Coverage Report

Created: 2025-10-13 06:08

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/libxml2/parser.c
Line
Count
Source
1
/*
2
 * parser.c : an XML 1.0 parser, namespaces and validity support are mostly
3
 *            implemented on top of the SAX interfaces
4
 *
5
 * References:
6
 *   The XML specification:
7
 *     http://www.w3.org/TR/REC-xml
8
 *   Original 1.0 version:
9
 *     http://www.w3.org/TR/1998/REC-xml-19980210
10
 *   XML second edition working draft
11
 *     http://www.w3.org/TR/2000/WD-xml-2e-20000814
12
 *
13
 * Okay this is a big file, the parser core is around 7000 lines, then it
14
 * is followed by the progressive parser top routines, then the various
15
 * high level APIs to call the parser and a few miscellaneous functions.
16
 * A number of helper functions and deprecated ones have been moved to
17
 * parserInternals.c to reduce this file size.
18
 * As much as possible the functions are associated with their relative
19
 * production in the XML specification. A few productions defining the
20
 * different ranges of character are actually implanted either in
21
 * parserInternals.h or parserInternals.c
22
 * The DOM tree build is realized from the default SAX callbacks in
23
 * the module SAX2.c.
24
 * The routines doing the validation checks are in valid.c and called either
25
 * from the SAX callbacks or as standalone functions using a preparsed
26
 * document.
27
 *
28
 * See Copyright for the status of this software.
29
 *
30
 * Author: Daniel Veillard
31
 */
32
33
/* To avoid EBCDIC trouble when parsing on zOS */
34
#if defined(__MVS__)
35
#pragma convert("ISO8859-1")
36
#endif
37
38
#define IN_LIBXML
39
#include "libxml.h"
40
41
#if defined(_WIN32)
42
#define XML_DIR_SEP '\\'
43
#else
44
#define XML_DIR_SEP '/'
45
#endif
46
47
#include <stdlib.h>
48
#include <limits.h>
49
#include <string.h>
50
#include <stdarg.h>
51
#include <stddef.h>
52
#include <ctype.h>
53
#include <stdlib.h>
54
#include <libxml/parser.h>
55
#include <libxml/xmlmemory.h>
56
#include <libxml/tree.h>
57
#include <libxml/parserInternals.h>
58
#include <libxml/valid.h>
59
#include <libxml/entities.h>
60
#include <libxml/xmlerror.h>
61
#include <libxml/encoding.h>
62
#include <libxml/xmlIO.h>
63
#include <libxml/uri.h>
64
#include <libxml/SAX2.h>
65
#include <libxml/HTMLparser.h>
66
#ifdef LIBXML_CATALOG_ENABLED
67
#include <libxml/catalog.h>
68
#endif
69
70
#include "private/buf.h"
71
#include "private/dict.h"
72
#include "private/entities.h"
73
#include "private/error.h"
74
#include "private/html.h"
75
#include "private/io.h"
76
#include "private/memory.h"
77
#include "private/parser.h"
78
#include "private/tree.h"
79
80
419k
#define NS_INDEX_EMPTY  INT_MAX
81
58.1k
#define NS_INDEX_XML    (INT_MAX - 1)
82
186k
#define URI_HASH_EMPTY  0xD943A04E
83
12.1k
#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
164k
#define XML_MAX_ATTRS 100000000 /* 100 million */
94
95
388k
#define XML_SPECIAL_EXTERNAL    (1 << 20)
96
338k
#define XML_SPECIAL_TYPE_MASK   (XML_SPECIAL_EXTERNAL - 1)
97
98
376k
#define XML_ATTVAL_ALLOC        (1 << 0)
99
1.02M
#define XML_ATTVAL_NORM_CHANGE  (1 << 1)
100
101
struct _xmlStartTag {
102
    const xmlChar *prefix;
103
    const xmlChar *URI;
104
    int line;
105
    int nsNr;
106
};
107
108
typedef struct {
109
    void *saxData;
110
    unsigned prefixHashValue;
111
    unsigned uriHashValue;
112
    unsigned elementId;
113
    int oldIndex;
114
} xmlParserNsExtra;
115
116
typedef struct {
117
    unsigned hashValue;
118
    int index;
119
} xmlParserNsBucket;
120
121
struct _xmlParserNsData {
122
    xmlParserNsExtra *extra;
123
124
    unsigned hashSize;
125
    unsigned hashElems;
126
    xmlParserNsBucket *hash;
127
128
    unsigned elementId;
129
    int defaultNsIndex;
130
    int minNsIndex;
131
};
132
133
static int
134
xmlParseElementStart(xmlParserCtxtPtr ctxt);
135
136
static void
137
xmlParseElementEnd(xmlParserCtxtPtr ctxt);
138
139
static xmlEntityPtr
140
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr);
141
142
static const xmlChar *
143
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt);
144
145
/************************************************************************
146
 *                  *
147
 *  Arbitrary limits set in the parser. See XML_PARSE_HUGE    *
148
 *                  *
149
 ************************************************************************/
150
151
#define XML_PARSER_BIG_ENTITY 1000
152
#define XML_PARSER_LOT_ENTITY 5000
153
154
/*
155
 * Constants for protection against abusive entity expansion
156
 * ("billion laughs").
157
 */
158
159
/*
160
 * A certain amount of entity expansion which is always allowed.
161
 */
162
1.47M
#define XML_PARSER_ALLOWED_EXPANSION 1000000
163
164
/*
165
 * Fixed cost for each entity reference. This crudely models processing time
166
 * as well to protect, for example, against exponential expansion of empty
167
 * or very short entities.
168
 */
169
1.48M
#define XML_ENT_FIXED_COST 20
170
171
82.2M
#define XML_PARSER_BIG_BUFFER_SIZE 300
172
589k
#define XML_PARSER_BUFFER_SIZE 100
173
74.2k
#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
2.15k
xmlErrMemory(xmlParserCtxtPtr ctxt) {
225
2.15k
    xmlCtxtErrMemory(ctxt);
226
2.15k
}
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
24.9k
{
239
24.9k
    if (prefix == NULL)
240
18.9k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
241
18.9k
                   XML_ERR_FATAL, localname, NULL, NULL, 0,
242
18.9k
                   "Attribute %s redefined\n", localname);
243
6.04k
    else
244
6.04k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
245
6.04k
                   XML_ERR_FATAL, prefix, localname, NULL, 0,
246
6.04k
                   "Attribute %s:%s redefined\n", prefix, localname);
247
24.9k
}
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
43.2M
{
260
43.2M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
261
43.2M
               NULL, NULL, NULL, 0, "%s", msg);
262
43.2M
}
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
17.0k
{
277
17.0k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING,
278
17.0k
               str1, str2, NULL, 0, msg, str1, str2);
279
17.0k
}
280
281
#ifdef LIBXML_VALID_ENABLED
282
/**
283
 * Handle a validity error.
284
 *
285
 * @param ctxt  an XML parser context
286
 * @param error  the error number
287
 * @param msg  the error message
288
 * @param str1  extra data
289
 * @param str2  extra data
290
 */
291
static void LIBXML_ATTR_FORMAT(3,0)
292
xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error,
293
              const char *msg, const xmlChar *str1, const xmlChar *str2)
294
0
{
295
0
    ctxt->valid = 0;
296
297
0
    xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR,
298
0
               str1, str2, NULL, 0, msg, str1, str2);
299
0
}
300
#endif
301
302
/**
303
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
304
 *
305
 * @param ctxt  an XML parser context
306
 * @param error  the error number
307
 * @param msg  the error message
308
 * @param val  an integer value
309
 */
310
static void LIBXML_ATTR_FORMAT(3,0)
311
xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error,
312
                  const char *msg, int val)
313
2.12M
{
314
2.12M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
315
2.12M
               NULL, NULL, NULL, val, msg, val);
316
2.12M
}
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
188k
{
333
188k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
334
188k
               str1, str2, NULL, val, msg, str1, val, str2);
335
188k
}
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
2.61M
{
349
2.61M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
350
2.61M
               val, NULL, NULL, 0, msg, val);
351
2.61M
}
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
74.6k
{
365
74.6k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR,
366
74.6k
               val, NULL, NULL, 0, msg, val);
367
74.6k
}
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
122k
{
385
122k
    ctxt->nsWellFormed = 0;
386
387
122k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR,
388
122k
               info1, info2, info3, 0, msg, info1, info2, info3);
389
122k
}
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
10.7k
{
407
10.7k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING,
408
10.7k
               info1, info2, info3, 0, msg, info1, info2, info3);
409
10.7k
}
410
411
/**
412
 * Check for non-linear entity expansion behaviour.
413
 *
414
 * In some cases like xmlExpandEntityInAttValue, this function is called
415
 * for each, possibly nested entity and its unexpanded content length.
416
 *
417
 * In other cases like #xmlParseReference, it's only called for each
418
 * top-level entity with its unexpanded content length plus the sum of
419
 * the unexpanded content lengths (plus fixed cost) of all nested
420
 * entities.
421
 *
422
 * Summing the unexpanded lengths also adds the length of the reference.
423
 * This is by design. Taking the length of the entity name into account
424
 * discourages attacks that try to waste CPU time with abusively long
425
 * entity names. See test/recurse/lol6.xml for example. Each call also
426
 * adds some fixed cost XML_ENT_FIXED_COST to discourage attacks with
427
 * short entities.
428
 *
429
 * @param ctxt  parser context
430
 * @param extra  sum of unexpanded entity sizes
431
 * @returns 1 on error, 0 on success.
432
 */
433
static int
434
xmlParserEntityCheck(xmlParserCtxtPtr ctxt, unsigned long extra)
435
1.59M
{
436
1.59M
    unsigned long consumed;
437
1.59M
    unsigned long *expandedSize;
438
1.59M
    xmlParserInputPtr input = ctxt->input;
439
1.59M
    xmlEntityPtr entity = input->entity;
440
441
1.59M
    if ((entity) && (entity->flags & XML_ENT_CHECKED))
442
116k
        return(0);
443
444
    /*
445
     * Compute total consumed bytes so far, including input streams of
446
     * external entities.
447
     */
448
1.47M
    consumed = input->consumed;
449
1.47M
    xmlSaturatedAddSizeT(&consumed, input->cur - input->base);
450
1.47M
    xmlSaturatedAdd(&consumed, ctxt->sizeentities);
451
452
1.47M
    if (entity)
453
35.6k
        expandedSize = &entity->expandedSize;
454
1.44M
    else
455
1.44M
        expandedSize = &ctxt->sizeentcopy;
456
457
    /*
458
     * Add extra cost and some fixed cost.
459
     */
460
1.47M
    xmlSaturatedAdd(expandedSize, extra);
461
1.47M
    xmlSaturatedAdd(expandedSize, XML_ENT_FIXED_COST);
462
463
    /*
464
     * It's important to always use saturation arithmetic when tracking
465
     * entity sizes to make the size checks reliable. If "sizeentcopy"
466
     * overflows, we have to abort.
467
     */
468
1.47M
    if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) &&
469
60.1k
        ((*expandedSize >= ULONG_MAX) ||
470
60.1k
         (*expandedSize / ctxt->maxAmpl > consumed))) {
471
1.05k
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
472
1.05k
                       "Maximum entity amplification factor exceeded, see "
473
1.05k
                       "xmlCtxtSetMaxAmplification.\n");
474
1.05k
        return(1);
475
1.05k
    }
476
477
1.47M
    return(0);
478
1.47M
}
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
381k
xmlSBufInit(xmlSBuf *buf, unsigned max) {
683
381k
    buf->mem = NULL;
684
381k
    buf->size = 0;
685
381k
    buf->cap = 0;
686
381k
    buf->max = max;
687
381k
    buf->code = XML_ERR_OK;
688
381k
}
689
690
static int
691
197k
xmlSBufGrow(xmlSBuf *buf, unsigned len) {
692
197k
    xmlChar *mem;
693
197k
    unsigned cap;
694
695
197k
    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
197k
    cap = (buf->size + len) * 2;
702
197k
    if (cap < 240)
703
153k
        cap = 240;
704
705
197k
    mem = xmlRealloc(buf->mem, cap);
706
197k
    if (mem == NULL) {
707
340
        buf->code = XML_ERR_NO_MEMORY;
708
340
        return(-1);
709
340
    }
710
711
196k
    buf->mem = mem;
712
196k
    buf->cap = cap;
713
714
196k
    return(0);
715
197k
}
716
717
static void
718
150M
xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) {
719
150M
    if (buf->max - buf->size < len) {
720
1.21M
        if (buf->code == XML_ERR_OK)
721
236
            buf->code = XML_ERR_RESOURCE_LIMIT;
722
1.21M
        return;
723
1.21M
    }
724
725
148M
    if (buf->cap - buf->size <= len) {
726
188k
        if (xmlSBufGrow(buf, len) < 0)
727
312
            return;
728
188k
    }
729
730
148M
    if (len > 0)
731
148M
        memcpy(buf->mem + buf->size, str, len);
732
148M
    buf->size += len;
733
148M
}
734
735
static void
736
148M
xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) {
737
148M
    xmlSBufAddString(buf, (const xmlChar *) str, len);
738
148M
}
739
740
static void
741
276k
xmlSBufAddChar(xmlSBuf *buf, int c) {
742
276k
    xmlChar *end;
743
744
276k
    if (buf->max - buf->size < 4) {
745
344
        if (buf->code == XML_ERR_OK)
746
6
            buf->code = XML_ERR_RESOURCE_LIMIT;
747
344
        return;
748
344
    }
749
750
275k
    if (buf->cap - buf->size <= 4) {
751
8.70k
        if (xmlSBufGrow(buf, 4) < 0)
752
28
            return;
753
8.70k
    }
754
755
275k
    end = buf->mem + buf->size;
756
757
275k
    if (c < 0x80) {
758
162k
        *end = (xmlChar) c;
759
162k
        buf->size += 1;
760
162k
    } else {
761
113k
        buf->size += xmlCopyCharMultiByte(end, c);
762
113k
    }
763
275k
}
764
765
static void
766
128M
xmlSBufAddReplChar(xmlSBuf *buf) {
767
128M
    xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3);
768
128M
}
769
770
static void
771
603
xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
772
603
    if (buf->code == XML_ERR_NO_MEMORY)
773
361
        xmlCtxtErrMemory(ctxt);
774
242
    else
775
242
        xmlFatalErr(ctxt, buf->code, errMsg);
776
603
}
777
778
static xmlChar *
779
xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt,
780
154k
              const char *errMsg) {
781
154k
    if (buf->mem == NULL) {
782
11.3k
        buf->mem = xmlMalloc(1);
783
11.3k
        if (buf->mem == NULL) {
784
21
            buf->code = XML_ERR_NO_MEMORY;
785
11.3k
        } else {
786
11.3k
            buf->mem[0] = 0;
787
11.3k
        }
788
143k
    } else {
789
143k
        buf->mem[buf->size] = 0;
790
143k
    }
791
792
154k
    if (buf->code == XML_ERR_OK) {
793
154k
        if (sizeOut != NULL)
794
59.9k
            *sizeOut = buf->size;
795
154k
        return(buf->mem);
796
154k
    }
797
798
315
    xmlSBufReportError(buf, ctxt, errMsg);
799
800
315
    xmlFree(buf->mem);
801
802
315
    if (sizeOut != NULL)
803
134
        *sizeOut = 0;
804
315
    return(NULL);
805
154k
}
806
807
static void
808
221k
xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
809
221k
    if (buf->code != XML_ERR_OK)
810
288
        xmlSBufReportError(buf, ctxt, errMsg);
811
812
221k
    xmlFree(buf->mem);
813
221k
}
814
815
static int
816
xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str,
817
499M
                    const char *errMsg) {
818
499M
    int c = str[0];
819
499M
    int c1 = str[1];
820
821
499M
    if ((c1 & 0xC0) != 0x80)
822
71.1M
        goto encoding_error;
823
824
428M
    if (c < 0xE0) {
825
        /* 2-byte sequence */
826
24.6M
        if (c < 0xC2)
827
14.9M
            goto encoding_error;
828
829
9.72M
        return(2);
830
403M
    } else {
831
403M
        int c2 = str[2];
832
833
403M
        if ((c2 & 0xC0) != 0x80)
834
13.8k
            goto encoding_error;
835
836
403M
        if (c < 0xF0) {
837
            /* 3-byte sequence */
838
403M
            if (c == 0xE0) {
839
                /* overlong */
840
3.50k
                if (c1 < 0xA0)
841
216
                    goto encoding_error;
842
403M
            } else if (c == 0xED) {
843
                /* surrogate */
844
710
                if (c1 >= 0xA0)
845
256
                    goto encoding_error;
846
403M
            } else if (c == 0xEF) {
847
                /* U+FFFE and U+FFFF are invalid Chars */
848
44.8M
                if ((c1 == 0xBF) && (c2 >= 0xBE))
849
329
                    xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg);
850
44.8M
            }
851
852
403M
            return(3);
853
403M
        } else {
854
            /* 4-byte sequence */
855
15.4k
            if ((str[3] & 0xC0) != 0x80)
856
3.47k
                goto encoding_error;
857
11.9k
            if (c == 0xF0) {
858
                /* overlong */
859
1.19k
                if (c1 < 0x90)
860
261
                    goto encoding_error;
861
10.7k
            } else if (c >= 0xF4) {
862
                /* greater than 0x10FFFF */
863
4.28k
                if ((c > 0xF4) || (c1 >= 0x90))
864
3.99k
                    goto encoding_error;
865
4.28k
            }
866
867
7.72k
            return(4);
868
11.9k
        }
869
403M
    }
870
871
86.0M
encoding_error:
872
    /* Only report the first error */
873
86.0M
    if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) {
874
14.2k
        xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL);
875
14.2k
        ctxt->input->flags |= XML_INPUT_ENCODING_ERROR;
876
14.2k
    }
877
878
86.0M
    return(0);
879
428M
}
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
62.2k
xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) {
897
62.2k
    xmlSAXHandlerPtr sax;
898
899
    /* Avoid unused variable warning if features are disabled. */
900
62.2k
    (void) sax;
901
902
    /*
903
     * Changing the SAX struct directly is still widespread practice
904
     * in internal and external code.
905
     */
906
62.2k
    if (ctxt == NULL) return;
907
62.2k
    sax = ctxt->sax;
908
62.2k
#ifdef LIBXML_SAX1_ENABLED
909
    /*
910
     * Only enable SAX2 if there SAX2 element handlers, except when there
911
     * are no element handlers at all.
912
     */
913
62.2k
    if (((ctxt->options & XML_PARSE_SAX1) == 0) &&
914
62.2k
        (sax) &&
915
62.2k
        (sax->initialized == XML_SAX2_MAGIC) &&
916
62.2k
        ((sax->startElementNs != NULL) ||
917
0
         (sax->endElementNs != NULL) ||
918
0
         ((sax->startElement == NULL) && (sax->endElement == NULL))))
919
62.2k
        ctxt->sax2 = 1;
920
#else
921
    ctxt->sax2 = 1;
922
#endif /* LIBXML_SAX1_ENABLED */
923
924
    /*
925
     * Some users replace the dictionary directly in the context struct.
926
     * We really need an API function to do that cleanly.
927
     */
928
62.2k
    ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3);
929
62.2k
    ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5);
930
62.2k
    ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36);
931
62.2k
    if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) ||
932
61.7k
    (ctxt->str_xml_ns == NULL)) {
933
502
        xmlErrMemory(ctxt);
934
502
    }
935
936
62.2k
    xmlDictSetLimit(ctxt->dict,
937
62.2k
                    (ctxt->options & XML_PARSE_HUGE) ?
938
27.5k
                        0 :
939
62.2k
                        XML_MAX_DICTIONARY_LIMIT);
940
941
62.2k
#ifdef LIBXML_VALID_ENABLED
942
62.2k
    if (ctxt->validate)
943
0
        ctxt->vctxt.flags |= XML_VCTXT_VALIDATE;
944
62.2k
    else
945
62.2k
        ctxt->vctxt.flags &= ~XML_VCTXT_VALIDATE;
946
62.2k
#endif /* LIBXML_VALID_ENABLED */
947
62.2k
}
948
949
typedef struct {
950
    xmlHashedString prefix;
951
    xmlHashedString name;
952
    xmlHashedString value;
953
    const xmlChar *valueEnd;
954
    int external;
955
    int expandedSize;
956
} xmlDefAttr;
957
958
typedef struct _xmlDefAttrs xmlDefAttrs;
959
typedef xmlDefAttrs *xmlDefAttrsPtr;
960
struct _xmlDefAttrs {
961
    int nbAttrs;  /* number of defaulted attributes on that element */
962
    int maxAttrs;       /* the size of the array */
963
#if __STDC_VERSION__ >= 199901L
964
    /* Using a C99 flexible array member avoids UBSan errors. */
965
    xmlDefAttr attrs[] ATTRIBUTE_COUNTED_BY(maxAttrs);
966
#else
967
    xmlDefAttr attrs[1];
968
#endif
969
};
970
971
/**
972
 * Normalize the space in non CDATA attribute values:
973
 * If the attribute type is not CDATA, then the XML processor MUST further
974
 * process the normalized attribute value by discarding any leading and
975
 * trailing space (\#x20) characters, and by replacing sequences of space
976
 * (\#x20) characters by a single space (\#x20) character.
977
 * Note that the size of dst need to be at least src, and if one doesn't need
978
 * to preserve dst (and it doesn't come from a dictionary or read-only) then
979
 * passing src as dst is just fine.
980
 *
981
 * @param src  the source string
982
 * @param dst  the target string
983
 * @returns a pointer to the normalized value (dst) or NULL if no conversion
984
 *         is needed.
985
 */
986
static xmlChar *
987
xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst)
988
51.9k
{
989
51.9k
    if ((src == NULL) || (dst == NULL))
990
0
        return(NULL);
991
992
52.7k
    while (*src == 0x20) src++;
993
77.7M
    while (*src != 0) {
994
77.6M
  if (*src == 0x20) {
995
327k
      while (*src == 0x20) src++;
996
10.5k
      if (*src != 0)
997
10.0k
    *dst++ = 0x20;
998
77.6M
  } else {
999
77.6M
      *dst++ = *src++;
1000
77.6M
  }
1001
77.6M
    }
1002
51.9k
    *dst = 0;
1003
51.9k
    if (dst == src)
1004
50.3k
       return(NULL);
1005
1.62k
    return(dst);
1006
51.9k
}
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
49.0k
               const xmlChar *value) {
1021
49.0k
    xmlDefAttrsPtr defaults;
1022
49.0k
    xmlDefAttr *attr;
1023
49.0k
    int len, expandedSize;
1024
49.0k
    xmlHashedString name;
1025
49.0k
    xmlHashedString prefix;
1026
49.0k
    xmlHashedString hvalue;
1027
49.0k
    const xmlChar *localname;
1028
1029
    /*
1030
     * Allows to detect attribute redefinitions
1031
     */
1032
49.0k
    if (ctxt->attsSpecial != NULL) {
1033
45.1k
        if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL)
1034
31.2k
      return;
1035
45.1k
    }
1036
1037
17.8k
    if (ctxt->attsDefault == NULL) {
1038
4.13k
        ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict);
1039
4.13k
  if (ctxt->attsDefault == NULL)
1040
14
      goto mem_error;
1041
4.13k
    }
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
17.7k
    localname = xmlSplitQName3(fullname, &len);
1048
17.7k
    if (localname == NULL) {
1049
17.2k
        name = xmlDictLookupHashed(ctxt->dict, fullname, -1);
1050
17.2k
  prefix.name = NULL;
1051
17.2k
    } else {
1052
568
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1053
568
  prefix = xmlDictLookupHashed(ctxt->dict, fullname, len);
1054
568
        if (prefix.name == NULL)
1055
1
            goto mem_error;
1056
568
    }
1057
17.7k
    if (name.name == NULL)
1058
3
        goto mem_error;
1059
1060
    /*
1061
     * make sure there is some storage
1062
     */
1063
17.7k
    defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name);
1064
17.7k
    if ((defaults == NULL) ||
1065
12.8k
        (defaults->nbAttrs >= defaults->maxAttrs)) {
1066
5.88k
        xmlDefAttrsPtr temp;
1067
5.88k
        int newSize;
1068
1069
5.88k
        if (defaults == NULL) {
1070
4.90k
            newSize = 4;
1071
4.90k
        } else {
1072
973
            if ((defaults->maxAttrs >= XML_MAX_ATTRS) ||
1073
973
                ((size_t) defaults->maxAttrs >
1074
973
                     SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults)))
1075
0
                goto mem_error;
1076
1077
973
            if (defaults->maxAttrs > XML_MAX_ATTRS / 2)
1078
0
                newSize = XML_MAX_ATTRS;
1079
973
            else
1080
973
                newSize = defaults->maxAttrs * 2;
1081
973
        }
1082
5.88k
        temp = xmlRealloc(defaults,
1083
5.88k
                          sizeof(*defaults) + newSize * sizeof(xmlDefAttr));
1084
5.88k
  if (temp == NULL)
1085
10
      goto mem_error;
1086
5.87k
        if (defaults == NULL)
1087
4.89k
            temp->nbAttrs = 0;
1088
5.87k
  temp->maxAttrs = newSize;
1089
5.87k
        defaults = temp;
1090
5.87k
  if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name,
1091
5.87k
                          defaults, NULL) < 0) {
1092
1
      xmlFree(defaults);
1093
1
      goto mem_error;
1094
1
  }
1095
5.87k
    }
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
17.7k
    localname = xmlSplitQName3(fullattr, &len);
1102
17.7k
    if (localname == NULL) {
1103
14.0k
        name = xmlDictLookupHashed(ctxt->dict, fullattr, -1);
1104
14.0k
  prefix.name = NULL;
1105
14.0k
    } else {
1106
3.75k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1107
3.75k
  prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len);
1108
3.75k
        if (prefix.name == NULL)
1109
3
            goto mem_error;
1110
3.75k
    }
1111
17.7k
    if (name.name == NULL)
1112
3
        goto mem_error;
1113
1114
    /* intern the string and precompute the end */
1115
17.7k
    len = strlen((const char *) value);
1116
17.7k
    hvalue = xmlDictLookupHashed(ctxt->dict, value, len);
1117
17.7k
    if (hvalue.name == NULL)
1118
3
        goto mem_error;
1119
1120
17.7k
    expandedSize = strlen((const char *) name.name);
1121
17.7k
    if (prefix.name != NULL)
1122
3.74k
        expandedSize += strlen((const char *) prefix.name);
1123
17.7k
    expandedSize += len;
1124
1125
17.7k
    attr = &defaults->attrs[defaults->nbAttrs++];
1126
17.7k
    attr->name = name;
1127
17.7k
    attr->prefix = prefix;
1128
17.7k
    attr->value = hvalue;
1129
17.7k
    attr->valueEnd = hvalue.name + len;
1130
17.7k
    attr->external = PARSER_EXTERNAL(ctxt);
1131
17.7k
    attr->expandedSize = expandedSize;
1132
1133
17.7k
    return;
1134
1135
38
mem_error:
1136
38
    xmlErrMemory(ctxt);
1137
38
}
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
83.9k
{
1153
83.9k
    if (ctxt->attsSpecial == NULL) {
1154
5.99k
        ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict);
1155
5.99k
  if (ctxt->attsSpecial == NULL)
1156
17
      goto mem_error;
1157
5.99k
    }
1158
1159
83.9k
    if (PARSER_EXTERNAL(ctxt))
1160
50.2k
        type |= XML_SPECIAL_EXTERNAL;
1161
1162
83.9k
    if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr,
1163
83.9k
                    XML_INT_TO_PTR(type)) < 0)
1164
6
        goto mem_error;
1165
83.9k
    return;
1166
1167
83.9k
mem_error:
1168
23
    xmlErrMemory(ctxt);
1169
23
}
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
43.6k
                            const xmlChar *unused ATTRIBUTE_UNUSED) {
1178
43.6k
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data;
1179
1180
43.6k
    if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) {
1181
4.32k
        xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL);
1182
4.32k
    }
1183
43.6k
}
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
26.7k
{
1195
26.7k
    if (ctxt->attsSpecial == NULL)
1196
20.7k
        return;
1197
1198
5.98k
    xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt);
1199
1200
5.98k
    if (xmlHashSize(ctxt->attsSpecial) == 0) {
1201
173
        xmlHashFree(ctxt->attsSpecial, NULL);
1202
173
        ctxt->attsSpecial = NULL;
1203
173
    }
1204
5.98k
}
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
12.7k
{
1268
12.7k
    const xmlChar *cur = lang, *nxt;
1269
1270
12.7k
    if (cur == NULL)
1271
0
        return (0);
1272
12.7k
    if (((cur[0] == 'i') && (cur[1] == '-')) ||
1273
12.4k
        ((cur[0] == 'I') && (cur[1] == '-')) ||
1274
12.2k
        ((cur[0] == 'x') && (cur[1] == '-')) ||
1275
12.0k
        ((cur[0] == 'X') && (cur[1] == '-'))) {
1276
        /*
1277
         * Still allow IANA code and user code which were coming
1278
         * from the previous version of the XML-1.0 specification
1279
         * it's deprecated but we should not fail
1280
         */
1281
1.25k
        cur += 2;
1282
3.55k
        while (((cur[0] >= 'A') && (cur[0] <= 'Z')) ||
1283
2.65k
               ((cur[0] >= 'a') && (cur[0] <= 'z')))
1284
2.29k
            cur++;
1285
1.25k
        return(cur[0] == 0);
1286
1.25k
    }
1287
11.4k
    nxt = cur;
1288
38.6k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1289
27.6k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1290
27.1k
           nxt++;
1291
11.4k
    if (nxt - cur >= 4) {
1292
        /*
1293
         * Reserved
1294
         */
1295
1.00k
        if ((nxt - cur > 8) || (nxt[0] != 0))
1296
797
            return(0);
1297
211
        return(1);
1298
1.00k
    }
1299
10.4k
    if (nxt - cur < 2)
1300
859
        return(0);
1301
    /* we got an ISO 639 code */
1302
9.59k
    if (nxt[0] == 0)
1303
393
        return(1);
1304
9.19k
    if (nxt[0] != '-')
1305
564
        return(0);
1306
1307
8.63k
    nxt++;
1308
8.63k
    cur = nxt;
1309
    /* now we can have extlang or script or region or variant */
1310
8.63k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1311
696
        goto region_m49;
1312
1313
47.8k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1314
27.5k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1315
39.9k
           nxt++;
1316
7.93k
    if (nxt - cur == 4)
1317
1.79k
        goto script;
1318
6.14k
    if (nxt - cur == 2)
1319
975
        goto region;
1320
5.17k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1321
2.23k
        goto variant;
1322
2.93k
    if (nxt - cur != 3)
1323
499
        return(0);
1324
    /* we parsed an extlang */
1325
2.43k
    if (nxt[0] == 0)
1326
294
        return(1);
1327
2.14k
    if (nxt[0] != '-')
1328
255
        return(0);
1329
1330
1.89k
    nxt++;
1331
1.89k
    cur = nxt;
1332
    /* now we can have script or region or variant */
1333
1.89k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1334
200
        goto region_m49;
1335
1336
9.68k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1337
6.23k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1338
7.99k
           nxt++;
1339
1.69k
    if (nxt - cur == 2)
1340
266
        goto region;
1341
1.42k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1342
334
        goto variant;
1343
1.09k
    if (nxt - cur != 4)
1344
566
        return(0);
1345
    /* we parsed a script */
1346
2.31k
script:
1347
2.31k
    if (nxt[0] == 0)
1348
409
        return(1);
1349
1.90k
    if (nxt[0] != '-')
1350
282
        return(0);
1351
1352
1.62k
    nxt++;
1353
1.62k
    cur = nxt;
1354
    /* now we can have region or variant */
1355
1.62k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1356
375
        goto region_m49;
1357
1358
7.73k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1359
5.04k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1360
6.48k
           nxt++;
1361
1362
1.25k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1363
334
        goto variant;
1364
916
    if (nxt - cur != 2)
1365
594
        return(0);
1366
    /* we parsed a region */
1367
1.78k
region:
1368
1.78k
    if (nxt[0] == 0)
1369
331
        return(1);
1370
1.45k
    if (nxt[0] != '-')
1371
817
        return(0);
1372
1373
640
    nxt++;
1374
640
    cur = nxt;
1375
    /* now we can just have a variant */
1376
4.41k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1377
3.45k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1378
3.77k
           nxt++;
1379
1380
640
    if ((nxt - cur < 5) || (nxt - cur > 8))
1381
432
        return(0);
1382
1383
    /* we parsed a variant */
1384
3.10k
variant:
1385
3.10k
    if (nxt[0] == 0)
1386
260
        return(1);
1387
2.84k
    if (nxt[0] != '-')
1388
2.41k
        return(0);
1389
    /* extensions and private use subtags not checked */
1390
436
    return (1);
1391
1392
1.27k
region_m49:
1393
1.27k
    if (((nxt[1] >= '0') && (nxt[1] <= '9')) &&
1394
666
        ((nxt[2] >= '0') && (nxt[2] <= '9'))) {
1395
225
        nxt += 3;
1396
225
        goto region;
1397
225
    }
1398
1.04k
    return(0);
1399
1.27k
}
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
50.6k
xmlParserNsCreate(void) {
1417
50.6k
    xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb));
1418
1419
50.6k
    if (nsdb == NULL)
1420
12
        return(NULL);
1421
50.6k
    memset(nsdb, 0, sizeof(*nsdb));
1422
50.6k
    nsdb->defaultNsIndex = INT_MAX;
1423
1424
50.6k
    return(nsdb);
1425
50.6k
}
1426
1427
/**
1428
 * Free a namespace database.
1429
 *
1430
 * @param nsdb  namespace database
1431
 */
1432
void
1433
50.6k
xmlParserNsFree(xmlParserNsData *nsdb) {
1434
50.6k
    if (nsdb == NULL)
1435
0
        return;
1436
1437
50.6k
    xmlFree(nsdb->extra);
1438
50.6k
    xmlFree(nsdb->hash);
1439
50.6k
    xmlFree(nsdb);
1440
50.6k
}
1441
1442
/**
1443
 * Reset a namespace database.
1444
 *
1445
 * @param nsdb  namespace database
1446
 */
1447
static void
1448
25.3k
xmlParserNsReset(xmlParserNsData *nsdb) {
1449
25.3k
    if (nsdb == NULL)
1450
0
        return;
1451
1452
25.3k
    nsdb->hashElems = 0;
1453
25.3k
    nsdb->elementId = 0;
1454
25.3k
    nsdb->defaultNsIndex = INT_MAX;
1455
1456
25.3k
    if (nsdb->hash)
1457
0
        memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0]));
1458
25.3k
}
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
638k
xmlParserNsStartElement(xmlParserNsData *nsdb) {
1468
638k
    if (nsdb->elementId == UINT_MAX)
1469
0
        return(-1);
1470
638k
    nsdb->elementId++;
1471
1472
638k
    return(0);
1473
638k
}
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
1.34M
                  xmlParserNsBucket **bucketPtr) {
1489
1.34M
    xmlParserNsBucket *bucket, *tombstone;
1490
1.34M
    unsigned index, hashValue;
1491
1492
1.34M
    if (prefix->name == NULL)
1493
656k
        return(ctxt->nsdb->defaultNsIndex);
1494
1495
683k
    if (ctxt->nsdb->hashSize == 0)
1496
22.6k
        return(INT_MAX);
1497
1498
660k
    hashValue = prefix->hashValue;
1499
660k
    index = hashValue & (ctxt->nsdb->hashSize - 1);
1500
660k
    bucket = &ctxt->nsdb->hash[index];
1501
660k
    tombstone = NULL;
1502
1503
861k
    while (bucket->hashValue) {
1504
794k
        if (bucket->index == INT_MAX) {
1505
40.7k
            if (tombstone == NULL)
1506
27.4k
                tombstone = bucket;
1507
753k
        } else if (bucket->hashValue == hashValue) {
1508
593k
            if (ctxt->nsTab[bucket->index * 2] == prefix->name) {
1509
593k
                if (bucketPtr != NULL)
1510
352k
                    *bucketPtr = bucket;
1511
593k
                return(bucket->index);
1512
593k
            }
1513
593k
        }
1514
1515
200k
        index++;
1516
200k
        bucket++;
1517
200k
        if (index == ctxt->nsdb->hashSize) {
1518
10.1k
            index = 0;
1519
10.1k
            bucket = ctxt->nsdb->hash;
1520
10.1k
        }
1521
200k
    }
1522
1523
67.4k
    if (bucketPtr != NULL)
1524
17.3k
        *bucketPtr = tombstone ? tombstone : bucket;
1525
67.4k
    return(INT_MAX);
1526
660k
}
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
573k
xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) {
1537
573k
    const xmlChar *ret;
1538
573k
    int nsIndex;
1539
1540
573k
    if (prefix->name == ctxt->str_xml)
1541
866
        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
572k
    nsIndex = xmlParserNsLookup(ctxt, prefix, NULL);
1548
572k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1549
401k
        return(NULL);
1550
1551
170k
    ret = ctxt->nsTab[nsIndex * 2 + 1];
1552
170k
    if (ret[0] == 0)
1553
7.07k
        ret = NULL;
1554
170k
    return(ret);
1555
572k
}
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
156k
xmlParserNsLookupSax(xmlParserCtxt *ctxt, const xmlChar *prefix) {
1567
156k
    xmlHashedString hprefix;
1568
156k
    int nsIndex;
1569
1570
156k
    if (prefix == ctxt->str_xml)
1571
27.0k
        return(NULL);
1572
1573
129k
    hprefix.name = prefix;
1574
129k
    if (prefix != NULL)
1575
32.3k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1576
96.8k
    else
1577
96.8k
        hprefix.hashValue = 0;
1578
129k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1579
129k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1580
0
        return(NULL);
1581
1582
129k
    return(ctxt->nsdb->extra[nsIndex].saxData);
1583
129k
}
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
187k
                     void *saxData) {
1598
187k
    xmlHashedString hprefix;
1599
187k
    int nsIndex;
1600
1601
187k
    if (prefix == ctxt->str_xml)
1602
0
        return(-1);
1603
1604
187k
    hprefix.name = prefix;
1605
187k
    if (prefix != NULL)
1606
158k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1607
28.5k
    else
1608
28.5k
        hprefix.hashValue = 0;
1609
187k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1610
187k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1611
0
        return(-1);
1612
1613
187k
    ctxt->nsdb->extra[nsIndex].saxData = saxData;
1614
187k
    return(0);
1615
187k
}
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
19.4k
xmlParserNsGrow(xmlParserCtxtPtr ctxt) {
1625
19.4k
    const xmlChar **table;
1626
19.4k
    xmlParserNsExtra *extra;
1627
19.4k
    int newSize;
1628
1629
19.4k
    newSize = xmlGrowCapacity(ctxt->nsMax,
1630
19.4k
                              sizeof(table[0]) + sizeof(extra[0]),
1631
19.4k
                              16, XML_MAX_ITEMS);
1632
19.4k
    if (newSize < 0)
1633
0
        goto error;
1634
1635
19.4k
    table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0]));
1636
19.4k
    if (table == NULL)
1637
29
        goto error;
1638
19.3k
    ctxt->nsTab = table;
1639
1640
19.3k
    extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0]));
1641
19.3k
    if (extra == NULL)
1642
36
        goto error;
1643
19.3k
    ctxt->nsdb->extra = extra;
1644
1645
19.3k
    ctxt->nsMax = newSize;
1646
19.3k
    return(0);
1647
1648
65
error:
1649
65
    xmlErrMemory(ctxt);
1650
65
    return(-1);
1651
19.3k
}
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
249k
                const xmlHashedString *uri, void *saxData, int defAttr) {
1667
249k
    xmlParserNsBucket *bucket = NULL;
1668
249k
    xmlParserNsExtra *extra;
1669
249k
    const xmlChar **ns;
1670
249k
    unsigned hashValue, nsIndex, oldIndex;
1671
1672
249k
    if ((prefix != NULL) && (prefix->name == ctxt->str_xml))
1673
218
        return(0);
1674
1675
249k
    if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) {
1676
65
        xmlErrMemory(ctxt);
1677
65
        return(-1);
1678
65
    }
1679
1680
    /*
1681
     * Default namespace and 'xml' namespace
1682
     */
1683
249k
    if ((prefix == NULL) || (prefix->name == NULL)) {
1684
47.6k
        oldIndex = ctxt->nsdb->defaultNsIndex;
1685
1686
47.6k
        if (oldIndex != INT_MAX) {
1687
37.5k
            extra = &ctxt->nsdb->extra[oldIndex];
1688
1689
37.5k
            if (extra->elementId == ctxt->nsdb->elementId) {
1690
1.27k
                if (defAttr == 0)
1691
1.07k
                    xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns");
1692
1.27k
                return(0);
1693
1.27k
            }
1694
1695
36.2k
            if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1696
19.5k
                (uri->name == ctxt->nsTab[oldIndex * 2 + 1]))
1697
9.72k
                return(0);
1698
36.2k
        }
1699
1700
36.6k
        ctxt->nsdb->defaultNsIndex = ctxt->nsNr;
1701
36.6k
        goto populate_entry;
1702
47.6k
    }
1703
1704
    /*
1705
     * Hash table lookup
1706
     */
1707
201k
    oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket);
1708
201k
    if (oldIndex != INT_MAX) {
1709
181k
        extra = &ctxt->nsdb->extra[oldIndex];
1710
1711
        /*
1712
         * Check for duplicate definitions on the same element.
1713
         */
1714
181k
        if (extra->elementId == ctxt->nsdb->elementId) {
1715
713
            if (defAttr == 0)
1716
503
                xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name);
1717
713
            return(0);
1718
713
        }
1719
1720
180k
        if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1721
11.7k
            (uri->name == ctxt->nsTab[bucket->index * 2 + 1]))
1722
10.3k
            return(0);
1723
1724
170k
        bucket->index = ctxt->nsNr;
1725
170k
        goto populate_entry;
1726
180k
    }
1727
1728
    /*
1729
     * Insert new bucket
1730
     */
1731
1732
20.3k
    hashValue = prefix->hashValue;
1733
1734
    /*
1735
     * Grow hash table, 50% fill factor
1736
     */
1737
20.3k
    if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) {
1738
3.81k
        xmlParserNsBucket *newHash;
1739
3.81k
        unsigned newSize, i, index;
1740
1741
3.81k
        if (ctxt->nsdb->hashSize > UINT_MAX / 2) {
1742
0
            xmlErrMemory(ctxt);
1743
0
            return(-1);
1744
0
        }
1745
3.81k
        newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16;
1746
3.81k
        newHash = xmlMalloc(newSize * sizeof(newHash[0]));
1747
3.81k
        if (newHash == NULL) {
1748
8
            xmlErrMemory(ctxt);
1749
8
            return(-1);
1750
8
        }
1751
3.80k
        memset(newHash, 0, newSize * sizeof(newHash[0]));
1752
1753
46.0k
        for (i = 0; i < ctxt->nsdb->hashSize; i++) {
1754
42.2k
            unsigned hv = ctxt->nsdb->hash[i].hashValue;
1755
42.2k
            unsigned newIndex;
1756
1757
42.2k
            if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX))
1758
40.0k
                continue;
1759
2.24k
            newIndex = hv & (newSize - 1);
1760
1761
3.35k
            while (newHash[newIndex].hashValue != 0) {
1762
1.10k
                newIndex++;
1763
1.10k
                if (newIndex == newSize)
1764
48
                    newIndex = 0;
1765
1.10k
            }
1766
1767
2.24k
            newHash[newIndex] = ctxt->nsdb->hash[i];
1768
2.24k
        }
1769
1770
3.80k
        xmlFree(ctxt->nsdb->hash);
1771
3.80k
        ctxt->nsdb->hash = newHash;
1772
3.80k
        ctxt->nsdb->hashSize = newSize;
1773
1774
        /*
1775
         * Relookup
1776
         */
1777
3.80k
        index = hashValue & (newSize - 1);
1778
1779
4.20k
        while (newHash[index].hashValue != 0) {
1780
404
            index++;
1781
404
            if (index == newSize)
1782
17
                index = 0;
1783
404
        }
1784
1785
3.80k
        bucket = &newHash[index];
1786
3.80k
    }
1787
1788
20.3k
    bucket->hashValue = hashValue;
1789
20.3k
    bucket->index = ctxt->nsNr;
1790
20.3k
    ctxt->nsdb->hashElems++;
1791
20.3k
    oldIndex = INT_MAX;
1792
1793
227k
populate_entry:
1794
227k
    nsIndex = ctxt->nsNr;
1795
1796
227k
    ns = &ctxt->nsTab[nsIndex * 2];
1797
227k
    ns[0] = prefix ? prefix->name : NULL;
1798
227k
    ns[1] = uri->name;
1799
1800
227k
    extra = &ctxt->nsdb->extra[nsIndex];
1801
227k
    extra->saxData = saxData;
1802
227k
    extra->prefixHashValue = prefix ? prefix->hashValue : 0;
1803
227k
    extra->uriHashValue = uri->hashValue;
1804
227k
    extra->elementId = ctxt->nsdb->elementId;
1805
227k
    extra->oldIndex = oldIndex;
1806
1807
227k
    ctxt->nsNr++;
1808
1809
227k
    return(1);
1810
20.3k
}
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
72.4k
{
1822
72.4k
    int i;
1823
1824
    /* assert(nr <= ctxt->nsNr); */
1825
1826
275k
    for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) {
1827
202k
        const xmlChar *prefix = ctxt->nsTab[i * 2];
1828
202k
        xmlParserNsExtra *extra = &ctxt->nsdb->extra[i];
1829
1830
202k
        if (prefix == NULL) {
1831
31.2k
            ctxt->nsdb->defaultNsIndex = extra->oldIndex;
1832
171k
        } else {
1833
171k
            xmlHashedString hprefix;
1834
171k
            xmlParserNsBucket *bucket = NULL;
1835
1836
171k
            hprefix.name = prefix;
1837
171k
            hprefix.hashValue = extra->prefixHashValue;
1838
171k
            xmlParserNsLookup(ctxt, &hprefix, &bucket);
1839
            /* assert(bucket && bucket->hashValue); */
1840
171k
            bucket->index = extra->oldIndex;
1841
171k
        }
1842
202k
    }
1843
1844
72.4k
    ctxt->nsNr -= nr;
1845
72.4k
    return(nr);
1846
72.4k
}
1847
1848
static int
1849
13.9k
xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) {
1850
13.9k
    const xmlChar **atts;
1851
13.9k
    unsigned *attallocs;
1852
13.9k
    int newSize;
1853
1854
13.9k
    newSize = xmlGrowCapacity(ctxt->maxatts / 5,
1855
13.9k
                              sizeof(atts[0]) * 5 + sizeof(attallocs[0]),
1856
13.9k
                              10, XML_MAX_ATTRS);
1857
13.9k
    if (newSize < 0) {
1858
0
        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
1859
0
                    "Maximum number of attributes exceeded");
1860
0
        return(-1);
1861
0
    }
1862
1863
13.9k
    atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5);
1864
13.9k
    if (atts == NULL)
1865
13
        goto mem_error;
1866
13.9k
    ctxt->atts = atts;
1867
1868
13.9k
    attallocs = xmlRealloc(ctxt->attallocs,
1869
13.9k
                           newSize * sizeof(attallocs[0]));
1870
13.9k
    if (attallocs == NULL)
1871
13
        goto mem_error;
1872
13.9k
    ctxt->attallocs = attallocs;
1873
1874
13.9k
    ctxt->maxatts = newSize * 5;
1875
1876
13.9k
    return(0);
1877
1878
26
mem_error:
1879
26
    xmlErrMemory(ctxt);
1880
26
    return(-1);
1881
13.9k
}
1882
1883
/**
1884
 * Pushes a new parser input on top of the input stack
1885
 *
1886
 * @param ctxt  an XML parser context
1887
 * @param value  the parser input
1888
 * @returns -1 in case of error, the index in the stack otherwise
1889
 */
1890
int
1891
xmlCtxtPushInput(xmlParserCtxt *ctxt, xmlParserInput *value)
1892
228k
{
1893
228k
    char *directory = NULL;
1894
228k
    int maxDepth;
1895
1896
228k
    if ((ctxt == NULL) || (value == NULL))
1897
2.56k
        return(-1);
1898
1899
226k
    maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
1900
1901
226k
    if (ctxt->inputNr >= ctxt->inputMax) {
1902
11.0k
        xmlParserInputPtr *tmp;
1903
11.0k
        int newSize;
1904
1905
11.0k
        newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]),
1906
11.0k
                                  5, maxDepth);
1907
11.0k
        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
11.0k
        tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0]));
1913
11.0k
        if (tmp == NULL) {
1914
48
            xmlErrMemory(ctxt);
1915
48
            return(-1);
1916
48
        }
1917
11.0k
        ctxt->inputTab = tmp;
1918
11.0k
        ctxt->inputMax = newSize;
1919
11.0k
    }
1920
1921
226k
    if ((ctxt->inputNr == 0) && (value->filename != NULL)) {
1922
70.7k
        directory = xmlParserGetDirectory(value->filename);
1923
70.7k
        if (directory == NULL) {
1924
82
            xmlErrMemory(ctxt);
1925
82
            return(-1);
1926
82
        }
1927
70.7k
    }
1928
1929
225k
    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
225k
    ctxt->inputTab[ctxt->inputNr] = value;
1935
225k
    ctxt->input = value;
1936
1937
225k
    if (ctxt->inputNr == 0) {
1938
70.6k
        xmlFree(ctxt->directory);
1939
70.6k
        ctxt->directory = directory;
1940
70.6k
    }
1941
1942
    /*
1943
     * The input ID is unused internally, but there are entity
1944
     * loaders in downstream code that detect the main document
1945
     * by checking for "input_id == 1".
1946
     */
1947
225k
    value->id = ctxt->input_id++;
1948
1949
225k
    return(ctxt->inputNr++);
1950
225k
}
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
351k
{
1961
351k
    xmlParserInputPtr ret;
1962
1963
351k
    if (ctxt == NULL)
1964
0
        return(NULL);
1965
351k
    if (ctxt->inputNr <= 0)
1966
126k
        return (NULL);
1967
224k
    ctxt->inputNr--;
1968
224k
    if (ctxt->inputNr > 0)
1969
155k
        ctxt->input = ctxt->inputTab[ctxt->inputNr - 1];
1970
69.1k
    else
1971
69.1k
        ctxt->input = NULL;
1972
224k
    ret = ctxt->inputTab[ctxt->inputNr];
1973
224k
    ctxt->inputTab[ctxt->inputNr] = NULL;
1974
224k
    return (ret);
1975
351k
}
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
556k
{
1989
556k
    if (ctxt == NULL)
1990
0
        return(0);
1991
1992
556k
    if (ctxt->nodeNr >= ctxt->nodeMax) {
1993
40.0k
        int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
1994
40.0k
        xmlNodePtr *tmp;
1995
40.0k
        int newSize;
1996
1997
40.0k
        newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]),
1998
40.0k
                                  10, maxDepth);
1999
40.0k
        if (newSize < 0) {
2000
18
            xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
2001
18
                    "Excessive depth in document: %d,"
2002
18
                    " use XML_PARSE_HUGE option\n",
2003
18
                    ctxt->nodeNr);
2004
18
            return(-1);
2005
18
        }
2006
2007
39.9k
  tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0]));
2008
39.9k
        if (tmp == NULL) {
2009
51
            xmlErrMemory(ctxt);
2010
51
            return (-1);
2011
51
        }
2012
39.9k
        ctxt->nodeTab = tmp;
2013
39.9k
  ctxt->nodeMax = newSize;
2014
39.9k
    }
2015
2016
556k
    ctxt->nodeTab[ctxt->nodeNr] = value;
2017
556k
    ctxt->node = value;
2018
556k
    return (ctxt->nodeNr++);
2019
556k
}
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
510k
{
2032
510k
    xmlNodePtr ret;
2033
2034
510k
    if (ctxt == NULL) return(NULL);
2035
510k
    if (ctxt->nodeNr <= 0)
2036
24.8k
        return (NULL);
2037
485k
    ctxt->nodeNr--;
2038
485k
    if (ctxt->nodeNr > 0)
2039
478k
        ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1];
2040
7.38k
    else
2041
7.38k
        ctxt->node = NULL;
2042
485k
    ret = ctxt->nodeTab[ctxt->nodeNr];
2043
485k
    ctxt->nodeTab[ctxt->nodeNr] = NULL;
2044
485k
    return (ret);
2045
510k
}
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
612k
{
2062
612k
    xmlStartTag *tag;
2063
2064
612k
    if (ctxt->nameNr >= ctxt->nameMax) {
2065
40.9k
        const xmlChar **tmp;
2066
40.9k
        xmlStartTag *tmp2;
2067
40.9k
        int newSize;
2068
2069
40.9k
        newSize = xmlGrowCapacity(ctxt->nameMax,
2070
40.9k
                                  sizeof(tmp[0]) + sizeof(tmp2[0]),
2071
40.9k
                                  10, XML_MAX_ITEMS);
2072
40.9k
        if (newSize < 0)
2073
0
            goto mem_error;
2074
2075
40.9k
        tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0]));
2076
40.9k
        if (tmp == NULL)
2077
21
      goto mem_error;
2078
40.9k
  ctxt->nameTab = tmp;
2079
2080
40.9k
        tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0]));
2081
40.9k
        if (tmp2 == NULL)
2082
25
      goto mem_error;
2083
40.9k
  ctxt->pushTab = tmp2;
2084
2085
40.9k
        ctxt->nameMax = newSize;
2086
571k
    } else if (ctxt->pushTab == NULL) {
2087
23.7k
        ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0]));
2088
23.7k
        if (ctxt->pushTab == NULL)
2089
55
            goto mem_error;
2090
23.7k
    }
2091
612k
    ctxt->nameTab[ctxt->nameNr] = value;
2092
612k
    ctxt->name = value;
2093
612k
    tag = &ctxt->pushTab[ctxt->nameNr];
2094
612k
    tag->prefix = prefix;
2095
612k
    tag->URI = URI;
2096
612k
    tag->line = line;
2097
612k
    tag->nsNr = nsNr;
2098
612k
    return (ctxt->nameNr++);
2099
101
mem_error:
2100
101
    xmlErrMemory(ctxt);
2101
101
    return (-1);
2102
612k
}
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
5.67k
{
2113
5.67k
    const xmlChar *ret;
2114
2115
5.67k
    if (ctxt->nameNr <= 0)
2116
0
        return (NULL);
2117
5.67k
    ctxt->nameNr--;
2118
5.67k
    if (ctxt->nameNr > 0)
2119
5.38k
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2120
289
    else
2121
289
        ctxt->name = NULL;
2122
5.67k
    ret = ctxt->nameTab[ctxt->nameNr];
2123
5.67k
    ctxt->nameTab[ctxt->nameNr] = NULL;
2124
5.67k
    return (ret);
2125
5.67k
}
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
534k
{
2139
534k
    const xmlChar *ret;
2140
2141
534k
    if ((ctxt == NULL) || (ctxt->nameNr <= 0))
2142
37
        return (NULL);
2143
534k
    ctxt->nameNr--;
2144
534k
    if (ctxt->nameNr > 0)
2145
527k
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2146
6.55k
    else
2147
6.55k
        ctxt->name = NULL;
2148
534k
    ret = ctxt->nameTab[ctxt->nameNr];
2149
534k
    ctxt->nameTab[ctxt->nameNr] = NULL;
2150
534k
    return (ret);
2151
534k
}
2152
2153
714k
static int spacePush(xmlParserCtxtPtr ctxt, int val) {
2154
714k
    if (ctxt->spaceNr >= ctxt->spaceMax) {
2155
56.5k
        int *tmp;
2156
56.5k
        int newSize;
2157
2158
56.5k
        newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]),
2159
56.5k
                                  10, XML_MAX_ITEMS);
2160
56.5k
        if (newSize < 0) {
2161
0
      xmlErrMemory(ctxt);
2162
0
      return(-1);
2163
0
        }
2164
2165
56.5k
        tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0]));
2166
56.5k
        if (tmp == NULL) {
2167
56
      xmlErrMemory(ctxt);
2168
56
      return(-1);
2169
56
  }
2170
56.5k
  ctxt->spaceTab = tmp;
2171
2172
56.5k
        ctxt->spaceMax = newSize;
2173
56.5k
    }
2174
714k
    ctxt->spaceTab[ctxt->spaceNr] = val;
2175
714k
    ctxt->space = &ctxt->spaceTab[ctxt->spaceNr];
2176
714k
    return(ctxt->spaceNr++);
2177
714k
}
2178
2179
641k
static int spacePop(xmlParserCtxtPtr ctxt) {
2180
641k
    int ret;
2181
641k
    if (ctxt->spaceNr <= 0) return(0);
2182
641k
    ctxt->spaceNr--;
2183
641k
    if (ctxt->spaceNr > 0)
2184
634k
  ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1];
2185
6.48k
    else
2186
6.48k
        ctxt->space = &ctxt->spaceTab[0];
2187
641k
    ret = ctxt->spaceTab[ctxt->spaceNr];
2188
641k
    ctxt->spaceTab[ctxt->spaceNr] = -1;
2189
641k
    return(ret);
2190
641k
}
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
12.7M
#define RAW (*ctxt->input->cur)
2225
1.61G
#define CUR (*ctxt->input->cur)
2226
8.99M
#define NXT(val) ctxt->input->cur[(val)]
2227
2.64G
#define CUR_PTR ctxt->input->cur
2228
1.67M
#define BASE_PTR ctxt->input->base
2229
2230
#define CMP4( s, c1, c2, c3, c4 ) \
2231
9.24M
  ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \
2232
4.66M
    ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 )
2233
#define CMP5( s, c1, c2, c3, c4, c5 ) \
2234
8.80M
  ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 )
2235
#define CMP6( s, c1, c2, c3, c4, c5, c6 ) \
2236
8.05M
  ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 )
2237
#define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \
2238
7.43M
  ( 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
6.99M
  ( 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
3.38M
  ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \
2243
3.38M
    ((unsigned char *) s)[ 8 ] == c9 )
2244
#define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \
2245
3.61k
  ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \
2246
3.61k
    ((unsigned char *) s)[ 9 ] == c10 )
2247
2248
2.33M
#define SKIP(val) do {             \
2249
2.33M
    ctxt->input->cur += (val),ctxt->input->col+=(val);      \
2250
2.33M
    if (*ctxt->input->cur == 0)           \
2251
2.33M
        xmlParserGrow(ctxt);           \
2252
2.33M
  } 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
11.0M
    if (!PARSER_PROGRESSIVE(ctxt)) \
2268
11.0M
  xmlParserShrink(ctxt);
2269
2270
#define GROW \
2271
21.2M
    if ((!PARSER_PROGRESSIVE(ctxt)) && \
2272
21.2M
        (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \
2273
2.90M
  xmlParserGrow(ctxt);
2274
2275
3.05M
#define SKIP_BLANKS xmlSkipBlankChars(ctxt)
2276
2277
845k
#define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt)
2278
2279
342M
#define NEXT xmlNextChar(ctxt)
2280
2281
940k
#define NEXT1 {               \
2282
940k
  ctxt->input->col++;           \
2283
940k
  ctxt->input->cur++;           \
2284
940k
  if (*ctxt->input->cur == 0)         \
2285
940k
      xmlParserGrow(ctxt);           \
2286
940k
    }
2287
2288
1.43G
#define NEXTL(l) do {             \
2289
1.43G
    if (*(ctxt->input->cur) == '\n') {         \
2290
14.4M
  ctxt->input->line++; ctxt->input->col = 1;      \
2291
1.41G
    } else ctxt->input->col++;           \
2292
1.43G
    ctxt->input->cur += l;        \
2293
1.43G
  } while (0)
2294
2295
#define COPY_BUF(b, i, v)           \
2296
134M
    if (v < 0x80) b[i++] = v;           \
2297
134M
    else i += xmlCopyCharMultiByte(&b[i],v)
2298
2299
static int
2300
126M
xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) {
2301
126M
    int c = xmlCurrentChar(ctxt, len);
2302
2303
126M
    if (c == XML_INVALID_CHAR)
2304
20.7M
        c = 0xFFFD; /* replacement character */
2305
2306
126M
    return(c);
2307
126M
}
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
3.27M
xmlSkipBlankChars(xmlParserCtxt *ctxt) {
2319
3.27M
    const xmlChar *cur;
2320
3.27M
    int res = 0;
2321
2322
3.27M
    cur = ctxt->input->cur;
2323
3.84M
    while (IS_BLANK_CH(*cur)) {
2324
3.84M
        if (*cur == '\n') {
2325
2.83M
            ctxt->input->line++; ctxt->input->col = 1;
2326
2.83M
        } else {
2327
1.01M
            ctxt->input->col++;
2328
1.01M
        }
2329
3.84M
        cur++;
2330
3.84M
        if (res < INT_MAX)
2331
3.84M
            res++;
2332
3.84M
        if (*cur == 0) {
2333
10.5k
            ctxt->input->cur = cur;
2334
10.5k
            xmlParserGrow(ctxt);
2335
10.5k
            cur = ctxt->input->cur;
2336
10.5k
        }
2337
3.84M
    }
2338
3.27M
    ctxt->input->cur = cur;
2339
2340
3.27M
    if (res > 4)
2341
33.7k
        GROW;
2342
2343
3.27M
    return(res);
2344
3.27M
}
2345
2346
static void
2347
78.1k
xmlPopPE(xmlParserCtxtPtr ctxt) {
2348
78.1k
    unsigned long consumed;
2349
78.1k
    xmlEntityPtr ent;
2350
2351
78.1k
    ent = ctxt->input->entity;
2352
2353
78.1k
    ent->flags &= ~XML_ENT_EXPANDING;
2354
2355
78.1k
    if ((ent->flags & XML_ENT_CHECKED) == 0) {
2356
7.38k
        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
7.59k
        do {
2363
7.59k
            ctxt->input->cur = ctxt->input->end;
2364
7.59k
            xmlParserShrink(ctxt);
2365
7.59k
            result = xmlParserGrow(ctxt);
2366
7.59k
        } while (result > 0);
2367
2368
7.38k
        consumed = ctxt->input->consumed;
2369
7.38k
        xmlSaturatedAddSizeT(&consumed,
2370
7.38k
                             ctxt->input->end - ctxt->input->base);
2371
2372
7.38k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
2373
2374
        /*
2375
         * Add to sizeentities when parsing an external entity
2376
         * for the first time.
2377
         */
2378
7.38k
        if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
2379
4.47k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
2380
4.47k
        }
2381
2382
7.38k
        ent->flags |= XML_ENT_CHECKED;
2383
7.38k
    }
2384
2385
78.1k
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
2386
2387
78.1k
    xmlParserEntityCheck(ctxt, ent->expandedSize);
2388
2389
78.1k
    GROW;
2390
78.1k
}
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
845k
xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) {
2401
845k
    int res = 0;
2402
845k
    int inParam;
2403
845k
    int expandParam;
2404
2405
845k
    inParam = PARSER_IN_PE(ctxt);
2406
845k
    expandParam = PARSER_EXTERNAL(ctxt);
2407
2408
845k
    if (!inParam && !expandParam)
2409
222k
        return(xmlSkipBlankChars(ctxt));
2410
2411
    /*
2412
     * It's Okay to use CUR/NEXT here since all the blanks are on
2413
     * the ASCII range.
2414
     */
2415
1.83M
    while (PARSER_STOPPED(ctxt) == 0) {
2416
1.83M
        if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */
2417
1.19M
            NEXT;
2418
1.19M
        } else if (CUR == '%') {
2419
24.8k
            if ((expandParam == 0) ||
2420
24.6k
                (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0))
2421
14.2k
                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
10.6k
            xmlParsePERefInternal(ctxt, 0);
2430
2431
10.6k
            inParam = PARSER_IN_PE(ctxt);
2432
10.6k
            expandParam = PARSER_EXTERNAL(ctxt);
2433
616k
        } else if (CUR == 0) {
2434
35.5k
            if (inParam == 0)
2435
93
                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
35.4k
            if (ctxt->input->flags & XML_INPUT_MARKUP_DECL)
2443
27.2k
                break;
2444
2445
8.26k
            xmlPopPE(ctxt);
2446
2447
8.26k
            inParam = PARSER_IN_PE(ctxt);
2448
8.26k
            expandParam = PARSER_EXTERNAL(ctxt);
2449
581k
        } else {
2450
581k
            break;
2451
581k
        }
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
1.21M
        if (res < INT_MAX)
2461
1.21M
            res++;
2462
1.21M
    }
2463
2464
623k
    return(res);
2465
845k
}
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
106k
xmlParseCharRef(xmlParserCtxt *ctxt) {
2530
106k
    int val = 0;
2531
106k
    int count = 0;
2532
2533
    /*
2534
     * Using RAW/CUR/NEXT is okay since we are working on ASCII range here
2535
     */
2536
106k
    if ((RAW == '&') && (NXT(1) == '#') &&
2537
106k
        (NXT(2) == 'x')) {
2538
57.9k
  SKIP(3);
2539
57.9k
  GROW;
2540
229k
  while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) {
2541
188k
      if (count++ > 20) {
2542
6.94k
    count = 0;
2543
6.94k
    GROW;
2544
6.94k
      }
2545
188k
      if ((RAW >= '0') && (RAW <= '9'))
2546
108k
          val = val * 16 + (CUR - '0');
2547
80.6k
      else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20))
2548
41.3k
          val = val * 16 + (CUR - 'a') + 10;
2549
39.3k
      else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20))
2550
21.7k
          val = val * 16 + (CUR - 'A') + 10;
2551
17.5k
      else {
2552
17.5k
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2553
17.5k
    val = 0;
2554
17.5k
    break;
2555
17.5k
      }
2556
171k
      if (val > 0x110000)
2557
81.6k
          val = 0x110000;
2558
2559
171k
      NEXT;
2560
171k
      count++;
2561
171k
  }
2562
57.9k
  if (RAW == ';') {
2563
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2564
40.3k
      ctxt->input->col++;
2565
40.3k
      ctxt->input->cur++;
2566
40.3k
  }
2567
57.9k
    } else if  ((RAW == '&') && (NXT(1) == '#')) {
2568
48.7k
  SKIP(2);
2569
48.7k
  GROW;
2570
152k
  while (RAW != ';') { /* loop blocked by count */
2571
109k
      if (count++ > 20) {
2572
1.02k
    count = 0;
2573
1.02k
    GROW;
2574
1.02k
      }
2575
109k
      if ((RAW >= '0') && (RAW <= '9'))
2576
103k
          val = val * 10 + (CUR - '0');
2577
5.99k
      else {
2578
5.99k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2579
5.99k
    val = 0;
2580
5.99k
    break;
2581
5.99k
      }
2582
103k
      if (val > 0x110000)
2583
10.9k
          val = 0x110000;
2584
2585
103k
      NEXT;
2586
103k
      count++;
2587
103k
  }
2588
48.7k
  if (RAW == ';') {
2589
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2590
42.7k
      ctxt->input->col++;
2591
42.7k
      ctxt->input->cur++;
2592
42.7k
  }
2593
48.7k
    } else {
2594
0
        if (RAW == '&')
2595
0
            SKIP(1);
2596
0
        xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2597
0
    }
2598
2599
    /*
2600
     * [ WFC: Legal Character ]
2601
     * Characters referred to using character references must match the
2602
     * production for Char.
2603
     */
2604
106k
    if (val >= 0x110000) {
2605
893
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2606
893
                "xmlParseCharRef: character reference out of bounds\n",
2607
893
          val);
2608
893
        val = 0xFFFD;
2609
105k
    } else if (!IS_CHAR(val)) {
2610
25.8k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2611
25.8k
                          "xmlParseCharRef: invalid xmlChar value %d\n",
2612
25.8k
                    val);
2613
25.8k
    }
2614
106k
    return(val);
2615
106k
}
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
264k
xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) {
2635
264k
    const xmlChar *ptr;
2636
264k
    xmlChar cur;
2637
264k
    int val = 0;
2638
2639
264k
    if ((str == NULL) || (*str == NULL)) return(0);
2640
264k
    ptr = *str;
2641
264k
    cur = *ptr;
2642
264k
    if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) {
2643
111k
  ptr += 3;
2644
111k
  cur = *ptr;
2645
657k
  while (cur != ';') { /* Non input consuming loop */
2646
546k
      if ((cur >= '0') && (cur <= '9'))
2647
113k
          val = val * 16 + (cur - '0');
2648
433k
      else if ((cur >= 'a') && (cur <= 'f'))
2649
216k
          val = val * 16 + (cur - 'a') + 10;
2650
216k
      else if ((cur >= 'A') && (cur <= 'F'))
2651
216k
          val = val * 16 + (cur - 'A') + 10;
2652
489
      else {
2653
489
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2654
489
    val = 0;
2655
489
    break;
2656
489
      }
2657
546k
      if (val > 0x110000)
2658
659
          val = 0x110000;
2659
2660
546k
      ptr++;
2661
546k
      cur = *ptr;
2662
546k
  }
2663
111k
  if (cur == ';')
2664
110k
      ptr++;
2665
153k
    } else if  ((cur == '&') && (ptr[1] == '#')){
2666
153k
  ptr += 2;
2667
153k
  cur = *ptr;
2668
477k
  while (cur != ';') { /* Non input consuming loops */
2669
324k
      if ((cur >= '0') && (cur <= '9'))
2670
323k
          val = val * 10 + (cur - '0');
2671
867
      else {
2672
867
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2673
867
    val = 0;
2674
867
    break;
2675
867
      }
2676
323k
      if (val > 0x110000)
2677
5.05k
          val = 0x110000;
2678
2679
323k
      ptr++;
2680
323k
      cur = *ptr;
2681
323k
  }
2682
153k
  if (cur == ';')
2683
152k
      ptr++;
2684
153k
    } else {
2685
0
  xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2686
0
  return(0);
2687
0
    }
2688
264k
    *str = ptr;
2689
2690
    /*
2691
     * [ WFC: Legal Character ]
2692
     * Characters referred to using character references must match the
2693
     * production for Char.
2694
     */
2695
264k
    if (val >= 0x110000) {
2696
238
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2697
238
                "xmlParseStringCharRef: character reference out of bounds\n",
2698
238
                val);
2699
264k
    } else if (IS_CHAR(val)) {
2700
262k
        return(val);
2701
262k
    } else {
2702
2.16k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2703
2.16k
        "xmlParseStringCharRef: invalid xmlChar value %d\n",
2704
2.16k
        val);
2705
2.16k
    }
2706
2.40k
    return(0);
2707
264k
}
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
618k
                     int blank_chars) {
2816
618k
    int i;
2817
618k
    xmlNodePtr lastChild;
2818
2819
    /*
2820
     * Check for xml:space value.
2821
     */
2822
618k
    if ((ctxt->space == NULL) || (*(ctxt->space) == 1) ||
2823
617k
        (*(ctxt->space) == -2))
2824
547k
  return(0);
2825
2826
    /*
2827
     * Check that the string is made of blanks
2828
     */
2829
70.8k
    if (blank_chars == 0) {
2830
113k
  for (i = 0;i < len;i++)
2831
109k
      if (!(IS_BLANK_CH(str[i]))) return(0);
2832
56.2k
    }
2833
2834
    /*
2835
     * Look if the element is mixed content in the DTD if available
2836
     */
2837
18.8k
    if (ctxt->node == NULL) return(0);
2838
18.8k
    if (ctxt->myDoc != NULL) {
2839
18.8k
        xmlElementPtr elemDecl = NULL;
2840
18.8k
        xmlDocPtr doc = ctxt->myDoc;
2841
18.8k
        const xmlChar *prefix = NULL;
2842
2843
18.8k
        if (ctxt->node->ns)
2844
9.78k
            prefix = ctxt->node->ns->prefix;
2845
18.8k
        if (doc->intSubset != NULL)
2846
11.2k
            elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name,
2847
11.2k
                                      prefix);
2848
18.8k
        if ((elemDecl == NULL) && (doc->extSubset != NULL))
2849
673
            elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name,
2850
673
                                      prefix);
2851
18.8k
        if (elemDecl != NULL) {
2852
1.82k
            if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)
2853
715
                return(1);
2854
1.10k
            if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) ||
2855
915
                (elemDecl->etype == XML_ELEMENT_TYPE_MIXED))
2856
396
                return(0);
2857
1.10k
        }
2858
18.8k
    }
2859
2860
    /*
2861
     * Otherwise, heuristic :-\
2862
     *
2863
     * When push parsing, we could be at the end of a chunk.
2864
     * This makes the look-ahead and consequently the NOBLANKS
2865
     * option unreliable.
2866
     */
2867
17.6k
    if ((RAW != '<') && (RAW != 0xD)) return(0);
2868
15.8k
    if ((ctxt->node->children == NULL) &&
2869
9.63k
  (RAW == '<') && (NXT(1) == '/')) return(0);
2870
2871
15.3k
    lastChild = xmlGetLastChild(ctxt->node);
2872
15.3k
    if (lastChild == NULL) {
2873
9.09k
        if ((ctxt->node->type != XML_ELEMENT_NODE) &&
2874
0
            (ctxt->node->content != NULL)) return(0);
2875
9.09k
    } else if (xmlNodeIsText(lastChild))
2876
360
        return(0);
2877
5.87k
    else if ((ctxt->node->children != NULL) &&
2878
5.87k
             (xmlNodeIsText(ctxt->node->children)))
2879
273
        return(0);
2880
14.6k
    return(1);
2881
15.3k
}
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
758k
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
758k
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2953
752k
        (((c >= 'a') && (c <= 'z')) ||
2954
197k
         ((c >= 'A') && (c <= 'Z')) ||
2955
186k
         (c == '_') || (c == ':') ||
2956
182k
         ((c >= 0xC0) && (c <= 0xD6)) ||
2957
181k
         ((c >= 0xD8) && (c <= 0xF6)) ||
2958
180k
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2959
178k
         ((c >= 0x370) && (c <= 0x37D)) ||
2960
178k
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2961
161k
         ((c >= 0x200C) && (c <= 0x200D)) ||
2962
161k
         ((c >= 0x2070) && (c <= 0x218F)) ||
2963
158k
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2964
157k
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2965
152k
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2966
151k
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2967
146k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2968
607k
        return(1);
2969
151k
    return(0);
2970
758k
}
2971
2972
static int
2973
27.6M
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
27.6M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2979
27.6M
        (((c >= 'a') && (c <= 'z')) ||
2980
18.1M
         ((c >= 'A') && (c <= 'Z')) ||
2981
17.9M
         ((c >= '0') && (c <= '9')) || /* !start */
2982
17.6M
         (c == '_') || (c == ':') ||
2983
17.5M
         (c == '-') || (c == '.') || (c == 0xB7) || /* !start */
2984
17.4M
         ((c >= 0xC0) && (c <= 0xD6)) ||
2985
17.4M
         ((c >= 0xD8) && (c <= 0xF6)) ||
2986
17.4M
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2987
17.3M
         ((c >= 0x300) && (c <= 0x36F)) || /* !start */
2988
17.3M
         ((c >= 0x370) && (c <= 0x37D)) ||
2989
17.3M
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2990
15.4M
         ((c >= 0x200C) && (c <= 0x200D)) ||
2991
15.4M
         ((c >= 0x203F) && (c <= 0x2040)) || /* !start */
2992
15.4M
         ((c >= 0x2070) && (c <= 0x218F)) ||
2993
5.78M
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2994
5.77M
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2995
3.84M
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2996
3.84M
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2997
607k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2998
26.9M
         return(1);
2999
617k
    return(0);
3000
27.6M
}
3001
3002
static int
3003
374k
xmlIsNameStartCharOld(int c) {
3004
374k
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3005
371k
        ((IS_LETTER(c) || (c == '_') || (c == ':'))))
3006
290k
        return(1);
3007
83.3k
    return(0);
3008
374k
}
3009
3010
static int
3011
10.0M
xmlIsNameCharOld(int c) {
3012
10.0M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3013
10.0M
        ((IS_LETTER(c)) || (IS_DIGIT(c)) ||
3014
453k
         (c == '.') || (c == '-') ||
3015
301k
         (c == '_') || (c == ':') ||
3016
296k
         (IS_COMBINING(c)) ||
3017
295k
         (IS_EXTENDER(c))))
3018
9.72M
        return(1);
3019
297k
    return(0);
3020
10.0M
}
3021
3022
static int
3023
1.13M
xmlIsNameStartChar(int c, int old10) {
3024
1.13M
    if (!old10)
3025
758k
        return(xmlIsNameStartCharNew(c));
3026
374k
    else
3027
374k
        return(xmlIsNameStartCharOld(c));
3028
1.13M
}
3029
3030
static int
3031
37.6M
xmlIsNameChar(int c, int old10) {
3032
37.6M
    if (!old10)
3033
27.6M
        return(xmlIsNameCharNew(c));
3034
10.0M
    else
3035
10.0M
        return(xmlIsNameCharOld(c));
3036
37.6M
}
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
797k
xmlScanName(const xmlChar *ptr, size_t maxSize, int flags) {
3052
797k
    int stop = flags & XML_SCAN_NC ? ':' : 0;
3053
797k
    int old10 = flags & XML_SCAN_OLD10 ? 1 : 0;
3054
3055
8.95M
    while (1) {
3056
8.95M
        int c, len;
3057
3058
8.95M
        c = *ptr;
3059
8.95M
        if (c < 0x80) {
3060
3.65M
            if (c == stop)
3061
1.06k
                break;
3062
3.65M
            len = 1;
3063
5.30M
        } else {
3064
5.30M
            len = 4;
3065
5.30M
            c = xmlGetUTF8Char(ptr, &len);
3066
5.30M
            if (c < 0)
3067
2.76k
                break;
3068
5.30M
        }
3069
3070
8.95M
        if (flags & XML_SCAN_NMTOKEN ?
3071
8.15M
                !xmlIsNameChar(c, old10) :
3072
8.95M
                !xmlIsNameStartChar(c, old10))
3073
793k
            break;
3074
3075
8.15M
        if ((size_t) len > maxSize)
3076
143
            return(NULL);
3077
8.15M
        ptr += len;
3078
8.15M
        maxSize -= len;
3079
8.15M
        flags |= XML_SCAN_NMTOKEN;
3080
8.15M
    }
3081
3082
797k
    return(ptr);
3083
797k
}
3084
3085
static const xmlChar *
3086
134k
xmlParseNameComplex(xmlParserCtxtPtr ctxt) {
3087
134k
    const xmlChar *ret;
3088
134k
    int len = 0, l;
3089
134k
    int c;
3090
134k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3091
64.5k
                    XML_MAX_TEXT_LENGTH :
3092
134k
                    XML_MAX_NAME_LENGTH;
3093
134k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3094
3095
    /*
3096
     * Handler for more complex cases
3097
     */
3098
134k
    c = xmlCurrentChar(ctxt, &l);
3099
134k
    if (!xmlIsNameStartChar(c, old10))
3100
87.3k
        return(NULL);
3101
47.0k
    len += l;
3102
47.0k
    NEXTL(l);
3103
47.0k
    c = xmlCurrentChar(ctxt, &l);
3104
5.08M
    while (xmlIsNameChar(c, old10)) {
3105
5.04M
        if (len <= INT_MAX - l)
3106
5.04M
            len += l;
3107
5.04M
        NEXTL(l);
3108
5.04M
        c = xmlCurrentChar(ctxt, &l);
3109
5.04M
    }
3110
47.0k
    if (len > maxLength) {
3111
89
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3112
89
        return(NULL);
3113
89
    }
3114
46.9k
    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
46.9k
    if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r'))
3125
478
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len);
3126
46.4k
    else
3127
46.4k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len);
3128
46.9k
    if (ret == NULL)
3129
6
        xmlErrMemory(ctxt);
3130
46.9k
    return(ret);
3131
46.9k
}
3132
3133
/**
3134
 * Parse an XML name.
3135
 *
3136
 * @deprecated Internal function, don't use.
3137
 *
3138
 *     [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' |
3139
 *                      CombiningChar | Extender
3140
 *
3141
 *     [5] Name ::= (Letter | '_' | ':') (NameChar)*
3142
 *
3143
 *     [6] Names ::= Name (#x20 Name)*
3144
 *
3145
 * @param ctxt  an XML parser context
3146
 * @returns the Name parsed or NULL
3147
 */
3148
3149
const xmlChar *
3150
1.66M
xmlParseName(xmlParserCtxt *ctxt) {
3151
1.66M
    const xmlChar *in;
3152
1.66M
    const xmlChar *ret;
3153
1.66M
    size_t count = 0;
3154
1.66M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3155
934k
                       XML_MAX_TEXT_LENGTH :
3156
1.66M
                       XML_MAX_NAME_LENGTH;
3157
3158
1.66M
    GROW;
3159
3160
    /*
3161
     * Accelerator for simple ASCII names
3162
     */
3163
1.66M
    in = ctxt->input->cur;
3164
1.66M
    if (((*in >= 0x61) && (*in <= 0x7A)) ||
3165
153k
  ((*in >= 0x41) && (*in <= 0x5A)) ||
3166
1.56M
  (*in == '_') || (*in == ':')) {
3167
1.56M
  in++;
3168
9.04M
  while (((*in >= 0x61) && (*in <= 0x7A)) ||
3169
2.27M
         ((*in >= 0x41) && (*in <= 0x5A)) ||
3170
2.07M
         ((*in >= 0x30) && (*in <= 0x39)) ||
3171
1.74M
         (*in == '_') || (*in == '-') ||
3172
1.69M
         (*in == ':') || (*in == '.'))
3173
7.48M
      in++;
3174
1.56M
  if ((*in > 0) && (*in < 0x80)) {
3175
1.53M
      count = in - ctxt->input->cur;
3176
1.53M
            if (count > maxLength) {
3177
95
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3178
95
                return(NULL);
3179
95
            }
3180
1.53M
      ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count);
3181
1.53M
      ctxt->input->cur = in;
3182
1.53M
      ctxt->input->col += count;
3183
1.53M
      if (ret == NULL)
3184
10
          xmlErrMemory(ctxt);
3185
1.53M
      return(ret);
3186
1.53M
  }
3187
1.56M
    }
3188
    /* accelerator for special cases */
3189
134k
    return(xmlParseNameComplex(ctxt));
3190
1.66M
}
3191
3192
static xmlHashedString
3193
216k
xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) {
3194
216k
    xmlHashedString ret;
3195
216k
    int len = 0, l;
3196
216k
    int c;
3197
216k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3198
116k
                    XML_MAX_TEXT_LENGTH :
3199
216k
                    XML_MAX_NAME_LENGTH;
3200
216k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3201
216k
    size_t startPosition = 0;
3202
3203
216k
    ret.name = NULL;
3204
216k
    ret.hashValue = 0;
3205
3206
    /*
3207
     * Handler for more complex cases
3208
     */
3209
216k
    startPosition = CUR_PTR - BASE_PTR;
3210
216k
    c = xmlCurrentChar(ctxt, &l);
3211
216k
    if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3212
203k
  (!xmlIsNameStartChar(c, old10) || (c == ':'))) {
3213
161k
  return(ret);
3214
161k
    }
3215
3216
14.9M
    while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */
3217
14.9M
     (xmlIsNameChar(c, old10) && (c != ':'))) {
3218
14.9M
        if (len <= INT_MAX - l)
3219
14.9M
      len += l;
3220
14.9M
  NEXTL(l);
3221
14.9M
  c = xmlCurrentChar(ctxt, &l);
3222
14.9M
    }
3223
54.6k
    if (len > maxLength) {
3224
226
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3225
226
        return(ret);
3226
226
    }
3227
54.3k
    ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len);
3228
54.3k
    if (ret.name == NULL)
3229
7
        xmlErrMemory(ctxt);
3230
54.3k
    return(ret);
3231
54.6k
}
3232
3233
/**
3234
 * Parse an XML name.
3235
 *
3236
 *     [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' |
3237
 *                          CombiningChar | Extender
3238
 *
3239
 *     [5NS] NCName ::= (Letter | '_') (NCNameChar)*
3240
 *
3241
 * @param ctxt  an XML parser context
3242
 * @returns the Name parsed or NULL
3243
 */
3244
3245
static xmlHashedString
3246
1.18M
xmlParseNCName(xmlParserCtxtPtr ctxt) {
3247
1.18M
    const xmlChar *in, *e;
3248
1.18M
    xmlHashedString ret;
3249
1.18M
    size_t count = 0;
3250
1.18M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3251
657k
                       XML_MAX_TEXT_LENGTH :
3252
1.18M
                       XML_MAX_NAME_LENGTH;
3253
3254
1.18M
    ret.name = NULL;
3255
3256
    /*
3257
     * Accelerator for simple ASCII names
3258
     */
3259
1.18M
    in = ctxt->input->cur;
3260
1.18M
    e = ctxt->input->end;
3261
1.18M
    if ((((*in >= 0x61) && (*in <= 0x7A)) ||
3262
220k
   ((*in >= 0x41) && (*in <= 0x5A)) ||
3263
995k
   (*in == '_')) && (in < e)) {
3264
995k
  in++;
3265
12.1M
  while ((((*in >= 0x61) && (*in <= 0x7A)) ||
3266
1.59M
          ((*in >= 0x41) && (*in <= 0x5A)) ||
3267
1.25M
          ((*in >= 0x30) && (*in <= 0x39)) ||
3268
1.19M
          (*in == '_') || (*in == '-') ||
3269
11.1M
          (*in == '.')) && (in < e))
3270
11.1M
      in++;
3271
995k
  if (in >= e)
3272
2.34k
      goto complex;
3273
993k
  if ((*in > 0) && (*in < 0x80)) {
3274
966k
      count = in - ctxt->input->cur;
3275
966k
            if (count > maxLength) {
3276
127
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3277
127
                return(ret);
3278
127
            }
3279
966k
      ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count);
3280
966k
      ctxt->input->cur = in;
3281
966k
      ctxt->input->col += count;
3282
966k
      if (ret.name == NULL) {
3283
4
          xmlErrMemory(ctxt);
3284
4
      }
3285
966k
      return(ret);
3286
966k
  }
3287
993k
    }
3288
216k
complex:
3289
216k
    return(xmlParseNCNameComplex(ctxt));
3290
1.18M
}
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
53.7k
xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) {
3304
53.7k
    register const xmlChar *cmp = other;
3305
53.7k
    register const xmlChar *in;
3306
53.7k
    const xmlChar *ret;
3307
3308
53.7k
    GROW;
3309
3310
53.7k
    in = ctxt->input->cur;
3311
128k
    while (*in != 0 && *in == *cmp) {
3312
75.1k
  ++in;
3313
75.1k
  ++cmp;
3314
75.1k
    }
3315
53.7k
    if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
3316
  /* success */
3317
29.7k
  ctxt->input->col += in - ctxt->input->cur;
3318
29.7k
  ctxt->input->cur = in;
3319
29.7k
  return (const xmlChar*) 1;
3320
29.7k
    }
3321
    /* failure (or end of input buffer), check with full function */
3322
23.9k
    ret = xmlParseName (ctxt);
3323
    /* strings coming from the dictionary direct compare possible */
3324
23.9k
    if (ret == other) {
3325
1.77k
  return (const xmlChar*) 1;
3326
1.77k
    }
3327
22.1k
    return ret;
3328
23.9k
}
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
792k
xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) {
3341
792k
    xmlChar *ret;
3342
792k
    const xmlChar *cur = *str;
3343
792k
    int flags = 0;
3344
792k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3345
255k
                    XML_MAX_TEXT_LENGTH :
3346
792k
                    XML_MAX_NAME_LENGTH;
3347
3348
792k
    if (ctxt->options & XML_PARSE_OLD10)
3349
266k
        flags |= XML_SCAN_OLD10;
3350
3351
792k
    cur = xmlScanName(*str, maxLength, flags);
3352
792k
    if (cur == NULL) {
3353
143
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3354
143
        return(NULL);
3355
143
    }
3356
792k
    if (cur == *str)
3357
4.49k
        return(NULL);
3358
3359
787k
    ret = xmlStrndup(*str, cur - *str);
3360
787k
    if (ret == NULL)
3361
89
        xmlErrMemory(ctxt);
3362
787k
    *str = cur;
3363
787k
    return(ret);
3364
792k
}
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
39.9k
xmlParseNmtoken(xmlParserCtxt *ctxt) {
3381
39.9k
    xmlChar buf[XML_MAX_NAMELEN + 5];
3382
39.9k
    xmlChar *ret;
3383
39.9k
    int len = 0, l;
3384
39.9k
    int c;
3385
39.9k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3386
16.3k
                    XML_MAX_TEXT_LENGTH :
3387
39.9k
                    XML_MAX_NAME_LENGTH;
3388
39.9k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3389
3390
39.9k
    c = xmlCurrentChar(ctxt, &l);
3391
3392
205k
    while (xmlIsNameChar(c, old10)) {
3393
166k
  COPY_BUF(buf, len, c);
3394
166k
  NEXTL(l);
3395
166k
  c = xmlCurrentChar(ctxt, &l);
3396
166k
  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
1.17k
      xmlChar *buffer;
3402
1.17k
      int max = len * 2;
3403
3404
1.17k
      buffer = xmlMalloc(max);
3405
1.17k
      if (buffer == NULL) {
3406
6
          xmlErrMemory(ctxt);
3407
6
    return(NULL);
3408
6
      }
3409
1.16k
      memcpy(buffer, buf, len);
3410
9.24M
      while (xmlIsNameChar(c, old10)) {
3411
9.24M
    if (len + 10 > max) {
3412
5.00k
        xmlChar *tmp;
3413
5.00k
                    int newSize;
3414
3415
5.00k
                    newSize = xmlGrowCapacity(max, 1, 1, maxLength);
3416
5.00k
                    if (newSize < 0) {
3417
163
                        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3418
163
                        xmlFree(buffer);
3419
163
                        return(NULL);
3420
163
                    }
3421
4.83k
        tmp = xmlRealloc(buffer, newSize);
3422
4.83k
        if (tmp == NULL) {
3423
6
      xmlErrMemory(ctxt);
3424
6
      xmlFree(buffer);
3425
6
      return(NULL);
3426
6
        }
3427
4.83k
        buffer = tmp;
3428
4.83k
                    max = newSize;
3429
4.83k
    }
3430
9.24M
    COPY_BUF(buffer, len, c);
3431
9.24M
    NEXTL(l);
3432
9.24M
    c = xmlCurrentChar(ctxt, &l);
3433
9.24M
      }
3434
998
      buffer[len] = 0;
3435
998
      return(buffer);
3436
1.16k
  }
3437
166k
    }
3438
38.7k
    if (len == 0)
3439
5.97k
        return(NULL);
3440
32.8k
    if (len > maxLength) {
3441
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3442
0
        return(NULL);
3443
0
    }
3444
32.8k
    ret = xmlStrndup(buf, len);
3445
32.8k
    if (ret == NULL)
3446
18
        xmlErrMemory(ctxt);
3447
32.8k
    return(ret);
3448
32.8k
}
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
75.1k
                          const xmlChar *str, int length, int depth) {
3462
75.1k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3463
75.1k
    const xmlChar *end, *chunk;
3464
75.1k
    int c, l;
3465
3466
75.1k
    if (str == NULL)
3467
19.0k
        return;
3468
3469
56.0k
    depth += 1;
3470
56.0k
    if (depth > maxDepth) {
3471
3
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3472
3
                       "Maximum entity nesting depth exceeded");
3473
3
  return;
3474
3
    }
3475
3476
56.0k
    end = str + length;
3477
56.0k
    chunk = str;
3478
3479
477M
    while ((str < end) && (!PARSER_STOPPED(ctxt))) {
3480
477M
        c = *str;
3481
3482
477M
        if (c >= 0x80) {
3483
406M
            l = xmlUTF8MultibyteLen(ctxt, str,
3484
406M
                    "invalid character in entity value\n");
3485
406M
            if (l == 0) {
3486
51.8M
                if (chunk < str)
3487
64.2k
                    xmlSBufAddString(buf, chunk, str - chunk);
3488
51.8M
                xmlSBufAddReplChar(buf);
3489
51.8M
                str += 1;
3490
51.8M
                chunk = str;
3491
354M
            } else {
3492
354M
                str += l;
3493
354M
            }
3494
406M
        } else if (c == '&') {
3495
50.3k
            if (str[1] == '#') {
3496
13.7k
                if (chunk < str)
3497
6.44k
                    xmlSBufAddString(buf, chunk, str - chunk);
3498
3499
13.7k
                c = xmlParseStringCharRef(ctxt, &str);
3500
13.7k
                if (c == 0)
3501
2.36k
                    return;
3502
3503
11.3k
                xmlSBufAddChar(buf, c);
3504
3505
11.3k
                chunk = str;
3506
36.5k
            } else {
3507
36.5k
                xmlChar *name;
3508
3509
                /*
3510
                 * General entity references are checked for
3511
                 * syntactic validity.
3512
                 */
3513
36.5k
                str++;
3514
36.5k
                name = xmlParseStringName(ctxt, &str);
3515
3516
36.5k
                if ((name == NULL) || (*str++ != ';')) {
3517
1.32k
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR,
3518
1.32k
                            "EntityValue: '&' forbidden except for entities "
3519
1.32k
                            "references\n");
3520
1.32k
                    xmlFree(name);
3521
1.32k
                    return;
3522
1.32k
                }
3523
3524
35.2k
                xmlFree(name);
3525
35.2k
            }
3526
71.4M
        } else if (c == '%') {
3527
44.3k
            xmlEntityPtr ent;
3528
3529
44.3k
            if (chunk < str)
3530
7.20k
                xmlSBufAddString(buf, chunk, str - chunk);
3531
3532
44.3k
            ent = xmlParseStringPEReference(ctxt, &str);
3533
44.3k
            if (ent == NULL)
3534
10.7k
                return;
3535
3536
33.5k
            if (!PARSER_EXTERNAL(ctxt)) {
3537
326
                xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL);
3538
326
                return;
3539
326
            }
3540
3541
33.2k
            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
19.3k
                if (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
3549
19.3k
                    ((ctxt->replaceEntities) ||
3550
18.9k
                     (ctxt->validate))) {
3551
18.9k
                    xmlLoadEntityContent(ctxt, ent);
3552
18.9k
                } else {
3553
336
                    xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
3554
336
                                  "not validating will not read content for "
3555
336
                                  "PE entity %s\n", ent->name, NULL);
3556
336
                }
3557
19.3k
            }
3558
3559
            /*
3560
             * TODO: Skip if ent->content is still NULL.
3561
             */
3562
3563
33.2k
            if (xmlParserEntityCheck(ctxt, ent->length))
3564
11
                return;
3565
3566
33.2k
            if (ent->flags & XML_ENT_EXPANDING) {
3567
143
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3568
143
                return;
3569
143
            }
3570
3571
33.1k
            ent->flags |= XML_ENT_EXPANDING;
3572
33.1k
            xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length,
3573
33.1k
                                      depth);
3574
33.1k
            ent->flags &= ~XML_ENT_EXPANDING;
3575
3576
33.1k
            chunk = str;
3577
71.4M
        } else {
3578
            /* Normal ASCII char */
3579
71.4M
            if (!IS_BYTE_CHAR(c)) {
3580
3.96M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3581
3.96M
                        "invalid character in entity value\n");
3582
3.96M
                if (chunk < str)
3583
20.5k
                    xmlSBufAddString(buf, chunk, str - chunk);
3584
3.96M
                xmlSBufAddReplChar(buf);
3585
3.96M
                str += 1;
3586
3.96M
                chunk = str;
3587
67.4M
            } else {
3588
67.4M
                str += 1;
3589
67.4M
            }
3590
71.4M
        }
3591
477M
    }
3592
3593
41.1k
    if (chunk < str)
3594
35.0k
        xmlSBufAddString(buf, chunk, str - chunk);
3595
41.1k
}
3596
3597
/**
3598
 * Parse a value for ENTITY declarations
3599
 *
3600
 * @deprecated Internal function, don't use.
3601
 *
3602
 *     [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' |
3603
 *                         "'" ([^%&'] | PEReference | Reference)* "'"
3604
 *
3605
 * @param ctxt  an XML parser context
3606
 * @param orig  if non-NULL store a copy of the original entity value
3607
 * @returns the EntityValue parsed with reference substituted or NULL
3608
 */
3609
xmlChar *
3610
43.6k
xmlParseEntityValue(xmlParserCtxt *ctxt, xmlChar **orig) {
3611
43.6k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3612
20.2k
                         XML_MAX_HUGE_LENGTH :
3613
43.6k
                         XML_MAX_TEXT_LENGTH;
3614
43.6k
    xmlSBuf buf;
3615
43.6k
    const xmlChar *start;
3616
43.6k
    int quote, length;
3617
3618
43.6k
    xmlSBufInit(&buf, maxLength);
3619
3620
43.6k
    GROW;
3621
3622
43.6k
    quote = CUR;
3623
43.6k
    if ((quote != '"') && (quote != '\'')) {
3624
0
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
3625
0
  return(NULL);
3626
0
    }
3627
43.6k
    CUR_PTR++;
3628
3629
43.6k
    length = 0;
3630
3631
    /*
3632
     * Copy raw content of the entity into a buffer
3633
     */
3634
1.12G
    while (1) {
3635
1.12G
        int c;
3636
3637
1.12G
        if (PARSER_STOPPED(ctxt))
3638
15
            goto error;
3639
3640
1.12G
        if (CUR_PTR >= ctxt->input->end) {
3641
1.35k
            xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL);
3642
1.35k
            goto error;
3643
1.35k
        }
3644
3645
1.12G
        c = CUR;
3646
3647
1.12G
        if (c == 0) {
3648
282
            xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3649
282
                    "invalid character in entity value\n");
3650
282
            goto error;
3651
282
        }
3652
1.12G
        if (c == quote)
3653
42.0k
            break;
3654
1.12G
        NEXTL(1);
3655
1.12G
        length += 1;
3656
3657
        /*
3658
         * TODO: Check growth threshold
3659
         */
3660
1.12G
        if (ctxt->input->end - CUR_PTR < 10)
3661
98.0k
            GROW;
3662
1.12G
    }
3663
3664
42.0k
    start = CUR_PTR - length;
3665
3666
42.0k
    if (orig != NULL) {
3667
42.0k
        *orig = xmlStrndup(start, length);
3668
42.0k
        if (*orig == NULL)
3669
52
            xmlErrMemory(ctxt);
3670
42.0k
    }
3671
3672
42.0k
    xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr);
3673
3674
42.0k
    NEXTL(1);
3675
3676
42.0k
    return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long"));
3677
3678
1.65k
error:
3679
1.65k
    xmlSBufCleanup(&buf, ctxt, "entity length too long");
3680
1.65k
    return(NULL);
3681
43.6k
}
3682
3683
/**
3684
 * Check an entity reference in an attribute value for validity
3685
 * without expanding it.
3686
 *
3687
 * @param ctxt  parser context
3688
 * @param pent  entity
3689
 * @param depth  nesting depth
3690
 */
3691
static void
3692
6.56k
xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) {
3693
6.56k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3694
6.56k
    const xmlChar *str;
3695
6.56k
    unsigned long expandedSize = pent->length;
3696
6.56k
    int c, flags;
3697
3698
6.56k
    depth += 1;
3699
6.56k
    if (depth > maxDepth) {
3700
3
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3701
3
                       "Maximum entity nesting depth exceeded");
3702
3
  return;
3703
3
    }
3704
3705
6.55k
    if (pent->flags & XML_ENT_EXPANDING) {
3706
21
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3707
21
        return;
3708
21
    }
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
6.53k
    if (ctxt->inSubset == 0)
3716
6.40k
        flags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
3717
131
    else
3718
131
        flags = XML_ENT_VALIDATED;
3719
3720
6.53k
    str = pent->content;
3721
6.53k
    if (str == NULL)
3722
3
        goto done;
3723
3724
    /*
3725
     * Note that entity values are already validated. We only check
3726
     * for illegal less-than signs and compute the expanded size
3727
     * of the entity. No special handling for multi-byte characters
3728
     * is needed.
3729
     */
3730
116M
    while (!PARSER_STOPPED(ctxt)) {
3731
116M
        c = *str;
3732
3733
116M
  if (c != '&') {
3734
116M
            if (c == 0)
3735
6.32k
                break;
3736
3737
116M
            if (c == '<')
3738
1.60k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3739
1.60k
                        "'<' in entity '%s' is not allowed in attributes "
3740
1.60k
                        "values\n", pent->name);
3741
3742
116M
            str += 1;
3743
116M
        } else if (str[1] == '#') {
3744
1.95k
            int val;
3745
3746
1.95k
      val = xmlParseStringCharRef(ctxt, &str);
3747
1.95k
      if (val == 0) {
3748
21
                pent->content[0] = 0;
3749
21
                break;
3750
21
            }
3751
13.8k
  } else {
3752
13.8k
            xmlChar *name;
3753
13.8k
            xmlEntityPtr ent;
3754
3755
13.8k
      name = xmlParseStringEntityRef(ctxt, &str);
3756
13.8k
      if (name == NULL) {
3757
29
                pent->content[0] = 0;
3758
29
                break;
3759
29
            }
3760
3761
13.8k
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3762
13.8k
            xmlFree(name);
3763
3764
13.8k
            if ((ent != NULL) &&
3765
12.3k
                (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) {
3766
11.0k
                if ((ent->flags & flags) != flags) {
3767
5.38k
                    pent->flags |= XML_ENT_EXPANDING;
3768
5.38k
                    xmlCheckEntityInAttValue(ctxt, ent, depth);
3769
5.38k
                    pent->flags &= ~XML_ENT_EXPANDING;
3770
5.38k
                }
3771
3772
11.0k
                xmlSaturatedAdd(&expandedSize, ent->expandedSize);
3773
11.0k
                xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST);
3774
11.0k
            }
3775
13.8k
        }
3776
116M
    }
3777
3778
6.53k
done:
3779
6.53k
    if (ctxt->inSubset == 0)
3780
6.40k
        pent->expandedSize = expandedSize;
3781
3782
6.53k
    pent->flags |= flags;
3783
6.53k
}
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
633k
                          int *inSpace, int depth, int check) {
3803
633k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3804
633k
    int c, chunkSize;
3805
633k
    int normChange = 0;
3806
3807
633k
    if (str == NULL)
3808
36
        return(0);
3809
3810
633k
    depth += 1;
3811
633k
    if (depth > maxDepth) {
3812
3
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3813
3
                       "Maximum entity nesting depth exceeded");
3814
3
  return(0);
3815
3
    }
3816
3817
633k
    if (pent != NULL) {
3818
633k
        if (pent->flags & XML_ENT_EXPANDING) {
3819
17
            xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3820
17
            return(0);
3821
17
        }
3822
3823
633k
        if (check) {
3824
633k
            if (xmlParserEntityCheck(ctxt, pent->length))
3825
419
                return(0);
3826
633k
        }
3827
633k
    }
3828
3829
632k
    chunkSize = 0;
3830
3831
    /*
3832
     * Note that entity values are already validated. No special
3833
     * handling for multi-byte characters is needed.
3834
     */
3835
4.67G
    while (!PARSER_STOPPED(ctxt)) {
3836
4.67G
        c = *str;
3837
3838
4.67G
  if (c != '&') {
3839
4.67G
            if (c == 0)
3840
527k
                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
4.67G
            if ((pent != NULL) && (c == '<')) {
3848
105k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3849
105k
                        "'<' in entity '%s' is not allowed in attributes "
3850
105k
                        "values\n", pent->name);
3851
105k
                break;
3852
105k
            }
3853
3854
4.67G
            if (c <= 0x20) {
3855
18.0M
                if ((normalize) && (*inSpace)) {
3856
                    /* Skip char */
3857
1.12M
                    if (chunkSize > 0) {
3858
197k
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3859
197k
                        chunkSize = 0;
3860
197k
                    }
3861
1.12M
                    normChange = 1;
3862
16.9M
                } else if (c < 0x20) {
3863
15.8M
                    if (chunkSize > 0) {
3864
81.3k
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3865
81.3k
                        chunkSize = 0;
3866
81.3k
                    }
3867
3868
15.8M
                    xmlSBufAddCString(buf, " ", 1);
3869
15.8M
                } else {
3870
1.05M
                    chunkSize += 1;
3871
1.05M
                }
3872
3873
18.0M
                *inSpace = 1;
3874
4.65G
            } else {
3875
4.65G
                chunkSize += 1;
3876
4.65G
                *inSpace = 0;
3877
4.65G
            }
3878
3879
4.67G
            str += 1;
3880
4.67G
        } else if (str[1] == '#') {
3881
249k
            int val;
3882
3883
249k
            if (chunkSize > 0) {
3884
246k
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3885
246k
                chunkSize = 0;
3886
246k
            }
3887
3888
249k
      val = xmlParseStringCharRef(ctxt, &str);
3889
249k
      if (val == 0) {
3890
18
                if (pent != NULL)
3891
18
                    pent->content[0] = 0;
3892
18
                break;
3893
18
            }
3894
3895
249k
            if (val == ' ') {
3896
4.28k
                if ((normalize) && (*inSpace))
3897
194
                    normChange = 1;
3898
4.08k
                else
3899
4.08k
                    xmlSBufAddCString(buf, " ", 1);
3900
4.28k
                *inSpace = 1;
3901
244k
            } else {
3902
244k
                xmlSBufAddChar(buf, val);
3903
244k
                *inSpace = 0;
3904
244k
            }
3905
697k
  } else {
3906
697k
            xmlChar *name;
3907
697k
            xmlEntityPtr ent;
3908
3909
697k
            if (chunkSize > 0) {
3910
159k
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3911
159k
                chunkSize = 0;
3912
159k
            }
3913
3914
697k
      name = xmlParseStringEntityRef(ctxt, &str);
3915
697k
            if (name == NULL) {
3916
35
                if (pent != NULL)
3917
35
                    pent->content[0] = 0;
3918
35
                break;
3919
35
            }
3920
3921
697k
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3922
697k
            xmlFree(name);
3923
3924
697k
      if ((ent != NULL) &&
3925
547k
    (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
3926
138k
    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
138k
                xmlSBufAddString(buf, ent->content, ent->length);
3933
3934
138k
                *inSpace = 0;
3935
558k
      } else if ((ent != NULL) && (ent->content != NULL)) {
3936
408k
                if (pent != NULL)
3937
408k
                    pent->flags |= XML_ENT_EXPANDING;
3938
408k
    normChange |= xmlExpandEntityInAttValue(ctxt, buf,
3939
408k
                        ent->content, ent, normalize, inSpace, depth, check);
3940
408k
                if (pent != NULL)
3941
408k
                    pent->flags &= ~XML_ENT_EXPANDING;
3942
408k
      }
3943
697k
        }
3944
4.67G
    }
3945
3946
632k
    if (chunkSize > 0)
3947
276k
        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3948
3949
632k
    return(normChange);
3950
633k
}
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
338k
                         int special, int isNamespace) {
4020
338k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4021
157k
                         XML_MAX_HUGE_LENGTH :
4022
338k
                         XML_MAX_TEXT_LENGTH;
4023
338k
    xmlSBuf buf;
4024
338k
    xmlChar *ret;
4025
338k
    int c, l, quote, entFlags, chunkSize;
4026
338k
    int inSpace = 1;
4027
338k
    int replaceEntities;
4028
338k
    int normalize = (special & XML_SPECIAL_TYPE_MASK) > XML_ATTRIBUTE_CDATA;
4029
338k
    int attvalFlags = 0;
4030
4031
    /* Always expand namespace URIs */
4032
338k
    replaceEntities = (ctxt->replaceEntities) || (isNamespace);
4033
4034
338k
    xmlSBufInit(&buf, maxLength);
4035
4036
338k
    GROW;
4037
4038
338k
    quote = CUR;
4039
338k
    if ((quote != '"') && (quote != '\'')) {
4040
5.67k
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
4041
5.67k
  return(NULL);
4042
5.67k
    }
4043
332k
    NEXTL(1);
4044
4045
332k
    if (ctxt->inSubset == 0)
4046
271k
        entFlags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
4047
60.9k
    else
4048
60.9k
        entFlags = XML_ENT_VALIDATED;
4049
4050
332k
    inSpace = 1;
4051
332k
    chunkSize = 0;
4052
4053
145M
    while (1) {
4054
145M
        if (PARSER_STOPPED(ctxt))
4055
590
            goto error;
4056
4057
145M
        if (CUR_PTR >= ctxt->input->end) {
4058
10.0k
            xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED,
4059
10.0k
                           "AttValue: ' expected\n");
4060
10.0k
            goto error;
4061
10.0k
        }
4062
4063
        /*
4064
         * TODO: Check growth threshold
4065
         */
4066
145M
        if (ctxt->input->end - CUR_PTR < 10)
4067
138k
            GROW;
4068
4069
145M
        c = CUR;
4070
4071
145M
        if (c >= 0x80) {
4072
93.3M
            l = xmlUTF8MultibyteLen(ctxt, CUR_PTR,
4073
93.3M
                    "invalid character in attribute value\n");
4074
93.3M
            if (l == 0) {
4075
34.1M
                if (chunkSize > 0) {
4076
94.5k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4077
94.5k
                    chunkSize = 0;
4078
94.5k
                }
4079
34.1M
                xmlSBufAddReplChar(&buf);
4080
34.1M
                NEXTL(1);
4081
59.1M
            } else {
4082
59.1M
                chunkSize += l;
4083
59.1M
                NEXTL(l);
4084
59.1M
            }
4085
4086
93.3M
            inSpace = 0;
4087
93.3M
        } else if (c != '&') {
4088
50.9M
            if (c > 0x20) {
4089
8.45M
                if (c == quote)
4090
316k
                    break;
4091
4092
8.14M
                if (c == '<')
4093
107k
                    xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL);
4094
4095
8.14M
                chunkSize += 1;
4096
8.14M
                inSpace = 0;
4097
42.4M
            } else if (!IS_BYTE_CHAR(c)) {
4098
38.8M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
4099
38.8M
                        "invalid character in attribute value\n");
4100
38.8M
                if (chunkSize > 0) {
4101
56.3k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4102
56.3k
                    chunkSize = 0;
4103
56.3k
                }
4104
38.8M
                xmlSBufAddReplChar(&buf);
4105
38.8M
                inSpace = 0;
4106
38.8M
            } else {
4107
                /* Whitespace */
4108
3.63M
                if ((normalize) && (inSpace)) {
4109
                    /* Skip char */
4110
954k
                    if (chunkSize > 0) {
4111
6.11k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4112
6.11k
                        chunkSize = 0;
4113
6.11k
                    }
4114
954k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4115
2.67M
                } else if (c < 0x20) {
4116
                    /* Convert to space */
4117
2.42M
                    if (chunkSize > 0) {
4118
50.8k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4119
50.8k
                        chunkSize = 0;
4120
50.8k
                    }
4121
4122
2.42M
                    xmlSBufAddCString(&buf, " ", 1);
4123
2.42M
                } else {
4124
258k
                    chunkSize += 1;
4125
258k
                }
4126
4127
3.63M
                inSpace = 1;
4128
4129
3.63M
                if ((c == 0xD) && (NXT(1) == 0xA))
4130
1.73k
                    CUR_PTR++;
4131
3.63M
            }
4132
4133
50.6M
            NEXTL(1);
4134
50.6M
        } else if (NXT(1) == '#') {
4135
42.9k
            int val;
4136
4137
42.9k
            if (chunkSize > 0) {
4138
20.1k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4139
20.1k
                chunkSize = 0;
4140
20.1k
            }
4141
4142
42.9k
            val = xmlParseCharRef(ctxt);
4143
42.9k
            if (val == 0)
4144
4.89k
                goto error;
4145
4146
38.0k
            if ((val == '&') && (!replaceEntities)) {
4147
                /*
4148
                 * The reparsing will be done in xmlNodeParseContent()
4149
                 * called from SAX2.c
4150
                 */
4151
10.4k
                xmlSBufAddCString(&buf, "&#38;", 5);
4152
10.4k
                inSpace = 0;
4153
27.5k
            } else if (val == ' ') {
4154
7.58k
                if ((normalize) && (inSpace))
4155
237
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4156
7.34k
                else
4157
7.34k
                    xmlSBufAddCString(&buf, " ", 1);
4158
7.58k
                inSpace = 1;
4159
20.0k
            } else {
4160
20.0k
                xmlSBufAddChar(&buf, val);
4161
20.0k
                inSpace = 0;
4162
20.0k
            }
4163
807k
        } else {
4164
807k
            const xmlChar *name;
4165
807k
            xmlEntityPtr ent;
4166
4167
807k
            if (chunkSize > 0) {
4168
89.9k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4169
89.9k
                chunkSize = 0;
4170
89.9k
            }
4171
4172
807k
            name = xmlParseEntityRefInternal(ctxt);
4173
807k
            if (name == NULL) {
4174
                /*
4175
                 * Probably a literal '&' which wasn't escaped.
4176
                 * TODO: Handle gracefully in recovery mode.
4177
                 */
4178
48.5k
                continue;
4179
48.5k
            }
4180
4181
758k
            ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1);
4182
758k
            if (ent == NULL)
4183
94.9k
                continue;
4184
4185
664k
            if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) {
4186
10.5k
                if ((ent->content[0] == '&') && (!replaceEntities))
4187
4.31k
                    xmlSBufAddCString(&buf, "&#38;", 5);
4188
6.25k
                else
4189
6.25k
                    xmlSBufAddString(&buf, ent->content, ent->length);
4190
10.5k
                inSpace = 0;
4191
653k
            } else if (replaceEntities) {
4192
224k
                if (xmlExpandEntityInAttValue(ctxt, &buf,
4193
224k
                        ent->content, ent, normalize, &inSpace, ctxt->inputNr,
4194
224k
                        /* check */ 1) > 0)
4195
67.3k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4196
428k
            } else {
4197
428k
                if ((ent->flags & entFlags) != entFlags)
4198
1.18k
                    xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr);
4199
4200
428k
                if (xmlParserEntityCheck(ctxt, ent->expandedSize)) {
4201
187
                    ent->content[0] = 0;
4202
187
                    goto error;
4203
187
                }
4204
4205
                /*
4206
                 * Just output the reference
4207
                 */
4208
428k
                xmlSBufAddCString(&buf, "&", 1);
4209
428k
                xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name));
4210
428k
                xmlSBufAddCString(&buf, ";", 1);
4211
4212
428k
                inSpace = 0;
4213
428k
            }
4214
664k
  }
4215
145M
    }
4216
4217
316k
    if ((buf.mem == NULL) && (outFlags != NULL)) {
4218
203k
        ret = (xmlChar *) CUR_PTR - chunkSize;
4219
4220
203k
        if (attlen != NULL)
4221
203k
            *attlen = chunkSize;
4222
203k
        if ((normalize) && (inSpace) && (chunkSize > 0)) {
4223
271
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4224
271
            *attlen -= 1;
4225
271
        }
4226
4227
        /* Report potential error */
4228
203k
        xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4229
203k
    } else {
4230
112k
        if (chunkSize > 0)
4231
85.5k
            xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4232
4233
112k
        if ((normalize) && (inSpace) && (buf.size > 0)) {
4234
2.37k
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4235
2.37k
            buf.size--;
4236
2.37k
        }
4237
4238
112k
        ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long");
4239
112k
        attvalFlags |= XML_ATTVAL_ALLOC;
4240
4241
112k
        if (ret != NULL) {
4242
112k
            if (attlen != NULL)
4243
59.9k
                *attlen = buf.size;
4244
112k
        }
4245
112k
    }
4246
4247
316k
    if (outFlags != NULL)
4248
263k
        *outFlags = attvalFlags;
4249
4250
316k
    NEXTL(1);
4251
4252
316k
    return(ret);
4253
4254
15.7k
error:
4255
15.7k
    xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4256
15.7k
    return(NULL);
4257
332k
}
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
62.0k
xmlParseAttValue(xmlParserCtxt *ctxt) {
4297
62.0k
    if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL);
4298
62.0k
    return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0));
4299
62.0k
}
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
28.5k
xmlParseSystemLiteral(xmlParserCtxt *ctxt) {
4314
28.5k
    xmlChar *buf = NULL;
4315
28.5k
    int len = 0;
4316
28.5k
    int size = XML_PARSER_BUFFER_SIZE;
4317
28.5k
    int cur, l;
4318
28.5k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4319
11.8k
                    XML_MAX_TEXT_LENGTH :
4320
28.5k
                    XML_MAX_NAME_LENGTH;
4321
28.5k
    xmlChar stop;
4322
4323
28.5k
    if (RAW == '"') {
4324
14.0k
        NEXT;
4325
14.0k
  stop = '"';
4326
14.5k
    } else if (RAW == '\'') {
4327
11.4k
        NEXT;
4328
11.4k
  stop = '\'';
4329
11.4k
    } else {
4330
3.13k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4331
3.13k
  return(NULL);
4332
3.13k
    }
4333
4334
25.4k
    buf = xmlMalloc(size);
4335
25.4k
    if (buf == NULL) {
4336
26
        xmlErrMemory(ctxt);
4337
26
  return(NULL);
4338
26
    }
4339
25.4k
    cur = xmlCurrentCharRecover(ctxt, &l);
4340
9.71M
    while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */
4341
9.68M
  if (len + 5 >= size) {
4342
6.52k
      xmlChar *tmp;
4343
6.52k
            int newSize;
4344
4345
6.52k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4346
6.52k
            if (newSize < 0) {
4347
3
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral");
4348
3
                xmlFree(buf);
4349
3
                return(NULL);
4350
3
            }
4351
6.51k
      tmp = xmlRealloc(buf, newSize);
4352
6.51k
      if (tmp == NULL) {
4353
8
          xmlFree(buf);
4354
8
    xmlErrMemory(ctxt);
4355
8
    return(NULL);
4356
8
      }
4357
6.51k
      buf = tmp;
4358
6.51k
            size = newSize;
4359
6.51k
  }
4360
9.68M
  COPY_BUF(buf, len, cur);
4361
9.68M
  NEXTL(l);
4362
9.68M
  cur = xmlCurrentCharRecover(ctxt, &l);
4363
9.68M
    }
4364
25.4k
    buf[len] = 0;
4365
25.4k
    if (!IS_CHAR(cur)) {
4366
2.12k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4367
23.3k
    } else {
4368
23.3k
  NEXT;
4369
23.3k
    }
4370
25.4k
    return(buf);
4371
25.4k
}
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
9.30k
xmlParsePubidLiteral(xmlParserCtxt *ctxt) {
4386
9.30k
    xmlChar *buf = NULL;
4387
9.30k
    int len = 0;
4388
9.30k
    int size = XML_PARSER_BUFFER_SIZE;
4389
9.30k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4390
4.24k
                    XML_MAX_TEXT_LENGTH :
4391
9.30k
                    XML_MAX_NAME_LENGTH;
4392
9.30k
    xmlChar cur;
4393
9.30k
    xmlChar stop;
4394
4395
9.30k
    if (RAW == '"') {
4396
5.43k
        NEXT;
4397
5.43k
  stop = '"';
4398
5.43k
    } else if (RAW == '\'') {
4399
3.58k
        NEXT;
4400
3.58k
  stop = '\'';
4401
3.58k
    } else {
4402
297
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4403
297
  return(NULL);
4404
297
    }
4405
9.01k
    buf = xmlMalloc(size);
4406
9.01k
    if (buf == NULL) {
4407
17
  xmlErrMemory(ctxt);
4408
17
  return(NULL);
4409
17
    }
4410
8.99k
    cur = CUR;
4411
504k
    while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) &&
4412
495k
           (PARSER_STOPPED(ctxt) == 0)) { /* checked */
4413
495k
  if (len + 1 >= size) {
4414
1.15k
      xmlChar *tmp;
4415
1.15k
            int newSize;
4416
4417
1.15k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4418
1.15k
            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
1.15k
      tmp = xmlRealloc(buf, newSize);
4424
1.15k
      if (tmp == NULL) {
4425
5
    xmlErrMemory(ctxt);
4426
5
    xmlFree(buf);
4427
5
    return(NULL);
4428
5
      }
4429
1.14k
      buf = tmp;
4430
1.14k
            size = newSize;
4431
1.14k
  }
4432
495k
  buf[len++] = cur;
4433
495k
  NEXT;
4434
495k
  cur = CUR;
4435
495k
    }
4436
8.98k
    buf[len] = 0;
4437
8.98k
    if (cur != stop) {
4438
1.28k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4439
7.70k
    } else {
4440
7.70k
  NEXTL(1);
4441
7.70k
    }
4442
8.98k
    return(buf);
4443
8.99k
}
4444
4445
static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial);
4446
4447
/*
4448
 * used for the test in the inner loop of the char data testing
4449
 */
4450
static const unsigned char test_char_data[256] = {
4451
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4452
    0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */
4453
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4454
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4455
    0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */
4456
    0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F,
4457
    0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37,
4458
    0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */
4459
    0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47,
4460
    0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F,
4461
    0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57,
4462
    0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */
4463
    0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67,
4464
    0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F,
4465
    0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77,
4466
    0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F,
4467
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */
4468
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4469
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4470
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4471
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4472
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4473
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4474
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4475
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4476
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4477
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4478
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4479
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4480
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4481
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
4482
    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
4483
};
4484
4485
static void
4486
xmlCharacters(xmlParserCtxtPtr ctxt, const xmlChar *buf, int size,
4487
1.42M
              int isBlank) {
4488
1.42M
    int checkBlanks;
4489
4490
1.42M
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
4491
132k
        return;
4492
4493
1.29M
    checkBlanks = (!ctxt->keepBlanks) ||
4494
676k
                  (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
1.29M
    if ((checkBlanks) &&
4502
618k
        (areBlanks(ctxt, buf, size, isBlank))) {
4503
15.4k
        if ((ctxt->sax->ignorableWhitespace != NULL) &&
4504
15.4k
            (ctxt->keepBlanks))
4505
0
            ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
4506
1.27M
    } else {
4507
1.27M
        if (ctxt->sax->characters != NULL)
4508
1.27M
            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
1.27M
        if ((checkBlanks) && (*ctxt->space == -1))
4515
54.0k
            *ctxt->space = -2;
4516
1.27M
    }
4517
1.29M
}
4518
4519
/**
4520
 * Parse character data. Always makes progress if the first char isn't
4521
 * '<' or '&'.
4522
 *
4523
 * The right angle bracket (>) may be represented using the string "&gt;",
4524
 * and must, for compatibility, be escaped using "&gt;" or a character
4525
 * reference when it appears in the string "]]>" in content, when that
4526
 * string is not marking the end of a CDATA section.
4527
 *
4528
 *     [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*)
4529
 * @param ctxt  an XML parser context
4530
 * @param partial  buffer may contain partial UTF-8 sequences
4531
 */
4532
static void
4533
2.49M
xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) {
4534
2.49M
    const xmlChar *in;
4535
2.49M
    int line = ctxt->input->line;
4536
2.49M
    int col = ctxt->input->col;
4537
2.49M
    int ccol;
4538
2.49M
    int terminate = 0;
4539
4540
2.49M
    GROW;
4541
    /*
4542
     * Accelerated common case where input don't need to be
4543
     * modified before passing it to the handler.
4544
     */
4545
2.49M
    in = ctxt->input->cur;
4546
2.50M
    do {
4547
2.57M
get_more_space:
4548
2.70M
        while (*in == 0x20) { in++; ctxt->input->col++; }
4549
2.57M
        if (*in == 0xA) {
4550
8.27M
            do {
4551
8.27M
                ctxt->input->line++; ctxt->input->col = 1;
4552
8.27M
                in++;
4553
8.27M
            } while (*in == 0xA);
4554
69.5k
            goto get_more_space;
4555
69.5k
        }
4556
2.50M
        if (*in == '<') {
4557
112k
            while (in > ctxt->input->cur) {
4558
56.4k
                const xmlChar *tmp = ctxt->input->cur;
4559
56.4k
                size_t nbchar = in - tmp;
4560
4561
56.4k
                if (nbchar > XML_MAX_ITEMS)
4562
0
                    nbchar = XML_MAX_ITEMS;
4563
56.4k
                ctxt->input->cur += nbchar;
4564
4565
56.4k
                xmlCharacters(ctxt, tmp, nbchar, 1);
4566
56.4k
            }
4567
56.4k
            return;
4568
56.4k
        }
4569
4570
3.63M
get_more:
4571
3.63M
        ccol = ctxt->input->col;
4572
9.09M
        while (test_char_data[*in]) {
4573
5.46M
            in++;
4574
5.46M
            ccol++;
4575
5.46M
        }
4576
3.63M
        ctxt->input->col = ccol;
4577
3.63M
        if (*in == 0xA) {
4578
4.65M
            do {
4579
4.65M
                ctxt->input->line++; ctxt->input->col = 1;
4580
4.65M
                in++;
4581
4.65M
            } while (*in == 0xA);
4582
82.8k
            goto get_more;
4583
82.8k
        }
4584
3.54M
        if (*in == ']') {
4585
1.09M
            size_t avail = ctxt->input->end - in;
4586
4587
1.09M
            if (partial && avail < 2) {
4588
296
                terminate = 1;
4589
296
                goto invoke_callback;
4590
296
            }
4591
1.09M
            if (in[1] == ']') {
4592
945k
                if (partial && avail < 3) {
4593
1.40k
                    terminate = 1;
4594
1.40k
                    goto invoke_callback;
4595
1.40k
                }
4596
944k
                if (in[2] == '>')
4597
4.39k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4598
944k
            }
4599
4600
1.09M
            in++;
4601
1.09M
            ctxt->input->col++;
4602
1.09M
            goto get_more;
4603
1.09M
        }
4604
4605
2.45M
invoke_callback:
4606
2.85M
        while (in > ctxt->input->cur) {
4607
400k
            const xmlChar *tmp = ctxt->input->cur;
4608
400k
            size_t nbchar = in - tmp;
4609
4610
400k
            if (nbchar > XML_MAX_ITEMS)
4611
0
                nbchar = XML_MAX_ITEMS;
4612
400k
            ctxt->input->cur += nbchar;
4613
4614
400k
            xmlCharacters(ctxt, tmp, nbchar, 0);
4615
4616
400k
            line = ctxt->input->line;
4617
400k
            col = ctxt->input->col;
4618
400k
        }
4619
2.45M
        ctxt->input->cur = in;
4620
2.45M
        if (*in == 0xD) {
4621
22.1k
            in++;
4622
22.1k
            if (*in == 0xA) {
4623
16.2k
                ctxt->input->cur = in;
4624
16.2k
                in++;
4625
16.2k
                ctxt->input->line++; ctxt->input->col = 1;
4626
16.2k
                continue; /* while */
4627
16.2k
            }
4628
5.86k
            in--;
4629
5.86k
        }
4630
2.43M
        if (*in == '<') {
4631
157k
            return;
4632
157k
        }
4633
2.27M
        if (*in == '&') {
4634
77.3k
            return;
4635
77.3k
        }
4636
2.20M
        if (terminate) {
4637
1.70k
            return;
4638
1.70k
        }
4639
2.19M
        SHRINK;
4640
2.19M
        GROW;
4641
2.19M
        in = ctxt->input->cur;
4642
2.21M
    } while (((*in >= 0x20) && (*in <= 0x7F)) ||
4643
2.20M
             (*in == 0x09) || (*in == 0x0a));
4644
2.19M
    ctxt->input->line = line;
4645
2.19M
    ctxt->input->col = col;
4646
2.19M
    xmlParseCharDataComplex(ctxt, partial);
4647
2.19M
}
4648
4649
/**
4650
 * Always makes progress if the first char isn't '<' or '&'.
4651
 *
4652
 * parse a CharData section.this is the fallback function
4653
 * of #xmlParseCharData when the parsing requires handling
4654
 * of non-ASCII characters.
4655
 *
4656
 * @param ctxt  an XML parser context
4657
 * @param partial  whether the input can end with truncated UTF-8
4658
 */
4659
static void
4660
2.19M
xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
4661
2.19M
    xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5];
4662
2.19M
    int nbchar = 0;
4663
2.19M
    int cur, l;
4664
4665
2.19M
    cur = xmlCurrentCharRecover(ctxt, &l);
4666
84.2M
    while ((cur != '<') && /* checked */
4667
84.1M
           (cur != '&') &&
4668
84.1M
     (IS_CHAR(cur))) {
4669
82.0M
        if (cur == ']') {
4670
849k
            size_t avail = ctxt->input->end - ctxt->input->cur;
4671
4672
849k
            if (partial && avail < 2)
4673
113
                break;
4674
849k
            if (NXT(1) == ']') {
4675
760k
                if (partial && avail < 3)
4676
571
                    break;
4677
759k
                if (NXT(2) == '>')
4678
1.41k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4679
759k
            }
4680
849k
        }
4681
4682
82.0M
  COPY_BUF(buf, nbchar, cur);
4683
  /* move current position before possible calling of ctxt->sax->characters */
4684
82.0M
  NEXTL(l);
4685
82.0M
  if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) {
4686
732k
      buf[nbchar] = 0;
4687
4688
732k
            xmlCharacters(ctxt, buf, nbchar, 0);
4689
732k
      nbchar = 0;
4690
732k
            SHRINK;
4691
732k
  }
4692
82.0M
  cur = xmlCurrentCharRecover(ctxt, &l);
4693
82.0M
    }
4694
2.19M
    if (nbchar != 0) {
4695
237k
        buf[nbchar] = 0;
4696
4697
237k
        xmlCharacters(ctxt, buf, nbchar, 0);
4698
237k
    }
4699
    /*
4700
     * cur == 0 can mean
4701
     *
4702
     * - End of buffer.
4703
     * - An actual 0 character.
4704
     * - An incomplete UTF-8 sequence. This is allowed if partial is set.
4705
     */
4706
2.19M
    if (ctxt->input->cur < ctxt->input->end) {
4707
2.18M
        if ((cur == 0) && (CUR != 0)) {
4708
1.50k
            if (partial == 0) {
4709
1.27k
                xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4710
1.27k
                        "Incomplete UTF-8 sequence starting with %02X\n", CUR);
4711
1.27k
                NEXTL(1);
4712
1.27k
            }
4713
2.18M
        } else if ((cur != '<') && (cur != '&') && (cur != ']')) {
4714
            /* Generate the error and skip the offending character */
4715
2.07M
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4716
2.07M
                              "PCDATA invalid Char value %d\n", cur);
4717
2.07M
            NEXTL(l);
4718
2.07M
        }
4719
2.18M
    }
4720
2.19M
}
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
55.7k
xmlParseExternalID(xmlParserCtxt *ctxt, xmlChar **publicId, int strict) {
4756
55.7k
    xmlChar *URI = NULL;
4757
4758
55.7k
    *publicId = NULL;
4759
55.7k
    if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) {
4760
21.5k
        SKIP(6);
4761
21.5k
  if (SKIP_BLANKS == 0) {
4762
980
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4763
980
                     "Space required after 'SYSTEM'\n");
4764
980
  }
4765
21.5k
  URI = xmlParseSystemLiteral(ctxt);
4766
21.5k
  if (URI == NULL) {
4767
1.07k
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4768
1.07k
        }
4769
34.2k
    } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) {
4770
9.30k
        SKIP(6);
4771
9.30k
  if (SKIP_BLANKS == 0) {
4772
1.13k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4773
1.13k
        "Space required after 'PUBLIC'\n");
4774
1.13k
  }
4775
9.30k
  *publicId = xmlParsePubidLiteral(ctxt);
4776
9.30k
  if (*publicId == NULL) {
4777
322
      xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL);
4778
322
  }
4779
9.30k
  if (strict) {
4780
      /*
4781
       * We don't handle [83] so "S SystemLiteral" is required.
4782
       */
4783
6.66k
      if (SKIP_BLANKS == 0) {
4784
2.51k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4785
2.51k
      "Space required after the Public Identifier\n");
4786
2.51k
      }
4787
6.66k
  } else {
4788
      /*
4789
       * We handle [83] so we return immediately, if
4790
       * "S SystemLiteral" is not detected. We skip blanks if no
4791
             * system literal was found, but this is harmless since we must
4792
             * be at the end of a NotationDecl.
4793
       */
4794
2.64k
      if (SKIP_BLANKS == 0) return(NULL);
4795
631
      if ((CUR != '\'') && (CUR != '"')) return(NULL);
4796
631
  }
4797
7.09k
  URI = xmlParseSystemLiteral(ctxt);
4798
7.09k
  if (URI == NULL) {
4799
2.09k
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4800
2.09k
        }
4801
7.09k
    }
4802
53.5k
    return(URI);
4803
55.7k
}
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
15.4k
                       size_t len, size_t size) {
4820
15.4k
    int q, ql;
4821
15.4k
    int r, rl;
4822
15.4k
    int cur, l;
4823
15.4k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4824
6.32k
                    XML_MAX_HUGE_LENGTH :
4825
15.4k
                    XML_MAX_TEXT_LENGTH;
4826
4827
15.4k
    if (buf == NULL) {
4828
4.63k
        len = 0;
4829
4.63k
  size = XML_PARSER_BUFFER_SIZE;
4830
4.63k
  buf = xmlMalloc(size);
4831
4.63k
  if (buf == NULL) {
4832
28
      xmlErrMemory(ctxt);
4833
28
      return;
4834
28
  }
4835
4.63k
    }
4836
15.3k
    q = xmlCurrentCharRecover(ctxt, &ql);
4837
15.3k
    if (q == 0)
4838
3.30k
        goto not_terminated;
4839
12.0k
    if (!IS_CHAR(q)) {
4840
723
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4841
723
                          "xmlParseComment: invalid xmlChar value %d\n",
4842
723
                    q);
4843
723
  xmlFree (buf);
4844
723
  return;
4845
723
    }
4846
11.3k
    NEXTL(ql);
4847
11.3k
    r = xmlCurrentCharRecover(ctxt, &rl);
4848
11.3k
    if (r == 0)
4849
359
        goto not_terminated;
4850
11.0k
    if (!IS_CHAR(r)) {
4851
703
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4852
703
                          "xmlParseComment: invalid xmlChar value %d\n",
4853
703
                    r);
4854
703
  xmlFree (buf);
4855
703
  return;
4856
703
    }
4857
10.3k
    NEXTL(rl);
4858
10.3k
    cur = xmlCurrentCharRecover(ctxt, &l);
4859
10.3k
    if (cur == 0)
4860
471
        goto not_terminated;
4861
17.8M
    while (IS_CHAR(cur) && /* checked */
4862
17.8M
           ((cur != '>') ||
4863
17.8M
      (r != '-') || (q != '-'))) {
4864
17.8M
  if ((r == '-') && (q == '-')) {
4865
2.15M
      xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL);
4866
2.15M
  }
4867
17.8M
  if (len + 5 >= size) {
4868
4.42k
      xmlChar *tmp;
4869
4.42k
            int newSize;
4870
4871
4.42k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4872
4.42k
            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
4.42k
      tmp = xmlRealloc(buf, newSize);
4879
4.42k
      if (tmp == NULL) {
4880
9
    xmlErrMemory(ctxt);
4881
9
    xmlFree(buf);
4882
9
    return;
4883
9
      }
4884
4.41k
      buf = tmp;
4885
4.41k
            size = newSize;
4886
4.41k
  }
4887
17.8M
  COPY_BUF(buf, len, q);
4888
4889
17.8M
  q = r;
4890
17.8M
  ql = rl;
4891
17.8M
  r = cur;
4892
17.8M
  rl = l;
4893
4894
17.8M
  NEXTL(l);
4895
17.8M
  cur = xmlCurrentCharRecover(ctxt, &l);
4896
4897
17.8M
    }
4898
9.82k
    buf[len] = 0;
4899
9.82k
    if (cur == 0) {
4900
2.50k
  xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4901
2.50k
                       "Comment not terminated \n<!--%.50s\n", buf);
4902
7.32k
    } else if (!IS_CHAR(cur)) {
4903
645
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4904
645
                          "xmlParseComment: invalid xmlChar value %d\n",
4905
645
                    cur);
4906
6.67k
    } else {
4907
6.67k
        NEXT;
4908
6.67k
  if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
4909
6.67k
      (!ctxt->disableSAX))
4910
5.46k
      ctxt->sax->comment(ctxt->userData, buf);
4911
6.67k
    }
4912
9.82k
    xmlFree(buf);
4913
9.82k
    return;
4914
4.13k
not_terminated:
4915
4.13k
    xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4916
4.13k
       "Comment not terminated\n", NULL);
4917
4.13k
    xmlFree(buf);
4918
4.13k
}
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
435k
xmlParseComment(xmlParserCtxt *ctxt) {
4933
435k
    xmlChar *buf = NULL;
4934
435k
    size_t size = XML_PARSER_BUFFER_SIZE;
4935
435k
    size_t len = 0;
4936
435k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4937
72.0k
                       XML_MAX_HUGE_LENGTH :
4938
435k
                       XML_MAX_TEXT_LENGTH;
4939
435k
    const xmlChar *in;
4940
435k
    size_t nbchar = 0;
4941
435k
    int ccol;
4942
4943
    /*
4944
     * Check that there is a comment right here.
4945
     */
4946
435k
    if ((RAW != '<') || (NXT(1) != '!'))
4947
0
        return;
4948
435k
    SKIP(2);
4949
435k
    if ((RAW != '-') || (NXT(1) != '-'))
4950
38
        return;
4951
435k
    SKIP(2);
4952
435k
    GROW;
4953
4954
    /*
4955
     * Accelerated common case where input don't need to be
4956
     * modified before passing it to the handler.
4957
     */
4958
435k
    in = ctxt->input->cur;
4959
436k
    do {
4960
436k
  if (*in == 0xA) {
4961
7.89k
      do {
4962
7.89k
    ctxt->input->line++; ctxt->input->col = 1;
4963
7.89k
    in++;
4964
7.89k
      } while (*in == 0xA);
4965
3.13k
  }
4966
4.03M
get_more:
4967
4.03M
        ccol = ctxt->input->col;
4968
12.5M
  while (((*in > '-') && (*in <= 0x7F)) ||
4969
5.81M
         ((*in >= 0x20) && (*in < '-')) ||
4970
8.47M
         (*in == 0x09)) {
4971
8.47M
        in++;
4972
8.47M
        ccol++;
4973
8.47M
  }
4974
4.03M
  ctxt->input->col = ccol;
4975
4.03M
  if (*in == 0xA) {
4976
166k
      do {
4977
166k
    ctxt->input->line++; ctxt->input->col = 1;
4978
166k
    in++;
4979
166k
      } while (*in == 0xA);
4980
20.7k
      goto get_more;
4981
20.7k
  }
4982
4.01M
  nbchar = in - ctxt->input->cur;
4983
  /*
4984
   * save current set of data
4985
   */
4986
4.01M
  if (nbchar > 0) {
4987
3.60M
            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
3.60M
            if (buf == NULL) {
4994
34.4k
                if ((*in == '-') && (in[1] == '-'))
4995
16.1k
                    size = nbchar + 1;
4996
18.2k
                else
4997
18.2k
                    size = XML_PARSER_BUFFER_SIZE + nbchar;
4998
34.4k
                buf = xmlMalloc(size);
4999
34.4k
                if (buf == NULL) {
5000
23
                    xmlErrMemory(ctxt);
5001
23
                    return;
5002
23
                }
5003
34.3k
                len = 0;
5004
3.56M
            } else if (len + nbchar + 1 >= size) {
5005
11.3k
                xmlChar *new_buf;
5006
11.3k
                size += len + nbchar + XML_PARSER_BUFFER_SIZE;
5007
11.3k
                new_buf = xmlRealloc(buf, size);
5008
11.3k
                if (new_buf == NULL) {
5009
3
                    xmlErrMemory(ctxt);
5010
3
                    xmlFree(buf);
5011
3
                    return;
5012
3
                }
5013
11.3k
                buf = new_buf;
5014
11.3k
            }
5015
3.60M
            memcpy(&buf[len], ctxt->input->cur, nbchar);
5016
3.60M
            len += nbchar;
5017
3.60M
            buf[len] = 0;
5018
3.60M
  }
5019
4.01M
  ctxt->input->cur = in;
5020
4.01M
  if (*in == 0xA) {
5021
0
      in++;
5022
0
      ctxt->input->line++; ctxt->input->col = 1;
5023
0
  }
5024
4.01M
  if (*in == 0xD) {
5025
5.34k
      in++;
5026
5.34k
      if (*in == 0xA) {
5027
4.50k
    ctxt->input->cur = in;
5028
4.50k
    in++;
5029
4.50k
    ctxt->input->line++; ctxt->input->col = 1;
5030
4.50k
    goto get_more;
5031
4.50k
      }
5032
832
      in--;
5033
832
  }
5034
4.01M
  SHRINK;
5035
4.01M
  GROW;
5036
4.01M
  in = ctxt->input->cur;
5037
4.01M
  if (*in == '-') {
5038
3.99M
      if (in[1] == '-') {
5039
2.50M
          if (in[2] == '>') {
5040
420k
        SKIP(3);
5041
420k
        if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
5042
420k
            (!ctxt->disableSAX)) {
5043
408k
      if (buf != NULL)
5044
19.1k
          ctxt->sax->comment(ctxt->userData, buf);
5045
389k
      else
5046
389k
          ctxt->sax->comment(ctxt->userData, BAD_CAST "");
5047
408k
        }
5048
420k
        if (buf != NULL)
5049
23.5k
            xmlFree(buf);
5050
420k
        return;
5051
420k
    }
5052
2.08M
    if (buf != NULL) {
5053
2.07M
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5054
2.07M
                          "Double hyphen within comment: "
5055
2.07M
                                      "<!--%.50s\n",
5056
2.07M
              buf);
5057
2.07M
    } else
5058
3.80k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5059
3.80k
                          "Double hyphen within comment\n", NULL);
5060
2.08M
    in++;
5061
2.08M
    ctxt->input->col++;
5062
2.08M
      }
5063
3.57M
      in++;
5064
3.57M
      ctxt->input->col++;
5065
3.57M
      goto get_more;
5066
3.99M
  }
5067
4.01M
    } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a));
5068
15.4k
    xmlParseCommentComplex(ctxt, buf, len, size);
5069
15.4k
}
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
58.0k
xmlParsePITarget(xmlParserCtxt *ctxt) {
5085
58.0k
    const xmlChar *name;
5086
5087
58.0k
    name = xmlParseName(ctxt);
5088
58.0k
    if ((name != NULL) &&
5089
52.0k
        ((name[0] == 'x') || (name[0] == 'X')) &&
5090
30.2k
        ((name[1] == 'm') || (name[1] == 'M')) &&
5091
10.5k
        ((name[2] == 'l') || (name[2] == 'L'))) {
5092
7.41k
  int i;
5093
7.41k
  if ((name[0] == 'x') && (name[1] == 'm') &&
5094
7.01k
      (name[2] == 'l') && (name[3] == 0)) {
5095
5.83k
      xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5096
5.83k
     "XML declaration allowed only at the start of the document\n");
5097
5.83k
      return(name);
5098
5.83k
  } else if (name[3] == 0) {
5099
542
      xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL);
5100
542
      return(name);
5101
542
  }
5102
2.59k
  for (i = 0;;i++) {
5103
2.59k
      if (xmlW3CPIs[i] == NULL) break;
5104
1.84k
      if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i]))
5105
282
          return(name);
5106
1.84k
  }
5107
750
  xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5108
750
          "xmlParsePITarget: invalid name prefix 'xml'\n",
5109
750
          NULL, NULL);
5110
750
    }
5111
51.3k
    if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) {
5112
394
  xmlNsErr(ctxt, XML_NS_ERR_COLON,
5113
394
     "colons are forbidden from PI names '%s'\n", name, NULL, NULL);
5114
394
    }
5115
51.3k
    return(name);
5116
58.0k
}
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
58.0k
xmlParsePI(xmlParserCtxt *ctxt) {
5199
58.0k
    xmlChar *buf = NULL;
5200
58.0k
    size_t len = 0;
5201
58.0k
    size_t size = XML_PARSER_BUFFER_SIZE;
5202
58.0k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5203
21.3k
                       XML_MAX_HUGE_LENGTH :
5204
58.0k
                       XML_MAX_TEXT_LENGTH;
5205
58.0k
    int cur, l;
5206
58.0k
    const xmlChar *target;
5207
5208
58.0k
    if ((RAW == '<') && (NXT(1) == '?')) {
5209
  /*
5210
   * this is a Processing Instruction.
5211
   */
5212
58.0k
  SKIP(2);
5213
5214
  /*
5215
   * Parse the target name and check for special support like
5216
   * namespace.
5217
   */
5218
58.0k
        target = xmlParsePITarget(ctxt);
5219
58.0k
  if (target != NULL) {
5220
52.0k
      if ((RAW == '?') && (NXT(1) == '>')) {
5221
11.5k
    SKIP(2);
5222
5223
    /*
5224
     * SAX: PI detected.
5225
     */
5226
11.5k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5227
11.2k
        (ctxt->sax->processingInstruction != NULL))
5228
11.2k
        ctxt->sax->processingInstruction(ctxt->userData,
5229
11.2k
                                         target, NULL);
5230
11.5k
    return;
5231
11.5k
      }
5232
40.4k
      buf = xmlMalloc(size);
5233
40.4k
      if (buf == NULL) {
5234
52
    xmlErrMemory(ctxt);
5235
52
    return;
5236
52
      }
5237
40.4k
      if (SKIP_BLANKS == 0) {
5238
23.0k
    xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED,
5239
23.0k
        "ParsePI: PI %s space expected\n", target);
5240
23.0k
      }
5241
40.4k
      cur = xmlCurrentCharRecover(ctxt, &l);
5242
10.9M
      while (IS_CHAR(cur) && /* checked */
5243
10.9M
       ((cur != '?') || (NXT(1) != '>'))) {
5244
10.8M
    if (len + 5 >= size) {
5245
11.2k
        xmlChar *tmp;
5246
11.2k
                    int newSize;
5247
5248
11.2k
                    newSize = xmlGrowCapacity(size, 1, 1, maxLength);
5249
11.2k
                    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
11.2k
        tmp = xmlRealloc(buf, newSize);
5256
11.2k
        if (tmp == NULL) {
5257
11
      xmlErrMemory(ctxt);
5258
11
      xmlFree(buf);
5259
11
      return;
5260
11
        }
5261
11.2k
        buf = tmp;
5262
11.2k
                    size = newSize;
5263
11.2k
    }
5264
10.8M
    COPY_BUF(buf, len, cur);
5265
10.8M
    NEXTL(l);
5266
10.8M
    cur = xmlCurrentCharRecover(ctxt, &l);
5267
10.8M
      }
5268
40.3k
      buf[len] = 0;
5269
40.3k
      if (cur != '?') {
5270
12.9k
    xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5271
12.9k
          "ParsePI: PI %s never end ...\n", target);
5272
27.4k
      } else {
5273
27.4k
    SKIP(2);
5274
5275
27.4k
#ifdef LIBXML_CATALOG_ENABLED
5276
27.4k
    if ((ctxt->inSubset == 0) &&
5277
18.7k
        (xmlStrEqual(target, XML_CATALOG_PI))) {
5278
3.27k
        xmlCatalogAllow allow = xmlCatalogGetDefaults();
5279
5280
3.27k
        if ((ctxt->options & XML_PARSE_CATALOG_PI) &&
5281
559
                        ((allow == XML_CATA_ALLOW_DOCUMENT) ||
5282
559
       (allow == XML_CATA_ALLOW_ALL)))
5283
0
      xmlParseCatalogPI(ctxt, buf);
5284
3.27k
    }
5285
27.4k
#endif
5286
5287
    /*
5288
     * SAX: PI detected.
5289
     */
5290
27.4k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5291
22.7k
        (ctxt->sax->processingInstruction != NULL))
5292
22.7k
        ctxt->sax->processingInstruction(ctxt->userData,
5293
22.7k
                                         target, buf);
5294
27.4k
      }
5295
40.3k
      xmlFree(buf);
5296
40.3k
  } else {
5297
6.02k
      xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL);
5298
6.02k
  }
5299
58.0k
    }
5300
58.0k
}
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
9.72k
xmlParseNotationDecl(xmlParserCtxt *ctxt) {
5323
9.72k
    const xmlChar *name;
5324
9.72k
    xmlChar *Pubid;
5325
9.72k
    xmlChar *Systemid;
5326
5327
9.72k
    if ((CUR != '<') || (NXT(1) != '!'))
5328
0
        return;
5329
9.72k
    SKIP(2);
5330
5331
9.72k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5332
9.68k
#ifdef LIBXML_VALID_ENABLED
5333
9.68k
  int oldInputNr = ctxt->inputNr;
5334
9.68k
#endif
5335
5336
9.68k
  SKIP(8);
5337
9.68k
  if (SKIP_BLANKS_PE == 0) {
5338
274
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5339
274
         "Space required after '<!NOTATION'\n");
5340
274
      return;
5341
274
  }
5342
5343
9.40k
        name = xmlParseName(ctxt);
5344
9.40k
  if (name == NULL) {
5345
500
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5346
500
      return;
5347
500
  }
5348
8.90k
  if (xmlStrchr(name, ':') != NULL) {
5349
231
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5350
231
         "colons are forbidden from notation names '%s'\n",
5351
231
         name, NULL, NULL);
5352
231
  }
5353
8.90k
  if (SKIP_BLANKS_PE == 0) {
5354
226
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5355
226
         "Space required after the NOTATION name'\n");
5356
226
      return;
5357
226
  }
5358
5359
  /*
5360
   * Parse the IDs.
5361
   */
5362
8.68k
  Systemid = xmlParseExternalID(ctxt, &Pubid, 0);
5363
8.68k
  SKIP_BLANKS_PE;
5364
5365
8.68k
  if (RAW == '>') {
5366
7.13k
#ifdef LIBXML_VALID_ENABLED
5367
7.13k
      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
7.13k
#endif
5374
7.13k
      NEXT;
5375
7.13k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5376
6.75k
    (ctxt->sax->notationDecl != NULL))
5377
6.75k
    ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid);
5378
7.13k
  } else {
5379
1.55k
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5380
1.55k
  }
5381
8.68k
  if (Systemid != NULL) xmlFree(Systemid);
5382
8.68k
  if (Pubid != NULL) xmlFree(Pubid);
5383
8.68k
    }
5384
9.72k
}
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
64.7k
xmlParseEntityDecl(xmlParserCtxt *ctxt) {
5411
64.7k
    const xmlChar *name = NULL;
5412
64.7k
    xmlChar *value = NULL;
5413
64.7k
    xmlChar *URI = NULL, *literal = NULL;
5414
64.7k
    const xmlChar *ndata = NULL;
5415
64.7k
    int isParameter = 0;
5416
64.7k
    xmlChar *orig = NULL;
5417
5418
64.7k
    if ((CUR != '<') || (NXT(1) != '!'))
5419
0
        return;
5420
64.7k
    SKIP(2);
5421
5422
    /* GROW; done in the caller */
5423
64.7k
    if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5424
64.7k
#ifdef LIBXML_VALID_ENABLED
5425
64.7k
  int oldInputNr = ctxt->inputNr;
5426
64.7k
#endif
5427
5428
64.7k
  SKIP(6);
5429
64.7k
  if (SKIP_BLANKS_PE == 0) {
5430
5.01k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5431
5.01k
         "Space required after '<!ENTITY'\n");
5432
5.01k
  }
5433
5434
64.7k
  if (RAW == '%') {
5435
23.6k
      NEXT;
5436
23.6k
      if (SKIP_BLANKS_PE == 0) {
5437
674
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5438
674
             "Space required after '%%'\n");
5439
674
      }
5440
23.6k
      isParameter = 1;
5441
23.6k
  }
5442
5443
64.7k
        name = xmlParseName(ctxt);
5444
64.7k
  if (name == NULL) {
5445
1.56k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5446
1.56k
                     "xmlParseEntityDecl: no name\n");
5447
1.56k
            return;
5448
1.56k
  }
5449
63.1k
  if (xmlStrchr(name, ':') != NULL) {
5450
500
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5451
500
         "colons are forbidden from entities names '%s'\n",
5452
500
         name, NULL, NULL);
5453
500
  }
5454
63.1k
  if (SKIP_BLANKS_PE == 0) {
5455
4.21k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5456
4.21k
         "Space required after the entity name\n");
5457
4.21k
  }
5458
5459
  /*
5460
   * handle the various case of definitions...
5461
   */
5462
63.1k
  if (isParameter) {
5463
23.2k
      if ((RAW == '"') || (RAW == '\'')) {
5464
14.6k
          value = xmlParseEntityValue(ctxt, &orig);
5465
14.6k
    if (value) {
5466
14.2k
        if ((ctxt->sax != NULL) &&
5467
14.2k
      (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5468
12.7k
      ctxt->sax->entityDecl(ctxt->userData, name,
5469
12.7k
                        XML_INTERNAL_PARAMETER_ENTITY,
5470
12.7k
            NULL, NULL, value);
5471
14.2k
    }
5472
14.6k
      } else {
5473
8.54k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5474
8.54k
    if ((URI == NULL) && (literal == NULL)) {
5475
572
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5476
572
    }
5477
8.54k
    if (URI) {
5478
7.72k
                    if (xmlStrchr(URI, '#')) {
5479
259
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5480
7.46k
                    } else {
5481
7.46k
                        if ((ctxt->sax != NULL) &&
5482
7.46k
                            (!ctxt->disableSAX) &&
5483
6.91k
                            (ctxt->sax->entityDecl != NULL))
5484
6.91k
                            ctxt->sax->entityDecl(ctxt->userData, name,
5485
6.91k
                                        XML_EXTERNAL_PARAMETER_ENTITY,
5486
6.91k
                                        literal, URI, NULL);
5487
7.46k
                    }
5488
7.72k
    }
5489
8.54k
      }
5490
39.9k
  } else {
5491
39.9k
      if ((RAW == '"') || (RAW == '\'')) {
5492
28.9k
          value = xmlParseEntityValue(ctxt, &orig);
5493
28.9k
    if ((ctxt->sax != NULL) &&
5494
28.9k
        (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5495
26.1k
        ctxt->sax->entityDecl(ctxt->userData, name,
5496
26.1k
        XML_INTERNAL_GENERAL_ENTITY,
5497
26.1k
        NULL, NULL, value);
5498
    /*
5499
     * For expat compatibility in SAX mode.
5500
     */
5501
28.9k
    if ((ctxt->myDoc == NULL) ||
5502
28.7k
        (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) {
5503
1.40k
        if (ctxt->myDoc == NULL) {
5504
241
      ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5505
241
      if (ctxt->myDoc == NULL) {
5506
1
          xmlErrMemory(ctxt);
5507
1
          goto done;
5508
1
      }
5509
240
      ctxt->myDoc->properties = XML_DOC_INTERNAL;
5510
240
        }
5511
1.40k
        if (ctxt->myDoc->intSubset == NULL) {
5512
240
      ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5513
240
              BAD_CAST "fake", NULL, NULL);
5514
240
                        if (ctxt->myDoc->intSubset == NULL) {
5515
1
                            xmlErrMemory(ctxt);
5516
1
                            goto done;
5517
1
                        }
5518
240
                    }
5519
5520
1.40k
        xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY,
5521
1.40k
                    NULL, NULL, value);
5522
1.40k
    }
5523
28.9k
      } else {
5524
10.9k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5525
10.9k
    if ((URI == NULL) && (literal == NULL)) {
5526
2.40k
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5527
2.40k
    }
5528
10.9k
    if (URI) {
5529
7.97k
                    if (xmlStrchr(URI, '#')) {
5530
393
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5531
393
                    }
5532
7.97k
    }
5533
10.9k
    if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) {
5534
2.48k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5535
2.48k
           "Space required before 'NDATA'\n");
5536
2.48k
    }
5537
10.9k
    if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) {
5538
1.04k
        SKIP(5);
5539
1.04k
        if (SKIP_BLANKS_PE == 0) {
5540
401
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5541
401
               "Space required after 'NDATA'\n");
5542
401
        }
5543
1.04k
        ndata = xmlParseName(ctxt);
5544
1.04k
        if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5545
798
            (ctxt->sax->unparsedEntityDecl != NULL))
5546
798
      ctxt->sax->unparsedEntityDecl(ctxt->userData, name,
5547
798
            literal, URI, ndata);
5548
9.89k
    } else {
5549
9.89k
        if ((ctxt->sax != NULL) &&
5550
9.89k
            (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5551
8.94k
      ctxt->sax->entityDecl(ctxt->userData, name,
5552
8.94k
            XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5553
8.94k
            literal, URI, NULL);
5554
        /*
5555
         * For expat compatibility in SAX mode.
5556
         * assuming the entity replacement was asked for
5557
         */
5558
9.89k
        if ((ctxt->replaceEntities != 0) &&
5559
6.52k
      ((ctxt->myDoc == NULL) ||
5560
6.51k
      (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) {
5561
214
      if (ctxt->myDoc == NULL) {
5562
15
          ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5563
15
          if (ctxt->myDoc == NULL) {
5564
1
              xmlErrMemory(ctxt);
5565
1
        goto done;
5566
1
          }
5567
14
          ctxt->myDoc->properties = XML_DOC_INTERNAL;
5568
14
      }
5569
5570
213
      if (ctxt->myDoc->intSubset == NULL) {
5571
14
          ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5572
14
            BAD_CAST "fake", NULL, NULL);
5573
14
                            if (ctxt->myDoc->intSubset == NULL) {
5574
1
                                xmlErrMemory(ctxt);
5575
1
                                goto done;
5576
1
                            }
5577
14
                        }
5578
212
      xmlSAX2EntityDecl(ctxt, name,
5579
212
                  XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5580
212
                  literal, URI, NULL);
5581
212
        }
5582
9.89k
    }
5583
10.9k
      }
5584
39.9k
  }
5585
63.1k
  SKIP_BLANKS_PE;
5586
63.1k
  if (RAW != '>') {
5587
7.90k
      xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED,
5588
7.90k
              "xmlParseEntityDecl: entity %s not terminated\n", name);
5589
55.2k
  } else {
5590
55.2k
#ifdef LIBXML_VALID_ENABLED
5591
55.2k
      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
55.2k
#endif
5598
55.2k
      NEXT;
5599
55.2k
  }
5600
63.1k
  if (orig != NULL) {
5601
      /*
5602
       * Ugly mechanism to save the raw entity value.
5603
       */
5604
41.9k
      xmlEntityPtr cur = NULL;
5605
5606
41.9k
      if (isParameter) {
5607
14.2k
          if ((ctxt->sax != NULL) &&
5608
14.2k
        (ctxt->sax->getParameterEntity != NULL))
5609
14.2k
        cur = ctxt->sax->getParameterEntity(ctxt->userData, name);
5610
27.7k
      } else {
5611
27.7k
          if ((ctxt->sax != NULL) &&
5612
27.7k
        (ctxt->sax->getEntity != NULL))
5613
27.7k
        cur = ctxt->sax->getEntity(ctxt->userData, name);
5614
27.7k
    if ((cur == NULL) && (ctxt->userData==ctxt)) {
5615
1.10k
        cur = xmlSAX2GetEntity(ctxt, name);
5616
1.10k
    }
5617
27.7k
      }
5618
41.9k
            if ((cur != NULL) && (cur->orig == NULL)) {
5619
23.1k
    cur->orig = orig;
5620
23.1k
                orig = NULL;
5621
23.1k
      }
5622
41.9k
  }
5623
5624
63.1k
done:
5625
63.1k
  if (value != NULL) xmlFree(value);
5626
63.1k
  if (URI != NULL) xmlFree(URI);
5627
63.1k
  if (literal != NULL) xmlFree(literal);
5628
63.1k
        if (orig != NULL) xmlFree(orig);
5629
63.1k
    }
5630
64.7k
}
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
96.9k
xmlParseDefaultDecl(xmlParserCtxt *ctxt, xmlChar **value) {
5663
96.9k
    int val;
5664
96.9k
    xmlChar *ret;
5665
5666
96.9k
    *value = NULL;
5667
96.9k
    if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) {
5668
6.22k
  SKIP(9);
5669
6.22k
  return(XML_ATTRIBUTE_REQUIRED);
5670
6.22k
    }
5671
90.7k
    if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) {
5672
28.6k
  SKIP(8);
5673
28.6k
  return(XML_ATTRIBUTE_IMPLIED);
5674
28.6k
    }
5675
62.0k
    val = XML_ATTRIBUTE_NONE;
5676
62.0k
    if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) {
5677
6.39k
  SKIP(6);
5678
6.39k
  val = XML_ATTRIBUTE_FIXED;
5679
6.39k
  if (SKIP_BLANKS_PE == 0) {
5680
904
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5681
904
         "Space required after '#FIXED'\n");
5682
904
  }
5683
6.39k
    }
5684
62.0k
    ret = xmlParseAttValue(ctxt);
5685
62.0k
    if (ret == NULL) {
5686
9.28k
  xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo,
5687
9.28k
           "Attribute default value declaration error\n");
5688
9.28k
    } else
5689
52.7k
        *value = ret;
5690
62.0k
    return(val);
5691
90.7k
}
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
2.15k
xmlParseNotationType(xmlParserCtxt *ctxt) {
5712
2.15k
    const xmlChar *name;
5713
2.15k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5714
5715
2.15k
    if (RAW != '(') {
5716
322
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5717
322
  return(NULL);
5718
322
    }
5719
2.17k
    do {
5720
2.17k
        NEXT;
5721
2.17k
  SKIP_BLANKS_PE;
5722
2.17k
        name = xmlParseName(ctxt);
5723
2.17k
  if (name == NULL) {
5724
226
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5725
226
         "Name expected in NOTATION declaration\n");
5726
226
            xmlFreeEnumeration(ret);
5727
226
      return(NULL);
5728
226
  }
5729
1.94k
        tmp = NULL;
5730
1.94k
#ifdef LIBXML_VALID_ENABLED
5731
1.94k
        if (ctxt->validate) {
5732
0
            tmp = ret;
5733
0
            while (tmp != NULL) {
5734
0
                if (xmlStrEqual(name, tmp->name)) {
5735
0
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5736
0
              "standalone: attribute notation value token %s duplicated\n",
5737
0
                                     name, NULL);
5738
0
                    if (!xmlDictOwns(ctxt->dict, name))
5739
0
                        xmlFree((xmlChar *) name);
5740
0
                    break;
5741
0
                }
5742
0
                tmp = tmp->next;
5743
0
            }
5744
0
        }
5745
1.94k
#endif /* LIBXML_VALID_ENABLED */
5746
1.94k
  if (tmp == NULL) {
5747
1.94k
      cur = xmlCreateEnumeration(name);
5748
1.94k
      if (cur == NULL) {
5749
6
                xmlErrMemory(ctxt);
5750
6
                xmlFreeEnumeration(ret);
5751
6
                return(NULL);
5752
6
            }
5753
1.94k
      if (last == NULL) ret = last = cur;
5754
344
      else {
5755
344
    last->next = cur;
5756
344
    last = cur;
5757
344
      }
5758
1.94k
  }
5759
1.94k
  SKIP_BLANKS_PE;
5760
1.94k
    } while (RAW == '|');
5761
1.59k
    if (RAW != ')') {
5762
270
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5763
270
        xmlFreeEnumeration(ret);
5764
270
  return(NULL);
5765
270
    }
5766
1.32k
    NEXT;
5767
1.32k
    return(ret);
5768
1.59k
}
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
17.6k
xmlParseEnumerationType(xmlParserCtxt *ctxt) {
5787
17.6k
    xmlChar *name;
5788
17.6k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5789
5790
17.6k
    if (RAW != '(') {
5791
1.73k
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL);
5792
1.73k
  return(NULL);
5793
1.73k
    }
5794
25.1k
    do {
5795
25.1k
        NEXT;
5796
25.1k
  SKIP_BLANKS_PE;
5797
25.1k
        name = xmlParseNmtoken(ctxt);
5798
25.1k
  if (name == NULL) {
5799
236
      xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL);
5800
236
      return(ret);
5801
236
  }
5802
24.9k
        tmp = NULL;
5803
24.9k
#ifdef LIBXML_VALID_ENABLED
5804
24.9k
        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
24.9k
#endif /* LIBXML_VALID_ENABLED */
5819
24.9k
  if (tmp == NULL) {
5820
24.9k
      cur = xmlCreateEnumeration(name);
5821
24.9k
      if (!xmlDictOwns(ctxt->dict, name))
5822
24.9k
    xmlFree(name);
5823
24.9k
      if (cur == NULL) {
5824
23
                xmlErrMemory(ctxt);
5825
23
                xmlFreeEnumeration(ret);
5826
23
                return(NULL);
5827
23
            }
5828
24.8k
      if (last == NULL) ret = last = cur;
5829
9.22k
      else {
5830
9.22k
    last->next = cur;
5831
9.22k
    last = cur;
5832
9.22k
      }
5833
24.8k
  }
5834
24.8k
  SKIP_BLANKS_PE;
5835
24.8k
    } while (RAW == '|');
5836
15.6k
    if (RAW != ')') {
5837
585
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL);
5838
585
  return(ret);
5839
585
    }
5840
15.0k
    NEXT;
5841
15.0k
    return(ret);
5842
15.6k
}
5843
5844
/**
5845
 * Parse an Enumerated attribute type.
5846
 *
5847
 * @deprecated Internal function, don't use.
5848
 *
5849
 *     [57] EnumeratedType ::= NotationType | Enumeration
5850
 *
5851
 *     [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
5852
 *
5853
 * @param ctxt  an XML parser context
5854
 * @param tree  the enumeration tree built while parsing
5855
 * @returns XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION
5856
 */
5857
5858
int
5859
20.0k
xmlParseEnumeratedType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5860
20.0k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5861
2.34k
  SKIP(8);
5862
2.34k
  if (SKIP_BLANKS_PE == 0) {
5863
196
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5864
196
         "Space required after 'NOTATION'\n");
5865
196
      return(0);
5866
196
  }
5867
2.15k
  *tree = xmlParseNotationType(ctxt);
5868
2.15k
  if (*tree == NULL) return(0);
5869
1.32k
  return(XML_ATTRIBUTE_NOTATION);
5870
2.15k
    }
5871
17.6k
    *tree = xmlParseEnumerationType(ctxt);
5872
17.6k
    if (*tree == NULL) return(0);
5873
15.6k
    return(XML_ATTRIBUTE_ENUMERATION);
5874
17.6k
}
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
102k
xmlParseAttributeType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5923
102k
    if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) {
5924
14.4k
  SKIP(5);
5925
14.4k
  return(XML_ATTRIBUTE_CDATA);
5926
88.1k
     } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) {
5927
1.03k
  SKIP(6);
5928
1.03k
  return(XML_ATTRIBUTE_IDREFS);
5929
87.1k
     } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) {
5930
3.40k
  SKIP(5);
5931
3.40k
  return(XML_ATTRIBUTE_IDREF);
5932
83.7k
     } else if ((RAW == 'I') && (NXT(1) == 'D')) {
5933
17.6k
        SKIP(2);
5934
17.6k
  return(XML_ATTRIBUTE_ID);
5935
66.0k
     } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5936
23.7k
  SKIP(6);
5937
23.7k
  return(XML_ATTRIBUTE_ENTITY);
5938
42.3k
     } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) {
5939
5.68k
  SKIP(8);
5940
5.68k
  return(XML_ATTRIBUTE_ENTITIES);
5941
36.6k
     } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) {
5942
6.53k
  SKIP(8);
5943
6.53k
  return(XML_ATTRIBUTE_NMTOKENS);
5944
30.1k
     } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) {
5945
10.0k
  SKIP(7);
5946
10.0k
  return(XML_ATTRIBUTE_NMTOKEN);
5947
10.0k
     }
5948
20.0k
     return(xmlParseEnumeratedType(ctxt, tree));
5949
102k
}
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
77.8k
xmlParseAttributeListDecl(xmlParserCtxt *ctxt) {
5963
77.8k
    const xmlChar *elemName;
5964
77.8k
    const xmlChar *attrName;
5965
77.8k
    xmlEnumerationPtr tree;
5966
5967
77.8k
    if ((CUR != '<') || (NXT(1) != '!'))
5968
0
        return;
5969
77.8k
    SKIP(2);
5970
5971
77.8k
    if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) {
5972
77.7k
#ifdef LIBXML_VALID_ENABLED
5973
77.7k
  int oldInputNr = ctxt->inputNr;
5974
77.7k
#endif
5975
5976
77.7k
  SKIP(7);
5977
77.7k
  if (SKIP_BLANKS_PE == 0) {
5978
13.7k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5979
13.7k
                     "Space required after '<!ATTLIST'\n");
5980
13.7k
  }
5981
77.7k
        elemName = xmlParseName(ctxt);
5982
77.7k
  if (elemName == NULL) {
5983
6.78k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5984
6.78k
         "ATTLIST: no name for Element\n");
5985
6.78k
      return;
5986
6.78k
  }
5987
70.9k
  SKIP_BLANKS_PE;
5988
70.9k
  GROW;
5989
154k
  while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) {
5990
109k
      int type;
5991
109k
      int def;
5992
109k
      xmlChar *defaultValue = NULL;
5993
5994
109k
      GROW;
5995
109k
            tree = NULL;
5996
109k
      attrName = xmlParseName(ctxt);
5997
109k
      if (attrName == NULL) {
5998
2.87k
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5999
2.87k
             "ATTLIST: no name for Attribute\n");
6000
2.87k
    break;
6001
2.87k
      }
6002
107k
      GROW;
6003
107k
      if (SKIP_BLANKS_PE == 0) {
6004
4.48k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6005
4.48k
            "Space required after the attribute name\n");
6006
4.48k
    break;
6007
4.48k
      }
6008
6009
102k
      type = xmlParseAttributeType(ctxt, &tree);
6010
102k
      if (type <= 0) {
6011
3.00k
          break;
6012
3.00k
      }
6013
6014
99.5k
      GROW;
6015
99.5k
      if (SKIP_BLANKS_PE == 0) {
6016
2.62k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6017
2.62k
             "Space required after the attribute type\n");
6018
2.62k
          if (tree != NULL)
6019
740
        xmlFreeEnumeration(tree);
6020
2.62k
    break;
6021
2.62k
      }
6022
6023
96.9k
      def = xmlParseDefaultDecl(ctxt, &defaultValue);
6024
96.9k
      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
96.9k
      if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL))
6032
51.9k
          xmlAttrNormalizeSpace(defaultValue, defaultValue);
6033
6034
96.9k
      GROW;
6035
96.9k
            if (RAW != '>') {
6036
60.9k
    if (SKIP_BLANKS_PE == 0) {
6037
13.0k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6038
13.0k
      "Space required after the attribute default value\n");
6039
13.0k
        if (defaultValue != NULL)
6040
3.69k
      xmlFree(defaultValue);
6041
13.0k
        if (tree != NULL)
6042
791
      xmlFreeEnumeration(tree);
6043
13.0k
        break;
6044
13.0k
    }
6045
60.9k
      }
6046
83.9k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6047
79.3k
    (ctxt->sax->attributeDecl != NULL))
6048
79.3k
    ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName,
6049
79.3k
                          type, def, defaultValue, tree);
6050
4.57k
      else if (tree != NULL)
6051
1.97k
    xmlFreeEnumeration(tree);
6052
6053
83.9k
      if ((ctxt->sax2) && (defaultValue != NULL) &&
6054
49.0k
          (def != XML_ATTRIBUTE_IMPLIED) &&
6055
49.0k
    (def != XML_ATTRIBUTE_REQUIRED)) {
6056
49.0k
    xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue);
6057
49.0k
      }
6058
83.9k
      if (ctxt->sax2) {
6059
83.9k
    xmlAddSpecialAttr(ctxt, elemName, attrName, type);
6060
83.9k
      }
6061
83.9k
      if (defaultValue != NULL)
6062
49.0k
          xmlFree(defaultValue);
6063
83.9k
      GROW;
6064
83.9k
  }
6065
70.9k
  if (RAW == '>') {
6066
45.9k
#ifdef LIBXML_VALID_ENABLED
6067
45.9k
      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
45.9k
#endif
6074
45.9k
      NEXT;
6075
45.9k
  }
6076
70.9k
    }
6077
77.8k
}
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
274k
xmlSkipBlankCharsPEBalanced(xmlParserCtxt *ctxt, int openInputNr) {
6088
274k
    SKIP_BLANKS;
6089
274k
    GROW;
6090
6091
274k
    (void) openInputNr;
6092
6093
274k
    if (!PARSER_EXTERNAL(ctxt) && !PARSER_IN_PE(ctxt))
6094
127k
        return;
6095
6096
160k
    while (!PARSER_STOPPED(ctxt)) {
6097
159k
        if (ctxt->input->cur >= ctxt->input->end) {
6098
5.27k
#ifdef LIBXML_VALID_ENABLED
6099
5.27k
            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
5.27k
#endif
6106
5.27k
            if (PARSER_IN_PE(ctxt))
6107
5.21k
                xmlPopPE(ctxt);
6108
54
            else
6109
54
                break;
6110
154k
        } else if (RAW == '%') {
6111
7.84k
            xmlParsePERefInternal(ctxt, 0);
6112
146k
        } else {
6113
146k
            break;
6114
146k
        }
6115
6116
13.0k
        SKIP_BLANKS;
6117
13.0k
        GROW;
6118
13.0k
    }
6119
147k
}
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
9.53k
xmlParseElementMixedContentDecl(xmlParserCtxt *ctxt, int openInputNr) {
6143
9.53k
    xmlElementContentPtr ret = NULL, cur = NULL, n;
6144
9.53k
    const xmlChar *elem = NULL;
6145
6146
9.53k
    GROW;
6147
9.53k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6148
9.53k
  SKIP(7);
6149
9.53k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6150
9.53k
  if (RAW == ')') {
6151
7.14k
#ifdef LIBXML_VALID_ENABLED
6152
7.14k
      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
7.14k
#endif
6159
7.14k
      NEXT;
6160
7.14k
      ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6161
7.14k
      if (ret == NULL)
6162
9
                goto mem_error;
6163
7.13k
      if (RAW == '*') {
6164
208
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6165
208
    NEXT;
6166
208
      }
6167
7.13k
      return(ret);
6168
7.14k
  }
6169
2.39k
  if ((RAW == '(') || (RAW == '|')) {
6170
2.00k
      ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6171
2.00k
      if (ret == NULL)
6172
8
                goto mem_error;
6173
2.00k
  }
6174
21.3k
  while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) {
6175
19.1k
      NEXT;
6176
19.1k
            n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6177
19.1k
            if (n == NULL)
6178
6
                goto mem_error;
6179
19.1k
      if (elem == NULL) {
6180
1.98k
    n->c1 = cur;
6181
1.98k
    if (cur != NULL)
6182
1.98k
        cur->parent = n;
6183
1.98k
    ret = cur = n;
6184
17.1k
      } else {
6185
17.1k
          cur->c2 = n;
6186
17.1k
    n->parent = cur;
6187
17.1k
    n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6188
17.1k
                if (n->c1 == NULL)
6189
3
                    goto mem_error;
6190
17.1k
    n->c1->parent = n;
6191
17.1k
    cur = n;
6192
17.1k
      }
6193
19.0k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6194
19.0k
      elem = xmlParseName(ctxt);
6195
19.0k
      if (elem == NULL) {
6196
151
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6197
151
      "xmlParseElementMixedContentDecl : Name expected\n");
6198
151
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6199
151
    return(NULL);
6200
151
      }
6201
18.9k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6202
18.9k
  }
6203
2.22k
  if ((RAW == ')') && (NXT(1) == '*')) {
6204
1.69k
      if (elem != NULL) {
6205
1.69k
    cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem,
6206
1.69k
                                   XML_ELEMENT_CONTENT_ELEMENT);
6207
1.69k
    if (cur->c2 == NULL)
6208
6
                    goto mem_error;
6209
1.69k
    cur->c2->parent = cur;
6210
1.69k
            }
6211
1.69k
            if (ret != NULL)
6212
1.69k
                ret->ocur = XML_ELEMENT_CONTENT_MULT;
6213
1.69k
#ifdef LIBXML_VALID_ENABLED
6214
1.69k
      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
1.69k
#endif
6221
1.69k
      SKIP(2);
6222
1.69k
  } else {
6223
523
      xmlFreeDocElementContent(ctxt->myDoc, ret);
6224
523
      xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL);
6225
523
      return(NULL);
6226
523
  }
6227
6228
2.22k
    } else {
6229
0
  xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL);
6230
0
    }
6231
1.69k
    return(ret);
6232
6233
32
mem_error:
6234
32
    xmlErrMemory(ctxt);
6235
32
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6236
32
    return(NULL);
6237
9.53k
}
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
94.0k
                                       int depth) {
6272
94.0k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
6273
94.0k
    xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL;
6274
94.0k
    const xmlChar *elem;
6275
94.0k
    xmlChar type = 0;
6276
6277
94.0k
    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
94.0k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6284
94.0k
    if (RAW == '(') {
6285
71.7k
        int newInputNr = ctxt->inputNr;
6286
6287
        /* Recurse on first child */
6288
71.7k
  NEXT;
6289
71.7k
        cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6290
71.7k
                                                           depth + 1);
6291
71.7k
        if (cur == NULL)
6292
57.7k
            return(NULL);
6293
71.7k
    } else {
6294
22.3k
  elem = xmlParseName(ctxt);
6295
22.3k
  if (elem == NULL) {
6296
465
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6297
465
      return(NULL);
6298
465
  }
6299
21.8k
        cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6300
21.8k
  if (cur == NULL) {
6301
36
      xmlErrMemory(ctxt);
6302
36
      return(NULL);
6303
36
  }
6304
21.8k
  GROW;
6305
21.8k
  if (RAW == '?') {
6306
3.16k
      cur->ocur = XML_ELEMENT_CONTENT_OPT;
6307
3.16k
      NEXT;
6308
18.6k
  } else if (RAW == '*') {
6309
1.71k
      cur->ocur = XML_ELEMENT_CONTENT_MULT;
6310
1.71k
      NEXT;
6311
16.9k
  } else if (RAW == '+') {
6312
1.98k
      cur->ocur = XML_ELEMENT_CONTENT_PLUS;
6313
1.98k
      NEXT;
6314
14.9k
  } else {
6315
14.9k
      cur->ocur = XML_ELEMENT_CONTENT_ONCE;
6316
14.9k
  }
6317
21.8k
  GROW;
6318
21.8k
    }
6319
70.4k
    while (!PARSER_STOPPED(ctxt)) {
6320
69.0k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6321
69.0k
        if (RAW == ')')
6322
24.9k
            break;
6323
        /*
6324
   * Each loop we parse one separator and one element.
6325
   */
6326
44.0k
        if (RAW == ',') {
6327
14.4k
      if (type == 0) type = CUR;
6328
6329
      /*
6330
       * Detect "Name | Name , Name" error
6331
       */
6332
7.49k
      else if (type != CUR) {
6333
4
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6334
4
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6335
4
                      type);
6336
4
    if ((last != NULL) && (last != ret))
6337
4
        xmlFreeDocElementContent(ctxt->myDoc, last);
6338
4
    if (ret != NULL)
6339
4
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6340
4
    return(NULL);
6341
4
      }
6342
14.4k
      NEXT;
6343
6344
14.4k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ);
6345
14.4k
      if (op == NULL) {
6346
9
                xmlErrMemory(ctxt);
6347
9
    if ((last != NULL) && (last != ret))
6348
5
        xmlFreeDocElementContent(ctxt->myDoc, last);
6349
9
          xmlFreeDocElementContent(ctxt->myDoc, ret);
6350
9
    return(NULL);
6351
9
      }
6352
14.4k
      if (last == NULL) {
6353
6.98k
    op->c1 = ret;
6354
6.98k
    if (ret != NULL)
6355
6.98k
        ret->parent = op;
6356
6.98k
    ret = cur = op;
6357
7.48k
      } else {
6358
7.48k
          cur->c2 = op;
6359
7.48k
    if (op != NULL)
6360
7.48k
        op->parent = cur;
6361
7.48k
    op->c1 = last;
6362
7.48k
    if (last != NULL)
6363
7.48k
        last->parent = op;
6364
7.48k
    cur =op;
6365
7.48k
    last = NULL;
6366
7.48k
      }
6367
29.5k
  } else if (RAW == '|') {
6368
24.0k
      if (type == 0) type = CUR;
6369
6370
      /*
6371
       * Detect "Name , Name | Name" error
6372
       */
6373
14.8k
      else if (type != CUR) {
6374
3
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6375
3
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6376
3
          type);
6377
3
    if ((last != NULL) && (last != ret))
6378
3
        xmlFreeDocElementContent(ctxt->myDoc, last);
6379
3
    if (ret != NULL)
6380
3
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6381
3
    return(NULL);
6382
3
      }
6383
24.0k
      NEXT;
6384
6385
24.0k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6386
24.0k
      if (op == NULL) {
6387
12
                xmlErrMemory(ctxt);
6388
12
    if ((last != NULL) && (last != ret))
6389
4
        xmlFreeDocElementContent(ctxt->myDoc, last);
6390
12
    if (ret != NULL)
6391
12
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6392
12
    return(NULL);
6393
12
      }
6394
24.0k
      if (last == NULL) {
6395
9.26k
    op->c1 = ret;
6396
9.26k
    if (ret != NULL)
6397
9.26k
        ret->parent = op;
6398
9.26k
    ret = cur = op;
6399
14.7k
      } else {
6400
14.7k
          cur->c2 = op;
6401
14.7k
    if (op != NULL)
6402
14.7k
        op->parent = cur;
6403
14.7k
    op->c1 = last;
6404
14.7k
    if (last != NULL)
6405
14.7k
        last->parent = op;
6406
14.7k
    cur =op;
6407
14.7k
    last = NULL;
6408
14.7k
      }
6409
24.0k
  } else {
6410
5.49k
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL);
6411
5.49k
      if ((last != NULL) && (last != ret))
6412
1.49k
          xmlFreeDocElementContent(ctxt->myDoc, last);
6413
5.49k
      if (ret != NULL)
6414
5.49k
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6415
5.49k
      return(NULL);
6416
5.49k
  }
6417
38.5k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6418
38.5k
        if (RAW == '(') {
6419
6.39k
            int newInputNr = ctxt->inputNr;
6420
6421
      /* Recurse on second child */
6422
6.39k
      NEXT;
6423
6.39k
      last = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6424
6.39k
                                                          depth + 1);
6425
6.39k
            if (last == NULL) {
6426
2.20k
    if (ret != NULL)
6427
2.20k
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6428
2.20k
    return(NULL);
6429
2.20k
            }
6430
32.1k
  } else {
6431
32.1k
      elem = xmlParseName(ctxt);
6432
32.1k
      if (elem == NULL) {
6433
1.68k
    xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6434
1.68k
    if (ret != NULL)
6435
1.68k
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6436
1.68k
    return(NULL);
6437
1.68k
      }
6438
30.4k
      last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6439
30.4k
      if (last == NULL) {
6440
18
                xmlErrMemory(ctxt);
6441
18
    if (ret != NULL)
6442
18
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6443
18
    return(NULL);
6444
18
      }
6445
30.4k
      if (RAW == '?') {
6446
5.98k
    last->ocur = XML_ELEMENT_CONTENT_OPT;
6447
5.98k
    NEXT;
6448
24.4k
      } else if (RAW == '*') {
6449
4.42k
    last->ocur = XML_ELEMENT_CONTENT_MULT;
6450
4.42k
    NEXT;
6451
20.0k
      } else if (RAW == '+') {
6452
914
    last->ocur = XML_ELEMENT_CONTENT_PLUS;
6453
914
    NEXT;
6454
19.1k
      } else {
6455
19.1k
    last->ocur = XML_ELEMENT_CONTENT_ONCE;
6456
19.1k
      }
6457
30.4k
  }
6458
38.5k
    }
6459
26.3k
    if ((cur != NULL) && (last != NULL)) {
6460
10.8k
        cur->c2 = last;
6461
10.8k
  if (last != NULL)
6462
10.8k
      last->parent = cur;
6463
10.8k
    }
6464
26.3k
#ifdef LIBXML_VALID_ENABLED
6465
26.3k
    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
26.3k
#endif
6472
26.3k
    NEXT;
6473
26.3k
    if (RAW == '?') {
6474
874
  if (ret != NULL) {
6475
874
      if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) ||
6476
835
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6477
105
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6478
769
      else
6479
769
          ret->ocur = XML_ELEMENT_CONTENT_OPT;
6480
874
  }
6481
874
  NEXT;
6482
25.4k
    } else if (RAW == '*') {
6483
2.71k
  if (ret != NULL) {
6484
2.71k
      ret->ocur = XML_ELEMENT_CONTENT_MULT;
6485
2.71k
      cur = ret;
6486
      /*
6487
       * Some normalization:
6488
       * (a | b* | c?)* == (a | b | c)*
6489
       */
6490
11.2k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6491
8.51k
    if ((cur->c1 != NULL) &&
6492
8.51k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6493
7.53k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT)))
6494
1.37k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6495
8.51k
    if ((cur->c2 != NULL) &&
6496
8.51k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6497
8.29k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT)))
6498
619
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6499
8.51k
    cur = cur->c2;
6500
8.51k
      }
6501
2.71k
  }
6502
2.71k
  NEXT;
6503
22.7k
    } else if (RAW == '+') {
6504
6.01k
  if (ret != NULL) {
6505
6.01k
      int found = 0;
6506
6507
6.01k
      if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) ||
6508
5.93k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6509
500
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6510
5.51k
      else
6511
5.51k
          ret->ocur = XML_ELEMENT_CONTENT_PLUS;
6512
      /*
6513
       * Some normalization:
6514
       * (a | b*)+ == (a | b)*
6515
       * (a | b?)+ == (a | b)*
6516
       */
6517
10.4k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6518
4.39k
    if ((cur->c1 != NULL) &&
6519
4.39k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6520
2.68k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) {
6521
2.34k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6522
2.34k
        found = 1;
6523
2.34k
    }
6524
4.39k
    if ((cur->c2 != NULL) &&
6525
4.39k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6526
4.32k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) {
6527
1.29k
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6528
1.29k
        found = 1;
6529
1.29k
    }
6530
4.39k
    cur = cur->c2;
6531
4.39k
      }
6532
6.01k
      if (found)
6533
2.43k
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6534
6.01k
  }
6535
6.01k
  NEXT;
6536
6.01k
    }
6537
26.3k
    return(ret);
6538
35.7k
}
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
25.4k
                           xmlElementContent **result) {
6593
6594
25.4k
    xmlElementContentPtr tree = NULL;
6595
25.4k
    int openInputNr = ctxt->inputNr;
6596
25.4k
    int res;
6597
6598
25.4k
    *result = NULL;
6599
6600
25.4k
    if (RAW != '(') {
6601
0
  xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6602
0
    "xmlParseElementContentDecl : %s '(' expected\n", name);
6603
0
  return(-1);
6604
0
    }
6605
25.4k
    NEXT;
6606
25.4k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6607
25.4k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6608
9.53k
        tree = xmlParseElementMixedContentDecl(ctxt, openInputNr);
6609
9.53k
  res = XML_ELEMENT_TYPE_MIXED;
6610
15.9k
    } else {
6611
15.9k
        tree = xmlParseElementChildrenContentDeclPriv(ctxt, openInputNr, 1);
6612
15.9k
  res = XML_ELEMENT_TYPE_ELEMENT;
6613
15.9k
    }
6614
25.4k
    if (tree == NULL)
6615
8.43k
        return(-1);
6616
17.0k
    SKIP_BLANKS_PE;
6617
17.0k
    *result = tree;
6618
17.0k
    return(res);
6619
25.4k
}
6620
6621
/**
6622
 * Parse an element declaration. Always consumes '<!'.
6623
 *
6624
 * @deprecated Internal function, don't use.
6625
 *
6626
 *     [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>'
6627
 *
6628
 * [ VC: Unique Element Type Declaration ]
6629
 * No element type may be declared more than once
6630
 *
6631
 * @param ctxt  an XML parser context
6632
 * @returns the type of the element, or -1 in case of error
6633
 */
6634
int
6635
32.6k
xmlParseElementDecl(xmlParserCtxt *ctxt) {
6636
32.6k
    const xmlChar *name;
6637
32.6k
    int ret = -1;
6638
32.6k
    xmlElementContentPtr content  = NULL;
6639
6640
32.6k
    if ((CUR != '<') || (NXT(1) != '!'))
6641
0
        return(ret);
6642
32.6k
    SKIP(2);
6643
6644
    /* GROW; done in the caller */
6645
32.6k
    if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) {
6646
32.5k
#ifdef LIBXML_VALID_ENABLED
6647
32.5k
  int oldInputNr = ctxt->inputNr;
6648
32.5k
#endif
6649
6650
32.5k
  SKIP(7);
6651
32.5k
  if (SKIP_BLANKS_PE == 0) {
6652
264
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6653
264
               "Space required after 'ELEMENT'\n");
6654
264
      return(-1);
6655
264
  }
6656
32.3k
        name = xmlParseName(ctxt);
6657
32.3k
  if (name == NULL) {
6658
815
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6659
815
         "xmlParseElementDecl: no name for Element\n");
6660
815
      return(-1);
6661
815
  }
6662
31.4k
  if (SKIP_BLANKS_PE == 0) {
6663
4.85k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6664
4.85k
         "Space required after the element name\n");
6665
4.85k
  }
6666
31.4k
  if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) {
6667
3.67k
      SKIP(5);
6668
      /*
6669
       * Element must always be empty.
6670
       */
6671
3.67k
      ret = XML_ELEMENT_TYPE_EMPTY;
6672
27.8k
  } else if ((RAW == 'A') && (NXT(1) == 'N') &&
6673
593
             (NXT(2) == 'Y')) {
6674
590
      SKIP(3);
6675
      /*
6676
       * Element is a generic container.
6677
       */
6678
590
      ret = XML_ELEMENT_TYPE_ANY;
6679
27.2k
  } else if (RAW == '(') {
6680
25.4k
      ret = xmlParseElementContentDecl(ctxt, name, &content);
6681
25.4k
            if (ret <= 0)
6682
8.43k
                return(-1);
6683
25.4k
  } else {
6684
      /*
6685
       * [ WFC: PEs in Internal Subset ] error handling.
6686
       */
6687
1.77k
            xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6688
1.77k
                  "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n");
6689
1.77k
      return(-1);
6690
1.77k
  }
6691
6692
21.2k
  SKIP_BLANKS_PE;
6693
6694
21.2k
  if (RAW != '>') {
6695
935
      xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
6696
935
      if (content != NULL) {
6697
625
    xmlFreeDocElementContent(ctxt->myDoc, content);
6698
625
      }
6699
20.3k
  } else {
6700
20.3k
#ifdef LIBXML_VALID_ENABLED
6701
20.3k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6702
0
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6703
0
                                 "Element declaration doesn't start and stop in"
6704
0
                                 " the same entity\n",
6705
0
                                 NULL, NULL);
6706
0
      }
6707
20.3k
#endif
6708
6709
20.3k
      NEXT;
6710
20.3k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6711
15.2k
    (ctxt->sax->elementDecl != NULL)) {
6712
15.2k
    if (content != NULL)
6713
11.8k
        content->parent = NULL;
6714
15.2k
          ctxt->sax->elementDecl(ctxt->userData, name, ret,
6715
15.2k
                           content);
6716
15.2k
    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
4.38k
        xmlFreeDocElementContent(ctxt->myDoc, content);
6724
4.38k
    }
6725
15.2k
      } else if (content != NULL) {
6726
4.60k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6727
4.60k
      }
6728
20.3k
  }
6729
21.2k
    }
6730
21.3k
    return(ret);
6731
32.6k
}
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
10.6k
xmlParseConditionalSections(xmlParserCtxtPtr ctxt) {
6747
10.6k
    size_t depth = 0;
6748
10.6k
    int isFreshPE = 0;
6749
10.6k
    int oldInputNr = ctxt->inputNr;
6750
10.6k
    int declInputNr = ctxt->inputNr;
6751
6752
27.2k
    while (!PARSER_STOPPED(ctxt)) {
6753
27.2k
        if (ctxt->input->cur >= ctxt->input->end) {
6754
1.28k
            if (ctxt->inputNr <= oldInputNr) {
6755
411
                xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6756
411
                return;
6757
411
            }
6758
6759
876
            xmlPopPE(ctxt);
6760
876
            declInputNr = ctxt->inputNr;
6761
25.9k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6762
10.7k
            SKIP(3);
6763
10.7k
            SKIP_BLANKS_PE;
6764
6765
10.7k
            isFreshPE = 0;
6766
6767
10.7k
            if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) {
6768
5.32k
                SKIP(7);
6769
5.32k
                SKIP_BLANKS_PE;
6770
5.32k
                if (RAW != '[') {
6771
234
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6772
234
                    return;
6773
234
                }
6774
5.09k
#ifdef LIBXML_VALID_ENABLED
6775
5.09k
                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
5.09k
#endif
6782
5.09k
                NEXT;
6783
6784
5.09k
                depth++;
6785
5.46k
            } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) {
6786
1.77k
                size_t ignoreDepth = 0;
6787
6788
1.77k
                SKIP(6);
6789
1.77k
                SKIP_BLANKS_PE;
6790
1.77k
                if (RAW != '[') {
6791
554
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6792
554
                    return;
6793
554
                }
6794
1.22k
#ifdef LIBXML_VALID_ENABLED
6795
1.22k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6796
0
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6797
0
                                     "All markup of the conditional section is"
6798
0
                                     " not in the same entity\n",
6799
0
                                     NULL, NULL);
6800
0
                }
6801
1.22k
#endif
6802
1.22k
                NEXT;
6803
6804
13.2k
                while (PARSER_STOPPED(ctxt) == 0) {
6805
13.2k
                    if (RAW == 0) {
6806
518
                        xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL);
6807
518
                        return;
6808
518
                    }
6809
12.7k
                    if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6810
329
                        SKIP(3);
6811
329
                        ignoreDepth++;
6812
                        /* Check for integer overflow */
6813
329
                        if (ignoreDepth == 0) {
6814
0
                            xmlErrMemory(ctxt);
6815
0
                            return;
6816
0
                        }
6817
12.3k
                    } else if ((RAW == ']') && (NXT(1) == ']') &&
6818
1.23k
                               (NXT(2) == '>')) {
6819
886
                        SKIP(3);
6820
886
                        if (ignoreDepth == 0)
6821
690
                            break;
6822
196
                        ignoreDepth--;
6823
11.5k
                    } else {
6824
11.5k
                        NEXT;
6825
11.5k
                    }
6826
12.7k
                }
6827
6828
703
#ifdef LIBXML_VALID_ENABLED
6829
703
                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
703
#endif
6836
3.69k
            } else {
6837
3.69k
                xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL);
6838
3.69k
                return;
6839
3.69k
            }
6840
15.1k
        } else if ((depth > 0) &&
6841
15.1k
                   (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) {
6842
4.48k
            if (isFreshPE) {
6843
3
                xmlFatalErrMsg(ctxt, XML_ERR_CONDSEC_INVALID,
6844
3
                               "Parameter entity must match "
6845
3
                               "extSubsetDecl\n");
6846
3
                return;
6847
3
            }
6848
6849
4.48k
            depth--;
6850
4.48k
#ifdef LIBXML_VALID_ENABLED
6851
4.48k
            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
4.48k
#endif
6858
4.48k
            SKIP(3);
6859
10.6k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
6860
8.98k
            isFreshPE = 0;
6861
8.98k
            xmlParseMarkupDecl(ctxt);
6862
8.98k
        } else if (RAW == '%') {
6863
1.62k
            xmlParsePERefInternal(ctxt, 1);
6864
1.62k
            if (ctxt->inputNr > declInputNr) {
6865
902
                isFreshPE = 1;
6866
902
                declInputNr = ctxt->inputNr;
6867
902
            }
6868
1.62k
        } else {
6869
42
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6870
42
            return;
6871
42
        }
6872
6873
21.7k
        if (depth == 0)
6874
5.17k
            break;
6875
6876
16.5k
        SKIP_BLANKS;
6877
16.5k
        SHRINK;
6878
16.5k
        GROW;
6879
16.5k
    }
6880
10.6k
}
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
613k
xmlParseMarkupDecl(xmlParserCtxt *ctxt) {
6907
613k
    GROW;
6908
613k
    if (CUR == '<') {
6909
613k
        if (NXT(1) == '!') {
6910
591k
      switch (NXT(2)) {
6911
97.4k
          case 'E':
6912
97.4k
        if (NXT(3) == 'L')
6913
32.6k
      xmlParseElementDecl(ctxt);
6914
64.7k
        else if (NXT(3) == 'N')
6915
64.7k
      xmlParseEntityDecl(ctxt);
6916
25
                    else
6917
25
                        SKIP(2);
6918
97.4k
        break;
6919
77.8k
          case 'A':
6920
77.8k
        xmlParseAttributeListDecl(ctxt);
6921
77.8k
        break;
6922
9.72k
          case 'N':
6923
9.72k
        xmlParseNotationDecl(ctxt);
6924
9.72k
        break;
6925
393k
          case '-':
6926
393k
        xmlParseComment(ctxt);
6927
393k
        break;
6928
13.4k
    default:
6929
13.4k
                    xmlFatalErr(ctxt,
6930
13.4k
                                ctxt->inSubset == 2 ?
6931
1.70k
                                    XML_ERR_EXT_SUBSET_NOT_FINISHED :
6932
13.4k
                                    XML_ERR_INT_SUBSET_NOT_FINISHED,
6933
13.4k
                                NULL);
6934
13.4k
                    SKIP(2);
6935
13.4k
        break;
6936
591k
      }
6937
591k
  } else if (NXT(1) == '?') {
6938
21.7k
      xmlParsePI(ctxt);
6939
21.7k
  }
6940
613k
    }
6941
613k
}
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
22.4k
xmlParseTextDecl(xmlParserCtxt *ctxt) {
6954
22.4k
    xmlChar *version;
6955
6956
    /*
6957
     * We know that '<?xml' is here.
6958
     */
6959
22.4k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
6960
22.4k
  SKIP(5);
6961
22.4k
    } else {
6962
6
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL);
6963
6
  return;
6964
6
    }
6965
6966
22.4k
    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
22.4k
    version = xmlParseVersionInfo(ctxt);
6975
22.4k
    if (version == NULL) {
6976
18.0k
  version = xmlCharStrdup(XML_DEFAULT_VERSION);
6977
18.0k
        if (version == NULL) {
6978
40
            xmlErrMemory(ctxt);
6979
40
            return;
6980
40
        }
6981
18.0k
    } else {
6982
4.35k
  if (SKIP_BLANKS == 0) {
6983
642
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6984
642
               "Space needed here\n");
6985
642
  }
6986
4.35k
    }
6987
22.3k
    ctxt->input->version = version;
6988
6989
    /*
6990
     * We must have the encoding declaration
6991
     */
6992
22.3k
    xmlParseEncodingDecl(ctxt);
6993
6994
22.3k
    SKIP_BLANKS;
6995
22.3k
    if ((RAW == '?') && (NXT(1) == '>')) {
6996
2.28k
        SKIP(2);
6997
20.1k
    } else if (RAW == '>') {
6998
        /* Deprecated old WD ... */
6999
786
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7000
786
  NEXT;
7001
19.3k
    } else {
7002
19.3k
        int c;
7003
7004
19.3k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7005
335M
        while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) {
7006
335M
            NEXT;
7007
335M
            if (c == '>')
7008
4.66k
                break;
7009
335M
        }
7010
19.3k
    }
7011
22.3k
}
7012
7013
/**
7014
 * Parse Markup declarations from an external subset
7015
 *
7016
 * @deprecated Internal function, don't use.
7017
 *
7018
 *     [30] extSubset ::= textDecl? extSubsetDecl
7019
 *
7020
 *     [31] extSubsetDecl ::= (markupdecl | conditionalSect |
7021
 *                             PEReference | S) *
7022
 * @param ctxt  an XML parser context
7023
 * @param publicId  the public identifier
7024
 * @param systemId  the system identifier (URL)
7025
 */
7026
void
7027
xmlParseExternalSubset(xmlParserCtxt *ctxt, const xmlChar *publicId,
7028
1.46k
                       const xmlChar *systemId) {
7029
1.46k
    int oldInputNr;
7030
7031
1.46k
    xmlCtxtInitializeLate(ctxt);
7032
7033
1.46k
    xmlDetectEncoding(ctxt);
7034
7035
1.46k
    if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) {
7036
115
  xmlParseTextDecl(ctxt);
7037
115
    }
7038
1.46k
    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
1.46k
    if ((ctxt->myDoc->intSubset == NULL) &&
7047
0
        (xmlCreateIntSubset(ctxt->myDoc, NULL, publicId, systemId) == NULL)) {
7048
0
        xmlErrMemory(ctxt);
7049
0
    }
7050
7051
1.46k
    ctxt->inSubset = 2;
7052
1.46k
    oldInputNr = ctxt->inputNr;
7053
7054
1.46k
    SKIP_BLANKS;
7055
421k
    while (!PARSER_STOPPED(ctxt)) {
7056
421k
        if (ctxt->input->cur >= ctxt->input->end) {
7057
946
            if (ctxt->inputNr <= oldInputNr) {
7058
327
                xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED);
7059
327
                break;
7060
327
            }
7061
7062
619
            xmlPopPE(ctxt);
7063
420k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
7064
7.54k
            xmlParseConditionalSections(ctxt);
7065
412k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
7066
410k
            xmlParseMarkupDecl(ctxt);
7067
410k
        } else if (RAW == '%') {
7068
1.68k
            xmlParsePERefInternal(ctxt, 1);
7069
1.68k
        } else {
7070
721
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
7071
7072
976
            while (ctxt->inputNr > oldInputNr)
7073
255
                xmlPopPE(ctxt);
7074
721
            break;
7075
721
        }
7076
420k
        SKIP_BLANKS;
7077
420k
        SHRINK;
7078
420k
        GROW;
7079
420k
    }
7080
1.46k
}
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
284k
xmlParseReference(xmlParserCtxt *ctxt) {
7097
284k
    xmlEntityPtr ent = NULL;
7098
284k
    const xmlChar *name;
7099
284k
    xmlChar *val;
7100
7101
284k
    if (RAW != '&')
7102
0
        return;
7103
7104
    /*
7105
     * Simple case of a CharRef
7106
     */
7107
284k
    if (NXT(1) == '#') {
7108
63.7k
  int i = 0;
7109
63.7k
  xmlChar out[16];
7110
63.7k
  int value = xmlParseCharRef(ctxt);
7111
7112
63.7k
  if (value == 0)
7113
19.2k
      return;
7114
7115
        /*
7116
         * Just encode the value in UTF-8
7117
         */
7118
44.4k
        COPY_BUF(out, i, value);
7119
44.4k
        out[i] = 0;
7120
44.4k
        if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7121
44.4k
            (!ctxt->disableSAX))
7122
31.5k
            ctxt->sax->characters(ctxt->userData, out, i);
7123
44.4k
  return;
7124
63.7k
    }
7125
7126
    /*
7127
     * We are seeing an entity reference
7128
     */
7129
220k
    name = xmlParseEntityRefInternal(ctxt);
7130
220k
    if (name == NULL)
7131
25.2k
        return;
7132
195k
    ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0);
7133
195k
    if (ent == NULL) {
7134
        /*
7135
         * Create a reference for undeclared entities.
7136
         */
7137
83.7k
        if ((ctxt->replaceEntities == 0) &&
7138
13.9k
            (ctxt->sax != NULL) &&
7139
13.9k
            (ctxt->disableSAX == 0) &&
7140
13.0k
            (ctxt->sax->reference != NULL)) {
7141
13.0k
            ctxt->sax->reference(ctxt->userData, name);
7142
13.0k
        }
7143
83.7k
        return;
7144
83.7k
    }
7145
111k
    if (!ctxt->wellFormed)
7146
25.6k
  return;
7147
7148
    /* special case of predefined entities */
7149
85.6k
    if ((ent->name == NULL) ||
7150
85.6k
        (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
7151
431
  val = ent->content;
7152
431
  if (val == NULL) return;
7153
  /*
7154
   * inline the entity.
7155
   */
7156
431
  if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7157
431
      (!ctxt->disableSAX))
7158
431
      ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val));
7159
431
  return;
7160
431
    }
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
85.2k
    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
85.2k
    if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) ||
7211
9.58k
        (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
7212
9.38k
         ((ctxt->replaceEntities) ||
7213
84.7k
          (ctxt->validate)))) {
7214
84.7k
        if ((ent->flags & XML_ENT_PARSED) == 0) {
7215
4.45k
            xmlCtxtParseEntity(ctxt, ent);
7216
80.3k
        } else if (ent->children == NULL) {
7217
            /*
7218
             * Probably running in SAX mode and the callbacks don't
7219
             * build the entity content. Parse the entity again.
7220
             *
7221
             * This will also be triggered in normal tree builder mode
7222
             * if an entity happens to be empty, causing unnecessary
7223
             * reloads. It's hard to come up with a reliable check in
7224
             * which mode we're running.
7225
             */
7226
72.0k
            xmlCtxtParseEntity(ctxt, ent);
7227
72.0k
        }
7228
84.7k
    }
7229
7230
    /*
7231
     * We also check for amplification if entities aren't substituted.
7232
     * They might be expanded later.
7233
     */
7234
85.2k
    if (xmlParserEntityCheck(ctxt, ent->expandedSize))
7235
119
        return;
7236
7237
85.1k
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
7238
876
        return;
7239
7240
84.2k
    if (ctxt->replaceEntities == 0) {
7241
  /*
7242
   * Create a reference
7243
   */
7244
769
        if (ctxt->sax->reference != NULL)
7245
769
      ctxt->sax->reference(ctxt->userData, ent->name);
7246
83.4k
    } else if ((ent->children != NULL) && (ctxt->node != NULL)) {
7247
11.0k
        xmlNodePtr copy, cur;
7248
7249
        /*
7250
         * Seems we are generating the DOM content, copy the tree
7251
   */
7252
11.0k
        cur = ent->children;
7253
7254
        /*
7255
         * Handle first text node with SAX to coalesce text efficiently
7256
         */
7257
11.0k
        if ((cur->type == XML_TEXT_NODE) ||
7258
10.4k
            (cur->type == XML_CDATA_SECTION_NODE)) {
7259
10.4k
            int len = xmlStrlen(cur->content);
7260
7261
10.4k
            if ((cur->type == XML_TEXT_NODE) ||
7262
9.82k
                (ctxt->options & XML_PARSE_NOCDATA)) {
7263
9.82k
                if (ctxt->sax->characters != NULL)
7264
9.82k
                    ctxt->sax->characters(ctxt, cur->content, len);
7265
9.82k
            } else {
7266
638
                if (ctxt->sax->cdataBlock != NULL)
7267
638
                    ctxt->sax->cdataBlock(ctxt, cur->content, len);
7268
638
            }
7269
7270
10.4k
            cur = cur->next;
7271
10.4k
        }
7272
7273
71.9k
        while (cur != NULL) {
7274
63.4k
            xmlNodePtr last;
7275
7276
            /*
7277
             * Handle last text node with SAX to coalesce text efficiently
7278
             */
7279
63.4k
            if ((cur->next == NULL) &&
7280
4.58k
                ((cur->type == XML_TEXT_NODE) ||
7281
2.39k
                 (cur->type == XML_CDATA_SECTION_NODE))) {
7282
2.26k
                int len = xmlStrlen(cur->content);
7283
7284
2.26k
                if ((cur->type == XML_TEXT_NODE) ||
7285
2.19k
                    (ctxt->options & XML_PARSE_NOCDATA)) {
7286
2.19k
                    if (ctxt->sax->characters != NULL)
7287
2.19k
                        ctxt->sax->characters(ctxt, cur->content, len);
7288
2.19k
                } else {
7289
75
                    if (ctxt->sax->cdataBlock != NULL)
7290
75
                        ctxt->sax->cdataBlock(ctxt, cur->content, len);
7291
75
                }
7292
7293
2.26k
                break;
7294
2.26k
            }
7295
7296
            /*
7297
             * Reset coalesce buffer stats only for non-text nodes.
7298
             */
7299
61.1k
            ctxt->nodemem = 0;
7300
61.1k
            ctxt->nodelen = 0;
7301
7302
61.1k
            copy = xmlDocCopyNode(cur, ctxt->myDoc, 1);
7303
7304
61.1k
            if (copy == NULL) {
7305
235
                xmlErrMemory(ctxt);
7306
235
                break;
7307
235
            }
7308
7309
60.9k
            if (ctxt->parseMode == XML_PARSE_READER) {
7310
                /* Needed for reader */
7311
0
                copy->extra = cur->extra;
7312
                /* Maybe needed for reader */
7313
0
                copy->_private = cur->_private;
7314
0
            }
7315
7316
60.9k
            copy->parent = ctxt->node;
7317
60.9k
            last = ctxt->node->last;
7318
60.9k
            if (last == NULL) {
7319
145
                ctxt->node->children = copy;
7320
60.7k
            } else {
7321
60.7k
                last->next = copy;
7322
60.7k
                copy->prev = last;
7323
60.7k
            }
7324
60.9k
            ctxt->node->last = copy;
7325
7326
60.9k
            cur = cur->next;
7327
60.9k
        }
7328
11.0k
    }
7329
84.2k
}
7330
7331
static void
7332
383k
xmlHandleUndeclaredEntity(xmlParserCtxtPtr ctxt, const xmlChar *name) {
7333
    /*
7334
     * [ WFC: Entity Declared ]
7335
     * In a document without any DTD, a document with only an
7336
     * internal DTD subset which contains no parameter entity
7337
     * references, or a document with "standalone='yes'", the
7338
     * Name given in the entity reference must match that in an
7339
     * entity declaration, except that well-formed documents
7340
     * need not declare any of the following entities: amp, lt,
7341
     * gt, apos, quot.
7342
     * The declaration of a parameter entity must precede any
7343
     * reference to it.
7344
     * Similarly, the declaration of a general entity must
7345
     * precede any reference to it which appears in a default
7346
     * value in an attribute-list declaration. Note that if
7347
     * entities are declared in the external subset or in
7348
     * external parameter entities, a non-validating processor
7349
     * is not obligated to read and process their declarations;
7350
     * for such documents, the rule that an entity must be
7351
     * declared is a well-formedness constraint only if
7352
     * standalone='yes'.
7353
     */
7354
383k
    if ((ctxt->standalone == 1) ||
7355
382k
        ((ctxt->hasExternalSubset == 0) &&
7356
379k
         (ctxt->hasPErefs == 0))) {
7357
304k
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY,
7358
304k
                          "Entity '%s' not defined\n", name);
7359
304k
#ifdef LIBXML_VALID_ENABLED
7360
304k
    } else if (ctxt->validate) {
7361
        /*
7362
         * [ VC: Entity Declared ]
7363
         * In a document with an external subset or external
7364
         * parameter entities with "standalone='no'", ...
7365
         * ... The declaration of a parameter entity must
7366
         * precede any reference to it...
7367
         */
7368
0
        xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY,
7369
0
                         "Entity '%s' not defined\n", name, NULL);
7370
0
#endif
7371
79.3k
    } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) ||
7372
6.01k
               ((ctxt->replaceEntities) &&
7373
74.6k
                ((ctxt->options & XML_PARSE_NO_XXE) == 0))) {
7374
        /*
7375
         * Also raise a non-fatal error
7376
         *
7377
         * - if the external subset is loaded and all entity declarations
7378
         *   should be available, or
7379
         * - entity substition was requested without restricting
7380
         *   external entity access.
7381
         */
7382
74.6k
        xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY,
7383
74.6k
                     "Entity '%s' not defined\n", name);
7384
74.6k
    } else {
7385
4.64k
        xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7386
4.64k
                      "Entity '%s' not defined\n", name, NULL);
7387
4.64k
    }
7388
7389
383k
    ctxt->valid = 0;
7390
383k
}
7391
7392
static xmlEntityPtr
7393
1.66M
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) {
7394
1.66M
    xmlEntityPtr ent = NULL;
7395
7396
    /*
7397
     * Predefined entities override any extra definition
7398
     */
7399
1.66M
    if ((ctxt->options & XML_PARSE_OLDSAX) == 0) {
7400
1.04M
        ent = xmlGetPredefinedEntity(name);
7401
1.04M
        if (ent != NULL)
7402
101k
            return(ent);
7403
1.04M
    }
7404
7405
    /*
7406
     * Ask first SAX for entity resolution, otherwise try the
7407
     * entities which may have stored in the parser context.
7408
     */
7409
1.56M
    if (ctxt->sax != NULL) {
7410
1.56M
  if (ctxt->sax->getEntity != NULL)
7411
1.56M
      ent = ctxt->sax->getEntity(ctxt->userData, name);
7412
1.56M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7413
17.7k
      (ctxt->options & XML_PARSE_OLDSAX))
7414
274
      ent = xmlGetPredefinedEntity(name);
7415
1.56M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7416
17.7k
      (ctxt->userData==ctxt)) {
7417
17.7k
      ent = xmlSAX2GetEntity(ctxt, name);
7418
17.7k
  }
7419
1.56M
    }
7420
7421
1.56M
    if (ent == NULL) {
7422
329k
        xmlHandleUndeclaredEntity(ctxt, name);
7423
329k
    }
7424
7425
    /*
7426
     * [ WFC: Parsed Entity ]
7427
     * An entity reference must not contain the name of an
7428
     * unparsed entity
7429
     */
7430
1.23M
    else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) {
7431
194
  xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY,
7432
194
     "Entity reference to unparsed entity %s\n", name);
7433
194
        ent = NULL;
7434
194
    }
7435
7436
    /*
7437
     * [ WFC: No External Entity References ]
7438
     * Attribute values cannot contain direct or indirect
7439
     * entity references to external entities.
7440
     */
7441
1.23M
    else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) {
7442
20.2k
        if (inAttr) {
7443
1.42k
            xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL,
7444
1.42k
                 "Attribute references external entity '%s'\n", name);
7445
1.42k
            ent = NULL;
7446
1.42k
        }
7447
20.2k
    }
7448
7449
1.56M
    return(ent);
7450
1.66M
}
7451
7452
/**
7453
 * Parse an entity reference. Always consumes '&'.
7454
 *
7455
 *     [68] EntityRef ::= '&' Name ';'
7456
 *
7457
 * @param ctxt  an XML parser context
7458
 * @returns the name, or NULL in case of error.
7459
 */
7460
static const xmlChar *
7461
1.02M
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) {
7462
1.02M
    const xmlChar *name;
7463
7464
1.02M
    GROW;
7465
7466
1.02M
    if (RAW != '&')
7467
0
        return(NULL);
7468
1.02M
    NEXT;
7469
1.02M
    name = xmlParseName(ctxt);
7470
1.02M
    if (name == NULL) {
7471
43.2k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7472
43.2k
           "xmlParseEntityRef: no name\n");
7473
43.2k
        return(NULL);
7474
43.2k
    }
7475
984k
    if (RAW != ';') {
7476
30.4k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7477
30.4k
  return(NULL);
7478
30.4k
    }
7479
954k
    NEXT;
7480
7481
954k
    return(name);
7482
984k
}
7483
7484
/**
7485
 * @deprecated Internal function, don't use.
7486
 *
7487
 * @param ctxt  an XML parser context
7488
 * @returns the xmlEntity if found, or NULL otherwise.
7489
 */
7490
xmlEntity *
7491
0
xmlParseEntityRef(xmlParserCtxt *ctxt) {
7492
0
    const xmlChar *name;
7493
7494
0
    if (ctxt == NULL)
7495
0
        return(NULL);
7496
7497
0
    name = xmlParseEntityRefInternal(ctxt);
7498
0
    if (name == NULL)
7499
0
        return(NULL);
7500
7501
0
    return(xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 0));
7502
0
}
7503
7504
/**
7505
 * Parse ENTITY references declarations, but this version parses it from
7506
 * a string value.
7507
 *
7508
 *     [68] EntityRef ::= '&' Name ';'
7509
 *
7510
 * [ WFC: Entity Declared ]
7511
 * In a document without any DTD, a document with only an internal DTD
7512
 * subset which contains no parameter entity references, or a document
7513
 * with "standalone='yes'", the Name given in the entity reference
7514
 * must match that in an entity declaration, except that well-formed
7515
 * documents need not declare any of the following entities: amp, lt,
7516
 * gt, apos, quot.  The declaration of a parameter entity must precede
7517
 * any reference to it.  Similarly, the declaration of a general entity
7518
 * must precede any reference to it which appears in a default value in an
7519
 * attribute-list declaration. Note that if entities are declared in the
7520
 * external subset or in external parameter entities, a non-validating
7521
 * processor is not obligated to read and process their declarations;
7522
 * for such documents, the rule that an entity must be declared is a
7523
 * well-formedness constraint only if standalone='yes'.
7524
 *
7525
 * [ WFC: Parsed Entity ]
7526
 * An entity reference must not contain the name of an unparsed entity
7527
 *
7528
 * @param ctxt  an XML parser context
7529
 * @param str  a pointer to an index in the string
7530
 * @returns the xmlEntity if found, or NULL otherwise. The str pointer
7531
 * is updated to the current location in the string.
7532
 */
7533
static xmlChar *
7534
711k
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) {
7535
711k
    xmlChar *name;
7536
711k
    const xmlChar *ptr;
7537
711k
    xmlChar cur;
7538
7539
711k
    if ((str == NULL) || (*str == NULL))
7540
0
        return(NULL);
7541
711k
    ptr = *str;
7542
711k
    cur = *ptr;
7543
711k
    if (cur != '&')
7544
0
  return(NULL);
7545
7546
711k
    ptr++;
7547
711k
    name = xmlParseStringName(ctxt, &ptr);
7548
711k
    if (name == NULL) {
7549
49
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7550
49
           "xmlParseStringEntityRef: no name\n");
7551
49
  *str = ptr;
7552
49
  return(NULL);
7553
49
    }
7554
711k
    if (*ptr != ';') {
7555
15
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7556
15
        xmlFree(name);
7557
15
  *str = ptr;
7558
15
  return(NULL);
7559
15
    }
7560
711k
    ptr++;
7561
7562
711k
    *str = ptr;
7563
711k
    return(name);
7564
711k
}
7565
7566
/**
7567
 * Parse a parameter entity reference. Always consumes '%'.
7568
 *
7569
 * The entity content is handled directly by pushing it's content as
7570
 * a new input stream.
7571
 *
7572
 *     [69] PEReference ::= '%' Name ';'
7573
 *
7574
 * [ WFC: No Recursion ]
7575
 * A parsed entity must not contain a recursive
7576
 * reference to itself, either directly or indirectly.
7577
 *
7578
 * [ WFC: Entity Declared ]
7579
 * In a document without any DTD, a document with only an internal DTD
7580
 * subset which contains no parameter entity references, or a document
7581
 * with "standalone='yes'", ...  ... The declaration of a parameter
7582
 * entity must precede any reference to it...
7583
 *
7584
 * [ VC: Entity Declared ]
7585
 * In a document with an external subset or external parameter entities
7586
 * with "standalone='no'", ...  ... The declaration of a parameter entity
7587
 * must precede any reference to it...
7588
 *
7589
 * [ WFC: In DTD ]
7590
 * Parameter-entity references may only appear in the DTD.
7591
 * NOTE: misleading but this is handled.
7592
 *
7593
 * @param ctxt  an XML parser context
7594
 * @param markupDecl  whether the PERef starts a markup declaration
7595
 */
7596
static void
7597
159k
xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl) {
7598
159k
    const xmlChar *name;
7599
159k
    xmlEntityPtr entity = NULL;
7600
159k
    xmlParserInputPtr input;
7601
7602
159k
    if (RAW != '%')
7603
0
        return;
7604
159k
    NEXT;
7605
159k
    name = xmlParseName(ctxt);
7606
159k
    if (name == NULL) {
7607
10.0k
  xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n");
7608
10.0k
  return;
7609
10.0k
    }
7610
149k
    if (RAW != ';') {
7611
14.3k
  xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL);
7612
14.3k
        return;
7613
14.3k
    }
7614
7615
134k
    NEXT;
7616
7617
    /* Must be set before xmlHandleUndeclaredEntity */
7618
134k
    ctxt->hasPErefs = 1;
7619
7620
    /*
7621
     * Request the entity from SAX
7622
     */
7623
134k
    if ((ctxt->sax != NULL) &&
7624
134k
  (ctxt->sax->getParameterEntity != NULL))
7625
134k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7626
7627
134k
    if (entity == NULL) {
7628
51.8k
        xmlHandleUndeclaredEntity(ctxt, name);
7629
82.8k
    } else {
7630
  /*
7631
   * Internal checking in case the entity quest barfed
7632
   */
7633
82.8k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7634
58.7k
      (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) {
7635
0
      xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7636
0
      "Internal: %%%s; is not a parameter entity\n",
7637
0
        name, NULL);
7638
82.8k
  } else {
7639
82.8k
      if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) &&
7640
58.7k
                ((ctxt->options & XML_PARSE_NO_XXE) ||
7641
58.3k
     (((ctxt->loadsubset & ~XML_SKIP_IDS) == 0) &&
7642
6.34k
      (ctxt->replaceEntities == 0) &&
7643
194
      (ctxt->validate == 0))))
7644
522
    return;
7645
7646
82.3k
            if (entity->flags & XML_ENT_EXPANDING) {
7647
20
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
7648
20
                return;
7649
20
            }
7650
7651
82.3k
      input = xmlNewEntityInputStream(ctxt, entity);
7652
82.3k
      if (xmlCtxtPushInput(ctxt, input) < 0) {
7653
2.59k
                xmlFreeInputStream(input);
7654
2.59k
    return;
7655
2.59k
            }
7656
7657
79.7k
            entity->flags |= XML_ENT_EXPANDING;
7658
7659
79.7k
            if (markupDecl)
7660
66.4k
                input->flags |= XML_INPUT_MARKUP_DECL;
7661
7662
79.7k
            GROW;
7663
7664
79.7k
      if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
7665
55.6k
                xmlDetectEncoding(ctxt);
7666
7667
55.6k
                if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
7668
11.3k
                    (IS_BLANK_CH(NXT(5)))) {
7669
10.5k
                    xmlParseTextDecl(ctxt);
7670
10.5k
                }
7671
55.6k
            }
7672
79.7k
  }
7673
82.8k
    }
7674
134k
}
7675
7676
/**
7677
 * Parse a parameter entity reference.
7678
 *
7679
 * @deprecated Internal function, don't use.
7680
 *
7681
 * @param ctxt  an XML parser context
7682
 */
7683
void
7684
0
xmlParsePEReference(xmlParserCtxt *ctxt) {
7685
0
    xmlParsePERefInternal(ctxt, 0);
7686
0
}
7687
7688
/**
7689
 * Load the content of an entity.
7690
 *
7691
 * @param ctxt  an XML parser context
7692
 * @param entity  an unloaded system entity
7693
 * @returns 0 in case of success and -1 in case of failure
7694
 */
7695
static int
7696
18.9k
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) {
7697
18.9k
    xmlParserInputPtr oldinput, input = NULL;
7698
18.9k
    xmlParserInputPtr *oldinputTab;
7699
18.9k
    xmlChar *oldencoding;
7700
18.9k
    xmlChar *content = NULL;
7701
18.9k
    xmlResourceType rtype;
7702
18.9k
    size_t length, i;
7703
18.9k
    int oldinputNr, oldinputMax;
7704
18.9k
    int ret = -1;
7705
18.9k
    int res;
7706
7707
18.9k
    if ((ctxt == NULL) || (entity == NULL) ||
7708
18.9k
        ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) &&
7709
0
   (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) ||
7710
18.9k
  (entity->content != NULL)) {
7711
0
  xmlFatalErr(ctxt, XML_ERR_ARGUMENT,
7712
0
              "xmlLoadEntityContent parameter error");
7713
0
        return(-1);
7714
0
    }
7715
7716
18.9k
    if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY)
7717
18.9k
        rtype = XML_RESOURCE_PARAMETER_ENTITY;
7718
0
    else
7719
0
        rtype = XML_RESOURCE_GENERAL_ENTITY;
7720
7721
18.9k
    input = xmlLoadResource(ctxt, (char *) entity->URI,
7722
18.9k
                            (char *) entity->ExternalID, rtype);
7723
18.9k
    if (input == NULL)
7724
412
        return(-1);
7725
7726
18.5k
    oldinput = ctxt->input;
7727
18.5k
    oldinputNr = ctxt->inputNr;
7728
18.5k
    oldinputMax = ctxt->inputMax;
7729
18.5k
    oldinputTab = ctxt->inputTab;
7730
18.5k
    oldencoding = ctxt->encoding;
7731
7732
18.5k
    ctxt->input = NULL;
7733
18.5k
    ctxt->inputNr = 0;
7734
18.5k
    ctxt->inputMax = 1;
7735
18.5k
    ctxt->encoding = NULL;
7736
18.5k
    ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr));
7737
18.5k
    if (ctxt->inputTab == NULL) {
7738
3
        xmlErrMemory(ctxt);
7739
3
        xmlFreeInputStream(input);
7740
3
        goto error;
7741
3
    }
7742
7743
18.5k
    xmlBufResetInput(input->buf->buffer, input);
7744
7745
18.5k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
7746
17
        xmlFreeInputStream(input);
7747
17
        goto error;
7748
17
    }
7749
7750
18.5k
    xmlDetectEncoding(ctxt);
7751
7752
    /*
7753
     * Parse a possible text declaration first
7754
     */
7755
18.5k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
7756
11.3k
  xmlParseTextDecl(ctxt);
7757
        /*
7758
         * An XML-1.0 document can't reference an entity not XML-1.0
7759
         */
7760
11.3k
        if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
7761
11.1k
            (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
7762
1.59k
            xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
7763
1.59k
                           "Version mismatch between document and entity\n");
7764
1.59k
        }
7765
11.3k
    }
7766
7767
18.5k
    length = input->cur - input->base;
7768
18.5k
    xmlBufShrink(input->buf->buffer, length);
7769
18.5k
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7770
7771
23.7k
    while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0)
7772
5.19k
        ;
7773
7774
18.5k
    xmlBufResetInput(input->buf->buffer, input);
7775
7776
18.5k
    if (res < 0) {
7777
1.74k
        xmlCtxtErrIO(ctxt, input->buf->error, NULL);
7778
1.74k
        goto error;
7779
1.74k
    }
7780
7781
16.8k
    length = xmlBufUse(input->buf->buffer);
7782
16.8k
    if (length > INT_MAX) {
7783
0
        xmlErrMemory(ctxt);
7784
0
        goto error;
7785
0
    }
7786
7787
16.8k
    content = xmlStrndup(xmlBufContent(input->buf->buffer), length);
7788
16.8k
    if (content == NULL) {
7789
23
        xmlErrMemory(ctxt);
7790
23
        goto error;
7791
23
    }
7792
7793
4.39M
    for (i = 0; i < length; ) {
7794
4.39M
        int clen = length - i;
7795
4.39M
        int c = xmlGetUTF8Char(content + i, &clen);
7796
7797
4.39M
        if ((c < 0) || (!IS_CHAR(c))) {
7798
16.6k
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
7799
16.6k
                              "xmlLoadEntityContent: invalid char value %d\n",
7800
16.6k
                              content[i]);
7801
16.6k
            goto error;
7802
16.6k
        }
7803
4.38M
        i += clen;
7804
4.38M
    }
7805
7806
145
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7807
145
    entity->content = content;
7808
145
    entity->length = length;
7809
145
    content = NULL;
7810
145
    ret = 0;
7811
7812
18.5k
error:
7813
37.1k
    while (ctxt->inputNr > 0)
7814
18.5k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
7815
18.5k
    xmlFree(ctxt->inputTab);
7816
18.5k
    xmlFree(ctxt->encoding);
7817
7818
18.5k
    ctxt->input = oldinput;
7819
18.5k
    ctxt->inputNr = oldinputNr;
7820
18.5k
    ctxt->inputMax = oldinputMax;
7821
18.5k
    ctxt->inputTab = oldinputTab;
7822
18.5k
    ctxt->encoding = oldencoding;
7823
7824
18.5k
    xmlFree(content);
7825
7826
18.5k
    return(ret);
7827
145
}
7828
7829
/**
7830
 * Parse PEReference declarations
7831
 *
7832
 *     [69] PEReference ::= '%' Name ';'
7833
 *
7834
 * [ WFC: No Recursion ]
7835
 * A parsed entity must not contain a recursive
7836
 * reference to itself, either directly or indirectly.
7837
 *
7838
 * [ WFC: Entity Declared ]
7839
 * In a document without any DTD, a document with only an internal DTD
7840
 * subset which contains no parameter entity references, or a document
7841
 * with "standalone='yes'", ...  ... The declaration of a parameter
7842
 * entity must precede any reference to it...
7843
 *
7844
 * [ VC: Entity Declared ]
7845
 * In a document with an external subset or external parameter entities
7846
 * with "standalone='no'", ...  ... The declaration of a parameter entity
7847
 * must precede any reference to it...
7848
 *
7849
 * [ WFC: In DTD ]
7850
 * Parameter-entity references may only appear in the DTD.
7851
 * NOTE: misleading but this is handled.
7852
 *
7853
 * @param ctxt  an XML parser context
7854
 * @param str  a pointer to an index in the string
7855
 * @returns the string of the entity content.
7856
 *         str is updated to the current value of the index
7857
 */
7858
static xmlEntityPtr
7859
44.3k
xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) {
7860
44.3k
    const xmlChar *ptr;
7861
44.3k
    xmlChar cur;
7862
44.3k
    xmlChar *name;
7863
44.3k
    xmlEntityPtr entity = NULL;
7864
7865
44.3k
    if ((str == NULL) || (*str == NULL)) return(NULL);
7866
44.3k
    ptr = *str;
7867
44.3k
    cur = *ptr;
7868
44.3k
    if (cur != '%')
7869
0
        return(NULL);
7870
44.3k
    ptr++;
7871
44.3k
    name = xmlParseStringName(ctxt, &ptr);
7872
44.3k
    if (name == NULL) {
7873
4.02k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7874
4.02k
           "xmlParseStringPEReference: no name\n");
7875
4.02k
  *str = ptr;
7876
4.02k
  return(NULL);
7877
4.02k
    }
7878
40.3k
    cur = *ptr;
7879
40.3k
    if (cur != ';') {
7880
4.26k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7881
4.26k
  xmlFree(name);
7882
4.26k
  *str = ptr;
7883
4.26k
  return(NULL);
7884
4.26k
    }
7885
36.0k
    ptr++;
7886
7887
    /* Must be set before xmlHandleUndeclaredEntity */
7888
36.0k
    ctxt->hasPErefs = 1;
7889
7890
    /*
7891
     * Request the entity from SAX
7892
     */
7893
36.0k
    if ((ctxt->sax != NULL) &&
7894
36.0k
  (ctxt->sax->getParameterEntity != NULL))
7895
36.0k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7896
7897
36.0k
    if (entity == NULL) {
7898
2.44k
        xmlHandleUndeclaredEntity(ctxt, name);
7899
33.5k
    } else {
7900
  /*
7901
   * Internal checking in case the entity quest barfed
7902
   */
7903
33.5k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7904
19.4k
      (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) {
7905
0
      xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7906
0
        "%%%s; is not a parameter entity\n",
7907
0
        name, NULL);
7908
0
  }
7909
33.5k
    }
7910
7911
36.0k
    xmlFree(name);
7912
36.0k
    *str = ptr;
7913
36.0k
    return(entity);
7914
40.3k
}
7915
7916
/**
7917
 * Parse a DOCTYPE declaration
7918
 *
7919
 * @deprecated Internal function, don't use.
7920
 *
7921
 *     [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S?
7922
 *                          ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
7923
 *
7924
 * [ VC: Root Element Type ]
7925
 * The Name in the document type declaration must match the element
7926
 * type of the root element.
7927
 *
7928
 * @param ctxt  an XML parser context
7929
 */
7930
7931
void
7932
27.5k
xmlParseDocTypeDecl(xmlParserCtxt *ctxt) {
7933
27.5k
    const xmlChar *name = NULL;
7934
27.5k
    xmlChar *publicId = NULL;
7935
27.5k
    xmlChar *URI = NULL;
7936
7937
    /*
7938
     * We know that '<!DOCTYPE' has been detected.
7939
     */
7940
27.5k
    SKIP(9);
7941
7942
27.5k
    if (SKIP_BLANKS == 0) {
7943
8.38k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7944
8.38k
                       "Space required after 'DOCTYPE'\n");
7945
8.38k
    }
7946
7947
    /*
7948
     * Parse the DOCTYPE name.
7949
     */
7950
27.5k
    name = xmlParseName(ctxt);
7951
27.5k
    if (name == NULL) {
7952
6.84k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7953
6.84k
           "xmlParseDocTypeDecl : no DOCTYPE name !\n");
7954
6.84k
    }
7955
27.5k
    ctxt->intSubName = name;
7956
7957
27.5k
    SKIP_BLANKS;
7958
7959
    /*
7960
     * Check for public and system identifier (URI)
7961
     */
7962
27.5k
    URI = xmlParseExternalID(ctxt, &publicId, 1);
7963
7964
27.5k
    if ((URI != NULL) || (publicId != NULL)) {
7965
5.32k
        ctxt->hasExternalSubset = 1;
7966
5.32k
    }
7967
27.5k
    ctxt->extSubURI = URI;
7968
27.5k
    ctxt->extSubSystem = publicId;
7969
7970
27.5k
    SKIP_BLANKS;
7971
7972
    /*
7973
     * Create and update the internal subset.
7974
     */
7975
27.5k
    if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) &&
7976
27.5k
  (!ctxt->disableSAX))
7977
25.3k
  ctxt->sax->internalSubset(ctxt->userData, name, publicId, URI);
7978
7979
27.5k
    if ((RAW != '[') && (RAW != '>')) {
7980
1.23k
  xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
7981
1.23k
    }
7982
27.5k
}
7983
7984
/**
7985
 * Parse the internal subset declaration
7986
 *
7987
 *     [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>'
7988
 * @param ctxt  an XML parser context
7989
 */
7990
7991
static void
7992
21.2k
xmlParseInternalSubset(xmlParserCtxtPtr ctxt) {
7993
    /*
7994
     * Is there any DTD definition ?
7995
     */
7996
21.2k
    if (RAW == '[') {
7997
21.2k
        int oldInputNr = ctxt->inputNr;
7998
7999
21.2k
        NEXT;
8000
  /*
8001
   * Parse the succession of Markup declarations and
8002
   * PEReferences.
8003
   * Subsequence (markupdecl | PEReference | S)*
8004
   */
8005
21.2k
  SKIP_BLANKS;
8006
418k
        while (1) {
8007
418k
            if (PARSER_STOPPED(ctxt)) {
8008
2.73k
                return;
8009
415k
            } else if (ctxt->input->cur >= ctxt->input->end) {
8010
65.6k
                if (ctxt->inputNr <= oldInputNr) {
8011
3.46k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8012
3.46k
                    return;
8013
3.46k
                }
8014
62.2k
                xmlPopPE(ctxt);
8015
349k
            } else if ((RAW == ']') && (ctxt->inputNr <= oldInputNr)) {
8016
8.53k
                NEXT;
8017
8.53k
                SKIP_BLANKS;
8018
8.53k
                break;
8019
341k
            } else if ((PARSER_EXTERNAL(ctxt)) &&
8020
223k
                       (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
8021
                /*
8022
                 * Conditional sections are allowed in external entities
8023
                 * included by PE References in the internal subset.
8024
                 */
8025
3.12k
                xmlParseConditionalSections(ctxt);
8026
338k
            } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
8027
194k
                xmlParseMarkupDecl(ctxt);
8028
194k
            } else if (RAW == '%') {
8029
137k
                xmlParsePERefInternal(ctxt, 1);
8030
137k
            } else {
8031
6.49k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8032
8033
7.21k
                while (ctxt->inputNr > oldInputNr)
8034
718
                    xmlPopPE(ctxt);
8035
6.49k
                return;
8036
6.49k
            }
8037
397k
            SKIP_BLANKS;
8038
397k
            SHRINK;
8039
397k
            GROW;
8040
397k
        }
8041
21.2k
    }
8042
8043
    /*
8044
     * We should be at the end of the DOCTYPE declaration.
8045
     */
8046
8.53k
    if (RAW != '>') {
8047
335
        xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8048
335
        return;
8049
335
    }
8050
8.20k
    NEXT;
8051
8.20k
}
8052
8053
#ifdef LIBXML_SAX1_ENABLED
8054
/**
8055
 * Parse an attribute
8056
 *
8057
 * @deprecated Internal function, don't use.
8058
 *
8059
 *     [41] Attribute ::= Name Eq AttValue
8060
 *
8061
 * [ WFC: No External Entity References ]
8062
 * Attribute values cannot contain direct or indirect entity references
8063
 * to external entities.
8064
 *
8065
 * [ WFC: No < in Attribute Values ]
8066
 * The replacement text of any entity referred to directly or indirectly in
8067
 * an attribute value (other than "&lt;") must not contain a <.
8068
 *
8069
 * [ VC: Attribute Value Type ]
8070
 * The attribute must have been declared; the value must be of the type
8071
 * declared for it.
8072
 *
8073
 *     [25] Eq ::= S? '=' S?
8074
 *
8075
 * With namespace:
8076
 *
8077
 *     [NS 11] Attribute ::= QName Eq AttValue
8078
 *
8079
 * Also the case QName == xmlns:??? is handled independently as a namespace
8080
 * definition.
8081
 *
8082
 * @param ctxt  an XML parser context
8083
 * @param value  a xmlChar ** used to store the value of the attribute
8084
 * @returns the attribute name, and the value in *value.
8085
 */
8086
8087
const xmlChar *
8088
0
xmlParseAttribute(xmlParserCtxt *ctxt, xmlChar **value) {
8089
0
    const xmlChar *name;
8090
0
    xmlChar *val;
8091
8092
0
    *value = NULL;
8093
0
    GROW;
8094
0
    name = xmlParseName(ctxt);
8095
0
    if (name == NULL) {
8096
0
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8097
0
                 "error parsing attribute name\n");
8098
0
        return(NULL);
8099
0
    }
8100
8101
    /*
8102
     * read the value
8103
     */
8104
0
    SKIP_BLANKS;
8105
0
    if (RAW == '=') {
8106
0
        NEXT;
8107
0
  SKIP_BLANKS;
8108
0
  val = xmlParseAttValue(ctxt);
8109
0
    } else {
8110
0
  xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8111
0
         "Specification mandates value for attribute %s\n", name);
8112
0
  return(name);
8113
0
    }
8114
8115
    /*
8116
     * Check that xml:lang conforms to the specification
8117
     * No more registered as an error, just generate a warning now
8118
     * since this was deprecated in XML second edition
8119
     */
8120
0
    if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) {
8121
0
  if (!xmlCheckLanguageID(val)) {
8122
0
      xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8123
0
              "Malformed value for xml:lang : %s\n",
8124
0
        val, NULL);
8125
0
  }
8126
0
    }
8127
8128
    /*
8129
     * Check that xml:space conforms to the specification
8130
     */
8131
0
    if (xmlStrEqual(name, BAD_CAST "xml:space")) {
8132
0
  if (xmlStrEqual(val, BAD_CAST "default"))
8133
0
      *(ctxt->space) = 0;
8134
0
  else if (xmlStrEqual(val, BAD_CAST "preserve"))
8135
0
      *(ctxt->space) = 1;
8136
0
  else {
8137
0
    xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8138
0
"Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8139
0
                                 val, NULL);
8140
0
  }
8141
0
    }
8142
8143
0
    *value = val;
8144
0
    return(name);
8145
0
}
8146
8147
/**
8148
 * Parse a start tag. Always consumes '<'.
8149
 *
8150
 * @deprecated Internal function, don't use.
8151
 *
8152
 *     [40] STag ::= '<' Name (S Attribute)* S? '>'
8153
 *
8154
 * [ WFC: Unique Att Spec ]
8155
 * No attribute name may appear more than once in the same start-tag or
8156
 * empty-element tag.
8157
 *
8158
 *     [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
8159
 *
8160
 * [ WFC: Unique Att Spec ]
8161
 * No attribute name may appear more than once in the same start-tag or
8162
 * empty-element tag.
8163
 *
8164
 * With namespace:
8165
 *
8166
 *     [NS 8] STag ::= '<' QName (S Attribute)* S? '>'
8167
 *
8168
 *     [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>'
8169
 *
8170
 * @param ctxt  an XML parser context
8171
 * @returns the element name parsed
8172
 */
8173
8174
const xmlChar *
8175
0
xmlParseStartTag(xmlParserCtxt *ctxt) {
8176
0
    const xmlChar *name;
8177
0
    const xmlChar *attname;
8178
0
    xmlChar *attvalue;
8179
0
    const xmlChar **atts = ctxt->atts;
8180
0
    int nbatts = 0;
8181
0
    int maxatts = ctxt->maxatts;
8182
0
    int i;
8183
8184
0
    if (RAW != '<') return(NULL);
8185
0
    NEXT1;
8186
8187
0
    name = xmlParseName(ctxt);
8188
0
    if (name == NULL) {
8189
0
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8190
0
       "xmlParseStartTag: invalid element name\n");
8191
0
        return(NULL);
8192
0
    }
8193
8194
    /*
8195
     * Now parse the attributes, it ends up with the ending
8196
     *
8197
     * (S Attribute)* S?
8198
     */
8199
0
    SKIP_BLANKS;
8200
0
    GROW;
8201
8202
0
    while (((RAW != '>') &&
8203
0
     ((RAW != '/') || (NXT(1) != '>')) &&
8204
0
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8205
0
  attname = xmlParseAttribute(ctxt, &attvalue);
8206
0
        if (attname == NULL)
8207
0
      break;
8208
0
        if (attvalue != NULL) {
8209
      /*
8210
       * [ WFC: Unique Att Spec ]
8211
       * No attribute name may appear more than once in the same
8212
       * start-tag or empty-element tag.
8213
       */
8214
0
      for (i = 0; i < nbatts;i += 2) {
8215
0
          if (xmlStrEqual(atts[i], attname)) {
8216
0
        xmlErrAttributeDup(ctxt, NULL, attname);
8217
0
        goto failed;
8218
0
    }
8219
0
      }
8220
      /*
8221
       * Add the pair to atts
8222
       */
8223
0
      if (nbatts + 4 > maxatts) {
8224
0
          const xmlChar **n;
8225
0
                int newSize;
8226
8227
0
                newSize = xmlGrowCapacity(maxatts, sizeof(n[0]) * 2,
8228
0
                                          11, XML_MAX_ATTRS);
8229
0
                if (newSize < 0) {
8230
0
        xmlErrMemory(ctxt);
8231
0
        goto failed;
8232
0
    }
8233
0
#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
8234
0
                if (newSize < 2)
8235
0
                    newSize = 2;
8236
0
#endif
8237
0
          n = xmlRealloc(atts, newSize * sizeof(n[0]) * 2);
8238
0
    if (n == NULL) {
8239
0
        xmlErrMemory(ctxt);
8240
0
        goto failed;
8241
0
    }
8242
0
    atts = n;
8243
0
                maxatts = newSize * 2;
8244
0
    ctxt->atts = atts;
8245
0
    ctxt->maxatts = maxatts;
8246
0
      }
8247
8248
0
      atts[nbatts++] = attname;
8249
0
      atts[nbatts++] = attvalue;
8250
0
      atts[nbatts] = NULL;
8251
0
      atts[nbatts + 1] = NULL;
8252
8253
0
            attvalue = NULL;
8254
0
  }
8255
8256
0
failed:
8257
8258
0
        if (attvalue != NULL)
8259
0
            xmlFree(attvalue);
8260
8261
0
  GROW
8262
0
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
8263
0
      break;
8264
0
  if (SKIP_BLANKS == 0) {
8265
0
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
8266
0
         "attributes construct error\n");
8267
0
  }
8268
0
  SHRINK;
8269
0
        GROW;
8270
0
    }
8271
8272
    /*
8273
     * SAX: Start of Element !
8274
     */
8275
0
    if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) &&
8276
0
  (!ctxt->disableSAX)) {
8277
0
  if (nbatts > 0)
8278
0
      ctxt->sax->startElement(ctxt->userData, name, atts);
8279
0
  else
8280
0
      ctxt->sax->startElement(ctxt->userData, name, NULL);
8281
0
    }
8282
8283
0
    if (atts != NULL) {
8284
        /* Free only the content strings */
8285
0
        for (i = 1;i < nbatts;i+=2)
8286
0
      if (atts[i] != NULL)
8287
0
         xmlFree((xmlChar *) atts[i]);
8288
0
    }
8289
0
    return(name);
8290
0
}
8291
8292
/**
8293
 * Parse an end tag. Always consumes '</'.
8294
 *
8295
 *     [42] ETag ::= '</' Name S? '>'
8296
 *
8297
 * With namespace
8298
 *
8299
 *     [NS 9] ETag ::= '</' QName S? '>'
8300
 * @param ctxt  an XML parser context
8301
 * @param line  line of the start tag
8302
 */
8303
8304
static void
8305
0
xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) {
8306
0
    const xmlChar *name;
8307
8308
0
    GROW;
8309
0
    if ((RAW != '<') || (NXT(1) != '/')) {
8310
0
  xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED,
8311
0
           "xmlParseEndTag: '</' not found\n");
8312
0
  return;
8313
0
    }
8314
0
    SKIP(2);
8315
8316
0
    name = xmlParseNameAndCompare(ctxt,ctxt->name);
8317
8318
    /*
8319
     * We should definitely be at the ending "S? '>'" part
8320
     */
8321
0
    GROW;
8322
0
    SKIP_BLANKS;
8323
0
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
8324
0
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
8325
0
    } else
8326
0
  NEXT1;
8327
8328
    /*
8329
     * [ WFC: Element Type Match ]
8330
     * The Name in an element's end-tag must match the element type in the
8331
     * start-tag.
8332
     *
8333
     */
8334
0
    if (name != (xmlChar*)1) {
8335
0
        if (name == NULL) name = BAD_CAST "unparsable";
8336
0
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
8337
0
         "Opening and ending tag mismatch: %s line %d and %s\n",
8338
0
                    ctxt->name, line, name);
8339
0
    }
8340
8341
    /*
8342
     * SAX: End of Tag
8343
     */
8344
0
    if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
8345
0
  (!ctxt->disableSAX))
8346
0
        ctxt->sax->endElement(ctxt->userData, ctxt->name);
8347
8348
0
    namePop(ctxt);
8349
0
    spacePop(ctxt);
8350
0
}
8351
8352
/**
8353
 * Parse an end of tag
8354
 *
8355
 * @deprecated Internal function, don't use.
8356
 *
8357
 *     [42] ETag ::= '</' Name S? '>'
8358
 *
8359
 * With namespace
8360
 *
8361
 *     [NS 9] ETag ::= '</' QName S? '>'
8362
 * @param ctxt  an XML parser context
8363
 */
8364
8365
void
8366
0
xmlParseEndTag(xmlParserCtxt *ctxt) {
8367
0
    xmlParseEndTag1(ctxt, 0);
8368
0
}
8369
#endif /* LIBXML_SAX1_ENABLED */
8370
8371
/************************************************************************
8372
 *                  *
8373
 *          SAX 2 specific operations       *
8374
 *                  *
8375
 ************************************************************************/
8376
8377
/**
8378
 * Parse an XML Namespace QName
8379
 *
8380
 *     [6]  QName  ::= (Prefix ':')? LocalPart
8381
 *     [7]  Prefix  ::= NCName
8382
 *     [8]  LocalPart  ::= NCName
8383
 *
8384
 * @param ctxt  an XML parser context
8385
 * @param prefix  pointer to store the prefix part
8386
 * @returns the Name parsed or NULL
8387
 */
8388
8389
static xmlHashedString
8390
1.04M
xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) {
8391
1.04M
    xmlHashedString l, p;
8392
1.04M
    int start, isNCName = 0;
8393
8394
1.04M
    l.name = NULL;
8395
1.04M
    p.name = NULL;
8396
8397
1.04M
    GROW;
8398
1.04M
    start = CUR_PTR - BASE_PTR;
8399
8400
1.04M
    l = xmlParseNCName(ctxt);
8401
1.04M
    if (l.name != NULL) {
8402
887k
        isNCName = 1;
8403
887k
        if (CUR == ':') {
8404
142k
            NEXT;
8405
142k
            p = l;
8406
142k
            l = xmlParseNCName(ctxt);
8407
142k
        }
8408
887k
    }
8409
1.04M
    if ((l.name == NULL) || (CUR == ':')) {
8410
163k
        xmlChar *tmp;
8411
8412
163k
        l.name = NULL;
8413
163k
        p.name = NULL;
8414
163k
        if ((isNCName == 0) && (CUR != ':'))
8415
148k
            return(l);
8416
14.7k
        tmp = xmlParseNmtoken(ctxt);
8417
14.7k
        if (tmp != NULL)
8418
8.86k
            xmlFree(tmp);
8419
14.7k
        l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start,
8420
14.7k
                                CUR_PTR - (BASE_PTR + start));
8421
14.7k
        if (l.name == NULL) {
8422
3
            xmlErrMemory(ctxt);
8423
3
            return(l);
8424
3
        }
8425
14.7k
        xmlNsErr(ctxt, XML_NS_ERR_QNAME,
8426
14.7k
                 "Failed to parse QName '%s'\n", l.name, NULL, NULL);
8427
14.7k
    }
8428
8429
891k
    *prefix = p;
8430
891k
    return(l);
8431
1.04M
}
8432
8433
/**
8434
 * Parse an XML Namespace QName
8435
 *
8436
 *     [6]  QName  ::= (Prefix ':')? LocalPart
8437
 *     [7]  Prefix  ::= NCName
8438
 *     [8]  LocalPart  ::= NCName
8439
 *
8440
 * @param ctxt  an XML parser context
8441
 * @param prefix  pointer to store the prefix part
8442
 * @returns the Name parsed or NULL
8443
 */
8444
8445
static const xmlChar *
8446
10.1k
xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) {
8447
10.1k
    xmlHashedString n, p;
8448
8449
10.1k
    n = xmlParseQNameHashed(ctxt, &p);
8450
10.1k
    if (n.name == NULL)
8451
1.70k
        return(NULL);
8452
8.41k
    *prefix = p.name;
8453
8.41k
    return(n.name);
8454
10.1k
}
8455
8456
/**
8457
 * Parse an XML name and compares for match
8458
 * (specialized for endtag parsing)
8459
 *
8460
 * @param ctxt  an XML parser context
8461
 * @param name  the localname
8462
 * @param prefix  the prefix, if any.
8463
 * @returns NULL for an illegal name, (xmlChar*) 1 for success
8464
 * and the name for mismatch
8465
 */
8466
8467
static const xmlChar *
8468
xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name,
8469
13.5k
                        xmlChar const *prefix) {
8470
13.5k
    const xmlChar *cmp;
8471
13.5k
    const xmlChar *in;
8472
13.5k
    const xmlChar *ret;
8473
13.5k
    const xmlChar *prefix2;
8474
8475
13.5k
    if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name));
8476
8477
13.5k
    GROW;
8478
13.5k
    in = ctxt->input->cur;
8479
8480
13.5k
    cmp = prefix;
8481
31.2k
    while (*in != 0 && *in == *cmp) {
8482
17.7k
  ++in;
8483
17.7k
  ++cmp;
8484
17.7k
    }
8485
13.5k
    if ((*cmp == 0) && (*in == ':')) {
8486
6.22k
        in++;
8487
6.22k
  cmp = name;
8488
30.4k
  while (*in != 0 && *in == *cmp) {
8489
24.1k
      ++in;
8490
24.1k
      ++cmp;
8491
24.1k
  }
8492
6.22k
  if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
8493
      /* success */
8494
3.40k
            ctxt->input->col += in - ctxt->input->cur;
8495
3.40k
      ctxt->input->cur = in;
8496
3.40k
      return((const xmlChar*) 1);
8497
3.40k
  }
8498
6.22k
    }
8499
    /*
8500
     * all strings coms from the dictionary, equality can be done directly
8501
     */
8502
10.1k
    ret = xmlParseQName (ctxt, &prefix2);
8503
10.1k
    if (ret == NULL)
8504
1.70k
        return(NULL);
8505
8.41k
    if ((ret == name) && (prefix == prefix2))
8506
789
  return((const xmlChar*) 1);
8507
7.62k
    return ret;
8508
8.41k
}
8509
8510
/**
8511
 * Parse an attribute in the new SAX2 framework.
8512
 *
8513
 * @param ctxt  an XML parser context
8514
 * @param pref  the element prefix
8515
 * @param elem  the element name
8516
 * @param hprefix  resulting attribute prefix
8517
 * @param value  resulting value of the attribute
8518
 * @param len  resulting length of the attribute
8519
 * @param alloc  resulting indicator if the attribute was allocated
8520
 * @returns the attribute name, and the value in *value, .
8521
 */
8522
8523
static xmlHashedString
8524
xmlParseAttribute2(xmlParserCtxtPtr ctxt,
8525
                   const xmlChar * pref, const xmlChar * elem,
8526
                   xmlHashedString * hprefix, xmlChar ** value,
8527
                   int *len, int *alloc)
8528
391k
{
8529
391k
    xmlHashedString hname;
8530
391k
    const xmlChar *prefix, *name;
8531
391k
    xmlChar *val = NULL, *internal_val = NULL;
8532
391k
    int special = 0;
8533
391k
    int isNamespace;
8534
391k
    int flags;
8535
8536
391k
    *value = NULL;
8537
391k
    GROW;
8538
391k
    hname = xmlParseQNameHashed(ctxt, hprefix);
8539
391k
    if (hname.name == NULL) {
8540
82.3k
        xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8541
82.3k
                       "error parsing attribute name\n");
8542
82.3k
        return(hname);
8543
82.3k
    }
8544
309k
    name = hname.name;
8545
309k
    prefix = hprefix->name;
8546
8547
    /*
8548
     * get the type if needed
8549
     */
8550
309k
    if (ctxt->attsSpecial != NULL) {
8551
86.8k
        special = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem,
8552
86.8k
                                              prefix, name));
8553
86.8k
    }
8554
8555
    /*
8556
     * read the value
8557
     */
8558
309k
    SKIP_BLANKS;
8559
309k
    if (RAW != '=') {
8560
32.9k
        xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8561
32.9k
                          "Specification mandates value for attribute %s\n",
8562
32.9k
                          name);
8563
32.9k
        goto error;
8564
32.9k
    }
8565
8566
8567
276k
    NEXT;
8568
276k
    SKIP_BLANKS;
8569
276k
    flags = 0;
8570
276k
    isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) ||
8571
233k
                   (prefix == ctxt->str_xmlns));
8572
276k
    val = xmlParseAttValueInternal(ctxt, len, &flags, special,
8573
276k
                                   isNamespace);
8574
276k
    if (val == NULL)
8575
12.3k
        goto error;
8576
8577
263k
    *alloc = (flags & XML_ATTVAL_ALLOC) != 0;
8578
8579
263k
#ifdef LIBXML_VALID_ENABLED
8580
263k
    if ((ctxt->validate) &&
8581
0
        (ctxt->standalone == 1) &&
8582
0
        (special & XML_SPECIAL_EXTERNAL) &&
8583
0
        (flags & XML_ATTVAL_NORM_CHANGE)) {
8584
0
        xmlValidityError(ctxt, XML_DTD_NOT_STANDALONE,
8585
0
                         "standalone: normalization of attribute %s on %s "
8586
0
                         "by external subset declaration\n",
8587
0
                         name, elem);
8588
0
    }
8589
263k
#endif
8590
8591
263k
    if (prefix == ctxt->str_xml) {
8592
        /*
8593
         * Check that xml:lang conforms to the specification
8594
         * No more registered as an error, just generate a warning now
8595
         * since this was deprecated in XML second edition
8596
         */
8597
21.7k
        if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) {
8598
12.7k
            internal_val = xmlStrndup(val, *len);
8599
12.7k
            if (internal_val == NULL)
8600
5
                goto mem_error;
8601
12.7k
            if (!xmlCheckLanguageID(internal_val)) {
8602
10.2k
                xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8603
10.2k
                              "Malformed value for xml:lang : %s\n",
8604
10.2k
                              internal_val, NULL);
8605
10.2k
            }
8606
12.7k
        }
8607
8608
        /*
8609
         * Check that xml:space conforms to the specification
8610
         */
8611
21.7k
        if (xmlStrEqual(name, BAD_CAST "space")) {
8612
681
            internal_val = xmlStrndup(val, *len);
8613
681
            if (internal_val == NULL)
8614
3
                goto mem_error;
8615
678
            if (xmlStrEqual(internal_val, BAD_CAST "default"))
8616
256
                *(ctxt->space) = 0;
8617
422
            else if (xmlStrEqual(internal_val, BAD_CAST "preserve"))
8618
212
                *(ctxt->space) = 1;
8619
210
            else {
8620
210
                xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8621
210
                              "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8622
210
                              internal_val, NULL);
8623
210
            }
8624
678
        }
8625
21.6k
        if (internal_val) {
8626
13.3k
            xmlFree(internal_val);
8627
13.3k
        }
8628
21.6k
    }
8629
8630
263k
    *value = val;
8631
263k
    return (hname);
8632
8633
8
mem_error:
8634
8
    xmlErrMemory(ctxt);
8635
45.2k
error:
8636
45.2k
    if ((val != NULL) && (*alloc != 0))
8637
3
        xmlFree(val);
8638
45.2k
    return(hname);
8639
8
}
8640
8641
/**
8642
 * Inserts a new attribute into the hash table.
8643
 *
8644
 * @param ctxt  parser context
8645
 * @param size  size of the hash table
8646
 * @param name  attribute name
8647
 * @param uri  namespace uri
8648
 * @param hashValue  combined hash value of name and uri
8649
 * @param aindex  attribute index (this is a multiple of 5)
8650
 * @returns INT_MAX if no existing attribute was found, the attribute
8651
 * index if an attribute was found, -1 if a memory allocation failed.
8652
 */
8653
static int
8654
xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name,
8655
184k
                  const xmlChar *uri, unsigned hashValue, int aindex) {
8656
184k
    xmlAttrHashBucket *table = ctxt->attrHash;
8657
184k
    xmlAttrHashBucket *bucket;
8658
184k
    unsigned hindex;
8659
8660
184k
    hindex = hashValue & (size - 1);
8661
184k
    bucket = &table[hindex];
8662
8663
220k
    while (bucket->index >= 0) {
8664
64.3k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8665
8666
64.3k
        if (name == atts[0]) {
8667
32.2k
            int nsIndex = XML_PTR_TO_INT(atts[2]);
8668
8669
32.2k
            if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) :
8670
32.2k
                (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) :
8671
10.1k
                (uri == ctxt->nsTab[nsIndex * 2 + 1]))
8672
28.3k
                return(bucket->index);
8673
32.2k
        }
8674
8675
36.0k
        hindex++;
8676
36.0k
        bucket++;
8677
36.0k
        if (hindex >= size) {
8678
2.85k
            hindex = 0;
8679
2.85k
            bucket = table;
8680
2.85k
        }
8681
36.0k
    }
8682
8683
156k
    bucket->index = aindex;
8684
8685
156k
    return(INT_MAX);
8686
184k
}
8687
8688
static int
8689
xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size,
8690
                       const xmlChar *name, const xmlChar *prefix,
8691
3.80k
                       unsigned hashValue, int aindex) {
8692
3.80k
    xmlAttrHashBucket *table = ctxt->attrHash;
8693
3.80k
    xmlAttrHashBucket *bucket;
8694
3.80k
    unsigned hindex;
8695
8696
3.80k
    hindex = hashValue & (size - 1);
8697
3.80k
    bucket = &table[hindex];
8698
8699
5.91k
    while (bucket->index >= 0) {
8700
3.23k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8701
8702
3.23k
        if ((name == atts[0]) && (prefix == atts[1]))
8703
1.12k
            return(bucket->index);
8704
8705
2.10k
        hindex++;
8706
2.10k
        bucket++;
8707
2.10k
        if (hindex >= size) {
8708
196
            hindex = 0;
8709
196
            bucket = table;
8710
196
        }
8711
2.10k
    }
8712
8713
2.68k
    bucket->index = aindex;
8714
8715
2.68k
    return(INT_MAX);
8716
3.80k
}
8717
/**
8718
 * Parse a start tag. Always consumes '<'.
8719
 *
8720
 * This routine is called when running SAX2 parsing
8721
 *
8722
 *     [40] STag ::= '<' Name (S Attribute)* S? '>'
8723
 *
8724
 * [ WFC: Unique Att Spec ]
8725
 * No attribute name may appear more than once in the same start-tag or
8726
 * empty-element tag.
8727
 *
8728
 *     [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>'
8729
 *
8730
 * [ WFC: Unique Att Spec ]
8731
 * No attribute name may appear more than once in the same start-tag or
8732
 * empty-element tag.
8733
 *
8734
 * With namespace:
8735
 *
8736
 *     [NS 8] STag ::= '<' QName (S Attribute)* S? '>'
8737
 *
8738
 *     [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>'
8739
 *
8740
 * @param ctxt  an XML parser context
8741
 * @param pref  resulting namespace prefix
8742
 * @param URI  resulting namespace URI
8743
 * @param nbNsPtr  resulting number of namespace declarations
8744
 * @returns the element name parsed
8745
 */
8746
8747
static const xmlChar *
8748
xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref,
8749
638k
                  const xmlChar **URI, int *nbNsPtr) {
8750
638k
    xmlHashedString hlocalname;
8751
638k
    xmlHashedString hprefix;
8752
638k
    xmlHashedString hattname;
8753
638k
    xmlHashedString haprefix;
8754
638k
    const xmlChar *localname;
8755
638k
    const xmlChar *prefix;
8756
638k
    const xmlChar *attname;
8757
638k
    const xmlChar *aprefix;
8758
638k
    const xmlChar *uri;
8759
638k
    xmlChar *attvalue = NULL;
8760
638k
    const xmlChar **atts = ctxt->atts;
8761
638k
    unsigned attrHashSize = 0;
8762
638k
    int maxatts = ctxt->maxatts;
8763
638k
    int nratts, nbatts, nbdef;
8764
638k
    int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts;
8765
638k
    int alloc = 0;
8766
638k
    int numNsErr = 0;
8767
638k
    int numDupErr = 0;
8768
8769
638k
    if (RAW != '<') return(NULL);
8770
638k
    NEXT1;
8771
8772
638k
    nbatts = 0;
8773
638k
    nratts = 0;
8774
638k
    nbdef = 0;
8775
638k
    nbNs = 0;
8776
638k
    nbTotalDef = 0;
8777
638k
    attval = 0;
8778
8779
638k
    if (xmlParserNsStartElement(ctxt->nsdb) < 0) {
8780
0
        xmlErrMemory(ctxt);
8781
0
        return(NULL);
8782
0
    }
8783
8784
638k
    hlocalname = xmlParseQNameHashed(ctxt, &hprefix);
8785
638k
    if (hlocalname.name == NULL) {
8786
64.9k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8787
64.9k
           "StartTag: invalid element name\n");
8788
64.9k
        return(NULL);
8789
64.9k
    }
8790
573k
    localname = hlocalname.name;
8791
573k
    prefix = hprefix.name;
8792
8793
    /*
8794
     * Now parse the attributes, it ends up with the ending
8795
     *
8796
     * (S Attribute)* S?
8797
     */
8798
573k
    SKIP_BLANKS;
8799
573k
    GROW;
8800
8801
    /*
8802
     * The ctxt->atts array will be ultimately passed to the SAX callback
8803
     * containing five xmlChar pointers for each attribute:
8804
     *
8805
     * [0] attribute name
8806
     * [1] attribute prefix
8807
     * [2] namespace URI
8808
     * [3] attribute value
8809
     * [4] end of attribute value
8810
     *
8811
     * To save memory, we reuse this array temporarily and store integers
8812
     * in these pointer variables.
8813
     *
8814
     * [0] attribute name
8815
     * [1] attribute prefix
8816
     * [2] hash value of attribute prefix, and later namespace index
8817
     * [3] for non-allocated values: ptrdiff_t offset into input buffer
8818
     * [4] for non-allocated values: ptrdiff_t offset into input buffer
8819
     *
8820
     * The ctxt->attallocs array contains an additional unsigned int for
8821
     * each attribute, containing the hash value of the attribute name
8822
     * and the alloc flag in bit 31.
8823
     */
8824
8825
714k
    while (((RAW != '>') &&
8826
455k
     ((RAW != '/') || (NXT(1) != '>')) &&
8827
405k
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8828
391k
  int len = -1;
8829
8830
391k
  hattname = xmlParseAttribute2(ctxt, prefix, localname,
8831
391k
                                          &haprefix, &attvalue, &len,
8832
391k
                                          &alloc);
8833
391k
        if (hattname.name == NULL)
8834
82.3k
      break;
8835
309k
        if (attvalue == NULL)
8836
45.2k
            goto next_attr;
8837
263k
        attname = hattname.name;
8838
263k
        aprefix = haprefix.name;
8839
263k
  if (len < 0) len = xmlStrlen(attvalue);
8840
8841
263k
        if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
8842
41.4k
            xmlHashedString huri;
8843
41.4k
            xmlURIPtr parsedUri;
8844
8845
41.4k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8846
41.4k
            uri = huri.name;
8847
41.4k
            if (uri == NULL) {
8848
5
                xmlErrMemory(ctxt);
8849
5
                goto next_attr;
8850
5
            }
8851
41.4k
            if (*uri != 0) {
8852
40.4k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8853
71
                    xmlErrMemory(ctxt);
8854
71
                    goto next_attr;
8855
71
                }
8856
40.4k
                if (parsedUri == NULL) {
8857
25.3k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8858
25.3k
                             "xmlns: '%s' is not a valid URI\n",
8859
25.3k
                                       uri, NULL, NULL);
8860
25.3k
                } else {
8861
15.0k
                    if (parsedUri->scheme == NULL) {
8862
9.41k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8863
9.41k
                                  "xmlns: URI %s is not absolute\n",
8864
9.41k
                                  uri, NULL, NULL);
8865
9.41k
                    }
8866
15.0k
                    xmlFreeURI(parsedUri);
8867
15.0k
                }
8868
40.4k
                if (uri == ctxt->str_xml_ns) {
8869
199
                    if (attname != ctxt->str_xml) {
8870
199
                        xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8871
199
                     "xml namespace URI cannot be the default namespace\n",
8872
199
                                 NULL, NULL, NULL);
8873
199
                    }
8874
199
                    goto next_attr;
8875
199
                }
8876
40.2k
                if ((len == 29) &&
8877
850
                    (xmlStrEqual(uri,
8878
850
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8879
325
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8880
325
                         "reuse of the xmlns namespace name is forbidden\n",
8881
325
                             NULL, NULL, NULL);
8882
325
                    goto next_attr;
8883
325
                }
8884
40.2k
            }
8885
8886
40.8k
            if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0)
8887
30.8k
                nbNs++;
8888
222k
        } else if (aprefix == ctxt->str_xmlns) {
8889
19.2k
            xmlHashedString huri;
8890
19.2k
            xmlURIPtr parsedUri;
8891
8892
19.2k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8893
19.2k
            uri = huri.name;
8894
19.2k
            if (uri == NULL) {
8895
3
                xmlErrMemory(ctxt);
8896
3
                goto next_attr;
8897
3
            }
8898
8899
19.2k
            if (attname == ctxt->str_xml) {
8900
196
                if (uri != ctxt->str_xml_ns) {
8901
196
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8902
196
                             "xml namespace prefix mapped to wrong URI\n",
8903
196
                             NULL, NULL, NULL);
8904
196
                }
8905
                /*
8906
                 * Do not keep a namespace definition node
8907
                 */
8908
196
                goto next_attr;
8909
196
            }
8910
19.0k
            if (uri == ctxt->str_xml_ns) {
8911
194
                if (attname != ctxt->str_xml) {
8912
194
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8913
194
                             "xml namespace URI mapped to wrong prefix\n",
8914
194
                             NULL, NULL, NULL);
8915
194
                }
8916
194
                goto next_attr;
8917
194
            }
8918
18.8k
            if (attname == ctxt->str_xmlns) {
8919
196
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8920
196
                         "redefinition of the xmlns prefix is forbidden\n",
8921
196
                         NULL, NULL, NULL);
8922
196
                goto next_attr;
8923
196
            }
8924
18.6k
            if ((len == 29) &&
8925
1.03k
                (xmlStrEqual(uri,
8926
1.03k
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8927
194
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8928
194
                         "reuse of the xmlns namespace name is forbidden\n",
8929
194
                         NULL, NULL, NULL);
8930
194
                goto next_attr;
8931
194
            }
8932
18.4k
            if ((uri == NULL) || (uri[0] == 0)) {
8933
225
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8934
225
                         "xmlns:%s: Empty XML namespace is not allowed\n",
8935
225
                              attname, NULL, NULL);
8936
225
                goto next_attr;
8937
18.2k
            } else {
8938
18.2k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8939
8
                    xmlErrMemory(ctxt);
8940
8
                    goto next_attr;
8941
8
                }
8942
18.2k
                if (parsedUri == NULL) {
8943
7.04k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8944
7.04k
                         "xmlns:%s: '%s' is not a valid URI\n",
8945
7.04k
                                       attname, uri, NULL);
8946
11.1k
                } else {
8947
11.1k
                    if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) {
8948
1.34k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8949
1.34k
                                  "xmlns:%s: URI %s is not absolute\n",
8950
1.34k
                                  attname, uri, NULL);
8951
1.34k
                    }
8952
11.1k
                    xmlFreeURI(parsedUri);
8953
11.1k
                }
8954
18.2k
            }
8955
8956
18.2k
            if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0)
8957
15.8k
                nbNs++;
8958
203k
        } else {
8959
            /*
8960
             * Populate attributes array, see above for repurposing
8961
             * of xmlChar pointers.
8962
             */
8963
203k
            if ((atts == NULL) || (nbatts + 5 > maxatts)) {
8964
11.4k
                int res = xmlCtxtGrowAttrs(ctxt);
8965
8966
11.4k
                maxatts = ctxt->maxatts;
8967
11.4k
                atts = ctxt->atts;
8968
8969
11.4k
                if (res < 0)
8970
18
                    goto next_attr;
8971
11.4k
            }
8972
203k
            ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) |
8973
203k
                                        ((unsigned) alloc << 31);
8974
203k
            atts[nbatts++] = attname;
8975
203k
            atts[nbatts++] = aprefix;
8976
203k
            atts[nbatts++] = XML_INT_TO_PTR(haprefix.hashValue);
8977
203k
            if (alloc) {
8978
38.3k
                atts[nbatts++] = attvalue;
8979
38.3k
                attvalue += len;
8980
38.3k
                atts[nbatts++] = attvalue;
8981
164k
            } else {
8982
                /*
8983
                 * attvalue points into the input buffer which can be
8984
                 * reallocated. Store differences to input->base instead.
8985
                 * The pointers will be reconstructed later.
8986
                 */
8987
164k
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8988
164k
                attvalue += len;
8989
164k
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8990
164k
            }
8991
            /*
8992
             * tag if some deallocation is needed
8993
             */
8994
203k
            if (alloc != 0) attval = 1;
8995
203k
            attvalue = NULL; /* moved into atts */
8996
203k
        }
8997
8998
309k
next_attr:
8999
309k
        if ((attvalue != NULL) && (alloc != 0)) {
9000
21.5k
            xmlFree(attvalue);
9001
21.5k
            attvalue = NULL;
9002
21.5k
        }
9003
9004
309k
  GROW
9005
309k
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
9006
79.8k
      break;
9007
229k
  if (SKIP_BLANKS == 0) {
9008
88.8k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
9009
88.8k
         "attributes construct error\n");
9010
88.8k
      break;
9011
88.8k
  }
9012
140k
        GROW;
9013
140k
    }
9014
9015
    /*
9016
     * Namespaces from default attributes
9017
     */
9018
573k
    if (ctxt->attsDefault != NULL) {
9019
106k
        xmlDefAttrsPtr defaults;
9020
9021
106k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9022
106k
  if (defaults != NULL) {
9023
427k
      for (i = 0; i < defaults->nbAttrs; i++) {
9024
338k
                xmlDefAttr *attr = &defaults->attrs[i];
9025
9026
338k
          attname = attr->name.name;
9027
338k
    aprefix = attr->prefix.name;
9028
9029
338k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
9030
6.80k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9031
9032
6.80k
                    if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0)
9033
5.80k
                        nbNs++;
9034
332k
    } else if (aprefix == ctxt->str_xmlns) {
9035
183k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9036
9037
183k
                    if (xmlParserNsPush(ctxt, &attr->name, &attr->value,
9038
183k
                                      NULL, 1) > 0)
9039
174k
                        nbNs++;
9040
183k
    } else {
9041
148k
                    if (nratts + nbTotalDef >= XML_MAX_ATTRS) {
9042
0
                        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
9043
0
                                    "Maximum number of attributes exceeded");
9044
0
                        break;
9045
0
                    }
9046
148k
                    nbTotalDef += 1;
9047
148k
                }
9048
338k
      }
9049
88.9k
  }
9050
106k
    }
9051
9052
    /*
9053
     * Resolve attribute namespaces
9054
     */
9055
776k
    for (i = 0; i < nbatts; i += 5) {
9056
203k
        attname = atts[i];
9057
203k
        aprefix = atts[i+1];
9058
9059
        /*
9060
  * The default namespace does not apply to attribute names.
9061
  */
9062
203k
  if (aprefix == NULL) {
9063
142k
            nsIndex = NS_INDEX_EMPTY;
9064
142k
        } else if (aprefix == ctxt->str_xml) {
9065
21.6k
            nsIndex = NS_INDEX_XML;
9066
39.3k
        } else {
9067
39.3k
            haprefix.name = aprefix;
9068
39.3k
            haprefix.hashValue = (size_t) atts[i+2];
9069
39.3k
            nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL);
9070
9071
39.3k
      if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) {
9072
21.3k
                xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9073
21.3k
        "Namespace prefix %s for %s on %s is not defined\n",
9074
21.3k
        aprefix, attname, localname);
9075
21.3k
                nsIndex = NS_INDEX_EMPTY;
9076
21.3k
            }
9077
39.3k
        }
9078
9079
203k
        atts[i+2] = XML_INT_TO_PTR(nsIndex);
9080
203k
    }
9081
9082
    /*
9083
     * Maximum number of attributes including default attributes.
9084
     */
9085
573k
    maxAtts = nratts + nbTotalDef;
9086
9087
    /*
9088
     * Verify that attribute names are unique.
9089
     */
9090
573k
    if (maxAtts > 1) {
9091
40.5k
        attrHashSize = 4;
9092
58.1k
        while (attrHashSize / 2 < (unsigned) maxAtts)
9093
17.6k
            attrHashSize *= 2;
9094
9095
40.5k
        if (attrHashSize > ctxt->attrHashMax) {
9096
2.32k
            xmlAttrHashBucket *tmp;
9097
9098
2.32k
            tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0]));
9099
2.32k
            if (tmp == NULL) {
9100
3
                xmlErrMemory(ctxt);
9101
3
                goto done;
9102
3
            }
9103
9104
2.32k
            ctxt->attrHash = tmp;
9105
2.32k
            ctxt->attrHashMax = attrHashSize;
9106
2.32k
        }
9107
9108
40.4k
        memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0]));
9109
9110
132k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9111
91.9k
            const xmlChar *nsuri;
9112
91.9k
            unsigned hashValue, nameHashValue, uriHashValue;
9113
91.9k
            int res;
9114
9115
91.9k
            attname = atts[i];
9116
91.9k
            aprefix = atts[i+1];
9117
91.9k
            nsIndex = XML_PTR_TO_INT(atts[i+2]);
9118
            /* Hash values always have bit 31 set, see dict.c */
9119
91.9k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9120
9121
91.9k
            if (nsIndex == NS_INDEX_EMPTY) {
9122
                /*
9123
                 * Prefix with empty namespace means an undeclared
9124
                 * prefix which was already reported above.
9125
                 */
9126
74.2k
                if (aprefix != NULL)
9127
20.1k
                    continue;
9128
54.0k
                nsuri = NULL;
9129
54.0k
                uriHashValue = URI_HASH_EMPTY;
9130
54.0k
            } else if (nsIndex == NS_INDEX_XML) {
9131
3.63k
                nsuri = ctxt->str_xml_ns;
9132
3.63k
                uriHashValue = URI_HASH_XML;
9133
14.0k
            } else {
9134
14.0k
                nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9135
14.0k
                uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9136
14.0k
            }
9137
9138
71.8k
            hashValue = xmlDictCombineHash(nameHashValue, uriHashValue);
9139
71.8k
            res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9140
71.8k
                                    hashValue, i);
9141
71.8k
            if (res < 0)
9142
0
                continue;
9143
9144
            /*
9145
             * [ WFC: Unique Att Spec ]
9146
             * No attribute name may appear more than once in the same
9147
             * start-tag or empty-element tag.
9148
             * As extended by the Namespace in XML REC.
9149
             */
9150
71.8k
            if (res < INT_MAX) {
9151
24.0k
                if (aprefix == atts[res+1]) {
9152
22.2k
                    xmlErrAttributeDup(ctxt, aprefix, attname);
9153
22.2k
                    numDupErr += 1;
9154
22.2k
                } else {
9155
1.81k
                    xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9156
1.81k
                             "Namespaced Attribute %s in '%s' redefined\n",
9157
1.81k
                             attname, nsuri, NULL);
9158
1.81k
                    numNsErr += 1;
9159
1.81k
                }
9160
24.0k
            }
9161
71.8k
        }
9162
40.4k
    }
9163
9164
    /*
9165
     * Default attributes
9166
     */
9167
573k
    if (ctxt->attsDefault != NULL) {
9168
106k
        xmlDefAttrsPtr defaults;
9169
9170
106k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9171
106k
  if (defaults != NULL) {
9172
427k
      for (i = 0; i < defaults->nbAttrs; i++) {
9173
338k
                xmlDefAttr *attr = &defaults->attrs[i];
9174
338k
                const xmlChar *nsuri = NULL;
9175
338k
                unsigned hashValue, uriHashValue = 0;
9176
338k
                int res;
9177
9178
338k
          attname = attr->name.name;
9179
338k
    aprefix = attr->prefix.name;
9180
9181
338k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL))
9182
6.80k
                    continue;
9183
332k
    if (aprefix == ctxt->str_xmlns)
9184
183k
                    continue;
9185
9186
148k
                if (aprefix == NULL) {
9187
101k
                    nsIndex = NS_INDEX_EMPTY;
9188
101k
                    nsuri = NULL;
9189
101k
                    uriHashValue = URI_HASH_EMPTY;
9190
101k
                } else if (aprefix == ctxt->str_xml) {
9191
8.52k
                    nsIndex = NS_INDEX_XML;
9192
8.52k
                    nsuri = ctxt->str_xml_ns;
9193
8.52k
                    uriHashValue = URI_HASH_XML;
9194
37.9k
                } else {
9195
37.9k
                    nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL);
9196
37.9k
                    if ((nsIndex == INT_MAX) ||
9197
30.1k
                        (nsIndex < ctxt->nsdb->minNsIndex)) {
9198
30.1k
                        xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9199
30.1k
                                 "Namespace prefix %s for %s on %s is not "
9200
30.1k
                                 "defined\n",
9201
30.1k
                                 aprefix, attname, localname);
9202
30.1k
                        nsIndex = NS_INDEX_EMPTY;
9203
30.1k
                        nsuri = NULL;
9204
30.1k
                        uriHashValue = URI_HASH_EMPTY;
9205
30.1k
                    } else {
9206
7.82k
                        nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9207
7.82k
                        uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9208
7.82k
                    }
9209
37.9k
                }
9210
9211
                /*
9212
                 * Check whether the attribute exists
9213
                 */
9214
148k
                if (maxAtts > 1) {
9215
112k
                    hashValue = xmlDictCombineHash(attr->name.hashValue,
9216
112k
                                                   uriHashValue);
9217
112k
                    res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9218
112k
                                            hashValue, nbatts);
9219
112k
                    if (res < 0)
9220
0
                        continue;
9221
112k
                    if (res < INT_MAX) {
9222
4.24k
                        if (aprefix == atts[res+1])
9223
3.22k
                            continue;
9224
1.02k
                        xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9225
1.02k
                                 "Namespaced Attribute %s in '%s' redefined\n",
9226
1.02k
                                 attname, nsuri, NULL);
9227
1.02k
                    }
9228
112k
                }
9229
9230
145k
                xmlParserEntityCheck(ctxt, attr->expandedSize);
9231
9232
145k
                if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9233
2.51k
                    res = xmlCtxtGrowAttrs(ctxt);
9234
9235
2.51k
                    maxatts = ctxt->maxatts;
9236
2.51k
                    atts = ctxt->atts;
9237
9238
2.51k
                    if (res < 0) {
9239
8
                        localname = NULL;
9240
8
                        goto done;
9241
8
                    }
9242
2.51k
                }
9243
9244
145k
                atts[nbatts++] = attname;
9245
145k
                atts[nbatts++] = aprefix;
9246
145k
                atts[nbatts++] = XML_INT_TO_PTR(nsIndex);
9247
145k
                atts[nbatts++] = attr->value.name;
9248
145k
                atts[nbatts++] = attr->valueEnd;
9249
9250
145k
#ifdef LIBXML_VALID_ENABLED
9251
                /*
9252
                 * This should be moved to valid.c, but we don't keep track
9253
                 * whether an attribute was defaulted.
9254
                 */
9255
145k
                if ((ctxt->validate) &&
9256
0
                    (ctxt->standalone == 1) &&
9257
0
                    (attr->external != 0)) {
9258
0
                    xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED,
9259
0
                            "standalone: attribute %s on %s defaulted "
9260
0
                            "from external subset\n",
9261
0
                            attname, localname);
9262
0
                }
9263
145k
#endif
9264
145k
                nbdef++;
9265
145k
      }
9266
88.9k
  }
9267
106k
    }
9268
9269
    /*
9270
     * Using a single hash table for nsUri/localName pairs cannot
9271
     * detect duplicate QNames reliably. The following example will
9272
     * only result in two namespace errors.
9273
     *
9274
     * <doc xmlns:a="a" xmlns:b="a">
9275
     *   <elem a:a="" b:a="" b:a=""/>
9276
     * </doc>
9277
     *
9278
     * If we saw more than one namespace error but no duplicate QNames
9279
     * were found, we have to scan for duplicate QNames.
9280
     */
9281
573k
    if ((numDupErr == 0) && (numNsErr > 1)) {
9282
700
        memset(ctxt->attrHash, -1,
9283
700
               attrHashSize * sizeof(ctxt->attrHash[0]));
9284
9285
4.79k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9286
4.09k
            unsigned hashValue, nameHashValue, prefixHashValue;
9287
4.09k
            int res;
9288
9289
4.09k
            aprefix = atts[i+1];
9290
4.09k
            if (aprefix == NULL)
9291
288
                continue;
9292
9293
3.80k
            attname = atts[i];
9294
            /* Hash values always have bit 31 set, see dict.c */
9295
3.80k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9296
3.80k
            prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix);
9297
9298
3.80k
            hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue);
9299
3.80k
            res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname,
9300
3.80k
                                         aprefix, hashValue, i);
9301
3.80k
            if (res < INT_MAX)
9302
1.12k
                xmlErrAttributeDup(ctxt, aprefix, attname);
9303
3.80k
        }
9304
700
    }
9305
9306
    /*
9307
     * Reconstruct attribute pointers
9308
     */
9309
922k
    for (i = 0, j = 0; i < nbatts; i += 5, j++) {
9310
        /* namespace URI */
9311
348k
        nsIndex = XML_PTR_TO_INT(atts[i+2]);
9312
348k
        if (nsIndex == INT_MAX)
9313
292k
            atts[i+2] = NULL;
9314
55.6k
        else if (nsIndex == INT_MAX - 1)
9315
30.0k
            atts[i+2] = ctxt->str_xml_ns;
9316
25.6k
        else
9317
25.6k
            atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1];
9318
9319
348k
        if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) {
9320
164k
            atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]);  /* value */
9321
164k
            atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]);  /* valuend */
9322
164k
        }
9323
348k
    }
9324
9325
573k
    uri = xmlParserNsLookupUri(ctxt, &hprefix);
9326
573k
    if ((prefix != NULL) && (uri == NULL)) {
9327
18.4k
  xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9328
18.4k
           "Namespace prefix %s on %s is not defined\n",
9329
18.4k
     prefix, localname, NULL);
9330
18.4k
    }
9331
573k
    *pref = prefix;
9332
573k
    *URI = uri;
9333
9334
    /*
9335
     * SAX callback
9336
     */
9337
573k
    if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) &&
9338
573k
  (!ctxt->disableSAX)) {
9339
481k
  if (nbNs > 0)
9340
62.7k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9341
62.7k
                          nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs),
9342
62.7k
        nbatts / 5, nbdef, atts);
9343
418k
  else
9344
418k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9345
418k
                          0, NULL, nbatts / 5, nbdef, atts);
9346
481k
    }
9347
9348
573k
done:
9349
    /*
9350
     * Free allocated attribute values
9351
     */
9352
573k
    if (attval != 0) {
9353
92.0k
  for (i = 0, j = 0; j < nratts; i += 5, j++)
9354
58.0k
      if (ctxt->attallocs[j] & 0x80000000)
9355
38.3k
          xmlFree((xmlChar *) atts[i+3]);
9356
34.0k
    }
9357
9358
573k
    *nbNsPtr = nbNs;
9359
573k
    return(localname);
9360
573k
}
9361
9362
/**
9363
 * Parse an end tag. Always consumes '</'.
9364
 *
9365
 *     [42] ETag ::= '</' Name S? '>'
9366
 *
9367
 * With namespace
9368
 *
9369
 *     [NS 9] ETag ::= '</' QName S? '>'
9370
 * @param ctxt  an XML parser context
9371
 * @param tag  the corresponding start tag
9372
 */
9373
9374
static void
9375
68.3k
xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) {
9376
68.3k
    const xmlChar *name;
9377
9378
68.3k
    GROW;
9379
68.3k
    if ((RAW != '<') || (NXT(1) != '/')) {
9380
1.11k
  xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL);
9381
1.11k
  return;
9382
1.11k
    }
9383
67.2k
    SKIP(2);
9384
9385
67.2k
    if (tag->prefix == NULL)
9386
53.7k
        name = xmlParseNameAndCompare(ctxt, ctxt->name);
9387
13.5k
    else
9388
13.5k
        name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix);
9389
9390
    /*
9391
     * We should definitely be at the ending "S? '>'" part
9392
     */
9393
67.2k
    GROW;
9394
67.2k
    SKIP_BLANKS;
9395
67.2k
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
9396
23.1k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
9397
23.1k
    } else
9398
44.1k
  NEXT1;
9399
9400
    /*
9401
     * [ WFC: Element Type Match ]
9402
     * The Name in an element's end-tag must match the element type in the
9403
     * start-tag.
9404
     *
9405
     */
9406
67.2k
    if (name != (xmlChar*)1) {
9407
31.4k
        if (name == NULL) name = BAD_CAST "unparsable";
9408
31.4k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
9409
31.4k
         "Opening and ending tag mismatch: %s line %d and %s\n",
9410
31.4k
                    ctxt->name, tag->line, name);
9411
31.4k
    }
9412
9413
    /*
9414
     * SAX: End of Tag
9415
     */
9416
67.2k
    if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9417
67.2k
  (!ctxt->disableSAX))
9418
56.9k
  ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix,
9419
56.9k
                                tag->URI);
9420
9421
67.2k
    spacePop(ctxt);
9422
67.2k
    if (tag->nsNr != 0)
9423
3.06k
  xmlParserNsPop(ctxt, tag->nsNr);
9424
67.2k
}
9425
9426
/**
9427
 * Parse escaped pure raw content. Always consumes '<!['.
9428
 *
9429
 * @deprecated Internal function, don't use.
9430
 *
9431
 *     [18] CDSect ::= CDStart CData CDEnd
9432
 *
9433
 *     [19] CDStart ::= '<![CDATA['
9434
 *
9435
 *     [20] Data ::= (Char* - (Char* ']]>' Char*))
9436
 *
9437
 *     [21] CDEnd ::= ']]>'
9438
 * @param ctxt  an XML parser context
9439
 */
9440
void
9441
23.5k
xmlParseCDSect(xmlParserCtxt *ctxt) {
9442
23.5k
    xmlChar *buf = NULL;
9443
23.5k
    int len = 0;
9444
23.5k
    int size = XML_PARSER_BUFFER_SIZE;
9445
23.5k
    int r, rl;
9446
23.5k
    int s, sl;
9447
23.5k
    int cur, l;
9448
23.5k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9449
9.35k
                    XML_MAX_HUGE_LENGTH :
9450
23.5k
                    XML_MAX_TEXT_LENGTH;
9451
9452
23.5k
    if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '['))
9453
0
        return;
9454
23.5k
    SKIP(3);
9455
9456
23.5k
    if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '['))
9457
0
        return;
9458
23.5k
    SKIP(6);
9459
9460
23.5k
    r = xmlCurrentCharRecover(ctxt, &rl);
9461
23.5k
    if (!IS_CHAR(r)) {
9462
598
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9463
598
        goto out;
9464
598
    }
9465
22.9k
    NEXTL(rl);
9466
22.9k
    s = xmlCurrentCharRecover(ctxt, &sl);
9467
22.9k
    if (!IS_CHAR(s)) {
9468
684
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9469
684
        goto out;
9470
684
    }
9471
22.2k
    NEXTL(sl);
9472
22.2k
    cur = xmlCurrentCharRecover(ctxt, &l);
9473
22.2k
    buf = xmlMalloc(size);
9474
22.2k
    if (buf == NULL) {
9475
3
  xmlErrMemory(ctxt);
9476
3
        goto out;
9477
3
    }
9478
4.12M
    while (IS_CHAR(cur) &&
9479
4.11M
           ((r != ']') || (s != ']') || (cur != '>'))) {
9480
4.09M
  if (len + 5 >= size) {
9481
4.00k
      xmlChar *tmp;
9482
4.00k
            int newSize;
9483
9484
4.00k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9485
4.00k
            if (newSize < 0) {
9486
0
                xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9487
0
                               "CData section too big found\n");
9488
0
                goto out;
9489
0
            }
9490
4.00k
      tmp = xmlRealloc(buf, newSize);
9491
4.00k
      if (tmp == NULL) {
9492
8
    xmlErrMemory(ctxt);
9493
8
                goto out;
9494
8
      }
9495
3.99k
      buf = tmp;
9496
3.99k
      size = newSize;
9497
3.99k
  }
9498
4.09M
  COPY_BUF(buf, len, r);
9499
4.09M
  r = s;
9500
4.09M
  rl = sl;
9501
4.09M
  s = cur;
9502
4.09M
  sl = l;
9503
4.09M
  NEXTL(l);
9504
4.09M
  cur = xmlCurrentCharRecover(ctxt, &l);
9505
4.09M
    }
9506
22.2k
    buf[len] = 0;
9507
22.2k
    if (cur != '>') {
9508
4.38k
  xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9509
4.38k
                       "CData section not finished\n%.50s\n", buf);
9510
4.38k
        goto out;
9511
4.38k
    }
9512
17.8k
    NEXTL(l);
9513
9514
    /*
9515
     * OK the buffer is to be consumed as cdata.
9516
     */
9517
17.8k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
9518
13.0k
        if ((ctxt->sax->cdataBlock != NULL) &&
9519
13.0k
            ((ctxt->options & XML_PARSE_NOCDATA) == 0)) {
9520
11.0k
            ctxt->sax->cdataBlock(ctxt->userData, buf, len);
9521
11.0k
        } else if (ctxt->sax->characters != NULL) {
9522
1.91k
            ctxt->sax->characters(ctxt->userData, buf, len);
9523
1.91k
        }
9524
13.0k
    }
9525
9526
23.5k
out:
9527
23.5k
    xmlFree(buf);
9528
23.5k
}
9529
9530
/**
9531
 * Parse a content sequence. Stops at EOF or '</'. Leaves checking of
9532
 * unexpected EOF to the caller.
9533
 *
9534
 * @param ctxt  an XML parser context
9535
 */
9536
9537
static void
9538
85.1k
xmlParseContentInternal(xmlParserCtxtPtr ctxt) {
9539
85.1k
    int oldNameNr = ctxt->nameNr;
9540
85.1k
    int oldSpaceNr = ctxt->spaceNr;
9541
85.1k
    int oldNodeNr = ctxt->nodeNr;
9542
9543
85.1k
    GROW;
9544
3.36M
    while ((ctxt->input->cur < ctxt->input->end) &&
9545
3.28M
     (PARSER_STOPPED(ctxt) == 0)) {
9546
3.27M
  const xmlChar *cur = ctxt->input->cur;
9547
9548
  /*
9549
   * First case : a Processing Instruction.
9550
   */
9551
3.27M
  if ((*cur == '<') && (cur[1] == '?')) {
9552
22.7k
      xmlParsePI(ctxt);
9553
22.7k
  }
9554
9555
  /*
9556
   * Second case : a CDSection
9557
   */
9558
  /* 2.6.0 test was *cur not RAW */
9559
3.25M
  else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) {
9560
21.5k
      xmlParseCDSect(ctxt);
9561
21.5k
  }
9562
9563
  /*
9564
   * Third case :  a comment
9565
   */
9566
3.23M
  else if ((*cur == '<') && (NXT(1) == '!') &&
9567
72.8k
     (NXT(2) == '-') && (NXT(3) == '-')) {
9568
36.8k
      xmlParseComment(ctxt);
9569
36.8k
  }
9570
9571
  /*
9572
   * Fourth case :  a sub-element.
9573
   */
9574
3.19M
  else if (*cur == '<') {
9575
574k
            if (NXT(1) == '/') {
9576
61.5k
                if (ctxt->nameNr <= oldNameNr)
9577
392
                    break;
9578
61.1k
          xmlParseElementEnd(ctxt);
9579
513k
            } else {
9580
513k
          xmlParseElementStart(ctxt);
9581
513k
            }
9582
574k
  }
9583
9584
  /*
9585
   * Fifth case : a reference. If if has not been resolved,
9586
   *    parsing returns it's Name, create the node
9587
   */
9588
9589
2.62M
  else if (*cur == '&') {
9590
265k
      xmlParseReference(ctxt);
9591
265k
  }
9592
9593
  /*
9594
   * Last case, text. Note that References are handled directly.
9595
   */
9596
2.35M
  else {
9597
2.35M
      xmlParseCharDataInternal(ctxt, 0);
9598
2.35M
  }
9599
9600
3.27M
  SHRINK;
9601
3.27M
  GROW;
9602
3.27M
    }
9603
9604
85.1k
    if ((ctxt->nameNr > oldNameNr) &&
9605
6.21k
        (ctxt->input->cur >= ctxt->input->end) &&
9606
5.64k
        (ctxt->wellFormed)) {
9607
130
        const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9608
130
        int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9609
130
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9610
130
                "Premature end of data in tag %s line %d\n",
9611
130
                name, line, NULL);
9612
130
    }
9613
9614
    /*
9615
     * Clean up in error case
9616
     */
9617
9618
227k
    while (ctxt->nodeNr > oldNodeNr)
9619
142k
        nodePop(ctxt);
9620
9621
272k
    while (ctxt->nameNr > oldNameNr) {
9622
186k
        xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1];
9623
9624
186k
        if (tag->nsNr != 0)
9625
39.2k
            xmlParserNsPop(ctxt, tag->nsNr);
9626
9627
186k
        namePop(ctxt);
9628
186k
    }
9629
9630
272k
    while (ctxt->spaceNr > oldSpaceNr)
9631
186k
        spacePop(ctxt);
9632
85.1k
}
9633
9634
/**
9635
 * Parse XML element content. This is useful if you're only interested
9636
 * in custom SAX callbacks. If you want a node list, use
9637
 * #xmlCtxtParseContent.
9638
 *
9639
 * @param ctxt  an XML parser context
9640
 */
9641
void
9642
0
xmlParseContent(xmlParserCtxt *ctxt) {
9643
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
9644
0
        return;
9645
9646
0
    xmlCtxtInitializeLate(ctxt);
9647
9648
0
    xmlParseContentInternal(ctxt);
9649
9650
0
    xmlParserCheckEOF(ctxt, XML_ERR_NOT_WELL_BALANCED);
9651
0
}
9652
9653
/**
9654
 * Parse an XML element
9655
 *
9656
 * @deprecated Internal function, don't use.
9657
 *
9658
 *     [39] element ::= EmptyElemTag | STag content ETag
9659
 *
9660
 * [ WFC: Element Type Match ]
9661
 * The Name in an element's end-tag must match the element type in the
9662
 * start-tag.
9663
 *
9664
 * @param ctxt  an XML parser context
9665
 */
9666
9667
void
9668
15.6k
xmlParseElement(xmlParserCtxt *ctxt) {
9669
15.6k
    if (xmlParseElementStart(ctxt) != 0)
9670
6.02k
        return;
9671
9672
9.62k
    xmlParseContentInternal(ctxt);
9673
9674
9.62k
    if (ctxt->input->cur >= ctxt->input->end) {
9675
8.07k
        if (ctxt->wellFormed) {
9676
367
            const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9677
367
            int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9678
367
            xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9679
367
                    "Premature end of data in tag %s line %d\n",
9680
367
                    name, line, NULL);
9681
367
        }
9682
8.07k
        return;
9683
8.07k
    }
9684
9685
1.55k
    xmlParseElementEnd(ctxt);
9686
1.55k
}
9687
9688
/**
9689
 * Parse the start of an XML element. Returns -1 in case of error, 0 if an
9690
 * opening tag was parsed, 1 if an empty element was parsed.
9691
 *
9692
 * Always consumes '<'.
9693
 *
9694
 * @param ctxt  an XML parser context
9695
 */
9696
static int
9697
528k
xmlParseElementStart(xmlParserCtxtPtr ctxt) {
9698
528k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
9699
528k
    const xmlChar *name;
9700
528k
    const xmlChar *prefix = NULL;
9701
528k
    const xmlChar *URI = NULL;
9702
528k
    xmlParserNodeInfo node_info;
9703
528k
    int line;
9704
528k
    xmlNodePtr cur;
9705
528k
    int nbNs = 0;
9706
9707
528k
    if (ctxt->nameNr > maxDepth) {
9708
12
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
9709
12
                "Excessive depth in document: %d use XML_PARSE_HUGE option\n",
9710
12
                ctxt->nameNr);
9711
12
  return(-1);
9712
12
    }
9713
9714
    /* Capture start position */
9715
528k
    if (ctxt->record_info) {
9716
0
        node_info.begin_pos = ctxt->input->consumed +
9717
0
                          (CUR_PTR - ctxt->input->base);
9718
0
  node_info.begin_line = ctxt->input->line;
9719
0
    }
9720
9721
528k
    if (ctxt->spaceNr == 0)
9722
15.6k
  spacePush(ctxt, -1);
9723
513k
    else if (*ctxt->space == -2)
9724
134k
  spacePush(ctxt, -1);
9725
379k
    else
9726
379k
  spacePush(ctxt, *ctxt->space);
9727
9728
528k
    line = ctxt->input->line;
9729
528k
#ifdef LIBXML_SAX1_ENABLED
9730
528k
    if (ctxt->sax2)
9731
528k
#endif /* LIBXML_SAX1_ENABLED */
9732
528k
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
9733
0
#ifdef LIBXML_SAX1_ENABLED
9734
0
    else
9735
0
  name = xmlParseStartTag(ctxt);
9736
528k
#endif /* LIBXML_SAX1_ENABLED */
9737
528k
    if (name == NULL) {
9738
61.9k
  spacePop(ctxt);
9739
61.9k
        return(-1);
9740
61.9k
    }
9741
466k
    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
9742
466k
    cur = ctxt->node;
9743
9744
466k
#ifdef LIBXML_VALID_ENABLED
9745
    /*
9746
     * [ VC: Root Element Type ]
9747
     * The Name in the document type declaration must match the element
9748
     * type of the root element.
9749
     */
9750
466k
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
9751
0
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
9752
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
9753
466k
#endif /* LIBXML_VALID_ENABLED */
9754
9755
    /*
9756
     * Check for an Empty Element.
9757
     */
9758
466k
    if ((RAW == '/') && (NXT(1) == '>')) {
9759
57.3k
        SKIP(2);
9760
57.3k
  if (ctxt->sax2) {
9761
57.3k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9762
57.3k
    (!ctxt->disableSAX))
9763
46.3k
    ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI);
9764
57.3k
#ifdef LIBXML_SAX1_ENABLED
9765
57.3k
  } else {
9766
0
      if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
9767
0
    (!ctxt->disableSAX))
9768
0
    ctxt->sax->endElement(ctxt->userData, name);
9769
0
#endif /* LIBXML_SAX1_ENABLED */
9770
0
  }
9771
57.3k
  namePop(ctxt);
9772
57.3k
  spacePop(ctxt);
9773
57.3k
  if (nbNs > 0)
9774
3.40k
      xmlParserNsPop(ctxt, nbNs);
9775
57.3k
  if (cur != NULL && ctxt->record_info) {
9776
0
            node_info.node = cur;
9777
0
            node_info.end_pos = ctxt->input->consumed +
9778
0
                                (CUR_PTR - ctxt->input->base);
9779
0
            node_info.end_line = ctxt->input->line;
9780
0
            xmlParserAddNodeInfo(ctxt, &node_info);
9781
0
  }
9782
57.3k
  return(1);
9783
57.3k
    }
9784
409k
    if (RAW == '>') {
9785
257k
        NEXT1;
9786
257k
        if (cur != NULL && ctxt->record_info) {
9787
0
            node_info.node = cur;
9788
0
            node_info.end_pos = 0;
9789
0
            node_info.end_line = 0;
9790
0
            xmlParserAddNodeInfo(ctxt, &node_info);
9791
0
        }
9792
257k
    } else {
9793
151k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED,
9794
151k
         "Couldn't find end of Start Tag %s line %d\n",
9795
151k
                    name, line, NULL);
9796
9797
  /*
9798
   * end of parsing of this node.
9799
   */
9800
151k
  nodePop(ctxt);
9801
151k
  namePop(ctxt);
9802
151k
  spacePop(ctxt);
9803
151k
  if (nbNs > 0)
9804
18.8k
      xmlParserNsPop(ctxt, nbNs);
9805
151k
  return(-1);
9806
151k
    }
9807
9808
257k
    return(0);
9809
409k
}
9810
9811
/**
9812
 * Parse the end of an XML element. Always consumes '</'.
9813
 *
9814
 * @param ctxt  an XML parser context
9815
 */
9816
static void
9817
62.6k
xmlParseElementEnd(xmlParserCtxtPtr ctxt) {
9818
62.6k
    xmlNodePtr cur = ctxt->node;
9819
9820
62.6k
    if (ctxt->nameNr <= 0) {
9821
12
        if ((RAW == '<') && (NXT(1) == '/'))
9822
2
            SKIP(2);
9823
12
        return;
9824
12
    }
9825
9826
    /*
9827
     * parse the end of tag: '</' should be here.
9828
     */
9829
62.6k
    if (ctxt->sax2) {
9830
62.6k
  xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
9831
62.6k
  namePop(ctxt);
9832
62.6k
    }
9833
0
#ifdef LIBXML_SAX1_ENABLED
9834
0
    else
9835
0
  xmlParseEndTag1(ctxt, 0);
9836
62.6k
#endif /* LIBXML_SAX1_ENABLED */
9837
9838
    /*
9839
     * Capture end position
9840
     */
9841
62.6k
    if (cur != NULL && ctxt->record_info) {
9842
0
        xmlParserNodeInfoPtr node_info;
9843
9844
0
        node_info = (xmlParserNodeInfoPtr) xmlParserFindNodeInfo(ctxt, cur);
9845
0
        if (node_info != NULL) {
9846
0
            node_info->end_pos = ctxt->input->consumed +
9847
0
                                 (CUR_PTR - ctxt->input->base);
9848
0
            node_info->end_line = ctxt->input->line;
9849
0
        }
9850
0
    }
9851
62.6k
}
9852
9853
/**
9854
 * Parse the XML version value.
9855
 *
9856
 * @deprecated Internal function, don't use.
9857
 *
9858
 *     [26] VersionNum ::= '1.' [0-9]+
9859
 *
9860
 * In practice allow [0-9].[0-9]+ at that level
9861
 *
9862
 * @param ctxt  an XML parser context
9863
 * @returns the string giving the XML version number, or NULL
9864
 */
9865
xmlChar *
9866
8.89k
xmlParseVersionNum(xmlParserCtxt *ctxt) {
9867
8.89k
    xmlChar *buf = NULL;
9868
8.89k
    int len = 0;
9869
8.89k
    int size = 10;
9870
8.89k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9871
3.68k
                    XML_MAX_TEXT_LENGTH :
9872
8.89k
                    XML_MAX_NAME_LENGTH;
9873
8.89k
    xmlChar cur;
9874
9875
8.89k
    buf = xmlMalloc(size);
9876
8.89k
    if (buf == NULL) {
9877
29
  xmlErrMemory(ctxt);
9878
29
  return(NULL);
9879
29
    }
9880
8.86k
    cur = CUR;
9881
8.86k
    if (!((cur >= '0') && (cur <= '9'))) {
9882
571
  xmlFree(buf);
9883
571
  return(NULL);
9884
571
    }
9885
8.29k
    buf[len++] = cur;
9886
8.29k
    NEXT;
9887
8.29k
    cur=CUR;
9888
8.29k
    if (cur != '.') {
9889
511
  xmlFree(buf);
9890
511
  return(NULL);
9891
511
    }
9892
7.78k
    buf[len++] = cur;
9893
7.78k
    NEXT;
9894
7.78k
    cur=CUR;
9895
598k
    while ((cur >= '0') && (cur <= '9')) {
9896
590k
  if (len + 1 >= size) {
9897
901
      xmlChar *tmp;
9898
901
            int newSize;
9899
9900
901
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9901
901
            if (newSize < 0) {
9902
3
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum");
9903
3
                xmlFree(buf);
9904
3
                return(NULL);
9905
3
            }
9906
898
      tmp = xmlRealloc(buf, newSize);
9907
898
      if (tmp == NULL) {
9908
4
    xmlErrMemory(ctxt);
9909
4
          xmlFree(buf);
9910
4
    return(NULL);
9911
4
      }
9912
894
      buf = tmp;
9913
894
            size = newSize;
9914
894
  }
9915
590k
  buf[len++] = cur;
9916
590k
  NEXT;
9917
590k
  cur=CUR;
9918
590k
    }
9919
7.77k
    buf[len] = 0;
9920
7.77k
    return(buf);
9921
7.78k
}
9922
9923
/**
9924
 * Parse the XML version.
9925
 *
9926
 * @deprecated Internal function, don't use.
9927
 *
9928
 *     [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ")
9929
 *
9930
 *     [25] Eq ::= S? '=' S?
9931
 *
9932
 * @param ctxt  an XML parser context
9933
 * @returns the version string, e.g. "1.0"
9934
 */
9935
9936
xmlChar *
9937
29.5k
xmlParseVersionInfo(xmlParserCtxt *ctxt) {
9938
29.5k
    xmlChar *version = NULL;
9939
9940
29.5k
    if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) {
9941
10.0k
  SKIP(7);
9942
10.0k
  SKIP_BLANKS;
9943
10.0k
  if (RAW != '=') {
9944
913
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
9945
913
      return(NULL);
9946
913
        }
9947
9.18k
  NEXT;
9948
9.18k
  SKIP_BLANKS;
9949
9.18k
  if (RAW == '"') {
9950
4.95k
      NEXT;
9951
4.95k
      version = xmlParseVersionNum(ctxt);
9952
4.95k
      if (RAW != '"') {
9953
974
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9954
974
      } else
9955
3.98k
          NEXT;
9956
4.95k
  } else if (RAW == '\''){
9957
3.93k
      NEXT;
9958
3.93k
      version = xmlParseVersionNum(ctxt);
9959
3.93k
      if (RAW != '\'') {
9960
545
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9961
545
      } else
9962
3.39k
          NEXT;
9963
3.93k
  } else {
9964
286
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
9965
286
  }
9966
9.18k
    }
9967
28.6k
    return(version);
9968
29.5k
}
9969
9970
/**
9971
 * Parse the XML encoding name
9972
 *
9973
 * @deprecated Internal function, don't use.
9974
 *
9975
 *     [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')*
9976
 *
9977
 * @param ctxt  an XML parser context
9978
 * @returns the encoding name value or NULL
9979
 */
9980
xmlChar *
9981
10.6k
xmlParseEncName(xmlParserCtxt *ctxt) {
9982
10.6k
    xmlChar *buf = NULL;
9983
10.6k
    int len = 0;
9984
10.6k
    int size = 10;
9985
10.6k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9986
2.03k
                    XML_MAX_TEXT_LENGTH :
9987
10.6k
                    XML_MAX_NAME_LENGTH;
9988
10.6k
    xmlChar cur;
9989
9990
10.6k
    cur = CUR;
9991
10.6k
    if (((cur >= 'a') && (cur <= 'z')) ||
9992
10.2k
        ((cur >= 'A') && (cur <= 'Z'))) {
9993
10.2k
  buf = xmlMalloc(size);
9994
10.2k
  if (buf == NULL) {
9995
32
      xmlErrMemory(ctxt);
9996
32
      return(NULL);
9997
32
  }
9998
9999
10.1k
  buf[len++] = cur;
10000
10.1k
  NEXT;
10001
10.1k
  cur = CUR;
10002
535k
  while (((cur >= 'a') && (cur <= 'z')) ||
10003
85.7k
         ((cur >= 'A') && (cur <= 'Z')) ||
10004
62.3k
         ((cur >= '0') && (cur <= '9')) ||
10005
27.9k
         (cur == '.') || (cur == '_') ||
10006
525k
         (cur == '-')) {
10007
525k
      if (len + 1 >= size) {
10008
5.66k
          xmlChar *tmp;
10009
5.66k
                int newSize;
10010
10011
5.66k
                newSize = xmlGrowCapacity(size, 1, 1, maxLength);
10012
5.66k
                if (newSize < 0) {
10013
3
                    xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName");
10014
3
                    xmlFree(buf);
10015
3
                    return(NULL);
10016
3
                }
10017
5.66k
    tmp = xmlRealloc(buf, newSize);
10018
5.66k
    if (tmp == NULL) {
10019
5
        xmlErrMemory(ctxt);
10020
5
        xmlFree(buf);
10021
5
        return(NULL);
10022
5
    }
10023
5.66k
    buf = tmp;
10024
5.66k
                size = newSize;
10025
5.66k
      }
10026
525k
      buf[len++] = cur;
10027
525k
      NEXT;
10028
525k
      cur = CUR;
10029
525k
        }
10030
10.1k
  buf[len] = 0;
10031
10.1k
    } else {
10032
420
  xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL);
10033
420
    }
10034
10.5k
    return(buf);
10035
10.6k
}
10036
10037
/**
10038
 * Parse the XML encoding declaration
10039
 *
10040
 * @deprecated Internal function, don't use.
10041
 *
10042
 *     [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | 
10043
 *                           "'" EncName "'")
10044
 *
10045
 * this setups the conversion filters.
10046
 *
10047
 * @param ctxt  an XML parser context
10048
 * @returns the encoding value or NULL
10049
 */
10050
10051
const xmlChar *
10052
28.7k
xmlParseEncodingDecl(xmlParserCtxt *ctxt) {
10053
28.7k
    xmlChar *encoding = NULL;
10054
10055
28.7k
    SKIP_BLANKS;
10056
28.7k
    if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0)
10057
17.4k
        return(NULL);
10058
10059
11.2k
    SKIP(8);
10060
11.2k
    SKIP_BLANKS;
10061
11.2k
    if (RAW != '=') {
10062
400
        xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10063
400
        return(NULL);
10064
400
    }
10065
10.8k
    NEXT;
10066
10.8k
    SKIP_BLANKS;
10067
10.8k
    if (RAW == '"') {
10068
6.08k
        NEXT;
10069
6.08k
        encoding = xmlParseEncName(ctxt);
10070
6.08k
        if (RAW != '"') {
10071
531
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10072
531
            xmlFree(encoding);
10073
531
            return(NULL);
10074
531
        } else
10075
5.55k
            NEXT;
10076
6.08k
    } else if (RAW == '\''){
10077
4.54k
        NEXT;
10078
4.54k
        encoding = xmlParseEncName(ctxt);
10079
4.54k
        if (RAW != '\'') {
10080
776
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10081
776
            xmlFree(encoding);
10082
776
            return(NULL);
10083
776
        } else
10084
3.76k
            NEXT;
10085
4.54k
    } else {
10086
234
        xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10087
234
    }
10088
10089
9.55k
    if (encoding == NULL)
10090
237
        return(NULL);
10091
10092
9.31k
    xmlSetDeclaredEncoding(ctxt, encoding);
10093
10094
9.31k
    return(ctxt->encoding);
10095
9.55k
}
10096
10097
/**
10098
 * Parse the XML standalone declaration
10099
 *
10100
 * @deprecated Internal function, don't use.
10101
 *
10102
 *     [32] SDDecl ::= S 'standalone' Eq
10103
 *                     (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"'))
10104
 *
10105
 * [ VC: Standalone Document Declaration ]
10106
 * TODO The standalone document declaration must have the value "no"
10107
 * if any external markup declarations contain declarations of:
10108
 *  - attributes with default values, if elements to which these
10109
 *    attributes apply appear in the document without specifications
10110
 *    of values for these attributes, or
10111
 *  - entities (other than amp, lt, gt, apos, quot), if references
10112
 *    to those entities appear in the document, or
10113
 *  - attributes with values subject to normalization, where the
10114
 *    attribute appears in the document with a value which will change
10115
 *    as a result of normalization, or
10116
 *  - element types with element content, if white space occurs directly
10117
 *    within any instance of those types.
10118
 *
10119
 * @param ctxt  an XML parser context
10120
 * @returns
10121
 *   1 if standalone="yes"
10122
 *   0 if standalone="no"
10123
 *  -2 if standalone attribute is missing or invalid
10124
 *    (A standalone value of -2 means that the XML declaration was found,
10125
 *     but no value was specified for the standalone attribute).
10126
 */
10127
10128
int
10129
3.61k
xmlParseSDDecl(xmlParserCtxt *ctxt) {
10130
3.61k
    int standalone = -2;
10131
10132
3.61k
    SKIP_BLANKS;
10133
3.61k
    if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) {
10134
183
  SKIP(10);
10135
183
        SKIP_BLANKS;
10136
183
  if (RAW != '=') {
10137
3
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10138
3
      return(standalone);
10139
3
        }
10140
180
  NEXT;
10141
180
  SKIP_BLANKS;
10142
180
        if (RAW == '\''){
10143
83
      NEXT;
10144
83
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10145
56
          standalone = 0;
10146
56
                SKIP(2);
10147
56
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10148
9
                 (NXT(2) == 's')) {
10149
6
          standalone = 1;
10150
6
    SKIP(3);
10151
21
            } else {
10152
21
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10153
21
      }
10154
83
      if (RAW != '\'') {
10155
29
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10156
29
      } else
10157
54
          NEXT;
10158
97
  } else if (RAW == '"'){
10159
92
      NEXT;
10160
92
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10161
8
          standalone = 0;
10162
8
    SKIP(2);
10163
84
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10164
65
                 (NXT(2) == 's')) {
10165
62
          standalone = 1;
10166
62
                SKIP(3);
10167
62
            } else {
10168
22
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10169
22
      }
10170
92
      if (RAW != '"') {
10171
56
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10172
56
      } else
10173
36
          NEXT;
10174
92
  } else {
10175
5
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10176
5
        }
10177
180
    }
10178
3.61k
    return(standalone);
10179
3.61k
}
10180
10181
/**
10182
 * Parse an XML declaration header
10183
 *
10184
 * @deprecated Internal function, don't use.
10185
 *
10186
 *     [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>'
10187
 * @param ctxt  an XML parser context
10188
 */
10189
10190
void
10191
7.16k
xmlParseXMLDecl(xmlParserCtxt *ctxt) {
10192
7.16k
    xmlChar *version;
10193
10194
    /*
10195
     * This value for standalone indicates that the document has an
10196
     * XML declaration but it does not have a standalone attribute.
10197
     * It will be overwritten later if a standalone attribute is found.
10198
     */
10199
10200
7.16k
    ctxt->standalone = -2;
10201
10202
    /*
10203
     * We know that '<?xml' is here.
10204
     */
10205
7.16k
    SKIP(5);
10206
10207
7.16k
    if (!IS_BLANK_CH(RAW)) {
10208
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
10209
0
                 "Blank needed after '<?xml'\n");
10210
0
    }
10211
7.16k
    SKIP_BLANKS;
10212
10213
    /*
10214
     * We must have the VersionInfo here.
10215
     */
10216
7.16k
    version = xmlParseVersionInfo(ctxt);
10217
7.16k
    if (version == NULL) {
10218
3.74k
  xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL);
10219
3.74k
    } else {
10220
3.42k
  if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) {
10221
      /*
10222
       * Changed here for XML-1.0 5th edition
10223
       */
10224
1.58k
      if (ctxt->options & XML_PARSE_OLD10) {
10225
238
    xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10226
238
                "Unsupported version '%s'\n",
10227
238
                version);
10228
1.34k
      } else {
10229
1.34k
          if ((version[0] == '1') && ((version[1] == '.'))) {
10230
767
        xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION,
10231
767
                      "Unsupported version '%s'\n",
10232
767
          version, NULL);
10233
767
    } else {
10234
577
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10235
577
              "Unsupported version '%s'\n",
10236
577
              version);
10237
577
    }
10238
1.34k
      }
10239
1.58k
  }
10240
3.42k
  if (ctxt->version != NULL)
10241
0
      xmlFree(ctxt->version);
10242
3.42k
  ctxt->version = version;
10243
3.42k
    }
10244
10245
    /*
10246
     * We may have the encoding declaration
10247
     */
10248
7.16k
    if (!IS_BLANK_CH(RAW)) {
10249
4.69k
        if ((RAW == '?') && (NXT(1) == '>')) {
10250
819
      SKIP(2);
10251
819
      return;
10252
819
  }
10253
3.88k
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10254
3.88k
    }
10255
6.34k
    xmlParseEncodingDecl(ctxt);
10256
10257
    /*
10258
     * We may have the standalone status.
10259
     */
10260
6.34k
    if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) {
10261
3.75k
        if ((RAW == '?') && (NXT(1) == '>')) {
10262
2.73k
      SKIP(2);
10263
2.73k
      return;
10264
2.73k
  }
10265
1.02k
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10266
1.02k
    }
10267
10268
    /*
10269
     * We can grow the input buffer freely at that point
10270
     */
10271
3.61k
    GROW;
10272
10273
3.61k
    SKIP_BLANKS;
10274
3.61k
    ctxt->standalone = xmlParseSDDecl(ctxt);
10275
10276
3.61k
    SKIP_BLANKS;
10277
3.61k
    if ((RAW == '?') && (NXT(1) == '>')) {
10278
151
        SKIP(2);
10279
3.46k
    } else if (RAW == '>') {
10280
        /* Deprecated old WD ... */
10281
270
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10282
270
  NEXT;
10283
3.19k
    } else {
10284
3.19k
        int c;
10285
10286
3.19k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10287
679k
        while ((PARSER_STOPPED(ctxt) == 0) &&
10288
678k
               ((c = CUR) != 0)) {
10289
678k
            NEXT;
10290
678k
            if (c == '>')
10291
2.21k
                break;
10292
678k
        }
10293
3.19k
    }
10294
3.61k
}
10295
10296
/**
10297
 * @since 2.14.0
10298
 *
10299
 * @param ctxt  parser context
10300
 * @returns the version from the XML declaration.
10301
 */
10302
const xmlChar *
10303
0
xmlCtxtGetVersion(xmlParserCtxt *ctxt) {
10304
0
    if (ctxt == NULL)
10305
0
        return(NULL);
10306
10307
0
    return(ctxt->version);
10308
0
}
10309
10310
/**
10311
 * @since 2.14.0
10312
 *
10313
 * @param ctxt  parser context
10314
 * @returns the value from the standalone document declaration.
10315
 */
10316
int
10317
0
xmlCtxtGetStandalone(xmlParserCtxt *ctxt) {
10318
0
    if (ctxt == NULL)
10319
0
        return(0);
10320
10321
0
    return(ctxt->standalone);
10322
0
}
10323
10324
/**
10325
 * Parse an XML Misc* optional field.
10326
 *
10327
 * @deprecated Internal function, don't use.
10328
 *
10329
 *     [27] Misc ::= Comment | PI |  S
10330
 * @param ctxt  an XML parser context
10331
 */
10332
10333
void
10334
54.6k
xmlParseMisc(xmlParserCtxt *ctxt) {
10335
60.8k
    while (PARSER_STOPPED(ctxt) == 0) {
10336
54.9k
        SKIP_BLANKS;
10337
54.9k
        GROW;
10338
54.9k
        if ((RAW == '<') && (NXT(1) == '?')) {
10339
4.46k
      xmlParsePI(ctxt);
10340
50.4k
        } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) {
10341
1.77k
      xmlParseComment(ctxt);
10342
48.6k
        } else {
10343
48.6k
            break;
10344
48.6k
        }
10345
54.9k
    }
10346
54.6k
}
10347
10348
static void
10349
39.7k
xmlFinishDocument(xmlParserCtxtPtr ctxt) {
10350
39.7k
    xmlDocPtr doc;
10351
10352
    /*
10353
     * SAX: end of the document processing.
10354
     */
10355
39.7k
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10356
39.7k
        ctxt->sax->endDocument(ctxt->userData);
10357
10358
    /*
10359
     * Remove locally kept entity definitions if the tree was not built
10360
     */
10361
39.7k
    doc = ctxt->myDoc;
10362
39.7k
    if ((doc != NULL) &&
10363
38.7k
        (xmlStrEqual(doc->version, SAX_COMPAT_MODE))) {
10364
254
        xmlFreeDoc(doc);
10365
254
        ctxt->myDoc = NULL;
10366
254
    }
10367
39.7k
}
10368
10369
/**
10370
 * Parse an XML document and invoke the SAX handlers. This is useful
10371
 * if you're only interested in custom SAX callbacks. If you want a
10372
 * document tree, use #xmlCtxtParseDocument.
10373
 *
10374
 * @param ctxt  an XML parser context
10375
 * @returns 0, -1 in case of error.
10376
 */
10377
10378
int
10379
25.3k
xmlParseDocument(xmlParserCtxt *ctxt) {
10380
25.3k
    if ((ctxt == NULL) || (ctxt->input == NULL))
10381
0
        return(-1);
10382
10383
25.3k
    GROW;
10384
10385
    /*
10386
     * SAX: detecting the level.
10387
     */
10388
25.3k
    xmlCtxtInitializeLate(ctxt);
10389
10390
25.3k
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10391
25.3k
        ctxt->sax->setDocumentLocator(ctxt->userData,
10392
25.3k
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10393
25.3k
    }
10394
10395
25.3k
    xmlDetectEncoding(ctxt);
10396
10397
25.3k
    if (CUR == 0) {
10398
283
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10399
283
  return(-1);
10400
283
    }
10401
10402
25.0k
    GROW;
10403
25.0k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
10404
10405
  /*
10406
   * Note that we will switch encoding on the fly.
10407
   */
10408
3.63k
  xmlParseXMLDecl(ctxt);
10409
3.63k
  SKIP_BLANKS;
10410
21.4k
    } else {
10411
21.4k
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10412
21.4k
        if (ctxt->version == NULL) {
10413
14
            xmlErrMemory(ctxt);
10414
14
            return(-1);
10415
14
        }
10416
21.4k
    }
10417
25.0k
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10418
24.0k
        ctxt->sax->startDocument(ctxt->userData);
10419
25.0k
    if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) &&
10420
23.9k
        (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) {
10421
0
  ctxt->myDoc->compression = ctxt->input->buf->compressed;
10422
0
    }
10423
10424
    /*
10425
     * The Misc part of the Prolog
10426
     */
10427
25.0k
    xmlParseMisc(ctxt);
10428
10429
    /*
10430
     * Then possibly doc type declaration(s) and more Misc
10431
     * (doctypedecl Misc*)?
10432
     */
10433
25.0k
    GROW;
10434
25.0k
    if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) {
10435
10436
13.9k
  ctxt->inSubset = 1;
10437
13.9k
  xmlParseDocTypeDecl(ctxt);
10438
13.9k
  if (RAW == '[') {
10439
11.1k
      xmlParseInternalSubset(ctxt);
10440
11.1k
  } else if (RAW == '>') {
10441
2.14k
            NEXT;
10442
2.14k
        }
10443
10444
  /*
10445
   * Create and update the external subset.
10446
   */
10447
13.9k
  ctxt->inSubset = 2;
10448
13.9k
  if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) &&
10449
13.9k
      (!ctxt->disableSAX))
10450
10.5k
      ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
10451
10.5k
                                ctxt->extSubSystem, ctxt->extSubURI);
10452
13.9k
  ctxt->inSubset = 0;
10453
10454
13.9k
        xmlCleanSpecialAttr(ctxt);
10455
10456
13.9k
  xmlParseMisc(ctxt);
10457
13.9k
    }
10458
10459
    /*
10460
     * Time to start parsing the tree itself
10461
     */
10462
25.0k
    GROW;
10463
25.0k
    if (RAW != '<') {
10464
9.36k
        if (ctxt->wellFormed)
10465
956
            xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10466
956
                           "Start tag expected, '<' not found\n");
10467
15.6k
    } else {
10468
15.6k
  xmlParseElement(ctxt);
10469
10470
  /*
10471
   * The Misc part at the end
10472
   */
10473
15.6k
  xmlParseMisc(ctxt);
10474
10475
15.6k
        xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
10476
15.6k
    }
10477
10478
25.0k
    ctxt->instate = XML_PARSER_EOF;
10479
25.0k
    xmlFinishDocument(ctxt);
10480
10481
25.0k
    if (! ctxt->wellFormed) {
10482
24.8k
  ctxt->valid = 0;
10483
24.8k
  return(-1);
10484
24.8k
    }
10485
10486
148
    return(0);
10487
25.0k
}
10488
10489
/**
10490
 * Parse a general parsed entity
10491
 * An external general parsed entity is well-formed if it matches the
10492
 * production labeled extParsedEnt.
10493
 *
10494
 * @deprecated Internal function, don't use.
10495
 *
10496
 *     [78] extParsedEnt ::= TextDecl? content
10497
 *
10498
 * @param ctxt  an XML parser context
10499
 * @returns 0, -1 in case of error. the parser context is augmented
10500
 *                as a result of the parsing.
10501
 */
10502
10503
int
10504
0
xmlParseExtParsedEnt(xmlParserCtxt *ctxt) {
10505
0
    if ((ctxt == NULL) || (ctxt->input == NULL))
10506
0
        return(-1);
10507
10508
0
    xmlCtxtInitializeLate(ctxt);
10509
10510
0
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10511
0
        ctxt->sax->setDocumentLocator(ctxt->userData,
10512
0
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10513
0
    }
10514
10515
0
    xmlDetectEncoding(ctxt);
10516
10517
0
    if (CUR == 0) {
10518
0
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10519
0
    }
10520
10521
    /*
10522
     * Check for the XMLDecl in the Prolog.
10523
     */
10524
0
    GROW;
10525
0
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
10526
10527
  /*
10528
   * Note that we will switch encoding on the fly.
10529
   */
10530
0
  xmlParseXMLDecl(ctxt);
10531
0
  SKIP_BLANKS;
10532
0
    } else {
10533
0
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10534
0
    }
10535
0
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10536
0
        ctxt->sax->startDocument(ctxt->userData);
10537
10538
    /*
10539
     * Doing validity checking on chunk doesn't make sense
10540
     */
10541
0
    ctxt->options &= ~XML_PARSE_DTDVALID;
10542
0
    ctxt->validate = 0;
10543
0
    ctxt->depth = 0;
10544
10545
0
    xmlParseContentInternal(ctxt);
10546
10547
0
    if (ctxt->input->cur < ctxt->input->end)
10548
0
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
10549
10550
    /*
10551
     * SAX: end of the document processing.
10552
     */
10553
0
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10554
0
        ctxt->sax->endDocument(ctxt->userData);
10555
10556
0
    if (! ctxt->wellFormed) return(-1);
10557
0
    return(0);
10558
0
}
10559
10560
#ifdef LIBXML_PUSH_ENABLED
10561
/************************************************************************
10562
 *                  *
10563
 *    Progressive parsing interfaces        *
10564
 *                  *
10565
 ************************************************************************/
10566
10567
/**
10568
 * Check whether the input buffer contains a character.
10569
 *
10570
 * @param ctxt  an XML parser context
10571
 * @param c  character
10572
 */
10573
static int
10574
16.8k
xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) {
10575
16.8k
    const xmlChar *cur;
10576
10577
16.8k
    if (ctxt->checkIndex == 0) {
10578
11.3k
        cur = ctxt->input->cur + 1;
10579
11.3k
    } else {
10580
5.51k
        cur = ctxt->input->cur + ctxt->checkIndex;
10581
5.51k
    }
10582
10583
16.8k
    if (memchr(cur, c, ctxt->input->end - cur) == NULL) {
10584
6.14k
        size_t index = ctxt->input->end - ctxt->input->cur;
10585
10586
6.14k
        if (index > LONG_MAX) {
10587
0
            ctxt->checkIndex = 0;
10588
0
            return(1);
10589
0
        }
10590
6.14k
        ctxt->checkIndex = index;
10591
6.14k
        return(0);
10592
10.6k
    } else {
10593
10.6k
        ctxt->checkIndex = 0;
10594
10.6k
        return(1);
10595
10.6k
    }
10596
16.8k
}
10597
10598
/**
10599
 * Check whether the input buffer contains a string.
10600
 *
10601
 * @param ctxt  an XML parser context
10602
 * @param startDelta  delta to apply at the start
10603
 * @param str  string
10604
 * @param strLen  length of string
10605
 */
10606
static const xmlChar *
10607
xmlParseLookupString(xmlParserCtxtPtr ctxt, size_t startDelta,
10608
84.2k
                     const char *str, size_t strLen) {
10609
84.2k
    const xmlChar *cur, *term;
10610
10611
84.2k
    if (ctxt->checkIndex == 0) {
10612
9.34k
        cur = ctxt->input->cur + startDelta;
10613
74.8k
    } else {
10614
74.8k
        cur = ctxt->input->cur + ctxt->checkIndex;
10615
74.8k
    }
10616
10617
84.2k
    term = BAD_CAST strstr((const char *) cur, str);
10618
84.2k
    if (term == NULL) {
10619
76.9k
        const xmlChar *end = ctxt->input->end;
10620
76.9k
        size_t index;
10621
10622
        /* Rescan (strLen - 1) characters. */
10623
76.9k
        if ((size_t) (end - cur) < strLen)
10624
3.58k
            end = cur;
10625
73.3k
        else
10626
73.3k
            end -= strLen - 1;
10627
76.9k
        index = end - ctxt->input->cur;
10628
76.9k
        if (index > LONG_MAX) {
10629
0
            ctxt->checkIndex = 0;
10630
0
            return(ctxt->input->end - strLen);
10631
0
        }
10632
76.9k
        ctxt->checkIndex = index;
10633
76.9k
    } else {
10634
7.25k
        ctxt->checkIndex = 0;
10635
7.25k
    }
10636
10637
84.2k
    return(term);
10638
84.2k
}
10639
10640
/**
10641
 * Check whether the input buffer contains terminated char data.
10642
 *
10643
 * @param ctxt  an XML parser context
10644
 */
10645
static int
10646
26.4k
xmlParseLookupCharData(xmlParserCtxtPtr ctxt) {
10647
26.4k
    const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex;
10648
26.4k
    const xmlChar *end = ctxt->input->end;
10649
26.4k
    size_t index;
10650
10651
937k
    while (cur < end) {
10652
924k
        if ((*cur == '<') || (*cur == '&')) {
10653
13.3k
            ctxt->checkIndex = 0;
10654
13.3k
            return(1);
10655
13.3k
        }
10656
911k
        cur++;
10657
911k
    }
10658
10659
13.1k
    index = cur - ctxt->input->cur;
10660
13.1k
    if (index > LONG_MAX) {
10661
0
        ctxt->checkIndex = 0;
10662
0
        return(1);
10663
0
    }
10664
13.1k
    ctxt->checkIndex = index;
10665
13.1k
    return(0);
10666
13.1k
}
10667
10668
/**
10669
 * Check whether there's enough data in the input buffer to finish parsing
10670
 * a start tag. This has to take quotes into account.
10671
 *
10672
 * @param ctxt  an XML parser context
10673
 */
10674
static int
10675
122k
xmlParseLookupGt(xmlParserCtxtPtr ctxt) {
10676
122k
    const xmlChar *cur;
10677
122k
    const xmlChar *end = ctxt->input->end;
10678
122k
    int state = ctxt->endCheckState;
10679
122k
    size_t index;
10680
10681
122k
    if (ctxt->checkIndex == 0)
10682
56.2k
        cur = ctxt->input->cur + 1;
10683
65.9k
    else
10684
65.9k
        cur = ctxt->input->cur + ctxt->checkIndex;
10685
10686
65.0M
    while (cur < end) {
10687
64.9M
        if (state) {
10688
59.0M
            if (*cur == state)
10689
58.7k
                state = 0;
10690
59.0M
        } else if (*cur == '\'' || *cur == '"') {
10691
62.4k
            state = *cur;
10692
5.85M
        } else if (*cur == '>') {
10693
48.7k
            ctxt->checkIndex = 0;
10694
48.7k
            ctxt->endCheckState = 0;
10695
48.7k
            return(1);
10696
48.7k
        }
10697
64.9M
        cur++;
10698
64.9M
    }
10699
10700
73.3k
    index = cur - ctxt->input->cur;
10701
73.3k
    if (index > LONG_MAX) {
10702
0
        ctxt->checkIndex = 0;
10703
0
        ctxt->endCheckState = 0;
10704
0
        return(1);
10705
0
    }
10706
73.3k
    ctxt->checkIndex = index;
10707
73.3k
    ctxt->endCheckState = state;
10708
73.3k
    return(0);
10709
73.3k
}
10710
10711
/**
10712
 * Check whether there's enough data in the input buffer to finish parsing
10713
 * the internal subset.
10714
 *
10715
 * @param ctxt  an XML parser context
10716
 */
10717
static int
10718
199k
xmlParseLookupInternalSubset(xmlParserCtxtPtr ctxt) {
10719
    /*
10720
     * Sorry, but progressive parsing of the internal subset is not
10721
     * supported. We first check that the full content of the internal
10722
     * subset is available and parsing is launched only at that point.
10723
     * Internal subset ends with "']' S? '>'" in an unescaped section and
10724
     * not in a ']]>' sequence which are conditional sections.
10725
     */
10726
199k
    const xmlChar *cur, *start;
10727
199k
    const xmlChar *end = ctxt->input->end;
10728
199k
    int state = ctxt->endCheckState;
10729
199k
    size_t index;
10730
10731
199k
    if (ctxt->checkIndex == 0) {
10732
5.08k
        cur = ctxt->input->cur + 1;
10733
194k
    } else {
10734
194k
        cur = ctxt->input->cur + ctxt->checkIndex;
10735
194k
    }
10736
199k
    start = cur;
10737
10738
325M
    while (cur < end) {
10739
324M
        if (state == '-') {
10740
19.6k
            if ((*cur == '-') &&
10741
3.53k
                (cur[1] == '-') &&
10742
2.51k
                (cur[2] == '>')) {
10743
1.73k
                state = 0;
10744
1.73k
                cur += 3;
10745
1.73k
                start = cur;
10746
1.73k
                continue;
10747
1.73k
            }
10748
19.6k
        }
10749
324M
        else if (state == ']') {
10750
4.52k
            if (*cur == '>') {
10751
2.47k
                ctxt->checkIndex = 0;
10752
2.47k
                ctxt->endCheckState = 0;
10753
2.47k
                return(1);
10754
2.47k
            }
10755
2.04k
            if (IS_BLANK_CH(*cur)) {
10756
1.05k
                state = ' ';
10757
1.05k
            } else if (*cur != ']') {
10758
556
                state = 0;
10759
556
                start = cur;
10760
556
                continue;
10761
556
            }
10762
2.04k
        }
10763
324M
        else if (state == ' ') {
10764
2.23k
            if (*cur == '>') {
10765
39
                ctxt->checkIndex = 0;
10766
39
                ctxt->endCheckState = 0;
10767
39
                return(1);
10768
39
            }
10769
2.19k
            if (!IS_BLANK_CH(*cur)) {
10770
993
                state = 0;
10771
993
                start = cur;
10772
993
                continue;
10773
993
            }
10774
2.19k
        }
10775
324M
        else if (state != 0) {
10776
323M
            if (*cur == state) {
10777
14.5k
                state = 0;
10778
14.5k
                start = cur + 1;
10779
14.5k
            }
10780
323M
        }
10781
1.70M
        else if (*cur == '<') {
10782
24.0k
            if ((cur[1] == '!') &&
10783
16.2k
                (cur[2] == '-') &&
10784
2.25k
                (cur[3] == '-')) {
10785
1.77k
                state = '-';
10786
1.77k
                cur += 4;
10787
                /* Don't treat <!--> as comment */
10788
1.77k
                start = cur;
10789
1.77k
                continue;
10790
1.77k
            }
10791
24.0k
        }
10792
1.67M
        else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) {
10793
19.3k
            state = *cur;
10794
19.3k
        }
10795
10796
324M
        cur++;
10797
324M
    }
10798
10799
    /*
10800
     * Rescan the three last characters to detect "<!--" and "-->"
10801
     * split across chunks.
10802
     */
10803
196k
    if ((state == 0) || (state == '-')) {
10804
10.4k
        if (cur - start < 3)
10805
1.74k
            cur = start;
10806
8.68k
        else
10807
8.68k
            cur -= 3;
10808
10.4k
    }
10809
196k
    index = cur - ctxt->input->cur;
10810
196k
    if (index > LONG_MAX) {
10811
0
        ctxt->checkIndex = 0;
10812
0
        ctxt->endCheckState = 0;
10813
0
        return(1);
10814
0
    }
10815
196k
    ctxt->checkIndex = index;
10816
196k
    ctxt->endCheckState = state;
10817
196k
    return(0);
10818
196k
}
10819
10820
/**
10821
 * Try to progress on parsing
10822
 *
10823
 * @param ctxt  an XML parser context
10824
 * @param terminate  last chunk indicator
10825
 * @returns zero if no parsing was possible
10826
 */
10827
static int
10828
424k
xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
10829
424k
    int ret = 0;
10830
424k
    size_t avail;
10831
424k
    xmlChar cur, next;
10832
10833
424k
    if (ctxt->input == NULL)
10834
0
        return(0);
10835
10836
424k
    if ((ctxt->input != NULL) &&
10837
424k
        (ctxt->input->cur - ctxt->input->base > 4096)) {
10838
829
        xmlParserShrink(ctxt);
10839
829
    }
10840
10841
896k
    while (ctxt->disableSAX == 0) {
10842
886k
        avail = ctxt->input->end - ctxt->input->cur;
10843
886k
        if (avail < 1)
10844
13.5k
      goto done;
10845
872k
        switch (ctxt->instate) {
10846
6.58k
            case XML_PARSER_EOF:
10847
          /*
10848
     * Document parsing is done !
10849
     */
10850
6.58k
          goto done;
10851
35.0k
            case XML_PARSER_START:
10852
                /*
10853
                 * Very first chars read from the document flow.
10854
                 */
10855
35.0k
                if ((!terminate) && (avail < 4))
10856
8.11k
                    goto done;
10857
10858
                /*
10859
                 * We need more bytes to detect EBCDIC code pages.
10860
                 * See xmlDetectEBCDIC.
10861
                 */
10862
26.9k
                if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) &&
10863
2.30k
                    (!terminate) && (avail < 200))
10864
1.99k
                    goto done;
10865
10866
24.9k
                xmlDetectEncoding(ctxt);
10867
24.9k
                ctxt->instate = XML_PARSER_XML_DECL;
10868
24.9k
    break;
10869
10870
48.0k
            case XML_PARSER_XML_DECL:
10871
48.0k
    if ((!terminate) && (avail < 2))
10872
208
        goto done;
10873
47.8k
    cur = ctxt->input->cur[0];
10874
47.8k
    next = ctxt->input->cur[1];
10875
47.8k
          if ((cur == '<') && (next == '?')) {
10876
        /* PI or XML decl */
10877
28.3k
        if ((!terminate) &&
10878
25.0k
                        (!xmlParseLookupString(ctxt, 2, "?>", 2)))
10879
23.0k
      goto done;
10880
5.30k
        if ((ctxt->input->cur[2] == 'x') &&
10881
4.21k
      (ctxt->input->cur[3] == 'm') &&
10882
3.94k
      (ctxt->input->cur[4] == 'l') &&
10883
3.63k
      (IS_BLANK_CH(ctxt->input->cur[5]))) {
10884
3.53k
      ret += 5;
10885
3.53k
      xmlParseXMLDecl(ctxt);
10886
3.53k
        } else {
10887
1.76k
      ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10888
1.76k
                        if (ctxt->version == NULL) {
10889
3
                            xmlErrMemory(ctxt);
10890
3
                            break;
10891
3
                        }
10892
1.76k
        }
10893
19.5k
    } else {
10894
19.5k
        ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10895
19.5k
        if (ctxt->version == NULL) {
10896
40
            xmlErrMemory(ctxt);
10897
40
      break;
10898
40
        }
10899
19.5k
    }
10900
24.8k
                if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10901
24.8k
                    ctxt->sax->setDocumentLocator(ctxt->userData,
10902
24.8k
                            (xmlSAXLocator *) &xmlDefaultSAXLocator);
10903
24.8k
                }
10904
24.8k
                if ((ctxt->sax) && (ctxt->sax->startDocument) &&
10905
24.8k
                    (!ctxt->disableSAX))
10906
24.1k
                    ctxt->sax->startDocument(ctxt->userData);
10907
24.8k
                ctxt->instate = XML_PARSER_MISC;
10908
24.8k
    break;
10909
147k
            case XML_PARSER_START_TAG: {
10910
147k
          const xmlChar *name;
10911
147k
    const xmlChar *prefix = NULL;
10912
147k
    const xmlChar *URI = NULL;
10913
147k
                int line = ctxt->input->line;
10914
147k
    int nbNs = 0;
10915
10916
147k
    if ((!terminate) && (avail < 2))
10917
23
        goto done;
10918
147k
    cur = ctxt->input->cur[0];
10919
147k
          if (cur != '<') {
10920
2.53k
        xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10921
2.53k
                                   "Start tag expected, '<' not found");
10922
2.53k
                    ctxt->instate = XML_PARSER_EOF;
10923
2.53k
                    xmlFinishDocument(ctxt);
10924
2.53k
        goto done;
10925
2.53k
    }
10926
144k
    if ((!terminate) && (!xmlParseLookupGt(ctxt)))
10927
34.8k
                    goto done;
10928
109k
    if (ctxt->spaceNr == 0)
10929
0
        spacePush(ctxt, -1);
10930
109k
    else if (*ctxt->space == -2)
10931
23.1k
        spacePush(ctxt, -1);
10932
86.7k
    else
10933
86.7k
        spacePush(ctxt, *ctxt->space);
10934
109k
#ifdef LIBXML_SAX1_ENABLED
10935
109k
    if (ctxt->sax2)
10936
109k
#endif /* LIBXML_SAX1_ENABLED */
10937
109k
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
10938
0
#ifdef LIBXML_SAX1_ENABLED
10939
0
    else
10940
0
        name = xmlParseStartTag(ctxt);
10941
109k
#endif /* LIBXML_SAX1_ENABLED */
10942
109k
    if (name == NULL) {
10943
2.97k
        spacePop(ctxt);
10944
2.97k
                    ctxt->instate = XML_PARSER_EOF;
10945
2.97k
                    xmlFinishDocument(ctxt);
10946
2.97k
        goto done;
10947
2.97k
    }
10948
106k
#ifdef LIBXML_VALID_ENABLED
10949
    /*
10950
     * [ VC: Root Element Type ]
10951
     * The Name in the document type declaration must match
10952
     * the element type of the root element.
10953
     */
10954
106k
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
10955
0
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
10956
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
10957
106k
#endif /* LIBXML_VALID_ENABLED */
10958
10959
    /*
10960
     * Check for an Empty Element.
10961
     */
10962
106k
    if ((RAW == '/') && (NXT(1) == '>')) {
10963
3.89k
        SKIP(2);
10964
10965
3.89k
        if (ctxt->sax2) {
10966
3.89k
      if ((ctxt->sax != NULL) &&
10967
3.89k
          (ctxt->sax->endElementNs != NULL) &&
10968
3.89k
          (!ctxt->disableSAX))
10969
3.87k
          ctxt->sax->endElementNs(ctxt->userData, name,
10970
3.87k
                                  prefix, URI);
10971
3.89k
      if (nbNs > 0)
10972
1.73k
          xmlParserNsPop(ctxt, nbNs);
10973
3.89k
#ifdef LIBXML_SAX1_ENABLED
10974
3.89k
        } else {
10975
0
      if ((ctxt->sax != NULL) &&
10976
0
          (ctxt->sax->endElement != NULL) &&
10977
0
          (!ctxt->disableSAX))
10978
0
          ctxt->sax->endElement(ctxt->userData, name);
10979
0
#endif /* LIBXML_SAX1_ENABLED */
10980
0
        }
10981
3.89k
        spacePop(ctxt);
10982
103k
    } else if (RAW == '>') {
10983
69.7k
        NEXT;
10984
69.7k
                    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
10985
69.7k
    } else {
10986
33.3k
        xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED,
10987
33.3k
           "Couldn't find end of Start Tag %s\n",
10988
33.3k
           name);
10989
33.3k
        nodePop(ctxt);
10990
33.3k
        spacePop(ctxt);
10991
33.3k
                    if (nbNs > 0)
10992
6.20k
                        xmlParserNsPop(ctxt, nbNs);
10993
33.3k
    }
10994
10995
106k
                if (ctxt->nameNr == 0)
10996
4.33k
                    ctxt->instate = XML_PARSER_EPILOG;
10997
102k
                else
10998
102k
                    ctxt->instate = XML_PARSER_CONTENT;
10999
106k
                break;
11000
109k
      }
11001
332k
            case XML_PARSER_CONTENT: {
11002
332k
    cur = ctxt->input->cur[0];
11003
11004
332k
    if (cur == '<') {
11005
162k
                    if ((!terminate) && (avail < 2))
11006
1.84k
                        goto done;
11007
160k
        next = ctxt->input->cur[1];
11008
11009
160k
                    if (next == '/') {
11010
5.67k
                        ctxt->instate = XML_PARSER_END_TAG;
11011
5.67k
                        break;
11012
155k
                    } else if (next == '?') {
11013
18.3k
                        if ((!terminate) &&
11014
14.1k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11015
13.3k
                            goto done;
11016
4.98k
                        xmlParsePI(ctxt);
11017
4.98k
                        ctxt->instate = XML_PARSER_CONTENT;
11018
4.98k
                        break;
11019
136k
                    } else if (next == '!') {
11020
41.4k
                        if ((!terminate) && (avail < 3))
11021
579
                            goto done;
11022
40.9k
                        next = ctxt->input->cur[2];
11023
11024
40.9k
                        if (next == '-') {
11025
4.90k
                            if ((!terminate) && (avail < 4))
11026
438
                                goto done;
11027
4.46k
                            if (ctxt->input->cur[3] == '-') {
11028
4.43k
                                if ((!terminate) &&
11029
3.43k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11030
2.28k
                                    goto done;
11031
2.15k
                                xmlParseComment(ctxt);
11032
2.15k
                                ctxt->instate = XML_PARSER_CONTENT;
11033
2.15k
                                break;
11034
4.43k
                            }
11035
36.0k
                        } else if (next == '[') {
11036
35.5k
                            if ((!terminate) && (avail < 9))
11037
698
                                goto done;
11038
34.8k
                            if ((ctxt->input->cur[2] == '[') &&
11039
34.8k
                                (ctxt->input->cur[3] == 'C') &&
11040
34.7k
                                (ctxt->input->cur[4] == 'D') &&
11041
34.7k
                                (ctxt->input->cur[5] == 'A') &&
11042
34.7k
                                (ctxt->input->cur[6] == 'T') &&
11043
34.7k
                                (ctxt->input->cur[7] == 'A') &&
11044
34.7k
                                (ctxt->input->cur[8] == '[')) {
11045
34.6k
                                if ((!terminate) &&
11046
33.2k
                                    (!xmlParseLookupString(ctxt, 9, "]]>", 3)))
11047
32.6k
                                    goto done;
11048
2.02k
                                ctxt->instate = XML_PARSER_CDATA_SECTION;
11049
2.02k
                                xmlParseCDSect(ctxt);
11050
2.02k
                                ctxt->instate = XML_PARSER_CONTENT;
11051
2.02k
                                break;
11052
34.6k
                            }
11053
34.8k
                        }
11054
40.9k
                    }
11055
170k
    } else if (cur == '&') {
11056
23.1k
        if ((!terminate) && (!xmlParseLookupChar(ctxt, ';')))
11057
4.61k
      goto done;
11058
18.5k
        xmlParseReference(ctxt);
11059
18.5k
                    break;
11060
147k
    } else {
11061
        /* TODO Avoid the extra copy, handle directly !!! */
11062
        /*
11063
         * Goal of the following test is:
11064
         *  - minimize calls to the SAX 'character' callback
11065
         *    when they are mergeable
11066
         *  - handle an problem for isBlank when we only parse
11067
         *    a sequence of blank chars and the next one is
11068
         *    not available to check against '<' presence.
11069
         *  - tries to homogenize the differences in SAX
11070
         *    callbacks between the push and pull versions
11071
         *    of the parser.
11072
         */
11073
147k
        if (avail < XML_PARSER_BIG_BUFFER_SIZE) {
11074
47.6k
      if ((!terminate) && (!xmlParseLookupCharData(ctxt)))
11075
13.1k
          goto done;
11076
47.6k
                    }
11077
134k
                    ctxt->checkIndex = 0;
11078
134k
        xmlParseCharDataInternal(ctxt, !terminate);
11079
134k
                    break;
11080
147k
    }
11081
11082
95.8k
                ctxt->instate = XML_PARSER_START_TAG;
11083
95.8k
    break;
11084
332k
      }
11085
7.20k
            case XML_PARSER_END_TAG:
11086
7.20k
    if ((!terminate) && (!xmlParseLookupChar(ctxt, '>')))
11087
1.52k
        goto done;
11088
5.67k
    if (ctxt->sax2) {
11089
5.67k
              xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
11090
5.67k
        nameNsPop(ctxt);
11091
5.67k
    }
11092
0
#ifdef LIBXML_SAX1_ENABLED
11093
0
      else
11094
0
        xmlParseEndTag1(ctxt, 0);
11095
5.67k
#endif /* LIBXML_SAX1_ENABLED */
11096
5.67k
    if (ctxt->nameNr == 0) {
11097
289
        ctxt->instate = XML_PARSER_EPILOG;
11098
5.38k
    } else {
11099
5.38k
        ctxt->instate = XML_PARSER_CONTENT;
11100
5.38k
    }
11101
5.67k
    break;
11102
75.5k
            case XML_PARSER_MISC:
11103
84.9k
            case XML_PARSER_PROLOG:
11104
88.8k
            case XML_PARSER_EPILOG:
11105
88.8k
    SKIP_BLANKS;
11106
88.8k
                avail = ctxt->input->end - ctxt->input->cur;
11107
88.8k
    if (avail < 1)
11108
812
        goto done;
11109
88.0k
    if (ctxt->input->cur[0] == '<') {
11110
84.7k
                    if ((!terminate) && (avail < 2))
11111
687
                        goto done;
11112
84.0k
                    next = ctxt->input->cur[1];
11113
84.0k
                    if (next == '?') {
11114
6.84k
                        if ((!terminate) &&
11115
4.56k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11116
2.71k
                            goto done;
11117
4.12k
                        xmlParsePI(ctxt);
11118
4.12k
                        break;
11119
77.2k
                    } else if (next == '!') {
11120
63.1k
                        if ((!terminate) && (avail < 3))
11121
281
                            goto done;
11122
11123
62.8k
                        if (ctxt->input->cur[2] == '-') {
11124
4.79k
                            if ((!terminate) && (avail < 4))
11125
145
                                goto done;
11126
4.64k
                            if (ctxt->input->cur[3] == '-') {
11127
4.63k
                                if ((!terminate) &&
11128
3.78k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11129
2.94k
                                    goto done;
11130
1.68k
                                xmlParseComment(ctxt);
11131
1.68k
                                break;
11132
4.63k
                            }
11133
58.0k
                        } else if (ctxt->instate == XML_PARSER_MISC) {
11134
58.0k
                            if ((!terminate) && (avail < 9))
11135
5.86k
                                goto done;
11136
52.1k
                            if ((ctxt->input->cur[2] == 'D') &&
11137
52.1k
                                (ctxt->input->cur[3] == 'O') &&
11138
52.1k
                                (ctxt->input->cur[4] == 'C') &&
11139
52.1k
                                (ctxt->input->cur[5] == 'T') &&
11140
52.1k
                                (ctxt->input->cur[6] == 'Y') &&
11141
52.1k
                                (ctxt->input->cur[7] == 'P') &&
11142
52.1k
                                (ctxt->input->cur[8] == 'E')) {
11143
52.1k
                                if ((!terminate) && (!xmlParseLookupGt(ctxt)))
11144
38.5k
                                    goto done;
11145
13.6k
                                ctxt->inSubset = 1;
11146
13.6k
                                xmlParseDocTypeDecl(ctxt);
11147
13.6k
                                if (RAW == '[') {
11148
10.8k
                                    ctxt->instate = XML_PARSER_DTD;
11149
10.8k
                                } else {
11150
2.72k
                                    if (RAW == '>')
11151
2.12k
                                        NEXT;
11152
                                    /*
11153
                                     * Create and update the external subset.
11154
                                     */
11155
2.72k
                                    ctxt->inSubset = 2;
11156
2.72k
                                    if ((ctxt->sax != NULL) &&
11157
2.72k
                                        (!ctxt->disableSAX) &&
11158
2.59k
                                        (ctxt->sax->externalSubset != NULL))
11159
2.59k
                                        ctxt->sax->externalSubset(
11160
2.59k
                                                ctxt->userData,
11161
2.59k
                                                ctxt->intSubName,
11162
2.59k
                                                ctxt->extSubSystem,
11163
2.59k
                                                ctxt->extSubURI);
11164
2.72k
                                    ctxt->inSubset = 0;
11165
2.72k
                                    xmlCleanSpecialAttr(ctxt);
11166
2.72k
                                    ctxt->instate = XML_PARSER_PROLOG;
11167
2.72k
                                }
11168
13.6k
                                break;
11169
52.1k
                            }
11170
52.1k
                        }
11171
62.8k
                    }
11172
84.0k
                }
11173
11174
17.4k
                if (ctxt->instate == XML_PARSER_EPILOG) {
11175
860
                    if (ctxt->errNo == XML_ERR_OK)
11176
24
                        xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL);
11177
860
        ctxt->instate = XML_PARSER_EOF;
11178
860
                    xmlFinishDocument(ctxt);
11179
16.5k
                } else {
11180
16.5k
        ctxt->instate = XML_PARSER_START_TAG;
11181
16.5k
    }
11182
17.4k
    break;
11183
206k
            case XML_PARSER_DTD: {
11184
206k
                if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt)))
11185
196k
                    goto done;
11186
10.0k
    xmlParseInternalSubset(ctxt);
11187
10.0k
    ctxt->inSubset = 2;
11188
10.0k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
11189
8.00k
        (ctxt->sax->externalSubset != NULL))
11190
8.00k
        ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
11191
8.00k
          ctxt->extSubSystem, ctxt->extSubURI);
11192
10.0k
    ctxt->inSubset = 0;
11193
10.0k
    xmlCleanSpecialAttr(ctxt);
11194
10.0k
    ctxt->instate = XML_PARSER_PROLOG;
11195
10.0k
                break;
11196
206k
      }
11197
0
            default:
11198
0
                xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
11199
0
      "PP: internal error\n");
11200
0
    ctxt->instate = XML_PARSER_EOF;
11201
0
    break;
11202
872k
  }
11203
872k
    }
11204
424k
done:
11205
424k
    return(ret);
11206
424k
}
11207
11208
/**
11209
 * Parse a chunk of memory in push parser mode.
11210
 *
11211
 * Assumes that the parser context was initialized with
11212
 * #xmlCreatePushParserCtxt.
11213
 *
11214
 * The last chunk, which will often be empty, must be marked with
11215
 * the `terminate` flag. With the default SAX callbacks, the resulting
11216
 * document will be available in ctxt->myDoc. This pointer will not
11217
 * be freed when calling #xmlFreeParserCtxt and must be freed by the
11218
 * caller. If the document isn't well-formed, it will still be returned
11219
 * in ctxt->myDoc.
11220
 *
11221
 * As an exception, #xmlCtxtResetPush will free the document in
11222
 * ctxt->myDoc. So ctxt->myDoc should be set to NULL after extracting
11223
 * the document.
11224
 *
11225
 * Since 2.14.0, #xmlCtxtGetDocument can be used to retrieve the
11226
 * result document.
11227
 *
11228
 * @param ctxt  an XML parser context
11229
 * @param chunk  chunk of memory
11230
 * @param size  size of chunk in bytes
11231
 * @param terminate  last chunk indicator
11232
 * @returns an xmlParserErrors code (0 on success).
11233
 */
11234
int
11235
xmlParseChunk(xmlParserCtxt *ctxt, const char *chunk, int size,
11236
520k
              int terminate) {
11237
520k
    size_t curBase;
11238
520k
    size_t maxLength;
11239
520k
    size_t pos;
11240
520k
    int end_in_lf = 0;
11241
520k
    int res;
11242
11243
520k
    if ((ctxt == NULL) || (size < 0))
11244
0
        return(XML_ERR_ARGUMENT);
11245
520k
    if ((chunk == NULL) && (size > 0))
11246
0
        return(XML_ERR_ARGUMENT);
11247
520k
    if ((ctxt->input == NULL) || (ctxt->input->buf == NULL))
11248
0
        return(XML_ERR_ARGUMENT);
11249
520k
    if (ctxt->disableSAX != 0)
11250
87.4k
        return(ctxt->errNo);
11251
11252
432k
    ctxt->input->flags |= XML_INPUT_PROGRESSIVE;
11253
432k
    if (ctxt->instate == XML_PARSER_START)
11254
35.4k
        xmlCtxtInitializeLate(ctxt);
11255
432k
    if ((size > 0) && (chunk != NULL) && (!terminate) &&
11256
408k
        (chunk[size - 1] == '\r')) {
11257
1.60k
  end_in_lf = 1;
11258
1.60k
  size--;
11259
1.60k
    }
11260
11261
    /*
11262
     * Also push an empty chunk to make sure that the raw buffer
11263
     * will be flushed if there is an encoder.
11264
     */
11265
432k
    pos = ctxt->input->cur - ctxt->input->base;
11266
432k
    res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
11267
432k
    xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11268
432k
    if (res < 0) {
11269
8.52k
        xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11270
8.52k
        return(ctxt->errNo);
11271
8.52k
    }
11272
11273
424k
    xmlParseTryOrFinish(ctxt, terminate);
11274
11275
424k
    curBase = ctxt->input->cur - ctxt->input->base;
11276
424k
    maxLength = (ctxt->options & XML_PARSE_HUGE) ?
11277
131k
                XML_MAX_HUGE_LENGTH :
11278
424k
                XML_MAX_LOOKUP_LIMIT;
11279
424k
    if (curBase > maxLength) {
11280
0
        xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT,
11281
0
                    "Buffer size limit exceeded, try XML_PARSE_HUGE\n");
11282
0
    }
11283
11284
424k
    if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0))
11285
11.1k
        return(ctxt->errNo);
11286
11287
412k
    if (end_in_lf == 1) {
11288
1.60k
  pos = ctxt->input->cur - ctxt->input->base;
11289
1.60k
  res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r");
11290
1.60k
  xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11291
1.60k
        if (res < 0) {
11292
2
            xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11293
2
            return(ctxt->errNo);
11294
2
        }
11295
1.60k
    }
11296
412k
    if (terminate) {
11297
  /*
11298
   * Check for termination
11299
   */
11300
14.1k
        if ((ctxt->instate != XML_PARSER_EOF) &&
11301
8.59k
            (ctxt->instate != XML_PARSER_EPILOG)) {
11302
6.82k
            if (ctxt->nameNr > 0) {
11303
4.11k
                const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
11304
4.11k
                int line = ctxt->pushTab[ctxt->nameNr - 1].line;
11305
4.11k
                xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
11306
4.11k
                        "Premature end of data in tag %s line %d\n",
11307
4.11k
                        name, line, NULL);
11308
4.11k
            } else if (ctxt->instate == XML_PARSER_START) {
11309
166
                xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
11310
2.54k
            } else {
11311
2.54k
                xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
11312
2.54k
                               "Start tag expected, '<' not found\n");
11313
2.54k
            }
11314
7.30k
        } else {
11315
7.30k
            xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
11316
7.30k
        }
11317
14.1k
  if (ctxt->instate != XML_PARSER_EOF) {
11318
8.40k
            ctxt->instate = XML_PARSER_EOF;
11319
8.40k
            xmlFinishDocument(ctxt);
11320
8.40k
  }
11321
14.1k
    }
11322
412k
    if (ctxt->wellFormed == 0)
11323
247k
  return((xmlParserErrors) ctxt->errNo);
11324
165k
    else
11325
165k
        return(0);
11326
412k
}
11327
11328
/************************************************************************
11329
 *                  *
11330
 *    I/O front end functions to the parser     *
11331
 *                  *
11332
 ************************************************************************/
11333
11334
/**
11335
 * Create a parser context for using the XML parser in push mode.
11336
 * See #xmlParseChunk.
11337
 *
11338
 * Passing an initial chunk is useless and deprecated.
11339
 *
11340
 * The push parser doesn't support recovery mode or the
11341
 * XML_PARSE_NOBLANKS option.
11342
 *
11343
 * `filename` is used as base URI to fetch external entities and for
11344
 * error reports.
11345
 *
11346
 * @param sax  a SAX handler (optional)
11347
 * @param user_data  user data for SAX callbacks (optional)
11348
 * @param chunk  initial chunk (optional, deprecated)
11349
 * @param size  size of initial chunk in bytes
11350
 * @param filename  file name or URI (optional)
11351
 * @returns the new parser context or NULL if a memory allocation
11352
 * failed.
11353
 */
11354
11355
xmlParserCtxt *
11356
xmlCreatePushParserCtxt(xmlSAXHandler *sax, void *user_data,
11357
25.3k
                        const char *chunk, int size, const char *filename) {
11358
25.3k
    xmlParserCtxtPtr ctxt;
11359
25.3k
    xmlParserInputPtr input;
11360
11361
25.3k
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11362
25.3k
    if (ctxt == NULL)
11363
26
  return(NULL);
11364
11365
25.3k
    ctxt->options &= ~XML_PARSE_NODICT;
11366
25.3k
    ctxt->dictNames = 1;
11367
11368
25.3k
    input = xmlNewPushInput(filename, chunk, size);
11369
25.3k
    if (input == NULL) {
11370
15
  xmlFreeParserCtxt(ctxt);
11371
15
  return(NULL);
11372
15
    }
11373
25.3k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11374
6
        xmlFreeInputStream(input);
11375
6
        xmlFreeParserCtxt(ctxt);
11376
6
        return(NULL);
11377
6
    }
11378
11379
25.3k
    return(ctxt);
11380
25.3k
}
11381
#endif /* LIBXML_PUSH_ENABLED */
11382
11383
/**
11384
 * Blocks further parser processing
11385
 *
11386
 * @param ctxt  an XML parser context
11387
 */
11388
void
11389
0
xmlStopParser(xmlParserCtxt *ctxt) {
11390
0
    if (ctxt == NULL)
11391
0
        return;
11392
11393
    /* This stops the parser */
11394
0
    ctxt->disableSAX = 2;
11395
11396
    /*
11397
     * xmlStopParser is often called from error handlers,
11398
     * so we can't raise an error here to avoid infinite
11399
     * loops. Just make sure that an error condition is
11400
     * reported.
11401
     */
11402
0
    if (ctxt->errNo == XML_ERR_OK) {
11403
0
        ctxt->errNo = XML_ERR_USER_STOP;
11404
0
        ctxt->lastError.code = XML_ERR_USER_STOP;
11405
0
        ctxt->wellFormed = 0;
11406
0
    }
11407
0
}
11408
11409
/**
11410
 * Create a parser context for using the XML parser with an existing
11411
 * I/O stream
11412
 *
11413
 * @param sax  a SAX handler (optional)
11414
 * @param user_data  user data for SAX callbacks (optional)
11415
 * @param ioread  an I/O read function
11416
 * @param ioclose  an I/O close function (optional)
11417
 * @param ioctx  an I/O handler
11418
 * @param enc  the charset encoding if known (deprecated)
11419
 * @returns the new parser context or NULL
11420
 */
11421
xmlParserCtxt *
11422
xmlCreateIOParserCtxt(xmlSAXHandler *sax, void *user_data,
11423
                      xmlInputReadCallback ioread,
11424
                      xmlInputCloseCallback ioclose,
11425
0
                      void *ioctx, xmlCharEncoding enc) {
11426
0
    xmlParserCtxtPtr ctxt;
11427
0
    xmlParserInputPtr input;
11428
0
    const char *encoding;
11429
11430
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11431
0
    if (ctxt == NULL)
11432
0
  return(NULL);
11433
11434
0
    encoding = xmlGetCharEncodingName(enc);
11435
0
    input = xmlCtxtNewInputFromIO(ctxt, NULL, ioread, ioclose, ioctx,
11436
0
                                  encoding, 0);
11437
0
    if (input == NULL) {
11438
0
  xmlFreeParserCtxt(ctxt);
11439
0
        return (NULL);
11440
0
    }
11441
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11442
0
        xmlFreeInputStream(input);
11443
0
        xmlFreeParserCtxt(ctxt);
11444
0
        return(NULL);
11445
0
    }
11446
11447
0
    return(ctxt);
11448
0
}
11449
11450
#ifdef LIBXML_VALID_ENABLED
11451
/************************************************************************
11452
 *                  *
11453
 *    Front ends when parsing a DTD       *
11454
 *                  *
11455
 ************************************************************************/
11456
11457
/**
11458
 * Parse a DTD.
11459
 *
11460
 * Option XML_PARSE_DTDLOAD should be enabled in the parser context
11461
 * to make external entities work.
11462
 *
11463
 * @since 2.14.0
11464
 *
11465
 * @param ctxt  a parser context
11466
 * @param input  a parser input
11467
 * @param publicId  public ID of the DTD (optional)
11468
 * @param systemId  system ID of the DTD (optional)
11469
 * @returns the resulting xmlDtd or NULL in case of error.
11470
 * `input` will be freed by the function in any case.
11471
 */
11472
xmlDtd *
11473
xmlCtxtParseDtd(xmlParserCtxt *ctxt, xmlParserInput *input,
11474
0
                const xmlChar *publicId, const xmlChar *systemId) {
11475
0
    xmlDtdPtr ret = NULL;
11476
11477
0
    if ((ctxt == NULL) || (input == NULL)) {
11478
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
11479
0
        xmlFreeInputStream(input);
11480
0
        return(NULL);
11481
0
    }
11482
11483
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11484
0
        xmlFreeInputStream(input);
11485
0
        return(NULL);
11486
0
    }
11487
11488
0
    if (publicId == NULL)
11489
0
        publicId = BAD_CAST "none";
11490
0
    if (systemId == NULL)
11491
0
        systemId = BAD_CAST "none";
11492
11493
0
    ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0");
11494
0
    if (ctxt->myDoc == NULL) {
11495
0
        xmlErrMemory(ctxt);
11496
0
        goto error;
11497
0
    }
11498
0
    ctxt->myDoc->properties = XML_DOC_INTERNAL;
11499
0
    ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none",
11500
0
                                       publicId, systemId);
11501
0
    if (ctxt->myDoc->extSubset == NULL) {
11502
0
        xmlErrMemory(ctxt);
11503
0
        xmlFreeDoc(ctxt->myDoc);
11504
0
        goto error;
11505
0
    }
11506
11507
0
    xmlParseExternalSubset(ctxt, publicId, systemId);
11508
11509
0
    if (ctxt->wellFormed) {
11510
0
        ret = ctxt->myDoc->extSubset;
11511
0
        ctxt->myDoc->extSubset = NULL;
11512
0
        if (ret != NULL) {
11513
0
            xmlNodePtr tmp;
11514
11515
0
            ret->doc = NULL;
11516
0
            tmp = ret->children;
11517
0
            while (tmp != NULL) {
11518
0
                tmp->doc = NULL;
11519
0
                tmp = tmp->next;
11520
0
            }
11521
0
        }
11522
0
    } else {
11523
0
        ret = NULL;
11524
0
    }
11525
0
    xmlFreeDoc(ctxt->myDoc);
11526
0
    ctxt->myDoc = NULL;
11527
11528
0
error:
11529
0
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
11530
11531
0
    return(ret);
11532
0
}
11533
11534
/**
11535
 * Load and parse a DTD
11536
 *
11537
 * @deprecated Use #xmlCtxtParseDtd.
11538
 *
11539
 * @param sax  the SAX handler block or NULL
11540
 * @param input  an Input Buffer
11541
 * @param enc  the charset encoding if known
11542
 * @returns the resulting xmlDtd or NULL in case of error.
11543
 * `input` will be freed by the function in any case.
11544
 */
11545
11546
xmlDtd *
11547
xmlIOParseDTD(xmlSAXHandler *sax, xmlParserInputBuffer *input,
11548
0
        xmlCharEncoding enc) {
11549
0
    xmlDtdPtr ret = NULL;
11550
0
    xmlParserCtxtPtr ctxt;
11551
0
    xmlParserInputPtr pinput = NULL;
11552
11553
0
    if (input == NULL)
11554
0
  return(NULL);
11555
11556
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11557
0
    if (ctxt == NULL) {
11558
0
        xmlFreeParserInputBuffer(input);
11559
0
  return(NULL);
11560
0
    }
11561
0
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11562
11563
    /*
11564
     * generate a parser input from the I/O handler
11565
     */
11566
11567
0
    pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE);
11568
0
    if (pinput == NULL) {
11569
0
        xmlFreeParserInputBuffer(input);
11570
0
  xmlFreeParserCtxt(ctxt);
11571
0
  return(NULL);
11572
0
    }
11573
11574
0
    if (enc != XML_CHAR_ENCODING_NONE) {
11575
0
        xmlSwitchEncoding(ctxt, enc);
11576
0
    }
11577
11578
0
    ret = xmlCtxtParseDtd(ctxt, pinput, NULL, NULL);
11579
11580
0
    xmlFreeParserCtxt(ctxt);
11581
0
    return(ret);
11582
0
}
11583
11584
/**
11585
 * Load and parse an external subset.
11586
 *
11587
 * @deprecated Use #xmlCtxtParseDtd.
11588
 *
11589
 * @param sax  the SAX handler block
11590
 * @param publicId  public identifier of the DTD (optional)
11591
 * @param systemId  system identifier (URL) of the DTD
11592
 * @returns the resulting xmlDtd or NULL in case of error.
11593
 */
11594
11595
xmlDtd *
11596
xmlSAXParseDTD(xmlSAXHandler *sax, const xmlChar *publicId,
11597
0
               const xmlChar *systemId) {
11598
0
    xmlDtdPtr ret = NULL;
11599
0
    xmlParserCtxtPtr ctxt;
11600
0
    xmlParserInputPtr input = NULL;
11601
0
    xmlChar* systemIdCanonic;
11602
11603
0
    if ((publicId == NULL) && (systemId == NULL)) return(NULL);
11604
11605
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
11606
0
    if (ctxt == NULL) {
11607
0
  return(NULL);
11608
0
    }
11609
0
    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
11610
11611
    /*
11612
     * Canonicalise the system ID
11613
     */
11614
0
    systemIdCanonic = xmlCanonicPath(systemId);
11615
0
    if ((systemId != NULL) && (systemIdCanonic == NULL)) {
11616
0
  xmlFreeParserCtxt(ctxt);
11617
0
  return(NULL);
11618
0
    }
11619
11620
    /*
11621
     * Ask the Entity resolver to load the damn thing
11622
     */
11623
11624
0
    if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL))
11625
0
  input = ctxt->sax->resolveEntity(ctxt->userData, publicId,
11626
0
                                   systemIdCanonic);
11627
0
    if (input == NULL) {
11628
0
  xmlFreeParserCtxt(ctxt);
11629
0
  if (systemIdCanonic != NULL)
11630
0
      xmlFree(systemIdCanonic);
11631
0
  return(NULL);
11632
0
    }
11633
11634
0
    if (input->filename == NULL)
11635
0
  input->filename = (char *) systemIdCanonic;
11636
0
    else
11637
0
  xmlFree(systemIdCanonic);
11638
11639
0
    ret = xmlCtxtParseDtd(ctxt, input, publicId, systemId);
11640
11641
0
    xmlFreeParserCtxt(ctxt);
11642
0
    return(ret);
11643
0
}
11644
11645
11646
/**
11647
 * Load and parse an external subset.
11648
 *
11649
 * @param publicId  public identifier of the DTD (optional)
11650
 * @param systemId  system identifier (URL) of the DTD
11651
 * @returns the resulting xmlDtd or NULL in case of error.
11652
 */
11653
11654
xmlDtd *
11655
0
xmlParseDTD(const xmlChar *publicId, const xmlChar *systemId) {
11656
0
    return(xmlSAXParseDTD(NULL, publicId, systemId));
11657
0
}
11658
#endif /* LIBXML_VALID_ENABLED */
11659
11660
/************************************************************************
11661
 *                  *
11662
 *    Front ends when parsing an Entity     *
11663
 *                  *
11664
 ************************************************************************/
11665
11666
static xmlNodePtr
11667
xmlCtxtParseContentInternal(xmlParserCtxtPtr ctxt, xmlParserInputPtr input,
11668
75.5k
                            int hasTextDecl, int buildTree) {
11669
75.5k
    xmlNodePtr root = NULL;
11670
75.5k
    xmlNodePtr list = NULL;
11671
75.5k
    xmlChar *rootName = BAD_CAST "#root";
11672
75.5k
    int result;
11673
11674
75.5k
    if (buildTree) {
11675
75.5k
        root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL);
11676
75.5k
        if (root == NULL) {
11677
28
            xmlErrMemory(ctxt);
11678
28
            goto error;
11679
28
        }
11680
75.5k
    }
11681
11682
75.5k
    if (xmlCtxtPushInput(ctxt, input) < 0)
11683
18
        goto error;
11684
11685
75.5k
    nameNsPush(ctxt, rootName, NULL, NULL, 0, 0);
11686
75.5k
    spacePush(ctxt, -1);
11687
11688
75.5k
    if (buildTree)
11689
75.5k
        nodePush(ctxt, root);
11690
11691
75.5k
    if (hasTextDecl) {
11692
2.95k
        xmlDetectEncoding(ctxt);
11693
11694
        /*
11695
         * Parse a possible text declaration first
11696
         */
11697
2.95k
        if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
11698
401
            (IS_BLANK_CH(NXT(5)))) {
11699
389
            xmlParseTextDecl(ctxt);
11700
            /*
11701
             * An XML-1.0 document can't reference an entity not XML-1.0
11702
             */
11703
389
            if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
11704
215
                (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
11705
5
                xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
11706
5
                               "Version mismatch between document and "
11707
5
                               "entity\n");
11708
5
            }
11709
389
        }
11710
2.95k
    }
11711
11712
75.5k
    xmlParseContentInternal(ctxt);
11713
11714
75.5k
    if (ctxt->input->cur < ctxt->input->end)
11715
437
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
11716
11717
75.5k
    if ((ctxt->wellFormed) ||
11718
74.7k
        ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) {
11719
74.7k
        if (root != NULL) {
11720
74.7k
            xmlNodePtr cur;
11721
11722
            /*
11723
             * Unlink newly created node list.
11724
             */
11725
74.7k
            list = root->children;
11726
74.7k
            root->children = NULL;
11727
74.7k
            root->last = NULL;
11728
105k
            for (cur = list; cur != NULL; cur = cur->next)
11729
30.8k
                cur->parent = NULL;
11730
74.7k
        }
11731
74.7k
    }
11732
11733
    /*
11734
     * Read the rest of the stream in case of errors. We want
11735
     * to account for the whole entity size.
11736
     */
11737
75.7k
    do {
11738
75.7k
        ctxt->input->cur = ctxt->input->end;
11739
75.7k
        xmlParserShrink(ctxt);
11740
75.7k
        result = xmlParserGrow(ctxt);
11741
75.7k
    } while (result > 0);
11742
11743
75.5k
    if (buildTree)
11744
75.5k
        nodePop(ctxt);
11745
11746
75.5k
    namePop(ctxt);
11747
75.5k
    spacePop(ctxt);
11748
11749
75.5k
    xmlCtxtPopInput(ctxt);
11750
11751
75.5k
error:
11752
75.5k
    xmlFreeNode(root);
11753
11754
75.5k
    return(list);
11755
75.5k
}
11756
11757
static void
11758
76.5k
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
11759
76.5k
    xmlParserInputPtr input;
11760
76.5k
    xmlNodePtr list;
11761
76.5k
    unsigned long consumed;
11762
76.5k
    int isExternal;
11763
76.5k
    int buildTree;
11764
76.5k
    int oldMinNsIndex;
11765
76.5k
    int oldNodelen, oldNodemem;
11766
11767
76.5k
    isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY);
11768
76.5k
    buildTree = (ctxt->node != NULL);
11769
11770
    /*
11771
     * Recursion check
11772
     */
11773
76.5k
    if (ent->flags & XML_ENT_EXPANDING) {
11774
10
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
11775
10
        goto error;
11776
10
    }
11777
11778
    /*
11779
     * Load entity
11780
     */
11781
76.5k
    input = xmlNewEntityInputStream(ctxt, ent);
11782
76.5k
    if (input == NULL)
11783
921
        goto error;
11784
11785
    /*
11786
     * When building a tree, we need to limit the scope of namespace
11787
     * declarations, so that entities don't reference xmlNs structs
11788
     * from the parent of a reference.
11789
     */
11790
75.5k
    oldMinNsIndex = ctxt->nsdb->minNsIndex;
11791
75.5k
    if (buildTree)
11792
75.5k
        ctxt->nsdb->minNsIndex = ctxt->nsNr;
11793
11794
75.5k
    oldNodelen = ctxt->nodelen;
11795
75.5k
    oldNodemem = ctxt->nodemem;
11796
75.5k
    ctxt->nodelen = 0;
11797
75.5k
    ctxt->nodemem = 0;
11798
11799
    /*
11800
     * Parse content
11801
     *
11802
     * This initiates a recursive call chain:
11803
     *
11804
     * - xmlCtxtParseContentInternal
11805
     * - xmlParseContentInternal
11806
     * - xmlParseReference
11807
     * - xmlCtxtParseEntity
11808
     *
11809
     * The nesting depth is limited by the maximum number of inputs,
11810
     * see xmlCtxtPushInput.
11811
     *
11812
     * It's possible to make this non-recursive (minNsIndex must be
11813
     * stored in the input struct) at the expense of code readability.
11814
     */
11815
11816
75.5k
    ent->flags |= XML_ENT_EXPANDING;
11817
11818
75.5k
    list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree);
11819
11820
75.5k
    ent->flags &= ~XML_ENT_EXPANDING;
11821
11822
75.5k
    ctxt->nsdb->minNsIndex = oldMinNsIndex;
11823
75.5k
    ctxt->nodelen = oldNodelen;
11824
75.5k
    ctxt->nodemem = oldNodemem;
11825
11826
    /*
11827
     * Entity size accounting
11828
     */
11829
75.5k
    consumed = input->consumed;
11830
75.5k
    xmlSaturatedAddSizeT(&consumed, input->end - input->base);
11831
11832
75.5k
    if ((ent->flags & XML_ENT_CHECKED) == 0)
11833
4.32k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
11834
11835
75.5k
    if ((ent->flags & XML_ENT_PARSED) == 0) {
11836
4.32k
        if (isExternal)
11837
2.74k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
11838
11839
4.32k
        ent->children = list;
11840
11841
35.1k
        while (list != NULL) {
11842
30.8k
            list->parent = (xmlNodePtr) ent;
11843
11844
            /*
11845
             * Downstream code like the nginx xslt module can set
11846
             * ctxt->myDoc->extSubset to a separate DTD, so the entity
11847
             * might have a different or a NULL document.
11848
             */
11849
30.8k
            if (list->doc != ent->doc)
11850
0
                xmlSetTreeDoc(list, ent->doc);
11851
11852
30.8k
            if (list->next == NULL)
11853
3.27k
                ent->last = list;
11854
30.8k
            list = list->next;
11855
30.8k
        }
11856
71.2k
    } else {
11857
71.2k
        xmlFreeNodeList(list);
11858
71.2k
    }
11859
11860
75.5k
    xmlFreeInputStream(input);
11861
11862
76.5k
error:
11863
76.5k
    ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED;
11864
76.5k
}
11865
11866
/**
11867
 * Parse an external general entity within an existing parsing context
11868
 * An external general parsed entity is well-formed if it matches the
11869
 * production labeled extParsedEnt.
11870
 *
11871
 *     [78] extParsedEnt ::= TextDecl? content
11872
 *
11873
 * @param ctxt  the existing parsing context
11874
 * @param URL  the URL for the entity to load
11875
 * @param ID  the System ID for the entity to load
11876
 * @param listOut  the return value for the set of parsed nodes
11877
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11878
 *    the parser error code otherwise
11879
 */
11880
11881
int
11882
xmlParseCtxtExternalEntity(xmlParserCtxt *ctxt, const xmlChar *URL,
11883
0
                           const xmlChar *ID, xmlNode **listOut) {
11884
0
    xmlParserInputPtr input;
11885
0
    xmlNodePtr list;
11886
11887
0
    if (listOut != NULL)
11888
0
        *listOut = NULL;
11889
11890
0
    if (ctxt == NULL)
11891
0
        return(XML_ERR_ARGUMENT);
11892
11893
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
11894
0
                            XML_RESOURCE_GENERAL_ENTITY);
11895
0
    if (input == NULL)
11896
0
        return(ctxt->errNo);
11897
11898
0
    xmlCtxtInitializeLate(ctxt);
11899
11900
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1);
11901
0
    if (listOut != NULL)
11902
0
        *listOut = list;
11903
0
    else
11904
0
        xmlFreeNodeList(list);
11905
11906
0
    xmlFreeInputStream(input);
11907
0
    return(ctxt->errNo);
11908
0
}
11909
11910
#ifdef LIBXML_SAX1_ENABLED
11911
/**
11912
 * Parse an external general entity
11913
 * An external general parsed entity is well-formed if it matches the
11914
 * production labeled extParsedEnt.
11915
 *
11916
 * This function uses deprecated global variables to set parser options
11917
 * which default to XML_PARSE_NODICT.
11918
 *
11919
 * @deprecated Use #xmlParseCtxtExternalEntity.
11920
 *
11921
 *     [78] extParsedEnt ::= TextDecl? content
11922
 *
11923
 * @param doc  the document the chunk pertains to
11924
 * @param sax  the SAX handler block (possibly NULL)
11925
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11926
 * @param depth  Used for loop detection, use 0
11927
 * @param URL  the URL for the entity to load
11928
 * @param ID  the System ID for the entity to load
11929
 * @param list  the return value for the set of parsed nodes
11930
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11931
 *    the parser error code otherwise
11932
 */
11933
11934
int
11935
xmlParseExternalEntity(xmlDoc *doc, xmlSAXHandler *sax, void *user_data,
11936
0
    int depth, const xmlChar *URL, const xmlChar *ID, xmlNode **list) {
11937
0
    xmlParserCtxtPtr ctxt;
11938
0
    int ret;
11939
11940
0
    if (list != NULL)
11941
0
        *list = NULL;
11942
11943
0
    if (doc == NULL)
11944
0
        return(XML_ERR_ARGUMENT);
11945
11946
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11947
0
    if (ctxt == NULL)
11948
0
        return(XML_ERR_NO_MEMORY);
11949
11950
0
    ctxt->depth = depth;
11951
0
    ctxt->myDoc = doc;
11952
0
    ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list);
11953
11954
0
    xmlFreeParserCtxt(ctxt);
11955
0
    return(ret);
11956
0
}
11957
11958
/**
11959
 * Parse a well-balanced chunk of an XML document
11960
 * called by the parser
11961
 * The allowed sequence for the Well Balanced Chunk is the one defined by
11962
 * the content production in the XML grammar:
11963
 *
11964
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
11965
 *                       Comment)*
11966
 *
11967
 * This function uses deprecated global variables to set parser options
11968
 * which default to XML_PARSE_NODICT.
11969
 *
11970
 * @param doc  the document the chunk pertains to (must not be NULL)
11971
 * @param sax  the SAX handler block (possibly NULL)
11972
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11973
 * @param depth  Used for loop detection, use 0
11974
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
11975
 * @param lst  the return value for the set of parsed nodes
11976
 * @returns 0 if the chunk is well balanced, -1 in case of args problem and
11977
 *    the parser error code otherwise
11978
 */
11979
11980
int
11981
xmlParseBalancedChunkMemory(xmlDoc *doc, xmlSAXHandler *sax,
11982
0
     void *user_data, int depth, const xmlChar *string, xmlNode **lst) {
11983
0
    return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data,
11984
0
                                                depth, string, lst, 0 );
11985
0
}
11986
#endif /* LIBXML_SAX1_ENABLED */
11987
11988
/**
11989
 * Parse a well-balanced chunk of XML matching the 'content' production.
11990
 *
11991
 * Namespaces in scope of `node` and entities of `node`'s document are
11992
 * recognized. When validating, the DTD of `node`'s document is used.
11993
 *
11994
 * Always consumes `input` even in error case.
11995
 *
11996
 * @since 2.14.0
11997
 *
11998
 * @param ctxt  parser context
11999
 * @param input  parser input
12000
 * @param node  target node or document
12001
 * @param hasTextDecl  whether to parse text declaration
12002
 * @returns a node list or NULL in case of error.
12003
 */
12004
xmlNode *
12005
xmlCtxtParseContent(xmlParserCtxt *ctxt, xmlParserInput *input,
12006
0
                    xmlNode *node, int hasTextDecl) {
12007
0
    xmlDocPtr doc;
12008
0
    xmlNodePtr cur, list = NULL;
12009
0
    int nsnr = 0;
12010
0
    xmlDictPtr oldDict;
12011
0
    int oldOptions, oldDictNames, oldLoadSubset;
12012
12013
0
    if ((ctxt == NULL) || (input == NULL) || (node == NULL)) {
12014
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12015
0
        goto exit;
12016
0
    }
12017
12018
0
    doc = node->doc;
12019
0
    if (doc == NULL) {
12020
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12021
0
        goto exit;
12022
0
    }
12023
12024
0
    switch (node->type) {
12025
0
        case XML_ELEMENT_NODE:
12026
0
        case XML_DOCUMENT_NODE:
12027
0
        case XML_HTML_DOCUMENT_NODE:
12028
0
            break;
12029
12030
0
        case XML_ATTRIBUTE_NODE:
12031
0
        case XML_TEXT_NODE:
12032
0
        case XML_CDATA_SECTION_NODE:
12033
0
        case XML_ENTITY_REF_NODE:
12034
0
        case XML_PI_NODE:
12035
0
        case XML_COMMENT_NODE:
12036
0
            for (cur = node->parent; cur != NULL; cur = node->parent) {
12037
0
                if ((cur->type == XML_ELEMENT_NODE) ||
12038
0
                    (cur->type == XML_DOCUMENT_NODE) ||
12039
0
                    (cur->type == XML_HTML_DOCUMENT_NODE)) {
12040
0
                    node = cur;
12041
0
                    break;
12042
0
                }
12043
0
            }
12044
0
            break;
12045
12046
0
        default:
12047
0
            xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12048
0
            goto exit;
12049
0
    }
12050
12051
0
    xmlCtxtReset(ctxt);
12052
12053
0
    oldDict = ctxt->dict;
12054
0
    oldOptions = ctxt->options;
12055
0
    oldDictNames = ctxt->dictNames;
12056
0
    oldLoadSubset = ctxt->loadsubset;
12057
12058
    /*
12059
     * Use input doc's dict if present, else assure XML_PARSE_NODICT is set.
12060
     */
12061
0
    if (doc->dict != NULL) {
12062
0
        ctxt->dict = doc->dict;
12063
0
    } else {
12064
0
        ctxt->options |= XML_PARSE_NODICT;
12065
0
        ctxt->dictNames = 0;
12066
0
    }
12067
12068
    /*
12069
     * Disable IDs
12070
     */
12071
0
    ctxt->loadsubset |= XML_SKIP_IDS;
12072
0
    ctxt->options |= XML_PARSE_SKIP_IDS;
12073
12074
0
    ctxt->myDoc = doc;
12075
12076
0
#ifdef LIBXML_HTML_ENABLED
12077
0
    if (ctxt->html) {
12078
        /*
12079
         * When parsing in context, it makes no sense to add implied
12080
         * elements like html/body/etc...
12081
         */
12082
0
        ctxt->options |= HTML_PARSE_NOIMPLIED;
12083
12084
0
        list = htmlCtxtParseContentInternal(ctxt, input);
12085
0
    } else
12086
0
#endif
12087
0
    {
12088
0
        xmlCtxtInitializeLate(ctxt);
12089
12090
        /*
12091
         * initialize the SAX2 namespaces stack
12092
         */
12093
0
        cur = node;
12094
0
        while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) {
12095
0
            xmlNsPtr ns = cur->nsDef;
12096
0
            xmlHashedString hprefix, huri;
12097
12098
0
            while (ns != NULL) {
12099
0
                hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1);
12100
0
                huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1);
12101
0
                if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0)
12102
0
                    nsnr++;
12103
0
                ns = ns->next;
12104
0
            }
12105
0
            cur = cur->parent;
12106
0
        }
12107
12108
0
        list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1);
12109
12110
0
        if (nsnr > 0)
12111
0
            xmlParserNsPop(ctxt, nsnr);
12112
0
    }
12113
12114
0
    ctxt->dict = oldDict;
12115
0
    ctxt->options = oldOptions;
12116
0
    ctxt->dictNames = oldDictNames;
12117
0
    ctxt->loadsubset = oldLoadSubset;
12118
0
    ctxt->myDoc = NULL;
12119
0
    ctxt->node = NULL;
12120
12121
0
exit:
12122
0
    xmlFreeInputStream(input);
12123
0
    return(list);
12124
0
}
12125
12126
/**
12127
 * Parse a well-balanced chunk of an XML document
12128
 * within the context (DTD, namespaces, etc ...) of the given node.
12129
 *
12130
 * The allowed sequence for the data is a Well Balanced Chunk defined by
12131
 * the content production in the XML grammar:
12132
 *
12133
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12134
 *                       Comment)*
12135
 *
12136
 * This function assumes the encoding of `node`'s document which is
12137
 * typically not what you want. A better alternative is
12138
 * #xmlCtxtParseContent.
12139
 *
12140
 * @param node  the context node
12141
 * @param data  the input string
12142
 * @param datalen  the input string length in bytes
12143
 * @param options  a combination of xmlParserOption
12144
 * @param listOut  the return value for the set of parsed nodes
12145
 * @returns XML_ERR_OK if the chunk is well balanced, and the parser
12146
 * error code otherwise
12147
 */
12148
xmlParserErrors
12149
xmlParseInNodeContext(xmlNode *node, const char *data, int datalen,
12150
0
                      int options, xmlNode **listOut) {
12151
0
    xmlParserCtxtPtr ctxt;
12152
0
    xmlParserInputPtr input;
12153
0
    xmlDocPtr doc;
12154
0
    xmlNodePtr list;
12155
0
    xmlParserErrors ret;
12156
12157
0
    if (listOut == NULL)
12158
0
        return(XML_ERR_INTERNAL_ERROR);
12159
0
    *listOut = NULL;
12160
12161
0
    if ((node == NULL) || (data == NULL) || (datalen < 0))
12162
0
        return(XML_ERR_INTERNAL_ERROR);
12163
12164
0
    doc = node->doc;
12165
0
    if (doc == NULL)
12166
0
        return(XML_ERR_INTERNAL_ERROR);
12167
12168
0
#ifdef LIBXML_HTML_ENABLED
12169
0
    if (doc->type == XML_HTML_DOCUMENT_NODE) {
12170
0
        ctxt = htmlNewParserCtxt();
12171
0
    }
12172
0
    else
12173
0
#endif
12174
0
        ctxt = xmlNewParserCtxt();
12175
12176
0
    if (ctxt == NULL)
12177
0
        return(XML_ERR_NO_MEMORY);
12178
12179
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen,
12180
0
                                      (const char *) doc->encoding,
12181
0
                                      XML_INPUT_BUF_STATIC);
12182
0
    if (input == NULL) {
12183
0
        xmlFreeParserCtxt(ctxt);
12184
0
        return(XML_ERR_NO_MEMORY);
12185
0
    }
12186
12187
0
    xmlCtxtUseOptions(ctxt, options);
12188
12189
0
    list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0);
12190
12191
0
    if (list == NULL) {
12192
0
        ret = ctxt->errNo;
12193
0
        if (ret == XML_ERR_ARGUMENT)
12194
0
            ret = XML_ERR_INTERNAL_ERROR;
12195
0
    } else {
12196
0
        ret = XML_ERR_OK;
12197
0
        *listOut = list;
12198
0
    }
12199
12200
0
    xmlFreeParserCtxt(ctxt);
12201
12202
0
    return(ret);
12203
0
}
12204
12205
#ifdef LIBXML_SAX1_ENABLED
12206
/**
12207
 * Parse a well-balanced chunk of an XML document
12208
 *
12209
 * The allowed sequence for the Well Balanced Chunk is the one defined by
12210
 * the content production in the XML grammar:
12211
 *
12212
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12213
 *                       Comment)*
12214
 *
12215
 * In case recover is set to 1, the nodelist will not be empty even if
12216
 * the parsed chunk is not well balanced, assuming the parsing succeeded to
12217
 * some extent.
12218
 *
12219
 * This function uses deprecated global variables to set parser options
12220
 * which default to XML_PARSE_NODICT.
12221
 *
12222
 * @param doc  the document the chunk pertains to (must not be NULL)
12223
 * @param sax  the SAX handler block (possibly NULL)
12224
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
12225
 * @param depth  Used for loop detection, use 0
12226
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
12227
 * @param listOut  the return value for the set of parsed nodes
12228
 * @param recover  return nodes even if the data is broken (use 0)
12229
 * @returns 0 if the chunk is well balanced, or thehe parser error code
12230
 * otherwise.
12231
 */
12232
int
12233
xmlParseBalancedChunkMemoryRecover(xmlDoc *doc, xmlSAXHandler *sax,
12234
     void *user_data, int depth, const xmlChar *string, xmlNode **listOut,
12235
0
     int recover) {
12236
0
    xmlParserCtxtPtr ctxt;
12237
0
    xmlParserInputPtr input;
12238
0
    xmlNodePtr list;
12239
0
    int ret;
12240
12241
0
    if (listOut != NULL)
12242
0
        *listOut = NULL;
12243
12244
0
    if (string == NULL)
12245
0
        return(XML_ERR_ARGUMENT);
12246
12247
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
12248
0
    if (ctxt == NULL)
12249
0
        return(XML_ERR_NO_MEMORY);
12250
12251
0
    xmlCtxtInitializeLate(ctxt);
12252
12253
0
    ctxt->depth = depth;
12254
0
    ctxt->myDoc = doc;
12255
0
    if (recover) {
12256
0
        ctxt->options |= XML_PARSE_RECOVER;
12257
0
        ctxt->recovery = 1;
12258
0
    }
12259
12260
0
    input = xmlNewStringInputStream(ctxt, string);
12261
0
    if (input == NULL) {
12262
0
        ret = ctxt->errNo;
12263
0
        goto error;
12264
0
    }
12265
12266
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1);
12267
0
    if (listOut != NULL)
12268
0
        *listOut = list;
12269
0
    else
12270
0
        xmlFreeNodeList(list);
12271
12272
0
    if (!ctxt->wellFormed)
12273
0
        ret = ctxt->errNo;
12274
0
    else
12275
0
        ret = XML_ERR_OK;
12276
12277
0
error:
12278
0
    xmlFreeInputStream(input);
12279
0
    xmlFreeParserCtxt(ctxt);
12280
0
    return(ret);
12281
0
}
12282
12283
/**
12284
 * Parse an XML external entity out of context and build a tree.
12285
 * It use the given SAX function block to handle the parsing callback.
12286
 * If sax is NULL, fallback to the default DOM tree building routines.
12287
 *
12288
 * @deprecated Don't use.
12289
 *
12290
 *     [78] extParsedEnt ::= TextDecl? content
12291
 *
12292
 * This correspond to a "Well Balanced" chunk
12293
 *
12294
 * This function uses deprecated global variables to set parser options
12295
 * which default to XML_PARSE_NODICT.
12296
 *
12297
 * @param sax  the SAX handler block
12298
 * @param filename  the filename
12299
 * @returns the resulting document tree
12300
 */
12301
12302
xmlDoc *
12303
0
xmlSAXParseEntity(xmlSAXHandler *sax, const char *filename) {
12304
0
    xmlDocPtr ret;
12305
0
    xmlParserCtxtPtr ctxt;
12306
12307
0
    ctxt = xmlCreateFileParserCtxt(filename);
12308
0
    if (ctxt == NULL) {
12309
0
  return(NULL);
12310
0
    }
12311
0
    if (sax != NULL) {
12312
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12313
0
            *ctxt->sax = *sax;
12314
0
        } else {
12315
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12316
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12317
0
        }
12318
0
        ctxt->userData = NULL;
12319
0
    }
12320
12321
0
    xmlParseExtParsedEnt(ctxt);
12322
12323
0
    if (ctxt->wellFormed) {
12324
0
  ret = ctxt->myDoc;
12325
0
    } else {
12326
0
        ret = NULL;
12327
0
        xmlFreeDoc(ctxt->myDoc);
12328
0
    }
12329
12330
0
    xmlFreeParserCtxt(ctxt);
12331
12332
0
    return(ret);
12333
0
}
12334
12335
/**
12336
 * Parse an XML external entity out of context and build a tree.
12337
 *
12338
 *     [78] extParsedEnt ::= TextDecl? content
12339
 *
12340
 * This correspond to a "Well Balanced" chunk
12341
 *
12342
 * This function uses deprecated global variables to set parser options
12343
 * which default to XML_PARSE_NODICT.
12344
 *
12345
 * @deprecated Don't use.
12346
 *
12347
 * @param filename  the filename
12348
 * @returns the resulting document tree
12349
 */
12350
12351
xmlDoc *
12352
0
xmlParseEntity(const char *filename) {
12353
0
    return(xmlSAXParseEntity(NULL, filename));
12354
0
}
12355
#endif /* LIBXML_SAX1_ENABLED */
12356
12357
/**
12358
 * Create a parser context for an external entity
12359
 * Automatic support for ZLIB/Compress compressed document is provided
12360
 * by default if found at compile-time.
12361
 *
12362
 * @deprecated Don't use.
12363
 *
12364
 * @param URL  the entity URL
12365
 * @param ID  the entity PUBLIC ID
12366
 * @param base  a possible base for the target URI
12367
 * @returns the new parser context or NULL
12368
 */
12369
xmlParserCtxt *
12370
xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID,
12371
0
                    const xmlChar *base) {
12372
0
    xmlParserCtxtPtr ctxt;
12373
0
    xmlParserInputPtr input;
12374
0
    xmlChar *uri = NULL;
12375
12376
0
    ctxt = xmlNewParserCtxt();
12377
0
    if (ctxt == NULL)
12378
0
  return(NULL);
12379
12380
0
    if (base != NULL) {
12381
0
        if (xmlBuildURISafe(URL, base, &uri) < 0)
12382
0
            goto error;
12383
0
        if (uri != NULL)
12384
0
            URL = uri;
12385
0
    }
12386
12387
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
12388
0
                            XML_RESOURCE_UNKNOWN);
12389
0
    if (input == NULL)
12390
0
        goto error;
12391
12392
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12393
0
        xmlFreeInputStream(input);
12394
0
        goto error;
12395
0
    }
12396
12397
0
    xmlFree(uri);
12398
0
    return(ctxt);
12399
12400
0
error:
12401
0
    xmlFree(uri);
12402
0
    xmlFreeParserCtxt(ctxt);
12403
0
    return(NULL);
12404
0
}
12405
12406
/************************************************************************
12407
 *                  *
12408
 *    Front ends when parsing from a file     *
12409
 *                  *
12410
 ************************************************************************/
12411
12412
/**
12413
 * Create a parser context for a file or URL content.
12414
 * Automatic support for ZLIB/Compress compressed document is provided
12415
 * by default if found at compile-time and for file accesses
12416
 *
12417
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12418
 *
12419
 * @param filename  the filename or URL
12420
 * @param options  a combination of xmlParserOption
12421
 * @returns the new parser context or NULL
12422
 */
12423
xmlParserCtxt *
12424
xmlCreateURLParserCtxt(const char *filename, int options)
12425
0
{
12426
0
    xmlParserCtxtPtr ctxt;
12427
0
    xmlParserInputPtr input;
12428
12429
0
    ctxt = xmlNewParserCtxt();
12430
0
    if (ctxt == NULL)
12431
0
  return(NULL);
12432
12433
0
    xmlCtxtUseOptions(ctxt, options);
12434
12435
0
    input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT);
12436
0
    if (input == NULL) {
12437
0
  xmlFreeParserCtxt(ctxt);
12438
0
  return(NULL);
12439
0
    }
12440
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12441
0
        xmlFreeInputStream(input);
12442
0
        xmlFreeParserCtxt(ctxt);
12443
0
        return(NULL);
12444
0
    }
12445
12446
0
    return(ctxt);
12447
0
}
12448
12449
/**
12450
 * Create a parser context for a file content.
12451
 * Automatic support for ZLIB/Compress compressed document is provided
12452
 * by default if found at compile-time.
12453
 *
12454
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12455
 *
12456
 * @param filename  the filename
12457
 * @returns the new parser context or NULL
12458
 */
12459
xmlParserCtxt *
12460
xmlCreateFileParserCtxt(const char *filename)
12461
0
{
12462
0
    return(xmlCreateURLParserCtxt(filename, 0));
12463
0
}
12464
12465
#ifdef LIBXML_SAX1_ENABLED
12466
/**
12467
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12468
 * compressed document is provided by default if found at compile-time.
12469
 * It use the given SAX function block to handle the parsing callback.
12470
 * If sax is NULL, fallback to the default DOM tree building routines.
12471
 *
12472
 * This function uses deprecated global variables to set parser options
12473
 * which default to XML_PARSE_NODICT.
12474
 *
12475
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12476
 *
12477
 * User data (void *) is stored within the parser context in the
12478
 * context's _private member, so it is available nearly everywhere in libxml
12479
 *
12480
 * @param sax  the SAX handler block
12481
 * @param filename  the filename
12482
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12483
 *             documents
12484
 * @param data  the userdata
12485
 * @returns the resulting document tree
12486
 */
12487
12488
xmlDoc *
12489
xmlSAXParseFileWithData(xmlSAXHandler *sax, const char *filename,
12490
0
                        int recovery, void *data) {
12491
0
    xmlDocPtr ret = NULL;
12492
0
    xmlParserCtxtPtr ctxt;
12493
0
    xmlParserInputPtr input;
12494
12495
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12496
0
    if (ctxt == NULL)
12497
0
  return(NULL);
12498
12499
0
    if (data != NULL)
12500
0
  ctxt->_private = data;
12501
12502
0
    if (recovery) {
12503
0
        ctxt->options |= XML_PARSE_RECOVER;
12504
0
        ctxt->recovery = 1;
12505
0
    }
12506
12507
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
12508
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0);
12509
0
    else
12510
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0);
12511
12512
0
    if (input != NULL)
12513
0
        ret = xmlCtxtParseDocument(ctxt, input);
12514
12515
0
    xmlFreeParserCtxt(ctxt);
12516
0
    return(ret);
12517
0
}
12518
12519
/**
12520
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12521
 * compressed document is provided by default if found at compile-time.
12522
 * It use the given SAX function block to handle the parsing callback.
12523
 * If sax is NULL, fallback to the default DOM tree building routines.
12524
 *
12525
 * This function uses deprecated global variables to set parser options
12526
 * which default to XML_PARSE_NODICT.
12527
 *
12528
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12529
 *
12530
 * @param sax  the SAX handler block
12531
 * @param filename  the filename
12532
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12533
 *             documents
12534
 * @returns the resulting document tree
12535
 */
12536
12537
xmlDoc *
12538
xmlSAXParseFile(xmlSAXHandler *sax, const char *filename,
12539
0
                          int recovery) {
12540
0
    return(xmlSAXParseFileWithData(sax,filename,recovery,NULL));
12541
0
}
12542
12543
/**
12544
 * Parse an XML in-memory document and build a tree.
12545
 * In the case the document is not Well Formed, a attempt to build a
12546
 * tree is tried anyway
12547
 *
12548
 * This function uses deprecated global variables to set parser options
12549
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12550
 *
12551
 * @deprecated Use #xmlReadDoc with XML_PARSE_RECOVER.
12552
 *
12553
 * @param cur  a pointer to an array of xmlChar
12554
 * @returns the resulting document tree or NULL in case of failure
12555
 */
12556
12557
xmlDoc *
12558
0
xmlRecoverDoc(const xmlChar *cur) {
12559
0
    return(xmlSAXParseDoc(NULL, cur, 1));
12560
0
}
12561
12562
/**
12563
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12564
 * compressed document is provided by default if found at compile-time.
12565
 *
12566
 * This function uses deprecated global variables to set parser options
12567
 * which default to XML_PARSE_NODICT.
12568
 *
12569
 * @deprecated Use #xmlReadFile.
12570
 *
12571
 * @param filename  the filename
12572
 * @returns the resulting document tree if the file was wellformed,
12573
 * NULL otherwise.
12574
 */
12575
12576
xmlDoc *
12577
0
xmlParseFile(const char *filename) {
12578
0
    return(xmlSAXParseFile(NULL, filename, 0));
12579
0
}
12580
12581
/**
12582
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12583
 * compressed document is provided by default if found at compile-time.
12584
 * In the case the document is not Well Formed, it attempts to build
12585
 * a tree anyway
12586
 *
12587
 * This function uses deprecated global variables to set parser options
12588
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12589
 *
12590
 * @deprecated Use #xmlReadFile with XML_PARSE_RECOVER.
12591
 *
12592
 * @param filename  the filename
12593
 * @returns the resulting document tree or NULL in case of failure
12594
 */
12595
12596
xmlDoc *
12597
0
xmlRecoverFile(const char *filename) {
12598
0
    return(xmlSAXParseFile(NULL, filename, 1));
12599
0
}
12600
12601
12602
/**
12603
 * Setup the parser context to parse a new buffer; Clears any prior
12604
 * contents from the parser context. The buffer parameter must not be
12605
 * NULL, but the filename parameter can be
12606
 *
12607
 * @deprecated Don't use.
12608
 *
12609
 * @param ctxt  an XML parser context
12610
 * @param buffer  a xmlChar * buffer
12611
 * @param filename  a file name
12612
 */
12613
void
12614
xmlSetupParserForBuffer(xmlParserCtxt *ctxt, const xmlChar* buffer,
12615
                             const char* filename)
12616
0
{
12617
0
    xmlParserInputPtr input;
12618
12619
0
    if ((ctxt == NULL) || (buffer == NULL))
12620
0
        return;
12621
12622
0
    xmlCtxtReset(ctxt);
12623
12624
0
    input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer,
12625
0
                                      NULL, 0);
12626
0
    if (input == NULL)
12627
0
        return;
12628
0
    if (xmlCtxtPushInput(ctxt, input) < 0)
12629
0
        xmlFreeInputStream(input);
12630
0
}
12631
12632
/**
12633
 * Parse an XML file and call the given SAX handler routines.
12634
 * Automatic support for ZLIB/Compress compressed document is provided
12635
 *
12636
 * This function uses deprecated global variables to set parser options
12637
 * which default to XML_PARSE_NODICT.
12638
 *
12639
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12640
 *
12641
 * @param sax  a SAX handler
12642
 * @param user_data  The user data returned on SAX callbacks
12643
 * @param filename  a file name
12644
 * @returns 0 in case of success or a error number otherwise
12645
 */
12646
int
12647
xmlSAXUserParseFile(xmlSAXHandler *sax, void *user_data,
12648
0
                    const char *filename) {
12649
0
    int ret = 0;
12650
0
    xmlParserCtxtPtr ctxt;
12651
12652
0
    ctxt = xmlCreateFileParserCtxt(filename);
12653
0
    if (ctxt == NULL) return -1;
12654
0
    if (sax != NULL) {
12655
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12656
0
            *ctxt->sax = *sax;
12657
0
        } else {
12658
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12659
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12660
0
        }
12661
0
  ctxt->userData = user_data;
12662
0
    }
12663
12664
0
    xmlParseDocument(ctxt);
12665
12666
0
    if (ctxt->wellFormed)
12667
0
  ret = 0;
12668
0
    else {
12669
0
        if (ctxt->errNo != 0)
12670
0
      ret = ctxt->errNo;
12671
0
  else
12672
0
      ret = -1;
12673
0
    }
12674
0
    if (ctxt->myDoc != NULL) {
12675
0
        xmlFreeDoc(ctxt->myDoc);
12676
0
  ctxt->myDoc = NULL;
12677
0
    }
12678
0
    xmlFreeParserCtxt(ctxt);
12679
12680
0
    return ret;
12681
0
}
12682
#endif /* LIBXML_SAX1_ENABLED */
12683
12684
/************************************************************************
12685
 *                  *
12686
 *    Front ends when parsing from memory     *
12687
 *                  *
12688
 ************************************************************************/
12689
12690
/**
12691
 * Create a parser context for an XML in-memory document. The input buffer
12692
 * must not contain a terminating null byte.
12693
 *
12694
 * @param buffer  a pointer to a char array
12695
 * @param size  the size of the array
12696
 * @returns the new parser context or NULL
12697
 */
12698
xmlParserCtxt *
12699
0
xmlCreateMemoryParserCtxt(const char *buffer, int size) {
12700
0
    xmlParserCtxtPtr ctxt;
12701
0
    xmlParserInputPtr input;
12702
12703
0
    if (size < 0)
12704
0
  return(NULL);
12705
12706
0
    ctxt = xmlNewParserCtxt();
12707
0
    if (ctxt == NULL)
12708
0
  return(NULL);
12709
12710
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0);
12711
0
    if (input == NULL) {
12712
0
  xmlFreeParserCtxt(ctxt);
12713
0
  return(NULL);
12714
0
    }
12715
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12716
0
        xmlFreeInputStream(input);
12717
0
        xmlFreeParserCtxt(ctxt);
12718
0
        return(NULL);
12719
0
    }
12720
12721
0
    return(ctxt);
12722
0
}
12723
12724
#ifdef LIBXML_SAX1_ENABLED
12725
/**
12726
 * Parse an XML in-memory block and use the given SAX function block
12727
 * to handle the parsing callback. If sax is NULL, fallback to the default
12728
 * DOM tree building routines.
12729
 *
12730
 * This function uses deprecated global variables to set parser options
12731
 * which default to XML_PARSE_NODICT.
12732
 *
12733
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12734
 *
12735
 * User data (void *) is stored within the parser context in the
12736
 * context's _private member, so it is available nearly everywhere in libxml
12737
 *
12738
 * @param sax  the SAX handler block
12739
 * @param buffer  an pointer to a char array
12740
 * @param size  the size of the array
12741
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12742
 *             documents
12743
 * @param data  the userdata
12744
 * @returns the resulting document tree
12745
 */
12746
12747
xmlDoc *
12748
xmlSAXParseMemoryWithData(xmlSAXHandler *sax, const char *buffer,
12749
0
                          int size, int recovery, void *data) {
12750
0
    xmlDocPtr ret = NULL;
12751
0
    xmlParserCtxtPtr ctxt;
12752
0
    xmlParserInputPtr input;
12753
12754
0
    if (size < 0)
12755
0
        return(NULL);
12756
12757
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12758
0
    if (ctxt == NULL)
12759
0
        return(NULL);
12760
12761
0
    if (data != NULL)
12762
0
  ctxt->_private=data;
12763
12764
0
    if (recovery) {
12765
0
        ctxt->options |= XML_PARSE_RECOVER;
12766
0
        ctxt->recovery = 1;
12767
0
    }
12768
12769
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL,
12770
0
                                      XML_INPUT_BUF_STATIC);
12771
12772
0
    if (input != NULL)
12773
0
        ret = xmlCtxtParseDocument(ctxt, input);
12774
12775
0
    xmlFreeParserCtxt(ctxt);
12776
0
    return(ret);
12777
0
}
12778
12779
/**
12780
 * Parse an XML in-memory block and use the given SAX function block
12781
 * to handle the parsing callback. If sax is NULL, fallback to the default
12782
 * DOM tree building routines.
12783
 *
12784
 * This function uses deprecated global variables to set parser options
12785
 * which default to XML_PARSE_NODICT.
12786
 *
12787
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12788
 *
12789
 * @param sax  the SAX handler block
12790
 * @param buffer  an pointer to a char array
12791
 * @param size  the size of the array
12792
 * @param recovery  work in recovery mode, i.e. tries to read not Well Formed
12793
 *             documents
12794
 * @returns the resulting document tree
12795
 */
12796
xmlDoc *
12797
xmlSAXParseMemory(xmlSAXHandler *sax, const char *buffer,
12798
0
            int size, int recovery) {
12799
0
    return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL);
12800
0
}
12801
12802
/**
12803
 * Parse an XML in-memory block and build a tree.
12804
 *
12805
 * This function uses deprecated global variables to set parser options
12806
 * which default to XML_PARSE_NODICT.
12807
 *
12808
 * @deprecated Use #xmlReadMemory.
12809
 *
12810
 * @param buffer  an pointer to a char array
12811
 * @param size  the size of the array
12812
 * @returns the resulting document tree
12813
 */
12814
12815
0
xmlDoc *xmlParseMemory(const char *buffer, int size) {
12816
0
   return(xmlSAXParseMemory(NULL, buffer, size, 0));
12817
0
}
12818
12819
/**
12820
 * Parse an XML in-memory block and build a tree.
12821
 * In the case the document is not Well Formed, an attempt to
12822
 * build a tree is tried anyway
12823
 *
12824
 * This function uses deprecated global variables to set parser options
12825
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12826
 *
12827
 * @deprecated Use #xmlReadMemory with XML_PARSE_RECOVER.
12828
 *
12829
 * @param buffer  an pointer to a char array
12830
 * @param size  the size of the array
12831
 * @returns the resulting document tree or NULL in case of error
12832
 */
12833
12834
0
xmlDoc *xmlRecoverMemory(const char *buffer, int size) {
12835
0
   return(xmlSAXParseMemory(NULL, buffer, size, 1));
12836
0
}
12837
12838
/**
12839
 * Parse an XML in-memory buffer and call the given SAX handler routines.
12840
 *
12841
 * This function uses deprecated global variables to set parser options
12842
 * which default to XML_PARSE_NODICT.
12843
 *
12844
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12845
 *
12846
 * @param sax  a SAX handler
12847
 * @param user_data  The user data returned on SAX callbacks
12848
 * @param buffer  an in-memory XML document input
12849
 * @param size  the length of the XML document in bytes
12850
 * @returns 0 in case of success or a error number otherwise
12851
 */
12852
int xmlSAXUserParseMemory(xmlSAXHandler *sax, void *user_data,
12853
0
        const char *buffer, int size) {
12854
0
    int ret = 0;
12855
0
    xmlParserCtxtPtr ctxt;
12856
12857
0
    ctxt = xmlCreateMemoryParserCtxt(buffer, size);
12858
0
    if (ctxt == NULL) return -1;
12859
0
    if (sax != NULL) {
12860
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12861
0
            *ctxt->sax = *sax;
12862
0
        } else {
12863
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12864
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12865
0
        }
12866
0
  ctxt->userData = user_data;
12867
0
    }
12868
12869
0
    xmlParseDocument(ctxt);
12870
12871
0
    if (ctxt->wellFormed)
12872
0
  ret = 0;
12873
0
    else {
12874
0
        if (ctxt->errNo != 0)
12875
0
      ret = ctxt->errNo;
12876
0
  else
12877
0
      ret = -1;
12878
0
    }
12879
0
    if (ctxt->myDoc != NULL) {
12880
0
        xmlFreeDoc(ctxt->myDoc);
12881
0
  ctxt->myDoc = NULL;
12882
0
    }
12883
0
    xmlFreeParserCtxt(ctxt);
12884
12885
0
    return ret;
12886
0
}
12887
#endif /* LIBXML_SAX1_ENABLED */
12888
12889
/**
12890
 * Creates a parser context for an XML in-memory document.
12891
 *
12892
 * @param str  a pointer to an array of xmlChar
12893
 * @returns the new parser context or NULL
12894
 */
12895
xmlParserCtxt *
12896
0
xmlCreateDocParserCtxt(const xmlChar *str) {
12897
0
    xmlParserCtxtPtr ctxt;
12898
0
    xmlParserInputPtr input;
12899
12900
0
    ctxt = xmlNewParserCtxt();
12901
0
    if (ctxt == NULL)
12902
0
  return(NULL);
12903
12904
0
    input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0);
12905
0
    if (input == NULL) {
12906
0
  xmlFreeParserCtxt(ctxt);
12907
0
  return(NULL);
12908
0
    }
12909
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12910
0
        xmlFreeInputStream(input);
12911
0
        xmlFreeParserCtxt(ctxt);
12912
0
        return(NULL);
12913
0
    }
12914
12915
0
    return(ctxt);
12916
0
}
12917
12918
#ifdef LIBXML_SAX1_ENABLED
12919
/**
12920
 * Parse an XML in-memory document and build a tree.
12921
 * It use the given SAX function block to handle the parsing callback.
12922
 * If sax is NULL, fallback to the default DOM tree building routines.
12923
 *
12924
 * This function uses deprecated global variables to set parser options
12925
 * which default to XML_PARSE_NODICT.
12926
 *
12927
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadDoc.
12928
 *
12929
 * @param sax  the SAX handler block
12930
 * @param cur  a pointer to an array of xmlChar
12931
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12932
 *             documents
12933
 * @returns the resulting document tree
12934
 */
12935
12936
xmlDoc *
12937
0
xmlSAXParseDoc(xmlSAXHandler *sax, const xmlChar *cur, int recovery) {
12938
0
    xmlDocPtr ret;
12939
0
    xmlParserCtxtPtr ctxt;
12940
0
    xmlSAXHandlerPtr oldsax = NULL;
12941
12942
0
    if (cur == NULL) return(NULL);
12943
12944
12945
0
    ctxt = xmlCreateDocParserCtxt(cur);
12946
0
    if (ctxt == NULL) return(NULL);
12947
0
    if (sax != NULL) {
12948
0
        oldsax = ctxt->sax;
12949
0
        ctxt->sax = sax;
12950
0
        ctxt->userData = NULL;
12951
0
    }
12952
12953
0
    xmlParseDocument(ctxt);
12954
0
    if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc;
12955
0
    else {
12956
0
       ret = NULL;
12957
0
       xmlFreeDoc(ctxt->myDoc);
12958
0
       ctxt->myDoc = NULL;
12959
0
    }
12960
0
    if (sax != NULL)
12961
0
  ctxt->sax = oldsax;
12962
0
    xmlFreeParserCtxt(ctxt);
12963
12964
0
    return(ret);
12965
0
}
12966
12967
/**
12968
 * Parse an XML in-memory document and build a tree.
12969
 *
12970
 * This function uses deprecated global variables to set parser options
12971
 * which default to XML_PARSE_NODICT.
12972
 *
12973
 * @deprecated Use #xmlReadDoc.
12974
 *
12975
 * @param cur  a pointer to an array of xmlChar
12976
 * @returns the resulting document tree
12977
 */
12978
12979
xmlDoc *
12980
0
xmlParseDoc(const xmlChar *cur) {
12981
0
    return(xmlSAXParseDoc(NULL, cur, 0));
12982
0
}
12983
#endif /* LIBXML_SAX1_ENABLED */
12984
12985
/************************************************************************
12986
 *                  *
12987
 *  New set (2.6.0) of simpler and more flexible APIs   *
12988
 *                  *
12989
 ************************************************************************/
12990
12991
/**
12992
 * Reset a parser context
12993
 *
12994
 * @param ctxt  an XML parser context
12995
 */
12996
void
12997
xmlCtxtReset(xmlParserCtxt *ctxt)
12998
25.3k
{
12999
25.3k
    xmlParserInputPtr input;
13000
13001
25.3k
    if (ctxt == NULL)
13002
0
        return;
13003
13004
25.3k
    while ((input = xmlCtxtPopInput(ctxt)) != NULL) { /* Non consuming */
13005
0
        xmlFreeInputStream(input);
13006
0
    }
13007
25.3k
    ctxt->inputNr = 0;
13008
25.3k
    ctxt->input = NULL;
13009
13010
25.3k
    ctxt->spaceNr = 0;
13011
25.3k
    if (ctxt->spaceTab != NULL) {
13012
25.3k
  ctxt->spaceTab[0] = -1;
13013
25.3k
  ctxt->space = &ctxt->spaceTab[0];
13014
25.3k
    } else {
13015
0
        ctxt->space = NULL;
13016
0
    }
13017
13018
13019
25.3k
    ctxt->nodeNr = 0;
13020
25.3k
    ctxt->node = NULL;
13021
13022
25.3k
    ctxt->nameNr = 0;
13023
25.3k
    ctxt->name = NULL;
13024
13025
25.3k
    ctxt->nsNr = 0;
13026
25.3k
    xmlParserNsReset(ctxt->nsdb);
13027
13028
25.3k
    if (ctxt->version != NULL) {
13029
0
        xmlFree(ctxt->version);
13030
0
        ctxt->version = NULL;
13031
0
    }
13032
25.3k
    if (ctxt->encoding != NULL) {
13033
0
        xmlFree(ctxt->encoding);
13034
0
        ctxt->encoding = NULL;
13035
0
    }
13036
25.3k
    if (ctxt->extSubURI != NULL) {
13037
0
        xmlFree(ctxt->extSubURI);
13038
0
        ctxt->extSubURI = NULL;
13039
0
    }
13040
25.3k
    if (ctxt->extSubSystem != NULL) {
13041
0
        xmlFree(ctxt->extSubSystem);
13042
0
        ctxt->extSubSystem = NULL;
13043
0
    }
13044
25.3k
    if (ctxt->directory != NULL) {
13045
0
        xmlFree(ctxt->directory);
13046
0
        ctxt->directory = NULL;
13047
0
    }
13048
13049
25.3k
    if (ctxt->myDoc != NULL)
13050
0
        xmlFreeDoc(ctxt->myDoc);
13051
25.3k
    ctxt->myDoc = NULL;
13052
13053
25.3k
    ctxt->standalone = -1;
13054
25.3k
    ctxt->hasExternalSubset = 0;
13055
25.3k
    ctxt->hasPErefs = 0;
13056
25.3k
    ctxt->html = ctxt->html ? 1 : 0;
13057
25.3k
    ctxt->instate = XML_PARSER_START;
13058
13059
25.3k
    ctxt->wellFormed = 1;
13060
25.3k
    ctxt->nsWellFormed = 1;
13061
25.3k
    ctxt->disableSAX = 0;
13062
25.3k
    ctxt->valid = 1;
13063
25.3k
    ctxt->record_info = 0;
13064
25.3k
    ctxt->checkIndex = 0;
13065
25.3k
    ctxt->endCheckState = 0;
13066
25.3k
    ctxt->inSubset = 0;
13067
25.3k
    ctxt->errNo = XML_ERR_OK;
13068
25.3k
    ctxt->depth = 0;
13069
25.3k
    ctxt->catalogs = NULL;
13070
25.3k
    ctxt->sizeentities = 0;
13071
25.3k
    ctxt->sizeentcopy = 0;
13072
25.3k
    xmlInitNodeInfoSeq(&ctxt->node_seq);
13073
13074
25.3k
    if (ctxt->attsDefault != NULL) {
13075
0
        xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator);
13076
0
        ctxt->attsDefault = NULL;
13077
0
    }
13078
25.3k
    if (ctxt->attsSpecial != NULL) {
13079
0
        xmlHashFree(ctxt->attsSpecial, NULL);
13080
0
        ctxt->attsSpecial = NULL;
13081
0
    }
13082
13083
25.3k
#ifdef LIBXML_CATALOG_ENABLED
13084
25.3k
    if (ctxt->catalogs != NULL)
13085
0
  xmlCatalogFreeLocal(ctxt->catalogs);
13086
25.3k
#endif
13087
25.3k
    ctxt->nbErrors = 0;
13088
25.3k
    ctxt->nbWarnings = 0;
13089
25.3k
    if (ctxt->lastError.code != XML_ERR_OK)
13090
0
        xmlResetError(&ctxt->lastError);
13091
25.3k
}
13092
13093
/**
13094
 * Reset a push parser context
13095
 *
13096
 * @param ctxt  an XML parser context
13097
 * @param chunk  a pointer to an array of chars
13098
 * @param size  number of chars in the array
13099
 * @param filename  an optional file name or URI
13100
 * @param encoding  the document encoding, or NULL
13101
 * @returns 0 in case of success and 1 in case of error
13102
 */
13103
int
13104
xmlCtxtResetPush(xmlParserCtxt *ctxt, const char *chunk,
13105
                 int size, const char *filename, const char *encoding)
13106
0
{
13107
0
    xmlParserInputPtr input;
13108
13109
0
    if (ctxt == NULL)
13110
0
        return(1);
13111
13112
0
    xmlCtxtReset(ctxt);
13113
13114
0
    input = xmlNewPushInput(filename, chunk, size);
13115
0
    if (input == NULL)
13116
0
        return(1);
13117
13118
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13119
0
        xmlFreeInputStream(input);
13120
0
        return(1);
13121
0
    }
13122
13123
0
    if (encoding != NULL)
13124
0
        xmlSwitchEncodingName(ctxt, encoding);
13125
13126
0
    return(0);
13127
0
}
13128
13129
static int
13130
xmlCtxtSetOptionsInternal(xmlParserCtxtPtr ctxt, int options, int keepMask)
13131
50.6k
{
13132
50.6k
    int allMask;
13133
13134
50.6k
    if (ctxt == NULL)
13135
0
        return(-1);
13136
13137
    /*
13138
     * XInclude options aren't handled by the parser.
13139
     *
13140
     * XML_PARSE_XINCLUDE
13141
     * XML_PARSE_NOXINCNODE
13142
     * XML_PARSE_NOBASEFIX
13143
     */
13144
50.6k
    allMask = XML_PARSE_RECOVER |
13145
50.6k
              XML_PARSE_NOENT |
13146
50.6k
              XML_PARSE_DTDLOAD |
13147
50.6k
              XML_PARSE_DTDATTR |
13148
50.6k
              XML_PARSE_DTDVALID |
13149
50.6k
              XML_PARSE_NOERROR |
13150
50.6k
              XML_PARSE_NOWARNING |
13151
50.6k
              XML_PARSE_PEDANTIC |
13152
50.6k
              XML_PARSE_NOBLANKS |
13153
50.6k
#ifdef LIBXML_SAX1_ENABLED
13154
50.6k
              XML_PARSE_SAX1 |
13155
50.6k
#endif
13156
50.6k
              XML_PARSE_NONET |
13157
50.6k
              XML_PARSE_NODICT |
13158
50.6k
              XML_PARSE_NSCLEAN |
13159
50.6k
              XML_PARSE_NOCDATA |
13160
50.6k
              XML_PARSE_COMPACT |
13161
50.6k
              XML_PARSE_OLD10 |
13162
50.6k
              XML_PARSE_HUGE |
13163
50.6k
              XML_PARSE_OLDSAX |
13164
50.6k
              XML_PARSE_IGNORE_ENC |
13165
50.6k
              XML_PARSE_BIG_LINES |
13166
50.6k
              XML_PARSE_NO_XXE |
13167
50.6k
              XML_PARSE_UNZIP |
13168
50.6k
              XML_PARSE_NO_SYS_CATALOG |
13169
50.6k
              XML_PARSE_CATALOG_PI;
13170
13171
50.6k
    ctxt->options = (ctxt->options & keepMask) | (options & allMask);
13172
13173
    /*
13174
     * For some options, struct members are historically the source
13175
     * of truth. The values are initalized from global variables and
13176
     * old code could also modify them directly. Several older API
13177
     * functions that don't take an options argument rely on these
13178
     * deprecated mechanisms.
13179
     *
13180
     * Once public access to struct members and the globals are
13181
     * disabled, we can use the options bitmask as source of
13182
     * truth, making all these struct members obsolete.
13183
     *
13184
     * The XML_DETECT_IDS flags is misnamed. It simply enables
13185
     * loading of the external subset.
13186
     */
13187
50.6k
    ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0;
13188
50.6k
    ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0;
13189
50.6k
    ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0;
13190
50.6k
    ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0;
13191
50.6k
    ctxt->loadsubset |= (options & XML_PARSE_SKIP_IDS) ? XML_SKIP_IDS : 0;
13192
50.6k
    ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0;
13193
50.6k
    ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0;
13194
50.6k
    ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1;
13195
50.6k
    ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1;
13196
13197
50.6k
    return(options & ~allMask);
13198
50.6k
}
13199
13200
/**
13201
 * Applies the options to the parser context. Unset options are
13202
 * cleared.
13203
 *
13204
 * @since 2.13.0
13205
 *
13206
 * With older versions, you can use #xmlCtxtUseOptions.
13207
 *
13208
 * @param ctxt  an XML parser context
13209
 * @param options  a bitmask of xmlParserOption values
13210
 * @returns 0 in case of success, the set of unknown or unimplemented options
13211
 *         in case of error.
13212
 */
13213
int
13214
xmlCtxtSetOptions(xmlParserCtxt *ctxt, int options)
13215
0
{
13216
0
#ifdef LIBXML_HTML_ENABLED
13217
0
    if ((ctxt != NULL) && (ctxt->html))
13218
0
        return(htmlCtxtSetOptions(ctxt, options));
13219
0
#endif
13220
13221
0
    return(xmlCtxtSetOptionsInternal(ctxt, options, 0));
13222
0
}
13223
13224
/**
13225
 * Get the current options of the parser context.
13226
 *
13227
 * @since 2.14.0
13228
 *
13229
 * @param ctxt  an XML parser context
13230
 * @returns the current options set in the parser context, or -1 if ctxt is NULL.
13231
 */
13232
int
13233
xmlCtxtGetOptions(xmlParserCtxt *ctxt)
13234
0
{
13235
0
    if (ctxt == NULL)
13236
0
        return(-1);
13237
13238
0
    return(ctxt->options);
13239
0
}
13240
13241
/**
13242
 * Applies the options to the parser context. The following options
13243
 * are never cleared and can only be enabled:
13244
 *
13245
 * - XML_PARSE_NOERROR
13246
 * - XML_PARSE_NOWARNING
13247
 * - XML_PARSE_NONET
13248
 * - XML_PARSE_NSCLEAN
13249
 * - XML_PARSE_NOCDATA
13250
 * - XML_PARSE_COMPACT
13251
 * - XML_PARSE_OLD10
13252
 * - XML_PARSE_HUGE
13253
 * - XML_PARSE_OLDSAX
13254
 * - XML_PARSE_IGNORE_ENC
13255
 * - XML_PARSE_BIG_LINES
13256
 *
13257
 * @deprecated Use #xmlCtxtSetOptions.
13258
 *
13259
 * @param ctxt  an XML parser context
13260
 * @param options  a combination of xmlParserOption
13261
 * @returns 0 in case of success, the set of unknown or unimplemented options
13262
 *         in case of error.
13263
 */
13264
int
13265
xmlCtxtUseOptions(xmlParserCtxt *ctxt, int options)
13266
50.6k
{
13267
50.6k
    int keepMask;
13268
13269
50.6k
#ifdef LIBXML_HTML_ENABLED
13270
50.6k
    if ((ctxt != NULL) && (ctxt->html))
13271
0
        return(htmlCtxtUseOptions(ctxt, options));
13272
50.6k
#endif
13273
13274
    /*
13275
     * For historic reasons, some options can only be enabled.
13276
     */
13277
50.6k
    keepMask = XML_PARSE_NOERROR |
13278
50.6k
               XML_PARSE_NOWARNING |
13279
50.6k
               XML_PARSE_NONET |
13280
50.6k
               XML_PARSE_NSCLEAN |
13281
50.6k
               XML_PARSE_NOCDATA |
13282
50.6k
               XML_PARSE_COMPACT |
13283
50.6k
               XML_PARSE_OLD10 |
13284
50.6k
               XML_PARSE_HUGE |
13285
50.6k
               XML_PARSE_OLDSAX |
13286
50.6k
               XML_PARSE_IGNORE_ENC |
13287
50.6k
               XML_PARSE_BIG_LINES;
13288
13289
50.6k
    return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask));
13290
50.6k
}
13291
13292
/**
13293
 * To protect against exponential entity expansion ("billion laughs"), the
13294
 * size of serialized output is (roughly) limited to the input size
13295
 * multiplied by this factor. The default value is 5.
13296
 *
13297
 * When working with documents making heavy use of entity expansion, it can
13298
 * be necessary to increase the value. For security reasons, this should only
13299
 * be considered when processing trusted input.
13300
 *
13301
 * @param ctxt  an XML parser context
13302
 * @param maxAmpl  maximum amplification factor
13303
 */
13304
void
13305
xmlCtxtSetMaxAmplification(xmlParserCtxt *ctxt, unsigned maxAmpl)
13306
0
{
13307
0
    if (ctxt == NULL)
13308
0
        return;
13309
0
    ctxt->maxAmpl = maxAmpl;
13310
0
}
13311
13312
/**
13313
 * Parse an XML document and return the resulting document tree.
13314
 * Takes ownership of the input object.
13315
 *
13316
 * @since 2.13.0
13317
 *
13318
 * @param ctxt  an XML parser context
13319
 * @param input  parser input
13320
 * @returns the resulting document tree or NULL
13321
 */
13322
xmlDoc *
13323
xmlCtxtParseDocument(xmlParserCtxt *ctxt, xmlParserInput *input)
13324
25.3k
{
13325
25.3k
    xmlDocPtr ret = NULL;
13326
13327
25.3k
    if ((ctxt == NULL) || (input == NULL)) {
13328
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
13329
0
        xmlFreeInputStream(input);
13330
0
        return(NULL);
13331
0
    }
13332
13333
    /* assert(ctxt->inputNr == 0); */
13334
25.3k
    while (ctxt->inputNr > 0)
13335
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13336
13337
25.3k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13338
5
        xmlFreeInputStream(input);
13339
5
        return(NULL);
13340
5
    }
13341
13342
25.3k
    xmlParseDocument(ctxt);
13343
13344
25.3k
    ret = xmlCtxtGetDocument(ctxt);
13345
13346
    /* assert(ctxt->inputNr == 1); */
13347
51.2k
    while (ctxt->inputNr > 0)
13348
25.9k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13349
13350
25.3k
    return(ret);
13351
25.3k
}
13352
13353
/**
13354
 * Convenience function to parse an XML document from a
13355
 * zero-terminated string.
13356
 *
13357
 * See #xmlCtxtReadDoc for details.
13358
 *
13359
 * @param cur  a pointer to a zero terminated string
13360
 * @param URL  base URL (optional)
13361
 * @param encoding  the document encoding (optional)
13362
 * @param options  a combination of xmlParserOption
13363
 * @returns the resulting document tree
13364
 */
13365
xmlDoc *
13366
xmlReadDoc(const xmlChar *cur, const char *URL, const char *encoding,
13367
           int options)
13368
0
{
13369
0
    xmlParserCtxtPtr ctxt;
13370
0
    xmlParserInputPtr input;
13371
0
    xmlDocPtr doc = NULL;
13372
13373
0
    ctxt = xmlNewParserCtxt();
13374
0
    if (ctxt == NULL)
13375
0
        return(NULL);
13376
13377
0
    xmlCtxtUseOptions(ctxt, options);
13378
13379
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding,
13380
0
                                      XML_INPUT_BUF_STATIC);
13381
13382
0
    if (input != NULL)
13383
0
        doc = xmlCtxtParseDocument(ctxt, input);
13384
13385
0
    xmlFreeParserCtxt(ctxt);
13386
0
    return(doc);
13387
0
}
13388
13389
/**
13390
 * Convenience function to parse an XML file from the filesystem
13391
 * or a global, user-defined resource loader.
13392
 *
13393
 * If a "-" filename is passed, the function will read from stdin.
13394
 * This feature is potentially insecure and might be removed from
13395
 * later versions.
13396
 *
13397
 * See #xmlCtxtReadFile for details.
13398
 *
13399
 * @param filename  a file or URL
13400
 * @param encoding  the document encoding (optional)
13401
 * @param options  a combination of xmlParserOption
13402
 * @returns the resulting document tree
13403
 */
13404
xmlDoc *
13405
xmlReadFile(const char *filename, const char *encoding, int options)
13406
0
{
13407
0
    xmlParserCtxtPtr ctxt;
13408
0
    xmlParserInputPtr input;
13409
0
    xmlDocPtr doc = NULL;
13410
13411
0
    ctxt = xmlNewParserCtxt();
13412
0
    if (ctxt == NULL)
13413
0
        return(NULL);
13414
13415
0
    xmlCtxtUseOptions(ctxt, options);
13416
13417
    /*
13418
     * Backward compatibility for users of command line utilities like
13419
     * xmlstarlet expecting "-" to mean stdin. This is dangerous and
13420
     * should be removed at some point.
13421
     */
13422
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
13423
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO,
13424
0
                                      encoding, 0);
13425
0
    else
13426
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13427
13428
0
    if (input != NULL)
13429
0
        doc = xmlCtxtParseDocument(ctxt, input);
13430
13431
0
    xmlFreeParserCtxt(ctxt);
13432
0
    return(doc);
13433
0
}
13434
13435
/**
13436
 * Parse an XML in-memory document and build a tree. The input buffer must
13437
 * not contain a terminating null byte.
13438
 *
13439
 * See #xmlCtxtReadMemory for details.
13440
 *
13441
 * @param buffer  a pointer to a char array
13442
 * @param size  the size of the array
13443
 * @param url  base URL (optional)
13444
 * @param encoding  the document encoding (optional)
13445
 * @param options  a combination of xmlParserOption
13446
 * @returns the resulting document tree
13447
 */
13448
xmlDoc *
13449
xmlReadMemory(const char *buffer, int size, const char *url,
13450
              const char *encoding, int options)
13451
0
{
13452
0
    xmlParserCtxtPtr ctxt;
13453
0
    xmlParserInputPtr input;
13454
0
    xmlDocPtr doc = NULL;
13455
13456
0
    if (size < 0)
13457
0
  return(NULL);
13458
13459
0
    ctxt = xmlNewParserCtxt();
13460
0
    if (ctxt == NULL)
13461
0
        return(NULL);
13462
13463
0
    xmlCtxtUseOptions(ctxt, options);
13464
13465
0
    input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding,
13466
0
                                      XML_INPUT_BUF_STATIC);
13467
13468
0
    if (input != NULL)
13469
0
        doc = xmlCtxtParseDocument(ctxt, input);
13470
13471
0
    xmlFreeParserCtxt(ctxt);
13472
0
    return(doc);
13473
0
}
13474
13475
/**
13476
 * Parse an XML from a file descriptor and build a tree.
13477
 *
13478
 * See #xmlCtxtReadFd for details.
13479
 *
13480
 * NOTE that the file descriptor will not be closed when the
13481
 * context is freed or reset.
13482
 *
13483
 * @param fd  an open file descriptor
13484
 * @param URL  base URL (optional)
13485
 * @param encoding  the document encoding (optional)
13486
 * @param options  a combination of xmlParserOption
13487
 * @returns the resulting document tree
13488
 */
13489
xmlDoc *
13490
xmlReadFd(int fd, const char *URL, const char *encoding, int options)
13491
0
{
13492
0
    xmlParserCtxtPtr ctxt;
13493
0
    xmlParserInputPtr input;
13494
0
    xmlDocPtr doc = NULL;
13495
13496
0
    ctxt = xmlNewParserCtxt();
13497
0
    if (ctxt == NULL)
13498
0
        return(NULL);
13499
13500
0
    xmlCtxtUseOptions(ctxt, options);
13501
13502
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13503
13504
0
    if (input != NULL)
13505
0
        doc = xmlCtxtParseDocument(ctxt, input);
13506
13507
0
    xmlFreeParserCtxt(ctxt);
13508
0
    return(doc);
13509
0
}
13510
13511
/**
13512
 * Parse an XML document from I/O functions and context and build a tree.
13513
 *
13514
 * See #xmlCtxtReadIO for details.
13515
 *
13516
 * @param ioread  an I/O read function
13517
 * @param ioclose  an I/O close function (optional)
13518
 * @param ioctx  an I/O handler
13519
 * @param URL  base URL (optional)
13520
 * @param encoding  the document encoding (optional)
13521
 * @param options  a combination of xmlParserOption
13522
 * @returns the resulting document tree
13523
 */
13524
xmlDoc *
13525
xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose,
13526
          void *ioctx, const char *URL, const char *encoding, int options)
13527
0
{
13528
0
    xmlParserCtxtPtr ctxt;
13529
0
    xmlParserInputPtr input;
13530
0
    xmlDocPtr doc = NULL;
13531
13532
0
    ctxt = xmlNewParserCtxt();
13533
0
    if (ctxt == NULL)
13534
0
        return(NULL);
13535
13536
0
    xmlCtxtUseOptions(ctxt, options);
13537
13538
0
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13539
0
                                  encoding, 0);
13540
13541
0
    if (input != NULL)
13542
0
        doc = xmlCtxtParseDocument(ctxt, input);
13543
13544
0
    xmlFreeParserCtxt(ctxt);
13545
0
    return(doc);
13546
0
}
13547
13548
/**
13549
 * Parse an XML in-memory document and build a tree.
13550
 *
13551
 * `URL` is used as base to resolve external entities and for error
13552
 * reporting.
13553
 *
13554
 * @param ctxt  an XML parser context
13555
 * @param str  a pointer to a zero terminated string
13556
 * @param URL  base URL (optional)
13557
 * @param encoding  the document encoding (optional)
13558
 * @param options  a combination of xmlParserOption
13559
 * @returns the resulting document tree
13560
 */
13561
xmlDoc *
13562
xmlCtxtReadDoc(xmlParserCtxt *ctxt, const xmlChar *str,
13563
               const char *URL, const char *encoding, int options)
13564
0
{
13565
0
    xmlParserInputPtr input;
13566
13567
0
    if (ctxt == NULL)
13568
0
        return(NULL);
13569
13570
0
    xmlCtxtReset(ctxt);
13571
0
    xmlCtxtUseOptions(ctxt, options);
13572
13573
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) str, encoding,
13574
0
                                      XML_INPUT_BUF_STATIC);
13575
0
    if (input == NULL)
13576
0
        return(NULL);
13577
13578
0
    return(xmlCtxtParseDocument(ctxt, input));
13579
0
}
13580
13581
/**
13582
 * Parse an XML file from the filesystem or a global, user-defined
13583
 * resource loader.
13584
 *
13585
 * @param ctxt  an XML parser context
13586
 * @param filename  a file or URL
13587
 * @param encoding  the document encoding (optional)
13588
 * @param options  a combination of xmlParserOption
13589
 * @returns the resulting document tree
13590
 */
13591
xmlDoc *
13592
xmlCtxtReadFile(xmlParserCtxt *ctxt, const char *filename,
13593
                const char *encoding, int options)
13594
0
{
13595
0
    xmlParserInputPtr input;
13596
13597
0
    if (ctxt == NULL)
13598
0
        return(NULL);
13599
13600
0
    xmlCtxtReset(ctxt);
13601
0
    xmlCtxtUseOptions(ctxt, options);
13602
13603
0
    input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13604
0
    if (input == NULL)
13605
0
        return(NULL);
13606
13607
0
    return(xmlCtxtParseDocument(ctxt, input));
13608
0
}
13609
13610
/**
13611
 * Parse an XML in-memory document and build a tree. The input buffer must
13612
 * not contain a terminating null byte.
13613
 *
13614
 * `URL` is used as base to resolve external entities and for error
13615
 * reporting.
13616
 *
13617
 * @param ctxt  an XML parser context
13618
 * @param buffer  a pointer to a char array
13619
 * @param size  the size of the array
13620
 * @param URL  base URL (optional)
13621
 * @param encoding  the document encoding (optional)
13622
 * @param options  a combination of xmlParserOption
13623
 * @returns the resulting document tree
13624
 */
13625
xmlDoc *
13626
xmlCtxtReadMemory(xmlParserCtxt *ctxt, const char *buffer, int size,
13627
                  const char *URL, const char *encoding, int options)
13628
25.3k
{
13629
25.3k
    xmlParserInputPtr input;
13630
13631
25.3k
    if ((ctxt == NULL) || (size < 0))
13632
0
        return(NULL);
13633
13634
25.3k
    xmlCtxtReset(ctxt);
13635
25.3k
    xmlCtxtUseOptions(ctxt, options);
13636
13637
25.3k
    input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding,
13638
25.3k
                                      XML_INPUT_BUF_STATIC);
13639
25.3k
    if (input == NULL)
13640
21
        return(NULL);
13641
13642
25.3k
    return(xmlCtxtParseDocument(ctxt, input));
13643
25.3k
}
13644
13645
/**
13646
 * Parse an XML document from a file descriptor and build a tree.
13647
 *
13648
 * NOTE that the file descriptor will not be closed when the
13649
 * context is freed or reset.
13650
 *
13651
 * `URL` is used as base to resolve external entities and for error
13652
 * reporting.
13653
 *
13654
 * @param ctxt  an XML parser context
13655
 * @param fd  an open file descriptor
13656
 * @param URL  base URL (optional)
13657
 * @param encoding  the document encoding (optional)
13658
 * @param options  a combination of xmlParserOption
13659
 * @returns the resulting document tree
13660
 */
13661
xmlDoc *
13662
xmlCtxtReadFd(xmlParserCtxt *ctxt, int fd,
13663
              const char *URL, const char *encoding, int options)
13664
0
{
13665
0
    xmlParserInputPtr input;
13666
13667
0
    if (ctxt == NULL)
13668
0
        return(NULL);
13669
13670
0
    xmlCtxtReset(ctxt);
13671
0
    xmlCtxtUseOptions(ctxt, options);
13672
13673
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13674
0
    if (input == NULL)
13675
0
        return(NULL);
13676
13677
0
    return(xmlCtxtParseDocument(ctxt, input));
13678
0
}
13679
13680
/**
13681
 * Parse an XML document from I/O functions and source and build a tree.
13682
 * This reuses the existing `ctxt` parser context
13683
 *
13684
 * `URL` is used as base to resolve external entities and for error
13685
 * reporting.
13686
 *
13687
 * @param ctxt  an XML parser context
13688
 * @param ioread  an I/O read function
13689
 * @param ioclose  an I/O close function
13690
 * @param ioctx  an I/O handler
13691
 * @param URL  the base URL to use for the document
13692
 * @param encoding  the document encoding, or NULL
13693
 * @param options  a combination of xmlParserOption
13694
 * @returns the resulting document tree
13695
 */
13696
xmlDoc *
13697
xmlCtxtReadIO(xmlParserCtxt *ctxt, xmlInputReadCallback ioread,
13698
              xmlInputCloseCallback ioclose, void *ioctx,
13699
        const char *URL,
13700
              const char *encoding, int options)
13701
0
{
13702
0
    xmlParserInputPtr input;
13703
13704
0
    if (ctxt == NULL)
13705
0
        return(NULL);
13706
13707
0
    xmlCtxtReset(ctxt);
13708
0
    xmlCtxtUseOptions(ctxt, options);
13709
13710
0
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13711
0
                                  encoding, 0);
13712
0
    if (input == NULL)
13713
0
        return(NULL);
13714
13715
0
    return(xmlCtxtParseDocument(ctxt, input));
13716
0
}
13717