NULL) return;
7153
  /*
7154
   * inline the entity.
7155
   */
7156
41.6k
  if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7157
41.6k
      (!ctxt->disableSAX))
7158
41.6k
      ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val));
7159
41.6k
  return;
7160
41.6k
    }
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
622k
    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
622k
    if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) ||
7211
419
        (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
7212
419
         ((ctxt->replaceEntities) ||
7213
621k
          (ctxt->validate)))) {
7214
621k
        if ((ent->flags & XML_ENT_PARSED) == 0) {
7215
7.43k
            xmlCtxtParseEntity(ctxt, ent);
7216
614k
        } 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
614k
            xmlCtxtParseEntity(ctxt, ent);
7227
614k
        }
7228
621k
    }
7229
7230
    /*
7231
     * We also check for amplification if entities aren't substituted.
7232
     * They might be expanded later.
7233
     */
7234
622k
    if (xmlParserEntityCheck(ctxt, ent->expandedSize))
7235
181
        return;
7236
7237
621k
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
7238
1.29k
        return;
7239
7240
620k
    if (ctxt->replaceEntities == 0) {
7241
  /*
7242
   * Create a reference
7243
   */
7244
620k
        if (ctxt->sax->reference != NULL)
7245
0
      ctxt->sax->reference(ctxt->userData, ent->name);
7246
620k
    } else if ((ent->children != NULL) && (ctxt->node != NULL)) {
7247
0
        xmlNodePtr copy, cur;
7248
7249
        /*
7250
         * Seems we are generating the DOM content, copy the tree
7251
   */
7252
0
        cur = ent->children;
7253
7254
        /*
7255
         * Handle first text node with SAX to coalesce text efficiently
7256
         */
7257
0
        if ((cur->type == XML_TEXT_NODE) ||
7258
0
            (cur->type == XML_CDATA_SECTION_NODE)) {
7259
0
            int len = xmlStrlen(cur->content);
7260
7261
0
            if ((cur->type == XML_TEXT_NODE) ||
7262
0
                (ctxt->options & XML_PARSE_NOCDATA)) {
7263
0
                if (ctxt->sax->characters != NULL)
7264
0
                    ctxt->sax->characters(ctxt->userData, cur->content, len);
7265
0
            } else {
7266
0
                if (ctxt->sax->cdataBlock != NULL)
7267
0
                    ctxt->sax->cdataBlock(ctxt->userData, cur->content, len);
7268
0
            }
7269
7270
0
            cur = cur->next;
7271
0
        }
7272
7273
0
        while (cur != NULL) {
7274
0
            xmlNodePtr last;
7275
7276
            /*
7277
             * Handle last text node with SAX to coalesce text efficiently
7278
             */
7279
0
            if ((cur->next == NULL) &&
7280
0
                ((cur->type == XML_TEXT_NODE) ||
7281
0
                 (cur->type == XML_CDATA_SECTION_NODE))) {
7282
0
                int len = xmlStrlen(cur->content);
7283
7284
0
                if ((cur->type == XML_TEXT_NODE) ||
7285
0
                    (ctxt->options & XML_PARSE_NOCDATA)) {
7286
0
                    if (ctxt->sax->characters != NULL)
7287
0
                        ctxt->sax->characters(ctxt->userData, cur->content,
7288
0
                                              len);
7289
0
                } else {
7290
0
                    if (ctxt->sax->cdataBlock != NULL)
7291
0
                        ctxt->sax->cdataBlock(ctxt->userData, cur->content,
7292
0
                                              len);
7293
0
                }
7294
7295
0
                break;
7296
0
            }
7297
7298
            /*
7299
             * Reset coalesce buffer stats only for non-text nodes.
7300
             */
7301
0
            ctxt->nodemem = 0;
7302
0
            ctxt->nodelen = 0;
7303
7304
0
            copy = xmlDocCopyNode(cur, ctxt->myDoc, 1);
7305
7306
0
            if (copy == NULL) {
7307
0
                xmlErrMemory(ctxt);
7308
0
                break;
7309
0
            }
7310
7311
0
            if (ctxt->parseMode == XML_PARSE_READER) {
7312
                /* Needed for reader */
7313
0
                copy->extra = cur->extra;
7314
                /* Maybe needed for reader */
7315
0
                copy->_private = cur->_private;
7316
0
            }
7317
7318
0
            copy->parent = ctxt->node;
7319
0
            last = ctxt->node->last;
7320
0
            if (last == NULL) {
7321
0
                ctxt->node->children = copy;
7322
0
            } else {
7323
0
                last->next = copy;
7324
0
                copy->prev = last;
7325
0
            }
7326
0
            ctxt->node->last = copy;
7327
7328
0
            cur = cur->next;
7329
0
        }
7330
0
    }
7331
620k
}
7332
7333
static void
7334
23.5k
xmlHandleUndeclaredEntity(xmlParserCtxtPtr ctxt, const xmlChar *name) {
7335
    /*
7336
     * [ WFC: Entity Declared ]
7337
     * In a document without any DTD, a document with only an
7338
     * internal DTD subset which contains no parameter entity
7339
     * references, or a document with "standalone='yes'", the
7340
     * Name given in the entity reference must match that in an
7341
     * entity declaration, except that well-formed documents
7342
     * need not declare any of the following entities: amp, lt,
7343
     * gt, apos, quot.
7344
     * The declaration of a parameter entity must precede any
7345
     * reference to it.
7346
     * Similarly, the declaration of a general entity must
7347
     * precede any reference to it which appears in a default
7348
     * value in an attribute-list declaration. Note that if
7349
     * entities are declared in the external subset or in
7350
     * external parameter entities, a non-validating processor
7351
     * is not obligated to read and process their declarations;
7352
     * for such documents, the rule that an entity must be
7353
     * declared is a well-formedness constraint only if
7354
     * standalone='yes'.
7355
     */
7356
23.5k
    if ((ctxt->standalone == 1) ||
7357
23.4k
        ((ctxt->hasExternalSubset == 0) &&
7358
18.3k
         (ctxt->hasPErefs == 0))) {
7359
524
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY,
7360
524
                          "Entity '%s' not defined\n", name);
7361
524
#ifdef LIBXML_VALID_ENABLED
7362
22.9k
    } else if (ctxt->validate) {
7363
        /*
7364
         * [ VC: Entity Declared ]
7365
         * In a document with an external subset or external
7366
         * parameter entities with "standalone='no'", ...
7367
         * ... The declaration of a parameter entity must
7368
         * precede any reference to it...
7369
         */
7370
0
        xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY,
7371
0
                         "Entity '%s' not defined\n", name, NULL);
7372
0
#endif
7373
22.9k
    } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) ||
7374
22.9k
               ((ctxt->replaceEntities) &&
7375
0
                ((ctxt->options & XML_PARSE_NO_XXE) == 0))) {
7376
        /*
7377
         * Also raise a non-fatal error
7378
         *
7379
         * - if the external subset is loaded and all entity declarations
7380
         *   should be available, or
7381
         * - entity substition was requested without restricting
7382
         *   external entity access.
7383
         */
7384
0
        xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY,
7385
0
                     "Entity '%s' not defined\n", name);
7386
22.9k
    } else {
7387
22.9k
        xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7388
22.9k
                      "Entity '%s' not defined\n", name, NULL);
7389
22.9k
    }
7390
7391
23.5k
    ctxt->valid = 0;
7392
23.5k
}
7393
7394
static xmlEntityPtr
7395
720k
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) {
7396
720k
    xmlEntityPtr ent = NULL;
7397
7398
    /*
7399
     * Predefined entities override any extra definition
7400
     */
7401
720k
    if ((ctxt->options & XML_PARSE_OLDSAX) == 0) {
7402
720k
        ent = xmlGetPredefinedEntity(name);
7403
720k
        if (ent != NULL)
7404
60.0k
            return(ent);
7405
720k
    }
7406
7407
    /*
7408
     * Ask first SAX for entity resolution, otherwise try the
7409
     * entities which may have stored in the parser context.
7410
     */
7411
660k
    if (ctxt->sax != NULL) {
7412
660k
  if (ctxt->sax->getEntity != NULL)
7413
660k
      ent = ctxt->sax->getEntity(ctxt->userData, name);
7414
660k
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7415
20.3k
      (ctxt->options & XML_PARSE_OLDSAX))
7416
0
      ent = xmlGetPredefinedEntity(name);
7417
660k
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7418
20.3k
      (ctxt->userData==ctxt)) {
7419
20.3k
      ent = xmlSAX2GetEntity(ctxt, name);
7420
20.3k
  }
7421
660k
    }
7422
7423
660k
    if (ent == NULL) {
7424
20.3k
        xmlHandleUndeclaredEntity(ctxt, name);
7425
20.3k
    }
7426
7427
    /*
7428
     * [ WFC: Parsed Entity ]
7429
     * An entity reference must not contain the name of an
7430
     * unparsed entity
7431
     */
7432
639k
    else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) {
7433
3
  xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY,
7434
3
     "Entity reference to unparsed entity %s\n", name);
7435
3
        ent = NULL;
7436
3
    }
7437
7438
    /*
7439
     * [ WFC: No External Entity References ]
7440
     * Attribute values cannot contain direct or indirect
7441
     * entity references to external entities.
7442
     */
7443
639k
    else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) {
7444
421
        if (inAttr) {
7445
2
            xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL,
7446
2
                 "Attribute references external entity '%s'\n", name);
7447
2
            ent = NULL;
7448
2
        }
7449
421
    }
7450
7451
660k
    return(ent);
7452
720k
}
7453
7454
/**
7455
 * Parse an entity reference. Always consumes '&'.
7456
 *
7457
 *     [68] EntityRef ::= '&' Name ';'
7458
 *
7459
 * @param ctxt  an XML parser context
7460
 * @returns the name, or NULL in case of error.
7461
 */
7462
static const xmlChar *
7463
713k
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) {
7464
713k
    const xmlChar *name;
7465
7466
713k
    GROW;
7467
7468
713k
    if (RAW != '&')
7469
0
        return(NULL);
7470
713k
    NEXT;
7471
713k
    name = xmlParseName(ctxt);
7472
713k
    if (name == NULL) {
7473
621
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7474
621
           "xmlParseEntityRef: no name\n");
7475
621
        return(NULL);
7476
621
    }
7477
712k
    if (RAW != ';') {
7478
403
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7479
403
  return(NULL);
7480
403
    }
7481
712k
    NEXT;
7482
7483
712k
    return(name);
7484
712k
}
7485
7486
/**
7487
 * @deprecated Internal function, don't use.
7488
 *
7489
 * @param ctxt  an XML parser context
7490
 * @returns the xmlEntity if found, or NULL otherwise.
7491
 */
7492
xmlEntity *
7493
0
xmlParseEntityRef(xmlParserCtxt *ctxt) {
7494
0
    const xmlChar *name;
7495
7496
0
    if (ctxt == NULL)
7497
0
        return(NULL);
7498
7499
0
    name = xmlParseEntityRefInternal(ctxt);
7500
0
    if (name == NULL)
7501
0
        return(NULL);
7502
7503
0
    return(xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 0));
7504
0
}
7505
7506
/**
7507
 * Parse ENTITY references declarations, but this version parses it from
7508
 * a string value.
7509
 *
7510
 *     [68] EntityRef ::= '&' Name ';'
7511
 *
7512
 * [ WFC: Entity Declared ]
7513
 * In a document without any DTD, a document with only an internal DTD
7514
 * subset which contains no parameter entity references, or a document
7515
 * with "standalone='yes'", the Name given in the entity reference
7516
 * must match that in an entity declaration, except that well-formed
7517
 * documents need not declare any of the following entities: amp, lt,
7518
 * gt, apos, quot.  The declaration of a parameter entity must precede
7519
 * any reference to it.  Similarly, the declaration of a general entity
7520
 * must precede any reference to it which appears in a default value in an
7521
 * attribute-list declaration. Note that if entities are declared in the
7522
 * external subset or in external parameter entities, a non-validating
7523
 * processor is not obligated to read and process their declarations;
7524
 * for such documents, the rule that an entity must be declared is a
7525
 * well-formedness constraint only if standalone='yes'.
7526
 *
7527
 * [ WFC: Parsed Entity ]
7528
 * An entity reference must not contain the name of an unparsed entity
7529
 *
7530
 * @param ctxt  an XML parser context
7531
 * @param str  a pointer to an index in the string
7532
 * @returns the xmlEntity if found, or NULL otherwise. The str pointer
7533
 * is updated to the current location in the string.
7534
 */
7535
static xmlChar *
7536
7.92k
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) {
7537
7.92k
    xmlChar *name;
7538
7.92k
    const xmlChar *ptr;
7539
7.92k
    xmlChar cur;
7540
7541
7.92k
    if ((str == NULL) || (*str == NULL))
7542
0
        return(NULL);
7543
7.92k
    ptr = *str;
7544
7.92k
    cur = *ptr;
7545
7.92k
    if (cur != '&')
7546
0
  return(NULL);
7547
7548
7.92k
    ptr++;
7549
7.92k
    name = xmlParseStringName(ctxt, &ptr);
7550
7.92k
    if (name == NULL) {
7551
3
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7552
3
           "xmlParseStringEntityRef: no name\n");
7553
3
  *str = ptr;
7554
3
  return(NULL);
7555
3
    }
7556
7.92k
    if (*ptr != ';') {
7557
2
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7558
2
        xmlFree(name);
7559
2
  *str = ptr;
7560
2
  return(NULL);
7561
2
    }
7562
7.91k
    ptr++;
7563
7564
7.91k
    *str = ptr;
7565
7.91k
    return(name);
7566
7.92k
}
7567
7568
/**
7569
 * Parse a parameter entity reference. Always consumes '%'.
7570
 *
7571
 * The entity content is handled directly by pushing it's content as
7572
 * a new input stream.
7573
 *
7574
 *     [69] PEReference ::= '%' Name ';'
7575
 *
7576
 * [ WFC: No Recursion ]
7577
 * A parsed entity must not contain a recursive
7578
 * reference to itself, either directly or indirectly.
7579
 *
7580
 * [ WFC: Entity Declared ]
7581
 * In a document without any DTD, a document with only an internal DTD
7582
 * subset which contains no parameter entity references, or a document
7583
 * with "standalone='yes'", ...  ... The declaration of a parameter
7584
 * entity must precede any reference to it...
7585
 *
7586
 * [ VC: Entity Declared ]
7587
 * In a document with an external subset or external parameter entities
7588
 * with "standalone='no'", ...  ... The declaration of a parameter entity
7589
 * must precede any reference to it...
7590
 *
7591
 * [ WFC: In DTD ]
7592
 * Parameter-entity references may only appear in the DTD.
7593
 * NOTE: misleading but this is handled.
7594
 *
7595
 * @param ctxt  an XML parser context
7596
 * @param markupDecl  whether the PERef starts a markup declaration
7597
 */
7598
static void
7599
2.56k
xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl) {
7600
2.56k
    const xmlChar *name;
7601
2.56k
    xmlEntityPtr entity = NULL;
7602
2.56k
    xmlParserInputPtr input;
7603
7604
2.56k
    if (RAW != '%')
7605
0
        return;
7606
2.56k
    NEXT;
7607
2.56k
    name = xmlParseName(ctxt);
7608
2.56k
    if (name == NULL) {
7609
27
  xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n");
7610
27
  return;
7611
27
    }
7612
2.53k
    if (RAW != ';') {
7613
11
  xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL);
7614
11
        return;
7615
11
    }
7616
7617
2.52k
    NEXT;
7618
7619
    /* Must be set before xmlHandleUndeclaredEntity */
7620
2.52k
    ctxt->hasPErefs = 1;
7621
7622
    /*
7623
     * Request the entity from SAX
7624
     */
7625
2.52k
    if ((ctxt->sax != NULL) &&
7626
2.52k
  (ctxt->sax->getParameterEntity != NULL))
7627
2.52k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7628
7629
2.52k
    if (entity == NULL) {
7630
473
        xmlHandleUndeclaredEntity(ctxt, name);
7631
2.05k
    } else {
7632
  /*
7633
   * Internal checking in case the entity quest barfed
7634
   */
7635
2.05k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7636
6
      (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) {
7637
0
      xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7638
0
      "Internal: %%%s; is not a parameter entity\n",
7639
0
        name, NULL);
7640
2.05k
  } else {
7641
2.05k
      if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) &&
7642
6
                ((ctxt->options & XML_PARSE_NO_XXE) ||
7643
6
     (((ctxt->loadsubset & ~XML_SKIP_IDS) == 0) &&
7644
6
      (ctxt->replaceEntities == 0) &&
7645
6
      (ctxt->validate == 0))))
7646
6
    return;
7647
7648
2.04k
            if (entity->flags & XML_ENT_EXPANDING) {
7649
0
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
7650
0
                return;
7651
0
            }
7652
7653
2.04k
      input = xmlNewEntityInputStream(ctxt, entity);
7654
2.04k
      if (xmlCtxtPushInput(ctxt, input) < 0) {
7655
0
                xmlFreeInputStream(input);
7656
0
    return;
7657
0
            }
7658
7659
2.04k
            entity->flags |= XML_ENT_EXPANDING;
7660
7661
2.04k
            if (markupDecl)
7662
2.04k
                input->flags |= XML_INPUT_MARKUP_DECL;
7663
7664
2.04k
            GROW;
7665
7666
2.04k
      if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
7667
0
                xmlDetectEncoding(ctxt);
7668
7669
0
                if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
7670
0
                    (IS_BLANK_CH(NXT(5)))) {
7671
0
                    xmlParseTextDecl(ctxt);
7672
0
                }
7673
0
            }
7674
2.04k
  }
7675
2.05k
    }
7676
2.52k
}
7677
7678
/**
7679
 * Parse a parameter entity reference.
7680
 *
7681
 * @deprecated Internal function, don't use.
7682
 *
7683
 * @param ctxt  an XML parser context
7684
 */
7685
void
7686
0
xmlParsePEReference(xmlParserCtxt *ctxt) {
7687
0
    xmlParsePERefInternal(ctxt, 0);
7688
0
}
7689
7690
/**
7691
 * Load the content of an entity.
7692
 *
7693
 * @param ctxt  an XML parser context
7694
 * @param entity  an unloaded system entity
7695
 * @returns 0 in case of success and -1 in case of failure
7696
 */
7697
static int
7698
0
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) {
7699
0
    xmlParserInputPtr oldinput, input = NULL;
7700
0
    xmlParserInputPtr *oldinputTab;
7701
0
    xmlChar *oldencoding;
7702
0
    xmlChar *content = NULL;
7703
0
    xmlResourceType rtype;
7704
0
    size_t length, i;
7705
0
    int oldinputNr, oldinputMax;
7706
0
    int ret = -1;
7707
0
    int res;
7708
7709
0
    if ((ctxt == NULL) || (entity == NULL) ||
7710
0
        ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) &&
7711
0
   (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) ||
7712
0
  (entity->content != NULL)) {
7713
0
  xmlFatalErr(ctxt, XML_ERR_ARGUMENT,
7714
0
              "xmlLoadEntityContent parameter error");
7715
0
        return(-1);
7716
0
    }
7717
7718
0
    if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY)
7719
0
        rtype = XML_RESOURCE_PARAMETER_ENTITY;
7720
0
    else
7721
0
        rtype = XML_RESOURCE_GENERAL_ENTITY;
7722
7723
0
    input = xmlLoadResource(ctxt, (char *) entity->URI,
7724
0
                            (char *) entity->ExternalID, rtype);
7725
0
    if (input == NULL)
7726
0
        return(-1);
7727
7728
0
    oldinput = ctxt->input;
7729
0
    oldinputNr = ctxt->inputNr;
7730
0
    oldinputMax = ctxt->inputMax;
7731
0
    oldinputTab = ctxt->inputTab;
7732
0
    oldencoding = ctxt->encoding;
7733
7734
0
    ctxt->input = NULL;
7735
0
    ctxt->inputNr = 0;
7736
0
    ctxt->inputMax = 1;
7737
0
    ctxt->encoding = NULL;
7738
0
    ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr));
7739
0
    if (ctxt->inputTab == NULL) {
7740
0
        xmlErrMemory(ctxt);
7741
0
        xmlFreeInputStream(input);
7742
0
        goto error;
7743
0
    }
7744
7745
0
    xmlBufResetInput(input->buf->buffer, input);
7746
7747
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
7748
0
        xmlFreeInputStream(input);
7749
0
        goto error;
7750
0
    }
7751
7752
0
    xmlDetectEncoding(ctxt);
7753
7754
    /*
7755
     * Parse a possible text declaration first
7756
     */
7757
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
7758
0
  xmlParseTextDecl(ctxt);
7759
        /*
7760
         * An XML-1.0 document can't reference an entity not XML-1.0
7761
         */
7762
0
        if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
7763
0
            (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
7764
0
            xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
7765
0
                           "Version mismatch between document and entity\n");
7766
0
        }
7767
0
    }
7768
7769
0
    length = input->cur - input->base;
7770
0
    xmlBufShrink(input->buf->buffer, length);
7771
0
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7772
7773
0
    while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0)
7774
0
        ;
7775
7776
0
    xmlBufResetInput(input->buf->buffer, input);
7777
7778
0
    if (res < 0) {
7779
0
        xmlCtxtErrIO(ctxt, input->buf->error, NULL);
7780
0
        goto error;
7781
0
    }
7782
7783
0
    length = xmlBufUse(input->buf->buffer);
7784
0
    if (length > INT_MAX) {
7785
0
        xmlErrMemory(ctxt);
7786
0
        goto error;
7787
0
    }
7788
7789
0
    content = xmlStrndup(xmlBufContent(input->buf->buffer), length);
7790
0
    if (content == NULL) {
7791
0
        xmlErrMemory(ctxt);
7792
0
        goto error;
7793
0
    }
7794
7795
0
    for (i = 0; i < length; ) {
7796
0
        int clen = length - i;
7797
0
        int c = xmlGetUTF8Char(content + i, &clen);
7798
7799
0
        if ((c < 0) || (!IS_CHAR(c))) {
7800
0
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
7801
0
                              "xmlLoadEntityContent: invalid char value %d\n",
7802
0
                              content[i]);
7803
0
            goto error;
7804
0
        }
7805
0
        i += clen;
7806
0
    }
7807
7808
0
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7809
0
    entity->content = content;
7810
0
    entity->length = length;
7811
0
    content = NULL;
7812
0
    ret = 0;
7813
7814
0
error:
7815
0
    while (ctxt->inputNr > 0)
7816
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
7817
0
    xmlFree(ctxt->inputTab);
7818
0
    xmlFree(ctxt->encoding);
7819
7820
0
    ctxt->input = oldinput;
7821
0
    ctxt->inputNr = oldinputNr;
7822
0
    ctxt->inputMax = oldinputMax;
7823
0
    ctxt->inputTab = oldinputTab;
7824
0
    ctxt->encoding = oldencoding;
7825
7826
0
    xmlFree(content);
7827
7828
0
    return(ret);
7829
0
}
7830
7831
/**
7832
 * Parse PEReference declarations
7833
 *
7834
 *     [69] PEReference ::= '%' Name ';'
7835
 *
7836
 * [ WFC: No Recursion ]
7837
 * A parsed entity must not contain a recursive
7838
 * reference to itself, either directly or indirectly.
7839
 *
7840
 * [ WFC: Entity Declared ]
7841
 * In a document without any DTD, a document with only an internal DTD
7842
 * subset which contains no parameter entity references, or a document
7843
 * with "standalone='yes'", ...  ... The declaration of a parameter
7844
 * entity must precede any reference to it...
7845
 *
7846
 * [ VC: Entity Declared ]
7847
 * In a document with an external subset or external parameter entities
7848
 * with "standalone='no'", ...  ... The declaration of a parameter entity
7849
 * must precede any reference to it...
7850
 *
7851
 * [ WFC: In DTD ]
7852
 * Parameter-entity references may only appear in the DTD.
7853
 * NOTE: misleading but this is handled.
7854
 *
7855
 * @param ctxt  an XML parser context
7856
 * @param str  a pointer to an index in the string
7857
 * @returns the string of the entity content.
7858
 *         str is updated to the current value of the index
7859
 */
7860
static xmlEntityPtr
7861
2.80k
xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) {
7862
2.80k
    const xmlChar *ptr;
7863
2.80k
    xmlChar cur;
7864
2.80k
    xmlChar *name;
7865
2.80k
    xmlEntityPtr entity = NULL;
7866
7867
2.80k
    if ((str == NULL) || (*str == NULL)) return(NULL);
7868
2.80k
    ptr = *str;
7869
2.80k
    cur = *ptr;
7870
2.80k
    if (cur != '%')
7871
0
        return(NULL);
7872
2.80k
    ptr++;
7873
2.80k
    name = xmlParseStringName(ctxt, &ptr);
7874
2.80k
    if (name == NULL) {
7875
60
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7876
60
           "xmlParseStringPEReference: no name\n");
7877
60
  *str = ptr;
7878
60
  return(NULL);
7879
60
    }
7880
2.74k
    cur = *ptr;
7881
2.74k
    if (cur != ';') {
7882
69
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7883
69
  xmlFree(name);
7884
69
  *str = ptr;
7885
69
  return(NULL);
7886
69
    }
7887
2.67k
    ptr++;
7888
7889
    /* Must be set before xmlHandleUndeclaredEntity */
7890
2.67k
    ctxt->hasPErefs = 1;
7891
7892
    /*
7893
     * Request the entity from SAX
7894
     */
7895
2.67k
    if ((ctxt->sax != NULL) &&
7896
2.67k
  (ctxt->sax->getParameterEntity != NULL))
7897
2.67k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7898
7899
2.67k
    if (entity == NULL) {
7900
2.67k
        xmlHandleUndeclaredEntity(ctxt, name);
7901
2.67k
    } else {
7902
  /*
7903
   * Internal checking in case the entity quest barfed
7904
   */
7905
3
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7906
0
      (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) {
7907
0
      xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7908
0
        "%%%s; is not a parameter entity\n",
7909
0
        name, NULL);
7910
0
  }
7911
3
    }
7912
7913
2.67k
    xmlFree(name);
7914
2.67k
    *str = ptr;
7915
2.67k
    return(entity);
7916
2.74k
}
7917
7918
/**
7919
 * Parse a DOCTYPE declaration
7920
 *
7921
 * @deprecated Internal function, don't use.
7922
 *
7923
 *     [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
7924
 *                          ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
7925
 *
7926
 * [ VC: Root Element Type ]
7927
 * The Name in the document type declaration must match the element
7928
 * type of the root element.
7929
 *
7930
 * @param ctxt  an XML parser context
7931
 */
7932
7933
void
7934
17.7k
xmlParseDocTypeDecl(xmlParserCtxt *ctxt) {
7935
17.7k
    const xmlChar *name = NULL;
7936
17.7k
    xmlChar *publicId = NULL;
7937
17.7k
    xmlChar *URI = NULL;
7938
7939
    /*
7940
     * We know that '<!DOCTYPE' has been detected.
7941
     */
7942
17.7k
    SKIP(9);
7943
7944
17.7k
    if (SKIP_BLANKS == 0) {
7945
109
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7946
109
                       "Space required after 'DOCTYPE'\n");
7947
109
    }
7948
7949
    /*
7950
     * Parse the DOCTYPE name.
7951
     */
7952
17.7k
    name = xmlParseName(ctxt);
7953
17.7k
    if (name == NULL) {
7954
77
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7955
77
           "xmlParseDocTypeDecl : no DOCTYPE name !\n");
7956
77
    }
7957
17.7k
    ctxt->intSubName = name;
7958
7959
17.7k
    SKIP_BLANKS;
7960
7961
    /*
7962
     * Check for public and system identifier (URI)
7963
     */
7964
17.7k
    URI = xmlParseExternalID(ctxt, &publicId, 1);
7965
7966
17.7k
    if ((URI != NULL) || (publicId != NULL)) {
7967
568
        ctxt->hasExternalSubset = 1;
7968
568
    }
7969
17.7k
    ctxt->extSubURI = URI;
7970
17.7k
    ctxt->extSubSystem = publicId;
7971
7972
17.7k
    SKIP_BLANKS;
7973
7974
    /*
7975
     * Create and update the internal subset.
7976
     */
7977
17.7k
    if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) &&
7978
17.7k
  (!ctxt->disableSAX))
7979
17.1k
  ctxt->sax->internalSubset(ctxt->userData, name, publicId, URI);
7980
7981
17.7k
    if ((RAW != '[') && (RAW != '>')) {
7982
999
  xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
7983
999
    }
7984
17.7k
}
7985
7986
/**
7987
 * Parse the internal subset declaration
7988
 *
7989
 *     [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
7990
 * @param ctxt  an XML parser context
7991
 */
7992
7993
static void
7994
13.8k
xmlParseInternalSubset(xmlParserCtxtPtr ctxt) {
7995
    /*
7996
     * Is there any DTD definition ?
7997
     */
7998
13.8k
    if (RAW == '[') {
7999
13.8k
        int oldInputNr = ctxt->inputNr;
8000
8001
13.8k
        NEXT;
8002
  /*
8003
   * Parse the succession of Markup declarations and
8004
   * PEReferences.
8005
   * Subsequence (markupdecl | PEReference | S)*
8006
   */
8007
13.8k
  SKIP_BLANKS;
8008
85.9k
        while (1) {
8009
85.9k
            if (PARSER_STOPPED(ctxt)) {
8010
4.75k
                return;
8011
81.1k
            } else if (ctxt->input->cur >= ctxt->input->end) {
8012
1.68k
                if (ctxt->inputNr <= oldInputNr) {
8013
3
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8014
3
                    return;
8015
3
                }
8016
1.68k
                xmlPopPE(ctxt);
8017
79.4k
            } else if ((RAW == ']') && (ctxt->inputNr <= oldInputNr)) {
8018
6.82k
                NEXT;
8019
6.82k
                SKIP_BLANKS;
8020
6.82k
                break;
8021
72.6k
            } else if ((PARSER_EXTERNAL(ctxt)) &&
8022
0
                       (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
8023
                /*
8024
                 * Conditional sections are allowed in external entities
8025
                 * included by PE References in the internal subset.
8026
                 */
8027
0
                xmlParseConditionalSections(ctxt);
8028
72.6k
            } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
8029
67.8k
                xmlParseMarkupDecl(ctxt);
8030
67.8k
            } else if (RAW == '%') {
8031
2.53k
                xmlParsePERefInternal(ctxt, 1);
8032
2.53k
            } else {
8033
2.27k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8034
8035
2.38k
                while (ctxt->inputNr > oldInputNr)
8036
112
                    xmlPopPE(ctxt);
8037
2.27k
                return;
8038
2.27k
            }
8039
72.0k
            SKIP_BLANKS;
8040
72.0k
            SHRINK;
8041
72.0k
            GROW;
8042
72.0k
        }
8043
13.8k
    }
8044
8045
    /*
8046
     * We should be at the end of the DOCTYPE declaration.
8047
     */
8048
6.82k
    if (RAW != '>') {
8049
34
        xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8050
34
        return;
8051
34
    }
8052
6.79k
    NEXT;
8053
6.79k
}
8054
8055
#ifdef LIBXML_SAX1_ENABLED
8056
/**
8057
 * Parse an attribute
8058
 *
8059
 * @deprecated Internal function, don't use.
8060
 *
8061
 *     [41] Attribute ::= Name Eq AttValue
8062
 *
8063
 * [ WFC: No External Entity References ]
8064
 * Attribute values cannot contain direct or indirect entity references
8065
 * to external entities.
8066
 *
8067
 * [ WFC: No < in Attribute Values ]
8068
 * The replacement text of any entity referred to directly or indirectly in
8069
 * an attribute value (other than "&lt;") must not contain a <.
8070
 *
8071
 * [ VC: Attribute Value Type ]
8072
 * The attribute must have been declared; the value must be of the type
8073
 * declared for it.
8074
 *
8075
 *     [25] Eq ::= S? '=' S?
8076
 *
8077
 * With namespace:
8078
 *
8079
 *     [NS 11] Attribute ::= QName Eq AttValue
8080
 *
8081
 * Also the case QName == xmlns:??? is handled independently as a namespace
8082
 * definition.
8083
 *
8084
 * @param ctxt  an XML parser context
8085
 * @param value  a xmlChar ** used to store the value of the attribute
8086
 * @returns the attribute name, and the value in *value.
8087
 */
8088
8089
const xmlChar *
8090
875k
xmlParseAttribute(xmlParserCtxt *ctxt, xmlChar **value) {
8091
875k
    const xmlChar *name;
8092
875k
    xmlChar *val;
8093
8094
875k
    *value = NULL;
8095
875k
    GROW;
8096
875k
    name = xmlParseName(ctxt);
8097
875k
    if (name == NULL) {
8098
2.85k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8099
2.85k
                 "error parsing attribute name\n");
8100
2.85k
        return(NULL);
8101
2.85k
    }
8102
8103
    /*
8104
     * read the value
8105
     */
8106
873k
    SKIP_BLANKS;
8107
873k
    if (RAW == '=') {
8108
870k
        NEXT;
8109
870k
  SKIP_BLANKS;
8110
870k
  val = xmlParseAttValue(ctxt);
8111
870k
    } else {
8112
2.43k
  xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8113
2.43k
         "Specification mandates value for attribute %s\n", name);
8114
2.43k
  return(name);
8115
2.43k
    }
8116
8117
    /*
8118
     * Check that xml:lang conforms to the specification
8119
     * No more registered as an error, just generate a warning now
8120
     * since this was deprecated in XML second edition
8121
     */
8122
870k
    if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) {
8123
0
  if (!xmlCheckLanguageID(val)) {
8124
0
      xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8125
0
              "Malformed value for xml:lang : %s\n",
8126
0
        val, NULL);
8127
0
  }
8128
0
    }
8129
8130
    /*
8131
     * Check that xml:space conforms to the specification
8132
     */
8133
870k
    if (xmlStrEqual(name, BAD_CAST "xml:space")) {
8134
2.21k
  if (xmlStrEqual(val, BAD_CAST "default"))
8135
663
      *(ctxt->space) = 0;
8136
1.55k
  else if (xmlStrEqual(val, BAD_CAST "preserve"))
8137
627
      *(ctxt->space) = 1;
8138
925
  else {
8139
925
    xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8140
925
"Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8141
925
                                 val, NULL);
8142
925
  }
8143
2.21k
    }
8144
8145
870k
    *value = val;
8146
870k
    return(name);
8147
873k
}
8148
8149
/**
8150
 * Parse a start tag. Always consumes '<'.
8151
 *
8152
 * @deprecated Internal function, don't use.
8153
 *
8154
 *     [40] STag ::= '<' Name (S Attribute)* S? '>'
8155
 *
8156
 * [ WFC: Unique Att Spec ]
8157
 * No attribute name may appear more than once in the same start-tag or
8158
 * empty-element tag.
8159
 *
8160
 *     [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
8161
 *
8162
 * [ WFC: Unique Att Spec ]
8163
 * No attribute name may appear more than once in the same start-tag or
8164
 * empty-element tag.
8165
 *
8166
 * With namespace:
8167
 *
8168
 *     [NS 8] STag ::= '<' QName (S Attribute)* S? '>'
8169
 *
8170
 *     [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>'
8171
 *
8172
 * @param ctxt  an XML parser context
8173
 * @returns the element name parsed
8174
 */
8175
8176
const xmlChar *
8177
2.21M
xmlParseStartTag(xmlParserCtxt *ctxt) {
8178
2.21M
    const xmlChar *name;
8179
2.21M
    const xmlChar *attname;
8180
2.21M
    xmlChar *attvalue;
8181
2.21M
    const xmlChar **atts = ctxt->atts;
8182
2.21M
    int nbatts = 0;
8183
2.21M
    int maxatts = ctxt->maxatts;
8184
2.21M
    int i;
8185
8186
2.21M
    if (RAW != '<') return(NULL);
8187
2.21M
    NEXT1;
8188
8189
2.21M
    name = xmlParseName(ctxt);
8190
2.21M
    if (name == NULL) {
8191
4.91k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8192
4.91k
       "xmlParseStartTag: invalid element name\n");
8193
4.91k
        return(NULL);
8194
4.91k
    }
8195
8196
    /*
8197
     * Now parse the attributes, it ends up with the ending
8198
     *
8199
     * (S Attribute)* S?
8200
     */
8201
2.20M
    SKIP_BLANKS;
8202
2.20M
    GROW;
8203
8204
2.77M
    while (((RAW != '>') &&
8205
1.58M
     ((RAW != '/') || (NXT(1) != '>')) &&
8206
888k
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8207
875k
  attname = xmlParseAttribute(ctxt, &attvalue);
8208
875k
        if (attname == NULL)
8209
2.85k
      break;
8210
873k
        if (attvalue != NULL) {
8211
      /*
8212
       * [ WFC: Unique Att Spec ]
8213
       * No attribute name may appear more than once in the same
8214
       * start-tag or empty-element tag.
8215
       */
8216
1.68M
      for (i = 0; i < nbatts;i += 2) {
8217
821k
          if (xmlStrEqual(atts[i], attname)) {
8218
26
        xmlErrAttributeDup(ctxt, NULL, attname);
8219
26
        goto failed;
8220
26
    }
8221
821k
      }
8222
      /*
8223
       * Add the pair to atts
8224
       */
8225
866k
      if (nbatts + 4 > maxatts) {
8226
30.9k
          const xmlChar **n;
8227
30.9k
                int newSize;
8228
8229
30.9k
                newSize = xmlGrowCapacity(maxatts, sizeof(n[0]) * 2,
8230
30.9k
                                          11, XML_MAX_ATTRS);
8231
30.9k
                if (newSize < 0) {
8232
0
        xmlErrMemory(ctxt);
8233
0
        goto failed;
8234
0
    }
8235
30.9k
#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
8236
30.9k
                if (newSize < 2)
8237
20.3k
                    newSize = 2;
8238
30.9k
#endif
8239
30.9k
          n = xmlRealloc(atts, newSize * sizeof(n[0]) * 2);
8240
30.9k
    if (n == NULL) {
8241
0
        xmlErrMemory(ctxt);
8242
0
        goto failed;
8243
0
    }
8244
30.9k
    atts = n;
8245
30.9k
                maxatts = newSize * 2;
8246
30.9k
    ctxt->atts = atts;
8247
30.9k
    ctxt->maxatts = maxatts;
8248
30.9k
      }
8249
8250
866k
      atts[nbatts++] = attname;
8251
866k
      atts[nbatts++] = attvalue;
8252
866k
      atts[nbatts] = NULL;
8253
866k
      atts[nbatts + 1] = NULL;
8254
8255
866k
            attvalue = NULL;
8256
866k
  }
8257
8258
873k
failed:
8259
8260
873k
        if (attvalue != NULL)
8261
26
            xmlFree(attvalue);
8262
8263
873k
  GROW
8264
873k
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
8265
304k
      break;
8266
568k
  if (SKIP_BLANKS == 0) {
8267
6.44k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
8268
6.44k
         "attributes construct error\n");
8269
6.44k
  }
8270
568k
  SHRINK;
8271
568k
        GROW;
8272
568k
    }
8273
8274
    /*
8275
     * SAX: Start of Element !
8276
     */
8277
2.20M
    if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) &&
8278
2.20M
  (!ctxt->disableSAX)) {
8279
2.19M
  if (nbatts > 0)
8280
477k
      ctxt->sax->startElement(ctxt->userData, name, atts);
8281
1.71M
  else
8282
1.71M
      ctxt->sax->startElement(ctxt->userData, name, NULL);
8283
2.19M
    }
8284
8285
2.20M
    if (atts != NULL) {
8286
        /* Free only the content strings */
8287
2.36M
        for (i = 1;i < nbatts;i+=2)
8288
866k
      if (atts[i] != NULL)
8289
866k
         xmlFree((xmlChar *) atts[i]);
8290
1.49M
    }
8291
2.20M
    return(name);
8292
2.20M
}
8293
8294
/**
8295
 * Parse an end tag. Always consumes '</'.
8296
 *
8297
 *     [42] ETag ::= '</' Name S? '>'
8298
 *
8299
 * With namespace
8300
 *
8301
 *     [NS 9] ETag ::= '</' QName S? '>'
8302
 * @param ctxt  an XML parser context
8303
 * @param line  line of the start tag
8304
 */
8305
8306
static void
8307
235k
xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) {
8308
235k
    const xmlChar *name;
8309
8310
235k
    GROW;
8311
235k
    if ((RAW != '<') || (NXT(1) != '/')) {
8312
0
  xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED,
8313
0
           "xmlParseEndTag: '</' not found\n");
8314
0
  return;
8315
0
    }
8316
235k
    SKIP(2);
8317
8318
235k
    name = xmlParseNameAndCompare(ctxt,ctxt->name);
8319
8320
    /*
8321
     * We should definitely be at the ending "S? '>'" part
8322
     */
8323
235k
    GROW;
8324
235k
    SKIP_BLANKS;
8325
235k
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
8326
2.21k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
8327
2.21k
    } else
8328
232k
  NEXT1;
8329
8330
    /*
8331
     * [ WFC: Element Type Match ]
8332
     * The Name in an element's end-tag must match the element type in the
8333
     * start-tag.
8334
     *
8335
     */
8336
235k
    if (name != (xmlChar*)1) {
8337
2.47k
        if (name == NULL) name = BAD_CAST "unparsable";
8338
2.47k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
8339
2.47k
         "Opening and ending tag mismatch: %s line %d and %s\n",
8340
2.47k
                    ctxt->name, line, name);
8341
2.47k
    }
8342
8343
    /*
8344
     * SAX: End of Tag
8345
     */
8346
235k
    if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
8347
235k
  (!ctxt->disableSAX))
8348
232k
        ctxt->sax->endElement(ctxt->userData, ctxt->name);
8349
8350
235k
    namePop(ctxt);
8351
235k
    spacePop(ctxt);
8352
235k
}
8353
8354
/**
8355
 * Parse an end of tag
8356
 *
8357
 * @deprecated Internal function, don't use.
8358
 *
8359
 *     [42] ETag ::= '</' Name S? '>'
8360
 *
8361
 * With namespace
8362
 *
8363
 *     [NS 9] ETag ::= '</' QName S? '>'
8364
 * @param ctxt  an XML parser context
8365
 */
8366
8367
void
8368
0
xmlParseEndTag(xmlParserCtxt *ctxt) {
8369
0
    xmlParseEndTag1(ctxt, 0);
8370
0
}
8371
#endif /* LIBXML_SAX1_ENABLED */
8372
8373
/************************************************************************
8374
 *                  *
8375
 *          SAX 2 specific operations       *
8376
 *                  *
8377
 ************************************************************************/
8378
8379
/**
8380
 * Parse an XML Namespace QName
8381
 *
8382
 *     [6]  QName  ::= (Prefix ':')? LocalPart
8383
 *     [7]  Prefix  ::= NCName
8384
 *     [8]  LocalPart  ::= NCName
8385
 *
8386
 * @param ctxt  an XML parser context
8387
 * @param prefix  pointer to store the prefix part
8388
 * @returns the Name parsed or NULL
8389
 */
8390
8391
static xmlHashedString
8392
0
xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) {
8393
0
    xmlHashedString l, p;
8394
0
    int start, isNCName = 0;
8395
8396
0
    l.name = NULL;
8397
0
    p.name = NULL;
8398
8399
0
    GROW;
8400
0
    start = CUR_PTR - BASE_PTR;
8401
8402
0
    l = xmlParseNCName(ctxt);
8403
0
    if (l.name != NULL) {
8404
0
        isNCName = 1;
8405
0
        if (CUR == ':') {
8406
0
            NEXT;
8407
0
            p = l;
8408
0
            l = xmlParseNCName(ctxt);
8409
0
        }
8410
0
    }
8411
0
    if ((l.name == NULL) || (CUR == ':')) {
8412
0
        xmlChar *tmp;
8413
8414
0
        l.name = NULL;
8415
0
        p.name = NULL;
8416
0
        if ((isNCName == 0) && (CUR != ':'))
8417
0
            return(l);
8418
0
        tmp = xmlParseNmtoken(ctxt);
8419
0
        if (tmp != NULL)
8420
0
            xmlFree(tmp);
8421
0
        l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start,
8422
0
                                CUR_PTR - (BASE_PTR + start));
8423
0
        if (l.name == NULL) {
8424
0
            xmlErrMemory(ctxt);
8425
0
            return(l);
8426
0
        }
8427
0
        xmlNsErr(ctxt, XML_NS_ERR_QNAME,
8428
0
                 "Failed to parse QName '%s'\n", l.name, NULL, NULL);
8429
0
    }
8430
8431
0
    *prefix = p;
8432
0
    return(l);
8433
0
}
8434
8435
/**
8436
 * Parse an XML Namespace QName
8437
 *
8438
 *     [6]  QName  ::= (Prefix ':')? LocalPart
8439
 *     [7]  Prefix  ::= NCName
8440
 *     [8]  LocalPart  ::= NCName
8441
 *
8442
 * @param ctxt  an XML parser context
8443
 * @param prefix  pointer to store the prefix part
8444
 * @returns the Name parsed or NULL
8445
 */
8446
8447
static const xmlChar *
8448
0
xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) {
8449
0
    xmlHashedString n, p;
8450
8451
0
    n = xmlParseQNameHashed(ctxt, &p);
8452
0
    if (n.name == NULL)
8453
0
        return(NULL);
8454
0
    *prefix = p.name;
8455
0
    return(n.name);
8456
0
}
8457
8458
/**
8459
 * Parse an XML name and compares for match
8460
 * (specialized for endtag parsing)
8461
 *
8462
 * @param ctxt  an XML parser context
8463
 * @param name  the localname
8464
 * @param prefix  the prefix, if any.
8465
 * @returns NULL for an illegal name, (xmlChar*) 1 for success
8466
 * and the name for mismatch
8467
 */
8468
8469
static const xmlChar *
8470
xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name,
8471
0
                        xmlChar const *prefix) {
8472
0
    const xmlChar *cmp;
8473
0
    const xmlChar *in;
8474
0
    const xmlChar *ret;
8475
0
    const xmlChar *prefix2;
8476
8477
0
    if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name));
8478
8479
0
    GROW;
8480
0
    in = ctxt->input->cur;
8481
8482
0
    cmp = prefix;
8483
0
    while (*in != 0 && *in == *cmp) {
8484
0
  ++in;
8485
0
  ++cmp;
8486
0
    }
8487
0
    if ((*cmp == 0) && (*in == ':')) {
8488
0
        in++;
8489
0
  cmp = name;
8490
0
  while (*in != 0 && *in == *cmp) {
8491
0
      ++in;
8492
0
      ++cmp;
8493
0
  }
8494
0
  if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
8495
      /* success */
8496
0
            ctxt->input->col += in - ctxt->input->cur;
8497
0
      ctxt->input->cur = in;
8498
0
      return((const xmlChar*) 1);
8499
0
  }
8500
0
    }
8501
    /*
8502
     * all strings coms from the dictionary, equality can be done directly
8503
     */
8504
0
    ret = xmlParseQName (ctxt, &prefix2);
8505
0
    if (ret == NULL)
8506
0
        return(NULL);
8507
0
    if ((ret == name) && (prefix == prefix2))
8508
0
  return((const xmlChar*) 1);
8509
0
    return ret;
8510
0
}
8511
8512
/**
8513
 * Parse an attribute in the new SAX2 framework.
8514
 *
8515
 * @param ctxt  an XML parser context
8516
 * @param pref  the element prefix
8517
 * @param elem  the element name
8518
 * @param hprefix  resulting attribute prefix
8519
 * @param value  resulting value of the attribute
8520
 * @param len  resulting length of the attribute
8521
 * @param alloc  resulting indicator if the attribute was allocated
8522
 * @returns the attribute name, and the value in *value, .
8523
 */
8524
8525
static xmlHashedString
8526
xmlParseAttribute2(xmlParserCtxtPtr ctxt,
8527
                   const xmlChar * pref, const xmlChar * elem,
8528
                   xmlHashedString * hprefix, xmlChar ** value,
8529
                   int *len, int *alloc)
8530
0
{
8531
0
    xmlHashedString hname;
8532
0
    const xmlChar *prefix, *name;
8533
0
    xmlChar *val = NULL, *internal_val = NULL;
8534
0
    int special = 0;
8535
0
    int isNamespace;
8536
0
    int flags;
8537
8538
0
    *value = NULL;
8539
0
    GROW;
8540
0
    hname = xmlParseQNameHashed(ctxt, hprefix);
8541
0
    if (hname.name == NULL) {
8542
0
        xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8543
0
                       "error parsing attribute name\n");
8544
0
        return(hname);
8545
0
    }
8546
0
    name = hname.name;
8547
0
    prefix = hprefix->name;
8548
8549
    /*
8550
     * get the type if needed
8551
     */
8552
0
    if (ctxt->attsSpecial != NULL) {
8553
0
        special = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem,
8554
0
                                              prefix, name));
8555
0
    }
8556
8557
    /*
8558
     * read the value
8559
     */
8560
0
    SKIP_BLANKS;
8561
0
    if (RAW != '=') {
8562
0
        xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8563
0
                          "Specification mandates value for attribute %s\n",
8564
0
                          name);
8565
0
        goto error;
8566
0
    }
8567
8568
8569
0
    NEXT;
8570
0
    SKIP_BLANKS;
8571
0
    flags = 0;
8572
0
    isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) ||
8573
0
                   (prefix == ctxt->str_xmlns));
8574
0
    val = xmlParseAttValueInternal(ctxt, len, &flags, special,
8575
0
                                   isNamespace);
8576
0
    if (val == NULL)
8577
0
        goto error;
8578
8579
0
    *alloc = (flags & XML_ATTVAL_ALLOC) != 0;
8580
8581
0
#ifdef LIBXML_VALID_ENABLED
8582
0
    if ((ctxt->validate) &&
8583
0
        (ctxt->standalone == 1) &&
8584
0
        (special & XML_SPECIAL_EXTERNAL) &&
8585
0
        (flags & XML_ATTVAL_NORM_CHANGE)) {
8586
0
        xmlValidityError(ctxt, XML_DTD_NOT_STANDALONE,
8587
0
                         "standalone: normalization of attribute %s on %s "
8588
0
                         "by external subset declaration\n",
8589
0
                         name, elem);
8590
0
    }
8591
0
#endif
8592
8593
0
    if (prefix == ctxt->str_xml) {
8594
        /*
8595
         * Check that xml:lang conforms to the specification
8596
         * No more registered as an error, just generate a warning now
8597
         * since this was deprecated in XML second edition
8598
         */
8599
0
        if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) {
8600
0
            internal_val = xmlStrndup(val, *len);
8601
0
            if (internal_val == NULL)
8602
0
                goto mem_error;
8603
0
            if (!xmlCheckLanguageID(internal_val)) {
8604
0
                xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8605
0
                              "Malformed value for xml:lang : %s\n",
8606
0
                              internal_val, NULL);
8607
0
            }
8608
0
        }
8609
8610
        /*
8611
         * Check that xml:space conforms to the specification
8612
         */
8613
0
        if (xmlStrEqual(name, BAD_CAST "space")) {
8614
0
            internal_val = xmlStrndup(val, *len);
8615
0
            if (internal_val == NULL)
8616
0
                goto mem_error;
8617
0
            if (xmlStrEqual(internal_val, BAD_CAST "default"))
8618
0
                *(ctxt->space) = 0;
8619
0
            else if (xmlStrEqual(internal_val, BAD_CAST "preserve"))
8620
0
                *(ctxt->space) = 1;
8621
0
            else {
8622
0
                xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8623
0
                              "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8624
0
                              internal_val, NULL);
8625
0
            }
8626
0
        }
8627
0
        if (internal_val) {
8628
0
            xmlFree(internal_val);
8629
0
        }
8630
0
    }
8631
8632
0
    *value = val;
8633
0
    return (hname);
8634
8635
0
mem_error:
8636
0
    xmlErrMemory(ctxt);
8637
0
error:
8638
0
    if ((val != NULL) && (*alloc != 0))
8639
0
        xmlFree(val);
8640
0
    return(hname);
8641
0
}
8642
8643
/**
8644
 * Inserts a new attribute into the hash table.
8645
 *
8646
 * @param ctxt  parser context
8647
 * @param size  size of the hash table
8648
 * @param name  attribute name
8649
 * @param uri  namespace uri
8650
 * @param hashValue  combined hash value of name and uri
8651
 * @param aindex  attribute index (this is a multiple of 5)
8652
 * @returns INT_MAX if no existing attribute was found, the attribute
8653
 * index if an attribute was found, -1 if a memory allocation failed.
8654
 */
8655
static int
8656
xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name,
8657
0
                  const xmlChar *uri, unsigned hashValue, int aindex) {
8658
0
    xmlAttrHashBucket *table = ctxt->attrHash;
8659
0
    xmlAttrHashBucket *bucket;
8660
0
    unsigned hindex;
8661
8662
0
    hindex = hashValue & (size - 1);
8663
0
    bucket = &table[hindex];
8664
8665
0
    while (bucket->index >= 0) {
8666
0
        const xmlChar **atts = &ctxt->atts[bucket->index];
8667
8668
0
        if (name == atts[0]) {
8669
0
            int nsIndex = XML_PTR_TO_INT(atts[2]);
8670
8671
0
            if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) :
8672
0
                (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) :
8673
0
                (uri == ctxt->nsTab[nsIndex * 2 + 1]))
8674
0
                return(bucket->index);
8675
0
        }
8676
8677
0
        hindex++;
8678
0
        bucket++;
8679
0
        if (hindex >= size) {
8680
0
            hindex = 0;
8681
0
            bucket = table;
8682
0
        }
8683
0
    }
8684
8685
0
    bucket->index = aindex;
8686
8687
0
    return(INT_MAX);
8688
0
}
8689
8690
static int
8691
xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size,
8692
                       const xmlChar *name, const xmlChar *prefix,
8693
0
                       unsigned hashValue, int aindex) {
8694
0
    xmlAttrHashBucket *table = ctxt->attrHash;
8695
0
    xmlAttrHashBucket *bucket;
8696
0
    unsigned hindex;
8697
8698
0
    hindex = hashValue & (size - 1);
8699
0
    bucket = &table[hindex];
8700
8701
0
    while (bucket->index >= 0) {
8702
0
        const xmlChar **atts = &ctxt->atts[bucket->index];
8703
8704
0
        if ((name == atts[0]) && (prefix == atts[1]))
8705
0
            return(bucket->index);
8706
8707
0
        hindex++;
8708
0
        bucket++;
8709
0
        if (hindex >= size) {
8710
0
            hindex = 0;
8711
0
            bucket = table;
8712
0
        }
8713
0
    }
8714
8715
0
    bucket->index = aindex;
8716
8717
0
    return(INT_MAX);
8718
0
}
8719
/**
8720
 * Parse a start tag. Always consumes '<'.
8721
 *
8722
 * This routine is called when running SAX2 parsing
8723
 *
8724
 *     [40] STag ::= '<' Name (S Attribute)* S? '>'
8725
 *
8726
 * [ WFC: Unique Att Spec ]
8727
 * No attribute name may appear more than once in the same start-tag or
8728
 * empty-element tag.
8729
 *
8730
 *     [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
8731
 *
8732
 * [ WFC: Unique Att Spec ]
8733
 * No attribute name may appear more than once in the same start-tag or
8734
 * empty-element tag.
8735
 *
8736
 * With namespace:
8737
 *
8738
 *     [NS 8] STag ::= '<' QName (S Attribute)* S? '>'
8739
 *
8740
 *     [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>'
8741
 *
8742
 * @param ctxt  an XML parser context
8743
 * @param pref  resulting namespace prefix
8744
 * @param URI  resulting namespace URI
8745
 * @param nbNsPtr  resulting number of namespace declarations
8746
 * @returns the element name parsed
8747
 */
8748
8749
static const xmlChar *
8750
xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref,
8751
0
                  const xmlChar **URI, int *nbNsPtr) {
8752
0
    xmlHashedString hlocalname;
8753
0
    xmlHashedString hprefix;
8754
0
    xmlHashedString hattname;
8755
0
    xmlHashedString haprefix;
8756
0
    const xmlChar *localname;
8757
0
    const xmlChar *prefix;
8758
0
    const xmlChar *attname;
8759
0
    const xmlChar *aprefix;
8760
0
    const xmlChar *uri;
8761
0
    xmlChar *attvalue = NULL;
8762
0
    const xmlChar **atts = ctxt->atts;
8763
0
    unsigned attrHashSize = 0;
8764
0
    int maxatts = ctxt->maxatts;
8765
0
    int nratts, nbatts, nbdef;
8766
0
    int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts;
8767
0
    int alloc = 0;
8768
0
    int numNsErr = 0;
8769
0
    int numDupErr = 0;
8770
8771
0
    if (RAW != '<') return(NULL);
8772
0
    NEXT1;
8773
8774
0
    nbatts = 0;
8775
0
    nratts = 0;
8776
0
    nbdef = 0;
8777
0
    nbNs = 0;
8778
0
    nbTotalDef = 0;
8779
0
    attval = 0;
8780
8781
0
    if (xmlParserNsStartElement(ctxt->nsdb) < 0) {
8782
0
        xmlErrMemory(ctxt);
8783
0
        return(NULL);
8784
0
    }
8785
8786
0
    hlocalname = xmlParseQNameHashed(ctxt, &hprefix);
8787
0
    if (hlocalname.name == NULL) {
8788
0
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8789
0
           "StartTag: invalid element name\n");
8790
0
        return(NULL);
8791
0
    }
8792
0
    localname = hlocalname.name;
8793
0
    prefix = hprefix.name;
8794
8795
    /*
8796
     * Now parse the attributes, it ends up with the ending
8797
     *
8798
     * (S Attribute)* S?
8799
     */
8800
0
    SKIP_BLANKS;
8801
0
    GROW;
8802
8803
    /*
8804
     * The ctxt->atts array will be ultimately passed to the SAX callback
8805
     * containing five xmlChar pointers for each attribute:
8806
     *
8807
     * [0] attribute name
8808
     * [1] attribute prefix
8809
     * [2] namespace URI
8810
     * [3] attribute value
8811
     * [4] end of attribute value
8812
     *
8813
     * To save memory, we reuse this array temporarily and store integers
8814
     * in these pointer variables.
8815
     *
8816
     * [0] attribute name
8817
     * [1] attribute prefix
8818
     * [2] hash value of attribute prefix, and later namespace index
8819
     * [3] for non-allocated values: ptrdiff_t offset into input buffer
8820
     * [4] for non-allocated values: ptrdiff_t offset into input buffer
8821
     *
8822
     * The ctxt->attallocs array contains an additional unsigned int for
8823
     * each attribute, containing the hash value of the attribute name
8824
     * and the alloc flag in bit 31.
8825
     */
8826
8827
0
    while (((RAW != '>') &&
8828
0
     ((RAW != '/') || (NXT(1) != '>')) &&
8829
0
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8830
0
  int len = -1;
8831
8832
0
  hattname = xmlParseAttribute2(ctxt, prefix, localname,
8833
0
                                          &haprefix, &attvalue, &len,
8834
0
                                          &alloc);
8835
0
        if (hattname.name == NULL)
8836
0
      break;
8837
0
        if (attvalue == NULL)
8838
0
            goto next_attr;
8839
0
        attname = hattname.name;
8840
0
        aprefix = haprefix.name;
8841
0
  if (len < 0) len = xmlStrlen(attvalue);
8842
8843
0
        if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
8844
0
            xmlHashedString huri;
8845
0
            xmlURIPtr parsedUri;
8846
8847
0
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8848
0
            uri = huri.name;
8849
0
            if (uri == NULL) {
8850
0
                xmlErrMemory(ctxt);
8851
0
                goto next_attr;
8852
0
            }
8853
0
            if (*uri != 0) {
8854
0
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8855
0
                    xmlErrMemory(ctxt);
8856
0
                    goto next_attr;
8857
0
                }
8858
0
                if (parsedUri == NULL) {
8859
0
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8860
0
                             "xmlns: '%s' is not a valid URI\n",
8861
0
                                       uri, NULL, NULL);
8862
0
                } else {
8863
0
                    if (parsedUri->scheme == NULL) {
8864
0
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8865
0
                                  "xmlns: URI %s is not absolute\n",
8866
0
                                  uri, NULL, NULL);
8867
0
                    }
8868
0
                    xmlFreeURI(parsedUri);
8869
0
                }
8870
0
                if (uri == ctxt->str_xml_ns) {
8871
0
                    if (attname != ctxt->str_xml) {
8872
0
                        xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8873
0
                     "xml namespace URI cannot be the default namespace\n",
8874
0
                                 NULL, NULL, NULL);
8875
0
                    }
8876
0
                    goto next_attr;
8877
0
                }
8878
0
                if ((len == 29) &&
8879
0
                    (xmlStrEqual(uri,
8880
0
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8881
0
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8882
0
                         "reuse of the xmlns namespace name is forbidden\n",
8883
0
                             NULL, NULL, NULL);
8884
0
                    goto next_attr;
8885
0
                }
8886
0
            }
8887
8888
0
            if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0)
8889
0
                nbNs++;
8890
0
        } else if (aprefix == ctxt->str_xmlns) {
8891
0
            xmlHashedString huri;
8892
0
            xmlURIPtr parsedUri;
8893
8894
0
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8895
0
            uri = huri.name;
8896
0
            if (uri == NULL) {
8897
0
                xmlErrMemory(ctxt);
8898
0
                goto next_attr;
8899
0
            }
8900
8901
0
            if (attname == ctxt->str_xml) {
8902
0
                if (uri != ctxt->str_xml_ns) {
8903
0
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8904
0
                             "xml namespace prefix mapped to wrong URI\n",
8905
0
                             NULL, NULL, NULL);
8906
0
                }
8907
                /*
8908
                 * Do not keep a namespace definition node
8909
                 */
8910
0
                goto next_attr;
8911
0
            }
8912
0
            if (uri == ctxt->str_xml_ns) {
8913
0
                if (attname != ctxt->str_xml) {
8914
0
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8915
0
                             "xml namespace URI mapped to wrong prefix\n",
8916
0
                             NULL, NULL, NULL);
8917
0
                }
8918
0
                goto next_attr;
8919
0
            }
8920
0
            if (attname == ctxt->str_xmlns) {
8921
0
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8922
0
                         "redefinition of the xmlns prefix is forbidden\n",
8923
0
                         NULL, NULL, NULL);
8924
0
                goto next_attr;
8925
0
            }
8926
0
            if ((len == 29) &&
8927
0
                (xmlStrEqual(uri,
8928
0
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8929
0
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8930
0
                         "reuse of the xmlns namespace name is forbidden\n",
8931
0
                         NULL, NULL, NULL);
8932
0
                goto next_attr;
8933
0
            }
8934
0
            if ((uri == NULL) || (uri[0] == 0)) {
8935
0
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8936
0
                         "xmlns:%s: Empty XML namespace is not allowed\n",
8937
0
                              attname, NULL, NULL);
8938
0
                goto next_attr;
8939
0
            } else {
8940
0
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8941
0
                    xmlErrMemory(ctxt);
8942
0
                    goto next_attr;
8943
0
                }
8944
0
                if (parsedUri == NULL) {
8945
0
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8946
0
                         "xmlns:%s: '%s' is not a valid URI\n",
8947
0
                                       attname, uri, NULL);
8948
0
                } else {
8949
0
                    if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) {
8950
0
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8951
0
                                  "xmlns:%s: URI %s is not absolute\n",
8952
0
                                  attname, uri, NULL);
8953
0
                    }
8954
0
                    xmlFreeURI(parsedUri);
8955
0
                }
8956
0
            }
8957
8958
0
            if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0)
8959
0
                nbNs++;
8960
0
        } else {
8961
            /*
8962
             * Populate attributes array, see above for repurposing
8963
             * of xmlChar pointers.
8964
             */
8965
0
            if ((atts == NULL) || (nbatts + 5 > maxatts)) {
8966
0
                int res = xmlCtxtGrowAttrs(ctxt);
8967
8968
0
                maxatts = ctxt->maxatts;
8969
0
                atts = ctxt->atts;
8970
8971
0
                if (res < 0)
8972
0
                    goto next_attr;
8973
0
            }
8974
0
            ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) |
8975
0
                                        ((unsigned) alloc << 31);
8976
0
            atts[nbatts++] = attname;
8977
0
            atts[nbatts++] = aprefix;
8978
0
            atts[nbatts++] = XML_INT_TO_PTR(haprefix.hashValue);
8979
0
            if (alloc) {
8980
0
                atts[nbatts++] = attvalue;
8981
0
                attvalue += len;
8982
0
                atts[nbatts++] = attvalue;
8983
0
            } else {
8984
                /*
8985
                 * attvalue points into the input buffer which can be
8986
                 * reallocated. Store differences to input->base instead.
8987
                 * The pointers will be reconstructed later.
8988
                 */
8989
0
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8990
0
                attvalue += len;
8991
0
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8992
0
            }
8993
            /*
8994
             * tag if some deallocation is needed
8995
             */
8996
0
            if (alloc != 0) attval = 1;
8997
0
            attvalue = NULL; /* moved into atts */
8998
0
        }
8999
9000
0
next_attr:
9001
0
        if ((attvalue != NULL) && (alloc != 0)) {
9002
0
            xmlFree(attvalue);
9003
0
            attvalue = NULL;
9004
0
        }
9005
9006
0
  GROW
9007
0
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
9008
0
      break;
9009
0
  if (SKIP_BLANKS == 0) {
9010
0
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
9011
0
         "attributes construct error\n");
9012
0
      break;
9013
0
  }
9014
0
        GROW;
9015
0
    }
9016
9017
    /*
9018
     * Namespaces from default attributes
9019
     */
9020
0
    if (ctxt->attsDefault != NULL) {
9021
0
        xmlDefAttrsPtr defaults;
9022
9023
0
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9024
0
  if (defaults != NULL) {
9025
0
      for (i = 0; i < defaults->nbAttrs; i++) {
9026
0
                xmlDefAttr *attr = &defaults->attrs[i];
9027
9028
0
          attname = attr->name.name;
9029
0
    aprefix = attr->prefix.name;
9030
9031
0
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
9032
0
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9033
9034
0
                    if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0)
9035
0
                        nbNs++;
9036
0
    } else if (aprefix == ctxt->str_xmlns) {
9037
0
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9038
9039
0
                    if (xmlParserNsPush(ctxt, &attr->name, &attr->value,
9040
0
                                      NULL, 1) > 0)
9041
0
                        nbNs++;
9042
0
    } else {
9043
0
                    if (nratts + nbTotalDef >= XML_MAX_ATTRS) {
9044
0
                        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
9045
0
                                    "Maximum number of attributes exceeded");
9046
0
                        break;
9047
0
                    }
9048
0
                    nbTotalDef += 1;
9049
0
                }
9050
0
      }
9051
0
  }
9052
0
    }
9053
9054
    /*
9055
     * Resolve attribute namespaces
9056
     */
9057
0
    for (i = 0; i < nbatts; i += 5) {
9058
0
        attname = atts[i];
9059
0
        aprefix = atts[i+1];
9060
9061
        /*
9062
  * The default namespace does not apply to attribute names.
9063
  */
9064
0
  if (aprefix == NULL) {
9065
0
            nsIndex = NS_INDEX_EMPTY;
9066
0
        } else if (aprefix == ctxt->str_xml) {
9067
0
            nsIndex = NS_INDEX_XML;
9068
0
        } else {
9069
0
            haprefix.name = aprefix;
9070
0
            haprefix.hashValue = (size_t) atts[i+2];
9071
0
            nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL);
9072
9073
0
      if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) {
9074
0
                xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9075
0
        "Namespace prefix %s for %s on %s is not defined\n",
9076
0
        aprefix, attname, localname);
9077
0
                nsIndex = NS_INDEX_EMPTY;
9078
0
            }
9079
0
        }
9080
9081
0
        atts[i+2] = XML_INT_TO_PTR(nsIndex);
9082
0
    }
9083
9084
    /*
9085
     * Maximum number of attributes including default attributes.
9086
     */
9087
0
    maxAtts = nratts + nbTotalDef;
9088
9089
    /*
9090
     * Verify that attribute names are unique.
9091
     */
9092
0
    if (maxAtts > 1) {
9093
0
        attrHashSize = 4;
9094
0
        while (attrHashSize / 2 < (unsigned) maxAtts)
9095
0
            attrHashSize *= 2;
9096
9097
0
        if (attrHashSize > ctxt->attrHashMax) {
9098
0
            xmlAttrHashBucket *tmp;
9099
9100
0
            tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0]));
9101
0
            if (tmp == NULL) {
9102
0
                xmlErrMemory(ctxt);
9103
0
                goto done;
9104
0
            }
9105
9106
0
            ctxt->attrHash = tmp;
9107
0
            ctxt->attrHashMax = attrHashSize;
9108
0
        }
9109
9110
0
        memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0]));
9111
9112
0
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9113
0
            const xmlChar *nsuri;
9114
0
            unsigned hashValue, nameHashValue, uriHashValue;
9115
0
            int res;
9116
9117
0
            attname = atts[i];
9118
0
            aprefix = atts[i+1];
9119
0
            nsIndex = XML_PTR_TO_INT(atts[i+2]);
9120
            /* Hash values always have bit 31 set, see dict.c */
9121
0
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9122
9123
0
            if (nsIndex == NS_INDEX_EMPTY) {
9124
                /*
9125
                 * Prefix with empty namespace means an undeclared
9126
                 * prefix which was already reported above.
9127
                 */
9128
0
                if (aprefix != NULL)
9129
0
                    continue;
9130
0
                nsuri = NULL;
9131
0
                uriHashValue = URI_HASH_EMPTY;
9132
0
            } else if (nsIndex == NS_INDEX_XML) {
9133
0
                nsuri = ctxt->str_xml_ns;
9134
0
                uriHashValue = URI_HASH_XML;
9135
0
            } else {
9136
0
                nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9137
0
                uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9138
0
            }
9139
9140
0
            hashValue = xmlDictCombineHash(nameHashValue, uriHashValue);
9141
0
            res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9142
0
                                    hashValue, i);
9143
0
            if (res < 0)
9144
0
                continue;
9145
9146
            /*
9147
             * [ WFC: Unique Att Spec ]
9148
             * No attribute name may appear more than once in the same
9149
             * start-tag or empty-element tag.
9150
             * As extended by the Namespace in XML REC.
9151
             */
9152
0
            if (res < INT_MAX) {
9153
0
                if (aprefix == atts[res+1]) {
9154
0
                    xmlErrAttributeDup(ctxt, aprefix, attname);
9155
0
                    numDupErr += 1;
9156
0
                } else {
9157
0
                    xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9158
0
                             "Namespaced Attribute %s in '%s' redefined\n",
9159
0
                             attname, nsuri, NULL);
9160
0
                    numNsErr += 1;
9161
0
                }
9162
0
            }
9163
0
        }
9164
0
    }
9165
9166
    /*
9167
     * Default attributes
9168
     */
9169
0
    if (ctxt->attsDefault != NULL) {
9170
0
        xmlDefAttrsPtr defaults;
9171
9172
0
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9173
0
  if (defaults != NULL) {
9174
0
      for (i = 0; i < defaults->nbAttrs; i++) {
9175
0
                xmlDefAttr *attr = &defaults->attrs[i];
9176
0
                const xmlChar *nsuri = NULL;
9177
0
                unsigned hashValue, uriHashValue = 0;
9178
0
                int res;
9179
9180
0
          attname = attr->name.name;
9181
0
    aprefix = attr->prefix.name;
9182
9183
0
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL))
9184
0
                    continue;
9185
0
    if (aprefix == ctxt->str_xmlns)
9186
0
                    continue;
9187
9188
0
                if (aprefix == NULL) {
9189
0
                    nsIndex = NS_INDEX_EMPTY;
9190
0
                    nsuri = NULL;
9191
0
                    uriHashValue = URI_HASH_EMPTY;
9192
0
                } else if (aprefix == ctxt->str_xml) {
9193
0
                    nsIndex = NS_INDEX_XML;
9194
0
                    nsuri = ctxt->str_xml_ns;
9195
0
                    uriHashValue = URI_HASH_XML;
9196
0
                } else {
9197
0
                    nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL);
9198
0
                    if ((nsIndex == INT_MAX) ||
9199
0
                        (nsIndex < ctxt->nsdb->minNsIndex)) {
9200
0
                        xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9201
0
                                 "Namespace prefix %s for %s on %s is not "
9202
0
                                 "defined\n",
9203
0
                                 aprefix, attname, localname);
9204
0
                        nsIndex = NS_INDEX_EMPTY;
9205
0
                        nsuri = NULL;
9206
0
                        uriHashValue = URI_HASH_EMPTY;
9207
0
                    } else {
9208
0
                        nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9209
0
                        uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9210
0
                    }
9211
0
                }
9212
9213
                /*
9214
                 * Check whether the attribute exists
9215
                 */
9216
0
                if (maxAtts > 1) {
9217
0
                    hashValue = xmlDictCombineHash(attr->name.hashValue,
9218
0
                                                   uriHashValue);
9219
0
                    res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9220
0
                                            hashValue, nbatts);
9221
0
                    if (res < 0)
9222
0
                        continue;
9223
0
                    if (res < INT_MAX) {
9224
0
                        if (aprefix == atts[res+1])
9225
0
                            continue;
9226
0
                        xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9227
0
                                 "Namespaced Attribute %s in '%s' redefined\n",
9228
0
                                 attname, nsuri, NULL);
9229
0
                    }
9230
0
                }
9231
9232
0
                xmlParserEntityCheck(ctxt, attr->expandedSize);
9233
9234
0
                if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9235
0
                    res = xmlCtxtGrowAttrs(ctxt);
9236
9237
0
                    maxatts = ctxt->maxatts;
9238
0
                    atts = ctxt->atts;
9239
9240
0
                    if (res < 0) {
9241
0
                        localname = NULL;
9242
0
                        goto done;
9243
0
                    }
9244
0
                }
9245
9246
0
                atts[nbatts++] = attname;
9247
0
                atts[nbatts++] = aprefix;
9248
0
                atts[nbatts++] = XML_INT_TO_PTR(nsIndex);
9249
0
                atts[nbatts++] = attr->value.name;
9250
0
                atts[nbatts++] = attr->valueEnd;
9251
9252
0
#ifdef LIBXML_VALID_ENABLED
9253
                /*
9254
                 * This should be moved to valid.c, but we don't keep track
9255
                 * whether an attribute was defaulted.
9256
                 */
9257
0
                if ((ctxt->validate) &&
9258
0
                    (ctxt->standalone == 1) &&
9259
0
                    (attr->external != 0)) {
9260
0
                    xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED,
9261
0
                            "standalone: attribute %s on %s defaulted "
9262
0
                            "from external subset\n",
9263
0
                            attname, localname);
9264
0
                }
9265
0
#endif
9266
0
                nbdef++;
9267
0
      }
9268
0
  }
9269
0
    }
9270
9271
    /*
9272
     * Using a single hash table for nsUri/localName pairs cannot
9273
     * detect duplicate QNames reliably. The following example will
9274
     * only result in two namespace errors.
9275
     *
9276
     * <doc xmlns:a="a" xmlns:b="a">
9277
     *   <elem a:a="" b:a="" b:a=""/>
9278
     * </doc>
9279
     *
9280
     * If we saw more than one namespace error but no duplicate QNames
9281
     * were found, we have to scan for duplicate QNames.
9282
     */
9283
0
    if ((numDupErr == 0) && (numNsErr > 1)) {
9284
0
        memset(ctxt->attrHash, -1,
9285
0
               attrHashSize * sizeof(ctxt->attrHash[0]));
9286
9287
0
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9288
0
            unsigned hashValue, nameHashValue, prefixHashValue;
9289
0
            int res;
9290
9291
0
            aprefix = atts[i+1];
9292
0
            if (aprefix == NULL)
9293
0
                continue;
9294
9295
0
            attname = atts[i];
9296
            /* Hash values always have bit 31 set, see dict.c */
9297
0
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9298
0
            prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix);
9299
9300
0
            hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue);
9301
0
            res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname,
9302
0
                                         aprefix, hashValue, i);
9303
0
            if (res < INT_MAX)
9304
0
                xmlErrAttributeDup(ctxt, aprefix, attname);
9305
0
        }
9306
0
    }
9307
9308
    /*
9309
     * Reconstruct attribute pointers
9310
     */
9311
0
    for (i = 0, j = 0; i < nbatts; i += 5, j++) {
9312
        /* namespace URI */
9313
0
        nsIndex = XML_PTR_TO_INT(atts[i+2]);
9314
0
        if (nsIndex == INT_MAX)
9315
0
            atts[i+2] = NULL;
9316
0
        else if (nsIndex == INT_MAX - 1)
9317
0
            atts[i+2] = ctxt->str_xml_ns;
9318
0
        else
9319
0
            atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1];
9320
9321
0
        if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) {
9322
0
            atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]);  /* value */
9323
0
            atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]);  /* valuend */
9324
0
        }
9325
0
    }
9326
9327
0
    uri = xmlParserNsLookupUri(ctxt, &hprefix);
9328
0
    if ((prefix != NULL) && (uri == NULL)) {
9329
0
  xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9330
0
           "Namespace prefix %s on %s is not defined\n",
9331
0
     prefix, localname, NULL);
9332
0
    }
9333
0
    *pref = prefix;
9334
0
    *URI = uri;
9335
9336
    /*
9337
     * SAX callback
9338
     */
9339
0
    if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) &&
9340
0
  (!ctxt->disableSAX)) {
9341
0
  if (nbNs > 0)
9342
0
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9343
0
                          nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs),
9344
0
        nbatts / 5, nbdef, atts);
9345
0
  else
9346
0
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9347
0
                          0, NULL, nbatts / 5, nbdef, atts);
9348
0
    }
9349
9350
0
done:
9351
    /*
9352
     * Free allocated attribute values
9353
     */
9354
0
    if (attval != 0) {
9355
0
  for (i = 0, j = 0; j < nratts; i += 5, j++)
9356
0
      if (ctxt->attallocs[j] & 0x80000000)
9357
0
          xmlFree((xmlChar *) atts[i+3]);
9358
0
    }
9359
9360
0
    *nbNsPtr = nbNs;
9361
0
    return(localname);
9362
0
}
9363
9364
/**
9365
 * Parse an end tag. Always consumes '</'.
9366
 *
9367
 *     [42] ETag ::= '</' Name S? '>'
9368
 *
9369
 * With namespace
9370
 *
9371
 *     [NS 9] ETag ::= '</' QName S? '>'
9372
 * @param ctxt  an XML parser context
9373
 * @param tag  the corresponding start tag
9374
 */
9375
9376
static void
9377
0
xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) {
9378
0
    const xmlChar *name;
9379
9380
0
    GROW;
9381
0
    if ((RAW != '<') || (NXT(1) != '/')) {
9382
0
  xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL);
9383
0
  return;
9384
0
    }
9385
0
    SKIP(2);
9386
9387
0
    if (tag->prefix == NULL)
9388
0
        name = xmlParseNameAndCompare(ctxt, ctxt->name);
9389
0
    else
9390
0
        name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix);
9391
9392
    /*
9393
     * We should definitely be at the ending "S? '>'" part
9394
     */
9395
0
    GROW;
9396
0
    SKIP_BLANKS;
9397
0
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
9398
0
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
9399
0
    } else
9400
0
  NEXT1;
9401
9402
    /*
9403
     * [ WFC: Element Type Match ]
9404
     * The Name in an element's end-tag must match the element type in the
9405
     * start-tag.
9406
     *
9407
     */
9408
0
    if (name != (xmlChar*)1) {
9409
0
        if (name == NULL) name = BAD_CAST "unparsable";
9410
0
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
9411
0
         "Opening and ending tag mismatch: %s line %d and %s\n",
9412
0
                    ctxt->name, tag->line, name);
9413
0
    }
9414
9415
    /*
9416
     * SAX: End of Tag
9417
     */
9418
0
    if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9419
0
  (!ctxt->disableSAX))
9420
0
  ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix,
9421
0
                                tag->URI);
9422
9423
0
    spacePop(ctxt);
9424
0
    if (tag->nsNr != 0)
9425
0
  xmlParserNsPop(ctxt, tag->nsNr);
9426
0
}
9427
9428
/**
9429
 * Parse escaped pure raw content. Always consumes '<!['.
9430
 *
9431
 * @deprecated Internal function, don't use.
9432
 *
9433
 *     [18] CDSect ::= CDStart CData CDEnd
9434
 *
9435
 *     [19] CDStart ::= '<![CDATA['
9436
 *
9437
 *     [20] Data ::= (Char* - (Char* ']]>' Char*))
9438
 *
9439
 *     [21] CDEnd ::= ']]>'
9440
 * @param ctxt  an XML parser context
9441
 */
9442
void
9443
281k
xmlParseCDSect(xmlParserCtxt *ctxt) {
9444
281k
    xmlChar *buf = NULL;
9445
281k
    int len = 0;
9446
281k
    int size = XML_PARSER_BUFFER_SIZE;
9447
281k
    int r, rl;
9448
281k
    int s, sl;
9449
281k
    int cur, l;
9450
281k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9451
0
                    XML_MAX_HUGE_LENGTH :
9452
281k
                    XML_MAX_TEXT_LENGTH;
9453
9454
281k
    if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '['))
9455
0
        return;
9456
281k
    SKIP(3);
9457
9458
281k
    if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '['))
9459
0
        return;
9460
281k
    SKIP(6);
9461
9462
281k
    r = xmlCurrentCharRecover(ctxt, &rl);
9463
281k
    if (!IS_CHAR(r)) {
9464
11
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9465
11
        goto out;
9466
11
    }
9467
281k
    NEXTL(rl);
9468
281k
    s = xmlCurrentCharRecover(ctxt, &sl);
9469
281k
    if (!IS_CHAR(s)) {
9470
13
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9471
13
        goto out;
9472
13
    }
9473
281k
    NEXTL(sl);
9474
281k
    cur = xmlCurrentCharRecover(ctxt, &l);
9475
281k
    buf = xmlMalloc(size);
9476
281k
    if (buf == NULL) {
9477
0
  xmlErrMemory(ctxt);
9478
0
        goto out;
9479
0
    }
9480
17.7M
    while (IS_CHAR(cur) &&
9481
17.7M
           ((r != ']') || (s != ']') || (cur != '>'))) {
9482
17.4M
  if (len + 5 >= size) {
9483
43.4k
      xmlChar *tmp;
9484
43.4k
            int newSize;
9485
9486
43.4k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9487
43.4k
            if (newSize < 0) {
9488
0
                xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9489
0
                               "CData section too big found\n");
9490
0
                goto out;
9491
0
            }
9492
43.4k
      tmp = xmlRealloc(buf, newSize);
9493
43.4k
      if (tmp == NULL) {
9494
0
    xmlErrMemory(ctxt);
9495
0
                goto out;
9496
0
      }
9497
43.4k
      buf = tmp;
9498
43.4k
      size = newSize;
9499
43.4k
  }
9500
17.4M
  COPY_BUF(buf, len, r);
9501
17.4M
  r = s;
9502
17.4M
  rl = sl;
9503
17.4M
  s = cur;
9504
17.4M
  sl = l;
9505
17.4M
  NEXTL(l);
9506
17.4M
  cur = xmlCurrentCharRecover(ctxt, &l);
9507
17.4M
    }
9508
281k
    buf[len] = 0;
9509
281k
    if (cur != '>') {
9510
111
  xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9511
111
                       "CData section not finished\n%.50s\n", buf);
9512
111
        goto out;
9513
111
    }
9514
281k
    NEXTL(l);
9515
9516
    /*
9517
     * OK the buffer is to be consumed as cdata.
9518
     */
9519
281k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
9520
281k
        if ((ctxt->sax->cdataBlock != NULL) &&
9521
281k
            ((ctxt->options & XML_PARSE_NOCDATA) == 0)) {
9522
281k
            ctxt->sax->cdataBlock(ctxt->userData, buf, len);
9523
281k
        } else if (ctxt->sax->characters != NULL) {
9524
0
            ctxt->sax->characters(ctxt->userData, buf, len);
9525
0
        }
9526
281k
    }
9527
9528
281k
out:
9529
281k
    xmlFree(buf);
9530
281k
}
9531
9532
/**
9533
 * Parse a content sequence. Stops at EOF or '</'. Leaves checking of
9534
 * unexpected EOF to the caller.
9535
 *
9536
 * @param ctxt  an XML parser context
9537
 */
9538
9539
static void
9540
621k
xmlParseContentInternal(xmlParserCtxtPtr ctxt) {
9541
621k
    int oldNameNr = ctxt->nameNr;
9542
621k
    int oldSpaceNr = ctxt->spaceNr;
9543
621k
    int oldNodeNr = ctxt->nodeNr;
9544
9545
621k
    GROW;
9546
2.31M
    while ((ctxt->input->cur < ctxt->input->end) &&
9547
1.69M
     (PARSER_STOPPED(ctxt) == 0)) {
9548
1.69M
  const xmlChar *cur = ctxt->input->cur;
9549
9550
  /*
9551
   * First case : a Processing Instruction.
9552
   */
9553
1.69M
  if ((*cur == '<') && (cur[1] == '?')) {
9554
12.6k
      xmlParsePI(ctxt);
9555
12.6k
  }
9556
9557
  /*
9558
   * Second case : a CDSection
9559
   */
9560
  /* 2.6.0 test was *cur not RAW */
9561
1.68M
  else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) {
9562
6.21k
      xmlParseCDSect(ctxt);
9563
6.21k
  }
9564
9565
  /*
9566
   * Third case :  a comment
9567
   */
9568
1.67M
  else if ((*cur == '<') && (NXT(1) == '!') &&
9569
41.9k
     (NXT(2) == '-') && (NXT(3) == '-')) {
9570
41.8k
      xmlParseComment(ctxt);
9571
41.8k
  }
9572
9573
  /*
9574
   * Fourth case :  a sub-element.
9575
   */
9576
1.63M
  else if (*cur == '<') {
9577
279k
            if (NXT(1) == '/') {
9578
11.4k
                if (ctxt->nameNr <= oldNameNr)
9579
6
                    break;
9580
11.4k
          xmlParseElementEnd(ctxt);
9581
268k
            } else {
9582
268k
          xmlParseElementStart(ctxt);
9583
268k
            }
9584
279k
  }
9585
9586
  /*
9587
   * Fifth case : a reference. If if has not been resolved,
9588
   *    parsing returns it's Name, create the node
9589
   */
9590
9591
1.35M
  else if (*cur == '&') {
9592
640k
      xmlParseReference(ctxt);
9593
640k
  }
9594
9595
  /*
9596
   * Last case, text. Note that References are handled directly.
9597
   */
9598
712k
  else {
9599
712k
      xmlParseCharDataInternal(ctxt, 0);
9600
712k
  }
9601
9602
1.69M
  SHRINK;
9603
1.69M
  GROW;
9604
1.69M
    }
9605
9606
621k
    if ((ctxt->nameNr > oldNameNr) &&
9607
364
        (ctxt->input->cur >= ctxt->input->end) &&
9608
138
        (ctxt->wellFormed)) {
9609
82
        const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9610
82
        int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9611
82
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9612
82
                "Premature end of data in tag %s line %d\n",
9613
82
                name, line, NULL);
9614
82
    }
9615
9616
    /*
9617
     * Clean up in error case
9618
     */
9619
9620
621k
    while (ctxt->nodeNr > oldNodeNr)
9621
0
        nodePop(ctxt);
9622
9623
635k
    while (ctxt->nameNr > oldNameNr) {
9624
13.4k
        xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1];
9625
9626
13.4k
        if (tag->nsNr != 0)
9627
0
            xmlParserNsPop(ctxt, tag->nsNr);
9628
9629
13.4k
        namePop(ctxt);
9630
13.4k
    }
9631
9632
635k
    while (ctxt->spaceNr > oldSpaceNr)
9633
13.4k
        spacePop(ctxt);
9634
621k
}
9635
9636
/**
9637
 * Parse XML element content. This is useful if you're only interested
9638
 * in custom SAX callbacks. If you want a node list, use
9639
 * #xmlCtxtParseContent.
9640
 *
9641
 * @param ctxt  an XML parser context
9642
 */
9643
void
9644
0
xmlParseContent(xmlParserCtxt *ctxt) {
9645
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
9646
0
        return;
9647
9648
0
    xmlCtxtInitializeLate(ctxt);
9649
9650
0
    xmlParseContentInternal(ctxt);
9651
9652
0
    xmlParserCheckEOF(ctxt, XML_ERR_NOT_WELL_BALANCED);
9653
0
}
9654
9655
/**
9656
 * Parse an XML element
9657
 *
9658
 * @deprecated Internal function, don't use.
9659
 *
9660
 *     [39] element ::= EmptyElemTag | STag content ETag
9661
 *
9662
 * [ WFC: Element Type Match ]
9663
 * The Name in an element's end-tag must match the element type in the
9664
 * start-tag.
9665
 *
9666
 * @param ctxt  an XML parser context
9667
 */
9668
9669
void
9670
0
xmlParseElement(xmlParserCtxt *ctxt) {
9671
0
    if (xmlParseElementStart(ctxt) != 0)
9672
0
        return;
9673
9674
0
    xmlParseContentInternal(ctxt);
9675
9676
0
    if (ctxt->input->cur >= ctxt->input->end) {
9677
0
        if (ctxt->wellFormed) {
9678
0
            const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9679
0
            int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9680
0
            xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9681
0
                    "Premature end of data in tag %s line %d\n",
9682
0
                    name, line, NULL);
9683
0
        }
9684
0
        return;
9685
0
    }
9686
9687
0
    xmlParseElementEnd(ctxt);
9688
0
}
9689
9690
/**
9691
 * Parse the start of an XML element. Returns -1 in case of error, 0 if an
9692
 * opening tag was parsed, 1 if an empty element was parsed.
9693
 *
9694
 * Always consumes '<'.
9695
 *
9696
 * @param ctxt  an XML parser context
9697
 */
9698
static int
9699
268k
xmlParseElementStart(xmlParserCtxtPtr ctxt) {
9700
268k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
9701
268k
    const xmlChar *name;
9702
268k
    const xmlChar *prefix = NULL;
9703
268k
    const xmlChar *URI = NULL;
9704
268k
    xmlParserNodeInfo node_info;
9705
268k
    int line;
9706
268k
    xmlNodePtr cur;
9707
268k
    int nbNs = 0;
9708
9709
268k
    if (ctxt->nameNr > maxDepth) {
9710
33
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
9711
33
                "Excessive depth in document: %d use XML_PARSE_HUGE option\n",
9712
33
                ctxt->nameNr);
9713
33
  return(-1);
9714
33
    }
9715
9716
    /* Capture start position */
9717
268k
    if (ctxt->record_info) {
9718
0
        node_info.begin_pos = ctxt->input->consumed +
9719
0
                          (CUR_PTR - ctxt->input->base);
9720
0
  node_info.begin_line = ctxt->input->line;
9721
0
    }
9722
9723
268k
    if (ctxt->spaceNr == 0)
9724
0
  spacePush(ctxt, -1);
9725
268k
    else if (*ctxt->space == -2)
9726
251k
  spacePush(ctxt, -1);
9727
16.5k
    else
9728
16.5k
  spacePush(ctxt, *ctxt->space);
9729
9730
268k
    line = ctxt->input->line;
9731
268k
#ifdef LIBXML_SAX1_ENABLED
9732
268k
    if (ctxt->sax2)
9733
0
#endif /* LIBXML_SAX1_ENABLED */
9734
0
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
9735
268k
#ifdef LIBXML_SAX1_ENABLED
9736
268k
    else
9737
268k
  name = xmlParseStartTag(ctxt);
9738
268k
#endif /* LIBXML_SAX1_ENABLED */
9739
268k
    if (name == NULL) {
9740
129
  spacePop(ctxt);
9741
129
        return(-1);
9742
129
    }
9743
268k
    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
9744
268k
    cur = ctxt->node;
9745
9746
268k
#ifdef LIBXML_VALID_ENABLED
9747
    /*
9748
     * [ VC: Root Element Type ]
9749
     * The Name in the document type declaration must match the element
9750
     * type of the root element.
9751
     */
9752
268k
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
9753
0
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
9754
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
9755
268k
#endif /* LIBXML_VALID_ENABLED */
9756
9757
    /*
9758
     * Check for an Empty Element.
9759
     */
9760
268k
    if ((RAW == '/') && (NXT(1) == '>')) {
9761
242k
        SKIP(2);
9762
242k
  if (ctxt->sax2) {
9763
0
      if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9764
0
    (!ctxt->disableSAX))
9765
0
    ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI);
9766
0
#ifdef LIBXML_SAX1_ENABLED
9767
242k
  } else {
9768
242k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
9769
242k
    (!ctxt->disableSAX))
9770
242k
    ctxt->sax->endElement(ctxt->userData, name);
9771
242k
#endif /* LIBXML_SAX1_ENABLED */
9772
242k
  }
9773
242k
  namePop(ctxt);
9774
242k
  spacePop(ctxt);
9775
242k
  if (nbNs > 0)
9776
0
      xmlParserNsPop(ctxt, nbNs);
9777
242k
  if (cur != NULL && ctxt->record_info) {
9778
0
            node_info.node = cur;
9779
0
            node_info.end_pos = ctxt->input->consumed +
9780
0
                                (CUR_PTR - ctxt->input->base);
9781
0
            node_info.end_line = ctxt->input->line;
9782
0
            xmlParserAddNodeInfo(ctxt, &node_info);
9783
0
  }
9784
242k
  return(1);
9785
242k
    }
9786
25.2k
    if (RAW == '>') {
9787
24.9k
        NEXT1;
9788
24.9k
        if (cur != NULL && ctxt->record_info) {
9789
0
            node_info.node = cur;
9790
0
            node_info.end_pos = 0;
9791
0
            node_info.end_line = 0;
9792
0
            xmlParserAddNodeInfo(ctxt, &node_info);
9793
0
        }
9794
24.9k
    } else {
9795
310
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED,
9796
310
         "Couldn't find end of Start Tag %s line %d\n",
9797
310
                    name, line, NULL);
9798
9799
  /*
9800
   * end of parsing of this node.
9801
   */
9802
310
  nodePop(ctxt);
9803
310
  namePop(ctxt);
9804
310
  spacePop(ctxt);
9805
310
  if (nbNs > 0)
9806
0
      xmlParserNsPop(ctxt, nbNs);
9807
310
  return(-1);
9808
310
    }
9809
9810
24.9k
    return(0);
9811
25.2k
}
9812
9813
/**
9814
 * Parse the end of an XML element. Always consumes '</'.
9815
 *
9816
 * @param ctxt  an XML parser context
9817
 */
9818
static void
9819
11.4k
xmlParseElementEnd(xmlParserCtxtPtr ctxt) {
9820
11.4k
    xmlNodePtr cur = ctxt->node;
9821
9822
11.4k
    if (ctxt->nameNr <= 0) {
9823
0
        if ((RAW == '<') && (NXT(1) == '/'))
9824
0
            SKIP(2);
9825
0
        return;
9826
0
    }
9827
9828
    /*
9829
     * parse the end of tag: '</' should be here.
9830
     */
9831
11.4k
    if (ctxt->sax2) {
9832
0
  xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
9833
0
  namePop(ctxt);
9834
0
    }
9835
11.4k
#ifdef LIBXML_SAX1_ENABLED
9836
11.4k
    else
9837
11.4k
  xmlParseEndTag1(ctxt, 0);
9838
11.4k
#endif /* LIBXML_SAX1_ENABLED */
9839
9840
    /*
9841
     * Capture end position
9842
     */
9843
11.4k
    if (cur != NULL && ctxt->record_info) {
9844
0
        xmlParserNodeInfoPtr node_info;
9845
9846
0
        node_info = (xmlParserNodeInfoPtr) xmlParserFindNodeInfo(ctxt, cur);
9847
0
        if (node_info != NULL) {
9848
0
            node_info->end_pos = ctxt->input->consumed +
9849
0
                                 (CUR_PTR - ctxt->input->base);
9850
0
            node_info->end_line = ctxt->input->line;
9851
0
        }
9852
0
    }
9853
11.4k
}
9854
9855
/**
9856
 * Parse the XML version value.
9857
 *
9858
 * @deprecated Internal function, don't use.
9859
 *
9860
 *     [26] VersionNum ::= '1.' [0-9]+
9861
 *
9862
 * In practice allow [0-9].[0-9]+ at that level
9863
 *
9864
 * @param ctxt  an XML parser context
9865
 * @returns the string giving the XML version number, or NULL
9866
 */
9867
xmlChar *
9868
31.2k
xmlParseVersionNum(xmlParserCtxt *ctxt) {
9869
31.2k
    xmlChar *buf = NULL;
9870
31.2k
    int len = 0;
9871
31.2k
    int size = 10;
9872
31.2k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9873
0
                    XML_MAX_TEXT_LENGTH :
9874
31.2k
                    XML_MAX_NAME_LENGTH;
9875
31.2k
    xmlChar cur;
9876
9877
31.2k
    buf = xmlMalloc(size);
9878
31.2k
    if (buf == NULL) {
9879
0
  xmlErrMemory(ctxt);
9880
0
  return(NULL);
9881
0
    }
9882
31.2k
    cur = CUR;
9883
31.2k
    if (!((cur >= '0') && (cur <= '9'))) {
9884
469
  xmlFree(buf);
9885
469
  return(NULL);
9886
469
    }
9887
30.8k
    buf[len++] = cur;
9888
30.8k
    NEXT;
9889
30.8k
    cur=CUR;
9890
30.8k
    if (cur != '.') {
9891
139
  xmlFree(buf);
9892
139
  return(NULL);
9893
139
    }
9894
30.6k
    buf[len++] = cur;
9895
30.6k
    NEXT;
9896
30.6k
    cur=CUR;
9897
401k
    while ((cur >= '0') && (cur <= '9')) {
9898
370k
  if (len + 1 >= size) {
9899
3.09k
      xmlChar *tmp;
9900
3.09k
            int newSize;
9901
9902
3.09k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9903
3.09k
            if (newSize < 0) {
9904
4
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum");
9905
4
                xmlFree(buf);
9906
4
                return(NULL);
9907
4
            }
9908
3.08k
      tmp = xmlRealloc(buf, newSize);
9909
3.08k
      if (tmp == NULL) {
9910
0
    xmlErrMemory(ctxt);
9911
0
          xmlFree(buf);
9912
0
    return(NULL);
9913
0
      }
9914
3.08k
      buf = tmp;
9915
3.08k
            size = newSize;
9916
3.08k
  }
9917
370k
  buf[len++] = cur;
9918
370k
  NEXT;
9919
370k
  cur=CUR;
9920
370k
    }
9921
30.6k
    buf[len] = 0;
9922
30.6k
    return(buf);
9923
30.6k
}
9924
9925
/**
9926
 * Parse the XML version.
9927
 *
9928
 * @deprecated Internal function, don't use.
9929
 *
9930
 *     [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ")
9931
 *
9932
 *     [25] Eq ::= S? '=' S?
9933
 *
9934
 * @param ctxt  an XML parser context
9935
 * @returns the version string, e.g. "1.0"
9936
 */
9937
9938
xmlChar *
9939
44.6k
xmlParseVersionInfo(xmlParserCtxt *ctxt) {
9940
44.6k
    xmlChar *version = NULL;
9941
9942
44.6k
    if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) {
9943
32.2k
  SKIP(7);
9944
32.2k
  SKIP_BLANKS;
9945
32.2k
  if (RAW != '=') {
9946
761
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
9947
761
      return(NULL);
9948
761
        }
9949
31.5k
  NEXT;
9950
31.5k
  SKIP_BLANKS;
9951
31.5k
  if (RAW == '"') {
9952
27.1k
      NEXT;
9953
27.1k
      version = xmlParseVersionNum(ctxt);
9954
27.1k
      if (RAW != '"') {
9955
4.64k
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9956
4.64k
      } else
9957
22.4k
          NEXT;
9958
27.1k
  } else if (RAW == '\''){
9959
4.17k
      NEXT;
9960
4.17k
      version = xmlParseVersionNum(ctxt);
9961
4.17k
      if (RAW != '\'') {
9962
46
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9963
46
      } else
9964
4.13k
          NEXT;
9965
4.17k
  } else {
9966
231
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
9967
231
  }
9968
31.5k
    }
9969
43.8k
    return(version);
9970
44.6k
}
9971
9972
/**
9973
 * Parse the XML encoding name
9974
 *
9975
 * @deprecated Internal function, don't use.
9976
 *
9977
 *     [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')*
9978
 *
9979
 * @param ctxt  an XML parser context
9980
 * @returns the encoding name value or NULL
9981
 */
9982
xmlChar *
9983
20.4k
xmlParseEncName(xmlParserCtxt *ctxt) {
9984
20.4k
    xmlChar *buf = NULL;
9985
20.4k
    int len = 0;
9986
20.4k
    int size = 10;
9987
20.4k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9988
0
                    XML_MAX_TEXT_LENGTH :
9989
20.4k
                    XML_MAX_NAME_LENGTH;
9990
20.4k
    xmlChar cur;
9991
9992
20.4k
    cur = CUR;
9993
20.4k
    if (((cur >= 'a') && (cur <= 'z')) ||
9994
19.3k
        ((cur >= 'A') && (cur <= 'Z'))) {
9995
19.3k
  buf = xmlMalloc(size);
9996
19.3k
  if (buf == NULL) {
9997
0
      xmlErrMemory(ctxt);
9998
0
      return(NULL);
9999
0
  }
10000
10001
19.3k
  buf[len++] = cur;
10002
19.3k
  NEXT;
10003
19.3k
  cur = CUR;
10004
512k
  while (((cur >= 'a') && (cur <= 'z')) ||
10005
180k
         ((cur >= 'A') && (cur <= 'Z')) ||
10006
115k
         ((cur >= '0') && (cur <= '9')) ||
10007
61.8k
         (cur == '.') || (cur == '_') ||
10008
492k
         (cur == '-')) {
10009
492k
      if (len + 1 >= size) {
10010
6.64k
          xmlChar *tmp;
10011
6.64k
                int newSize;
10012
10013
6.64k
                newSize = xmlGrowCapacity(size, 1, 1, maxLength);
10014
6.64k
                if (newSize < 0) {
10015
4
                    xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName");
10016
4
                    xmlFree(buf);
10017
4
                    return(NULL);
10018
4
                }
10019
6.64k
    tmp = xmlRealloc(buf, newSize);
10020
6.64k
    if (tmp == NULL) {
10021
0
        xmlErrMemory(ctxt);
10022
0
        xmlFree(buf);
10023
0
        return(NULL);
10024
0
    }
10025
6.64k
    buf = tmp;
10026
6.64k
                size = newSize;
10027
6.64k
      }
10028
492k
      buf[len++] = cur;
10029
492k
      NEXT;
10030
492k
      cur = CUR;
10031
492k
        }
10032
19.3k
  buf[len] = 0;
10033
19.3k
    } else {
10034
1.06k
  xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL);
10035
1.06k
    }
10036
20.4k
    return(buf);
10037
20.4k
}
10038
10039
/**
10040
 * Parse the XML encoding declaration
10041
 *
10042
 * @deprecated Internal function, don't use.
10043
 *
10044
 *     [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | 
10045
 *                           "'" EncName "'")
10046
 *
10047
 * this setups the conversion filters.
10048
 *
10049
 * @param ctxt  an XML parser context
10050
 * @returns the encoding value or NULL
10051
 */
10052
10053
const xmlChar *
10054
41.8k
xmlParseEncodingDecl(xmlParserCtxt *ctxt) {
10055
41.8k
    xmlChar *encoding = NULL;
10056
10057
41.8k
    SKIP_BLANKS;
10058
41.8k
    if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0)
10059
19.8k
        return(NULL);
10060
10061
21.9k
    SKIP(8);
10062
21.9k
    SKIP_BLANKS;
10063
21.9k
    if (RAW != '=') {
10064
1.32k
        xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10065
1.32k
        return(NULL);
10066
1.32k
    }
10067
20.6k
    NEXT;
10068
20.6k
    SKIP_BLANKS;
10069
20.6k
    if (RAW == '"') {
10070
16.2k
        NEXT;
10071
16.2k
        encoding = xmlParseEncName(ctxt);
10072
16.2k
        if (RAW != '"') {
10073
1.25k
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10074
1.25k
            xmlFree(encoding);
10075
1.25k
            return(NULL);
10076
1.25k
        } else
10077
15.0k
            NEXT;
10078
16.2k
    } else if (RAW == '\''){
10079
4.13k
        NEXT;
10080
4.13k
        encoding = xmlParseEncName(ctxt);
10081
4.13k
        if (RAW != '\'') {
10082
42
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10083
42
            xmlFree(encoding);
10084
42
            return(NULL);
10085
42
        } else
10086
4.09k
            NEXT;
10087
4.13k
    } else {
10088
235
        xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10089
235
    }
10090
10091
19.3k
    if (encoding == NULL)
10092
238
        return(NULL);
10093
10094
19.1k
    xmlSetDeclaredEncoding(ctxt, encoding);
10095
10096
19.1k
    return(ctxt->encoding);
10097
19.3k
}
10098
10099
/**
10100
 * Parse the XML standalone declaration
10101
 *
10102
 * @deprecated Internal function, don't use.
10103
 *
10104
 *     [32] SDDecl ::= S 'standalone' Eq
10105
 *                     (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"'))
10106
 *
10107
 * [ VC: Standalone Document Declaration ]
10108
 * TODO The standalone document declaration must have the value "no"
10109
 * if any external markup declarations contain declarations of:
10110
 *  - attributes with default values, if elements to which these
10111
 *    attributes apply appear in the document without specifications
10112
 *    of values for these attributes, or
10113
 *  - entities (other than amp, lt, gt, apos, quot), if references
10114
 *    to those entities appear in the document, or
10115
 *  - attributes with values subject to normalization, where the
10116
 *    attribute appears in the document with a value which will change
10117
 *    as a result of normalization, or
10118
 *  - element types with element content, if white space occurs directly
10119
 *    within any instance of those types.
10120
 *
10121
 * @param ctxt  an XML parser context
10122
 * @returns
10123
 *   1 if standalone="yes"
10124
 *   0 if standalone="no"
10125
 *  -2 if standalone attribute is missing or invalid
10126
 *    (A standalone value of -2 means that the XML declaration was found,
10127
 *     but no value was specified for the standalone attribute).
10128
 */
10129
10130
int
10131
29.7k
xmlParseSDDecl(xmlParserCtxt *ctxt) {
10132
29.7k
    int standalone = -2;
10133
10134
29.7k
    SKIP_BLANKS;
10135
29.7k
    if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) {
10136
796
  SKIP(10);
10137
796
        SKIP_BLANKS;
10138
796
  if (RAW != '=') {
10139
7
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10140
7
      return(standalone);
10141
7
        }
10142
789
  NEXT;
10143
789
  SKIP_BLANKS;
10144
789
        if (RAW == '\''){
10145
73
      NEXT;
10146
73
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10147
49
          standalone = 0;
10148
49
                SKIP(2);
10149
49
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10150
9
                 (NXT(2) == 's')) {
10151
5
          standalone = 1;
10152
5
    SKIP(3);
10153
19
            } else {
10154
19
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10155
19
      }
10156
73
      if (RAW != '\'') {
10157
24
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10158
24
      } else
10159
49
          NEXT;
10160
716
  } else if (RAW == '"'){
10161
711
      NEXT;
10162
711
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10163
475
          standalone = 0;
10164
475
    SKIP(2);
10165
475
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10166
218
                 (NXT(2) == 's')) {
10167
214
          standalone = 1;
10168
214
                SKIP(3);
10169
214
            } else {
10170
22
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10171
22
      }
10172
711
      if (RAW != '"') {
10173
28
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10174
28
      } else
10175
683
          NEXT;
10176
711
  } else {
10177
5
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10178
5
        }
10179
789
    }
10180
29.7k
    return(standalone);
10181
29.7k
}
10182
10183
/**
10184
 * Parse an XML declaration header
10185
 *
10186
 * @deprecated Internal function, don't use.
10187
 *
10188
 *     [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
10189
 * @param ctxt  an XML parser context
10190
 */
10191
10192
void
10193
44.6k
xmlParseXMLDecl(xmlParserCtxt *ctxt) {
10194
44.6k
    xmlChar *version;
10195
10196
    /*
10197
     * This value for standalone indicates that the document has an
10198
     * XML declaration but it does not have a standalone attribute.
10199
     * It will be overwritten later if a standalone attribute is found.
10200
     */
10201
10202
44.6k
    ctxt->standalone = -2;
10203
10204
    /*
10205
     * We know that '<?xml' is here.
10206
     */
10207
44.6k
    SKIP(5);
10208
10209
44.6k
    if (!IS_BLANK_CH(RAW)) {
10210
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
10211
0
                 "Blank needed after '<?xml'\n");
10212
0
    }
10213
44.6k
    SKIP_BLANKS;
10214
10215
    /*
10216
     * We must have the VersionInfo here.
10217
     */
10218
44.6k
    version = xmlParseVersionInfo(ctxt);
10219
44.6k
    if (version == NULL) {
10220
13.9k
  xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL);
10221
30.6k
    } else {
10222
30.6k
  if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) {
10223
      /*
10224
       * Changed here for XML-1.0 5th edition
10225
       */
10226
18.6k
      if (ctxt->options & XML_PARSE_OLD10) {
10227
0
    xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10228
0
                "Unsupported version '%s'\n",
10229
0
                version);
10230
18.6k
      } else {
10231
18.6k
          if ((version[0] == '1') && ((version[1] == '.'))) {
10232
12.4k
        xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION,
10233
12.4k
                      "Unsupported version '%s'\n",
10234
12.4k
          version, NULL);
10235
12.4k
    } else {
10236
6.16k
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10237
6.16k
              "Unsupported version '%s'\n",
10238
6.16k
              version);
10239
6.16k
    }
10240
18.6k
      }
10241
18.6k
  }
10242
30.6k
  if (ctxt->version != NULL)
10243
0
      xmlFree(ctxt->version);
10244
30.6k
  ctxt->version = version;
10245
30.6k
    }
10246
10247
    /*
10248
     * We may have the encoding declaration
10249
     */
10250
44.6k
    if (!IS_BLANK_CH(RAW)) {
10251
20.8k
        if ((RAW == '?') && (NXT(1) == '>')) {
10252
2.79k
      SKIP(2);
10253
2.79k
      return;
10254
2.79k
  }
10255
18.0k
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10256
18.0k
    }
10257
41.8k
    xmlParseEncodingDecl(ctxt);
10258
10259
    /*
10260
     * We may have the standalone status.
10261
     */
10262
41.8k
    if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) {
10263
12.8k
        if ((RAW == '?') && (NXT(1) == '>')) {
10264
12.1k
      SKIP(2);
10265
12.1k
      return;
10266
12.1k
  }
10267
747
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10268
747
    }
10269
10270
    /*
10271
     * We can grow the input buffer freely at that point
10272
     */
10273
29.7k
    GROW;
10274
10275
29.7k
    SKIP_BLANKS;
10276
29.7k
    ctxt->standalone = xmlParseSDDecl(ctxt);
10277
10278
29.7k
    SKIP_BLANKS;
10279
29.7k
    if ((RAW == '?') && (NXT(1) == '>')) {
10280
5.54k
        SKIP(2);
10281
24.1k
    } else if (RAW == '>') {
10282
        /* Deprecated old WD ... */
10283
31
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10284
31
  NEXT;
10285
24.1k
    } else {
10286
24.1k
        int c;
10287
10288
24.1k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10289
24.1k
        while ((PARSER_STOPPED(ctxt) == 0) &&
10290
0
               ((c = CUR) != 0)) {
10291
0
            NEXT;
10292
0
            if (c == '>')
10293
0
                break;
10294
0
        }
10295
24.1k
    }
10296
29.7k
}
10297
10298
/**
10299
 * @since 2.14.0
10300
 *
10301
 * @param ctxt  parser context
10302
 * @returns the version from the XML declaration.
10303
 */
10304
const xmlChar *
10305
0
xmlCtxtGetVersion(xmlParserCtxt *ctxt) {
10306
0
    if (ctxt == NULL)
10307
0
        return(NULL);
10308
10309
0
    return(ctxt->version);
10310
0
}
10311
10312
/**
10313
 * @since 2.14.0
10314
 *
10315
 * @param ctxt  parser context
10316
 * @returns the value from the standalone document declaration.
10317
 */
10318
int
10319
0
xmlCtxtGetStandalone(xmlParserCtxt *ctxt) {
10320
0
    if (ctxt == NULL)
10321
0
        return(0);
10322
10323
0
    return(ctxt->standalone);
10324
0
}
10325
10326
/**
10327
 * Parse an XML Misc* optional field.
10328
 *
10329
 * @deprecated Internal function, don't use.
10330
 *
10331
 *     [27] Misc ::= Comment | PI |  S
10332
 * @param ctxt  an XML parser context
10333
 */
10334
10335
void
10336
0
xmlParseMisc(xmlParserCtxt *ctxt) {
10337
0
    while (PARSER_STOPPED(ctxt) == 0) {
10338
0
        SKIP_BLANKS;
10339
0
        GROW;
10340
0
        if ((RAW == '<') && (NXT(1) == '?')) {
10341
0
      xmlParsePI(ctxt);
10342
0
        } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) {
10343
0
      xmlParseComment(ctxt);
10344
0
        } else {
10345
0
            break;
10346
0
        }
10347
0
    }
10348
0
}
10349
10350
static void
10351
13.1k
xmlFinishDocument(xmlParserCtxtPtr ctxt) {
10352
13.1k
    xmlDocPtr doc;
10353
10354
    /*
10355
     * SAX: end of the document processing.
10356
     */
10357
13.1k
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10358
13.1k
        ctxt->sax->endDocument(ctxt->userData);
10359
10360
    /*
10361
     * Remove locally kept entity definitions if the tree was not built
10362
     */
10363
13.1k
    doc = ctxt->myDoc;
10364
13.1k
    if ((doc != NULL) &&
10365
13.1k
        (xmlStrEqual(doc->version, SAX_COMPAT_MODE))) {
10366
0
        xmlFreeDoc(doc);
10367
0
        ctxt->myDoc = NULL;
10368
0
    }
10369
13.1k
}
10370
10371
/**
10372
 * Parse an XML document and invoke the SAX handlers. This is useful
10373
 * if you're only interested in custom SAX callbacks. If you want a
10374
 * document tree, use #xmlCtxtParseDocument.
10375
 *
10376
 * @param ctxt  an XML parser context
10377
 * @returns 0, -1 in case of error.
10378
 */
10379
10380
int
10381
0
xmlParseDocument(xmlParserCtxt *ctxt) {
10382
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
10383
0
        return(-1);
10384
10385
0
    GROW;
10386
10387
    /*
10388
     * SAX: detecting the level.
10389
     */
10390
0
    xmlCtxtInitializeLate(ctxt);
10391
10392
0
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10393
0
        ctxt->sax->setDocumentLocator(ctxt->userData,
10394
0
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10395
0
    }
10396
10397
0
    xmlDetectEncoding(ctxt);
10398
10399
0
    if (CUR == 0) {
10400
0
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10401
0
  return(-1);
10402
0
    }
10403
10404
0
    GROW;
10405
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
10406
10407
  /*
10408
   * Note that we will switch encoding on the fly.
10409
   */
10410
0
  xmlParseXMLDecl(ctxt);
10411
0
  SKIP_BLANKS;
10412
0
    } else {
10413
0
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10414
0
        if (ctxt->version == NULL) {
10415
0
            xmlErrMemory(ctxt);
10416
0
            return(-1);
10417
0
        }
10418
0
    }
10419
0
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10420
0
        ctxt->sax->startDocument(ctxt->userData);
10421
0
    if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) &&
10422
0
        (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) {
10423
0
  ctxt->myDoc->compression = ctxt->input->buf->compressed;
10424
0
    }
10425
10426
    /*
10427
     * The Misc part of the Prolog
10428
     */
10429
0
    xmlParseMisc(ctxt);
10430
10431
    /*
10432
     * Then possibly doc type declaration(s) and more Misc
10433
     * (doctypedecl Misc*)?
10434
     */
10435
0
    GROW;
10436
0
    if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) {
10437
10438
0
  ctxt->inSubset = 1;
10439
0
  xmlParseDocTypeDecl(ctxt);
10440
0
  if (RAW == '[') {
10441
0
      xmlParseInternalSubset(ctxt);
10442
0
  } else if (RAW == '>') {
10443
0
            NEXT;
10444
0
        }
10445
10446
  /*
10447
   * Create and update the external subset.
10448
   */
10449
0
  ctxt->inSubset = 2;
10450
0
  if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) &&
10451
0
      (!ctxt->disableSAX))
10452
0
      ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
10453
0
                                ctxt->extSubSystem, ctxt->extSubURI);
10454
0
  ctxt->inSubset = 0;
10455
10456
0
        xmlCleanSpecialAttr(ctxt);
10457
10458
0
  xmlParseMisc(ctxt);
10459
0
    }
10460
10461
    /*
10462
     * Time to start parsing the tree itself
10463
     */
10464
0
    GROW;
10465
0
    if (RAW != '<') {
10466
0
        if (ctxt->wellFormed)
10467
0
            xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10468
0
                           "Start tag expected, '<' not found\n");
10469
0
    } else {
10470
0
  xmlParseElement(ctxt);
10471
10472
  /*
10473
   * The Misc part at the end
10474
   */
10475
0
  xmlParseMisc(ctxt);
10476
10477
0
        xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
10478
0
    }
10479
10480
0
    ctxt->instate = XML_PARSER_EOF;
10481
0
    xmlFinishDocument(ctxt);
10482
10483
0
    if (! ctxt->wellFormed) {
10484
0
  ctxt->valid = 0;
10485
0
  return(-1);
10486
0
    }
10487
10488
0
    return(0);
10489
0
}
10490
10491
/**
10492
 * Parse a general parsed entity
10493
 * An external general parsed entity is well-formed if it matches the
10494
 * production labeled extParsedEnt.
10495
 *
10496
 * @deprecated Internal function, don't use.
10497
 *
10498
 *     [78] extParsedEnt ::= TextDecl? content
10499
 *
10500
 * @param ctxt  an XML parser context
10501
 * @returns 0, -1 in case of error. the parser context is augmented
10502
 *                as a result of the parsing.
10503
 */
10504
10505
int
10506
0
xmlParseExtParsedEnt(xmlParserCtxt *ctxt) {
10507
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
10508
0
        return(-1);
10509
10510
0
    xmlCtxtInitializeLate(ctxt);
10511
10512
0
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10513
0
        ctxt->sax->setDocumentLocator(ctxt->userData,
10514
0
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10515
0
    }
10516
10517
0
    xmlDetectEncoding(ctxt);
10518
10519
0
    if (CUR == 0) {
10520
0
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10521
0
    }
10522
10523
    /*
10524
     * Check for the XMLDecl in the Prolog.
10525
     */
10526
0
    GROW;
10527
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
10528
10529
  /*
10530
   * Note that we will switch encoding on the fly.
10531
   */
10532
0
  xmlParseXMLDecl(ctxt);
10533
0
  SKIP_BLANKS;
10534
0
    } else {
10535
0
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10536
0
    }
10537
0
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10538
0
        ctxt->sax->startDocument(ctxt->userData);
10539
10540
    /*
10541
     * Doing validity checking on chunk doesn't make sense
10542
     */
10543
0
    ctxt->options &= ~XML_PARSE_DTDVALID;
10544
0
    ctxt->validate = 0;
10545
0
    ctxt->depth = 0;
10546
10547
0
    xmlParseContentInternal(ctxt);
10548
10549
0
    if (ctxt->input->cur < ctxt->input->end)
10550
0
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
10551
10552
    /*
10553
     * SAX: end of the document processing.
10554
     */
10555
0
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10556
0
        ctxt->sax->endDocument(ctxt->userData);
10557
10558
0
    if (! ctxt->wellFormed) return(-1);
10559
0
    return(0);
10560
0
}
10561
10562
#ifdef LIBXML_PUSH_ENABLED
10563
/************************************************************************
10564
 *                  *
10565
 *    Progressive parsing interfaces        *
10566
 *                  *
10567
 ************************************************************************/
10568
10569
/**
10570
 * Check whether the input buffer contains a character.
10571
 *
10572
 * @param ctxt  an XML parser context
10573
 * @param c  character
10574
 */
10575
static int
10576
286k
xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) {
10577
286k
    const xmlChar *cur;
10578
10579
286k
    if (ctxt->checkIndex == 0) {
10580
280k
        cur = ctxt->input->cur + 1;
10581
280k
    } else {
10582
6.40k
        cur = ctxt->input->cur + ctxt->checkIndex;
10583
6.40k
    }
10584
10585
286k
    if (memchr(cur, c, ctxt->input->end - cur) == NULL) {
10586
9.97k
        size_t index = ctxt->input->end - ctxt->input->cur;
10587
10588
9.97k
        if (index > LONG_MAX) {
10589
0
            ctxt->checkIndex = 0;
10590
0
            return(1);
10591
0
        }
10592
9.97k
        ctxt->checkIndex = index;
10593
9.97k
        return(0);
10594
276k
    } else {
10595
276k
        ctxt->checkIndex = 0;
10596
276k
        return(1);
10597
276k
    }
10598
286k
}
10599
10600
/**
10601
 * Check whether the input buffer contains a string.
10602
 *
10603
 * @param ctxt  an XML parser context
10604
 * @param startDelta  delta to apply at the start
10605
 * @param str  string
10606
 * @param strLen  length of string
10607
 */
10608
static const xmlChar *
10609
xmlParseLookupString(xmlParserCtxtPtr ctxt, size_t startDelta,
10610
1.11M
                     const char *str, size_t strLen) {
10611
1.11M
    const xmlChar *cur, *term;
10612
10613
1.11M
    if (ctxt->checkIndex == 0) {
10614
1.07M
        cur = ctxt->input->cur + startDelta;
10615
1.07M
    } else {
10616
44.3k
        cur = ctxt->input->cur + ctxt->checkIndex;
10617
44.3k
    }
10618
10619
1.11M
    term = BAD_CAST strstr((const char *) cur, str);
10620
1.11M
    if (term == NULL) {
10621
84.0k
        const xmlChar *end = ctxt->input->end;
10622
84.0k
        size_t index;
10623
10624
        /* Rescan (strLen - 1) characters. */
10625
84.0k
        if ((size_t) (end - cur) < strLen)
10626
1.48k
            end = cur;
10627
82.5k
        else
10628
82.5k
            end -= strLen - 1;
10629
84.0k
        index = end - ctxt->input->cur;
10630
84.0k
        if (index > LONG_MAX) {
10631
0
            ctxt->checkIndex = 0;
10632
0
            return(ctxt->input->end - strLen);
10633
0
        }
10634
84.0k
        ctxt->checkIndex = index;
10635
1.03M
    } else {
10636
1.03M
        ctxt->checkIndex = 0;
10637
1.03M
    }
10638
10639
1.11M
    return(term);
10640
1.11M
}
10641
10642
/**
10643
 * Check whether the input buffer contains terminated char data.
10644
 *
10645
 * @param ctxt  an XML parser context
10646
 */
10647
static int
10648
231k
xmlParseLookupCharData(xmlParserCtxtPtr ctxt) {
10649
231k
    const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex;
10650
231k
    const xmlChar *end = ctxt->input->end;
10651
231k
    size_t index;
10652
10653
4.23M
    while (cur < end) {
10654
4.21M
        if ((*cur == '<') || (*cur == '&')) {
10655
208k
            ctxt->checkIndex = 0;
10656
208k
            return(1);
10657
208k
        }
10658
4.00M
        cur++;
10659
4.00M
    }
10660
10661
22.8k
    index = cur - ctxt->input->cur;
10662
22.8k
    if (index > LONG_MAX) {
10663
0
        ctxt->checkIndex = 0;
10664
0
        return(1);
10665
0
    }
10666
22.8k
    ctxt->checkIndex = index;
10667
22.8k
    return(0);
10668
22.8k
}
10669
10670
/**
10671
 * Check whether there's enough data in the input buffer to finish parsing
10672
 * a start tag. This has to take quotes into account.
10673
 *
10674
 * @param ctxt  an XML parser context
10675
 */
10676
static int
10677
2.07M
xmlParseLookupGt(xmlParserCtxtPtr ctxt) {
10678
2.07M
    const xmlChar *cur;
10679
2.07M
    const xmlChar *end = ctxt->input->end;
10680
2.07M
    int state = ctxt->endCheckState;
10681
2.07M
    size_t index;
10682
10683
2.07M
    if (ctxt->checkIndex == 0)
10684
1.97M
        cur = ctxt->input->cur + 1;
10685
101k
    else
10686
101k
        cur = ctxt->input->cur + ctxt->checkIndex;
10687
10688
453M
    while (cur < end) {
10689
453M
        if (state) {
10690
413M
            if (*cur == state)
10691
583k
                state = 0;
10692
413M
        } else if (*cur == '\'' || *cur == '"') {
10693
590k
            state = *cur;
10694
39.3M
        } else if (*cur == '>') {
10695
1.95M
            ctxt->checkIndex = 0;
10696
1.95M
            ctxt->endCheckState = 0;
10697
1.95M
            return(1);
10698
1.95M
        }
10699
451M
        cur++;
10700
451M
    }
10701
10702
121k
    index = cur - ctxt->input->cur;
10703
121k
    if (index > LONG_MAX) {
10704
0
        ctxt->checkIndex = 0;
10705
0
        ctxt->endCheckState = 0;
10706
0
        return(1);
10707
0
    }
10708
121k
    ctxt->checkIndex = index;
10709
121k
    ctxt->endCheckState = state;
10710
121k
    return(0);
10711
121k
}
10712
10713
/**
10714
 * Check whether there's enough data in the input buffer to finish parsing
10715
 * the internal subset.
10716
 *
10717
 * @param ctxt  an XML parser context
10718
 */
10719
static int
10720
46.7k
xmlParseLookupInternalSubset(xmlParserCtxtPtr ctxt) {
10721
    /*
10722
     * Sorry, but progressive parsing of the internal subset is not
10723
     * supported. We first check that the full content of the internal
10724
     * subset is available and parsing is launched only at that point.
10725
     * Internal subset ends with "']' S? '>'" in an unescaped section and
10726
     * not in a ']]>' sequence which are conditional sections.
10727
     */
10728
46.7k
    const xmlChar *cur, *start;
10729
46.7k
    const xmlChar *end = ctxt->input->end;
10730
46.7k
    int state = ctxt->endCheckState;
10731
46.7k
    size_t index;
10732
10733
46.7k
    if (ctxt->checkIndex == 0) {
10734
14.4k
        cur = ctxt->input->cur + 1;
10735
32.2k
    } else {
10736
32.2k
        cur = ctxt->input->cur + ctxt->checkIndex;
10737
32.2k
    }
10738
46.7k
    start = cur;
10739
10740
301M
    while (cur < end) {
10741
301M
        if (state == '-') {
10742
3.87M
            if ((*cur == '-') &&
10743
207k
                (cur[1] == '-') &&
10744
127k
                (cur[2] == '>')) {
10745
116k
                state = 0;
10746
116k
                cur += 3;
10747
116k
                start = cur;
10748
116k
                continue;
10749
116k
            }
10750
3.87M
        }
10751
297M
        else if (state == ']') {
10752
45.4k
            if (*cur == '>') {
10753
12.7k
                ctxt->checkIndex = 0;
10754
12.7k
                ctxt->endCheckState = 0;
10755
12.7k
                return(1);
10756
12.7k
            }
10757
32.7k
            if (IS_BLANK_CH(*cur)) {
10758
9.73k
                state = ' ';
10759
22.9k
            } else if (*cur != ']') {
10760
5.89k
                state = 0;
10761
5.89k
                start = cur;
10762
5.89k
                continue;
10763
5.89k
            }
10764
32.7k
        }
10765
297M
        else if (state == ' ') {
10766
39.8k
            if (*cur == '>') {
10767
385
                ctxt->checkIndex = 0;
10768
385
                ctxt->endCheckState = 0;
10769
385
                return(1);
10770
385
            }
10771
39.4k
            if (!IS_BLANK_CH(*cur)) {
10772
9.33k
                state = 0;
10773
9.33k
                start = cur;
10774
9.33k
                continue;
10775
9.33k
            }
10776
39.4k
        }
10777
297M
        else if (state != 0) {
10778
277M
            if (*cur == state) {
10779
188k
                state = 0;
10780
188k
                start = cur + 1;
10781
188k
            }
10782
277M
        }
10783
20.3M
        else if (*cur == '<') {
10784
222k
            if ((cur[1] == '!') &&
10785
184k
                (cur[2] == '-') &&
10786
117k
                (cur[3] == '-')) {
10787
116k
                state = '-';
10788
116k
                cur += 4;
10789
                /* Don't treat <!--> as comment */
10790
116k
                start = cur;
10791
116k
                continue;
10792
116k
            }
10793
222k
        }
10794
20.1M
        else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) {
10795
217k
            state = *cur;
10796
217k
        }
10797
10798
301M
        cur++;
10799
301M
    }
10800
10801
    /*
10802
     * Rescan the three last characters to detect "<!--" and "-->"
10803
     * split across chunks.
10804
     */
10805
33.6k
    if ((state == 0) || (state == '-')) {
10806
7.39k
        if (cur - start < 3)
10807
703
            cur = start;
10808
6.69k
        else
10809
6.69k
            cur -= 3;
10810
7.39k
    }
10811
33.6k
    index = cur - ctxt->input->cur;
10812
33.6k
    if (index > LONG_MAX) {
10813
0
        ctxt->checkIndex = 0;
10814
0
        ctxt->endCheckState = 0;
10815
0
        return(1);
10816
0
    }
10817
33.6k
    ctxt->checkIndex = index;
10818
33.6k
    ctxt->endCheckState = state;
10819
33.6k
    return(0);
10820
33.6k
}
10821
10822
/**
10823
 * Try to progress on parsing
10824
 *
10825
 * @param ctxt  an XML parser context
10826
 * @param terminate  last chunk indicator
10827
 * @returns zero if no parsing was possible
10828
 */
10829
static int
10830
395k
xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
10831
395k
    int ret = 0;
10832
395k
    size_t avail;
10833
395k
    xmlChar cur, next;
10834
10835
395k
    if (ctxt->input == NULL)
10836
0
        return(0);
10837
10838
395k
    if ((ctxt->input != NULL) &&
10839
395k
        (ctxt->input->cur - ctxt->input->base > 4096)) {
10840
24.3k
        xmlParserShrink(ctxt);
10841
24.3k
    }
10842
10843
7.29M
    while (ctxt->disableSAX == 0) {
10844
7.20M
        avail = ctxt->input->end - ctxt->input->cur;
10845
7.20M
        if (avail < 1)
10846
22.8k
      goto done;
10847
7.18M
        switch (ctxt->instate) {
10848
0
            case XML_PARSER_EOF:
10849
          /*
10850
     * Document parsing is done !
10851
     */
10852
0
          goto done;
10853
122k
            case XML_PARSER_START:
10854
                /*
10855
                 * Very first chars read from the document flow.
10856
                 */
10857
122k
                if ((!terminate) && (avail < 4))
10858
30
                    goto done;
10859
10860
                /*
10861
                 * We need more bytes to detect EBCDIC code pages.
10862
                 * See xmlDetectEBCDIC.
10863
                 */
10864
122k
                if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) &&
10865
16
                    (!terminate) && (avail < 200))
10866
2
                    goto done;
10867
10868
122k
                xmlDetectEncoding(ctxt);
10869
122k
                ctxt->instate = XML_PARSER_XML_DECL;
10870
122k
    break;
10871
10872
173k
            case XML_PARSER_XML_DECL:
10873
173k
    if ((!terminate) && (avail < 2))
10874
32
        goto done;
10875
173k
    cur = ctxt->input->cur[0];
10876
173k
    next = ctxt->input->cur[1];
10877
173k
          if ((cur == '<') && (next == '?')) {
10878
        /* PI or XML decl */
10879
141k
        if ((!terminate) &&
10880
103k
                        (!xmlParseLookupString(ctxt, 2, "?>", 2)))
10881
51.4k
      goto done;
10882
89.6k
        if ((ctxt->input->cur[2] == 'x') &&
10883
53.5k
      (ctxt->input->cur[3] == 'm') &&
10884
53.0k
      (ctxt->input->cur[4] == 'l') &&
10885
52.7k
      (IS_BLANK_CH(ctxt->input->cur[5]))) {
10886
44.6k
      ret += 5;
10887
44.6k
      xmlParseXMLDecl(ctxt);
10888
45.0k
        } else {
10889
45.0k
      ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10890
45.0k
                        if (ctxt->version == NULL) {
10891
0
                            xmlErrMemory(ctxt);
10892
0
                            break;
10893
0
                        }
10894
45.0k
        }
10895
89.6k
    } else {
10896
31.8k
        ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10897
31.8k
        if (ctxt->version == NULL) {
10898
0
            xmlErrMemory(ctxt);
10899
0
      break;
10900
0
        }
10901
31.8k
    }
10902
121k
                if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10903
121k
                    ctxt->sax->setDocumentLocator(ctxt->userData,
10904
121k
                            (xmlSAXLocator *) &xmlDefaultSAXLocator);
10905
121k
                }
10906
121k
                if ((ctxt->sax) && (ctxt->sax->startDocument) &&
10907
121k
                    (!ctxt->disableSAX))
10908
91.6k
                    ctxt->sax->startDocument(ctxt->userData);
10909
121k
                ctxt->instate = XML_PARSER_MISC;
10910
121k
    break;
10911
2.03M
            case XML_PARSER_START_TAG: {
10912
2.03M
          const xmlChar *name;
10913
2.03M
    const xmlChar *prefix = NULL;
10914
2.03M
    const xmlChar *URI = NULL;
10915
2.03M
                int line = ctxt->input->line;
10916
2.03M
    int nbNs = 0;
10917
10918
2.03M
    if ((!terminate) && (avail < 2))
10919
273
        goto done;
10920
2.03M
    cur = ctxt->input->cur[0];
10921
2.03M
          if (cur != '<') {
10922
5.13k
        xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10923
5.13k
                                   "Start tag expected, '<' not found");
10924
5.13k
                    ctxt->instate = XML_PARSER_EOF;
10925
5.13k
                    xmlFinishDocument(ctxt);
10926
5.13k
        goto done;
10927
5.13k
    }
10928
2.02M
    if ((!terminate) && (!xmlParseLookupGt(ctxt)))
10929
80.2k
                    goto done;
10930
1.94M
    if (ctxt->spaceNr == 0)
10931
0
        spacePush(ctxt, -1);
10932
1.94M
    else if (*ctxt->space == -2)
10933
1.11M
        spacePush(ctxt, -1);
10934
832k
    else
10935
832k
        spacePush(ctxt, *ctxt->space);
10936
1.94M
#ifdef LIBXML_SAX1_ENABLED
10937
1.94M
    if (ctxt->sax2)
10938
0
#endif /* LIBXML_SAX1_ENABLED */
10939
0
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
10940
1.94M
#ifdef LIBXML_SAX1_ENABLED
10941
1.94M
    else
10942
1.94M
        name = xmlParseStartTag(ctxt);
10943
1.94M
#endif /* LIBXML_SAX1_ENABLED */
10944
1.94M
    if (name == NULL) {
10945
4.78k
        spacePop(ctxt);
10946
4.78k
                    ctxt->instate = XML_PARSER_EOF;
10947
4.78k
                    xmlFinishDocument(ctxt);
10948
4.78k
        goto done;
10949
4.78k
    }
10950
1.94M
#ifdef LIBXML_VALID_ENABLED
10951
    /*
10952
     * [ VC: Root Element Type ]
10953
     * The Name in the document type declaration must match
10954
     * the element type of the root element.
10955
     */
10956
1.94M
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
10957
0
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
10958
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
10959
1.94M
#endif /* LIBXML_VALID_ENABLED */
10960
10961
    /*
10962
     * Check for an Empty Element.
10963
     */
10964
1.94M
    if ((RAW == '/') && (NXT(1) == '>')) {
10965
547k
        SKIP(2);
10966
10967
547k
        if (ctxt->sax2) {
10968
0
      if ((ctxt->sax != NULL) &&
10969
0
          (ctxt->sax->endElementNs != NULL) &&
10970
0
          (!ctxt->disableSAX))
10971
0
          ctxt->sax->endElementNs(ctxt->userData, name,
10972
0
                                  prefix, URI);
10973
0
      if (nbNs > 0)
10974
0
          xmlParserNsPop(ctxt, nbNs);
10975
0
#ifdef LIBXML_SAX1_ENABLED
10976
547k
        } else {
10977
547k
      if ((ctxt->sax != NULL) &&
10978
547k
          (ctxt->sax->endElement != NULL) &&
10979
547k
          (!ctxt->disableSAX))
10980
547k
          ctxt->sax->endElement(ctxt->userData, name);
10981
547k
#endif /* LIBXML_SAX1_ENABLED */
10982
547k
        }
10983
547k
        spacePop(ctxt);
10984
1.39M
    } else if (RAW == '>') {
10985
1.37M
        NEXT;
10986
1.37M
                    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
10987
1.37M
    } else {
10988
15.3k
        xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED,
10989
15.3k
           "Couldn't find end of Start Tag %s\n",
10990
15.3k
           name);
10991
15.3k
        nodePop(ctxt);
10992
15.3k
        spacePop(ctxt);
10993
15.3k
                    if (nbNs > 0)
10994
0
                        xmlParserNsPop(ctxt, nbNs);
10995
15.3k
    }
10996
10997
1.94M
                if (ctxt->nameNr == 0)
10998
7.81k
                    ctxt->instate = XML_PARSER_EPILOG;
10999
1.93M
                else
11000
1.93M
                    ctxt->instate = XML_PARSER_CONTENT;
11001
1.94M
                break;
11002
1.94M
      }
11003
4.07M
            case XML_PARSER_CONTENT: {
11004
4.07M
    cur = ctxt->input->cur[0];
11005
11006
4.07M
    if (cur == '<') {
11007
2.77M
                    if ((!terminate) && (avail < 2))
11008
2.20k
                        goto done;
11009
2.77M
        next = ctxt->input->cur[1];
11010
11011
2.77M
                    if (next == '/') {
11012
223k
                        ctxt->instate = XML_PARSER_END_TAG;
11013
223k
                        break;
11014
2.54M
                    } else if (next == '?') {
11015
128k
                        if ((!terminate) &&
11016
128k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11017
3.73k
                            goto done;
11018
125k
                        xmlParsePI(ctxt);
11019
125k
                        ctxt->instate = XML_PARSER_CONTENT;
11020
125k
                        break;
11021
2.41M
                    } else if (next == '!') {
11022
525k
                        if ((!terminate) && (avail < 3))
11023
471
                            goto done;
11024
524k
                        next = ctxt->input->cur[2];
11025
11026
524k
                        if (next == '-') {
11027
236k
                            if ((!terminate) && (avail < 4))
11028
187
                                goto done;
11029
235k
                            if (ctxt->input->cur[3] == '-') {
11030
235k
                                if ((!terminate) &&
11031
235k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11032
4.17k
                                    goto done;
11033
231k
                                xmlParseComment(ctxt);
11034
231k
                                ctxt->instate = XML_PARSER_CONTENT;
11035
231k
                                break;
11036
235k
                            }
11037
288k
                        } else if (next == '[') {
11038
288k
                            if ((!terminate) && (avail < 9))
11039
663
                                goto done;
11040
287k
                            if ((ctxt->input->cur[2] == '[') &&
11041
287k
                                (ctxt->input->cur[3] == 'C') &&
11042
287k
                                (ctxt->input->cur[4] == 'D') &&
11043
287k
                                (ctxt->input->cur[5] == 'A') &&
11044
287k
                                (ctxt->input->cur[6] == 'T') &&
11045
287k
                                (ctxt->input->cur[7] == 'A') &&
11046
287k
                                (ctxt->input->cur[8] == '[')) {
11047
287k
                                if ((!terminate) &&
11048
287k
                                    (!xmlParseLookupString(ctxt, 9, "]]>", 3)))
11049
12.0k
                                    goto done;
11050
275k
                                ctxt->instate = XML_PARSER_CDATA_SECTION;
11051
275k
                                xmlParseCDSect(ctxt);
11052
275k
                                ctxt->instate = XML_PARSER_CONTENT;
11053
275k
                                break;
11054
287k
                            }
11055
287k
                        }
11056
524k
                    }
11057
2.77M
    } else if (cur == '&') {
11058
58.7k
        if ((!terminate) && (!xmlParseLookupChar(ctxt, ';')))
11059
3.31k
      goto done;
11060
55.4k
        xmlParseReference(ctxt);
11061
55.4k
                    break;
11062
1.24M
    } else {
11063
        /* TODO Avoid the extra copy, handle directly !!! */
11064
        /*
11065
         * Goal of the following test is:
11066
         *  - minimize calls to the SAX 'character' callback
11067
         *    when they are mergeable
11068
         *  - handle an problem for isBlank when we only parse
11069
         *    a sequence of blank chars and the next one is
11070
         *    not available to check against '<' presence.
11071
         *  - tries to homogenize the differences in SAX
11072
         *    callbacks between the push and pull versions
11073
         *    of the parser.
11074
         */
11075
1.24M
        if (avail < XML_PARSER_BIG_BUFFER_SIZE) {
11076
235k
      if ((!terminate) && (!xmlParseLookupCharData(ctxt)))
11077
22.8k
          goto done;
11078
235k
                    }
11079
1.21M
                    ctxt->checkIndex = 0;
11080
1.21M
        xmlParseCharDataInternal(ctxt, !terminate);
11081
1.21M
                    break;
11082
1.24M
    }
11083
11084
1.89M
                ctxt->instate = XML_PARSER_START_TAG;
11085
1.89M
    break;
11086
4.07M
      }
11087
230k
            case XML_PARSER_END_TAG:
11088
230k
    if ((!terminate) && (!xmlParseLookupChar(ctxt, '>')))
11089
6.66k
        goto done;
11090
223k
    if (ctxt->sax2) {
11091
0
              xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
11092
0
        nameNsPop(ctxt);
11093
0
    }
11094
223k
#ifdef LIBXML_SAX1_ENABLED
11095
223k
      else
11096
223k
        xmlParseEndTag1(ctxt, 0);
11097
223k
#endif /* LIBXML_SAX1_ENABLED */
11098
223k
    if (ctxt->nameNr == 0) {
11099
2.78k
        ctxt->instate = XML_PARSER_EPILOG;
11100
220k
    } else {
11101
220k
        ctxt->instate = XML_PARSER_CONTENT;
11102
220k
    }
11103
223k
    break;
11104
485k
            case XML_PARSER_MISC:
11105
500k
            case XML_PARSER_PROLOG:
11106
502k
            case XML_PARSER_EPILOG:
11107
502k
    SKIP_BLANKS;
11108
502k
                avail = ctxt->input->end - ctxt->input->cur;
11109
502k
    if (avail < 1)
11110
822
        goto done;
11111
501k
    if (ctxt->input->cur[0] == '<') {
11112
496k
                    if ((!terminate) && (avail < 2))
11113
614
                        goto done;
11114
495k
                    next = ctxt->input->cur[1];
11115
495k
                    if (next == '?') {
11116
253k
                        if ((!terminate) &&
11117
237k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11118
6.11k
                            goto done;
11119
246k
                        xmlParsePI(ctxt);
11120
246k
                        break;
11121
253k
                    } else if (next == '!') {
11122
185k
                        if ((!terminate) && (avail < 3))
11123
138
                            goto done;
11124
11125
185k
                        if (ctxt->input->cur[2] == '-') {
11126
124k
                            if ((!terminate) && (avail < 4))
11127
158
                                goto done;
11128
124k
                            if (ctxt->input->cur[3] == '-') {
11129
124k
                                if ((!terminate) &&
11130
124k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11131
6.46k
                                    goto done;
11132
118k
                                xmlParseComment(ctxt);
11133
118k
                                break;
11134
124k
                            }
11135
124k
                        } else if (ctxt->instate == XML_PARSER_MISC) {
11136
60.7k
                            if ((!terminate) && (avail < 9))
11137
73
                                goto done;
11138
60.7k
                            if ((ctxt->input->cur[2] == 'D') &&
11139
59.7k
                                (ctxt->input->cur[3] == 'O') &&
11140
59.6k
                                (ctxt->input->cur[4] == 'C') &&
11141
59.5k
                                (ctxt->input->cur[5] == 'T') &&
11142
59.5k
                                (ctxt->input->cur[6] == 'Y') &&
11143
59.4k
                                (ctxt->input->cur[7] == 'P') &&
11144
59.4k
                                (ctxt->input->cur[8] == 'E')) {
11145
59.3k
                                if ((!terminate) && (!xmlParseLookupGt(ctxt)))
11146
41.5k
                                    goto done;
11147
17.7k
                                ctxt->inSubset = 1;
11148
17.7k
                                xmlParseDocTypeDecl(ctxt);
11149
17.7k
                                if (RAW == '[') {
11150
14.5k
                                    ctxt->instate = XML_PARSER_DTD;
11151
14.5k
                                } else {
11152
3.19k
                                    if (RAW == '>')
11153
2.19k
                                        NEXT;
11154
                                    /*
11155
                                     * Create and update the external subset.
11156
                                     */
11157
3.19k
                                    ctxt->inSubset = 2;
11158
3.19k
                                    if ((ctxt->sax != NULL) &&
11159
3.19k
                                        (!ctxt->disableSAX) &&
11160
2.09k
                                        (ctxt->sax->externalSubset != NULL))
11161
2.09k
                                        ctxt->sax->externalSubset(
11162
2.09k
                                                ctxt->userData,
11163
2.09k
                                                ctxt->intSubName,
11164
2.09k
                                                ctxt->extSubSystem,
11165
2.09k
                                                ctxt->extSubURI);
11166
3.19k
                                    ctxt->inSubset = 0;
11167
3.19k
                                    xmlCleanSpecialAttr(ctxt);
11168
3.19k
                                    ctxt->instate = XML_PARSER_PROLOG;
11169
3.19k
                                }
11170
17.7k
                                break;
11171
59.3k
                            }
11172
60.7k
                        }
11173
185k
                    }
11174
495k
                }
11175
11176
63.8k
                if (ctxt->instate == XML_PARSER_EPILOG) {
11177
323
                    if (ctxt->errNo == XML_ERR_OK)
11178
323
                        xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL);
11179
323
        ctxt->instate = XML_PARSER_EOF;
11180
323
                    xmlFinishDocument(ctxt);
11181
63.5k
                } else {
11182
63.5k
        ctxt->instate = XML_PARSER_START_TAG;
11183
63.5k
    }
11184
63.8k
    break;
11185
47.4k
            case XML_PARSER_DTD: {
11186
47.4k
                if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt)))
11187
33.6k
                    goto done;
11188
13.8k
    xmlParseInternalSubset(ctxt);
11189
13.8k
    ctxt->inSubset = 2;
11190
13.8k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
11191
6.79k
        (ctxt->sax->externalSubset != NULL))
11192
6.79k
        ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
11193
6.79k
          ctxt->extSubSystem, ctxt->extSubURI);
11194
13.8k
    ctxt->inSubset = 0;
11195
13.8k
    xmlCleanSpecialAttr(ctxt);
11196
13.8k
    ctxt->instate = XML_PARSER_PROLOG;
11197
13.8k
                break;
11198
47.4k
      }
11199
0
            default:
11200
0
                xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
11201
0
      "PP: internal error\n");
11202
0
    ctxt->instate = XML_PARSER_EOF;
11203
0
    break;
11204
7.18M
  }
11205
7.18M
    }
11206
395k
done:
11207
395k
    return(ret);
11208
395k
}
11209
11210
/**
11211
 * Parse a chunk of memory in push parser mode.
11212
 *
11213
 * Assumes that the parser context was initialized with
11214
 * #xmlCreatePushParserCtxt.
11215
 *
11216
 * The last chunk, which will often be empty, must be marked with
11217
 * the `terminate` flag. With the default SAX callbacks, the resulting
11218
 * document will be available in ctxt->myDoc. This pointer will not
11219
 * be freed when calling #xmlFreeParserCtxt and must be freed by the
11220
 * caller. If the document isn't well-formed, it will still be returned
11221
 * in ctxt->myDoc.
11222
 *
11223
 * As an exception, #xmlCtxtResetPush will free the document in
11224
 * ctxt->myDoc. So ctxt->myDoc should be set to NULL after extracting
11225
 * the document.
11226
 *
11227
 * Since 2.14.0, #xmlCtxtGetDocument can be used to retrieve the
11228
 * result document.
11229
 *
11230
 * @param ctxt  an XML parser context
11231
 * @param chunk  chunk of memory
11232
 * @param size  size of chunk in bytes
11233
 * @param terminate  last chunk indicator
11234
 * @returns an xmlParserErrors code (0 on success).
11235
 */
11236
int
11237
xmlParseChunk(xmlParserCtxt *ctxt, const char *chunk, int size,
11238
433k
              int terminate) {
11239
433k
    size_t curBase;
11240
433k
    size_t maxLength;
11241
433k
    size_t pos;
11242
433k
    int end_in_lf = 0;
11243
433k
    int res;
11244
11245
433k
    if ((ctxt == NULL) || (size < 0))
11246
0
        return(XML_ERR_ARGUMENT);
11247
433k
    if ((chunk == NULL) && (size > 0))
11248
0
        return(XML_ERR_ARGUMENT);
11249
433k
    if ((ctxt->input == NULL) || (ctxt->input->buf == NULL))
11250
0
        return(XML_ERR_ARGUMENT);
11251
433k
    if (ctxt->disableSAX != 0)
11252
37.8k
        return(ctxt->errNo);
11253
11254
395k
    ctxt->input->flags |= XML_INPUT_PROGRESSIVE;
11255
395k
    if (ctxt->instate == XML_PARSER_START)
11256
122k
        xmlCtxtInitializeLate(ctxt);
11257
395k
    if ((size > 0) && (chunk != NULL) && (!terminate) &&
11258
335k
        (chunk[size - 1] == '\r')) {
11259
5.75k
  end_in_lf = 1;
11260
5.75k
  size--;
11261
5.75k
    }
11262
11263
    /*
11264
     * Also push an empty chunk to make sure that the raw buffer
11265
     * will be flushed if there is an encoder.
11266
     */
11267
395k
    pos = ctxt->input->cur - ctxt->input->base;
11268
395k
    res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
11269
395k
    xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11270
395k
    if (res < 0) {
11271
74
        xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11272
74
        return(ctxt->errNo);
11273
74
    }
11274
11275
395k
    xmlParseTryOrFinish(ctxt, terminate);
11276
11277
395k
    curBase = ctxt->input->cur - ctxt->input->base;
11278
395k
    maxLength = (ctxt->options & XML_PARSE_HUGE) ?
11279
0
                XML_MAX_HUGE_LENGTH :
11280
395k
                XML_MAX_LOOKUP_LIMIT;
11281
395k
    if (curBase > maxLength) {
11282
0
        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
11283
0
                    "Buffer size limit exceeded, try XML_PARSE_HUGE\n");
11284
0
    }
11285
11286
395k
    if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0))
11287
94.9k
        return(ctxt->errNo);
11288
11289
300k
    if (end_in_lf == 1) {
11290
5.58k
  pos = ctxt->input->cur - ctxt->input->base;
11291
5.58k
  res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r");
11292
5.58k
  xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11293
5.58k
        if (res < 0) {
11294
4
            xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11295
4
            return(ctxt->errNo);
11296
4
        }
11297
5.58k
    }
11298
300k
    if (terminate) {
11299
  /*
11300
   * Check for termination
11301
   */
11302
2.87k
        if ((ctxt->instate != XML_PARSER_EOF) &&
11303
2.87k
            (ctxt->instate != XML_PARSER_EPILOG)) {
11304
2.79k
            if (ctxt->nameNr > 0) {
11305
2.64k
                const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
11306
2.64k
                int line = ctxt->pushTab[ctxt->nameNr - 1].line;
11307
2.64k
                xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
11308
2.64k
                        "Premature end of data in tag %s line %d\n",
11309
2.64k
                        name, line, NULL);
11310
2.64k
            } else if (ctxt->instate == XML_PARSER_START) {
11311
0
                xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
11312
156
            } else {
11313
156
                xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
11314
156
                               "Start tag expected, '<' not found\n");
11315
156
            }
11316
2.79k
        } else {
11317
73
            xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
11318
73
        }
11319
2.87k
  if (ctxt->instate != XML_PARSER_EOF) {
11320
2.87k
            ctxt->instate = XML_PARSER_EOF;
11321
2.87k
            xmlFinishDocument(ctxt);
11322
2.87k
  }
11323
2.87k
    }
11324
300k
    if (ctxt->wellFormed == 0)
11325
2.79k
  return((xmlParserErrors) ctxt->errNo);
11326
298k
    else
11327
298k
        return(0);
11328
300k
}
11329
11330
/************************************************************************
11331
 *                  *
11332
 *    I/O front end functions to the parser     *
11333
 *                  *
11334
 ************************************************************************/
11335
11336
/**
11337
 * Create a parser context for using the XML parser in push mode.
11338
 * See #xmlParseChunk.
11339
 *
11340
 * Passing an initial chunk is useless and deprecated.
11341
 *
11342
 * The push parser doesn't support recovery mode or the
11343
 * XML_PARSE_NOBLANKS option.
11344
 *
11345
 * `filename` is used as base URI to fetch external entities and for
11346
 * error reports.
11347
 *
11348
 * @param sax  a SAX handler (optional)
11349
 * @param user_data  user data for SAX callbacks (optional)
11350
 * @param chunk  initial chunk (optional, deprecated)
11351
 * @param size  size of initial chunk in bytes
11352
 * @param filename  file name or URI (optional)
11353
 * @returns the new parser context or NULL if a memory allocation
11354
 * failed.
11355
 */
11356
11357
xmlParserCtxt *
11358
xmlCreatePushParserCtxt(xmlSAXHandler *sax, void *user_data,
11359
122k
                        const char *chunk, int size, const char *filename) {
11360
122k
    xmlParserCtxtPtr ctxt;
11361
122k
    xmlParserInputPtr input;
11362
11363
122k
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11364
122k
    if (ctxt == NULL)
11365
0
  return(NULL);
11366
11367
122k
    ctxt->options &= ~XML_PARSE_NODICT;
11368
122k
    ctxt->dictNames = 1;
11369
11370
122k
    input = xmlNewPushInput(filename, chunk, size);
11371
122k
    if (input == NULL) {
11372
0
  xmlFreeParserCtxt(ctxt);
11373
0
  return(NULL);
11374
0
    }
11375
122k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11376
0
        xmlFreeInputStream(input);
11377
0
        xmlFreeParserCtxt(ctxt);
11378
0
        return(NULL);
11379
0
    }
11380
11381
122k
    return(ctxt);
11382
122k
}
11383
#endif /* LIBXML_PUSH_ENABLED */
11384
11385
/**
11386
 * Blocks further parser processing
11387
 *
11388
 * @param ctxt  an XML parser context
11389
 */
11390
void
11391
97.8k
xmlStopParser(xmlParserCtxt *ctxt) {
11392
97.8k
    if (ctxt == NULL)
11393
0
        return;
11394
11395
    /* This stops the parser */
11396
97.8k
    ctxt->disableSAX = 2;
11397
11398
    /*
11399
     * xmlStopParser is often called from error handlers,
11400
     * so we can't raise an error here to avoid infinite
11401
     * loops. Just make sure that an error condition is
11402
     * reported.
11403
     */
11404
97.8k
    if (ctxt->errNo == XML_ERR_OK) {
11405
0
        ctxt->errNo = XML_ERR_USER_STOP;
11406
0
        ctxt->lastError.code = XML_ERR_USER_STOP;
11407
0
        ctxt->wellFormed = 0;
11408
0
    }
11409
97.8k
}
11410
11411
/**
11412
 * Create a parser context for using the XML parser with an existing
11413
 * I/O stream
11414
 *
11415
 * @param sax  a SAX handler (optional)
11416
 * @param user_data  user data for SAX callbacks (optional)
11417
 * @param ioread  an I/O read function
11418
 * @param ioclose  an I/O close function (optional)
11419
 * @param ioctx  an I/O handler
11420
 * @param enc  the charset encoding if known (deprecated)
11421
 * @returns the new parser context or NULL
11422
 */
11423
xmlParserCtxt *
11424
xmlCreateIOParserCtxt(xmlSAXHandler *sax, void *user_data,
11425
                      xmlInputReadCallback ioread,
11426
                      xmlInputCloseCallback ioclose,
11427
0
                      void *ioctx, xmlCharEncoding enc) {
11428
0
    xmlParserCtxtPtr ctxt;
11429
0
    xmlParserInputPtr input;
11430
0
    const char *encoding;
11431
11432
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11433
0
    if (ctxt == NULL)
11434
0
  return(NULL);
11435
11436
0
    encoding = xmlGetCharEncodingName(enc);
11437
0
    input = xmlCtxtNewInputFromIO(ctxt, NULL, ioread, ioclose, ioctx,
11438
0
                                  encoding, 0);
11439
0
    if (input == NULL) {
11440
0
  xmlFreeParserCtxt(ctxt);
11441
0
        return (NULL);
11442
0
    }
11443
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11444
0
        xmlFreeInputStream(input);
11445
0
        xmlFreeParserCtxt(ctxt);
11446
0
        return(NULL);
11447
0
    }
11448
11449
0
    return(ctxt);
11450
0
}
11451
11452
#ifdef LIBXML_VALID_ENABLED
11453
/************************************************************************
11454
 *                  *
11455
 *    Front ends when parsing a DTD       *
11456
 *                  *
11457
 ************************************************************************/
11458
11459
/**
11460
 * Parse a DTD.
11461
 *
11462
 * Option XML_PARSE_DTDLOAD should be enabled in the parser context
11463
 * to make external entities work.
11464
 *
11465
 * @since 2.14.0
11466
 *
11467
 * @param ctxt  a parser context
11468
 * @param input  a parser input
11469
 * @param publicId  public ID of the DTD (optional)
11470
 * @param systemId  system ID of the DTD (optional)
11471
 * @returns the resulting xmlDtd or NULL in case of error.
11472
 * `input` will be freed by the function in any case.
11473
 */
11474
xmlDtd *
11475
xmlCtxtParseDtd(xmlParserCtxt *ctxt, xmlParserInput *input,
11476
0
                const xmlChar *publicId, const xmlChar *systemId) {
11477
0
    xmlDtdPtr ret = NULL;
11478
11479
0
    if ((ctxt == NULL) || (input == NULL)) {
11480
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
11481
0
        xmlFreeInputStream(input);
11482
0
        return(NULL);
11483
0
    }
11484
11485
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11486
0
        xmlFreeInputStream(input);
11487
0
        return(NULL);
11488
0
    }
11489
11490
0
    if (publicId == NULL)
11491
0
        publicId = BAD_CAST "none";
11492
0
    if (systemId == NULL)
11493
0
        systemId = BAD_CAST "none";
11494
11495
0
    ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0");
11496
0
    if (ctxt->myDoc == NULL) {
11497
0
        xmlErrMemory(ctxt);
11498
0
        goto error;
11499
0
    }
11500
0
    ctxt->myDoc->properties = XML_DOC_INTERNAL;
11501
0
    ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none",
11502
0
                                       publicId, systemId);
11503
0
    if (ctxt->myDoc->extSubset == NULL) {
11504
0
        xmlErrMemory(ctxt);
11505
0
        xmlFreeDoc(ctxt->myDoc);
11506
0
        goto error;
11507
0
    }
11508
11509
0
    xmlParseExternalSubset(ctxt, publicId, systemId);
11510
11511
0
    if (ctxt->wellFormed) {
11512
0
        ret = ctxt->myDoc->extSubset;
11513
0
        ctxt->myDoc->extSubset = NULL;
11514
0
        if (ret != NULL) {
11515
0
            xmlNodePtr tmp;
11516
11517
0
            ret->doc = NULL;
11518
0
            tmp = ret->children;
11519
0
            while (tmp != NULL) {
11520
0
                tmp->doc = NULL;
11521
0
                tmp = tmp->next;
11522
0
            }
11523
0
        }
11524
0
    } else {
11525
0
        ret = NULL;
11526
0
    }
11527
0
    xmlFreeDoc(ctxt->myDoc);
11528
0
    ctxt->myDoc = NULL;
11529
11530
0
error:
11531
0
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
11532
11533
0
    return(ret);
11534
0
}
11535
11536
/**
11537
 * Load and parse a DTD
11538
 *
11539
 * @deprecated Use #xmlCtxtParseDtd.
11540
 *
11541
 * @param sax  the SAX handler block or NULL
11542
 * @param input  an Input Buffer
11543
 * @param enc  the charset encoding if known
11544
 * @returns the resulting xmlDtd or NULL in case of error.
11545
 * `input` will be freed by the function in any case.
11546
 */
11547
11548
xmlDtd *
11549
xmlIOParseDTD(xmlSAXHandler *sax, xmlParserInputBuffer *input,
11550
0
        xmlCharEncoding enc) {
11551
0
    xmlDtdPtr ret = NULL;
11552
0
    xmlParserCtxtPtr ctxt;
11553
0
    xmlParserInputPtr pinput = NULL;
11554
11555
0
    if (input == NULL)
11556
0
  return(NULL);
11557
11558
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11559
0
    if (ctxt == NULL) {
11560
0
        xmlFreeParserInputBuffer(input);
11561
0
  return(NULL);
11562
0
    }
11563
0
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11564
11565
    /*
11566
     * generate a parser input from the I/O handler
11567
     */
11568
11569
0
    pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE);
11570
0
    if (pinput == NULL) {
11571
0
  xmlFreeParserCtxt(ctxt);
11572
0
  return(NULL);
11573
0
    }
11574
11575
0
    if (enc != XML_CHAR_ENCODING_NONE) {
11576
0
        xmlSwitchEncoding(ctxt, enc);
11577
0
    }
11578
11579
0
    ret = xmlCtxtParseDtd(ctxt, pinput, NULL, NULL);
11580
11581
0
    xmlFreeParserCtxt(ctxt);
11582
0
    return(ret);
11583
0
}
11584
11585
/**
11586
 * Load and parse an external subset.
11587
 *
11588
 * @deprecated Use #xmlCtxtParseDtd.
11589
 *
11590
 * @param sax  the SAX handler block
11591
 * @param publicId  public identifier of the DTD (optional)
11592
 * @param systemId  system identifier (URL) of the DTD
11593
 * @returns the resulting xmlDtd or NULL in case of error.
11594
 */
11595
11596
xmlDtd *
11597
xmlSAXParseDTD(xmlSAXHandler *sax, const xmlChar *publicId,
11598
0
               const xmlChar *systemId) {
11599
0
    xmlDtdPtr ret = NULL;
11600
0
    xmlParserCtxtPtr ctxt;
11601
0
    xmlParserInputPtr input = NULL;
11602
0
    xmlChar* systemIdCanonic;
11603
11604
0
    if ((publicId == NULL) && (systemId == NULL)) return(NULL);
11605
11606
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11607
0
    if (ctxt == NULL) {
11608
0
  return(NULL);
11609
0
    }
11610
0
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11611
11612
    /*
11613
     * Canonicalise the system ID
11614
     */
11615
0
    systemIdCanonic = xmlCanonicPath(systemId);
11616
0
    if ((systemId != NULL) && (systemIdCanonic == NULL)) {
11617
0
  xmlFreeParserCtxt(ctxt);
11618
0
  return(NULL);
11619
0
    }
11620
11621
    /*
11622
     * Ask the Entity resolver to load the damn thing
11623
     */
11624
11625
0
    if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL))
11626
0
  input = ctxt->sax->resolveEntity(ctxt->userData, publicId,
11627
0
                                   systemIdCanonic);
11628
0
    if (input == NULL) {
11629
0
  xmlFreeParserCtxt(ctxt);
11630
0
  if (systemIdCanonic != NULL)
11631
0
      xmlFree(systemIdCanonic);
11632
0
  return(NULL);
11633
0
    }
11634
11635
0
    if (input->filename == NULL)
11636
0
  input->filename = (char *) systemIdCanonic;
11637
0
    else
11638
0
  xmlFree(systemIdCanonic);
11639
11640
0
    ret = xmlCtxtParseDtd(ctxt, input, publicId, systemId);
11641
11642
0
    xmlFreeParserCtxt(ctxt);
11643
0
    return(ret);
11644
0
}
11645
11646
11647
/**
11648
 * Load and parse an external subset.
11649
 *
11650
 * @param publicId  public identifier of the DTD (optional)
11651
 * @param systemId  system identifier (URL) of the DTD
11652
 * @returns the resulting xmlDtd or NULL in case of error.
11653
 */
11654
11655
xmlDtd *
11656
0
xmlParseDTD(const xmlChar *publicId, const xmlChar *systemId) {
11657
0
    return(xmlSAXParseDTD(NULL, publicId, systemId));
11658
0
}
11659
#endif /* LIBXML_VALID_ENABLED */
11660
11661
/************************************************************************
11662
 *                  *
11663
 *    Front ends when parsing an Entity     *
11664
 *                  *
11665
 ************************************************************************/
11666
11667
static xmlNodePtr
11668
xmlCtxtParseContentInternal(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
11669
621k
                            int hasTextDecl, int buildTree) {
11670
621k
    xmlNodePtr root = NULL;
11671
621k
    xmlNodePtr list = NULL;
11672
621k
    xmlChar *rootName = BAD_CAST "#root";
11673
621k
    int result;
11674
11675
621k
    if (buildTree) {
11676
0
        root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL);
11677
0
        if (root == NULL) {
11678
0
            xmlErrMemory(ctxt);
11679
0
            goto error;
11680
0
        }
11681
0
    }
11682
11683
621k
    if (xmlCtxtPushInput(ctxt, input) < 0)
11684
3
        goto error;
11685
11686
621k
    nameNsPush(ctxt, rootName, NULL, NULL, 0, 0);
11687
621k
    spacePush(ctxt, -1);
11688
11689
621k
    if (buildTree)
11690
0
        nodePush(ctxt, root);
11691
11692
621k
    if (hasTextDecl) {
11693
0
        xmlDetectEncoding(ctxt);
11694
11695
        /*
11696
         * Parse a possible text declaration first
11697
         */
11698
0
        if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
11699
0
            (IS_BLANK_CH(NXT(5)))) {
11700
0
            xmlParseTextDecl(ctxt);
11701
            /*
11702
             * An XML-1.0 document can't reference an entity not XML-1.0
11703
             */
11704
0
            if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
11705
0
                (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
11706
0
                xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
11707
0
                               "Version mismatch between document and "
11708
0
                               "entity\n");
11709
0
            }
11710
0
        }
11711
0
    }
11712
11713
621k
    xmlParseContentInternal(ctxt);
11714
11715
621k
    if (ctxt->input->cur < ctxt->input->end)
11716
1.00k
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
11717
11718
621k
    if ((ctxt->wellFormed) ||
11719
620k
        ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) {
11720
620k
        if (root != NULL) {
11721
0
            xmlNodePtr cur;
11722
11723
            /*
11724
             * Unlink newly created node list.
11725
             */
11726
0
            list = root->children;
11727
0
            root->children = NULL;
11728
0
            root->last = NULL;
11729
0
            for (cur = list; cur != NULL; cur = cur->next)
11730
0
                cur->parent = NULL;
11731
0
        }
11732
620k
    }
11733
11734
    /*
11735
     * Read the rest of the stream in case of errors. We want
11736
     * to account for the whole entity size.
11737
     */
11738
621k
    do {
11739
621k
        ctxt->input->cur = ctxt->input->end;
11740
621k
        xmlParserShrink(ctxt);
11741
621k
        result = xmlParserGrow(ctxt);
11742
621k
    } while (result > 0);
11743
11744
621k
    if (buildTree)
11745
0
        nodePop(ctxt);
11746
11747
621k
    namePop(ctxt);
11748
621k
    spacePop(ctxt);
11749
11750
621k
    xmlCtxtPopInput(ctxt);
11751
11752
621k
error:
11753
621k
    xmlFreeNode(root);
11754
11755
621k
    return(list);
11756
621k
}
11757
11758
static void
11759
621k
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
11760
621k
    xmlParserInputPtr input;
11761
621k
    xmlNodePtr list;
11762
621k
    unsigned long consumed;
11763
621k
    int isExternal;
11764
621k
    int buildTree;
11765
621k
    int oldMinNsIndex;
11766
621k
    int oldNodelen, oldNodemem;
11767
11768
621k
    isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY);
11769
621k
    buildTree = (ctxt->node != NULL);
11770
11771
    /*
11772
     * Recursion check
11773
     */
11774
621k
    if (ent->flags & XML_ENT_EXPANDING) {
11775
31
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
11776
31
        goto error;
11777
31
    }
11778
11779
    /*
11780
     * Load entity
11781
     */
11782
621k
    input = xmlNewEntityInputStream(ctxt, ent);
11783
621k
    if (input == NULL)
11784
0
        goto error;
11785
11786
    /*
11787
     * When building a tree, we need to limit the scope of namespace
11788
     * declarations, so that entities don't reference xmlNs structs
11789
     * from the parent of a reference.
11790
     */
11791
621k
    oldMinNsIndex = ctxt->nsdb->minNsIndex;
11792
621k
    if (buildTree)
11793
0
        ctxt->nsdb->minNsIndex = ctxt->nsNr;
11794
11795
621k
    oldNodelen = ctxt->nodelen;
11796
621k
    oldNodemem = ctxt->nodemem;
11797
621k
    ctxt->nodelen = 0;
11798
621k
    ctxt->nodemem = 0;
11799
11800
    /*
11801
     * Parse content
11802
     *
11803
     * This initiates a recursive call chain:
11804
     *
11805
     * - xmlCtxtParseContentInternal
11806
     * - xmlParseContentInternal
11807
     * - xmlParseReference
11808
     * - xmlCtxtParseEntity
11809
     *
11810
     * The nesting depth is limited by the maximum number of inputs,
11811
     * see xmlCtxtPushInput.
11812
     *
11813
     * It's possible to make this non-recursive (minNsIndex must be
11814
     * stored in the input struct) at the expense of code readability.
11815
     */
11816
11817
621k
    ent->flags |= XML_ENT_EXPANDING;
11818
11819
621k
    list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree);
11820
11821
621k
    ent->flags &= ~XML_ENT_EXPANDING;
11822
11823
621k
    ctxt->nsdb->minNsIndex = oldMinNsIndex;
11824
621k
    ctxt->nodelen = oldNodelen;
11825
621k
    ctxt->nodemem = oldNodemem;
11826
11827
    /*
11828
     * Entity size accounting
11829
     */
11830
621k
    consumed = input->consumed;
11831
621k
    xmlSaturatedAddSizeT(&consumed, input->end - input->base);
11832
11833
621k
    if ((ent->flags & XML_ENT_CHECKED) == 0)
11834
7.24k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
11835
11836
621k
    if ((ent->flags & XML_ENT_PARSED) == 0) {
11837
7.37k
        if (isExternal)
11838
0
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
11839
11840
7.37k
        ent->children = list;
11841
11842
7.37k
        while (list != NULL) {
11843
0
            list->parent = (xmlNodePtr) ent;
11844
11845
            /*
11846
             * Downstream code like the nginx xslt module can set
11847
             * ctxt->myDoc->extSubset to a separate DTD, so the entity
11848
             * might have a different or a NULL document.
11849
             */
11850
0
            if (list->doc != ent->doc)
11851
0
                xmlSetTreeDoc(list, ent->doc);
11852
11853
0
            if (list->next == NULL)
11854
0
                ent->last = list;
11855
0
            list = list->next;
11856
0
        }
11857
614k
    } else {
11858
614k
        xmlFreeNodeList(list);
11859
614k
    }
11860
11861
621k
    xmlFreeInputStream(input);
11862
11863
621k
error:
11864
621k
    ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED;
11865
621k
}
11866
11867
/**
11868
 * Parse an external general entity within an existing parsing context
11869
 * An external general parsed entity is well-formed if it matches the
11870
 * production labeled extParsedEnt.
11871
 *
11872
 *     [78] extParsedEnt ::= TextDecl? content
11873
 *
11874
 * @param ctxt  the existing parsing context
11875
 * @param URL  the URL for the entity to load
11876
 * @param ID  the System ID for the entity to load
11877
 * @param listOut  the return value for the set of parsed nodes
11878
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11879
 *    the parser error code otherwise
11880
 */
11881
11882
int
11883
xmlParseCtxtExternalEntity(xmlParserCtxt *ctxt, const xmlChar *URL,
11884
0
                           const xmlChar *ID, xmlNode **listOut) {
11885
0
    xmlParserInputPtr input;
11886
0
    xmlNodePtr list;
11887
11888
0
    if (listOut != NULL)
11889
0
        *listOut = NULL;
11890
11891
0
    if (ctxt == NULL)
11892
0
        return(XML_ERR_ARGUMENT);
11893
11894
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
11895
0
                            XML_RESOURCE_GENERAL_ENTITY);
11896
0
    if (input == NULL)
11897
0
        return(ctxt->errNo);
11898
11899
0
    xmlCtxtInitializeLate(ctxt);
11900
11901
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1);
11902
0
    if (listOut != NULL)
11903
0
        *listOut = list;
11904
0
    else
11905
0
        xmlFreeNodeList(list);
11906
11907
0
    xmlFreeInputStream(input);
11908
0
    return(ctxt->errNo);
11909
0
}
11910
11911
#ifdef LIBXML_SAX1_ENABLED
11912
/**
11913
 * Parse an external general entity
11914
 * An external general parsed entity is well-formed if it matches the
11915
 * production labeled extParsedEnt.
11916
 *
11917
 * This function uses deprecated global variables to set parser options
11918
 * which default to XML_PARSE_NODICT.
11919
 *
11920
 * @deprecated Use #xmlParseCtxtExternalEntity.
11921
 *
11922
 *     [78] extParsedEnt ::= TextDecl? content
11923
 *
11924
 * @param doc  the document the chunk pertains to
11925
 * @param sax  the SAX handler block (possibly NULL)
11926
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11927
 * @param depth  Used for loop detection, use 0
11928
 * @param URL  the URL for the entity to load
11929
 * @param ID  the System ID for the entity to load
11930
 * @param list  the return value for the set of parsed nodes
11931
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11932
 *    the parser error code otherwise
11933
 */
11934
11935
int
11936
xmlParseExternalEntity(xmlDoc *doc, xmlSAXHandler *sax, void *user_data,
11937
0
    int depth, const xmlChar *URL, const xmlChar *ID, xmlNode **list) {
11938
0
    xmlParserCtxtPtr ctxt;
11939
0
    int ret;
11940
11941
0
    if (list != NULL)
11942
0
        *list = NULL;
11943
11944
0
    if (doc == NULL)
11945
0
        return(XML_ERR_ARGUMENT);
11946
11947
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11948
0
    if (ctxt == NULL)
11949
0
        return(XML_ERR_NO_MEMORY);
11950
11951
0
    ctxt->depth = depth;
11952
0
    ctxt->myDoc = doc;
11953
0
    ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list);
11954
11955
0
    xmlFreeParserCtxt(ctxt);
11956
0
    return(ret);
11957
0
}
11958
11959
/**
11960
 * Parse a well-balanced chunk of an XML document
11961
 * called by the parser
11962
 * The allowed sequence for the Well Balanced Chunk is the one defined by
11963
 * the content production in the XML grammar:
11964
 *
11965
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
11966
 *                       Comment)*
11967
 *
11968
 * This function uses deprecated global variables to set parser options
11969
 * which default to XML_PARSE_NODICT.
11970
 *
11971
 * @param doc  the document the chunk pertains to (must not be NULL)
11972
 * @param sax  the SAX handler block (possibly NULL)
11973
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11974
 * @param depth  Used for loop detection, use 0
11975
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
11976
 * @param lst  the return value for the set of parsed nodes
11977
 * @returns 0 if the chunk is well balanced, -1 in case of args problem and
11978
 *    the parser error code otherwise
11979
 */
11980
11981
int
11982
xmlParseBalancedChunkMemory(xmlDoc *doc, xmlSAXHandler *sax,
11983
0
     void *user_data, int depth, const xmlChar *string, xmlNode **lst) {
11984
0
    return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data,
11985
0
                                                depth, string, lst, 0 );
11986
0
}
11987
#endif /* LIBXML_SAX1_ENABLED */
11988
11989
/**
11990
 * Parse a well-balanced chunk of XML matching the 'content' production.
11991
 *
11992
 * Namespaces in scope of `node` and entities of `node`'s document are
11993
 * recognized. When validating, the DTD of `node`'s document is used.
11994
 *
11995
 * Always consumes `input` even in error case.
11996
 *
11997
 * @since 2.14.0
11998
 *
11999
 * @param ctxt  parser context
12000
 * @param input  parser input
12001
 * @param node  target node or document
12002
 * @param hasTextDecl  whether to parse text declaration
12003
 * @returns a node list or NULL in case of error.
12004
 */
12005
xmlNode *
12006
xmlCtxtParseContent(xmlParserCtxt *ctxt, xmlParserInput *input,
12007
0
                    xmlNode *node, int hasTextDecl) {
12008
0
    xmlDocPtr doc;
12009
0
    xmlNodePtr cur, list = NULL;
12010
0
    int nsnr = 0;
12011
0
    xmlDictPtr oldDict;
12012
0
    int oldOptions, oldDictNames, oldLoadSubset;
12013
12014
0
    if ((ctxt == NULL) || (input == NULL) || (node == NULL)) {
12015
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12016
0
        goto exit;
12017
0
    }
12018
12019
0
    doc = node->doc;
12020
0
    if (doc == NULL) {
12021
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12022
0
        goto exit;
12023
0
    }
12024
12025
0
    switch (node->type) {
12026
0
        case XML_ELEMENT_NODE:
12027
0
        case XML_DOCUMENT_NODE:
12028
0
        case XML_HTML_DOCUMENT_NODE:
12029
0
            break;
12030
12031
0
        case XML_ATTRIBUTE_NODE:
12032
0
        case XML_TEXT_NODE:
12033
0
        case XML_CDATA_SECTION_NODE:
12034
0
        case XML_ENTITY_REF_NODE:
12035
0
        case XML_PI_NODE:
12036
0
        case XML_COMMENT_NODE:
12037
0
            for (cur = node->parent; cur != NULL; cur = cur->parent) {
12038
0
                if ((cur->type == XML_ELEMENT_NODE) ||
12039
0
                    (cur->type == XML_DOCUMENT_NODE) ||
12040
0
                    (cur->type == XML_HTML_DOCUMENT_NODE)) {
12041
0
                    node = cur;
12042
0
                    break;
12043
0
                }
12044
0
            }
12045
0
            break;
12046
12047
0
        default:
12048
0
            xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12049
0
            goto exit;
12050
0
    }
12051
12052
0
    xmlCtxtReset(ctxt);
12053
12054
0
    oldDict = ctxt->dict;
12055
0
    oldOptions = ctxt->options;
12056
0
    oldDictNames = ctxt->dictNames;
12057
0
    oldLoadSubset = ctxt->loadsubset;
12058
12059
    /*
12060
     * Use input doc's dict if present, else assure XML_PARSE_NODICT is set.
12061
     */
12062
0
    if (doc->dict != NULL) {
12063
0
        ctxt->dict = doc->dict;
12064
0
    } else {
12065
0
        ctxt->options |= XML_PARSE_NODICT;
12066
0
        ctxt->dictNames = 0;
12067
0
    }
12068
12069
    /*
12070
     * Disable IDs
12071
     */
12072
0
    ctxt->loadsubset |= XML_SKIP_IDS;
12073
0
    ctxt->options |= XML_PARSE_SKIP_IDS;
12074
12075
0
    ctxt->myDoc = doc;
12076
12077
0
#ifdef LIBXML_HTML_ENABLED
12078
0
    if (ctxt->html) {
12079
        /*
12080
         * When parsing in context, it makes no sense to add implied
12081
         * elements like html/body/etc...
12082
         */
12083
0
        ctxt->options |= HTML_PARSE_NOIMPLIED;
12084
12085
0
        list = htmlCtxtParseContentInternal(ctxt, input);
12086
0
    } else
12087
0
#endif
12088
0
    {
12089
0
        xmlCtxtInitializeLate(ctxt);
12090
12091
        /*
12092
         * initialize the SAX2 namespaces stack
12093
         */
12094
0
        cur = node;
12095
0
        while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) {
12096
0
            xmlNsPtr ns = cur->nsDef;
12097
0
            xmlHashedString hprefix, huri;
12098
12099
0
            while (ns != NULL) {
12100
0
                hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1);
12101
0
                huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1);
12102
0
                if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0)
12103
0
                    nsnr++;
12104
0
                ns = ns->next;
12105
0
            }
12106
0
            cur = cur->parent;
12107
0
        }
12108
12109
0
        list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1);
12110
12111
0
        if (nsnr > 0)
12112
0
            xmlParserNsPop(ctxt, nsnr);
12113
0
    }
12114
12115
0
    ctxt->dict = oldDict;
12116
0
    ctxt->options = oldOptions;
12117
0
    ctxt->dictNames = oldDictNames;
12118
0
    ctxt->loadsubset = oldLoadSubset;
12119
0
    ctxt->myDoc = NULL;
12120
0
    ctxt->node = NULL;
12121
12122
0
exit:
12123
0
    xmlFreeInputStream(input);
12124
0
    return(list);
12125
0
}
12126
12127
/**
12128
 * Parse a well-balanced chunk of an XML document
12129
 * within the context (DTD, namespaces, etc ...) of the given node.
12130
 *
12131
 * The allowed sequence for the data is a Well Balanced Chunk defined by
12132
 * the content production in the XML grammar:
12133
 *
12134
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12135
 *                       Comment)*
12136
 *
12137
 * This function assumes the encoding of `node`'s document which is
12138
 * typically not what you want. A better alternative is
12139
 * #xmlCtxtParseContent.
12140
 *
12141
 * @param node  the context node
12142
 * @param data  the input string
12143
 * @param datalen  the input string length in bytes
12144
 * @param options  a combination of xmlParserOption
12145
 * @param listOut  the return value for the set of parsed nodes
12146
 * @returns XML_ERR_OK if the chunk is well balanced, and the parser
12147
 * error code otherwise
12148
 */
12149
xmlParserErrors
12150
xmlParseInNodeContext(xmlNode *node, const char *data, int datalen,
12151
0
                      int options, xmlNode **listOut) {
12152
0
    xmlParserCtxtPtr ctxt;
12153
0
    xmlParserInputPtr input;
12154
0
    xmlDocPtr doc;
12155
0
    xmlNodePtr list;
12156
0
    xmlParserErrors ret;
12157
12158
0
    if (listOut == NULL)
12159
0
        return(XML_ERR_INTERNAL_ERROR);
12160
0
    *listOut = NULL;
12161
12162
0
    if ((node == NULL) || (data == NULL) || (datalen < 0))
12163
0
        return(XML_ERR_INTERNAL_ERROR);
12164
12165
0
    doc = node->doc;
12166
0
    if (doc == NULL)
12167
0
        return(XML_ERR_INTERNAL_ERROR);
12168
12169
0
#ifdef LIBXML_HTML_ENABLED
12170
0
    if (doc->type == XML_HTML_DOCUMENT_NODE) {
12171
0
        ctxt = htmlNewParserCtxt();
12172
0
    }
12173
0
    else
12174
0
#endif
12175
0
        ctxt = xmlNewParserCtxt();
12176
12177
0
    if (ctxt == NULL)
12178
0
        return(XML_ERR_NO_MEMORY);
12179
12180
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen,
12181
0
                                      (const char *) doc->encoding,
12182
0
                                      XML_INPUT_BUF_STATIC);
12183
0
    if (input == NULL) {
12184
0
        xmlFreeParserCtxt(ctxt);
12185
0
        return(XML_ERR_NO_MEMORY);
12186
0
    }
12187
12188
0
    xmlCtxtUseOptions(ctxt, options);
12189
12190
0
    list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0);
12191
12192
0
    if (list == NULL) {
12193
0
        ret = ctxt->errNo;
12194
0
        if (ret == XML_ERR_ARGUMENT)
12195
0
            ret = XML_ERR_INTERNAL_ERROR;
12196
0
    } else {
12197
0
        ret = XML_ERR_OK;
12198
0
        *listOut = list;
12199
0
    }
12200
12201
0
    xmlFreeParserCtxt(ctxt);
12202
12203
0
    return(ret);
12204
0
}
12205
12206
#ifdef LIBXML_SAX1_ENABLED
12207
/**
12208
 * Parse a well-balanced chunk of an XML document
12209
 *
12210
 * The allowed sequence for the Well Balanced Chunk is the one defined by
12211
 * the content production in the XML grammar:
12212
 *
12213
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12214
 *                       Comment)*
12215
 *
12216
 * In case recover is set to 1, the nodelist will not be empty even if
12217
 * the parsed chunk is not well balanced, assuming the parsing succeeded to
12218
 * some extent.
12219
 *
12220
 * This function uses deprecated global variables to set parser options
12221
 * which default to XML_PARSE_NODICT.
12222
 *
12223
 * @param doc  the document the chunk pertains to (must not be NULL)
12224
 * @param sax  the SAX handler block (possibly NULL)
12225
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
12226
 * @param depth  Used for loop detection, use 0
12227
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
12228
 * @param listOut  the return value for the set of parsed nodes
12229
 * @param recover  return nodes even if the data is broken (use 0)
12230
 * @returns 0 if the chunk is well balanced, or thehe parser error code
12231
 * otherwise.
12232
 */
12233
int
12234
xmlParseBalancedChunkMemoryRecover(xmlDoc *doc, xmlSAXHandler *sax,
12235
     void *user_data, int depth, const xmlChar *string, xmlNode **listOut,
12236
0
     int recover) {
12237
0
    xmlParserCtxtPtr ctxt;
12238
0
    xmlParserInputPtr input;
12239
0
    xmlNodePtr list;
12240
0
    int ret;
12241
12242
0
    if (listOut != NULL)
12243
0
        *listOut = NULL;
12244
12245
0
    if (string == NULL)
12246
0
        return(XML_ERR_ARGUMENT);
12247
12248
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
12249
0
    if (ctxt == NULL)
12250
0
        return(XML_ERR_NO_MEMORY);
12251
12252
0
    xmlCtxtInitializeLate(ctxt);
12253
12254
0
    ctxt->depth = depth;
12255
0
    ctxt->myDoc = doc;
12256
0
    if (recover) {
12257
0
        ctxt->options |= XML_PARSE_RECOVER;
12258
0
        ctxt->recovery = 1;
12259
0
    }
12260
12261
0
    input = xmlNewStringInputStream(ctxt, string);
12262
0
    if (input == NULL) {
12263
0
        ret = ctxt->errNo;
12264
0
        goto error;
12265
0
    }
12266
12267
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1);
12268
0
    if (listOut != NULL)
12269
0
        *listOut = list;
12270
0
    else
12271
0
        xmlFreeNodeList(list);
12272
12273
0
    if (!ctxt->wellFormed)
12274
0
        ret = ctxt->errNo;
12275
0
    else
12276
0
        ret = XML_ERR_OK;
12277
12278
0
error:
12279
0
    xmlFreeInputStream(input);
12280
0
    xmlFreeParserCtxt(ctxt);
12281
0
    return(ret);
12282
0
}
12283
12284
/**
12285
 * Parse an XML external entity out of context and build a tree.
12286
 * It use the given SAX function block to handle the parsing callback.
12287
 * If sax is NULL, fallback to the default DOM tree building routines.
12288
 *
12289
 * @deprecated Don't use.
12290
 *
12291
 *     [78] extParsedEnt ::= TextDecl? content
12292
 *
12293
 * This correspond to a "Well Balanced" chunk
12294
 *
12295
 * This function uses deprecated global variables to set parser options
12296
 * which default to XML_PARSE_NODICT.
12297
 *
12298
 * @param sax  the SAX handler block
12299
 * @param filename  the filename
12300
 * @returns the resulting document tree
12301
 */
12302
12303
xmlDoc *
12304
0
xmlSAXParseEntity(xmlSAXHandler *sax, const char *filename) {
12305
0
    xmlDocPtr ret;
12306
0
    xmlParserCtxtPtr ctxt;
12307
12308
0
    ctxt = xmlCreateFileParserCtxt(filename);
12309
0
    if (ctxt == NULL) {
12310
0
  return(NULL);
12311
0
    }
12312
0
    if (sax != NULL) {
12313
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12314
0
            *ctxt->sax = *sax;
12315
0
        } else {
12316
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12317
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12318
0
        }
12319
0
        ctxt->userData = NULL;
12320
0
    }
12321
12322
0
    xmlParseExtParsedEnt(ctxt);
12323
12324
0
    if (ctxt->wellFormed) {
12325
0
  ret = ctxt->myDoc;
12326
0
    } else {
12327
0
        ret = NULL;
12328
0
        xmlFreeDoc(ctxt->myDoc);
12329
0
    }
12330
12331
0
    xmlFreeParserCtxt(ctxt);
12332
12333
0
    return(ret);
12334
0
}
12335
12336
/**
12337
 * Parse an XML external entity out of context and build a tree.
12338
 *
12339
 *     [78] extParsedEnt ::= TextDecl? content
12340
 *
12341
 * This correspond to a "Well Balanced" chunk
12342
 *
12343
 * This function uses deprecated global variables to set parser options
12344
 * which default to XML_PARSE_NODICT.
12345
 *
12346
 * @deprecated Don't use.
12347
 *
12348
 * @param filename  the filename
12349
 * @returns the resulting document tree
12350
 */
12351
12352
xmlDoc *
12353
0
xmlParseEntity(const char *filename) {
12354
0
    return(xmlSAXParseEntity(NULL, filename));
12355
0
}
12356
#endif /* LIBXML_SAX1_ENABLED */
12357
12358
/**
12359
 * Create a parser context for an external entity
12360
 * Automatic support for ZLIB/Compress compressed document is provided
12361
 * by default if found at compile-time.
12362
 *
12363
 * @deprecated Don't use.
12364
 *
12365
 * @param URL  the entity URL
12366
 * @param ID  the entity PUBLIC ID
12367
 * @param base  a possible base for the target URI
12368
 * @returns the new parser context or NULL
12369
 */
12370
xmlParserCtxt *
12371
xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID,
12372
0
                    const xmlChar *base) {
12373
0
    xmlParserCtxtPtr ctxt;
12374
0
    xmlParserInputPtr input;
12375
0
    xmlChar *uri = NULL;
12376
12377
0
    ctxt = xmlNewParserCtxt();
12378
0
    if (ctxt == NULL)
12379
0
  return(NULL);
12380
12381
0
    if (base != NULL) {
12382
0
        if (xmlBuildURISafe(URL, base, &uri) < 0)
12383
0
            goto error;
12384
0
        if (uri != NULL)
12385
0
            URL = uri;
12386
0
    }
12387
12388
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
12389
0
                            XML_RESOURCE_UNKNOWN);
12390
0
    if (input == NULL)
12391
0
        goto error;
12392
12393
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12394
0
        xmlFreeInputStream(input);
12395
0
        goto error;
12396
0
    }
12397
12398
0
    xmlFree(uri);
12399
0
    return(ctxt);
12400
12401
0
error:
12402
0
    xmlFree(uri);
12403
0
    xmlFreeParserCtxt(ctxt);
12404
0
    return(NULL);
12405
0
}
12406
12407
/************************************************************************
12408
 *                  *
12409
 *    Front ends when parsing from a file     *
12410
 *                  *
12411
 ************************************************************************/
12412
12413
/**
12414
 * Create a parser context for a file or URL content.
12415
 * Automatic support for ZLIB/Compress compressed document is provided
12416
 * by default if found at compile-time and for file accesses
12417
 *
12418
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12419
 *
12420
 * @param filename  the filename or URL
12421
 * @param options  a combination of xmlParserOption
12422
 * @returns the new parser context or NULL
12423
 */
12424
xmlParserCtxt *
12425
xmlCreateURLParserCtxt(const char *filename, int options)
12426
0
{
12427
0
    xmlParserCtxtPtr ctxt;
12428
0
    xmlParserInputPtr input;
12429
12430
0
    ctxt = xmlNewParserCtxt();
12431
0
    if (ctxt == NULL)
12432
0
  return(NULL);
12433
12434
0
    xmlCtxtUseOptions(ctxt, options);
12435
12436
0
    input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT);
12437
0
    if (input == NULL) {
12438
0
  xmlFreeParserCtxt(ctxt);
12439
0
  return(NULL);
12440
0
    }
12441
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12442
0
        xmlFreeInputStream(input);
12443
0
        xmlFreeParserCtxt(ctxt);
12444
0
        return(NULL);
12445
0
    }
12446
12447
0
    return(ctxt);
12448
0
}
12449
12450
/**
12451
 * Create a parser context for a file content.
12452
 * Automatic support for ZLIB/Compress compressed document is provided
12453
 * by default if found at compile-time.
12454
 *
12455
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12456
 *
12457
 * @param filename  the filename
12458
 * @returns the new parser context or NULL
12459
 */
12460
xmlParserCtxt *
12461
xmlCreateFileParserCtxt(const char *filename)
12462
0
{
12463
0
    return(xmlCreateURLParserCtxt(filename, 0));
12464
0
}
12465
12466
#ifdef LIBXML_SAX1_ENABLED
12467
/**
12468
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12469
 * compressed document is provided by default if found at compile-time.
12470
 * It use the given SAX function block to handle the parsing callback.
12471
 * If sax is NULL, fallback to the default DOM tree building routines.
12472
 *
12473
 * This function uses deprecated global variables to set parser options
12474
 * which default to XML_PARSE_NODICT.
12475
 *
12476
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12477
 *
12478
 * User data (void *) is stored within the parser context in the
12479
 * context's _private member, so it is available nearly everywhere in libxml
12480
 *
12481
 * @param sax  the SAX handler block
12482
 * @param filename  the filename
12483
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12484
 *             documents
12485
 * @param data  the userdata
12486
 * @returns the resulting document tree
12487
 */
12488
12489
xmlDoc *
12490
xmlSAXParseFileWithData(xmlSAXHandler *sax, const char *filename,
12491
0
                        int recovery, void *data) {
12492
0
    xmlDocPtr ret = NULL;
12493
0
    xmlParserCtxtPtr ctxt;
12494
0
    xmlParserInputPtr input;
12495
12496
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12497
0
    if (ctxt == NULL)
12498
0
  return(NULL);
12499
12500
0
    if (data != NULL)
12501
0
  ctxt->_private = data;
12502
12503
0
    if (recovery) {
12504
0
        ctxt->options |= XML_PARSE_RECOVER;
12505
0
        ctxt->recovery = 1;
12506
0
    }
12507
12508
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
12509
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0);
12510
0
    else
12511
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0);
12512
12513
0
    if (input != NULL)
12514
0
        ret = xmlCtxtParseDocument(ctxt, input);
12515
12516
0
    xmlFreeParserCtxt(ctxt);
12517
0
    return(ret);
12518
0
}
12519
12520
/**
12521
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12522
 * compressed document is provided by default if found at compile-time.
12523
 * It use the given SAX function block to handle the parsing callback.
12524
 * If sax is NULL, fallback to the default DOM tree building routines.
12525
 *
12526
 * This function uses deprecated global variables to set parser options
12527
 * which default to XML_PARSE_NODICT.
12528
 *
12529
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12530
 *
12531
 * @param sax  the SAX handler block
12532
 * @param filename  the filename
12533
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12534
 *             documents
12535
 * @returns the resulting document tree
12536
 */
12537
12538
xmlDoc *
12539
xmlSAXParseFile(xmlSAXHandler *sax, const char *filename,
12540
0
                          int recovery) {
12541
0
    return(xmlSAXParseFileWithData(sax,filename,recovery,NULL));
12542
0
}
12543
12544
/**
12545
 * Parse an XML in-memory document and build a tree.
12546
 * In the case the document is not Well Formed, a attempt to build a
12547
 * tree is tried anyway
12548
 *
12549
 * This function uses deprecated global variables to set parser options
12550
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12551
 *
12552
 * @deprecated Use #xmlReadDoc with XML_PARSE_RECOVER.
12553
 *
12554
 * @param cur  a pointer to an array of xmlChar
12555
 * @returns the resulting document tree or NULL in case of failure
12556
 */
12557
12558
xmlDoc *
12559
0
xmlRecoverDoc(const xmlChar *cur) {
12560
0
    return(xmlSAXParseDoc(NULL, cur, 1));
12561
0
}
12562
12563
/**
12564
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12565
 * compressed document is provided by default if found at compile-time.
12566
 *
12567
 * This function uses deprecated global variables to set parser options
12568
 * which default to XML_PARSE_NODICT.
12569
 *
12570
 * @deprecated Use #xmlReadFile.
12571
 *
12572
 * @param filename  the filename
12573
 * @returns the resulting document tree if the file was wellformed,
12574
 * NULL otherwise.
12575
 */
12576
12577
xmlDoc *
12578
0
xmlParseFile(const char *filename) {
12579
0
    return(xmlSAXParseFile(NULL, filename, 0));
12580
0
}
12581
12582
/**
12583
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12584
 * compressed document is provided by default if found at compile-time.
12585
 * In the case the document is not Well Formed, it attempts to build
12586
 * a tree anyway
12587
 *
12588
 * This function uses deprecated global variables to set parser options
12589
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12590
 *
12591
 * @deprecated Use #xmlReadFile with XML_PARSE_RECOVER.
12592
 *
12593
 * @param filename  the filename
12594
 * @returns the resulting document tree or NULL in case of failure
12595
 */
12596
12597
xmlDoc *
12598
0
xmlRecoverFile(const char *filename) {
12599
0
    return(xmlSAXParseFile(NULL, filename, 1));
12600
0
}
12601
12602
12603
/**
12604
 * Setup the parser context to parse a new buffer; Clears any prior
12605
 * contents from the parser context. The buffer parameter must not be
12606
 * NULL, but the filename parameter can be
12607
 *
12608
 * @deprecated Don't use.
12609
 *
12610
 * @param ctxt  an XML parser context
12611
 * @param buffer  a xmlChar * buffer
12612
 * @param filename  a file name
12613
 */
12614
void
12615
xmlSetupParserForBuffer(xmlParserCtxt *ctxt, const xmlChar* buffer,
12616
                             const char* filename)
12617
0
{
12618
0
    xmlParserInputPtr input;
12619
12620
0
    if ((ctxt == NULL) || (buffer == NULL))
12621
0
        return;
12622
12623
0
    xmlCtxtReset(ctxt);
12624
12625
0
    input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer,
12626
0
                                      NULL, 0);
12627
0
    if (input == NULL)
12628
0
        return;
12629
0
    if (xmlCtxtPushInput(ctxt, input) < 0)
12630
0
        xmlFreeInputStream(input);
12631
0
}
12632
12633
/**
12634
 * Parse an XML file and call the given SAX handler routines.
12635
 * Automatic support for ZLIB/Compress compressed document is provided
12636
 *
12637
 * This function uses deprecated global variables to set parser options
12638
 * which default to XML_PARSE_NODICT.
12639
 *
12640
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12641
 *
12642
 * @param sax  a SAX handler
12643
 * @param user_data  The user data returned on SAX callbacks
12644
 * @param filename  a file name
12645
 * @returns 0 in case of success or a error number otherwise
12646
 */
12647
int
12648
xmlSAXUserParseFile(xmlSAXHandler *sax, void *user_data,
12649
0
                    const char *filename) {
12650
0
    int ret = 0;
12651
0
    xmlParserCtxtPtr ctxt;
12652
12653
0
    ctxt = xmlCreateFileParserCtxt(filename);
12654
0
    if (ctxt == NULL) return -1;
12655
0
    if (sax != NULL) {
12656
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12657
0
            *ctxt->sax = *sax;
12658
0
        } else {
12659
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12660
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12661
0
        }
12662
0
  ctxt->userData = user_data;
12663
0
    }
12664
12665
0
    xmlParseDocument(ctxt);
12666
12667
0
    if (ctxt->wellFormed)
12668
0
  ret = 0;
12669
0
    else {
12670
0
        if (ctxt->errNo != 0)
12671
0
      ret = ctxt->errNo;
12672
0
  else
12673
0
      ret = -1;
12674
0
    }
12675
0
    if (ctxt->myDoc != NULL) {
12676
0
        xmlFreeDoc(ctxt->myDoc);
12677
0
  ctxt->myDoc = NULL;
12678
0
    }
12679
0
    xmlFreeParserCtxt(ctxt);
12680
12681
0
    return ret;
12682
0
}
12683
#endif /* LIBXML_SAX1_ENABLED */
12684
12685
/************************************************************************
12686
 *                  *
12687
 *    Front ends when parsing from memory     *
12688
 *                  *
12689
 ************************************************************************/
12690
12691
/**
12692
 * Create a parser context for an XML in-memory document. The input buffer
12693
 * must not contain a terminating null byte.
12694
 *
12695
 * @param buffer  a pointer to a char array
12696
 * @param size  the size of the array
12697
 * @returns the new parser context or NULL
12698
 */
12699
xmlParserCtxt *
12700
0
xmlCreateMemoryParserCtxt(const char *buffer, int size) {
12701
0
    xmlParserCtxtPtr ctxt;
12702
0
    xmlParserInputPtr input;
12703
12704
0
    if (size < 0)
12705
0
  return(NULL);
12706
12707
0
    ctxt = xmlNewParserCtxt();
12708
0
    if (ctxt == NULL)
12709
0
  return(NULL);
12710
12711
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0);
12712
0
    if (input == NULL) {
12713
0
  xmlFreeParserCtxt(ctxt);
12714
0
  return(NULL);
12715
0
    }
12716
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12717
0
        xmlFreeInputStream(input);
12718
0
        xmlFreeParserCtxt(ctxt);
12719
0
        return(NULL);
12720
0
    }
12721
12722
0
    return(ctxt);
12723
0
}
12724
12725
#ifdef LIBXML_SAX1_ENABLED
12726
/**
12727
 * Parse an XML in-memory block and use the given SAX function block
12728
 * to handle the parsing callback. If sax is NULL, fallback to the default
12729
 * DOM tree building routines.
12730
 *
12731
 * This function uses deprecated global variables to set parser options
12732
 * which default to XML_PARSE_NODICT.
12733
 *
12734
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12735
 *
12736
 * User data (void *) is stored within the parser context in the
12737
 * context's _private member, so it is available nearly everywhere in libxml
12738
 *
12739
 * @param sax  the SAX handler block
12740
 * @param buffer  an pointer to a char array
12741
 * @param size  the size of the array
12742
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12743
 *             documents
12744
 * @param data  the userdata
12745
 * @returns the resulting document tree
12746
 */
12747
12748
xmlDoc *
12749
xmlSAXParseMemoryWithData(xmlSAXHandler *sax, const char *buffer,
12750
0
                          int size, int recovery, void *data) {
12751
0
    xmlDocPtr ret = NULL;
12752
0
    xmlParserCtxtPtr ctxt;
12753
0
    xmlParserInputPtr input;
12754
12755
0
    if (size < 0)
12756
0
        return(NULL);
12757
12758
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12759
0
    if (ctxt == NULL)
12760
0
        return(NULL);
12761
12762
0
    if (data != NULL)
12763
0
  ctxt->_private=data;
12764
12765
0
    if (recovery) {
12766
0
        ctxt->options |= XML_PARSE_RECOVER;
12767
0
        ctxt->recovery = 1;
12768
0
    }
12769
12770
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL,
12771
0
                                      XML_INPUT_BUF_STATIC);
12772
12773
0
    if (input != NULL)
12774
0
        ret = xmlCtxtParseDocument(ctxt, input);
12775
12776
0
    xmlFreeParserCtxt(ctxt);
12777
0
    return(ret);
12778
0
}
12779
12780
/**
12781
 * Parse an XML in-memory block and use the given SAX function block
12782
 * to handle the parsing callback. If sax is NULL, fallback to the default
12783
 * DOM tree building routines.
12784
 *
12785
 * This function uses deprecated global variables to set parser options
12786
 * which default to XML_PARSE_NODICT.
12787
 *
12788
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12789
 *
12790
 * @param sax  the SAX handler block
12791
 * @param buffer  an pointer to a char array
12792
 * @param size  the size of the array
12793
 * @param recovery  work in recovery mode, i.e. tries to read not Well Formed
12794
 *             documents
12795
 * @returns the resulting document tree
12796
 */
12797
xmlDoc *
12798
xmlSAXParseMemory(xmlSAXHandler *sax, const char *buffer,
12799
0
            int size, int recovery) {
12800
0
    return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL);
12801
0
}
12802
12803
/**
12804
 * Parse an XML in-memory block and build a tree.
12805
 *
12806
 * This function uses deprecated global variables to set parser options
12807
 * which default to XML_PARSE_NODICT.
12808
 *
12809
 * @deprecated Use #xmlReadMemory.
12810
 *
12811
 * @param buffer  an pointer to a char array
12812
 * @param size  the size of the array
12813
 * @returns the resulting document tree
12814
 */
12815
12816
0
xmlDoc *xmlParseMemory(const char *buffer, int size) {
12817
0
   return(xmlSAXParseMemory(NULL, buffer, size, 0));
12818
0
}
12819
12820
/**
12821
 * Parse an XML in-memory block and build a tree.
12822
 * In the case the document is not Well Formed, an attempt to
12823
 * build a tree is tried anyway
12824
 *
12825
 * This function uses deprecated global variables to set parser options
12826
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12827
 *
12828
 * @deprecated Use #xmlReadMemory with XML_PARSE_RECOVER.
12829
 *
12830
 * @param buffer  an pointer to a char array
12831
 * @param size  the size of the array
12832
 * @returns the resulting document tree or NULL in case of error
12833
 */
12834
12835
0
xmlDoc *xmlRecoverMemory(const char *buffer, int size) {
12836
0
   return(xmlSAXParseMemory(NULL, buffer, size, 1));
12837
0
}
12838
12839
/**
12840
 * Parse an XML in-memory buffer and call the given SAX handler routines.
12841
 *
12842
 * This function uses deprecated global variables to set parser options
12843
 * which default to XML_PARSE_NODICT.
12844
 *
12845
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12846
 *
12847
 * @param sax  a SAX handler
12848
 * @param user_data  The user data returned on SAX callbacks
12849
 * @param buffer  an in-memory XML document input
12850
 * @param size  the length of the XML document in bytes
12851
 * @returns 0 in case of success or a error number otherwise
12852
 */
12853
int xmlSAXUserParseMemory(xmlSAXHandler *sax, void *user_data,
12854
0
        const char *buffer, int size) {
12855
0
    int ret = 0;
12856
0
    xmlParserCtxtPtr ctxt;
12857
12858
0
    ctxt = xmlCreateMemoryParserCtxt(buffer, size);
12859
0
    if (ctxt == NULL) return -1;
12860
0
    if (sax != NULL) {
12861
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12862
0
            *ctxt->sax = *sax;
12863
0
        } else {
12864
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12865
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12866
0
        }
12867
0
  ctxt->userData = user_data;
12868
0
    }
12869
12870
0
    xmlParseDocument(ctxt);
12871
12872
0
    if (ctxt->wellFormed)
12873
0
  ret = 0;
12874
0
    else {
12875
0
        if (ctxt->errNo != 0)
12876
0
      ret = ctxt->errNo;
12877
0
  else
12878
0
      ret = -1;
12879
0
    }
12880
0
    if (ctxt->myDoc != NULL) {
12881
0
        xmlFreeDoc(ctxt->myDoc);
12882
0
  ctxt->myDoc = NULL;
12883
0
    }
12884
0
    xmlFreeParserCtxt(ctxt);
12885
12886
0
    return ret;
12887
0
}
12888
#endif /* LIBXML_SAX1_ENABLED */
12889
12890
/**
12891
 * Creates a parser context for an XML in-memory document.
12892
 *
12893
 * @param str  a pointer to an array of xmlChar
12894
 * @returns the new parser context or NULL
12895
 */
12896
xmlParserCtxt *
12897
0
xmlCreateDocParserCtxt(const xmlChar *str) {
12898
0
    xmlParserCtxtPtr ctxt;
12899
0
    xmlParserInputPtr input;
12900
12901
0
    ctxt = xmlNewParserCtxt();
12902
0
    if (ctxt == NULL)
12903
0
  return(NULL);
12904
12905
0
    input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0);
12906
0
    if (input == NULL) {
12907
0
  xmlFreeParserCtxt(ctxt);
12908
0
  return(NULL);
12909
0
    }
12910
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12911
0
        xmlFreeInputStream(input);
12912
0
        xmlFreeParserCtxt(ctxt);
12913
0
        return(NULL);
12914
0
    }
12915
12916
0
    return(ctxt);
12917
0
}
12918
12919
#ifdef LIBXML_SAX1_ENABLED
12920
/**
12921
 * Parse an XML in-memory document and build a tree.
12922
 * It use the given SAX function block to handle the parsing callback.
12923
 * If sax is NULL, fallback to the default DOM tree building routines.
12924
 *
12925
 * This function uses deprecated global variables to set parser options
12926
 * which default to XML_PARSE_NODICT.
12927
 *
12928
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadDoc.
12929
 *
12930
 * @param sax  the SAX handler block
12931
 * @param cur  a pointer to an array of xmlChar
12932
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12933
 *             documents
12934
 * @returns the resulting document tree
12935
 */
12936
12937
xmlDoc *
12938
0
xmlSAXParseDoc(xmlSAXHandler *sax, const xmlChar *cur, int recovery) {
12939
0
    xmlDocPtr ret;
12940
0
    xmlParserCtxtPtr ctxt;
12941
0
    xmlSAXHandlerPtr oldsax = NULL;
12942
12943
0
    if (cur == NULL) return(NULL);
12944
12945
12946
0
    ctxt = xmlCreateDocParserCtxt(cur);
12947
0
    if (ctxt == NULL) return(NULL);
12948
0
    if (sax != NULL) {
12949
0
        oldsax = ctxt->sax;
12950
0
        ctxt->sax = sax;
12951
0
        ctxt->userData = NULL;
12952
0
    }
12953
12954
0
    xmlParseDocument(ctxt);
12955
0
    if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc;
12956
0
    else {
12957
0
       ret = NULL;
12958
0
       xmlFreeDoc(ctxt->myDoc);
12959
0
       ctxt->myDoc = NULL;
12960
0
    }
12961
0
    if (sax != NULL)
12962
0
  ctxt->sax = oldsax;
12963
0
    xmlFreeParserCtxt(ctxt);
12964
12965
0
    return(ret);
12966
0
}
12967
12968
/**
12969
 * Parse an XML in-memory document and build a tree.
12970
 *
12971
 * This function uses deprecated global variables to set parser options
12972
 * which default to XML_PARSE_NODICT.
12973
 *
12974
 * @deprecated Use #xmlReadDoc.
12975
 *
12976
 * @param cur  a pointer to an array of xmlChar
12977
 * @returns the resulting document tree
12978
 */
12979
12980
xmlDoc *
12981
0
xmlParseDoc(const xmlChar *cur) {
12982
0
    return(xmlSAXParseDoc(NULL, cur, 0));
12983
0
}
12984
#endif /* LIBXML_SAX1_ENABLED */
12985
12986
/************************************************************************
12987
 *                  *
12988
 *  New set (2.6.0) of simpler and more flexible APIs   *
12989
 *                  *
12990
 ************************************************************************/
12991
12992
/**
12993
 * Reset a parser context
12994
 *
12995
 * @param ctxt  an XML parser context
12996
 */
12997
void
12998
xmlCtxtReset(xmlParserCtxt *ctxt)
12999
0
{
13000
0
    xmlParserInputPtr input;
13001
13002
0
    if (ctxt == NULL)
13003
0
        return;
13004
13005
0
    while ((input = xmlCtxtPopInput(ctxt)) != NULL) { /* Non consuming */
13006
0
        xmlFreeInputStream(input);
13007
0
    }
13008
0
    ctxt->inputNr = 0;
13009
0
    ctxt->input = NULL;
13010
13011
0
    ctxt->spaceNr = 0;
13012
0
    if (ctxt->spaceTab != NULL) {
13013
0
  ctxt->spaceTab[0] = -1;
13014
0
  ctxt->space = &ctxt->spaceTab[0];
13015
0
    } else {
13016
0
        ctxt->space = NULL;
13017
0
    }
13018
13019
13020
0
    ctxt->nodeNr = 0;
13021
0
    ctxt->node = NULL;
13022
13023
0
    ctxt->nameNr = 0;
13024
0
    ctxt->name = NULL;
13025
13026
0
    ctxt->nsNr = 0;
13027
0
    xmlParserNsReset(ctxt->nsdb);
13028
13029
0
    if (ctxt->version != NULL) {
13030
0
        xmlFree(ctxt->version);
13031
0
        ctxt->version = NULL;
13032
0
    }
13033
0
    if (ctxt->encoding != NULL) {
13034
0
        xmlFree(ctxt->encoding);
13035
0
        ctxt->encoding = NULL;
13036
0
    }
13037
0
    if (ctxt->extSubURI != NULL) {
13038
0
        xmlFree(ctxt->extSubURI);
13039
0
        ctxt->extSubURI = NULL;
13040
0
    }
13041
0
    if (ctxt->extSubSystem != NULL) {
13042
0
        xmlFree(ctxt->extSubSystem);
13043
0
        ctxt->extSubSystem = NULL;
13044
0
    }
13045
0
    if (ctxt->directory != NULL) {
13046
0
        xmlFree(ctxt->directory);
13047
0
        ctxt->directory = NULL;
13048
0
    }
13049
13050
0
    if (ctxt->myDoc != NULL)
13051
0
        xmlFreeDoc(ctxt->myDoc);
13052
0
    ctxt->myDoc = NULL;
13053
13054
0
    ctxt->standalone = -1;
13055
0
    ctxt->hasExternalSubset = 0;
13056
0
    ctxt->hasPErefs = 0;
13057
0
    ctxt->html = ctxt->html ? 1 : 0;
13058
0
    ctxt->instate = XML_PARSER_START;
13059
13060
0
    ctxt->wellFormed = 1;
13061
0
    ctxt->nsWellFormed = 1;
13062
0
    ctxt->disableSAX = 0;
13063
0
    ctxt->valid = 1;
13064
0
    ctxt->record_info = 0;
13065
0
    ctxt->checkIndex = 0;
13066
0
    ctxt->endCheckState = 0;
13067
0
    ctxt->inSubset = 0;
13068
0
    ctxt->errNo = XML_ERR_OK;
13069
0
    ctxt->depth = 0;
13070
0
    ctxt->catalogs = NULL;
13071
0
    ctxt->sizeentities = 0;
13072
0
    ctxt->sizeentcopy = 0;
13073
0
    xmlInitNodeInfoSeq(&ctxt->node_seq);
13074
13075
0
    if (ctxt->attsDefault != NULL) {
13076
0
        xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator);
13077
0
        ctxt->attsDefault = NULL;
13078
0
    }
13079
0
    if (ctxt->attsSpecial != NULL) {
13080
0
        xmlHashFree(ctxt->attsSpecial, NULL);
13081
0
        ctxt->attsSpecial = NULL;
13082
0
    }
13083
13084
0
#ifdef LIBXML_CATALOG_ENABLED
13085
0
    if (ctxt->catalogs != NULL)
13086
0
  xmlCatalogFreeLocal(ctxt->catalogs);
13087
0
#endif
13088
0
    ctxt->nbErrors = 0;
13089
0
    ctxt->nbWarnings = 0;
13090
0
    if (ctxt->lastError.code != XML_ERR_OK)
13091
0
        xmlResetError(&ctxt->lastError);
13092
0
}
13093
13094
/**
13095
 * Reset a push parser context
13096
 *
13097
 * @param ctxt  an XML parser context
13098
 * @param chunk  a pointer to an array of chars
13099
 * @param size  number of chars in the array
13100
 * @param filename  an optional file name or URI
13101
 * @param encoding  the document encoding, or NULL
13102
 * @returns 0 in case of success and 1 in case of error
13103
 */
13104
int
13105
xmlCtxtResetPush(xmlParserCtxt *ctxt, const char *chunk,
13106
                 int size, const char *filename, const char *encoding)
13107
0
{
13108
0
    xmlParserInputPtr input;
13109
13110
0
    if (ctxt == NULL)
13111
0
        return(1);
13112
13113
0
    xmlCtxtReset(ctxt);
13114
13115
0
    input = xmlNewPushInput(filename, chunk, size);
13116
0
    if (input == NULL)
13117
0
        return(1);
13118
13119
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13120
0
        xmlFreeInputStream(input);
13121
0
        return(1);
13122
0
    }
13123
13124
0
    if (encoding != NULL)
13125
0
        xmlSwitchEncodingName(ctxt, encoding);
13126
13127
0
    return(0);
13128
0
}
13129
13130
static int
13131
xmlCtxtSetOptionsInternal(xmlParserCtxtPtr ctxt, int options, int keepMask)
13132
0
{
13133
0
    int allMask;
13134
13135
0
    if (ctxt == NULL)
13136
0
        return(-1);
13137
13138
    /*
13139
     * XInclude options aren't handled by the parser.
13140
     *
13141
     * XML_PARSE_XINCLUDE
13142
     * XML_PARSE_NOXINCNODE
13143
     * XML_PARSE_NOBASEFIX
13144
     */
13145
0
    allMask = XML_PARSE_RECOVER |
13146
0
              XML_PARSE_NOENT |
13147
0
              XML_PARSE_DTDLOAD |
13148
0
              XML_PARSE_DTDATTR |
13149
0
              XML_PARSE_DTDVALID |
13150
0
              XML_PARSE_NOERROR |
13151
0
              XML_PARSE_NOWARNING |
13152
0
              XML_PARSE_PEDANTIC |
13153
0
              XML_PARSE_NOBLANKS |
13154
0
#ifdef LIBXML_SAX1_ENABLED
13155
0
              XML_PARSE_SAX1 |
13156
0
#endif
13157
0
              XML_PARSE_NONET |
13158
0
              XML_PARSE_NODICT |
13159
0
              XML_PARSE_NSCLEAN |
13160
0
              XML_PARSE_NOCDATA |
13161
0
              XML_PARSE_COMPACT |
13162
0
              XML_PARSE_OLD10 |
13163
0
              XML_PARSE_HUGE |
13164
0
              XML_PARSE_OLDSAX |
13165
0
              XML_PARSE_IGNORE_ENC |
13166
0
              XML_PARSE_BIG_LINES |
13167
0
              XML_PARSE_NO_XXE |
13168
0
              XML_PARSE_UNZIP |
13169
0
              XML_PARSE_NO_SYS_CATALOG |
13170
0
              XML_PARSE_CATALOG_PI;
13171
13172
0
    ctxt->options = (ctxt->options & keepMask) | (options & allMask);
13173
13174
    /*
13175
     * For some options, struct members are historically the source
13176
     * of truth. The values are initalized from global variables and
13177
     * old code could also modify them directly. Several older API
13178
     * functions that don't take an options argument rely on these
13179
     * deprecated mechanisms.
13180
     *
13181
     * Once public access to struct members and the globals are
13182
     * disabled, we can use the options bitmask as source of
13183
     * truth, making all these struct members obsolete.
13184
     *
13185
     * The XML_DETECT_IDS flags is misnamed. It simply enables
13186
     * loading of the external subset.
13187
     */
13188
0
    ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0;
13189
0
    ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0;
13190
0
    ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0;
13191
0
    ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0;
13192
0
    ctxt->loadsubset |= (options & XML_PARSE_SKIP_IDS) ? XML_SKIP_IDS : 0;
13193
0
    ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0;
13194
0
    ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0;
13195
0
    ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1;
13196
0
    ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1;
13197
13198
0
    return(options & ~allMask);
13199
0
}
13200
13201
/**
13202
 * Applies the options to the parser context. Unset options are
13203
 * cleared.
13204
 *
13205
 * @since 2.13.0
13206
 *
13207
 * With older versions, you can use #xmlCtxtUseOptions.
13208
 *
13209
 * @param ctxt  an XML parser context
13210
 * @param options  a bitmask of xmlParserOption values
13211
 * @returns 0 in case of success, the set of unknown or unimplemented options
13212
 *         in case of error.
13213
 */
13214
int
13215
xmlCtxtSetOptions(xmlParserCtxt *ctxt, int options)
13216
0
{
13217
0
#ifdef LIBXML_HTML_ENABLED
13218
0
    if ((ctxt != NULL) && (ctxt->html))
13219
0
        return(htmlCtxtSetOptions(ctxt, options));
13220
0
#endif
13221
13222
0
    return(xmlCtxtSetOptionsInternal(ctxt, options, 0));
13223
0
}
13224
13225
/**
13226
 * Get the current options of the parser context.
13227
 *
13228
 * @since 2.14.0
13229
 *
13230
 * @param ctxt  an XML parser context
13231
 * @returns the current options set in the parser context, or -1 if ctxt is NULL.
13232
 */
13233
int
13234
xmlCtxtGetOptions(xmlParserCtxt *ctxt)
13235
0
{
13236
0
    if (ctxt == NULL)
13237
0
        return(-1);
13238
13239
0
    return(ctxt->options);
13240
0
}
13241
13242
/**
13243
 * Applies the options to the parser context. The following options
13244
 * are never cleared and can only be enabled:
13245
 *
13246
 * - XML_PARSE_NOERROR
13247
 * - XML_PARSE_NOWARNING
13248
 * - XML_PARSE_NONET
13249
 * - XML_PARSE_NSCLEAN
13250
 * - XML_PARSE_NOCDATA
13251
 * - XML_PARSE_COMPACT
13252
 * - XML_PARSE_OLD10
13253
 * - XML_PARSE_HUGE
13254
 * - XML_PARSE_OLDSAX
13255
 * - XML_PARSE_IGNORE_ENC
13256
 * - XML_PARSE_BIG_LINES
13257
 *
13258
 * @deprecated Use #xmlCtxtSetOptions.
13259
 *
13260
 * @param ctxt  an XML parser context
13261
 * @param options  a combination of xmlParserOption
13262
 * @returns 0 in case of success, the set of unknown or unimplemented options
13263
 *         in case of error.
13264
 */
13265
int
13266
xmlCtxtUseOptions(xmlParserCtxt *ctxt, int options)
13267
0
{
13268
0
    int keepMask;
13269
13270
0
#ifdef LIBXML_HTML_ENABLED
13271
0
    if ((ctxt != NULL) && (ctxt->html))
13272
0
        return(htmlCtxtUseOptions(ctxt, options));
13273
0
#endif
13274
13275
    /*
13276
     * For historic reasons, some options can only be enabled.
13277
     */
13278
0
    keepMask = XML_PARSE_NOERROR |
13279
0
               XML_PARSE_NOWARNING |
13280
0
               XML_PARSE_NONET |
13281
0
               XML_PARSE_NSCLEAN |
13282
0
               XML_PARSE_NOCDATA |
13283
0
               XML_PARSE_COMPACT |
13284
0
               XML_PARSE_OLD10 |
13285
0
               XML_PARSE_HUGE |
13286
0
               XML_PARSE_OLDSAX |
13287
0
               XML_PARSE_IGNORE_ENC |
13288
0
               XML_PARSE_BIG_LINES;
13289
13290
0
    return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask));
13291
0
}
13292
13293
/**
13294
 * To protect against exponential entity expansion ("billion laughs"), the
13295
 * size of serialized output is (roughly) limited to the input size
13296
 * multiplied by this factor. The default value is 5.
13297
 *
13298
 * When working with documents making heavy use of entity expansion, it can
13299
 * be necessary to increase the value. For security reasons, this should only
13300
 * be considered when processing trusted input.
13301
 *
13302
 * @param ctxt  an XML parser context
13303
 * @param maxAmpl  maximum amplification factor
13304
 */
13305
void
13306
xmlCtxtSetMaxAmplification(xmlParserCtxt *ctxt, unsigned maxAmpl)
13307
0
{
13308
0
    if (ctxt == NULL)
13309
0
        return;
13310
0
    ctxt->maxAmpl = maxAmpl;
13311
0
}
13312
13313
/**
13314
 * Parse an XML document and return the resulting document tree.
13315
 * Takes ownership of the input object.
13316
 *
13317
 * @since 2.13.0
13318
 *
13319
 * @param ctxt  an XML parser context
13320
 * @param input  parser input
13321
 * @returns the resulting document tree or NULL
13322
 */
13323
xmlDoc *
13324
xmlCtxtParseDocument(xmlParserCtxt *ctxt, xmlParserInput *input)
13325
0
{
13326
0
    xmlDocPtr ret = NULL;
13327
13328
0
    if ((ctxt == NULL) || (input == NULL)) {
13329
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
13330
0
        xmlFreeInputStream(input);
13331
0
        return(NULL);
13332
0
    }
13333
13334
    /* assert(ctxt->inputNr == 0); */
13335
0
    while (ctxt->inputNr > 0)
13336
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13337
13338
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13339
0
        xmlFreeInputStream(input);
13340
0
        return(NULL);
13341
0
    }
13342
13343
0
    xmlParseDocument(ctxt);
13344
13345
0
    ret = xmlCtxtGetDocument(ctxt);
13346
13347
    /* assert(ctxt->inputNr == 1); */
13348
0
    while (ctxt->inputNr > 0)
13349
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13350
13351
0
    return(ret);
13352
0
}
13353
13354
/**
13355
 * Convenience function to parse an XML document from a
13356
 * zero-terminated string.
13357
 *
13358
 * See #xmlCtxtReadDoc for details.
13359
 *
13360
 * @param cur  a pointer to a zero terminated string
13361
 * @param URL  base URL (optional)
13362
 * @param encoding  the document encoding (optional)
13363
 * @param options  a combination of xmlParserOption
13364
 * @returns the resulting document tree
13365
 */
13366
xmlDoc *
13367
xmlReadDoc(const xmlChar *cur, const char *URL, const char *encoding,
13368
           int options)
13369
0
{
13370
0
    xmlParserCtxtPtr ctxt;
13371
0
    xmlParserInputPtr input;
13372
0
    xmlDocPtr doc = NULL;
13373
13374
0
    ctxt = xmlNewParserCtxt();
13375
0
    if (ctxt == NULL)
13376
0
        return(NULL);
13377
13378
0
    xmlCtxtUseOptions(ctxt, options);
13379
13380
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding,
13381
0
                                      XML_INPUT_BUF_STATIC);
13382
13383
0
    if (input != NULL)
13384
0
        doc = xmlCtxtParseDocument(ctxt, input);
13385
13386
0
    xmlFreeParserCtxt(ctxt);
13387
0
    return(doc);
13388
0
}
13389
13390
/**
13391
 * Convenience function to parse an XML file from the filesystem
13392
 * or a global, user-defined resource loader.
13393
 *
13394
 * If a "-" filename is passed, the function will read from stdin.
13395
 * This feature is potentially insecure and might be removed from
13396
 * later versions.
13397
 *
13398
 * See #xmlCtxtReadFile for details.
13399
 *
13400
 * @param filename  a file or URL
13401
 * @param encoding  the document encoding (optional)
13402
 * @param options  a combination of xmlParserOption
13403
 * @returns the resulting document tree
13404
 */
13405
xmlDoc *
13406
xmlReadFile(const char *filename, const char *encoding, int options)
13407
0
{
13408
0
    xmlParserCtxtPtr ctxt;
13409
0
    xmlParserInputPtr input;
13410
0
    xmlDocPtr doc = NULL;
13411
13412
0
    ctxt = xmlNewParserCtxt();
13413
0
    if (ctxt == NULL)
13414
0
        return(NULL);
13415
13416
0
    xmlCtxtUseOptions(ctxt, options);
13417
13418
    /*
13419
     * Backward compatibility for users of command line utilities like
13420
     * xmlstarlet expecting "-" to mean stdin. This is dangerous and
13421
     * should be removed at some point.
13422
     */
13423
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
13424
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO,
13425
0
                                      encoding, 0);
13426
0
    else
13427
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13428
13429
0
    if (input != NULL)
13430
0
        doc = xmlCtxtParseDocument(ctxt, input);
13431
13432
0
    xmlFreeParserCtxt(ctxt);
13433
0
    return(doc);
13434
0
}
13435
13436
/**
13437
 * Parse an XML in-memory document and build a tree. The input buffer must
13438
 * not contain a terminating null byte.
13439
 *
13440
 * See #xmlCtxtReadMemory for details.
13441
 *
13442
 * @param buffer  a pointer to a char array
13443
 * @param size  the size of the array
13444
 * @param url  base URL (optional)
13445
 * @param encoding  the document encoding (optional)
13446
 * @param options  a combination of xmlParserOption
13447
 * @returns the resulting document tree
13448
 */
13449
xmlDoc *
13450
xmlReadMemory(const char *buffer, int size, const char *url,
13451
              const char *encoding, int options)
13452
0
{
13453
0
    xmlParserCtxtPtr ctxt;
13454
0
    xmlParserInputPtr input;
13455
0
    xmlDocPtr doc = NULL;
13456
13457
0
    if (size < 0)
13458
0
  return(NULL);
13459
13460
0
    ctxt = xmlNewParserCtxt();
13461
0
    if (ctxt == NULL)
13462
0
        return(NULL);
13463
13464
0
    xmlCtxtUseOptions(ctxt, options);
13465
13466
0
    input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding,
13467
0
                                      XML_INPUT_BUF_STATIC);
13468
13469
0
    if (input != NULL)
13470
0
        doc = xmlCtxtParseDocument(ctxt, input);
13471
13472
0
    xmlFreeParserCtxt(ctxt);
13473
0
    return(doc);
13474
0
}
13475
13476
/**
13477
 * Parse an XML from a file descriptor and build a tree.
13478
 *
13479
 * See #xmlCtxtReadFd for details.
13480
 *
13481
 * NOTE that the file descriptor will not be closed when the
13482
 * context is freed or reset.
13483
 *
13484
 * @param fd  an open file descriptor
13485
 * @param URL  base URL (optional)
13486
 * @param encoding  the document encoding (optional)
13487
 * @param options  a combination of xmlParserOption
13488
 * @returns the resulting document tree
13489
 */
13490
xmlDoc *
13491
xmlReadFd(int fd, const char *URL, const char *encoding, int options)
13492
0
{
13493
0
    xmlParserCtxtPtr ctxt;
13494
0
    xmlParserInputPtr input;
13495
0
    xmlDocPtr doc = NULL;
13496
13497
0
    ctxt = xmlNewParserCtxt();
13498
0
    if (ctxt == NULL)
13499
0
        return(NULL);
13500
13501
0
    xmlCtxtUseOptions(ctxt, options);
13502
13503
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13504
13505
0
    if (input != NULL)
13506
0
        doc = xmlCtxtParseDocument(ctxt, input);
13507
13508
0
    xmlFreeParserCtxt(ctxt);
13509
0
    return(doc);
13510
0
}
13511
13512
/**
13513
 * Parse an XML document from I/O functions and context and build a tree.
13514
 *
13515
 * See #xmlCtxtReadIO for details.
13516
 *
13517
 * @param ioread  an I/O read function
13518
 * @param ioclose  an I/O close function (optional)
13519
 * @param ioctx  an I/O handler
13520
 * @param URL  base URL (optional)
13521
 * @param encoding  the document encoding (optional)
13522
 * @param options  a combination of xmlParserOption
13523
 * @returns the resulting document tree
13524
 */
13525
xmlDoc *
13526
xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose,
13527
          void *ioctx, const char *URL, const char *encoding, int options)
13528
0
{
13529
0
    xmlParserCtxtPtr ctxt;
13530
0
    xmlParserInputPtr input;
13531
0
    xmlDocPtr doc = NULL;
13532
13533
0
    ctxt = xmlNewParserCtxt();
13534
0
    if (ctxt == NULL)
13535
0
        return(NULL);
13536
13537
0
    xmlCtxtUseOptions(ctxt, options);
13538
13539
0
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13540
0
                                  encoding, 0);
13541
13542
0
    if (input != NULL)
13543
0
        doc = xmlCtxtParseDocument(ctxt, input);
13544
13545
0
    xmlFreeParserCtxt(ctxt);
13546
0
    return(doc);
13547
0
}
13548
13549
/**
13550
 * Parse an XML in-memory document and build a tree.
13551
 *
13552
 * `URL` is used as base to resolve external entities and for error
13553
 * reporting.
13554
 *
13555
 * @param ctxt  an XML parser context
13556
 * @param str  a pointer to a zero terminated string
13557
 * @param URL  base URL (optional)
13558
 * @param encoding  the document encoding (optional)
13559
 * @param options  a combination of xmlParserOption
13560
 * @returns the resulting document tree
13561
 */
13562
xmlDoc *
13563
xmlCtxtReadDoc(xmlParserCtxt *ctxt, const xmlChar *str,
13564
               const char *URL, const char *encoding, int options)
13565
0
{
13566
0
    xmlParserInputPtr input;
13567
13568
0
    if (ctxt == NULL)
13569
0
        return(NULL);
13570
13571
0
    xmlCtxtReset(ctxt);
13572
0
    xmlCtxtUseOptions(ctxt, options);
13573
13574
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) str, encoding,
13575
0
                                      XML_INPUT_BUF_STATIC);
13576
0
    if (input == NULL)
13577
0
        return(NULL);
13578
13579
0
    return(xmlCtxtParseDocument(ctxt, input));
13580
0
}
13581
13582
/**
13583
 * Parse an XML file from the filesystem or a global, user-defined
13584
 * resource loader.
13585
 *
13586
 * @param ctxt  an XML parser context
13587
 * @param filename  a file or URL
13588
 * @param encoding  the document encoding (optional)
13589
 * @param options  a combination of xmlParserOption
13590
 * @returns the resulting document tree
13591
 */
13592
xmlDoc *
13593
xmlCtxtReadFile(xmlParserCtxt *ctxt, const char *filename,
13594
                const char *encoding, int options)
13595
0
{
13596
0
    xmlParserInputPtr input;
13597
13598
0
    if (ctxt == NULL)
13599
0
        return(NULL);
13600
13601
0
    xmlCtxtReset(ctxt);
13602
0
    xmlCtxtUseOptions(ctxt, options);
13603
13604
0
    input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13605
0
    if (input == NULL)
13606
0
        return(NULL);
13607
13608
0
    return(xmlCtxtParseDocument(ctxt, input));
13609
0
}
13610
13611
/**
13612
 * Parse an XML in-memory document and build a tree. The input buffer must
13613
 * not contain a terminating null byte.
13614
 *
13615
 * `URL` is used as base to resolve external entities and for error
13616
 * reporting.
13617
 *
13618
 * @param ctxt  an XML parser context
13619
 * @param buffer  a pointer to a char array
13620
 * @param size  the size of the array
13621
 * @param URL  base URL (optional)
13622
 * @param encoding  the document encoding (optional)
13623
 * @param options  a combination of xmlParserOption
13624
 * @returns the resulting document tree
13625
 */
13626
xmlDoc *
13627
xmlCtxtReadMemory(xmlParserCtxt *ctxt, const char *buffer, int size,
13628
                  const char *URL, const char *encoding, int options)
13629
0
{
13630
0
    xmlParserInputPtr input;
13631
13632
0
    if ((ctxt == NULL) || (size < 0))
13633
0
        return(NULL);
13634
13635
0
    xmlCtxtReset(ctxt);
13636
0
    xmlCtxtUseOptions(ctxt, options);
13637
13638
0
    input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding,
13639
0
                                      XML_INPUT_BUF_STATIC);
13640
0
    if (input == NULL)
13641
0
        return(NULL);
13642
13643
0
    return(xmlCtxtParseDocument(ctxt, input));
13644
0
}
13645
13646
/**
13647
 * Parse an XML document from a file descriptor and build a tree.
13648
 *
13649
 * NOTE that the file descriptor will not be closed when the
13650
 * context is freed or reset.
13651
 *
13652
 * `URL` is used as base to resolve external entities and for error
13653
 * reporting.
13654
 *
13655
 * @param ctxt  an XML parser context
13656
 * @param fd  an open file descriptor
13657
 * @param URL  base URL (optional)
13658
 * @param encoding  the document encoding (optional)
13659
 * @param options  a combination of xmlParserOption
13660
 * @returns the resulting document tree
13661
 */
13662
xmlDoc *
13663
xmlCtxtReadFd(xmlParserCtxt *ctxt, int fd,
13664
              const char *URL, const char *encoding, int options)
13665
0
{
13666
0
    xmlParserInputPtr input;
13667
13668
0
    if (ctxt == NULL)
13669
0
        return(NULL);
13670
13671
0
    xmlCtxtReset(ctxt);
13672
0
    xmlCtxtUseOptions(ctxt, options);
13673
13674
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13675
0
    if (input == NULL)
13676
0
        return(NULL);
13677
13678
0
    return(xmlCtxtParseDocument(ctxt, input));
13679
0
}
13680
13681
/**
13682
 * Parse an XML document from I/O functions and source and build a tree.
13683
 * This reuses the existing `ctxt` parser context
13684
 *
13685
 * `URL` is used as base to resolve external entities and for error
13686
 * reporting.
13687
 *
13688
 * @param ctxt  an XML parser context
13689
 * @param ioread  an I/O read function
13690
 * @param ioclose  an I/O close function
13691
 * @param ioctx  an I/O handler
13692
 * @param URL  the base URL to use for the document
13693
 * @param encoding  the document encoding, or NULL
13694
 * @param options  a combination of xmlParserOption
13695
 * @returns the resulting document tree
13696
 */
13697
xmlDoc *
13698
xmlCtxtReadIO(xmlParserCtxt *ctxt, xmlInputReadCallback ioread,
13699
              xmlInputCloseCallback ioclose, void *ioctx,
13700
        const char *URL,
13701
              const char *encoding, int options)
13702
0
{
13703
0
    xmlParserInputPtr input;
13704
13705
0
    if (ctxt == NULL)
13706
0
        return(NULL);
13707
13708
0
    xmlCtxtReset(ctxt);
13709
0
    xmlCtxtUseOptions(ctxt, options);
13710
13711
0
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13712
0
                                  encoding, 0);
13713
0
    if (input == NULL)
13714
0
        return(NULL);
13715
13716
0
    return(xmlCtxtParseDocument(ctxt, input));
13717
0
}
13718