Coverage Report

Created: 2025-12-14 06:11

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
297k
#define NS_INDEX_EMPTY  INT_MAX
81
95.6k
#define NS_INDEX_XML    (INT_MAX - 1)
82
182k
#define URI_HASH_EMPTY  0xD943A04E
83
40.7k
#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
222k
#define XML_MAX_ATTRS 100000000 /* 100 million */
94
95
327k
#define XML_SPECIAL_EXTERNAL    (1 << 20)
96
299k
#define XML_SPECIAL_TYPE_MASK   (XML_SPECIAL_EXTERNAL - 1)
97
98
275k
#define XML_ATTVAL_ALLOC        (1 << 0)
99
55.9k
#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
2.34M
#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
2.34M
#define XML_ENT_FIXED_COST 20
170
171
11.2M
#define XML_PARSER_BIG_BUFFER_SIZE 300
172
850k
#define XML_PARSER_BUFFER_SIZE 100
173
78.0k
#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
412
xmlErrMemory(xmlParserCtxtPtr ctxt) {
225
412
    xmlCtxtErrMemory(ctxt);
226
412
}
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
33.5k
{
239
33.5k
    if (prefix == NULL)
240
10.2k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
241
10.2k
                   XML_ERR_FATAL, localname, NULL, NULL, 0,
242
10.2k
                   "Attribute %s redefined\n", localname);
243
23.3k
    else
244
23.3k
        xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED,
245
23.3k
                   XML_ERR_FATAL, prefix, localname, NULL, 0,
246
23.3k
                   "Attribute %s:%s redefined\n", prefix, localname);
247
33.5k
}
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
6.88M
{
260
6.88M
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
261
6.88M
               NULL, NULL, NULL, 0, "%s", msg);
262
6.88M
}
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
19.4k
{
277
19.4k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING,
278
19.4k
               str1, str2, NULL, 0, msg, str1, str2);
279
19.4k
}
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
325k
{
295
325k
    ctxt->valid = 0;
296
297
325k
    xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR,
298
325k
               str1, str2, NULL, 0, msg, str1, str2);
299
325k
}
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
615k
{
314
615k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
315
615k
               NULL, NULL, NULL, val, msg, val);
316
615k
}
317
318
/**
319
 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
320
 *
321
 * @param ctxt  an XML parser context
322
 * @param error  the error number
323
 * @param msg  the error message
324
 * @param str1  an string info
325
 * @param val  an integer value
326
 * @param str2  an string info
327
 */
328
static void LIBXML_ATTR_FORMAT(3,0)
329
xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error,
330
                  const char *msg, const xmlChar *str1, int val,
331
      const xmlChar *str2)
332
75.3k
{
333
75.3k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
334
75.3k
               str1, str2, NULL, val, msg, str1, val, str2);
335
75.3k
}
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
689k
{
349
689k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL,
350
689k
               val, NULL, NULL, 0, msg, val);
351
689k
}
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
45.3k
{
365
45.3k
    xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR,
366
45.3k
               val, NULL, NULL, 0, msg, val);
367
45.3k
}
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
108k
{
385
108k
    ctxt->nsWellFormed = 0;
386
387
108k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR,
388
108k
               info1, info2, info3, 0, msg, info1, info2, info3);
389
108k
}
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
5.13k
{
407
5.13k
    xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING,
408
5.13k
               info1, info2, info3, 0, msg, info1, info2, info3);
409
5.13k
}
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
2.68M
{
436
2.68M
    unsigned long consumed;
437
2.68M
    unsigned long *expandedSize;
438
2.68M
    xmlParserInputPtr input = ctxt->input;
439
2.68M
    xmlEntityPtr entity = input->entity;
440
441
2.68M
    if ((entity) && (entity->flags & XML_ENT_CHECKED))
442
343k
        return(0);
443
444
    /*
445
     * Compute total consumed bytes so far, including input streams of
446
     * external entities.
447
     */
448
2.34M
    consumed = input->consumed;
449
2.34M
    xmlSaturatedAddSizeT(&consumed, input->cur - input->base);
450
2.34M
    xmlSaturatedAdd(&consumed, ctxt->sizeentities);
451
452
2.34M
    if (entity)
453
55.7k
        expandedSize = &entity->expandedSize;
454
2.28M
    else
455
2.28M
        expandedSize = &ctxt->sizeentcopy;
456
457
    /*
458
     * Add extra cost and some fixed cost.
459
     */
460
2.34M
    xmlSaturatedAdd(expandedSize, extra);
461
2.34M
    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
2.34M
    if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) &&
469
982k
        ((*expandedSize >= ULONG_MAX) ||
470
982k
         (*expandedSize / ctxt->maxAmpl > consumed))) {
471
233
        xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT,
472
233
                       "Maximum entity amplification factor exceeded, see "
473
233
                       "xmlCtxtSetMaxAmplification.\n");
474
233
        return(1);
475
233
    }
476
477
2.34M
    return(0);
478
2.34M
}
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
427k
xmlSBufInit(xmlSBuf *buf, unsigned max) {
683
427k
    buf->mem = NULL;
684
427k
    buf->size = 0;
685
427k
    buf->cap = 0;
686
427k
    buf->max = max;
687
427k
    buf->code = XML_ERR_OK;
688
427k
}
689
690
static int
691
406k
xmlSBufGrow(xmlSBuf *buf, unsigned len) {
692
406k
    xmlChar *mem;
693
406k
    unsigned cap;
694
695
406k
    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
406k
    cap = (buf->size + len) * 2;
702
406k
    if (cap < 240)
703
265k
        cap = 240;
704
705
406k
    mem = xmlRealloc(buf->mem, cap);
706
406k
    if (mem == NULL) {
707
89
        buf->code = XML_ERR_NO_MEMORY;
708
89
        return(-1);
709
89
    }
710
711
406k
    buf->mem = mem;
712
406k
    buf->cap = cap;
713
714
406k
    return(0);
715
406k
}
716
717
static void
718
82.9M
xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) {
719
82.9M
    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
82.9M
    if (buf->cap - buf->size <= len) {
726
400k
        if (xmlSBufGrow(buf, len) < 0)
727
86
            return;
728
400k
    }
729
730
82.9M
    if (len > 0)
731
82.9M
        memcpy(buf->mem + buf->size, str, len);
732
82.9M
    buf->size += len;
733
82.9M
}
734
735
static void
736
74.9M
xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) {
737
74.9M
    xmlSBufAddString(buf, (const xmlChar *) str, len);
738
74.9M
}
739
740
static void
741
1.20M
xmlSBufAddChar(xmlSBuf *buf, int c) {
742
1.20M
    xmlChar *end;
743
744
1.20M
    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
1.20M
    if (buf->cap - buf->size <= 4) {
751
5.97k
        if (xmlSBufGrow(buf, 4) < 0)
752
3
            return;
753
5.97k
    }
754
755
1.20M
    end = buf->mem + buf->size;
756
757
1.20M
    if (c < 0x80) {
758
1.20M
        *end = (xmlChar) c;
759
1.20M
        buf->size += 1;
760
1.20M
    } else {
761
3.47k
        buf->size += xmlCopyCharMultiByte(end, c);
762
3.47k
    }
763
1.20M
}
764
765
static void
766
62.0M
xmlSBufAddReplChar(xmlSBuf *buf) {
767
62.0M
    xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3);
768
62.0M
}
769
770
static void
771
95
xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
772
95
    if (buf->code == XML_ERR_NO_MEMORY)
773
95
        xmlCtxtErrMemory(ctxt);
774
0
    else
775
0
        xmlFatalErr(ctxt, buf->code, errMsg);
776
95
}
777
778
static xmlChar *
779
xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt,
780
258k
              const char *errMsg) {
781
258k
    if (buf->mem == NULL) {
782
14.5k
        buf->mem = xmlMalloc(1);
783
14.5k
        if (buf->mem == NULL) {
784
6
            buf->code = XML_ERR_NO_MEMORY;
785
14.5k
        } else {
786
14.5k
            buf->mem[0] = 0;
787
14.5k
        }
788
244k
    } else {
789
244k
        buf->mem[buf->size] = 0;
790
244k
    }
791
792
258k
    if (buf->code == XML_ERR_OK) {
793
258k
        if (sizeOut != NULL)
794
20.2k
            *sizeOut = buf->size;
795
258k
        return(buf->mem);
796
258k
    }
797
798
59
    xmlSBufReportError(buf, ctxt, errMsg);
799
800
59
    xmlFree(buf->mem);
801
802
59
    if (sizeOut != NULL)
803
9
        *sizeOut = 0;
804
59
    return(NULL);
805
258k
}
806
807
static void
808
161k
xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) {
809
161k
    if (buf->code != XML_ERR_OK)
810
36
        xmlSBufReportError(buf, ctxt, errMsg);
811
812
161k
    xmlFree(buf->mem);
813
161k
}
814
815
static int
816
xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str,
817
78.7M
                    const char *errMsg) {
818
78.7M
    int c = str[0];
819
78.7M
    int c1 = str[1];
820
821
78.7M
    if ((c1 & 0xC0) != 0x80)
822
35.3M
        goto encoding_error;
823
824
43.3M
    if (c < 0xE0) {
825
        /* 2-byte sequence */
826
20.6M
        if (c < 0xC2)
827
20.4M
            goto encoding_error;
828
829
213k
        return(2);
830
22.6M
    } else {
831
22.6M
        int c2 = str[2];
832
833
22.6M
        if ((c2 & 0xC0) != 0x80)
834
9.74k
            goto encoding_error;
835
836
22.6M
        if (c < 0xF0) {
837
            /* 3-byte sequence */
838
22.6M
            if (c == 0xE0) {
839
                /* overlong */
840
2.92M
                if (c1 < 0xA0)
841
1.73k
                    goto encoding_error;
842
19.7M
            } else if (c == 0xED) {
843
                /* surrogate */
844
27.2k
                if (c1 >= 0xA0)
845
276
                    goto encoding_error;
846
19.6M
            } else if (c == 0xEF) {
847
                /* U+FFFE and U+FFFF are invalid Chars */
848
16.6M
                if ((c1 == 0xBF) && (c2 >= 0xBE))
849
514
                    xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg);
850
16.6M
            }
851
852
22.6M
            return(3);
853
22.6M
        } else {
854
            /* 4-byte sequence */
855
25.3k
            if ((str[3] & 0xC0) != 0x80)
856
979
                goto encoding_error;
857
24.3k
            if (c == 0xF0) {
858
                /* overlong */
859
1.38k
                if (c1 < 0x90)
860
856
                    goto encoding_error;
861
22.9k
            } else if (c >= 0xF4) {
862
                /* greater than 0x10FFFF */
863
4.97k
                if ((c > 0xF4) || (c1 >= 0x90))
864
4.63k
                    goto encoding_error;
865
4.97k
            }
866
867
18.8k
            return(4);
868
24.3k
        }
869
22.6M
    }
870
871
55.8M
encoding_error:
872
    /* Only report the first error */
873
55.8M
    if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) {
874
15.8k
        xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL);
875
15.8k
        ctxt->input->flags |= XML_INPUT_ENCODING_ERROR;
876
15.8k
    }
877
878
55.8M
    return(0);
879
43.3M
}
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
40.7k
xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) {
897
40.7k
    xmlSAXHandlerPtr sax;
898
899
    /* Avoid unused variable warning if features are disabled. */
900
40.7k
    (void) sax;
901
902
    /*
903
     * Changing the SAX struct directly is still widespread practice
904
     * in internal and external code.
905
     */
906
40.7k
    if (ctxt == NULL) return;
907
40.7k
    sax = ctxt->sax;
908
40.7k
#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
40.7k
    if (((ctxt->options & XML_PARSE_SAX1) == 0) &&
914
23.4k
        (sax) &&
915
23.4k
        (sax->initialized == XML_SAX2_MAGIC) &&
916
23.4k
        ((sax->startElementNs != NULL) ||
917
0
         (sax->endElementNs != NULL) ||
918
0
         ((sax->startElement == NULL) && (sax->endElement == NULL))))
919
23.4k
        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
40.7k
    ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3);
929
40.7k
    ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5);
930
40.7k
    ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36);
931
40.7k
    if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) ||
932
40.7k
    (ctxt->str_xml_ns == NULL)) {
933
9
        xmlErrMemory(ctxt);
934
9
    }
935
936
40.7k
    xmlDictSetLimit(ctxt->dict,
937
40.7k
                    (ctxt->options & XML_PARSE_HUGE) ?
938
19.6k
                        0 :
939
40.7k
                        XML_MAX_DICTIONARY_LIMIT);
940
941
40.7k
#ifdef LIBXML_VALID_ENABLED
942
40.7k
    if (ctxt->validate)
943
19.5k
        ctxt->vctxt.flags |= XML_VCTXT_VALIDATE;
944
21.1k
    else
945
21.1k
        ctxt->vctxt.flags &= ~XML_VCTXT_VALIDATE;
946
40.7k
#endif /* LIBXML_VALID_ENABLED */
947
40.7k
}
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
27.2k
{
989
27.2k
    if ((src == NULL) || (dst == NULL))
990
0
        return(NULL);
991
992
40.2k
    while (*src == 0x20) src++;
993
6.76M
    while (*src != 0) {
994
6.73M
  if (*src == 0x20) {
995
261k
      while (*src == 0x20) src++;
996
23.1k
      if (*src != 0)
997
22.5k
    *dst++ = 0x20;
998
6.71M
  } else {
999
6.71M
      *dst++ = *src++;
1000
6.71M
  }
1001
6.73M
    }
1002
27.2k
    *dst = 0;
1003
27.2k
    if (dst == src)
1004
21.6k
       return(NULL);
1005
5.63k
    return(dst);
1006
27.2k
}
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
23.3k
               const xmlChar *value) {
1021
23.3k
    xmlDefAttrsPtr defaults;
1022
23.3k
    xmlDefAttr *attr;
1023
23.3k
    int len, expandedSize;
1024
23.3k
    xmlHashedString name;
1025
23.3k
    xmlHashedString prefix;
1026
23.3k
    xmlHashedString hvalue;
1027
23.3k
    const xmlChar *localname;
1028
1029
    /*
1030
     * Allows to detect attribute redefinitions
1031
     */
1032
23.3k
    if (ctxt->attsSpecial != NULL) {
1033
21.3k
        if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL)
1034
14.5k
      return;
1035
21.3k
    }
1036
1037
8.83k
    if (ctxt->attsDefault == NULL) {
1038
2.13k
        ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict);
1039
2.13k
  if (ctxt->attsDefault == NULL)
1040
1
      goto mem_error;
1041
2.13k
    }
1042
1043
    /*
1044
     * split the element name into prefix:localname , the string found
1045
     * are within the DTD and then not associated to namespace names.
1046
     */
1047
8.83k
    localname = xmlSplitQName3(fullname, &len);
1048
8.83k
    if (localname == NULL) {
1049
8.37k
        name = xmlDictLookupHashed(ctxt->dict, fullname, -1);
1050
8.37k
  prefix.name = NULL;
1051
8.37k
    } else {
1052
461
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1053
461
  prefix = xmlDictLookupHashed(ctxt->dict, fullname, len);
1054
461
        if (prefix.name == NULL)
1055
1
            goto mem_error;
1056
461
    }
1057
8.83k
    if (name.name == NULL)
1058
1
        goto mem_error;
1059
1060
    /*
1061
     * make sure there is some storage
1062
     */
1063
8.83k
    defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name);
1064
8.83k
    if ((defaults == NULL) ||
1065
6.24k
        (defaults->nbAttrs >= defaults->maxAttrs)) {
1066
3.16k
        xmlDefAttrsPtr temp;
1067
3.16k
        int newSize;
1068
1069
3.16k
        if (defaults == NULL) {
1070
2.58k
            newSize = 4;
1071
2.58k
        } else {
1072
579
            if ((defaults->maxAttrs >= XML_MAX_ATTRS) ||
1073
579
                ((size_t) defaults->maxAttrs >
1074
579
                     SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults)))
1075
0
                goto mem_error;
1076
1077
579
            if (defaults->maxAttrs > XML_MAX_ATTRS / 2)
1078
0
                newSize = XML_MAX_ATTRS;
1079
579
            else
1080
579
                newSize = defaults->maxAttrs * 2;
1081
579
        }
1082
3.16k
        temp = xmlRealloc(defaults,
1083
3.16k
                          sizeof(*defaults) + newSize * sizeof(xmlDefAttr));
1084
3.16k
  if (temp == NULL)
1085
1
      goto mem_error;
1086
3.16k
        if (defaults == NULL)
1087
2.58k
            temp->nbAttrs = 0;
1088
3.16k
  temp->maxAttrs = newSize;
1089
3.16k
        defaults = temp;
1090
3.16k
  if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name,
1091
3.16k
                          defaults, NULL) < 0) {
1092
0
      xmlFree(defaults);
1093
0
      goto mem_error;
1094
0
  }
1095
3.16k
    }
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
8.83k
    localname = xmlSplitQName3(fullattr, &len);
1102
8.83k
    if (localname == NULL) {
1103
6.25k
        name = xmlDictLookupHashed(ctxt->dict, fullattr, -1);
1104
6.25k
  prefix.name = NULL;
1105
6.25k
    } else {
1106
2.58k
        name = xmlDictLookupHashed(ctxt->dict, localname, -1);
1107
2.58k
  prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len);
1108
2.58k
        if (prefix.name == NULL)
1109
1
            goto mem_error;
1110
2.58k
    }
1111
8.83k
    if (name.name == NULL)
1112
1
        goto mem_error;
1113
1114
    /* intern the string and precompute the end */
1115
8.83k
    len = strlen((const char *) value);
1116
8.83k
    hvalue = xmlDictLookupHashed(ctxt->dict, value, len);
1117
8.83k
    if (hvalue.name == NULL)
1118
1
        goto mem_error;
1119
1120
8.83k
    expandedSize = strlen((const char *) name.name);
1121
8.83k
    if (prefix.name != NULL)
1122
2.58k
        expandedSize += strlen((const char *) prefix.name);
1123
8.83k
    expandedSize += len;
1124
1125
8.83k
    attr = &defaults->attrs[defaults->nbAttrs++];
1126
8.83k
    attr->name = name;
1127
8.83k
    attr->prefix = prefix;
1128
8.83k
    attr->value = hvalue;
1129
8.83k
    attr->valueEnd = hvalue.name + len;
1130
8.83k
    attr->external = PARSER_EXTERNAL(ctxt);
1131
8.83k
    attr->expandedSize = expandedSize;
1132
1133
8.83k
    return;
1134
1135
7
mem_error:
1136
7
    xmlErrMemory(ctxt);
1137
7
}
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
41.3k
{
1153
41.3k
    if (ctxt->attsSpecial == NULL) {
1154
2.72k
        ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict);
1155
2.72k
  if (ctxt->attsSpecial == NULL)
1156
4
      goto mem_error;
1157
2.72k
    }
1158
1159
41.3k
    if (PARSER_EXTERNAL(ctxt))
1160
27.5k
        type |= XML_SPECIAL_EXTERNAL;
1161
1162
41.3k
    if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr,
1163
41.3k
                    XML_INT_TO_PTR(type)) < 0)
1164
1
        goto mem_error;
1165
41.3k
    return;
1166
1167
41.3k
mem_error:
1168
5
    xmlErrMemory(ctxt);
1169
5
}
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
22.7k
                            const xmlChar *unused ATTRIBUTE_UNUSED) {
1178
22.7k
    xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data;
1179
1180
22.7k
    if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) {
1181
1.99k
        xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL);
1182
1.99k
    }
1183
22.7k
}
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
16.8k
{
1195
16.8k
    if (ctxt->attsSpecial == NULL)
1196
14.1k
        return;
1197
1198
2.71k
    xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt);
1199
1200
2.71k
    if (xmlHashSize(ctxt->attsSpecial) == 0) {
1201
180
        xmlHashFree(ctxt->attsSpecial, NULL);
1202
180
        ctxt->attsSpecial = NULL;
1203
180
    }
1204
2.71k
}
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
21.6k
{
1268
21.6k
    const xmlChar *cur = lang, *nxt;
1269
1270
21.6k
    if (cur == NULL)
1271
248
        return (0);
1272
21.4k
    if (((cur[0] == 'i') && (cur[1] == '-')) ||
1273
21.3k
        ((cur[0] == 'I') && (cur[1] == '-')) ||
1274
20.8k
        ((cur[0] == 'x') && (cur[1] == '-')) ||
1275
20.6k
        ((cur[0] == 'X') && (cur[1] == '-'))) {
1276
        /*
1277
         * Still allow IANA code and user code which were coming
1278
         * from the previous version of the XML-1.0 specification
1279
         * it's deprecated but we should not fail
1280
         */
1281
1.03k
        cur += 2;
1282
3.55k
        while (((cur[0] >= 'A') && (cur[0] <= 'Z')) ||
1283
1.83k
               ((cur[0] >= 'a') && (cur[0] <= 'z')))
1284
2.52k
            cur++;
1285
1.03k
        return(cur[0] == 0);
1286
1.03k
    }
1287
20.3k
    nxt = cur;
1288
93.1k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1289
26.0k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1290
72.7k
           nxt++;
1291
20.3k
    if (nxt - cur >= 4) {
1292
        /*
1293
         * Reserved
1294
         */
1295
1.36k
        if ((nxt - cur > 8) || (nxt[0] != 0))
1296
1.06k
            return(0);
1297
303
        return(1);
1298
1.36k
    }
1299
19.0k
    if (nxt - cur < 2)
1300
924
        return(0);
1301
    /* we got an ISO 639 code */
1302
18.1k
    if (nxt[0] == 0)
1303
779
        return(1);
1304
17.3k
    if (nxt[0] != '-')
1305
1.32k
        return(0);
1306
1307
16.0k
    nxt++;
1308
16.0k
    cur = nxt;
1309
    /* now we can have extlang or script or region or variant */
1310
16.0k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1311
3.07k
        goto region_m49;
1312
1313
55.2k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1314
16.0k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1315
42.3k
           nxt++;
1316
12.9k
    if (nxt - cur == 4)
1317
1.75k
        goto script;
1318
11.1k
    if (nxt - cur == 2)
1319
1.18k
        goto region;
1320
9.98k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1321
738
        goto variant;
1322
9.24k
    if (nxt - cur != 3)
1323
1.64k
        return(0);
1324
    /* we parsed an extlang */
1325
7.60k
    if (nxt[0] == 0)
1326
518
        return(1);
1327
7.08k
    if (nxt[0] != '-')
1328
2.25k
        return(0);
1329
1330
4.83k
    nxt++;
1331
4.83k
    cur = nxt;
1332
    /* now we can have script or region or variant */
1333
4.83k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1334
665
        goto region_m49;
1335
1336
17.8k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1337
6.23k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1338
13.6k
           nxt++;
1339
4.17k
    if (nxt - cur == 2)
1340
2.61k
        goto region;
1341
1.55k
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1342
589
        goto variant;
1343
963
    if (nxt - cur != 4)
1344
290
        return(0);
1345
    /* we parsed a script */
1346
2.42k
script:
1347
2.42k
    if (nxt[0] == 0)
1348
550
        return(1);
1349
1.87k
    if (nxt[0] != '-')
1350
303
        return(0);
1351
1352
1.57k
    nxt++;
1353
1.57k
    cur = nxt;
1354
    /* now we can have region or variant */
1355
1.57k
    if ((nxt[0] >= '0') && (nxt[0] <= '9'))
1356
572
        goto region_m49;
1357
1358
3.66k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1359
1.93k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1360
2.67k
           nxt++;
1361
1362
998
    if ((nxt - cur >= 5) && (nxt - cur <= 8))
1363
111
        goto variant;
1364
887
    if (nxt - cur != 2)
1365
506
        return(0);
1366
    /* we parsed a region */
1367
4.59k
region:
1368
4.59k
    if (nxt[0] == 0)
1369
351
        return(1);
1370
4.24k
    if (nxt[0] != '-')
1371
3.05k
        return(0);
1372
1373
1.18k
    nxt++;
1374
1.18k
    cur = nxt;
1375
    /* now we can just have a variant */
1376
8.37k
    while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) ||
1377
5.00k
           ((nxt[0] >= 'a') && (nxt[0] <= 'z')))
1378
7.19k
           nxt++;
1379
1380
1.18k
    if ((nxt - cur < 5) || (nxt - cur > 8))
1381
742
        return(0);
1382
1383
    /* we parsed a variant */
1384
1.88k
variant:
1385
1.88k
    if (nxt[0] == 0)
1386
376
        return(1);
1387
1.50k
    if (nxt[0] != '-')
1388
773
        return(0);
1389
    /* extensions and private use subtags not checked */
1390
733
    return (1);
1391
1392
4.31k
region_m49:
1393
4.31k
    if (((nxt[1] >= '0') && (nxt[1] <= '9')) &&
1394
2.97k
        ((nxt[2] >= '0') && (nxt[2] <= '9'))) {
1395
405
        nxt += 3;
1396
405
        goto region;
1397
405
    }
1398
3.91k
    return(0);
1399
4.31k
}
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
40.1k
xmlParserNsCreate(void) {
1417
40.1k
    xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb));
1418
1419
40.1k
    if (nsdb == NULL)
1420
3
        return(NULL);
1421
40.1k
    memset(nsdb, 0, sizeof(*nsdb));
1422
40.1k
    nsdb->defaultNsIndex = INT_MAX;
1423
1424
40.1k
    return(nsdb);
1425
40.1k
}
1426
1427
/**
1428
 * Free a namespace database.
1429
 *
1430
 * @param nsdb  namespace database
1431
 */
1432
void
1433
40.1k
xmlParserNsFree(xmlParserNsData *nsdb) {
1434
40.1k
    if (nsdb == NULL)
1435
0
        return;
1436
1437
40.1k
    xmlFree(nsdb->extra);
1438
40.1k
    xmlFree(nsdb->hash);
1439
40.1k
    xmlFree(nsdb);
1440
40.1k
}
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
1.01M
xmlParserNsStartElement(xmlParserNsData *nsdb) {
1468
1.01M
    if (nsdb->elementId == UINT_MAX)
1469
0
        return(-1);
1470
1.01M
    nsdb->elementId++;
1471
1472
1.01M
    return(0);
1473
1.01M
}
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
1.05M
                  xmlParserNsBucket **bucketPtr) {
1489
1.05M
    xmlParserNsBucket *bucket, *tombstone;
1490
1.05M
    unsigned index, hashValue;
1491
1492
1.05M
    if (prefix->name == NULL)
1493
610k
        return(ctxt->nsdb->defaultNsIndex);
1494
1495
445k
    if (ctxt->nsdb->hashSize == 0)
1496
30.2k
        return(INT_MAX);
1497
1498
415k
    hashValue = prefix->hashValue;
1499
415k
    index = hashValue & (ctxt->nsdb->hashSize - 1);
1500
415k
    bucket = &ctxt->nsdb->hash[index];
1501
415k
    tombstone = NULL;
1502
1503
591k
    while (bucket->hashValue) {
1504
486k
        if (bucket->index == INT_MAX) {
1505
62.0k
            if (tombstone == NULL)
1506
53.3k
                tombstone = bucket;
1507
424k
        } else if (bucket->hashValue == hashValue) {
1508
310k
            if (ctxt->nsTab[bucket->index * 2] == prefix->name) {
1509
310k
                if (bucketPtr != NULL)
1510
128k
                    *bucketPtr = bucket;
1511
310k
                return(bucket->index);
1512
310k
            }
1513
310k
        }
1514
1515
175k
        index++;
1516
175k
        bucket++;
1517
175k
        if (index == ctxt->nsdb->hashSize) {
1518
62.9k
            index = 0;
1519
62.9k
            bucket = ctxt->nsdb->hash;
1520
62.9k
        }
1521
175k
    }
1522
1523
104k
    if (bucketPtr != NULL)
1524
43.0k
        *bucketPtr = tombstone ? tombstone : bucket;
1525
104k
    return(INT_MAX);
1526
415k
}
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
652k
xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) {
1537
652k
    const xmlChar *ret;
1538
652k
    int nsIndex;
1539
1540
652k
    if (prefix->name == ctxt->str_xml)
1541
348
        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
652k
    nsIndex = xmlParserNsLookup(ctxt, prefix, NULL);
1548
652k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1549
569k
        return(NULL);
1550
1551
82.9k
    ret = ctxt->nsTab[nsIndex * 2 + 1];
1552
82.9k
    if (ret[0] == 0)
1553
4.95k
        ret = NULL;
1554
82.9k
    return(ret);
1555
652k
}
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
77.6k
xmlParserNsLookupSax(xmlParserCtxt *ctxt, const xmlChar *prefix) {
1567
77.6k
    xmlHashedString hprefix;
1568
77.6k
    int nsIndex;
1569
1570
77.6k
    if (prefix == ctxt->str_xml)
1571
40.9k
        return(NULL);
1572
1573
36.6k
    hprefix.name = prefix;
1574
36.6k
    if (prefix != NULL)
1575
29.4k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1576
7.22k
    else
1577
7.22k
        hprefix.hashValue = 0;
1578
36.6k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1579
36.6k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1580
0
        return(NULL);
1581
1582
36.6k
    return(ctxt->nsdb->extra[nsIndex].saxData);
1583
36.6k
}
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
109k
                     void *saxData) {
1598
109k
    xmlHashedString hprefix;
1599
109k
    int nsIndex;
1600
1601
109k
    if (prefix == ctxt->str_xml)
1602
0
        return(-1);
1603
1604
109k
    hprefix.name = prefix;
1605
109k
    if (prefix != NULL)
1606
87.5k
        hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix);
1607
21.6k
    else
1608
21.6k
        hprefix.hashValue = 0;
1609
109k
    nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL);
1610
109k
    if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex))
1611
0
        return(-1);
1612
1613
109k
    ctxt->nsdb->extra[nsIndex].saxData = saxData;
1614
109k
    return(0);
1615
109k
}
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
12.6k
xmlParserNsGrow(xmlParserCtxtPtr ctxt) {
1625
12.6k
    const xmlChar **table;
1626
12.6k
    xmlParserNsExtra *extra;
1627
12.6k
    int newSize;
1628
1629
12.6k
    newSize = xmlGrowCapacity(ctxt->nsMax,
1630
12.6k
                              sizeof(table[0]) + sizeof(extra[0]),
1631
12.6k
                              16, XML_MAX_ITEMS);
1632
12.6k
    if (newSize < 0)
1633
0
        goto error;
1634
1635
12.6k
    table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0]));
1636
12.6k
    if (table == NULL)
1637
10
        goto error;
1638
12.6k
    ctxt->nsTab = table;
1639
1640
12.6k
    extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0]));
1641
12.6k
    if (extra == NULL)
1642
12
        goto error;
1643
12.6k
    ctxt->nsdb->extra = extra;
1644
1645
12.6k
    ctxt->nsMax = newSize;
1646
12.6k
    return(0);
1647
1648
22
error:
1649
22
    xmlErrMemory(ctxt);
1650
22
    return(-1);
1651
12.6k
}
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
135k
                const xmlHashedString *uri, void *saxData, int defAttr) {
1667
135k
    xmlParserNsBucket *bucket = NULL;
1668
135k
    xmlParserNsExtra *extra;
1669
135k
    const xmlChar **ns;
1670
135k
    unsigned hashValue, nsIndex, oldIndex;
1671
1672
135k
    if ((prefix != NULL) && (prefix->name == ctxt->str_xml))
1673
233
        return(0);
1674
1675
135k
    if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) {
1676
22
        xmlErrMemory(ctxt);
1677
22
        return(-1);
1678
22
    }
1679
1680
    /*
1681
     * Default namespace and 'xml' namespace
1682
     */
1683
135k
    if ((prefix == NULL) || (prefix->name == NULL)) {
1684
25.7k
        oldIndex = ctxt->nsdb->defaultNsIndex;
1685
1686
25.7k
        if (oldIndex != INT_MAX) {
1687
22.0k
            extra = &ctxt->nsdb->extra[oldIndex];
1688
1689
22.0k
            if (extra->elementId == ctxt->nsdb->elementId) {
1690
436
                if (defAttr == 0)
1691
227
                    xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns");
1692
436
                return(0);
1693
436
            }
1694
1695
21.6k
            if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1696
3.74k
                (uri->name == ctxt->nsTab[oldIndex * 2 + 1]))
1697
1.73k
                return(0);
1698
21.6k
        }
1699
1700
23.6k
        ctxt->nsdb->defaultNsIndex = ctxt->nsNr;
1701
23.6k
        goto populate_entry;
1702
25.7k
    }
1703
1704
    /*
1705
     * Hash table lookup
1706
     */
1707
109k
    oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket);
1708
109k
    if (oldIndex != INT_MAX) {
1709
59.9k
        extra = &ctxt->nsdb->extra[oldIndex];
1710
1711
        /*
1712
         * Check for duplicate definitions on the same element.
1713
         */
1714
59.9k
        if (extra->elementId == ctxt->nsdb->elementId) {
1715
1.12k
            if (defAttr == 0)
1716
926
                xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name);
1717
1.12k
            return(0);
1718
1.12k
        }
1719
1720
58.8k
        if ((ctxt->options & XML_PARSE_NSCLEAN) &&
1721
5.99k
            (uri->name == ctxt->nsTab[bucket->index * 2 + 1]))
1722
4.56k
            return(0);
1723
1724
54.2k
        bucket->index = ctxt->nsNr;
1725
54.2k
        goto populate_entry;
1726
58.8k
    }
1727
1728
    /*
1729
     * Insert new bucket
1730
     */
1731
1732
49.5k
    hashValue = prefix->hashValue;
1733
1734
    /*
1735
     * Grow hash table, 50% fill factor
1736
     */
1737
49.5k
    if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) {
1738
7.09k
        xmlParserNsBucket *newHash;
1739
7.09k
        unsigned newSize, i, index;
1740
1741
7.09k
        if (ctxt->nsdb->hashSize > UINT_MAX / 2) {
1742
0
            xmlErrMemory(ctxt);
1743
0
            return(-1);
1744
0
        }
1745
7.09k
        newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16;
1746
7.09k
        newHash = xmlMalloc(newSize * sizeof(newHash[0]));
1747
7.09k
        if (newHash == NULL) {
1748
1
            xmlErrMemory(ctxt);
1749
1
            return(-1);
1750
1
        }
1751
7.08k
        memset(newHash, 0, newSize * sizeof(newHash[0]));
1752
1753
140k
        for (i = 0; i < ctxt->nsdb->hashSize; i++) {
1754
132k
            unsigned hv = ctxt->nsdb->hash[i].hashValue;
1755
132k
            unsigned newIndex;
1756
1757
132k
            if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX))
1758
131k
                continue;
1759
1.04k
            newIndex = hv & (newSize - 1);
1760
1761
1.61k
            while (newHash[newIndex].hashValue != 0) {
1762
562
                newIndex++;
1763
562
                if (newIndex == newSize)
1764
119
                    newIndex = 0;
1765
562
            }
1766
1767
1.04k
            newHash[newIndex] = ctxt->nsdb->hash[i];
1768
1.04k
        }
1769
1770
7.08k
        xmlFree(ctxt->nsdb->hash);
1771
7.08k
        ctxt->nsdb->hash = newHash;
1772
7.08k
        ctxt->nsdb->hashSize = newSize;
1773
1774
        /*
1775
         * Relookup
1776
         */
1777
7.08k
        index = hashValue & (newSize - 1);
1778
1779
7.30k
        while (newHash[index].hashValue != 0) {
1780
213
            index++;
1781
213
            if (index == newSize)
1782
39
                index = 0;
1783
213
        }
1784
1785
7.08k
        bucket = &newHash[index];
1786
7.08k
    }
1787
1788
49.5k
    bucket->hashValue = hashValue;
1789
49.5k
    bucket->index = ctxt->nsNr;
1790
49.5k
    ctxt->nsdb->hashElems++;
1791
49.5k
    oldIndex = INT_MAX;
1792
1793
127k
populate_entry:
1794
127k
    nsIndex = ctxt->nsNr;
1795
1796
127k
    ns = &ctxt->nsTab[nsIndex * 2];
1797
127k
    ns[0] = prefix ? prefix->name : NULL;
1798
127k
    ns[1] = uri->name;
1799
1800
127k
    extra = &ctxt->nsdb->extra[nsIndex];
1801
127k
    extra->saxData = saxData;
1802
127k
    extra->prefixHashValue = prefix ? prefix->hashValue : 0;
1803
127k
    extra->uriHashValue = uri->hashValue;
1804
127k
    extra->elementId = ctxt->nsdb->elementId;
1805
127k
    extra->oldIndex = oldIndex;
1806
1807
127k
    ctxt->nsNr++;
1808
1809
127k
    return(1);
1810
49.5k
}
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
56.4k
{
1822
56.4k
    int i;
1823
1824
    /* assert(nr <= ctxt->nsNr); */
1825
1826
135k
    for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) {
1827
79.2k
        const xmlChar *prefix = ctxt->nsTab[i * 2];
1828
79.2k
        xmlParserNsExtra *extra = &ctxt->nsdb->extra[i];
1829
1830
79.2k
        if (prefix == NULL) {
1831
10.3k
            ctxt->nsdb->defaultNsIndex = extra->oldIndex;
1832
68.9k
        } else {
1833
68.9k
            xmlHashedString hprefix;
1834
68.9k
            xmlParserNsBucket *bucket = NULL;
1835
1836
68.9k
            hprefix.name = prefix;
1837
68.9k
            hprefix.hashValue = extra->prefixHashValue;
1838
68.9k
            xmlParserNsLookup(ctxt, &hprefix, &bucket);
1839
            /* assert(bucket && bucket->hashValue); */
1840
68.9k
            bucket->index = extra->oldIndex;
1841
68.9k
        }
1842
79.2k
    }
1843
1844
56.4k
    ctxt->nsNr -= nr;
1845
56.4k
    return(nr);
1846
56.4k
}
1847
1848
static int
1849
20.0k
xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) {
1850
20.0k
    const xmlChar **atts;
1851
20.0k
    unsigned *attallocs;
1852
20.0k
    int newSize;
1853
1854
20.0k
    newSize = xmlGrowCapacity(ctxt->maxatts / 5,
1855
20.0k
                              sizeof(atts[0]) * 5 + sizeof(attallocs[0]),
1856
20.0k
                              10, XML_MAX_ATTRS);
1857
20.0k
    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
20.0k
    atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5);
1864
20.0k
    if (atts == NULL)
1865
4
        goto mem_error;
1866
20.0k
    ctxt->atts = atts;
1867
1868
20.0k
    attallocs = xmlRealloc(ctxt->attallocs,
1869
20.0k
                           newSize * sizeof(attallocs[0]));
1870
20.0k
    if (attallocs == NULL)
1871
6
        goto mem_error;
1872
20.0k
    ctxt->attallocs = attallocs;
1873
1874
20.0k
    ctxt->maxatts = newSize * 5;
1875
1876
20.0k
    return(0);
1877
1878
10
mem_error:
1879
10
    xmlErrMemory(ctxt);
1880
10
    return(-1);
1881
20.0k
}
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
293k
{
1893
293k
    char *directory = NULL;
1894
293k
    int maxDepth;
1895
1896
293k
    if ((ctxt == NULL) || (value == NULL))
1897
3.15k
        return(-1);
1898
1899
289k
    maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
1900
1901
289k
    if (ctxt->inputNr >= ctxt->inputMax) {
1902
6.82k
        xmlParserInputPtr *tmp;
1903
6.82k
        int newSize;
1904
1905
6.82k
        newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]),
1906
6.82k
                                  5, maxDepth);
1907
6.82k
        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
6.82k
        tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0]));
1913
6.82k
        if (tmp == NULL) {
1914
1
            xmlErrMemory(ctxt);
1915
1
            return(-1);
1916
1
        }
1917
6.81k
        ctxt->inputTab = tmp;
1918
6.81k
        ctxt->inputMax = newSize;
1919
6.81k
    }
1920
1921
289k
    if ((ctxt->inputNr == 0) && (value->filename != NULL)) {
1922
146k
        directory = xmlParserGetDirectory(value->filename);
1923
146k
        if (directory == NULL) {
1924
8
            xmlErrMemory(ctxt);
1925
8
            return(-1);
1926
8
        }
1927
146k
    }
1928
1929
289k
    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
289k
    ctxt->inputTab[ctxt->inputNr] = value;
1935
289k
    ctxt->input = value;
1936
1937
289k
    if (ctxt->inputNr == 0) {
1938
179k
        xmlFree(ctxt->directory);
1939
179k
        ctxt->directory = directory;
1940
179k
    }
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
289k
    value->id = ctxt->input_id++;
1948
1949
289k
    return(ctxt->inputNr++);
1950
289k
}
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
368k
{
1961
368k
    xmlParserInputPtr ret;
1962
1963
368k
    if (ctxt == NULL)
1964
0
        return(NULL);
1965
368k
    if (ctxt->inputNr <= 0)
1966
80.2k
        return (NULL);
1967
288k
    ctxt->inputNr--;
1968
288k
    if (ctxt->inputNr > 0)
1969
110k
        ctxt->input = ctxt->inputTab[ctxt->inputNr - 1];
1970
178k
    else
1971
178k
        ctxt->input = NULL;
1972
288k
    ret = ctxt->inputTab[ctxt->inputNr];
1973
288k
    ctxt->inputTab[ctxt->inputNr] = NULL;
1974
288k
    return (ret);
1975
368k
}
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
801k
{
1989
801k
    if (ctxt == NULL)
1990
0
        return(0);
1991
1992
801k
    if (ctxt->nodeNr >= ctxt->nodeMax) {
1993
42.0k
        int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
1994
42.0k
        xmlNodePtr *tmp;
1995
42.0k
        int newSize;
1996
1997
42.0k
        newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]),
1998
42.0k
                                  10, maxDepth);
1999
42.0k
        if (newSize < 0) {
2000
23
            xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
2001
23
                    "Excessive depth in document: %d,"
2002
23
                    " use XML_PARSE_HUGE option\n",
2003
23
                    ctxt->nodeNr);
2004
23
            return(-1);
2005
23
        }
2006
2007
41.9k
  tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0]));
2008
41.9k
        if (tmp == NULL) {
2009
15
            xmlErrMemory(ctxt);
2010
15
            return (-1);
2011
15
        }
2012
41.9k
        ctxt->nodeTab = tmp;
2013
41.9k
  ctxt->nodeMax = newSize;
2014
41.9k
    }
2015
2016
801k
    ctxt->nodeTab[ctxt->nodeNr] = value;
2017
801k
    ctxt->node = value;
2018
801k
    return (ctxt->nodeNr++);
2019
801k
}
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
685k
{
2032
685k
    xmlNodePtr ret;
2033
2034
685k
    if (ctxt == NULL) return(NULL);
2035
685k
    if (ctxt->nodeNr <= 0)
2036
9.33k
        return (NULL);
2037
675k
    ctxt->nodeNr--;
2038
675k
    if (ctxt->nodeNr > 0)
2039
662k
        ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1];
2040
13.3k
    else
2041
13.3k
        ctxt->node = NULL;
2042
675k
    ret = ctxt->nodeTab[ctxt->nodeNr];
2043
675k
    ctxt->nodeTab[ctxt->nodeNr] = NULL;
2044
675k
    return (ret);
2045
685k
}
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
811k
{
2062
811k
    xmlStartTag *tag;
2063
2064
811k
    if (ctxt->nameNr >= ctxt->nameMax) {
2065
38.0k
        const xmlChar **tmp;
2066
38.0k
        xmlStartTag *tmp2;
2067
38.0k
        int newSize;
2068
2069
38.0k
        newSize = xmlGrowCapacity(ctxt->nameMax,
2070
38.0k
                                  sizeof(tmp[0]) + sizeof(tmp2[0]),
2071
38.0k
                                  10, XML_MAX_ITEMS);
2072
38.0k
        if (newSize < 0)
2073
0
            goto mem_error;
2074
2075
38.0k
        tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0]));
2076
38.0k
        if (tmp == NULL)
2077
21
      goto mem_error;
2078
38.0k
  ctxt->nameTab = tmp;
2079
2080
38.0k
        tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0]));
2081
38.0k
        if (tmp2 == NULL)
2082
18
      goto mem_error;
2083
38.0k
  ctxt->pushTab = tmp2;
2084
2085
38.0k
        ctxt->nameMax = newSize;
2086
773k
    } else if (ctxt->pushTab == NULL) {
2087
24.1k
        ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0]));
2088
24.1k
        if (ctxt->pushTab == NULL)
2089
10
            goto mem_error;
2090
24.1k
    }
2091
811k
    ctxt->nameTab[ctxt->nameNr] = value;
2092
811k
    ctxt->name = value;
2093
811k
    tag = &ctxt->pushTab[ctxt->nameNr];
2094
811k
    tag->prefix = prefix;
2095
811k
    tag->URI = URI;
2096
811k
    tag->line = line;
2097
811k
    tag->nsNr = nsNr;
2098
811k
    return (ctxt->nameNr++);
2099
49
mem_error:
2100
49
    xmlErrMemory(ctxt);
2101
49
    return (-1);
2102
811k
}
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
24.4k
{
2113
24.4k
    const xmlChar *ret;
2114
2115
24.4k
    if (ctxt->nameNr <= 0)
2116
0
        return (NULL);
2117
24.4k
    ctxt->nameNr--;
2118
24.4k
    if (ctxt->nameNr > 0)
2119
23.0k
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2120
1.37k
    else
2121
1.37k
        ctxt->name = NULL;
2122
24.4k
    ret = ctxt->nameTab[ctxt->nameNr];
2123
24.4k
    ctxt->nameTab[ctxt->nameNr] = NULL;
2124
24.4k
    return (ret);
2125
24.4k
}
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
661k
{
2139
661k
    const xmlChar *ret;
2140
2141
661k
    if ((ctxt == NULL) || (ctxt->nameNr <= 0))
2142
2
        return (NULL);
2143
661k
    ctxt->nameNr--;
2144
661k
    if (ctxt->nameNr > 0)
2145
654k
        ctxt->name = ctxt->nameTab[ctxt->nameNr - 1];
2146
6.94k
    else
2147
6.94k
        ctxt->name = NULL;
2148
661k
    ret = ctxt->nameTab[ctxt->nameNr];
2149
661k
    ctxt->nameTab[ctxt->nameNr] = NULL;
2150
661k
    return (ret);
2151
661k
}
2152
2153
1.31M
static int spacePush(xmlParserCtxtPtr ctxt, int val) {
2154
1.31M
    if (ctxt->spaceNr >= ctxt->spaceMax) {
2155
71.1k
        int *tmp;
2156
71.1k
        int newSize;
2157
2158
71.1k
        newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]),
2159
71.1k
                                  10, XML_MAX_ITEMS);
2160
71.1k
        if (newSize < 0) {
2161
0
      xmlErrMemory(ctxt);
2162
0
      return(-1);
2163
0
        }
2164
2165
71.1k
        tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0]));
2166
71.1k
        if (tmp == NULL) {
2167
30
      xmlErrMemory(ctxt);
2168
30
      return(-1);
2169
30
  }
2170
71.1k
  ctxt->spaceTab = tmp;
2171
2172
71.1k
        ctxt->spaceMax = newSize;
2173
71.1k
    }
2174
1.31M
    ctxt->spaceTab[ctxt->spaceNr] = val;
2175
1.31M
    ctxt->space = &ctxt->spaceTab[ctxt->spaceNr];
2176
1.31M
    return(ctxt->spaceNr++);
2177
1.31M
}
2178
2179
1.18M
static int spacePop(xmlParserCtxtPtr ctxt) {
2180
1.18M
    int ret;
2181
1.18M
    if (ctxt->spaceNr <= 0) return(0);
2182
1.18M
    ctxt->spaceNr--;
2183
1.18M
    if (ctxt->spaceNr > 0)
2184
1.18M
  ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1];
2185
15
    else
2186
15
        ctxt->space = &ctxt->spaceTab[0];
2187
1.18M
    ret = ctxt->spaceTab[ctxt->spaceNr];
2188
1.18M
    ctxt->spaceTab[ctxt->spaceNr] = -1;
2189
1.18M
    return(ret);
2190
1.18M
}
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
16.8M
#define RAW (*ctxt->input->cur)
2225
382M
#define CUR (*ctxt->input->cur)
2226
14.0M
#define NXT(val) ctxt->input->cur[(val)]
2227
369M
#define CUR_PTR ctxt->input->cur
2228
1.80M
#define BASE_PTR ctxt->input->base
2229
2230
#define CMP4( s, c1, c2, c3, c4 ) \
2231
9.07M
  ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \
2232
4.56M
    ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 )
2233
#define CMP5( s, c1, c2, c3, c4, c5 ) \
2234
8.36M
  ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 )
2235
#define CMP6( s, c1, c2, c3, c4, c5, c6 ) \
2236
7.12M
  ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 )
2237
#define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \
2238
6.17M
  ( 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
5.38M
  ( 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
2.48M
  ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \
2243
2.48M
    ((unsigned char *) s)[ 8 ] == c9 )
2244
#define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \
2245
1.17k
  ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \
2246
1.17k
    ((unsigned char *) s)[ 9 ] == c10 )
2247
2248
3.45M
#define SKIP(val) do {             \
2249
3.45M
    ctxt->input->cur += (val),ctxt->input->col+=(val);      \
2250
3.45M
    if (*ctxt->input->cur == 0)           \
2251
3.45M
        xmlParserGrow(ctxt);           \
2252
3.45M
  } while (0)
2253
2254
#define SKIPL(val) do {             \
2255
    int skipl;                \
2256
    for(skipl=0; skipl<val; skipl++) {          \
2257
  if (*(ctxt->input->cur) == '\n') {        \
2258
  ctxt->input->line++; ctxt->input->col = 1;      \
2259
  } else ctxt->input->col++;          \
2260
  ctxt->input->cur++;           \
2261
    }                 \
2262
    if (*ctxt->input->cur == 0)           \
2263
        xmlParserGrow(ctxt);            \
2264
  } while (0)
2265
2266
#define SHRINK \
2267
5.03M
    if (!PARSER_PROGRESSIVE(ctxt)) \
2268
5.03M
  xmlParserShrink(ctxt);
2269
2270
#define GROW \
2271
17.8M
    if ((!PARSER_PROGRESSIVE(ctxt)) && \
2272
17.8M
        (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \
2273
2.62M
  xmlParserGrow(ctxt);
2274
2275
4.42M
#define SKIP_BLANKS xmlSkipBlankChars(ctxt)
2276
2277
1.17M
#define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt)
2278
2279
223M
#define NEXT xmlNextChar(ctxt)
2280
2281
2.03M
#define NEXT1 {               \
2282
2.03M
  ctxt->input->col++;           \
2283
2.03M
  ctxt->input->cur++;           \
2284
2.03M
  if (*ctxt->input->cur == 0)         \
2285
2.03M
      xmlParserGrow(ctxt);           \
2286
2.03M
    }
2287
2288
205M
#define NEXTL(l) do {             \
2289
205M
    if (*(ctxt->input->cur) == '\n') {         \
2290
8.06M
  ctxt->input->line++; ctxt->input->col = 1;      \
2291
197M
    } else ctxt->input->col++;           \
2292
205M
    ctxt->input->cur += l;        \
2293
205M
  } while (0)
2294
2295
#define COPY_BUF(b, i, v)           \
2296
38.2M
    if (v < 0x80) b[i++] = v;           \
2297
38.2M
    else i += xmlCopyCharMultiByte(&b[i],v)
2298
2299
static int
2300
36.9M
xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) {
2301
36.9M
    int c = xmlCurrentChar(ctxt, len);
2302
2303
36.9M
    if (c == XML_INVALID_CHAR)
2304
7.23M
        c = 0xFFFD; /* replacement character */
2305
2306
36.9M
    return(c);
2307
36.9M
}
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
4.56M
xmlSkipBlankChars(xmlParserCtxt *ctxt) {
2319
4.56M
    const xmlChar *cur;
2320
4.56M
    int res = 0;
2321
2322
4.56M
    cur = ctxt->input->cur;
2323
4.56M
    while (IS_BLANK_CH(*cur)) {
2324
1.91M
        if (*cur == '\n') {
2325
643k
            ctxt->input->line++; ctxt->input->col = 1;
2326
1.27M
        } else {
2327
1.27M
            ctxt->input->col++;
2328
1.27M
        }
2329
1.91M
        cur++;
2330
1.91M
        if (res < INT_MAX)
2331
1.91M
            res++;
2332
1.91M
        if (*cur == 0) {
2333
24.9k
            ctxt->input->cur = cur;
2334
24.9k
            xmlParserGrow(ctxt);
2335
24.9k
            cur = ctxt->input->cur;
2336
24.9k
        }
2337
1.91M
    }
2338
4.56M
    ctxt->input->cur = cur;
2339
2340
4.56M
    if (res > 4)
2341
13.9k
        GROW;
2342
2343
4.56M
    return(res);
2344
4.56M
}
2345
2346
static void
2347
105k
xmlPopPE(xmlParserCtxtPtr ctxt) {
2348
105k
    unsigned long consumed;
2349
105k
    xmlEntityPtr ent;
2350
2351
105k
    ent = ctxt->input->entity;
2352
2353
105k
    ent->flags &= ~XML_ENT_EXPANDING;
2354
2355
105k
    if ((ent->flags & XML_ENT_CHECKED) == 0) {
2356
4.44k
        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
4.65k
        do {
2363
4.65k
            ctxt->input->cur = ctxt->input->end;
2364
4.65k
            xmlParserShrink(ctxt);
2365
4.65k
            result = xmlParserGrow(ctxt);
2366
4.65k
        } while (result > 0);
2367
2368
4.44k
        consumed = ctxt->input->consumed;
2369
4.44k
        xmlSaturatedAddSizeT(&consumed,
2370
4.44k
                             ctxt->input->end - ctxt->input->base);
2371
2372
4.44k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
2373
2374
        /*
2375
         * Add to sizeentities when parsing an external entity
2376
         * for the first time.
2377
         */
2378
4.44k
        if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
2379
3.31k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
2380
3.31k
        }
2381
2382
4.44k
        ent->flags |= XML_ENT_CHECKED;
2383
4.44k
    }
2384
2385
105k
    xmlFreeInputStream(xmlCtxtPopInput(ctxt));
2386
2387
105k
    xmlParserEntityCheck(ctxt, ent->expandedSize);
2388
2389
105k
    GROW;
2390
105k
}
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
1.17M
xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) {
2401
1.17M
    int res = 0;
2402
1.17M
    int inParam;
2403
1.17M
    int expandParam;
2404
2405
1.17M
    inParam = PARSER_IN_PE(ctxt);
2406
1.17M
    expandParam = PARSER_EXTERNAL(ctxt);
2407
2408
1.17M
    if (!inParam && !expandParam)
2409
144k
        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
2.86M
    while (PARSER_STOPPED(ctxt) == 0) {
2416
2.86M
        if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */
2417
1.81M
            NEXT;
2418
1.81M
        } else if (CUR == '%') {
2419
78.6k
            if ((expandParam == 0) ||
2420
78.5k
                (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0))
2421
65.8k
                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
12.8k
            xmlParsePERefInternal(ctxt, 0);
2430
2431
12.8k
            inParam = PARSER_IN_PE(ctxt);
2432
12.8k
            expandParam = PARSER_EXTERNAL(ctxt);
2433
966k
        } else if (CUR == 0) {
2434
60.4k
            if (inParam == 0)
2435
108
                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
60.3k
            if (ctxt->input->flags & XML_INPUT_MARKUP_DECL)
2443
53.9k
                break;
2444
2445
6.31k
            xmlPopPE(ctxt);
2446
2447
6.31k
            inParam = PARSER_IN_PE(ctxt);
2448
6.31k
            expandParam = PARSER_EXTERNAL(ctxt);
2449
905k
        } else {
2450
905k
            break;
2451
905k
        }
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
1.83M
        if (res < INT_MAX)
2461
1.83M
            res++;
2462
1.83M
    }
2463
2464
1.02M
    return(res);
2465
1.17M
}
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
46.4k
xmlParseCharRef(xmlParserCtxt *ctxt) {
2530
46.4k
    int val = 0;
2531
46.4k
    int count = 0;
2532
2533
    /*
2534
     * Using RAW/CUR/NEXT is okay since we are working on ASCII range here
2535
     */
2536
46.4k
    if ((RAW == '&') && (NXT(1) == '#') &&
2537
46.4k
        (NXT(2) == 'x')) {
2538
13.1k
  SKIP(3);
2539
13.1k
  GROW;
2540
50.3k
  while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) {
2541
38.7k
      if (count++ > 20) {
2542
738
    count = 0;
2543
738
    GROW;
2544
738
      }
2545
38.7k
      if ((RAW >= '0') && (RAW <= '9'))
2546
24.4k
          val = val * 16 + (CUR - '0');
2547
14.2k
      else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20))
2548
9.92k
          val = val * 16 + (CUR - 'a') + 10;
2549
4.33k
      else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20))
2550
2.89k
          val = val * 16 + (CUR - 'A') + 10;
2551
1.44k
      else {
2552
1.44k
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2553
1.44k
    val = 0;
2554
1.44k
    break;
2555
1.44k
      }
2556
37.2k
      if (val > 0x110000)
2557
8.45k
          val = 0x110000;
2558
2559
37.2k
      NEXT;
2560
37.2k
      count++;
2561
37.2k
  }
2562
13.1k
  if (RAW == ';') {
2563
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2564
11.6k
      ctxt->input->col++;
2565
11.6k
      ctxt->input->cur++;
2566
11.6k
  }
2567
33.3k
    } else if  ((RAW == '&') && (NXT(1) == '#')) {
2568
33.3k
  SKIP(2);
2569
33.3k
  GROW;
2570
127k
  while (RAW != ';') { /* loop blocked by count */
2571
95.9k
      if (count++ > 20) {
2572
2.27k
    count = 0;
2573
2.27k
    GROW;
2574
2.27k
      }
2575
95.9k
      if ((RAW >= '0') && (RAW <= '9'))
2576
93.7k
          val = val * 10 + (CUR - '0');
2577
2.22k
      else {
2578
2.22k
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2579
2.22k
    val = 0;
2580
2.22k
    break;
2581
2.22k
      }
2582
93.7k
      if (val > 0x110000)
2583
20.9k
          val = 0x110000;
2584
2585
93.7k
      NEXT;
2586
93.7k
      count++;
2587
93.7k
  }
2588
33.3k
  if (RAW == ';') {
2589
      /* on purpose to avoid reentrancy problems with NEXT and SKIP */
2590
31.0k
      ctxt->input->col++;
2591
31.0k
      ctxt->input->cur++;
2592
31.0k
  }
2593
33.3k
    } 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
46.4k
    if (val >= 0x110000) {
2605
619
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2606
619
                "xmlParseCharRef: character reference out of bounds\n",
2607
619
          val);
2608
619
        val = 0xFFFD;
2609
45.8k
    } else if (!IS_CHAR(val)) {
2610
5.79k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2611
5.79k
                          "xmlParseCharRef: invalid xmlChar value %d\n",
2612
5.79k
                    val);
2613
5.79k
    }
2614
46.4k
    return(val);
2615
46.4k
}
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
1.19M
xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) {
2635
1.19M
    const xmlChar *ptr;
2636
1.19M
    xmlChar cur;
2637
1.19M
    int val = 0;
2638
2639
1.19M
    if ((str == NULL) || (*str == NULL)) return(0);
2640
1.19M
    ptr = *str;
2641
1.19M
    cur = *ptr;
2642
1.19M
    if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) {
2643
4.12k
  ptr += 3;
2644
4.12k
  cur = *ptr;
2645
13.4k
  while (cur != ';') { /* Non input consuming loop */
2646
9.74k
      if ((cur >= '0') && (cur <= '9'))
2647
4.82k
          val = val * 16 + (cur - '0');
2648
4.92k
      else if ((cur >= 'a') && (cur <= 'f'))
2649
1.42k
          val = val * 16 + (cur - 'a') + 10;
2650
3.49k
      else if ((cur >= 'A') && (cur <= 'F'))
2651
3.10k
          val = val * 16 + (cur - 'A') + 10;
2652
389
      else {
2653
389
    xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL);
2654
389
    val = 0;
2655
389
    break;
2656
389
      }
2657
9.35k
      if (val > 0x110000)
2658
419
          val = 0x110000;
2659
2660
9.35k
      ptr++;
2661
9.35k
      cur = *ptr;
2662
9.35k
  }
2663
4.12k
  if (cur == ';')
2664
3.73k
      ptr++;
2665
1.19M
    } else if  ((cur == '&') && (ptr[1] == '#')){
2666
1.19M
  ptr += 2;
2667
1.19M
  cur = *ptr;
2668
3.59M
  while (cur != ';') { /* Non input consuming loops */
2669
2.39M
      if ((cur >= '0') && (cur <= '9'))
2670
2.39M
          val = val * 10 + (cur - '0');
2671
658
      else {
2672
658
    xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL);
2673
658
    val = 0;
2674
658
    break;
2675
658
      }
2676
2.39M
      if (val > 0x110000)
2677
1.63k
          val = 0x110000;
2678
2679
2.39M
      ptr++;
2680
2.39M
      cur = *ptr;
2681
2.39M
  }
2682
1.19M
  if (cur == ';')
2683
1.19M
      ptr++;
2684
1.19M
    } else {
2685
0
  xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL);
2686
0
  return(0);
2687
0
    }
2688
1.19M
    *str = ptr;
2689
2690
    /*
2691
     * [ WFC: Legal Character ]
2692
     * Characters referred to using character references must match the
2693
     * production for Char.
2694
     */
2695
1.19M
    if (val >= 0x110000) {
2696
215
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2697
215
                "xmlParseStringCharRef: character reference out of bounds\n",
2698
215
                val);
2699
1.19M
    } else if (IS_CHAR(val)) {
2700
1.19M
        return(val);
2701
1.19M
    } else {
2702
2.15k
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
2703
2.15k
        "xmlParseStringCharRef: invalid xmlChar value %d\n",
2704
2.15k
        val);
2705
2.15k
    }
2706
2.37k
    return(0);
2707
1.19M
}
2708
2709
/**
2710
 *     [69] PEReference ::= '%' Name ';'
2711
 *
2712
 * @deprecated Internal function, do not use.
2713
 *
2714
 * [ WFC: No Recursion ]
2715
 * A parsed entity must not contain a recursive
2716
 * reference to itself, either directly or indirectly.
2717
 *
2718
 * [ WFC: Entity Declared ]
2719
 * In a document without any DTD, a document with only an internal DTD
2720
 * subset which contains no parameter entity references, or a document
2721
 * with "standalone='yes'", ...  ... The declaration of a parameter
2722
 * entity must precede any reference to it...
2723
 *
2724
 * [ VC: Entity Declared ]
2725
 * In a document with an external subset or external parameter entities
2726
 * with "standalone='no'", ...  ... The declaration of a parameter entity
2727
 * must precede any reference to it...
2728
 *
2729
 * [ WFC: In DTD ]
2730
 * Parameter-entity references may only appear in the DTD.
2731
 * NOTE: misleading but this is handled.
2732
 *
2733
 * A PEReference may have been detected in the current input stream
2734
 * the handling is done accordingly to
2735
 *      http://www.w3.org/TR/REC-xml#entproc
2736
 * i.e.
2737
 *   - Included in literal in entity values
2738
 *   - Included as Parameter Entity reference within DTDs
2739
 * @param ctxt  the parser context
2740
 */
2741
void
2742
0
xmlParserHandlePEReference(xmlParserCtxt *ctxt) {
2743
0
    xmlParsePERefInternal(ctxt, 0);
2744
0
}
2745
2746
/**
2747
 * @deprecated Internal function, don't use.
2748
 *
2749
 * @param ctxt  the parser context
2750
 * @param str  the input string
2751
 * @param len  the string length
2752
 * @param what  combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF
2753
 * @param end  an end marker xmlChar, 0 if none
2754
 * @param end2  an end marker xmlChar, 0 if none
2755
 * @param end3  an end marker xmlChar, 0 if none
2756
 * @returns A newly allocated string with the substitution done. The caller
2757
 *      must deallocate it !
2758
 */
2759
xmlChar *
2760
xmlStringLenDecodeEntities(xmlParserCtxt *ctxt, const xmlChar *str, int len,
2761
                           int what ATTRIBUTE_UNUSED,
2762
0
                           xmlChar end, xmlChar end2, xmlChar end3) {
2763
0
    if ((ctxt == NULL) || (str == NULL) || (len < 0))
2764
0
        return(NULL);
2765
2766
0
    if ((str[len] != 0) ||
2767
0
        (end != 0) || (end2 != 0) || (end3 != 0))
2768
0
        return(NULL);
2769
2770
0
    return(xmlExpandEntitiesInAttValue(ctxt, str, 0));
2771
0
}
2772
2773
/**
2774
 * @deprecated Internal function, don't use.
2775
 *
2776
 * @param ctxt  the parser context
2777
 * @param str  the input string
2778
 * @param what  combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF
2779
 * @param end  an end marker xmlChar, 0 if none
2780
 * @param end2  an end marker xmlChar, 0 if none
2781
 * @param end3  an end marker xmlChar, 0 if none
2782
 * @returns A newly allocated string with the substitution done. The caller
2783
 *      must deallocate it !
2784
 */
2785
xmlChar *
2786
xmlStringDecodeEntities(xmlParserCtxt *ctxt, const xmlChar *str,
2787
                        int what ATTRIBUTE_UNUSED,
2788
0
            xmlChar end, xmlChar  end2, xmlChar end3) {
2789
0
    if ((ctxt == NULL) || (str == NULL))
2790
0
        return(NULL);
2791
2792
0
    if ((end != 0) || (end2 != 0) || (end3 != 0))
2793
0
        return(NULL);
2794
2795
0
    return(xmlExpandEntitiesInAttValue(ctxt, str, 0));
2796
0
}
2797
2798
/************************************************************************
2799
 *                  *
2800
 *    Commodity functions, cleanup needed ?     *
2801
 *                  *
2802
 ************************************************************************/
2803
2804
/**
2805
 * Is this a sequence of blank chars that one can ignore ?
2806
 *
2807
 * @param ctxt  an XML parser context
2808
 * @param str  a xmlChar *
2809
 * @param len  the size of `str`
2810
 * @param blank_chars  we know the chars are blanks
2811
 * @returns 1 if ignorable 0 otherwise.
2812
 */
2813
2814
static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len,
2815
479k
                     int blank_chars) {
2816
479k
    int i;
2817
479k
    xmlNodePtr lastChild;
2818
2819
    /*
2820
     * Check for xml:space value.
2821
     */
2822
479k
    if ((ctxt->space == NULL) || (*(ctxt->space) == 1) ||
2823
479k
        (*(ctxt->space) == -2))
2824
335k
  return(0);
2825
2826
    /*
2827
     * Check that the string is made of blanks
2828
     */
2829
144k
    if (blank_chars == 0) {
2830
209k
  for (i = 0;i < len;i++)
2831
204k
      if (!(IS_BLANK_CH(str[i]))) return(0);
2832
130k
    }
2833
2834
    /*
2835
     * Look if the element is mixed content in the DTD if available
2836
     */
2837
18.2k
    if (ctxt->node == NULL) return(0);
2838
18.2k
    if (ctxt->myDoc != NULL) {
2839
18.2k
        xmlElementPtr elemDecl = NULL;
2840
18.2k
        xmlDocPtr doc = ctxt->myDoc;
2841
18.2k
        const xmlChar *prefix = NULL;
2842
2843
18.2k
        if (ctxt->node->ns)
2844
2.11k
            prefix = ctxt->node->ns->prefix;
2845
18.2k
        if (doc->intSubset != NULL)
2846
9.27k
            elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name,
2847
9.27k
                                      prefix);
2848
18.2k
        if ((elemDecl == NULL) && (doc->extSubset != NULL))
2849
361
            elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name,
2850
361
                                      prefix);
2851
18.2k
        if (elemDecl != NULL) {
2852
2.43k
            if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT)
2853
1.38k
                return(1);
2854
1.04k
            if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) ||
2855
771
                (elemDecl->etype == XML_ELEMENT_TYPE_MIXED))
2856
589
                return(0);
2857
1.04k
        }
2858
18.2k
    }
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
16.2k
    if ((RAW != '<') && (RAW != 0xD)) return(0);
2868
15.1k
    if ((ctxt->node->children == NULL) &&
2869
11.9k
  (RAW == '<') && (NXT(1) == '/')) return(0);
2870
2871
14.4k
    lastChild = xmlGetLastChild(ctxt->node);
2872
14.4k
    if (lastChild == NULL) {
2873
11.2k
        if ((ctxt->node->type != XML_ELEMENT_NODE) &&
2874
0
            (ctxt->node->content != NULL)) return(0);
2875
11.2k
    } else if (xmlNodeIsText(lastChild))
2876
268
        return(0);
2877
2.89k
    else if ((ctxt->node->children != NULL) &&
2878
2.89k
             (xmlNodeIsText(ctxt->node->children)))
2879
450
        return(0);
2880
13.7k
    return(1);
2881
14.4k
}
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
2.74M
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
2.74M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2953
2.73M
        (((c >= 'a') && (c <= 'z')) ||
2954
223k
         ((c >= 'A') && (c <= 'Z')) ||
2955
167k
         (c == '_') || (c == ':') ||
2956
157k
         ((c >= 0xC0) && (c <= 0xD6)) ||
2957
157k
         ((c >= 0xD8) && (c <= 0xF6)) ||
2958
155k
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2959
155k
         ((c >= 0x370) && (c <= 0x37D)) ||
2960
147k
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2961
146k
         ((c >= 0x200C) && (c <= 0x200D)) ||
2962
146k
         ((c >= 0x2070) && (c <= 0x218F)) ||
2963
145k
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2964
145k
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2965
143k
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2966
143k
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2967
132k
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2968
2.60M
        return(1);
2969
145k
    return(0);
2970
2.74M
}
2971
2972
static int
2973
19.8M
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
19.8M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
2979
19.7M
        (((c >= 'a') && (c <= 'z')) ||
2980
17.6M
         ((c >= 'A') && (c <= 'Z')) ||
2981
14.4M
         ((c >= '0') && (c <= '9')) || /* !start */
2982
13.9M
         (c == '_') || (c == ':') ||
2983
13.9M
         (c == '-') || (c == '.') || (c == 0xB7) || /* !start */
2984
13.8M
         ((c >= 0xC0) && (c <= 0xD6)) ||
2985
9.64M
         ((c >= 0xD8) && (c <= 0xF6)) ||
2986
9.64M
         ((c >= 0xF8) && (c <= 0x2FF)) ||
2987
8.32M
         ((c >= 0x300) && (c <= 0x36F)) || /* !start */
2988
8.32M
         ((c >= 0x370) && (c <= 0x37D)) ||
2989
8.31M
         ((c >= 0x37F) && (c <= 0x1FFF)) ||
2990
7.62M
         ((c >= 0x200C) && (c <= 0x200D)) ||
2991
7.62M
         ((c >= 0x203F) && (c <= 0x2040)) || /* !start */
2992
7.62M
         ((c >= 0x2070) && (c <= 0x218F)) ||
2993
7.62M
         ((c >= 0x2C00) && (c <= 0x2FEF)) ||
2994
7.62M
         ((c >= 0x3001) && (c <= 0xD7FF)) ||
2995
7.48M
         ((c >= 0xF900) && (c <= 0xFDCF)) ||
2996
7.48M
         ((c >= 0xFDF0) && (c <= 0xFFFD)) ||
2997
2.52M
         ((c >= 0x10000) && (c <= 0xEFFFF))))
2998
17.2M
         return(1);
2999
2.60M
    return(0);
3000
19.8M
}
3001
3002
static int
3003
1.10M
xmlIsNameStartCharOld(int c) {
3004
1.10M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3005
1.10M
        ((IS_LETTER(c) || (c == '_') || (c == ':'))))
3006
703k
        return(1);
3007
403k
    return(0);
3008
1.10M
}
3009
3010
static int
3011
2.68M
xmlIsNameCharOld(int c) {
3012
2.68M
    if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */
3013
2.68M
        ((IS_LETTER(c)) || (IS_DIGIT(c)) ||
3014
726k
         (c == '.') || (c == '-') ||
3015
720k
         (c == '_') || (c == ':') ||
3016
708k
         (IS_COMBINING(c)) ||
3017
705k
         (IS_EXTENDER(c))))
3018
1.97M
        return(1);
3019
709k
    return(0);
3020
2.68M
}
3021
3022
static int
3023
3.85M
xmlIsNameStartChar(int c, int old10) {
3024
3.85M
    if (!old10)
3025
2.74M
        return(xmlIsNameStartCharNew(c));
3026
1.10M
    else
3027
1.10M
        return(xmlIsNameStartCharOld(c));
3028
3.85M
}
3029
3030
static int
3031
22.5M
xmlIsNameChar(int c, int old10) {
3032
22.5M
    if (!old10)
3033
19.8M
        return(xmlIsNameCharNew(c));
3034
2.68M
    else
3035
2.68M
        return(xmlIsNameCharOld(c));
3036
22.5M
}
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
3.26M
xmlScanName(const xmlChar *ptr, size_t maxSize, int flags) {
3052
3.26M
    int stop = flags & XML_SCAN_NC ? ':' : 0;
3053
3.26M
    int old10 = flags & XML_SCAN_OLD10 ? 1 : 0;
3054
3055
14.5M
    while (1) {
3056
14.5M
        int c, len;
3057
3058
14.5M
        c = *ptr;
3059
14.5M
        if (c < 0x80) {
3060
10.9M
            if (c == stop)
3061
64.5k
                break;
3062
10.8M
            len = 1;
3063
10.8M
        } else {
3064
3.65M
            len = 4;
3065
3.65M
            c = xmlGetUTF8Char(ptr, &len);
3066
3.65M
            if (c < 0)
3067
6.55k
                break;
3068
3.65M
        }
3069
3070
14.5M
        if (flags & XML_SCAN_NMTOKEN ?
3071
11.3M
                !xmlIsNameChar(c, old10) :
3072
14.5M
                !xmlIsNameStartChar(c, old10))
3073
3.19M
            break;
3074
3075
11.3M
        if ((size_t) len > maxSize)
3076
68
            return(NULL);
3077
11.3M
        ptr += len;
3078
11.3M
        maxSize -= len;
3079
11.3M
        flags |= XML_SCAN_NMTOKEN;
3080
11.3M
    }
3081
3082
3.26M
    return(ptr);
3083
3.26M
}
3084
3085
static const xmlChar *
3086
232k
xmlParseNameComplex(xmlParserCtxtPtr ctxt) {
3087
232k
    const xmlChar *ret;
3088
232k
    int len = 0, l;
3089
232k
    int c;
3090
232k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3091
91.1k
                    XML_MAX_TEXT_LENGTH :
3092
232k
                    XML_MAX_NAME_LENGTH;
3093
232k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3094
3095
    /*
3096
     * Handler for more complex cases
3097
     */
3098
232k
    c = xmlCurrentChar(ctxt, &l);
3099
232k
    if (!xmlIsNameStartChar(c, old10))
3100
146k
        return(NULL);
3101
85.5k
    len += l;
3102
85.5k
    NEXTL(l);
3103
85.5k
    c = xmlCurrentChar(ctxt, &l);
3104
5.32M
    while (xmlIsNameChar(c, old10)) {
3105
5.24M
        if (len <= INT_MAX - l)
3106
5.24M
            len += l;
3107
5.24M
        NEXTL(l);
3108
5.24M
        c = xmlCurrentChar(ctxt, &l);
3109
5.24M
    }
3110
85.5k
    if (len > maxLength) {
3111
55
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3112
55
        return(NULL);
3113
55
    }
3114
85.4k
    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
85.4k
    if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r'))
3125
424
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len);
3126
85.0k
    else
3127
85.0k
        ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len);
3128
85.4k
    if (ret == NULL)
3129
1
        xmlErrMemory(ctxt);
3130
85.4k
    return(ret);
3131
85.4k
}
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
2.87M
xmlParseName(xmlParserCtxt *ctxt) {
3151
2.87M
    const xmlChar *in;
3152
2.87M
    const xmlChar *ret;
3153
2.87M
    size_t count = 0;
3154
2.87M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3155
1.11M
                       XML_MAX_TEXT_LENGTH :
3156
2.87M
                       XML_MAX_NAME_LENGTH;
3157
3158
2.87M
    GROW;
3159
3160
    /*
3161
     * Accelerator for simple ASCII names
3162
     */
3163
2.87M
    in = ctxt->input->cur;
3164
2.87M
    if (((*in >= 0x61) && (*in <= 0x7A)) ||
3165
289k
  ((*in >= 0x41) && (*in <= 0x5A)) ||
3166
2.70M
  (*in == '_') || (*in == ':')) {
3167
2.70M
  in++;
3168
14.5M
  while (((*in >= 0x61) && (*in <= 0x7A)) ||
3169
10.6M
         ((*in >= 0x41) && (*in <= 0x5A)) ||
3170
4.11M
         ((*in >= 0x30) && (*in <= 0x39)) ||
3171
2.99M
         (*in == '_') || (*in == '-') ||
3172
2.86M
         (*in == ':') || (*in == '.'))
3173
11.8M
      in++;
3174
2.70M
  if ((*in > 0) && (*in < 0x80)) {
3175
2.63M
      count = in - ctxt->input->cur;
3176
2.63M
            if (count > maxLength) {
3177
77
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name");
3178
77
                return(NULL);
3179
77
            }
3180
2.63M
      ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count);
3181
2.63M
      ctxt->input->cur = in;
3182
2.63M
      ctxt->input->col += count;
3183
2.63M
      if (ret == NULL)
3184
5
          xmlErrMemory(ctxt);
3185
2.63M
      return(ret);
3186
2.63M
  }
3187
2.70M
    }
3188
    /* accelerator for special cases */
3189
232k
    return(xmlParseNameComplex(ctxt));
3190
2.87M
}
3191
3192
static xmlHashedString
3193
420k
xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) {
3194
420k
    xmlHashedString ret;
3195
420k
    int len = 0, l;
3196
420k
    int c;
3197
420k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3198
388k
                    XML_MAX_TEXT_LENGTH :
3199
420k
                    XML_MAX_NAME_LENGTH;
3200
420k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3201
420k
    size_t startPosition = 0;
3202
3203
420k
    ret.name = NULL;
3204
420k
    ret.hashValue = 0;
3205
3206
    /*
3207
     * Handler for more complex cases
3208
     */
3209
420k
    startPosition = CUR_PTR - BASE_PTR;
3210
420k
    c = xmlCurrentChar(ctxt, &l);
3211
420k
    if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */
3212
411k
  (!xmlIsNameStartChar(c, old10) || (c == ':'))) {
3213
402k
  return(ret);
3214
402k
    }
3215
3216
3.51M
    while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */
3217
3.51M
     (xmlIsNameChar(c, old10) && (c != ':'))) {
3218
3.49M
        if (len <= INT_MAX - l)
3219
3.49M
      len += l;
3220
3.49M
  NEXTL(l);
3221
3.49M
  c = xmlCurrentChar(ctxt, &l);
3222
3.49M
    }
3223
17.8k
    if (len > maxLength) {
3224
24
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3225
24
        return(ret);
3226
24
    }
3227
17.8k
    ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len);
3228
17.8k
    if (ret.name == NULL)
3229
2
        xmlErrMemory(ctxt);
3230
17.8k
    return(ret);
3231
17.8k
}
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
1.38M
xmlParseNCName(xmlParserCtxtPtr ctxt) {
3247
1.38M
    const xmlChar *in, *e;
3248
1.38M
    xmlHashedString ret;
3249
1.38M
    size_t count = 0;
3250
1.38M
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3251
982k
                       XML_MAX_TEXT_LENGTH :
3252
1.38M
                       XML_MAX_NAME_LENGTH;
3253
3254
1.38M
    ret.name = NULL;
3255
3256
    /*
3257
     * Accelerator for simple ASCII names
3258
     */
3259
1.38M
    in = ctxt->input->cur;
3260
1.38M
    e = ctxt->input->end;
3261
1.38M
    if ((((*in >= 0x61) && (*in <= 0x7A)) ||
3262
418k
   ((*in >= 0x41) && (*in <= 0x5A)) ||
3263
976k
   (*in == '_')) && (in < e)) {
3264
976k
  in++;
3265
2.83M
  while ((((*in >= 0x61) && (*in <= 0x7A)) ||
3266
1.50M
          ((*in >= 0x41) && (*in <= 0x5A)) ||
3267
1.01M
          ((*in >= 0x30) && (*in <= 0x39)) ||
3268
989k
          (*in == '_') || (*in == '-') ||
3269
1.85M
          (*in == '.')) && (in < e))
3270
1.85M
      in++;
3271
976k
  if (in >= e)
3272
465
      goto complex;
3273
975k
  if ((*in > 0) && (*in < 0x80)) {
3274
960k
      count = in - ctxt->input->cur;
3275
960k
            if (count > maxLength) {
3276
8
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3277
8
                return(ret);
3278
8
            }
3279
960k
      ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count);
3280
960k
      ctxt->input->cur = in;
3281
960k
      ctxt->input->col += count;
3282
960k
      if (ret.name == NULL) {
3283
1
          xmlErrMemory(ctxt);
3284
1
      }
3285
960k
      return(ret);
3286
960k
  }
3287
975k
    }
3288
420k
complex:
3289
420k
    return(xmlParseNCNameComplex(ctxt));
3290
1.38M
}
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
336k
xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) {
3304
336k
    register const xmlChar *cmp = other;
3305
336k
    register const xmlChar *in;
3306
336k
    const xmlChar *ret;
3307
3308
336k
    GROW;
3309
3310
336k
    in = ctxt->input->cur;
3311
738k
    while (*in != 0 && *in == *cmp) {
3312
402k
  ++in;
3313
402k
  ++cmp;
3314
402k
    }
3315
336k
    if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
3316
  /* success */
3317
309k
  ctxt->input->col += in - ctxt->input->cur;
3318
309k
  ctxt->input->cur = in;
3319
309k
  return (const xmlChar*) 1;
3320
309k
    }
3321
    /* failure (or end of input buffer), check with full function */
3322
27.0k
    ret = xmlParseName (ctxt);
3323
    /* strings coming from the dictionary direct compare possible */
3324
27.0k
    if (ret == other) {
3325
8.17k
  return (const xmlChar*) 1;
3326
8.17k
    }
3327
18.8k
    return ret;
3328
27.0k
}
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
3.01M
xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) {
3341
3.01M
    xmlChar *ret;
3342
3.01M
    const xmlChar *cur = *str;
3343
3.01M
    int flags = 0;
3344
3.01M
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3345
1.44M
                    XML_MAX_TEXT_LENGTH :
3346
3.01M
                    XML_MAX_NAME_LENGTH;
3347
3348
3.01M
    if (ctxt->options & XML_PARSE_OLD10)
3349
670k
        flags |= XML_SCAN_OLD10;
3350
3351
3.01M
    cur = xmlScanName(*str, maxLength, flags);
3352
3.01M
    if (cur == NULL) {
3353
59
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName");
3354
59
        return(NULL);
3355
59
    }
3356
3.01M
    if (cur == *str)
3357
4.03k
        return(NULL);
3358
3359
3.01M
    ret = xmlStrndup(*str, cur - *str);
3360
3.01M
    if (ret == NULL)
3361
22
        xmlErrMemory(ctxt);
3362
3.01M
    *str = cur;
3363
3.01M
    return(ret);
3364
3.01M
}
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
31.1k
xmlParseNmtoken(xmlParserCtxt *ctxt) {
3381
31.1k
    xmlChar buf[XML_MAX_NAMELEN + 5];
3382
31.1k
    xmlChar *ret;
3383
31.1k
    int len = 0, l;
3384
31.1k
    int c;
3385
31.1k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3386
15.5k
                    XML_MAX_TEXT_LENGTH :
3387
31.1k
                    XML_MAX_NAME_LENGTH;
3388
31.1k
    int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0;
3389
3390
31.1k
    c = xmlCurrentChar(ctxt, &l);
3391
3392
160k
    while (xmlIsNameChar(c, old10)) {
3393
129k
  COPY_BUF(buf, len, c);
3394
129k
  NEXTL(l);
3395
129k
  c = xmlCurrentChar(ctxt, &l);
3396
129k
  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
924
      xmlChar *buffer;
3402
924
      int max = len * 2;
3403
3404
924
      buffer = xmlMalloc(max);
3405
924
      if (buffer == NULL) {
3406
1
          xmlErrMemory(ctxt);
3407
1
    return(NULL);
3408
1
      }
3409
923
      memcpy(buffer, buf, len);
3410
2.25M
      while (xmlIsNameChar(c, old10)) {
3411
2.24M
    if (len + 10 > max) {
3412
1.88k
        xmlChar *tmp;
3413
1.88k
                    int newSize;
3414
3415
1.88k
                    newSize = xmlGrowCapacity(max, 1, 1, maxLength);
3416
1.88k
                    if (newSize < 0) {
3417
49
                        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3418
49
                        xmlFree(buffer);
3419
49
                        return(NULL);
3420
49
                    }
3421
1.84k
        tmp = xmlRealloc(buffer, newSize);
3422
1.84k
        if (tmp == NULL) {
3423
1
      xmlErrMemory(ctxt);
3424
1
      xmlFree(buffer);
3425
1
      return(NULL);
3426
1
        }
3427
1.83k
        buffer = tmp;
3428
1.83k
                    max = newSize;
3429
1.83k
    }
3430
2.24M
    COPY_BUF(buffer, len, c);
3431
2.24M
    NEXTL(l);
3432
2.24M
    c = xmlCurrentChar(ctxt, &l);
3433
2.24M
      }
3434
873
      buffer[len] = 0;
3435
873
      return(buffer);
3436
923
  }
3437
129k
    }
3438
30.2k
    if (len == 0)
3439
3.91k
        return(NULL);
3440
26.3k
    if (len > maxLength) {
3441
0
        xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken");
3442
0
        return(NULL);
3443
0
    }
3444
26.3k
    ret = xmlStrndup(buf, len);
3445
26.3k
    if (ret == NULL)
3446
2
        xmlErrMemory(ctxt);
3447
26.3k
    return(ret);
3448
26.3k
}
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
256k
                          const xmlChar *str, int length, int depth) {
3462
256k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3463
256k
    const xmlChar *end, *chunk;
3464
256k
    int c, l;
3465
3466
256k
    if (str == NULL)
3467
139k
        return;
3468
3469
117k
    depth += 1;
3470
117k
    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
117k
    end = str + length;
3477
117k
    chunk = str;
3478
3479
86.3M
    while ((str < end) && (!PARSER_STOPPED(ctxt))) {
3480
86.2M
        c = *str;
3481
3482
86.2M
        if (c >= 0x80) {
3483
30.8M
            l = xmlUTF8MultibyteLen(ctxt, str,
3484
30.8M
                    "invalid character in entity value\n");
3485
30.8M
            if (l == 0) {
3486
11.3M
                if (chunk < str)
3487
70.2k
                    xmlSBufAddString(buf, chunk, str - chunk);
3488
11.3M
                xmlSBufAddReplChar(buf);
3489
11.3M
                str += 1;
3490
11.3M
                chunk = str;
3491
19.4M
            } else {
3492
19.4M
                str += l;
3493
19.4M
            }
3494
55.4M
        } else if (c == '&') {
3495
78.8k
            if (str[1] == '#') {
3496
12.4k
                if (chunk < str)
3497
6.72k
                    xmlSBufAddString(buf, chunk, str - chunk);
3498
3499
12.4k
                c = xmlParseStringCharRef(ctxt, &str);
3500
12.4k
                if (c == 0)
3501
2.36k
                    return;
3502
3503
10.1k
                xmlSBufAddChar(buf, c);
3504
3505
10.1k
                chunk = str;
3506
66.3k
            } else {
3507
66.3k
                xmlChar *name;
3508
3509
                /*
3510
                 * General entity references are checked for
3511
                 * syntactic validity.
3512
                 */
3513
66.3k
                str++;
3514
66.3k
                name = xmlParseStringName(ctxt, &str);
3515
3516
66.3k
                if ((name == NULL) || (*str++ != ';')) {
3517
3.90k
                    xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR,
3518
3.90k
                            "EntityValue: '&' forbidden except for entities "
3519
3.90k
                            "references\n");
3520
3.90k
                    xmlFree(name);
3521
3.90k
                    return;
3522
3.90k
                }
3523
3524
62.4k
                xmlFree(name);
3525
62.4k
            }
3526
55.3M
        } else if (c == '%') {
3527
198k
            xmlEntityPtr ent;
3528
3529
198k
            if (chunk < str)
3530
93.5k
                xmlSBufAddString(buf, chunk, str - chunk);
3531
3532
198k
            ent = xmlParseStringPEReference(ctxt, &str);
3533
198k
            if (ent == NULL)
3534
50.4k
                return;
3535
3536
148k
            if (!PARSER_EXTERNAL(ctxt)) {
3537
67
                xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL);
3538
67
                return;
3539
67
            }
3540
3541
148k
            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
139k
                if (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
3549
139k
                    ((ctxt->replaceEntities) ||
3550
139k
                     (ctxt->validate))) {
3551
139k
                    xmlLoadEntityContent(ctxt, ent);
3552
139k
                } else {
3553
200
                    xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
3554
200
                                  "not validating will not read content for "
3555
200
                                  "PE entity %s\n", ent->name, NULL);
3556
200
                }
3557
139k
            }
3558
3559
            /*
3560
             * TODO: Skip if ent->content is still NULL.
3561
             */
3562
3563
148k
            if (xmlParserEntityCheck(ctxt, ent->length))
3564
20
                return;
3565
3566
148k
            if (ent->flags & XML_ENT_EXPANDING) {
3567
62
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3568
62
                return;
3569
62
            }
3570
3571
148k
            ent->flags |= XML_ENT_EXPANDING;
3572
148k
            xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length,
3573
148k
                                      depth);
3574
148k
            ent->flags &= ~XML_ENT_EXPANDING;
3575
3576
148k
            chunk = str;
3577
55.1M
        } else {
3578
            /* Normal ASCII char */
3579
55.1M
            if (!IS_BYTE_CHAR(c)) {
3580
418k
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3581
418k
                        "invalid character in entity value\n");
3582
418k
                if (chunk < str)
3583
4.29k
                    xmlSBufAddString(buf, chunk, str - chunk);
3584
418k
                xmlSBufAddReplChar(buf);
3585
418k
                str += 1;
3586
418k
                chunk = str;
3587
54.7M
            } else {
3588
54.7M
                str += 1;
3589
54.7M
            }
3590
55.1M
        }
3591
86.2M
    }
3592
3593
60.2k
    if (chunk < str)
3594
55.7k
        xmlSBufAddString(buf, chunk, str - chunk);
3595
60.2k
}
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
111k
xmlParseEntityValue(xmlParserCtxt *ctxt, xmlChar **orig) {
3611
111k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3612
37.8k
                         XML_MAX_HUGE_LENGTH :
3613
111k
                         XML_MAX_TEXT_LENGTH;
3614
111k
    xmlSBuf buf;
3615
111k
    const xmlChar *start;
3616
111k
    int quote, length;
3617
3618
111k
    xmlSBufInit(&buf, maxLength);
3619
3620
111k
    GROW;
3621
3622
111k
    quote = CUR;
3623
111k
    if ((quote != '"') && (quote != '\'')) {
3624
0
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
3625
0
  return(NULL);
3626
0
    }
3627
111k
    CUR_PTR++;
3628
3629
111k
    length = 0;
3630
3631
    /*
3632
     * Copy raw content of the entity into a buffer
3633
     */
3634
40.7M
    while (1) {
3635
40.7M
        int c;
3636
3637
40.7M
        if (PARSER_STOPPED(ctxt))
3638
3
            goto error;
3639
3640
40.7M
        if (CUR_PTR >= ctxt->input->end) {
3641
2.05k
            xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL);
3642
2.05k
            goto error;
3643
2.05k
        }
3644
3645
40.7M
        c = CUR;
3646
3647
40.7M
        if (c == 0) {
3648
225
            xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
3649
225
                    "invalid character in entity value\n");
3650
225
            goto error;
3651
225
        }
3652
40.7M
        if (c == quote)
3653
108k
            break;
3654
40.6M
        NEXTL(1);
3655
40.6M
        length += 1;
3656
3657
        /*
3658
         * TODO: Check growth threshold
3659
         */
3660
40.6M
        if (ctxt->input->end - CUR_PTR < 10)
3661
71.2k
            GROW;
3662
40.6M
    }
3663
3664
108k
    start = CUR_PTR - length;
3665
3666
108k
    if (orig != NULL) {
3667
108k
        *orig = xmlStrndup(start, length);
3668
108k
        if (*orig == NULL)
3669
1
            xmlErrMemory(ctxt);
3670
108k
    }
3671
3672
108k
    xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr);
3673
3674
108k
    NEXTL(1);
3675
3676
108k
    return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long"));
3677
3678
2.28k
error:
3679
2.28k
    xmlSBufCleanup(&buf, ctxt, "entity length too long");
3680
2.28k
    return(NULL);
3681
111k
}
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
3.41k
xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) {
3693
3.41k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3694
3.41k
    const xmlChar *str;
3695
3.41k
    unsigned long expandedSize = pent->length;
3696
3.41k
    int c, flags;
3697
3698
3.41k
    depth += 1;
3699
3.41k
    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
3.41k
    if (pent->flags & XML_ENT_EXPANDING) {
3706
14
        xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3707
14
        return;
3708
14
    }
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
3.39k
    if (ctxt->inSubset == 0)
3716
3.26k
        flags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
3717
133
    else
3718
133
        flags = XML_ENT_VALIDATED;
3719
3720
3.39k
    str = pent->content;
3721
3.39k
    if (str == NULL)
3722
3
        goto done;
3723
3724
    /*
3725
     * Note that entity values are already validated. We only check
3726
     * for illegal less-than signs and compute the expanded size
3727
     * of the entity. No special handling for multi-byte characters
3728
     * is needed.
3729
     */
3730
13.3M
    while (!PARSER_STOPPED(ctxt)) {
3731
13.3M
        c = *str;
3732
3733
13.3M
  if (c != '&') {
3734
13.3M
            if (c == 0)
3735
3.27k
                break;
3736
3737
13.3M
            if (c == '<')
3738
1.19k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3739
1.19k
                        "'<' in entity '%s' is not allowed in attributes "
3740
1.19k
                        "values\n", pent->name);
3741
3742
13.3M
            str += 1;
3743
13.3M
        } else if (str[1] == '#') {
3744
1.15k
            int val;
3745
3746
1.15k
      val = xmlParseStringCharRef(ctxt, &str);
3747
1.15k
      if (val == 0) {
3748
3
                pent->content[0] = 0;
3749
3
                break;
3750
3
            }
3751
7.65k
  } else {
3752
7.65k
            xmlChar *name;
3753
7.65k
            xmlEntityPtr ent;
3754
3755
7.65k
      name = xmlParseStringEntityRef(ctxt, &str);
3756
7.65k
      if (name == NULL) {
3757
13
                pent->content[0] = 0;
3758
13
                break;
3759
13
            }
3760
3761
7.64k
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3762
7.64k
            xmlFree(name);
3763
3764
7.64k
            if ((ent != NULL) &&
3765
6.26k
                (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) {
3766
5.60k
                if ((ent->flags & flags) != flags) {
3767
2.63k
                    pent->flags |= XML_ENT_EXPANDING;
3768
2.63k
                    xmlCheckEntityInAttValue(ctxt, ent, depth);
3769
2.63k
                    pent->flags &= ~XML_ENT_EXPANDING;
3770
2.63k
                }
3771
3772
5.60k
                xmlSaturatedAdd(&expandedSize, ent->expandedSize);
3773
5.60k
                xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST);
3774
5.60k
            }
3775
7.64k
        }
3776
13.3M
    }
3777
3778
3.39k
done:
3779
3.39k
    if (ctxt->inSubset == 0)
3780
3.26k
        pent->expandedSize = expandedSize;
3781
3782
3.39k
    pent->flags |= flags;
3783
3.39k
}
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
2.08M
                          int *inSpace, int depth, int check) {
3803
2.08M
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20;
3804
2.08M
    int c, chunkSize;
3805
2.08M
    int normChange = 0;
3806
3807
2.08M
    if (str == NULL)
3808
262
        return(0);
3809
3810
2.08M
    depth += 1;
3811
2.08M
    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
2.08M
    if (pent != NULL) {
3818
2.06M
        if (pent->flags & XML_ENT_EXPANDING) {
3819
4
            xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
3820
4
            return(0);
3821
4
        }
3822
3823
2.06M
        if (check) {
3824
2.06M
            if (xmlParserEntityCheck(ctxt, pent->length))
3825
14
                return(0);
3826
2.06M
        }
3827
2.06M
    }
3828
3829
2.08M
    chunkSize = 0;
3830
3831
    /*
3832
     * Note that entity values are already validated. No special
3833
     * handling for multi-byte characters is needed.
3834
     */
3835
195M
    while (!PARSER_STOPPED(ctxt)) {
3836
195M
        c = *str;
3837
3838
195M
  if (c != '&') {
3839
191M
            if (c == 0)
3840
1.95M
                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
189M
            if ((pent != NULL) && (c == '<')) {
3848
131k
                xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE,
3849
131k
                        "'<' in entity '%s' is not allowed in attributes "
3850
131k
                        "values\n", pent->name);
3851
131k
                break;
3852
131k
            }
3853
3854
189M
            if (c <= 0x20) {
3855
16.6M
                if ((normalize) && (*inSpace)) {
3856
                    /* Skip char */
3857
45.1k
                    if (chunkSize > 0) {
3858
22.0k
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3859
22.0k
                        chunkSize = 0;
3860
22.0k
                    }
3861
45.1k
                    normChange = 1;
3862
16.6M
                } else if (c < 0x20) {
3863
6.59M
                    if (chunkSize > 0) {
3864
1.12M
                        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3865
1.12M
                        chunkSize = 0;
3866
1.12M
                    }
3867
3868
6.59M
                    xmlSBufAddCString(buf, " ", 1);
3869
10.0M
                } else {
3870
10.0M
                    chunkSize += 1;
3871
10.0M
                }
3872
3873
16.6M
                *inSpace = 1;
3874
172M
            } else {
3875
172M
                chunkSize += 1;
3876
172M
                *inSpace = 0;
3877
172M
            }
3878
3879
189M
            str += 1;
3880
189M
        } else if (str[1] == '#') {
3881
1.18M
            int val;
3882
3883
1.18M
            if (chunkSize > 0) {
3884
1.18M
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3885
1.18M
                chunkSize = 0;
3886
1.18M
            }
3887
3888
1.18M
      val = xmlParseStringCharRef(ctxt, &str);
3889
1.18M
      if (val == 0) {
3890
4
                if (pent != NULL)
3891
4
                    pent->content[0] = 0;
3892
4
                break;
3893
4
            }
3894
3895
1.18M
            if (val == ' ') {
3896
17.3k
                if ((normalize) && (*inSpace))
3897
1.41k
                    normChange = 1;
3898
15.9k
                else
3899
15.9k
                    xmlSBufAddCString(buf, " ", 1);
3900
17.3k
                *inSpace = 1;
3901
1.16M
            } else {
3902
1.16M
                xmlSBufAddChar(buf, val);
3903
1.16M
                *inSpace = 0;
3904
1.16M
            }
3905
2.74M
  } else {
3906
2.74M
            xmlChar *name;
3907
2.74M
            xmlEntityPtr ent;
3908
3909
2.74M
            if (chunkSize > 0) {
3910
1.30M
                xmlSBufAddString(buf, str - chunkSize, chunkSize);
3911
1.30M
                chunkSize = 0;
3912
1.30M
            }
3913
3914
2.74M
      name = xmlParseStringEntityRef(ctxt, &str);
3915
2.74M
            if (name == NULL) {
3916
28
                if (pent != NULL)
3917
27
                    pent->content[0] = 0;
3918
28
                break;
3919
28
            }
3920
3921
2.74M
            ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1);
3922
2.74M
            xmlFree(name);
3923
3924
2.74M
      if ((ent != NULL) &&
3925
2.68M
    (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
3926
1.28M
    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
1.28M
                xmlSBufAddString(buf, ent->content, ent->length);
3933
3934
1.28M
                *inSpace = 0;
3935
1.45M
      } else if ((ent != NULL) && (ent->content != NULL)) {
3936
1.40M
                if (pent != NULL)
3937
1.39M
                    pent->flags |= XML_ENT_EXPANDING;
3938
1.40M
    normChange |= xmlExpandEntityInAttValue(ctxt, buf,
3939
1.40M
                        ent->content, ent, normalize, inSpace, depth, check);
3940
1.40M
                if (pent != NULL)
3941
1.39M
                    pent->flags &= ~XML_ENT_EXPANDING;
3942
1.40M
      }
3943
2.74M
        }
3944
195M
    }
3945
3946
2.08M
    if (chunkSize > 0)
3947
1.35M
        xmlSBufAddString(buf, str - chunkSize, chunkSize);
3948
3949
2.08M
    return(normChange);
3950
2.08M
}
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
16.9k
                            int normalize) {
3964
16.9k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
3965
9.77k
                         XML_MAX_HUGE_LENGTH :
3966
16.9k
                         XML_MAX_TEXT_LENGTH;
3967
16.9k
    xmlSBuf buf;
3968
16.9k
    int inSpace = 1;
3969
3970
16.9k
    xmlSBufInit(&buf, maxLength);
3971
3972
16.9k
    xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace,
3973
16.9k
                              ctxt->inputNr, /* check */ 0);
3974
3975
16.9k
    if ((normalize) && (inSpace) && (buf.size > 0))
3976
0
        buf.size--;
3977
3978
16.9k
    return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long"));
3979
16.9k
}
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
299k
                         int special, int isNamespace) {
4020
299k
    unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4021
136k
                         XML_MAX_HUGE_LENGTH :
4022
299k
                         XML_MAX_TEXT_LENGTH;
4023
299k
    xmlSBuf buf;
4024
299k
    xmlChar *ret;
4025
299k
    int c, l, quote, entFlags, chunkSize;
4026
299k
    int inSpace = 1;
4027
299k
    int replaceEntities;
4028
299k
    int normalize = (special & XML_SPECIAL_TYPE_MASK) > XML_ATTRIBUTE_CDATA;
4029
299k
    int attvalFlags = 0;
4030
4031
    /* Always expand namespace URIs */
4032
299k
    replaceEntities = (ctxt->replaceEntities) || (isNamespace);
4033
4034
299k
    xmlSBufInit(&buf, maxLength);
4035
4036
299k
    GROW;
4037
4038
299k
    quote = CUR;
4039
299k
    if ((quote != '"') && (quote != '\'')) {
4040
7.69k
  xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL);
4041
7.69k
  return(NULL);
4042
7.69k
    }
4043
291k
    NEXTL(1);
4044
4045
291k
    if (ctxt->inSubset == 0)
4046
226k
        entFlags = XML_ENT_CHECKED | XML_ENT_VALIDATED;
4047
65.7k
    else
4048
65.7k
        entFlags = XML_ENT_VALIDATED;
4049
4050
291k
    inSpace = 1;
4051
291k
    chunkSize = 0;
4052
4053
118M
    while (1) {
4054
118M
        if (PARSER_STOPPED(ctxt))
4055
82
            goto error;
4056
4057
118M
        if (CUR_PTR >= ctxt->input->end) {
4058
36.2k
            xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED,
4059
36.2k
                           "AttValue: ' expected\n");
4060
36.2k
            goto error;
4061
36.2k
        }
4062
4063
        /*
4064
         * TODO: Check growth threshold
4065
         */
4066
118M
        if (ctxt->input->end - CUR_PTR < 10)
4067
376k
            GROW;
4068
4069
118M
        c = CUR;
4070
4071
118M
        if (c >= 0x80) {
4072
47.8M
            l = xmlUTF8MultibyteLen(ctxt, CUR_PTR,
4073
47.8M
                    "invalid character in attribute value\n");
4074
47.8M
            if (l == 0) {
4075
44.5M
                if (chunkSize > 0) {
4076
253k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4077
253k
                    chunkSize = 0;
4078
253k
                }
4079
44.5M
                xmlSBufAddReplChar(&buf);
4080
44.5M
                NEXTL(1);
4081
44.5M
            } else {
4082
3.36M
                chunkSize += l;
4083
3.36M
                NEXTL(l);
4084
3.36M
            }
4085
4086
47.8M
            inSpace = 0;
4087
70.2M
        } else if (c != '&') {
4088
69.0M
            if (c > 0x20) {
4089
55.4M
                if (c == quote)
4090
254k
                    break;
4091
4092
55.1M
                if (c == '<')
4093
827k
                    xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL);
4094
4095
55.1M
                chunkSize += 1;
4096
55.1M
                inSpace = 0;
4097
55.1M
            } else if (!IS_BYTE_CHAR(c)) {
4098
5.74M
                xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR,
4099
5.74M
                        "invalid character in attribute value\n");
4100
5.74M
                if (chunkSize > 0) {
4101
56.1k
                    xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4102
56.1k
                    chunkSize = 0;
4103
56.1k
                }
4104
5.74M
                xmlSBufAddReplChar(&buf);
4105
5.74M
                inSpace = 0;
4106
7.84M
            } else {
4107
                /* Whitespace */
4108
7.84M
                if ((normalize) && (inSpace)) {
4109
                    /* Skip char */
4110
48.6k
                    if (chunkSize > 0) {
4111
24.7k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4112
24.7k
                        chunkSize = 0;
4113
24.7k
                    }
4114
48.6k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4115
7.79M
                } else if (c < 0x20) {
4116
                    /* Convert to space */
4117
6.10M
                    if (chunkSize > 0) {
4118
239k
                        xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4119
239k
                        chunkSize = 0;
4120
239k
                    }
4121
4122
6.10M
                    xmlSBufAddCString(&buf, " ", 1);
4123
6.10M
                } else {
4124
1.69M
                    chunkSize += 1;
4125
1.69M
                }
4126
4127
7.84M
                inSpace = 1;
4128
4129
7.84M
                if ((c == 0xD) && (NXT(1) == 0xA))
4130
71.6k
                    CUR_PTR++;
4131
7.84M
            }
4132
4133
68.7M
            NEXTL(1);
4134
68.7M
        } else if (NXT(1) == '#') {
4135
31.2k
            int val;
4136
4137
31.2k
            if (chunkSize > 0) {
4138
24.3k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4139
24.3k
                chunkSize = 0;
4140
24.3k
            }
4141
4142
31.2k
            val = xmlParseCharRef(ctxt);
4143
31.2k
            if (val == 0)
4144
531
                goto error;
4145
4146
30.7k
            if ((val == '&') && (!replaceEntities)) {
4147
                /*
4148
                 * The reparsing will be done in xmlNodeParseContent()
4149
                 * called from SAX2.c
4150
                 */
4151
3.89k
                xmlSBufAddCString(&buf, "&#38;", 5);
4152
3.89k
                inSpace = 0;
4153
26.8k
            } else if (val == ' ') {
4154
1.08k
                if ((normalize) && (inSpace))
4155
200
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4156
885
                else
4157
885
                    xmlSBufAddCString(&buf, " ", 1);
4158
1.08k
                inSpace = 1;
4159
25.7k
            } else {
4160
25.7k
                xmlSBufAddChar(&buf, val);
4161
25.7k
                inSpace = 0;
4162
25.7k
            }
4163
1.21M
        } else {
4164
1.21M
            const xmlChar *name;
4165
1.21M
            xmlEntityPtr ent;
4166
4167
1.21M
            if (chunkSize > 0) {
4168
729k
                xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4169
729k
                chunkSize = 0;
4170
729k
            }
4171
4172
1.21M
            name = xmlParseEntityRefInternal(ctxt);
4173
1.21M
            if (name == NULL) {
4174
                /*
4175
                 * Probably a literal '&' which wasn't escaped.
4176
                 * TODO: Handle gracefully in recovery mode.
4177
                 */
4178
126k
                continue;
4179
126k
            }
4180
4181
1.08M
            ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1);
4182
1.08M
            if (ent == NULL)
4183
313k
                continue;
4184
4185
771k
            if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) {
4186
14.9k
                if ((ent->content[0] == '&') && (!replaceEntities))
4187
2.03k
                    xmlSBufAddCString(&buf, "&#38;", 5);
4188
12.8k
                else
4189
12.8k
                    xmlSBufAddString(&buf, ent->content, ent->length);
4190
14.9k
                inSpace = 0;
4191
756k
            } else if (replaceEntities) {
4192
667k
                if (xmlExpandEntityInAttValue(ctxt, &buf,
4193
667k
                        ent->content, ent, normalize, &inSpace, ctxt->inputNr,
4194
667k
                        /* check */ 1) > 0)
4195
6.29k
                    attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4196
667k
            } else {
4197
88.7k
                if ((ent->flags & entFlags) != entFlags)
4198
775
                    xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr);
4199
4200
88.7k
                if (xmlParserEntityCheck(ctxt, ent->expandedSize)) {
4201
72
                    ent->content[0] = 0;
4202
72
                    goto error;
4203
72
                }
4204
4205
                /*
4206
                 * Just output the reference
4207
                 */
4208
88.6k
                xmlSBufAddCString(&buf, "&", 1);
4209
88.6k
                xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name));
4210
88.6k
                xmlSBufAddCString(&buf, ";", 1);
4211
4212
88.6k
                inSpace = 0;
4213
88.6k
            }
4214
771k
  }
4215
118M
    }
4216
4217
254k
    if ((buf.mem == NULL) && (outFlags != NULL)) {
4218
121k
        ret = (xmlChar *) CUR_PTR - chunkSize;
4219
4220
121k
        if (attlen != NULL)
4221
121k
            *attlen = chunkSize;
4222
121k
        if ((normalize) && (inSpace) && (chunkSize > 0)) {
4223
299
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4224
299
            *attlen -= 1;
4225
299
        }
4226
4227
        /* Report potential error */
4228
121k
        xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4229
133k
    } else {
4230
133k
        if (chunkSize > 0)
4231
116k
            xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize);
4232
4233
133k
        if ((normalize) && (inSpace) && (buf.size > 0)) {
4234
420
            attvalFlags |= XML_ATTVAL_NORM_CHANGE;
4235
420
            buf.size--;
4236
420
        }
4237
4238
133k
        ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long");
4239
133k
        attvalFlags |= XML_ATTVAL_ALLOC;
4240
4241
133k
        if (ret != NULL) {
4242
133k
            if (attlen != NULL)
4243
20.2k
                *attlen = buf.size;
4244
133k
        }
4245
133k
    }
4246
4247
254k
    if (outFlags != NULL)
4248
142k
        *outFlags = attvalFlags;
4249
4250
254k
    NEXTL(1);
4251
4252
254k
    return(ret);
4253
4254
36.9k
error:
4255
36.9k
    xmlSBufCleanup(&buf, ctxt, "AttValue length too long");
4256
36.9k
    return(NULL);
4257
291k
}
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
154k
xmlParseAttValue(xmlParserCtxt *ctxt) {
4297
154k
    if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL);
4298
154k
    return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0));
4299
154k
}
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
26.8k
xmlParseSystemLiteral(xmlParserCtxt *ctxt) {
4314
26.8k
    xmlChar *buf = NULL;
4315
26.8k
    int len = 0;
4316
26.8k
    int size = XML_PARSER_BUFFER_SIZE;
4317
26.8k
    int cur, l;
4318
26.8k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4319
7.88k
                    XML_MAX_TEXT_LENGTH :
4320
26.8k
                    XML_MAX_NAME_LENGTH;
4321
26.8k
    xmlChar stop;
4322
4323
26.8k
    if (RAW == '"') {
4324
11.1k
        NEXT;
4325
11.1k
  stop = '"';
4326
15.7k
    } else if (RAW == '\'') {
4327
13.9k
        NEXT;
4328
13.9k
  stop = '\'';
4329
13.9k
    } else {
4330
1.81k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4331
1.81k
  return(NULL);
4332
1.81k
    }
4333
4334
25.0k
    buf = xmlMalloc(size);
4335
25.0k
    if (buf == NULL) {
4336
3
        xmlErrMemory(ctxt);
4337
3
  return(NULL);
4338
3
    }
4339
25.0k
    cur = xmlCurrentCharRecover(ctxt, &l);
4340
2.02M
    while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */
4341
1.99M
  if (len + 5 >= size) {
4342
1.74k
      xmlChar *tmp;
4343
1.74k
            int newSize;
4344
4345
1.74k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4346
1.74k
            if (newSize < 0) {
4347
1
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral");
4348
1
                xmlFree(buf);
4349
1
                return(NULL);
4350
1
            }
4351
1.74k
      tmp = xmlRealloc(buf, newSize);
4352
1.74k
      if (tmp == NULL) {
4353
1
          xmlFree(buf);
4354
1
    xmlErrMemory(ctxt);
4355
1
    return(NULL);
4356
1
      }
4357
1.74k
      buf = tmp;
4358
1.74k
            size = newSize;
4359
1.74k
  }
4360
1.99M
  COPY_BUF(buf, len, cur);
4361
1.99M
  NEXTL(l);
4362
1.99M
  cur = xmlCurrentCharRecover(ctxt, &l);
4363
1.99M
    }
4364
25.0k
    buf[len] = 0;
4365
25.0k
    if (!IS_CHAR(cur)) {
4366
2.15k
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4367
22.8k
    } else {
4368
22.8k
  NEXT;
4369
22.8k
    }
4370
25.0k
    return(buf);
4371
25.0k
}
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
14.0k
xmlParsePubidLiteral(xmlParserCtxt *ctxt) {
4386
14.0k
    xmlChar *buf = NULL;
4387
14.0k
    int len = 0;
4388
14.0k
    int size = XML_PARSER_BUFFER_SIZE;
4389
14.0k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4390
1.28k
                    XML_MAX_TEXT_LENGTH :
4391
14.0k
                    XML_MAX_NAME_LENGTH;
4392
14.0k
    xmlChar cur;
4393
14.0k
    xmlChar stop;
4394
4395
14.0k
    if (RAW == '"') {
4396
915
        NEXT;
4397
915
  stop = '"';
4398
13.1k
    } else if (RAW == '\'') {
4399
12.7k
        NEXT;
4400
12.7k
  stop = '\'';
4401
12.7k
    } else {
4402
382
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL);
4403
382
  return(NULL);
4404
382
    }
4405
13.6k
    buf = xmlMalloc(size);
4406
13.6k
    if (buf == NULL) {
4407
2
  xmlErrMemory(ctxt);
4408
2
  return(NULL);
4409
2
    }
4410
13.6k
    cur = CUR;
4411
171k
    while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) &&
4412
157k
           (PARSER_STOPPED(ctxt) == 0)) { /* checked */
4413
157k
  if (len + 1 >= size) {
4414
350
      xmlChar *tmp;
4415
350
            int newSize;
4416
4417
350
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4418
350
            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
349
      tmp = xmlRealloc(buf, newSize);
4424
349
      if (tmp == NULL) {
4425
1
    xmlErrMemory(ctxt);
4426
1
    xmlFree(buf);
4427
1
    return(NULL);
4428
1
      }
4429
348
      buf = tmp;
4430
348
            size = newSize;
4431
348
  }
4432
157k
  buf[len++] = cur;
4433
157k
  NEXT;
4434
157k
  cur = CUR;
4435
157k
    }
4436
13.6k
    buf[len] = 0;
4437
13.6k
    if (cur != stop) {
4438
794
  xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL);
4439
12.8k
    } else {
4440
12.8k
  NEXTL(1);
4441
12.8k
    }
4442
13.6k
    return(buf);
4443
13.6k
}
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
951k
              int isBlank) {
4488
951k
    int checkBlanks;
4489
4490
951k
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
4491
136k
        return;
4492
4493
815k
    checkBlanks = (!ctxt->keepBlanks) ||
4494
335k
                  (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
815k
    if ((checkBlanks) &&
4502
479k
        (areBlanks(ctxt, buf, size, isBlank))) {
4503
15.1k
        if ((ctxt->sax->ignorableWhitespace != NULL) &&
4504
15.1k
            (ctxt->keepBlanks))
4505
0
            ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
4506
800k
    } else {
4507
800k
        if (ctxt->sax->characters != NULL)
4508
800k
            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
800k
        if ((checkBlanks) && (*ctxt->space == -1))
4515
128k
            *ctxt->space = -2;
4516
800k
    }
4517
815k
}
4518
4519
/**
4520
 * Parse character data. Always makes progress if the first char isn't
4521
 * '<' or '&'.
4522
 *
4523
 * The right angle bracket (>) may be represented using the string "&gt;",
4524
 * and must, for compatibility, be escaped using "&gt;" or a character
4525
 * reference when it appears in the string "]]>" in content, when that
4526
 * string is not marking the end of a CDATA section.
4527
 *
4528
 *     [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*)
4529
 * @param ctxt  an XML parser context
4530
 * @param partial  buffer may contain partial UTF-8 sequences
4531
 */
4532
static void
4533
1.19M
xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) {
4534
1.19M
    const xmlChar *in;
4535
1.19M
    int line = ctxt->input->line;
4536
1.19M
    int col = ctxt->input->col;
4537
1.19M
    int ccol;
4538
1.19M
    int terminate = 0;
4539
4540
1.19M
    GROW;
4541
    /*
4542
     * Accelerated common case where input don't need to be
4543
     * modified before passing it to the handler.
4544
     */
4545
1.19M
    in = ctxt->input->cur;
4546
1.20M
    do {
4547
1.27M
get_more_space:
4548
1.44M
        while (*in == 0x20) { in++; ctxt->input->col++; }
4549
1.27M
        if (*in == 0xA) {
4550
4.57M
            do {
4551
4.57M
                ctxt->input->line++; ctxt->input->col = 1;
4552
4.57M
                in++;
4553
4.57M
            } while (*in == 0xA);
4554
70.3k
            goto get_more_space;
4555
70.3k
        }
4556
1.20M
        if (*in == '<') {
4557
82.0k
            while (in > ctxt->input->cur) {
4558
41.0k
                const xmlChar *tmp = ctxt->input->cur;
4559
41.0k
                size_t nbchar = in - tmp;
4560
4561
41.0k
                if (nbchar > XML_MAX_ITEMS)
4562
0
                    nbchar = XML_MAX_ITEMS;
4563
41.0k
                ctxt->input->cur += nbchar;
4564
4565
41.0k
                xmlCharacters(ctxt, tmp, nbchar, 1);
4566
41.0k
            }
4567
41.0k
            return;
4568
41.0k
        }
4569
4570
1.54M
get_more:
4571
1.54M
        ccol = ctxt->input->col;
4572
11.3M
        while (test_char_data[*in]) {
4573
9.81M
            in++;
4574
9.81M
            ccol++;
4575
9.81M
        }
4576
1.54M
        ctxt->input->col = ccol;
4577
1.54M
        if (*in == 0xA) {
4578
3.87M
            do {
4579
3.87M
                ctxt->input->line++; ctxt->input->col = 1;
4580
3.87M
                in++;
4581
3.87M
            } while (*in == 0xA);
4582
64.3k
            goto get_more;
4583
64.3k
        }
4584
1.48M
        if (*in == ']') {
4585
319k
            size_t avail = ctxt->input->end - in;
4586
4587
319k
            if (partial && avail < 2) {
4588
80
                terminate = 1;
4589
80
                goto invoke_callback;
4590
80
            }
4591
319k
            if (in[1] == ']') {
4592
197k
                if (partial && avail < 3) {
4593
361
                    terminate = 1;
4594
361
                    goto invoke_callback;
4595
361
                }
4596
197k
                if (in[2] == '>')
4597
2.56k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4598
197k
            }
4599
4600
319k
            in++;
4601
319k
            ctxt->input->col++;
4602
319k
            goto get_more;
4603
319k
        }
4604
4605
1.16M
invoke_callback:
4606
1.89M
        while (in > ctxt->input->cur) {
4607
725k
            const xmlChar *tmp = ctxt->input->cur;
4608
725k
            size_t nbchar = in - tmp;
4609
4610
725k
            if (nbchar > XML_MAX_ITEMS)
4611
0
                nbchar = XML_MAX_ITEMS;
4612
725k
            ctxt->input->cur += nbchar;
4613
4614
725k
            xmlCharacters(ctxt, tmp, nbchar, 0);
4615
4616
725k
            line = ctxt->input->line;
4617
725k
            col = ctxt->input->col;
4618
725k
        }
4619
1.16M
        ctxt->input->cur = in;
4620
1.16M
        if (*in == 0xD) {
4621
14.5k
            in++;
4622
14.5k
            if (*in == 0xA) {
4623
7.79k
                ctxt->input->cur = in;
4624
7.79k
                in++;
4625
7.79k
                ctxt->input->line++; ctxt->input->col = 1;
4626
7.79k
                continue; /* while */
4627
7.79k
            }
4628
6.75k
            in--;
4629
6.75k
        }
4630
1.15M
        if (*in == '<') {
4631
594k
            return;
4632
594k
        }
4633
562k
        if (*in == '&') {
4634
29.1k
            return;
4635
29.1k
        }
4636
533k
        if (terminate) {
4637
441
            return;
4638
441
        }
4639
533k
        SHRINK;
4640
533k
        GROW;
4641
533k
        in = ctxt->input->cur;
4642
541k
    } while (((*in >= 0x20) && (*in <= 0x7F)) ||
4643
538k
             (*in == 0x09) || (*in == 0x0a));
4644
533k
    ctxt->input->line = line;
4645
533k
    ctxt->input->col = col;
4646
533k
    xmlParseCharDataComplex(ctxt, partial);
4647
533k
}
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
533k
xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) {
4661
533k
    xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5];
4662
533k
    int nbchar = 0;
4663
533k
    int cur, l;
4664
4665
533k
    cur = xmlCurrentCharRecover(ctxt, &l);
4666
11.5M
    while ((cur != '<') && /* checked */
4667
11.4M
           (cur != '&') &&
4668
11.4M
     (IS_CHAR(cur))) {
4669
10.9M
        if (cur == ']') {
4670
85.9k
            size_t avail = ctxt->input->end - ctxt->input->cur;
4671
4672
85.9k
            if (partial && avail < 2)
4673
244
                break;
4674
85.7k
            if (NXT(1) == ']') {
4675
25.2k
                if (partial && avail < 3)
4676
69
                    break;
4677
25.1k
                if (NXT(2) == '>')
4678
1.06k
                    xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL);
4679
25.1k
            }
4680
85.7k
        }
4681
4682
10.9M
  COPY_BUF(buf, nbchar, cur);
4683
  /* move current position before possible calling of ctxt->sax->characters */
4684
10.9M
  NEXTL(l);
4685
10.9M
  if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) {
4686
63.0k
      buf[nbchar] = 0;
4687
4688
63.0k
            xmlCharacters(ctxt, buf, nbchar, 0);
4689
63.0k
      nbchar = 0;
4690
63.0k
            SHRINK;
4691
63.0k
  }
4692
10.9M
  cur = xmlCurrentCharRecover(ctxt, &l);
4693
10.9M
    }
4694
533k
    if (nbchar != 0) {
4695
122k
        buf[nbchar] = 0;
4696
4697
122k
        xmlCharacters(ctxt, buf, nbchar, 0);
4698
122k
    }
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
533k
    if (ctxt->input->cur < ctxt->input->end) {
4707
529k
        if ((cur == 0) && (CUR != 0)) {
4708
457
            if (partial == 0) {
4709
382
                xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4710
382
                        "Incomplete UTF-8 sequence starting with %02X\n", CUR);
4711
382
                NEXTL(1);
4712
382
            }
4713
529k
        } else if ((cur != '<') && (cur != '&') && (cur != ']')) {
4714
            /* Generate the error and skip the offending character */
4715
466k
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4716
466k
                              "PCDATA invalid Char value %d\n", cur);
4717
466k
            NEXTL(l);
4718
466k
        }
4719
529k
    }
4720
533k
}
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
46.2k
xmlParseExternalID(xmlParserCtxt *ctxt, xmlChar **publicId, int strict) {
4756
46.2k
    xmlChar *URI = NULL;
4757
4758
46.2k
    *publicId = NULL;
4759
46.2k
    if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) {
4760
14.2k
        SKIP(6);
4761
14.2k
  if (SKIP_BLANKS == 0) {
4762
579
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4763
579
                     "Space required after 'SYSTEM'\n");
4764
579
  }
4765
14.2k
  URI = xmlParseSystemLiteral(ctxt);
4766
14.2k
  if (URI == NULL) {
4767
1.27k
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4768
1.27k
        }
4769
31.9k
    } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) {
4770
14.0k
        SKIP(6);
4771
14.0k
  if (SKIP_BLANKS == 0) {
4772
465
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4773
465
        "Space required after 'PUBLIC'\n");
4774
465
  }
4775
14.0k
  *publicId = xmlParsePubidLiteral(ctxt);
4776
14.0k
  if (*publicId == NULL) {
4777
386
      xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL);
4778
386
  }
4779
14.0k
  if (strict) {
4780
      /*
4781
       * We don't handle [83] so "S SystemLiteral" is required.
4782
       */
4783
12.4k
      if (SKIP_BLANKS == 0) {
4784
533
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
4785
533
      "Space required after the Public Identifier\n");
4786
533
      }
4787
12.4k
  } 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
1.60k
      if (SKIP_BLANKS == 0) return(NULL);
4795
364
      if ((CUR != '\'') && (CUR != '"')) return(NULL);
4796
364
  }
4797
12.5k
  URI = xmlParseSystemLiteral(ctxt);
4798
12.5k
  if (URI == NULL) {
4799
547
      xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL);
4800
547
        }
4801
12.5k
    }
4802
44.6k
    return(URI);
4803
46.2k
}
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
69.8k
                       size_t len, size_t size) {
4820
69.8k
    int q, ql;
4821
69.8k
    int r, rl;
4822
69.8k
    int cur, l;
4823
69.8k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4824
35.5k
                    XML_MAX_HUGE_LENGTH :
4825
69.8k
                    XML_MAX_TEXT_LENGTH;
4826
4827
69.8k
    if (buf == NULL) {
4828
22.1k
        len = 0;
4829
22.1k
  size = XML_PARSER_BUFFER_SIZE;
4830
22.1k
  buf = xmlMalloc(size);
4831
22.1k
  if (buf == NULL) {
4832
1
      xmlErrMemory(ctxt);
4833
1
      return;
4834
1
  }
4835
22.1k
    }
4836
69.8k
    q = xmlCurrentCharRecover(ctxt, &ql);
4837
69.8k
    if (q == 0)
4838
1.55k
        goto not_terminated;
4839
68.2k
    if (!IS_CHAR(q)) {
4840
663
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4841
663
                          "xmlParseComment: invalid xmlChar value %d\n",
4842
663
                    q);
4843
663
  xmlFree (buf);
4844
663
  return;
4845
663
    }
4846
67.6k
    NEXTL(ql);
4847
67.6k
    r = xmlCurrentCharRecover(ctxt, &rl);
4848
67.6k
    if (r == 0)
4849
326
        goto not_terminated;
4850
67.3k
    if (!IS_CHAR(r)) {
4851
518
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4852
518
                          "xmlParseComment: invalid xmlChar value %d\n",
4853
518
                    r);
4854
518
  xmlFree (buf);
4855
518
  return;
4856
518
    }
4857
66.7k
    NEXTL(rl);
4858
66.7k
    cur = xmlCurrentCharRecover(ctxt, &l);
4859
66.7k
    if (cur == 0)
4860
447
        goto not_terminated;
4861
14.4M
    while (IS_CHAR(cur) && /* checked */
4862
14.4M
           ((cur != '>') ||
4863
14.3M
      (r != '-') || (q != '-'))) {
4864
14.3M
  if ((r == '-') && (q == '-')) {
4865
548k
      xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL);
4866
548k
  }
4867
14.3M
  if (len + 5 >= size) {
4868
31.2k
      xmlChar *tmp;
4869
31.2k
            int newSize;
4870
4871
31.2k
      newSize = xmlGrowCapacity(size, 1, 1, maxLength);
4872
31.2k
            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
31.2k
      tmp = xmlRealloc(buf, newSize);
4879
31.2k
      if (tmp == NULL) {
4880
1
    xmlErrMemory(ctxt);
4881
1
    xmlFree(buf);
4882
1
    return;
4883
1
      }
4884
31.2k
      buf = tmp;
4885
31.2k
            size = newSize;
4886
31.2k
  }
4887
14.3M
  COPY_BUF(buf, len, q);
4888
4889
14.3M
  q = r;
4890
14.3M
  ql = rl;
4891
14.3M
  r = cur;
4892
14.3M
  rl = l;
4893
4894
14.3M
  NEXTL(l);
4895
14.3M
  cur = xmlCurrentCharRecover(ctxt, &l);
4896
4897
14.3M
    }
4898
66.3k
    buf[len] = 0;
4899
66.3k
    if (cur == 0) {
4900
4.49k
  xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4901
4.49k
                       "Comment not terminated \n<!--%.50s\n", buf);
4902
61.8k
    } else if (!IS_CHAR(cur)) {
4903
784
        xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
4904
784
                          "xmlParseComment: invalid xmlChar value %d\n",
4905
784
                    cur);
4906
61.0k
    } else {
4907
61.0k
        NEXT;
4908
61.0k
  if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
4909
61.0k
      (!ctxt->disableSAX))
4910
57.2k
      ctxt->sax->comment(ctxt->userData, buf);
4911
61.0k
    }
4912
66.3k
    xmlFree(buf);
4913
66.3k
    return;
4914
2.32k
not_terminated:
4915
2.32k
    xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED,
4916
2.32k
       "Comment not terminated\n", NULL);
4917
2.32k
    xmlFree(buf);
4918
2.32k
}
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
383k
xmlParseComment(xmlParserCtxt *ctxt) {
4933
383k
    xmlChar *buf = NULL;
4934
383k
    size_t size = XML_PARSER_BUFFER_SIZE;
4935
383k
    size_t len = 0;
4936
383k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
4937
134k
                       XML_MAX_HUGE_LENGTH :
4938
383k
                       XML_MAX_TEXT_LENGTH;
4939
383k
    const xmlChar *in;
4940
383k
    size_t nbchar = 0;
4941
383k
    int ccol;
4942
4943
    /*
4944
     * Check that there is a comment right here.
4945
     */
4946
383k
    if ((RAW != '<') || (NXT(1) != '!'))
4947
0
        return;
4948
383k
    SKIP(2);
4949
383k
    if ((RAW != '-') || (NXT(1) != '-'))
4950
9
        return;
4951
383k
    SKIP(2);
4952
383k
    GROW;
4953
4954
    /*
4955
     * Accelerated common case where input don't need to be
4956
     * modified before passing it to the handler.
4957
     */
4958
383k
    in = ctxt->input->cur;
4959
383k
    do {
4960
383k
  if (*in == 0xA) {
4961
2.86k
      do {
4962
2.86k
    ctxt->input->line++; ctxt->input->col = 1;
4963
2.86k
    in++;
4964
2.86k
      } while (*in == 0xA);
4965
806
  }
4966
984k
get_more:
4967
984k
        ccol = ctxt->input->col;
4968
7.69M
  while (((*in > '-') && (*in <= 0x7F)) ||
4969
1.98M
         ((*in >= 0x20) && (*in < '-')) ||
4970
6.70M
         (*in == 0x09)) {
4971
6.70M
        in++;
4972
6.70M
        ccol++;
4973
6.70M
  }
4974
984k
  ctxt->input->col = ccol;
4975
984k
  if (*in == 0xA) {
4976
119k
      do {
4977
119k
    ctxt->input->line++; ctxt->input->col = 1;
4978
119k
    in++;
4979
119k
      } while (*in == 0xA);
4980
7.03k
      goto get_more;
4981
7.03k
  }
4982
977k
  nbchar = in - ctxt->input->cur;
4983
  /*
4984
   * save current set of data
4985
   */
4986
977k
  if (nbchar > 0) {
4987
662k
            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
662k
            if (buf == NULL) {
4994
78.4k
                if ((*in == '-') && (in[1] == '-'))
4995
29.7k
                    size = nbchar + 1;
4996
48.7k
                else
4997
48.7k
                    size = XML_PARSER_BUFFER_SIZE + nbchar;
4998
78.4k
                buf = xmlMalloc(size);
4999
78.4k
                if (buf == NULL) {
5000
9
                    xmlErrMemory(ctxt);
5001
9
                    return;
5002
9
                }
5003
78.4k
                len = 0;
5004
584k
            } else if (len + nbchar + 1 >= size) {
5005
11.7k
                xmlChar *new_buf;
5006
11.7k
                size += len + nbchar + XML_PARSER_BUFFER_SIZE;
5007
11.7k
                new_buf = xmlRealloc(buf, size);
5008
11.7k
                if (new_buf == NULL) {
5009
1
                    xmlErrMemory(ctxt);
5010
1
                    xmlFree(buf);
5011
1
                    return;
5012
1
                }
5013
11.7k
                buf = new_buf;
5014
11.7k
            }
5015
662k
            memcpy(&buf[len], ctxt->input->cur, nbchar);
5016
662k
            len += nbchar;
5017
662k
            buf[len] = 0;
5018
662k
  }
5019
977k
  ctxt->input->cur = in;
5020
977k
  if (*in == 0xA) {
5021
0
      in++;
5022
0
      ctxt->input->line++; ctxt->input->col = 1;
5023
0
  }
5024
977k
  if (*in == 0xD) {
5025
50.9k
      in++;
5026
50.9k
      if (*in == 0xA) {
5027
21.2k
    ctxt->input->cur = in;
5028
21.2k
    in++;
5029
21.2k
    ctxt->input->line++; ctxt->input->col = 1;
5030
21.2k
    goto get_more;
5031
21.2k
      }
5032
29.6k
      in--;
5033
29.6k
  }
5034
955k
  SHRINK;
5035
955k
  GROW;
5036
955k
  in = ctxt->input->cur;
5037
955k
  if (*in == '-') {
5038
885k
      if (in[1] == '-') {
5039
634k
          if (in[2] == '>') {
5040
313k
        SKIP(3);
5041
313k
        if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) &&
5042
313k
            (!ctxt->disableSAX)) {
5043
305k
      if (buf != NULL)
5044
28.8k
          ctxt->sax->comment(ctxt->userData, buf);
5045
277k
      else
5046
277k
          ctxt->sax->comment(ctxt->userData, BAD_CAST "");
5047
305k
        }
5048
313k
        if (buf != NULL)
5049
30.7k
            xmlFree(buf);
5050
313k
        return;
5051
313k
    }
5052
320k
    if (buf != NULL) {
5053
317k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5054
317k
                          "Double hyphen within comment: "
5055
317k
                                      "<!--%.50s\n",
5056
317k
              buf);
5057
317k
    } else
5058
2.54k
        xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT,
5059
2.54k
                          "Double hyphen within comment\n", NULL);
5060
320k
    in++;
5061
320k
    ctxt->input->col++;
5062
320k
      }
5063
572k
      in++;
5064
572k
      ctxt->input->col++;
5065
572k
      goto get_more;
5066
885k
  }
5067
955k
    } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a));
5068
69.8k
    xmlParseCommentComplex(ctxt, buf, len, size);
5069
69.8k
}
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
232k
xmlParsePITarget(xmlParserCtxt *ctxt) {
5085
232k
    const xmlChar *name;
5086
5087
232k
    name = xmlParseName(ctxt);
5088
232k
    if ((name != NULL) &&
5089
228k
        ((name[0] == 'x') || (name[0] == 'X')) &&
5090
15.3k
        ((name[1] == 'm') || (name[1] == 'M')) &&
5091
12.5k
        ((name[2] == 'l') || (name[2] == 'L'))) {
5092
7.51k
  int i;
5093
7.51k
  if ((name[0] == 'x') && (name[1] == 'm') &&
5094
7.14k
      (name[2] == 'l') && (name[3] == 0)) {
5095
6.16k
      xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5096
6.16k
     "XML declaration allowed only at the start of the document\n");
5097
6.16k
      return(name);
5098
6.16k
  } else if (name[3] == 0) {
5099
554
      xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL);
5100
554
      return(name);
5101
554
  }
5102
1.99k
  for (i = 0;;i++) {
5103
1.99k
      if (xmlW3CPIs[i] == NULL) break;
5104
1.39k
      if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i]))
5105
200
          return(name);
5106
1.39k
  }
5107
598
  xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME,
5108
598
          "xmlParsePITarget: invalid name prefix 'xml'\n",
5109
598
          NULL, NULL);
5110
598
    }
5111
226k
    if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) {
5112
1.04k
  xmlNsErr(ctxt, XML_NS_ERR_COLON,
5113
1.04k
     "colons are forbidden from PI names '%s'\n", name, NULL, NULL);
5114
1.04k
    }
5115
226k
    return(name);
5116
232k
}
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
232k
xmlParsePI(xmlParserCtxt *ctxt) {
5199
232k
    xmlChar *buf = NULL;
5200
232k
    size_t len = 0;
5201
232k
    size_t size = XML_PARSER_BUFFER_SIZE;
5202
232k
    size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ?
5203
50.7k
                       XML_MAX_HUGE_LENGTH :
5204
232k
                       XML_MAX_TEXT_LENGTH;
5205
232k
    int cur, l;
5206
232k
    const xmlChar *target;
5207
5208
232k
    if ((RAW == '<') && (NXT(1) == '?')) {
5209
  /*
5210
   * this is a Processing Instruction.
5211
   */
5212
232k
  SKIP(2);
5213
5214
  /*
5215
   * Parse the target name and check for special support like
5216
   * namespace.
5217
   */
5218
232k
        target = xmlParsePITarget(ctxt);
5219
232k
  if (target != NULL) {
5220
228k
      if ((RAW == '?') && (NXT(1) == '>')) {
5221
203k
    SKIP(2);
5222
5223
    /*
5224
     * SAX: PI detected.
5225
     */
5226
203k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5227
196k
        (ctxt->sax->processingInstruction != NULL))
5228
196k
        ctxt->sax->processingInstruction(ctxt->userData,
5229
196k
                                         target, NULL);
5230
203k
    return;
5231
203k
      }
5232
25.1k
      buf = xmlMalloc(size);
5233
25.1k
      if (buf == NULL) {
5234
11
    xmlErrMemory(ctxt);
5235
11
    return;
5236
11
      }
5237
25.1k
      if (SKIP_BLANKS == 0) {
5238
8.83k
    xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED,
5239
8.83k
        "ParsePI: PI %s space expected\n", target);
5240
8.83k
      }
5241
25.1k
      cur = xmlCurrentCharRecover(ctxt, &l);
5242
5.80M
      while (IS_CHAR(cur) && /* checked */
5243
5.80M
       ((cur != '?') || (NXT(1) != '>'))) {
5244
5.78M
    if (len + 5 >= size) {
5245
15.5k
        xmlChar *tmp;
5246
15.5k
                    int newSize;
5247
5248
15.5k
                    newSize = xmlGrowCapacity(size, 1, 1, maxLength);
5249
15.5k
                    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
15.5k
        tmp = xmlRealloc(buf, newSize);
5256
15.5k
        if (tmp == NULL) {
5257
8
      xmlErrMemory(ctxt);
5258
8
      xmlFree(buf);
5259
8
      return;
5260
8
        }
5261
15.5k
        buf = tmp;
5262
15.5k
                    size = newSize;
5263
15.5k
    }
5264
5.78M
    COPY_BUF(buf, len, cur);
5265
5.78M
    NEXTL(l);
5266
5.78M
    cur = xmlCurrentCharRecover(ctxt, &l);
5267
5.78M
      }
5268
25.1k
      buf[len] = 0;
5269
25.1k
      if (cur != '?') {
5270
9.74k
    xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED,
5271
9.74k
          "ParsePI: PI %s never end ...\n", target);
5272
15.3k
      } else {
5273
15.3k
    SKIP(2);
5274
5275
15.3k
#ifdef LIBXML_CATALOG_ENABLED
5276
15.3k
    if ((ctxt->inSubset == 0) &&
5277
12.9k
        (xmlStrEqual(target, XML_CATALOG_PI))) {
5278
404
        xmlCatalogAllow allow = xmlCatalogGetDefaults();
5279
5280
404
        if ((ctxt->options & XML_PARSE_CATALOG_PI) &&
5281
201
                        ((allow == XML_CATA_ALLOW_DOCUMENT) ||
5282
201
       (allow == XML_CATA_ALLOW_ALL)))
5283
0
      xmlParseCatalogPI(ctxt, buf);
5284
404
    }
5285
15.3k
#endif
5286
5287
    /*
5288
     * SAX: PI detected.
5289
     */
5290
15.3k
    if ((ctxt->sax) && (!ctxt->disableSAX) &&
5291
14.4k
        (ctxt->sax->processingInstruction != NULL))
5292
14.4k
        ctxt->sax->processingInstruction(ctxt->userData,
5293
14.4k
                                         target, buf);
5294
15.3k
      }
5295
25.1k
      xmlFree(buf);
5296
25.1k
  } else {
5297
4.80k
      xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL);
5298
4.80k
  }
5299
232k
    }
5300
232k
}
5301
5302
/**
5303
 * Parse a notation declaration. Always consumes '<!'.
5304
 *
5305
 * @deprecated Internal function, don't use.
5306
 *
5307
 *     [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID |  PublicID)
5308
 *                           S? '>'
5309
 *
5310
 * Hence there is actually 3 choices:
5311
 *
5312
 *     'PUBLIC' S PubidLiteral
5313
 *     'PUBLIC' S PubidLiteral S SystemLiteral
5314
 *     'SYSTEM' S SystemLiteral
5315
 *
5316
 * See the NOTE on #xmlParseExternalID.
5317
 *
5318
 * @param ctxt  an XML parser context
5319
 */
5320
5321
void
5322
3.95k
xmlParseNotationDecl(xmlParserCtxt *ctxt) {
5323
3.95k
    const xmlChar *name;
5324
3.95k
    xmlChar *Pubid;
5325
3.95k
    xmlChar *Systemid;
5326
5327
3.95k
    if ((CUR != '<') || (NXT(1) != '!'))
5328
0
        return;
5329
3.95k
    SKIP(2);
5330
5331
3.95k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5332
3.93k
#ifdef LIBXML_VALID_ENABLED
5333
3.93k
  int oldInputNr = ctxt->inputNr;
5334
3.93k
#endif
5335
5336
3.93k
  SKIP(8);
5337
3.93k
  if (SKIP_BLANKS_PE == 0) {
5338
255
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5339
255
         "Space required after '<!NOTATION'\n");
5340
255
      return;
5341
255
  }
5342
5343
3.68k
        name = xmlParseName(ctxt);
5344
3.68k
  if (name == NULL) {
5345
285
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5346
285
      return;
5347
285
  }
5348
3.39k
  if (xmlStrchr(name, ':') != NULL) {
5349
135
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5350
135
         "colons are forbidden from notation names '%s'\n",
5351
135
         name, NULL, NULL);
5352
135
  }
5353
3.39k
  if (SKIP_BLANKS_PE == 0) {
5354
107
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5355
107
         "Space required after the NOTATION name'\n");
5356
107
      return;
5357
107
  }
5358
5359
  /*
5360
   * Parse the IDs.
5361
   */
5362
3.28k
  Systemid = xmlParseExternalID(ctxt, &Pubid, 0);
5363
3.28k
  SKIP_BLANKS_PE;
5364
5365
3.28k
  if (RAW == '>') {
5366
2.68k
#ifdef LIBXML_VALID_ENABLED
5367
2.68k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
5368
19
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
5369
19
                           "Notation declaration doesn't start and stop"
5370
19
                                 " in the same entity\n",
5371
19
                                 NULL, NULL);
5372
19
      }
5373
2.68k
#endif
5374
2.68k
      NEXT;
5375
2.68k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5376
2.16k
    (ctxt->sax->notationDecl != NULL))
5377
2.16k
    ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid);
5378
2.68k
  } else {
5379
600
      xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5380
600
  }
5381
3.28k
  if (Systemid != NULL) xmlFree(Systemid);
5382
3.28k
  if (Pubid != NULL) xmlFree(Pubid);
5383
3.28k
    }
5384
3.95k
}
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
168k
xmlParseEntityDecl(xmlParserCtxt *ctxt) {
5411
168k
    const xmlChar *name = NULL;
5412
168k
    xmlChar *value = NULL;
5413
168k
    xmlChar *URI = NULL, *literal = NULL;
5414
168k
    const xmlChar *ndata = NULL;
5415
168k
    int isParameter = 0;
5416
168k
    xmlChar *orig = NULL;
5417
5418
168k
    if ((CUR != '<') || (NXT(1) != '!'))
5419
0
        return;
5420
168k
    SKIP(2);
5421
5422
    /* GROW; done in the caller */
5423
168k
    if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5424
168k
#ifdef LIBXML_VALID_ENABLED
5425
168k
  int oldInputNr = ctxt->inputNr;
5426
168k
#endif
5427
5428
168k
  SKIP(6);
5429
168k
  if (SKIP_BLANKS_PE == 0) {
5430
2.42k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5431
2.42k
         "Space required after '<!ENTITY'\n");
5432
2.42k
  }
5433
5434
168k
  if (RAW == '%') {
5435
69.5k
      NEXT;
5436
69.5k
      if (SKIP_BLANKS_PE == 0) {
5437
329
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5438
329
             "Space required after '%%'\n");
5439
329
      }
5440
69.5k
      isParameter = 1;
5441
69.5k
  }
5442
5443
168k
        name = xmlParseName(ctxt);
5444
168k
  if (name == NULL) {
5445
31.5k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5446
31.5k
                     "xmlParseEntityDecl: no name\n");
5447
31.5k
            return;
5448
31.5k
  }
5449
136k
  if (xmlStrchr(name, ':') != NULL) {
5450
2.06k
      xmlNsErr(ctxt, XML_NS_ERR_COLON,
5451
2.06k
         "colons are forbidden from entities names '%s'\n",
5452
2.06k
         name, NULL, NULL);
5453
2.06k
  }
5454
136k
  if (SKIP_BLANKS_PE == 0) {
5455
3.90k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5456
3.90k
         "Space required after the entity name\n");
5457
3.90k
  }
5458
5459
  /*
5460
   * handle the various case of definitions...
5461
   */
5462
136k
  if (isParameter) {
5463
68.6k
      if ((RAW == '"') || (RAW == '\'')) {
5464
59.7k
          value = xmlParseEntityValue(ctxt, &orig);
5465
59.7k
    if (value) {
5466
58.5k
        if ((ctxt->sax != NULL) &&
5467
58.5k
      (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5468
52.5k
      ctxt->sax->entityDecl(ctxt->userData, name,
5469
52.5k
                        XML_INTERNAL_PARAMETER_ENTITY,
5470
52.5k
            NULL, NULL, value);
5471
58.5k
    }
5472
59.7k
      } else {
5473
8.95k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5474
8.95k
    if ((URI == NULL) && (literal == NULL)) {
5475
1.52k
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5476
1.52k
    }
5477
8.95k
    if (URI) {
5478
7.21k
                    if (xmlStrchr(URI, '#')) {
5479
647
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5480
6.56k
                    } else {
5481
6.56k
                        if ((ctxt->sax != NULL) &&
5482
6.56k
                            (!ctxt->disableSAX) &&
5483
5.53k
                            (ctxt->sax->entityDecl != NULL))
5484
5.53k
                            ctxt->sax->entityDecl(ctxt->userData, name,
5485
5.53k
                                        XML_EXTERNAL_PARAMETER_ENTITY,
5486
5.53k
                                        literal, URI, NULL);
5487
6.56k
                    }
5488
7.21k
    }
5489
8.95k
      }
5490
68.6k
  } else {
5491
68.2k
      if ((RAW == '"') || (RAW == '\'')) {
5492
51.2k
          value = xmlParseEntityValue(ctxt, &orig);
5493
51.2k
    if ((ctxt->sax != NULL) &&
5494
51.2k
        (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5495
49.7k
        ctxt->sax->entityDecl(ctxt->userData, name,
5496
49.7k
        XML_INTERNAL_GENERAL_ENTITY,
5497
49.7k
        NULL, NULL, value);
5498
    /*
5499
     * For expat compatibility in SAX mode.
5500
     */
5501
51.2k
    if ((ctxt->myDoc == NULL) ||
5502
51.2k
        (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) {
5503
113
        if (ctxt->myDoc == NULL) {
5504
17
      ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5505
17
      if (ctxt->myDoc == NULL) {
5506
1
          xmlErrMemory(ctxt);
5507
1
          goto done;
5508
1
      }
5509
16
      ctxt->myDoc->properties = XML_DOC_INTERNAL;
5510
16
        }
5511
112
        if (ctxt->myDoc->intSubset == NULL) {
5512
16
      ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5513
16
              BAD_CAST "fake", NULL, NULL);
5514
16
                        if (ctxt->myDoc->intSubset == NULL) {
5515
0
                            xmlErrMemory(ctxt);
5516
0
                            goto done;
5517
0
                        }
5518
16
                    }
5519
5520
112
        xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY,
5521
112
                    NULL, NULL, value);
5522
112
    }
5523
51.2k
      } else {
5524
17.0k
          URI = xmlParseExternalID(ctxt, &literal, 1);
5525
17.0k
    if ((URI == NULL) && (literal == NULL)) {
5526
2.90k
        xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL);
5527
2.90k
    }
5528
17.0k
    if (URI) {
5529
13.8k
                    if (xmlStrchr(URI, '#')) {
5530
249
                        xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL);
5531
249
                    }
5532
13.8k
    }
5533
17.0k
    if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) {
5534
2.92k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5535
2.92k
           "Space required before 'NDATA'\n");
5536
2.92k
    }
5537
17.0k
    if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) {
5538
818
        SKIP(5);
5539
818
        if (SKIP_BLANKS_PE == 0) {
5540
348
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5541
348
               "Space required after 'NDATA'\n");
5542
348
        }
5543
818
        ndata = xmlParseName(ctxt);
5544
818
        if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
5545
546
            (ctxt->sax->unparsedEntityDecl != NULL))
5546
546
      ctxt->sax->unparsedEntityDecl(ctxt->userData, name,
5547
546
            literal, URI, ndata);
5548
16.2k
    } else {
5549
16.2k
        if ((ctxt->sax != NULL) &&
5550
16.2k
            (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL))
5551
15.0k
      ctxt->sax->entityDecl(ctxt->userData, name,
5552
15.0k
            XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5553
15.0k
            literal, URI, NULL);
5554
        /*
5555
         * For expat compatibility in SAX mode.
5556
         * assuming the entity replacement was asked for
5557
         */
5558
16.2k
        if ((ctxt->replaceEntities != 0) &&
5559
6.35k
      ((ctxt->myDoc == NULL) ||
5560
6.34k
      (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) {
5561
224
      if (ctxt->myDoc == NULL) {
5562
4
          ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE);
5563
4
          if (ctxt->myDoc == NULL) {
5564
1
              xmlErrMemory(ctxt);
5565
1
        goto done;
5566
1
          }
5567
3
          ctxt->myDoc->properties = XML_DOC_INTERNAL;
5568
3
      }
5569
5570
223
      if (ctxt->myDoc->intSubset == NULL) {
5571
3
          ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc,
5572
3
            BAD_CAST "fake", NULL, NULL);
5573
3
                            if (ctxt->myDoc->intSubset == NULL) {
5574
1
                                xmlErrMemory(ctxt);
5575
1
                                goto done;
5576
1
                            }
5577
3
                        }
5578
222
      xmlSAX2EntityDecl(ctxt, name,
5579
222
                  XML_EXTERNAL_GENERAL_PARSED_ENTITY,
5580
222
                  literal, URI, NULL);
5581
222
        }
5582
16.2k
    }
5583
17.0k
      }
5584
68.2k
  }
5585
136k
  SKIP_BLANKS_PE;
5586
136k
  if (RAW != '>') {
5587
9.33k
      xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED,
5588
9.33k
              "xmlParseEntityDecl: entity %s not terminated\n", name);
5589
127k
  } else {
5590
127k
#ifdef LIBXML_VALID_ENABLED
5591
127k
      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
127k
#endif
5598
127k
      NEXT;
5599
127k
  }
5600
136k
  if (orig != NULL) {
5601
      /*
5602
       * Ugly mechanism to save the raw entity value.
5603
       */
5604
108k
      xmlEntityPtr cur = NULL;
5605
5606
108k
      if (isParameter) {
5607
58.5k
          if ((ctxt->sax != NULL) &&
5608
58.5k
        (ctxt->sax->getParameterEntity != NULL))
5609
58.5k
        cur = ctxt->sax->getParameterEntity(ctxt->userData, name);
5610
58.5k
      } else {
5611
50.1k
          if ((ctxt->sax != NULL) &&
5612
50.1k
        (ctxt->sax->getEntity != NULL))
5613
50.1k
        cur = ctxt->sax->getEntity(ctxt->userData, name);
5614
50.1k
    if ((cur == NULL) && (ctxt->userData==ctxt)) {
5615
1.24k
        cur = xmlSAX2GetEntity(ctxt, name);
5616
1.24k
    }
5617
50.1k
      }
5618
108k
            if ((cur != NULL) && (cur->orig == NULL)) {
5619
11.3k
    cur->orig = orig;
5620
11.3k
                orig = NULL;
5621
11.3k
      }
5622
108k
  }
5623
5624
136k
done:
5625
136k
  if (value != NULL) xmlFree(value);
5626
136k
  if (URI != NULL) xmlFree(URI);
5627
136k
  if (literal != NULL) xmlFree(literal);
5628
136k
        if (orig != NULL) xmlFree(orig);
5629
136k
    }
5630
168k
}
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
94.1k
xmlParseDefaultDecl(xmlParserCtxt *ctxt, xmlChar **value) {
5663
94.1k
    int val;
5664
94.1k
    xmlChar *ret;
5665
5666
94.1k
    *value = NULL;
5667
94.1k
    if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) {
5668
2.81k
  SKIP(9);
5669
2.81k
  return(XML_ATTRIBUTE_REQUIRED);
5670
2.81k
    }
5671
91.2k
    if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) {
5672
21.2k
  SKIP(8);
5673
21.2k
  return(XML_ATTRIBUTE_IMPLIED);
5674
21.2k
    }
5675
70.0k
    val = XML_ATTRIBUTE_NONE;
5676
70.0k
    if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) {
5677
12.8k
  SKIP(6);
5678
12.8k
  val = XML_ATTRIBUTE_FIXED;
5679
12.8k
  if (SKIP_BLANKS_PE == 0) {
5680
246
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5681
246
         "Space required after '#FIXED'\n");
5682
246
  }
5683
12.8k
    }
5684
70.0k
    ret = xmlParseAttValue(ctxt);
5685
70.0k
    if (ret == NULL) {
5686
39.7k
  xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo,
5687
39.7k
           "Attribute default value declaration error\n");
5688
39.7k
    } else
5689
30.3k
        *value = ret;
5690
70.0k
    return(val);
5691
91.2k
}
5692
5693
/**
5694
 * Parse an Notation attribute type.
5695
 *
5696
 * @deprecated Internal function, don't use.
5697
 *
5698
 * Note: the leading 'NOTATION' S part has already being parsed...
5699
 *
5700
 *     [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')'
5701
 *
5702
 * [ VC: Notation Attributes ]
5703
 * Values of this type must match one of the notation names included
5704
 * in the declaration; all notation names in the declaration must be declared.
5705
 *
5706
 * @param ctxt  an XML parser context
5707
 * @returns the notation attribute tree built while parsing
5708
 */
5709
5710
xmlEnumeration *
5711
2.21k
xmlParseNotationType(xmlParserCtxt *ctxt) {
5712
2.21k
    const xmlChar *name;
5713
2.21k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5714
5715
2.21k
    if (RAW != '(') {
5716
217
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL);
5717
217
  return(NULL);
5718
217
    }
5719
2.85k
    do {
5720
2.85k
        NEXT;
5721
2.85k
  SKIP_BLANKS_PE;
5722
2.85k
        name = xmlParseName(ctxt);
5723
2.85k
  if (name == NULL) {
5724
214
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5725
214
         "Name expected in NOTATION declaration\n");
5726
214
            xmlFreeEnumeration(ret);
5727
214
      return(NULL);
5728
214
  }
5729
2.63k
        tmp = NULL;
5730
2.63k
#ifdef LIBXML_VALID_ENABLED
5731
2.63k
        if (ctxt->validate) {
5732
2.00k
            tmp = ret;
5733
6.65k
            while (tmp != NULL) {
5734
4.93k
                if (xmlStrEqual(name, tmp->name)) {
5735
296
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5736
296
              "standalone: attribute notation value token %s duplicated\n",
5737
296
                                     name, NULL);
5738
296
                    if (!xmlDictOwns(ctxt->dict, name))
5739
0
                        xmlFree((xmlChar *) name);
5740
296
                    break;
5741
296
                }
5742
4.64k
                tmp = tmp->next;
5743
4.64k
            }
5744
2.00k
        }
5745
2.63k
#endif /* LIBXML_VALID_ENABLED */
5746
2.63k
  if (tmp == NULL) {
5747
2.34k
      cur = xmlCreateEnumeration(name);
5748
2.34k
      if (cur == NULL) {
5749
1
                xmlErrMemory(ctxt);
5750
1
                xmlFreeEnumeration(ret);
5751
1
                return(NULL);
5752
1
            }
5753
2.33k
      if (last == NULL) ret = last = cur;
5754
554
      else {
5755
554
    last->next = cur;
5756
554
    last = cur;
5757
554
      }
5758
2.33k
  }
5759
2.63k
  SKIP_BLANKS_PE;
5760
2.63k
    } while (RAW == '|');
5761
1.78k
    if (RAW != ')') {
5762
441
  xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL);
5763
441
        xmlFreeEnumeration(ret);
5764
441
  return(NULL);
5765
441
    }
5766
1.34k
    NEXT;
5767
1.34k
    return(ret);
5768
1.78k
}
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
14.7k
xmlParseEnumerationType(xmlParserCtxt *ctxt) {
5787
14.7k
    xmlChar *name;
5788
14.7k
    xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp;
5789
5790
14.7k
    if (RAW != '(') {
5791
1.52k
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL);
5792
1.52k
  return(NULL);
5793
1.52k
    }
5794
20.7k
    do {
5795
20.7k
        NEXT;
5796
20.7k
  SKIP_BLANKS_PE;
5797
20.7k
        name = xmlParseNmtoken(ctxt);
5798
20.7k
  if (name == NULL) {
5799
234
      xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL);
5800
234
      return(ret);
5801
234
  }
5802
20.5k
        tmp = NULL;
5803
20.5k
#ifdef LIBXML_VALID_ENABLED
5804
20.5k
        if (ctxt->validate) {
5805
13.6k
            tmp = ret;
5806
31.9k
            while (tmp != NULL) {
5807
18.5k
                if (xmlStrEqual(name, tmp->name)) {
5808
205
                    xmlValidityError(ctxt, XML_DTD_DUP_TOKEN,
5809
205
              "standalone: attribute enumeration value token %s duplicated\n",
5810
205
                                     name, NULL);
5811
205
                    if (!xmlDictOwns(ctxt->dict, name))
5812
205
                        xmlFree(name);
5813
205
                    break;
5814
205
                }
5815
18.3k
                tmp = tmp->next;
5816
18.3k
            }
5817
13.6k
        }
5818
20.5k
#endif /* LIBXML_VALID_ENABLED */
5819
20.5k
  if (tmp == NULL) {
5820
20.3k
      cur = xmlCreateEnumeration(name);
5821
20.3k
      if (!xmlDictOwns(ctxt->dict, name))
5822
20.3k
    xmlFree(name);
5823
20.3k
      if (cur == NULL) {
5824
4
                xmlErrMemory(ctxt);
5825
4
                xmlFreeEnumeration(ret);
5826
4
                return(NULL);
5827
4
            }
5828
20.3k
      if (last == NULL) ret = last = cur;
5829
7.29k
      else {
5830
7.29k
    last->next = cur;
5831
7.29k
    last = cur;
5832
7.29k
      }
5833
20.3k
  }
5834
20.5k
  SKIP_BLANKS_PE;
5835
20.5k
    } while (RAW == '|');
5836
13.0k
    if (RAW != ')') {
5837
677
  xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL);
5838
677
  return(ret);
5839
677
    }
5840
12.3k
    NEXT;
5841
12.3k
    return(ret);
5842
13.0k
}
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
17.2k
xmlParseEnumeratedType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5860
17.2k
    if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) {
5861
2.50k
  SKIP(8);
5862
2.50k
  if (SKIP_BLANKS_PE == 0) {
5863
288
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5864
288
         "Space required after 'NOTATION'\n");
5865
288
      return(0);
5866
288
  }
5867
2.21k
  *tree = xmlParseNotationType(ctxt);
5868
2.21k
  if (*tree == NULL) return(0);
5869
1.34k
  return(XML_ATTRIBUTE_NOTATION);
5870
2.21k
    }
5871
14.7k
    *tree = xmlParseEnumerationType(ctxt);
5872
14.7k
    if (*tree == NULL) return(0);
5873
13.0k
    return(XML_ATTRIBUTE_ENUMERATION);
5874
14.7k
}
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
99.6k
xmlParseAttributeType(xmlParserCtxt *ctxt, xmlEnumeration **tree) {
5923
99.6k
    if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) {
5924
14.6k
  SKIP(5);
5925
14.6k
  return(XML_ATTRIBUTE_CDATA);
5926
84.9k
     } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) {
5927
2.87k
  SKIP(6);
5928
2.87k
  return(XML_ATTRIBUTE_IDREFS);
5929
82.0k
     } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) {
5930
1.72k
  SKIP(5);
5931
1.72k
  return(XML_ATTRIBUTE_IDREF);
5932
80.3k
     } else if ((RAW == 'I') && (NXT(1) == 'D')) {
5933
11.1k
        SKIP(2);
5934
11.1k
  return(XML_ATTRIBUTE_ID);
5935
69.1k
     } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) {
5936
4.84k
  SKIP(6);
5937
4.84k
  return(XML_ATTRIBUTE_ENTITY);
5938
64.3k
     } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) {
5939
1.19k
  SKIP(8);
5940
1.19k
  return(XML_ATTRIBUTE_ENTITIES);
5941
63.1k
     } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) {
5942
37.2k
  SKIP(8);
5943
37.2k
  return(XML_ATTRIBUTE_NMTOKENS);
5944
37.2k
     } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) {
5945
8.57k
  SKIP(7);
5946
8.57k
  return(XML_ATTRIBUTE_NMTOKEN);
5947
8.57k
     }
5948
17.2k
     return(xmlParseEnumeratedType(ctxt, tree));
5949
99.6k
}
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
88.8k
xmlParseAttributeListDecl(xmlParserCtxt *ctxt) {
5963
88.8k
    const xmlChar *elemName;
5964
88.8k
    const xmlChar *attrName;
5965
88.8k
    xmlEnumerationPtr tree;
5966
5967
88.8k
    if ((CUR != '<') || (NXT(1) != '!'))
5968
0
        return;
5969
88.8k
    SKIP(2);
5970
5971
88.8k
    if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) {
5972
88.8k
#ifdef LIBXML_VALID_ENABLED
5973
88.8k
  int oldInputNr = ctxt->inputNr;
5974
88.8k
#endif
5975
5976
88.8k
  SKIP(7);
5977
88.8k
  if (SKIP_BLANKS_PE == 0) {
5978
2.07k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
5979
2.07k
                     "Space required after '<!ATTLIST'\n");
5980
2.07k
  }
5981
88.8k
        elemName = xmlParseName(ctxt);
5982
88.8k
  if (elemName == NULL) {
5983
2.34k
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5984
2.34k
         "ATTLIST: no name for Element\n");
5985
2.34k
      return;
5986
2.34k
  }
5987
86.4k
  SKIP_BLANKS_PE;
5988
86.4k
  GROW;
5989
140k
  while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) {
5990
110k
      int type;
5991
110k
      int def;
5992
110k
      xmlChar *defaultValue = NULL;
5993
5994
110k
      GROW;
5995
110k
            tree = NULL;
5996
110k
      attrName = xmlParseName(ctxt);
5997
110k
      if (attrName == NULL) {
5998
7.70k
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
5999
7.70k
             "ATTLIST: no name for Attribute\n");
6000
7.70k
    break;
6001
7.70k
      }
6002
102k
      GROW;
6003
102k
      if (SKIP_BLANKS_PE == 0) {
6004
3.01k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6005
3.01k
            "Space required after the attribute name\n");
6006
3.01k
    break;
6007
3.01k
      }
6008
6009
99.6k
      type = xmlParseAttributeType(ctxt, &tree);
6010
99.6k
      if (type <= 0) {
6011
2.91k
          break;
6012
2.91k
      }
6013
6014
96.7k
      GROW;
6015
96.7k
      if (SKIP_BLANKS_PE == 0) {
6016
2.61k
    xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6017
2.61k
             "Space required after the attribute type\n");
6018
2.61k
          if (tree != NULL)
6019
753
        xmlFreeEnumeration(tree);
6020
2.61k
    break;
6021
2.61k
      }
6022
6023
94.1k
      def = xmlParseDefaultDecl(ctxt, &defaultValue);
6024
94.1k
      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
94.1k
      if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL))
6032
27.2k
          xmlAttrNormalizeSpace(defaultValue, defaultValue);
6033
6034
94.1k
      GROW;
6035
94.1k
            if (RAW != '>') {
6036
71.6k
    if (SKIP_BLANKS_PE == 0) {
6037
40.1k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6038
40.1k
      "Space required after the attribute default value\n");
6039
40.1k
        if (defaultValue != NULL)
6040
300
      xmlFree(defaultValue);
6041
40.1k
        if (tree != NULL)
6042
1.15k
      xmlFreeEnumeration(tree);
6043
40.1k
        break;
6044
40.1k
    }
6045
71.6k
      }
6046
53.9k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6047
51.5k
    (ctxt->sax->attributeDecl != NULL))
6048
51.5k
    ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName,
6049
51.5k
                          type, def, defaultValue, tree);
6050
2.40k
      else if (tree != NULL)
6051
1.15k
    xmlFreeEnumeration(tree);
6052
6053
53.9k
      if ((ctxt->sax2) && (defaultValue != NULL) &&
6054
23.3k
          (def != XML_ATTRIBUTE_IMPLIED) &&
6055
23.3k
    (def != XML_ATTRIBUTE_REQUIRED)) {
6056
23.3k
    xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue);
6057
23.3k
      }
6058
53.9k
      if (ctxt->sax2) {
6059
41.3k
    xmlAddSpecialAttr(ctxt, elemName, attrName, type);
6060
41.3k
      }
6061
53.9k
      if (defaultValue != NULL)
6062
30.0k
          xmlFree(defaultValue);
6063
53.9k
      GROW;
6064
53.9k
  }
6065
86.4k
  if (RAW == '>') {
6066
30.9k
#ifdef LIBXML_VALID_ENABLED
6067
30.9k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6068
86
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6069
86
                                 "Attribute list declaration doesn't start and"
6070
86
                                 " stop in the same entity\n",
6071
86
                                 NULL, NULL);
6072
86
      }
6073
30.9k
#endif
6074
30.9k
      NEXT;
6075
30.9k
  }
6076
86.4k
    }
6077
88.8k
}
6078
6079
/**
6080
 * Handle PEs and check that we don't pop the entity that started
6081
 * a balanced group.
6082
 *
6083
 * @param ctxt  parser context
6084
 * @param openInputNr  input nr of the entity with opening '('
6085
 */
6086
static void
6087
725k
xmlSkipBlankCharsPEBalanced(xmlParserCtxt *ctxt, int openInputNr) {
6088
725k
    SKIP_BLANKS;
6089
725k
    GROW;
6090
6091
725k
    (void) openInputNr;
6092
6093
725k
    if (!PARSER_EXTERNAL(ctxt) && !PARSER_IN_PE(ctxt))
6094
599k
        return;
6095
6096
134k
    while (!PARSER_STOPPED(ctxt)) {
6097
133k
        if (ctxt->input->cur >= ctxt->input->end) {
6098
2.93k
#ifdef LIBXML_VALID_ENABLED
6099
2.93k
            if ((ctxt->validate) && (ctxt->inputNr <= openInputNr)) {
6100
852
                xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6101
852
                                 "Element content declaration doesn't start "
6102
852
                                 "and stop in the same entity\n",
6103
852
                                 NULL, NULL);
6104
852
            }
6105
2.93k
#endif
6106
2.93k
            if (PARSER_IN_PE(ctxt))
6107
2.91k
                xmlPopPE(ctxt);
6108
22
            else
6109
22
                break;
6110
130k
        } else if (RAW == '%') {
6111
5.66k
            xmlParsePERefInternal(ctxt, 0);
6112
125k
        } else {
6113
125k
            break;
6114
125k
        }
6115
6116
8.58k
        SKIP_BLANKS;
6117
8.58k
        GROW;
6118
8.58k
    }
6119
125k
}
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
8.08k
xmlParseElementMixedContentDecl(xmlParserCtxt *ctxt, int openInputNr) {
6143
8.08k
    xmlElementContentPtr ret = NULL, cur = NULL, n;
6144
8.08k
    const xmlChar *elem = NULL;
6145
6146
8.08k
    GROW;
6147
8.08k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6148
8.08k
  SKIP(7);
6149
8.08k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6150
8.08k
  if (RAW == ')') {
6151
4.08k
#ifdef LIBXML_VALID_ENABLED
6152
4.08k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6153
1
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6154
1
                                 "Element content declaration doesn't start "
6155
1
                                 "and stop in the same entity\n",
6156
1
                                 NULL, NULL);
6157
1
      }
6158
4.08k
#endif
6159
4.08k
      NEXT;
6160
4.08k
      ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6161
4.08k
      if (ret == NULL)
6162
2
                goto mem_error;
6163
4.08k
      if (RAW == '*') {
6164
436
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6165
436
    NEXT;
6166
436
      }
6167
4.08k
      return(ret);
6168
4.08k
  }
6169
4.00k
  if ((RAW == '(') || (RAW == '|')) {
6170
2.93k
      ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA);
6171
2.93k
      if (ret == NULL)
6172
1
                goto mem_error;
6173
2.93k
  }
6174
23.9k
  while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) {
6175
20.0k
      NEXT;
6176
20.0k
            n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6177
20.0k
            if (n == NULL)
6178
2
                goto mem_error;
6179
20.0k
      if (elem == NULL) {
6180
2.93k
    n->c1 = cur;
6181
2.93k
    if (cur != NULL)
6182
2.93k
        cur->parent = n;
6183
2.93k
    ret = cur = n;
6184
17.0k
      } else {
6185
17.0k
          cur->c2 = n;
6186
17.0k
    n->parent = cur;
6187
17.0k
    n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6188
17.0k
                if (n->c1 == NULL)
6189
1
                    goto mem_error;
6190
17.0k
    n->c1->parent = n;
6191
17.0k
    cur = n;
6192
17.0k
      }
6193
20.0k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6194
20.0k
      elem = xmlParseName(ctxt);
6195
20.0k
      if (elem == NULL) {
6196
86
    xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6197
86
      "xmlParseElementMixedContentDecl : Name expected\n");
6198
86
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6199
86
    return(NULL);
6200
86
      }
6201
19.9k
            xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6202
19.9k
  }
6203
3.91k
  if ((RAW == ')') && (NXT(1) == '*')) {
6204
2.53k
      if (elem != NULL) {
6205
2.53k
    cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem,
6206
2.53k
                                   XML_ELEMENT_CONTENT_ELEMENT);
6207
2.53k
    if (cur->c2 == NULL)
6208
1
                    goto mem_error;
6209
2.53k
    cur->c2->parent = cur;
6210
2.53k
            }
6211
2.53k
            if (ret != NULL)
6212
2.53k
                ret->ocur = XML_ELEMENT_CONTENT_MULT;
6213
2.53k
#ifdef LIBXML_VALID_ENABLED
6214
2.53k
      if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6215
11
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6216
11
                                 "Element content declaration doesn't start "
6217
11
                                 "and stop in the same entity\n",
6218
11
                                 NULL, NULL);
6219
11
      }
6220
2.53k
#endif
6221
2.53k
      SKIP(2);
6222
2.53k
  } else {
6223
1.37k
      xmlFreeDocElementContent(ctxt->myDoc, ret);
6224
1.37k
      xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL);
6225
1.37k
      return(NULL);
6226
1.37k
  }
6227
6228
3.91k
    } else {
6229
0
  xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL);
6230
0
    }
6231
2.53k
    return(ret);
6232
6233
7
mem_error:
6234
7
    xmlErrMemory(ctxt);
6235
7
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6236
7
    return(NULL);
6237
8.08k
}
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
91.1k
                                       int depth) {
6272
91.1k
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
6273
91.1k
    xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL;
6274
91.1k
    const xmlChar *elem;
6275
91.1k
    xmlChar type = 0;
6276
6277
91.1k
    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
91.1k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6284
91.1k
    if (RAW == '(') {
6285
26.3k
        int newInputNr = ctxt->inputNr;
6286
6287
        /* Recurse on first child */
6288
26.3k
  NEXT;
6289
26.3k
        cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6290
26.3k
                                                           depth + 1);
6291
26.3k
        if (cur == NULL)
6292
25.0k
            return(NULL);
6293
64.7k
    } else {
6294
64.7k
  elem = xmlParseName(ctxt);
6295
64.7k
  if (elem == NULL) {
6296
401
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6297
401
      return(NULL);
6298
401
  }
6299
64.3k
        cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6300
64.3k
  if (cur == NULL) {
6301
5
      xmlErrMemory(ctxt);
6302
5
      return(NULL);
6303
5
  }
6304
64.3k
  GROW;
6305
64.3k
  if (RAW == '?') {
6306
951
      cur->ocur = XML_ELEMENT_CONTENT_OPT;
6307
951
      NEXT;
6308
63.4k
  } else if (RAW == '*') {
6309
1.14k
      cur->ocur = XML_ELEMENT_CONTENT_MULT;
6310
1.14k
      NEXT;
6311
62.2k
  } else if (RAW == '+') {
6312
744
      cur->ocur = XML_ELEMENT_CONTENT_PLUS;
6313
744
      NEXT;
6314
61.5k
  } else {
6315
61.5k
      cur->ocur = XML_ELEMENT_CONTENT_ONCE;
6316
61.5k
  }
6317
64.3k
  GROW;
6318
64.3k
    }
6319
312k
    while (!PARSER_STOPPED(ctxt)) {
6320
311k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6321
311k
        if (RAW == ')')
6322
62.1k
            break;
6323
        /*
6324
   * Each loop we parse one separator and one element.
6325
   */
6326
249k
        if (RAW == ',') {
6327
166k
      if (type == 0) type = CUR;
6328
6329
      /*
6330
       * Detect "Name | Name , Name" error
6331
       */
6332
154k
      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
166k
      NEXT;
6343
6344
166k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ);
6345
166k
      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
166k
      if (last == NULL) {
6353
11.8k
    op->c1 = ret;
6354
11.8k
    if (ret != NULL)
6355
11.8k
        ret->parent = op;
6356
11.8k
    ret = cur = op;
6357
154k
      } else {
6358
154k
          cur->c2 = op;
6359
154k
    if (op != NULL)
6360
154k
        op->parent = cur;
6361
154k
    op->c1 = last;
6362
154k
    if (last != NULL)
6363
154k
        last->parent = op;
6364
154k
    cur =op;
6365
154k
    last = NULL;
6366
154k
      }
6367
166k
  } else if (RAW == '|') {
6368
81.4k
      if (type == 0) type = CUR;
6369
6370
      /*
6371
       * Detect "Name , Name | Name" error
6372
       */
6373
47.7k
      else if (type != CUR) {
6374
2
    xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED,
6375
2
        "xmlParseElementChildrenContentDecl : '%c' expected\n",
6376
2
          type);
6377
2
    if ((last != NULL) && (last != ret))
6378
2
        xmlFreeDocElementContent(ctxt->myDoc, last);
6379
2
    if (ret != NULL)
6380
2
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6381
2
    return(NULL);
6382
2
      }
6383
81.4k
      NEXT;
6384
6385
81.4k
      op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR);
6386
81.4k
      if (op == NULL) {
6387
5
                xmlErrMemory(ctxt);
6388
5
    if ((last != NULL) && (last != ret))
6389
3
        xmlFreeDocElementContent(ctxt->myDoc, last);
6390
5
    if (ret != NULL)
6391
5
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6392
5
    return(NULL);
6393
5
      }
6394
81.4k
      if (last == NULL) {
6395
33.6k
    op->c1 = ret;
6396
33.6k
    if (ret != NULL)
6397
33.6k
        ret->parent = op;
6398
33.6k
    ret = cur = op;
6399
47.7k
      } else {
6400
47.7k
          cur->c2 = op;
6401
47.7k
    if (op != NULL)
6402
47.7k
        op->parent = cur;
6403
47.7k
    op->c1 = last;
6404
47.7k
    if (last != NULL)
6405
47.7k
        last->parent = op;
6406
47.7k
    cur =op;
6407
47.7k
    last = NULL;
6408
47.7k
      }
6409
81.4k
  } else {
6410
1.70k
      xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL);
6411
1.70k
      if ((last != NULL) && (last != ret))
6412
794
          xmlFreeDocElementContent(ctxt->myDoc, last);
6413
1.70k
      if (ret != NULL)
6414
1.70k
    xmlFreeDocElementContent(ctxt->myDoc, ret);
6415
1.70k
      return(NULL);
6416
1.70k
  }
6417
247k
        xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6418
247k
        if (RAW == '(') {
6419
45.7k
            int newInputNr = ctxt->inputNr;
6420
6421
      /* Recurse on second child */
6422
45.7k
      NEXT;
6423
45.7k
      last = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr,
6424
45.7k
                                                          depth + 1);
6425
45.7k
            if (last == NULL) {
6426
436
    if (ret != NULL)
6427
436
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6428
436
    return(NULL);
6429
436
            }
6430
201k
  } else {
6431
201k
      elem = xmlParseName(ctxt);
6432
201k
      if (elem == NULL) {
6433
338
    xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL);
6434
338
    if (ret != NULL)
6435
338
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6436
338
    return(NULL);
6437
338
      }
6438
201k
      last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT);
6439
201k
      if (last == NULL) {
6440
2
                xmlErrMemory(ctxt);
6441
2
    if (ret != NULL)
6442
2
        xmlFreeDocElementContent(ctxt->myDoc, ret);
6443
2
    return(NULL);
6444
2
      }
6445
201k
      if (RAW == '?') {
6446
24.4k
    last->ocur = XML_ELEMENT_CONTENT_OPT;
6447
24.4k
    NEXT;
6448
177k
      } else if (RAW == '*') {
6449
11.7k
    last->ocur = XML_ELEMENT_CONTENT_MULT;
6450
11.7k
    NEXT;
6451
165k
      } else if (RAW == '+') {
6452
22.3k
    last->ocur = XML_ELEMENT_CONTENT_PLUS;
6453
22.3k
    NEXT;
6454
143k
      } else {
6455
143k
    last->ocur = XML_ELEMENT_CONTENT_ONCE;
6456
143k
      }
6457
201k
  }
6458
247k
    }
6459
63.2k
    if ((cur != NULL) && (last != NULL)) {
6460
43.9k
        cur->c2 = last;
6461
43.9k
  if (last != NULL)
6462
43.9k
      last->parent = cur;
6463
43.9k
    }
6464
63.2k
#ifdef LIBXML_VALID_ENABLED
6465
63.2k
    if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) {
6466
236
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6467
236
                         "Element content declaration doesn't start "
6468
236
                         "and stop in the same entity\n",
6469
236
                         NULL, NULL);
6470
236
    }
6471
63.2k
#endif
6472
63.2k
    NEXT;
6473
63.2k
    if (RAW == '?') {
6474
9.69k
  if (ret != NULL) {
6475
9.69k
      if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) ||
6476
9.38k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6477
349
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6478
9.34k
      else
6479
9.34k
          ret->ocur = XML_ELEMENT_CONTENT_OPT;
6480
9.69k
  }
6481
9.69k
  NEXT;
6482
53.5k
    } else if (RAW == '*') {
6483
25.1k
  if (ret != NULL) {
6484
25.1k
      ret->ocur = XML_ELEMENT_CONTENT_MULT;
6485
25.1k
      cur = ret;
6486
      /*
6487
       * Some normalization:
6488
       * (a | b* | c?)* == (a | b | c)*
6489
       */
6490
83.8k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6491
58.6k
    if ((cur->c1 != NULL) &&
6492
58.6k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6493
57.6k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT)))
6494
2.65k
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6495
58.6k
    if ((cur->c2 != NULL) &&
6496
58.6k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6497
58.3k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT)))
6498
513
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6499
58.6k
    cur = cur->c2;
6500
58.6k
      }
6501
25.1k
  }
6502
25.1k
  NEXT;
6503
28.3k
    } else if (RAW == '+') {
6504
8.51k
  if (ret != NULL) {
6505
8.51k
      int found = 0;
6506
6507
8.51k
      if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) ||
6508
8.45k
          (ret->ocur == XML_ELEMENT_CONTENT_MULT))
6509
134
          ret->ocur = XML_ELEMENT_CONTENT_MULT;
6510
8.37k
      else
6511
8.37k
          ret->ocur = XML_ELEMENT_CONTENT_PLUS;
6512
      /*
6513
       * Some normalization:
6514
       * (a | b*)+ == (a | b)*
6515
       * (a | b?)+ == (a | b)*
6516
       */
6517
16.5k
      while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) {
6518
8.06k
    if ((cur->c1 != NULL) &&
6519
8.06k
              ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) ||
6520
7.82k
         (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) {
6521
350
        cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE;
6522
350
        found = 1;
6523
350
    }
6524
8.06k
    if ((cur->c2 != NULL) &&
6525
8.06k
              ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) ||
6526
7.66k
         (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) {
6527
653
        cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE;
6528
653
        found = 1;
6529
653
    }
6530
8.06k
    cur = cur->c2;
6531
8.06k
      }
6532
8.51k
      if (found)
6533
757
    ret->ocur = XML_ELEMENT_CONTENT_MULT;
6534
8.51k
  }
6535
8.51k
  NEXT;
6536
8.51k
    }
6537
63.2k
    return(ret);
6538
65.6k
}
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
27.0k
                           xmlElementContent **result) {
6593
6594
27.0k
    xmlElementContentPtr tree = NULL;
6595
27.0k
    int openInputNr = ctxt->inputNr;
6596
27.0k
    int res;
6597
6598
27.0k
    *result = NULL;
6599
6600
27.0k
    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
27.0k
    NEXT;
6606
27.0k
    xmlSkipBlankCharsPEBalanced(ctxt, openInputNr);
6607
27.0k
    if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) {
6608
8.08k
        tree = xmlParseElementMixedContentDecl(ctxt, openInputNr);
6609
8.08k
  res = XML_ELEMENT_TYPE_MIXED;
6610
19.0k
    } else {
6611
19.0k
        tree = xmlParseElementChildrenContentDeclPriv(ctxt, openInputNr, 1);
6612
19.0k
  res = XML_ELEMENT_TYPE_ELEMENT;
6613
19.0k
    }
6614
27.0k
    if (tree == NULL)
6615
3.92k
        return(-1);
6616
23.1k
    SKIP_BLANKS_PE;
6617
23.1k
    *result = tree;
6618
23.1k
    return(res);
6619
27.0k
}
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
37.4k
xmlParseElementDecl(xmlParserCtxt *ctxt) {
6636
37.4k
    const xmlChar *name;
6637
37.4k
    int ret = -1;
6638
37.4k
    xmlElementContentPtr content  = NULL;
6639
6640
37.4k
    if ((CUR != '<') || (NXT(1) != '!'))
6641
0
        return(ret);
6642
37.4k
    SKIP(2);
6643
6644
    /* GROW; done in the caller */
6645
37.4k
    if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) {
6646
37.3k
#ifdef LIBXML_VALID_ENABLED
6647
37.3k
  int oldInputNr = ctxt->inputNr;
6648
37.3k
#endif
6649
6650
37.3k
  SKIP(7);
6651
37.3k
  if (SKIP_BLANKS_PE == 0) {
6652
381
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6653
381
               "Space required after 'ELEMENT'\n");
6654
381
      return(-1);
6655
381
  }
6656
36.9k
        name = xmlParseName(ctxt);
6657
36.9k
  if (name == NULL) {
6658
659
      xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
6659
659
         "xmlParseElementDecl: no name for Element\n");
6660
659
      return(-1);
6661
659
  }
6662
36.3k
  if (SKIP_BLANKS_PE == 0) {
6663
1.61k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6664
1.61k
         "Space required after the element name\n");
6665
1.61k
  }
6666
36.3k
  if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) {
6667
5.55k
      SKIP(5);
6668
      /*
6669
       * Element must always be empty.
6670
       */
6671
5.55k
      ret = XML_ELEMENT_TYPE_EMPTY;
6672
30.7k
  } else if ((RAW == 'A') && (NXT(1) == 'N') &&
6673
2.18k
             (NXT(2) == 'Y')) {
6674
2.18k
      SKIP(3);
6675
      /*
6676
       * Element is a generic container.
6677
       */
6678
2.18k
      ret = XML_ELEMENT_TYPE_ANY;
6679
28.6k
  } else if (RAW == '(') {
6680
27.0k
      ret = xmlParseElementContentDecl(ctxt, name, &content);
6681
27.0k
            if (ret <= 0)
6682
3.92k
                return(-1);
6683
27.0k
  } else {
6684
      /*
6685
       * [ WFC: PEs in Internal Subset ] error handling.
6686
       */
6687
1.50k
            xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED,
6688
1.50k
                  "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n");
6689
1.50k
      return(-1);
6690
1.50k
  }
6691
6692
30.9k
  SKIP_BLANKS_PE;
6693
6694
30.9k
  if (RAW != '>') {
6695
974
      xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
6696
974
      if (content != NULL) {
6697
526
    xmlFreeDocElementContent(ctxt->myDoc, content);
6698
526
      }
6699
29.9k
  } else {
6700
29.9k
#ifdef LIBXML_VALID_ENABLED
6701
29.9k
      if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) {
6702
12
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6703
12
                                 "Element declaration doesn't start and stop in"
6704
12
                                 " the same entity\n",
6705
12
                                 NULL, NULL);
6706
12
      }
6707
29.9k
#endif
6708
6709
29.9k
      NEXT;
6710
29.9k
      if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
6711
27.8k
    (ctxt->sax->elementDecl != NULL)) {
6712
27.8k
    if (content != NULL)
6713
20.8k
        content->parent = NULL;
6714
27.8k
          ctxt->sax->elementDecl(ctxt->userData, name, ret,
6715
27.8k
                           content);
6716
27.8k
    if ((content != NULL) && (content->parent == NULL)) {
6717
        /*
6718
         * this is a trick: if xmlAddElementDecl is called,
6719
         * instead of copying the full tree it is plugged directly
6720
         * if called from the parser. Avoid duplicating the
6721
         * interfaces or change the API/ABI
6722
         */
6723
12.5k
        xmlFreeDocElementContent(ctxt->myDoc, content);
6724
12.5k
    }
6725
27.8k
      } else if (content != NULL) {
6726
1.78k
    xmlFreeDocElementContent(ctxt->myDoc, content);
6727
1.78k
      }
6728
29.9k
  }
6729
30.9k
    }
6730
30.9k
    return(ret);
6731
37.4k
}
6732
6733
/**
6734
 * Parse a conditional section. Always consumes '<!['.
6735
 *
6736
 *     [61] conditionalSect ::= includeSect | ignoreSect
6737
 *     [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>'
6738
 *     [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>'
6739
 *     [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>'
6740
 *                                 Ignore)*
6741
 *     [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*)
6742
 * @param ctxt  an XML parser context
6743
 */
6744
6745
static void
6746
3.71k
xmlParseConditionalSections(xmlParserCtxtPtr ctxt) {
6747
3.71k
    size_t depth = 0;
6748
3.71k
    int isFreshPE = 0;
6749
3.71k
    int oldInputNr = ctxt->inputNr;
6750
3.71k
    int declInputNr = ctxt->inputNr;
6751
6752
6.41k
    while (!PARSER_STOPPED(ctxt)) {
6753
6.41k
        if (ctxt->input->cur >= ctxt->input->end) {
6754
159
            if (ctxt->inputNr <= oldInputNr) {
6755
147
                xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6756
147
                return;
6757
147
            }
6758
6759
12
            xmlPopPE(ctxt);
6760
12
            declInputNr = ctxt->inputNr;
6761
6.25k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6762
3.89k
            SKIP(3);
6763
3.89k
            SKIP_BLANKS_PE;
6764
6765
3.89k
            isFreshPE = 0;
6766
6767
3.89k
            if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) {
6768
1.39k
                SKIP(7);
6769
1.39k
                SKIP_BLANKS_PE;
6770
1.39k
                if (RAW != '[') {
6771
298
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6772
298
                    return;
6773
298
                }
6774
1.10k
#ifdef LIBXML_VALID_ENABLED
6775
1.10k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6776
25
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6777
25
                                     "All markup of the conditional section is"
6778
25
                                     " not in the same entity\n",
6779
25
                                     NULL, NULL);
6780
25
                }
6781
1.10k
#endif
6782
1.10k
                NEXT;
6783
6784
1.10k
                depth++;
6785
2.49k
            } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) {
6786
2.22k
                size_t ignoreDepth = 0;
6787
6788
2.22k
                SKIP(6);
6789
2.22k
                SKIP_BLANKS_PE;
6790
2.22k
                if (RAW != '[') {
6791
236
                    xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL);
6792
236
                    return;
6793
236
                }
6794
1.99k
#ifdef LIBXML_VALID_ENABLED
6795
1.99k
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6796
10
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6797
10
                                     "All markup of the conditional section is"
6798
10
                                     " not in the same entity\n",
6799
10
                                     NULL, NULL);
6800
10
                }
6801
1.99k
#endif
6802
1.99k
                NEXT;
6803
6804
73.0k
                while (PARSER_STOPPED(ctxt) == 0) {
6805
73.0k
                    if (RAW == 0) {
6806
1.36k
                        xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL);
6807
1.36k
                        return;
6808
1.36k
                    }
6809
71.7k
                    if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
6810
740
                        SKIP(3);
6811
740
                        ignoreDepth++;
6812
                        /* Check for integer overflow */
6813
740
                        if (ignoreDepth == 0) {
6814
0
                            xmlErrMemory(ctxt);
6815
0
                            return;
6816
0
                        }
6817
70.9k
                    } else if ((RAW == ']') && (NXT(1) == ']') &&
6818
1.47k
                               (NXT(2) == '>')) {
6819
947
                        SKIP(3);
6820
947
                        if (ignoreDepth == 0)
6821
626
                            break;
6822
321
                        ignoreDepth--;
6823
70.0k
                    } else {
6824
70.0k
                        NEXT;
6825
70.0k
                    }
6826
71.7k
                }
6827
6828
628
#ifdef LIBXML_VALID_ENABLED
6829
628
                if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6830
10
        xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6831
10
                                     "All markup of the conditional section is"
6832
10
                                     " not in the same entity\n",
6833
10
                                     NULL, NULL);
6834
10
                }
6835
628
#endif
6836
628
            } else {
6837
264
                xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL);
6838
264
                return;
6839
264
            }
6840
3.89k
        } else if ((depth > 0) &&
6841
2.36k
                   (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) {
6842
753
            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
752
            depth--;
6850
752
#ifdef LIBXML_VALID_ENABLED
6851
752
            if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) {
6852
12
    xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY,
6853
12
                                 "All markup of the conditional section is not"
6854
12
                                 " in the same entity\n",
6855
12
                                 NULL, NULL);
6856
12
            }
6857
752
#endif
6858
752
            SKIP(3);
6859
1.60k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
6860
1.06k
            isFreshPE = 0;
6861
1.06k
            xmlParseMarkupDecl(ctxt);
6862
1.06k
        } else if (RAW == '%') {
6863
504
            xmlParsePERefInternal(ctxt, 1);
6864
504
            if (ctxt->inputNr > declInputNr) {
6865
10
                isFreshPE = 1;
6866
10
                declInputNr = ctxt->inputNr;
6867
10
            }
6868
504
        } else {
6869
39
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
6870
39
            return;
6871
39
        }
6872
6873
4.06k
        if (depth == 0)
6874
1.36k
            break;
6875
6876
2.70k
        SKIP_BLANKS;
6877
2.70k
        SHRINK;
6878
2.70k
        GROW;
6879
2.70k
    }
6880
3.71k
}
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
643k
xmlParseMarkupDecl(xmlParserCtxt *ctxt) {
6907
643k
    GROW;
6908
643k
    if (CUR == '<') {
6909
643k
        if (NXT(1) == '!') {
6910
631k
      switch (NXT(2)) {
6911
206k
          case 'E':
6912
206k
        if (NXT(3) == 'L')
6913
37.4k
      xmlParseElementDecl(ctxt);
6914
168k
        else if (NXT(3) == 'N')
6915
168k
      xmlParseEntityDecl(ctxt);
6916
30
                    else
6917
30
                        SKIP(2);
6918
206k
        break;
6919
88.8k
          case 'A':
6920
88.8k
        xmlParseAttributeListDecl(ctxt);
6921
88.8k
        break;
6922
3.95k
          case 'N':
6923
3.95k
        xmlParseNotationDecl(ctxt);
6924
3.95k
        break;
6925
320k
          case '-':
6926
320k
        xmlParseComment(ctxt);
6927
320k
        break;
6928
12.1k
    default:
6929
12.1k
                    xmlFatalErr(ctxt,
6930
12.1k
                                ctxt->inSubset == 2 ?
6931
1.06k
                                    XML_ERR_EXT_SUBSET_NOT_FINISHED :
6932
12.1k
                                    XML_ERR_INT_SUBSET_NOT_FINISHED,
6933
12.1k
                                NULL);
6934
12.1k
                    SKIP(2);
6935
12.1k
        break;
6936
631k
      }
6937
631k
  } else if (NXT(1) == '?') {
6938
11.6k
      xmlParsePI(ctxt);
6939
11.6k
  }
6940
643k
    }
6941
643k
}
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
175k
xmlParseTextDecl(xmlParserCtxt *ctxt) {
6954
175k
    xmlChar *version;
6955
6956
    /*
6957
     * We know that '<?xml' is here.
6958
     */
6959
175k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
6960
175k
  SKIP(5);
6961
175k
    } else {
6962
14
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL);
6963
14
  return;
6964
14
    }
6965
6966
175k
    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
175k
    version = xmlParseVersionInfo(ctxt);
6975
175k
    if (version == NULL) {
6976
170k
  version = xmlCharStrdup(XML_DEFAULT_VERSION);
6977
170k
        if (version == NULL) {
6978
10
            xmlErrMemory(ctxt);
6979
10
            return;
6980
10
        }
6981
170k
    } else {
6982
5.12k
  if (SKIP_BLANKS == 0) {
6983
833
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
6984
833
               "Space needed here\n");
6985
833
  }
6986
5.12k
    }
6987
175k
    ctxt->input->version = version;
6988
6989
    /*
6990
     * We must have the encoding declaration
6991
     */
6992
175k
    xmlParseEncodingDecl(ctxt);
6993
6994
175k
    SKIP_BLANKS;
6995
175k
    if ((RAW == '?') && (NXT(1) == '>')) {
6996
2.02k
        SKIP(2);
6997
173k
    } else if (RAW == '>') {
6998
        /* Deprecated old WD ... */
6999
210
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7000
210
  NEXT;
7001
173k
    } else {
7002
173k
        int c;
7003
7004
173k
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
7005
216M
        while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) {
7006
216M
            NEXT;
7007
216M
            if (c == '>')
7008
164k
                break;
7009
216M
        }
7010
173k
    }
7011
175k
}
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
1.31k
                       const xmlChar *systemId) {
7029
1.31k
    int oldInputNr;
7030
7031
1.31k
    xmlCtxtInitializeLate(ctxt);
7032
7033
1.31k
    xmlDetectEncoding(ctxt);
7034
7035
1.31k
    if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) {
7036
94
  xmlParseTextDecl(ctxt);
7037
94
    }
7038
1.31k
    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
1.31k
    if ((ctxt->myDoc->intSubset == NULL) &&
7047
0
        (xmlCreateIntSubset(ctxt->myDoc, NULL, publicId, systemId) == NULL)) {
7048
0
        xmlErrMemory(ctxt);
7049
0
    }
7050
7051
1.31k
    ctxt->inSubset = 2;
7052
1.31k
    oldInputNr = ctxt->inputNr;
7053
7054
1.31k
    SKIP_BLANKS;
7055
355k
    while (!PARSER_STOPPED(ctxt)) {
7056
354k
        if (ctxt->input->cur >= ctxt->input->end) {
7057
1.33k
            if (ctxt->inputNr <= oldInputNr) {
7058
576
                xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED);
7059
576
                break;
7060
576
            }
7061
7062
755
            xmlPopPE(ctxt);
7063
353k
        } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
7064
662
            xmlParseConditionalSections(ctxt);
7065
352k
        } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
7066
350k
            xmlParseMarkupDecl(ctxt);
7067
350k
        } else if (RAW == '%') {
7068
1.64k
            xmlParsePERefInternal(ctxt, 1);
7069
1.64k
        } else {
7070
600
            xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL);
7071
7072
781
            while (ctxt->inputNr > oldInputNr)
7073
181
                xmlPopPE(ctxt);
7074
600
            break;
7075
600
        }
7076
353k
        SKIP_BLANKS;
7077
353k
        SHRINK;
7078
353k
        GROW;
7079
353k
    }
7080
1.31k
}
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
109k
xmlParseReference(xmlParserCtxt *ctxt) {
7097
109k
    xmlEntityPtr ent = NULL;
7098
109k
    const xmlChar *name;
7099
109k
    xmlChar *val;
7100
7101
109k
    if (RAW != '&')
7102
0
        return;
7103
7104
    /*
7105
     * Simple case of a CharRef
7106
     */
7107
109k
    if (NXT(1) == '#') {
7108
15.1k
  int i = 0;
7109
15.1k
  xmlChar out[16];
7110
15.1k
  int value = xmlParseCharRef(ctxt);
7111
7112
15.1k
  if (value == 0)
7113
3.39k
      return;
7114
7115
        /*
7116
         * Just encode the value in UTF-8
7117
         */
7118
11.7k
        COPY_BUF(out, i, value);
7119
11.7k
        out[i] = 0;
7120
11.7k
        if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7121
11.7k
            (!ctxt->disableSAX))
7122
10.5k
            ctxt->sax->characters(ctxt->userData, out, i);
7123
11.7k
  return;
7124
15.1k
    }
7125
7126
    /*
7127
     * We are seeing an entity reference
7128
     */
7129
93.9k
    name = xmlParseEntityRefInternal(ctxt);
7130
93.9k
    if (name == NULL)
7131
26.9k
        return;
7132
67.0k
    ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0);
7133
67.0k
    if (ent == NULL) {
7134
        /*
7135
         * Create a reference for undeclared entities.
7136
         */
7137
41.3k
        if ((ctxt->replaceEntities == 0) &&
7138
30.9k
            (ctxt->sax != NULL) &&
7139
30.9k
            (ctxt->disableSAX == 0) &&
7140
25.8k
            (ctxt->sax->reference != NULL)) {
7141
25.8k
            ctxt->sax->reference(ctxt->userData, name);
7142
25.8k
        }
7143
41.3k
        return;
7144
41.3k
    }
7145
25.6k
    if (!ctxt->wellFormed)
7146
7.31k
  return;
7147
7148
    /* special case of predefined entities */
7149
18.3k
    if ((ent->name == NULL) ||
7150
18.3k
        (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) {
7151
1.59k
  val = ent->content;
7152
1.59k
  if (val == NULL) return;
7153
  /*
7154
   * inline the entity.
7155
   */
7156
1.59k
  if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) &&
7157
1.59k
      (!ctxt->disableSAX))
7158
1.59k
      ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val));
7159
1.59k
  return;
7160
1.59k
    }
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
16.7k
    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
16.7k
    if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) ||
7211
8.04k
        (((ctxt->options & XML_PARSE_NO_XXE) == 0) &&
7212
7.77k
         ((ctxt->replaceEntities) ||
7213
16.2k
          (ctxt->validate)))) {
7214
16.2k
        if ((ent->flags & XML_ENT_PARSED) == 0) {
7215
3.21k
            xmlCtxtParseEntity(ctxt, ent);
7216
13.0k
        } 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
3.06k
            xmlCtxtParseEntity(ctxt, ent);
7227
3.06k
        }
7228
16.2k
    }
7229
7230
    /*
7231
     * We also check for amplification if entities aren't substituted.
7232
     * They might be expanded later.
7233
     */
7234
16.7k
    if (xmlParserEntityCheck(ctxt, ent->expandedSize))
7235
74
        return;
7236
7237
16.6k
    if ((ctxt->sax == NULL) || (ctxt->disableSAX))
7238
694
        return;
7239
7240
15.9k
    if (ctxt->replaceEntities == 0) {
7241
  /*
7242
   * Create a reference
7243
   */
7244
5.56k
        if (ctxt->sax->reference != NULL)
7245
5.56k
      ctxt->sax->reference(ctxt->userData, ent->name);
7246
10.4k
    } else if ((ent->children != NULL) && (ctxt->node != NULL)) {
7247
7.48k
        xmlNodePtr copy, cur;
7248
7249
        /*
7250
         * Seems we are generating the DOM content, copy the tree
7251
   */
7252
7.48k
        cur = ent->children;
7253
7254
        /*
7255
         * Handle first text node with SAX to coalesce text efficiently
7256
         */
7257
7.48k
        if ((cur->type == XML_TEXT_NODE) ||
7258
6.52k
            (cur->type == XML_CDATA_SECTION_NODE)) {
7259
6.52k
            int len = xmlStrlen(cur->content);
7260
7261
6.52k
            if ((cur->type == XML_TEXT_NODE) ||
7262
6.29k
                (ctxt->options & XML_PARSE_NOCDATA)) {
7263
6.29k
                if (ctxt->sax->characters != NULL)
7264
6.29k
                    ctxt->sax->characters(ctxt, cur->content, len);
7265
6.29k
            } else {
7266
232
                if (ctxt->sax->cdataBlock != NULL)
7267
232
                    ctxt->sax->cdataBlock(ctxt, cur->content, len);
7268
232
            }
7269
7270
6.52k
            cur = cur->next;
7271
6.52k
        }
7272
7273
59.7k
        while (cur != NULL) {
7274
54.3k
            xmlNodePtr last;
7275
7276
            /*
7277
             * Handle last text node with SAX to coalesce text efficiently
7278
             */
7279
54.3k
            if ((cur->next == NULL) &&
7280
3.34k
                ((cur->type == XML_TEXT_NODE) ||
7281
2.03k
                 (cur->type == XML_CDATA_SECTION_NODE))) {
7282
2.03k
                int len = xmlStrlen(cur->content);
7283
7284
2.03k
                if ((cur->type == XML_TEXT_NODE) ||
7285
1.93k
                    (ctxt->options & XML_PARSE_NOCDATA)) {
7286
1.93k
                    if (ctxt->sax->characters != NULL)
7287
1.93k
                        ctxt->sax->characters(ctxt, cur->content, len);
7288
1.93k
                } else {
7289
104
                    if (ctxt->sax->cdataBlock != NULL)
7290
104
                        ctxt->sax->cdataBlock(ctxt, cur->content, len);
7291
104
                }
7292
7293
2.03k
                break;
7294
2.03k
            }
7295
7296
            /*
7297
             * Reset coalesce buffer stats only for non-text nodes.
7298
             */
7299
52.3k
            ctxt->nodemem = 0;
7300
52.3k
            ctxt->nodelen = 0;
7301
7302
52.3k
            copy = xmlDocCopyNode(cur, ctxt->myDoc, 1);
7303
7304
52.3k
            if (copy == NULL) {
7305
50
                xmlErrMemory(ctxt);
7306
50
                break;
7307
50
            }
7308
7309
52.2k
            if (ctxt->parseMode == XML_PARSE_READER) {
7310
                /* Needed for reader */
7311
51.4k
                copy->extra = cur->extra;
7312
                /* Maybe needed for reader */
7313
51.4k
                copy->_private = cur->_private;
7314
51.4k
            }
7315
7316
52.2k
            copy->parent = ctxt->node;
7317
52.2k
            last = ctxt->node->last;
7318
52.2k
            if (last == NULL) {
7319
247
                ctxt->node->children = copy;
7320
52.0k
            } else {
7321
52.0k
                last->next = copy;
7322
52.0k
                copy->prev = last;
7323
52.0k
            }
7324
52.2k
            ctxt->node->last = copy;
7325
7326
52.2k
            cur = cur->next;
7327
52.2k
        }
7328
7.48k
    }
7329
15.9k
}
7330
7331
static void
7332
501k
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
501k
    if ((ctxt->standalone == 1) ||
7355
501k
        ((ctxt->hasExternalSubset == 0) &&
7356
495k
         (ctxt->hasPErefs == 0))) {
7357
132k
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY,
7358
132k
                          "Entity '%s' not defined\n", name);
7359
132k
#ifdef LIBXML_VALID_ENABLED
7360
369k
    } 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
323k
        xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY,
7369
323k
                         "Entity '%s' not defined\n", name, NULL);
7370
323k
#endif
7371
323k
    } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) ||
7372
4.73k
               ((ctxt->replaceEntities) &&
7373
45.3k
                ((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
45.3k
        xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY,
7383
45.3k
                     "Entity '%s' not defined\n", name);
7384
45.3k
    } else {
7385
657
        xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY,
7386
657
                      "Entity '%s' not defined\n", name, NULL);
7387
657
    }
7388
7389
501k
    ctxt->valid = 0;
7390
501k
}
7391
7392
static xmlEntityPtr
7393
3.90M
xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) {
7394
3.90M
    xmlEntityPtr ent = NULL;
7395
7396
    /*
7397
     * Predefined entities override any extra definition
7398
     */
7399
3.90M
    if ((ctxt->options & XML_PARSE_OLDSAX) == 0) {
7400
1.79M
        ent = xmlGetPredefinedEntity(name);
7401
1.79M
        if (ent != NULL)
7402
599k
            return(ent);
7403
1.79M
    }
7404
7405
    /*
7406
     * Ask first SAX for entity resolution, otherwise try the
7407
     * entities which may have stored in the parser context.
7408
     */
7409
3.30M
    if (ctxt->sax != NULL) {
7410
3.30M
  if (ctxt->sax->getEntity != NULL)
7411
3.30M
      ent = ctxt->sax->getEntity(ctxt->userData, name);
7412
3.30M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7413
8.07k
      (ctxt->options & XML_PARSE_OLDSAX))
7414
452
      ent = xmlGetPredefinedEntity(name);
7415
3.30M
  if ((ctxt->wellFormed == 1 ) && (ent == NULL) &&
7416
8.07k
      (ctxt->userData==ctxt)) {
7417
8.07k
      ent = xmlSAX2GetEntity(ctxt, name);
7418
8.07k
  }
7419
3.30M
    }
7420
7421
3.30M
    if (ent == NULL) {
7422
413k
        xmlHandleUndeclaredEntity(ctxt, name);
7423
413k
    }
7424
7425
    /*
7426
     * [ WFC: Parsed Entity ]
7427
     * An entity reference must not contain the name of an
7428
     * unparsed entity
7429
     */
7430
2.89M
    else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) {
7431
44
  xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY,
7432
44
     "Entity reference to unparsed entity %s\n", name);
7433
44
        ent = NULL;
7434
44
    }
7435
7436
    /*
7437
     * [ WFC: No External Entity References ]
7438
     * Attribute values cannot contain direct or indirect
7439
     * entity references to external entities.
7440
     */
7441
2.89M
    else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) {
7442
12.7k
        if (inAttr) {
7443
515
            xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL,
7444
515
                 "Attribute references external entity '%s'\n", name);
7445
515
            ent = NULL;
7446
515
        }
7447
12.7k
    }
7448
7449
3.30M
    return(ent);
7450
3.90M
}
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
1.30M
xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) {
7462
1.30M
    const xmlChar *name;
7463
7464
1.30M
    GROW;
7465
7466
1.30M
    if (RAW != '&')
7467
0
        return(NULL);
7468
1.30M
    NEXT;
7469
1.30M
    name = xmlParseName(ctxt);
7470
1.30M
    if (name == NULL) {
7471
45.1k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7472
45.1k
           "xmlParseEntityRef: no name\n");
7473
45.1k
        return(NULL);
7474
45.1k
    }
7475
1.26M
    if (RAW != ';') {
7476
108k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7477
108k
  return(NULL);
7478
108k
    }
7479
1.15M
    NEXT;
7480
7481
1.15M
    return(name);
7482
1.26M
}
7483
7484
/**
7485
 * @deprecated Internal function, don't use.
7486
 *
7487
 * @param ctxt  an XML parser context
7488
 * @returns the xmlEntity if found, or NULL otherwise.
7489
 */
7490
xmlEntity *
7491
0
xmlParseEntityRef(xmlParserCtxt *ctxt) {
7492
0
    const xmlChar *name;
7493
7494
0
    if (ctxt == NULL)
7495
0
        return(NULL);
7496
7497
0
    name = xmlParseEntityRefInternal(ctxt);
7498
0
    if (name == NULL)
7499
0
        return(NULL);
7500
7501
0
    return(xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 0));
7502
0
}
7503
7504
/**
7505
 * Parse ENTITY references declarations, but this version parses it from
7506
 * a string value.
7507
 *
7508
 *     [68] EntityRef ::= '&' Name ';'
7509
 *
7510
 * [ WFC: Entity Declared ]
7511
 * In a document without any DTD, a document with only an internal DTD
7512
 * subset which contains no parameter entity references, or a document
7513
 * with "standalone='yes'", the Name given in the entity reference
7514
 * must match that in an entity declaration, except that well-formed
7515
 * documents need not declare any of the following entities: amp, lt,
7516
 * gt, apos, quot.  The declaration of a parameter entity must precede
7517
 * any reference to it.  Similarly, the declaration of a general entity
7518
 * must precede any reference to it which appears in a default value in an
7519
 * attribute-list declaration. Note that if entities are declared in the
7520
 * external subset or in external parameter entities, a non-validating
7521
 * processor is not obligated to read and process their declarations;
7522
 * for such documents, the rule that an entity must be declared is a
7523
 * well-formedness constraint only if standalone='yes'.
7524
 *
7525
 * [ WFC: Parsed Entity ]
7526
 * An entity reference must not contain the name of an unparsed entity
7527
 *
7528
 * @param ctxt  an XML parser context
7529
 * @param str  a pointer to an index in the string
7530
 * @returns the xmlEntity if found, or NULL otherwise. The str pointer
7531
 * is updated to the current location in the string.
7532
 */
7533
static xmlChar *
7534
2.75M
xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) {
7535
2.75M
    xmlChar *name;
7536
2.75M
    const xmlChar *ptr;
7537
2.75M
    xmlChar cur;
7538
7539
2.75M
    if ((str == NULL) || (*str == NULL))
7540
0
        return(NULL);
7541
2.75M
    ptr = *str;
7542
2.75M
    cur = *ptr;
7543
2.75M
    if (cur != '&')
7544
0
  return(NULL);
7545
7546
2.75M
    ptr++;
7547
2.75M
    name = xmlParseStringName(ctxt, &ptr);
7548
2.75M
    if (name == NULL) {
7549
25
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7550
25
           "xmlParseStringEntityRef: no name\n");
7551
25
  *str = ptr;
7552
25
  return(NULL);
7553
25
    }
7554
2.75M
    if (*ptr != ';') {
7555
16
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7556
16
        xmlFree(name);
7557
16
  *str = ptr;
7558
16
  return(NULL);
7559
16
    }
7560
2.75M
    ptr++;
7561
7562
2.75M
    *str = ptr;
7563
2.75M
    return(name);
7564
2.75M
}
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
10.1k
  xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n");
7608
10.1k
  return;
7609
10.1k
    }
7610
166k
    if (RAW != ';') {
7611
11.9k
  xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL);
7612
11.9k
        return;
7613
11.9k
    }
7614
7615
154k
    NEXT;
7616
7617
    /* Must be set before xmlHandleUndeclaredEntity */
7618
154k
    ctxt->hasPErefs = 1;
7619
7620
    /*
7621
     * Request the entity from SAX
7622
     */
7623
154k
    if ((ctxt->sax != NULL) &&
7624
154k
  (ctxt->sax->getParameterEntity != NULL))
7625
154k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7626
7627
154k
    if (entity == NULL) {
7628
44.9k
        xmlHandleUndeclaredEntity(ctxt, name);
7629
109k
    } else {
7630
  /*
7631
   * Internal checking in case the entity quest barfed
7632
   */
7633
109k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7634
101k
      (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
109k
  } else {
7639
109k
      if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) &&
7640
101k
                ((ctxt->options & XML_PARSE_NO_XXE) ||
7641
101k
     (((ctxt->loadsubset & ~XML_SKIP_IDS) == 0) &&
7642
21.4k
      (ctxt->replaceEntities == 0) &&
7643
9.43k
      (ctxt->validate == 0))))
7644
263
    return;
7645
7646
109k
            if (entity->flags & XML_ENT_EXPANDING) {
7647
16
                xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL);
7648
16
                return;
7649
16
            }
7650
7651
109k
      input = xmlNewEntityInputStream(ctxt, entity);
7652
109k
      if (xmlCtxtPushInput(ctxt, input) < 0) {
7653
3.15k
                xmlFreeInputStream(input);
7654
3.15k
    return;
7655
3.15k
            }
7656
7657
106k
            entity->flags |= XML_ENT_EXPANDING;
7658
7659
106k
            if (markupDecl)
7660
96.7k
                input->flags |= XML_INPUT_MARKUP_DECL;
7661
7662
106k
            GROW;
7663
7664
106k
      if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) {
7665
98.0k
                xmlDetectEncoding(ctxt);
7666
7667
98.0k
                if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
7668
44.3k
                    (IS_BLANK_CH(NXT(5)))) {
7669
43.5k
                    xmlParseTextDecl(ctxt);
7670
43.5k
                }
7671
98.0k
            }
7672
106k
  }
7673
109k
    }
7674
154k
}
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
139k
xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) {
7697
139k
    xmlParserInputPtr oldinput, input = NULL;
7698
139k
    xmlParserInputPtr *oldinputTab;
7699
139k
    xmlChar *oldencoding;
7700
139k
    xmlChar *content = NULL;
7701
139k
    xmlResourceType rtype;
7702
139k
    size_t length, i;
7703
139k
    int oldinputNr, oldinputMax;
7704
139k
    int ret = -1;
7705
139k
    int res;
7706
7707
139k
    if ((ctxt == NULL) || (entity == NULL) ||
7708
139k
        ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) &&
7709
0
   (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) ||
7710
139k
  (entity->content != NULL)) {
7711
0
  xmlFatalErr(ctxt, XML_ERR_ARGUMENT,
7712
0
              "xmlLoadEntityContent parameter error");
7713
0
        return(-1);
7714
0
    }
7715
7716
139k
    if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY)
7717
139k
        rtype = XML_RESOURCE_PARAMETER_ENTITY;
7718
0
    else
7719
0
        rtype = XML_RESOURCE_GENERAL_ENTITY;
7720
7721
139k
    input = xmlLoadResource(ctxt, (char *) entity->URI,
7722
139k
                            (char *) entity->ExternalID, rtype);
7723
139k
    if (input == NULL)
7724
738
        return(-1);
7725
7726
138k
    oldinput = ctxt->input;
7727
138k
    oldinputNr = ctxt->inputNr;
7728
138k
    oldinputMax = ctxt->inputMax;
7729
138k
    oldinputTab = ctxt->inputTab;
7730
138k
    oldencoding = ctxt->encoding;
7731
7732
138k
    ctxt->input = NULL;
7733
138k
    ctxt->inputNr = 0;
7734
138k
    ctxt->inputMax = 1;
7735
138k
    ctxt->encoding = NULL;
7736
138k
    ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr));
7737
138k
    if (ctxt->inputTab == NULL) {
7738
7
        xmlErrMemory(ctxt);
7739
7
        xmlFreeInputStream(input);
7740
7
        goto error;
7741
7
    }
7742
7743
138k
    xmlBufResetInput(input->buf->buffer, input);
7744
7745
138k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
7746
6
        xmlFreeInputStream(input);
7747
6
        goto error;
7748
6
    }
7749
7750
138k
    xmlDetectEncoding(ctxt);
7751
7752
    /*
7753
     * Parse a possible text declaration first
7754
     */
7755
138k
    if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) {
7756
131k
  xmlParseTextDecl(ctxt);
7757
        /*
7758
         * An XML-1.0 document can't reference an entity not XML-1.0
7759
         */
7760
131k
        if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
7761
3.88k
            (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
7762
481
            xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
7763
481
                           "Version mismatch between document and entity\n");
7764
481
        }
7765
131k
    }
7766
7767
138k
    length = input->cur - input->base;
7768
138k
    xmlBufShrink(input->buf->buffer, length);
7769
138k
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7770
7771
140k
    while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0)
7772
1.70k
        ;
7773
7774
138k
    xmlBufResetInput(input->buf->buffer, input);
7775
7776
138k
    if (res < 0) {
7777
1.01k
        xmlCtxtErrIO(ctxt, input->buf->error, NULL);
7778
1.01k
        goto error;
7779
1.01k
    }
7780
7781
137k
    length = xmlBufUse(input->buf->buffer);
7782
137k
    if (length > INT_MAX) {
7783
0
        xmlErrMemory(ctxt);
7784
0
        goto error;
7785
0
    }
7786
7787
137k
    content = xmlStrndup(xmlBufContent(input->buf->buffer), length);
7788
137k
    if (content == NULL) {
7789
9
        xmlErrMemory(ctxt);
7790
9
        goto error;
7791
9
    }
7792
7793
14.6M
    for (i = 0; i < length; ) {
7794
14.6M
        int clen = length - i;
7795
14.6M
        int c = xmlGetUTF8Char(content + i, &clen);
7796
7797
14.6M
        if ((c < 0) || (!IS_CHAR(c))) {
7798
137k
            xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR,
7799
137k
                              "xmlLoadEntityContent: invalid char value %d\n",
7800
137k
                              content[i]);
7801
137k
            goto error;
7802
137k
        }
7803
14.5M
        i += clen;
7804
14.5M
    }
7805
7806
146
    xmlSaturatedAdd(&ctxt->sizeentities, length);
7807
146
    entity->content = content;
7808
146
    entity->length = length;
7809
146
    content = NULL;
7810
146
    ret = 0;
7811
7812
138k
error:
7813
277k
    while (ctxt->inputNr > 0)
7814
138k
        xmlFreeInputStream(xmlCtxtPopInput(ctxt));
7815
138k
    xmlFree(ctxt->inputTab);
7816
138k
    xmlFree(ctxt->encoding);
7817
7818
138k
    ctxt->input = oldinput;
7819
138k
    ctxt->inputNr = oldinputNr;
7820
138k
    ctxt->inputMax = oldinputMax;
7821
138k
    ctxt->inputTab = oldinputTab;
7822
138k
    ctxt->encoding = oldencoding;
7823
7824
138k
    xmlFree(content);
7825
7826
138k
    return(ret);
7827
146
}
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
198k
xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) {
7860
198k
    const xmlChar *ptr;
7861
198k
    xmlChar cur;
7862
198k
    xmlChar *name;
7863
198k
    xmlEntityPtr entity = NULL;
7864
7865
198k
    if ((str == NULL) || (*str == NULL)) return(NULL);
7866
198k
    ptr = *str;
7867
198k
    cur = *ptr;
7868
198k
    if (cur != '%')
7869
0
        return(NULL);
7870
198k
    ptr++;
7871
198k
    name = xmlParseStringName(ctxt, &ptr);
7872
198k
    if (name == NULL) {
7873
2.12k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7874
2.12k
           "xmlParseStringPEReference: no name\n");
7875
2.12k
  *str = ptr;
7876
2.12k
  return(NULL);
7877
2.12k
    }
7878
196k
    cur = *ptr;
7879
196k
    if (cur != ';') {
7880
4.63k
  xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL);
7881
4.63k
  xmlFree(name);
7882
4.63k
  *str = ptr;
7883
4.63k
  return(NULL);
7884
4.63k
    }
7885
191k
    ptr++;
7886
7887
    /* Must be set before xmlHandleUndeclaredEntity */
7888
191k
    ctxt->hasPErefs = 1;
7889
7890
    /*
7891
     * Request the entity from SAX
7892
     */
7893
191k
    if ((ctxt->sax != NULL) &&
7894
191k
  (ctxt->sax->getParameterEntity != NULL))
7895
191k
  entity = ctxt->sax->getParameterEntity(ctxt->userData, name);
7896
7897
191k
    if (entity == NULL) {
7898
43.7k
        xmlHandleUndeclaredEntity(ctxt, name);
7899
148k
    } else {
7900
  /*
7901
   * Internal checking in case the entity quest barfed
7902
   */
7903
148k
  if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) &&
7904
140k
      (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
148k
    }
7910
7911
191k
    xmlFree(name);
7912
191k
    *str = ptr;
7913
191k
    return(entity);
7914
196k
}
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
16.9k
xmlParseDocTypeDecl(xmlParserCtxt *ctxt) {
7933
16.9k
    const xmlChar *name = NULL;
7934
16.9k
    xmlChar *publicId = NULL;
7935
16.9k
    xmlChar *URI = NULL;
7936
7937
    /*
7938
     * We know that '<!DOCTYPE' has been detected.
7939
     */
7940
16.9k
    SKIP(9);
7941
7942
16.9k
    if (SKIP_BLANKS == 0) {
7943
1.04k
        xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
7944
1.04k
                       "Space required after 'DOCTYPE'\n");
7945
1.04k
    }
7946
7947
    /*
7948
     * Parse the DOCTYPE name.
7949
     */
7950
16.9k
    name = xmlParseName(ctxt);
7951
16.9k
    if (name == NULL) {
7952
781
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
7953
781
           "xmlParseDocTypeDecl : no DOCTYPE name !\n");
7954
781
    }
7955
16.9k
    ctxt->intSubName = name;
7956
7957
16.9k
    SKIP_BLANKS;
7958
7959
    /*
7960
     * Check for public and system identifier (URI)
7961
     */
7962
16.9k
    URI = xmlParseExternalID(ctxt, &publicId, 1);
7963
7964
16.9k
    if ((URI != NULL) || (publicId != NULL)) {
7965
2.44k
        ctxt->hasExternalSubset = 1;
7966
2.44k
    }
7967
16.9k
    ctxt->extSubURI = URI;
7968
16.9k
    ctxt->extSubSystem = publicId;
7969
7970
16.9k
    SKIP_BLANKS;
7971
7972
    /*
7973
     * Create and update the internal subset.
7974
     */
7975
16.9k
    if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) &&
7976
16.9k
  (!ctxt->disableSAX))
7977
16.7k
  ctxt->sax->internalSubset(ctxt->userData, name, publicId, URI);
7978
7979
16.9k
    if ((RAW != '[') && (RAW != '>')) {
7980
221
  xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
7981
221
    }
7982
16.9k
}
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
11.5k
xmlParseInternalSubset(xmlParserCtxtPtr ctxt) {
7993
    /*
7994
     * Is there any DTD definition ?
7995
     */
7996
11.5k
    if (RAW == '[') {
7997
11.5k
        int oldInputNr = ctxt->inputNr;
7998
7999
11.5k
        NEXT;
8000
  /*
8001
   * Parse the succession of Markup declarations and
8002
   * PEReferences.
8003
   * Subsequence (markupdecl | PEReference | S)*
8004
   */
8005
11.5k
  SKIP_BLANKS;
8006
556k
        while (1) {
8007
556k
            if (PARSER_STOPPED(ctxt)) {
8008
626
                return;
8009
556k
            } else if (ctxt->input->cur >= ctxt->input->end) {
8010
95.8k
                if (ctxt->inputNr <= oldInputNr) {
8011
1.22k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8012
1.22k
                    return;
8013
1.22k
                }
8014
94.5k
                xmlPopPE(ctxt);
8015
460k
            } else if ((RAW == ']') && (ctxt->inputNr <= oldInputNr)) {
8016
7.07k
                NEXT;
8017
7.07k
                SKIP_BLANKS;
8018
7.07k
                break;
8019
453k
            } else if ((PARSER_EXTERNAL(ctxt)) &&
8020
402k
                       (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) {
8021
                /*
8022
                 * Conditional sections are allowed in external entities
8023
                 * included by PE References in the internal subset.
8024
                 */
8025
3.05k
                xmlParseConditionalSections(ctxt);
8026
450k
            } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) {
8027
291k
                xmlParseMarkupDecl(ctxt);
8028
291k
            } else if (RAW == '%') {
8029
155k
                xmlParsePERefInternal(ctxt, 1);
8030
155k
            } else {
8031
2.60k
                xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL);
8032
8033
3.19k
                while (ctxt->inputNr > oldInputNr)
8034
594
                    xmlPopPE(ctxt);
8035
2.60k
                return;
8036
2.60k
            }
8037
545k
            SKIP_BLANKS;
8038
545k
            SHRINK;
8039
545k
            GROW;
8040
545k
        }
8041
11.5k
    }
8042
8043
    /*
8044
     * We should be at the end of the DOCTYPE declaration.
8045
     */
8046
7.07k
    if (RAW != '>') {
8047
118
        xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL);
8048
118
        return;
8049
118
    }
8050
6.95k
    NEXT;
8051
6.95k
}
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
118k
xmlParseAttribute(xmlParserCtxt *ctxt, xmlChar **value) {
8089
118k
    const xmlChar *name;
8090
118k
    xmlChar *val;
8091
8092
118k
    *value = NULL;
8093
118k
    GROW;
8094
118k
    name = xmlParseName(ctxt);
8095
118k
    if (name == NULL) {
8096
22.9k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8097
22.9k
                 "error parsing attribute name\n");
8098
22.9k
        return(NULL);
8099
22.9k
    }
8100
8101
    /*
8102
     * read the value
8103
     */
8104
95.7k
    SKIP_BLANKS;
8105
95.7k
    if (RAW == '=') {
8106
84.8k
        NEXT;
8107
84.8k
  SKIP_BLANKS;
8108
84.8k
  val = xmlParseAttValue(ctxt);
8109
84.8k
    } else {
8110
10.8k
  xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8111
10.8k
         "Specification mandates value for attribute %s\n", name);
8112
10.8k
  return(name);
8113
10.8k
    }
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
84.8k
    if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) {
8121
6.12k
  if (!xmlCheckLanguageID(val)) {
8122
3.99k
      xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8123
3.99k
              "Malformed value for xml:lang : %s\n",
8124
3.99k
        val, NULL);
8125
3.99k
  }
8126
6.12k
    }
8127
8128
    /*
8129
     * Check that xml:space conforms to the specification
8130
     */
8131
84.8k
    if (xmlStrEqual(name, BAD_CAST "xml:space")) {
8132
538
  if (xmlStrEqual(val, BAD_CAST "default"))
8133
239
      *(ctxt->space) = 0;
8134
299
  else if (xmlStrEqual(val, BAD_CAST "preserve"))
8135
210
      *(ctxt->space) = 1;
8136
89
  else {
8137
89
    xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8138
89
"Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8139
89
                                 val, NULL);
8140
89
  }
8141
538
    }
8142
8143
84.8k
    *value = val;
8144
84.8k
    return(name);
8145
95.7k
}
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
294k
xmlParseStartTag(xmlParserCtxt *ctxt) {
8176
294k
    const xmlChar *name;
8177
294k
    const xmlChar *attname;
8178
294k
    xmlChar *attvalue;
8179
294k
    const xmlChar **atts = ctxt->atts;
8180
294k
    int nbatts = 0;
8181
294k
    int maxatts = ctxt->maxatts;
8182
294k
    int i;
8183
8184
294k
    if (RAW != '<') return(NULL);
8185
294k
    NEXT1;
8186
8187
294k
    name = xmlParseName(ctxt);
8188
294k
    if (name == NULL) {
8189
9.86k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8190
9.86k
       "xmlParseStartTag: invalid element name\n");
8191
9.86k
        return(NULL);
8192
9.86k
    }
8193
8194
    /*
8195
     * Now parse the attributes, it ends up with the ending
8196
     *
8197
     * (S Attribute)* S?
8198
     */
8199
284k
    SKIP_BLANKS;
8200
284k
    GROW;
8201
8202
328k
    while (((RAW != '>') &&
8203
192k
     ((RAW != '/') || (NXT(1) != '>')) &&
8204
124k
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8205
118k
  attname = xmlParseAttribute(ctxt, &attvalue);
8206
118k
        if (attname == NULL)
8207
22.9k
      break;
8208
95.7k
        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
242k
      for (i = 0; i < nbatts;i += 2) {
8215
166k
          if (xmlStrEqual(atts[i], attname)) {
8216
6.84k
        xmlErrAttributeDup(ctxt, NULL, attname);
8217
6.84k
        goto failed;
8218
6.84k
    }
8219
166k
      }
8220
      /*
8221
       * Add the pair to atts
8222
       */
8223
75.6k
      if (nbatts + 4 > maxatts) {
8224
15.3k
          const xmlChar **n;
8225
15.3k
                int newSize;
8226
8227
15.3k
                newSize = xmlGrowCapacity(maxatts, sizeof(n[0]) * 2,
8228
15.3k
                                          11, XML_MAX_ATTRS);
8229
15.3k
                if (newSize < 0) {
8230
0
        xmlErrMemory(ctxt);
8231
0
        goto failed;
8232
0
    }
8233
15.3k
#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
8234
15.3k
                if (newSize < 2)
8235
9.62k
                    newSize = 2;
8236
15.3k
#endif
8237
15.3k
          n = xmlRealloc(atts, newSize * sizeof(n[0]) * 2);
8238
15.3k
    if (n == NULL) {
8239
2
        xmlErrMemory(ctxt);
8240
2
        goto failed;
8241
2
    }
8242
15.3k
    atts = n;
8243
15.3k
                maxatts = newSize * 2;
8244
15.3k
    ctxt->atts = atts;
8245
15.3k
    ctxt->maxatts = maxatts;
8246
15.3k
      }
8247
8248
75.6k
      atts[nbatts++] = attname;
8249
75.6k
      atts[nbatts++] = attvalue;
8250
75.6k
      atts[nbatts] = NULL;
8251
75.6k
      atts[nbatts + 1] = NULL;
8252
8253
75.6k
            attvalue = NULL;
8254
75.6k
  }
8255
8256
95.7k
failed:
8257
8258
95.7k
        if (attvalue != NULL)
8259
6.84k
            xmlFree(attvalue);
8260
8261
95.7k
  GROW
8262
95.7k
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
8263
52.3k
      break;
8264
43.3k
  if (SKIP_BLANKS == 0) {
8265
16.8k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
8266
16.8k
         "attributes construct error\n");
8267
16.8k
  }
8268
43.3k
  SHRINK;
8269
43.3k
        GROW;
8270
43.3k
    }
8271
8272
    /*
8273
     * SAX: Start of Element !
8274
     */
8275
284k
    if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) &&
8276
284k
  (!ctxt->disableSAX)) {
8277
272k
  if (nbatts > 0)
8278
58.4k
      ctxt->sax->startElement(ctxt->userData, name, atts);
8279
214k
  else
8280
214k
      ctxt->sax->startElement(ctxt->userData, name, NULL);
8281
272k
    }
8282
8283
284k
    if (atts != NULL) {
8284
        /* Free only the content strings */
8285
257k
        for (i = 1;i < nbatts;i+=2)
8286
75.6k
      if (atts[i] != NULL)
8287
75.6k
         xmlFree((xmlChar *) atts[i]);
8288
182k
    }
8289
284k
    return(name);
8290
284k
}
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
74.6k
xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) {
8306
74.6k
    const xmlChar *name;
8307
8308
74.6k
    GROW;
8309
74.6k
    if ((RAW != '<') || (NXT(1) != '/')) {
8310
14
  xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED,
8311
14
           "xmlParseEndTag: '</' not found\n");
8312
14
  return;
8313
14
    }
8314
74.6k
    SKIP(2);
8315
8316
74.6k
    name = xmlParseNameAndCompare(ctxt,ctxt->name);
8317
8318
    /*
8319
     * We should definitely be at the ending "S? '>'" part
8320
     */
8321
74.6k
    GROW;
8322
74.6k
    SKIP_BLANKS;
8323
74.6k
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
8324
8.34k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
8325
8.34k
    } else
8326
66.3k
  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
74.6k
    if (name != (xmlChar*)1) {
8335
6.09k
        if (name == NULL) name = BAD_CAST "unparsable";
8336
6.09k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
8337
6.09k
         "Opening and ending tag mismatch: %s line %d and %s\n",
8338
6.09k
                    ctxt->name, line, name);
8339
6.09k
    }
8340
8341
    /*
8342
     * SAX: End of Tag
8343
     */
8344
74.6k
    if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
8345
74.6k
  (!ctxt->disableSAX))
8346
72.9k
        ctxt->sax->endElement(ctxt->userData, ctxt->name);
8347
8348
74.6k
    namePop(ctxt);
8349
74.6k
    spacePop(ctxt);
8350
74.6k
}
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
1.21M
xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) {
8391
1.21M
    xmlHashedString l, p;
8392
1.21M
    int start, isNCName = 0;
8393
8394
1.21M
    l.name = NULL;
8395
1.21M
    p.name = NULL;
8396
8397
1.21M
    GROW;
8398
1.21M
    start = CUR_PTR - BASE_PTR;
8399
8400
1.21M
    l = xmlParseNCName(ctxt);
8401
1.21M
    if (l.name != NULL) {
8402
811k
        isNCName = 1;
8403
811k
        if (CUR == ':') {
8404
171k
            NEXT;
8405
171k
            p = l;
8406
171k
            l = xmlParseNCName(ctxt);
8407
171k
        }
8408
811k
    }
8409
1.21M
    if ((l.name == NULL) || (CUR == ':')) {
8410
404k
        xmlChar *tmp;
8411
8412
404k
        l.name = NULL;
8413
404k
        p.name = NULL;
8414
404k
        if ((isNCName == 0) && (CUR != ':'))
8415
393k
            return(l);
8416
10.3k
        tmp = xmlParseNmtoken(ctxt);
8417
10.3k
        if (tmp != NULL)
8418
6.65k
            xmlFree(tmp);
8419
10.3k
        l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start,
8420
10.3k
                                CUR_PTR - (BASE_PTR + start));
8421
10.3k
        if (l.name == NULL) {
8422
1
            xmlErrMemory(ctxt);
8423
1
            return(l);
8424
1
        }
8425
10.3k
        xmlNsErr(ctxt, XML_NS_ERR_QNAME,
8426
10.3k
                 "Failed to parse QName '%s'\n", l.name, NULL, NULL);
8427
10.3k
    }
8428
8429
816k
    *prefix = p;
8430
816k
    return(l);
8431
1.21M
}
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
5.09k
xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) {
8447
5.09k
    xmlHashedString n, p;
8448
8449
5.09k
    n = xmlParseQNameHashed(ctxt, &p);
8450
5.09k
    if (n.name == NULL)
8451
686
        return(NULL);
8452
4.40k
    *prefix = p.name;
8453
4.40k
    return(n.name);
8454
5.09k
}
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
22.9k
                        xmlChar const *prefix) {
8470
22.9k
    const xmlChar *cmp;
8471
22.9k
    const xmlChar *in;
8472
22.9k
    const xmlChar *ret;
8473
22.9k
    const xmlChar *prefix2;
8474
8475
22.9k
    if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name));
8476
8477
22.9k
    GROW;
8478
22.9k
    in = ctxt->input->cur;
8479
8480
22.9k
    cmp = prefix;
8481
46.3k
    while (*in != 0 && *in == *cmp) {
8482
23.3k
  ++in;
8483
23.3k
  ++cmp;
8484
23.3k
    }
8485
22.9k
    if ((*cmp == 0) && (*in == ':')) {
8486
20.4k
        in++;
8487
20.4k
  cmp = name;
8488
49.4k
  while (*in != 0 && *in == *cmp) {
8489
28.9k
      ++in;
8490
28.9k
      ++cmp;
8491
28.9k
  }
8492
20.4k
  if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) {
8493
      /* success */
8494
17.8k
            ctxt->input->col += in - ctxt->input->cur;
8495
17.8k
      ctxt->input->cur = in;
8496
17.8k
      return((const xmlChar*) 1);
8497
17.8k
  }
8498
20.4k
    }
8499
    /*
8500
     * all strings coms from the dictionary, equality can be done directly
8501
     */
8502
5.09k
    ret = xmlParseQName (ctxt, &prefix2);
8503
5.09k
    if (ret == NULL)
8504
686
        return(NULL);
8505
4.40k
    if ((ret == name) && (prefix == prefix2))
8506
1.87k
  return((const xmlChar*) 1);
8507
2.53k
    return ret;
8508
4.40k
}
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
189k
{
8529
189k
    xmlHashedString hname;
8530
189k
    const xmlChar *prefix, *name;
8531
189k
    xmlChar *val = NULL, *internal_val = NULL;
8532
189k
    int special = 0;
8533
189k
    int isNamespace;
8534
189k
    int flags;
8535
8536
189k
    *value = NULL;
8537
189k
    GROW;
8538
189k
    hname = xmlParseQNameHashed(ctxt, hprefix);
8539
189k
    if (hname.name == NULL) {
8540
30.8k
        xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8541
30.8k
                       "error parsing attribute name\n");
8542
30.8k
        return(hname);
8543
30.8k
    }
8544
158k
    name = hname.name;
8545
158k
    prefix = hprefix->name;
8546
8547
    /*
8548
     * get the type if needed
8549
     */
8550
158k
    if (ctxt->attsSpecial != NULL) {
8551
21.7k
        special = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem,
8552
21.7k
                                              prefix, name));
8553
21.7k
    }
8554
8555
    /*
8556
     * read the value
8557
     */
8558
158k
    SKIP_BLANKS;
8559
158k
    if (RAW != '=') {
8560
14.1k
        xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE,
8561
14.1k
                          "Specification mandates value for attribute %s\n",
8562
14.1k
                          name);
8563
14.1k
        goto error;
8564
14.1k
    }
8565
8566
8567
144k
    NEXT;
8568
144k
    SKIP_BLANKS;
8569
144k
    flags = 0;
8570
144k
    isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) ||
8571
133k
                   (prefix == ctxt->str_xmlns));
8572
144k
    val = xmlParseAttValueInternal(ctxt, len, &flags, special,
8573
144k
                                   isNamespace);
8574
144k
    if (val == NULL)
8575
2.53k
        goto error;
8576
8577
142k
    *alloc = (flags & XML_ATTVAL_ALLOC) != 0;
8578
8579
142k
#ifdef LIBXML_VALID_ENABLED
8580
142k
    if ((ctxt->validate) &&
8581
42.5k
        (ctxt->standalone == 1) &&
8582
680
        (special & XML_SPECIAL_EXTERNAL) &&
8583
95
        (flags & XML_ATTVAL_NORM_CHANGE)) {
8584
10
        xmlValidityError(ctxt, XML_DTD_NOT_STANDALONE,
8585
10
                         "standalone: normalization of attribute %s on %s "
8586
10
                         "by external subset declaration\n",
8587
10
                         name, elem);
8588
10
    }
8589
142k
#endif
8590
8591
142k
    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
26.8k
        if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) {
8598
15.5k
            internal_val = xmlStrndup(val, *len);
8599
15.5k
            if (internal_val == NULL)
8600
1
                goto mem_error;
8601
15.5k
            if (!xmlCheckLanguageID(internal_val)) {
8602
13.5k
                xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE,
8603
13.5k
                              "Malformed value for xml:lang : %s\n",
8604
13.5k
                              internal_val, NULL);
8605
13.5k
            }
8606
15.5k
        }
8607
8608
        /*
8609
         * Check that xml:space conforms to the specification
8610
         */
8611
26.8k
        if (xmlStrEqual(name, BAD_CAST "space")) {
8612
871
            internal_val = xmlStrndup(val, *len);
8613
871
            if (internal_val == NULL)
8614
1
                goto mem_error;
8615
870
            if (xmlStrEqual(internal_val, BAD_CAST "default"))
8616
218
                *(ctxt->space) = 0;
8617
652
            else if (xmlStrEqual(internal_val, BAD_CAST "preserve"))
8618
539
                *(ctxt->space) = 1;
8619
113
            else {
8620
113
                xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE,
8621
113
                              "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n",
8622
113
                              internal_val, NULL);
8623
113
            }
8624
870
        }
8625
26.8k
        if (internal_val) {
8626
16.4k
            xmlFree(internal_val);
8627
16.4k
        }
8628
26.8k
    }
8629
8630
142k
    *value = val;
8631
142k
    return (hname);
8632
8633
2
mem_error:
8634
2
    xmlErrMemory(ctxt);
8635
16.6k
error:
8636
16.6k
    if ((val != NULL) && (*alloc != 0))
8637
1
        xmlFree(val);
8638
16.6k
    return(hname);
8639
2
}
8640
8641
/**
8642
 * Inserts a new attribute into the hash table.
8643
 *
8644
 * @param ctxt  parser context
8645
 * @param size  size of the hash table
8646
 * @param name  attribute name
8647
 * @param uri  namespace uri
8648
 * @param hashValue  combined hash value of name and uri
8649
 * @param aindex  attribute index (this is a multiple of 5)
8650
 * @returns INT_MAX if no existing attribute was found, the attribute
8651
 * index if an attribute was found, -1 if a memory allocation failed.
8652
 */
8653
static int
8654
xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name,
8655
184k
                  const xmlChar *uri, unsigned hashValue, int aindex) {
8656
184k
    xmlAttrHashBucket *table = ctxt->attrHash;
8657
184k
    xmlAttrHashBucket *bucket;
8658
184k
    unsigned hindex;
8659
8660
184k
    hindex = hashValue & (size - 1);
8661
184k
    bucket = &table[hindex];
8662
8663
205k
    while (bucket->index >= 0) {
8664
49.8k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8665
8666
49.8k
        if (name == atts[0]) {
8667
29.8k
            int nsIndex = XML_PTR_TO_INT(atts[2]);
8668
8669
29.8k
            if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) :
8670
29.8k
                (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) :
8671
23.8k
                (uri == ctxt->nsTab[nsIndex * 2 + 1]))
8672
28.7k
                return(bucket->index);
8673
29.8k
        }
8674
8675
21.0k
        hindex++;
8676
21.0k
        bucket++;
8677
21.0k
        if (hindex >= size) {
8678
2.38k
            hindex = 0;
8679
2.38k
            bucket = table;
8680
2.38k
        }
8681
21.0k
    }
8682
8683
155k
    bucket->index = aindex;
8684
8685
155k
    return(INT_MAX);
8686
184k
}
8687
8688
static int
8689
xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size,
8690
                       const xmlChar *name, const xmlChar *prefix,
8691
3.12k
                       unsigned hashValue, int aindex) {
8692
3.12k
    xmlAttrHashBucket *table = ctxt->attrHash;
8693
3.12k
    xmlAttrHashBucket *bucket;
8694
3.12k
    unsigned hindex;
8695
8696
3.12k
    hindex = hashValue & (size - 1);
8697
3.12k
    bucket = &table[hindex];
8698
8699
3.79k
    while (bucket->index >= 0) {
8700
2.09k
        const xmlChar **atts = &ctxt->atts[bucket->index];
8701
8702
2.09k
        if ((name == atts[0]) && (prefix == atts[1]))
8703
1.42k
            return(bucket->index);
8704
8705
670
        hindex++;
8706
670
        bucket++;
8707
670
        if (hindex >= size) {
8708
215
            hindex = 0;
8709
215
            bucket = table;
8710
215
        }
8711
670
    }
8712
8713
1.70k
    bucket->index = aindex;
8714
8715
1.70k
    return(INT_MAX);
8716
3.12k
}
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
1.01M
                  const xmlChar **URI, int *nbNsPtr) {
8750
1.01M
    xmlHashedString hlocalname;
8751
1.01M
    xmlHashedString hprefix;
8752
1.01M
    xmlHashedString hattname;
8753
1.01M
    xmlHashedString haprefix;
8754
1.01M
    const xmlChar *localname;
8755
1.01M
    const xmlChar *prefix;
8756
1.01M
    const xmlChar *attname;
8757
1.01M
    const xmlChar *aprefix;
8758
1.01M
    const xmlChar *uri;
8759
1.01M
    xmlChar *attvalue = NULL;
8760
1.01M
    const xmlChar **atts = ctxt->atts;
8761
1.01M
    unsigned attrHashSize = 0;
8762
1.01M
    int maxatts = ctxt->maxatts;
8763
1.01M
    int nratts, nbatts, nbdef;
8764
1.01M
    int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts;
8765
1.01M
    int alloc = 0;
8766
1.01M
    int numNsErr = 0;
8767
1.01M
    int numDupErr = 0;
8768
8769
1.01M
    if (RAW != '<') return(NULL);
8770
1.01M
    NEXT1;
8771
8772
1.01M
    nbatts = 0;
8773
1.01M
    nratts = 0;
8774
1.01M
    nbdef = 0;
8775
1.01M
    nbNs = 0;
8776
1.01M
    nbTotalDef = 0;
8777
1.01M
    attval = 0;
8778
8779
1.01M
    if (xmlParserNsStartElement(ctxt->nsdb) < 0) {
8780
0
        xmlErrMemory(ctxt);
8781
0
        return(NULL);
8782
0
    }
8783
8784
1.01M
    hlocalname = xmlParseQNameHashed(ctxt, &hprefix);
8785
1.01M
    if (hlocalname.name == NULL) {
8786
362k
  xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED,
8787
362k
           "StartTag: invalid element name\n");
8788
362k
        return(NULL);
8789
362k
    }
8790
652k
    localname = hlocalname.name;
8791
652k
    prefix = hprefix.name;
8792
8793
    /*
8794
     * Now parse the attributes, it ends up with the ending
8795
     *
8796
     * (S Attribute)* S?
8797
     */
8798
652k
    SKIP_BLANKS;
8799
652k
    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
716k
    while (((RAW != '>') &&
8826
352k
     ((RAW != '/') || (NXT(1) != '>')) &&
8827
199k
     (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) {
8828
189k
  int len = -1;
8829
8830
189k
  hattname = xmlParseAttribute2(ctxt, prefix, localname,
8831
189k
                                          &haprefix, &attvalue, &len,
8832
189k
                                          &alloc);
8833
189k
        if (hattname.name == NULL)
8834
30.8k
      break;
8835
158k
        if (attvalue == NULL)
8836
16.6k
            goto next_attr;
8837
142k
        attname = hattname.name;
8838
142k
        aprefix = haprefix.name;
8839
142k
  if (len < 0) len = xmlStrlen(attvalue);
8840
8841
142k
        if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
8842
10.4k
            xmlHashedString huri;
8843
10.4k
            xmlURIPtr parsedUri;
8844
8845
10.4k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8846
10.4k
            uri = huri.name;
8847
10.4k
            if (uri == NULL) {
8848
1
                xmlErrMemory(ctxt);
8849
1
                goto next_attr;
8850
1
            }
8851
10.4k
            if (*uri != 0) {
8852
7.16k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8853
18
                    xmlErrMemory(ctxt);
8854
18
                    goto next_attr;
8855
18
                }
8856
7.14k
                if (parsedUri == NULL) {
8857
1.49k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8858
1.49k
                             "xmlns: '%s' is not a valid URI\n",
8859
1.49k
                                       uri, NULL, NULL);
8860
5.65k
                } else {
8861
5.65k
                    if (parsedUri->scheme == NULL) {
8862
3.68k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8863
3.68k
                                  "xmlns: URI %s is not absolute\n",
8864
3.68k
                                  uri, NULL, NULL);
8865
3.68k
                    }
8866
5.65k
                    xmlFreeURI(parsedUri);
8867
5.65k
                }
8868
7.14k
                if (uri == ctxt->str_xml_ns) {
8869
70
                    if (attname != ctxt->str_xml) {
8870
70
                        xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8871
70
                     "xml namespace URI cannot be the default namespace\n",
8872
70
                                 NULL, NULL, NULL);
8873
70
                    }
8874
70
                    goto next_attr;
8875
70
                }
8876
7.07k
                if ((len == 29) &&
8877
1.26k
                    (xmlStrEqual(uri,
8878
1.26k
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8879
492
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8880
492
                         "reuse of the xmlns namespace name is forbidden\n",
8881
492
                             NULL, NULL, NULL);
8882
492
                    goto next_attr;
8883
492
                }
8884
7.07k
            }
8885
8886
9.84k
            if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0)
8887
8.67k
                nbNs++;
8888
131k
        } else if (aprefix == ctxt->str_xmlns) {
8889
47.8k
            xmlHashedString huri;
8890
47.8k
            xmlURIPtr parsedUri;
8891
8892
47.8k
            huri = xmlDictLookupHashed(ctxt->dict, attvalue, len);
8893
47.8k
            uri = huri.name;
8894
47.8k
            if (uri == NULL) {
8895
1
                xmlErrMemory(ctxt);
8896
1
                goto next_attr;
8897
1
            }
8898
8899
47.8k
            if (attname == ctxt->str_xml) {
8900
198
                if (uri != ctxt->str_xml_ns) {
8901
198
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8902
198
                             "xml namespace prefix mapped to wrong URI\n",
8903
198
                             NULL, NULL, NULL);
8904
198
                }
8905
                /*
8906
                 * Do not keep a namespace definition node
8907
                 */
8908
198
                goto next_attr;
8909
198
            }
8910
47.6k
            if (uri == ctxt->str_xml_ns) {
8911
10
                if (attname != ctxt->str_xml) {
8912
10
                    xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8913
10
                             "xml namespace URI mapped to wrong prefix\n",
8914
10
                             NULL, NULL, NULL);
8915
10
                }
8916
10
                goto next_attr;
8917
10
            }
8918
47.6k
            if (attname == ctxt->str_xmlns) {
8919
266
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8920
266
                         "redefinition of the xmlns prefix is forbidden\n",
8921
266
                         NULL, NULL, NULL);
8922
266
                goto next_attr;
8923
266
            }
8924
47.4k
            if ((len == 29) &&
8925
602
                (xmlStrEqual(uri,
8926
602
                             BAD_CAST "http://www.w3.org/2000/xmlns/"))) {
8927
365
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8928
365
                         "reuse of the xmlns namespace name is forbidden\n",
8929
365
                         NULL, NULL, NULL);
8930
365
                goto next_attr;
8931
365
            }
8932
47.0k
            if ((uri == NULL) || (uri[0] == 0)) {
8933
228
                xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE,
8934
228
                         "xmlns:%s: Empty XML namespace is not allowed\n",
8935
228
                              attname, NULL, NULL);
8936
228
                goto next_attr;
8937
46.8k
            } else {
8938
46.8k
                if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) {
8939
2
                    xmlErrMemory(ctxt);
8940
2
                    goto next_attr;
8941
2
                }
8942
46.8k
                if (parsedUri == NULL) {
8943
3.36k
                    xmlNsErr(ctxt, XML_WAR_NS_URI,
8944
3.36k
                         "xmlns:%s: '%s' is not a valid URI\n",
8945
3.36k
                                       attname, uri, NULL);
8946
43.4k
                } else {
8947
43.4k
                    if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) {
8948
1.45k
                        xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE,
8949
1.45k
                                  "xmlns:%s: URI %s is not absolute\n",
8950
1.45k
                                  attname, uri, NULL);
8951
1.45k
                    }
8952
43.4k
                    xmlFreeURI(parsedUri);
8953
43.4k
                }
8954
46.8k
            }
8955
8956
46.8k
            if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0)
8957
44.9k
                nbNs++;
8958
83.7k
        } else {
8959
            /*
8960
             * Populate attributes array, see above for repurposing
8961
             * of xmlChar pointers.
8962
             */
8963
83.7k
            if ((atts == NULL) || (nbatts + 5 > maxatts)) {
8964
18.0k
                int res = xmlCtxtGrowAttrs(ctxt);
8965
8966
18.0k
                maxatts = ctxt->maxatts;
8967
18.0k
                atts = ctxt->atts;
8968
8969
18.0k
                if (res < 0)
8970
7
                    goto next_attr;
8971
18.0k
            }
8972
83.7k
            ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) |
8973
83.7k
                                        ((unsigned) alloc << 31);
8974
83.7k
            atts[nbatts++] = attname;
8975
83.7k
            atts[nbatts++] = aprefix;
8976
83.7k
            atts[nbatts++] = XML_INT_TO_PTR(haprefix.hashValue);
8977
83.7k
            if (alloc) {
8978
16.3k
                atts[nbatts++] = attvalue;
8979
16.3k
                attvalue += len;
8980
16.3k
                atts[nbatts++] = attvalue;
8981
67.3k
            } 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
67.3k
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8988
67.3k
                attvalue += len;
8989
67.3k
                atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR);
8990
67.3k
            }
8991
            /*
8992
             * tag if some deallocation is needed
8993
             */
8994
83.7k
            if (alloc != 0) attval = 1;
8995
83.7k
            attvalue = NULL; /* moved into atts */
8996
83.7k
        }
8997
8998
158k
next_attr:
8999
158k
        if ((attvalue != NULL) && (alloc != 0)) {
9000
3.83k
            xmlFree(attvalue);
9001
3.83k
            attvalue = NULL;
9002
3.83k
        }
9003
9004
158k
  GROW
9005
158k
  if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>'))))
9006
74.8k
      break;
9007
83.8k
  if (SKIP_BLANKS == 0) {
9008
20.5k
      xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
9009
20.5k
         "attributes construct error\n");
9010
20.5k
      break;
9011
20.5k
  }
9012
63.3k
        GROW;
9013
63.3k
    }
9014
9015
    /*
9016
     * Namespaces from default attributes
9017
     */
9018
652k
    if (ctxt->attsDefault != NULL) {
9019
126k
        xmlDefAttrsPtr defaults;
9020
9021
126k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9022
126k
  if (defaults != NULL) {
9023
350k
      for (i = 0; i < defaults->nbAttrs; i++) {
9024
265k
                xmlDefAttr *attr = &defaults->attrs[i];
9025
9026
265k
          attname = attr->name.name;
9027
265k
    aprefix = attr->prefix.name;
9028
9029
265k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) {
9030
15.9k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9031
9032
15.9k
                    if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0)
9033
14.9k
                        nbNs++;
9034
249k
    } else if (aprefix == ctxt->str_xmlns) {
9035
62.9k
                    xmlParserEntityCheck(ctxt, attr->expandedSize);
9036
9037
62.9k
                    if (xmlParserNsPush(ctxt, &attr->name, &attr->value,
9038
62.9k
                                      NULL, 1) > 0)
9039
58.8k
                        nbNs++;
9040
186k
    } else {
9041
186k
                    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
186k
                    nbTotalDef += 1;
9047
186k
                }
9048
265k
      }
9049
85.6k
  }
9050
126k
    }
9051
9052
    /*
9053
     * Resolve attribute namespaces
9054
     */
9055
736k
    for (i = 0; i < nbatts; i += 5) {
9056
83.7k
        attname = atts[i];
9057
83.7k
        aprefix = atts[i+1];
9058
9059
        /*
9060
  * The default namespace does not apply to attribute names.
9061
  */
9062
83.7k
  if (aprefix == NULL) {
9063
42.4k
            nsIndex = NS_INDEX_EMPTY;
9064
42.4k
        } else if (aprefix == ctxt->str_xml) {
9065
26.8k
            nsIndex = NS_INDEX_XML;
9066
26.8k
        } else {
9067
14.4k
            haprefix.name = aprefix;
9068
14.4k
            haprefix.hashValue = (size_t) atts[i+2];
9069
14.4k
            nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL);
9070
9071
14.4k
      if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) {
9072
10.0k
                xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9073
10.0k
        "Namespace prefix %s for %s on %s is not defined\n",
9074
10.0k
        aprefix, attname, localname);
9075
10.0k
                nsIndex = NS_INDEX_EMPTY;
9076
10.0k
            }
9077
14.4k
        }
9078
9079
83.7k
        atts[i+2] = XML_INT_TO_PTR(nsIndex);
9080
83.7k
    }
9081
9082
    /*
9083
     * Maximum number of attributes including default attributes.
9084
     */
9085
652k
    maxAtts = nratts + nbTotalDef;
9086
9087
    /*
9088
     * Verify that attribute names are unique.
9089
     */
9090
652k
    if (maxAtts > 1) {
9091
34.6k
        attrHashSize = 4;
9092
70.1k
        while (attrHashSize / 2 < (unsigned) maxAtts)
9093
35.5k
            attrHashSize *= 2;
9094
9095
34.6k
        if (attrHashSize > ctxt->attrHashMax) {
9096
5.52k
            xmlAttrHashBucket *tmp;
9097
9098
5.52k
            tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0]));
9099
5.52k
            if (tmp == NULL) {
9100
3
                xmlErrMemory(ctxt);
9101
3
                goto done;
9102
3
            }
9103
9104
5.52k
            ctxt->attrHash = tmp;
9105
5.52k
            ctxt->attrHashMax = attrHashSize;
9106
5.52k
        }
9107
9108
34.6k
        memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0]));
9109
9110
92.9k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9111
58.3k
            const xmlChar *nsuri;
9112
58.3k
            unsigned hashValue, nameHashValue, uriHashValue;
9113
58.3k
            int res;
9114
9115
58.3k
            attname = atts[i];
9116
58.3k
            aprefix = atts[i+1];
9117
58.3k
            nsIndex = XML_PTR_TO_INT(atts[i+2]);
9118
            /* Hash values always have bit 31 set, see dict.c */
9119
58.3k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9120
9121
58.3k
            if (nsIndex == NS_INDEX_EMPTY) {
9122
                /*
9123
                 * Prefix with empty namespace means an undeclared
9124
                 * prefix which was already reported above.
9125
                 */
9126
32.3k
                if (aprefix != NULL)
9127
6.40k
                    continue;
9128
25.9k
                nsuri = NULL;
9129
25.9k
                uriHashValue = URI_HASH_EMPTY;
9130
25.9k
            } else if (nsIndex == NS_INDEX_XML) {
9131
21.7k
                nsuri = ctxt->str_xml_ns;
9132
21.7k
                uriHashValue = URI_HASH_XML;
9133
21.7k
            } else {
9134
4.19k
                nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9135
4.19k
                uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9136
4.19k
            }
9137
9138
51.9k
            hashValue = xmlDictCombineHash(nameHashValue, uriHashValue);
9139
51.9k
            res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9140
51.9k
                                    hashValue, i);
9141
51.9k
            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
51.9k
            if (res < INT_MAX) {
9151
26.0k
                if (aprefix == atts[res+1]) {
9152
24.1k
                    xmlErrAttributeDup(ctxt, aprefix, attname);
9153
24.1k
                    numDupErr += 1;
9154
24.1k
                } else {
9155
1.93k
                    xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9156
1.93k
                             "Namespaced Attribute %s in '%s' redefined\n",
9157
1.93k
                             attname, nsuri, NULL);
9158
1.93k
                    numNsErr += 1;
9159
1.93k
                }
9160
26.0k
            }
9161
51.9k
        }
9162
34.6k
    }
9163
9164
    /*
9165
     * Default attributes
9166
     */
9167
652k
    if (ctxt->attsDefault != NULL) {
9168
126k
        xmlDefAttrsPtr defaults;
9169
9170
126k
  defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix);
9171
126k
  if (defaults != NULL) {
9172
350k
      for (i = 0; i < defaults->nbAttrs; i++) {
9173
265k
                xmlDefAttr *attr = &defaults->attrs[i];
9174
265k
                const xmlChar *nsuri = NULL;
9175
265k
                unsigned hashValue, uriHashValue = 0;
9176
265k
                int res;
9177
9178
265k
          attname = attr->name.name;
9179
265k
    aprefix = attr->prefix.name;
9180
9181
265k
    if ((attname == ctxt->str_xmlns) && (aprefix == NULL))
9182
15.9k
                    continue;
9183
249k
    if (aprefix == ctxt->str_xmlns)
9184
62.9k
                    continue;
9185
9186
186k
                if (aprefix == NULL) {
9187
102k
                    nsIndex = NS_INDEX_EMPTY;
9188
102k
                    nsuri = NULL;
9189
102k
                    uriHashValue = URI_HASH_EMPTY;
9190
102k
                } else if (aprefix == ctxt->str_xml) {
9191
18.9k
                    nsIndex = NS_INDEX_XML;
9192
18.9k
                    nsuri = ctxt->str_xml_ns;
9193
18.9k
                    uriHashValue = URI_HASH_XML;
9194
64.5k
                } else {
9195
64.5k
                    nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL);
9196
64.5k
                    if ((nsIndex == INT_MAX) ||
9197
54.0k
                        (nsIndex < ctxt->nsdb->minNsIndex)) {
9198
54.0k
                        xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9199
54.0k
                                 "Namespace prefix %s for %s on %s is not "
9200
54.0k
                                 "defined\n",
9201
54.0k
                                 aprefix, attname, localname);
9202
54.0k
                        nsIndex = NS_INDEX_EMPTY;
9203
54.0k
                        nsuri = NULL;
9204
54.0k
                        uriHashValue = URI_HASH_EMPTY;
9205
54.0k
                    } else {
9206
10.4k
                        nsuri = ctxt->nsTab[nsIndex * 2 + 1];
9207
10.4k
                        uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue;
9208
10.4k
                    }
9209
64.5k
                }
9210
9211
                /*
9212
                 * Check whether the attribute exists
9213
                 */
9214
186k
                if (maxAtts > 1) {
9215
132k
                    hashValue = xmlDictCombineHash(attr->name.hashValue,
9216
132k
                                                   uriHashValue);
9217
132k
                    res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri,
9218
132k
                                            hashValue, nbatts);
9219
132k
                    if (res < 0)
9220
0
                        continue;
9221
132k
                    if (res < INT_MAX) {
9222
2.70k
                        if (aprefix == atts[res+1])
9223
1.32k
                            continue;
9224
1.37k
                        xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
9225
1.37k
                                 "Namespaced Attribute %s in '%s' redefined\n",
9226
1.37k
                                 attname, nsuri, NULL);
9227
1.37k
                    }
9228
132k
                }
9229
9230
184k
                xmlParserEntityCheck(ctxt, attr->expandedSize);
9231
9232
184k
                if ((atts == NULL) || (nbatts + 5 > maxatts)) {
9233
1.93k
                    res = xmlCtxtGrowAttrs(ctxt);
9234
9235
1.93k
                    maxatts = ctxt->maxatts;
9236
1.93k
                    atts = ctxt->atts;
9237
9238
1.93k
                    if (res < 0) {
9239
3
                        localname = NULL;
9240
3
                        goto done;
9241
3
                    }
9242
1.93k
                }
9243
9244
184k
                atts[nbatts++] = attname;
9245
184k
                atts[nbatts++] = aprefix;
9246
184k
                atts[nbatts++] = XML_INT_TO_PTR(nsIndex);
9247
184k
                atts[nbatts++] = attr->value.name;
9248
184k
                atts[nbatts++] = attr->valueEnd;
9249
9250
184k
#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
184k
                if ((ctxt->validate) &&
9256
141k
                    (ctxt->standalone == 1) &&
9257
1.06k
                    (attr->external != 0)) {
9258
834
                    xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED,
9259
834
                            "standalone: attribute %s on %s defaulted "
9260
834
                            "from external subset\n",
9261
834
                            attname, localname);
9262
834
                }
9263
184k
#endif
9264
184k
                nbdef++;
9265
184k
      }
9266
85.6k
  }
9267
126k
    }
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
652k
    if ((numDupErr == 0) && (numNsErr > 1)) {
9282
563
        memset(ctxt->attrHash, -1,
9283
563
               attrHashSize * sizeof(ctxt->attrHash[0]));
9284
9285
4.16k
        for (i = 0, j = 0; j < nratts; i += 5, j++) {
9286
3.60k
            unsigned hashValue, nameHashValue, prefixHashValue;
9287
3.60k
            int res;
9288
9289
3.60k
            aprefix = atts[i+1];
9290
3.60k
            if (aprefix == NULL)
9291
475
                continue;
9292
9293
3.12k
            attname = atts[i];
9294
            /* Hash values always have bit 31 set, see dict.c */
9295
3.12k
            nameHashValue = ctxt->attallocs[j] | 0x80000000;
9296
3.12k
            prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix);
9297
9298
3.12k
            hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue);
9299
3.12k
            res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname,
9300
3.12k
                                         aprefix, hashValue, i);
9301
3.12k
            if (res < INT_MAX)
9302
1.42k
                xmlErrAttributeDup(ctxt, aprefix, attname);
9303
3.12k
        }
9304
563
    }
9305
9306
    /*
9307
     * Reconstruct attribute pointers
9308
     */
9309
921k
    for (i = 0, j = 0; i < nbatts; i += 5, j++) {
9310
        /* namespace URI */
9311
268k
        nsIndex = XML_PTR_TO_INT(atts[i+2]);
9312
268k
        if (nsIndex == INT_MAX)
9313
208k
            atts[i+2] = NULL;
9314
60.5k
        else if (nsIndex == INT_MAX - 1)
9315
45.5k
            atts[i+2] = ctxt->str_xml_ns;
9316
14.9k
        else
9317
14.9k
            atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1];
9318
9319
268k
        if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) {
9320
67.3k
            atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]);  /* value */
9321
67.3k
            atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]);  /* valuend */
9322
67.3k
        }
9323
268k
    }
9324
9325
652k
    uri = xmlParserNsLookupUri(ctxt, &hprefix);
9326
652k
    if ((prefix != NULL) && (uri == NULL)) {
9327
21.4k
  xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
9328
21.4k
           "Namespace prefix %s on %s is not defined\n",
9329
21.4k
     prefix, localname, NULL);
9330
21.4k
    }
9331
652k
    *pref = prefix;
9332
652k
    *URI = uri;
9333
9334
    /*
9335
     * SAX callback
9336
     */
9337
652k
    if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) &&
9338
652k
  (!ctxt->disableSAX)) {
9339
524k
  if (nbNs > 0)
9340
65.0k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9341
65.0k
                          nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs),
9342
65.0k
        nbatts / 5, nbdef, atts);
9343
459k
  else
9344
459k
      ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri,
9345
459k
                          0, NULL, nbatts / 5, nbdef, atts);
9346
524k
    }
9347
9348
652k
done:
9349
    /*
9350
     * Free allocated attribute values
9351
     */
9352
652k
    if (attval != 0) {
9353
47.6k
  for (i = 0, j = 0; j < nratts; i += 5, j++)
9354
38.6k
      if (ctxt->attallocs[j] & 0x80000000)
9355
16.3k
          xmlFree((xmlChar *) atts[i+3]);
9356
8.93k
    }
9357
9358
652k
    *nbNsPtr = nbNs;
9359
652k
    return(localname);
9360
652k
}
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
285k
xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) {
9376
285k
    const xmlChar *name;
9377
9378
285k
    GROW;
9379
285k
    if ((RAW != '<') || (NXT(1) != '/')) {
9380
34
  xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL);
9381
34
  return;
9382
34
    }
9383
285k
    SKIP(2);
9384
9385
285k
    if (tag->prefix == NULL)
9386
262k
        name = xmlParseNameAndCompare(ctxt, ctxt->name);
9387
22.9k
    else
9388
22.9k
        name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix);
9389
9390
    /*
9391
     * We should definitely be at the ending "S? '>'" part
9392
     */
9393
285k
    GROW;
9394
285k
    SKIP_BLANKS;
9395
285k
    if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) {
9396
22.1k
  xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL);
9397
22.1k
    } else
9398
262k
  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
285k
    if (name != (xmlChar*)1) {
9407
15.9k
        if (name == NULL) name = BAD_CAST "unparsable";
9408
15.9k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH,
9409
15.9k
         "Opening and ending tag mismatch: %s line %d and %s\n",
9410
15.9k
                    ctxt->name, tag->line, name);
9411
15.9k
    }
9412
9413
    /*
9414
     * SAX: End of Tag
9415
     */
9416
285k
    if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9417
285k
  (!ctxt->disableSAX))
9418
188k
  ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix,
9419
188k
                                tag->URI);
9420
9421
285k
    spacePop(ctxt);
9422
285k
    if (tag->nsNr != 0)
9423
16.7k
  xmlParserNsPop(ctxt, tag->nsNr);
9424
285k
}
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
110k
xmlParseCDSect(xmlParserCtxt *ctxt) {
9442
110k
    xmlChar *buf = NULL;
9443
110k
    int len = 0;
9444
110k
    int size = XML_PARSER_BUFFER_SIZE;
9445
110k
    int r, rl;
9446
110k
    int s, sl;
9447
110k
    int cur, l;
9448
110k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9449
56.5k
                    XML_MAX_HUGE_LENGTH :
9450
110k
                    XML_MAX_TEXT_LENGTH;
9451
9452
110k
    if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '['))
9453
0
        return;
9454
110k
    SKIP(3);
9455
9456
110k
    if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '['))
9457
0
        return;
9458
110k
    SKIP(6);
9459
9460
110k
    r = xmlCurrentCharRecover(ctxt, &rl);
9461
110k
    if (!IS_CHAR(r)) {
9462
420
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9463
420
        goto out;
9464
420
    }
9465
109k
    NEXTL(rl);
9466
109k
    s = xmlCurrentCharRecover(ctxt, &sl);
9467
109k
    if (!IS_CHAR(s)) {
9468
599
  xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL);
9469
599
        goto out;
9470
599
    }
9471
109k
    NEXTL(sl);
9472
109k
    cur = xmlCurrentCharRecover(ctxt, &l);
9473
109k
    buf = xmlMalloc(size);
9474
109k
    if (buf == NULL) {
9475
2
  xmlErrMemory(ctxt);
9476
2
        goto out;
9477
2
    }
9478
2.82M
    while (IS_CHAR(cur) &&
9479
2.82M
           ((r != ']') || (s != ']') || (cur != '>'))) {
9480
2.72M
  if (len + 5 >= size) {
9481
2.03k
      xmlChar *tmp;
9482
2.03k
            int newSize;
9483
9484
2.03k
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9485
2.03k
            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
2.03k
      tmp = xmlRealloc(buf, newSize);
9491
2.03k
      if (tmp == NULL) {
9492
1
    xmlErrMemory(ctxt);
9493
1
                goto out;
9494
1
      }
9495
2.02k
      buf = tmp;
9496
2.02k
      size = newSize;
9497
2.02k
  }
9498
2.72M
  COPY_BUF(buf, len, r);
9499
2.72M
  r = s;
9500
2.72M
  rl = sl;
9501
2.72M
  s = cur;
9502
2.72M
  sl = l;
9503
2.72M
  NEXTL(l);
9504
2.72M
  cur = xmlCurrentCharRecover(ctxt, &l);
9505
2.72M
    }
9506
109k
    buf[len] = 0;
9507
109k
    if (cur != '>') {
9508
2.97k
  xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED,
9509
2.97k
                       "CData section not finished\n%.50s\n", buf);
9510
2.97k
        goto out;
9511
2.97k
    }
9512
106k
    NEXTL(l);
9513
9514
    /*
9515
     * OK the buffer is to be consumed as cdata.
9516
     */
9517
106k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) {
9518
101k
        if ((ctxt->sax->cdataBlock != NULL) &&
9519
101k
            ((ctxt->options & XML_PARSE_NOCDATA) == 0)) {
9520
100k
            ctxt->sax->cdataBlock(ctxt->userData, buf, len);
9521
100k
        } else if (ctxt->sax->characters != NULL) {
9522
1.35k
            ctxt->sax->characters(ctxt->userData, buf, len);
9523
1.35k
        }
9524
101k
    }
9525
9526
110k
out:
9527
110k
    xmlFree(buf);
9528
110k
}
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
7.97k
xmlParseContentInternal(xmlParserCtxtPtr ctxt) {
9539
7.97k
    int oldNameNr = ctxt->nameNr;
9540
7.97k
    int oldSpaceNr = ctxt->spaceNr;
9541
7.97k
    int oldNodeNr = ctxt->nodeNr;
9542
9543
7.97k
    GROW;
9544
2.54M
    while ((ctxt->input->cur < ctxt->input->end) &&
9545
2.53M
     (PARSER_STOPPED(ctxt) == 0)) {
9546
2.53M
  const xmlChar *cur = ctxt->input->cur;
9547
9548
  /*
9549
   * First case : a Processing Instruction.
9550
   */
9551
2.53M
  if ((*cur == '<') && (cur[1] == '?')) {
9552
155k
      xmlParsePI(ctxt);
9553
155k
  }
9554
9555
  /*
9556
   * Second case : a CDSection
9557
   */
9558
  /* 2.6.0 test was *cur not RAW */
9559
2.38M
  else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) {
9560
79.0k
      xmlParseCDSect(ctxt);
9561
79.0k
  }
9562
9563
  /*
9564
   * Third case :  a comment
9565
   */
9566
2.30M
  else if ((*cur == '<') && (NXT(1) == '!') &&
9567
22.7k
     (NXT(2) == '-') && (NXT(3) == '-')) {
9568
2.43k
      xmlParseComment(ctxt);
9569
2.43k
  }
9570
9571
  /*
9572
   * Fourth case :  a sub-element.
9573
   */
9574
2.30M
  else if (*cur == '<') {
9575
1.33M
            if (NXT(1) == '/') {
9576
325k
                if (ctxt->nameNr <= oldNameNr)
9577
3.33k
                    break;
9578
321k
          xmlParseElementEnd(ctxt);
9579
1.01M
            } else {
9580
1.01M
          xmlParseElementStart(ctxt);
9581
1.01M
            }
9582
1.33M
  }
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
964k
  else if (*cur == '&') {
9590
76.7k
      xmlParseReference(ctxt);
9591
76.7k
  }
9592
9593
  /*
9594
   * Last case, text. Note that References are handled directly.
9595
   */
9596
887k
  else {
9597
887k
      xmlParseCharDataInternal(ctxt, 0);
9598
887k
  }
9599
9600
2.53M
  SHRINK;
9601
2.53M
  GROW;
9602
2.53M
    }
9603
9604
7.97k
    if ((ctxt->nameNr > oldNameNr) &&
9605
1.06k
        (ctxt->input->cur >= ctxt->input->end) &&
9606
939
        (ctxt->wellFormed)) {
9607
32
        const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9608
32
        int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9609
32
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9610
32
                "Premature end of data in tag %s line %d\n",
9611
32
                name, line, NULL);
9612
32
    }
9613
9614
    /*
9615
     * Clean up in error case
9616
     */
9617
9618
57.9k
    while (ctxt->nodeNr > oldNodeNr)
9619
49.9k
        nodePop(ctxt);
9620
9621
81.8k
    while (ctxt->nameNr > oldNameNr) {
9622
73.8k
        xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1];
9623
9624
73.8k
        if (tag->nsNr != 0)
9625
7.89k
            xmlParserNsPop(ctxt, tag->nsNr);
9626
9627
73.8k
        namePop(ctxt);
9628
73.8k
    }
9629
9630
81.8k
    while (ctxt->spaceNr > oldSpaceNr)
9631
73.8k
        spacePop(ctxt);
9632
7.97k
}
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
6.21k
xmlParseElement(xmlParserCtxt *ctxt) {
9669
6.21k
    if (xmlParseElementStart(ctxt) != 0)
9670
2.55k
        return;
9671
9672
3.66k
    xmlParseContentInternal(ctxt);
9673
9674
3.66k
    if (ctxt->input->cur >= ctxt->input->end) {
9675
301
        if (ctxt->wellFormed) {
9676
3
            const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
9677
3
            int line = ctxt->pushTab[ctxt->nameNr - 1].line;
9678
3
            xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
9679
3
                    "Premature end of data in tag %s line %d\n",
9680
3
                    name, line, NULL);
9681
3
        }
9682
301
        return;
9683
301
    }
9684
9685
3.36k
    xmlParseElementEnd(ctxt);
9686
3.36k
}
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
1.01M
xmlParseElementStart(xmlParserCtxtPtr ctxt) {
9698
1.01M
    int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256;
9699
1.01M
    const xmlChar *name;
9700
1.01M
    const xmlChar *prefix = NULL;
9701
1.01M
    const xmlChar *URI = NULL;
9702
1.01M
    xmlParserNodeInfo node_info;
9703
1.01M
    int line;
9704
1.01M
    xmlNodePtr cur;
9705
1.01M
    int nbNs = 0;
9706
9707
1.01M
    if (ctxt->nameNr > maxDepth) {
9708
5
        xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT,
9709
5
                "Excessive depth in document: %d use XML_PARSE_HUGE option\n",
9710
5
                ctxt->nameNr);
9711
5
  return(-1);
9712
5
    }
9713
9714
    /* Capture start position */
9715
1.01M
    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
1.01M
    if (ctxt->spaceNr == 0)
9722
0
  spacePush(ctxt, -1);
9723
1.01M
    else if (*ctxt->space == -2)
9724
215k
  spacePush(ctxt, -1);
9725
801k
    else
9726
801k
  spacePush(ctxt, *ctxt->space);
9727
9728
1.01M
    line = ctxt->input->line;
9729
1.01M
#ifdef LIBXML_SAX1_ENABLED
9730
1.01M
    if (ctxt->sax2)
9731
832k
#endif /* LIBXML_SAX1_ENABLED */
9732
832k
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
9733
185k
#ifdef LIBXML_SAX1_ENABLED
9734
185k
    else
9735
185k
  name = xmlParseStartTag(ctxt);
9736
1.01M
#endif /* LIBXML_SAX1_ENABLED */
9737
1.01M
    if (name == NULL) {
9738
370k
  spacePop(ctxt);
9739
370k
        return(-1);
9740
370k
    }
9741
647k
    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
9742
647k
    cur = ctxt->node;
9743
9744
647k
#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
647k
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
9751
6.76k
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
9752
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
9753
647k
#endif /* LIBXML_VALID_ENABLED */
9754
9755
    /*
9756
     * Check for an Empty Element.
9757
     */
9758
647k
    if ((RAW == '/') && (NXT(1) == '>')) {
9759
198k
        SKIP(2);
9760
198k
  if (ctxt->sax2) {
9761
130k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) &&
9762
130k
    (!ctxt->disableSAX))
9763
122k
    ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI);
9764
130k
#ifdef LIBXML_SAX1_ENABLED
9765
130k
  } else {
9766
67.5k
      if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) &&
9767
67.5k
    (!ctxt->disableSAX))
9768
67.2k
    ctxt->sax->endElement(ctxt->userData, name);
9769
67.5k
#endif /* LIBXML_SAX1_ENABLED */
9770
67.5k
  }
9771
198k
  namePop(ctxt);
9772
198k
  spacePop(ctxt);
9773
198k
  if (nbNs > 0)
9774
19.0k
      xmlParserNsPop(ctxt, nbNs);
9775
198k
  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
198k
  return(1);
9783
198k
    }
9784
449k
    if (RAW == '>') {
9785
399k
        NEXT1;
9786
399k
        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
399k
    } else {
9793
49.6k
        xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED,
9794
49.6k
         "Couldn't find end of Start Tag %s line %d\n",
9795
49.6k
                    name, line, NULL);
9796
9797
  /*
9798
   * end of parsing of this node.
9799
   */
9800
49.6k
  nodePop(ctxt);
9801
49.6k
  namePop(ctxt);
9802
49.6k
  spacePop(ctxt);
9803
49.6k
  if (nbNs > 0)
9804
4.40k
      xmlParserNsPop(ctxt, nbNs);
9805
49.6k
  return(-1);
9806
49.6k
    }
9807
9808
399k
    return(0);
9809
449k
}
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
325k
xmlParseElementEnd(xmlParserCtxtPtr ctxt) {
9818
325k
    xmlNodePtr cur = ctxt->node;
9819
9820
325k
    if (ctxt->nameNr <= 0) {
9821
4
        if ((RAW == '<') && (NXT(1) == '/'))
9822
2
            SKIP(2);
9823
4
        return;
9824
4
    }
9825
9826
    /*
9827
     * parse the end of tag: '</' should be here.
9828
     */
9829
325k
    if (ctxt->sax2) {
9830
260k
  xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
9831
260k
  namePop(ctxt);
9832
260k
    }
9833
64.5k
#ifdef LIBXML_SAX1_ENABLED
9834
64.5k
    else
9835
64.5k
  xmlParseEndTag1(ctxt, 0);
9836
325k
#endif /* LIBXML_SAX1_ENABLED */
9837
9838
    /*
9839
     * Capture end position
9840
     */
9841
325k
    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
325k
}
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
12.2k
xmlParseVersionNum(xmlParserCtxt *ctxt) {
9867
12.2k
    xmlChar *buf = NULL;
9868
12.2k
    int len = 0;
9869
12.2k
    int size = 10;
9870
12.2k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9871
5.37k
                    XML_MAX_TEXT_LENGTH :
9872
12.2k
                    XML_MAX_NAME_LENGTH;
9873
12.2k
    xmlChar cur;
9874
9875
12.2k
    buf = xmlMalloc(size);
9876
12.2k
    if (buf == NULL) {
9877
5
  xmlErrMemory(ctxt);
9878
5
  return(NULL);
9879
5
    }
9880
12.2k
    cur = CUR;
9881
12.2k
    if (!((cur >= '0') && (cur <= '9'))) {
9882
449
  xmlFree(buf);
9883
449
  return(NULL);
9884
449
    }
9885
11.7k
    buf[len++] = cur;
9886
11.7k
    NEXT;
9887
11.7k
    cur=CUR;
9888
11.7k
    if (cur != '.') {
9889
3.79k
  xmlFree(buf);
9890
3.79k
  return(NULL);
9891
3.79k
    }
9892
7.97k
    buf[len++] = cur;
9893
7.97k
    NEXT;
9894
7.97k
    cur=CUR;
9895
336k
    while ((cur >= '0') && (cur <= '9')) {
9896
328k
  if (len + 1 >= size) {
9897
446
      xmlChar *tmp;
9898
446
            int newSize;
9899
9900
446
            newSize = xmlGrowCapacity(size, 1, 1, maxLength);
9901
446
            if (newSize < 0) {
9902
1
                xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum");
9903
1
                xmlFree(buf);
9904
1
                return(NULL);
9905
1
            }
9906
445
      tmp = xmlRealloc(buf, newSize);
9907
445
      if (tmp == NULL) {
9908
1
    xmlErrMemory(ctxt);
9909
1
          xmlFree(buf);
9910
1
    return(NULL);
9911
1
      }
9912
444
      buf = tmp;
9913
444
            size = newSize;
9914
444
  }
9915
328k
  buf[len++] = cur;
9916
328k
  NEXT;
9917
328k
  cur=CUR;
9918
328k
    }
9919
7.97k
    buf[len] = 0;
9920
7.97k
    return(buf);
9921
7.97k
}
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
179k
xmlParseVersionInfo(xmlParserCtxt *ctxt) {
9938
179k
    xmlChar *version = NULL;
9939
9940
179k
    if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) {
9941
13.2k
  SKIP(7);
9942
13.2k
  SKIP_BLANKS;
9943
13.2k
  if (RAW != '=') {
9944
789
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
9945
789
      return(NULL);
9946
789
        }
9947
12.4k
  NEXT;
9948
12.4k
  SKIP_BLANKS;
9949
12.4k
  if (RAW == '"') {
9950
3.63k
      NEXT;
9951
3.63k
      version = xmlParseVersionNum(ctxt);
9952
3.63k
      if (RAW != '"') {
9953
627
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9954
627
      } else
9955
3.01k
          NEXT;
9956
8.81k
  } else if (RAW == '\''){
9957
8.59k
      NEXT;
9958
8.59k
      version = xmlParseVersionNum(ctxt);
9959
8.59k
      if (RAW != '\'') {
9960
4.00k
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
9961
4.00k
      } else
9962
4.58k
          NEXT;
9963
8.59k
  } else {
9964
222
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
9965
222
  }
9966
12.4k
    }
9967
178k
    return(version);
9968
179k
}
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
6.43k
xmlParseEncName(xmlParserCtxt *ctxt) {
9982
6.43k
    xmlChar *buf = NULL;
9983
6.43k
    int len = 0;
9984
6.43k
    int size = 10;
9985
6.43k
    int maxLength = (ctxt->options & XML_PARSE_HUGE) ?
9986
2.69k
                    XML_MAX_TEXT_LENGTH :
9987
6.43k
                    XML_MAX_NAME_LENGTH;
9988
6.43k
    xmlChar cur;
9989
9990
6.43k
    cur = CUR;
9991
6.43k
    if (((cur >= 'a') && (cur <= 'z')) ||
9992
5.89k
        ((cur >= 'A') && (cur <= 'Z'))) {
9993
5.89k
  buf = xmlMalloc(size);
9994
5.89k
  if (buf == NULL) {
9995
2
      xmlErrMemory(ctxt);
9996
2
      return(NULL);
9997
2
  }
9998
9999
5.89k
  buf[len++] = cur;
10000
5.89k
  NEXT;
10001
5.89k
  cur = CUR;
10002
126k
  while (((cur >= 'a') && (cur <= 'z')) ||
10003
75.0k
         ((cur >= 'A') && (cur <= 'Z')) ||
10004
25.2k
         ((cur >= '0') && (cur <= '9')) ||
10005
12.7k
         (cur == '.') || (cur == '_') ||
10006
120k
         (cur == '-')) {
10007
120k
      if (len + 1 >= size) {
10008
2.46k
          xmlChar *tmp;
10009
2.46k
                int newSize;
10010
10011
2.46k
                newSize = xmlGrowCapacity(size, 1, 1, maxLength);
10012
2.46k
                if (newSize < 0) {
10013
1
                    xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName");
10014
1
                    xmlFree(buf);
10015
1
                    return(NULL);
10016
1
                }
10017
2.46k
    tmp = xmlRealloc(buf, newSize);
10018
2.46k
    if (tmp == NULL) {
10019
2
        xmlErrMemory(ctxt);
10020
2
        xmlFree(buf);
10021
2
        return(NULL);
10022
2
    }
10023
2.46k
    buf = tmp;
10024
2.46k
                size = newSize;
10025
2.46k
      }
10026
120k
      buf[len++] = cur;
10027
120k
      NEXT;
10028
120k
      cur = CUR;
10029
120k
        }
10030
5.88k
  buf[len] = 0;
10031
5.88k
    } else {
10032
537
  xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL);
10033
537
    }
10034
6.42k
    return(buf);
10035
6.43k
}
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
177k
xmlParseEncodingDecl(xmlParserCtxt *ctxt) {
10053
177k
    xmlChar *encoding = NULL;
10054
10055
177k
    SKIP_BLANKS;
10056
177k
    if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0)
10057
170k
        return(NULL);
10058
10059
7.58k
    SKIP(8);
10060
7.58k
    SKIP_BLANKS;
10061
7.58k
    if (RAW != '=') {
10062
557
        xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10063
557
        return(NULL);
10064
557
    }
10065
7.03k
    NEXT;
10066
7.03k
    SKIP_BLANKS;
10067
7.03k
    if (RAW == '"') {
10068
1.31k
        NEXT;
10069
1.31k
        encoding = xmlParseEncName(ctxt);
10070
1.31k
        if (RAW != '"') {
10071
361
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10072
361
            xmlFree(encoding);
10073
361
            return(NULL);
10074
361
        } else
10075
949
            NEXT;
10076
5.72k
    } else if (RAW == '\''){
10077
5.12k
        NEXT;
10078
5.12k
        encoding = xmlParseEncName(ctxt);
10079
5.12k
        if (RAW != '\'') {
10080
524
            xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10081
524
            xmlFree(encoding);
10082
524
            return(NULL);
10083
524
        } else
10084
4.59k
            NEXT;
10085
5.12k
    } else {
10086
600
        xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10087
600
    }
10088
10089
6.14k
    if (encoding == NULL)
10090
606
        return(NULL);
10091
10092
5.53k
    xmlSetDeclaredEncoding(ctxt, encoding);
10093
10094
5.53k
    return(ctxt->encoding);
10095
6.14k
}
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
1.17k
xmlParseSDDecl(xmlParserCtxt *ctxt) {
10130
1.17k
    int standalone = -2;
10131
10132
1.17k
    SKIP_BLANKS;
10133
1.17k
    if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) {
10134
246
  SKIP(10);
10135
246
        SKIP_BLANKS;
10136
246
  if (RAW != '=') {
10137
1
      xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL);
10138
1
      return(standalone);
10139
1
        }
10140
245
  NEXT;
10141
245
  SKIP_BLANKS;
10142
245
        if (RAW == '\''){
10143
235
      NEXT;
10144
235
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10145
8
          standalone = 0;
10146
8
                SKIP(2);
10147
227
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10148
224
                 (NXT(2) == 's')) {
10149
223
          standalone = 1;
10150
223
    SKIP(3);
10151
223
            } else {
10152
4
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10153
4
      }
10154
235
      if (RAW != '\'') {
10155
14
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10156
14
      } else
10157
221
          NEXT;
10158
235
  } else if (RAW == '"'){
10159
9
      NEXT;
10160
9
      if ((RAW == 'n') && (NXT(1) == 'o')) {
10161
2
          standalone = 0;
10162
2
    SKIP(2);
10163
7
      } else if ((RAW == 'y') && (NXT(1) == 'e') &&
10164
3
                 (NXT(2) == 's')) {
10165
2
          standalone = 1;
10166
2
                SKIP(3);
10167
5
            } else {
10168
5
    xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL);
10169
5
      }
10170
9
      if (RAW != '"') {
10171
8
    xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL);
10172
8
      } else
10173
1
          NEXT;
10174
9
  } else {
10175
1
      xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL);
10176
1
        }
10177
245
    }
10178
1.17k
    return(standalone);
10179
1.17k
}
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
3.66k
xmlParseXMLDecl(xmlParserCtxt *ctxt) {
10192
3.66k
    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
3.66k
    ctxt->standalone = -2;
10201
10202
    /*
10203
     * We know that '<?xml' is here.
10204
     */
10205
3.66k
    SKIP(5);
10206
10207
3.66k
    if (!IS_BLANK_CH(RAW)) {
10208
0
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED,
10209
0
                 "Blank needed after '<?xml'\n");
10210
0
    }
10211
3.66k
    SKIP_BLANKS;
10212
10213
    /*
10214
     * We must have the VersionInfo here.
10215
     */
10216
3.66k
    version = xmlParseVersionInfo(ctxt);
10217
3.66k
    if (version == NULL) {
10218
810
  xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL);
10219
2.85k
    } else {
10220
2.85k
  if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) {
10221
      /*
10222
       * Changed here for XML-1.0 5th edition
10223
       */
10224
261
      if (ctxt->options & XML_PARSE_OLD10) {
10225
60
    xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10226
60
                "Unsupported version '%s'\n",
10227
60
                version);
10228
201
      } else {
10229
201
          if ((version[0] == '1') && ((version[1] == '.'))) {
10230
152
        xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION,
10231
152
                      "Unsupported version '%s'\n",
10232
152
          version, NULL);
10233
152
    } else {
10234
49
        xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION,
10235
49
              "Unsupported version '%s'\n",
10236
49
              version);
10237
49
    }
10238
201
      }
10239
261
  }
10240
2.85k
  if (ctxt->version != NULL)
10241
0
      xmlFree(ctxt->version);
10242
2.85k
  ctxt->version = version;
10243
2.85k
    }
10244
10245
    /*
10246
     * We may have the encoding declaration
10247
     */
10248
3.66k
    if (!IS_BLANK_CH(RAW)) {
10249
2.60k
        if ((RAW == '?') && (NXT(1) == '>')) {
10250
1.73k
      SKIP(2);
10251
1.73k
      return;
10252
1.73k
  }
10253
872
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10254
872
    }
10255
1.93k
    xmlParseEncodingDecl(ctxt);
10256
10257
    /*
10258
     * We may have the standalone status.
10259
     */
10260
1.93k
    if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) {
10261
826
        if ((RAW == '?') && (NXT(1) == '>')) {
10262
759
      SKIP(2);
10263
759
      return;
10264
759
  }
10265
67
  xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n");
10266
67
    }
10267
10268
    /*
10269
     * We can grow the input buffer freely at that point
10270
     */
10271
1.17k
    GROW;
10272
10273
1.17k
    SKIP_BLANKS;
10274
1.17k
    ctxt->standalone = xmlParseSDDecl(ctxt);
10275
10276
1.17k
    SKIP_BLANKS;
10277
1.17k
    if ((RAW == '?') && (NXT(1) == '>')) {
10278
274
        SKIP(2);
10279
901
    } else if (RAW == '>') {
10280
        /* Deprecated old WD ... */
10281
32
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10282
32
  NEXT;
10283
869
    } else {
10284
869
        int c;
10285
10286
869
  xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL);
10287
102k
        while ((PARSER_STOPPED(ctxt) == 0) &&
10288
102k
               ((c = CUR) != 0)) {
10289
102k
            NEXT;
10290
102k
            if (c == '>')
10291
462
                break;
10292
102k
        }
10293
869
    }
10294
1.17k
}
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
14.6k
xmlParseMisc(xmlParserCtxt *ctxt) {
10335
16.4k
    while (PARSER_STOPPED(ctxt) == 0) {
10336
16.3k
        SKIP_BLANKS;
10337
16.3k
        GROW;
10338
16.3k
        if ((RAW == '<') && (NXT(1) == '?')) {
10339
822
      xmlParsePI(ctxt);
10340
15.5k
        } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) {
10341
1.03k
      xmlParseComment(ctxt);
10342
14.5k
        } else {
10343
14.5k
            break;
10344
14.5k
        }
10345
16.3k
    }
10346
14.6k
}
10347
10348
static void
10349
20.5k
xmlFinishDocument(xmlParserCtxtPtr ctxt) {
10350
20.5k
    xmlDocPtr doc;
10351
10352
    /*
10353
     * SAX: end of the document processing.
10354
     */
10355
20.5k
    if ((ctxt->sax) && (ctxt->sax->endDocument != NULL))
10356
20.5k
        ctxt->sax->endDocument(ctxt->userData);
10357
10358
    /*
10359
     * Remove locally kept entity definitions if the tree was not built
10360
     */
10361
20.5k
    doc = ctxt->myDoc;
10362
20.5k
    if ((doc != NULL) &&
10363
20.4k
        (xmlStrEqual(doc->version, SAX_COMPAT_MODE))) {
10364
19
        xmlFreeDoc(doc);
10365
19
        ctxt->myDoc = NULL;
10366
19
    }
10367
20.5k
}
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
6.30k
xmlParseDocument(xmlParserCtxt *ctxt) {
10380
6.30k
    if ((ctxt == NULL) || (ctxt->input == NULL))
10381
0
        return(-1);
10382
10383
6.30k
    GROW;
10384
10385
    /*
10386
     * SAX: detecting the level.
10387
     */
10388
6.30k
    xmlCtxtInitializeLate(ctxt);
10389
10390
6.30k
    if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10391
6.30k
        ctxt->sax->setDocumentLocator(ctxt->userData,
10392
6.30k
                (xmlSAXLocator *) &xmlDefaultSAXLocator);
10393
6.30k
    }
10394
10395
6.30k
    xmlDetectEncoding(ctxt);
10396
10397
6.30k
    if (CUR == 0) {
10398
3
  xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
10399
3
  return(-1);
10400
3
    }
10401
10402
6.30k
    GROW;
10403
6.30k
    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
168
  xmlParseXMLDecl(ctxt);
10409
168
  SKIP_BLANKS;
10410
6.13k
    } else {
10411
6.13k
  ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10412
6.13k
        if (ctxt->version == NULL) {
10413
1
            xmlErrMemory(ctxt);
10414
1
            return(-1);
10415
1
        }
10416
6.13k
    }
10417
6.30k
    if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX))
10418
6.23k
        ctxt->sax->startDocument(ctxt->userData);
10419
6.30k
    if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) &&
10420
6.23k
        (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
6.30k
    xmlParseMisc(ctxt);
10428
10429
    /*
10430
     * Then possibly doc type declaration(s) and more Misc
10431
     * (doctypedecl Misc*)?
10432
     */
10433
6.30k
    GROW;
10434
6.30k
    if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) {
10435
10436
2.11k
  ctxt->inSubset = 1;
10437
2.11k
  xmlParseDocTypeDecl(ctxt);
10438
2.11k
  if (RAW == '[') {
10439
513
      xmlParseInternalSubset(ctxt);
10440
1.60k
  } else if (RAW == '>') {
10441
1.59k
            NEXT;
10442
1.59k
        }
10443
10444
  /*
10445
   * Create and update the external subset.
10446
   */
10447
2.11k
  ctxt->inSubset = 2;
10448
2.11k
  if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) &&
10449
2.11k
      (!ctxt->disableSAX))
10450
2.03k
      ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
10451
2.03k
                                ctxt->extSubSystem, ctxt->extSubURI);
10452
2.11k
  ctxt->inSubset = 0;
10453
10454
2.11k
        xmlCleanSpecialAttr(ctxt);
10455
10456
2.11k
  xmlParseMisc(ctxt);
10457
2.11k
    }
10458
10459
    /*
10460
     * Time to start parsing the tree itself
10461
     */
10462
6.30k
    GROW;
10463
6.30k
    if (RAW != '<') {
10464
85
        if (ctxt->wellFormed)
10465
4
            xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10466
4
                           "Start tag expected, '<' not found\n");
10467
6.21k
    } else {
10468
6.21k
  xmlParseElement(ctxt);
10469
10470
  /*
10471
   * The Misc part at the end
10472
   */
10473
6.21k
  xmlParseMisc(ctxt);
10474
10475
6.21k
        xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
10476
6.21k
    }
10477
10478
6.30k
    ctxt->instate = XML_PARSER_EOF;
10479
6.30k
    xmlFinishDocument(ctxt);
10480
10481
6.30k
    if (! ctxt->wellFormed) {
10482
562
  ctxt->valid = 0;
10483
562
  return(-1);
10484
562
    }
10485
10486
5.74k
    return(0);
10487
6.30k
}
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
52.2k
xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) {
10575
52.2k
    const xmlChar *cur;
10576
10577
52.2k
    if (ctxt->checkIndex == 0) {
10578
45.2k
        cur = ctxt->input->cur + 1;
10579
45.2k
    } else {
10580
6.98k
        cur = ctxt->input->cur + ctxt->checkIndex;
10581
6.98k
    }
10582
10583
52.2k
    if (memchr(cur, c, ctxt->input->end - cur) == NULL) {
10584
10.1k
        size_t index = ctxt->input->end - ctxt->input->cur;
10585
10586
10.1k
        if (index > LONG_MAX) {
10587
0
            ctxt->checkIndex = 0;
10588
0
            return(1);
10589
0
        }
10590
10.1k
        ctxt->checkIndex = index;
10591
10.1k
        return(0);
10592
42.1k
    } else {
10593
42.1k
        ctxt->checkIndex = 0;
10594
42.1k
        return(1);
10595
42.1k
    }
10596
52.2k
}
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
179k
                     const char *str, size_t strLen) {
10609
179k
    const xmlChar *cur, *term;
10610
10611
179k
    if (ctxt->checkIndex == 0) {
10612
148k
        cur = ctxt->input->cur + startDelta;
10613
148k
    } else {
10614
31.8k
        cur = ctxt->input->cur + ctxt->checkIndex;
10615
31.8k
    }
10616
10617
179k
    term = BAD_CAST strstr((const char *) cur, str);
10618
179k
    if (term == NULL) {
10619
33.7k
        const xmlChar *end = ctxt->input->end;
10620
33.7k
        size_t index;
10621
10622
        /* Rescan (strLen - 1) characters. */
10623
33.7k
        if ((size_t) (end - cur) < strLen)
10624
771
            end = cur;
10625
32.9k
        else
10626
32.9k
            end -= strLen - 1;
10627
33.7k
        index = end - ctxt->input->cur;
10628
33.7k
        if (index > LONG_MAX) {
10629
0
            ctxt->checkIndex = 0;
10630
0
            return(ctxt->input->end - strLen);
10631
0
        }
10632
33.7k
        ctxt->checkIndex = index;
10633
146k
    } else {
10634
146k
        ctxt->checkIndex = 0;
10635
146k
    }
10636
10637
179k
    return(term);
10638
179k
}
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
83.0k
xmlParseLookupCharData(xmlParserCtxtPtr ctxt) {
10647
83.0k
    const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex;
10648
83.0k
    const xmlChar *end = ctxt->input->end;
10649
83.0k
    size_t index;
10650
10651
1.16M
    while (cur < end) {
10652
1.15M
        if ((*cur == '<') || (*cur == '&')) {
10653
76.4k
            ctxt->checkIndex = 0;
10654
76.4k
            return(1);
10655
76.4k
        }
10656
1.07M
        cur++;
10657
1.07M
    }
10658
10659
6.60k
    index = cur - ctxt->input->cur;
10660
6.60k
    if (index > LONG_MAX) {
10661
0
        ctxt->checkIndex = 0;
10662
0
        return(1);
10663
0
    }
10664
6.60k
    ctxt->checkIndex = index;
10665
6.60k
    return(0);
10666
6.60k
}
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
433k
xmlParseLookupGt(xmlParserCtxtPtr ctxt) {
10676
433k
    const xmlChar *cur;
10677
433k
    const xmlChar *end = ctxt->input->end;
10678
433k
    int state = ctxt->endCheckState;
10679
433k
    size_t index;
10680
10681
433k
    if (ctxt->checkIndex == 0)
10682
259k
        cur = ctxt->input->cur + 1;
10683
174k
    else
10684
174k
        cur = ctxt->input->cur + ctxt->checkIndex;
10685
10686
124M
    while (cur < end) {
10687
124M
        if (state) {
10688
110M
            if (*cur == state)
10689
141k
                state = 0;
10690
110M
        } else if (*cur == '\'' || *cur == '"') {
10691
143k
            state = *cur;
10692
13.5M
        } else if (*cur == '>') {
10693
251k
            ctxt->checkIndex = 0;
10694
251k
            ctxt->endCheckState = 0;
10695
251k
            return(1);
10696
251k
        }
10697
124M
        cur++;
10698
124M
    }
10699
10700
182k
    index = cur - ctxt->input->cur;
10701
182k
    if (index > LONG_MAX) {
10702
0
        ctxt->checkIndex = 0;
10703
0
        ctxt->endCheckState = 0;
10704
0
        return(1);
10705
0
    }
10706
182k
    ctxt->checkIndex = index;
10707
182k
    ctxt->endCheckState = state;
10708
182k
    return(0);
10709
182k
}
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
32.1k
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
32.1k
    const xmlChar *cur, *start;
10727
32.1k
    const xmlChar *end = ctxt->input->end;
10728
32.1k
    int state = ctxt->endCheckState;
10729
32.1k
    size_t index;
10730
10731
32.1k
    if (ctxt->checkIndex == 0) {
10732
9.80k
        cur = ctxt->input->cur + 1;
10733
22.3k
    } else {
10734
22.3k
        cur = ctxt->input->cur + ctxt->checkIndex;
10735
22.3k
    }
10736
32.1k
    start = cur;
10737
10738
16.5M
    while (cur < end) {
10739
16.4M
        if (state == '-') {
10740
1.25M
            if ((*cur == '-') &&
10741
6.61k
                (cur[1] == '-') &&
10742
5.39k
                (cur[2] == '>')) {
10743
4.27k
                state = 0;
10744
4.27k
                cur += 3;
10745
4.27k
                start = cur;
10746
4.27k
                continue;
10747
4.27k
            }
10748
1.25M
        }
10749
15.2M
        else if (state == ']') {
10750
9.12k
            if (*cur == '>') {
10751
6.79k
                ctxt->checkIndex = 0;
10752
6.79k
                ctxt->endCheckState = 0;
10753
6.79k
                return(1);
10754
6.79k
            }
10755
2.32k
            if (IS_BLANK_CH(*cur)) {
10756
1.36k
                state = ' ';
10757
1.36k
            } else if (*cur != ']') {
10758
593
                state = 0;
10759
593
                start = cur;
10760
593
                continue;
10761
593
            }
10762
2.32k
        }
10763
15.2M
        else if (state == ' ') {
10764
6.23k
            if (*cur == '>') {
10765
192
                ctxt->checkIndex = 0;
10766
192
                ctxt->endCheckState = 0;
10767
192
                return(1);
10768
192
            }
10769
6.03k
            if (!IS_BLANK_CH(*cur)) {
10770
1.16k
                state = 0;
10771
1.16k
                start = cur;
10772
1.16k
                continue;
10773
1.16k
            }
10774
6.03k
        }
10775
15.2M
        else if (state != 0) {
10776
11.5M
            if (*cur == state) {
10777
30.6k
                state = 0;
10778
30.6k
                start = cur + 1;
10779
30.6k
            }
10780
11.5M
        }
10781
3.64M
        else if (*cur == '<') {
10782
38.0k
            if ((cur[1] == '!') &&
10783
31.1k
                (cur[2] == '-') &&
10784
4.55k
                (cur[3] == '-')) {
10785
4.32k
                state = '-';
10786
4.32k
                cur += 4;
10787
                /* Don't treat <!--> as comment */
10788
4.32k
                start = cur;
10789
4.32k
                continue;
10790
4.32k
            }
10791
38.0k
        }
10792
3.61M
        else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) {
10793
39.7k
            state = *cur;
10794
39.7k
        }
10795
10796
16.4M
        cur++;
10797
16.4M
    }
10798
10799
    /*
10800
     * Rescan the three last characters to detect "<!--" and "-->"
10801
     * split across chunks.
10802
     */
10803
25.2k
    if ((state == 0) || (state == '-')) {
10804
7.08k
        if (cur - start < 3)
10805
361
            cur = start;
10806
6.72k
        else
10807
6.72k
            cur -= 3;
10808
7.08k
    }
10809
25.2k
    index = cur - ctxt->input->cur;
10810
25.2k
    if (index > LONG_MAX) {
10811
0
        ctxt->checkIndex = 0;
10812
0
        ctxt->endCheckState = 0;
10813
0
        return(1);
10814
0
    }
10815
25.2k
    ctxt->checkIndex = index;
10816
25.2k
    ctxt->endCheckState = state;
10817
25.2k
    return(0);
10818
25.2k
}
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
301k
xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) {
10829
301k
    int ret = 0;
10830
301k
    size_t avail;
10831
301k
    xmlChar cur, next;
10832
10833
301k
    if (ctxt->input == NULL)
10834
0
        return(0);
10835
10836
301k
    if ((ctxt->input != NULL) &&
10837
301k
        (ctxt->input->cur - ctxt->input->base > 4096)) {
10838
1.88k
        xmlParserShrink(ctxt);
10839
1.88k
    }
10840
10841
1.54M
    while (ctxt->disableSAX == 0) {
10842
1.53M
        avail = ctxt->input->end - ctxt->input->cur;
10843
1.53M
        if (avail < 1)
10844
19.5k
      goto done;
10845
1.51M
        switch (ctxt->instate) {
10846
4.16k
            case XML_PARSER_EOF:
10847
          /*
10848
     * Document parsing is done !
10849
     */
10850
4.16k
          goto done;
10851
33.0k
            case XML_PARSER_START:
10852
                /*
10853
                 * Very first chars read from the document flow.
10854
                 */
10855
33.0k
                if ((!terminate) && (avail < 4))
10856
416
                    goto done;
10857
10858
                /*
10859
                 * We need more bytes to detect EBCDIC code pages.
10860
                 * See xmlDetectEBCDIC.
10861
                 */
10862
32.6k
                if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) &&
10863
712
                    (!terminate) && (avail < 200))
10864
342
                    goto done;
10865
10866
32.3k
                xmlDetectEncoding(ctxt);
10867
32.3k
                ctxt->instate = XML_PARSER_XML_DECL;
10868
32.3k
    break;
10869
10870
40.7k
            case XML_PARSER_XML_DECL:
10871
40.7k
    if ((!terminate) && (avail < 2))
10872
12
        goto done;
10873
40.7k
    cur = ctxt->input->cur[0];
10874
40.7k
    next = ctxt->input->cur[1];
10875
40.7k
          if ((cur == '<') && (next == '?')) {
10876
        /* PI or XML decl */
10877
17.2k
        if ((!terminate) &&
10878
15.9k
                        (!xmlParseLookupString(ctxt, 2, "?>", 2)))
10879
8.43k
      goto done;
10880
8.79k
        if ((ctxt->input->cur[2] == 'x') &&
10881
7.44k
      (ctxt->input->cur[3] == 'm') &&
10882
6.59k
      (ctxt->input->cur[4] == 'l') &&
10883
3.57k
      (IS_BLANK_CH(ctxt->input->cur[5]))) {
10884
3.49k
      ret += 5;
10885
3.49k
      xmlParseXMLDecl(ctxt);
10886
5.29k
        } else {
10887
5.29k
      ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10888
5.29k
                        if (ctxt->version == NULL) {
10889
3
                            xmlErrMemory(ctxt);
10890
3
                            break;
10891
3
                        }
10892
5.29k
        }
10893
23.5k
    } else {
10894
23.5k
        ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION);
10895
23.5k
        if (ctxt->version == NULL) {
10896
5
            xmlErrMemory(ctxt);
10897
5
      break;
10898
5
        }
10899
23.5k
    }
10900
32.2k
                if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) {
10901
32.2k
                    ctxt->sax->setDocumentLocator(ctxt->userData,
10902
32.2k
                            (xmlSAXLocator *) &xmlDefaultSAXLocator);
10903
32.2k
                }
10904
32.2k
                if ((ctxt->sax) && (ctxt->sax->startDocument) &&
10905
32.2k
                    (!ctxt->disableSAX))
10906
31.9k
                    ctxt->sax->startDocument(ctxt->userData);
10907
32.2k
                ctxt->instate = XML_PARSER_MISC;
10908
32.2k
    break;
10909
465k
            case XML_PARSER_START_TAG: {
10910
465k
          const xmlChar *name;
10911
465k
    const xmlChar *prefix = NULL;
10912
465k
    const xmlChar *URI = NULL;
10913
465k
                int line = ctxt->input->line;
10914
465k
    int nbNs = 0;
10915
10916
465k
    if ((!terminate) && (avail < 2))
10917
89
        goto done;
10918
464k
    cur = ctxt->input->cur[0];
10919
464k
          if (cur != '<') {
10920
1.88k
        xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
10921
1.88k
                                   "Start tag expected, '<' not found");
10922
1.88k
                    ctxt->instate = XML_PARSER_EOF;
10923
1.88k
                    xmlFinishDocument(ctxt);
10924
1.88k
        goto done;
10925
1.88k
    }
10926
463k
    if ((!terminate) && (!xmlParseLookupGt(ctxt)))
10927
170k
                    goto done;
10928
292k
    if (ctxt->spaceNr == 0)
10929
0
        spacePush(ctxt, -1);
10930
292k
    else if (*ctxt->space == -2)
10931
71.7k
        spacePush(ctxt, -1);
10932
220k
    else
10933
220k
        spacePush(ctxt, *ctxt->space);
10934
292k
#ifdef LIBXML_SAX1_ENABLED
10935
292k
    if (ctxt->sax2)
10936
183k
#endif /* LIBXML_SAX1_ENABLED */
10937
183k
        name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs);
10938
108k
#ifdef LIBXML_SAX1_ENABLED
10939
108k
    else
10940
108k
        name = xmlParseStartTag(ctxt);
10941
292k
#endif /* LIBXML_SAX1_ENABLED */
10942
292k
    if (name == NULL) {
10943
1.51k
        spacePop(ctxt);
10944
1.51k
                    ctxt->instate = XML_PARSER_EOF;
10945
1.51k
                    xmlFinishDocument(ctxt);
10946
1.51k
        goto done;
10947
1.51k
    }
10948
290k
#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
290k
    if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc &&
10955
97.1k
        ctxt->node && (ctxt->node == ctxt->myDoc->children))
10956
0
        ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
10957
290k
#endif /* LIBXML_VALID_ENABLED */
10958
10959
    /*
10960
     * Check for an Empty Element.
10961
     */
10962
290k
    if ((RAW == '/') && (NXT(1) == '>')) {
10963
90.1k
        SKIP(2);
10964
10965
90.1k
        if (ctxt->sax2) {
10966
53.3k
      if ((ctxt->sax != NULL) &&
10967
53.3k
          (ctxt->sax->endElementNs != NULL) &&
10968
53.3k
          (!ctxt->disableSAX))
10969
53.2k
          ctxt->sax->endElementNs(ctxt->userData, name,
10970
53.2k
                                  prefix, URI);
10971
53.3k
      if (nbNs > 0)
10972
4.05k
          xmlParserNsPop(ctxt, nbNs);
10973
53.3k
#ifdef LIBXML_SAX1_ENABLED
10974
53.3k
        } else {
10975
36.7k
      if ((ctxt->sax != NULL) &&
10976
36.7k
          (ctxt->sax->endElement != NULL) &&
10977
36.7k
          (!ctxt->disableSAX))
10978
36.7k
          ctxt->sax->endElement(ctxt->userData, name);
10979
36.7k
#endif /* LIBXML_SAX1_ENABLED */
10980
36.7k
        }
10981
90.1k
        spacePop(ctxt);
10982
200k
    } else if (RAW == '>') {
10983
160k
        NEXT;
10984
160k
                    nameNsPush(ctxt, name, prefix, URI, line, nbNs);
10985
160k
    } else {
10986
40.2k
        xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED,
10987
40.2k
           "Couldn't find end of Start Tag %s\n",
10988
40.2k
           name);
10989
40.2k
        nodePop(ctxt);
10990
40.2k
        spacePop(ctxt);
10991
40.2k
                    if (nbNs > 0)
10992
4.26k
                        xmlParserNsPop(ctxt, nbNs);
10993
40.2k
    }
10994
10995
290k
                if (ctxt->nameNr == 0)
10996
7.18k
                    ctxt->instate = XML_PARSER_EPILOG;
10997
283k
                else
10998
283k
                    ctxt->instate = XML_PARSER_CONTENT;
10999
290k
                break;
11000
292k
      }
11001
763k
            case XML_PARSER_CONTENT: {
11002
763k
    cur = ctxt->input->cur[0];
11003
11004
763k
    if (cur == '<') {
11005
406k
                    if ((!terminate) && (avail < 2))
11006
1.13k
                        goto done;
11007
405k
        next = ctxt->input->cur[1];
11008
11009
405k
                    if (next == '/') {
11010
35.5k
                        ctxt->instate = XML_PARSER_END_TAG;
11011
35.5k
                        break;
11012
369k
                    } else if (next == '?') {
11013
44.9k
                        if ((!terminate) &&
11014
39.8k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11015
7.13k
                            goto done;
11016
37.8k
                        xmlParsePI(ctxt);
11017
37.8k
                        ctxt->instate = XML_PARSER_CONTENT;
11018
37.8k
                        break;
11019
324k
                    } else if (next == '!') {
11020
57.1k
                        if ((!terminate) && (avail < 3))
11021
112
                            goto done;
11022
57.0k
                        next = ctxt->input->cur[2];
11023
11024
57.0k
                        if (next == '-') {
11025
17.0k
                            if ((!terminate) && (avail < 4))
11026
34
                                goto done;
11027
17.0k
                            if (ctxt->input->cur[3] == '-') {
11028
17.0k
                                if ((!terminate) &&
11029
10.4k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11030
764
                                    goto done;
11031
16.2k
                                xmlParseComment(ctxt);
11032
16.2k
                                ctxt->instate = XML_PARSER_CONTENT;
11033
16.2k
                                break;
11034
17.0k
                            }
11035
39.9k
                        } else if (next == '[') {
11036
39.7k
                            if ((!terminate) && (avail < 9))
11037
357
                                goto done;
11038
39.3k
                            if ((ctxt->input->cur[2] == '[') &&
11039
39.3k
                                (ctxt->input->cur[3] == 'C') &&
11040
39.3k
                                (ctxt->input->cur[4] == 'D') &&
11041
39.3k
                                (ctxt->input->cur[5] == 'A') &&
11042
39.3k
                                (ctxt->input->cur[6] == 'T') &&
11043
39.3k
                                (ctxt->input->cur[7] == 'A') &&
11044
39.3k
                                (ctxt->input->cur[8] == '[')) {
11045
39.3k
                                if ((!terminate) &&
11046
35.6k
                                    (!xmlParseLookupString(ctxt, 9, "]]>", 3)))
11047
8.33k
                                    goto done;
11048
30.9k
                                ctxt->instate = XML_PARSER_CDATA_SECTION;
11049
30.9k
                                xmlParseCDSect(ctxt);
11050
30.9k
                                ctxt->instate = XML_PARSER_CONTENT;
11051
30.9k
                                break;
11052
39.3k
                            }
11053
39.3k
                        }
11054
57.0k
                    }
11055
405k
    } else if (cur == '&') {
11056
39.2k
        if ((!terminate) && (!xmlParseLookupChar(ctxt, ';')))
11057
6.88k
      goto done;
11058
32.3k
        xmlParseReference(ctxt);
11059
32.3k
                    break;
11060
317k
    } 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
317k
        if (avail < XML_PARSER_BIG_BUFFER_SIZE) {
11074
101k
      if ((!terminate) && (!xmlParseLookupCharData(ctxt)))
11075
6.60k
          goto done;
11076
101k
                    }
11077
311k
                    ctxt->checkIndex = 0;
11078
311k
        xmlParseCharDataInternal(ctxt, !terminate);
11079
311k
                    break;
11080
317k
    }
11081
11082
267k
                ctxt->instate = XML_PARSER_START_TAG;
11083
267k
    break;
11084
763k
      }
11085
37.7k
            case XML_PARSER_END_TAG:
11086
37.7k
    if ((!terminate) && (!xmlParseLookupChar(ctxt, '>')))
11087
3.23k
        goto done;
11088
34.5k
    if (ctxt->sax2) {
11089
24.4k
              xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]);
11090
24.4k
        nameNsPop(ctxt);
11091
24.4k
    }
11092
10.0k
#ifdef LIBXML_SAX1_ENABLED
11093
10.0k
      else
11094
10.0k
        xmlParseEndTag1(ctxt, 0);
11095
34.5k
#endif /* LIBXML_SAX1_ENABLED */
11096
34.5k
    if (ctxt->nameNr == 0) {
11097
2.44k
        ctxt->instate = XML_PARSER_EPILOG;
11098
32.1k
    } else {
11099
32.1k
        ctxt->instate = XML_PARSER_CONTENT;
11100
32.1k
    }
11101
34.5k
    break;
11102
74.0k
            case XML_PARSER_MISC:
11103
133k
            case XML_PARSER_PROLOG:
11104
139k
            case XML_PARSER_EPILOG:
11105
139k
    SKIP_BLANKS;
11106
139k
                avail = ctxt->input->end - ctxt->input->cur;
11107
139k
    if (avail < 1)
11108
3.11k
        goto done;
11109
135k
    if (ctxt->input->cur[0] == '<') {
11110
132k
                    if ((!terminate) && (avail < 2))
11111
185
                        goto done;
11112
132k
                    next = ctxt->input->cur[1];
11113
132k
                    if (next == '?') {
11114
29.4k
                        if ((!terminate) &&
11115
28.2k
                            (!xmlParseLookupString(ctxt, 2, "?>", 2)))
11116
1.80k
                            goto done;
11117
27.6k
                        xmlParsePI(ctxt);
11118
27.6k
                        break;
11119
102k
                    } else if (next == '!') {
11120
77.1k
                        if ((!terminate) && (avail < 3))
11121
122
                            goto done;
11122
11123
77.0k
                        if (ctxt->input->cur[2] == '-') {
11124
50.4k
                            if ((!terminate) && (avail < 4))
11125
106
                                goto done;
11126
50.3k
                            if (ctxt->input->cur[3] == '-') {
11127
50.2k
                                if ((!terminate) &&
11128
49.8k
                                    (!xmlParseLookupString(ctxt, 4, "-->", 3)))
11129
7.28k
                                    goto done;
11130
43.0k
                                xmlParseComment(ctxt);
11131
43.0k
                                break;
11132
50.2k
                            }
11133
50.3k
                        } else if (ctxt->instate == XML_PARSER_MISC) {
11134
26.5k
                            if ((!terminate) && (avail < 9))
11135
21
                                goto done;
11136
26.5k
                            if ((ctxt->input->cur[2] == 'D') &&
11137
26.5k
                                (ctxt->input->cur[3] == 'O') &&
11138
26.5k
                                (ctxt->input->cur[4] == 'C') &&
11139
26.5k
                                (ctxt->input->cur[5] == 'T') &&
11140
26.4k
                                (ctxt->input->cur[6] == 'Y') &&
11141
26.4k
                                (ctxt->input->cur[7] == 'P') &&
11142
26.4k
                                (ctxt->input->cur[8] == 'E')) {
11143
26.4k
                                if ((!terminate) && (!xmlParseLookupGt(ctxt)))
11144
11.6k
                                    goto done;
11145
14.8k
                                ctxt->inSubset = 1;
11146
14.8k
                                xmlParseDocTypeDecl(ctxt);
11147
14.8k
                                if (RAW == '[') {
11148
11.0k
                                    ctxt->instate = XML_PARSER_DTD;
11149
11.0k
                                } else {
11150
3.73k
                                    if (RAW == '>')
11151
3.52k
                                        NEXT;
11152
                                    /*
11153
                                     * Create and update the external subset.
11154
                                     */
11155
3.73k
                                    ctxt->inSubset = 2;
11156
3.73k
                                    if ((ctxt->sax != NULL) &&
11157
3.73k
                                        (!ctxt->disableSAX) &&
11158
3.61k
                                        (ctxt->sax->externalSubset != NULL))
11159
3.61k
                                        ctxt->sax->externalSubset(
11160
3.61k
                                                ctxt->userData,
11161
3.61k
                                                ctxt->intSubName,
11162
3.61k
                                                ctxt->extSubSystem,
11163
3.61k
                                                ctxt->extSubURI);
11164
3.73k
                                    ctxt->inSubset = 0;
11165
3.73k
                                    xmlCleanSpecialAttr(ctxt);
11166
3.73k
                                    ctxt->instate = XML_PARSER_PROLOG;
11167
3.73k
                                }
11168
14.8k
                                break;
11169
26.4k
                            }
11170
26.5k
                        }
11171
77.0k
                    }
11172
132k
                }
11173
11174
29.2k
                if (ctxt->instate == XML_PARSER_EPILOG) {
11175
1.74k
                    if (ctxt->errNo == XML_ERR_OK)
11176
28
                        xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL);
11177
1.74k
        ctxt->instate = XML_PARSER_EOF;
11178
1.74k
                    xmlFinishDocument(ctxt);
11179
27.4k
                } else {
11180
27.4k
        ctxt->instate = XML_PARSER_START_TAG;
11181
27.4k
    }
11182
29.2k
    break;
11183
36.2k
            case XML_PARSER_DTD: {
11184
36.2k
                if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt)))
11185
25.2k
                    goto done;
11186
11.0k
    xmlParseInternalSubset(ctxt);
11187
11.0k
    ctxt->inSubset = 2;
11188
11.0k
    if ((ctxt->sax != NULL) && (!ctxt->disableSAX) &&
11189
8.95k
        (ctxt->sax->externalSubset != NULL))
11190
8.95k
        ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName,
11191
8.95k
          ctxt->extSubSystem, ctxt->extSubURI);
11192
11.0k
    ctxt->inSubset = 0;
11193
11.0k
    xmlCleanSpecialAttr(ctxt);
11194
11.0k
    ctxt->instate = XML_PARSER_PROLOG;
11195
11.0k
                break;
11196
36.2k
      }
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
1.51M
  }
11203
1.51M
    }
11204
301k
done:
11205
301k
    return(ret);
11206
301k
}
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
304k
              int terminate) {
11237
304k
    size_t curBase;
11238
304k
    size_t maxLength;
11239
304k
    size_t pos;
11240
304k
    int end_in_lf = 0;
11241
304k
    int res;
11242
11243
304k
    if ((ctxt == NULL) || (size < 0))
11244
0
        return(XML_ERR_ARGUMENT);
11245
304k
    if ((chunk == NULL) && (size > 0))
11246
0
        return(XML_ERR_ARGUMENT);
11247
304k
    if ((ctxt->input == NULL) || (ctxt->input->buf == NULL))
11248
0
        return(XML_ERR_ARGUMENT);
11249
304k
    if (ctxt->disableSAX != 0)
11250
2.84k
        return(ctxt->errNo);
11251
11252
301k
    ctxt->input->flags |= XML_INPUT_PROGRESSIVE;
11253
301k
    if (ctxt->instate == XML_PARSER_START)
11254
33.1k
        xmlCtxtInitializeLate(ctxt);
11255
301k
    if ((size > 0) && (chunk != NULL) && (!terminate) &&
11256
275k
        (chunk[size - 1] == '\r')) {
11257
4.63k
  end_in_lf = 1;
11258
4.63k
  size--;
11259
4.63k
    }
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
301k
    pos = ctxt->input->cur - ctxt->input->base;
11266
301k
    res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
11267
301k
    xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11268
301k
    if (res < 0) {
11269
223
        xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11270
223
        return(ctxt->errNo);
11271
223
    }
11272
11273
301k
    xmlParseTryOrFinish(ctxt, terminate);
11274
11275
301k
    curBase = ctxt->input->cur - ctxt->input->base;
11276
301k
    maxLength = (ctxt->options & XML_PARSE_HUGE) ?
11277
127k
                XML_MAX_HUGE_LENGTH :
11278
301k
                XML_MAX_LOOKUP_LIMIT;
11279
301k
    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
301k
    if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0))
11285
10.4k
        return(ctxt->errNo);
11286
11287
290k
    if (end_in_lf == 1) {
11288
4.63k
  pos = ctxt->input->cur - ctxt->input->base;
11289
4.63k
  res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r");
11290
4.63k
  xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos);
11291
4.63k
        if (res < 0) {
11292
3
            xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL);
11293
3
            return(ctxt->errNo);
11294
3
        }
11295
4.63k
    }
11296
290k
    if (terminate) {
11297
  /*
11298
   * Check for termination
11299
   */
11300
17.8k
        if ((ctxt->instate != XML_PARSER_EOF) &&
11301
9.29k
            (ctxt->instate != XML_PARSER_EPILOG)) {
11302
4.74k
            if (ctxt->nameNr > 0) {
11303
3.55k
                const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1];
11304
3.55k
                int line = ctxt->pushTab[ctxt->nameNr - 1].line;
11305
3.55k
                xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED,
11306
3.55k
                        "Premature end of data in tag %s line %d\n",
11307
3.55k
                        name, line, NULL);
11308
3.55k
            } else if (ctxt->instate == XML_PARSER_START) {
11309
23
                xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL);
11310
1.17k
            } else {
11311
1.17k
                xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY,
11312
1.17k
                               "Start tag expected, '<' not found\n");
11313
1.17k
            }
11314
13.1k
        } else {
11315
13.1k
            xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END);
11316
13.1k
        }
11317
17.8k
  if (ctxt->instate != XML_PARSER_EOF) {
11318
9.07k
            ctxt->instate = XML_PARSER_EOF;
11319
9.07k
            xmlFinishDocument(ctxt);
11320
9.07k
  }
11321
17.8k
    }
11322
290k
    if (ctxt->wellFormed == 0)
11323
20.2k
  return((xmlParserErrors) ctxt->errNo);
11324
270k
    else
11325
270k
        return(0);
11326
290k
}
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
32.9k
                        const char *chunk, int size, const char *filename) {
11358
32.9k
    xmlParserCtxtPtr ctxt;
11359
32.9k
    xmlParserInputPtr input;
11360
11361
32.9k
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11362
32.9k
    if (ctxt == NULL)
11363
24
  return(NULL);
11364
11365
32.9k
    ctxt->options &= ~XML_PARSE_NODICT;
11366
32.9k
    ctxt->dictNames = 1;
11367
11368
32.9k
    input = xmlNewPushInput(filename, chunk, size);
11369
32.9k
    if (input == NULL) {
11370
13
  xmlFreeParserCtxt(ctxt);
11371
13
  return(NULL);
11372
13
    }
11373
32.9k
    if (xmlCtxtPushInput(ctxt, input) < 0) {
11374
0
        xmlFreeInputStream(input);
11375
0
        xmlFreeParserCtxt(ctxt);
11376
0
        return(NULL);
11377
0
    }
11378
11379
32.9k
    return(ctxt);
11380
32.9k
}
11381
#endif /* LIBXML_PUSH_ENABLED */
11382
11383
/**
11384
 * Blocks further parser processing
11385
 *
11386
 * @param ctxt  an XML parser context
11387
 */
11388
void
11389
35.0k
xmlStopParser(xmlParserCtxt *ctxt) {
11390
35.0k
    if (ctxt == NULL)
11391
0
        return;
11392
11393
    /* This stops the parser */
11394
35.0k
    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
35.0k
    if (ctxt->errNo == XML_ERR_OK) {
11403
3.18k
        ctxt->errNo = XML_ERR_USER_STOP;
11404
3.18k
        ctxt->lastError.code = XML_ERR_USER_STOP;
11405
3.18k
        ctxt->wellFormed = 0;
11406
3.18k
    }
11407
35.0k
}
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
4.31k
                            int hasTextDecl, int buildTree) {
11669
4.31k
    xmlNodePtr root = NULL;
11670
4.31k
    xmlNodePtr list = NULL;
11671
4.31k
    xmlChar *rootName = BAD_CAST "#root";
11672
4.31k
    int result;
11673
11674
4.31k
    if (buildTree) {
11675
4.31k
        root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL);
11676
4.31k
        if (root == NULL) {
11677
1
            xmlErrMemory(ctxt);
11678
1
            goto error;
11679
1
        }
11680
4.31k
    }
11681
11682
4.31k
    if (xmlCtxtPushInput(ctxt, input) < 0)
11683
2
        goto error;
11684
11685
4.31k
    nameNsPush(ctxt, rootName, NULL, NULL, 0, 0);
11686
4.31k
    spacePush(ctxt, -1);
11687
11688
4.31k
    if (buildTree)
11689
4.31k
        nodePush(ctxt, root);
11690
11691
4.31k
    if (hasTextDecl) {
11692
1.65k
        xmlDetectEncoding(ctxt);
11693
11694
        /*
11695
         * Parse a possible text declaration first
11696
         */
11697
1.65k
        if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) &&
11698
395
            (IS_BLANK_CH(NXT(5)))) {
11699
390
            xmlParseTextDecl(ctxt);
11700
            /*
11701
             * An XML-1.0 document can't reference an entity not XML-1.0
11702
             */
11703
390
            if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) &&
11704
385
                (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) {
11705
77
                xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH,
11706
77
                               "Version mismatch between document and "
11707
77
                               "entity\n");
11708
77
            }
11709
390
        }
11710
1.65k
    }
11711
11712
4.31k
    xmlParseContentInternal(ctxt);
11713
11714
4.31k
    if (ctxt->input->cur < ctxt->input->end)
11715
333
  xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL);
11716
11717
4.31k
    if ((ctxt->wellFormed) ||
11718
3.59k
        ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) {
11719
3.59k
        if (root != NULL) {
11720
3.59k
            xmlNodePtr cur;
11721
11722
            /*
11723
             * Unlink newly created node list.
11724
             */
11725
3.59k
            list = root->children;
11726
3.59k
            root->children = NULL;
11727
3.59k
            root->last = NULL;
11728
32.2k
            for (cur = list; cur != NULL; cur = cur->next)
11729
28.6k
                cur->parent = NULL;
11730
3.59k
        }
11731
3.59k
    }
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
4.59k
    do {
11738
4.59k
        ctxt->input->cur = ctxt->input->end;
11739
4.59k
        xmlParserShrink(ctxt);
11740
4.59k
        result = xmlParserGrow(ctxt);
11741
4.59k
    } while (result > 0);
11742
11743
4.31k
    if (buildTree)
11744
4.31k
        nodePop(ctxt);
11745
11746
4.31k
    namePop(ctxt);
11747
4.31k
    spacePop(ctxt);
11748
11749
4.31k
    xmlCtxtPopInput(ctxt);
11750
11751
4.31k
error:
11752
4.31k
    xmlFreeNode(root);
11753
11754
4.31k
    return(list);
11755
4.31k
}
11756
11757
static void
11758
6.28k
xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
11759
6.28k
    xmlParserInputPtr input;
11760
6.28k
    xmlNodePtr list;
11761
6.28k
    unsigned long consumed;
11762
6.28k
    int isExternal;
11763
6.28k
    int buildTree;
11764
6.28k
    int oldMinNsIndex;
11765
6.28k
    int oldNodelen, oldNodemem;
11766
11767
6.28k
    isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY);
11768
6.28k
    buildTree = (ctxt->node != NULL);
11769
11770
    /*
11771
     * Recursion check
11772
     */
11773
6.28k
    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
6.27k
    input = xmlNewEntityInputStream(ctxt, ent);
11782
6.27k
    if (input == NULL)
11783
1.96k
        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
4.31k
    oldMinNsIndex = ctxt->nsdb->minNsIndex;
11791
4.31k
    if (buildTree)
11792
4.31k
        ctxt->nsdb->minNsIndex = ctxt->nsNr;
11793
11794
4.31k
    oldNodelen = ctxt->nodelen;
11795
4.31k
    oldNodemem = ctxt->nodemem;
11796
4.31k
    ctxt->nodelen = 0;
11797
4.31k
    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
4.31k
    ent->flags |= XML_ENT_EXPANDING;
11817
11818
4.31k
    list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree);
11819
11820
4.31k
    ent->flags &= ~XML_ENT_EXPANDING;
11821
11822
4.31k
    ctxt->nsdb->minNsIndex = oldMinNsIndex;
11823
4.31k
    ctxt->nodelen = oldNodelen;
11824
4.31k
    ctxt->nodemem = oldNodemem;
11825
11826
    /*
11827
     * Entity size accounting
11828
     */
11829
4.31k
    consumed = input->consumed;
11830
4.31k
    xmlSaturatedAddSizeT(&consumed, input->end - input->base);
11831
11832
4.31k
    if ((ent->flags & XML_ENT_CHECKED) == 0)
11833
2.97k
        xmlSaturatedAdd(&ent->expandedSize, consumed);
11834
11835
4.31k
    if ((ent->flags & XML_ENT_PARSED) == 0) {
11836
2.97k
        if (isExternal)
11837
1.26k
            xmlSaturatedAdd(&ctxt->sizeentities, consumed);
11838
11839
2.97k
        ent->children = list;
11840
11841
31.6k
        while (list != NULL) {
11842
28.6k
            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
28.6k
            if (list->doc != ent->doc)
11850
0
                xmlSetTreeDoc(list, ent->doc);
11851
11852
28.6k
            if (list->next == NULL)
11853
2.14k
                ent->last = list;
11854
28.6k
            list = list->next;
11855
28.6k
        }
11856
2.97k
    } else {
11857
1.33k
        xmlFreeNodeList(list);
11858
1.33k
    }
11859
11860
4.31k
    xmlFreeInputStream(input);
11861
11862
6.28k
error:
11863
6.28k
    ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED;
11864
6.28k
}
11865
11866
/**
11867
 * Parse an external general entity within an existing parsing context
11868
 * An external general parsed entity is well-formed if it matches the
11869
 * production labeled extParsedEnt.
11870
 *
11871
 *     [78] extParsedEnt ::= TextDecl? content
11872
 *
11873
 * @param ctxt  the existing parsing context
11874
 * @param URL  the URL for the entity to load
11875
 * @param ID  the System ID for the entity to load
11876
 * @param listOut  the return value for the set of parsed nodes
11877
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11878
 *    the parser error code otherwise
11879
 */
11880
11881
int
11882
xmlParseCtxtExternalEntity(xmlParserCtxt *ctxt, const xmlChar *URL,
11883
0
                           const xmlChar *ID, xmlNode **listOut) {
11884
0
    xmlParserInputPtr input;
11885
0
    xmlNodePtr list;
11886
11887
0
    if (listOut != NULL)
11888
0
        *listOut = NULL;
11889
11890
0
    if (ctxt == NULL)
11891
0
        return(XML_ERR_ARGUMENT);
11892
11893
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
11894
0
                            XML_RESOURCE_GENERAL_ENTITY);
11895
0
    if (input == NULL)
11896
0
        return(ctxt->errNo);
11897
11898
0
    xmlCtxtInitializeLate(ctxt);
11899
11900
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1);
11901
0
    if (listOut != NULL)
11902
0
        *listOut = list;
11903
0
    else
11904
0
        xmlFreeNodeList(list);
11905
11906
0
    xmlFreeInputStream(input);
11907
0
    return(ctxt->errNo);
11908
0
}
11909
11910
#ifdef LIBXML_SAX1_ENABLED
11911
/**
11912
 * Parse an external general entity
11913
 * An external general parsed entity is well-formed if it matches the
11914
 * production labeled extParsedEnt.
11915
 *
11916
 * This function uses deprecated global variables to set parser options
11917
 * which default to XML_PARSE_NODICT.
11918
 *
11919
 * @deprecated Use #xmlParseCtxtExternalEntity.
11920
 *
11921
 *     [78] extParsedEnt ::= TextDecl? content
11922
 *
11923
 * @param doc  the document the chunk pertains to
11924
 * @param sax  the SAX handler block (possibly NULL)
11925
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11926
 * @param depth  Used for loop detection, use 0
11927
 * @param URL  the URL for the entity to load
11928
 * @param ID  the System ID for the entity to load
11929
 * @param list  the return value for the set of parsed nodes
11930
 * @returns 0 if the entity is well formed, -1 in case of args problem and
11931
 *    the parser error code otherwise
11932
 */
11933
11934
int
11935
xmlParseExternalEntity(xmlDoc *doc, xmlSAXHandler *sax, void *user_data,
11936
0
    int depth, const xmlChar *URL, const xmlChar *ID, xmlNode **list) {
11937
0
    xmlParserCtxtPtr ctxt;
11938
0
    int ret;
11939
11940
0
    if (list != NULL)
11941
0
        *list = NULL;
11942
11943
0
    if (doc == NULL)
11944
0
        return(XML_ERR_ARGUMENT);
11945
11946
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
11947
0
    if (ctxt == NULL)
11948
0
        return(XML_ERR_NO_MEMORY);
11949
11950
0
    ctxt->depth = depth;
11951
0
    ctxt->myDoc = doc;
11952
0
    ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list);
11953
11954
0
    xmlFreeParserCtxt(ctxt);
11955
0
    return(ret);
11956
0
}
11957
11958
/**
11959
 * Parse a well-balanced chunk of an XML document
11960
 * called by the parser
11961
 * The allowed sequence for the Well Balanced Chunk is the one defined by
11962
 * the content production in the XML grammar:
11963
 *
11964
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
11965
 *                       Comment)*
11966
 *
11967
 * This function uses deprecated global variables to set parser options
11968
 * which default to XML_PARSE_NODICT.
11969
 *
11970
 * @param doc  the document the chunk pertains to (must not be NULL)
11971
 * @param sax  the SAX handler block (possibly NULL)
11972
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
11973
 * @param depth  Used for loop detection, use 0
11974
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
11975
 * @param lst  the return value for the set of parsed nodes
11976
 * @returns 0 if the chunk is well balanced, -1 in case of args problem and
11977
 *    the parser error code otherwise
11978
 */
11979
11980
int
11981
xmlParseBalancedChunkMemory(xmlDoc *doc, xmlSAXHandler *sax,
11982
0
     void *user_data, int depth, const xmlChar *string, xmlNode **lst) {
11983
0
    return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data,
11984
0
                                                depth, string, lst, 0 );
11985
0
}
11986
#endif /* LIBXML_SAX1_ENABLED */
11987
11988
/**
11989
 * Parse a well-balanced chunk of XML matching the 'content' production.
11990
 *
11991
 * Namespaces in scope of `node` and entities of `node`'s document are
11992
 * recognized. When validating, the DTD of `node`'s document is used.
11993
 *
11994
 * Always consumes `input` even in error case.
11995
 *
11996
 * @since 2.14.0
11997
 *
11998
 * @param ctxt  parser context
11999
 * @param input  parser input
12000
 * @param node  target node or document
12001
 * @param hasTextDecl  whether to parse text declaration
12002
 * @returns a node list or NULL in case of error.
12003
 */
12004
xmlNode *
12005
xmlCtxtParseContent(xmlParserCtxt *ctxt, xmlParserInput *input,
12006
0
                    xmlNode *node, int hasTextDecl) {
12007
0
    xmlDocPtr doc;
12008
0
    xmlNodePtr cur, list = NULL;
12009
0
    int nsnr = 0;
12010
0
    xmlDictPtr oldDict;
12011
0
    int oldOptions, oldDictNames, oldLoadSubset;
12012
12013
0
    if ((ctxt == NULL) || (input == NULL) || (node == NULL)) {
12014
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12015
0
        goto exit;
12016
0
    }
12017
12018
0
    doc = node->doc;
12019
0
    if (doc == NULL) {
12020
0
        xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12021
0
        goto exit;
12022
0
    }
12023
12024
0
    switch (node->type) {
12025
0
        case XML_ELEMENT_NODE:
12026
0
        case XML_DOCUMENT_NODE:
12027
0
        case XML_HTML_DOCUMENT_NODE:
12028
0
            break;
12029
12030
0
        case XML_ATTRIBUTE_NODE:
12031
0
        case XML_TEXT_NODE:
12032
0
        case XML_CDATA_SECTION_NODE:
12033
0
        case XML_ENTITY_REF_NODE:
12034
0
        case XML_PI_NODE:
12035
0
        case XML_COMMENT_NODE:
12036
0
            for (cur = node->parent; cur != NULL; cur = node->parent) {
12037
0
                if ((cur->type == XML_ELEMENT_NODE) ||
12038
0
                    (cur->type == XML_DOCUMENT_NODE) ||
12039
0
                    (cur->type == XML_HTML_DOCUMENT_NODE)) {
12040
0
                    node = cur;
12041
0
                    break;
12042
0
                }
12043
0
            }
12044
0
            break;
12045
12046
0
        default:
12047
0
            xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL);
12048
0
            goto exit;
12049
0
    }
12050
12051
0
    xmlCtxtReset(ctxt);
12052
12053
0
    oldDict = ctxt->dict;
12054
0
    oldOptions = ctxt->options;
12055
0
    oldDictNames = ctxt->dictNames;
12056
0
    oldLoadSubset = ctxt->loadsubset;
12057
12058
    /*
12059
     * Use input doc's dict if present, else assure XML_PARSE_NODICT is set.
12060
     */
12061
0
    if (doc->dict != NULL) {
12062
0
        ctxt->dict = doc->dict;
12063
0
    } else {
12064
0
        ctxt->options |= XML_PARSE_NODICT;
12065
0
        ctxt->dictNames = 0;
12066
0
    }
12067
12068
    /*
12069
     * Disable IDs
12070
     */
12071
0
    ctxt->loadsubset |= XML_SKIP_IDS;
12072
0
    ctxt->options |= XML_PARSE_SKIP_IDS;
12073
12074
0
    ctxt->myDoc = doc;
12075
12076
0
#ifdef LIBXML_HTML_ENABLED
12077
0
    if (ctxt->html) {
12078
        /*
12079
         * When parsing in context, it makes no sense to add implied
12080
         * elements like html/body/etc...
12081
         */
12082
0
        ctxt->options |= HTML_PARSE_NOIMPLIED;
12083
12084
0
        list = htmlCtxtParseContentInternal(ctxt, input);
12085
0
    } else
12086
0
#endif
12087
0
    {
12088
0
        xmlCtxtInitializeLate(ctxt);
12089
12090
        /*
12091
         * initialize the SAX2 namespaces stack
12092
         */
12093
0
        cur = node;
12094
0
        while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) {
12095
0
            xmlNsPtr ns = cur->nsDef;
12096
0
            xmlHashedString hprefix, huri;
12097
12098
0
            while (ns != NULL) {
12099
0
                hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1);
12100
0
                huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1);
12101
0
                if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0)
12102
0
                    nsnr++;
12103
0
                ns = ns->next;
12104
0
            }
12105
0
            cur = cur->parent;
12106
0
        }
12107
12108
0
        list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1);
12109
12110
0
        if (nsnr > 0)
12111
0
            xmlParserNsPop(ctxt, nsnr);
12112
0
    }
12113
12114
0
    ctxt->dict = oldDict;
12115
0
    ctxt->options = oldOptions;
12116
0
    ctxt->dictNames = oldDictNames;
12117
0
    ctxt->loadsubset = oldLoadSubset;
12118
0
    ctxt->myDoc = NULL;
12119
0
    ctxt->node = NULL;
12120
12121
0
exit:
12122
0
    xmlFreeInputStream(input);
12123
0
    return(list);
12124
0
}
12125
12126
/**
12127
 * Parse a well-balanced chunk of an XML document
12128
 * within the context (DTD, namespaces, etc ...) of the given node.
12129
 *
12130
 * The allowed sequence for the data is a Well Balanced Chunk defined by
12131
 * the content production in the XML grammar:
12132
 *
12133
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12134
 *                       Comment)*
12135
 *
12136
 * This function assumes the encoding of `node`'s document which is
12137
 * typically not what you want. A better alternative is
12138
 * #xmlCtxtParseContent.
12139
 *
12140
 * @param node  the context node
12141
 * @param data  the input string
12142
 * @param datalen  the input string length in bytes
12143
 * @param options  a combination of xmlParserOption
12144
 * @param listOut  the return value for the set of parsed nodes
12145
 * @returns XML_ERR_OK if the chunk is well balanced, and the parser
12146
 * error code otherwise
12147
 */
12148
xmlParserErrors
12149
xmlParseInNodeContext(xmlNode *node, const char *data, int datalen,
12150
0
                      int options, xmlNode **listOut) {
12151
0
    xmlParserCtxtPtr ctxt;
12152
0
    xmlParserInputPtr input;
12153
0
    xmlDocPtr doc;
12154
0
    xmlNodePtr list;
12155
0
    xmlParserErrors ret;
12156
12157
0
    if (listOut == NULL)
12158
0
        return(XML_ERR_INTERNAL_ERROR);
12159
0
    *listOut = NULL;
12160
12161
0
    if ((node == NULL) || (data == NULL) || (datalen < 0))
12162
0
        return(XML_ERR_INTERNAL_ERROR);
12163
12164
0
    doc = node->doc;
12165
0
    if (doc == NULL)
12166
0
        return(XML_ERR_INTERNAL_ERROR);
12167
12168
0
#ifdef LIBXML_HTML_ENABLED
12169
0
    if (doc->type == XML_HTML_DOCUMENT_NODE) {
12170
0
        ctxt = htmlNewParserCtxt();
12171
0
    }
12172
0
    else
12173
0
#endif
12174
0
        ctxt = xmlNewParserCtxt();
12175
12176
0
    if (ctxt == NULL)
12177
0
        return(XML_ERR_NO_MEMORY);
12178
12179
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen,
12180
0
                                      (const char *) doc->encoding,
12181
0
                                      XML_INPUT_BUF_STATIC);
12182
0
    if (input == NULL) {
12183
0
        xmlFreeParserCtxt(ctxt);
12184
0
        return(XML_ERR_NO_MEMORY);
12185
0
    }
12186
12187
0
    xmlCtxtUseOptions(ctxt, options);
12188
12189
0
    list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0);
12190
12191
0
    if (list == NULL) {
12192
0
        ret = ctxt->errNo;
12193
0
        if (ret == XML_ERR_ARGUMENT)
12194
0
            ret = XML_ERR_INTERNAL_ERROR;
12195
0
    } else {
12196
0
        ret = XML_ERR_OK;
12197
0
        *listOut = list;
12198
0
    }
12199
12200
0
    xmlFreeParserCtxt(ctxt);
12201
12202
0
    return(ret);
12203
0
}
12204
12205
#ifdef LIBXML_SAX1_ENABLED
12206
/**
12207
 * Parse a well-balanced chunk of an XML document
12208
 *
12209
 * The allowed sequence for the Well Balanced Chunk is the one defined by
12210
 * the content production in the XML grammar:
12211
 *
12212
 *     [43] content ::= (element | CharData | Reference | CDSect | PI |
12213
 *                       Comment)*
12214
 *
12215
 * In case recover is set to 1, the nodelist will not be empty even if
12216
 * the parsed chunk is not well balanced, assuming the parsing succeeded to
12217
 * some extent.
12218
 *
12219
 * This function uses deprecated global variables to set parser options
12220
 * which default to XML_PARSE_NODICT.
12221
 *
12222
 * @param doc  the document the chunk pertains to (must not be NULL)
12223
 * @param sax  the SAX handler block (possibly NULL)
12224
 * @param user_data  The user data returned on SAX callbacks (possibly NULL)
12225
 * @param depth  Used for loop detection, use 0
12226
 * @param string  the input string in UTF8 or ISO-Latin (zero terminated)
12227
 * @param listOut  the return value for the set of parsed nodes
12228
 * @param recover  return nodes even if the data is broken (use 0)
12229
 * @returns 0 if the chunk is well balanced, or thehe parser error code
12230
 * otherwise.
12231
 */
12232
int
12233
xmlParseBalancedChunkMemoryRecover(xmlDoc *doc, xmlSAXHandler *sax,
12234
     void *user_data, int depth, const xmlChar *string, xmlNode **listOut,
12235
0
     int recover) {
12236
0
    xmlParserCtxtPtr ctxt;
12237
0
    xmlParserInputPtr input;
12238
0
    xmlNodePtr list;
12239
0
    int ret;
12240
12241
0
    if (listOut != NULL)
12242
0
        *listOut = NULL;
12243
12244
0
    if (string == NULL)
12245
0
        return(XML_ERR_ARGUMENT);
12246
12247
0
    ctxt = xmlNewSAXParserCtxt(sax, user_data);
12248
0
    if (ctxt == NULL)
12249
0
        return(XML_ERR_NO_MEMORY);
12250
12251
0
    xmlCtxtInitializeLate(ctxt);
12252
12253
0
    ctxt->depth = depth;
12254
0
    ctxt->myDoc = doc;
12255
0
    if (recover) {
12256
0
        ctxt->options |= XML_PARSE_RECOVER;
12257
0
        ctxt->recovery = 1;
12258
0
    }
12259
12260
0
    input = xmlNewStringInputStream(ctxt, string);
12261
0
    if (input == NULL) {
12262
0
        ret = ctxt->errNo;
12263
0
        goto error;
12264
0
    }
12265
12266
0
    list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1);
12267
0
    if (listOut != NULL)
12268
0
        *listOut = list;
12269
0
    else
12270
0
        xmlFreeNodeList(list);
12271
12272
0
    if (!ctxt->wellFormed)
12273
0
        ret = ctxt->errNo;
12274
0
    else
12275
0
        ret = XML_ERR_OK;
12276
12277
0
error:
12278
0
    xmlFreeInputStream(input);
12279
0
    xmlFreeParserCtxt(ctxt);
12280
0
    return(ret);
12281
0
}
12282
12283
/**
12284
 * Parse an XML external entity out of context and build a tree.
12285
 * It use the given SAX function block to handle the parsing callback.
12286
 * If sax is NULL, fallback to the default DOM tree building routines.
12287
 *
12288
 * @deprecated Don't use.
12289
 *
12290
 *     [78] extParsedEnt ::= TextDecl? content
12291
 *
12292
 * This correspond to a "Well Balanced" chunk
12293
 *
12294
 * This function uses deprecated global variables to set parser options
12295
 * which default to XML_PARSE_NODICT.
12296
 *
12297
 * @param sax  the SAX handler block
12298
 * @param filename  the filename
12299
 * @returns the resulting document tree
12300
 */
12301
12302
xmlDoc *
12303
0
xmlSAXParseEntity(xmlSAXHandler *sax, const char *filename) {
12304
0
    xmlDocPtr ret;
12305
0
    xmlParserCtxtPtr ctxt;
12306
12307
0
    ctxt = xmlCreateFileParserCtxt(filename);
12308
0
    if (ctxt == NULL) {
12309
0
  return(NULL);
12310
0
    }
12311
0
    if (sax != NULL) {
12312
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12313
0
            *ctxt->sax = *sax;
12314
0
        } else {
12315
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12316
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12317
0
        }
12318
0
        ctxt->userData = NULL;
12319
0
    }
12320
12321
0
    xmlParseExtParsedEnt(ctxt);
12322
12323
0
    if (ctxt->wellFormed) {
12324
0
  ret = ctxt->myDoc;
12325
0
    } else {
12326
0
        ret = NULL;
12327
0
        xmlFreeDoc(ctxt->myDoc);
12328
0
    }
12329
12330
0
    xmlFreeParserCtxt(ctxt);
12331
12332
0
    return(ret);
12333
0
}
12334
12335
/**
12336
 * Parse an XML external entity out of context and build a tree.
12337
 *
12338
 *     [78] extParsedEnt ::= TextDecl? content
12339
 *
12340
 * This correspond to a "Well Balanced" chunk
12341
 *
12342
 * This function uses deprecated global variables to set parser options
12343
 * which default to XML_PARSE_NODICT.
12344
 *
12345
 * @deprecated Don't use.
12346
 *
12347
 * @param filename  the filename
12348
 * @returns the resulting document tree
12349
 */
12350
12351
xmlDoc *
12352
0
xmlParseEntity(const char *filename) {
12353
0
    return(xmlSAXParseEntity(NULL, filename));
12354
0
}
12355
#endif /* LIBXML_SAX1_ENABLED */
12356
12357
/**
12358
 * Create a parser context for an external entity
12359
 * Automatic support for ZLIB/Compress compressed document is provided
12360
 * by default if found at compile-time.
12361
 *
12362
 * @deprecated Don't use.
12363
 *
12364
 * @param URL  the entity URL
12365
 * @param ID  the entity PUBLIC ID
12366
 * @param base  a possible base for the target URI
12367
 * @returns the new parser context or NULL
12368
 */
12369
xmlParserCtxt *
12370
xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID,
12371
0
                    const xmlChar *base) {
12372
0
    xmlParserCtxtPtr ctxt;
12373
0
    xmlParserInputPtr input;
12374
0
    xmlChar *uri = NULL;
12375
12376
0
    ctxt = xmlNewParserCtxt();
12377
0
    if (ctxt == NULL)
12378
0
  return(NULL);
12379
12380
0
    if (base != NULL) {
12381
0
        if (xmlBuildURISafe(URL, base, &uri) < 0)
12382
0
            goto error;
12383
0
        if (uri != NULL)
12384
0
            URL = uri;
12385
0
    }
12386
12387
0
    input = xmlLoadResource(ctxt, (char *) URL, (char *) ID,
12388
0
                            XML_RESOURCE_UNKNOWN);
12389
0
    if (input == NULL)
12390
0
        goto error;
12391
12392
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12393
0
        xmlFreeInputStream(input);
12394
0
        goto error;
12395
0
    }
12396
12397
0
    xmlFree(uri);
12398
0
    return(ctxt);
12399
12400
0
error:
12401
0
    xmlFree(uri);
12402
0
    xmlFreeParserCtxt(ctxt);
12403
0
    return(NULL);
12404
0
}
12405
12406
/************************************************************************
12407
 *                  *
12408
 *    Front ends when parsing from a file     *
12409
 *                  *
12410
 ************************************************************************/
12411
12412
/**
12413
 * Create a parser context for a file or URL content.
12414
 * Automatic support for ZLIB/Compress compressed document is provided
12415
 * by default if found at compile-time and for file accesses
12416
 *
12417
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12418
 *
12419
 * @param filename  the filename or URL
12420
 * @param options  a combination of xmlParserOption
12421
 * @returns the new parser context or NULL
12422
 */
12423
xmlParserCtxt *
12424
xmlCreateURLParserCtxt(const char *filename, int options)
12425
0
{
12426
0
    xmlParserCtxtPtr ctxt;
12427
0
    xmlParserInputPtr input;
12428
12429
0
    ctxt = xmlNewParserCtxt();
12430
0
    if (ctxt == NULL)
12431
0
  return(NULL);
12432
12433
0
    xmlCtxtUseOptions(ctxt, options);
12434
12435
0
    input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT);
12436
0
    if (input == NULL) {
12437
0
  xmlFreeParserCtxt(ctxt);
12438
0
  return(NULL);
12439
0
    }
12440
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12441
0
        xmlFreeInputStream(input);
12442
0
        xmlFreeParserCtxt(ctxt);
12443
0
        return(NULL);
12444
0
    }
12445
12446
0
    return(ctxt);
12447
0
}
12448
12449
/**
12450
 * Create a parser context for a file content.
12451
 * Automatic support for ZLIB/Compress compressed document is provided
12452
 * by default if found at compile-time.
12453
 *
12454
 * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile.
12455
 *
12456
 * @param filename  the filename
12457
 * @returns the new parser context or NULL
12458
 */
12459
xmlParserCtxt *
12460
xmlCreateFileParserCtxt(const char *filename)
12461
0
{
12462
0
    return(xmlCreateURLParserCtxt(filename, 0));
12463
0
}
12464
12465
#ifdef LIBXML_SAX1_ENABLED
12466
/**
12467
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12468
 * compressed document is provided by default if found at compile-time.
12469
 * It use the given SAX function block to handle the parsing callback.
12470
 * If sax is NULL, fallback to the default DOM tree building routines.
12471
 *
12472
 * This function uses deprecated global variables to set parser options
12473
 * which default to XML_PARSE_NODICT.
12474
 *
12475
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12476
 *
12477
 * User data (void *) is stored within the parser context in the
12478
 * context's _private member, so it is available nearly everywhere in libxml
12479
 *
12480
 * @param sax  the SAX handler block
12481
 * @param filename  the filename
12482
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12483
 *             documents
12484
 * @param data  the userdata
12485
 * @returns the resulting document tree
12486
 */
12487
12488
xmlDoc *
12489
xmlSAXParseFileWithData(xmlSAXHandler *sax, const char *filename,
12490
0
                        int recovery, void *data) {
12491
0
    xmlDocPtr ret = NULL;
12492
0
    xmlParserCtxtPtr ctxt;
12493
0
    xmlParserInputPtr input;
12494
12495
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12496
0
    if (ctxt == NULL)
12497
0
  return(NULL);
12498
12499
0
    if (data != NULL)
12500
0
  ctxt->_private = data;
12501
12502
0
    if (recovery) {
12503
0
        ctxt->options |= XML_PARSE_RECOVER;
12504
0
        ctxt->recovery = 1;
12505
0
    }
12506
12507
0
    if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0))
12508
0
        input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0);
12509
0
    else
12510
0
        input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0);
12511
12512
0
    if (input != NULL)
12513
0
        ret = xmlCtxtParseDocument(ctxt, input);
12514
12515
0
    xmlFreeParserCtxt(ctxt);
12516
0
    return(ret);
12517
0
}
12518
12519
/**
12520
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12521
 * compressed document is provided by default if found at compile-time.
12522
 * It use the given SAX function block to handle the parsing callback.
12523
 * If sax is NULL, fallback to the default DOM tree building routines.
12524
 *
12525
 * This function uses deprecated global variables to set parser options
12526
 * which default to XML_PARSE_NODICT.
12527
 *
12528
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12529
 *
12530
 * @param sax  the SAX handler block
12531
 * @param filename  the filename
12532
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12533
 *             documents
12534
 * @returns the resulting document tree
12535
 */
12536
12537
xmlDoc *
12538
xmlSAXParseFile(xmlSAXHandler *sax, const char *filename,
12539
0
                          int recovery) {
12540
0
    return(xmlSAXParseFileWithData(sax,filename,recovery,NULL));
12541
0
}
12542
12543
/**
12544
 * Parse an XML in-memory document and build a tree.
12545
 * In the case the document is not Well Formed, a attempt to build a
12546
 * tree is tried anyway
12547
 *
12548
 * This function uses deprecated global variables to set parser options
12549
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12550
 *
12551
 * @deprecated Use #xmlReadDoc with XML_PARSE_RECOVER.
12552
 *
12553
 * @param cur  a pointer to an array of xmlChar
12554
 * @returns the resulting document tree or NULL in case of failure
12555
 */
12556
12557
xmlDoc *
12558
0
xmlRecoverDoc(const xmlChar *cur) {
12559
0
    return(xmlSAXParseDoc(NULL, cur, 1));
12560
0
}
12561
12562
/**
12563
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12564
 * compressed document is provided by default if found at compile-time.
12565
 *
12566
 * This function uses deprecated global variables to set parser options
12567
 * which default to XML_PARSE_NODICT.
12568
 *
12569
 * @deprecated Use #xmlReadFile.
12570
 *
12571
 * @param filename  the filename
12572
 * @returns the resulting document tree if the file was wellformed,
12573
 * NULL otherwise.
12574
 */
12575
12576
xmlDoc *
12577
0
xmlParseFile(const char *filename) {
12578
0
    return(xmlSAXParseFile(NULL, filename, 0));
12579
0
}
12580
12581
/**
12582
 * Parse an XML file and build a tree. Automatic support for ZLIB/Compress
12583
 * compressed document is provided by default if found at compile-time.
12584
 * In the case the document is not Well Formed, it attempts to build
12585
 * a tree anyway
12586
 *
12587
 * This function uses deprecated global variables to set parser options
12588
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12589
 *
12590
 * @deprecated Use #xmlReadFile with XML_PARSE_RECOVER.
12591
 *
12592
 * @param filename  the filename
12593
 * @returns the resulting document tree or NULL in case of failure
12594
 */
12595
12596
xmlDoc *
12597
0
xmlRecoverFile(const char *filename) {
12598
0
    return(xmlSAXParseFile(NULL, filename, 1));
12599
0
}
12600
12601
12602
/**
12603
 * Setup the parser context to parse a new buffer; Clears any prior
12604
 * contents from the parser context. The buffer parameter must not be
12605
 * NULL, but the filename parameter can be
12606
 *
12607
 * @deprecated Don't use.
12608
 *
12609
 * @param ctxt  an XML parser context
12610
 * @param buffer  a xmlChar * buffer
12611
 * @param filename  a file name
12612
 */
12613
void
12614
xmlSetupParserForBuffer(xmlParserCtxt *ctxt, const xmlChar* buffer,
12615
                             const char* filename)
12616
0
{
12617
0
    xmlParserInputPtr input;
12618
12619
0
    if ((ctxt == NULL) || (buffer == NULL))
12620
0
        return;
12621
12622
0
    xmlCtxtReset(ctxt);
12623
12624
0
    input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer,
12625
0
                                      NULL, 0);
12626
0
    if (input == NULL)
12627
0
        return;
12628
0
    if (xmlCtxtPushInput(ctxt, input) < 0)
12629
0
        xmlFreeInputStream(input);
12630
0
}
12631
12632
/**
12633
 * Parse an XML file and call the given SAX handler routines.
12634
 * Automatic support for ZLIB/Compress compressed document is provided
12635
 *
12636
 * This function uses deprecated global variables to set parser options
12637
 * which default to XML_PARSE_NODICT.
12638
 *
12639
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile.
12640
 *
12641
 * @param sax  a SAX handler
12642
 * @param user_data  The user data returned on SAX callbacks
12643
 * @param filename  a file name
12644
 * @returns 0 in case of success or a error number otherwise
12645
 */
12646
int
12647
xmlSAXUserParseFile(xmlSAXHandler *sax, void *user_data,
12648
0
                    const char *filename) {
12649
0
    int ret = 0;
12650
0
    xmlParserCtxtPtr ctxt;
12651
12652
0
    ctxt = xmlCreateFileParserCtxt(filename);
12653
0
    if (ctxt == NULL) return -1;
12654
0
    if (sax != NULL) {
12655
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12656
0
            *ctxt->sax = *sax;
12657
0
        } else {
12658
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12659
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12660
0
        }
12661
0
  ctxt->userData = user_data;
12662
0
    }
12663
12664
0
    xmlParseDocument(ctxt);
12665
12666
0
    if (ctxt->wellFormed)
12667
0
  ret = 0;
12668
0
    else {
12669
0
        if (ctxt->errNo != 0)
12670
0
      ret = ctxt->errNo;
12671
0
  else
12672
0
      ret = -1;
12673
0
    }
12674
0
    if (ctxt->myDoc != NULL) {
12675
0
        xmlFreeDoc(ctxt->myDoc);
12676
0
  ctxt->myDoc = NULL;
12677
0
    }
12678
0
    xmlFreeParserCtxt(ctxt);
12679
12680
0
    return ret;
12681
0
}
12682
#endif /* LIBXML_SAX1_ENABLED */
12683
12684
/************************************************************************
12685
 *                  *
12686
 *    Front ends when parsing from memory     *
12687
 *                  *
12688
 ************************************************************************/
12689
12690
/**
12691
 * Create a parser context for an XML in-memory document. The input buffer
12692
 * must not contain a terminating null byte.
12693
 *
12694
 * @param buffer  a pointer to a char array
12695
 * @param size  the size of the array
12696
 * @returns the new parser context or NULL
12697
 */
12698
xmlParserCtxt *
12699
0
xmlCreateMemoryParserCtxt(const char *buffer, int size) {
12700
0
    xmlParserCtxtPtr ctxt;
12701
0
    xmlParserInputPtr input;
12702
12703
0
    if (size < 0)
12704
0
  return(NULL);
12705
12706
0
    ctxt = xmlNewParserCtxt();
12707
0
    if (ctxt == NULL)
12708
0
  return(NULL);
12709
12710
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0);
12711
0
    if (input == NULL) {
12712
0
  xmlFreeParserCtxt(ctxt);
12713
0
  return(NULL);
12714
0
    }
12715
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12716
0
        xmlFreeInputStream(input);
12717
0
        xmlFreeParserCtxt(ctxt);
12718
0
        return(NULL);
12719
0
    }
12720
12721
0
    return(ctxt);
12722
0
}
12723
12724
#ifdef LIBXML_SAX1_ENABLED
12725
/**
12726
 * Parse an XML in-memory block and use the given SAX function block
12727
 * to handle the parsing callback. If sax is NULL, fallback to the default
12728
 * DOM tree building routines.
12729
 *
12730
 * This function uses deprecated global variables to set parser options
12731
 * which default to XML_PARSE_NODICT.
12732
 *
12733
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12734
 *
12735
 * User data (void *) is stored within the parser context in the
12736
 * context's _private member, so it is available nearly everywhere in libxml
12737
 *
12738
 * @param sax  the SAX handler block
12739
 * @param buffer  an pointer to a char array
12740
 * @param size  the size of the array
12741
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12742
 *             documents
12743
 * @param data  the userdata
12744
 * @returns the resulting document tree
12745
 */
12746
12747
xmlDoc *
12748
xmlSAXParseMemoryWithData(xmlSAXHandler *sax, const char *buffer,
12749
0
                          int size, int recovery, void *data) {
12750
0
    xmlDocPtr ret = NULL;
12751
0
    xmlParserCtxtPtr ctxt;
12752
0
    xmlParserInputPtr input;
12753
12754
0
    if (size < 0)
12755
0
        return(NULL);
12756
12757
0
    ctxt = xmlNewSAXParserCtxt(sax, NULL);
12758
0
    if (ctxt == NULL)
12759
0
        return(NULL);
12760
12761
0
    if (data != NULL)
12762
0
  ctxt->_private=data;
12763
12764
0
    if (recovery) {
12765
0
        ctxt->options |= XML_PARSE_RECOVER;
12766
0
        ctxt->recovery = 1;
12767
0
    }
12768
12769
0
    input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL,
12770
0
                                      XML_INPUT_BUF_STATIC);
12771
12772
0
    if (input != NULL)
12773
0
        ret = xmlCtxtParseDocument(ctxt, input);
12774
12775
0
    xmlFreeParserCtxt(ctxt);
12776
0
    return(ret);
12777
0
}
12778
12779
/**
12780
 * Parse an XML in-memory block and use the given SAX function block
12781
 * to handle the parsing callback. If sax is NULL, fallback to the default
12782
 * DOM tree building routines.
12783
 *
12784
 * This function uses deprecated global variables to set parser options
12785
 * which default to XML_PARSE_NODICT.
12786
 *
12787
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12788
 *
12789
 * @param sax  the SAX handler block
12790
 * @param buffer  an pointer to a char array
12791
 * @param size  the size of the array
12792
 * @param recovery  work in recovery mode, i.e. tries to read not Well Formed
12793
 *             documents
12794
 * @returns the resulting document tree
12795
 */
12796
xmlDoc *
12797
xmlSAXParseMemory(xmlSAXHandler *sax, const char *buffer,
12798
0
            int size, int recovery) {
12799
0
    return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL);
12800
0
}
12801
12802
/**
12803
 * Parse an XML in-memory block and build a tree.
12804
 *
12805
 * This function uses deprecated global variables to set parser options
12806
 * which default to XML_PARSE_NODICT.
12807
 *
12808
 * @deprecated Use #xmlReadMemory.
12809
 *
12810
 * @param buffer  an pointer to a char array
12811
 * @param size  the size of the array
12812
 * @returns the resulting document tree
12813
 */
12814
12815
0
xmlDoc *xmlParseMemory(const char *buffer, int size) {
12816
0
   return(xmlSAXParseMemory(NULL, buffer, size, 0));
12817
0
}
12818
12819
/**
12820
 * Parse an XML in-memory block and build a tree.
12821
 * In the case the document is not Well Formed, an attempt to
12822
 * build a tree is tried anyway
12823
 *
12824
 * This function uses deprecated global variables to set parser options
12825
 * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER.
12826
 *
12827
 * @deprecated Use #xmlReadMemory with XML_PARSE_RECOVER.
12828
 *
12829
 * @param buffer  an pointer to a char array
12830
 * @param size  the size of the array
12831
 * @returns the resulting document tree or NULL in case of error
12832
 */
12833
12834
0
xmlDoc *xmlRecoverMemory(const char *buffer, int size) {
12835
0
   return(xmlSAXParseMemory(NULL, buffer, size, 1));
12836
0
}
12837
12838
/**
12839
 * Parse an XML in-memory buffer and call the given SAX handler routines.
12840
 *
12841
 * This function uses deprecated global variables to set parser options
12842
 * which default to XML_PARSE_NODICT.
12843
 *
12844
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory.
12845
 *
12846
 * @param sax  a SAX handler
12847
 * @param user_data  The user data returned on SAX callbacks
12848
 * @param buffer  an in-memory XML document input
12849
 * @param size  the length of the XML document in bytes
12850
 * @returns 0 in case of success or a error number otherwise
12851
 */
12852
int xmlSAXUserParseMemory(xmlSAXHandler *sax, void *user_data,
12853
0
        const char *buffer, int size) {
12854
0
    int ret = 0;
12855
0
    xmlParserCtxtPtr ctxt;
12856
12857
0
    ctxt = xmlCreateMemoryParserCtxt(buffer, size);
12858
0
    if (ctxt == NULL) return -1;
12859
0
    if (sax != NULL) {
12860
0
        if (sax->initialized == XML_SAX2_MAGIC) {
12861
0
            *ctxt->sax = *sax;
12862
0
        } else {
12863
0
            memset(ctxt->sax, 0, sizeof(*ctxt->sax));
12864
0
            memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1));
12865
0
        }
12866
0
  ctxt->userData = user_data;
12867
0
    }
12868
12869
0
    xmlParseDocument(ctxt);
12870
12871
0
    if (ctxt->wellFormed)
12872
0
  ret = 0;
12873
0
    else {
12874
0
        if (ctxt->errNo != 0)
12875
0
      ret = ctxt->errNo;
12876
0
  else
12877
0
      ret = -1;
12878
0
    }
12879
0
    if (ctxt->myDoc != NULL) {
12880
0
        xmlFreeDoc(ctxt->myDoc);
12881
0
  ctxt->myDoc = NULL;
12882
0
    }
12883
0
    xmlFreeParserCtxt(ctxt);
12884
12885
0
    return ret;
12886
0
}
12887
#endif /* LIBXML_SAX1_ENABLED */
12888
12889
/**
12890
 * Creates a parser context for an XML in-memory document.
12891
 *
12892
 * @param str  a pointer to an array of xmlChar
12893
 * @returns the new parser context or NULL
12894
 */
12895
xmlParserCtxt *
12896
0
xmlCreateDocParserCtxt(const xmlChar *str) {
12897
0
    xmlParserCtxtPtr ctxt;
12898
0
    xmlParserInputPtr input;
12899
12900
0
    ctxt = xmlNewParserCtxt();
12901
0
    if (ctxt == NULL)
12902
0
  return(NULL);
12903
12904
0
    input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0);
12905
0
    if (input == NULL) {
12906
0
  xmlFreeParserCtxt(ctxt);
12907
0
  return(NULL);
12908
0
    }
12909
0
    if (xmlCtxtPushInput(ctxt, input) < 0) {
12910
0
        xmlFreeInputStream(input);
12911
0
        xmlFreeParserCtxt(ctxt);
12912
0
        return(NULL);
12913
0
    }
12914
12915
0
    return(ctxt);
12916
0
}
12917
12918
#ifdef LIBXML_SAX1_ENABLED
12919
/**
12920
 * Parse an XML in-memory document and build a tree.
12921
 * It use the given SAX function block to handle the parsing callback.
12922
 * If sax is NULL, fallback to the default DOM tree building routines.
12923
 *
12924
 * This function uses deprecated global variables to set parser options
12925
 * which default to XML_PARSE_NODICT.
12926
 *
12927
 * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadDoc.
12928
 *
12929
 * @param sax  the SAX handler block
12930
 * @param cur  a pointer to an array of xmlChar
12931
 * @param recovery  work in recovery mode, i.e. tries to read no Well Formed
12932
 *             documents
12933
 * @returns the resulting document tree
12934
 */
12935
12936
xmlDoc *
12937
0
xmlSAXParseDoc(xmlSAXHandler *sax, const xmlChar *cur, int recovery) {
12938
0
    xmlDocPtr ret;
12939
0
    xmlParserCtxtPtr ctxt;
12940
0
    xmlSAXHandlerPtr oldsax = NULL;
12941
12942
0
    if (cur == NULL) return(NULL);
12943
12944
12945
0
    ctxt = xmlCreateDocParserCtxt(cur);
12946
0
    if (ctxt == NULL) return(NULL);
12947
0
    if (sax != NULL) {
12948
0
        oldsax = ctxt->sax;
12949
0
        ctxt->sax = sax;
12950
0
        ctxt->userData = NULL;
12951
0
    }
12952
12953
0
    xmlParseDocument(ctxt);
12954
0
    if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc;
12955
0
    else {
12956
0
       ret = NULL;
12957
0
       xmlFreeDoc(ctxt->myDoc);
12958
0
       ctxt->myDoc = NULL;
12959
0
    }
12960
0
    if (sax != NULL)
12961
0
  ctxt->sax = oldsax;
12962
0
    xmlFreeParserCtxt(ctxt);
12963
12964
0
    return(ret);
12965
0
}
12966
12967
/**
12968
 * Parse an XML in-memory document and build a tree.
12969
 *
12970
 * This function uses deprecated global variables to set parser options
12971
 * which default to XML_PARSE_NODICT.
12972
 *
12973
 * @deprecated Use #xmlReadDoc.
12974
 *
12975
 * @param cur  a pointer to an array of xmlChar
12976
 * @returns the resulting document tree
12977
 */
12978
12979
xmlDoc *
12980
0
xmlParseDoc(const xmlChar *cur) {
12981
0
    return(xmlSAXParseDoc(NULL, cur, 0));
12982
0
}
12983
#endif /* LIBXML_SAX1_ENABLED */
12984
12985
/************************************************************************
12986
 *                  *
12987
 *  New set (2.6.0) of simpler and more flexible APIs   *
12988
 *                  *
12989
 ************************************************************************/
12990
12991
/**
12992
 * Reset a parser context
12993
 *
12994
 * @param ctxt  an XML parser context
12995
 */
12996
void
12997
xmlCtxtReset(xmlParserCtxt *ctxt)
12998
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
39.9k
{
13132
39.9k
    int allMask;
13133
13134
39.9k
    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
39.9k
    allMask = XML_PARSE_RECOVER |
13145
39.9k
              XML_PARSE_NOENT |
13146
39.9k
              XML_PARSE_DTDLOAD |
13147
39.9k
              XML_PARSE_DTDATTR |
13148
39.9k
              XML_PARSE_DTDVALID |
13149
39.9k
              XML_PARSE_NOERROR |
13150
39.9k
              XML_PARSE_NOWARNING |
13151
39.9k
              XML_PARSE_PEDANTIC |
13152
39.9k
              XML_PARSE_NOBLANKS |
13153
39.9k
#ifdef LIBXML_SAX1_ENABLED
13154
39.9k
              XML_PARSE_SAX1 |
13155
39.9k
#endif
13156
39.9k
              XML_PARSE_NONET |
13157
39.9k
              XML_PARSE_NODICT |
13158
39.9k
              XML_PARSE_NSCLEAN |
13159
39.9k
              XML_PARSE_NOCDATA |
13160
39.9k
              XML_PARSE_COMPACT |
13161
39.9k
              XML_PARSE_OLD10 |
13162
39.9k
              XML_PARSE_HUGE |
13163
39.9k
              XML_PARSE_OLDSAX |
13164
39.9k
              XML_PARSE_IGNORE_ENC |
13165
39.9k
              XML_PARSE_BIG_LINES |
13166
39.9k
              XML_PARSE_NO_XXE |
13167
39.9k
              XML_PARSE_UNZIP |
13168
39.9k
              XML_PARSE_NO_SYS_CATALOG |
13169
39.9k
              XML_PARSE_CATALOG_PI;
13170
13171
39.9k
    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
39.9k
    ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0;
13188
39.9k
    ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0;
13189
39.9k
    ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0;
13190
39.9k
    ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0;
13191
39.9k
    ctxt->loadsubset |= (options & XML_PARSE_SKIP_IDS) ? XML_SKIP_IDS : 0;
13192
39.9k
    ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0;
13193
39.9k
    ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0;
13194
39.9k
    ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1;
13195
39.9k
    ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1;
13196
13197
39.9k
    return(options & ~allMask);
13198
39.9k
}
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
39.9k
{
13267
39.9k
    int keepMask;
13268
13269
39.9k
#ifdef LIBXML_HTML_ENABLED
13270
39.9k
    if ((ctxt != NULL) && (ctxt->html))
13271
0
        return(htmlCtxtUseOptions(ctxt, options));
13272
39.9k
#endif
13273
13274
    /*
13275
     * For historic reasons, some options can only be enabled.
13276
     */
13277
39.9k
    keepMask = XML_PARSE_NOERROR |
13278
39.9k
               XML_PARSE_NOWARNING |
13279
39.9k
               XML_PARSE_NONET |
13280
39.9k
               XML_PARSE_NSCLEAN |
13281
39.9k
               XML_PARSE_NOCDATA |
13282
39.9k
               XML_PARSE_COMPACT |
13283
39.9k
               XML_PARSE_OLD10 |
13284
39.9k
               XML_PARSE_HUGE |
13285
39.9k
               XML_PARSE_OLDSAX |
13286
39.9k
               XML_PARSE_IGNORE_ENC |
13287
39.9k
               XML_PARSE_BIG_LINES;
13288
13289
39.9k
    return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask));
13290
39.9k
}
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