Coverage Report

Created: 2026-04-12 06:49

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
489k
#define NS_INDEX_EMPTY  INT_MAX
81
249k
#define NS_INDEX_XML    (INT_MAX - 1)
82
255k
#define URI_HASH_EMPTY  0xD943A04E
83
100k
#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
319k
#define XML_MAX_ATTRS 100000000 /* 100 million */
94
95
394k
#define XML_SPECIAL_EXTERNAL    (1 << 20)
96
367k
#define XML_SPECIAL_TYPE_MASK   (XML_SPECIAL_EXTERNAL - 1)
97
98
353k
#define XML_ATTVAL_ALLOC        (1 << 0)
99
59.7k
#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
2.54M
#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
2.55M
#define XML_ENT_FIXED_COST 20
170
171
12.3M
#define XML_PARSER_BIG_BUFFER_SIZE 300
172
790k
#define XML_PARSER_BUFFER_SIZE 100
173
79.2k
#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
446
xmlErrMemory(xmlParserCtxtPtr ctxt) {
225
446
    xmlCtxtErrMemory(ctxt);
226
446
}
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
83.3k
{
239
83.3k
    if (prefix == NULL)
240
17.7k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
241
17.7k
                   XML_ERR_FATAL, localname, NULL, NULL, 0,
242
17.7k
                   "Attribute %s redefined\n", localname);
243
65.6k
    else
244
65.6k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
245
65.6k
                   XML_ERR_FATAL, prefix, localname, NULL, 0,
246
65.6k
                   "Attribute %s:%s redefined\n", prefix, localname);
247
83.3k
}
248
249
/**
250
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
251
 *
252
 * @param ctxt  an XML parser context
253
 * @param error  the error number
254
 * @param msg  the error message
255
 */
256
static void LIBXML_ATTR_FORMAT(3,0)
257
xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
258
               const char *msg)
259
34.0M
{
260
34.0M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
261
34.0M
               NULL, NULL, NULL, 0, "%s", msg);
262
34.0M
}
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
67.5k
{
277
67.5k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING,
278
67.5k
               str1, str2, NULL, 0, msg, str1, str2);
279
67.5k
}
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
250k
{
295
250k
    ctxt->valid = 0;
296
297
250k
    xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR,
298
250k
               str1, str2, NULL, 0, msg, str1, str2);
299
250k
}
300
#endif
301
302
/**
303
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
304
 *
305
 * @param ctxt  an XML parser context
306
 * @param error  the error number
307
 * @param msg  the error message
308
 * @param val  an integer value
309
 */
310
static void LIBXML_ATTR_FORMAT(3,0)
311
xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error,
312
                  const char *msg, int val)
313
7.94M
{
314
7.94M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
315
7.94M
               NULL, NULL, NULL, val, msg, val);
316
7.94M
}
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
124k
{
333
124k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
334
124k
               str1, str2, NULL, val, msg, str1, val, str2);
335
124k
}
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
511k
{
349
511k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
350
511k
               val, NULL, NULL, 0, msg, val);
351
511k
}
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
31.5k
{
365
31.5k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR,
366
31.5k
               val, NULL, NULL, 0, msg, val);
367
31.5k
}
368
369
/**
370
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
371
 *
372
 * @param ctxt  an XML parser context
373
 * @param error  the error number
374
 * @param msg  the message
375
 * @param info1  extra information string
376
 * @param info2  extra information string
377
 * @param info3  extra information string
378
 */
379
static void LIBXML_ATTR_FORMAT(3,0)
380
xmlNsErr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
381
         const char *msg,
382
         const xmlChar * info1, const xmlChar * info2,
383
         const xmlChar * info3)
384
168k
{
385
168k
    ctxt->nsWellFormed = 0;
386
387
168k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR,
388
168k
               info1, info2, info3, 0, msg, info1, info2, info3);
389
168k
}
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
6.46k
{
407
6.46k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING,
408
6.46k
               info1, info2, info3, 0, msg, info1, info2, info3);
409
6.46k
}
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
2.80M
{
436
2.80M
    unsigned long consumed;
437
2.80M
    unsigned long *expandedSize;
438
2.80M
    xmlParserInputPtr input = ctxt->input;
439
2.80M
    xmlEntityPtr entity = input->entity;
440
441
2.80M
    if ((entity) && (entity->flags & XML_ENT_CHECKED))
442
259k
        return(0);
443
444
    /*
445
     * Compute total consumed bytes so far, including input streams of
446
     * external entities.
447
     */
448
2.54M
    consumed = input->consumed;
449
2.54M
    xmlSaturatedAddSizeT(&consumed, input->cur - input->base);
450
2.54M
    xmlSaturatedAdd(&consumed, ctxt->sizeentities);
451
452
2.54M
    if (entity)
453
105k
        expandedSize = &entity->expandedSize;
454
2.44M
    else
455
2.44M
        expandedSize = &ctxt->sizeentcopy;
456
457
    /*
458
     * Add extra cost and some fixed cost.
459
     */
460
2.54M
    xmlSaturatedAdd(expandedSize, extra);
461
2.54M
    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
2.54M
    if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) &&
469
853k
        ((*expandedSize >= ULONG_MAX) ||
470
853k
         (*expandedSize / ctxt->maxAmpl > consumed))) {
471
238
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
472
238
                       "Maximum entity amplification factor exceeded, see "
473
238
                       "xmlCtxtSetMaxAmplification.\n");
474
238
        return(1);
475
238
    }
476
477
2.54M
    return(0);
478
2.54M
}
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
513k
xmlSBufInit(xmlSBuf *buf, unsigned max) {
683
513k
    buf->mem = NULL;
684
513k
    buf->size = 0;
685
513k
    buf->cap = 0;
686
513k
    buf->max = max;
687
513k
    buf->code = XML_ERR_OK;
688
513k
}
689
690
static int
691
399k
xmlSBufGrow(xmlSBuf *buf, unsigned len) {
692
399k
    xmlChar *mem;
693
399k
    unsigned cap;
694
695
399k
    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
399k
    cap = (buf->size + len) * 2;
702
399k
    if (cap < 240)
703
302k
        cap = 240;
704
705
399k
    mem = xmlRealloc(buf->mem, cap);
706
399k
    if (mem == NULL) {
707
99
        buf->code = XML_ERR_NO_MEMORY;
708
99
        return(-1);
709
99
    }
710
711
399k
    buf->mem = mem;
712
399k
    buf->cap = cap;
713
714
399k
    return(0);
715
399k
}
716
717
static void
718
118M
xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) {
719
118M
    if (buf->max - buf->size < len) {
720
0
        if (buf->code == XML_ERR_OK)
721
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
722
0
        return;
723
0
    }
724
725
118M
    if (buf->cap - buf->size <= len) {
726
392k
        if (xmlSBufGrow(buf, len) < 0)
727
92
            return;
728
392k
    }
729
730
118M
    if (len > 0)
731
118M
        memcpy(buf->mem + buf->size, str, len);
732
118M
    buf->size += len;
733
118M
}
734
735
static void
736
110M
xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) {
737
110M
    xmlSBufAddString(buf, (const xmlChar *) str, len);
738
110M
}
739
740
static void
741
1.25M
xmlSBufAddChar(xmlSBuf *buf, int c) {
742
1.25M
    xmlChar *end;
743
744
1.25M
    if (buf->max - buf->size < 4) {
745
0
        if (buf->code == XML_ERR_OK)
746
0
            buf->code = XML_ERR_RESOURCE_LIMIT;
747
0
        return;
748
0
    }
749
750
1.25M
    if (buf->cap - buf->size <= 4) {
751
6.49k
        if (xmlSBufGrow(buf, 4) < 0)
752
7
            return;
753
6.49k
    }
754
755
1.25M
    end = buf->mem + buf->size;
756
757
1.25M
    if (c < 0x80) {
758
1.24M
        *end = (xmlChar) c;
759
1.24M
        buf->size += 1;
760
1.24M
    } else {
761
3.35k
        buf->size += xmlCopyCharMultiByte(end, c);
762
3.35k
    }
763
1.25M
}
764
765
static void
766
102M
xmlSBufAddReplChar(xmlSBuf *buf) {
767
102M
    xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3);
768
102M
}
769
770
static void
771
104
xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
772
104
    if (buf->code == XML_ERR_NO_MEMORY)
773
104
        xmlCtxtErrMemory(ctxt);
774
0
    else
775
0
        xmlFatalErr(ctxt, buf->code, errMsg);
776
104
}
777
778
static xmlChar *
779
xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt,
780
298k
              const char *errMsg) {
781
298k
    if (buf->mem == NULL) {
782
24.6k
        buf->mem = xmlMalloc(1);
783
24.6k
        if (buf->mem == NULL) {
784
5
            buf->code = XML_ERR_NO_MEMORY;
785
24.6k
        } else {
786
24.6k
            buf->mem[0] = 0;
787
24.6k
        }
788
273k
    } else {
789
273k
        buf->mem[buf->size] = 0;
790
273k
    }
791
792
298k
    if (buf->code == XML_ERR_OK) {
793
298k
        if (sizeOut != NULL)
794
30.6k
            *sizeOut = buf->size;
795
298k
        return(buf->mem);
796
298k
    }
797
798
64
    xmlSBufReportError(buf, ctxt, errMsg);
799
800
64
    xmlFree(buf->mem);
801
802
64
    if (sizeOut != NULL)
803
10
        *sizeOut = 0;
804
64
    return(NULL);
805
298k
}
806
807
static void
808
206k
xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
809
206k
    if (buf->code != XML_ERR_OK)
810
40
        xmlSBufReportError(buf, ctxt, errMsg);
811
812
206k
    xmlFree(buf->mem);
813
206k
}
814
815
static int
816
xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str,
817
99.3M
                    const char *errMsg) {
818
99.3M
    int c = str[0];
819
99.3M
    int c1 = str[1];
820
821
99.3M
    if ((c1 & 0xC0) != 0x80)
822
47.4M
        goto encoding_error;
823
824
51.9M
    if (c < 0xE0) {
825
        /* 2-byte sequence */
826
22.0M
        if (c < 0xC2)
827
21.0M
            goto encoding_error;
828
829
902k
        return(2);
830
29.9M
    } else {
831
29.9M
        int c2 = str[2];
832
833
29.9M
        if ((c2 & 0xC0) != 0x80)
834
9.83k
            goto encoding_error;
835
836
29.8M
        if (c < 0xF0) {
837
            /* 3-byte sequence */
838
29.8M
            if (c == 0xE0) {
839
                /* overlong */
840
5.00M
                if (c1 < 0xA0)
841
1.02k
                    goto encoding_error;
842
24.8M
            } else if (c == 0xED) {
843
                /* surrogate */
844
7.99k
                if (c1 >= 0xA0)
845
206
                    goto encoding_error;
846
24.8M
            } else if (c == 0xEF) {
847
                /* U+FFFE and U+FFFF are invalid Chars */
848
19.6M
                if ((c1 == 0xBF) && (c2 >= 0xBE))
849
284
                    xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg);
850
19.6M
            }
851
852
29.8M
            return(3);
853
29.8M
        } else {
854
            /* 4-byte sequence */
855
25.9k
            if ((str[3] & 0xC0) != 0x80)
856
2.75k
                goto encoding_error;
857
23.1k
            if (c == 0xF0) {
858
                /* overlong */
859
1.61k
                if (c1 < 0x90)
860
803
                    goto encoding_error;
861
21.5k
            } else if (c >= 0xF4) {
862
                /* greater than 0x10FFFF */
863
3.72k
                if ((c > 0xF4) || (c1 >= 0x90))
864
3.41k
                    goto encoding_error;
865
3.72k
            }
866
867
18.9k
            return(4);
868
23.1k
        }
869
29.8M
    }
870
871
68.5M
encoding_error:
872
    /* Only report the first error */
873
68.5M
    if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) {
874
17.2k
        xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL);
875
17.2k
        ctxt->input->flags |= XML_INPUT_ENCODING_ERROR;
876
17.2k
    }
877
878
68.5M
    return(0);
879
51.9M
}
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
44.5k
xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) {
897
44.5k
    xmlSAXHandlerPtr sax;
898
899
    /* Avoid unused variable warning if features are disabled. */
900
44.5k
    (void) sax;
901
902
    /*
903
     * Changing the SAX struct directly is still widespread practice
904
     * in internal and external code.
905
     */
906
44.5k
    if (ctxt == NULL) return;
907
44.5k
    sax = ctxt->sax;
908
44.5k
#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
44.5k
    if (((ctxt->options & XML_PARSE_SAX1) == 0) &&
914
25.2k
        (sax) &&
915
25.2k
        (sax->initialized == XML_SAX2_MAGIC) &&
916
25.2k
        ((sax->startElementNs != NULL) ||
917
0
         (sax->endElementNs != NULL) ||
918
0
         ((sax->startElement == NULL) && (sax->endElement == NULL))))
919
25.2k
        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
44.5k
    ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3);
929
44.5k
    ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5);
930
44.5k
    ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36);
931
44.5k
    if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) ||
932
44.5k
    (ctxt->str_xml_ns == NULL)) {
933
11
        xmlErrMemory(ctxt);
934
11
    }
935
936
44.5k
    xmlDictSetLimit(ctxt->dict,
937
44.5k
                    (ctxt->options & XML_PARSE_HUGE) ?
938
21.8k
                        0 :
939
44.5k
                        XML_MAX_DICTIONARY_LIMIT);
940
941
44.5k
#ifdef LIBXML_VALID_ENABLED
942
44.5k
    if (ctxt->validate)
943
21.1k
        ctxt->vctxt.flags |= XML_VCTXT_VALIDATE;
944
23.3k
    else
945
23.3k
        ctxt->vctxt.flags &= ~XML_VCTXT_VALIDATE;
946
44.5k
#endif /* LIBXML_VALID_ENABLED */
947
44.5k
}
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
26.6k
{
989
26.6k
    if ((src == NULL) || (dst == NULL))
990
0
        return(NULL);
991
992
37.4k
    while (*src == 0x20) src++;
993
5.93M
    while (*src != 0) {
994
5.91M
  if (*src == 0x20) {
995
309k
      while (*src == 0x20) src++;
996
29.0k
      if (*src != 0)
997
28.4k
    *dst++ = 0x20;
998
5.88M
  } else {
999
5.88M
      *dst++ = *src++;
1000
5.88M
  }
1001
5.91M
    }
1002
26.6k
    *dst = 0;
1003
26.6k
    if (dst == src)
1004
21.7k
       return(NULL);
1005
4.93k
    return(dst);
1006
26.6k
}
1007
1008
/**
1009
 * Add a defaulted attribute for an element
1010
 *
1011
 * @param ctxt  an XML parser context
1012
 * @param fullname  the element fullname
1013
 * @param fullattr  the attribute fullname
1014
 * @param value  the attribute value
1015
 */
1016
static void
1017
xmlAddDefAttrs(xmlParserCtxtPtr ctxt,
1018
               const xmlChar *fullname,
1019
               const xmlChar *fullattr,
1020
22.0k
               const xmlChar *value) {
1021
22.0k
    xmlDefAttrsPtr defaults;
1022
22.0k
    xmlDefAttr *attr;
1023
22.0k
    int len, expandedSize;
1024
22.0k
    xmlHashedString name;
1025
22.0k
    xmlHashedString prefix;
1026
22.0k
    xmlHashedString hvalue;
1027
22.0k
    const xmlChar *localname;
1028
1029
    /*
1030
     * Allows to detect attribute redefinitions
1031
     */
1032
22.0k
    if (ctxt->attsSpecial != NULL) {
1033
19.7k
        if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL)
1034
13.0k
      return;
1035
19.7k
    }
1036
1037
8.99k
    if (ctxt->attsDefault == NULL) {
1038
2.39k
        ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict);
1039
2.39k
  if (ctxt->attsDefault == NULL)
1040
1
      goto mem_error;
1041
2.39k
    }
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
8.99k
    localname = xmlSplitQName3(fullname, &len);
1048
8.99k
    if (localname == NULL) {
1049
8.62k
        name = xmlDictLookupHashed(ctxt->dict, fullname, -1);
1050
8.62k
  prefix.name = NULL;
1051
8.62k
    } else {
1052
366
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1053
366
  prefix = xmlDictLookupHashed(ctxt->dict, fullname, len);
1054
366
        if (prefix.name == NULL)
1055
1
            goto mem_error;
1056
366
    }
1057
8.98k
    if (name.name == NULL)
1058
1
        goto mem_error;
1059
1060
    /*
1061
     * make sure there is some storage
1062
     */
1063
8.98k
    defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name);
1064
8.98k
    if ((defaults == NULL) ||
1065
6.16k
        (defaults->nbAttrs >= defaults->maxAttrs)) {
1066
3.44k
        xmlDefAttrsPtr temp;
1067
3.44k
        int newSize;
1068
1069
3.44k
        if (defaults == NULL) {
1070
2.82k
            newSize = 4;
1071
2.82k
        } else {
1072
619
            if ((defaults->maxAttrs >= XML_MAX_ATTRS) ||
1073
619
                ((size_t) defaults->maxAttrs >
1074
619
                     SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults)))
1075
0
                goto mem_error;
1076
1077
619
            if (defaults->maxAttrs > XML_MAX_ATTRS / 2)
1078
0
                newSize = XML_MAX_ATTRS;
1079
619
            else
1080
619
                newSize = defaults->maxAttrs * 2;
1081
619
        }
1082
3.44k
        temp = xmlRealloc(defaults,
1083
3.44k
                          sizeof(*defaults) + newSize * sizeof(xmlDefAttr));
1084
3.44k
  if (temp == NULL)
1085
2
      goto mem_error;
1086
3.44k
        if (defaults == NULL)
1087
2.82k
            temp->nbAttrs = 0;
1088
3.44k
  temp->maxAttrs = newSize;
1089
3.44k
        defaults = temp;
1090
3.44k
  if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name,
1091
3.44k
                          defaults, NULL) < 0) {
1092
0
      xmlFree(defaults);
1093
0
      goto mem_error;
1094
0
  }
1095
3.44k
    }
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
8.98k
    localname = xmlSplitQName3(fullattr, &len);
1102
8.98k
    if (localname == NULL) {
1103
6.15k
        name = xmlDictLookupHashed(ctxt->dict, fullattr, -1);
1104
6.15k
  prefix.name = NULL;
1105
6.15k
    } else {
1106
2.83k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1107
2.83k
  prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len);
1108
2.83k
        if (prefix.name == NULL)
1109
1
            goto mem_error;
1110
2.83k
    }
1111
8.98k
    if (name.name == NULL)
1112
2
        goto mem_error;
1113
1114
    /* intern the string and precompute the end */
1115
8.98k
    len = strlen((const char *) value);
1116
8.98k
    hvalue = xmlDictLookupHashed(ctxt->dict, value, len);
1117
8.98k
    if (hvalue.name == NULL)
1118
1
        goto mem_error;
1119
1120
8.98k
    expandedSize = strlen((const char *) name.name);
1121
8.98k
    if (prefix.name != NULL)
1122
2.83k
        expandedSize += strlen((const char *) prefix.name);
1123
8.98k
    expandedSize += len;
1124
1125
8.98k
    attr = &defaults->attrs[defaults->nbAttrs++];
1126
8.98k
    attr->name = name;
1127
8.98k
    attr->prefix = prefix;
1128
8.98k
    attr->value = hvalue;
1129
8.98k
    attr->valueEnd = hvalue.name + len;
1130
8.98k
    attr->external = PARSER_EXTERNAL(ctxt);
1131
8.98k
    attr->expandedSize = expandedSize;
1132
1133
8.98k
    return;
1134
1135
9
mem_error:
1136
9
    xmlErrMemory(ctxt);
1137
9
}
1138
1139
/**
1140
 * Register this attribute type
1141
 *
1142
 * @param ctxt  an XML parser context
1143
 * @param fullname  the element fullname
1144
 * @param fullattr  the attribute fullname
1145
 * @param type  the attribute type
1146
 */
1147
static void
1148
xmlAddSpecialAttr(xmlParserCtxtPtr ctxt,
1149
      const xmlChar *fullname,
1150
      const xmlChar *fullattr,
1151
      int type)
1152
42.5k
{
1153
42.5k
    if (ctxt->attsSpecial == NULL) {
1154
2.96k
        ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict);
1155
2.96k
  if (ctxt->attsSpecial == NULL)
1156
5
      goto mem_error;
1157
2.96k
    }
1158
1159
42.5k
    if (PARSER_EXTERNAL(ctxt))
1160
26.7k
        type |= XML_SPECIAL_EXTERNAL;
1161
1162
42.5k
    if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr,
1163
42.5k
                    XML_INT_TO_PTR(type)) < 0)
1164
1
        goto mem_error;
1165
42.5k
    return;
1166
1167
42.5k
mem_error:
1168
6
    xmlErrMemory(ctxt);
1169
6
}
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
21.5k
                            const xmlChar *unused ATTRIBUTE_UNUSED) {
1178
21.5k
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data;
1179
1180
21.5k
    if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) {
1181
2.09k
        xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL);
1182
2.09k
    }
1183
21.5k
}
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
18.1k
{
1195
18.1k
    if (ctxt->attsSpecial == NULL)
1196
15.1k
        return;
1197
1198
2.96k
    xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt);
1199
1200
2.96k
    if (xmlHashSize(ctxt->attsSpecial) == 0) {
1201
179
        xmlHashFree(ctxt->attsSpecial, NULL);
1202
179
        ctxt->attsSpecial = NULL;
1203
179
    }
1204
2.96k
}
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
69.7k
{
1268
69.7k
    const xmlChar *cur = lang, *nxt;
1269
1270
69.7k
    if (cur == NULL)
1271
283
        return (0);
1272
69.5k
    if (((cur[0] == 'i') && (cur[1] == '-')) ||
1273
69.3k
        ((cur[0] == 'I') && (cur[1] == '-')) ||
1274
66.9k
        ((cur[0] == 'x') && (cur[1] == '-')) ||
1275
66.7k
        ((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
2.79k
        cur += 2;
1282
6.82k
        while (((cur[0] >= 'A') && (cur[0] <= 'Z')) ||
1283
3.18k
               ((cur[0] >= 'a') && (cur[0] <= 'z')))
1284
4.03k
            cur++;
1285
2.79k
        return(cur[0] == 0);
1286
2.79k
    }
1287
66.7k
    nxt = cur;
1288
331k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1289
74.1k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1290
264k
           nxt++;
1291
66.7k
    if (nxt - cur >= 4) {
1292
        /*
1293
         * Reserved
1294
         */
1295
3.28k
        if ((nxt - cur > 8) || (nxt[0] != 0))
1296
2.84k
            return(0);
1297
442
        return(1);
1298
3.28k
    }
1299
63.4k
    if (nxt - cur < 2)
1300
3.64k
        return(0);
1301
    /* we got an ISO 639 code */
1302
59.7k
    if (nxt[0] == 0)
1303
2.36k
        return(1);
1304
57.4k
    if (nxt[0] != '-')
1305
4.17k
        return(0);
1306
1307
53.2k
    nxt++;
1308
53.2k
    cur = nxt;
1309
    /* now we can have extlang or script or region or variant */
1310
53.2k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1311
15.8k
        goto region_m49;
1312
1313
139k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1314
40.8k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1315
102k
           nxt++;
1316
37.3k
    if (nxt - cur == 4)
1317
1.39k
        goto script;
1318
35.9k
    if (nxt - cur == 2)
1319
888
        goto region;
1320
35.0k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1321
297
        goto variant;
1322
34.8k
    if (nxt - cur != 3)
1323
4.86k
        return(0);
1324
    /* we parsed an extlang */
1325
29.9k
    if (nxt[0] == 0)
1326
608
        return(1);
1327
29.3k
    if (nxt[0] != '-')
1328
11.7k
        return(0);
1329
1330
17.6k
    nxt++;
1331
17.6k
    cur = nxt;
1332
    /* now we can have script or region or variant */
1333
17.6k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1334
1.06k
        goto region_m49;
1335
1336
66.4k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1337
28.3k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1338
49.9k
           nxt++;
1339
16.5k
    if (nxt - cur == 2)
1340
13.9k
        goto region;
1341
2.59k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1342
911
        goto variant;
1343
1.68k
    if (nxt - cur != 4)
1344
816
        return(0);
1345
    /* we parsed a script */
1346
2.26k
script:
1347
2.26k
    if (nxt[0] == 0)
1348
752
        return(1);
1349
1.51k
    if (nxt[0] != '-')
1350
179
        return(0);
1351
1352
1.33k
    nxt++;
1353
1.33k
    cur = nxt;
1354
    /* now we can have region or variant */
1355
1.33k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1356
91
        goto region_m49;
1357
1358
6.37k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1359
3.69k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1360
5.13k
           nxt++;
1361
1362
1.24k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1363
208
        goto variant;
1364
1.03k
    if (nxt - cur != 2)
1365
413
        return(0);
1366
    /* we parsed a region */
1367
15.8k
region:
1368
15.8k
    if (nxt[0] == 0)
1369
164
        return(1);
1370
15.6k
    if (nxt[0] != '-')
1371
14.2k
        return(0);
1372
1373
1.44k
    nxt++;
1374
1.44k
    cur = nxt;
1375
    /* now we can just have a variant */
1376
8.21k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1377
5.11k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1378
6.76k
           nxt++;
1379
1380
1.44k
    if ((nxt - cur < 5) || (nxt - cur > 8))
1381
1.05k
        return(0);
1382
1383
    /* we parsed a variant */
1384
1.81k
variant:
1385
1.81k
    if (nxt[0] == 0)
1386
331
        return(1);
1387
1.47k
    if (nxt[0] != '-')
1388
1.05k
        return(0);
1389
    /* extensions and private use subtags not checked */
1390
424
    return (1);
1391
1392
17.0k
region_m49:
1393
17.0k
    if (((nxt[1] >= '0') && (nxt[1] <= '9')) &&
1394
12.5k
        ((nxt[2] >= '0') && (nxt[2] <= '9'))) {
1395
353
        nxt += 3;
1396
353
        goto region;
1397
353
    }
1398
16.6k
    return(0);
1399
17.0k
}
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
43.8k
xmlParserNsCreate(void) {
1417
43.8k
    xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb));
1418
1419
43.8k
    if (nsdb == NULL)
1420
1
        return(NULL);
1421
43.8k
    memset(nsdb, 0, sizeof(*nsdb));
1422
43.8k
    nsdb->defaultNsIndex = INT_MAX;
1423
1424
43.8k
    return(nsdb);
1425
43.8k
}
1426
1427
/**
1428
 * Free a namespace database.
1429
 *
1430
 * @param nsdb  namespace database
1431
 */
1432
void
1433
43.8k
xmlParserNsFree(xmlParserNsData *nsdb) {
1434
43.8k
    if (nsdb == NULL)
1435
0
        return;
1436
1437
43.8k
    xmlFree(nsdb->extra);
1438
43.8k
    xmlFree(nsdb->hash);
1439
43.8k
    xmlFree(nsdb);
1440
43.8k
}
1441
1442
/**
1443
 * Reset a namespace database.
1444
 *
1445
 * @param nsdb  namespace database
1446
 */
1447
static void
1448
0
xmlParserNsReset(xmlParserNsData *nsdb) {
1449
0
    if (nsdb == NULL)
1450
0
        return;
1451
1452
0
    nsdb->hashElems = 0;
1453
0
    nsdb->elementId = 0;
1454
0
    nsdb->defaultNsIndex = INT_MAX;
1455
1456
0
    if (nsdb->hash)
1457
0
        memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0]));
1458
0
}
1459
1460
/**
1461
 * Signal that a new element has started.
1462
 *
1463
 * @param nsdb  namespace database
1464
 * @returns 0 on success, -1 if the element counter overflowed.
1465
 */
1466
static int
1467
827k
xmlParserNsStartElement(xmlParserNsData *nsdb) {
1468
827k
    if (nsdb->elementId == UINT_MAX)
1469
0
        return(-1);
1470
827k
    nsdb->elementId++;
1471
1472
827k
    return(0);
1473
827k
}
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
1.59M
                  xmlParserNsBucket **bucketPtr) {
1489
1.59M
    xmlParserNsBucket *bucket, *tombstone;
1490
1.59M
    unsigned index, hashValue;
1491
1492
1.59M
    if (prefix->name == NULL)
1493
796k
        return(ctxt->nsdb->defaultNsIndex);
1494
1495
794k
    if (ctxt->nsdb->hashSize == 0)
1496
25.1k
        return(INT_MAX);
1497
1498
768k
    hashValue = prefix->hashValue;
1499
768k
    index = hashValue & (ctxt->nsdb->hashSize - 1);
1500
768k
    bucket = &ctxt->nsdb->hash[index];
1501
768k
    tombstone = NULL;
1502
1503
1.17M
    while (bucket->hashValue) {
1504
1.02M
        if (bucket->index == INT_MAX) {
1505
61.7k
            if (tombstone == NULL)
1506
53.7k
                tombstone = bucket;
1507
958k
        } else if (bucket->hashValue == hashValue) {
1508
614k
            if (ctxt->nsTab[bucket->index * 2] == prefix->name) {
1509
614k
                if (bucketPtr != NULL)
1510
380k
                    *bucketPtr = bucket;
1511
614k
                return(bucket->index);
1512
614k
            }
1513
614k
        }
1514
1515
405k
        index++;
1516
405k
        bucket++;
1517
405k
        if (index == ctxt->nsdb->hashSize) {
1518
95.6k
            index = 0;
1519
95.6k
            bucket = ctxt->nsdb->hash;
1520
95.6k
        }
1521
405k
    }
1522
1523
154k
    if (bucketPtr != NULL)
1524
39.7k
        *bucketPtr = tombstone ? tombstone : bucket;
1525
154k
    return(INT_MAX);
1526
768k
}
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
800k
xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) {
1537
800k
    const xmlChar *ret;
1538
800k
    int nsIndex;
1539
1540
800k
    if (prefix->name == ctxt->str_xml)
1541
467
        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
800k
    nsIndex = xmlParserNsLookup(ctxt, prefix, NULL);
1548
800k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1549
701k
        return(NULL);
1550
1551
98.7k
    ret = ctxt->nsTab[nsIndex * 2 + 1];
1552
98.7k
    if (ret[0] == 0)
1553
8.99k
        ret = NULL;
1554
98.7k
    return(ret);
1555
800k
}
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
142k
xmlParserNsLookupSax(xmlParserCtxt *ctxt, const xmlChar *prefix) {
1567
142k
    xmlHashedString hprefix;
1568
142k
    int nsIndex;
1569
1570
142k
    if (prefix == ctxt->str_xml)
1571
91.8k
        return(NULL);
1572
1573
50.4k
    hprefix.name = prefix;
1574
50.4k
    if (prefix != NULL)
1575
26.7k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1576
23.7k
    else
1577
23.7k
        hprefix.hashValue = 0;
1578
50.4k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1579
50.4k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1580
0
        return(NULL);
1581
1582
50.4k
    return(ctxt->nsdb->extra[nsIndex].saxData);
1583
50.4k
}
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
178k
                     void *saxData) {
1598
178k
    xmlHashedString hprefix;
1599
178k
    int nsIndex;
1600
1601
178k
    if (prefix == ctxt->str_xml)
1602
0
        return(-1);
1603
1604
178k
    hprefix.name = prefix;
1605
178k
    if (prefix != NULL)
1606
148k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1607
29.7k
    else
1608
29.7k
        hprefix.hashValue = 0;
1609
178k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1610
178k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1611
0
        return(-1);
1612
1613
178k
    ctxt->nsdb->extra[nsIndex].saxData = saxData;
1614
178k
    return(0);
1615
178k
}
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
14.3k
xmlParserNsGrow(xmlParserCtxtPtr ctxt) {
1625
14.3k
    const xmlChar **table;
1626
14.3k
    xmlParserNsExtra *extra;
1627
14.3k
    int newSize;
1628
1629
14.3k
    newSize = xmlGrowCapacity(ctxt->nsMax,
1630
14.3k
                              sizeof(table[0]) + sizeof(extra[0]),
1631
14.3k
                              16, XML_MAX_ITEMS);
1632
14.3k
    if (newSize < 0)
1633
0
        goto error;
1634
1635
14.3k
    table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0]));
1636
14.3k
    if (table == NULL)
1637
15
        goto error;
1638
14.3k
    ctxt->nsTab = table;
1639
1640
14.3k
    extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0]));
1641
14.3k
    if (extra == NULL)
1642
11
        goto error;
1643
14.3k
    ctxt->nsdb->extra = extra;
1644
1645
14.3k
    ctxt->nsMax = newSize;
1646
14.3k
    return(0);
1647
1648
26
error:
1649
26
    xmlErrMemory(ctxt);
1650
26
    return(-1);
1651
14.3k
}
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
288k
                const xmlHashedString *uri, void *saxData, int defAttr) {
1667
288k
    xmlParserNsBucket *bucket = NULL;
1668
288k
    xmlParserNsExtra *extra;
1669
288k
    const xmlChar **ns;
1670
288k
    unsigned hashValue, nsIndex, oldIndex;
1671
1672
288k
    if ((prefix != NULL) && (prefix->name == ctxt->str_xml))
1673
196
        return(0);
1674
1675
287k
    if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) {
1676
26
        xmlErrMemory(ctxt);
1677
26
        return(-1);
1678
26
    }
1679
1680
    /*
1681
     * Default namespace and 'xml' namespace
1682
     */
1683
287k
    if ((prefix == NULL) || (prefix->name == NULL)) {
1684
45.6k
        oldIndex = ctxt->nsdb->defaultNsIndex;
1685
1686
45.6k
        if (oldIndex != INT_MAX) {
1687
42.8k
            extra = &ctxt->nsdb->extra[oldIndex];
1688
1689
42.8k
            if (extra->elementId == ctxt->nsdb->elementId) {
1690
475
                if (defAttr == 0)
1691
181
                    xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns");
1692
475
                return(0);
1693
475
            }
1694
1695
42.3k
            if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1696
14.7k
                (uri->name == ctxt->nsTab[oldIndex * 2 + 1]))
1697
9.79k
                return(0);
1698
42.3k
        }
1699
1700
35.4k
        ctxt->nsdb->defaultNsIndex = ctxt->nsNr;
1701
35.4k
        goto populate_entry;
1702
45.6k
    }
1703
1704
    /*
1705
     * Hash table lookup
1706
     */
1707
242k
    oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket);
1708
242k
    if (oldIndex != INT_MAX) {
1709
195k
        extra = &ctxt->nsdb->extra[oldIndex];
1710
1711
        /*
1712
         * Check for duplicate definitions on the same element.
1713
         */
1714
195k
        if (extra->elementId == ctxt->nsdb->elementId) {
1715
985
            if (defAttr == 0)
1716
790
                xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name);
1717
985
            return(0);
1718
985
        }
1719
1720
194k
        if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1721
7.89k
            (uri->name == ctxt->nsTab[bucket->index * 2 + 1]))
1722
6.71k
            return(0);
1723
1724
187k
        bucket->index = ctxt->nsNr;
1725
187k
        goto populate_entry;
1726
194k
    }
1727
1728
    /*
1729
     * Insert new bucket
1730
     */
1731
1732
46.7k
    hashValue = prefix->hashValue;
1733
1734
    /*
1735
     * Grow hash table, 50% fill factor
1736
     */
1737
46.7k
    if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) {
1738
7.70k
        xmlParserNsBucket *newHash;
1739
7.70k
        unsigned newSize, i, index;
1740
1741
7.70k
        if (ctxt->nsdb->hashSize > UINT_MAX / 2) {
1742
0
            xmlErrMemory(ctxt);
1743
0
            return(-1);
1744
0
        }
1745
7.70k
        newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16;
1746
7.70k
        newHash = xmlMalloc(newSize * sizeof(newHash[0]));
1747
7.70k
        if (newHash == NULL) {
1748
4
            xmlErrMemory(ctxt);
1749
4
            return(-1);
1750
4
        }
1751
7.69k
        memset(newHash, 0, newSize * sizeof(newHash[0]));
1752
1753
135k
        for (i = 0; i < ctxt->nsdb->hashSize; i++) {
1754
127k
            unsigned hv = ctxt->nsdb->hash[i].hashValue;
1755
127k
            unsigned newIndex;
1756
1757
127k
            if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX))
1758
126k
                continue;
1759
1.50k
            newIndex = hv & (newSize - 1);
1760
1761
2.40k
            while (newHash[newIndex].hashValue != 0) {
1762
900
                newIndex++;
1763
900
                if (newIndex == newSize)
1764
174
                    newIndex = 0;
1765
900
            }
1766
1767
1.50k
            newHash[newIndex] = ctxt->nsdb->hash[i];
1768
1.50k
        }
1769
1770
7.69k
        xmlFree(ctxt->nsdb->hash);
1771
7.69k
        ctxt->nsdb->hash = newHash;
1772
7.69k
        ctxt->nsdb->hashSize = newSize;
1773
1774
        /*
1775
         * Relookup
1776
         */
1777
7.69k
        index = hashValue & (newSize - 1);
1778
1779
8.10k
        while (newHash[index].hashValue != 0) {
1780
411
            index++;
1781
411
            if (index == newSize)
1782
67
                index = 0;
1783
411
        }
1784
1785
7.69k
        bucket = &newHash[index];
1786
7.69k
    }
1787
1788
46.7k
    bucket->hashValue = hashValue;
1789
46.7k
    bucket->index = ctxt->nsNr;
1790
46.7k
    ctxt->nsdb->hashElems++;
1791
46.7k
    oldIndex = INT_MAX;
1792
1793
269k
populate_entry:
1794
269k
    nsIndex = ctxt->nsNr;
1795
1796
269k
    ns = &ctxt->nsTab[nsIndex * 2];
1797
269k
    ns[0] = prefix ? prefix->name : NULL;
1798
269k
    ns[1] = uri->name;
1799
1800
269k
    extra = &ctxt->nsdb->extra[nsIndex];
1801
269k
    extra->saxData = saxData;
1802
269k
    extra->prefixHashValue = prefix ? prefix->hashValue : 0;
1803
269k
    extra->uriHashValue = uri->hashValue;
1804
269k
    extra->elementId = ctxt->nsdb->elementId;
1805
269k
    extra->oldIndex = oldIndex;
1806
1807
269k
    ctxt->nsNr++;
1808
1809
269k
    return(1);
1810
46.7k
}
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
103k
{
1822
103k
    int i;
1823
1824
    /* assert(nr <= ctxt->nsNr); */
1825
1826
307k
    for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) {
1827
203k
        const xmlChar *prefix = ctxt->nsTab[i * 2];
1828
203k
        xmlParserNsExtra *extra = &ctxt->nsdb->extra[i];
1829
1830
203k
        if (prefix == NULL) {
1831
18.1k
            ctxt->nsdb->defaultNsIndex = extra->oldIndex;
1832
185k
        } else {
1833
185k
            xmlHashedString hprefix;
1834
185k
            xmlParserNsBucket *bucket = NULL;
1835
1836
185k
            hprefix.name = prefix;
1837
185k
            hprefix.hashValue = extra->prefixHashValue;
1838
185k
            xmlParserNsLookup(ctxt, &hprefix, &bucket);
1839
            /* assert(bucket && bucket->hashValue); */
1840
185k
            bucket->index = extra->oldIndex;
1841
185k
        }
1842
203k
    }
1843
1844
103k
    ctxt->nsNr -= nr;
1845
103k
    return(nr);
1846
103k
}
1847
1848
static int
1849
21.4k
xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) {
1850
21.4k
    const xmlChar **atts;
1851
21.4k
    unsigned *attallocs;
1852
21.4k
    int newSize;
1853
1854
21.4k
    newSize = xmlGrowCapacity(ctxt->maxatts / 5,
1855
21.4k
                              sizeof(atts[0]) * 5 + sizeof(attallocs[0]),
1856
21.4k
                              10, XML_MAX_ATTRS);
1857
21.4k
    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
21.4k
    atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5);
1864
21.4k
    if (atts == NULL)
1865
6
        goto mem_error;
1866
21.4k
    ctxt->atts = atts;
1867
1868
21.4k
    attallocs = xmlRealloc(ctxt->attallocs,
1869
21.4k
                           newSize * sizeof(attallocs[0]));
1870
21.4k
    if (attallocs == NULL)
1871
6
        goto mem_error;
1872
21.4k
    ctxt->attallocs = attallocs;
1873
1874
21.4k
    ctxt->maxatts = newSize * 5;
1875
1876
21.4k
    return(0);
1877
1878
12
mem_error:
1879
12
    xmlErrMemory(ctxt);
1880
12
    return(-1);
1881
21.4k
}
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
271k
{
1893
271k
    char *directory = NULL;
1894
271k
    int maxDepth;
1895
1896
271k
    if ((ctxt == NULL) || (value == NULL))
1897
3.22k
        return(-1);
1898
1899
268k
    maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
1900
1901
268k
    if (ctxt->inputNr >= ctxt->inputMax) {
1902
7.21k
        xmlParserInputPtr *tmp;
1903
7.21k
        int newSize;
1904
1905
7.21k
        newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]),
1906
7.21k
                                  5, maxDepth);
1907
7.21k
        if (newSize < 0) {
1908
1
            xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
1909
1
                           "Maximum entity nesting depth exceeded");
1910
1
            return(-1);
1911
1
        }
1912
7.20k
        tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0]));
1913
7.20k
        if (tmp == NULL) {
1914
1
            xmlErrMemory(ctxt);
1915
1
            return(-1);
1916
1
        }
1917
7.20k
        ctxt->inputTab = tmp;
1918
7.20k
        ctxt->inputMax = newSize;
1919
7.20k
    }
1920
1921
268k
    if ((ctxt->inputNr == 0) && (value->filename != NULL)) {
1922
122k
        directory = xmlParserGetDirectory(value->filename);
1923
122k
        if (directory == NULL) {
1924
8
            xmlErrMemory(ctxt);
1925
8
            return(-1);
1926
8
        }
1927
122k
    }
1928
1929
268k
    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
268k
    ctxt->inputTab[ctxt->inputNr] = value;
1935
268k
    ctxt->input = value;
1936
1937
268k
    if (ctxt->inputNr == 0) {
1938
158k
        xmlFree(ctxt->directory);
1939
158k
        ctxt->directory = directory;
1940
158k
    }
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
268k
    value->id = ctxt->input_id++;
1948
1949
268k
    return(ctxt->inputNr++);
1950
268k
}
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
354k
{
1961
354k
    xmlParserInputPtr ret;
1962
1963
354k
    if (ctxt == NULL)
1964
0
        return(NULL);
1965
354k
    if (ctxt->inputNr <= 0)
1966
87.6k
        return (NULL);
1967
267k
    ctxt->inputNr--;
1968
267k
    if (ctxt->inputNr > 0)
1969
110k
        ctxt->input = ctxt->inputTab[ctxt->inputNr - 1];
1970
156k
    else
1971
156k
        ctxt->input = NULL;
1972
267k
    ret = ctxt->inputTab[ctxt->inputNr];
1973
267k
    ctxt->inputTab[ctxt->inputNr] = NULL;
1974
267k
    return (ret);
1975
354k
}
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
838k
{
1989
838k
    if (ctxt == NULL)
1990
0
        return(0);
1991
1992
838k
    if (ctxt->nodeNr >= ctxt->nodeMax) {
1993
46.5k
        int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
1994
46.5k
        xmlNodePtr *tmp;
1995
46.5k
        int newSize;
1996
1997
46.5k
        newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]),
1998
46.5k
                                  10, maxDepth);
1999
46.5k
        if (newSize < 0) {
2000
49
            xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
2001
49
                    "Excessive depth in document: %d,"
2002
49
                    " use XML_PARSE_HUGE option\n",
2003
49
                    ctxt->nodeNr);
2004
49
            return(-1);
2005
49
        }
2006
2007
46.4k
  tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0]));
2008
46.4k
        if (tmp == NULL) {
2009
13
            xmlErrMemory(ctxt);
2010
13
            return (-1);
2011
13
        }
2012
46.4k
        ctxt->nodeTab = tmp;
2013
46.4k
  ctxt->nodeMax = newSize;
2014
46.4k
    }
2015
2016
838k
    ctxt->nodeTab[ctxt->nodeNr] = value;
2017
838k
    ctxt->node = value;
2018
838k
    return (ctxt->nodeNr++);
2019
838k
}
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
752k
{
2032
752k
    xmlNodePtr ret;
2033
2034
752k
    if (ctxt == NULL) return(NULL);
2035
752k
    if (ctxt->nodeNr <= 0)
2036
54.9k
        return (NULL);
2037
697k
    ctxt->nodeNr--;
2038
697k
    if (ctxt->nodeNr > 0)
2039
682k
        ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1];
2040
15.0k
    else
2041
15.0k
        ctxt->node = NULL;
2042
697k
    ret = ctxt->nodeTab[ctxt->nodeNr];
2043
697k
    ctxt->nodeTab[ctxt->nodeNr] = NULL;
2044
697k
    return (ret);
2045
752k
}
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
944k
{
2062
944k
    xmlStartTag *tag;
2063
2064
944k
    if (ctxt->nameNr >= ctxt->nameMax) {
2065
42.4k
        const xmlChar **tmp;
2066
42.4k
        xmlStartTag *tmp2;
2067
42.4k
        int newSize;
2068
2069
42.4k
        newSize = xmlGrowCapacity(ctxt->nameMax,
2070
42.4k
                                  sizeof(tmp[0]) + sizeof(tmp2[0]),
2071
42.4k
                                  10, XML_MAX_ITEMS);
2072
42.4k
        if (newSize < 0)
2073
0
            goto mem_error;
2074
2075
42.4k
        tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0]));
2076
42.4k
        if (tmp == NULL)
2077
13
      goto mem_error;
2078
42.4k
  ctxt->nameTab = tmp;
2079
2080
42.4k
        tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0]));
2081
42.4k
        if (tmp2 == NULL)
2082
20
      goto mem_error;
2083
42.3k
  ctxt->pushTab = tmp2;
2084
2085
42.3k
        ctxt->nameMax = newSize;
2086
902k
    } else if (ctxt->pushTab == NULL) {
2087
26.6k
        ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0]));
2088
26.6k
        if (ctxt->pushTab == NULL)
2089
8
            goto mem_error;
2090
26.6k
    }
2091
944k
    ctxt->nameTab[ctxt->nameNr] = value;
2092
944k
    ctxt->name = value;
2093
944k
    tag = &ctxt->pushTab[ctxt->nameNr];
2094
944k
    tag->prefix = prefix;
2095
944k
    tag->URI = URI;
2096
944k
    tag->line = line;
2097
944k
    tag->nsNr = nsNr;
2098
944k
    return (ctxt->nameNr++);
2099
41
mem_error:
2100
41
    xmlErrMemory(ctxt);
2101
41
    return (-1);
2102
944k
}
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
32.1k
{
2113
32.1k
    const xmlChar *ret;
2114
2115
32.1k
    if (ctxt->nameNr <= 0)
2116
0
        return (NULL);
2117
32.1k
    ctxt->nameNr--;
2118
32.1k
    if (ctxt->nameNr > 0)
2119
30.8k
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2120
1.37k
    else
2121
1.37k
        ctxt->name = NULL;
2122
32.1k
    ret = ctxt->nameTab[ctxt->nameNr];
2123
32.1k
    ctxt->nameTab[ctxt->nameNr] = NULL;
2124
32.1k
    return (ret);
2125
32.1k
}
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
771k
{
2139
771k
    const xmlChar *ret;
2140
2141
771k
    if ((ctxt == NULL) || (ctxt->nameNr <= 0))
2142
2
        return (NULL);
2143
771k
    ctxt->nameNr--;
2144
771k
    if (ctxt->nameNr > 0)
2145
763k
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2146
7.90k
    else
2147
7.90k
        ctxt->name = NULL;
2148
771k
    ret = ctxt->nameTab[ctxt->nameNr];
2149
771k
    ctxt->nameTab[ctxt->nameNr] = NULL;
2150
771k
    return (ret);
2151
771k
}
2152
2153
1.13M
static int spacePush(xmlParserCtxtPtr ctxt, int val) {
2154
1.13M
    if (ctxt->spaceNr >= ctxt->spaceMax) {
2155
78.9k
        int *tmp;
2156
78.9k
        int newSize;
2157
2158
78.9k
        newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]),
2159
78.9k
                                  10, XML_MAX_ITEMS);
2160
78.9k
        if (newSize < 0) {
2161
0
      xmlErrMemory(ctxt);
2162
0
      return(-1);
2163
0
        }
2164
2165
78.9k
        tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0]));
2166
78.9k
        if (tmp == NULL) {
2167
36
      xmlErrMemory(ctxt);
2168
36
      return(-1);
2169
36
  }
2170
78.9k
  ctxt->spaceTab = tmp;
2171
2172
78.9k
        ctxt->spaceMax = newSize;
2173
78.9k
    }
2174
1.13M
    ctxt->spaceTab[ctxt->spaceNr] = val;
2175
1.13M
    ctxt->space = &ctxt->spaceTab[ctxt->spaceNr];
2176
1.13M
    return(ctxt->spaceNr++);
2177
1.13M
}
2178
2179
997k
static int spacePop(xmlParserCtxtPtr ctxt) {
2180
997k
    int ret;
2181
997k
    if (ctxt->spaceNr <= 0) return(0);
2182
997k
    ctxt->spaceNr--;
2183
997k
    if (ctxt->spaceNr > 0)
2184
997k
  ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1];
2185
14
    else
2186
14
        ctxt->space = &ctxt->spaceTab[0];
2187
997k
    ret = ctxt->spaceTab[ctxt->spaceNr];
2188
997k
    ctxt->spaceTab[ctxt->spaceNr] = -1;
2189
997k
    return(ret);
2190
997k
}
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
16.9M
#define RAW (*ctxt->input->cur)
2225
451M
#define CUR (*ctxt->input->cur)
2226
11.8M
#define NXT(val) ctxt->input->cur[(val)]
2227
436M
#define CUR_PTR ctxt->input->cur
2228
1.53M
#define BASE_PTR ctxt->input->base
2229
2230
#define CMP4( s, c1, c2, c3, c4 ) \
2231
18.6M
  ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \
2232
9.35M
    ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 )
2233
#define CMP5( s, c1, c2, c3, c4, c5 ) \
2234
18.0M
  ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 )
2235
#define CMP6( s, c1, c2, c3, c4, c5, c6 ) \
2236
16.9M
  ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 )
2237
#define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \
2238
16.1M
  ( 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
15.4M
  ( 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
7.51M
  ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \
2243
7.51M
    ((unsigned char *) s)[ 8 ] == c9 )
2244
#define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \
2245
1.20k
  ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \
2246
1.20k
    ((unsigned char *) s)[ 9 ] == c10 )
2247
2248
3.32M
#define SKIP(val) do {             \
2249
3.32M
    ctxt->input->cur += (val),ctxt->input->col+=(val);      \
2250
3.32M
    if (*ctxt->input->cur == 0)           \
2251
3.32M
        xmlParserGrow(ctxt);           \
2252
3.32M
  } 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
17.2M
    if (!PARSER_PROGRESSIVE(ctxt)) \
2268
17.2M
  xmlParserShrink(ctxt);
2269
2270
#define GROW \
2271
37.3M
    if ((!PARSER_PROGRESSIVE(ctxt)) && \
2272
37.3M
        (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \
2273
3.05M
  xmlParserGrow(ctxt);
2274
2275
4.61M
#define SKIP_BLANKS xmlSkipBlankChars(ctxt)
2276
2277
1.08M
#define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt)
2278
2279
257M
#define NEXT xmlNextChar(ctxt)
2280
2281
2.15M
#define NEXT1 {               \
2282
2.15M
  ctxt->input->col++;           \
2283
2.15M
  ctxt->input->cur++;           \
2284
2.15M
  if (*ctxt->input->cur == 0)         \
2285
2.15M
      xmlParserGrow(ctxt);           \
2286
2.15M
    }
2287
2288
241M
#define NEXTL(l) do {             \
2289
241M
    if (*(ctxt->input->cur) == '\n') {         \
2290
7.17M
  ctxt->input->line++; ctxt->input->col = 1;      \
2291
234M
    } else ctxt->input->col++;           \
2292
241M
    ctxt->input->cur += l;        \
2293
241M
  } while (0)
2294
2295
#define COPY_BUF(b, i, v)           \
2296
38.5M
    if (v < 0x80) b[i++] = v;           \
2297
38.5M
    else i += xmlCopyCharMultiByte(&b[i],v)
2298
2299
static int
2300
43.6M
xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) {
2301
43.6M
    int c = xmlCurrentChar(ctxt, len);
2302
2303
43.6M
    if (c == XML_INVALID_CHAR)
2304
8.52M
        c = 0xFFFD; /* replacement character */
2305
2306
43.6M
    return(c);
2307
43.6M
}
2308
2309
/**
2310
 * Skip whitespace in the input stream.
2311
 *
2312
 * @deprecated Internal function, do not use.
2313
 *
2314
 * @param ctxt  the XML parser context
2315
 * @returns the number of space chars skipped
2316
 */
2317
int
2318
4.76M
xmlSkipBlankChars(xmlParserCtxt *ctxt) {
2319
4.76M
    const xmlChar *cur;
2320
4.76M
    int res = 0;
2321
2322
4.76M
    cur = ctxt->input->cur;
2323
4.76M
    while (IS_BLANK_CH(*cur)) {
2324
1.80M
        if (*cur == '\n') {
2325
601k
            ctxt->input->line++; ctxt->input->col = 1;
2326
1.20M
        } else {
2327
1.20M
            ctxt->input->col++;
2328
1.20M
        }
2329
1.80M
        cur++;
2330
1.80M
        if (res < INT_MAX)
2331
1.80M
            res++;
2332
1.80M
        if (*cur == 0) {
2333
25.6k
            ctxt->input->cur = cur;
2334
25.6k
            xmlParserGrow(ctxt);
2335
25.6k
            cur = ctxt->input->cur;
2336
25.6k
        }
2337
1.80M
    }
2338
4.76M
    ctxt->input->cur = cur;
2339
2340
4.76M
    if (res > 4)
2341
27.2k
        GROW;
2342
2343
4.76M
    return(res);
2344
4.76M
}
2345
2346
static void
2347
104k
xmlPopPE(xmlParserCtxtPtr ctxt) {
2348
104k
    unsigned long consumed;
2349
104k
    xmlEntityPtr ent;
2350
2351
104k
    ent = ctxt->input->entity;
2352
2353
104k
    ent->flags &= ~XML_ENT_EXPANDING;
2354
2355
104k
    if ((ent->flags & XML_ENT_CHECKED) == 0) {
2356
4.64k
        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
4.85k
        do {
2363
4.85k
            ctxt->input->cur = ctxt->input->end;
2364
4.85k
            xmlParserShrink(ctxt);
2365
4.85k
            result = xmlParserGrow(ctxt);
2366
4.85k
        } while (result > 0);
2367
2368
4.64k
        consumed = ctxt->input->consumed;
2369
4.64k
        xmlSaturatedAddSizeT(&consumed,
2370
4.64k
                             ctxt->input->end - ctxt->input->base);
2371
2372
4.64k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
2373
2374
        /*
2375
         * Add to sizeentities when parsing an external entity
2376
         * for the first time.
2377
         */
2378
4.64k
        if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
2379
3.45k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
2380
3.45k
        }
2381
2382
4.64k
        ent->flags |= XML_ENT_CHECKED;
2383
4.64k
    }
2384
2385
104k
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
2386
2387
104k
    xmlParserEntityCheck(ctxt, ent->expandedSize);
2388
2389
104k
    GROW;
2390
104k
}
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
1.08M
xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) {
2401
1.08M
    int res = 0;
2402
1.08M
    int inParam;
2403
1.08M
    int expandParam;
2404
2405
1.08M
    inParam = PARSER_IN_PE(ctxt);
2406
1.08M
    expandParam = PARSER_EXTERNAL(ctxt);
2407
2408
1.08M
    if (!inParam && !expandParam)
2409
153k
        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
2.72M
    while (PARSER_STOPPED(ctxt) == 0) {
2416
2.72M
        if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */
2417
1.76M
            NEXT;
2418
1.76M
        } else if (CUR == '%') {
2419
73.1k
            if ((expandParam == 0) ||
2420
73.1k
                (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0))
2421
55.7k
                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
17.4k
            xmlParsePERefInternal(ctxt, 0);
2430
2431
17.4k
            inParam = PARSER_IN_PE(ctxt);
2432
17.4k
            expandParam = PARSER_EXTERNAL(ctxt);
2433
887k
        } else if (CUR == 0) {
2434
58.9k
            if (inParam == 0)
2435
133
                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
58.8k
            if (ctxt->input->flags & XML_INPUT_MARKUP_DECL)
2443
51.1k
                break;
2444
2445
7.63k
            xmlPopPE(ctxt);
2446
2447
7.63k
            inParam = PARSER_IN_PE(ctxt);
2448
7.63k
            expandParam = PARSER_EXTERNAL(ctxt);
2449
828k
        } else {
2450
828k
            break;
2451
828k
        }
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
1.78M
        if (res < INT_MAX)
2461
1.78M
            res++;
2462
1.78M
    }
2463
2464
935k
    return(res);
2465
1.08M
}
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
90.4k
xmlParseCharRef(xmlParserCtxt *ctxt) {
2530
90.4k
    int val = 0;
2531
90.4k
    int count = 0;
2532
2533
    /*
2534
     * Using RAW/CUR/NEXT is okay since we are working on ASCII range here
2535
     */
2536
90.4k
    if ((RAW == '&') && (NXT(1) == '#') &&
2537
90.4k
        (NXT(2) == 'x')) {
2538
18.1k
  SKIP(3);
2539
18.1k
  GROW;
2540
89.1k
  while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) {
2541
76.5k
      if (count++ > 20) {
2542
1.08k
    count = 0;
2543
1.08k
    GROW;
2544
1.08k
      }
2545
76.5k
      if ((RAW >= '0') && (RAW <= '9'))
2546
36.4k
          val = val * 16 + (CUR - '0');
2547
40.0k
      else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20))
2548
8.71k
          val = val * 16 + (CUR - 'a') + 10;
2549
31.3k
      else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20))
2550
25.8k
          val = val * 16 + (CUR - 'A') + 10;
2551
5.51k
      else {
2552
5.51k
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2553
5.51k
    val = 0;
2554
5.51k
    break;
2555
5.51k
      }
2556
71.0k
      if (val > 0x110000)
2557
20.0k
          val = 0x110000;
2558
2559
71.0k
      NEXT;
2560
71.0k
      count++;
2561
71.0k
  }
2562
18.1k
  if (RAW == ';') {
2563
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2564
12.6k
      ctxt->input->col++;
2565
12.6k
      ctxt->input->cur++;
2566
12.6k
  }
2567
72.3k
    } else if  ((RAW == '&') && (NXT(1) == '#')) {
2568
72.3k
  SKIP(2);
2569
72.3k
  GROW;
2570
225k
  while (RAW != ';') { /* loop blocked by count */
2571
156k
      if (count++ > 20) {
2572
1.01k
    count = 0;
2573
1.01k
    GROW;
2574
1.01k
      }
2575
156k
      if ((RAW >= '0') && (RAW <= '9'))
2576
152k
          val = val * 10 + (CUR - '0');
2577
3.54k
      else {
2578
3.54k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2579
3.54k
    val = 0;
2580
3.54k
    break;
2581
3.54k
      }
2582
152k
      if (val > 0x110000)
2583
10.5k
          val = 0x110000;
2584
2585
152k
      NEXT;
2586
152k
      count++;
2587
152k
  }
2588
72.3k
  if (RAW == ';') {
2589
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2590
68.8k
      ctxt->input->col++;
2591
68.8k
      ctxt->input->cur++;
2592
68.8k
  }
2593
72.3k
    } 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
90.4k
    if (val >= 0x110000) {
2605
443
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2606
443
                "xmlParseCharRef: character reference out of bounds\n",
2607
443
          val);
2608
443
        val = 0xFFFD;
2609
90.0k
    } else if (!IS_CHAR(val)) {
2610
11.3k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2611
11.3k
                          "xmlParseCharRef: invalid xmlChar value %d\n",
2612
11.3k
                    val);
2613
11.3k
    }
2614
90.4k
    return(val);
2615
90.4k
}
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
1.21M
xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) {
2635
1.21M
    const xmlChar *ptr;
2636
1.21M
    xmlChar cur;
2637
1.21M
    int val = 0;
2638
2639
1.21M
    if ((str == NULL) || (*str == NULL)) return(0);
2640
1.21M
    ptr = *str;
2641
1.21M
    cur = *ptr;
2642
1.21M
    if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) {
2643
10.6k
  ptr += 3;
2644
10.6k
  cur = *ptr;
2645
37.2k
  while (cur != ';') { /* Non input consuming loop */
2646
27.1k
      if ((cur >= '0') && (cur <= '9'))
2647
13.9k
          val = val * 16 + (cur - '0');
2648
13.2k
      else if ((cur >= 'a') && (cur <= 'f'))
2649
3.93k
          val = val * 16 + (cur - 'a') + 10;
2650
9.28k
      else if ((cur >= 'A') && (cur <= 'F'))
2651
8.65k
          val = val * 16 + (cur - 'A') + 10;
2652
635
      else {
2653
635
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2654
635
    val = 0;
2655
635
    break;
2656
635
      }
2657
26.5k
      if (val > 0x110000)
2658
585
          val = 0x110000;
2659
2660
26.5k
      ptr++;
2661
26.5k
      cur = *ptr;
2662
26.5k
  }
2663
10.6k
  if (cur == ';')
2664
10.0k
      ptr++;
2665
1.20M
    } else if  ((cur == '&') && (ptr[1] == '#')){
2666
1.20M
  ptr += 2;
2667
1.20M
  cur = *ptr;
2668
3.61M
  while (cur != ';') { /* Non input consuming loops */
2669
2.41M
      if ((cur >= '0') && (cur <= '9'))
2670
2.41M
          val = val * 10 + (cur - '0');
2671
700
      else {
2672
700
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2673
700
    val = 0;
2674
700
    break;
2675
700
      }
2676
2.41M
      if (val > 0x110000)
2677
2.40k
          val = 0x110000;
2678
2679
2.41M
      ptr++;
2680
2.41M
      cur = *ptr;
2681
2.41M
  }
2682
1.20M
  if (cur == ';')
2683
1.20M
      ptr++;
2684
1.20M
    } else {
2685
0
  xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2686
0
  return(0);
2687
0
    }
2688
1.21M
    *str = ptr;
2689
2690
    /*
2691
     * [ WFC: Legal Character ]
2692
     * Characters referred to using character references must match the
2693
     * production for Char.
2694
     */
2695
1.21M
    if (val >= 0x110000) {
2696
255
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2697
255
                "xmlParseStringCharRef: character reference out of bounds\n",
2698
255
                val);
2699
1.21M
    } else if (IS_CHAR(val)) {
2700
1.21M
        return(val);
2701
1.21M
    } else {
2702
2.27k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2703
2.27k
        "xmlParseStringCharRef: invalid xmlChar value %d\n",
2704
2.27k
        val);
2705
2.27k
    }
2706
2.53k
    return(0);
2707
1.21M
}
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
471k
                     int blank_chars) {
2816
471k
    int i;
2817
471k
    xmlNodePtr lastChild;
2818
2819
    /*
2820
     * Check for xml:space value.
2821
     */
2822
471k
    if ((ctxt->space == NULL) || (*(ctxt->space) == 1) ||
2823
470k
        (*(ctxt->space) == -2))
2824
264k
  return(0);
2825
2826
    /*
2827
     * Check that the string is made of blanks
2828
     */
2829
207k
    if (blank_chars == 0) {
2830
253k
  for (i = 0;i < len;i++)
2831
247k
      if (!(IS_BLANK_CH(str[i]))) return(0);
2832
198k
    }
2833
2834
    /*
2835
     * Look if the element is mixed content in the DTD if available
2836
     */
2837
14.8k
    if (ctxt->node == NULL) return(0);
2838
14.8k
    if (ctxt->myDoc != NULL) {
2839
14.8k
        xmlElementPtr elemDecl = NULL;
2840
14.8k
        xmlDocPtr doc = ctxt->myDoc;
2841
14.8k
        const xmlChar *prefix = NULL;
2842
2843
14.8k
        if (ctxt->node->ns)
2844
3.07k
            prefix = ctxt->node->ns->prefix;
2845
14.8k
        if (doc->intSubset != NULL)
2846
8.95k
            elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name,
2847
8.95k
                                      prefix);
2848
14.8k
        if ((elemDecl == NULL) && (doc->extSubset != NULL))
2849
371
            elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name,
2850
371
                                      prefix);
2851
14.8k
        if (elemDecl != NULL) {
2852
2.18k
            if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)
2853
1.12k
                return(1);
2854
1.05k
            if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) ||
2855
817
                (elemDecl->etype == XML_ELEMENT_TYPE_MIXED))
2856
312
                return(0);
2857
1.05k
        }
2858
14.8k
    }
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
13.4k
    if ((RAW != '<') && (RAW != 0xD)) return(0);
2868
10.9k
    if ((ctxt->node->children == NULL) &&
2869
8.57k
  (RAW == '<') && (NXT(1) == '/')) return(0);
2870
2871
10.4k
    lastChild = xmlGetLastChild(ctxt->node);
2872
10.4k
    if (lastChild == NULL) {
2873
8.01k
        if ((ctxt->node->type != XML_ELEMENT_NODE) &&
2874
0
            (ctxt->node->content != NULL)) return(0);
2875
8.01k
    } else if (xmlNodeIsText(lastChild))
2876
360
        return(0);
2877
2.03k
    else if ((ctxt->node->children != NULL) &&
2878
2.03k
             (xmlNodeIsText(ctxt->node->children)))
2879
300
        return(0);
2880
9.75k
    return(1);
2881
10.4k
}
2882
2883
/************************************************************************
2884
 *                  *
2885
 *    Extra stuff for namespace support     *
2886
 *  Relates to http://www.w3.org/TR/WD-xml-names      *
2887
 *                  *
2888
 ************************************************************************/
2889
2890
/**
2891
 * Parse an UTF8 encoded XML qualified name string
2892
 *
2893
 * @deprecated Don't use.
2894
 *
2895
 * @param ctxt  an XML parser context
2896
 * @param name  an XML parser context
2897
 * @param prefixOut  a xmlChar **
2898
 * @returns the local part, and prefix is updated
2899
 *   to get the Prefix if any.
2900
 */
2901
2902
xmlChar *
2903
0
xmlSplitQName(xmlParserCtxt *ctxt, const xmlChar *name, xmlChar **prefixOut) {
2904
0
    xmlChar *ret;
2905
0
    const xmlChar *localname;
2906
2907
0
    localname = xmlSplitQName4(name, prefixOut);
2908
0
    if (localname == NULL) {
2909
0
        xmlCtxtErrMemory(ctxt);
2910
0
        return(NULL);
2911
0
    }
2912
2913
0
    ret = xmlStrdup(localname);
2914
0
    if (ret == NULL) {
2915
0
        xmlCtxtErrMemory(ctxt);
2916
0
        xmlFree(*prefixOut);
2917
0
    }
2918
2919
0
    return(ret);
2920
0
}
2921
2922
/************************************************************************
2923
 *                  *
2924
 *      The parser itself       *
2925
 *  Relates to http://www.w3.org/TR/REC-xml       *
2926
 *                  *
2927
 ************************************************************************/
2928
2929
/************************************************************************
2930
 *                  *
2931
 *  Routines to parse Name, NCName and NmToken      *
2932
 *                  *
2933
 ************************************************************************/
2934
2935
/*
2936
 * The two following functions are related to the change of accepted
2937
 * characters for Name and NmToken in the Revision 5 of XML-1.0
2938
 * They correspond to the modified production [4] and the new production [4a]
2939
 * changes in that revision. Also note that the macros used for the
2940
 * productions Letter, Digit, CombiningChar and Extender are not needed
2941
 * anymore.
2942
 * We still keep compatibility to pre-revision5 parsing semantic if the
2943
 * new XML_PARSE_OLD10 option is given to the parser.
2944
 */
2945
2946
static int
2947
2.76M
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
2.76M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2953
2.75M
        (((c >= 'a') && (c <= 'z')) ||
2954
254k
         ((c >= 'A') && (c <= 'Z')) ||
2955
200k
         (c == '_') || (c == ':') ||
2956
192k
         ((c >= 0xC0) && (c <= 0xD6)) ||
2957
191k
         ((c >= 0xD8) && (c <= 0xF6)) ||
2958
190k
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2959
189k
         ((c >= 0x370) && (c <= 0x37D)) ||
2960
189k
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2961
186k
         ((c >= 0x200C) && (c <= 0x200D)) ||
2962
186k
         ((c >= 0x2070) && (c <= 0x218F)) ||
2963
186k
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2964
186k
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2965
183k
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2966
183k
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2967
172k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2968
2.58M
        return(1);
2969
181k
    return(0);
2970
2.76M
}
2971
2972
static int
2973
18.0M
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
18.0M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2979
17.9M
        (((c >= 'a') && (c <= 'z')) ||
2980
15.5M
         ((c >= 'A') && (c <= 'Z')) ||
2981
13.1M
         ((c >= '0') && (c <= '9')) || /* !start */
2982
12.6M
         (c == '_') || (c == ':') ||
2983
12.6M
         (c == '-') || (c == '.') || (c == 0xB7) || /* !start */
2984
12.5M
         ((c >= 0xC0) && (c <= 0xD6)) ||
2985
9.78M
         ((c >= 0xD8) && (c <= 0xF6)) ||
2986
9.76M
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2987
8.18M
         ((c >= 0x300) && (c <= 0x36F)) || /* !start */
2988
8.18M
         ((c >= 0x370) && (c <= 0x37D)) ||
2989
8.18M
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2990
6.72M
         ((c >= 0x200C) && (c <= 0x200D)) ||
2991
6.72M
         ((c >= 0x203F) && (c <= 0x2040)) || /* !start */
2992
6.72M
         ((c >= 0x2070) && (c <= 0x218F)) ||
2993
6.72M
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2994
6.71M
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2995
6.66M
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2996
6.65M
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2997
2.45M
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2998
15.4M
         return(1);
2999
2.58M
    return(0);
3000
18.0M
}
3001
3002
static int
3003
805k
xmlIsNameStartCharOld(int c) {
3004
805k
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3005
799k
        ((IS_LETTER(c) || (c == '_') || (c == ':'))))
3006
722k
        return(1);
3007
83.0k
    return(0);
3008
805k
}
3009
3010
static int
3011
4.07M
xmlIsNameCharOld(int c) {
3012
4.07M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3013
4.07M
        ((IS_LETTER(c)) || (IS_DIGIT(c)) ||
3014
1.38M
         (c == '.') || (c == '-') ||
3015
1.38M
         (c == '_') || (c == ':') ||
3016
1.37M
         (IS_COMBINING(c)) ||
3017
725k
         (IS_EXTENDER(c))))
3018
3.34M
        return(1);
3019
731k
    return(0);
3020
4.07M
}
3021
3022
static int
3023
3.57M
xmlIsNameStartChar(int c, int old10) {
3024
3.57M
    if (!old10)
3025
2.76M
        return(xmlIsNameStartCharNew(c));
3026
805k
    else
3027
805k
        return(xmlIsNameStartCharOld(c));
3028
3.57M
}
3029
3030
static int
3031
22.1M
xmlIsNameChar(int c, int old10) {
3032
22.1M
    if (!old10)
3033
18.0M
        return(xmlIsNameCharNew(c));
3034
4.07M
    else
3035
4.07M
        return(xmlIsNameCharOld(c));
3036
22.1M
}
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
3.29M
xmlScanName(const xmlChar *ptr, size_t maxSize, int flags) {
3052
3.29M
    int stop = flags & XML_SCAN_NC ? ':' : 0;
3053
3.29M
    int old10 = flags & XML_SCAN_OLD10 ? 1 : 0;
3054
3055
13.4M
    while (1) {
3056
13.4M
        int c, len;
3057
3058
13.4M
        c = *ptr;
3059
13.4M
        if (c < 0x80) {
3060
10.0M
            if (c == stop)
3061
67.4k
                break;
3062
9.98M
            len = 1;
3063
9.98M
        } else {
3064
3.37M
            len = 4;
3065
3.37M
            c = xmlGetUTF8Char(ptr, &len);
3066
3.37M
            if (c < 0)
3067
7.76k
                break;
3068
3.37M
        }
3069
3070
13.3M
        if (flags & XML_SCAN_NMTOKEN ?
3071
10.1M
                !xmlIsNameChar(c, old10) :
3072
13.3M
                !xmlIsNameStartChar(c, old10))
3073
3.21M
            break;
3074
3075
10.1M
        if ((size_t) len > maxSize)
3076
79
            return(NULL);
3077
10.1M
        ptr += len;
3078
10.1M
        maxSize -= len;
3079
10.1M
        flags |= XML_SCAN_NMTOKEN;
3080
10.1M
    }
3081
3082
3.29M
    return(ptr);
3083
3.29M
}
3084
3085
static const xmlChar *
3086
221k
xmlParseNameComplex(xmlParserCtxtPtr ctxt) {
3087
221k
    const xmlChar *ret;
3088
221k
    int len = 0, l;
3089
221k
    int c;
3090
221k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3091
102k
                    XML_MAX_TEXT_LENGTH :
3092
221k
                    XML_MAX_NAME_LENGTH;
3093
221k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3094
3095
    /*
3096
     * Handler for more complex cases
3097
     */
3098
221k
    c = xmlCurrentChar(ctxt, &l);
3099
221k
    if (!xmlIsNameStartChar(c, old10))
3100
157k
        return(NULL);
3101
64.4k
    len += l;
3102
64.4k
    NEXTL(l);
3103
64.4k
    c = xmlCurrentChar(ctxt, &l);
3104
5.07M
    while (xmlIsNameChar(c, old10)) {
3105
5.01M
        if (len <= INT_MAX - l)
3106
5.01M
            len += l;
3107
5.01M
        NEXTL(l);
3108
5.01M
        c = xmlCurrentChar(ctxt, &l);
3109
5.01M
    }
3110
64.4k
    if (len > maxLength) {
3111
49
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3112
49
        return(NULL);
3113
49
    }
3114
64.4k
    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
64.4k
    if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r'))
3125
425
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len);
3126
63.9k
    else
3127
63.9k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len);
3128
64.4k
    if (ret == NULL)
3129
1
        xmlErrMemory(ctxt);
3130
64.4k
    return(ret);
3131
64.4k
}
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
2.42M
xmlParseName(xmlParserCtxt *ctxt) {
3151
2.42M
    const xmlChar *in;
3152
2.42M
    const xmlChar *ret;
3153
2.42M
    size_t count = 0;
3154
2.42M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3155
983k
                       XML_MAX_TEXT_LENGTH :
3156
2.42M
                       XML_MAX_NAME_LENGTH;
3157
3158
2.42M
    GROW;
3159
3160
    /*
3161
     * Accelerator for simple ASCII names
3162
     */
3163
2.42M
    in = ctxt->input->cur;
3164
2.42M
    if (((*in >= 0x61) && (*in <= 0x7A)) ||
3165
263k
  ((*in >= 0x41) && (*in <= 0x5A)) ||
3166
2.25M
  (*in == '_') || (*in == ':')) {
3167
2.25M
  in++;
3168
14.2M
  while (((*in >= 0x61) && (*in <= 0x7A)) ||
3169
8.74M
         ((*in >= 0x41) && (*in <= 0x5A)) ||
3170
3.92M
         ((*in >= 0x30) && (*in <= 0x39)) ||
3171
2.51M
         (*in == '_') || (*in == '-') ||
3172
2.40M
         (*in == ':') || (*in == '.'))
3173
12.0M
      in++;
3174
2.25M
  if ((*in > 0) && (*in < 0x80)) {
3175
2.20M
      count = in - ctxt->input->cur;
3176
2.20M
            if (count > maxLength) {
3177
71
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3178
71
                return(NULL);
3179
71
            }
3180
2.19M
      ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count);
3181
2.19M
      ctxt->input->cur = in;
3182
2.19M
      ctxt->input->col += count;
3183
2.19M
      if (ret == NULL)
3184
3
          xmlErrMemory(ctxt);
3185
2.19M
      return(ret);
3186
2.20M
  }
3187
2.25M
    }
3188
    /* accelerator for special cases */
3189
221k
    return(xmlParseNameComplex(ctxt));
3190
2.42M
}
3191
3192
static xmlHashedString
3193
123k
xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) {
3194
123k
    xmlHashedString ret;
3195
123k
    int len = 0, l;
3196
123k
    int c;
3197
123k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3198
84.1k
                    XML_MAX_TEXT_LENGTH :
3199
123k
                    XML_MAX_NAME_LENGTH;
3200
123k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3201
123k
    size_t startPosition = 0;
3202
3203
123k
    ret.name = NULL;
3204
123k
    ret.hashValue = 0;
3205
3206
    /*
3207
     * Handler for more complex cases
3208
     */
3209
123k
    startPosition = CUR_PTR - BASE_PTR;
3210
123k
    c = xmlCurrentChar(ctxt, &l);
3211
123k
    if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3212
115k
  (!xmlIsNameStartChar(c, old10) || (c == ':'))) {
3213
107k
  return(ret);
3214
107k
    }
3215
3216
3.63M
    while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */
3217
3.63M
     (xmlIsNameChar(c, old10) && (c != ':'))) {
3218
3.62M
        if (len <= INT_MAX - l)
3219
3.62M
      len += l;
3220
3.62M
  NEXTL(l);
3221
3.62M
  c = xmlCurrentChar(ctxt, &l);
3222
3.62M
    }
3223
16.6k
    if (len > maxLength) {
3224
24
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3225
24
        return(ret);
3226
24
    }
3227
16.5k
    ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len);
3228
16.5k
    if (ret.name == NULL)
3229
2
        xmlErrMemory(ctxt);
3230
16.5k
    return(ret);
3231
16.6k
}
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
1.32M
xmlParseNCName(xmlParserCtxtPtr ctxt) {
3247
1.32M
    const xmlChar *in, *e;
3248
1.32M
    xmlHashedString ret;
3249
1.32M
    size_t count = 0;
3250
1.32M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3251
883k
                       XML_MAX_TEXT_LENGTH :
3252
1.32M
                       XML_MAX_NAME_LENGTH;
3253
3254
1.32M
    ret.name = NULL;
3255
3256
    /*
3257
     * Accelerator for simple ASCII names
3258
     */
3259
1.32M
    in = ctxt->input->cur;
3260
1.32M
    e = ctxt->input->end;
3261
1.32M
    if ((((*in >= 0x61) && (*in <= 0x7A)) ||
3262
128k
   ((*in >= 0x41) && (*in <= 0x5A)) ||
3263
1.21M
   (*in == '_')) && (in < e)) {
3264
1.21M
  in++;
3265
3.47M
  while ((((*in >= 0x61) && (*in <= 0x7A)) ||
3266
1.79M
          ((*in >= 0x41) && (*in <= 0x5A)) ||
3267
1.35M
          ((*in >= 0x30) && (*in <= 0x39)) ||
3268
1.22M
          (*in == '_') || (*in == '-') ||
3269
2.26M
          (*in == '.')) && (in < e))
3270
2.26M
      in++;
3271
1.21M
  if (in >= e)
3272
475
      goto complex;
3273
1.21M
  if ((*in > 0) && (*in < 0x80)) {
3274
1.20M
      count = in - ctxt->input->cur;
3275
1.20M
            if (count > maxLength) {
3276
10
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3277
10
                return(ret);
3278
10
            }
3279
1.20M
      ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count);
3280
1.20M
      ctxt->input->cur = in;
3281
1.20M
      ctxt->input->col += count;
3282
1.20M
      if (ret.name == NULL) {
3283
2
          xmlErrMemory(ctxt);
3284
2
      }
3285
1.20M
      return(ret);
3286
1.20M
  }
3287
1.21M
    }
3288
123k
complex:
3289
123k
    return(xmlParseNCNameComplex(ctxt));
3290
1.32M
}
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
486k
xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) {
3304
486k
    register const xmlChar *cmp = other;
3305
486k
    register const xmlChar *in;
3306
486k
    const xmlChar *ret;
3307
3308
486k
    GROW;
3309
3310
486k
    in = ctxt->input->cur;
3311
1.04M
    while (*in != 0 && *in == *cmp) {
3312
554k
  ++in;
3313
554k
  ++cmp;
3314
554k
    }
3315
486k
    if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
3316
  /* success */
3317
462k
  ctxt->input->col += in - ctxt->input->cur;
3318
462k
  ctxt->input->cur = in;
3319
462k
  return (const xmlChar*) 1;
3320
462k
    }
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
10.4k
  return (const xmlChar*) 1;
3326
10.4k
    }
3327
13.5k
    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
3.00M
xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) {
3341
3.00M
    xmlChar *ret;
3342
3.00M
    const xmlChar *cur = *str;
3343
3.00M
    int flags = 0;
3344
3.00M
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3345
1.06M
                    XML_MAX_TEXT_LENGTH :
3346
3.00M
                    XML_MAX_NAME_LENGTH;
3347
3348
3.00M
    if (ctxt->options & XML_PARSE_OLD10)
3349
691k
        flags |= XML_SCAN_OLD10;
3350
3351
3.00M
    cur = xmlScanName(*str, maxLength, flags);
3352
3.00M
    if (cur == NULL) {
3353
74
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3354
74
        return(NULL);
3355
74
    }
3356
3.00M
    if (cur == *str)
3357
2.93k
        return(NULL);
3358
3359
2.99M
    ret = xmlStrndup(*str, cur - *str);
3360
2.99M
    if (ret == NULL)
3361
25
        xmlErrMemory(ctxt);
3362
2.99M
    *str = cur;
3363
2.99M
    return(ret);
3364
3.00M
}
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
35.1k
xmlParseNmtoken(xmlParserCtxt *ctxt) {
3381
35.1k
    xmlChar buf[XML_MAX_NAMELEN + 5];
3382
35.1k
    xmlChar *ret;
3383
35.1k
    int len = 0, l;
3384
35.1k
    int c;
3385
35.1k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3386
17.9k
                    XML_MAX_TEXT_LENGTH :
3387
35.1k
                    XML_MAX_NAME_LENGTH;
3388
35.1k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3389
3390
35.1k
    c = xmlCurrentChar(ctxt, &l);
3391
3392
161k
    while (xmlIsNameChar(c, old10)) {
3393
127k
  COPY_BUF(buf, len, c);
3394
127k
  NEXTL(l);
3395
127k
  c = xmlCurrentChar(ctxt, &l);
3396
127k
  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
751
      xmlChar *buffer;
3402
751
      int max = len * 2;
3403
3404
751
      buffer = xmlMalloc(max);
3405
751
      if (buffer == NULL) {
3406
1
          xmlErrMemory(ctxt);
3407
1
    return(NULL);
3408
1
      }
3409
750
      memcpy(buffer, buf, len);
3410
3.15M
      while (xmlIsNameChar(c, old10)) {
3411
3.15M
    if (len + 10 > max) {
3412
2.06k
        xmlChar *tmp;
3413
2.06k
                    int newSize;
3414
3415
2.06k
                    newSize = xmlGrowCapacity(max, 1, 1, maxLength);
3416
2.06k
                    if (newSize < 0) {
3417
55
                        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3418
55
                        xmlFree(buffer);
3419
55
                        return(NULL);
3420
55
                    }
3421
2.00k
        tmp = xmlRealloc(buffer, newSize);
3422
2.00k
        if (tmp == NULL) {
3423
1
      xmlErrMemory(ctxt);
3424
1
      xmlFree(buffer);
3425
1
      return(NULL);
3426
1
        }
3427
2.00k
        buffer = tmp;
3428
2.00k
                    max = newSize;
3429
2.00k
    }
3430
3.15M
    COPY_BUF(buffer, len, c);
3431
3.15M
    NEXTL(l);
3432
3.15M
    c = xmlCurrentChar(ctxt, &l);
3433
3.15M
      }
3434
694
      buffer[len] = 0;
3435
694
      return(buffer);
3436
750
  }
3437
127k
    }
3438
34.3k
    if (len == 0)
3439
5.16k
        return(NULL);
3440
29.1k
    if (len > maxLength) {
3441
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3442
0
        return(NULL);
3443
0
    }
3444
29.1k
    ret = xmlStrndup(buf, len);
3445
29.1k
    if (ret == NULL)
3446
5
        xmlErrMemory(ctxt);
3447
29.1k
    return(ret);
3448
29.1k
}
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
223k
                          const xmlChar *str, int length, int depth) {
3462
223k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3463
223k
    const xmlChar *end, *chunk;
3464
223k
    int c, l;
3465
3466
223k
    if (str == NULL)
3467
115k
        return;
3468
3469
108k
    depth += 1;
3470
108k
    if (depth > maxDepth) {
3471
0
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3472
0
                       "Maximum entity nesting depth exceeded");
3473
0
  return;
3474
0
    }
3475
3476
108k
    end = str + length;
3477
108k
    chunk = str;
3478
3479
79.9M
    while ((str < end) && (!PARSER_STOPPED(ctxt))) {
3480
79.8M
        c = *str;
3481
3482
79.8M
        if (c >= 0x80) {
3483
38.8M
            l = xmlUTF8MultibyteLen(ctxt, str,
3484
38.8M
                    "invalid character in entity value\n");
3485
38.8M
            if (l == 0) {
3486
13.8M
                if (chunk < str)
3487
65.4k
                    xmlSBufAddString(buf, chunk, str - chunk);
3488
13.8M
                xmlSBufAddReplChar(buf);
3489
13.8M
                str += 1;
3490
13.8M
                chunk = str;
3491
24.9M
            } else {
3492
24.9M
                str += l;
3493
24.9M
            }
3494
41.0M
        } else if (c == '&') {
3495
77.1k
            if (str[1] == '#') {
3496
18.9k
                if (chunk < str)
3497
13.2k
                    xmlSBufAddString(buf, chunk, str - chunk);
3498
3499
18.9k
                c = xmlParseStringCharRef(ctxt, &str);
3500
18.9k
                if (c == 0)
3501
2.52k
                    return;
3502
3503
16.3k
                xmlSBufAddChar(buf, c);
3504
3505
16.3k
                chunk = str;
3506
58.2k
            } else {
3507
58.2k
                xmlChar *name;
3508
3509
                /*
3510
                 * General entity references are checked for
3511
                 * syntactic validity.
3512
                 */
3513
58.2k
                str++;
3514
58.2k
                name = xmlParseStringName(ctxt, &str);
3515
3516
58.2k
                if ((name == NULL) || (*str++ != ';')) {
3517
10.9k
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR,
3518
10.9k
                            "EntityValue: '&' forbidden except for entities "
3519
10.9k
                            "references\n");
3520
10.9k
                    xmlFree(name);
3521
10.9k
                    return;
3522
10.9k
                }
3523
3524
47.3k
                xmlFree(name);
3525
47.3k
            }
3526
40.9M
        } else if (c == '%') {
3527
164k
            xmlEntityPtr ent;
3528
3529
164k
            if (chunk < str)
3530
73.4k
                xmlSBufAddString(buf, chunk, str - chunk);
3531
3532
164k
            ent = xmlParseStringPEReference(ctxt, &str);
3533
164k
            if (ent == NULL)
3534
41.6k
                return;
3535
3536
123k
            if (!PARSER_EXTERNAL(ctxt)) {
3537
67
                xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL);
3538
67
                return;
3539
67
            }
3540
3541
123k
            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
115k
                if (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
3549
115k
                    ((ctxt->replaceEntities) ||
3550
114k
                     (ctxt->validate))) {
3551
114k
                    xmlLoadEntityContent(ctxt, ent);
3552
114k
                } else {
3553
771
                    xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
3554
771
                                  "not validating will not read content for "
3555
771
                                  "PE entity %s\n", ent->name, NULL);
3556
771
                }
3557
115k
            }
3558
3559
            /*
3560
             * TODO: Skip if ent->content is still NULL.
3561
             */
3562
3563
123k
            if (xmlParserEntityCheck(ctxt, ent->length))
3564
17
                return;
3565
3566
123k
            if (ent->flags & XML_ENT_EXPANDING) {
3567
65
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3568
65
                return;
3569
65
            }
3570
3571
122k
            ent->flags |= XML_ENT_EXPANDING;
3572
122k
            xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length,
3573
122k
                                      depth);
3574
122k
            ent->flags &= ~XML_ENT_EXPANDING;
3575
3576
122k
            chunk = str;
3577
40.8M
        } else {
3578
            /* Normal ASCII char */
3579
40.8M
            if (!IS_BYTE_CHAR(c)) {
3580
535k
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3581
535k
                        "invalid character in entity value\n");
3582
535k
                if (chunk < str)
3583
7.91k
                    xmlSBufAddString(buf, chunk, str - chunk);
3584
535k
                xmlSBufAddReplChar(buf);
3585
535k
                str += 1;
3586
535k
                chunk = str;
3587
40.2M
            } else {
3588
40.2M
                str += 1;
3589
40.2M
            }
3590
40.8M
        }
3591
79.8M
    }
3592
3593
52.7k
    if (chunk < str)
3594
48.0k
        xmlSBufAddString(buf, chunk, str - chunk);
3595
52.7k
}
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
103k
xmlParseEntityValue(xmlParserCtxt *ctxt, xmlChar **orig) {
3611
103k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3612
35.2k
                         XML_MAX_HUGE_LENGTH :
3613
103k
                         XML_MAX_TEXT_LENGTH;
3614
103k
    xmlSBuf buf;
3615
103k
    const xmlChar *start;
3616
103k
    int quote, length;
3617
3618
103k
    xmlSBufInit(&buf, maxLength);
3619
3620
103k
    GROW;
3621
3622
103k
    quote = CUR;
3623
103k
    if ((quote != '"') && (quote != '\'')) {
3624
0
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
3625
0
  return(NULL);
3626
0
    }
3627
103k
    CUR_PTR++;
3628
3629
103k
    length = 0;
3630
3631
    /*
3632
     * Copy raw content of the entity into a buffer
3633
     */
3634
34.7M
    while (1) {
3635
34.7M
        int c;
3636
3637
34.7M
        if (PARSER_STOPPED(ctxt))
3638
1
            goto error;
3639
3640
34.7M
        if (CUR_PTR >= ctxt->input->end) {
3641
2.80k
            xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL);
3642
2.80k
            goto error;
3643
2.80k
        }
3644
3645
34.7M
        c = CUR;
3646
3647
34.7M
        if (c == 0) {
3648
241
            xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3649
241
                    "invalid character in entity value\n");
3650
241
            goto error;
3651
241
        }
3652
34.7M
        if (c == quote)
3653
100k
            break;
3654
34.6M
        NEXTL(1);
3655
34.6M
        length += 1;
3656
3657
        /*
3658
         * TODO: Check growth threshold
3659
         */
3660
34.6M
        if (ctxt->input->end - CUR_PTR < 10)
3661
71.5k
            GROW;
3662
34.6M
    }
3663
3664
100k
    start = CUR_PTR - length;
3665
3666
100k
    if (orig != NULL) {
3667
100k
        *orig = xmlStrndup(start, length);
3668
100k
        if (*orig == NULL)
3669
3
            xmlErrMemory(ctxt);
3670
100k
    }
3671
3672
100k
    xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr);
3673
3674
100k
    NEXTL(1);
3675
3676
100k
    return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long"));
3677
3678
3.04k
error:
3679
3.04k
    xmlSBufCleanup(&buf, ctxt, "entity length too long");
3680
3.04k
    return(NULL);
3681
103k
}
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
4.19k
xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) {
3693
4.19k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3694
4.19k
    const xmlChar *str;
3695
4.19k
    unsigned long expandedSize = pent->length;
3696
4.19k
    int c, flags;
3697
3698
4.19k
    depth += 1;
3699
4.19k
    if (depth > maxDepth) {
3700
1
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3701
1
                       "Maximum entity nesting depth exceeded");
3702
1
  return;
3703
1
    }
3704
3705
4.19k
    if (pent->flags & XML_ENT_EXPANDING) {
3706
13
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3707
13
        return;
3708
13
    }
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
4.18k
    if (ctxt->inSubset == 0)
3716
3.96k
        flags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
3717
219
    else
3718
219
        flags = XML_ENT_VALIDATED;
3719
3720
4.18k
    str = pent->content;
3721
4.18k
    if (str == NULL)
3722
1
        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
14.6M
    while (!PARSER_STOPPED(ctxt)) {
3731
14.6M
        c = *str;
3732
3733
14.6M
  if (c != '&') {
3734
14.6M
            if (c == 0)
3735
4.05k
                break;
3736
3737
14.6M
            if (c == '<')
3738
1.22k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3739
1.22k
                        "'<' in entity '%s' is not allowed in attributes "
3740
1.22k
                        "values\n", pent->name);
3741
3742
14.6M
            str += 1;
3743
14.6M
        } else if (str[1] == '#') {
3744
3.44k
            int val;
3745
3746
3.44k
      val = xmlParseStringCharRef(ctxt, &str);
3747
3.44k
      if (val == 0) {
3748
5
                pent->content[0] = 0;
3749
5
                break;
3750
5
            }
3751
10.3k
  } else {
3752
10.3k
            xmlChar *name;
3753
10.3k
            xmlEntityPtr ent;
3754
3755
10.3k
      name = xmlParseStringEntityRef(ctxt, &str);
3756
10.3k
      if (name == NULL) {
3757
10
                pent->content[0] = 0;
3758
10
                break;
3759
10
            }
3760
3761
10.3k
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3762
10.3k
            xmlFree(name);
3763
3764
10.3k
            if ((ent != NULL) &&
3765
7.20k
                (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) {
3766
6.56k
                if ((ent->flags & flags) != flags) {
3767
3.30k
                    pent->flags |= XML_ENT_EXPANDING;
3768
3.30k
                    xmlCheckEntityInAttValue(ctxt, ent, depth);
3769
3.30k
                    pent->flags &= ~XML_ENT_EXPANDING;
3770
3.30k
                }
3771
3772
6.56k
                xmlSaturatedAdd(&expandedSize, ent->expandedSize);
3773
6.56k
                xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST);
3774
6.56k
            }
3775
10.3k
        }
3776
14.6M
    }
3777
3778
4.18k
done:
3779
4.18k
    if (ctxt->inSubset == 0)
3780
3.96k
        pent->expandedSize = expandedSize;
3781
3782
4.18k
    pent->flags |= flags;
3783
4.18k
}
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
2.00M
                          int *inSpace, int depth, int check) {
3803
2.00M
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3804
2.00M
    int c, chunkSize;
3805
2.00M
    int normChange = 0;
3806
3807
2.00M
    if (str == NULL)
3808
320
        return(0);
3809
3810
2.00M
    depth += 1;
3811
2.00M
    if (depth > maxDepth) {
3812
1
  xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
3813
1
                       "Maximum entity nesting depth exceeded");
3814
1
  return(0);
3815
1
    }
3816
3817
2.00M
    if (pent != NULL) {
3818
1.96M
        if (pent->flags & XML_ENT_EXPANDING) {
3819
2
            xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3820
2
            return(0);
3821
2
        }
3822
3823
1.96M
        if (check) {
3824
1.95M
            if (xmlParserEntityCheck(ctxt, pent->length))
3825
35
                return(0);
3826
1.95M
        }
3827
1.96M
    }
3828
3829
2.00M
    chunkSize = 0;
3830
3831
    /*
3832
     * Note that entity values are already validated. No special
3833
     * handling for multi-byte characters is needed.
3834
     */
3835
215M
    while (!PARSER_STOPPED(ctxt)) {
3836
215M
        c = *str;
3837
3838
215M
  if (c != '&') {
3839
211M
            if (c == 0)
3840
1.92M
                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
209M
            if ((pent != NULL) && (c == '<')) {
3848
76.0k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3849
76.0k
                        "'<' in entity '%s' is not allowed in attributes "
3850
76.0k
                        "values\n", pent->name);
3851
76.0k
                break;
3852
76.0k
            }
3853
3854
209M
            if (c <= 0x20) {
3855
15.3M
                if ((normalize) && (*inSpace)) {
3856
                    /* Skip char */
3857
109k
                    if (chunkSize > 0) {
3858
91.3k
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3859
91.3k
                        chunkSize = 0;
3860
91.3k
                    }
3861
109k
                    normChange = 1;
3862
15.2M
                } else if (c < 0x20) {
3863
5.05M
                    if (chunkSize > 0) {
3864
990k
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3865
990k
                        chunkSize = 0;
3866
990k
                    }
3867
3868
5.05M
                    xmlSBufAddCString(buf, " ", 1);
3869
10.1M
                } else {
3870
10.1M
                    chunkSize += 1;
3871
10.1M
                }
3872
3873
15.3M
                *inSpace = 1;
3874
194M
            } else {
3875
194M
                chunkSize += 1;
3876
194M
                *inSpace = 0;
3877
194M
            }
3878
3879
209M
            str += 1;
3880
209M
        } else if (str[1] == '#') {
3881
1.19M
            int val;
3882
3883
1.19M
            if (chunkSize > 0) {
3884
1.18M
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3885
1.18M
                chunkSize = 0;
3886
1.18M
            }
3887
3888
1.19M
      val = xmlParseStringCharRef(ctxt, &str);
3889
1.19M
      if (val == 0) {
3890
4
                if (pent != NULL)
3891
4
                    pent->content[0] = 0;
3892
4
                break;
3893
4
            }
3894
3895
1.19M
            if (val == ' ') {
3896
15.0k
                if ((normalize) && (*inSpace))
3897
9.84k
                    normChange = 1;
3898
5.20k
                else
3899
5.20k
                    xmlSBufAddCString(buf, " ", 1);
3900
15.0k
                *inSpace = 1;
3901
1.17M
            } else {
3902
1.17M
                xmlSBufAddChar(buf, val);
3903
1.17M
                *inSpace = 0;
3904
1.17M
            }
3905
2.76M
  } else {
3906
2.76M
            xmlChar *name;
3907
2.76M
            xmlEntityPtr ent;
3908
3909
2.76M
            if (chunkSize > 0) {
3910
1.37M
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3911
1.37M
                chunkSize = 0;
3912
1.37M
            }
3913
3914
2.76M
      name = xmlParseStringEntityRef(ctxt, &str);
3915
2.76M
            if (name == NULL) {
3916
20
                if (pent != NULL)
3917
19
                    pent->content[0] = 0;
3918
20
                break;
3919
20
            }
3920
3921
2.76M
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3922
2.76M
            xmlFree(name);
3923
3924
2.76M
      if ((ent != NULL) &&
3925
2.72M
    (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
3926
1.32M
    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
1.32M
                xmlSBufAddString(buf, ent->content, ent->length);
3933
3934
1.32M
                *inSpace = 0;
3935
1.44M
      } else if ((ent != NULL) && (ent->content != NULL)) {
3936
1.39M
                if (pent != NULL)
3937
1.39M
                    pent->flags |= XML_ENT_EXPANDING;
3938
1.39M
    normChange |= xmlExpandEntityInAttValue(ctxt, buf,
3939
1.39M
                        ent->content, ent, normalize, inSpace, depth, check);
3940
1.39M
                if (pent != NULL)
3941
1.39M
                    pent->flags &= ~XML_ENT_EXPANDING;
3942
1.39M
      }
3943
2.76M
        }
3944
215M
    }
3945
3946
2.00M
    if (chunkSize > 0)
3947
1.26M
        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3948
3949
2.00M
    return(normChange);
3950
2.00M
}
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
42.1k
                            int normalize) {
3964
42.1k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3965
35.2k
                         XML_MAX_HUGE_LENGTH :
3966
42.1k
                         XML_MAX_TEXT_LENGTH;
3967
42.1k
    xmlSBuf buf;
3968
42.1k
    int inSpace = 1;
3969
3970
42.1k
    xmlSBufInit(&buf, maxLength);
3971
3972
42.1k
    xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace,
3973
42.1k
                              ctxt->inputNr, /* check */ 0);
3974
3975
42.1k
    if ((normalize) && (inSpace) && (buf.size > 0))
3976
0
        buf.size--;
3977
3978
42.1k
    return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long"));
3979
42.1k
}
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
367k
                         int special, int isNamespace) {
4020
367k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4021
227k
                         XML_MAX_HUGE_LENGTH :
4022
367k
                         XML_MAX_TEXT_LENGTH;
4023
367k
    xmlSBuf buf;
4024
367k
    xmlChar *ret;
4025
367k
    int c, l, quote, entFlags, chunkSize;
4026
367k
    int inSpace = 1;
4027
367k
    int replaceEntities;
4028
367k
    int normalize = (special & XML_SPECIAL_TYPE_MASK) > XML_ATTRIBUTE_CDATA;
4029
367k
    int attvalFlags = 0;
4030
4031
    /* Always expand namespace URIs */
4032
367k
    replaceEntities = (ctxt->replaceEntities) || (isNamespace);
4033
4034
367k
    xmlSBufInit(&buf, maxLength);
4035
4036
367k
    GROW;
4037
4038
367k
    quote = CUR;
4039
367k
    if ((quote != '"') && (quote != '\'')) {
4040
8.57k
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
4041
8.57k
  return(NULL);
4042
8.57k
    }
4043
359k
    NEXTL(1);
4044
4045
359k
    if (ctxt->inSubset == 0)
4046
296k
        entFlags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
4047
63.1k
    else
4048
63.1k
        entFlags = XML_ENT_VALIDATED;
4049
4050
359k
    inSpace = 1;
4051
359k
    chunkSize = 0;
4052
4053
151M
    while (1) {
4054
151M
        if (PARSER_STOPPED(ctxt))
4055
101
            goto error;
4056
4057
151M
        if (CUR_PTR >= ctxt->input->end) {
4058
34.7k
            xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED,
4059
34.7k
                           "AttValue: ' expected\n");
4060
34.7k
            goto error;
4061
34.7k
        }
4062
4063
        /*
4064
         * TODO: Check growth threshold
4065
         */
4066
151M
        if (ctxt->input->end - CUR_PTR < 10)
4067
374k
            GROW;
4068
4069
151M
        c = CUR;
4070
4071
151M
        if (c >= 0x80) {
4072
60.4M
            l = xmlUTF8MultibyteLen(ctxt, CUR_PTR,
4073
60.4M
                    "invalid character in attribute value\n");
4074
60.4M
            if (l == 0) {
4075
54.6M
                if (chunkSize > 0) {
4076
241k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4077
241k
                    chunkSize = 0;
4078
241k
                }
4079
54.6M
                xmlSBufAddReplChar(&buf);
4080
54.6M
                NEXTL(1);
4081
54.6M
            } else {
4082
5.82M
                chunkSize += l;
4083
5.82M
                NEXTL(l);
4084
5.82M
            }
4085
4086
60.4M
            inSpace = 0;
4087
91.3M
        } else if (c != '&') {
4088
90.2M
            if (c > 0x20) {
4089
51.5M
                if (c == quote)
4090
322k
                    break;
4091
4092
51.2M
                if (c == '<')
4093
803k
                    xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL);
4094
4095
51.2M
                chunkSize += 1;
4096
51.2M
                inSpace = 0;
4097
51.2M
            } else if (!IS_BYTE_CHAR(c)) {
4098
33.0M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
4099
33.0M
                        "invalid character in attribute value\n");
4100
33.0M
                if (chunkSize > 0) {
4101
78.6k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4102
78.6k
                    chunkSize = 0;
4103
78.6k
                }
4104
33.0M
                xmlSBufAddReplChar(&buf);
4105
33.0M
                inSpace = 0;
4106
33.0M
            } else {
4107
                /* Whitespace */
4108
5.63M
                if ((normalize) && (inSpace)) {
4109
                    /* Skip char */
4110
50.6k
                    if (chunkSize > 0) {
4111
37.5k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4112
37.5k
                        chunkSize = 0;
4113
37.5k
                    }
4114
50.6k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4115
5.58M
                } else if (c < 0x20) {
4116
                    /* Convert to space */
4117
3.14M
                    if (chunkSize > 0) {
4118
232k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4119
232k
                        chunkSize = 0;
4120
232k
                    }
4121
4122
3.14M
                    xmlSBufAddCString(&buf, " ", 1);
4123
3.14M
                } else {
4124
2.43M
                    chunkSize += 1;
4125
2.43M
                }
4126
4127
5.63M
                inSpace = 1;
4128
4129
5.63M
                if ((c == 0xD) && (NXT(1) == 0xA))
4130
59.4k
                    CUR_PTR++;
4131
5.63M
            }
4132
4133
89.9M
            NEXTL(1);
4134
89.9M
        } else if (NXT(1) == '#') {
4135
66.8k
            int val;
4136
4137
66.8k
            if (chunkSize > 0) {
4138
56.4k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4139
56.4k
                chunkSize = 0;
4140
56.4k
            }
4141
4142
66.8k
            val = xmlParseCharRef(ctxt);
4143
66.8k
            if (val == 0)
4144
1.52k
                goto error;
4145
4146
65.2k
            if ((val == '&') && (!replaceEntities)) {
4147
                /*
4148
                 * The reparsing will be done in xmlNodeParseContent()
4149
                 * called from SAX2.c
4150
                 */
4151
2.84k
                xmlSBufAddCString(&buf, "&#38;", 5);
4152
2.84k
                inSpace = 0;
4153
62.4k
            } else if (val == ' ') {
4154
1.65k
                if ((normalize) && (inSpace))
4155
201
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4156
1.45k
                else
4157
1.45k
                    xmlSBufAddCString(&buf, " ", 1);
4158
1.65k
                inSpace = 1;
4159
60.7k
            } else {
4160
60.7k
                xmlSBufAddChar(&buf, val);
4161
60.7k
                inSpace = 0;
4162
60.7k
            }
4163
1.00M
        } else {
4164
1.00M
            const xmlChar *name;
4165
1.00M
            xmlEntityPtr ent;
4166
4167
1.00M
            if (chunkSize > 0) {
4168
620k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4169
620k
                chunkSize = 0;
4170
620k
            }
4171
4172
1.00M
            name = xmlParseEntityRefInternal(ctxt);
4173
1.00M
            if (name == NULL) {
4174
                /*
4175
                 * Probably a literal '&' which wasn't escaped.
4176
                 * TODO: Handle gracefully in recovery mode.
4177
                 */
4178
111k
                continue;
4179
111k
            }
4180
4181
898k
            ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1);
4182
898k
            if (ent == NULL)
4183
229k
                continue;
4184
4185
668k
            if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) {
4186
17.7k
                if ((ent->content[0] == '&') && (!replaceEntities))
4187
2.10k
                    xmlSBufAddCString(&buf, "&#38;", 5);
4188
15.6k
                else
4189
15.6k
                    xmlSBufAddString(&buf, ent->content, ent->length);
4190
17.7k
                inSpace = 0;
4191
650k
            } else if (replaceEntities) {
4192
565k
                if (xmlExpandEntityInAttValue(ctxt, &buf,
4193
565k
                        ent->content, ent, normalize, &inSpace, ctxt->inputNr,
4194
565k
                        /* check */ 1) > 0)
4195
7.94k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4196
565k
            } else {
4197
85.1k
                if ((ent->flags & entFlags) != entFlags)
4198
887
                    xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr);
4199
4200
85.1k
                if (xmlParserEntityCheck(ctxt, ent->expandedSize)) {
4201
88
                    ent->content[0] = 0;
4202
88
                    goto error;
4203
88
                }
4204
4205
                /*
4206
                 * Just output the reference
4207
                 */
4208
85.0k
                xmlSBufAddCString(&buf, "&", 1);
4209
85.0k
                xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name));
4210
85.0k
                xmlSBufAddCString(&buf, ";", 1);
4211
4212
85.0k
                inSpace = 0;
4213
85.0k
            }
4214
668k
  }
4215
151M
    }
4216
4217
322k
    if ((buf.mem == NULL) && (outFlags != NULL)) {
4218
166k
        ret = (xmlChar *) CUR_PTR - chunkSize;
4219
4220
166k
        if (attlen != NULL)
4221
166k
            *attlen = chunkSize;
4222
166k
        if ((normalize) && (inSpace) && (chunkSize > 0)) {
4223
290
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4224
290
            *attlen -= 1;
4225
290
        }
4226
4227
        /* Report potential error */
4228
166k
        xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4229
166k
    } else {
4230
156k
        if (chunkSize > 0)
4231
132k
            xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4232
4233
156k
        if ((normalize) && (inSpace) && (buf.size > 0)) {
4234
489
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4235
489
            buf.size--;
4236
489
        }
4237
4238
156k
        ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long");
4239
156k
        attvalFlags |= XML_ATTVAL_ALLOC;
4240
4241
156k
        if (ret != NULL) {
4242
155k
            if (attlen != NULL)
4243
30.6k
                *attlen = buf.size;
4244
155k
        }
4245
156k
    }
4246
4247
322k
    if (outFlags != NULL)
4248
197k
        *outFlags = attvalFlags;
4249
4250
322k
    NEXTL(1);
4251
4252
322k
    return(ret);
4253
4254
36.4k
error:
4255
36.4k
    xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4256
36.4k
    return(NULL);
4257
359k
}
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
166k
xmlParseAttValue(xmlParserCtxt *ctxt) {
4297
166k
    if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL);
4298
166k
    return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0));
4299
166k
}
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
22.6k
xmlParseSystemLiteral(xmlParserCtxt *ctxt) {
4314
22.6k
    xmlChar *buf = NULL;
4315
22.6k
    int len = 0;
4316
22.6k
    int size = XML_PARSER_BUFFER_SIZE;
4317
22.6k
    int cur, l;
4318
22.6k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4319
12.2k
                    XML_MAX_TEXT_LENGTH :
4320
22.6k
                    XML_MAX_NAME_LENGTH;
4321
22.6k
    xmlChar stop;
4322
4323
22.6k
    if (RAW == '"') {
4324
10.7k
        NEXT;
4325
10.7k
  stop = '"';
4326
11.9k
    } else if (RAW == '\'') {
4327
7.03k
        NEXT;
4328
7.03k
  stop = '\'';
4329
7.03k
    } else {
4330
4.89k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4331
4.89k
  return(NULL);
4332
4.89k
    }
4333
4334
17.8k
    buf = xmlMalloc(size);
4335
17.8k
    if (buf == NULL) {
4336
3
        xmlErrMemory(ctxt);
4337
3
  return(NULL);
4338
3
    }
4339
17.8k
    cur = xmlCurrentCharRecover(ctxt, &l);
4340
4.37M
    while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */
4341
4.35M
  if (len + 5 >= size) {
4342
3.80k
      xmlChar *tmp;
4343
3.80k
            int newSize;
4344
4345
3.80k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4346
3.80k
            if (newSize < 0) {
4347
3
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral");
4348
3
                xmlFree(buf);
4349
3
                return(NULL);
4350
3
            }
4351
3.80k
      tmp = xmlRealloc(buf, newSize);
4352
3.80k
      if (tmp == NULL) {
4353
2
          xmlFree(buf);
4354
2
    xmlErrMemory(ctxt);
4355
2
    return(NULL);
4356
2
      }
4357
3.80k
      buf = tmp;
4358
3.80k
            size = newSize;
4359
3.80k
  }
4360
4.35M
  COPY_BUF(buf, len, cur);
4361
4.35M
  NEXTL(l);
4362
4.35M
  cur = xmlCurrentCharRecover(ctxt, &l);
4363
4.35M
    }
4364
17.7k
    buf[len] = 0;
4365
17.7k
    if (!IS_CHAR(cur)) {
4366
2.20k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4367
15.5k
    } else {
4368
15.5k
  NEXT;
4369
15.5k
    }
4370
17.7k
    return(buf);
4371
17.8k
}
4372
4373
/**
4374
 * Parse an XML public literal
4375
 *
4376
 * @deprecated Internal function, don't use.
4377
 *
4378
 *     [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'"
4379
 *
4380
 * @param ctxt  an XML parser context
4381
 * @returns the PubidLiteral parsed or NULL.
4382
 */
4383
4384
xmlChar *
4385
10.3k
xmlParsePubidLiteral(xmlParserCtxt *ctxt) {
4386
10.3k
    xmlChar *buf = NULL;
4387
10.3k
    int len = 0;
4388
10.3k
    int size = XML_PARSER_BUFFER_SIZE;
4389
10.3k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4390
5.81k
                    XML_MAX_TEXT_LENGTH :
4391
10.3k
                    XML_MAX_NAME_LENGTH;
4392
10.3k
    xmlChar cur;
4393
10.3k
    xmlChar stop;
4394
4395
10.3k
    if (RAW == '"') {
4396
1.32k
        NEXT;
4397
1.32k
  stop = '"';
4398
8.99k
    } else if (RAW == '\'') {
4399
5.06k
        NEXT;
4400
5.06k
  stop = '\'';
4401
5.06k
    } else {
4402
3.93k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4403
3.93k
  return(NULL);
4404
3.93k
    }
4405
6.38k
    buf = xmlMalloc(size);
4406
6.38k
    if (buf == NULL) {
4407
2
  xmlErrMemory(ctxt);
4408
2
  return(NULL);
4409
2
    }
4410
6.38k
    cur = CUR;
4411
250k
    while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) &&
4412
244k
           (PARSER_STOPPED(ctxt) == 0)) { /* checked */
4413
244k
  if (len + 1 >= size) {
4414
293
      xmlChar *tmp;
4415
293
            int newSize;
4416
4417
293
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4418
293
            if (newSize < 0) {
4419
1
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID");
4420
1
                xmlFree(buf);
4421
1
                return(NULL);
4422
1
            }
4423
292
      tmp = xmlRealloc(buf, newSize);
4424
292
      if (tmp == NULL) {
4425
1
    xmlErrMemory(ctxt);
4426
1
    xmlFree(buf);
4427
1
    return(NULL);
4428
1
      }
4429
291
      buf = tmp;
4430
291
            size = newSize;
4431
291
  }
4432
244k
  buf[len++] = cur;
4433
244k
  NEXT;
4434
244k
  cur = CUR;
4435
244k
    }
4436
6.38k
    buf[len] = 0;
4437
6.38k
    if (cur != stop) {
4438
937
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4439
5.44k
    } else {
4440
5.44k
  NEXTL(1);
4441
5.44k
    }
4442
6.38k
    return(buf);
4443
6.38k
}
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
914k
              int isBlank) {
4488
914k
    int checkBlanks;
4489
4490
914k
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
4491
187k
        return;
4492
4493
727k
    checkBlanks = (!ctxt->keepBlanks) ||
4494
256k
                  (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
727k
    if ((checkBlanks) &&
4502
471k
        (areBlanks(ctxt, buf, size, isBlank))) {
4503
10.8k
        if ((ctxt->sax->ignorableWhitespace != NULL) &&
4504
10.8k
            (ctxt->keepBlanks))
4505
0
            ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
4506
716k
    } else {
4507
716k
        if (ctxt->sax->characters != NULL)
4508
716k
            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
716k
        if ((checkBlanks) && (*ctxt->space == -1))
4515
195k
            *ctxt->space = -2;
4516
716k
    }
4517
727k
}
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
8.54M
xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) {
4534
8.54M
    const xmlChar *in;
4535
8.54M
    int line = ctxt->input->line;
4536
8.54M
    int col = ctxt->input->col;
4537
8.54M
    int ccol;
4538
8.54M
    int terminate = 0;
4539
4540
8.54M
    GROW;
4541
    /*
4542
     * Accelerated common case where input don't need to be
4543
     * modified before passing it to the handler.
4544
     */
4545
8.54M
    in = ctxt->input->cur;
4546
8.54M
    do {
4547
8.65M
get_more_space:
4548
8.87M
        while (*in == 0x20) { in++; ctxt->input->col++; }
4549
8.65M
        if (*in == 0xA) {
4550
3.50M
            do {
4551
3.50M
                ctxt->input->line++; ctxt->input->col = 1;
4552
3.50M
                in++;
4553
3.50M
            } while (*in == 0xA);
4554
108k
            goto get_more_space;
4555
108k
        }
4556
8.54M
        if (*in == '<') {
4557
171k
            while (in > ctxt->input->cur) {
4558
85.9k
                const xmlChar *tmp = ctxt->input->cur;
4559
85.9k
                size_t nbchar = in - tmp;
4560
4561
85.9k
                if (nbchar > XML_MAX_ITEMS)
4562
0
                    nbchar = XML_MAX_ITEMS;
4563
85.9k
                ctxt->input->cur += nbchar;
4564
4565
85.9k
                xmlCharacters(ctxt, tmp, nbchar, 1);
4566
85.9k
            }
4567
85.9k
            return;
4568
85.9k
        }
4569
4570
8.96M
get_more:
4571
8.96M
        ccol = ctxt->input->col;
4572
16.5M
        while (test_char_data[*in]) {
4573
7.55M
            in++;
4574
7.55M
            ccol++;
4575
7.55M
        }
4576
8.96M
        ctxt->input->col = ccol;
4577
8.96M
        if (*in == 0xA) {
4578
2.98M
            do {
4579
2.98M
                ctxt->input->line++; ctxt->input->col = 1;
4580
2.98M
                in++;
4581
2.98M
            } while (*in == 0xA);
4582
52.5k
            goto get_more;
4583
52.5k
        }
4584
8.91M
        if (*in == ']') {
4585
452k
            size_t avail = ctxt->input->end - in;
4586
4587
452k
            if (partial && avail < 2) {
4588
36
                terminate = 1;
4589
36
                goto invoke_callback;
4590
36
            }
4591
452k
            if (in[1] == ']') {
4592
305k
                if (partial && avail < 3) {
4593
418
                    terminate = 1;
4594
418
                    goto invoke_callback;
4595
418
                }
4596
305k
                if (in[2] == '>')
4597
2.54k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4598
305k
            }
4599
4600
452k
            in++;
4601
452k
            ctxt->input->col++;
4602
452k
            goto get_more;
4603
452k
        }
4604
4605
8.46M
invoke_callback:
4606
9.12M
        while (in > ctxt->input->cur) {
4607
665k
            const xmlChar *tmp = ctxt->input->cur;
4608
665k
            size_t nbchar = in - tmp;
4609
4610
665k
            if (nbchar > XML_MAX_ITEMS)
4611
0
                nbchar = XML_MAX_ITEMS;
4612
665k
            ctxt->input->cur += nbchar;
4613
4614
665k
            xmlCharacters(ctxt, tmp, nbchar, 0);
4615
4616
665k
            line = ctxt->input->line;
4617
665k
            col = ctxt->input->col;
4618
665k
        }
4619
8.46M
        ctxt->input->cur = in;
4620
8.46M
        if (*in == 0xD) {
4621
18.9k
            in++;
4622
18.9k
            if (*in == 0xA) {
4623
6.07k
                ctxt->input->cur = in;
4624
6.07k
                in++;
4625
6.07k
                ctxt->input->line++; ctxt->input->col = 1;
4626
6.07k
                continue; /* while */
4627
6.07k
            }
4628
12.8k
            in--;
4629
12.8k
        }
4630
8.45M
        if (*in == '<') {
4631
537k
            return;
4632
537k
        }
4633
7.91M
        if (*in == '&') {
4634
39.5k
            return;
4635
39.5k
        }
4636
7.88M
        if (terminate) {
4637
454
            return;
4638
454
        }
4639
7.87M
        SHRINK;
4640
7.87M
        GROW;
4641
7.87M
        in = ctxt->input->cur;
4642
7.88M
    } while (((*in >= 0x20) && (*in <= 0x7F)) ||
4643
7.88M
             (*in == 0x09) || (*in == 0x0a));
4644
7.88M
    ctxt->input->line = line;
4645
7.88M
    ctxt->input->col = col;
4646
7.88M
    xmlParseCharDataComplex(ctxt, partial);
4647
7.88M
}
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
7.88M
xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
4661
7.88M
    xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5];
4662
7.88M
    int nbchar = 0;
4663
7.88M
    int cur, l;
4664
4665
7.88M
    cur = xmlCurrentCharRecover(ctxt, &l);
4666
17.7M
    while ((cur != '<') && /* checked */
4667
17.7M
           (cur != '&') &&
4668
17.7M
     (IS_CHAR(cur))) {
4669
9.89M
        if (cur == ']') {
4670
545k
            size_t avail = ctxt->input->end - ctxt->input->cur;
4671
4672
545k
            if (partial && avail < 2)
4673
55
                break;
4674
545k
            if (NXT(1) == ']') {
4675
529k
                if (partial && avail < 3)
4676
1.21k
                    break;
4677
528k
                if (NXT(2) == '>')
4678
828
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4679
528k
            }
4680
545k
        }
4681
4682
9.88M
  COPY_BUF(buf, nbchar, cur);
4683
  /* move current position before possible calling of ctxt->sax->characters */
4684
9.88M
  NEXTL(l);
4685
9.88M
  if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) {
4686
55.9k
      buf[nbchar] = 0;
4687
4688
55.9k
            xmlCharacters(ctxt, buf, nbchar, 0);
4689
55.9k
      nbchar = 0;
4690
55.9k
            SHRINK;
4691
55.9k
  }
4692
9.88M
  cur = xmlCurrentCharRecover(ctxt, &l);
4693
9.88M
    }
4694
7.88M
    if (nbchar != 0) {
4695
107k
        buf[nbchar] = 0;
4696
4697
107k
        xmlCharacters(ctxt, buf, nbchar, 0);
4698
107k
    }
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
7.88M
    if (ctxt->input->cur < ctxt->input->end) {
4707
7.87M
        if ((cur == 0) && (CUR != 0)) {
4708
444
            if (partial == 0) {
4709
350
                xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4710
350
                        "Incomplete UTF-8 sequence starting with %02X\n", CUR);
4711
350
                NEXTL(1);
4712
350
            }
4713
7.87M
        } else if ((cur != '<') && (cur != '&') && (cur != ']')) {
4714
            /* Generate the error and skip the offending character */
4715
7.81M
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4716
7.81M
                              "PCDATA invalid Char value %d\n", cur);
4717
7.81M
            NEXTL(l);
4718
7.81M
        }
4719
7.87M
    }
4720
7.88M
}
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
42.3k
xmlParseExternalID(xmlParserCtxt *ctxt, xmlChar **publicId, int strict) {
4756
42.3k
    xmlChar *URI = NULL;
4757
4758
42.3k
    *publicId = NULL;
4759
42.3k
    if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) {
4760
13.8k
        SKIP(6);
4761
13.8k
  if (SKIP_BLANKS == 0) {
4762
301
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4763
301
                     "Space required after 'SYSTEM'\n");
4764
301
  }
4765
13.8k
  URI = xmlParseSystemLiteral(ctxt);
4766
13.8k
  if (URI == NULL) {
4767
548
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4768
548
        }
4769
28.4k
    } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) {
4770
10.3k
        SKIP(6);
4771
10.3k
  if (SKIP_BLANKS == 0) {
4772
817
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4773
817
        "Space required after 'PUBLIC'\n");
4774
817
  }
4775
10.3k
  *publicId = xmlParsePubidLiteral(ctxt);
4776
10.3k
  if (*publicId == NULL) {
4777
3.93k
      xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL);
4778
3.93k
  }
4779
10.3k
  if (strict) {
4780
      /*
4781
       * We don't handle [83] so "S SystemLiteral" is required.
4782
       */
4783
8.68k
      if (SKIP_BLANKS == 0) {
4784
4.34k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4785
4.34k
      "Space required after the Public Identifier\n");
4786
4.34k
      }
4787
8.68k
  } else {
4788
      /*
4789
       * We handle [83] so we return immediately, if
4790
       * "S SystemLiteral" is not detected. We skip blanks if no
4791
             * system literal was found, but this is harmless since we must
4792
             * be at the end of a NotationDecl.
4793
       */
4794
1.63k
      if (SKIP_BLANKS == 0) return(NULL);
4795
536
      if ((CUR != '\'') && (CUR != '"')) return(NULL);
4796
536
  }
4797
8.79k
  URI = xmlParseSystemLiteral(ctxt);
4798
8.79k
  if (URI == NULL) {
4799
4.35k
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4800
4.35k
        }
4801
8.79k
    }
4802
40.8k
    return(URI);
4803
42.3k
}
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
117k
                       size_t len, size_t size) {
4820
117k
    int q, ql;
4821
117k
    int r, rl;
4822
117k
    int cur, l;
4823
117k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4824
44.3k
                    XML_MAX_HUGE_LENGTH :
4825
117k
                    XML_MAX_TEXT_LENGTH;
4826
4827
117k
    if (buf == NULL) {
4828
24.4k
        len = 0;
4829
24.4k
  size = XML_PARSER_BUFFER_SIZE;
4830
24.4k
  buf = xmlMalloc(size);
4831
24.4k
  if (buf == NULL) {
4832
2
      xmlErrMemory(ctxt);
4833
2
      return;
4834
2
  }
4835
24.4k
    }
4836
117k
    q = xmlCurrentCharRecover(ctxt, &ql);
4837
117k
    if (q == 0)
4838
1.05k
        goto not_terminated;
4839
116k
    if (!IS_CHAR(q)) {
4840
476
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4841
476
                          "xmlParseComment: invalid xmlChar value %d\n",
4842
476
                    q);
4843
476
  xmlFree (buf);
4844
476
  return;
4845
476
    }
4846
115k
    NEXTL(ql);
4847
115k
    r = xmlCurrentCharRecover(ctxt, &rl);
4848
115k
    if (r == 0)
4849
352
        goto not_terminated;
4850
115k
    if (!IS_CHAR(r)) {
4851
428
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4852
428
                          "xmlParseComment: invalid xmlChar value %d\n",
4853
428
                    r);
4854
428
  xmlFree (buf);
4855
428
  return;
4856
428
    }
4857
114k
    NEXTL(rl);
4858
114k
    cur = xmlCurrentCharRecover(ctxt, &l);
4859
114k
    if (cur == 0)
4860
549
        goto not_terminated;
4861
13.4M
    while (IS_CHAR(cur) && /* checked */
4862
13.4M
           ((cur != '>') ||
4863
13.3M
      (r != '-') || (q != '-'))) {
4864
13.3M
  if ((r == '-') && (q == '-')) {
4865
715k
      xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL);
4866
715k
  }
4867
13.3M
  if (len + 5 >= size) {
4868
41.4k
      xmlChar *tmp;
4869
41.4k
            int newSize;
4870
4871
41.4k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4872
41.4k
            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
41.4k
      tmp = xmlRealloc(buf, newSize);
4879
41.4k
      if (tmp == NULL) {
4880
6
    xmlErrMemory(ctxt);
4881
6
    xmlFree(buf);
4882
6
    return;
4883
6
      }
4884
41.4k
      buf = tmp;
4885
41.4k
            size = newSize;
4886
41.4k
  }
4887
13.3M
  COPY_BUF(buf, len, q);
4888
4889
13.3M
  q = r;
4890
13.3M
  ql = rl;
4891
13.3M
  r = cur;
4892
13.3M
  rl = l;
4893
4894
13.3M
  NEXTL(l);
4895
13.3M
  cur = xmlCurrentCharRecover(ctxt, &l);
4896
4897
13.3M
    }
4898
114k
    buf[len] = 0;
4899
114k
    if (cur == 0) {
4900
4.77k
  xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4901
4.77k
                       "Comment not terminated \n<!--%.50s\n", buf);
4902
109k
    } else if (!IS_CHAR(cur)) {
4903
426
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4904
426
                          "xmlParseComment: invalid xmlChar value %d\n",
4905
426
                    cur);
4906
109k
    } else {
4907
109k
        NEXT;
4908
109k
  if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
4909
109k
      (!ctxt->disableSAX))
4910
106k
      ctxt->sax->comment(ctxt->userData, buf);
4911
109k
    }
4912
114k
    xmlFree(buf);
4913
114k
    return;
4914
1.95k
not_terminated:
4915
1.95k
    xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4916
1.95k
       "Comment not terminated\n", NULL);
4917
1.95k
    xmlFree(buf);
4918
1.95k
}
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
485k
xmlParseComment(xmlParserCtxt *ctxt) {
4933
485k
    xmlChar *buf = NULL;
4934
485k
    size_t size = XML_PARSER_BUFFER_SIZE;
4935
485k
    size_t len = 0;
4936
485k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4937
165k
                       XML_MAX_HUGE_LENGTH :
4938
485k
                       XML_MAX_TEXT_LENGTH;
4939
485k
    const xmlChar *in;
4940
485k
    size_t nbchar = 0;
4941
485k
    int ccol;
4942
4943
    /*
4944
     * Check that there is a comment right here.
4945
     */
4946
485k
    if ((RAW != '<') || (NXT(1) != '!'))
4947
0
        return;
4948
485k
    SKIP(2);
4949
485k
    if ((RAW != '-') || (NXT(1) != '-'))
4950
10
        return;
4951
485k
    SKIP(2);
4952
485k
    GROW;
4953
4954
    /*
4955
     * Accelerated common case where input don't need to be
4956
     * modified before passing it to the handler.
4957
     */
4958
485k
    in = ctxt->input->cur;
4959
485k
    do {
4960
485k
  if (*in == 0xA) {
4961
4.09k
      do {
4962
4.09k
    ctxt->input->line++; ctxt->input->col = 1;
4963
4.09k
    in++;
4964
4.09k
      } while (*in == 0xA);
4965
1.03k
  }
4966
934k
get_more:
4967
934k
        ccol = ctxt->input->col;
4968
6.45M
  while (((*in > '-') && (*in <= 0x7F)) ||
4969
1.99M
         ((*in >= 0x20) && (*in < '-')) ||
4970
5.51M
         (*in == 0x09)) {
4971
5.51M
        in++;
4972
5.51M
        ccol++;
4973
5.51M
  }
4974
934k
  ctxt->input->col = ccol;
4975
934k
  if (*in == 0xA) {
4976
296k
      do {
4977
296k
    ctxt->input->line++; ctxt->input->col = 1;
4978
296k
    in++;
4979
296k
      } while (*in == 0xA);
4980
19.4k
      goto get_more;
4981
19.4k
  }
4982
915k
  nbchar = in - ctxt->input->cur;
4983
  /*
4984
   * save current set of data
4985
   */
4986
915k
  if (nbchar > 0) {
4987
559k
            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
559k
            if (buf == NULL) {
4994
147k
                if ((*in == '-') && (in[1] == '-'))
4995
53.5k
                    size = nbchar + 1;
4996
93.5k
                else
4997
93.5k
                    size = XML_PARSER_BUFFER_SIZE + nbchar;
4998
147k
                buf = xmlMalloc(size);
4999
147k
                if (buf == NULL) {
5000
4
                    xmlErrMemory(ctxt);
5001
4
                    return;
5002
4
                }
5003
147k
                len = 0;
5004
412k
            } else if (len + nbchar + 1 >= size) {
5005
10.4k
                xmlChar *new_buf;
5006
10.4k
                size += len + nbchar + XML_PARSER_BUFFER_SIZE;
5007
10.4k
                new_buf = xmlRealloc(buf, size);
5008
10.4k
                if (new_buf == NULL) {
5009
1
                    xmlErrMemory(ctxt);
5010
1
                    xmlFree(buf);
5011
1
                    return;
5012
1
                }
5013
10.4k
                buf = new_buf;
5014
10.4k
            }
5015
559k
            memcpy(&buf[len], ctxt->input->cur, nbchar);
5016
559k
            len += nbchar;
5017
559k
            buf[len] = 0;
5018
559k
  }
5019
915k
  ctxt->input->cur = in;
5020
915k
  if (*in == 0xA) {
5021
0
      in++;
5022
0
      ctxt->input->line++; ctxt->input->col = 1;
5023
0
  }
5024
915k
  if (*in == 0xD) {
5025
86.0k
      in++;
5026
86.0k
      if (*in == 0xA) {
5027
16.3k
    ctxt->input->cur = in;
5028
16.3k
    in++;
5029
16.3k
    ctxt->input->line++; ctxt->input->col = 1;
5030
16.3k
    goto get_more;
5031
16.3k
      }
5032
69.7k
      in--;
5033
69.7k
  }
5034
898k
  SHRINK;
5035
898k
  GROW;
5036
898k
  in = ctxt->input->cur;
5037
898k
  if (*in == '-') {
5038
781k
      if (in[1] == '-') {
5039
577k
          if (in[2] == '>') {
5040
368k
        SKIP(3);
5041
368k
        if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
5042
368k
            (!ctxt->disableSAX)) {
5043
364k
      if (buf != NULL)
5044
52.7k
          ctxt->sax->comment(ctxt->userData, buf);
5045
311k
      else
5046
311k
          ctxt->sax->comment(ctxt->userData, BAD_CAST "");
5047
364k
        }
5048
368k
        if (buf != NULL)
5049
54.3k
            xmlFree(buf);
5050
368k
        return;
5051
368k
    }
5052
209k
    if (buf != NULL) {
5053
197k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5054
197k
                          "Double hyphen within comment: "
5055
197k
                                      "<!--%.50s\n",
5056
197k
              buf);
5057
197k
    } else
5058
11.9k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5059
11.9k
                          "Double hyphen within comment\n", NULL);
5060
209k
    in++;
5061
209k
    ctxt->input->col++;
5062
209k
      }
5063
412k
      in++;
5064
412k
      ctxt->input->col++;
5065
412k
      goto get_more;
5066
781k
  }
5067
898k
    } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a));
5068
117k
    xmlParseCommentComplex(ctxt, buf, len, size);
5069
117k
}
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
104k
xmlParsePITarget(xmlParserCtxt *ctxt) {
5085
104k
    const xmlChar *name;
5086
5087
104k
    name = xmlParseName(ctxt);
5088
104k
    if ((name != NULL) &&
5089
100k
        ((name[0] == 'x') || (name[0] == 'X')) &&
5090
13.1k
        ((name[1] == 'm') || (name[1] == 'M')) &&
5091
11.0k
        ((name[2] == 'l') || (name[2] == 'L'))) {
5092
6.41k
  int i;
5093
6.41k
  if ((name[0] == 'x') && (name[1] == 'm') &&
5094
5.85k
      (name[2] == 'l') && (name[3] == 0)) {
5095
5.13k
      xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5096
5.13k
     "XML declaration allowed only at the start of the document\n");
5097
5.13k
      return(name);
5098
5.13k
  } else if (name[3] == 0) {
5099
435
      xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL);
5100
435
      return(name);
5101
435
  }
5102
2.43k
  for (i = 0;;i++) {
5103
2.43k
      if (xmlW3CPIs[i] == NULL) break;
5104
1.68k
      if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i]))
5105
90
          return(name);
5106
1.68k
  }
5107
757
  xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5108
757
          "xmlParsePITarget: invalid name prefix 'xml'\n",
5109
757
          NULL, NULL);
5110
757
    }
5111
98.7k
    if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) {
5112
771
  xmlNsErr(ctxt, XML_NS_ERR_COLON,
5113
771
     "colons are forbidden from PI names '%s'\n", name, NULL, NULL);
5114
771
    }
5115
98.7k
    return(name);
5116
104k
}
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
104k
xmlParsePI(xmlParserCtxt *ctxt) {
5199
104k
    xmlChar *buf = NULL;
5200
104k
    size_t len = 0;
5201
104k
    size_t size = XML_PARSER_BUFFER_SIZE;
5202
104k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5203
57.0k
                       XML_MAX_HUGE_LENGTH :
5204
104k
                       XML_MAX_TEXT_LENGTH;
5205
104k
    int cur, l;
5206
104k
    const xmlChar *target;
5207
5208
104k
    if ((RAW == '<') && (NXT(1) == '?')) {
5209
  /*
5210
   * this is a Processing Instruction.
5211
   */
5212
104k
  SKIP(2);
5213
5214
  /*
5215
   * Parse the target name and check for special support like
5216
   * namespace.
5217
   */
5218
104k
        target = xmlParsePITarget(ctxt);
5219
104k
  if (target != NULL) {
5220
100k
      if ((RAW == '?') && (NXT(1) == '>')) {
5221
81.3k
    SKIP(2);
5222
5223
    /*
5224
     * SAX: PI detected.
5225
     */
5226
81.3k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5227
80.5k
        (ctxt->sax->processingInstruction != NULL))
5228
80.5k
        ctxt->sax->processingInstruction(ctxt->userData,
5229
80.5k
                                         target, NULL);
5230
81.3k
    return;
5231
81.3k
      }
5232
18.8k
      buf = xmlMalloc(size);
5233
18.8k
      if (buf == NULL) {
5234
10
    xmlErrMemory(ctxt);
5235
10
    return;
5236
10
      }
5237
18.8k
      if (SKIP_BLANKS == 0) {
5238
7.40k
    xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED,
5239
7.40k
        "ParsePI: PI %s space expected\n", target);
5240
7.40k
      }
5241
18.8k
      cur = xmlCurrentCharRecover(ctxt, &l);
5242
5.04M
      while (IS_CHAR(cur) && /* checked */
5243
5.04M
       ((cur != '?') || (NXT(1) != '>'))) {
5244
5.03M
    if (len + 5 >= size) {
5245
15.7k
        xmlChar *tmp;
5246
15.7k
                    int newSize;
5247
5248
15.7k
                    newSize = xmlGrowCapacity(size, 1, 1, maxLength);
5249
15.7k
                    if (newSize < 0) {
5250
0
                        xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5251
0
                                          "PI %s too big found", target);
5252
0
                        xmlFree(buf);
5253
0
                        return;
5254
0
                    }
5255
15.7k
        tmp = xmlRealloc(buf, newSize);
5256
15.7k
        if (tmp == NULL) {
5257
8
      xmlErrMemory(ctxt);
5258
8
      xmlFree(buf);
5259
8
      return;
5260
8
        }
5261
15.7k
        buf = tmp;
5262
15.7k
                    size = newSize;
5263
15.7k
    }
5264
5.03M
    COPY_BUF(buf, len, cur);
5265
5.03M
    NEXTL(l);
5266
5.03M
    cur = xmlCurrentCharRecover(ctxt, &l);
5267
5.03M
      }
5268
18.8k
      buf[len] = 0;
5269
18.8k
      if (cur != '?') {
5270
8.05k
    xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5271
8.05k
          "ParsePI: PI %s never end ...\n", target);
5272
10.7k
      } else {
5273
10.7k
    SKIP(2);
5274
5275
10.7k
#ifdef LIBXML_CATALOG_ENABLED
5276
10.7k
    if ((ctxt->inSubset == 0) &&
5277
7.34k
        (xmlStrEqual(target, XML_CATALOG_PI))) {
5278
240
        xmlCatalogAllow allow = xmlCatalogGetDefaults();
5279
5280
240
        if ((ctxt->options & XML_PARSE_CATALOG_PI) &&
5281
163
                        ((allow == XML_CATA_ALLOW_DOCUMENT) ||
5282
163
       (allow == XML_CATA_ALLOW_ALL)))
5283
0
      xmlParseCatalogPI(ctxt, buf);
5284
240
    }
5285
10.7k
#endif
5286
5287
    /*
5288
     * SAX: PI detected.
5289
     */
5290
10.7k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5291
10.1k
        (ctxt->sax->processingInstruction != NULL))
5292
10.1k
        ctxt->sax->processingInstruction(ctxt->userData,
5293
10.1k
                                         target, buf);
5294
10.7k
      }
5295
18.8k
      xmlFree(buf);
5296
18.8k
  } else {
5297
4.23k
      xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL);
5298
4.23k
  }
5299
104k
    }
5300
104k
}
5301
5302
/**
5303
 * Parse a notation declaration. Always consumes '<!'.
5304
 *
5305
 * @deprecated Internal function, don't use.
5306
 *
5307
 *     [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID |  PublicID)
5308
 *                           S? '>'
5309
 *
5310
 * Hence there is actually 3 choices:
5311
 *
5312
 *     'PUBLIC' S PubidLiteral
5313
 *     'PUBLIC' S PubidLiteral S SystemLiteral
5314
 *     'SYSTEM' S SystemLiteral
5315
 *
5316
 * See the NOTE on #xmlParseExternalID.
5317
 *
5318
 * @param ctxt  an XML parser context
5319
 */
5320
5321
void
5322
4.38k
xmlParseNotationDecl(xmlParserCtxt *ctxt) {
5323
4.38k
    const xmlChar *name;
5324
4.38k
    xmlChar *Pubid;
5325
4.38k
    xmlChar *Systemid;
5326
5327
4.38k
    if ((CUR != '<') || (NXT(1) != '!'))
5328
0
        return;
5329
4.38k
    SKIP(2);
5330
5331
4.38k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5332
4.35k
#ifdef LIBXML_VALID_ENABLED
5333
4.35k
  int oldInputNr = ctxt->inputNr;
5334
4.35k
#endif
5335
5336
4.35k
  SKIP(8);
5337
4.35k
  if (SKIP_BLANKS_PE == 0) {
5338
110
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5339
110
         "Space required after '<!NOTATION'\n");
5340
110
      return;
5341
110
  }
5342
5343
4.24k
        name = xmlParseName(ctxt);
5344
4.24k
  if (name == NULL) {
5345
353
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5346
353
      return;
5347
353
  }
5348
3.89k
  if (xmlStrchr(name, ':') != NULL) {
5349
106
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5350
106
         "colons are forbidden from notation names '%s'\n",
5351
106
         name, NULL, NULL);
5352
106
  }
5353
3.89k
  if (SKIP_BLANKS_PE == 0) {
5354
469
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5355
469
         "Space required after the NOTATION name'\n");
5356
469
      return;
5357
469
  }
5358
5359
  /*
5360
   * Parse the IDs.
5361
   */
5362
3.42k
  Systemid = xmlParseExternalID(ctxt, &Pubid, 0);
5363
3.42k
  SKIP_BLANKS_PE;
5364
5365
3.42k
  if (RAW == '>') {
5366
2.59k
#ifdef LIBXML_VALID_ENABLED
5367
2.59k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5368
10
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5369
10
                           "Notation declaration doesn't start and stop"
5370
10
                                 " in the same entity\n",
5371
10
                                 NULL, NULL);
5372
10
      }
5373
2.59k
#endif
5374
2.59k
      NEXT;
5375
2.59k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5376
1.90k
    (ctxt->sax->notationDecl != NULL))
5377
1.90k
    ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid);
5378
2.59k
  } else {
5379
827
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5380
827
  }
5381
3.42k
  if (Systemid != NULL) xmlFree(Systemid);
5382
3.42k
  if (Pubid != NULL) xmlFree(Pubid);
5383
3.42k
    }
5384
4.38k
}
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
151k
xmlParseEntityDecl(xmlParserCtxt *ctxt) {
5411
151k
    const xmlChar *name = NULL;
5412
151k
    xmlChar *value = NULL;
5413
151k
    xmlChar *URI = NULL, *literal = NULL;
5414
151k
    const xmlChar *ndata = NULL;
5415
151k
    int isParameter = 0;
5416
151k
    xmlChar *orig = NULL;
5417
5418
151k
    if ((CUR != '<') || (NXT(1) != '!'))
5419
0
        return;
5420
151k
    SKIP(2);
5421
5422
    /* GROW; done in the caller */
5423
151k
    if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5424
151k
#ifdef LIBXML_VALID_ENABLED
5425
151k
  int oldInputNr = ctxt->inputNr;
5426
151k
#endif
5427
5428
151k
  SKIP(6);
5429
151k
  if (SKIP_BLANKS_PE == 0) {
5430
1.89k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5431
1.89k
         "Space required after '<!ENTITY'\n");
5432
1.89k
  }
5433
5434
151k
  if (RAW == '%') {
5435
59.0k
      NEXT;
5436
59.0k
      if (SKIP_BLANKS_PE == 0) {
5437
269
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5438
269
             "Space required after '%%'\n");
5439
269
      }
5440
59.0k
      isParameter = 1;
5441
59.0k
  }
5442
5443
151k
        name = xmlParseName(ctxt);
5444
151k
  if (name == NULL) {
5445
27.4k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5446
27.4k
                     "xmlParseEntityDecl: no name\n");
5447
27.4k
            return;
5448
27.4k
  }
5449
124k
  if (xmlStrchr(name, ':') != NULL) {
5450
965
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5451
965
         "colons are forbidden from entities names '%s'\n",
5452
965
         name, NULL, NULL);
5453
965
  }
5454
124k
  if (SKIP_BLANKS_PE == 0) {
5455
3.20k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5456
3.20k
         "Space required after the entity name\n");
5457
3.20k
  }
5458
5459
  /*
5460
   * handle the various case of definitions...
5461
   */
5462
124k
  if (isParameter) {
5463
58.6k
      if ((RAW == '"') || (RAW == '\'')) {
5464
51.0k
          value = xmlParseEntityValue(ctxt, &orig);
5465
51.0k
    if (value) {
5466
49.4k
        if ((ctxt->sax != NULL) &&
5467
49.4k
      (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5468
45.3k
      ctxt->sax->entityDecl(ctxt->userData, name,
5469
45.3k
                        XML_INTERNAL_PARAMETER_ENTITY,
5470
45.3k
            NULL, NULL, value);
5471
49.4k
    }
5472
51.0k
      } else {
5473
7.56k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5474
7.56k
    if ((URI == NULL) && (literal == NULL)) {
5475
549
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5476
549
    }
5477
7.56k
    if (URI) {
5478
6.75k
                    if (xmlStrchr(URI, '#')) {
5479
395
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5480
6.35k
                    } else {
5481
6.35k
                        if ((ctxt->sax != NULL) &&
5482
6.35k
                            (!ctxt->disableSAX) &&
5483
5.86k
                            (ctxt->sax->entityDecl != NULL))
5484
5.86k
                            ctxt->sax->entityDecl(ctxt->userData, name,
5485
5.86k
                                        XML_EXTERNAL_PARAMETER_ENTITY,
5486
5.86k
                                        literal, URI, NULL);
5487
6.35k
                    }
5488
6.75k
    }
5489
7.56k
      }
5490
65.5k
  } else {
5491
65.5k
      if ((RAW == '"') || (RAW == '\'')) {
5492
52.4k
          value = xmlParseEntityValue(ctxt, &orig);
5493
52.4k
    if ((ctxt->sax != NULL) &&
5494
52.4k
        (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5495
51.0k
        ctxt->sax->entityDecl(ctxt->userData, name,
5496
51.0k
        XML_INTERNAL_GENERAL_ENTITY,
5497
51.0k
        NULL, NULL, value);
5498
    /*
5499
     * For expat compatibility in SAX mode.
5500
     */
5501
52.4k
    if ((ctxt->myDoc == NULL) ||
5502
52.4k
        (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) {
5503
80
        if (ctxt->myDoc == NULL) {
5504
10
      ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5505
10
      if (ctxt->myDoc == NULL) {
5506
1
          xmlErrMemory(ctxt);
5507
1
          goto done;
5508
1
      }
5509
9
      ctxt->myDoc->properties = XML_DOC_INTERNAL;
5510
9
        }
5511
79
        if (ctxt->myDoc->intSubset == NULL) {
5512
9
      ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5513
9
              BAD_CAST "fake", NULL, NULL);
5514
9
                        if (ctxt->myDoc->intSubset == NULL) {
5515
1
                            xmlErrMemory(ctxt);
5516
1
                            goto done;
5517
1
                        }
5518
9
                    }
5519
5520
78
        xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY,
5521
78
                    NULL, NULL, value);
5522
78
    }
5523
52.4k
      } else {
5524
13.1k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5525
13.1k
    if ((URI == NULL) && (literal == NULL)) {
5526
5.80k
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5527
5.80k
    }
5528
13.1k
    if (URI) {
5529
6.77k
                    if (xmlStrchr(URI, '#')) {
5530
312
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5531
312
                    }
5532
6.77k
    }
5533
13.1k
    if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) {
5534
5.67k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5535
5.67k
           "Space required before 'NDATA'\n");
5536
5.67k
    }
5537
13.1k
    if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) {
5538
826
        SKIP(5);
5539
826
        if (SKIP_BLANKS_PE == 0) {
5540
254
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5541
254
               "Space required after 'NDATA'\n");
5542
254
        }
5543
826
        ndata = xmlParseName(ctxt);
5544
826
        if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5545
545
            (ctxt->sax->unparsedEntityDecl != NULL))
5546
545
      ctxt->sax->unparsedEntityDecl(ctxt->userData, name,
5547
545
            literal, URI, ndata);
5548
12.2k
    } else {
5549
12.2k
        if ((ctxt->sax != NULL) &&
5550
12.2k
            (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5551
11.2k
      ctxt->sax->entityDecl(ctxt->userData, name,
5552
11.2k
            XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5553
11.2k
            literal, URI, NULL);
5554
        /*
5555
         * For expat compatibility in SAX mode.
5556
         * assuming the entity replacement was asked for
5557
         */
5558
12.2k
        if ((ctxt->replaceEntities != 0) &&
5559
4.06k
      ((ctxt->myDoc == NULL) ||
5560
4.05k
      (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) {
5561
223
      if (ctxt->myDoc == NULL) {
5562
11
          ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5563
11
          if (ctxt->myDoc == NULL) {
5564
1
              xmlErrMemory(ctxt);
5565
1
        goto done;
5566
1
          }
5567
10
          ctxt->myDoc->properties = XML_DOC_INTERNAL;
5568
10
      }
5569
5570
222
      if (ctxt->myDoc->intSubset == NULL) {
5571
10
          ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5572
10
            BAD_CAST "fake", NULL, NULL);
5573
10
                            if (ctxt->myDoc->intSubset == NULL) {
5574
1
                                xmlErrMemory(ctxt);
5575
1
                                goto done;
5576
1
                            }
5577
10
                        }
5578
221
      xmlSAX2EntityDecl(ctxt, name,
5579
221
                  XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5580
221
                  literal, URI, NULL);
5581
221
        }
5582
12.2k
    }
5583
13.1k
      }
5584
65.5k
  }
5585
124k
  SKIP_BLANKS_PE;
5586
124k
  if (RAW != '>') {
5587
12.0k
      xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED,
5588
12.0k
              "xmlParseEntityDecl: entity %s not terminated\n", name);
5589
112k
  } else {
5590
112k
#ifdef LIBXML_VALID_ENABLED
5591
112k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5592
19
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5593
19
                           "Entity declaration doesn't start and stop in"
5594
19
                                 " the same entity\n",
5595
19
                                 NULL, NULL);
5596
19
      }
5597
112k
#endif
5598
112k
      NEXT;
5599
112k
  }
5600
124k
  if (orig != NULL) {
5601
      /*
5602
       * Ugly mechanism to save the raw entity value.
5603
       */
5604
100k
      xmlEntityPtr cur = NULL;
5605
5606
100k
      if (isParameter) {
5607
49.4k
          if ((ctxt->sax != NULL) &&
5608
49.4k
        (ctxt->sax->getParameterEntity != NULL))
5609
49.4k
        cur = ctxt->sax->getParameterEntity(ctxt->userData, name);
5610
50.9k
      } else {
5611
50.9k
          if ((ctxt->sax != NULL) &&
5612
50.9k
        (ctxt->sax->getEntity != NULL))
5613
50.9k
        cur = ctxt->sax->getEntity(ctxt->userData, name);
5614
50.9k
    if ((cur == NULL) && (ctxt->userData==ctxt)) {
5615
1.08k
        cur = xmlSAX2GetEntity(ctxt, name);
5616
1.08k
    }
5617
50.9k
      }
5618
100k
            if ((cur != NULL) && (cur->orig == NULL)) {
5619
12.5k
    cur->orig = orig;
5620
12.5k
                orig = NULL;
5621
12.5k
      }
5622
100k
  }
5623
5624
124k
done:
5625
124k
  if (value != NULL) xmlFree(value);
5626
124k
  if (URI != NULL) xmlFree(URI);
5627
124k
  if (literal != NULL) xmlFree(literal);
5628
124k
        if (orig != NULL) xmlFree(orig);
5629
124k
    }
5630
151k
}
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
93.2k
xmlParseDefaultDecl(xmlParserCtxt *ctxt, xmlChar **value) {
5663
93.2k
    int val;
5664
93.2k
    xmlChar *ret;
5665
5666
93.2k
    *value = NULL;
5667
93.2k
    if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) {
5668
3.44k
  SKIP(9);
5669
3.44k
  return(XML_ATTRIBUTE_REQUIRED);
5670
3.44k
    }
5671
89.7k
    if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) {
5672
22.4k
  SKIP(8);
5673
22.4k
  return(XML_ATTRIBUTE_IMPLIED);
5674
22.4k
    }
5675
67.3k
    val = XML_ATTRIBUTE_NONE;
5676
67.3k
    if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) {
5677
11.9k
  SKIP(6);
5678
11.9k
  val = XML_ATTRIBUTE_FIXED;
5679
11.9k
  if (SKIP_BLANKS_PE == 0) {
5680
380
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5681
380
         "Space required after '#FIXED'\n");
5682
380
  }
5683
11.9k
    }
5684
67.3k
    ret = xmlParseAttValue(ctxt);
5685
67.3k
    if (ret == NULL) {
5686
38.2k
  xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo,
5687
38.2k
           "Attribute default value declaration error\n");
5688
38.2k
    } else
5689
29.1k
        *value = ret;
5690
67.3k
    return(val);
5691
89.7k
}
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
3.16k
xmlParseNotationType(xmlParserCtxt *ctxt) {
5712
3.16k
    const xmlChar *name;
5713
3.16k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5714
5715
3.16k
    if (RAW != '(') {
5716
444
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5717
444
  return(NULL);
5718
444
    }
5719
3.60k
    do {
5720
3.60k
        NEXT;
5721
3.60k
  SKIP_BLANKS_PE;
5722
3.60k
        name = xmlParseName(ctxt);
5723
3.60k
  if (name == NULL) {
5724
317
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5725
317
         "Name expected in NOTATION declaration\n");
5726
317
            xmlFreeEnumeration(ret);
5727
317
      return(NULL);
5728
317
  }
5729
3.29k
        tmp = NULL;
5730
3.29k
#ifdef LIBXML_VALID_ENABLED
5731
3.29k
        if (ctxt->validate) {
5732
1.77k
            tmp = ret;
5733
5.32k
            while (tmp != NULL) {
5734
3.89k
                if (xmlStrEqual(name, tmp->name)) {
5735
348
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5736
348
              "standalone: attribute notation value token %s duplicated\n",
5737
348
                                     name, NULL);
5738
348
                    if (!xmlDictOwns(ctxt->dict, name))
5739
0
                        xmlFree((xmlChar *) name);
5740
348
                    break;
5741
348
                }
5742
3.55k
                tmp = tmp->next;
5743
3.55k
            }
5744
1.77k
        }
5745
3.29k
#endif /* LIBXML_VALID_ENABLED */
5746
3.29k
  if (tmp == NULL) {
5747
2.94k
      cur = xmlCreateEnumeration(name);
5748
2.94k
      if (cur == NULL) {
5749
1
                xmlErrMemory(ctxt);
5750
1
                xmlFreeEnumeration(ret);
5751
1
                return(NULL);
5752
1
            }
5753
2.94k
      if (last == NULL) ret = last = cur;
5754
536
      else {
5755
536
    last->next = cur;
5756
536
    last = cur;
5757
536
      }
5758
2.94k
  }
5759
3.28k
  SKIP_BLANKS_PE;
5760
3.28k
    } while (RAW == '|');
5761
2.40k
    if (RAW != ')') {
5762
513
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5763
513
        xmlFreeEnumeration(ret);
5764
513
  return(NULL);
5765
513
    }
5766
1.89k
    NEXT;
5767
1.89k
    return(ret);
5768
2.40k
}
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
17.9k
xmlParseEnumerationType(xmlParserCtxt *ctxt) {
5787
17.9k
    xmlChar *name;
5788
17.9k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5789
5790
17.9k
    if (RAW != '(') {
5791
3.27k
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL);
5792
3.27k
  return(NULL);
5793
3.27k
    }
5794
22.9k
    do {
5795
22.9k
        NEXT;
5796
22.9k
  SKIP_BLANKS_PE;
5797
22.9k
        name = xmlParseNmtoken(ctxt);
5798
22.9k
  if (name == NULL) {
5799
319
      xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL);
5800
319
      return(ret);
5801
319
  }
5802
22.6k
        tmp = NULL;
5803
22.6k
#ifdef LIBXML_VALID_ENABLED
5804
22.6k
        if (ctxt->validate) {
5805
15.1k
            tmp = ret;
5806
38.2k
            while (tmp != NULL) {
5807
23.4k
                if (xmlStrEqual(name, tmp->name)) {
5808
332
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5809
332
              "standalone: attribute enumeration value token %s duplicated\n",
5810
332
                                     name, NULL);
5811
332
                    if (!xmlDictOwns(ctxt->dict, name))
5812
332
                        xmlFree(name);
5813
332
                    break;
5814
332
                }
5815
23.0k
                tmp = tmp->next;
5816
23.0k
            }
5817
15.1k
        }
5818
22.6k
#endif /* LIBXML_VALID_ENABLED */
5819
22.6k
  if (tmp == NULL) {
5820
22.3k
      cur = xmlCreateEnumeration(name);
5821
22.3k
      if (!xmlDictOwns(ctxt->dict, name))
5822
22.3k
    xmlFree(name);
5823
22.3k
      if (cur == NULL) {
5824
3
                xmlErrMemory(ctxt);
5825
3
                xmlFreeEnumeration(ret);
5826
3
                return(NULL);
5827
3
            }
5828
22.3k
      if (last == NULL) ret = last = cur;
5829
7.96k
      else {
5830
7.96k
    last->next = cur;
5831
7.96k
    last = cur;
5832
7.96k
      }
5833
22.3k
  }
5834
22.6k
  SKIP_BLANKS_PE;
5835
22.6k
    } while (RAW == '|');
5836
14.3k
    if (RAW != ')') {
5837
793
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL);
5838
793
  return(ret);
5839
793
    }
5840
13.5k
    NEXT;
5841
13.5k
    return(ret);
5842
14.3k
}
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
21.3k
xmlParseEnumeratedType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5860
21.3k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5861
3.38k
  SKIP(8);
5862
3.38k
  if (SKIP_BLANKS_PE == 0) {
5863
223
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5864
223
         "Space required after 'NOTATION'\n");
5865
223
      return(0);
5866
223
  }
5867
3.16k
  *tree = xmlParseNotationType(ctxt);
5868
3.16k
  if (*tree == NULL) return(0);
5869
1.89k
  return(XML_ATTRIBUTE_NOTATION);
5870
3.16k
    }
5871
17.9k
    *tree = xmlParseEnumerationType(ctxt);
5872
17.9k
    if (*tree == NULL) return(0);
5873
14.3k
    return(XML_ATTRIBUTE_ENUMERATION);
5874
17.9k
}
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
101k
xmlParseAttributeType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5923
101k
    if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) {
5924
16.1k
  SKIP(5);
5925
16.1k
  return(XML_ATTRIBUTE_CDATA);
5926
85.0k
     } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) {
5927
2.03k
  SKIP(6);
5928
2.03k
  return(XML_ATTRIBUTE_IDREFS);
5929
82.9k
     } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) {
5930
1.62k
  SKIP(5);
5931
1.62k
  return(XML_ATTRIBUTE_IDREF);
5932
81.3k
     } else if ((RAW == 'I') && (NXT(1) == 'D')) {
5933
11.5k
        SKIP(2);
5934
11.5k
  return(XML_ATTRIBUTE_ID);
5935
69.7k
     } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5936
5.06k
  SKIP(6);
5937
5.06k
  return(XML_ATTRIBUTE_ENTITY);
5938
64.7k
     } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) {
5939
1.53k
  SKIP(8);
5940
1.53k
  return(XML_ATTRIBUTE_ENTITIES);
5941
63.1k
     } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) {
5942
34.3k
  SKIP(8);
5943
34.3k
  return(XML_ATTRIBUTE_NMTOKENS);
5944
34.3k
     } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) {
5945
7.50k
  SKIP(7);
5946
7.50k
  return(XML_ATTRIBUTE_NMTOKEN);
5947
7.50k
     }
5948
21.3k
     return(xmlParseEnumeratedType(ctxt, tree));
5949
101k
}
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
87.3k
xmlParseAttributeListDecl(xmlParserCtxt *ctxt) {
5963
87.3k
    const xmlChar *elemName;
5964
87.3k
    const xmlChar *attrName;
5965
87.3k
    xmlEnumerationPtr tree;
5966
5967
87.3k
    if ((CUR != '<') || (NXT(1) != '!'))
5968
0
        return;
5969
87.3k
    SKIP(2);
5970
5971
87.3k
    if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) {
5972
87.3k
#ifdef LIBXML_VALID_ENABLED
5973
87.3k
  int oldInputNr = ctxt->inputNr;
5974
87.3k
#endif
5975
5976
87.3k
  SKIP(7);
5977
87.3k
  if (SKIP_BLANKS_PE == 0) {
5978
1.90k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5979
1.90k
                     "Space required after '<!ATTLIST'\n");
5980
1.90k
  }
5981
87.3k
        elemName = xmlParseName(ctxt);
5982
87.3k
  if (elemName == NULL) {
5983
1.47k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5984
1.47k
         "ATTLIST: no name for Element\n");
5985
1.47k
      return;
5986
1.47k
  }
5987
85.8k
  SKIP_BLANKS_PE;
5988
85.8k
  GROW;
5989
141k
  while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) {
5990
112k
      int type;
5991
112k
      int def;
5992
112k
      xmlChar *defaultValue = NULL;
5993
5994
112k
      GROW;
5995
112k
            tree = NULL;
5996
112k
      attrName = xmlParseName(ctxt);
5997
112k
      if (attrName == NULL) {
5998
5.63k
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5999
5.63k
             "ATTLIST: no name for Attribute\n");
6000
5.63k
    break;
6001
5.63k
      }
6002
106k
      GROW;
6003
106k
      if (SKIP_BLANKS_PE == 0) {
6004
5.43k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6005
5.43k
            "Space required after the attribute name\n");
6006
5.43k
    break;
6007
5.43k
      }
6008
6009
101k
      type = xmlParseAttributeType(ctxt, &tree);
6010
101k
      if (type <= 0) {
6011
5.07k
          break;
6012
5.07k
      }
6013
6014
96.0k
      GROW;
6015
96.0k
      if (SKIP_BLANKS_PE == 0) {
6016
2.86k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6017
2.86k
             "Space required after the attribute type\n");
6018
2.86k
          if (tree != NULL)
6019
859
        xmlFreeEnumeration(tree);
6020
2.86k
    break;
6021
2.86k
      }
6022
6023
93.2k
      def = xmlParseDefaultDecl(ctxt, &defaultValue);
6024
93.2k
      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
93.2k
      if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL))
6032
26.6k
          xmlAttrNormalizeSpace(defaultValue, defaultValue);
6033
6034
93.2k
      GROW;
6035
93.2k
            if (RAW != '>') {
6036
72.0k
    if (SKIP_BLANKS_PE == 0) {
6037
38.0k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6038
38.0k
      "Space required after the attribute default value\n");
6039
38.0k
        if (defaultValue != NULL)
6040
403
      xmlFree(defaultValue);
6041
38.0k
        if (tree != NULL)
6042
1.99k
      xmlFreeEnumeration(tree);
6043
38.0k
        break;
6044
38.0k
    }
6045
72.0k
      }
6046
55.1k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6047
52.3k
    (ctxt->sax->attributeDecl != NULL))
6048
52.3k
    ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName,
6049
52.3k
                          type, def, defaultValue, tree);
6050
2.79k
      else if (tree != NULL)
6051
1.03k
    xmlFreeEnumeration(tree);
6052
6053
55.1k
      if ((ctxt->sax2) && (defaultValue != NULL) &&
6054
22.0k
          (def != XML_ATTRIBUTE_IMPLIED) &&
6055
22.0k
    (def != XML_ATTRIBUTE_REQUIRED)) {
6056
22.0k
    xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue);
6057
22.0k
      }
6058
55.1k
      if (ctxt->sax2) {
6059
42.5k
    xmlAddSpecialAttr(ctxt, elemName, attrName, type);
6060
42.5k
      }
6061
55.1k
      if (defaultValue != NULL)
6062
28.7k
          xmlFree(defaultValue);
6063
55.1k
      GROW;
6064
55.1k
  }
6065
85.8k
  if (RAW == '>') {
6066
29.9k
#ifdef LIBXML_VALID_ENABLED
6067
29.9k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6068
294
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6069
294
                                 "Attribute list declaration doesn't start and"
6070
294
                                 " stop in the same entity\n",
6071
294
                                 NULL, NULL);
6072
294
      }
6073
29.9k
#endif
6074
29.9k
      NEXT;
6075
29.9k
  }
6076
85.8k
    }
6077
87.3k
}
6078
6079
/**
6080
 * Handle PEs and check that we don't pop the entity that started
6081
 * a balanced group.
6082
 *
6083
 * @param ctxt  parser context
6084
 * @param openInputNr  input nr of the entity with opening '('
6085
 */
6086
static void
6087
441k
xmlSkipBlankCharsPEBalanced(xmlParserCtxt *ctxt, int openInputNr) {
6088
441k
    SKIP_BLANKS;
6089
441k
    GROW;
6090
6091
441k
    (void) openInputNr;
6092
6093
441k
    if (!PARSER_EXTERNAL(ctxt) && !PARSER_IN_PE(ctxt))
6094
354k
        return;
6095
6096
93.5k
    while (!PARSER_STOPPED(ctxt)) {
6097
93.1k
        if (ctxt->input->cur >= ctxt->input->end) {
6098
2.63k
#ifdef LIBXML_VALID_ENABLED
6099
2.63k
            if ((ctxt->validate) && (ctxt->inputNr <= openInputNr)) {
6100
966
                xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6101
966
                                 "Element content declaration doesn't start "
6102
966
                                 "and stop in the same entity\n",
6103
966
                                 NULL, NULL);
6104
966
            }
6105
2.63k
#endif
6106
2.63k
            if (PARSER_IN_PE(ctxt))
6107
2.60k
                xmlPopPE(ctxt);
6108
32
            else
6109
32
                break;
6110
90.5k
        } else if (RAW == '%') {
6111
3.89k
            xmlParsePERefInternal(ctxt, 0);
6112
86.6k
        } else {
6113
86.6k
            break;
6114
86.6k
        }
6115
6116
6.50k
        SKIP_BLANKS;
6117
6.50k
        GROW;
6118
6.50k
    }
6119
87.0k
}
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
7.39k
xmlParseElementMixedContentDecl(xmlParserCtxt *ctxt, int openInputNr) {
6143
7.39k
    xmlElementContentPtr ret = NULL, cur = NULL, n;
6144
7.39k
    const xmlChar *elem = NULL;
6145
6146
7.39k
    GROW;
6147
7.39k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6148
7.39k
  SKIP(7);
6149
7.39k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6150
7.39k
  if (RAW == ')') {
6151
3.70k
#ifdef LIBXML_VALID_ENABLED
6152
3.70k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6153
10
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6154
10
                                 "Element content declaration doesn't start "
6155
10
                                 "and stop in the same entity\n",
6156
10
                                 NULL, NULL);
6157
10
      }
6158
3.70k
#endif
6159
3.70k
      NEXT;
6160
3.70k
      ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6161
3.70k
      if (ret == NULL)
6162
2
                goto mem_error;
6163
3.70k
      if (RAW == '*') {
6164
317
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6165
317
    NEXT;
6166
317
      }
6167
3.70k
      return(ret);
6168
3.70k
  }
6169
3.68k
  if ((RAW == '(') || (RAW == '|')) {
6170
2.95k
      ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6171
2.95k
      if (ret == NULL)
6172
2
                goto mem_error;
6173
2.95k
  }
6174
24.5k
  while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) {
6175
21.1k
      NEXT;
6176
21.1k
            n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6177
21.1k
            if (n == NULL)
6178
1
                goto mem_error;
6179
21.1k
      if (elem == NULL) {
6180
2.94k
    n->c1 = cur;
6181
2.94k
    if (cur != NULL)
6182
2.94k
        cur->parent = n;
6183
2.94k
    ret = cur = n;
6184
18.1k
      } else {
6185
18.1k
          cur->c2 = n;
6186
18.1k
    n->parent = cur;
6187
18.1k
    n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6188
18.1k
                if (n->c1 == NULL)
6189
1
                    goto mem_error;
6190
18.1k
    n->c1->parent = n;
6191
18.1k
    cur = n;
6192
18.1k
      }
6193
21.1k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6194
21.1k
      elem = xmlParseName(ctxt);
6195
21.1k
      if (elem == NULL) {
6196
297
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6197
297
      "xmlParseElementMixedContentDecl : Name expected\n");
6198
297
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6199
297
    return(NULL);
6200
297
      }
6201
20.8k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6202
20.8k
  }
6203
3.38k
  if ((RAW == ')') && (NXT(1) == '*')) {
6204
2.40k
      if (elem != NULL) {
6205
2.40k
    cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem,
6206
2.40k
                                   XML_ELEMENT_CONTENT_ELEMENT);
6207
2.40k
    if (cur->c2 == NULL)
6208
1
                    goto mem_error;
6209
2.40k
    cur->c2->parent = cur;
6210
2.40k
            }
6211
2.40k
            if (ret != NULL)
6212
2.40k
                ret->ocur = XML_ELEMENT_CONTENT_MULT;
6213
2.40k
#ifdef LIBXML_VALID_ENABLED
6214
2.40k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6215
3
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6216
3
                                 "Element content declaration doesn't start "
6217
3
                                 "and stop in the same entity\n",
6218
3
                                 NULL, NULL);
6219
3
      }
6220
2.40k
#endif
6221
2.40k
      SKIP(2);
6222
2.40k
  } else {
6223
974
      xmlFreeDocElementContent(ctxt->myDoc, ret);
6224
974
      xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL);
6225
974
      return(NULL);
6226
974
  }
6227
6228
3.38k
    } else {
6229
0
  xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL);
6230
0
    }
6231
2.40k
    return(ret);
6232
6233
7
mem_error:
6234
7
    xmlErrMemory(ctxt);
6235
7
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6236
7
    return(NULL);
6237
7.39k
}
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
67.1k
                                       int depth) {
6272
67.1k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
6273
67.1k
    xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL;
6274
67.1k
    const xmlChar *elem;
6275
67.1k
    xmlChar type = 0;
6276
6277
67.1k
    if (depth > maxDepth) {
6278
2
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
6279
2
                "xmlParseElementChildrenContentDecl : depth %d too deep, "
6280
2
                "use XML_PARSE_HUGE\n", depth);
6281
2
  return(NULL);
6282
2
    }
6283
67.1k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6284
67.1k
    if (RAW == '(') {
6285
28.1k
        int newInputNr = ctxt->inputNr;
6286
6287
        /* Recurse on first child */
6288
28.1k
  NEXT;
6289
28.1k
        cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6290
28.1k
                                                           depth + 1);
6291
28.1k
        if (cur == NULL)
6292
27.0k
            return(NULL);
6293
38.9k
    } else {
6294
38.9k
  elem = xmlParseName(ctxt);
6295
38.9k
  if (elem == NULL) {
6296
402
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6297
402
      return(NULL);
6298
402
  }
6299
38.5k
        cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6300
38.5k
  if (cur == NULL) {
6301
4
      xmlErrMemory(ctxt);
6302
4
      return(NULL);
6303
4
  }
6304
38.5k
  GROW;
6305
38.5k
  if (RAW == '?') {
6306
1.29k
      cur->ocur = XML_ELEMENT_CONTENT_OPT;
6307
1.29k
      NEXT;
6308
37.2k
  } else if (RAW == '*') {
6309
1.36k
      cur->ocur = XML_ELEMENT_CONTENT_MULT;
6310
1.36k
      NEXT;
6311
35.8k
  } else if (RAW == '+') {
6312
456
      cur->ocur = XML_ELEMENT_CONTENT_PLUS;
6313
456
      NEXT;
6314
35.4k
  } else {
6315
35.4k
      cur->ocur = XML_ELEMENT_CONTENT_ONCE;
6316
35.4k
  }
6317
38.5k
  GROW;
6318
38.5k
    }
6319
174k
    while (!PARSER_STOPPED(ctxt)) {
6320
173k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6321
173k
        if (RAW == ')')
6322
36.5k
            break;
6323
        /*
6324
   * Each loop we parse one separator and one element.
6325
   */
6326
136k
        if (RAW == ',') {
6327
75.3k
      if (type == 0) type = CUR;
6328
6329
      /*
6330
       * Detect "Name | Name , Name" error
6331
       */
6332
66.7k
      else if (type != CUR) {
6333
2
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6334
2
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6335
2
                      type);
6336
2
    if ((last != NULL) && (last != ret))
6337
2
        xmlFreeDocElementContent(ctxt->myDoc, last);
6338
2
    if (ret != NULL)
6339
2
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6340
2
    return(NULL);
6341
2
      }
6342
75.3k
      NEXT;
6343
6344
75.3k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ);
6345
75.3k
      if (op == NULL) {
6346
3
                xmlErrMemory(ctxt);
6347
3
    if ((last != NULL) && (last != ret))
6348
2
        xmlFreeDocElementContent(ctxt->myDoc, last);
6349
3
          xmlFreeDocElementContent(ctxt->myDoc, ret);
6350
3
    return(NULL);
6351
3
      }
6352
75.3k
      if (last == NULL) {
6353
8.62k
    op->c1 = ret;
6354
8.62k
    if (ret != NULL)
6355
8.62k
        ret->parent = op;
6356
8.62k
    ret = cur = op;
6357
66.7k
      } else {
6358
66.7k
          cur->c2 = op;
6359
66.7k
    if (op != NULL)
6360
66.7k
        op->parent = cur;
6361
66.7k
    op->c1 = last;
6362
66.7k
    if (last != NULL)
6363
66.7k
        last->parent = op;
6364
66.7k
    cur =op;
6365
66.7k
    last = NULL;
6366
66.7k
      }
6367
75.3k
  } else if (RAW == '|') {
6368
59.6k
      if (type == 0) type = CUR;
6369
6370
      /*
6371
       * Detect "Name , Name | Name" error
6372
       */
6373
37.9k
      else if (type != CUR) {
6374
1
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6375
1
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6376
1
          type);
6377
1
    if ((last != NULL) && (last != ret))
6378
1
        xmlFreeDocElementContent(ctxt->myDoc, last);
6379
1
    if (ret != NULL)
6380
1
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6381
1
    return(NULL);
6382
1
      }
6383
59.6k
      NEXT;
6384
6385
59.6k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6386
59.6k
      if (op == NULL) {
6387
2
                xmlErrMemory(ctxt);
6388
2
    if ((last != NULL) && (last != ret))
6389
1
        xmlFreeDocElementContent(ctxt->myDoc, last);
6390
2
    if (ret != NULL)
6391
2
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6392
2
    return(NULL);
6393
2
      }
6394
59.6k
      if (last == NULL) {
6395
21.7k
    op->c1 = ret;
6396
21.7k
    if (ret != NULL)
6397
21.7k
        ret->parent = op;
6398
21.7k
    ret = cur = op;
6399
37.9k
      } else {
6400
37.9k
          cur->c2 = op;
6401
37.9k
    if (op != NULL)
6402
37.9k
        op->parent = cur;
6403
37.9k
    op->c1 = last;
6404
37.9k
    if (last != NULL)
6405
37.9k
        last->parent = op;
6406
37.9k
    cur =op;
6407
37.9k
    last = NULL;
6408
37.9k
      }
6409
59.6k
  } else {
6410
1.66k
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL);
6411
1.66k
      if ((last != NULL) && (last != ret))
6412
833
          xmlFreeDocElementContent(ctxt->myDoc, last);
6413
1.66k
      if (ret != NULL)
6414
1.66k
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6415
1.66k
      return(NULL);
6416
1.66k
  }
6417
134k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6418
134k
        if (RAW == '(') {
6419
29.3k
            int newInputNr = ctxt->inputNr;
6420
6421
      /* Recurse on second child */
6422
29.3k
      NEXT;
6423
29.3k
      last = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6424
29.3k
                                                          depth + 1);
6425
29.3k
            if (last == NULL) {
6426
286
    if (ret != NULL)
6427
286
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6428
286
    return(NULL);
6429
286
            }
6430
105k
  } else {
6431
105k
      elem = xmlParseName(ctxt);
6432
105k
      if (elem == NULL) {
6433
329
    xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6434
329
    if (ret != NULL)
6435
329
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6436
329
    return(NULL);
6437
329
      }
6438
105k
      last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6439
105k
      if (last == NULL) {
6440
1
                xmlErrMemory(ctxt);
6441
1
    if (ret != NULL)
6442
1
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6443
1
    return(NULL);
6444
1
      }
6445
105k
      if (RAW == '?') {
6446
16.4k
    last->ocur = XML_ELEMENT_CONTENT_OPT;
6447
16.4k
    NEXT;
6448
88.8k
      } else if (RAW == '*') {
6449
8.16k
    last->ocur = XML_ELEMENT_CONTENT_MULT;
6450
8.16k
    NEXT;
6451
80.7k
      } else if (RAW == '+') {
6452
7.77k
    last->ocur = XML_ELEMENT_CONTENT_PLUS;
6453
7.77k
    NEXT;
6454
72.9k
      } else {
6455
72.9k
    last->ocur = XML_ELEMENT_CONTENT_ONCE;
6456
72.9k
      }
6457
105k
  }
6458
134k
    }
6459
37.3k
    if ((cur != NULL) && (last != NULL)) {
6460
28.8k
        cur->c2 = last;
6461
28.8k
  if (last != NULL)
6462
28.8k
      last->parent = cur;
6463
28.8k
    }
6464
37.3k
#ifdef LIBXML_VALID_ENABLED
6465
37.3k
    if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6466
121
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6467
121
                         "Element content declaration doesn't start "
6468
121
                         "and stop in the same entity\n",
6469
121
                         NULL, NULL);
6470
121
    }
6471
37.3k
#endif
6472
37.3k
    NEXT;
6473
37.3k
    if (RAW == '?') {
6474
6.57k
  if (ret != NULL) {
6475
6.57k
      if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) ||
6476
6.48k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6477
109
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6478
6.46k
      else
6479
6.46k
          ret->ocur = XML_ELEMENT_CONTENT_OPT;
6480
6.57k
  }
6481
6.57k
  NEXT;
6482
30.7k
    } else if (RAW == '*') {
6483
16.4k
  if (ret != NULL) {
6484
16.4k
      ret->ocur = XML_ELEMENT_CONTENT_MULT;
6485
16.4k
      cur = ret;
6486
      /*
6487
       * Some normalization:
6488
       * (a | b* | c?)* == (a | b | c)*
6489
       */
6490
56.7k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6491
40.2k
    if ((cur->c1 != NULL) &&
6492
40.2k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6493
39.5k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT)))
6494
2.98k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6495
40.2k
    if ((cur->c2 != NULL) &&
6496
40.2k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6497
39.9k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT)))
6498
372
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6499
40.2k
    cur = cur->c2;
6500
40.2k
      }
6501
16.4k
  }
6502
16.4k
  NEXT;
6503
16.4k
    } else if (RAW == '+') {
6504
5.97k
  if (ret != NULL) {
6505
5.97k
      int found = 0;
6506
6507
5.97k
      if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) ||
6508
5.67k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6509
395
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6510
5.58k
      else
6511
5.58k
          ret->ocur = XML_ELEMENT_CONTENT_PLUS;
6512
      /*
6513
       * Some normalization:
6514
       * (a | b*)+ == (a | b)*
6515
       * (a | b?)+ == (a | b)*
6516
       */
6517
11.1k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6518
5.18k
    if ((cur->c1 != NULL) &&
6519
5.18k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6520
4.86k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) {
6521
613
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6522
613
        found = 1;
6523
613
    }
6524
5.18k
    if ((cur->c2 != NULL) &&
6525
5.18k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6526
4.95k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) {
6527
251
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6528
251
        found = 1;
6529
251
    }
6530
5.18k
    cur = cur->c2;
6531
5.18k
      }
6532
5.97k
      if (found)
6533
547
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6534
5.97k
  }
6535
5.97k
  NEXT;
6536
5.97k
    }
6537
37.3k
    return(ret);
6538
39.6k
}
6539
6540
/**
6541
 * Parse the declaration for a Mixed Element content
6542
 * The leading '(' and spaces have been skipped in #xmlParseElementContentDecl
6543
 *
6544
 * @deprecated Internal function, don't use.
6545
 *
6546
 *     [47] children ::= (choice | seq) ('?' | '*' | '+')?
6547
 *
6548
 *     [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')?
6549
 *
6550
 *     [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')'
6551
 *
6552
 *     [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')'
6553
 *
6554
 * [ VC: Proper Group/PE Nesting ] applies to [49] and [50]
6555
 * TODO Parameter-entity replacement text must be properly nested
6556
 *  with parenthesized groups. That is to say, if either of the
6557
 *  opening or closing parentheses in a choice, seq, or Mixed
6558
 *  construct is contained in the replacement text for a parameter
6559
 *  entity, both must be contained in the same replacement text. For
6560
 *  interoperability, if a parameter-entity reference appears in a
6561
 *  choice, seq, or Mixed construct, its replacement text should not
6562
 *  be empty, and neither the first nor last non-blank character of
6563
 *  the replacement text should be a connector (| or ,).
6564
 *
6565
 * @param ctxt  an XML parser context
6566
 * @param inputchk  the input used for the current entity, needed for boundary checks
6567
 * @returns the tree of xmlElementContent describing the element
6568
 *          hierarchy.
6569
 */
6570
xmlElementContent *
6571
0
xmlParseElementChildrenContentDecl(xmlParserCtxt *ctxt, int inputchk) {
6572
    /* stub left for API/ABI compat */
6573
0
    return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1));
6574
0
}
6575
6576
/**
6577
 * Parse the declaration for an Element content either Mixed or Children,
6578
 * the cases EMPTY and ANY are handled directly in #xmlParseElementDecl
6579
 *
6580
 * @deprecated Internal function, don't use.
6581
 *
6582
 *     [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children
6583
 *
6584
 * @param ctxt  an XML parser context
6585
 * @param name  the name of the element being defined.
6586
 * @param result  the Element Content pointer will be stored here if any
6587
 * @returns an xmlElementTypeVal value or -1 on error
6588
 */
6589
6590
int
6591
xmlParseElementContentDecl(xmlParserCtxt *ctxt, const xmlChar *name,
6592
16.9k
                           xmlElementContent **result) {
6593
6594
16.9k
    xmlElementContentPtr tree = NULL;
6595
16.9k
    int openInputNr = ctxt->inputNr;
6596
16.9k
    int res;
6597
6598
16.9k
    *result = NULL;
6599
6600
16.9k
    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
16.9k
    NEXT;
6606
16.9k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6607
16.9k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6608
7.39k
        tree = xmlParseElementMixedContentDecl(ctxt, openInputNr);
6609
7.39k
  res = XML_ELEMENT_TYPE_MIXED;
6610
9.56k
    } else {
6611
9.56k
        tree = xmlParseElementChildrenContentDeclPriv(ctxt, openInputNr, 1);
6612
9.56k
  res = XML_ELEMENT_TYPE_ELEMENT;
6613
9.56k
    }
6614
16.9k
    if (tree == NULL)
6615
3.68k
        return(-1);
6616
13.2k
    SKIP_BLANKS_PE;
6617
13.2k
    *result = tree;
6618
13.2k
    return(res);
6619
16.9k
}
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
26.2k
xmlParseElementDecl(xmlParserCtxt *ctxt) {
6636
26.2k
    const xmlChar *name;
6637
26.2k
    int ret = -1;
6638
26.2k
    xmlElementContentPtr content  = NULL;
6639
6640
26.2k
    if ((CUR != '<') || (NXT(1) != '!'))
6641
0
        return(ret);
6642
26.2k
    SKIP(2);
6643
6644
    /* GROW; done in the caller */
6645
26.2k
    if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) {
6646
26.1k
#ifdef LIBXML_VALID_ENABLED
6647
26.1k
  int oldInputNr = ctxt->inputNr;
6648
26.1k
#endif
6649
6650
26.1k
  SKIP(7);
6651
26.1k
  if (SKIP_BLANKS_PE == 0) {
6652
316
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6653
316
               "Space required after 'ELEMENT'\n");
6654
316
      return(-1);
6655
316
  }
6656
25.8k
        name = xmlParseName(ctxt);
6657
25.8k
  if (name == NULL) {
6658
502
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6659
502
         "xmlParseElementDecl: no name for Element\n");
6660
502
      return(-1);
6661
502
  }
6662
25.3k
  if (SKIP_BLANKS_PE == 0) {
6663
1.29k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6664
1.29k
         "Space required after the element name\n");
6665
1.29k
  }
6666
25.3k
  if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) {
6667
5.34k
      SKIP(5);
6668
      /*
6669
       * Element must always be empty.
6670
       */
6671
5.34k
      ret = XML_ELEMENT_TYPE_EMPTY;
6672
20.0k
  } else if ((RAW == 'A') && (NXT(1) == 'N') &&
6673
2.16k
             (NXT(2) == 'Y')) {
6674
2.16k
      SKIP(3);
6675
      /*
6676
       * Element is a generic container.
6677
       */
6678
2.16k
      ret = XML_ELEMENT_TYPE_ANY;
6679
17.8k
  } else if (RAW == '(') {
6680
16.9k
      ret = xmlParseElementContentDecl(ctxt, name, &content);
6681
16.9k
            if (ret <= 0)
6682
3.68k
                return(-1);
6683
16.9k
  } else {
6684
      /*
6685
       * [ WFC: PEs in Internal Subset ] error handling.
6686
       */
6687
887
            xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6688
887
                  "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n");
6689
887
      return(-1);
6690
887
  }
6691
6692
20.7k
  SKIP_BLANKS_PE;
6693
6694
20.7k
  if (RAW != '>') {
6695
1.01k
      xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
6696
1.01k
      if (content != NULL) {
6697
623
    xmlFreeDocElementContent(ctxt->myDoc, content);
6698
623
      }
6699
19.7k
  } else {
6700
19.7k
#ifdef LIBXML_VALID_ENABLED
6701
19.7k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6702
77
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6703
77
                                 "Element declaration doesn't start and stop in"
6704
77
                                 " the same entity\n",
6705
77
                                 NULL, NULL);
6706
77
      }
6707
19.7k
#endif
6708
6709
19.7k
      NEXT;
6710
19.7k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6711
17.7k
    (ctxt->sax->elementDecl != NULL)) {
6712
17.7k
    if (content != NULL)
6713
11.4k
        content->parent = NULL;
6714
17.7k
          ctxt->sax->elementDecl(ctxt->userData, name, ret,
6715
17.7k
                           content);
6716
17.7k
    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
3.26k
        xmlFreeDocElementContent(ctxt->myDoc, content);
6724
3.26k
    }
6725
17.7k
      } else if (content != NULL) {
6726
1.20k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6727
1.20k
      }
6728
19.7k
  }
6729
20.7k
    }
6730
20.8k
    return(ret);
6731
26.2k
}
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
4.93k
xmlParseConditionalSections(xmlParserCtxtPtr ctxt) {
6747
4.93k
    size_t depth = 0;
6748
4.93k
    int isFreshPE = 0;
6749
4.93k
    int oldInputNr = ctxt->inputNr;
6750
4.93k
    int declInputNr = ctxt->inputNr;
6751
6752
10.1k
    while (!PARSER_STOPPED(ctxt)) {
6753
10.1k
        if (ctxt->input->cur >= ctxt->input->end) {
6754
678
            if (ctxt->inputNr <= oldInputNr) {
6755
319
                xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6756
319
                return;
6757
319
            }
6758
6759
359
            xmlPopPE(ctxt);
6760
359
            declInputNr = ctxt->inputNr;
6761
9.50k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6762
5.20k
            SKIP(3);
6763
5.20k
            SKIP_BLANKS_PE;
6764
6765
5.20k
            isFreshPE = 0;
6766
6767
5.20k
            if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) {
6768
2.04k
                SKIP(7);
6769
2.04k
                SKIP_BLANKS_PE;
6770
2.04k
                if (RAW != '[') {
6771
224
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6772
224
                    return;
6773
224
                }
6774
1.81k
#ifdef LIBXML_VALID_ENABLED
6775
1.81k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6776
31
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6777
31
                                     "All markup of the conditional section is"
6778
31
                                     " not in the same entity\n",
6779
31
                                     NULL, NULL);
6780
31
                }
6781
1.81k
#endif
6782
1.81k
                NEXT;
6783
6784
1.81k
                depth++;
6785
3.16k
            } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) {
6786
2.56k
                size_t ignoreDepth = 0;
6787
6788
2.56k
                SKIP(6);
6789
2.56k
                SKIP_BLANKS_PE;
6790
2.56k
                if (RAW != '[') {
6791
527
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6792
527
                    return;
6793
527
                }
6794
2.04k
#ifdef LIBXML_VALID_ENABLED
6795
2.04k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6796
6
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6797
6
                                     "All markup of the conditional section is"
6798
6
                                     " not in the same entity\n",
6799
6
                                     NULL, NULL);
6800
6
                }
6801
2.04k
#endif
6802
2.04k
                NEXT;
6803
6804
63.5k
                while (PARSER_STOPPED(ctxt) == 0) {
6805
63.5k
                    if (RAW == 0) {
6806
1.27k
                        xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL);
6807
1.27k
                        return;
6808
1.27k
                    }
6809
62.2k
                    if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6810
892
                        SKIP(3);
6811
892
                        ignoreDepth++;
6812
                        /* Check for integer overflow */
6813
892
                        if (ignoreDepth == 0) {
6814
0
                            xmlErrMemory(ctxt);
6815
0
                            return;
6816
0
                        }
6817
61.3k
                    } else if ((RAW == ']') && (NXT(1) == ']') &&
6818
1.85k
                               (NXT(2) == '>')) {
6819
1.26k
                        SKIP(3);
6820
1.26k
                        if (ignoreDepth == 0)
6821
758
                            break;
6822
506
                        ignoreDepth--;
6823
60.1k
                    } else {
6824
60.1k
                        NEXT;
6825
60.1k
                    }
6826
62.2k
                }
6827
6828
761
#ifdef LIBXML_VALID_ENABLED
6829
761
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6830
6
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6831
6
                                     "All markup of the conditional section is"
6832
6
                                     " not in the same entity\n",
6833
6
                                     NULL, NULL);
6834
6
                }
6835
761
#endif
6836
761
            } else {
6837
600
                xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL);
6838
600
                return;
6839
600
            }
6840
5.20k
        } else if ((depth > 0) &&
6841
4.29k
                   (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) {
6842
1.19k
            if (isFreshPE) {
6843
1
                xmlFatalErrMsg(ctxt, XML_ERR_CONDSEC_INVALID,
6844
1
                               "Parameter entity must match "
6845
1
                               "extSubsetDecl\n");
6846
1
                return;
6847
1
            }
6848
6849
1.19k
            depth--;
6850
1.19k
#ifdef LIBXML_VALID_ENABLED
6851
1.19k
            if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6852
18
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6853
18
                                 "All markup of the conditional section is not"
6854
18
                                 " in the same entity\n",
6855
18
                                 NULL, NULL);
6856
18
            }
6857
1.19k
#endif
6858
1.19k
            SKIP(3);
6859
3.09k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
6860
1.73k
            isFreshPE = 0;
6861
1.73k
            xmlParseMarkupDecl(ctxt);
6862
1.73k
        } else if (RAW == '%') {
6863
1.31k
            xmlParsePERefInternal(ctxt, 1);
6864
1.31k
            if (ctxt->inputNr > declInputNr) {
6865
361
                isFreshPE = 1;
6866
361
                declInputNr = ctxt->inputNr;
6867
361
            }
6868
1.31k
        } else {
6869
47
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6870
47
            return;
6871
47
        }
6872
6873
7.18k
        if (depth == 0)
6874
1.93k
            break;
6875
6876
5.24k
        SKIP_BLANKS;
6877
5.24k
        SHRINK;
6878
5.24k
        GROW;
6879
5.24k
    }
6880
4.93k
}
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
648k
xmlParseMarkupDecl(xmlParserCtxt *ctxt) {
6907
648k
    GROW;
6908
648k
    if (CUR == '<') {
6909
648k
        if (NXT(1) == '!') {
6910
635k
      switch (NXT(2)) {
6911
177k
          case 'E':
6912
177k
        if (NXT(3) == 'L')
6913
26.2k
      xmlParseElementDecl(ctxt);
6914
151k
        else if (NXT(3) == 'N')
6915
151k
      xmlParseEntityDecl(ctxt);
6916
30
                    else
6917
30
                        SKIP(2);
6918
177k
        break;
6919
87.3k
          case 'A':
6920
87.3k
        xmlParseAttributeListDecl(ctxt);
6921
87.3k
        break;
6922
4.38k
          case 'N':
6923
4.38k
        xmlParseNotationDecl(ctxt);
6924
4.38k
        break;
6925
354k
          case '-':
6926
354k
        xmlParseComment(ctxt);
6927
354k
        break;
6928
11.0k
    default:
6929
11.0k
                    xmlFatalErr(ctxt,
6930
11.0k
                                ctxt->inSubset == 2 ?
6931
994
                                    XML_ERR_EXT_SUBSET_NOT_FINISHED :
6932
11.0k
                                    XML_ERR_INT_SUBSET_NOT_FINISHED,
6933
11.0k
                                NULL);
6934
11.0k
                    SKIP(2);
6935
11.0k
        break;
6936
635k
      }
6937
635k
  } else if (NXT(1) == '?') {
6938
13.3k
      xmlParsePI(ctxt);
6939
13.3k
  }
6940
648k
    }
6941
648k
}
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
147k
xmlParseTextDecl(xmlParserCtxt *ctxt) {
6954
147k
    xmlChar *version;
6955
6956
    /*
6957
     * We know that '<?xml' is here.
6958
     */
6959
147k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
6960
147k
  SKIP(5);
6961
147k
    } else {
6962
4
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL);
6963
4
  return;
6964
4
    }
6965
6966
147k
    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
147k
    version = xmlParseVersionInfo(ctxt);
6975
147k
    if (version == NULL) {
6976
139k
  version = xmlCharStrdup(XML_DEFAULT_VERSION);
6977
139k
        if (version == NULL) {
6978
8
            xmlErrMemory(ctxt);
6979
8
            return;
6980
8
        }
6981
139k
    } else {
6982
7.83k
  if (SKIP_BLANKS == 0) {
6983
389
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6984
389
               "Space needed here\n");
6985
389
  }
6986
7.83k
    }
6987
147k
    ctxt->input->version = version;
6988
6989
    /*
6990
     * We must have the encoding declaration
6991
     */
6992
147k
    xmlParseEncodingDecl(ctxt);
6993
6994
147k
    SKIP_BLANKS;
6995
147k
    if ((RAW == '?') && (NXT(1) == '>')) {
6996
2.76k
        SKIP(2);
6997
144k
    } else if (RAW == '>') {
6998
        /* Deprecated old WD ... */
6999
305
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7000
305
  NEXT;
7001
144k
    } else {
7002
144k
        int c;
7003
7004
144k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7005
250M
        while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) {
7006
250M
            NEXT;
7007
250M
            if (c == '>')
7008
134k
                break;
7009
250M
        }
7010
144k
    }
7011
147k
}
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
1.37k
                       const xmlChar *systemId) {
7029
1.37k
    int oldInputNr;
7030
7031
1.37k
    xmlCtxtInitializeLate(ctxt);
7032
7033
1.37k
    xmlDetectEncoding(ctxt);
7034
7035
1.37k
    if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) {
7036
97
  xmlParseTextDecl(ctxt);
7037
97
    }
7038
1.37k
    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
1.37k
    if ((ctxt->myDoc->intSubset == NULL) &&
7047
0
        (xmlCreateIntSubset(ctxt->myDoc, NULL, publicId, systemId) == NULL)) {
7048
0
        xmlErrMemory(ctxt);
7049
0
    }
7050
7051
1.37k
    ctxt->inSubset = 2;
7052
1.37k
    oldInputNr = ctxt->inputNr;
7053
7054
1.37k
    SKIP_BLANKS;
7055
258k
    while (!PARSER_STOPPED(ctxt)) {
7056
258k
        if (ctxt->input->cur >= ctxt->input->end) {
7057
1.22k
            if (ctxt->inputNr <= oldInputNr) {
7058
597
                xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED);
7059
597
                break;
7060
597
            }
7061
7062
628
            xmlPopPE(ctxt);
7063
257k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
7064
924
            xmlParseConditionalSections(ctxt);
7065
256k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
7066
254k
            xmlParseMarkupDecl(ctxt);
7067
254k
        } else if (RAW == '%') {
7068
1.41k
            xmlParsePERefInternal(ctxt, 1);
7069
1.41k
        } else {
7070
646
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
7071
7072
799
            while (ctxt->inputNr > oldInputNr)
7073
153
                xmlPopPE(ctxt);
7074
646
            break;
7075
646
        }
7076
256k
        SKIP_BLANKS;
7077
256k
        SHRINK;
7078
256k
        GROW;
7079
256k
    }
7080
1.37k
}
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
101k
xmlParseReference(xmlParserCtxt *ctxt) {
7097
101k
    xmlEntityPtr ent = NULL;
7098
101k
    const xmlChar *name;
7099
101k
    xmlChar *val;
7100
7101
101k
    if (RAW != '&')
7102
0
        return;
7103
7104
    /*
7105
     * Simple case of a CharRef
7106
     */
7107
101k
    if (NXT(1) == '#') {
7108
23.6k
  int i = 0;
7109
23.6k
  xmlChar out[16];
7110
23.6k
  int value = xmlParseCharRef(ctxt);
7111
7112
23.6k
  if (value == 0)
7113
8.18k
      return;
7114
7115
        /*
7116
         * Just encode the value in UTF-8
7117
         */
7118
15.4k
        COPY_BUF(out, i, value);
7119
15.4k
        out[i] = 0;
7120
15.4k
        if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7121
15.4k
            (!ctxt->disableSAX))
7122
8.85k
            ctxt->sax->characters(ctxt->userData, out, i);
7123
15.4k
  return;
7124
23.6k
    }
7125
7126
    /*
7127
     * We are seeing an entity reference
7128
     */
7129
77.8k
    name = xmlParseEntityRefInternal(ctxt);
7130
77.8k
    if (name == NULL)
7131
16.1k
        return;
7132
61.7k
    ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0);
7133
61.7k
    if (ent == NULL) {
7134
        /*
7135
         * Create a reference for undeclared entities.
7136
         */
7137
36.9k
        if ((ctxt->replaceEntities == 0) &&
7138
23.9k
            (ctxt->sax != NULL) &&
7139
23.9k
            (ctxt->disableSAX == 0) &&
7140
22.9k
            (ctxt->sax->reference != NULL)) {
7141
22.9k
            ctxt->sax->reference(ctxt->userData, name);
7142
22.9k
        }
7143
36.9k
        return;
7144
36.9k
    }
7145
24.8k
    if (!ctxt->wellFormed)
7146
6.11k
  return;
7147
7148
    /* special case of predefined entities */
7149
18.7k
    if ((ent->name == NULL) ||
7150
18.7k
        (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
7151
1.04k
  val = ent->content;
7152
1.04k
  if (val == NULL) return;
7153
  /*
7154
   * inline the entity.
7155
   */
7156
1.04k
  if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7157
1.04k
      (!ctxt->disableSAX))
7158
1.04k
      ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val));
7159
1.04k
  return;
7160
1.04k
    }
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
17.6k
    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
17.6k
    if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) ||
7211
9.61k
        (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
7212
9.32k
         ((ctxt->replaceEntities) ||
7213
17.0k
          (ctxt->validate)))) {
7214
17.0k
        if ((ent->flags & XML_ENT_PARSED) == 0) {
7215
3.52k
            xmlCtxtParseEntity(ctxt, ent);
7216
13.5k
        } 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
4.00k
            xmlCtxtParseEntity(ctxt, ent);
7227
4.00k
        }
7228
17.0k
    }
7229
7230
    /*
7231
     * We also check for amplification if entities aren't substituted.
7232
     * They might be expanded later.
7233
     */
7234
17.6k
    if (xmlParserEntityCheck(ctxt, ent->expandedSize))
7235
46
        return;
7236
7237
17.6k
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
7238
838
        return;
7239
7240
16.7k
    if (ctxt->replaceEntities == 0) {
7241
  /*
7242
   * Create a reference
7243
   */
7244
6.05k
        if (ctxt->sax->reference != NULL)
7245
6.05k
      ctxt->sax->reference(ctxt->userData, ent->name);
7246
10.7k
    } else if ((ent->children != NULL) && (ctxt->node != NULL)) {
7247
7.44k
        xmlNodePtr copy, cur;
7248
7249
        /*
7250
         * Seems we are generating the DOM content, copy the tree
7251
   */
7252
7.44k
        cur = ent->children;
7253
7254
        /*
7255
         * Handle first text node with SAX to coalesce text efficiently
7256
         */
7257
7.44k
        if ((cur->type == XML_TEXT_NODE) ||
7258
6.70k
            (cur->type == XML_CDATA_SECTION_NODE)) {
7259
6.70k
            int len = xmlStrlen(cur->content);
7260
7261
6.70k
            if ((cur->type == XML_TEXT_NODE) ||
7262
6.59k
                (ctxt->options & XML_PARSE_NOCDATA)) {
7263
6.59k
                if (ctxt->sax->characters != NULL)
7264
6.59k
                    ctxt->sax->characters(ctxt, cur->content, len);
7265
6.59k
            } else {
7266
112
                if (ctxt->sax->cdataBlock != NULL)
7267
112
                    ctxt->sax->cdataBlock(ctxt, cur->content, len);
7268
112
            }
7269
7270
6.70k
            cur = cur->next;
7271
6.70k
        }
7272
7273
39.9k
        while (cur != NULL) {
7274
34.5k
            xmlNodePtr last;
7275
7276
            /*
7277
             * Handle last text node with SAX to coalesce text efficiently
7278
             */
7279
34.5k
            if ((cur->next == NULL) &&
7280
3.19k
                ((cur->type == XML_TEXT_NODE) ||
7281
1.91k
                 (cur->type == XML_CDATA_SECTION_NODE))) {
7282
1.91k
                int len = xmlStrlen(cur->content);
7283
7284
1.91k
                if ((cur->type == XML_TEXT_NODE) ||
7285
1.78k
                    (ctxt->options & XML_PARSE_NOCDATA)) {
7286
1.78k
                    if (ctxt->sax->characters != NULL)
7287
1.78k
                        ctxt->sax->characters(ctxt, cur->content, len);
7288
1.78k
                } else {
7289
137
                    if (ctxt->sax->cdataBlock != NULL)
7290
137
                        ctxt->sax->cdataBlock(ctxt, cur->content, len);
7291
137
                }
7292
7293
1.91k
                break;
7294
1.91k
            }
7295
7296
            /*
7297
             * Reset coalesce buffer stats only for non-text nodes.
7298
             */
7299
32.6k
            ctxt->nodemem = 0;
7300
32.6k
            ctxt->nodelen = 0;
7301
7302
32.6k
            copy = xmlDocCopyNode(cur, ctxt->myDoc, 1);
7303
7304
32.6k
            if (copy == NULL) {
7305
61
                xmlErrMemory(ctxt);
7306
61
                break;
7307
61
            }
7308
7309
32.5k
            if (ctxt->parseMode == XML_PARSE_READER) {
7310
                /* Needed for reader */
7311
31.3k
                copy->extra = cur->extra;
7312
                /* Maybe needed for reader */
7313
31.3k
                copy->_private = cur->_private;
7314
31.3k
            }
7315
7316
32.5k
            copy->parent = ctxt->node;
7317
32.5k
            last = ctxt->node->last;
7318
32.5k
            if (last == NULL) {
7319
194
                ctxt->node->children = copy;
7320
32.3k
            } else {
7321
32.3k
                last->next = copy;
7322
32.3k
                copy->prev = last;
7323
32.3k
            }
7324
32.5k
            ctxt->node->last = copy;
7325
7326
32.5k
            cur = cur->next;
7327
32.5k
        }
7328
7.44k
    }
7329
16.7k
}
7330
7331
static void
7332
388k
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
388k
    if ((ctxt->standalone == 1) ||
7355
388k
        ((ctxt->hasExternalSubset == 0) &&
7356
375k
         (ctxt->hasPErefs == 0))) {
7357
106k
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY,
7358
106k
                          "Entity '%s' not defined\n", name);
7359
106k
#ifdef LIBXML_VALID_ENABLED
7360
282k
    } 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
248k
        xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY,
7369
248k
                         "Entity '%s' not defined\n", name, NULL);
7370
248k
#endif
7371
248k
    } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) ||
7372
4.12k
               ((ctxt->replaceEntities) &&
7373
31.5k
                ((ctxt->options & XML_PARSE_NO_XXE) == 0))) {
7374
        /*
7375
         * Also raise a non-fatal error
7376
         *
7377
         * - if the external subset is loaded and all entity declarations
7378
         *   should be available, or
7379
         * - entity substition was requested without restricting
7380
         *   external entity access.
7381
         */
7382
31.5k
        xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY,
7383
31.5k
                     "Entity '%s' not defined\n", name);
7384
31.5k
    } else {
7385
2.94k
        xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7386
2.94k
                      "Entity '%s' not defined\n", name, NULL);
7387
2.94k
    }
7388
7389
388k
    ctxt->valid = 0;
7390
388k
}
7391
7392
static xmlEntityPtr
7393
3.73M
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) {
7394
3.73M
    xmlEntityPtr ent = NULL;
7395
7396
    /*
7397
     * Predefined entities override any extra definition
7398
     */
7399
3.73M
    if ((ctxt->options & XML_PARSE_OLDSAX) == 0) {
7400
1.80M
        ent = xmlGetPredefinedEntity(name);
7401
1.80M
        if (ent != NULL)
7402
637k
            return(ent);
7403
1.80M
    }
7404
7405
    /*
7406
     * Ask first SAX for entity resolution, otherwise try the
7407
     * entities which may have stored in the parser context.
7408
     */
7409
3.10M
    if (ctxt->sax != NULL) {
7410
3.10M
  if (ctxt->sax->getEntity != NULL)
7411
3.10M
      ent = ctxt->sax->getEntity(ctxt->userData, name);
7412
3.10M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7413
2.36k
      (ctxt->options & XML_PARSE_OLDSAX))
7414
989
      ent = xmlGetPredefinedEntity(name);
7415
3.10M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7416
2.36k
      (ctxt->userData==ctxt)) {
7417
2.36k
      ent = xmlSAX2GetEntity(ctxt, name);
7418
2.36k
  }
7419
3.10M
    }
7420
7421
3.10M
    if (ent == NULL) {
7422
312k
        xmlHandleUndeclaredEntity(ctxt, name);
7423
312k
    }
7424
7425
    /*
7426
     * [ WFC: Parsed Entity ]
7427
     * An entity reference must not contain the name of an
7428
     * unparsed entity
7429
     */
7430
2.78M
    else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) {
7431
262
  xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY,
7432
262
     "Entity reference to unparsed entity %s\n", name);
7433
262
        ent = NULL;
7434
262
    }
7435
7436
    /*
7437
     * [ WFC: No External Entity References ]
7438
     * Attribute values cannot contain direct or indirect
7439
     * entity references to external entities.
7440
     */
7441
2.78M
    else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) {
7442
13.1k
        if (inAttr) {
7443
916
            xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL,
7444
916
                 "Attribute references external entity '%s'\n", name);
7445
916
            ent = NULL;
7446
916
        }
7447
13.1k
    }
7448
7449
3.10M
    return(ent);
7450
3.73M
}
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
1.08M
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) {
7462
1.08M
    const xmlChar *name;
7463
7464
1.08M
    GROW;
7465
7466
1.08M
    if (RAW != '&')
7467
0
        return(NULL);
7468
1.08M
    NEXT;
7469
1.08M
    name = xmlParseName(ctxt);
7470
1.08M
    if (name == NULL) {
7471
45.5k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7472
45.5k
           "xmlParseEntityRef: no name\n");
7473
45.5k
        return(NULL);
7474
45.5k
    }
7475
1.04M
    if (RAW != ';') {
7476
82.3k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7477
82.3k
  return(NULL);
7478
82.3k
    }
7479
959k
    NEXT;
7480
7481
959k
    return(name);
7482
1.04M
}
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
2.77M
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) {
7535
2.77M
    xmlChar *name;
7536
2.77M
    const xmlChar *ptr;
7537
2.77M
    xmlChar cur;
7538
7539
2.77M
    if ((str == NULL) || (*str == NULL))
7540
0
        return(NULL);
7541
2.77M
    ptr = *str;
7542
2.77M
    cur = *ptr;
7543
2.77M
    if (cur != '&')
7544
0
  return(NULL);
7545
7546
2.77M
    ptr++;
7547
2.77M
    name = xmlParseStringName(ctxt, &ptr);
7548
2.77M
    if (name == NULL) {
7549
26
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7550
26
           "xmlParseStringEntityRef: no name\n");
7551
26
  *str = ptr;
7552
26
  return(NULL);
7553
26
    }
7554
2.77M
    if (*ptr != ';') {
7555
4
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7556
4
        xmlFree(name);
7557
4
  *str = ptr;
7558
4
  return(NULL);
7559
4
    }
7560
2.77M
    ptr++;
7561
7562
2.77M
    *str = ptr;
7563
2.77M
    return(name);
7564
2.77M
}
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
176k
xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl) {
7598
176k
    const xmlChar *name;
7599
176k
    xmlEntityPtr entity = NULL;
7600
176k
    xmlParserInputPtr input;
7601
7602
176k
    if (RAW != '%')
7603
0
        return;
7604
176k
    NEXT;
7605
176k
    name = xmlParseName(ctxt);
7606
176k
    if (name == NULL) {
7607
11.7k
  xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n");
7608
11.7k
  return;
7609
11.7k
    }
7610
164k
    if (RAW != ';') {
7611
11.9k
  xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL);
7612
11.9k
        return;
7613
11.9k
    }
7614
7615
152k
    NEXT;
7616
7617
    /* Must be set before xmlHandleUndeclaredEntity */
7618
152k
    ctxt->hasPErefs = 1;
7619
7620
    /*
7621
     * Request the entity from SAX
7622
     */
7623
152k
    if ((ctxt->sax != NULL) &&
7624
152k
  (ctxt->sax->getParameterEntity != NULL))
7625
152k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7626
7627
152k
    if (entity == NULL) {
7628
43.8k
        xmlHandleUndeclaredEntity(ctxt, name);
7629
108k
    } else {
7630
  /*
7631
   * Internal checking in case the entity quest barfed
7632
   */
7633
108k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7634
100k
      (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
108k
  } else {
7639
108k
      if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) &&
7640
100k
                ((ctxt->options & XML_PARSE_NO_XXE) ||
7641
100k
     (((ctxt->loadsubset & ~XML_SKIP_IDS) == 0) &&
7642
29.6k
      (ctxt->replaceEntities == 0) &&
7643
9.62k
      (ctxt->validate == 0))))
7644
396
    return;
7645
7646
108k
            if (entity->flags & XML_ENT_EXPANDING) {
7647
14
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
7648
14
                return;
7649
14
            }
7650
7651
108k
      input = xmlNewEntityInputStream(ctxt, entity);
7652
108k
      if (xmlCtxtPushInput(ctxt, input) < 0) {
7653
3.22k
                xmlFreeInputStream(input);
7654
3.22k
    return;
7655
3.22k
            }
7656
7657
105k
            entity->flags |= XML_ENT_EXPANDING;
7658
7659
105k
            if (markupDecl)
7660
94.6k
                input->flags |= XML_INPUT_MARKUP_DECL;
7661
7662
105k
            GROW;
7663
7664
105k
      if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
7665
96.6k
                xmlDetectEncoding(ctxt);
7666
7667
96.6k
                if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
7668
42.1k
                    (IS_BLANK_CH(NXT(5)))) {
7669
41.4k
                    xmlParseTextDecl(ctxt);
7670
41.4k
                }
7671
96.6k
            }
7672
105k
  }
7673
108k
    }
7674
152k
}
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
114k
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) {
7697
114k
    xmlParserInputPtr oldinput, input = NULL;
7698
114k
    xmlParserInputPtr *oldinputTab;
7699
114k
    xmlChar *oldencoding;
7700
114k
    xmlChar *content = NULL;
7701
114k
    xmlResourceType rtype;
7702
114k
    size_t length, i;
7703
114k
    int oldinputNr, oldinputMax;
7704
114k
    int ret = -1;
7705
114k
    int res;
7706
7707
114k
    if ((ctxt == NULL) || (entity == NULL) ||
7708
114k
        ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) &&
7709
0
   (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) ||
7710
114k
  (entity->content != NULL)) {
7711
0
  xmlFatalErr(ctxt, XML_ERR_ARGUMENT,
7712
0
              "xmlLoadEntityContent parameter error");
7713
0
        return(-1);
7714
0
    }
7715
7716
114k
    if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY)
7717
114k
        rtype = XML_RESOURCE_PARAMETER_ENTITY;
7718
0
    else
7719
0
        rtype = XML_RESOURCE_GENERAL_ENTITY;
7720
7721
114k
    input = xmlLoadResource(ctxt, (char *) entity->URI,
7722
114k
                            (char *) entity->ExternalID, rtype);
7723
114k
    if (input == NULL)
7724
810
        return(-1);
7725
7726
113k
    oldinput = ctxt->input;
7727
113k
    oldinputNr = ctxt->inputNr;
7728
113k
    oldinputMax = ctxt->inputMax;
7729
113k
    oldinputTab = ctxt->inputTab;
7730
113k
    oldencoding = ctxt->encoding;
7731
7732
113k
    ctxt->input = NULL;
7733
113k
    ctxt->inputNr = 0;
7734
113k
    ctxt->inputMax = 1;
7735
113k
    ctxt->encoding = NULL;
7736
113k
    ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr));
7737
113k
    if (ctxt->inputTab == NULL) {
7738
5
        xmlErrMemory(ctxt);
7739
5
        xmlFreeInputStream(input);
7740
5
        goto error;
7741
5
    }
7742
7743
113k
    xmlBufResetInput(input->buf->buffer, input);
7744
7745
113k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
7746
6
        xmlFreeInputStream(input);
7747
6
        goto error;
7748
6
    }
7749
7750
113k
    xmlDetectEncoding(ctxt);
7751
7752
    /*
7753
     * Parse a possible text declaration first
7754
     */
7755
113k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
7756
105k
  xmlParseTextDecl(ctxt);
7757
        /*
7758
         * An XML-1.0 document can't reference an entity not XML-1.0
7759
         */
7760
105k
        if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
7761
1.70k
            (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
7762
919
            xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
7763
919
                           "Version mismatch between document and entity\n");
7764
919
        }
7765
105k
    }
7766
7767
113k
    length = input->cur - input->base;
7768
113k
    xmlBufShrink(input->buf->buffer, length);
7769
113k
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7770
7771
122k
    while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0)
7772
8.21k
        ;
7773
7774
113k
    xmlBufResetInput(input->buf->buffer, input);
7775
7776
113k
    if (res < 0) {
7777
1.30k
        xmlCtxtErrIO(ctxt, input->buf->error, NULL);
7778
1.30k
        goto error;
7779
1.30k
    }
7780
7781
112k
    length = xmlBufUse(input->buf->buffer);
7782
112k
    if (length > INT_MAX) {
7783
0
        xmlErrMemory(ctxt);
7784
0
        goto error;
7785
0
    }
7786
7787
112k
    content = xmlStrndup(xmlBufContent(input->buf->buffer), length);
7788
112k
    if (content == NULL) {
7789
4
        xmlErrMemory(ctxt);
7790
4
        goto error;
7791
4
    }
7792
7793
10.8M
    for (i = 0; i < length; ) {
7794
10.8M
        int clen = length - i;
7795
10.8M
        int c = xmlGetUTF8Char(content + i, &clen);
7796
7797
10.8M
        if ((c < 0) || (!IS_CHAR(c))) {
7798
112k
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
7799
112k
                              "xmlLoadEntityContent: invalid char value %d\n",
7800
112k
                              content[i]);
7801
112k
            goto error;
7802
112k
        }
7803
10.7M
        i += clen;
7804
10.7M
    }
7805
7806
157
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7807
157
    entity->content = content;
7808
157
    entity->length = length;
7809
157
    content = NULL;
7810
157
    ret = 0;
7811
7812
113k
error:
7813
227k
    while (ctxt->inputNr > 0)
7814
113k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
7815
113k
    xmlFree(ctxt->inputTab);
7816
113k
    xmlFree(ctxt->encoding);
7817
7818
113k
    ctxt->input = oldinput;
7819
113k
    ctxt->inputNr = oldinputNr;
7820
113k
    ctxt->inputMax = oldinputMax;
7821
113k
    ctxt->inputTab = oldinputTab;
7822
113k
    ctxt->encoding = oldencoding;
7823
7824
113k
    xmlFree(content);
7825
7826
113k
    return(ret);
7827
157
}
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
164k
xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) {
7860
164k
    const xmlChar *ptr;
7861
164k
    xmlChar cur;
7862
164k
    xmlChar *name;
7863
164k
    xmlEntityPtr entity = NULL;
7864
7865
164k
    if ((str == NULL) || (*str == NULL)) return(NULL);
7866
164k
    ptr = *str;
7867
164k
    cur = *ptr;
7868
164k
    if (cur != '%')
7869
0
        return(NULL);
7870
164k
    ptr++;
7871
164k
    name = xmlParseStringName(ctxt, &ptr);
7872
164k
    if (name == NULL) {
7873
1.67k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7874
1.67k
           "xmlParseStringPEReference: no name\n");
7875
1.67k
  *str = ptr;
7876
1.67k
  return(NULL);
7877
1.67k
    }
7878
163k
    cur = *ptr;
7879
163k
    if (cur != ';') {
7880
7.72k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7881
7.72k
  xmlFree(name);
7882
7.72k
  *str = ptr;
7883
7.72k
  return(NULL);
7884
7.72k
    }
7885
155k
    ptr++;
7886
7887
    /* Must be set before xmlHandleUndeclaredEntity */
7888
155k
    ctxt->hasPErefs = 1;
7889
7890
    /*
7891
     * Request the entity from SAX
7892
     */
7893
155k
    if ((ctxt->sax != NULL) &&
7894
155k
  (ctxt->sax->getParameterEntity != NULL))
7895
155k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7896
7897
155k
    if (entity == NULL) {
7898
32.2k
        xmlHandleUndeclaredEntity(ctxt, name);
7899
123k
    } else {
7900
  /*
7901
   * Internal checking in case the entity quest barfed
7902
   */
7903
123k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7904
116k
      (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
123k
    }
7910
7911
155k
    xmlFree(name);
7912
155k
    *str = ptr;
7913
155k
    return(entity);
7914
163k
}
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
18.2k
xmlParseDocTypeDecl(xmlParserCtxt *ctxt) {
7933
18.2k
    const xmlChar *name = NULL;
7934
18.2k
    xmlChar *publicId = NULL;
7935
18.2k
    xmlChar *URI = NULL;
7936
7937
    /*
7938
     * We know that '<!DOCTYPE' has been detected.
7939
     */
7940
18.2k
    SKIP(9);
7941
7942
18.2k
    if (SKIP_BLANKS == 0) {
7943
1.12k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7944
1.12k
                       "Space required after 'DOCTYPE'\n");
7945
1.12k
    }
7946
7947
    /*
7948
     * Parse the DOCTYPE name.
7949
     */
7950
18.2k
    name = xmlParseName(ctxt);
7951
18.2k
    if (name == NULL) {
7952
906
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7953
906
           "xmlParseDocTypeDecl : no DOCTYPE name !\n");
7954
906
    }
7955
18.2k
    ctxt->intSubName = name;
7956
7957
18.2k
    SKIP_BLANKS;
7958
7959
    /*
7960
     * Check for public and system identifier (URI)
7961
     */
7962
18.2k
    URI = xmlParseExternalID(ctxt, &publicId, 1);
7963
7964
18.2k
    if ((URI != NULL) || (publicId != NULL)) {
7965
2.68k
        ctxt->hasExternalSubset = 1;
7966
2.68k
    }
7967
18.2k
    ctxt->extSubURI = URI;
7968
18.2k
    ctxt->extSubSystem = publicId;
7969
7970
18.2k
    SKIP_BLANKS;
7971
7972
    /*
7973
     * Create and update the internal subset.
7974
     */
7975
18.2k
    if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) &&
7976
18.2k
  (!ctxt->disableSAX))
7977
18.1k
  ctxt->sax->internalSubset(ctxt->userData, name, publicId, URI);
7978
7979
18.2k
    if ((RAW != '[') && (RAW != '>')) {
7980
235
  xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
7981
235
    }
7982
18.2k
}
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
12.5k
xmlParseInternalSubset(xmlParserCtxtPtr ctxt) {
7993
    /*
7994
     * Is there any DTD definition ?
7995
     */
7996
12.5k
    if (RAW == '[') {
7997
12.5k
        int oldInputNr = ctxt->inputNr;
7998
7999
12.5k
        NEXT;
8000
  /*
8001
   * Parse the succession of Markup declarations and
8002
   * PEReferences.
8003
   * Subsequence (markupdecl | PEReference | S)*
8004
   */
8005
12.5k
  SKIP_BLANKS;
8006
654k
        while (1) {
8007
654k
            if (PARSER_STOPPED(ctxt)) {
8008
635
                return;
8009
653k
            } else if (ctxt->input->cur >= ctxt->input->end) {
8010
93.5k
                if (ctxt->inputNr <= oldInputNr) {
8011
1.34k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8012
1.34k
                    return;
8013
1.34k
                }
8014
92.2k
                xmlPopPE(ctxt);
8015
559k
            } else if ((RAW == ']') && (ctxt->inputNr <= oldInputNr)) {
8016
7.73k
                NEXT;
8017
7.73k
                SKIP_BLANKS;
8018
7.73k
                break;
8019
552k
            } else if ((PARSER_EXTERNAL(ctxt)) &&
8020
496k
                       (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
4.01k
                xmlParseConditionalSections(ctxt);
8026
548k
            } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
8027
393k
                xmlParseMarkupDecl(ctxt);
8028
393k
            } else if (RAW == '%') {
8029
152k
                xmlParsePERefInternal(ctxt, 1);
8030
152k
            } else {
8031
2.83k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8032
8033
3.45k
                while (ctxt->inputNr > oldInputNr)
8034
617
                    xmlPopPE(ctxt);
8035
2.83k
                return;
8036
2.83k
            }
8037
641k
            SKIP_BLANKS;
8038
641k
            SHRINK;
8039
641k
            GROW;
8040
641k
        }
8041
12.5k
    }
8042
8043
    /*
8044
     * We should be at the end of the DOCTYPE declaration.
8045
     */
8046
7.73k
    if (RAW != '>') {
8047
106
        xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8048
106
        return;
8049
106
    }
8050
7.62k
    NEXT;
8051
7.62k
}
8052
8053
#ifdef LIBXML_SAX1_ENABLED
8054
/**
8055
 * Parse an attribute
8056
 *
8057
 * @deprecated Internal function, don't use.
8058
 *
8059
 *     [41] Attribute ::= Name Eq AttValue
8060
 *
8061
 * [ WFC: No External Entity References ]
8062
 * Attribute values cannot contain direct or indirect entity references
8063
 * to external entities.
8064
 *
8065
 * [ WFC: No < in Attribute Values ]
8066
 * The replacement text of any entity referred to directly or indirectly in
8067
 * an attribute value (other than "&lt;") must not contain a <.
8068
 *
8069
 * [ VC: Attribute Value Type ]
8070
 * The attribute must have been declared; the value must be of the type
8071
 * declared for it.
8072
 *
8073
 *     [25] Eq ::= S? '=' S?
8074
 *
8075
 * With namespace:
8076
 *
8077
 *     [NS 11] Attribute ::= QName Eq AttValue
8078
 *
8079
 * Also the case QName == xmlns:??? is handled independently as a namespace
8080
 * definition.
8081
 *
8082
 * @param ctxt  an XML parser context
8083
 * @param value  a xmlChar ** used to store the value of the attribute
8084
 * @returns the attribute name, and the value in *value.
8085
 */
8086
8087
const xmlChar *
8088
154k
xmlParseAttribute(xmlParserCtxt *ctxt, xmlChar **value) {
8089
154k
    const xmlChar *name;
8090
154k
    xmlChar *val;
8091
8092
154k
    *value = NULL;
8093
154k
    GROW;
8094
154k
    name = xmlParseName(ctxt);
8095
154k
    if (name == NULL) {
8096
36.2k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8097
36.2k
                 "error parsing attribute name\n");
8098
36.2k
        return(NULL);
8099
36.2k
    }
8100
8101
    /*
8102
     * read the value
8103
     */
8104
117k
    SKIP_BLANKS;
8105
117k
    if (RAW == '=') {
8106
99.4k
        NEXT;
8107
99.4k
  SKIP_BLANKS;
8108
99.4k
  val = xmlParseAttValue(ctxt);
8109
99.4k
    } else {
8110
18.4k
  xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8111
18.4k
         "Specification mandates value for attribute %s\n", name);
8112
18.4k
  return(name);
8113
18.4k
    }
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
99.4k
    if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) {
8121
10.6k
  if (!xmlCheckLanguageID(val)) {
8122
7.78k
      xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8123
7.78k
              "Malformed value for xml:lang : %s\n",
8124
7.78k
        val, NULL);
8125
7.78k
  }
8126
10.6k
    }
8127
8128
    /*
8129
     * Check that xml:space conforms to the specification
8130
     */
8131
99.4k
    if (xmlStrEqual(name, BAD_CAST "xml:space")) {
8132
1.04k
  if (xmlStrEqual(val, BAD_CAST "default"))
8133
227
      *(ctxt->space) = 0;
8134
822
  else if (xmlStrEqual(val, BAD_CAST "preserve"))
8135
539
      *(ctxt->space) = 1;
8136
283
  else {
8137
283
    xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8138
283
"Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8139
283
                                 val, NULL);
8140
283
  }
8141
1.04k
    }
8142
8143
99.4k
    *value = val;
8144
99.4k
    return(name);
8145
117k
}
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
305k
xmlParseStartTag(xmlParserCtxt *ctxt) {
8176
305k
    const xmlChar *name;
8177
305k
    const xmlChar *attname;
8178
305k
    xmlChar *attvalue;
8179
305k
    const xmlChar **atts = ctxt->atts;
8180
305k
    int nbatts = 0;
8181
305k
    int maxatts = ctxt->maxatts;
8182
305k
    int i;
8183
8184
305k
    if (RAW != '<') return(NULL);
8185
305k
    NEXT1;
8186
8187
305k
    name = xmlParseName(ctxt);
8188
305k
    if (name == NULL) {
8189
17.1k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8190
17.1k
       "xmlParseStartTag: invalid element name\n");
8191
17.1k
        return(NULL);
8192
17.1k
    }
8193
8194
    /*
8195
     * Now parse the attributes, it ends up with the ending
8196
     *
8197
     * (S Attribute)* S?
8198
     */
8199
288k
    SKIP_BLANKS;
8200
288k
    GROW;
8201
8202
346k
    while (((RAW != '>') &&
8203
202k
     ((RAW != '/') || (NXT(1) != '>')) &&
8204
163k
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8205
154k
  attname = xmlParseAttribute(ctxt, &attvalue);
8206
154k
        if (attname == NULL)
8207
36.2k
      break;
8208
117k
        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
212k
      for (i = 0; i < nbatts;i += 2) {
8215
128k
          if (xmlStrEqual(atts[i], attname)) {
8216
12.1k
        xmlErrAttributeDup(ctxt, NULL, attname);
8217
12.1k
        goto failed;
8218
12.1k
    }
8219
128k
      }
8220
      /*
8221
       * Add the pair to atts
8222
       */
8223
84.0k
      if (nbatts + 4 > maxatts) {
8224
17.8k
          const xmlChar **n;
8225
17.8k
                int newSize;
8226
8227
17.8k
                newSize = xmlGrowCapacity(maxatts, sizeof(n[0]) * 2,
8228
17.8k
                                          11, XML_MAX_ATTRS);
8229
17.8k
                if (newSize < 0) {
8230
0
        xmlErrMemory(ctxt);
8231
0
        goto failed;
8232
0
    }
8233
17.8k
#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
8234
17.8k
                if (newSize < 2)
8235
11.1k
                    newSize = 2;
8236
17.8k
#endif
8237
17.8k
          n = xmlRealloc(atts, newSize * sizeof(n[0]) * 2);
8238
17.8k
    if (n == NULL) {
8239
4
        xmlErrMemory(ctxt);
8240
4
        goto failed;
8241
4
    }
8242
17.8k
    atts = n;
8243
17.8k
                maxatts = newSize * 2;
8244
17.8k
    ctxt->atts = atts;
8245
17.8k
    ctxt->maxatts = maxatts;
8246
17.8k
      }
8247
8248
84.0k
      atts[nbatts++] = attname;
8249
84.0k
      atts[nbatts++] = attvalue;
8250
84.0k
      atts[nbatts] = NULL;
8251
84.0k
      atts[nbatts + 1] = NULL;
8252
8253
84.0k
            attvalue = NULL;
8254
84.0k
  }
8255
8256
117k
failed:
8257
8258
117k
        if (attvalue != NULL)
8259
12.1k
            xmlFree(attvalue);
8260
8261
117k
  GROW
8262
117k
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
8263
59.7k
      break;
8264
58.1k
  if (SKIP_BLANKS == 0) {
8265
25.7k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
8266
25.7k
         "attributes construct error\n");
8267
25.7k
  }
8268
58.1k
  SHRINK;
8269
58.1k
        GROW;
8270
58.1k
    }
8271
8272
    /*
8273
     * SAX: Start of Element !
8274
     */
8275
288k
    if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) &&
8276
288k
  (!ctxt->disableSAX)) {
8277
271k
  if (nbatts > 0)
8278
64.8k
      ctxt->sax->startElement(ctxt->userData, name, atts);
8279
207k
  else
8280
207k
      ctxt->sax->startElement(ctxt->userData, name, NULL);
8281
271k
    }
8282
8283
288k
    if (atts != NULL) {
8284
        /* Free only the content strings */
8285
282k
        for (i = 1;i < nbatts;i+=2)
8286
84.0k
      if (atts[i] != NULL)
8287
84.0k
         xmlFree((xmlChar *) atts[i]);
8288
198k
    }
8289
288k
    return(name);
8290
288k
}
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
79.5k
xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) {
8306
79.5k
    const xmlChar *name;
8307
8308
79.5k
    GROW;
8309
79.5k
    if ((RAW != '<') || (NXT(1) != '/')) {
8310
39
  xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED,
8311
39
           "xmlParseEndTag: '</' not found\n");
8312
39
  return;
8313
39
    }
8314
79.5k
    SKIP(2);
8315
8316
79.5k
    name = xmlParseNameAndCompare(ctxt,ctxt->name);
8317
8318
    /*
8319
     * We should definitely be at the ending "S? '>'" part
8320
     */
8321
79.5k
    GROW;
8322
79.5k
    SKIP_BLANKS;
8323
79.5k
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
8324
12.1k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
8325
12.1k
    } else
8326
67.3k
  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
79.5k
    if (name != (xmlChar*)1) {
8335
5.43k
        if (name == NULL) name = BAD_CAST "unparsable";
8336
5.43k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
8337
5.43k
         "Opening and ending tag mismatch: %s line %d and %s\n",
8338
5.43k
                    ctxt->name, line, name);
8339
5.43k
    }
8340
8341
    /*
8342
     * SAX: End of Tag
8343
     */
8344
79.5k
    if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
8345
79.5k
  (!ctxt->disableSAX))
8346
70.8k
        ctxt->sax->endElement(ctxt->userData, ctxt->name);
8347
8348
79.5k
    namePop(ctxt);
8349
79.5k
    spacePop(ctxt);
8350
79.5k
}
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
1.12M
xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) {
8391
1.12M
    xmlHashedString l, p;
8392
1.12M
    int start, isNCName = 0;
8393
8394
1.12M
    l.name = NULL;
8395
1.12M
    p.name = NULL;
8396
8397
1.12M
    GROW;
8398
1.12M
    start = CUR_PTR - BASE_PTR;
8399
8400
1.12M
    l = xmlParseNCName(ctxt);
8401
1.12M
    if (l.name != NULL) {
8402
1.01M
        isNCName = 1;
8403
1.01M
        if (CUR == ':') {
8404
203k
            NEXT;
8405
203k
            p = l;
8406
203k
            l = xmlParseNCName(ctxt);
8407
203k
        }
8408
1.01M
    }
8409
1.12M
    if ((l.name == NULL) || (CUR == ':')) {
8410
109k
        xmlChar *tmp;
8411
8412
109k
        l.name = NULL;
8413
109k
        p.name = NULL;
8414
109k
        if ((isNCName == 0) && (CUR != ':'))
8415
97.6k
            return(l);
8416
12.1k
        tmp = xmlParseNmtoken(ctxt);
8417
12.1k
        if (tmp != NULL)
8418
7.20k
            xmlFree(tmp);
8419
12.1k
        l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start,
8420
12.1k
                                CUR_PTR - (BASE_PTR + start));
8421
12.1k
        if (l.name == NULL) {
8422
1
            xmlErrMemory(ctxt);
8423
1
            return(l);
8424
1
        }
8425
12.1k
        xmlNsErr(ctxt, XML_NS_ERR_QNAME,
8426
12.1k
                 "Failed to parse QName '%s'\n", l.name, NULL, NULL);
8427
12.1k
    }
8428
8429
1.02M
    *prefix = p;
8430
1.02M
    return(l);
8431
1.12M
}
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
7.67k
xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) {
8447
7.67k
    xmlHashedString n, p;
8448
8449
7.67k
    n = xmlParseQNameHashed(ctxt, &p);
8450
7.67k
    if (n.name == NULL)
8451
637
        return(NULL);
8452
7.04k
    *prefix = p.name;
8453
7.04k
    return(n.name);
8454
7.67k
}
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
12.4k
                        xmlChar const *prefix) {
8470
12.4k
    const xmlChar *cmp;
8471
12.4k
    const xmlChar *in;
8472
12.4k
    const xmlChar *ret;
8473
12.4k
    const xmlChar *prefix2;
8474
8475
12.4k
    if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name));
8476
8477
12.4k
    GROW;
8478
12.4k
    in = ctxt->input->cur;
8479
8480
12.4k
    cmp = prefix;
8481
25.4k
    while (*in != 0 && *in == *cmp) {
8482
13.0k
  ++in;
8483
13.0k
  ++cmp;
8484
13.0k
    }
8485
12.4k
    if ((*cmp == 0) && (*in == ':')) {
8486
10.2k
        in++;
8487
10.2k
  cmp = name;
8488
29.1k
  while (*in != 0 && *in == *cmp) {
8489
18.8k
      ++in;
8490
18.8k
      ++cmp;
8491
18.8k
  }
8492
10.2k
  if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
8493
      /* success */
8494
4.75k
            ctxt->input->col += in - ctxt->input->cur;
8495
4.75k
      ctxt->input->cur = in;
8496
4.75k
      return((const xmlChar*) 1);
8497
4.75k
  }
8498
10.2k
    }
8499
    /*
8500
     * all strings coms from the dictionary, equality can be done directly
8501
     */
8502
7.67k
    ret = xmlParseQName (ctxt, &prefix2);
8503
7.67k
    if (ret == NULL)
8504
637
        return(NULL);
8505
7.04k
    if ((ret == name) && (prefix == prefix2))
8506
3.79k
  return((const xmlChar*) 1);
8507
3.24k
    return ret;
8508
7.04k
}
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
285k
{
8529
285k
    xmlHashedString hname;
8530
285k
    const xmlChar *prefix, *name;
8531
285k
    xmlChar *val = NULL, *internal_val = NULL;
8532
285k
    int special = 0;
8533
285k
    int isNamespace;
8534
285k
    int flags;
8535
8536
285k
    *value = NULL;
8537
285k
    GROW;
8538
285k
    hname = xmlParseQNameHashed(ctxt, hprefix);
8539
285k
    if (hname.name == NULL) {
8540
69.9k
        xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8541
69.9k
                       "error parsing attribute name\n");
8542
69.9k
        return(hname);
8543
69.9k
    }
8544
215k
    name = hname.name;
8545
215k
    prefix = hprefix->name;
8546
8547
    /*
8548
     * get the type if needed
8549
     */
8550
215k
    if (ctxt->attsSpecial != NULL) {
8551
23.0k
        special = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem,
8552
23.0k
                                              prefix, name));
8553
23.0k
    }
8554
8555
    /*
8556
     * read the value
8557
     */
8558
215k
    SKIP_BLANKS;
8559
215k
    if (RAW != '=') {
8560
14.6k
        xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8561
14.6k
                          "Specification mandates value for attribute %s\n",
8562
14.6k
                          name);
8563
14.6k
        goto error;
8564
14.6k
    }
8565
8566
8567
200k
    NEXT;
8568
200k
    SKIP_BLANKS;
8569
200k
    flags = 0;
8570
200k
    isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) ||
8571
182k
                   (prefix == ctxt->str_xmlns));
8572
200k
    val = xmlParseAttValueInternal(ctxt, len, &flags, special,
8573
200k
                                   isNamespace);
8574
200k
    if (val == NULL)
8575
3.61k
        goto error;
8576
8577
197k
    *alloc = (flags & XML_ATTVAL_ALLOC) != 0;
8578
8579
197k
#ifdef LIBXML_VALID_ENABLED
8580
197k
    if ((ctxt->validate) &&
8581
83.8k
        (ctxt->standalone == 1) &&
8582
468
        (special & XML_SPECIAL_EXTERNAL) &&
8583
108
        (flags & XML_ATTVAL_NORM_CHANGE)) {
8584
35
        xmlValidityError(ctxt, XML_DTD_NOT_STANDALONE,
8585
35
                         "standalone: normalization of attribute %s on %s "
8586
35
                         "by external subset declaration\n",
8587
35
                         name, elem);
8588
35
    }
8589
197k
#endif
8590
8591
197k
    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
71.9k
        if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) {
8598
59.1k
            internal_val = xmlStrndup(val, *len);
8599
59.1k
            if (internal_val == NULL)
8600
2
                goto mem_error;
8601
59.1k
            if (!xmlCheckLanguageID(internal_val)) {
8602
54.6k
                xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8603
54.6k
                              "Malformed value for xml:lang : %s\n",
8604
54.6k
                              internal_val, NULL);
8605
54.6k
            }
8606
59.1k
        }
8607
8608
        /*
8609
         * Check that xml:space conforms to the specification
8610
         */
8611
71.9k
        if (xmlStrEqual(name, BAD_CAST "space")) {
8612
1.20k
            internal_val = xmlStrndup(val, *len);
8613
1.20k
            if (internal_val == NULL)
8614
1
                goto mem_error;
8615
1.20k
            if (xmlStrEqual(internal_val, BAD_CAST "default"))
8616
308
                *(ctxt->space) = 0;
8617
896
            else if (xmlStrEqual(internal_val, BAD_CAST "preserve"))
8618
647
                *(ctxt->space) = 1;
8619
249
            else {
8620
249
                xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8621
249
                              "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8622
249
                              internal_val, NULL);
8623
249
            }
8624
1.20k
        }
8625
71.9k
        if (internal_val) {
8626
60.3k
            xmlFree(internal_val);
8627
60.3k
        }
8628
71.9k
    }
8629
8630
197k
    *value = val;
8631
197k
    return (hname);
8632
8633
3
mem_error:
8634
3
    xmlErrMemory(ctxt);
8635
18.2k
error:
8636
18.2k
    if ((val != NULL) && (*alloc != 0))
8637
1
        xmlFree(val);
8638
18.2k
    return(hname);
8639
3
}
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
349k
                  const xmlChar *uri, unsigned hashValue, int aindex) {
8656
349k
    xmlAttrHashBucket *table = ctxt->attrHash;
8657
349k
    xmlAttrHashBucket *bucket;
8658
349k
    unsigned hindex;
8659
8660
349k
    hindex = hashValue & (size - 1);
8661
349k
    bucket = &table[hindex];
8662
8663
408k
    while (bucket->index >= 0) {
8664
138k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8665
8666
138k
        if (name == atts[0]) {
8667
82.0k
            int nsIndex = XML_PTR_TO_INT(atts[2]);
8668
8669
82.0k
            if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) :
8670
82.0k
                (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) :
8671
67.0k
                (uri == ctxt->nsTab[nsIndex * 2 + 1]))
8672
80.4k
                return(bucket->index);
8673
82.0k
        }
8674
8675
58.3k
        hindex++;
8676
58.3k
        bucket++;
8677
58.3k
        if (hindex >= size) {
8678
2.64k
            hindex = 0;
8679
2.64k
            bucket = table;
8680
2.64k
        }
8681
58.3k
    }
8682
8683
269k
    bucket->index = aindex;
8684
8685
269k
    return(INT_MAX);
8686
349k
}
8687
8688
static int
8689
xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size,
8690
                       const xmlChar *name, const xmlChar *prefix,
8691
5.79k
                       unsigned hashValue, int aindex) {
8692
5.79k
    xmlAttrHashBucket *table = ctxt->attrHash;
8693
5.79k
    xmlAttrHashBucket *bucket;
8694
5.79k
    unsigned hindex;
8695
8696
5.79k
    hindex = hashValue & (size - 1);
8697
5.79k
    bucket = &table[hindex];
8698
8699
8.86k
    while (bucket->index >= 0) {
8700
6.00k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8701
8702
6.00k
        if ((name == atts[0]) && (prefix == atts[1]))
8703
2.94k
            return(bucket->index);
8704
8705
3.06k
        hindex++;
8706
3.06k
        bucket++;
8707
3.06k
        if (hindex >= size) {
8708
202
            hindex = 0;
8709
202
            bucket = table;
8710
202
        }
8711
3.06k
    }
8712
8713
2.85k
    bucket->index = aindex;
8714
8715
2.85k
    return(INT_MAX);
8716
5.79k
}
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
827k
                  const xmlChar **URI, int *nbNsPtr) {
8750
827k
    xmlHashedString hlocalname;
8751
827k
    xmlHashedString hprefix;
8752
827k
    xmlHashedString hattname;
8753
827k
    xmlHashedString haprefix;
8754
827k
    const xmlChar *localname;
8755
827k
    const xmlChar *prefix;
8756
827k
    const xmlChar *attname;
8757
827k
    const xmlChar *aprefix;
8758
827k
    const xmlChar *uri;
8759
827k
    xmlChar *attvalue = NULL;
8760
827k
    const xmlChar **atts = ctxt->atts;
8761
827k
    unsigned attrHashSize = 0;
8762
827k
    int maxatts = ctxt->maxatts;
8763
827k
    int nratts, nbatts, nbdef;
8764
827k
    int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts;
8765
827k
    int alloc = 0;
8766
827k
    int numNsErr = 0;
8767
827k
    int numDupErr = 0;
8768
8769
827k
    if (RAW != '<') return(NULL);
8770
827k
    NEXT1;
8771
8772
827k
    nbatts = 0;
8773
827k
    nratts = 0;
8774
827k
    nbdef = 0;
8775
827k
    nbNs = 0;
8776
827k
    nbTotalDef = 0;
8777
827k
    attval = 0;
8778
8779
827k
    if (xmlParserNsStartElement(ctxt->nsdb) < 0) {
8780
0
        xmlErrMemory(ctxt);
8781
0
        return(NULL);
8782
0
    }
8783
8784
827k
    hlocalname = xmlParseQNameHashed(ctxt, &hprefix);
8785
827k
    if (hlocalname.name == NULL) {
8786
26.9k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8787
26.9k
           "StartTag: invalid element name\n");
8788
26.9k
        return(NULL);
8789
26.9k
    }
8790
800k
    localname = hlocalname.name;
8791
800k
    prefix = hprefix.name;
8792
8793
    /*
8794
     * Now parse the attributes, it ends up with the ending
8795
     *
8796
     * (S Attribute)* S?
8797
     */
8798
800k
    SKIP_BLANKS;
8799
800k
    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
923k
    while (((RAW != '>') &&
8826
409k
     ((RAW != '/') || (NXT(1) != '>')) &&
8827
296k
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8828
285k
  int len = -1;
8829
8830
285k
  hattname = xmlParseAttribute2(ctxt, prefix, localname,
8831
285k
                                          &haprefix, &attvalue, &len,
8832
285k
                                          &alloc);
8833
285k
        if (hattname.name == NULL)
8834
69.9k
      break;
8835
215k
        if (attvalue == NULL)
8836
18.2k
            goto next_attr;
8837
197k
        attname = hattname.name;
8838
197k
        aprefix = haprefix.name;
8839
197k
  if (len < 0) len = xmlStrlen(attvalue);
8840
8841
197k
        if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
8842
17.6k
            xmlHashedString huri;
8843
17.6k
            xmlURIPtr parsedUri;
8844
8845
17.6k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8846
17.6k
            uri = huri.name;
8847
17.6k
            if (uri == NULL) {
8848
1
                xmlErrMemory(ctxt);
8849
1
                goto next_attr;
8850
1
            }
8851
17.6k
            if (*uri != 0) {
8852
10.8k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8853
20
                    xmlErrMemory(ctxt);
8854
20
                    goto next_attr;
8855
20
                }
8856
10.7k
                if (parsedUri == NULL) {
8857
2.82k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8858
2.82k
                             "xmlns: '%s' is not a valid URI\n",
8859
2.82k
                                       uri, NULL, NULL);
8860
7.96k
                } else {
8861
7.96k
                    if (parsedUri->scheme == NULL) {
8862
4.76k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8863
4.76k
                                  "xmlns: URI %s is not absolute\n",
8864
4.76k
                                  uri, NULL, NULL);
8865
4.76k
                    }
8866
7.96k
                    xmlFreeURI(parsedUri);
8867
7.96k
                }
8868
10.7k
                if (uri == ctxt->str_xml_ns) {
8869
402
                    if (attname != ctxt->str_xml) {
8870
402
                        xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8871
402
                     "xml namespace URI cannot be the default namespace\n",
8872
402
                                 NULL, NULL, NULL);
8873
402
                    }
8874
402
                    goto next_attr;
8875
402
                }
8876
10.3k
                if ((len == 29) &&
8877
1.32k
                    (xmlStrEqual(uri,
8878
1.32k
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8879
449
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8880
449
                         "reuse of the xmlns namespace name is forbidden\n",
8881
449
                             NULL, NULL, NULL);
8882
449
                    goto next_attr;
8883
449
                }
8884
10.3k
            }
8885
8886
16.7k
            if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0)
8887
10.8k
                nbNs++;
8888
179k
        } else if (aprefix == ctxt->str_xmlns) {
8889
31.5k
            xmlHashedString huri;
8890
31.5k
            xmlURIPtr parsedUri;
8891
8892
31.5k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8893
31.5k
            uri = huri.name;
8894
31.5k
            if (uri == NULL) {
8895
1
                xmlErrMemory(ctxt);
8896
1
                goto next_attr;
8897
1
            }
8898
8899
31.5k
            if (attname == ctxt->str_xml) {
8900
272
                if (uri != ctxt->str_xml_ns) {
8901
202
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8902
202
                             "xml namespace prefix mapped to wrong URI\n",
8903
202
                             NULL, NULL, NULL);
8904
202
                }
8905
                /*
8906
                 * Do not keep a namespace definition node
8907
                 */
8908
272
                goto next_attr;
8909
272
            }
8910
31.2k
            if (uri == ctxt->str_xml_ns) {
8911
417
                if (attname != ctxt->str_xml) {
8912
417
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8913
417
                             "xml namespace URI mapped to wrong prefix\n",
8914
417
                             NULL, NULL, NULL);
8915
417
                }
8916
417
                goto next_attr;
8917
417
            }
8918
30.8k
            if (attname == ctxt->str_xmlns) {
8919
405
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8920
405
                         "redefinition of the xmlns prefix is forbidden\n",
8921
405
                         NULL, NULL, NULL);
8922
405
                goto next_attr;
8923
405
            }
8924
30.4k
            if ((len == 29) &&
8925
300
                (xmlStrEqual(uri,
8926
300
                             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
30.2k
            if ((uri == NULL) || (uri[0] == 0)) {
8933
232
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8934
232
                         "xmlns:%s: Empty XML namespace is not allowed\n",
8935
232
                              attname, NULL, NULL);
8936
232
                goto next_attr;
8937
30.0k
            } else {
8938
30.0k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8939
7
                    xmlErrMemory(ctxt);
8940
7
                    goto next_attr;
8941
7
                }
8942
30.0k
                if (parsedUri == NULL) {
8943
4.06k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8944
4.06k
                         "xmlns:%s: '%s' is not a valid URI\n",
8945
4.06k
                                       attname, uri, NULL);
8946
25.9k
                } else {
8947
25.9k
                    if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) {
8948
1.70k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8949
1.70k
                                  "xmlns:%s: URI %s is not absolute\n",
8950
1.70k
                                  attname, uri, NULL);
8951
1.70k
                    }
8952
25.9k
                    xmlFreeURI(parsedUri);
8953
25.9k
                }
8954
30.0k
            }
8955
8956
30.0k
            if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0)
8957
28.6k
                nbNs++;
8958
148k
        } else {
8959
            /*
8960
             * Populate attributes array, see above for repurposing
8961
             * of xmlChar pointers.
8962
             */
8963
148k
            if ((atts == NULL) || (nbatts + 5 > maxatts)) {
8964
19.4k
                int res = xmlCtxtGrowAttrs(ctxt);
8965
8966
19.4k
                maxatts = ctxt->maxatts;
8967
19.4k
                atts = ctxt->atts;
8968
8969
19.4k
                if (res < 0)
8970
7
                    goto next_attr;
8971
19.4k
            }
8972
148k
            ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) |
8973
148k
                                        ((unsigned) alloc << 31);
8974
148k
            atts[nbatts++] = attname;
8975
148k
            atts[nbatts++] = aprefix;
8976
148k
            atts[nbatts++] = XML_INT_TO_PTR(haprefix.hashValue);
8977
148k
            if (alloc) {
8978
25.4k
                atts[nbatts++] = attvalue;
8979
25.4k
                attvalue += len;
8980
25.4k
                atts[nbatts++] = attvalue;
8981
122k
            } 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
122k
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8988
122k
                attvalue += len;
8989
122k
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8990
122k
            }
8991
            /*
8992
             * tag if some deallocation is needed
8993
             */
8994
148k
            if (alloc != 0) attval = 1;
8995
148k
            attvalue = NULL; /* moved into atts */
8996
148k
        }
8997
8998
215k
next_attr:
8999
215k
        if ((attvalue != NULL) && (alloc != 0)) {
9000
5.19k
            xmlFree(attvalue);
9001
5.19k
            attvalue = NULL;
9002
5.19k
        }
9003
9004
215k
  GROW
9005
215k
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
9006
67.9k
      break;
9007
147k
  if (SKIP_BLANKS == 0) {
9008
24.2k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
9009
24.2k
         "attributes construct error\n");
9010
24.2k
      break;
9011
24.2k
  }
9012
123k
        GROW;
9013
123k
    }
9014
9015
    /*
9016
     * Namespaces from default attributes
9017
     */
9018
800k
    if (ctxt->attsDefault != NULL) {
9019
179k
        xmlDefAttrsPtr defaults;
9020
9021
179k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9022
179k
  if (defaults != NULL) {
9023
658k
      for (i = 0; i < defaults->nbAttrs; i++) {
9024
520k
                xmlDefAttr *attr = &defaults->attrs[i];
9025
9026
520k
          attname = attr->name.name;
9027
520k
    aprefix = attr->prefix.name;
9028
9029
520k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
9030
28.9k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9031
9032
28.9k
                    if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0)
9033
24.5k
                        nbNs++;
9034
491k
    } else if (aprefix == ctxt->str_xmlns) {
9035
212k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9036
9037
212k
                    if (xmlParserNsPush(ctxt, &attr->name, &attr->value,
9038
212k
                                      NULL, 1) > 0)
9039
205k
                        nbNs++;
9040
279k
    } else {
9041
279k
                    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
279k
                    nbTotalDef += 1;
9047
279k
                }
9048
520k
      }
9049
138k
  }
9050
179k
    }
9051
9052
    /*
9053
     * Resolve attribute namespaces
9054
     */
9055
948k
    for (i = 0; i < nbatts; i += 5) {
9056
148k
        attname = atts[i];
9057
148k
        aprefix = atts[i+1];
9058
9059
        /*
9060
  * The default namespace does not apply to attribute names.
9061
  */
9062
148k
  if (aprefix == NULL) {
9063
49.2k
            nsIndex = NS_INDEX_EMPTY;
9064
98.8k
        } else if (aprefix == ctxt->str_xml) {
9065
71.9k
            nsIndex = NS_INDEX_XML;
9066
71.9k
        } else {
9067
26.8k
            haprefix.name = aprefix;
9068
26.8k
            haprefix.hashValue = (size_t) atts[i+2];
9069
26.8k
            nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL);
9070
9071
26.8k
      if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) {
9072
16.3k
                xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9073
16.3k
        "Namespace prefix %s for %s on %s is not defined\n",
9074
16.3k
        aprefix, attname, localname);
9075
16.3k
                nsIndex = NS_INDEX_EMPTY;
9076
16.3k
            }
9077
26.8k
        }
9078
9079
148k
        atts[i+2] = XML_INT_TO_PTR(nsIndex);
9080
148k
    }
9081
9082
    /*
9083
     * Maximum number of attributes including default attributes.
9084
     */
9085
800k
    maxAtts = nratts + nbTotalDef;
9086
9087
    /*
9088
     * Verify that attribute names are unique.
9089
     */
9090
800k
    if (maxAtts > 1) {
9091
45.5k
        attrHashSize = 4;
9092
113k
        while (attrHashSize / 2 < (unsigned) maxAtts)
9093
67.7k
            attrHashSize *= 2;
9094
9095
45.5k
        if (attrHashSize > ctxt->attrHashMax) {
9096
6.00k
            xmlAttrHashBucket *tmp;
9097
9098
6.00k
            tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0]));
9099
6.00k
            if (tmp == NULL) {
9100
2
                xmlErrMemory(ctxt);
9101
2
                goto done;
9102
2
            }
9103
9104
6.00k
            ctxt->attrHash = tmp;
9105
6.00k
            ctxt->attrHashMax = attrHashSize;
9106
6.00k
        }
9107
9108
45.5k
        memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0]));
9109
9110
161k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9111
116k
            const xmlChar *nsuri;
9112
116k
            unsigned hashValue, nameHashValue, uriHashValue;
9113
116k
            int res;
9114
9115
116k
            attname = atts[i];
9116
116k
            aprefix = atts[i+1];
9117
116k
            nsIndex = XML_PTR_TO_INT(atts[i+2]);
9118
            /* Hash values always have bit 31 set, see dict.c */
9119
116k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9120
9121
116k
            if (nsIndex == NS_INDEX_EMPTY) {
9122
                /*
9123
                 * Prefix with empty namespace means an undeclared
9124
                 * prefix which was already reported above.
9125
                 */
9126
44.4k
                if (aprefix != NULL)
9127
13.6k
                    continue;
9128
30.7k
                nsuri = NULL;
9129
30.7k
                uriHashValue = URI_HASH_EMPTY;
9130
71.9k
            } else if (nsIndex == NS_INDEX_XML) {
9131
61.6k
                nsuri = ctxt->str_xml_ns;
9132
61.6k
                uriHashValue = URI_HASH_XML;
9133
61.6k
            } else {
9134
10.3k
                nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9135
10.3k
                uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9136
10.3k
            }
9137
9138
102k
            hashValue = xmlDictCombineHash(nameHashValue, uriHashValue);
9139
102k
            res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9140
102k
                                    hashValue, i);
9141
102k
            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
102k
            if (res < INT_MAX) {
9151
71.1k
                if (aprefix == atts[res+1]) {
9152
67.3k
                    xmlErrAttributeDup(ctxt, aprefix, attname);
9153
67.3k
                    numDupErr += 1;
9154
67.3k
                } else {
9155
3.83k
                    xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9156
3.83k
                             "Namespaced Attribute %s in '%s' redefined\n",
9157
3.83k
                             attname, nsuri, NULL);
9158
3.83k
                    numNsErr += 1;
9159
3.83k
                }
9160
71.1k
            }
9161
102k
        }
9162
45.5k
    }
9163
9164
    /*
9165
     * Default attributes
9166
     */
9167
800k
    if (ctxt->attsDefault != NULL) {
9168
179k
        xmlDefAttrsPtr defaults;
9169
9170
179k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9171
179k
  if (defaults != NULL) {
9172
658k
      for (i = 0; i < defaults->nbAttrs; i++) {
9173
520k
                xmlDefAttr *attr = &defaults->attrs[i];
9174
520k
                const xmlChar *nsuri = NULL;
9175
520k
                unsigned hashValue, uriHashValue = 0;
9176
520k
                int res;
9177
9178
520k
          attname = attr->name.name;
9179
520k
    aprefix = attr->prefix.name;
9180
9181
520k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL))
9182
28.9k
                    continue;
9183
491k
    if (aprefix == ctxt->str_xmlns)
9184
212k
                    continue;
9185
9186
279k
                if (aprefix == NULL) {
9187
133k
                    nsIndex = NS_INDEX_EMPTY;
9188
133k
                    nsuri = NULL;
9189
133k
                    uriHashValue = URI_HASH_EMPTY;
9190
145k
                } else if (aprefix == ctxt->str_xml) {
9191
38.5k
                    nsIndex = NS_INDEX_XML;
9192
38.5k
                    nsuri = ctxt->str_xml_ns;
9193
38.5k
                    uriHashValue = URI_HASH_XML;
9194
106k
                } else {
9195
106k
                    nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL);
9196
106k
                    if ((nsIndex == INT_MAX) ||
9197
91.2k
                        (nsIndex < ctxt->nsdb->minNsIndex)) {
9198
91.2k
                        xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9199
91.2k
                                 "Namespace prefix %s for %s on %s is not "
9200
91.2k
                                 "defined\n",
9201
91.2k
                                 aprefix, attname, localname);
9202
91.2k
                        nsIndex = NS_INDEX_EMPTY;
9203
91.2k
                        nsuri = NULL;
9204
91.2k
                        uriHashValue = URI_HASH_EMPTY;
9205
91.2k
                    } else {
9206
15.6k
                        nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9207
15.6k
                        uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9208
15.6k
                    }
9209
106k
                }
9210
9211
                /*
9212
                 * Check whether the attribute exists
9213
                 */
9214
279k
                if (maxAtts > 1) {
9215
247k
                    hashValue = xmlDictCombineHash(attr->name.hashValue,
9216
247k
                                                   uriHashValue);
9217
247k
                    res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9218
247k
                                            hashValue, nbatts);
9219
247k
                    if (res < 0)
9220
0
                        continue;
9221
247k
                    if (res < INT_MAX) {
9222
9.28k
                        if (aprefix == atts[res+1])
9223
1.04k
                            continue;
9224
8.23k
                        xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9225
8.23k
                                 "Namespaced Attribute %s in '%s' redefined\n",
9226
8.23k
                                 attname, nsuri, NULL);
9227
8.23k
                    }
9228
247k
                }
9229
9230
278k
                xmlParserEntityCheck(ctxt, attr->expandedSize);
9231
9232
278k
                if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9233
2.06k
                    res = xmlCtxtGrowAttrs(ctxt);
9234
9235
2.06k
                    maxatts = ctxt->maxatts;
9236
2.06k
                    atts = ctxt->atts;
9237
9238
2.06k
                    if (res < 0) {
9239
5
                        localname = NULL;
9240
5
                        goto done;
9241
5
                    }
9242
2.06k
                }
9243
9244
278k
                atts[nbatts++] = attname;
9245
278k
                atts[nbatts++] = aprefix;
9246
278k
                atts[nbatts++] = XML_INT_TO_PTR(nsIndex);
9247
278k
                atts[nbatts++] = attr->value.name;
9248
278k
                atts[nbatts++] = attr->valueEnd;
9249
9250
278k
#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
278k
                if ((ctxt->validate) &&
9256
142k
                    (ctxt->standalone == 1) &&
9257
750
                    (attr->external != 0)) {
9258
548
                    xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED,
9259
548
                            "standalone: attribute %s on %s defaulted "
9260
548
                            "from external subset\n",
9261
548
                            attname, localname);
9262
548
                }
9263
278k
#endif
9264
278k
                nbdef++;
9265
278k
      }
9266
138k
  }
9267
179k
    }
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
800k
    if ((numDupErr == 0) && (numNsErr > 1)) {
9282
848
        memset(ctxt->attrHash, -1,
9283
848
               attrHashSize * sizeof(ctxt->attrHash[0]));
9284
9285
6.92k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9286
6.07k
            unsigned hashValue, nameHashValue, prefixHashValue;
9287
6.07k
            int res;
9288
9289
6.07k
            aprefix = atts[i+1];
9290
6.07k
            if (aprefix == NULL)
9291
282
                continue;
9292
9293
5.79k
            attname = atts[i];
9294
            /* Hash values always have bit 31 set, see dict.c */
9295
5.79k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9296
5.79k
            prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix);
9297
9298
5.79k
            hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue);
9299
5.79k
            res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname,
9300
5.79k
                                         aprefix, hashValue, i);
9301
5.79k
            if (res < INT_MAX)
9302
2.94k
                xmlErrAttributeDup(ctxt, aprefix, attname);
9303
5.79k
        }
9304
848
    }
9305
9306
    /*
9307
     * Reconstruct attribute pointers
9308
     */
9309
1.22M
    for (i = 0, j = 0; i < nbatts; i += 5, j++) {
9310
        /* namespace URI */
9311
426k
        nsIndex = XML_PTR_TO_INT(atts[i+2]);
9312
426k
        if (nsIndex == INT_MAX)
9313
289k
            atts[i+2] = NULL;
9314
136k
        else if (nsIndex == INT_MAX - 1)
9315
110k
            atts[i+2] = ctxt->str_xml_ns;
9316
26.2k
        else
9317
26.2k
            atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1];
9318
9319
426k
        if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) {
9320
122k
            atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]);  /* value */
9321
122k
            atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]);  /* valuend */
9322
122k
        }
9323
426k
    }
9324
9325
800k
    uri = xmlParserNsLookupUri(ctxt, &hprefix);
9326
800k
    if ((prefix != NULL) && (uri == NULL)) {
9327
25.3k
  xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9328
25.3k
           "Namespace prefix %s on %s is not defined\n",
9329
25.3k
     prefix, localname, NULL);
9330
25.3k
    }
9331
800k
    *pref = prefix;
9332
800k
    *URI = uri;
9333
9334
    /*
9335
     * SAX callback
9336
     */
9337
800k
    if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) &&
9338
800k
  (!ctxt->disableSAX)) {
9339
561k
  if (nbNs > 0)
9340
63.1k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9341
63.1k
                          nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs),
9342
63.1k
        nbatts / 5, nbdef, atts);
9343
497k
  else
9344
497k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9345
497k
                          0, NULL, nbatts / 5, nbdef, atts);
9346
561k
    }
9347
9348
800k
done:
9349
    /*
9350
     * Free allocated attribute values
9351
     */
9352
800k
    if (attval != 0) {
9353
96.6k
  for (i = 0, j = 0; j < nratts; i += 5, j++)
9354
84.8k
      if (ctxt->attallocs[j] & 0x80000000)
9355
25.4k
          xmlFree((xmlChar *) atts[i+3]);
9356
11.7k
    }
9357
9358
800k
    *nbNsPtr = nbNs;
9359
800k
    return(localname);
9360
800k
}
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
418k
xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) {
9376
418k
    const xmlChar *name;
9377
9378
418k
    GROW;
9379
418k
    if ((RAW != '<') || (NXT(1) != '/')) {
9380
48
  xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL);
9381
48
  return;
9382
48
    }
9383
418k
    SKIP(2);
9384
9385
418k
    if (tag->prefix == NULL)
9386
406k
        name = xmlParseNameAndCompare(ctxt, ctxt->name);
9387
12.4k
    else
9388
12.4k
        name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix);
9389
9390
    /*
9391
     * We should definitely be at the ending "S? '>'" part
9392
     */
9393
418k
    GROW;
9394
418k
    SKIP_BLANKS;
9395
418k
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
9396
17.6k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
9397
17.6k
    } else
9398
401k
  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
418k
    if (name != (xmlChar*)1) {
9407
11.9k
        if (name == NULL) name = BAD_CAST "unparsable";
9408
11.9k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
9409
11.9k
         "Opening and ending tag mismatch: %s line %d and %s\n",
9410
11.9k
                    ctxt->name, tag->line, name);
9411
11.9k
    }
9412
9413
    /*
9414
     * SAX: End of Tag
9415
     */
9416
418k
    if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9417
418k
  (!ctxt->disableSAX))
9418
271k
  ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix,
9419
271k
                                tag->URI);
9420
9421
418k
    spacePop(ctxt);
9422
418k
    if (tag->nsNr != 0)
9423
20.2k
  xmlParserNsPop(ctxt, tag->nsNr);
9424
418k
}
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
39.3k
xmlParseCDSect(xmlParserCtxt *ctxt) {
9442
39.3k
    xmlChar *buf = NULL;
9443
39.3k
    int len = 0;
9444
39.3k
    int size = XML_PARSER_BUFFER_SIZE;
9445
39.3k
    int r, rl;
9446
39.3k
    int s, sl;
9447
39.3k
    int cur, l;
9448
39.3k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9449
19.5k
                    XML_MAX_HUGE_LENGTH :
9450
39.3k
                    XML_MAX_TEXT_LENGTH;
9451
9452
39.3k
    if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '['))
9453
0
        return;
9454
39.3k
    SKIP(3);
9455
9456
39.3k
    if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '['))
9457
0
        return;
9458
39.3k
    SKIP(6);
9459
9460
39.3k
    r = xmlCurrentCharRecover(ctxt, &rl);
9461
39.3k
    if (!IS_CHAR(r)) {
9462
345
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9463
345
        goto out;
9464
345
    }
9465
39.0k
    NEXTL(rl);
9466
39.0k
    s = xmlCurrentCharRecover(ctxt, &sl);
9467
39.0k
    if (!IS_CHAR(s)) {
9468
1.10k
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9469
1.10k
        goto out;
9470
1.10k
    }
9471
37.9k
    NEXTL(sl);
9472
37.9k
    cur = xmlCurrentCharRecover(ctxt, &l);
9473
37.9k
    buf = xmlMalloc(size);
9474
37.9k
    if (buf == NULL) {
9475
3
  xmlErrMemory(ctxt);
9476
3
        goto out;
9477
3
    }
9478
2.66M
    while (IS_CHAR(cur) &&
9479
2.65M
           ((r != ']') || (s != ']') || (cur != '>'))) {
9480
2.62M
  if (len + 5 >= size) {
9481
1.47k
      xmlChar *tmp;
9482
1.47k
            int newSize;
9483
9484
1.47k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9485
1.47k
            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
1.47k
      tmp = xmlRealloc(buf, newSize);
9491
1.47k
      if (tmp == NULL) {
9492
1
    xmlErrMemory(ctxt);
9493
1
                goto out;
9494
1
      }
9495
1.46k
      buf = tmp;
9496
1.46k
      size = newSize;
9497
1.46k
  }
9498
2.62M
  COPY_BUF(buf, len, r);
9499
2.62M
  r = s;
9500
2.62M
  rl = sl;
9501
2.62M
  s = cur;
9502
2.62M
  sl = l;
9503
2.62M
  NEXTL(l);
9504
2.62M
  cur = xmlCurrentCharRecover(ctxt, &l);
9505
2.62M
    }
9506
37.9k
    buf[len] = 0;
9507
37.9k
    if (cur != '>') {
9508
2.95k
  xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9509
2.95k
                       "CData section not finished\n%.50s\n", buf);
9510
2.95k
        goto out;
9511
2.95k
    }
9512
34.9k
    NEXTL(l);
9513
9514
    /*
9515
     * OK the buffer is to be consumed as cdata.
9516
     */
9517
34.9k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
9518
34.2k
        if ((ctxt->sax->cdataBlock != NULL) &&
9519
34.2k
            ((ctxt->options & XML_PARSE_NOCDATA) == 0)) {
9520
27.8k
            ctxt->sax->cdataBlock(ctxt->userData, buf, len);
9521
27.8k
        } else if (ctxt->sax->characters != NULL) {
9522
6.42k
            ctxt->sax->characters(ctxt->userData, buf, len);
9523
6.42k
        }
9524
34.2k
    }
9525
9526
39.3k
out:
9527
39.3k
    xmlFree(buf);
9528
39.3k
}
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
9.46k
xmlParseContentInternal(xmlParserCtxtPtr ctxt) {
9539
9.46k
    int oldNameNr = ctxt->nameNr;
9540
9.46k
    int oldSpaceNr = ctxt->spaceNr;
9541
9.46k
    int oldNodeNr = ctxt->nodeNr;
9542
9543
9.46k
    GROW;
9544
7.45M
    while ((ctxt->input->cur < ctxt->input->end) &&
9545
7.44M
     (PARSER_STOPPED(ctxt) == 0)) {
9546
7.44M
  const xmlChar *cur = ctxt->input->cur;
9547
9548
  /*
9549
   * First case : a Processing Instruction.
9550
   */
9551
7.44M
  if ((*cur == '<') && (cur[1] == '?')) {
9552
35.9k
      xmlParsePI(ctxt);
9553
35.9k
  }
9554
9555
  /*
9556
   * Second case : a CDSection
9557
   */
9558
  /* 2.6.0 test was *cur not RAW */
9559
7.40M
  else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) {
9560
25.7k
      xmlParseCDSect(ctxt);
9561
25.7k
  }
9562
9563
  /*
9564
   * Third case :  a comment
9565
   */
9566
7.38M
  else if ((*cur == '<') && (NXT(1) == '!') &&
9567
26.6k
     (NXT(2) == '-') && (NXT(3) == '-')) {
9568
3.16k
      xmlParseComment(ctxt);
9569
3.16k
  }
9570
9571
  /*
9572
   * Fourth case :  a sub-element.
9573
   */
9574
7.38M
  else if (*cur == '<') {
9575
1.24M
            if (NXT(1) == '/') {
9576
455k
                if (ctxt->nameNr <= oldNameNr)
9577
3.76k
                    break;
9578
451k
          xmlParseElementEnd(ctxt);
9579
791k
            } else {
9580
791k
          xmlParseElementStart(ctxt);
9581
791k
            }
9582
1.24M
  }
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
6.13M
  else if (*cur == '&') {
9590
70.9k
      xmlParseReference(ctxt);
9591
70.9k
  }
9592
9593
  /*
9594
   * Last case, text. Note that References are handled directly.
9595
   */
9596
6.06M
  else {
9597
6.06M
      xmlParseCharDataInternal(ctxt, 0);
9598
6.06M
  }
9599
9600
7.44M
  SHRINK;
9601
7.44M
  GROW;
9602
7.44M
    }
9603
9604
9.46k
    if ((ctxt->nameNr > oldNameNr) &&
9605
1.37k
        (ctxt->input->cur >= ctxt->input->end) &&
9606
1.19k
        (ctxt->wellFormed)) {
9607
35
        const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9608
35
        int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9609
35
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9610
35
                "Premature end of data in tag %s line %d\n",
9611
35
                name, line, NULL);
9612
35
    }
9613
9614
    /*
9615
     * Clean up in error case
9616
     */
9617
9618
79.2k
    while (ctxt->nodeNr > oldNodeNr)
9619
69.7k
        nodePop(ctxt);
9620
9621
109k
    while (ctxt->nameNr > oldNameNr) {
9622
99.9k
        xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1];
9623
9624
99.9k
        if (tag->nsNr != 0)
9625
16.0k
            xmlParserNsPop(ctxt, tag->nsNr);
9626
9627
99.9k
        namePop(ctxt);
9628
99.9k
    }
9629
9630
109k
    while (ctxt->spaceNr > oldSpaceNr)
9631
99.9k
        spacePop(ctxt);
9632
9.46k
}
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
7.21k
xmlParseElement(xmlParserCtxt *ctxt) {
9669
7.21k
    if (xmlParseElementStart(ctxt) != 0)
9670
3.00k
        return;
9671
9672
4.20k
    xmlParseContentInternal(ctxt);
9673
9674
4.20k
    if (ctxt->input->cur >= ctxt->input->end) {
9675
383
        if (ctxt->wellFormed) {
9676
2
            const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9677
2
            int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9678
2
            xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9679
2
                    "Premature end of data in tag %s line %d\n",
9680
2
                    name, line, NULL);
9681
2
        }
9682
383
        return;
9683
383
    }
9684
9685
3.82k
    xmlParseElementEnd(ctxt);
9686
3.82k
}
9687
9688
/**
9689
 * Parse the start of an XML element. Returns -1 in case of error, 0 if an
9690
 * opening tag was parsed, 1 if an empty element was parsed.
9691
 *
9692
 * Always consumes '<'.
9693
 *
9694
 * @param ctxt  an XML parser context
9695
 */
9696
static int
9697
798k
xmlParseElementStart(xmlParserCtxtPtr ctxt) {
9698
798k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
9699
798k
    const xmlChar *name;
9700
798k
    const xmlChar *prefix = NULL;
9701
798k
    const xmlChar *URI = NULL;
9702
798k
    xmlParserNodeInfo node_info;
9703
798k
    int line;
9704
798k
    xmlNodePtr cur;
9705
798k
    int nbNs = 0;
9706
9707
798k
    if (ctxt->nameNr > maxDepth) {
9708
15
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
9709
15
                "Excessive depth in document: %d use XML_PARSE_HUGE option\n",
9710
15
                ctxt->nameNr);
9711
15
  return(-1);
9712
15
    }
9713
9714
    /* Capture start position */
9715
798k
    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
798k
    if (ctxt->spaceNr == 0)
9722
0
  spacePush(ctxt, -1);
9723
798k
    else if (*ctxt->space == -2)
9724
138k
  spacePush(ctxt, -1);
9725
660k
    else
9726
660k
  spacePush(ctxt, *ctxt->space);
9727
9728
798k
    line = ctxt->input->line;
9729
798k
#ifdef LIBXML_SAX1_ENABLED
9730
798k
    if (ctxt->sax2)
9731
605k
#endif /* LIBXML_SAX1_ENABLED */
9732
605k
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
9733
192k
#ifdef LIBXML_SAX1_ENABLED
9734
192k
    else
9735
192k
  name = xmlParseStartTag(ctxt);
9736
798k
#endif /* LIBXML_SAX1_ENABLED */
9737
798k
    if (name == NULL) {
9738
42.4k
  spacePop(ctxt);
9739
42.4k
        return(-1);
9740
42.4k
    }
9741
755k
    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
9742
755k
    cur = ctxt->node;
9743
9744
755k
#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
755k
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
9751
8.11k
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
9752
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
9753
755k
#endif /* LIBXML_VALID_ENABLED */
9754
9755
    /*
9756
     * Check for an Empty Element.
9757
     */
9758
755k
    if ((RAW == '/') && (NXT(1) == '>')) {
9759
96.5k
        SKIP(2);
9760
96.5k
  if (ctxt->sax2) {
9761
55.6k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9762
55.6k
    (!ctxt->disableSAX))
9763
41.9k
    ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI);
9764
55.6k
#ifdef LIBXML_SAX1_ENABLED
9765
55.6k
  } else {
9766
40.9k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
9767
40.9k
    (!ctxt->disableSAX))
9768
40.3k
    ctxt->sax->endElement(ctxt->userData, name);
9769
40.9k
#endif /* LIBXML_SAX1_ENABLED */
9770
40.9k
  }
9771
96.5k
  namePop(ctxt);
9772
96.5k
  spacePop(ctxt);
9773
96.5k
  if (nbNs > 0)
9774
10.8k
      xmlParserNsPop(ctxt, nbNs);
9775
96.5k
  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
96.5k
  return(1);
9783
96.5k
    }
9784
659k
    if (RAW == '>') {
9785
555k
        NEXT1;
9786
555k
        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
555k
    } else {
9793
103k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED,
9794
103k
         "Couldn't find end of Start Tag %s line %d\n",
9795
103k
                    name, line, NULL);
9796
9797
  /*
9798
   * end of parsing of this node.
9799
   */
9800
103k
  nodePop(ctxt);
9801
103k
  namePop(ctxt);
9802
103k
  spacePop(ctxt);
9803
103k
  if (nbNs > 0)
9804
42.0k
      xmlParserNsPop(ctxt, nbNs);
9805
103k
  return(-1);
9806
103k
    }
9807
9808
555k
    return(0);
9809
659k
}
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
455k
xmlParseElementEnd(xmlParserCtxtPtr ctxt) {
9818
455k
    xmlNodePtr cur = ctxt->node;
9819
9820
455k
    if (ctxt->nameNr <= 0) {
9821
5
        if ((RAW == '<') && (NXT(1) == '/'))
9822
2
            SKIP(2);
9823
5
        return;
9824
5
    }
9825
9826
    /*
9827
     * parse the end of tag: '</' should be here.
9828
     */
9829
455k
    if (ctxt->sax2) {
9830
386k
  xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
9831
386k
  namePop(ctxt);
9832
386k
    }
9833
68.6k
#ifdef LIBXML_SAX1_ENABLED
9834
68.6k
    else
9835
68.6k
  xmlParseEndTag1(ctxt, 0);
9836
455k
#endif /* LIBXML_SAX1_ENABLED */
9837
9838
    /*
9839
     * Capture end position
9840
     */
9841
455k
    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
455k
}
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
13.9k
xmlParseVersionNum(xmlParserCtxt *ctxt) {
9867
13.9k
    xmlChar *buf = NULL;
9868
13.9k
    int len = 0;
9869
13.9k
    int size = 10;
9870
13.9k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9871
4.76k
                    XML_MAX_TEXT_LENGTH :
9872
13.9k
                    XML_MAX_NAME_LENGTH;
9873
13.9k
    xmlChar cur;
9874
9875
13.9k
    buf = xmlMalloc(size);
9876
13.9k
    if (buf == NULL) {
9877
3
  xmlErrMemory(ctxt);
9878
3
  return(NULL);
9879
3
    }
9880
13.9k
    cur = CUR;
9881
13.9k
    if (!((cur >= '0') && (cur <= '9'))) {
9882
528
  xmlFree(buf);
9883
528
  return(NULL);
9884
528
    }
9885
13.3k
    buf[len++] = cur;
9886
13.3k
    NEXT;
9887
13.3k
    cur=CUR;
9888
13.3k
    if (cur != '.') {
9889
2.91k
  xmlFree(buf);
9890
2.91k
  return(NULL);
9891
2.91k
    }
9892
10.4k
    buf[len++] = cur;
9893
10.4k
    NEXT;
9894
10.4k
    cur=CUR;
9895
637k
    while ((cur >= '0') && (cur <= '9')) {
9896
627k
  if (len + 1 >= size) {
9897
1.50k
      xmlChar *tmp;
9898
1.50k
            int newSize;
9899
9900
1.50k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9901
1.50k
            if (newSize < 0) {
9902
1
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum");
9903
1
                xmlFree(buf);
9904
1
                return(NULL);
9905
1
            }
9906
1.50k
      tmp = xmlRealloc(buf, newSize);
9907
1.50k
      if (tmp == NULL) {
9908
2
    xmlErrMemory(ctxt);
9909
2
          xmlFree(buf);
9910
2
    return(NULL);
9911
2
      }
9912
1.50k
      buf = tmp;
9913
1.50k
            size = newSize;
9914
1.50k
  }
9915
627k
  buf[len++] = cur;
9916
627k
  NEXT;
9917
627k
  cur=CUR;
9918
627k
    }
9919
10.4k
    buf[len] = 0;
9920
10.4k
    return(buf);
9921
10.4k
}
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
150k
xmlParseVersionInfo(xmlParserCtxt *ctxt) {
9938
150k
    xmlChar *version = NULL;
9939
9940
150k
    if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) {
9941
14.9k
  SKIP(7);
9942
14.9k
  SKIP_BLANKS;
9943
14.9k
  if (RAW != '=') {
9944
660
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
9945
660
      return(NULL);
9946
660
        }
9947
14.2k
  NEXT;
9948
14.2k
  SKIP_BLANKS;
9949
14.2k
  if (RAW == '"') {
9950
2.84k
      NEXT;
9951
2.84k
      version = xmlParseVersionNum(ctxt);
9952
2.84k
      if (RAW != '"') {
9953
252
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9954
252
      } else
9955
2.59k
          NEXT;
9956
11.4k
  } else if (RAW == '\''){
9957
11.0k
      NEXT;
9958
11.0k
      version = xmlParseVersionNum(ctxt);
9959
11.0k
      if (RAW != '\'') {
9960
3.39k
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9961
3.39k
      } else
9962
7.66k
          NEXT;
9963
11.0k
  } else {
9964
379
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
9965
379
  }
9966
14.2k
    }
9967
150k
    return(version);
9968
150k
}
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
8.18k
xmlParseEncName(xmlParserCtxt *ctxt) {
9982
8.18k
    xmlChar *buf = NULL;
9983
8.18k
    int len = 0;
9984
8.18k
    int size = 10;
9985
8.18k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9986
2.75k
                    XML_MAX_TEXT_LENGTH :
9987
8.18k
                    XML_MAX_NAME_LENGTH;
9988
8.18k
    xmlChar cur;
9989
9990
8.18k
    cur = CUR;
9991
8.18k
    if (((cur >= 'a') && (cur <= 'z')) ||
9992
7.59k
        ((cur >= 'A') && (cur <= 'Z'))) {
9993
7.59k
  buf = xmlMalloc(size);
9994
7.59k
  if (buf == NULL) {
9995
4
      xmlErrMemory(ctxt);
9996
4
      return(NULL);
9997
4
  }
9998
9999
7.58k
  buf[len++] = cur;
10000
7.58k
  NEXT;
10001
7.58k
  cur = CUR;
10002
159k
  while (((cur >= 'a') && (cur <= 'z')) ||
10003
106k
         ((cur >= 'A') && (cur <= 'Z')) ||
10004
39.4k
         ((cur >= '0') && (cur <= '9')) ||
10005
17.9k
         (cur == '.') || (cur == '_') ||
10006
151k
         (cur == '-')) {
10007
151k
      if (len + 1 >= size) {
10008
4.15k
          xmlChar *tmp;
10009
4.15k
                int newSize;
10010
10011
4.15k
                newSize = xmlGrowCapacity(size, 1, 1, maxLength);
10012
4.15k
                if (newSize < 0) {
10013
1
                    xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName");
10014
1
                    xmlFree(buf);
10015
1
                    return(NULL);
10016
1
                }
10017
4.15k
    tmp = xmlRealloc(buf, newSize);
10018
4.15k
    if (tmp == NULL) {
10019
4
        xmlErrMemory(ctxt);
10020
4
        xmlFree(buf);
10021
4
        return(NULL);
10022
4
    }
10023
4.14k
    buf = tmp;
10024
4.14k
                size = newSize;
10025
4.14k
      }
10026
151k
      buf[len++] = cur;
10027
151k
      NEXT;
10028
151k
      cur = CUR;
10029
151k
        }
10030
7.58k
  buf[len] = 0;
10031
7.58k
    } else {
10032
594
  xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL);
10033
594
    }
10034
8.17k
    return(buf);
10035
8.18k
}
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
149k
xmlParseEncodingDecl(xmlParserCtxt *ctxt) {
10053
149k
    xmlChar *encoding = NULL;
10054
10055
149k
    SKIP_BLANKS;
10056
149k
    if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0)
10057
139k
        return(NULL);
10058
10059
9.20k
    SKIP(8);
10060
9.20k
    SKIP_BLANKS;
10061
9.20k
    if (RAW != '=') {
10062
558
        xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10063
558
        return(NULL);
10064
558
    }
10065
8.65k
    NEXT;
10066
8.65k
    SKIP_BLANKS;
10067
8.65k
    if (RAW == '"') {
10068
1.56k
        NEXT;
10069
1.56k
        encoding = xmlParseEncName(ctxt);
10070
1.56k
        if (RAW != '"') {
10071
485
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10072
485
            xmlFree(encoding);
10073
485
            return(NULL);
10074
485
        } else
10075
1.07k
            NEXT;
10076
7.08k
    } else if (RAW == '\''){
10077
6.62k
        NEXT;
10078
6.62k
        encoding = xmlParseEncName(ctxt);
10079
6.62k
        if (RAW != '\'') {
10080
979
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10081
979
            xmlFree(encoding);
10082
979
            return(NULL);
10083
979
        } else
10084
5.64k
            NEXT;
10085
6.62k
    } else {
10086
465
        xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10087
465
    }
10088
10089
7.18k
    if (encoding == NULL)
10090
474
        return(NULL);
10091
10092
6.71k
    xmlSetDeclaredEncoding(ctxt, encoding);
10093
10094
6.71k
    return(ctxt->encoding);
10095
7.18k
}
10096
10097
/**
10098
 * Parse the XML standalone declaration
10099
 *
10100
 * @deprecated Internal function, don't use.
10101
 *
10102
 *     [32] SDDecl ::= S 'standalone' Eq
10103
 *                     (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"'))
10104
 *
10105
 * [ VC: Standalone Document Declaration ]
10106
 * TODO The standalone document declaration must have the value "no"
10107
 * if any external markup declarations contain declarations of:
10108
 *  - attributes with default values, if elements to which these
10109
 *    attributes apply appear in the document without specifications
10110
 *    of values for these attributes, or
10111
 *  - entities (other than amp, lt, gt, apos, quot), if references
10112
 *    to those entities appear in the document, or
10113
 *  - attributes with values subject to normalization, where the
10114
 *    attribute appears in the document with a value which will change
10115
 *    as a result of normalization, or
10116
 *  - element types with element content, if white space occurs directly
10117
 *    within any instance of those types.
10118
 *
10119
 * @param ctxt  an XML parser context
10120
 * @returns
10121
 *   1 if standalone="yes"
10122
 *   0 if standalone="no"
10123
 *  -2 if standalone attribute is missing or invalid
10124
 *    (A standalone value of -2 means that the XML declaration was found,
10125
 *     but no value was specified for the standalone attribute).
10126
 */
10127
10128
int
10129
1.20k
xmlParseSDDecl(xmlParserCtxt *ctxt) {
10130
1.20k
    int standalone = -2;
10131
10132
1.20k
    SKIP_BLANKS;
10133
1.20k
    if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) {
10134
267
  SKIP(10);
10135
267
        SKIP_BLANKS;
10136
267
  if (RAW != '=') {
10137
1
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10138
1
      return(standalone);
10139
1
        }
10140
266
  NEXT;
10141
266
  SKIP_BLANKS;
10142
266
        if (RAW == '\''){
10143
257
      NEXT;
10144
257
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10145
9
          standalone = 0;
10146
9
                SKIP(2);
10147
248
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10148
243
                 (NXT(2) == 's')) {
10149
242
          standalone = 1;
10150
242
    SKIP(3);
10151
242
            } else {
10152
6
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10153
6
      }
10154
257
      if (RAW != '\'') {
10155
13
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10156
13
      } else
10157
244
          NEXT;
10158
257
  } else if (RAW == '"'){
10159
8
      NEXT;
10160
8
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10161
2
          standalone = 0;
10162
2
    SKIP(2);
10163
6
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10164
3
                 (NXT(2) == 's')) {
10165
2
          standalone = 1;
10166
2
                SKIP(3);
10167
4
            } else {
10168
4
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10169
4
      }
10170
8
      if (RAW != '"') {
10171
7
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10172
7
      } else
10173
1
          NEXT;
10174
8
  } else {
10175
1
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10176
1
        }
10177
266
    }
10178
1.20k
    return(standalone);
10179
1.20k
}
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
3.44k
xmlParseXMLDecl(xmlParserCtxt *ctxt) {
10192
3.44k
    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
3.44k
    ctxt->standalone = -2;
10201
10202
    /*
10203
     * We know that '<?xml' is here.
10204
     */
10205
3.44k
    SKIP(5);
10206
10207
3.44k
    if (!IS_BLANK_CH(RAW)) {
10208
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
10209
0
                 "Blank needed after '<?xml'\n");
10210
0
    }
10211
3.44k
    SKIP_BLANKS;
10212
10213
    /*
10214
     * We must have the VersionInfo here.
10215
     */
10216
3.44k
    version = xmlParseVersionInfo(ctxt);
10217
3.44k
    if (version == NULL) {
10218
825
  xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL);
10219
2.61k
    } else {
10220
2.61k
  if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) {
10221
      /*
10222
       * Changed here for XML-1.0 5th edition
10223
       */
10224
195
      if (ctxt->options & XML_PARSE_OLD10) {
10225
29
    xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10226
29
                "Unsupported version '%s'\n",
10227
29
                version);
10228
166
      } else {
10229
166
          if ((version[0] == '1') && ((version[1] == '.'))) {
10230
130
        xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION,
10231
130
                      "Unsupported version '%s'\n",
10232
130
          version, NULL);
10233
130
    } else {
10234
36
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10235
36
              "Unsupported version '%s'\n",
10236
36
              version);
10237
36
    }
10238
166
      }
10239
195
  }
10240
2.61k
  if (ctxt->version != NULL)
10241
0
      xmlFree(ctxt->version);
10242
2.61k
  ctxt->version = version;
10243
2.61k
    }
10244
10245
    /*
10246
     * We may have the encoding declaration
10247
     */
10248
3.44k
    if (!IS_BLANK_CH(RAW)) {
10249
2.34k
        if ((RAW == '?') && (NXT(1) == '>')) {
10250
1.48k
      SKIP(2);
10251
1.48k
      return;
10252
1.48k
  }
10253
861
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10254
861
    }
10255
1.95k
    xmlParseEncodingDecl(ctxt);
10256
10257
    /*
10258
     * We may have the standalone status.
10259
     */
10260
1.95k
    if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) {
10261
814
        if ((RAW == '?') && (NXT(1) == '>')) {
10262
754
      SKIP(2);
10263
754
      return;
10264
754
  }
10265
60
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10266
60
    }
10267
10268
    /*
10269
     * We can grow the input buffer freely at that point
10270
     */
10271
1.20k
    GROW;
10272
10273
1.20k
    SKIP_BLANKS;
10274
1.20k
    ctxt->standalone = xmlParseSDDecl(ctxt);
10275
10276
1.20k
    SKIP_BLANKS;
10277
1.20k
    if ((RAW == '?') && (NXT(1) == '>')) {
10278
302
        SKIP(2);
10279
902
    } else if (RAW == '>') {
10280
        /* Deprecated old WD ... */
10281
29
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10282
29
  NEXT;
10283
873
    } else {
10284
873
        int c;
10285
10286
873
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10287
497k
        while ((PARSER_STOPPED(ctxt) == 0) &&
10288
497k
               ((c = CUR) != 0)) {
10289
496k
            NEXT;
10290
496k
            if (c == '>')
10291
509
                break;
10292
496k
        }
10293
873
    }
10294
1.20k
}
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
16.8k
xmlParseMisc(xmlParserCtxt *ctxt) {
10335
18.7k
    while (PARSER_STOPPED(ctxt) == 0) {
10336
18.5k
        SKIP_BLANKS;
10337
18.5k
        GROW;
10338
18.5k
        if ((RAW == '<') && (NXT(1) == '?')) {
10339
731
      xmlParsePI(ctxt);
10340
17.8k
        } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) {
10341
1.12k
      xmlParseComment(ctxt);
10342
16.7k
        } else {
10343
16.7k
            break;
10344
16.7k
        }
10345
18.5k
    }
10346
16.8k
}
10347
10348
static void
10349
22.8k
xmlFinishDocument(xmlParserCtxtPtr ctxt) {
10350
22.8k
    xmlDocPtr doc;
10351
10352
    /*
10353
     * SAX: end of the document processing.
10354
     */
10355
22.8k
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10356
22.8k
        ctxt->sax->endDocument(ctxt->userData);
10357
10358
    /*
10359
     * Remove locally kept entity definitions if the tree was not built
10360
     */
10361
22.8k
    doc = ctxt->myDoc;
10362
22.8k
    if ((doc != NULL) &&
10363
22.7k
        (xmlStrEqual(doc->version, SAX_COMPAT_MODE))) {
10364
19
        xmlFreeDoc(doc);
10365
19
        ctxt->myDoc = NULL;
10366
19
    }
10367
22.8k
}
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
7.30k
xmlParseDocument(xmlParserCtxt *ctxt) {
10380
7.30k
    if ((ctxt == NULL) || (ctxt->input == NULL))
10381
0
        return(-1);
10382
10383
7.30k
    GROW;
10384
10385
    /*
10386
     * SAX: detecting the level.
10387
     */
10388
7.30k
    xmlCtxtInitializeLate(ctxt);
10389
10390
7.30k
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10391
7.30k
        ctxt->sax->setDocumentLocator(ctxt->userData,
10392
7.30k
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10393
7.30k
    }
10394
10395
7.30k
    xmlDetectEncoding(ctxt);
10396
10397
7.30k
    if (CUR == 0) {
10398
8
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10399
8
  return(-1);
10400
8
    }
10401
10402
7.29k
    GROW;
10403
7.29k
    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
175
  xmlParseXMLDecl(ctxt);
10409
175
  SKIP_BLANKS;
10410
7.12k
    } else {
10411
7.12k
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10412
7.12k
        if (ctxt->version == NULL) {
10413
2
            xmlErrMemory(ctxt);
10414
2
            return(-1);
10415
2
        }
10416
7.12k
    }
10417
7.29k
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10418
7.24k
        ctxt->sax->startDocument(ctxt->userData);
10419
7.29k
    if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) &&
10420
7.23k
        (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
7.29k
    xmlParseMisc(ctxt);
10428
10429
    /*
10430
     * Then possibly doc type declaration(s) and more Misc
10431
     * (doctypedecl Misc*)?
10432
     */
10433
7.29k
    GROW;
10434
7.29k
    if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) {
10435
10436
2.37k
  ctxt->inSubset = 1;
10437
2.37k
  xmlParseDocTypeDecl(ctxt);
10438
2.37k
  if (RAW == '[') {
10439
639
      xmlParseInternalSubset(ctxt);
10440
1.73k
  } else if (RAW == '>') {
10441
1.70k
            NEXT;
10442
1.70k
        }
10443
10444
  /*
10445
   * Create and update the external subset.
10446
   */
10447
2.37k
  ctxt->inSubset = 2;
10448
2.37k
  if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) &&
10449
2.37k
      (!ctxt->disableSAX))
10450
2.28k
      ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
10451
2.28k
                                ctxt->extSubSystem, ctxt->extSubURI);
10452
2.37k
  ctxt->inSubset = 0;
10453
10454
2.37k
        xmlCleanSpecialAttr(ctxt);
10455
10456
2.37k
  xmlParseMisc(ctxt);
10457
2.37k
    }
10458
10459
    /*
10460
     * Time to start parsing the tree itself
10461
     */
10462
7.29k
    GROW;
10463
7.29k
    if (RAW != '<') {
10464
84
        if (ctxt->wellFormed)
10465
5
            xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10466
5
                           "Start tag expected, '<' not found\n");
10467
7.21k
    } else {
10468
7.21k
  xmlParseElement(ctxt);
10469
10470
  /*
10471
   * The Misc part at the end
10472
   */
10473
7.21k
  xmlParseMisc(ctxt);
10474
10475
7.21k
        xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
10476
7.21k
    }
10477
10478
7.29k
    ctxt->instate = XML_PARSER_EOF;
10479
7.29k
    xmlFinishDocument(ctxt);
10480
10481
7.29k
    if (! ctxt->wellFormed) {
10482
742
  ctxt->valid = 0;
10483
742
  return(-1);
10484
742
    }
10485
10486
6.55k
    return(0);
10487
7.29k
}
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
83.3k
xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) {
10575
83.3k
    const xmlChar *cur;
10576
10577
83.3k
    if (ctxt->checkIndex == 0) {
10578
62.1k
        cur = ctxt->input->cur + 1;
10579
62.1k
    } else {
10580
21.2k
        cur = ctxt->input->cur + ctxt->checkIndex;
10581
21.2k
    }
10582
10583
83.3k
    if (memchr(cur, c, ctxt->input->end - cur) == NULL) {
10584
24.7k
        size_t index = ctxt->input->end - ctxt->input->cur;
10585
10586
24.7k
        if (index > LONG_MAX) {
10587
0
            ctxt->checkIndex = 0;
10588
0
            return(1);
10589
0
        }
10590
24.7k
        ctxt->checkIndex = index;
10591
24.7k
        return(0);
10592
58.6k
    } else {
10593
58.6k
        ctxt->checkIndex = 0;
10594
58.6k
        return(1);
10595
58.6k
    }
10596
83.3k
}
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
236k
                     const char *str, size_t strLen) {
10609
236k
    const xmlChar *cur, *term;
10610
10611
236k
    if (ctxt->checkIndex == 0) {
10612
189k
        cur = ctxt->input->cur + startDelta;
10613
189k
    } else {
10614
47.2k
        cur = ctxt->input->cur + ctxt->checkIndex;
10615
47.2k
    }
10616
10617
236k
    term = BAD_CAST strstr((const char *) cur, str);
10618
236k
    if (term == NULL) {
10619
49.2k
        const xmlChar *end = ctxt->input->end;
10620
49.2k
        size_t index;
10621
10622
        /* Rescan (strLen - 1) characters. */
10623
49.2k
        if ((size_t) (end - cur) < strLen)
10624
1.12k
            end = cur;
10625
48.1k
        else
10626
48.1k
            end -= strLen - 1;
10627
49.2k
        index = end - ctxt->input->cur;
10628
49.2k
        if (index > LONG_MAX) {
10629
0
            ctxt->checkIndex = 0;
10630
0
            return(ctxt->input->end - strLen);
10631
0
        }
10632
49.2k
        ctxt->checkIndex = index;
10633
187k
    } else {
10634
187k
        ctxt->checkIndex = 0;
10635
187k
    }
10636
10637
236k
    return(term);
10638
236k
}
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
119k
xmlParseLookupCharData(xmlParserCtxtPtr ctxt) {
10647
119k
    const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex;
10648
119k
    const xmlChar *end = ctxt->input->end;
10649
119k
    size_t index;
10650
10651
1.28M
    while (cur < end) {
10652
1.28M
        if ((*cur == '<') || (*cur == '&')) {
10653
112k
            ctxt->checkIndex = 0;
10654
112k
            return(1);
10655
112k
        }
10656
1.16M
        cur++;
10657
1.16M
    }
10658
10659
7.43k
    index = cur - ctxt->input->cur;
10660
7.43k
    if (index > LONG_MAX) {
10661
0
        ctxt->checkIndex = 0;
10662
0
        return(1);
10663
0
    }
10664
7.43k
    ctxt->checkIndex = index;
10665
7.43k
    return(0);
10666
7.43k
}
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
562k
xmlParseLookupGt(xmlParserCtxtPtr ctxt) {
10676
562k
    const xmlChar *cur;
10677
562k
    const xmlChar *end = ctxt->input->end;
10678
562k
    int state = ctxt->endCheckState;
10679
562k
    size_t index;
10680
10681
562k
    if (ctxt->checkIndex == 0)
10682
293k
        cur = ctxt->input->cur + 1;
10683
268k
    else
10684
268k
        cur = ctxt->input->cur + ctxt->checkIndex;
10685
10686
164M
    while (cur < end) {
10687
164M
        if (state) {
10688
142M
            if (*cur == state)
10689
219k
                state = 0;
10690
142M
        } else if (*cur == '\'' || *cur == '"') {
10691
221k
            state = *cur;
10692
21.5M
        } else if (*cur == '>') {
10693
285k
            ctxt->checkIndex = 0;
10694
285k
            ctxt->endCheckState = 0;
10695
285k
            return(1);
10696
285k
        }
10697
164M
        cur++;
10698
164M
    }
10699
10700
276k
    index = cur - ctxt->input->cur;
10701
276k
    if (index > LONG_MAX) {
10702
0
        ctxt->checkIndex = 0;
10703
0
        ctxt->endCheckState = 0;
10704
0
        return(1);
10705
0
    }
10706
276k
    ctxt->checkIndex = index;
10707
276k
    ctxt->endCheckState = state;
10708
276k
    return(0);
10709
276k
}
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
27.8k
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
27.8k
    const xmlChar *cur, *start;
10727
27.8k
    const xmlChar *end = ctxt->input->end;
10728
27.8k
    int state = ctxt->endCheckState;
10729
27.8k
    size_t index;
10730
10731
27.8k
    if (ctxt->checkIndex == 0) {
10732
10.6k
        cur = ctxt->input->cur + 1;
10733
17.2k
    } else {
10734
17.2k
        cur = ctxt->input->cur + ctxt->checkIndex;
10735
17.2k
    }
10736
27.8k
    start = cur;
10737
10738
18.6M
    while (cur < end) {
10739
18.6M
        if (state == '-') {
10740
619k
            if ((*cur == '-') &&
10741
30.2k
                (cur[1] == '-') &&
10742
23.8k
                (cur[2] == '>')) {
10743
13.3k
                state = 0;
10744
13.3k
                cur += 3;
10745
13.3k
                start = cur;
10746
13.3k
                continue;
10747
13.3k
            }
10748
619k
        }
10749
18.0M
        else if (state == ']') {
10750
10.7k
            if (*cur == '>') {
10751
7.37k
                ctxt->checkIndex = 0;
10752
7.37k
                ctxt->endCheckState = 0;
10753
7.37k
                return(1);
10754
7.37k
            }
10755
3.38k
            if (IS_BLANK_CH(*cur)) {
10756
1.61k
                state = ' ';
10757
1.76k
            } else if (*cur != ']') {
10758
1.24k
                state = 0;
10759
1.24k
                start = cur;
10760
1.24k
                continue;
10761
1.24k
            }
10762
3.38k
        }
10763
18.0M
        else if (state == ' ') {
10764
5.86k
            if (*cur == '>') {
10765
199
                ctxt->checkIndex = 0;
10766
199
                ctxt->endCheckState = 0;
10767
199
                return(1);
10768
199
            }
10769
5.66k
            if (!IS_BLANK_CH(*cur)) {
10770
1.41k
                state = 0;
10771
1.41k
                start = cur;
10772
1.41k
                continue;
10773
1.41k
            }
10774
5.66k
        }
10775
18.0M
        else if (state != 0) {
10776
14.7M
            if (*cur == state) {
10777
32.7k
                state = 0;
10778
32.7k
                start = cur + 1;
10779
32.7k
            }
10780
14.7M
        }
10781
3.27M
        else if (*cur == '<') {
10782
54.8k
            if ((cur[1] == '!') &&
10783
42.0k
                (cur[2] == '-') &&
10784
13.6k
                (cur[3] == '-')) {
10785
13.3k
                state = '-';
10786
13.3k
                cur += 4;
10787
                /* Don't treat <!--> as comment */
10788
13.3k
                start = cur;
10789
13.3k
                continue;
10790
13.3k
            }
10791
54.8k
        }
10792
3.22M
        else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) {
10793
43.3k
            state = *cur;
10794
43.3k
        }
10795
10796
18.6M
        cur++;
10797
18.6M
    }
10798
10799
    /*
10800
     * Rescan the three last characters to detect "<!--" and "-->"
10801
     * split across chunks.
10802
     */
10803
20.3k
    if ((state == 0) || (state == '-')) {
10804
5.78k
        if (cur - start < 3)
10805
534
            cur = start;
10806
5.25k
        else
10807
5.25k
            cur -= 3;
10808
5.78k
    }
10809
20.3k
    index = cur - ctxt->input->cur;
10810
20.3k
    if (index > LONG_MAX) {
10811
0
        ctxt->checkIndex = 0;
10812
0
        ctxt->endCheckState = 0;
10813
0
        return(1);
10814
0
    }
10815
20.3k
    ctxt->checkIndex = index;
10816
20.3k
    ctxt->endCheckState = state;
10817
20.3k
    return(0);
10818
20.3k
}
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
422k
xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
10829
422k
    int ret = 0;
10830
422k
    size_t avail;
10831
422k
    xmlChar cur, next;
10832
10833
422k
    if (ctxt->input == NULL)
10834
0
        return(0);
10835
10836
422k
    if ((ctxt->input != NULL) &&
10837
422k
        (ctxt->input->cur - ctxt->input->base > 4096)) {
10838
2.55k
        xmlParserShrink(ctxt);
10839
2.55k
    }
10840
10841
3.98M
    while (ctxt->disableSAX == 0) {
10842
3.97M
        avail = ctxt->input->end - ctxt->input->cur;
10843
3.97M
        if (avail < 1)
10844
19.7k
      goto done;
10845
3.95M
        switch (ctxt->instate) {
10846
4.14k
            case XML_PARSER_EOF:
10847
          /*
10848
     * Document parsing is done !
10849
     */
10850
4.14k
          goto done;
10851
35.8k
            case XML_PARSER_START:
10852
                /*
10853
                 * Very first chars read from the document flow.
10854
                 */
10855
35.8k
                if ((!terminate) && (avail < 4))
10856
465
                    goto done;
10857
10858
                /*
10859
                 * We need more bytes to detect EBCDIC code pages.
10860
                 * See xmlDetectEBCDIC.
10861
                 */
10862
35.3k
                if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) &&
10863
704
                    (!terminate) && (avail < 200))
10864
334
                    goto done;
10865
10866
35.0k
                xmlDetectEncoding(ctxt);
10867
35.0k
                ctxt->instate = XML_PARSER_XML_DECL;
10868
35.0k
    break;
10869
10870
51.4k
            case XML_PARSER_XML_DECL:
10871
51.4k
    if ((!terminate) && (avail < 2))
10872
6
        goto done;
10873
51.4k
    cur = ctxt->input->cur[0];
10874
51.4k
    next = ctxt->input->cur[1];
10875
51.4k
          if ((cur == '<') && (next == '?')) {
10876
        /* PI or XML decl */
10877
25.2k
        if ((!terminate) &&
10878
23.9k
                        (!xmlParseLookupString(ctxt, 2, "?>", 2)))
10879
16.4k
      goto done;
10880
8.80k
        if ((ctxt->input->cur[2] == 'x') &&
10881
7.17k
      (ctxt->input->cur[3] == 'm') &&
10882
6.30k
      (ctxt->input->cur[4] == 'l') &&
10883
3.35k
      (IS_BLANK_CH(ctxt->input->cur[5]))) {
10884
3.26k
      ret += 5;
10885
3.26k
      xmlParseXMLDecl(ctxt);
10886
5.54k
        } else {
10887
5.54k
      ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10888
5.54k
                        if (ctxt->version == NULL) {
10889
3
                            xmlErrMemory(ctxt);
10890
3
                            break;
10891
3
                        }
10892
5.54k
        }
10893
26.1k
    } else {
10894
26.1k
        ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10895
26.1k
        if (ctxt->version == NULL) {
10896
4
            xmlErrMemory(ctxt);
10897
4
      break;
10898
4
        }
10899
26.1k
    }
10900
34.9k
                if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10901
34.9k
                    ctxt->sax->setDocumentLocator(ctxt->userData,
10902
34.9k
                            (xmlSAXLocator *) &xmlDefaultSAXLocator);
10903
34.9k
                }
10904
34.9k
                if ((ctxt->sax) && (ctxt->sax->startDocument) &&
10905
34.9k
                    (!ctxt->disableSAX))
10906
34.7k
                    ctxt->sax->startDocument(ctxt->userData);
10907
34.9k
                ctxt->instate = XML_PARSER_MISC;
10908
34.9k
    break;
10909
588k
            case XML_PARSER_START_TAG: {
10910
588k
          const xmlChar *name;
10911
588k
    const xmlChar *prefix = NULL;
10912
588k
    const xmlChar *URI = NULL;
10913
588k
                int line = ctxt->input->line;
10914
588k
    int nbNs = 0;
10915
10916
588k
    if ((!terminate) && (avail < 2))
10917
109
        goto done;
10918
588k
    cur = ctxt->input->cur[0];
10919
588k
          if (cur != '<') {
10920
2.01k
        xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10921
2.01k
                                   "Start tag expected, '<' not found");
10922
2.01k
                    ctxt->instate = XML_PARSER_EOF;
10923
2.01k
                    xmlFinishDocument(ctxt);
10924
2.01k
        goto done;
10925
2.01k
    }
10926
586k
    if ((!terminate) && (!xmlParseLookupGt(ctxt)))
10927
251k
                    goto done;
10928
334k
    if (ctxt->spaceNr == 0)
10929
0
        spacePush(ctxt, -1);
10930
334k
    else if (*ctxt->space == -2)
10931
58.5k
        spacePush(ctxt, -1);
10932
276k
    else
10933
276k
        spacePush(ctxt, *ctxt->space);
10934
334k
#ifdef LIBXML_SAX1_ENABLED
10935
334k
    if (ctxt->sax2)
10936
222k
#endif /* LIBXML_SAX1_ENABLED */
10937
222k
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
10938
112k
#ifdef LIBXML_SAX1_ENABLED
10939
112k
    else
10940
112k
        name = xmlParseStartTag(ctxt);
10941
334k
#endif /* LIBXML_SAX1_ENABLED */
10942
334k
    if (name == NULL) {
10943
1.70k
        spacePop(ctxt);
10944
1.70k
                    ctxt->instate = XML_PARSER_EOF;
10945
1.70k
                    xmlFinishDocument(ctxt);
10946
1.70k
        goto done;
10947
1.70k
    }
10948
333k
#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
333k
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
10955
114k
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
10956
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
10957
333k
#endif /* LIBXML_VALID_ENABLED */
10958
10959
    /*
10960
     * Check for an Empty Element.
10961
     */
10962
333k
    if ((RAW == '/') && (NXT(1) == '>')) {
10963
102k
        SKIP(2);
10964
10965
102k
        if (ctxt->sax2) {
10966
82.8k
      if ((ctxt->sax != NULL) &&
10967
82.8k
          (ctxt->sax->endElementNs != NULL) &&
10968
82.8k
          (!ctxt->disableSAX))
10969
82.8k
          ctxt->sax->endElementNs(ctxt->userData, name,
10970
82.8k
                                  prefix, URI);
10971
82.8k
      if (nbNs > 0)
10972
10.2k
          xmlParserNsPop(ctxt, nbNs);
10973
82.8k
#ifdef LIBXML_SAX1_ENABLED
10974
82.8k
        } else {
10975
19.4k
      if ((ctxt->sax != NULL) &&
10976
19.4k
          (ctxt->sax->endElement != NULL) &&
10977
19.4k
          (!ctxt->disableSAX))
10978
19.3k
          ctxt->sax->endElement(ctxt->userData, name);
10979
19.4k
#endif /* LIBXML_SAX1_ENABLED */
10980
19.4k
        }
10981
102k
        spacePop(ctxt);
10982
230k
    } else if (RAW == '>') {
10983
183k
        NEXT;
10984
183k
                    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
10985
183k
    } else {
10986
47.0k
        xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED,
10987
47.0k
           "Couldn't find end of Start Tag %s\n",
10988
47.0k
           name);
10989
47.0k
        nodePop(ctxt);
10990
47.0k
        spacePop(ctxt);
10991
47.0k
                    if (nbNs > 0)
10992
4.54k
                        xmlParserNsPop(ctxt, nbNs);
10993
47.0k
    }
10994
10995
333k
                if (ctxt->nameNr == 0)
10996
7.88k
                    ctxt->instate = XML_PARSER_EPILOG;
10997
325k
                else
10998
325k
                    ctxt->instate = XML_PARSER_CONTENT;
10999
333k
                break;
11000
334k
      }
11001
3.01M
            case XML_PARSER_CONTENT: {
11002
3.01M
    cur = ctxt->input->cur[0];
11003
11004
3.01M
    if (cur == '<') {
11005
478k
                    if ((!terminate) && (avail < 2))
11006
1.26k
                        goto done;
11007
477k
        next = ctxt->input->cur[1];
11008
11009
477k
                    if (next == '/') {
11010
44.1k
                        ctxt->instate = XML_PARSER_END_TAG;
11011
44.1k
                        break;
11012
433k
                    } else if (next == '?') {
11013
31.2k
                        if ((!terminate) &&
11014
25.7k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11015
7.60k
                            goto done;
11016
23.6k
                        xmlParsePI(ctxt);
11017
23.6k
                        ctxt->instate = XML_PARSER_CONTENT;
11018
23.6k
                        break;
11019
401k
                    } else if (next == '!') {
11020
93.6k
                        if ((!terminate) && (avail < 3))
11021
184
                            goto done;
11022
93.4k
                        next = ctxt->input->cur[2];
11023
11024
93.4k
                        if (next == '-') {
11025
73.0k
                            if ((!terminate) && (avail < 4))
11026
170
                                goto done;
11027
72.8k
                            if (ctxt->input->cur[3] == '-') {
11028
72.8k
                                if ((!terminate) &&
11029
70.9k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11030
5.18k
                                    goto done;
11031
67.6k
                                xmlParseComment(ctxt);
11032
67.6k
                                ctxt->instate = XML_PARSER_CONTENT;
11033
67.6k
                                break;
11034
72.8k
                            }
11035
72.8k
                        } else if (next == '[') {
11036
20.2k
                            if ((!terminate) && (avail < 9))
11037
119
                                goto done;
11038
20.1k
                            if ((ctxt->input->cur[2] == '[') &&
11039
20.1k
                                (ctxt->input->cur[3] == 'C') &&
11040
20.0k
                                (ctxt->input->cur[4] == 'D') &&
11041
20.0k
                                (ctxt->input->cur[5] == 'A') &&
11042
20.0k
                                (ctxt->input->cur[6] == 'T') &&
11043
20.0k
                                (ctxt->input->cur[7] == 'A') &&
11044
20.0k
                                (ctxt->input->cur[8] == '[')) {
11045
20.0k
                                if ((!terminate) &&
11046
15.0k
                                    (!xmlParseLookupString(ctxt, 9, "]]>", 3)))
11047
6.34k
                                    goto done;
11048
13.6k
                                ctxt->instate = XML_PARSER_CDATA_SECTION;
11049
13.6k
                                xmlParseCDSect(ctxt);
11050
13.6k
                                ctxt->instate = XML_PARSER_CONTENT;
11051
13.6k
                                break;
11052
20.0k
                            }
11053
20.1k
                        }
11054
93.4k
                    }
11055
2.53M
    } else if (cur == '&') {
11056
44.3k
        if ((!terminate) && (!xmlParseLookupChar(ctxt, ';')))
11057
13.7k
      goto done;
11058
30.5k
        xmlParseReference(ctxt);
11059
30.5k
                    break;
11060
2.48M
    } 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
2.48M
        if (avail < XML_PARSER_BIG_BUFFER_SIZE) {
11074
141k
      if ((!terminate) && (!xmlParseLookupCharData(ctxt)))
11075
7.43k
          goto done;
11076
141k
                    }
11077
2.47M
                    ctxt->checkIndex = 0;
11078
2.47M
        xmlParseCharDataInternal(ctxt, !terminate);
11079
2.47M
                    break;
11080
2.48M
    }
11081
11082
308k
                ctxt->instate = XML_PARSER_START_TAG;
11083
308k
    break;
11084
3.01M
      }
11085
54.0k
            case XML_PARSER_END_TAG:
11086
54.0k
    if ((!terminate) && (!xmlParseLookupChar(ctxt, '>')))
11087
10.9k
        goto done;
11088
43.1k
    if (ctxt->sax2) {
11089
32.1k
              xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
11090
32.1k
        nameNsPop(ctxt);
11091
32.1k
    }
11092
10.9k
#ifdef LIBXML_SAX1_ENABLED
11093
10.9k
      else
11094
10.9k
        xmlParseEndTag1(ctxt, 0);
11095
43.1k
#endif /* LIBXML_SAX1_ENABLED */
11096
43.1k
    if (ctxt->nameNr == 0) {
11097
2.55k
        ctxt->instate = XML_PARSER_EPILOG;
11098
40.5k
    } else {
11099
40.5k
        ctxt->instate = XML_PARSER_CONTENT;
11100
40.5k
    }
11101
43.1k
    break;
11102
98.5k
            case XML_PARSER_MISC:
11103
174k
            case XML_PARSER_PROLOG:
11104
179k
            case XML_PARSER_EPILOG:
11105
179k
    SKIP_BLANKS;
11106
179k
                avail = ctxt->input->end - ctxt->input->cur;
11107
179k
    if (avail < 1)
11108
3.09k
        goto done;
11109
176k
    if (ctxt->input->cur[0] == '<') {
11110
172k
                    if ((!terminate) && (avail < 2))
11111
224
                        goto done;
11112
172k
                    next = ctxt->input->cur[1];
11113
172k
                    if (next == '?') {
11114
38.2k
                        if ((!terminate) &&
11115
37.0k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11116
7.44k
                            goto done;
11117
30.7k
                        xmlParsePI(ctxt);
11118
30.7k
                        break;
11119
134k
                    } else if (next == '!') {
11120
106k
                        if ((!terminate) && (avail < 3))
11121
129
                            goto done;
11122
11123
106k
                        if (ctxt->input->cur[2] == '-') {
11124
65.0k
                            if ((!terminate) && (avail < 4))
11125
134
                                goto done;
11126
64.8k
                            if (ctxt->input->cur[3] == '-') {
11127
64.8k
                                if ((!terminate) &&
11128
64.2k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11129
6.26k
                                    goto done;
11130
58.6k
                                xmlParseComment(ctxt);
11131
58.6k
                                break;
11132
64.8k
                            }
11133
64.8k
                        } else if (ctxt->instate == XML_PARSER_MISC) {
11134
40.9k
                            if ((!terminate) && (avail < 9))
11135
22
                                goto done;
11136
40.9k
                            if ((ctxt->input->cur[2] == 'D') &&
11137
40.9k
                                (ctxt->input->cur[3] == 'O') &&
11138
40.9k
                                (ctxt->input->cur[4] == 'C') &&
11139
40.9k
                                (ctxt->input->cur[5] == 'T') &&
11140
40.9k
                                (ctxt->input->cur[6] == 'Y') &&
11141
40.9k
                                (ctxt->input->cur[7] == 'P') &&
11142
40.9k
                                (ctxt->input->cur[8] == 'E')) {
11143
40.9k
                                if ((!terminate) && (!xmlParseLookupGt(ctxt)))
11144
25.0k
                                    goto done;
11145
15.8k
                                ctxt->inSubset = 1;
11146
15.8k
                                xmlParseDocTypeDecl(ctxt);
11147
15.8k
                                if (RAW == '[') {
11148
12.0k
                                    ctxt->instate = XML_PARSER_DTD;
11149
12.0k
                                } else {
11150
3.86k
                                    if (RAW == '>')
11151
3.65k
                                        NEXT;
11152
                                    /*
11153
                                     * Create and update the external subset.
11154
                                     */
11155
3.86k
                                    ctxt->inSubset = 2;
11156
3.86k
                                    if ((ctxt->sax != NULL) &&
11157
3.86k
                                        (!ctxt->disableSAX) &&
11158
3.76k
                                        (ctxt->sax->externalSubset != NULL))
11159
3.76k
                                        ctxt->sax->externalSubset(
11160
3.76k
                                                ctxt->userData,
11161
3.76k
                                                ctxt->intSubName,
11162
3.76k
                                                ctxt->extSubSystem,
11163
3.76k
                                                ctxt->extSubURI);
11164
3.86k
                                    ctxt->inSubset = 0;
11165
3.86k
                                    xmlCleanSpecialAttr(ctxt);
11166
3.86k
                                    ctxt->instate = XML_PARSER_PROLOG;
11167
3.86k
                                }
11168
15.8k
                                break;
11169
40.9k
                            }
11170
40.9k
                        }
11171
106k
                    }
11172
172k
                }
11173
11174
31.9k
                if (ctxt->instate == XML_PARSER_EPILOG) {
11175
2.03k
                    if (ctxt->errNo == XML_ERR_OK)
11176
25
                        xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL);
11177
2.03k
        ctxt->instate = XML_PARSER_EOF;
11178
2.03k
                    xmlFinishDocument(ctxt);
11179
29.9k
                } else {
11180
29.9k
        ctxt->instate = XML_PARSER_START_TAG;
11181
29.9k
    }
11182
31.9k
    break;
11183
32.2k
            case XML_PARSER_DTD: {
11184
32.2k
                if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt)))
11185
20.3k
                    goto done;
11186
11.9k
    xmlParseInternalSubset(ctxt);
11187
11.9k
    ctxt->inSubset = 2;
11188
11.9k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
11189
9.78k
        (ctxt->sax->externalSubset != NULL))
11190
9.78k
        ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
11191
9.78k
          ctxt->extSubSystem, ctxt->extSubURI);
11192
11.9k
    ctxt->inSubset = 0;
11193
11.9k
    xmlCleanSpecialAttr(ctxt);
11194
11.9k
    ctxt->instate = XML_PARSER_PROLOG;
11195
11.9k
                break;
11196
32.2k
      }
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
3.95M
  }
11203
3.95M
    }
11204
422k
done:
11205
422k
    return(ret);
11206
422k
}
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
426k
              int terminate) {
11237
426k
    size_t curBase;
11238
426k
    size_t maxLength;
11239
426k
    size_t pos;
11240
426k
    int end_in_lf = 0;
11241
426k
    int res;
11242
11243
426k
    if ((ctxt == NULL) || (size < 0))
11244
0
        return(XML_ERR_ARGUMENT);
11245
426k
    if ((chunk == NULL) && (size > 0))
11246
0
        return(XML_ERR_ARGUMENT);
11247
426k
    if ((ctxt->input == NULL) || (ctxt->input->buf == NULL))
11248
0
        return(XML_ERR_ARGUMENT);
11249
426k
    if (ctxt->disableSAX != 0)
11250
3.77k
        return(ctxt->errNo);
11251
11252
422k
    ctxt->input->flags |= XML_INPUT_PROGRESSIVE;
11253
422k
    if (ctxt->instate == XML_PARSER_START)
11254
35.8k
        xmlCtxtInitializeLate(ctxt);
11255
422k
    if ((size > 0) && (chunk != NULL) && (!terminate) &&
11256
396k
        (chunk[size - 1] == '\r')) {
11257
2.70k
  end_in_lf = 1;
11258
2.70k
  size--;
11259
2.70k
    }
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
422k
    pos = ctxt->input->cur - ctxt->input->base;
11266
422k
    res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
11267
422k
    xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11268
422k
    if (res < 0) {
11269
229
        xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11270
229
        return(ctxt->errNo);
11271
229
    }
11272
11273
422k
    xmlParseTryOrFinish(ctxt, terminate);
11274
11275
422k
    curBase = ctxt->input->cur - ctxt->input->base;
11276
422k
    maxLength = (ctxt->options & XML_PARSE_HUGE) ?
11277
198k
                XML_MAX_HUGE_LENGTH :
11278
422k
                XML_MAX_LOOKUP_LIMIT;
11279
422k
    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
422k
    if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0))
11285
10.9k
        return(ctxt->errNo);
11286
11287
411k
    if (end_in_lf == 1) {
11288
2.69k
  pos = ctxt->input->cur - ctxt->input->base;
11289
2.69k
  res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r");
11290
2.69k
  xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11291
2.69k
        if (res < 0) {
11292
3
            xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11293
3
            return(ctxt->errNo);
11294
3
        }
11295
2.69k
    }
11296
411k
    if (terminate) {
11297
  /*
11298
   * Check for termination
11299
   */
11300
18.2k
        if ((ctxt->instate != XML_PARSER_EOF) &&
11301
10.0k
            (ctxt->instate != XML_PARSER_EPILOG)) {
11302
4.99k
            if (ctxt->nameNr > 0) {
11303
3.74k
                const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
11304
3.74k
                int line = ctxt->pushTab[ctxt->nameNr - 1].line;
11305
3.74k
                xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
11306
3.74k
                        "Premature end of data in tag %s line %d\n",
11307
3.74k
                        name, line, NULL);
11308
3.74k
            } else if (ctxt->instate == XML_PARSER_START) {
11309
24
                xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
11310
1.22k
            } else {
11311
1.22k
                xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
11312
1.22k
                               "Start tag expected, '<' not found\n");
11313
1.22k
            }
11314
13.2k
        } else {
11315
13.2k
            xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
11316
13.2k
        }
11317
18.2k
  if (ctxt->instate != XML_PARSER_EOF) {
11318
9.81k
            ctxt->instate = XML_PARSER_EOF;
11319
9.81k
            xmlFinishDocument(ctxt);
11320
9.81k
  }
11321
18.2k
    }
11322
411k
    if (ctxt->wellFormed == 0)
11323
20.2k
  return((xmlParserErrors) ctxt->errNo);
11324
391k
    else
11325
391k
        return(0);
11326
411k
}
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
35.6k
                        const char *chunk, int size, const char *filename) {
11358
35.6k
    xmlParserCtxtPtr ctxt;
11359
35.6k
    xmlParserInputPtr input;
11360
11361
35.6k
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11362
35.6k
    if (ctxt == NULL)
11363
20
  return(NULL);
11364
11365
35.6k
    ctxt->options &= ~XML_PARSE_NODICT;
11366
35.6k
    ctxt->dictNames = 1;
11367
11368
35.6k
    input = xmlNewPushInput(filename, chunk, size);
11369
35.6k
    if (input == NULL) {
11370
15
  xmlFreeParserCtxt(ctxt);
11371
15
  return(NULL);
11372
15
    }
11373
35.5k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11374
0
        xmlFreeInputStream(input);
11375
0
        xmlFreeParserCtxt(ctxt);
11376
0
        return(NULL);
11377
0
    }
11378
11379
35.5k
    return(ctxt);
11380
35.5k
}
11381
#endif /* LIBXML_PUSH_ENABLED */
11382
11383
/**
11384
 * Blocks further parser processing
11385
 *
11386
 * @param ctxt  an XML parser context
11387
 */
11388
void
11389
37.7k
xmlStopParser(xmlParserCtxt *ctxt) {
11390
37.7k
    if (ctxt == NULL)
11391
0
        return;
11392
11393
    /* This stops the parser */
11394
37.7k
    ctxt->disableSAX = 2;
11395
11396
    /*
11397
     * xmlStopParser is often called from error handlers,
11398
     * so we can't raise an error here to avoid infinite
11399
     * loops. Just make sure that an error condition is
11400
     * reported.
11401
     */
11402
37.7k
    if (ctxt->errNo == XML_ERR_OK) {
11403
3.38k
        ctxt->errNo = XML_ERR_USER_STOP;
11404
3.38k
        ctxt->lastError.code = XML_ERR_USER_STOP;
11405
3.38k
        ctxt->wellFormed = 0;
11406
3.38k
    }
11407
37.7k
}
11408
11409
/**
11410
 * Create a parser context for using the XML parser with an existing
11411
 * I/O stream
11412
 *
11413
 * @param sax  a SAX handler (optional)
11414
 * @param user_data  user data for SAX callbacks (optional)
11415
 * @param ioread  an I/O read function
11416
 * @param ioclose  an I/O close function (optional)
11417
 * @param ioctx  an I/O handler
11418
 * @param enc  the charset encoding if known (deprecated)
11419
 * @returns the new parser context or NULL
11420
 */
11421
xmlParserCtxt *
11422
xmlCreateIOParserCtxt(xmlSAXHandler *sax, void *user_data,
11423
                      xmlInputReadCallback ioread,
11424
                      xmlInputCloseCallback ioclose,
11425
0
                      void *ioctx, xmlCharEncoding enc) {
11426
0
    xmlParserCtxtPtr ctxt;
11427
0
    xmlParserInputPtr input;
11428
0
    const char *encoding;
11429
11430
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11431
0
    if (ctxt == NULL)
11432
0
  return(NULL);
11433
11434
0
    encoding = xmlGetCharEncodingName(enc);
11435
0
    input = xmlCtxtNewInputFromIO(ctxt, NULL, ioread, ioclose, ioctx,
11436
0
                                  encoding, 0);
11437
0
    if (input == NULL) {
11438
0
  xmlFreeParserCtxt(ctxt);
11439
0
        return (NULL);
11440
0
    }
11441
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11442
0
        xmlFreeInputStream(input);
11443
0
        xmlFreeParserCtxt(ctxt);
11444
0
        return(NULL);
11445
0
    }
11446
11447
0
    return(ctxt);
11448
0
}
11449
11450
#ifdef LIBXML_VALID_ENABLED
11451
/************************************************************************
11452
 *                  *
11453
 *    Front ends when parsing a DTD       *
11454
 *                  *
11455
 ************************************************************************/
11456
11457
/**
11458
 * Parse a DTD.
11459
 *
11460
 * Option XML_PARSE_DTDLOAD should be enabled in the parser context
11461
 * to make external entities work.
11462
 *
11463
 * @since 2.14.0
11464
 *
11465
 * @param ctxt  a parser context
11466
 * @param input  a parser input
11467
 * @param publicId  public ID of the DTD (optional)
11468
 * @param systemId  system ID of the DTD (optional)
11469
 * @returns the resulting xmlDtd or NULL in case of error.
11470
 * `input` will be freed by the function in any case.
11471
 */
11472
xmlDtd *
11473
xmlCtxtParseDtd(xmlParserCtxt *ctxt, xmlParserInput *input,
11474
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
5.25k
                            int hasTextDecl, int buildTree) {
11669
5.25k
    xmlNodePtr root = NULL;
11670
5.25k
    xmlNodePtr list = NULL;
11671
5.25k
    xmlChar *rootName = BAD_CAST "#root";
11672
5.25k
    int result;
11673
11674
5.25k
    if (buildTree) {
11675
5.25k
        root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL);
11676
5.25k
        if (root == NULL) {
11677
3
            xmlErrMemory(ctxt);
11678
3
            goto error;
11679
3
        }
11680
5.25k
    }
11681
11682
5.25k
    if (xmlCtxtPushInput(ctxt, input) < 0)
11683
2
        goto error;
11684
11685
5.25k
    nameNsPush(ctxt, rootName, NULL, NULL, 0, 0);
11686
5.25k
    spacePush(ctxt, -1);
11687
11688
5.25k
    if (buildTree)
11689
5.25k
        nodePush(ctxt, root);
11690
11691
5.25k
    if (hasTextDecl) {
11692
1.87k
        xmlDetectEncoding(ctxt);
11693
11694
        /*
11695
         * Parse a possible text declaration first
11696
         */
11697
1.87k
        if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
11698
273
            (IS_BLANK_CH(NXT(5)))) {
11699
271
            xmlParseTextDecl(ctxt);
11700
            /*
11701
             * An XML-1.0 document can't reference an entity not XML-1.0
11702
             */
11703
271
            if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
11704
267
                (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
11705
54
                xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
11706
54
                               "Version mismatch between document and "
11707
54
                               "entity\n");
11708
54
            }
11709
271
        }
11710
1.87k
    }
11711
11712
5.25k
    xmlParseContentInternal(ctxt);
11713
11714
5.25k
    if (ctxt->input->cur < ctxt->input->end)
11715
344
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
11716
11717
5.25k
    if ((ctxt->wellFormed) ||
11718
4.41k
        ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) {
11719
4.41k
        if (root != NULL) {
11720
4.41k
            xmlNodePtr cur;
11721
11722
            /*
11723
             * Unlink newly created node list.
11724
             */
11725
4.41k
            list = root->children;
11726
4.41k
            root->children = NULL;
11727
4.41k
            root->last = NULL;
11728
20.4k
            for (cur = list; cur != NULL; cur = cur->next)
11729
16.0k
                cur->parent = NULL;
11730
4.41k
        }
11731
4.41k
    }
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
5.58k
    do {
11738
5.58k
        ctxt->input->cur = ctxt->input->end;
11739
5.58k
        xmlParserShrink(ctxt);
11740
5.58k
        result = xmlParserGrow(ctxt);
11741
5.58k
    } while (result > 0);
11742
11743
5.25k
    if (buildTree)
11744
5.25k
        nodePop(ctxt);
11745
11746
5.25k
    namePop(ctxt);
11747
5.25k
    spacePop(ctxt);
11748
11749
5.25k
    xmlCtxtPopInput(ctxt);
11750
11751
5.25k
error:
11752
5.25k
    xmlFreeNode(root);
11753
11754
5.25k
    return(list);
11755
5.25k
}
11756
11757
static void
11758
7.52k
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
11759
7.52k
    xmlParserInputPtr input;
11760
7.52k
    xmlNodePtr list;
11761
7.52k
    unsigned long consumed;
11762
7.52k
    int isExternal;
11763
7.52k
    int buildTree;
11764
7.52k
    int oldMinNsIndex;
11765
7.52k
    int oldNodelen, oldNodemem;
11766
11767
7.52k
    isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY);
11768
7.52k
    buildTree = (ctxt->node != NULL);
11769
11770
    /*
11771
     * Recursion check
11772
     */
11773
7.52k
    if (ent->flags & XML_ENT_EXPANDING) {
11774
4
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
11775
4
        goto error;
11776
4
    }
11777
11778
    /*
11779
     * Load entity
11780
     */
11781
7.52k
    input = xmlNewEntityInputStream(ctxt, ent);
11782
7.52k
    if (input == NULL)
11783
2.26k
        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
5.25k
    oldMinNsIndex = ctxt->nsdb->minNsIndex;
11791
5.25k
    if (buildTree)
11792
5.25k
        ctxt->nsdb->minNsIndex = ctxt->nsNr;
11793
11794
5.25k
    oldNodelen = ctxt->nodelen;
11795
5.25k
    oldNodemem = ctxt->nodemem;
11796
5.25k
    ctxt->nodelen = 0;
11797
5.25k
    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
5.25k
    ent->flags |= XML_ENT_EXPANDING;
11817
11818
5.25k
    list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree);
11819
11820
5.25k
    ent->flags &= ~XML_ENT_EXPANDING;
11821
11822
5.25k
    ctxt->nsdb->minNsIndex = oldMinNsIndex;
11823
5.25k
    ctxt->nodelen = oldNodelen;
11824
5.25k
    ctxt->nodemem = oldNodemem;
11825
11826
    /*
11827
     * Entity size accounting
11828
     */
11829
5.25k
    consumed = input->consumed;
11830
5.25k
    xmlSaturatedAddSizeT(&consumed, input->end - input->base);
11831
11832
5.25k
    if ((ent->flags & XML_ENT_CHECKED) == 0)
11833
3.24k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
11834
11835
5.25k
    if ((ent->flags & XML_ENT_PARSED) == 0) {
11836
3.24k
        if (isExternal)
11837
1.49k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
11838
11839
3.24k
        ent->children = list;
11840
11841
19.3k
        while (list != NULL) {
11842
16.0k
            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
16.0k
            if (list->doc != ent->doc)
11850
0
                xmlSetTreeDoc(list, ent->doc);
11851
11852
16.0k
            if (list->next == NULL)
11853
2.29k
                ent->last = list;
11854
16.0k
            list = list->next;
11855
16.0k
        }
11856
3.24k
    } else {
11857
2.01k
        xmlFreeNodeList(list);
11858
2.01k
    }
11859
11860
5.25k
    xmlFreeInputStream(input);
11861
11862
7.52k
error:
11863
7.52k
    ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED;
11864
7.52k
}
11865
11866
/**
11867
 * Parse an external general entity within an existing parsing context
11868
 * An external general parsed entity is well-formed if it matches the
11869
 * production labeled extParsedEnt.
11870
 *
11871
 *     [78] extParsedEnt ::= TextDecl? content
11872
 *
11873
 * @param ctxt  the existing parsing context
11874
 * @param URL  the URL for the entity to load
11875
 * @param ID  the System ID for the entity to load
11876
 * @param listOut  the return value for the set of parsed nodes
11877
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11878
 *    the parser error code otherwise
11879
 */
11880
11881
int
11882
xmlParseCtxtExternalEntity(xmlParserCtxt *ctxt, const xmlChar *URL,
11883
0
                           const xmlChar *ID, xmlNode **listOut) {
11884
0
    xmlParserInputPtr input;
11885
0
    xmlNodePtr list;
11886
11887
0
    if (listOut != NULL)
11888
0
        *listOut = NULL;
11889
11890
0
    if (ctxt == NULL)
11891
0
        return(XML_ERR_ARGUMENT);
11892
11893
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
11894
0
                            XML_RESOURCE_GENERAL_ENTITY);
11895
0
    if (input == NULL)
11896
0
        return(ctxt->errNo);
11897
11898
0
    xmlCtxtInitializeLate(ctxt);
11899
11900
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1);
11901
0
    if (listOut != NULL)
11902
0
        *listOut = list;
11903
0
    else
11904
0
        xmlFreeNodeList(list);
11905
11906
0
    xmlFreeInputStream(input);
11907
0
    return(ctxt->errNo);
11908
0
}
11909
11910
#ifdef LIBXML_SAX1_ENABLED
11911
/**
11912
 * Parse an external general entity
11913
 * An external general parsed entity is well-formed if it matches the
11914
 * production labeled extParsedEnt.
11915
 *
11916
 * This function uses deprecated global variables to set parser options
11917
 * which default to XML_PARSE_NODICT.
11918
 *
11919
 * @deprecated Use #xmlParseCtxtExternalEntity.
11920
 *
11921
 *     [78] extParsedEnt ::= TextDecl? content
11922
 *
11923
 * @param doc  the document the chunk pertains to
11924
 * @param sax  the SAX handler block (possibly NULL)
11925
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11926
 * @param depth  Used for loop detection, use 0
11927
 * @param URL  the URL for the entity to load
11928
 * @param ID  the System ID for the entity to load
11929
 * @param list  the return value for the set of parsed nodes
11930
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11931
 *    the parser error code otherwise
11932
 */
11933
11934
int
11935
xmlParseExternalEntity(xmlDoc *doc, xmlSAXHandler *sax, void *user_data,
11936
0
    int depth, const xmlChar *URL, const xmlChar *ID, xmlNode **list) {
11937
0
    xmlParserCtxtPtr ctxt;
11938
0
    int ret;
11939
11940
0
    if (list != NULL)
11941
0
        *list = NULL;
11942
11943
0
    if (doc == NULL)
11944
0
        return(XML_ERR_ARGUMENT);
11945
11946
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11947
0
    if (ctxt == NULL)
11948
0
        return(XML_ERR_NO_MEMORY);
11949
11950
0
    ctxt->depth = depth;
11951
0
    ctxt->myDoc = doc;
11952
0
    ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list);
11953
11954
0
    xmlFreeParserCtxt(ctxt);
11955
0
    return(ret);
11956
0
}
11957
11958
/**
11959
 * Parse a well-balanced chunk of an XML document
11960
 * called by the parser
11961
 * The allowed sequence for the Well Balanced Chunk is the one defined by
11962
 * the content production in the XML grammar:
11963
 *
11964
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
11965
 *                       Comment)*
11966
 *
11967
 * This function uses deprecated global variables to set parser options
11968
 * which default to XML_PARSE_NODICT.
11969
 *
11970
 * @param doc  the document the chunk pertains to (must not be NULL)
11971
 * @param sax  the SAX handler block (possibly NULL)
11972
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11973
 * @param depth  Used for loop detection, use 0
11974
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
11975
 * @param lst  the return value for the set of parsed nodes
11976
 * @returns 0 if the chunk is well balanced, -1 in case of args problem and
11977
 *    the parser error code otherwise
11978
 */
11979
11980
int
11981
xmlParseBalancedChunkMemory(xmlDoc *doc, xmlSAXHandler *sax,
11982
0
     void *user_data, int depth, const xmlChar *string, xmlNode **lst) {
11983
0
    return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data,
11984
0
                                                depth, string, lst, 0 );
11985
0
}
11986
#endif /* LIBXML_SAX1_ENABLED */
11987
11988
/**
11989
 * Parse a well-balanced chunk of XML matching the 'content' production.
11990
 *
11991
 * Namespaces in scope of `node` and entities of `node`'s document are
11992
 * recognized. When validating, the DTD of `node`'s document is used.
11993
 *
11994
 * Always consumes `input` even in error case.
11995
 *
11996
 * @since 2.14.0
11997
 *
11998
 * @param ctxt  parser context
11999
 * @param input  parser input
12000
 * @param node  target node or document
12001
 * @param hasTextDecl  whether to parse text declaration
12002
 * @returns a node list or NULL in case of error.
12003
 */
12004
xmlNode *
12005
xmlCtxtParseContent(xmlParserCtxt *ctxt, xmlParserInput *input,
12006
0
                    xmlNode *node, int hasTextDecl) {
12007
0
    xmlDocPtr doc;
12008
0
    xmlNodePtr cur, list = NULL;
12009
0
    int nsnr = 0;
12010
0
    xmlDictPtr oldDict;
12011
0
    int oldOptions, oldDictNames, oldLoadSubset;
12012
12013
0
    if ((ctxt == NULL) || (input == NULL) || (node == NULL)) {
12014
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12015
0
        goto exit;
12016
0
    }
12017
12018
0
    doc = node->doc;
12019
0
    if (doc == NULL) {
12020
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12021
0
        goto exit;
12022
0
    }
12023
12024
0
    switch (node->type) {
12025
0
        case XML_ELEMENT_NODE:
12026
0
        case XML_DOCUMENT_NODE:
12027
0
        case XML_HTML_DOCUMENT_NODE:
12028
0
            break;
12029
12030
0
        case XML_ATTRIBUTE_NODE:
12031
0
        case XML_TEXT_NODE:
12032
0
        case XML_CDATA_SECTION_NODE:
12033
0
        case XML_ENTITY_REF_NODE:
12034
0
        case XML_PI_NODE:
12035
0
        case XML_COMMENT_NODE:
12036
0
            for (cur = node->parent; cur != NULL; cur = cur->parent) {
12037
0
                if ((cur->type == XML_ELEMENT_NODE) ||
12038
0
                    (cur->type == XML_DOCUMENT_NODE) ||
12039
0
                    (cur->type == XML_HTML_DOCUMENT_NODE)) {
12040
0
                    node = cur;
12041
0
                    break;
12042
0
                }
12043
0
            }
12044
0
            break;
12045
12046
0
        default:
12047
0
            xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12048
0
            goto exit;
12049
0
    }
12050
12051
0
    xmlCtxtReset(ctxt);
12052
12053
0
    oldDict = ctxt->dict;
12054
0
    oldOptions = ctxt->options;
12055
0
    oldDictNames = ctxt->dictNames;
12056
0
    oldLoadSubset = ctxt->loadsubset;
12057
12058
    /*
12059
     * Use input doc's dict if present, else assure XML_PARSE_NODICT is set.
12060
     */
12061
0
    if (doc->dict != NULL) {
12062
0
        ctxt->dict = doc->dict;
12063
0
    } else {
12064
0
        ctxt->options |= XML_PARSE_NODICT;
12065
0
        ctxt->dictNames = 0;
12066
0
    }
12067
12068
    /*
12069
     * Disable IDs
12070
     */
12071
0
    ctxt->loadsubset |= XML_SKIP_IDS;
12072
0
    ctxt->options |= XML_PARSE_SKIP_IDS;
12073
12074
0
    ctxt->myDoc = doc;
12075
12076
0
#ifdef LIBXML_HTML_ENABLED
12077
0
    if (ctxt->html) {
12078
        /*
12079
         * When parsing in context, it makes no sense to add implied
12080
         * elements like html/body/etc...
12081
         */
12082
0
        ctxt->options |= HTML_PARSE_NOIMPLIED;
12083
12084
0
        list = htmlCtxtParseContentInternal(ctxt, input);
12085
0
    } else
12086
0
#endif
12087
0
    {
12088
0
        xmlCtxtInitializeLate(ctxt);
12089
12090
        /*
12091
         * initialize the SAX2 namespaces stack
12092
         */
12093
0
        cur = node;
12094
0
        while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) {
12095
0
            xmlNsPtr ns = cur->nsDef;
12096
0
            xmlHashedString hprefix, huri;
12097
12098
0
            while (ns != NULL) {
12099
0
                hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1);
12100
0
                huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1);
12101
0
                if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0)
12102
0
                    nsnr++;
12103
0
                ns = ns->next;
12104
0
            }
12105
0
            cur = cur->parent;
12106
0
        }
12107
12108
0
        list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1);
12109
12110
0
        if (nsnr > 0)
12111
0
            xmlParserNsPop(ctxt, nsnr);
12112
0
    }
12113
12114
0
    ctxt->dict = oldDict;
12115
0
    ctxt->options = oldOptions;
12116
0
    ctxt->dictNames = oldDictNames;
12117
0
    ctxt->loadsubset = oldLoadSubset;
12118
0
    ctxt->myDoc = NULL;
12119
0
    ctxt->node = NULL;
12120
12121
0
exit:
12122
0
    xmlFreeInputStream(input);
12123
0
    return(list);
12124
0
}
12125
12126
/**
12127
 * Parse a well-balanced chunk of an XML document
12128
 * within the context (DTD, namespaces, etc ...) of the given node.
12129
 *
12130
 * The allowed sequence for the data is a Well Balanced Chunk defined by
12131
 * the content production in the XML grammar:
12132
 *
12133
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12134
 *                       Comment)*
12135
 *
12136
 * This function assumes the encoding of `node`'s document which is
12137
 * typically not what you want. A better alternative is
12138
 * #xmlCtxtParseContent.
12139
 *
12140
 * @param node  the context node
12141
 * @param data  the input string
12142
 * @param datalen  the input string length in bytes
12143
 * @param options  a combination of xmlParserOption
12144
 * @param listOut  the return value for the set of parsed nodes
12145
 * @returns XML_ERR_OK if the chunk is well balanced, and the parser
12146
 * error code otherwise
12147
 */
12148
xmlParserErrors
12149
xmlParseInNodeContext(xmlNode *node, const char *data, int datalen,
12150
0
                      int options, xmlNode **listOut) {
12151
0
    xmlParserCtxtPtr ctxt;
12152
0
    xmlParserInputPtr input;
12153
0
    xmlDocPtr doc;
12154
0
    xmlNodePtr list;
12155
0
    xmlParserErrors ret;
12156
12157
0
    if (listOut == NULL)
12158
0
        return(XML_ERR_INTERNAL_ERROR);
12159
0
    *listOut = NULL;
12160
12161
0
    if ((node == NULL) || (data == NULL) || (datalen < 0))
12162
0
        return(XML_ERR_INTERNAL_ERROR);
12163
12164
0
    doc = node->doc;
12165
0
    if (doc == NULL)
12166
0
        return(XML_ERR_INTERNAL_ERROR);
12167
12168
0
#ifdef LIBXML_HTML_ENABLED
12169
0
    if (doc->type == XML_HTML_DOCUMENT_NODE) {
12170
0
        ctxt = htmlNewParserCtxt();
12171
0
    }
12172
0
    else
12173
0
#endif
12174
0
        ctxt = xmlNewParserCtxt();
12175
12176
0
    if (ctxt == NULL)
12177
0
        return(XML_ERR_NO_MEMORY);
12178
12179
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen,
12180
0
                                      (const char *) doc->encoding,
12181
0
                                      XML_INPUT_BUF_STATIC);
12182
0
    if (input == NULL) {
12183
0
        xmlFreeParserCtxt(ctxt);
12184
0
        return(XML_ERR_NO_MEMORY);
12185
0
    }
12186
12187
0
    xmlCtxtUseOptions(ctxt, options);
12188
12189
0
    list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0);
12190
12191
0
    if (list == NULL) {
12192
0
        ret = ctxt->errNo;
12193
0
        if (ret == XML_ERR_ARGUMENT)
12194
0
            ret = XML_ERR_INTERNAL_ERROR;
12195
0
    } else {
12196
0
        ret = XML_ERR_OK;
12197
0
        *listOut = list;
12198
0
    }
12199
12200
0
    xmlFreeParserCtxt(ctxt);
12201
12202
0
    return(ret);
12203
0
}
12204
12205
#ifdef LIBXML_SAX1_ENABLED
12206
/**
12207
 * Parse a well-balanced chunk of an XML document
12208
 *
12209
 * The allowed sequence for the Well Balanced Chunk is the one defined by
12210
 * the content production in the XML grammar:
12211
 *
12212
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12213
 *                       Comment)*
12214
 *
12215
 * In case recover is set to 1, the nodelist will not be empty even if
12216
 * the parsed chunk is not well balanced, assuming the parsing succeeded to
12217
 * some extent.
12218
 *
12219
 * This function uses deprecated global variables to set parser options
12220
 * which default to XML_PARSE_NODICT.
12221
 *
12222
 * @param doc  the document the chunk pertains to (must not be NULL)
12223
 * @param sax  the SAX handler block (possibly NULL)
12224
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
12225
 * @param depth  Used for loop detection, use 0
12226
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
12227
 * @param listOut  the return value for the set of parsed nodes
12228
 * @param recover  return nodes even if the data is broken (use 0)
12229
 * @returns 0 if the chunk is well balanced, or thehe parser error code
12230
 * otherwise.
12231
 */
12232
int
12233
xmlParseBalancedChunkMemoryRecover(xmlDoc *doc, xmlSAXHandler *sax,
12234
     void *user_data, int depth, const xmlChar *string, xmlNode **listOut,
12235
0
     int recover) {
12236
0
    xmlParserCtxtPtr ctxt;
12237
0
    xmlParserInputPtr input;
12238
0
    xmlNodePtr list;
12239
0
    int ret;
12240
12241
0
    if (listOut != NULL)
12242
0
        *listOut = NULL;
12243
12244
0
    if (string == NULL)
12245
0
        return(XML_ERR_ARGUMENT);
12246
12247
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
12248
0
    if (ctxt == NULL)
12249
0
        return(XML_ERR_NO_MEMORY);
12250
12251
0
    xmlCtxtInitializeLate(ctxt);
12252
12253
0
    ctxt->depth = depth;
12254
0
    ctxt->myDoc = doc;
12255
0
    if (recover) {
12256
0
        ctxt->options |= XML_PARSE_RECOVER;
12257
0
        ctxt->recovery = 1;
12258
0
    }
12259
12260
0
    input = xmlNewStringInputStream(ctxt, string);
12261
0
    if (input == NULL) {
12262
0
        ret = ctxt->errNo;
12263
0
        goto error;
12264
0
    }
12265
12266
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1);
12267
0
    if (listOut != NULL)
12268
0
        *listOut = list;
12269
0
    else
12270
0
        xmlFreeNodeList(list);
12271
12272
0
    if (!ctxt->wellFormed)
12273
0
        ret = ctxt->errNo;
12274
0
    else
12275
0
        ret = XML_ERR_OK;
12276
12277
0
error:
12278
0
    xmlFreeInputStream(input);
12279
0
    xmlFreeParserCtxt(ctxt);
12280
0
    return(ret);
12281
0
}
12282
12283
/**
12284
 * Parse an XML external entity out of context and build a tree.
12285
 * It use the given SAX function block to handle the parsing callback.
12286
 * If sax is NULL, fallback to the default DOM tree building routines.
12287
 *
12288
 * @deprecated Don't use.
12289
 *
12290
 *     [78] extParsedEnt ::= TextDecl? content
12291
 *
12292
 * This correspond to a "Well Balanced" chunk
12293
 *
12294
 * This function uses deprecated global variables to set parser options
12295
 * which default to XML_PARSE_NODICT.
12296
 *
12297
 * @param sax  the SAX handler block
12298
 * @param filename  the filename
12299
 * @returns the resulting document tree
12300
 */
12301
12302
xmlDoc *
12303
0
xmlSAXParseEntity(xmlSAXHandler *sax, const char *filename) {
12304
0
    xmlDocPtr ret;
12305
0
    xmlParserCtxtPtr ctxt;
12306
12307
0
    ctxt = xmlCreateFileParserCtxt(filename);
12308
0
    if (ctxt == NULL) {
12309
0
  return(NULL);
12310
0
    }
12311
0
    if (sax != NULL) {
12312
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12313
0
            *ctxt->sax = *sax;
12314
0
        } else {
12315
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12316
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12317
0
        }
12318
0
        ctxt->userData = NULL;
12319
0
    }
12320
12321
0
    xmlParseExtParsedEnt(ctxt);
12322
12323
0
    if (ctxt->wellFormed) {
12324
0
  ret = ctxt->myDoc;
12325
0
    } else {
12326
0
        ret = NULL;
12327
0
        xmlFreeDoc(ctxt->myDoc);
12328
0
    }
12329
12330
0
    xmlFreeParserCtxt(ctxt);
12331
12332
0
    return(ret);
12333
0
}
12334
12335
/**
12336
 * Parse an XML external entity out of context and build a tree.
12337
 *
12338
 *     [78] extParsedEnt ::= TextDecl? content
12339
 *
12340
 * This correspond to a "Well Balanced" chunk
12341
 *
12342
 * This function uses deprecated global variables to set parser options
12343
 * which default to XML_PARSE_NODICT.
12344
 *
12345
 * @deprecated Don't use.
12346
 *
12347
 * @param filename  the filename
12348
 * @returns the resulting document tree
12349
 */
12350
12351
xmlDoc *
12352
0
xmlParseEntity(const char *filename) {
12353
0
    return(xmlSAXParseEntity(NULL, filename));
12354
0
}
12355
#endif /* LIBXML_SAX1_ENABLED */
12356
12357
/**
12358
 * Create a parser context for an external entity
12359
 * Automatic support for ZLIB/Compress compressed document is provided
12360
 * by default if found at compile-time.
12361
 *
12362
 * @deprecated Don't use.
12363
 *
12364
 * @param URL  the entity URL
12365
 * @param ID  the entity PUBLIC ID
12366
 * @param base  a possible base for the target URI
12367
 * @returns the new parser context or NULL
12368
 */
12369
xmlParserCtxt *
12370
xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID,
12371
0
                    const xmlChar *base) {
12372
0
    xmlParserCtxtPtr ctxt;
12373
0
    xmlParserInputPtr input;
12374
0
    xmlChar *uri = NULL;
12375
12376
0
    ctxt = xmlNewParserCtxt();
12377
0
    if (ctxt == NULL)
12378
0
  return(NULL);
12379
12380
0
    if (base != NULL) {
12381
0
        if (xmlBuildURISafe(URL, base, &uri) < 0)
12382
0
            goto error;
12383
0
        if (uri != NULL)
12384
0
            URL = uri;
12385
0
    }
12386
12387
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
12388
0
                            XML_RESOURCE_UNKNOWN);
12389
0
    if (input == NULL)
12390
0
        goto error;
12391
12392
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12393
0
        xmlFreeInputStream(input);
12394
0
        goto error;
12395
0
    }
12396
12397
0
    xmlFree(uri);
12398
0
    return(ctxt);
12399
12400
0
error:
12401
0
    xmlFree(uri);
12402
0
    xmlFreeParserCtxt(ctxt);
12403
0
    return(NULL);
12404
0
}
12405
12406
/************************************************************************
12407
 *                  *
12408
 *    Front ends when parsing from a file     *
12409
 *                  *
12410
 ************************************************************************/
12411
12412
/**
12413
 * Create a parser context for a file or URL content.
12414
 * Automatic support for ZLIB/Compress compressed document is provided
12415
 * by default if found at compile-time and for file accesses
12416
 *
12417
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12418
 *
12419
 * @param filename  the filename or URL
12420
 * @param options  a combination of xmlParserOption
12421
 * @returns the new parser context or NULL
12422
 */
12423
xmlParserCtxt *
12424
xmlCreateURLParserCtxt(const char *filename, int options)
12425
0
{
12426
0
    xmlParserCtxtPtr ctxt;
12427
0
    xmlParserInputPtr input;
12428
12429
0
    ctxt = xmlNewParserCtxt();
12430
0
    if (ctxt == NULL)
12431
0
  return(NULL);
12432
12433
0
    xmlCtxtUseOptions(ctxt, options);
12434
12435
0
    input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT);
12436
0
    if (input == NULL) {
12437
0
  xmlFreeParserCtxt(ctxt);
12438
0
  return(NULL);
12439
0
    }
12440
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12441
0
        xmlFreeInputStream(input);
12442
0
        xmlFreeParserCtxt(ctxt);
12443
0
        return(NULL);
12444
0
    }
12445
12446
0
    return(ctxt);
12447
0
}
12448
12449
/**
12450
 * Create a parser context for a file content.
12451
 * Automatic support for ZLIB/Compress compressed document is provided
12452
 * by default if found at compile-time.
12453
 *
12454
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12455
 *
12456
 * @param filename  the filename
12457
 * @returns the new parser context or NULL
12458
 */
12459
xmlParserCtxt *
12460
xmlCreateFileParserCtxt(const char *filename)
12461
0
{
12462
0
    return(xmlCreateURLParserCtxt(filename, 0));
12463
0
}
12464
12465
#ifdef LIBXML_SAX1_ENABLED
12466
/**
12467
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12468
 * compressed document is provided by default if found at compile-time.
12469
 * It use the given SAX function block to handle the parsing callback.
12470
 * If sax is NULL, fallback to the default DOM tree building routines.
12471
 *
12472
 * This function uses deprecated global variables to set parser options
12473
 * which default to XML_PARSE_NODICT.
12474
 *
12475
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12476
 *
12477
 * User data (void *) is stored within the parser context in the
12478
 * context's _private member, so it is available nearly everywhere in libxml
12479
 *
12480
 * @param sax  the SAX handler block
12481
 * @param filename  the filename
12482
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12483
 *             documents
12484
 * @param data  the userdata
12485
 * @returns the resulting document tree
12486
 */
12487
12488
xmlDoc *
12489
xmlSAXParseFileWithData(xmlSAXHandler *sax, const char *filename,
12490
0
                        int recovery, void *data) {
12491
0
    xmlDocPtr ret = NULL;
12492
0
    xmlParserCtxtPtr ctxt;
12493
0
    xmlParserInputPtr input;
12494
12495
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12496
0
    if (ctxt == NULL)
12497
0
  return(NULL);
12498
12499
0
    if (data != NULL)
12500
0
  ctxt->_private = data;
12501
12502
0
    if (recovery) {
12503
0
        ctxt->options |= XML_PARSE_RECOVER;
12504
0
        ctxt->recovery = 1;
12505
0
    }
12506
12507
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
12508
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0);
12509
0
    else
12510
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0);
12511
12512
0
    if (input != NULL)
12513
0
        ret = xmlCtxtParseDocument(ctxt, input);
12514
12515
0
    xmlFreeParserCtxt(ctxt);
12516
0
    return(ret);
12517
0
}
12518
12519
/**
12520
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12521
 * compressed document is provided by default if found at compile-time.
12522
 * It use the given SAX function block to handle the parsing callback.
12523
 * If sax is NULL, fallback to the default DOM tree building routines.
12524
 *
12525
 * This function uses deprecated global variables to set parser options
12526
 * which default to XML_PARSE_NODICT.
12527
 *
12528
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12529
 *
12530
 * @param sax  the SAX handler block
12531
 * @param filename  the filename
12532
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12533
 *             documents
12534
 * @returns the resulting document tree
12535
 */
12536
12537
xmlDoc *
12538
xmlSAXParseFile(xmlSAXHandler *sax, const char *filename,
12539
0
                          int recovery) {
12540
0
    return(xmlSAXParseFileWithData(sax,filename,recovery,NULL));
12541
0
}
12542
12543
/**
12544
 * Parse an XML in-memory document and build a tree.
12545
 * In the case the document is not Well Formed, a attempt to build a
12546
 * tree is tried anyway
12547
 *
12548
 * This function uses deprecated global variables to set parser options
12549
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12550
 *
12551
 * @deprecated Use #xmlReadDoc with XML_PARSE_RECOVER.
12552
 *
12553
 * @param cur  a pointer to an array of xmlChar
12554
 * @returns the resulting document tree or NULL in case of failure
12555
 */
12556
12557
xmlDoc *
12558
0
xmlRecoverDoc(const xmlChar *cur) {
12559
0
    return(xmlSAXParseDoc(NULL, cur, 1));
12560
0
}
12561
12562
/**
12563
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12564
 * compressed document is provided by default if found at compile-time.
12565
 *
12566
 * This function uses deprecated global variables to set parser options
12567
 * which default to XML_PARSE_NODICT.
12568
 *
12569
 * @deprecated Use #xmlReadFile.
12570
 *
12571
 * @param filename  the filename
12572
 * @returns the resulting document tree if the file was wellformed,
12573
 * NULL otherwise.
12574
 */
12575
12576
xmlDoc *
12577
0
xmlParseFile(const char *filename) {
12578
0
    return(xmlSAXParseFile(NULL, filename, 0));
12579
0
}
12580
12581
/**
12582
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12583
 * compressed document is provided by default if found at compile-time.
12584
 * In the case the document is not Well Formed, it attempts to build
12585
 * a tree anyway
12586
 *
12587
 * This function uses deprecated global variables to set parser options
12588
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12589
 *
12590
 * @deprecated Use #xmlReadFile with XML_PARSE_RECOVER.
12591
 *
12592
 * @param filename  the filename
12593
 * @returns the resulting document tree or NULL in case of failure
12594
 */
12595
12596
xmlDoc *
12597
0
xmlRecoverFile(const char *filename) {
12598
0
    return(xmlSAXParseFile(NULL, filename, 1));
12599
0
}
12600
12601
12602
/**
12603
 * Setup the parser context to parse a new buffer; Clears any prior
12604
 * contents from the parser context. The buffer parameter must not be
12605
 * NULL, but the filename parameter can be
12606
 *
12607
 * @deprecated Don't use.
12608
 *
12609
 * @param ctxt  an XML parser context
12610
 * @param buffer  a xmlChar * buffer
12611
 * @param filename  a file name
12612
 */
12613
void
12614
xmlSetupParserForBuffer(xmlParserCtxt *ctxt, const xmlChar* buffer,
12615
                             const char* filename)
12616
0
{
12617
0
    xmlParserInputPtr input;
12618
12619
0
    if ((ctxt == NULL) || (buffer == NULL))
12620
0
        return;
12621
12622
0
    xmlCtxtReset(ctxt);
12623
12624
0
    input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer,
12625
0
                                      NULL, 0);
12626
0
    if (input == NULL)
12627
0
        return;
12628
0
    if (xmlCtxtPushInput(ctxt, input) < 0)
12629
0
        xmlFreeInputStream(input);
12630
0
}
12631
12632
/**
12633
 * Parse an XML file and call the given SAX handler routines.
12634
 * Automatic support for ZLIB/Compress compressed document is provided
12635
 *
12636
 * This function uses deprecated global variables to set parser options
12637
 * which default to XML_PARSE_NODICT.
12638
 *
12639
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12640
 *
12641
 * @param sax  a SAX handler
12642
 * @param user_data  The user data returned on SAX callbacks
12643
 * @param filename  a file name
12644
 * @returns 0 in case of success or a error number otherwise
12645
 */
12646
int
12647
xmlSAXUserParseFile(xmlSAXHandler *sax, void *user_data,
12648
0
                    const char *filename) {
12649
0
    int ret = 0;
12650
0
    xmlParserCtxtPtr ctxt;
12651
12652
0
    ctxt = xmlCreateFileParserCtxt(filename);
12653
0
    if (ctxt == NULL) return -1;
12654
0
    if (sax != NULL) {
12655
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12656
0
            *ctxt->sax = *sax;
12657
0
        } else {
12658
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12659
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12660
0
        }
12661
0
  ctxt->userData = user_data;
12662
0
    }
12663
12664
0
    xmlParseDocument(ctxt);
12665
12666
0
    if (ctxt->wellFormed)
12667
0
  ret = 0;
12668
0
    else {
12669
0
        if (ctxt->errNo != 0)
12670
0
      ret = ctxt->errNo;
12671
0
  else
12672
0
      ret = -1;
12673
0
    }
12674
0
    if (ctxt->myDoc != NULL) {
12675
0
        xmlFreeDoc(ctxt->myDoc);
12676
0
  ctxt->myDoc = NULL;
12677
0
    }
12678
0
    xmlFreeParserCtxt(ctxt);
12679
12680
0
    return ret;
12681
0
}
12682
#endif /* LIBXML_SAX1_ENABLED */
12683
12684
/************************************************************************
12685
 *                  *
12686
 *    Front ends when parsing from memory     *
12687
 *                  *
12688
 ************************************************************************/
12689
12690
/**
12691
 * Create a parser context for an XML in-memory document. The input buffer
12692
 * must not contain a terminating null byte.
12693
 *
12694
 * @param buffer  a pointer to a char array
12695
 * @param size  the size of the array
12696
 * @returns the new parser context or NULL
12697
 */
12698
xmlParserCtxt *
12699
0
xmlCreateMemoryParserCtxt(const char *buffer, int size) {
12700
0
    xmlParserCtxtPtr ctxt;
12701
0
    xmlParserInputPtr input;
12702
12703
0
    if (size < 0)
12704
0
  return(NULL);
12705
12706
0
    ctxt = xmlNewParserCtxt();
12707
0
    if (ctxt == NULL)
12708
0
  return(NULL);
12709
12710
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0);
12711
0
    if (input == NULL) {
12712
0
  xmlFreeParserCtxt(ctxt);
12713
0
  return(NULL);
12714
0
    }
12715
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12716
0
        xmlFreeInputStream(input);
12717
0
        xmlFreeParserCtxt(ctxt);
12718
0
        return(NULL);
12719
0
    }
12720
12721
0
    return(ctxt);
12722
0
}
12723
12724
#ifdef LIBXML_SAX1_ENABLED
12725
/**
12726
 * Parse an XML in-memory block and use the given SAX function block
12727
 * to handle the parsing callback. If sax is NULL, fallback to the default
12728
 * DOM tree building routines.
12729
 *
12730
 * This function uses deprecated global variables to set parser options
12731
 * which default to XML_PARSE_NODICT.
12732
 *
12733
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12734
 *
12735
 * User data (void *) is stored within the parser context in the
12736
 * context's _private member, so it is available nearly everywhere in libxml
12737
 *
12738
 * @param sax  the SAX handler block
12739
 * @param buffer  an pointer to a char array
12740
 * @param size  the size of the array
12741
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12742
 *             documents
12743
 * @param data  the userdata
12744
 * @returns the resulting document tree
12745
 */
12746
12747
xmlDoc *
12748
xmlSAXParseMemoryWithData(xmlSAXHandler *sax, const char *buffer,
12749
0
                          int size, int recovery, void *data) {
12750
0
    xmlDocPtr ret = NULL;
12751
0
    xmlParserCtxtPtr ctxt;
12752
0
    xmlParserInputPtr input;
12753
12754
0
    if (size < 0)
12755
0
        return(NULL);
12756
12757
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12758
0
    if (ctxt == NULL)
12759
0
        return(NULL);
12760
12761
0
    if (data != NULL)
12762
0
  ctxt->_private=data;
12763
12764
0
    if (recovery) {
12765
0
        ctxt->options |= XML_PARSE_RECOVER;
12766
0
        ctxt->recovery = 1;
12767
0
    }
12768
12769
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL,
12770
0
                                      XML_INPUT_BUF_STATIC);
12771
12772
0
    if (input != NULL)
12773
0
        ret = xmlCtxtParseDocument(ctxt, input);
12774
12775
0
    xmlFreeParserCtxt(ctxt);
12776
0
    return(ret);
12777
0
}
12778
12779
/**
12780
 * Parse an XML in-memory block and use the given SAX function block
12781
 * to handle the parsing callback. If sax is NULL, fallback to the default
12782
 * DOM tree building routines.
12783
 *
12784
 * This function uses deprecated global variables to set parser options
12785
 * which default to XML_PARSE_NODICT.
12786
 *
12787
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12788
 *
12789
 * @param sax  the SAX handler block
12790
 * @param buffer  an pointer to a char array
12791
 * @param size  the size of the array
12792
 * @param recovery  work in recovery mode, i.e. tries to read not Well Formed
12793
 *             documents
12794
 * @returns the resulting document tree
12795
 */
12796
xmlDoc *
12797
xmlSAXParseMemory(xmlSAXHandler *sax, const char *buffer,
12798
0
            int size, int recovery) {
12799
0
    return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL);
12800
0
}
12801
12802
/**
12803
 * Parse an XML in-memory block and build a tree.
12804
 *
12805
 * This function uses deprecated global variables to set parser options
12806
 * which default to XML_PARSE_NODICT.
12807
 *
12808
 * @deprecated Use #xmlReadMemory.
12809
 *
12810
 * @param buffer  an pointer to a char array
12811
 * @param size  the size of the array
12812
 * @returns the resulting document tree
12813
 */
12814
12815
0
xmlDoc *xmlParseMemory(const char *buffer, int size) {
12816
0
   return(xmlSAXParseMemory(NULL, buffer, size, 0));
12817
0
}
12818
12819
/**
12820
 * Parse an XML in-memory block and build a tree.
12821
 * In the case the document is not Well Formed, an attempt to
12822
 * build a tree is tried anyway
12823
 *
12824
 * This function uses deprecated global variables to set parser options
12825
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12826
 *
12827
 * @deprecated Use #xmlReadMemory with XML_PARSE_RECOVER.
12828
 *
12829
 * @param buffer  an pointer to a char array
12830
 * @param size  the size of the array
12831
 * @returns the resulting document tree or NULL in case of error
12832
 */
12833
12834
0
xmlDoc *xmlRecoverMemory(const char *buffer, int size) {
12835
0
   return(xmlSAXParseMemory(NULL, buffer, size, 1));
12836
0
}
12837
12838
/**
12839
 * Parse an XML in-memory buffer and call the given SAX handler routines.
12840
 *
12841
 * This function uses deprecated global variables to set parser options
12842
 * which default to XML_PARSE_NODICT.
12843
 *
12844
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12845
 *
12846
 * @param sax  a SAX handler
12847
 * @param user_data  The user data returned on SAX callbacks
12848
 * @param buffer  an in-memory XML document input
12849
 * @param size  the length of the XML document in bytes
12850
 * @returns 0 in case of success or a error number otherwise
12851
 */
12852
int xmlSAXUserParseMemory(xmlSAXHandler *sax, void *user_data,
12853
0
        const char *buffer, int size) {
12854
0
    int ret = 0;
12855
0
    xmlParserCtxtPtr ctxt;
12856
12857
0
    ctxt = xmlCreateMemoryParserCtxt(buffer, size);
12858
0
    if (ctxt == NULL) return -1;
12859
0
    if (sax != NULL) {
12860
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12861
0
            *ctxt->sax = *sax;
12862
0
        } else {
12863
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12864
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12865
0
        }
12866
0
  ctxt->userData = user_data;
12867
0
    }
12868
12869
0
    xmlParseDocument(ctxt);
12870
12871
0
    if (ctxt->wellFormed)
12872
0
  ret = 0;
12873
0
    else {
12874
0
        if (ctxt->errNo != 0)
12875
0
      ret = ctxt->errNo;
12876
0
  else
12877
0
      ret = -1;
12878
0
    }
12879
0
    if (ctxt->myDoc != NULL) {
12880
0
        xmlFreeDoc(ctxt->myDoc);
12881
0
  ctxt->myDoc = NULL;
12882
0
    }
12883
0
    xmlFreeParserCtxt(ctxt);
12884
12885
0
    return ret;
12886
0
}
12887
#endif /* LIBXML_SAX1_ENABLED */
12888
12889
/**
12890
 * Creates a parser context for an XML in-memory document.
12891
 *
12892
 * @param str  a pointer to an array of xmlChar
12893
 * @returns the new parser context or NULL
12894
 */
12895
xmlParserCtxt *
12896
0
xmlCreateDocParserCtxt(const xmlChar *str) {
12897
0
    xmlParserCtxtPtr ctxt;
12898
0
    xmlParserInputPtr input;
12899
12900
0
    ctxt = xmlNewParserCtxt();
12901
0
    if (ctxt == NULL)
12902
0
  return(NULL);
12903
12904
0
    input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0);
12905
0
    if (input == NULL) {
12906
0
  xmlFreeParserCtxt(ctxt);
12907
0
  return(NULL);
12908
0
    }
12909
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12910
0
        xmlFreeInputStream(input);
12911
0
        xmlFreeParserCtxt(ctxt);
12912
0
        return(NULL);
12913
0
    }
12914
12915
0
    return(ctxt);
12916
0
}
12917
12918
#ifdef LIBXML_SAX1_ENABLED
12919
/**
12920
 * Parse an XML in-memory document and build a tree.
12921
 * It use the given SAX function block to handle the parsing callback.
12922
 * If sax is NULL, fallback to the default DOM tree building routines.
12923
 *
12924
 * This function uses deprecated global variables to set parser options
12925
 * which default to XML_PARSE_NODICT.
12926
 *
12927
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadDoc.
12928
 *
12929
 * @param sax  the SAX handler block
12930
 * @param cur  a pointer to an array of xmlChar
12931
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12932
 *             documents
12933
 * @returns the resulting document tree
12934
 */
12935
12936
xmlDoc *
12937
0
xmlSAXParseDoc(xmlSAXHandler *sax, const xmlChar *cur, int recovery) {
12938
0
    xmlDocPtr ret;
12939
0
    xmlParserCtxtPtr ctxt;
12940
0
    xmlSAXHandlerPtr oldsax = NULL;
12941
12942
0
    if (cur == NULL) return(NULL);
12943
12944
12945
0
    ctxt = xmlCreateDocParserCtxt(cur);
12946
0
    if (ctxt == NULL) return(NULL);
12947
0
    if (sax != NULL) {
12948
0
        oldsax = ctxt->sax;
12949
0
        ctxt->sax = sax;
12950
0
        ctxt->userData = NULL;
12951
0
    }
12952
12953
0
    xmlParseDocument(ctxt);
12954
0
    if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc;
12955
0
    else {
12956
0
       ret = NULL;
12957
0
       xmlFreeDoc(ctxt->myDoc);
12958
0
       ctxt->myDoc = NULL;
12959
0
    }
12960
0
    if (sax != NULL)
12961
0
  ctxt->sax = oldsax;
12962
0
    xmlFreeParserCtxt(ctxt);
12963
12964
0
    return(ret);
12965
0
}
12966
12967
/**
12968
 * Parse an XML in-memory document and build a tree.
12969
 *
12970
 * This function uses deprecated global variables to set parser options
12971
 * which default to XML_PARSE_NODICT.
12972
 *
12973
 * @deprecated Use #xmlReadDoc.
12974
 *
12975
 * @param cur  a pointer to an array of xmlChar
12976
 * @returns the resulting document tree
12977
 */
12978
12979
xmlDoc *
12980
0
xmlParseDoc(const xmlChar *cur) {
12981
0
    return(xmlSAXParseDoc(NULL, cur, 0));
12982
0
}
12983
#endif /* LIBXML_SAX1_ENABLED */
12984
12985
/************************************************************************
12986
 *                  *
12987
 *  New set (2.6.0) of simpler and more flexible APIs   *
12988
 *                  *
12989
 ************************************************************************/
12990
12991
/**
12992
 * Reset a parser context
12993
 *
12994
 * @param ctxt  an XML parser context
12995
 */
12996
void
12997
xmlCtxtReset(xmlParserCtxt *ctxt)
12998
0
{
12999
0
    xmlParserInputPtr input;
13000
13001
0
    if (ctxt == NULL)
13002
0
        return;
13003
13004
0
    while ((input = xmlCtxtPopInput(ctxt)) != NULL) { /* Non consuming */
13005
0
        xmlFreeInputStream(input);
13006
0
    }
13007
0
    ctxt->inputNr = 0;
13008
0
    ctxt->input = NULL;
13009
13010
0
    ctxt->spaceNr = 0;
13011
0
    if (ctxt->spaceTab != NULL) {
13012
0
  ctxt->spaceTab[0] = -1;
13013
0
  ctxt->space = &ctxt->spaceTab[0];
13014
0
    } else {
13015
0
        ctxt->space = NULL;
13016
0
    }
13017
13018
13019
0
    ctxt->nodeNr = 0;
13020
0
    ctxt->node = NULL;
13021
13022
0
    ctxt->nameNr = 0;
13023
0
    ctxt->name = NULL;
13024
13025
0
    ctxt->nsNr = 0;
13026
0
    xmlParserNsReset(ctxt->nsdb);
13027
13028
0
    if (ctxt->version != NULL) {
13029
0
        xmlFree(ctxt->version);
13030
0
        ctxt->version = NULL;
13031
0
    }
13032
0
    if (ctxt->encoding != NULL) {
13033
0
        xmlFree(ctxt->encoding);
13034
0
        ctxt->encoding = NULL;
13035
0
    }
13036
0
    if (ctxt->extSubURI != NULL) {
13037
0
        xmlFree(ctxt->extSubURI);
13038
0
        ctxt->extSubURI = NULL;
13039
0
    }
13040
0
    if (ctxt->extSubSystem != NULL) {
13041
0
        xmlFree(ctxt->extSubSystem);
13042
0
        ctxt->extSubSystem = NULL;
13043
0
    }
13044
0
    if (ctxt->directory != NULL) {
13045
0
        xmlFree(ctxt->directory);
13046
0
        ctxt->directory = NULL;
13047
0
    }
13048
13049
0
    if (ctxt->myDoc != NULL)
13050
0
        xmlFreeDoc(ctxt->myDoc);
13051
0
    ctxt->myDoc = NULL;
13052
13053
0
    ctxt->standalone = -1;
13054
0
    ctxt->hasExternalSubset = 0;
13055
0
    ctxt->hasPErefs = 0;
13056
0
    ctxt->html = ctxt->html ? 1 : 0;
13057
0
    ctxt->instate = XML_PARSER_START;
13058
13059
0
    ctxt->wellFormed = 1;
13060
0
    ctxt->nsWellFormed = 1;
13061
0
    ctxt->disableSAX = 0;
13062
0
    ctxt->valid = 1;
13063
0
    ctxt->record_info = 0;
13064
0
    ctxt->checkIndex = 0;
13065
0
    ctxt->endCheckState = 0;
13066
0
    ctxt->inSubset = 0;
13067
0
    ctxt->errNo = XML_ERR_OK;
13068
0
    ctxt->depth = 0;
13069
0
    ctxt->catalogs = NULL;
13070
0
    ctxt->sizeentities = 0;
13071
0
    ctxt->sizeentcopy = 0;
13072
0
    xmlInitNodeInfoSeq(&ctxt->node_seq);
13073
13074
0
    if (ctxt->attsDefault != NULL) {
13075
0
        xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator);
13076
0
        ctxt->attsDefault = NULL;
13077
0
    }
13078
0
    if (ctxt->attsSpecial != NULL) {
13079
0
        xmlHashFree(ctxt->attsSpecial, NULL);
13080
0
        ctxt->attsSpecial = NULL;
13081
0
    }
13082
13083
0
#ifdef LIBXML_CATALOG_ENABLED
13084
0
    if (ctxt->catalogs != NULL)
13085
0
  xmlCatalogFreeLocal(ctxt->catalogs);
13086
0
#endif
13087
0
    ctxt->nbErrors = 0;
13088
0
    ctxt->nbWarnings = 0;
13089
0
    if (ctxt->lastError.code != XML_ERR_OK)
13090
0
        xmlResetError(&ctxt->lastError);
13091
0
}
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
43.6k
{
13132
43.6k
    int allMask;
13133
13134
43.6k
    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
43.6k
    allMask = XML_PARSE_RECOVER |
13145
43.6k
              XML_PARSE_NOENT |
13146
43.6k
              XML_PARSE_DTDLOAD |
13147
43.6k
              XML_PARSE_DTDATTR |
13148
43.6k
              XML_PARSE_DTDVALID |
13149
43.6k
              XML_PARSE_NOERROR |
13150
43.6k
              XML_PARSE_NOWARNING |
13151
43.6k
              XML_PARSE_PEDANTIC |
13152
43.6k
              XML_PARSE_NOBLANKS |
13153
43.6k
#ifdef LIBXML_SAX1_ENABLED
13154
43.6k
              XML_PARSE_SAX1 |
13155
43.6k
#endif
13156
43.6k
              XML_PARSE_NONET |
13157
43.6k
              XML_PARSE_NODICT |
13158
43.6k
              XML_PARSE_NSCLEAN |
13159
43.6k
              XML_PARSE_NOCDATA |
13160
43.6k
              XML_PARSE_COMPACT |
13161
43.6k
              XML_PARSE_OLD10 |
13162
43.6k
              XML_PARSE_HUGE |
13163
43.6k
              XML_PARSE_OLDSAX |
13164
43.6k
              XML_PARSE_IGNORE_ENC |
13165
43.6k
              XML_PARSE_BIG_LINES |
13166
43.6k
              XML_PARSE_NO_XXE |
13167
43.6k
              XML_PARSE_UNZIP |
13168
43.6k
              XML_PARSE_NO_SYS_CATALOG |
13169
43.6k
              XML_PARSE_CATALOG_PI;
13170
13171
43.6k
    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
43.6k
    ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0;
13188
43.6k
    ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0;
13189
43.6k
    ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0;
13190
43.6k
    ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0;
13191
43.6k
    ctxt->loadsubset |= (options & XML_PARSE_SKIP_IDS) ? XML_SKIP_IDS : 0;
13192
43.6k
    ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0;
13193
43.6k
    ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0;
13194
43.6k
    ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1;
13195
43.6k
    ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1;
13196
13197
43.6k
    return(options & ~allMask);
13198
43.6k
}
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
43.6k
{
13267
43.6k
    int keepMask;
13268
13269
43.6k
#ifdef LIBXML_HTML_ENABLED
13270
43.6k
    if ((ctxt != NULL) && (ctxt->html))
13271
0
        return(htmlCtxtUseOptions(ctxt, options));
13272
43.6k
#endif
13273
13274
    /*
13275
     * For historic reasons, some options can only be enabled.
13276
     */
13277
43.6k
    keepMask = XML_PARSE_NOERROR |
13278
43.6k
               XML_PARSE_NOWARNING |
13279
43.6k
               XML_PARSE_NONET |
13280
43.6k
               XML_PARSE_NSCLEAN |
13281
43.6k
               XML_PARSE_NOCDATA |
13282
43.6k
               XML_PARSE_COMPACT |
13283
43.6k
               XML_PARSE_OLD10 |
13284
43.6k
               XML_PARSE_HUGE |
13285
43.6k
               XML_PARSE_OLDSAX |
13286
43.6k
               XML_PARSE_IGNORE_ENC |
13287
43.6k
               XML_PARSE_BIG_LINES;
13288
13289
43.6k
    return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask));
13290
43.6k
}
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
0
{
13325
0
    xmlDocPtr ret = NULL;
13326
13327
0
    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
0
    while (ctxt->inputNr > 0)
13335
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13336
13337
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
13338
0
        xmlFreeInputStream(input);
13339
0
        return(NULL);
13340
0
    }
13341
13342
0
    xmlParseDocument(ctxt);
13343
13344
0
    ret = xmlCtxtGetDocument(ctxt);
13345
13346
    /* assert(ctxt->inputNr == 1); */
13347
0
    while (ctxt->inputNr > 0)
13348
0
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
13349
13350
0
    return(ret);
13351
0
}
13352
13353
/**
13354
 * Convenience function to parse an XML document from a
13355
 * zero-terminated string.
13356
 *
13357
 * See #xmlCtxtReadDoc for details.
13358
 *
13359
 * @param cur  a pointer to a zero terminated string
13360
 * @param URL  base URL (optional)
13361
 * @param encoding  the document encoding (optional)
13362
 * @param options  a combination of xmlParserOption
13363
 * @returns the resulting document tree
13364
 */
13365
xmlDoc *
13366
xmlReadDoc(const xmlChar *cur, const char *URL, const char *encoding,
13367
           int options)
13368
0
{
13369
0
    xmlParserCtxtPtr ctxt;
13370
0
    xmlParserInputPtr input;
13371
0
    xmlDocPtr doc = NULL;
13372
13373
0
    ctxt = xmlNewParserCtxt();
13374
0
    if (ctxt == NULL)
13375
0
        return(NULL);
13376
13377
0
    xmlCtxtUseOptions(ctxt, options);
13378
13379
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding,
13380
0
                                      XML_INPUT_BUF_STATIC);
13381
13382
0
    if (input != NULL)
13383
0
        doc = xmlCtxtParseDocument(ctxt, input);
13384
13385
0
    xmlFreeParserCtxt(ctxt);
13386
0
    return(doc);
13387
0
}
13388
13389
/**
13390
 * Convenience function to parse an XML file from the filesystem
13391
 * or a global, user-defined resource loader.
13392
 *
13393
 * If a "-" filename is passed, the function will read from stdin.
13394
 * This feature is potentially insecure and might be removed from
13395
 * later versions.
13396
 *
13397
 * See #xmlCtxtReadFile for details.
13398
 *
13399
 * @param filename  a file or URL
13400
 * @param encoding  the document encoding (optional)
13401
 * @param options  a combination of xmlParserOption
13402
 * @returns the resulting document tree
13403
 */
13404
xmlDoc *
13405
xmlReadFile(const char *filename, const char *encoding, int options)
13406
0
{
13407
0
    xmlParserCtxtPtr ctxt;
13408
0
    xmlParserInputPtr input;
13409
0
    xmlDocPtr doc = NULL;
13410
13411
0
    ctxt = xmlNewParserCtxt();
13412
0
    if (ctxt == NULL)
13413
0
        return(NULL);
13414
13415
0
    xmlCtxtUseOptions(ctxt, options);
13416
13417
    /*
13418
     * Backward compatibility for users of command line utilities like
13419
     * xmlstarlet expecting "-" to mean stdin. This is dangerous and
13420
     * should be removed at some point.
13421
     */
13422
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
13423
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO,
13424
0
                                      encoding, 0);
13425
0
    else
13426
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13427
13428
0
    if (input != NULL)
13429
0
        doc = xmlCtxtParseDocument(ctxt, input);
13430
13431
0
    xmlFreeParserCtxt(ctxt);
13432
0
    return(doc);
13433
0
}
13434
13435
/**
13436
 * Parse an XML in-memory document and build a tree. The input buffer must
13437
 * not contain a terminating null byte.
13438
 *
13439
 * See #xmlCtxtReadMemory for details.
13440
 *
13441
 * @param buffer  a pointer to a char array
13442
 * @param size  the size of the array
13443
 * @param url  base URL (optional)
13444
 * @param encoding  the document encoding (optional)
13445
 * @param options  a combination of xmlParserOption
13446
 * @returns the resulting document tree
13447
 */
13448
xmlDoc *
13449
xmlReadMemory(const char *buffer, int size, const char *url,
13450
              const char *encoding, int options)
13451
0
{
13452
0
    xmlParserCtxtPtr ctxt;
13453
0
    xmlParserInputPtr input;
13454
0
    xmlDocPtr doc = NULL;
13455
13456
0
    if (size < 0)
13457
0
  return(NULL);
13458
13459
0
    ctxt = xmlNewParserCtxt();
13460
0
    if (ctxt == NULL)
13461
0
        return(NULL);
13462
13463
0
    xmlCtxtUseOptions(ctxt, options);
13464
13465
0
    input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding,
13466
0
                                      XML_INPUT_BUF_STATIC);
13467
13468
0
    if (input != NULL)
13469
0
        doc = xmlCtxtParseDocument(ctxt, input);
13470
13471
0
    xmlFreeParserCtxt(ctxt);
13472
0
    return(doc);
13473
0
}
13474
13475
/**
13476
 * Parse an XML from a file descriptor and build a tree.
13477
 *
13478
 * See #xmlCtxtReadFd for details.
13479
 *
13480
 * NOTE that the file descriptor will not be closed when the
13481
 * context is freed or reset.
13482
 *
13483
 * @param fd  an open file descriptor
13484
 * @param URL  base URL (optional)
13485
 * @param encoding  the document encoding (optional)
13486
 * @param options  a combination of xmlParserOption
13487
 * @returns the resulting document tree
13488
 */
13489
xmlDoc *
13490
xmlReadFd(int fd, const char *URL, const char *encoding, int options)
13491
0
{
13492
0
    xmlParserCtxtPtr ctxt;
13493
0
    xmlParserInputPtr input;
13494
0
    xmlDocPtr doc = NULL;
13495
13496
0
    ctxt = xmlNewParserCtxt();
13497
0
    if (ctxt == NULL)
13498
0
        return(NULL);
13499
13500
0
    xmlCtxtUseOptions(ctxt, options);
13501
13502
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13503
13504
0
    if (input != NULL)
13505
0
        doc = xmlCtxtParseDocument(ctxt, input);
13506
13507
0
    xmlFreeParserCtxt(ctxt);
13508
0
    return(doc);
13509
0
}
13510
13511
/**
13512
 * Parse an XML document from I/O functions and context and build a tree.
13513
 *
13514
 * See #xmlCtxtReadIO for details.
13515
 *
13516
 * @param ioread  an I/O read function
13517
 * @param ioclose  an I/O close function (optional)
13518
 * @param ioctx  an I/O handler
13519
 * @param URL  base URL (optional)
13520
 * @param encoding  the document encoding (optional)
13521
 * @param options  a combination of xmlParserOption
13522
 * @returns the resulting document tree
13523
 */
13524
xmlDoc *
13525
xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose,
13526
          void *ioctx, const char *URL, const char *encoding, int options)
13527
0
{
13528
0
    xmlParserCtxtPtr ctxt;
13529
0
    xmlParserInputPtr input;
13530
0
    xmlDocPtr doc = NULL;
13531
13532
0
    ctxt = xmlNewParserCtxt();
13533
0
    if (ctxt == NULL)
13534
0
        return(NULL);
13535
13536
0
    xmlCtxtUseOptions(ctxt, options);
13537
13538
0
    input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx,
13539
0
                                  encoding, 0);
13540
13541
0
    if (input != NULL)
13542
0
        doc = xmlCtxtParseDocument(ctxt, input);
13543
13544
0
    xmlFreeParserCtxt(ctxt);
13545
0
    return(doc);
13546
0
}
13547
13548
/**
13549
 * Parse an XML in-memory document and build a tree.
13550
 *
13551
 * `URL` is used as base to resolve external entities and for error
13552
 * reporting.
13553
 *
13554
 * @param ctxt  an XML parser context
13555
 * @param str  a pointer to a zero terminated string
13556
 * @param URL  base URL (optional)
13557
 * @param encoding  the document encoding (optional)
13558
 * @param options  a combination of xmlParserOption
13559
 * @returns the resulting document tree
13560
 */
13561
xmlDoc *
13562
xmlCtxtReadDoc(xmlParserCtxt *ctxt, const xmlChar *str,
13563
               const char *URL, const char *encoding, int options)
13564
0
{
13565
0
    xmlParserInputPtr input;
13566
13567
0
    if (ctxt == NULL)
13568
0
        return(NULL);
13569
13570
0
    xmlCtxtReset(ctxt);
13571
0
    xmlCtxtUseOptions(ctxt, options);
13572
13573
0
    input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) str, encoding,
13574
0
                                      XML_INPUT_BUF_STATIC);
13575
0
    if (input == NULL)
13576
0
        return(NULL);
13577
13578
0
    return(xmlCtxtParseDocument(ctxt, input));
13579
0
}
13580
13581
/**
13582
 * Parse an XML file from the filesystem or a global, user-defined
13583
 * resource loader.
13584
 *
13585
 * @param ctxt  an XML parser context
13586
 * @param filename  a file or URL
13587
 * @param encoding  the document encoding (optional)
13588
 * @param options  a combination of xmlParserOption
13589
 * @returns the resulting document tree
13590
 */
13591
xmlDoc *
13592
xmlCtxtReadFile(xmlParserCtxt *ctxt, const char *filename,
13593
                const char *encoding, int options)
13594
0
{
13595
0
    xmlParserInputPtr input;
13596
13597
0
    if (ctxt == NULL)
13598
0
        return(NULL);
13599
13600
0
    xmlCtxtReset(ctxt);
13601
0
    xmlCtxtUseOptions(ctxt, options);
13602
13603
0
    input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0);
13604
0
    if (input == NULL)
13605
0
        return(NULL);
13606
13607
0
    return(xmlCtxtParseDocument(ctxt, input));
13608
0
}
13609
13610
/**
13611
 * Parse an XML in-memory document and build a tree. The input buffer must
13612
 * not contain a terminating null byte.
13613
 *
13614
 * `URL` is used as base to resolve external entities and for error
13615
 * reporting.
13616
 *
13617
 * @param ctxt  an XML parser context
13618
 * @param buffer  a pointer to a char array
13619
 * @param size  the size of the array
13620
 * @param URL  base URL (optional)
13621
 * @param encoding  the document encoding (optional)
13622
 * @param options  a combination of xmlParserOption
13623
 * @returns the resulting document tree
13624
 */
13625
xmlDoc *
13626
xmlCtxtReadMemory(xmlParserCtxt *ctxt, const char *buffer, int size,
13627
                  const char *URL, const char *encoding, int options)
13628
0
{
13629
0
    xmlParserInputPtr input;
13630
13631
0
    if ((ctxt == NULL) || (size < 0))
13632
0
        return(NULL);
13633
13634
0
    xmlCtxtReset(ctxt);
13635
0
    xmlCtxtUseOptions(ctxt, options);
13636
13637
0
    input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding,
13638
0
                                      XML_INPUT_BUF_STATIC);
13639
0
    if (input == NULL)
13640
0
        return(NULL);
13641
13642
0
    return(xmlCtxtParseDocument(ctxt, input));
13643
0
}
13644
13645
/**
13646
 * Parse an XML document from a file descriptor and build a tree.
13647
 *
13648
 * NOTE that the file descriptor will not be closed when the
13649
 * context is freed or reset.
13650
 *
13651
 * `URL` is used as base to resolve external entities and for error
13652
 * reporting.
13653
 *
13654
 * @param ctxt  an XML parser context
13655
 * @param fd  an open file descriptor
13656
 * @param URL  base URL (optional)
13657
 * @param encoding  the document encoding (optional)
13658
 * @param options  a combination of xmlParserOption
13659
 * @returns the resulting document tree
13660
 */
13661
xmlDoc *
13662
xmlCtxtReadFd(xmlParserCtxt *ctxt, int fd,
13663
              const char *URL, const char *encoding, int options)
13664
0
{
13665
0
    xmlParserInputPtr input;
13666
13667
0
    if (ctxt == NULL)
13668
0
        return(NULL);
13669
13670
0
    xmlCtxtReset(ctxt);
13671
0
    xmlCtxtUseOptions(ctxt, options);
13672
13673
0
    input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0);
13674
0
    if (input == NULL)
13675
0
        return(NULL);
13676
13677
0
    return(xmlCtxtParseDocument(ctxt, input));
13678
0
}
13679
13680
/**
13681
 * Parse an XML document from I/O functions and source and build a tree.
13682
 * This reuses the existing `ctxt` parser context
13683
 *
13684
 * `URL` is used as base to resolve external entities and for error
13685
 * reporting.
13686
 *
13687
 * @param ctxt  an XML parser context
13688
 * @param ioread  an I/O read function
13689
 * @param ioclose  an I/O close function
13690
 * @param ioctx  an I/O handler
13691
 * @param URL  the base URL to use for the document
13692
 * @param encoding  the document encoding, or NULL
13693
 * @param options  a combination of xmlParserOption
13694
 * @returns the resulting document tree
13695
 */
13696
xmlDoc *
13697
xmlCtxtReadIO(xmlParserCtxt *ctxt, xmlInputReadCallback ioread,
13698
              xmlInputCloseCallback ioclose, void *ioctx,
13699
        const char *URL,
13700
              const char *encoding, int options)
13701
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