Line | Count | Source (jump to first uncovered line) |
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 | | * daniel@veillard.com |
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 | | #ifdef LIBXML_CATALOG_ENABLED |
66 | | #include <libxml/catalog.h> |
67 | | #endif |
68 | | |
69 | | #include "private/buf.h" |
70 | | #include "private/dict.h" |
71 | | #include "private/entities.h" |
72 | | #include "private/error.h" |
73 | | #include "private/html.h" |
74 | | #include "private/io.h" |
75 | | #include "private/parser.h" |
76 | | |
77 | 4.84M | #define NS_INDEX_EMPTY INT_MAX |
78 | 755k | #define NS_INDEX_XML (INT_MAX - 1) |
79 | 2.47M | #define URI_HASH_EMPTY 0xD943A04E |
80 | 132k | #define URI_HASH_XML 0xF0451F02 |
81 | | |
82 | | #ifndef STDIN_FILENO |
83 | 0 | #define STDIN_FILENO 0 |
84 | | #endif |
85 | | |
86 | | struct _xmlStartTag { |
87 | | const xmlChar *prefix; |
88 | | const xmlChar *URI; |
89 | | int line; |
90 | | int nsNr; |
91 | | }; |
92 | | |
93 | | typedef struct { |
94 | | void *saxData; |
95 | | unsigned prefixHashValue; |
96 | | unsigned uriHashValue; |
97 | | unsigned elementId; |
98 | | int oldIndex; |
99 | | } xmlParserNsExtra; |
100 | | |
101 | | typedef struct { |
102 | | unsigned hashValue; |
103 | | int index; |
104 | | } xmlParserNsBucket; |
105 | | |
106 | | struct _xmlParserNsData { |
107 | | xmlParserNsExtra *extra; |
108 | | |
109 | | unsigned hashSize; |
110 | | unsigned hashElems; |
111 | | xmlParserNsBucket *hash; |
112 | | |
113 | | unsigned elementId; |
114 | | int defaultNsIndex; |
115 | | int minNsIndex; |
116 | | }; |
117 | | |
118 | | struct _xmlAttrHashBucket { |
119 | | int index; |
120 | | }; |
121 | | |
122 | | static int |
123 | | xmlParseElementStart(xmlParserCtxtPtr ctxt); |
124 | | |
125 | | static void |
126 | | xmlParseElementEnd(xmlParserCtxtPtr ctxt); |
127 | | |
128 | | static xmlEntityPtr |
129 | | xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr); |
130 | | |
131 | | static const xmlChar * |
132 | | xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt); |
133 | | |
134 | | /************************************************************************ |
135 | | * * |
136 | | * Arbitrary limits set in the parser. See XML_PARSE_HUGE * |
137 | | * * |
138 | | ************************************************************************/ |
139 | | |
140 | | #define XML_PARSER_BIG_ENTITY 1000 |
141 | | #define XML_PARSER_LOT_ENTITY 5000 |
142 | | |
143 | | /* |
144 | | * Constants for protection against abusive entity expansion |
145 | | * ("billion laughs"). |
146 | | */ |
147 | | |
148 | | /* |
149 | | * A certain amount of entity expansion which is always allowed. |
150 | | */ |
151 | 10.0M | #define XML_PARSER_ALLOWED_EXPANSION 1000000 |
152 | | |
153 | | /* |
154 | | * Fixed cost for each entity reference. This crudely models processing time |
155 | | * as well to protect, for example, against exponential expansion of empty |
156 | | * or very short entities. |
157 | | */ |
158 | 10.0M | #define XML_ENT_FIXED_COST 20 |
159 | | |
160 | | /** |
161 | | * xmlParserMaxDepth: |
162 | | * |
163 | | * arbitrary depth limit for the XML documents that we allow to |
164 | | * process. This is not a limitation of the parser but a safety |
165 | | * boundary feature. It can be disabled with the XML_PARSE_HUGE |
166 | | * parser option. |
167 | | */ |
168 | | const unsigned int xmlParserMaxDepth = 256; |
169 | | |
170 | | |
171 | | |
172 | 272M | #define XML_PARSER_BIG_BUFFER_SIZE 300 |
173 | 5.72M | #define XML_PARSER_BUFFER_SIZE 100 |
174 | 619k | #define SAX_COMPAT_MODE BAD_CAST "SAX compatibility mode document" |
175 | | |
176 | | /** |
177 | | * XML_PARSER_CHUNK_SIZE |
178 | | * |
179 | | * When calling GROW that's the minimal amount of data |
180 | | * the parser expected to have received. It is not a hard |
181 | | * limit but an optimization when reading strings like Names |
182 | | * It is not strictly needed as long as inputs available characters |
183 | | * are followed by 0, which should be provided by the I/O level |
184 | | */ |
185 | | #define XML_PARSER_CHUNK_SIZE 100 |
186 | | |
187 | | /** |
188 | | * xmlParserVersion: |
189 | | * |
190 | | * Constant string describing the internal version of the library |
191 | | */ |
192 | | const char *const |
193 | | xmlParserVersion = LIBXML_VERSION_STRING LIBXML_VERSION_EXTRA; |
194 | | |
195 | | /* |
196 | | * List of XML prefixed PI allowed by W3C specs |
197 | | */ |
198 | | |
199 | | static const char* const xmlW3CPIs[] = { |
200 | | "xml-stylesheet", |
201 | | "xml-model", |
202 | | NULL |
203 | | }; |
204 | | |
205 | | |
206 | | /* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */ |
207 | | static xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt, |
208 | | const xmlChar **str); |
209 | | |
210 | | static void |
211 | | xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent); |
212 | | |
213 | | static int |
214 | | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity); |
215 | | |
216 | | /************************************************************************ |
217 | | * * |
218 | | * Some factorized error routines * |
219 | | * * |
220 | | ************************************************************************/ |
221 | | |
222 | | static void |
223 | 6.63k | xmlErrMemory(xmlParserCtxtPtr ctxt) { |
224 | 6.63k | xmlCtxtErrMemory(ctxt); |
225 | 6.63k | } |
226 | | |
227 | | /** |
228 | | * xmlErrAttributeDup: |
229 | | * @ctxt: an XML parser context |
230 | | * @prefix: the attribute prefix |
231 | | * @localname: the attribute localname |
232 | | * |
233 | | * Handle a redefinition of attribute error |
234 | | */ |
235 | | static void |
236 | | xmlErrAttributeDup(xmlParserCtxtPtr ctxt, const xmlChar * prefix, |
237 | | const xmlChar * localname) |
238 | 139k | { |
239 | 139k | if (prefix == NULL) |
240 | 89.1k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED, |
241 | 89.1k | XML_ERR_FATAL, localname, NULL, NULL, 0, |
242 | 89.1k | "Attribute %s redefined\n", localname); |
243 | 50.5k | else |
244 | 50.5k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED, |
245 | 50.5k | XML_ERR_FATAL, prefix, localname, NULL, 0, |
246 | 50.5k | "Attribute %s:%s redefined\n", prefix, localname); |
247 | 139k | } |
248 | | |
249 | | /** |
250 | | * xmlFatalErrMsg: |
251 | | * @ctxt: an XML parser context |
252 | | * @error: the error number |
253 | | * @msg: the error message |
254 | | * |
255 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
256 | | */ |
257 | | static void LIBXML_ATTR_FORMAT(3,0) |
258 | | xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
259 | | const char *msg) |
260 | 271M | { |
261 | 271M | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
262 | 271M | NULL, NULL, NULL, 0, "%s", msg); |
263 | 271M | } |
264 | | |
265 | | /** |
266 | | * xmlWarningMsg: |
267 | | * @ctxt: an XML parser context |
268 | | * @error: the error number |
269 | | * @msg: the error message |
270 | | * @str1: extra data |
271 | | * @str2: extra data |
272 | | * |
273 | | * Handle a warning. |
274 | | */ |
275 | | void LIBXML_ATTR_FORMAT(3,0) |
276 | | xmlWarningMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
277 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
278 | 287k | { |
279 | 287k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING, |
280 | 287k | str1, str2, NULL, 0, msg, str1, str2); |
281 | 287k | } |
282 | | |
283 | | /** |
284 | | * xmlValidityError: |
285 | | * @ctxt: an XML parser context |
286 | | * @error: the error number |
287 | | * @msg: the error message |
288 | | * @str1: extra data |
289 | | * |
290 | | * Handle a validity error. |
291 | | */ |
292 | | static void LIBXML_ATTR_FORMAT(3,0) |
293 | | xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
294 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
295 | 165k | { |
296 | 165k | ctxt->valid = 0; |
297 | | |
298 | 165k | xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR, |
299 | 165k | str1, str2, NULL, 0, msg, str1, str2); |
300 | 165k | } |
301 | | |
302 | | /** |
303 | | * xmlFatalErrMsgInt: |
304 | | * @ctxt: an XML parser context |
305 | | * @error: the error number |
306 | | * @msg: the error message |
307 | | * @val: an integer value |
308 | | * |
309 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
310 | | */ |
311 | | static void LIBXML_ATTR_FORMAT(3,0) |
312 | | xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
313 | | const char *msg, int val) |
314 | 45.0M | { |
315 | 45.0M | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
316 | 45.0M | NULL, NULL, NULL, val, msg, val); |
317 | 45.0M | } |
318 | | |
319 | | /** |
320 | | * xmlFatalErrMsgStrIntStr: |
321 | | * @ctxt: an XML parser context |
322 | | * @error: the error number |
323 | | * @msg: the error message |
324 | | * @str1: an string info |
325 | | * @val: an integer value |
326 | | * @str2: an string info |
327 | | * |
328 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
329 | | */ |
330 | | static void LIBXML_ATTR_FORMAT(3,0) |
331 | | xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
332 | | const char *msg, const xmlChar *str1, int val, |
333 | | const xmlChar *str2) |
334 | 3.99M | { |
335 | 3.99M | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
336 | 3.99M | str1, str2, NULL, val, msg, str1, val, str2); |
337 | 3.99M | } |
338 | | |
339 | | /** |
340 | | * xmlFatalErrMsgStr: |
341 | | * @ctxt: an XML parser context |
342 | | * @error: the error number |
343 | | * @msg: the error message |
344 | | * @val: a string value |
345 | | * |
346 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
347 | | */ |
348 | | static void LIBXML_ATTR_FORMAT(3,0) |
349 | | xmlFatalErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
350 | | const char *msg, const xmlChar * val) |
351 | 4.27M | { |
352 | 4.27M | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
353 | 4.27M | val, NULL, NULL, 0, msg, val); |
354 | 4.27M | } |
355 | | |
356 | | /** |
357 | | * xmlErrMsgStr: |
358 | | * @ctxt: an XML parser context |
359 | | * @error: the error number |
360 | | * @msg: the error message |
361 | | * @val: a string value |
362 | | * |
363 | | * Handle a non fatal parser error |
364 | | */ |
365 | | static void LIBXML_ATTR_FORMAT(3,0) |
366 | | xmlErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
367 | | const char *msg, const xmlChar * val) |
368 | 741k | { |
369 | 741k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR, |
370 | 741k | val, NULL, NULL, 0, msg, val); |
371 | 741k | } |
372 | | |
373 | | /** |
374 | | * xmlNsErr: |
375 | | * @ctxt: an XML parser context |
376 | | * @error: the error number |
377 | | * @msg: the message |
378 | | * @info1: extra information string |
379 | | * @info2: extra information string |
380 | | * |
381 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
382 | | */ |
383 | | static void LIBXML_ATTR_FORMAT(3,0) |
384 | | xmlNsErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
385 | | const char *msg, |
386 | | const xmlChar * info1, const xmlChar * info2, |
387 | | const xmlChar * info3) |
388 | 1.61M | { |
389 | 1.61M | ctxt->nsWellFormed = 0; |
390 | | |
391 | 1.61M | xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR, |
392 | 1.61M | info1, info2, info3, 0, msg, info1, info2, info3); |
393 | 1.61M | } |
394 | | |
395 | | /** |
396 | | * xmlNsWarn |
397 | | * @ctxt: an XML parser context |
398 | | * @error: the error number |
399 | | * @msg: the message |
400 | | * @info1: extra information string |
401 | | * @info2: extra information string |
402 | | * |
403 | | * Handle a namespace warning error |
404 | | */ |
405 | | static void LIBXML_ATTR_FORMAT(3,0) |
406 | | xmlNsWarn(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
407 | | const char *msg, |
408 | | const xmlChar * info1, const xmlChar * info2, |
409 | | const xmlChar * info3) |
410 | 175k | { |
411 | 175k | xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING, |
412 | 175k | info1, info2, info3, 0, msg, info1, info2, info3); |
413 | 175k | } |
414 | | |
415 | | static void |
416 | 30.2M | xmlSaturatedAdd(unsigned long *dst, unsigned long val) { |
417 | 30.2M | if (val > ULONG_MAX - *dst) |
418 | 0 | *dst = ULONG_MAX; |
419 | 30.2M | else |
420 | 30.2M | *dst += val; |
421 | 30.2M | } |
422 | | |
423 | | static void |
424 | 10.1M | xmlSaturatedAddSizeT(unsigned long *dst, unsigned long val) { |
425 | 10.1M | if (val > ULONG_MAX - *dst) |
426 | 0 | *dst = ULONG_MAX; |
427 | 10.1M | else |
428 | 10.1M | *dst += val; |
429 | 10.1M | } |
430 | | |
431 | | /** |
432 | | * xmlParserEntityCheck: |
433 | | * @ctxt: parser context |
434 | | * @extra: sum of unexpanded entity sizes |
435 | | * |
436 | | * Check for non-linear entity expansion behaviour. |
437 | | * |
438 | | * In some cases like xmlExpandEntityInAttValue, this function is called |
439 | | * for each, possibly nested entity and its unexpanded content length. |
440 | | * |
441 | | * In other cases like xmlParseReference, it's only called for each |
442 | | * top-level entity with its unexpanded content length plus the sum of |
443 | | * the unexpanded content lengths (plus fixed cost) of all nested |
444 | | * entities. |
445 | | * |
446 | | * Summing the unexpanded lengths also adds the length of the reference. |
447 | | * This is by design. Taking the length of the entity name into account |
448 | | * discourages attacks that try to waste CPU time with abusively long |
449 | | * entity names. See test/recurse/lol6.xml for example. Each call also |
450 | | * adds some fixed cost XML_ENT_FIXED_COST to discourage attacks with |
451 | | * short entities. |
452 | | * |
453 | | * Returns 1 on error, 0 on success. |
454 | | */ |
455 | | static int |
456 | | xmlParserEntityCheck(xmlParserCtxtPtr ctxt, unsigned long extra) |
457 | 10.5M | { |
458 | 10.5M | unsigned long consumed; |
459 | 10.5M | unsigned long *expandedSize; |
460 | 10.5M | xmlParserInputPtr input = ctxt->input; |
461 | 10.5M | xmlEntityPtr entity = input->entity; |
462 | | |
463 | 10.5M | if ((entity) && (entity->flags & XML_ENT_CHECKED)) |
464 | 538k | return(0); |
465 | | |
466 | | /* |
467 | | * Compute total consumed bytes so far, including input streams of |
468 | | * external entities. |
469 | | */ |
470 | 10.0M | consumed = input->consumed; |
471 | 10.0M | xmlSaturatedAddSizeT(&consumed, input->cur - input->base); |
472 | 10.0M | xmlSaturatedAdd(&consumed, ctxt->sizeentities); |
473 | | |
474 | 10.0M | if (entity) |
475 | 191k | expandedSize = &entity->expandedSize; |
476 | 9.84M | else |
477 | 9.84M | expandedSize = &ctxt->sizeentcopy; |
478 | | |
479 | | /* |
480 | | * Add extra cost and some fixed cost. |
481 | | */ |
482 | 10.0M | xmlSaturatedAdd(expandedSize, extra); |
483 | 10.0M | xmlSaturatedAdd(expandedSize, XML_ENT_FIXED_COST); |
484 | | |
485 | | /* |
486 | | * It's important to always use saturation arithmetic when tracking |
487 | | * entity sizes to make the size checks reliable. If "sizeentcopy" |
488 | | * overflows, we have to abort. |
489 | | */ |
490 | 10.0M | if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) && |
491 | 10.0M | ((*expandedSize >= ULONG_MAX) || |
492 | 2.49M | (*expandedSize / ctxt->maxAmpl > consumed))) { |
493 | 5.06k | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
494 | 5.06k | "Maximum entity amplification factor exceeded, see " |
495 | 5.06k | "xmlCtxtSetMaxAmplification.\n"); |
496 | 5.06k | xmlHaltParser(ctxt); |
497 | 5.06k | return(1); |
498 | 5.06k | } |
499 | | |
500 | 10.0M | return(0); |
501 | 10.0M | } |
502 | | |
503 | | /************************************************************************ |
504 | | * * |
505 | | * Library wide options * |
506 | | * * |
507 | | ************************************************************************/ |
508 | | |
509 | | /** |
510 | | * xmlHasFeature: |
511 | | * @feature: the feature to be examined |
512 | | * |
513 | | * Examines if the library has been compiled with a given feature. |
514 | | * |
515 | | * Returns a non-zero value if the feature exist, otherwise zero. |
516 | | * Returns zero (0) if the feature does not exist or an unknown |
517 | | * unknown feature is requested, non-zero otherwise. |
518 | | */ |
519 | | int |
520 | | xmlHasFeature(xmlFeature feature) |
521 | 97.3k | { |
522 | 97.3k | switch (feature) { |
523 | 3.24k | case XML_WITH_THREAD: |
524 | 3.24k | #ifdef LIBXML_THREAD_ENABLED |
525 | 3.24k | return(1); |
526 | | #else |
527 | | return(0); |
528 | | #endif |
529 | 3.24k | case XML_WITH_TREE: |
530 | 3.24k | return(1); |
531 | 3.24k | case XML_WITH_OUTPUT: |
532 | 3.24k | #ifdef LIBXML_OUTPUT_ENABLED |
533 | 3.24k | return(1); |
534 | | #else |
535 | | return(0); |
536 | | #endif |
537 | 3.24k | case XML_WITH_PUSH: |
538 | 3.24k | #ifdef LIBXML_PUSH_ENABLED |
539 | 3.24k | return(1); |
540 | | #else |
541 | | return(0); |
542 | | #endif |
543 | 3.24k | case XML_WITH_READER: |
544 | 3.24k | #ifdef LIBXML_READER_ENABLED |
545 | 3.24k | return(1); |
546 | | #else |
547 | | return(0); |
548 | | #endif |
549 | 3.24k | case XML_WITH_PATTERN: |
550 | 3.24k | #ifdef LIBXML_PATTERN_ENABLED |
551 | 3.24k | return(1); |
552 | | #else |
553 | | return(0); |
554 | | #endif |
555 | 3.24k | case XML_WITH_WRITER: |
556 | 3.24k | #ifdef LIBXML_WRITER_ENABLED |
557 | 3.24k | return(1); |
558 | | #else |
559 | | return(0); |
560 | | #endif |
561 | 3.24k | case XML_WITH_SAX1: |
562 | 3.24k | #ifdef LIBXML_SAX1_ENABLED |
563 | 3.24k | return(1); |
564 | | #else |
565 | | return(0); |
566 | | #endif |
567 | 3.24k | case XML_WITH_HTTP: |
568 | | #ifdef LIBXML_HTTP_ENABLED |
569 | | return(1); |
570 | | #else |
571 | 3.24k | return(0); |
572 | 0 | #endif |
573 | 3.24k | case XML_WITH_VALID: |
574 | 3.24k | #ifdef LIBXML_VALID_ENABLED |
575 | 3.24k | return(1); |
576 | | #else |
577 | | return(0); |
578 | | #endif |
579 | 3.24k | case XML_WITH_HTML: |
580 | 3.24k | #ifdef LIBXML_HTML_ENABLED |
581 | 3.24k | return(1); |
582 | | #else |
583 | | return(0); |
584 | | #endif |
585 | 3.24k | case XML_WITH_LEGACY: |
586 | | #ifdef LIBXML_LEGACY_ENABLED |
587 | | return(1); |
588 | | #else |
589 | 3.24k | return(0); |
590 | 0 | #endif |
591 | 3.24k | case XML_WITH_C14N: |
592 | 3.24k | #ifdef LIBXML_C14N_ENABLED |
593 | 3.24k | return(1); |
594 | | #else |
595 | | return(0); |
596 | | #endif |
597 | 3.24k | case XML_WITH_CATALOG: |
598 | 3.24k | #ifdef LIBXML_CATALOG_ENABLED |
599 | 3.24k | return(1); |
600 | | #else |
601 | | return(0); |
602 | | #endif |
603 | 3.24k | case XML_WITH_XPATH: |
604 | 3.24k | #ifdef LIBXML_XPATH_ENABLED |
605 | 3.24k | return(1); |
606 | | #else |
607 | | return(0); |
608 | | #endif |
609 | 3.24k | case XML_WITH_XPTR: |
610 | 3.24k | #ifdef LIBXML_XPTR_ENABLED |
611 | 3.24k | return(1); |
612 | | #else |
613 | | return(0); |
614 | | #endif |
615 | 3.24k | case XML_WITH_XINCLUDE: |
616 | 3.24k | #ifdef LIBXML_XINCLUDE_ENABLED |
617 | 3.24k | return(1); |
618 | | #else |
619 | | return(0); |
620 | | #endif |
621 | 3.24k | case XML_WITH_ICONV: |
622 | 3.24k | #ifdef LIBXML_ICONV_ENABLED |
623 | 3.24k | return(1); |
624 | | #else |
625 | | return(0); |
626 | | #endif |
627 | 3.24k | case XML_WITH_ISO8859X: |
628 | 3.24k | #ifdef LIBXML_ISO8859X_ENABLED |
629 | 3.24k | return(1); |
630 | | #else |
631 | | return(0); |
632 | | #endif |
633 | 3.24k | case XML_WITH_UNICODE: |
634 | 3.24k | #ifdef LIBXML_UNICODE_ENABLED |
635 | 3.24k | return(1); |
636 | | #else |
637 | | return(0); |
638 | | #endif |
639 | 3.24k | case XML_WITH_REGEXP: |
640 | 3.24k | #ifdef LIBXML_REGEXP_ENABLED |
641 | 3.24k | return(1); |
642 | | #else |
643 | | return(0); |
644 | | #endif |
645 | 3.24k | case XML_WITH_AUTOMATA: |
646 | 3.24k | #ifdef LIBXML_REGEXP_ENABLED |
647 | 3.24k | return(1); |
648 | | #else |
649 | | return(0); |
650 | | #endif |
651 | 3.24k | case XML_WITH_EXPR: |
652 | | #ifdef LIBXML_EXPR_ENABLED |
653 | | return(1); |
654 | | #else |
655 | 3.24k | return(0); |
656 | 0 | #endif |
657 | 3.24k | case XML_WITH_SCHEMAS: |
658 | 3.24k | #ifdef LIBXML_SCHEMAS_ENABLED |
659 | 3.24k | return(1); |
660 | | #else |
661 | | return(0); |
662 | | #endif |
663 | 3.24k | case XML_WITH_SCHEMATRON: |
664 | 3.24k | #ifdef LIBXML_SCHEMATRON_ENABLED |
665 | 3.24k | return(1); |
666 | | #else |
667 | | return(0); |
668 | | #endif |
669 | 3.24k | case XML_WITH_MODULES: |
670 | 3.24k | #ifdef LIBXML_MODULES_ENABLED |
671 | 3.24k | return(1); |
672 | | #else |
673 | | return(0); |
674 | | #endif |
675 | 3.24k | case XML_WITH_DEBUG: |
676 | | #ifdef LIBXML_DEBUG_ENABLED |
677 | | return(1); |
678 | | #else |
679 | 3.24k | return(0); |
680 | 0 | #endif |
681 | 0 | case XML_WITH_DEBUG_MEM: |
682 | 0 | return(0); |
683 | 3.24k | case XML_WITH_ZLIB: |
684 | 3.24k | #ifdef LIBXML_ZLIB_ENABLED |
685 | 3.24k | return(1); |
686 | | #else |
687 | | return(0); |
688 | | #endif |
689 | 3.24k | case XML_WITH_LZMA: |
690 | 3.24k | #ifdef LIBXML_LZMA_ENABLED |
691 | 3.24k | return(1); |
692 | | #else |
693 | | return(0); |
694 | | #endif |
695 | 3.24k | case XML_WITH_ICU: |
696 | | #ifdef LIBXML_ICU_ENABLED |
697 | | return(1); |
698 | | #else |
699 | 3.24k | return(0); |
700 | 0 | #endif |
701 | 0 | default: |
702 | 0 | break; |
703 | 97.3k | } |
704 | 0 | return(0); |
705 | 97.3k | } |
706 | | |
707 | | /************************************************************************ |
708 | | * * |
709 | | * Simple string buffer * |
710 | | * * |
711 | | ************************************************************************/ |
712 | | |
713 | | typedef struct { |
714 | | xmlChar *mem; |
715 | | unsigned size; |
716 | | unsigned cap; /* size < cap */ |
717 | | unsigned max; /* size <= max */ |
718 | | xmlParserErrors code; |
719 | | } xmlSBuf; |
720 | | |
721 | | static void |
722 | 5.58M | xmlSBufInit(xmlSBuf *buf, unsigned max) { |
723 | 5.58M | buf->mem = NULL; |
724 | 5.58M | buf->size = 0; |
725 | 5.58M | buf->cap = 0; |
726 | 5.58M | buf->max = max; |
727 | 5.58M | buf->code = XML_ERR_OK; |
728 | 5.58M | } |
729 | | |
730 | | static int |
731 | 4.64M | xmlSBufGrow(xmlSBuf *buf, unsigned len) { |
732 | 4.64M | xmlChar *mem; |
733 | 4.64M | unsigned cap; |
734 | | |
735 | 4.64M | if (len >= UINT_MAX / 2 - buf->size) { |
736 | 0 | if (buf->code == XML_ERR_OK) |
737 | 0 | buf->code = XML_ERR_RESOURCE_LIMIT; |
738 | 0 | return(-1); |
739 | 0 | } |
740 | | |
741 | 4.64M | cap = (buf->size + len) * 2; |
742 | 4.64M | if (cap < 240) |
743 | 2.44M | cap = 240; |
744 | | |
745 | 4.64M | mem = xmlRealloc(buf->mem, cap); |
746 | 4.64M | if (mem == NULL) { |
747 | 1.78M | buf->code = XML_ERR_NO_MEMORY; |
748 | 1.78M | return(-1); |
749 | 1.78M | } |
750 | | |
751 | 2.86M | buf->mem = mem; |
752 | 2.86M | buf->cap = cap; |
753 | | |
754 | 2.86M | return(0); |
755 | 4.64M | } |
756 | | |
757 | | static void |
758 | 675M | xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) { |
759 | 675M | if (buf->max - buf->size < len) { |
760 | 84.3k | if (buf->code == XML_ERR_OK) |
761 | 177 | buf->code = XML_ERR_RESOURCE_LIMIT; |
762 | 84.3k | return; |
763 | 84.3k | } |
764 | | |
765 | 675M | if (buf->cap - buf->size <= len) { |
766 | 4.54M | if (xmlSBufGrow(buf, len) < 0) |
767 | 1.78M | return; |
768 | 4.54M | } |
769 | | |
770 | 673M | if (len > 0) |
771 | 673M | memcpy(buf->mem + buf->size, str, len); |
772 | 673M | buf->size += len; |
773 | 673M | } |
774 | | |
775 | | static void |
776 | 646M | xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) { |
777 | 646M | xmlSBufAddString(buf, (const xmlChar *) str, len); |
778 | 646M | } |
779 | | |
780 | | static void |
781 | 3.80M | xmlSBufAddChar(xmlSBuf *buf, int c) { |
782 | 3.80M | xmlChar *end; |
783 | | |
784 | 3.80M | if (buf->max - buf->size < 4) { |
785 | 182 | if (buf->code == XML_ERR_OK) |
786 | 0 | buf->code = XML_ERR_RESOURCE_LIMIT; |
787 | 182 | return; |
788 | 182 | } |
789 | | |
790 | 3.80M | if (buf->cap - buf->size <= 4) { |
791 | 103k | if (xmlSBufGrow(buf, 4) < 0) |
792 | 1.86k | return; |
793 | 103k | } |
794 | | |
795 | 3.80M | end = buf->mem + buf->size; |
796 | | |
797 | 3.80M | if (c < 0x80) { |
798 | 3.63M | *end = (xmlChar) c; |
799 | 3.63M | buf->size += 1; |
800 | 3.63M | } else { |
801 | 167k | buf->size += xmlCopyCharMultiByte(end, c); |
802 | 167k | } |
803 | 3.80M | } |
804 | | |
805 | | static void |
806 | 574M | xmlSBufAddReplChar(xmlSBuf *buf) { |
807 | 574M | xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3); |
808 | 574M | } |
809 | | |
810 | | static void |
811 | 1.56k | xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) { |
812 | 1.56k | if (buf->code == XML_ERR_NO_MEMORY) |
813 | 1.38k | xmlCtxtErrMemory(ctxt); |
814 | 177 | else |
815 | 177 | xmlFatalErr(ctxt, buf->code, errMsg); |
816 | 1.56k | } |
817 | | |
818 | | static xmlChar * |
819 | | xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt, |
820 | 2.84M | const char *errMsg) { |
821 | 2.84M | if (buf->mem == NULL) { |
822 | 427k | buf->mem = xmlMalloc(1); |
823 | 427k | if (buf->mem == NULL) { |
824 | 65 | buf->code = XML_ERR_NO_MEMORY; |
825 | 427k | } else { |
826 | 427k | buf->mem[0] = 0; |
827 | 427k | } |
828 | 2.41M | } else { |
829 | 2.41M | buf->mem[buf->size] = 0; |
830 | 2.41M | } |
831 | | |
832 | 2.84M | if (buf->code == XML_ERR_OK) { |
833 | 2.84M | if (sizeOut != NULL) |
834 | 654k | *sizeOut = buf->size; |
835 | 2.84M | return(buf->mem); |
836 | 2.84M | } |
837 | | |
838 | 1.11k | xmlSBufReportError(buf, ctxt, errMsg); |
839 | | |
840 | 1.11k | xmlFree(buf->mem); |
841 | | |
842 | 1.11k | if (sizeOut != NULL) |
843 | 175 | *sizeOut = 0; |
844 | 1.11k | return(NULL); |
845 | 2.84M | } |
846 | | |
847 | | static void |
848 | 2.59M | xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) { |
849 | 2.59M | if (buf->code != XML_ERR_OK) |
850 | 449 | xmlSBufReportError(buf, ctxt, errMsg); |
851 | | |
852 | 2.59M | xmlFree(buf->mem); |
853 | 2.59M | } |
854 | | |
855 | | static int |
856 | | xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str, |
857 | 672M | const char *errMsg) { |
858 | 672M | int c = str[0]; |
859 | 672M | int c1 = str[1]; |
860 | | |
861 | 672M | if ((c1 & 0xC0) != 0x80) |
862 | 175M | goto encoding_error; |
863 | | |
864 | 496M | if (c < 0xE0) { |
865 | | /* 2-byte sequence */ |
866 | 176M | if (c < 0xC2) |
867 | 134M | goto encoding_error; |
868 | | |
869 | 42.1M | return(2); |
870 | 320M | } else { |
871 | 320M | int c2 = str[2]; |
872 | | |
873 | 320M | if ((c2 & 0xC0) != 0x80) |
874 | 105k | goto encoding_error; |
875 | | |
876 | 320M | if (c < 0xF0) { |
877 | | /* 3-byte sequence */ |
878 | 319M | if (c == 0xE0) { |
879 | | /* overlong */ |
880 | 19.4M | if (c1 < 0xA0) |
881 | 9.04k | goto encoding_error; |
882 | 300M | } else if (c == 0xED) { |
883 | | /* surrogate */ |
884 | 1.40M | if (c1 >= 0xA0) |
885 | 6.22k | goto encoding_error; |
886 | 298M | } else if (c == 0xEF) { |
887 | | /* U+FFFE and U+FFFF are invalid Chars */ |
888 | 25.4M | if ((c1 == 0xBF) && (c2 >= 0xBE)) |
889 | 17.6k | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg); |
890 | 25.4M | } |
891 | | |
892 | 319M | return(3); |
893 | 319M | } else { |
894 | | /* 4-byte sequence */ |
895 | 245k | if ((str[3] & 0xC0) != 0x80) |
896 | 22.7k | goto encoding_error; |
897 | 222k | if (c == 0xF0) { |
898 | | /* overlong */ |
899 | 88.4k | if (c1 < 0x90) |
900 | 9.53k | goto encoding_error; |
901 | 133k | } else if (c >= 0xF4) { |
902 | | /* greater than 0x10FFFF */ |
903 | 35.2k | if ((c > 0xF4) || (c1 >= 0x90)) |
904 | 31.1k | goto encoding_error; |
905 | 35.2k | } |
906 | | |
907 | 181k | return(4); |
908 | 222k | } |
909 | 320M | } |
910 | | |
911 | 310M | encoding_error: |
912 | | /* Only report the first error */ |
913 | 310M | if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) { |
914 | 79.6k | xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL); |
915 | 79.6k | ctxt->input->flags |= XML_INPUT_ENCODING_ERROR; |
916 | 79.6k | } |
917 | | |
918 | 310M | return(0); |
919 | 496M | } |
920 | | |
921 | | /************************************************************************ |
922 | | * * |
923 | | * SAX2 defaulted attributes handling * |
924 | | * * |
925 | | ************************************************************************/ |
926 | | |
927 | | /** |
928 | | * xmlCtxtInitializeLate: |
929 | | * @ctxt: an XML parser context |
930 | | * |
931 | | * Final initialization of the parser context before starting to parse. |
932 | | * |
933 | | * This accounts for users modifying struct members of parser context |
934 | | * directly. |
935 | | */ |
936 | | static void |
937 | 480k | xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) { |
938 | 480k | xmlSAXHandlerPtr sax; |
939 | | |
940 | | /* Avoid unused variable warning if features are disabled. */ |
941 | 480k | (void) sax; |
942 | | |
943 | | /* |
944 | | * Changing the SAX struct directly is still widespread practice |
945 | | * in internal and external code. |
946 | | */ |
947 | 480k | if (ctxt == NULL) return; |
948 | 480k | sax = ctxt->sax; |
949 | 480k | #ifdef LIBXML_SAX1_ENABLED |
950 | | /* |
951 | | * Only enable SAX2 if there SAX2 element handlers, except when there |
952 | | * are no element handlers at all. |
953 | | */ |
954 | 480k | if (((ctxt->options & XML_PARSE_SAX1) == 0) && |
955 | 480k | (sax) && |
956 | 480k | (sax->initialized == XML_SAX2_MAGIC) && |
957 | 480k | ((sax->startElementNs != NULL) || |
958 | 324k | (sax->endElementNs != NULL) || |
959 | 324k | ((sax->startElement == NULL) && (sax->endElement == NULL)))) |
960 | 324k | ctxt->sax2 = 1; |
961 | | #else |
962 | | ctxt->sax2 = 1; |
963 | | #endif /* LIBXML_SAX1_ENABLED */ |
964 | | |
965 | | /* |
966 | | * Some users replace the dictionary directly in the context struct. |
967 | | * We really need an API function to do that cleanly. |
968 | | */ |
969 | 480k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
970 | 480k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
971 | 480k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
972 | 480k | if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) || |
973 | 480k | (ctxt->str_xml_ns == NULL)) { |
974 | 1.20k | xmlErrMemory(ctxt); |
975 | 1.20k | } |
976 | | |
977 | 480k | xmlDictSetLimit(ctxt->dict, |
978 | 480k | (ctxt->options & XML_PARSE_HUGE) ? |
979 | 135k | 0 : |
980 | 480k | XML_MAX_DICTIONARY_LIMIT); |
981 | 480k | } |
982 | | |
983 | | typedef struct { |
984 | | xmlHashedString prefix; |
985 | | xmlHashedString name; |
986 | | xmlHashedString value; |
987 | | const xmlChar *valueEnd; |
988 | | int external; |
989 | | int expandedSize; |
990 | | } xmlDefAttr; |
991 | | |
992 | | typedef struct _xmlDefAttrs xmlDefAttrs; |
993 | | typedef xmlDefAttrs *xmlDefAttrsPtr; |
994 | | struct _xmlDefAttrs { |
995 | | int nbAttrs; /* number of defaulted attributes on that element */ |
996 | | int maxAttrs; /* the size of the array */ |
997 | | #if __STDC_VERSION__ >= 199901L |
998 | | /* Using a C99 flexible array member avoids UBSan errors. */ |
999 | | xmlDefAttr attrs[]; /* array of localname/prefix/values/external */ |
1000 | | #else |
1001 | | xmlDefAttr attrs[1]; |
1002 | | #endif |
1003 | | }; |
1004 | | |
1005 | | /** |
1006 | | * xmlAttrNormalizeSpace: |
1007 | | * @src: the source string |
1008 | | * @dst: the target string |
1009 | | * |
1010 | | * Normalize the space in non CDATA attribute values: |
1011 | | * If the attribute type is not CDATA, then the XML processor MUST further |
1012 | | * process the normalized attribute value by discarding any leading and |
1013 | | * trailing space (#x20) characters, and by replacing sequences of space |
1014 | | * (#x20) characters by a single space (#x20) character. |
1015 | | * Note that the size of dst need to be at least src, and if one doesn't need |
1016 | | * to preserve dst (and it doesn't come from a dictionary or read-only) then |
1017 | | * passing src as dst is just fine. |
1018 | | * |
1019 | | * Returns a pointer to the normalized value (dst) or NULL if no conversion |
1020 | | * is needed. |
1021 | | */ |
1022 | | static xmlChar * |
1023 | | xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst) |
1024 | 264k | { |
1025 | 264k | if ((src == NULL) || (dst == NULL)) |
1026 | 0 | return(NULL); |
1027 | | |
1028 | 283k | while (*src == 0x20) src++; |
1029 | 323M | while (*src != 0) { |
1030 | 322M | if (*src == 0x20) { |
1031 | 17.6M | while (*src == 0x20) src++; |
1032 | 461k | if (*src != 0) |
1033 | 446k | *dst++ = 0x20; |
1034 | 322M | } else { |
1035 | 322M | *dst++ = *src++; |
1036 | 322M | } |
1037 | 322M | } |
1038 | 264k | *dst = 0; |
1039 | 264k | if (dst == src) |
1040 | 228k | return(NULL); |
1041 | 35.7k | return(dst); |
1042 | 264k | } |
1043 | | |
1044 | | /** |
1045 | | * xmlAddDefAttrs: |
1046 | | * @ctxt: an XML parser context |
1047 | | * @fullname: the element fullname |
1048 | | * @fullattr: the attribute fullname |
1049 | | * @value: the attribute value |
1050 | | * |
1051 | | * Add a defaulted attribute for an element |
1052 | | */ |
1053 | | static void |
1054 | | xmlAddDefAttrs(xmlParserCtxtPtr ctxt, |
1055 | | const xmlChar *fullname, |
1056 | | const xmlChar *fullattr, |
1057 | 282k | const xmlChar *value) { |
1058 | 282k | xmlDefAttrsPtr defaults; |
1059 | 282k | xmlDefAttr *attr; |
1060 | 282k | int len, expandedSize; |
1061 | 282k | xmlHashedString name; |
1062 | 282k | xmlHashedString prefix; |
1063 | 282k | xmlHashedString hvalue; |
1064 | 282k | const xmlChar *localname; |
1065 | | |
1066 | | /* |
1067 | | * Allows to detect attribute redefinitions |
1068 | | */ |
1069 | 282k | if (ctxt->attsSpecial != NULL) { |
1070 | 238k | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1071 | 147k | return; |
1072 | 238k | } |
1073 | | |
1074 | 134k | if (ctxt->attsDefault == NULL) { |
1075 | 44.9k | ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict); |
1076 | 44.9k | if (ctxt->attsDefault == NULL) |
1077 | 63 | goto mem_error; |
1078 | 44.9k | } |
1079 | | |
1080 | | /* |
1081 | | * split the element name into prefix:localname , the string found |
1082 | | * are within the DTD and then not associated to namespace names. |
1083 | | */ |
1084 | 134k | localname = xmlSplitQName3(fullname, &len); |
1085 | 134k | if (localname == NULL) { |
1086 | 106k | name = xmlDictLookupHashed(ctxt->dict, fullname, -1); |
1087 | 106k | prefix.name = NULL; |
1088 | 106k | } else { |
1089 | 28.2k | name = xmlDictLookupHashed(ctxt->dict, localname, -1); |
1090 | 28.2k | prefix = xmlDictLookupHashed(ctxt->dict, fullname, len); |
1091 | 28.2k | if (prefix.name == NULL) |
1092 | 7 | goto mem_error; |
1093 | 28.2k | } |
1094 | 134k | if (name.name == NULL) |
1095 | 7 | goto mem_error; |
1096 | | |
1097 | | /* |
1098 | | * make sure there is some storage |
1099 | | */ |
1100 | 134k | defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name); |
1101 | 134k | if ((defaults == NULL) || |
1102 | 134k | (defaults->nbAttrs >= defaults->maxAttrs)) { |
1103 | 58.8k | xmlDefAttrsPtr temp; |
1104 | 58.8k | int newSize; |
1105 | | |
1106 | 58.8k | newSize = (defaults != NULL) ? 2 * defaults->maxAttrs : 4; |
1107 | 58.8k | temp = xmlRealloc(defaults, |
1108 | 58.8k | sizeof(*defaults) + newSize * sizeof(xmlDefAttr)); |
1109 | 58.8k | if (temp == NULL) |
1110 | 42 | goto mem_error; |
1111 | 58.7k | if (defaults == NULL) |
1112 | 51.6k | temp->nbAttrs = 0; |
1113 | 58.7k | temp->maxAttrs = newSize; |
1114 | 58.7k | defaults = temp; |
1115 | 58.7k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name, |
1116 | 58.7k | defaults, NULL) < 0) { |
1117 | 0 | xmlFree(defaults); |
1118 | 0 | goto mem_error; |
1119 | 0 | } |
1120 | 58.7k | } |
1121 | | |
1122 | | /* |
1123 | | * Split the attribute name into prefix:localname , the string found |
1124 | | * are within the DTD and hen not associated to namespace names. |
1125 | | */ |
1126 | 134k | localname = xmlSplitQName3(fullattr, &len); |
1127 | 134k | if (localname == NULL) { |
1128 | 96.7k | name = xmlDictLookupHashed(ctxt->dict, fullattr, -1); |
1129 | 96.7k | prefix.name = NULL; |
1130 | 96.7k | } else { |
1131 | 37.8k | name = xmlDictLookupHashed(ctxt->dict, localname, -1); |
1132 | 37.8k | prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len); |
1133 | 37.8k | if (prefix.name == NULL) |
1134 | 7 | goto mem_error; |
1135 | 37.8k | } |
1136 | 134k | if (name.name == NULL) |
1137 | 8 | goto mem_error; |
1138 | | |
1139 | | /* intern the string and precompute the end */ |
1140 | 134k | len = strlen((const char *) value); |
1141 | 134k | hvalue = xmlDictLookupHashed(ctxt->dict, value, len); |
1142 | 134k | if (hvalue.name == NULL) |
1143 | 22 | goto mem_error; |
1144 | | |
1145 | 134k | expandedSize = strlen((const char *) name.name); |
1146 | 134k | if (prefix.name != NULL) |
1147 | 37.8k | expandedSize += strlen((const char *) prefix.name); |
1148 | 134k | expandedSize += len; |
1149 | | |
1150 | 134k | attr = &defaults->attrs[defaults->nbAttrs++]; |
1151 | 134k | attr->name = name; |
1152 | 134k | attr->prefix = prefix; |
1153 | 134k | attr->value = hvalue; |
1154 | 134k | attr->valueEnd = hvalue.name + len; |
1155 | 134k | attr->external = PARSER_EXTERNAL(ctxt); |
1156 | 134k | attr->expandedSize = expandedSize; |
1157 | | |
1158 | 134k | return; |
1159 | | |
1160 | 156 | mem_error: |
1161 | 156 | xmlErrMemory(ctxt); |
1162 | 156 | } |
1163 | | |
1164 | | /** |
1165 | | * xmlAddSpecialAttr: |
1166 | | * @ctxt: an XML parser context |
1167 | | * @fullname: the element fullname |
1168 | | * @fullattr: the attribute fullname |
1169 | | * @type: the attribute type |
1170 | | * |
1171 | | * Register this attribute type |
1172 | | */ |
1173 | | static void |
1174 | | xmlAddSpecialAttr(xmlParserCtxtPtr ctxt, |
1175 | | const xmlChar *fullname, |
1176 | | const xmlChar *fullattr, |
1177 | | int type) |
1178 | 511k | { |
1179 | 511k | if (ctxt->attsSpecial == NULL) { |
1180 | 55.4k | ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict); |
1181 | 55.4k | if (ctxt->attsSpecial == NULL) |
1182 | 79 | goto mem_error; |
1183 | 55.4k | } |
1184 | | |
1185 | 511k | if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr, |
1186 | 511k | (void *) (ptrdiff_t) type) < 0) |
1187 | 10 | goto mem_error; |
1188 | 511k | return; |
1189 | | |
1190 | 511k | mem_error: |
1191 | 89 | xmlErrMemory(ctxt); |
1192 | 89 | } |
1193 | | |
1194 | | /** |
1195 | | * xmlCleanSpecialAttrCallback: |
1196 | | * |
1197 | | * Removes CDATA attributes from the special attribute table |
1198 | | */ |
1199 | | static void |
1200 | | xmlCleanSpecialAttrCallback(void *payload, void *data, |
1201 | | const xmlChar *fullname, const xmlChar *fullattr, |
1202 | 275k | const xmlChar *unused ATTRIBUTE_UNUSED) { |
1203 | 275k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data; |
1204 | | |
1205 | 275k | if (((ptrdiff_t) payload) == XML_ATTRIBUTE_CDATA) { |
1206 | 71.7k | xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL); |
1207 | 71.7k | } |
1208 | 275k | } |
1209 | | |
1210 | | /** |
1211 | | * xmlCleanSpecialAttr: |
1212 | | * @ctxt: an XML parser context |
1213 | | * |
1214 | | * Trim the list of attributes defined to remove all those of type |
1215 | | * CDATA as they are not special. This call should be done when finishing |
1216 | | * to parse the DTD and before starting to parse the document root. |
1217 | | */ |
1218 | | static void |
1219 | | xmlCleanSpecialAttr(xmlParserCtxtPtr ctxt) |
1220 | 258k | { |
1221 | 258k | if (ctxt->attsSpecial == NULL) |
1222 | 202k | return; |
1223 | | |
1224 | 55.2k | xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt); |
1225 | | |
1226 | 55.2k | if (xmlHashSize(ctxt->attsSpecial) == 0) { |
1227 | 12.5k | xmlHashFree(ctxt->attsSpecial, NULL); |
1228 | 12.5k | ctxt->attsSpecial = NULL; |
1229 | 12.5k | } |
1230 | 55.2k | } |
1231 | | |
1232 | | /** |
1233 | | * xmlCheckLanguageID: |
1234 | | * @lang: pointer to the string value |
1235 | | * |
1236 | | * DEPRECATED: Internal function, do not use. |
1237 | | * |
1238 | | * Checks that the value conforms to the LanguageID production: |
1239 | | * |
1240 | | * NOTE: this is somewhat deprecated, those productions were removed from |
1241 | | * the XML Second edition. |
1242 | | * |
1243 | | * [33] LanguageID ::= Langcode ('-' Subcode)* |
1244 | | * [34] Langcode ::= ISO639Code | IanaCode | UserCode |
1245 | | * [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z]) |
1246 | | * [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+ |
1247 | | * [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+ |
1248 | | * [38] Subcode ::= ([a-z] | [A-Z])+ |
1249 | | * |
1250 | | * The current REC reference the successors of RFC 1766, currently 5646 |
1251 | | * |
1252 | | * http://www.rfc-editor.org/rfc/rfc5646.txt |
1253 | | * langtag = language |
1254 | | * ["-" script] |
1255 | | * ["-" region] |
1256 | | * *("-" variant) |
1257 | | * *("-" extension) |
1258 | | * ["-" privateuse] |
1259 | | * language = 2*3ALPHA ; shortest ISO 639 code |
1260 | | * ["-" extlang] ; sometimes followed by |
1261 | | * ; extended language subtags |
1262 | | * / 4ALPHA ; or reserved for future use |
1263 | | * / 5*8ALPHA ; or registered language subtag |
1264 | | * |
1265 | | * extlang = 3ALPHA ; selected ISO 639 codes |
1266 | | * *2("-" 3ALPHA) ; permanently reserved |
1267 | | * |
1268 | | * script = 4ALPHA ; ISO 15924 code |
1269 | | * |
1270 | | * region = 2ALPHA ; ISO 3166-1 code |
1271 | | * / 3DIGIT ; UN M.49 code |
1272 | | * |
1273 | | * variant = 5*8alphanum ; registered variants |
1274 | | * / (DIGIT 3alphanum) |
1275 | | * |
1276 | | * extension = singleton 1*("-" (2*8alphanum)) |
1277 | | * |
1278 | | * ; Single alphanumerics |
1279 | | * ; "x" reserved for private use |
1280 | | * singleton = DIGIT ; 0 - 9 |
1281 | | * / %x41-57 ; A - W |
1282 | | * / %x59-5A ; Y - Z |
1283 | | * / %x61-77 ; a - w |
1284 | | * / %x79-7A ; y - z |
1285 | | * |
1286 | | * it sounds right to still allow Irregular i-xxx IANA and user codes too |
1287 | | * The parser below doesn't try to cope with extension or privateuse |
1288 | | * that could be added but that's not interoperable anyway |
1289 | | * |
1290 | | * Returns 1 if correct 0 otherwise |
1291 | | **/ |
1292 | | int |
1293 | | xmlCheckLanguageID(const xmlChar * lang) |
1294 | 279k | { |
1295 | 279k | const xmlChar *cur = lang, *nxt; |
1296 | | |
1297 | 279k | if (cur == NULL) |
1298 | 9.03k | return (0); |
1299 | 270k | if (((cur[0] == 'i') && (cur[1] == '-')) || |
1300 | 270k | ((cur[0] == 'I') && (cur[1] == '-')) || |
1301 | 270k | ((cur[0] == 'x') && (cur[1] == '-')) || |
1302 | 270k | ((cur[0] == 'X') && (cur[1] == '-'))) { |
1303 | | /* |
1304 | | * Still allow IANA code and user code which were coming |
1305 | | * from the previous version of the XML-1.0 specification |
1306 | | * it's deprecated but we should not fail |
1307 | | */ |
1308 | 14.6k | cur += 2; |
1309 | 102k | while (((cur[0] >= 'A') && (cur[0] <= 'Z')) || |
1310 | 102k | ((cur[0] >= 'a') && (cur[0] <= 'z'))) |
1311 | 87.9k | cur++; |
1312 | 14.6k | return(cur[0] == 0); |
1313 | 14.6k | } |
1314 | 255k | nxt = cur; |
1315 | 918k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1316 | 918k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1317 | 662k | nxt++; |
1318 | 255k | if (nxt - cur >= 4) { |
1319 | | /* |
1320 | | * Reserved |
1321 | | */ |
1322 | 10.9k | if ((nxt - cur > 8) || (nxt[0] != 0)) |
1323 | 7.67k | return(0); |
1324 | 3.29k | return(1); |
1325 | 10.9k | } |
1326 | 244k | if (nxt - cur < 2) |
1327 | 19.6k | return(0); |
1328 | | /* we got an ISO 639 code */ |
1329 | 224k | if (nxt[0] == 0) |
1330 | 13.3k | return(1); |
1331 | 211k | if (nxt[0] != '-') |
1332 | 13.4k | return(0); |
1333 | | |
1334 | 198k | nxt++; |
1335 | 198k | cur = nxt; |
1336 | | /* now we can have extlang or script or region or variant */ |
1337 | 198k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1338 | 7.04k | goto region_m49; |
1339 | | |
1340 | 826k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1341 | 826k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1342 | 635k | nxt++; |
1343 | 191k | if (nxt - cur == 4) |
1344 | 46.8k | goto script; |
1345 | 144k | if (nxt - cur == 2) |
1346 | 39.4k | goto region; |
1347 | 104k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1348 | 4.74k | goto variant; |
1349 | 100k | if (nxt - cur != 3) |
1350 | 13.5k | return(0); |
1351 | | /* we parsed an extlang */ |
1352 | 86.5k | if (nxt[0] == 0) |
1353 | 4.47k | return(1); |
1354 | 82.0k | if (nxt[0] != '-') |
1355 | 7.42k | return(0); |
1356 | | |
1357 | 74.6k | nxt++; |
1358 | 74.6k | cur = nxt; |
1359 | | /* now we can have script or region or variant */ |
1360 | 74.6k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1361 | 4.99k | goto region_m49; |
1362 | | |
1363 | 359k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1364 | 359k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1365 | 289k | nxt++; |
1366 | 69.6k | if (nxt - cur == 2) |
1367 | 7.95k | goto region; |
1368 | 61.6k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1369 | 4.05k | goto variant; |
1370 | 57.6k | if (nxt - cur != 4) |
1371 | 21.0k | return(0); |
1372 | | /* we parsed a script */ |
1373 | 83.3k | script: |
1374 | 83.3k | if (nxt[0] == 0) |
1375 | 1.61k | return(1); |
1376 | 81.7k | if (nxt[0] != '-') |
1377 | 11.5k | return(0); |
1378 | | |
1379 | 70.2k | nxt++; |
1380 | 70.2k | cur = nxt; |
1381 | | /* now we can have region or variant */ |
1382 | 70.2k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1383 | 5.71k | goto region_m49; |
1384 | | |
1385 | 353k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1386 | 353k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1387 | 289k | nxt++; |
1388 | | |
1389 | 64.5k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1390 | 15.9k | goto variant; |
1391 | 48.5k | if (nxt - cur != 2) |
1392 | 44.3k | return(0); |
1393 | | /* we parsed a region */ |
1394 | 59.4k | region: |
1395 | 59.4k | if (nxt[0] == 0) |
1396 | 8.66k | return(1); |
1397 | 50.7k | if (nxt[0] != '-') |
1398 | 12.8k | return(0); |
1399 | | |
1400 | 37.9k | nxt++; |
1401 | 37.9k | cur = nxt; |
1402 | | /* now we can just have a variant */ |
1403 | 269k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1404 | 269k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1405 | 231k | nxt++; |
1406 | | |
1407 | 37.9k | if ((nxt - cur < 5) || (nxt - cur > 8)) |
1408 | 18.4k | return(0); |
1409 | | |
1410 | | /* we parsed a variant */ |
1411 | 44.2k | variant: |
1412 | 44.2k | if (nxt[0] == 0) |
1413 | 1.80k | return(1); |
1414 | 42.4k | if (nxt[0] != '-') |
1415 | 38.5k | return(0); |
1416 | | /* extensions and private use subtags not checked */ |
1417 | 3.95k | return (1); |
1418 | | |
1419 | 17.7k | region_m49: |
1420 | 17.7k | if (((nxt[1] >= '0') && (nxt[1] <= '9')) && |
1421 | 17.7k | ((nxt[2] >= '0') && (nxt[2] <= '9'))) { |
1422 | 7.92k | nxt += 3; |
1423 | 7.92k | goto region; |
1424 | 7.92k | } |
1425 | 9.83k | return(0); |
1426 | 17.7k | } |
1427 | | |
1428 | | /************************************************************************ |
1429 | | * * |
1430 | | * Parser stacks related functions and macros * |
1431 | | * * |
1432 | | ************************************************************************/ |
1433 | | |
1434 | | static xmlChar * |
1435 | | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar **str); |
1436 | | |
1437 | | /** |
1438 | | * xmlParserNsCreate: |
1439 | | * |
1440 | | * Create a new namespace database. |
1441 | | * |
1442 | | * Returns the new obejct. |
1443 | | */ |
1444 | | xmlParserNsData * |
1445 | 512k | xmlParserNsCreate(void) { |
1446 | 512k | xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb)); |
1447 | | |
1448 | 512k | if (nsdb == NULL) |
1449 | 84 | return(NULL); |
1450 | 512k | memset(nsdb, 0, sizeof(*nsdb)); |
1451 | 512k | nsdb->defaultNsIndex = INT_MAX; |
1452 | | |
1453 | 512k | return(nsdb); |
1454 | 512k | } |
1455 | | |
1456 | | /** |
1457 | | * xmlParserNsFree: |
1458 | | * @nsdb: namespace database |
1459 | | * |
1460 | | * Free a namespace database. |
1461 | | */ |
1462 | | void |
1463 | 512k | xmlParserNsFree(xmlParserNsData *nsdb) { |
1464 | 512k | if (nsdb == NULL) |
1465 | 0 | return; |
1466 | | |
1467 | 512k | xmlFree(nsdb->extra); |
1468 | 512k | xmlFree(nsdb->hash); |
1469 | 512k | xmlFree(nsdb); |
1470 | 512k | } |
1471 | | |
1472 | | /** |
1473 | | * xmlParserNsReset: |
1474 | | * @nsdb: namespace database |
1475 | | * |
1476 | | * Reset a namespace database. |
1477 | | */ |
1478 | | static void |
1479 | 121k | xmlParserNsReset(xmlParserNsData *nsdb) { |
1480 | 121k | if (nsdb == NULL) |
1481 | 0 | return; |
1482 | | |
1483 | 121k | nsdb->hashElems = 0; |
1484 | 121k | nsdb->elementId = 0; |
1485 | 121k | nsdb->defaultNsIndex = INT_MAX; |
1486 | | |
1487 | 121k | if (nsdb->hash) |
1488 | 0 | memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0])); |
1489 | 121k | } |
1490 | | |
1491 | | /** |
1492 | | * xmlParserStartElement: |
1493 | | * @nsdb: namespace database |
1494 | | * |
1495 | | * Signal that a new element has started. |
1496 | | * |
1497 | | * Returns 0 on success, -1 if the element counter overflowed. |
1498 | | */ |
1499 | | static int |
1500 | 8.23M | xmlParserNsStartElement(xmlParserNsData *nsdb) { |
1501 | 8.23M | if (nsdb->elementId == UINT_MAX) |
1502 | 0 | return(-1); |
1503 | 8.23M | nsdb->elementId++; |
1504 | | |
1505 | 8.23M | return(0); |
1506 | 8.23M | } |
1507 | | |
1508 | | /** |
1509 | | * xmlParserNsLookup: |
1510 | | * @ctxt: parser context |
1511 | | * @prefix: namespace prefix |
1512 | | * @bucketPtr: optional bucket (return value) |
1513 | | * |
1514 | | * Lookup namespace with given prefix. If @bucketPtr is non-NULL, it will |
1515 | | * be set to the matching bucket, or the first empty bucket if no match |
1516 | | * was found. |
1517 | | * |
1518 | | * Returns the namespace index on success, INT_MAX if no namespace was |
1519 | | * found. |
1520 | | */ |
1521 | | static int |
1522 | | xmlParserNsLookup(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix, |
1523 | 11.4M | xmlParserNsBucket **bucketPtr) { |
1524 | 11.4M | xmlParserNsBucket *bucket, *tombstone; |
1525 | 11.4M | unsigned index, hashValue; |
1526 | | |
1527 | 11.4M | if (prefix->name == NULL) |
1528 | 6.86M | return(ctxt->nsdb->defaultNsIndex); |
1529 | | |
1530 | 4.62M | if (ctxt->nsdb->hashSize == 0) |
1531 | 295k | return(INT_MAX); |
1532 | | |
1533 | 4.33M | hashValue = prefix->hashValue; |
1534 | 4.33M | index = hashValue & (ctxt->nsdb->hashSize - 1); |
1535 | 4.33M | bucket = &ctxt->nsdb->hash[index]; |
1536 | 4.33M | tombstone = NULL; |
1537 | | |
1538 | 5.17M | while (bucket->hashValue) { |
1539 | 4.35M | if (bucket->index == INT_MAX) { |
1540 | 397k | if (tombstone == NULL) |
1541 | 371k | tombstone = bucket; |
1542 | 3.96M | } else if (bucket->hashValue == hashValue) { |
1543 | 3.51M | if (ctxt->nsTab[bucket->index * 2] == prefix->name) { |
1544 | 3.51M | if (bucketPtr != NULL) |
1545 | 932k | *bucketPtr = bucket; |
1546 | 3.51M | return(bucket->index); |
1547 | 3.51M | } |
1548 | 3.51M | } |
1549 | | |
1550 | 844k | index++; |
1551 | 844k | bucket++; |
1552 | 844k | if (index == ctxt->nsdb->hashSize) { |
1553 | 83.6k | index = 0; |
1554 | 83.6k | bucket = ctxt->nsdb->hash; |
1555 | 83.6k | } |
1556 | 844k | } |
1557 | | |
1558 | 817k | if (bucketPtr != NULL) |
1559 | 245k | *bucketPtr = tombstone ? tombstone : bucket; |
1560 | 817k | return(INT_MAX); |
1561 | 4.33M | } |
1562 | | |
1563 | | /** |
1564 | | * xmlParserNsLookupUri: |
1565 | | * @ctxt: parser context |
1566 | | * @prefix: namespace prefix |
1567 | | * |
1568 | | * Lookup namespace URI with given prefix. |
1569 | | * |
1570 | | * Returns the namespace URI on success, NULL if no namespace was found. |
1571 | | */ |
1572 | | static const xmlChar * |
1573 | 6.60M | xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) { |
1574 | 6.60M | const xmlChar *ret; |
1575 | 6.60M | int nsIndex; |
1576 | | |
1577 | 6.60M | if (prefix->name == ctxt->str_xml) |
1578 | 14.0k | return(ctxt->str_xml_ns); |
1579 | | |
1580 | | /* |
1581 | | * minNsIndex is used when building an entity tree. We must |
1582 | | * ignore namespaces declared outside the entity. |
1583 | | */ |
1584 | 6.58M | nsIndex = xmlParserNsLookup(ctxt, prefix, NULL); |
1585 | 6.58M | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1586 | 4.37M | return(NULL); |
1587 | | |
1588 | 2.21M | ret = ctxt->nsTab[nsIndex * 2 + 1]; |
1589 | 2.21M | if (ret[0] == 0) |
1590 | 55.7k | ret = NULL; |
1591 | 2.21M | return(ret); |
1592 | 6.58M | } |
1593 | | |
1594 | | /** |
1595 | | * xmlParserNsLookupSax: |
1596 | | * @ctxt: parser context |
1597 | | * @prefix: namespace prefix |
1598 | | * |
1599 | | * Lookup extra data for the given prefix. This returns data stored |
1600 | | * with xmlParserNsUdpateSax. |
1601 | | * |
1602 | | * Returns the data on success, NULL if no namespace was found. |
1603 | | */ |
1604 | | void * |
1605 | 2.21M | xmlParserNsLookupSax(xmlParserCtxtPtr ctxt, const xmlChar *prefix) { |
1606 | 2.21M | xmlHashedString hprefix; |
1607 | 2.21M | int nsIndex; |
1608 | | |
1609 | 2.21M | if (prefix == ctxt->str_xml) |
1610 | 350k | return(NULL); |
1611 | | |
1612 | 1.86M | hprefix.name = prefix; |
1613 | 1.86M | if (prefix != NULL) |
1614 | 935k | hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix); |
1615 | 932k | else |
1616 | 932k | hprefix.hashValue = 0; |
1617 | 1.86M | nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL); |
1618 | 1.86M | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1619 | 0 | return(NULL); |
1620 | | |
1621 | 1.86M | return(ctxt->nsdb->extra[nsIndex].saxData); |
1622 | 1.86M | } |
1623 | | |
1624 | | /** |
1625 | | * xmlParserNsUpdateSax: |
1626 | | * @ctxt: parser context |
1627 | | * @prefix: namespace prefix |
1628 | | * @saxData: extra data for SAX handler |
1629 | | * |
1630 | | * Sets or updates extra data for the given prefix. This value will be |
1631 | | * returned by xmlParserNsLookupSax as long as the namespace with the |
1632 | | * given prefix is in scope. |
1633 | | * |
1634 | | * Returns the data on success, NULL if no namespace was found. |
1635 | | */ |
1636 | | int |
1637 | | xmlParserNsUpdateSax(xmlParserCtxtPtr ctxt, const xmlChar *prefix, |
1638 | 929k | void *saxData) { |
1639 | 929k | xmlHashedString hprefix; |
1640 | 929k | int nsIndex; |
1641 | | |
1642 | 929k | if (prefix == ctxt->str_xml) |
1643 | 0 | return(-1); |
1644 | | |
1645 | 929k | hprefix.name = prefix; |
1646 | 929k | if (prefix != NULL) |
1647 | 526k | hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix); |
1648 | 403k | else |
1649 | 403k | hprefix.hashValue = 0; |
1650 | 929k | nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL); |
1651 | 929k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1652 | 0 | return(-1); |
1653 | | |
1654 | 929k | ctxt->nsdb->extra[nsIndex].saxData = saxData; |
1655 | 929k | return(0); |
1656 | 929k | } |
1657 | | |
1658 | | /** |
1659 | | * xmlParserNsGrow: |
1660 | | * @ctxt: parser context |
1661 | | * |
1662 | | * Grows the namespace tables. |
1663 | | * |
1664 | | * Returns 0 on success, -1 if a memory allocation failed. |
1665 | | */ |
1666 | | static int |
1667 | 70.6k | xmlParserNsGrow(xmlParserCtxtPtr ctxt) { |
1668 | 70.6k | const xmlChar **table; |
1669 | 70.6k | xmlParserNsExtra *extra; |
1670 | 70.6k | int newSize; |
1671 | | |
1672 | 70.6k | if (ctxt->nsMax > INT_MAX / 2) |
1673 | 0 | goto error; |
1674 | 70.6k | newSize = ctxt->nsMax ? ctxt->nsMax * 2 : 16; |
1675 | | |
1676 | 70.6k | table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0])); |
1677 | 70.6k | if (table == NULL) |
1678 | 53 | goto error; |
1679 | 70.5k | ctxt->nsTab = table; |
1680 | | |
1681 | 70.5k | extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0])); |
1682 | 70.5k | if (extra == NULL) |
1683 | 53 | goto error; |
1684 | 70.5k | ctxt->nsdb->extra = extra; |
1685 | | |
1686 | 70.5k | ctxt->nsMax = newSize; |
1687 | 70.5k | return(0); |
1688 | | |
1689 | 106 | error: |
1690 | 106 | xmlErrMemory(ctxt); |
1691 | 106 | return(-1); |
1692 | 70.5k | } |
1693 | | |
1694 | | /** |
1695 | | * xmlParserNsPush: |
1696 | | * @ctxt: parser context |
1697 | | * @prefix: prefix with hash value |
1698 | | * @uri: uri with hash value |
1699 | | * @saxData: extra data for SAX handler |
1700 | | * @defAttr: whether the namespace comes from a default attribute |
1701 | | * |
1702 | | * Push a new namespace on the table. |
1703 | | * |
1704 | | * Returns 1 if the namespace was pushed, 0 if the namespace was ignored, |
1705 | | * -1 if a memory allocation failed. |
1706 | | */ |
1707 | | static int |
1708 | | xmlParserNsPush(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix, |
1709 | 1.14M | const xmlHashedString *uri, void *saxData, int defAttr) { |
1710 | 1.14M | xmlParserNsBucket *bucket = NULL; |
1711 | 1.14M | xmlParserNsExtra *extra; |
1712 | 1.14M | const xmlChar **ns; |
1713 | 1.14M | unsigned hashValue, nsIndex, oldIndex; |
1714 | | |
1715 | 1.14M | if ((prefix != NULL) && (prefix->name == ctxt->str_xml)) |
1716 | 1.09k | return(0); |
1717 | | |
1718 | 1.14M | if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) { |
1719 | 106 | xmlErrMemory(ctxt); |
1720 | 106 | return(-1); |
1721 | 106 | } |
1722 | | |
1723 | | /* |
1724 | | * Default namespace and 'xml' namespace |
1725 | | */ |
1726 | 1.14M | if ((prefix == NULL) || (prefix->name == NULL)) { |
1727 | 500k | oldIndex = ctxt->nsdb->defaultNsIndex; |
1728 | | |
1729 | 500k | if (oldIndex != INT_MAX) { |
1730 | 440k | extra = &ctxt->nsdb->extra[oldIndex]; |
1731 | | |
1732 | 440k | if (extra->elementId == ctxt->nsdb->elementId) { |
1733 | 11.5k | if (defAttr == 0) |
1734 | 8.54k | xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns"); |
1735 | 11.5k | return(0); |
1736 | 11.5k | } |
1737 | | |
1738 | 428k | if ((ctxt->options & XML_PARSE_NSCLEAN) && |
1739 | 428k | (uri->name == ctxt->nsTab[oldIndex * 2 + 1])) |
1740 | 23.8k | return(0); |
1741 | 428k | } |
1742 | | |
1743 | 464k | ctxt->nsdb->defaultNsIndex = ctxt->nsNr; |
1744 | 464k | goto populate_entry; |
1745 | 500k | } |
1746 | | |
1747 | | /* |
1748 | | * Hash table lookup |
1749 | | */ |
1750 | 645k | oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket); |
1751 | 645k | if (oldIndex != INT_MAX) { |
1752 | 349k | extra = &ctxt->nsdb->extra[oldIndex]; |
1753 | | |
1754 | | /* |
1755 | | * Check for duplicate definitions on the same element. |
1756 | | */ |
1757 | 349k | if (extra->elementId == ctxt->nsdb->elementId) { |
1758 | 10.2k | if (defAttr == 0) |
1759 | 8.51k | xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name); |
1760 | 10.2k | return(0); |
1761 | 10.2k | } |
1762 | | |
1763 | 339k | if ((ctxt->options & XML_PARSE_NSCLEAN) && |
1764 | 339k | (uri->name == ctxt->nsTab[bucket->index * 2 + 1])) |
1765 | 28.5k | return(0); |
1766 | | |
1767 | 310k | bucket->index = ctxt->nsNr; |
1768 | 310k | goto populate_entry; |
1769 | 339k | } |
1770 | | |
1771 | | /* |
1772 | | * Insert new bucket |
1773 | | */ |
1774 | | |
1775 | 295k | hashValue = prefix->hashValue; |
1776 | | |
1777 | | /* |
1778 | | * Grow hash table, 50% fill factor |
1779 | | */ |
1780 | 295k | if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) { |
1781 | 59.2k | xmlParserNsBucket *newHash; |
1782 | 59.2k | unsigned newSize, i, index; |
1783 | | |
1784 | 59.2k | if (ctxt->nsdb->hashSize > UINT_MAX / 2) { |
1785 | 0 | xmlErrMemory(ctxt); |
1786 | 0 | return(-1); |
1787 | 0 | } |
1788 | 59.2k | newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16; |
1789 | 59.2k | newHash = xmlMalloc(newSize * sizeof(newHash[0])); |
1790 | 59.2k | if (newHash == NULL) { |
1791 | 37 | xmlErrMemory(ctxt); |
1792 | 37 | return(-1); |
1793 | 37 | } |
1794 | 59.2k | memset(newHash, 0, newSize * sizeof(newHash[0])); |
1795 | | |
1796 | 668k | for (i = 0; i < ctxt->nsdb->hashSize; i++) { |
1797 | 609k | unsigned hv = ctxt->nsdb->hash[i].hashValue; |
1798 | 609k | unsigned newIndex; |
1799 | | |
1800 | 609k | if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX)) |
1801 | 595k | continue; |
1802 | 13.7k | newIndex = hv & (newSize - 1); |
1803 | | |
1804 | 17.7k | while (newHash[newIndex].hashValue != 0) { |
1805 | 3.97k | newIndex++; |
1806 | 3.97k | if (newIndex == newSize) |
1807 | 827 | newIndex = 0; |
1808 | 3.97k | } |
1809 | | |
1810 | 13.7k | newHash[newIndex] = ctxt->nsdb->hash[i]; |
1811 | 13.7k | } |
1812 | | |
1813 | 59.2k | xmlFree(ctxt->nsdb->hash); |
1814 | 59.2k | ctxt->nsdb->hash = newHash; |
1815 | 59.2k | ctxt->nsdb->hashSize = newSize; |
1816 | | |
1817 | | /* |
1818 | | * Relookup |
1819 | | */ |
1820 | 59.2k | index = hashValue & (newSize - 1); |
1821 | | |
1822 | 61.6k | while (newHash[index].hashValue != 0) { |
1823 | 2.43k | index++; |
1824 | 2.43k | if (index == newSize) |
1825 | 242 | index = 0; |
1826 | 2.43k | } |
1827 | | |
1828 | 59.2k | bucket = &newHash[index]; |
1829 | 59.2k | } |
1830 | | |
1831 | 295k | bucket->hashValue = hashValue; |
1832 | 295k | bucket->index = ctxt->nsNr; |
1833 | 295k | ctxt->nsdb->hashElems++; |
1834 | 295k | oldIndex = INT_MAX; |
1835 | | |
1836 | 1.07M | populate_entry: |
1837 | 1.07M | nsIndex = ctxt->nsNr; |
1838 | | |
1839 | 1.07M | ns = &ctxt->nsTab[nsIndex * 2]; |
1840 | 1.07M | ns[0] = prefix ? prefix->name : NULL; |
1841 | 1.07M | ns[1] = uri->name; |
1842 | | |
1843 | 1.07M | extra = &ctxt->nsdb->extra[nsIndex]; |
1844 | 1.07M | extra->saxData = saxData; |
1845 | 1.07M | extra->prefixHashValue = prefix ? prefix->hashValue : 0; |
1846 | 1.07M | extra->uriHashValue = uri->hashValue; |
1847 | 1.07M | extra->elementId = ctxt->nsdb->elementId; |
1848 | 1.07M | extra->oldIndex = oldIndex; |
1849 | | |
1850 | 1.07M | ctxt->nsNr++; |
1851 | | |
1852 | 1.07M | return(1); |
1853 | 295k | } |
1854 | | |
1855 | | /** |
1856 | | * xmlParserNsPop: |
1857 | | * @ctxt: an XML parser context |
1858 | | * @nr: the number to pop |
1859 | | * |
1860 | | * Pops the top @nr namespaces and restores the hash table. |
1861 | | * |
1862 | | * Returns the number of namespaces popped. |
1863 | | */ |
1864 | | static int |
1865 | | xmlParserNsPop(xmlParserCtxtPtr ctxt, int nr) |
1866 | 817k | { |
1867 | 817k | int i; |
1868 | | |
1869 | | /* assert(nr <= ctxt->nsNr); */ |
1870 | | |
1871 | 1.84M | for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) { |
1872 | 1.02M | const xmlChar *prefix = ctxt->nsTab[i * 2]; |
1873 | 1.02M | xmlParserNsExtra *extra = &ctxt->nsdb->extra[i]; |
1874 | | |
1875 | 1.02M | if (prefix == NULL) { |
1876 | 443k | ctxt->nsdb->defaultNsIndex = extra->oldIndex; |
1877 | 582k | } else { |
1878 | 582k | xmlHashedString hprefix; |
1879 | 582k | xmlParserNsBucket *bucket = NULL; |
1880 | | |
1881 | 582k | hprefix.name = prefix; |
1882 | 582k | hprefix.hashValue = extra->prefixHashValue; |
1883 | 582k | xmlParserNsLookup(ctxt, &hprefix, &bucket); |
1884 | | /* assert(bucket && bucket->hashValue); */ |
1885 | 582k | bucket->index = extra->oldIndex; |
1886 | 582k | } |
1887 | 1.02M | } |
1888 | | |
1889 | 817k | ctxt->nsNr -= nr; |
1890 | 817k | return(nr); |
1891 | 817k | } |
1892 | | |
1893 | | static int |
1894 | 96.4k | xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt, int nr) { |
1895 | 96.4k | const xmlChar **atts; |
1896 | 96.4k | unsigned *attallocs; |
1897 | 96.4k | int maxatts; |
1898 | | |
1899 | 96.4k | if (nr + 5 > ctxt->maxatts) { |
1900 | 96.4k | maxatts = ctxt->maxatts == 0 ? 55 : (nr + 5) * 2; |
1901 | 96.4k | atts = (const xmlChar **) xmlMalloc( |
1902 | 96.4k | maxatts * sizeof(const xmlChar *)); |
1903 | 96.4k | if (atts == NULL) goto mem_error; |
1904 | 96.4k | attallocs = xmlRealloc(ctxt->attallocs, |
1905 | 96.4k | (maxatts / 5) * sizeof(attallocs[0])); |
1906 | 96.4k | if (attallocs == NULL) { |
1907 | 63 | xmlFree(atts); |
1908 | 63 | goto mem_error; |
1909 | 63 | } |
1910 | 96.3k | if (ctxt->maxatts > 0) |
1911 | 8.02k | memcpy(atts, ctxt->atts, ctxt->maxatts * sizeof(const xmlChar *)); |
1912 | 96.3k | xmlFree(ctxt->atts); |
1913 | 96.3k | ctxt->atts = atts; |
1914 | 96.3k | ctxt->attallocs = attallocs; |
1915 | 96.3k | ctxt->maxatts = maxatts; |
1916 | 96.3k | } |
1917 | 96.3k | return(ctxt->maxatts); |
1918 | 119 | mem_error: |
1919 | 119 | xmlErrMemory(ctxt); |
1920 | 119 | return(-1); |
1921 | 96.4k | } |
1922 | | |
1923 | | /** |
1924 | | * inputPush: |
1925 | | * @ctxt: an XML parser context |
1926 | | * @value: the parser input |
1927 | | * |
1928 | | * Pushes a new parser input on top of the input stack |
1929 | | * |
1930 | | * Returns -1 in case of error, the index in the stack otherwise |
1931 | | */ |
1932 | | int |
1933 | | inputPush(xmlParserCtxtPtr ctxt, xmlParserInputPtr value) |
1934 | 1.26M | { |
1935 | 1.26M | char *directory = NULL; |
1936 | | |
1937 | 1.26M | if ((ctxt == NULL) || (value == NULL)) |
1938 | 0 | return(-1); |
1939 | | |
1940 | 1.26M | if (ctxt->inputNr >= ctxt->inputMax) { |
1941 | 40.5k | size_t newSize = ctxt->inputMax * 2; |
1942 | 40.5k | xmlParserInputPtr *tmp; |
1943 | | |
1944 | 40.5k | tmp = (xmlParserInputPtr *) xmlRealloc(ctxt->inputTab, |
1945 | 40.5k | newSize * sizeof(*tmp)); |
1946 | 40.5k | if (tmp == NULL) { |
1947 | 81 | xmlErrMemory(ctxt); |
1948 | 81 | return (-1); |
1949 | 81 | } |
1950 | 40.4k | ctxt->inputTab = tmp; |
1951 | 40.4k | ctxt->inputMax = newSize; |
1952 | 40.4k | } |
1953 | | |
1954 | 1.26M | if ((ctxt->inputNr == 0) && (value->filename != NULL)) { |
1955 | 413k | directory = xmlParserGetDirectory(value->filename); |
1956 | 413k | if (directory == NULL) { |
1957 | 112 | xmlErrMemory(ctxt); |
1958 | 112 | return(-1); |
1959 | 112 | } |
1960 | 413k | } |
1961 | | |
1962 | 1.26M | ctxt->inputTab[ctxt->inputNr] = value; |
1963 | 1.26M | ctxt->input = value; |
1964 | | |
1965 | 1.26M | if (ctxt->inputNr == 0) { |
1966 | 536k | xmlFree(ctxt->directory); |
1967 | 536k | ctxt->directory = directory; |
1968 | 536k | } |
1969 | | |
1970 | 1.26M | return(ctxt->inputNr++); |
1971 | 1.26M | } |
1972 | | /** |
1973 | | * inputPop: |
1974 | | * @ctxt: an XML parser context |
1975 | | * |
1976 | | * Pops the top parser input from the input stack |
1977 | | * |
1978 | | * Returns the input just removed |
1979 | | */ |
1980 | | xmlParserInputPtr |
1981 | | inputPop(xmlParserCtxtPtr ctxt) |
1982 | 2.44M | { |
1983 | 2.44M | xmlParserInputPtr ret; |
1984 | | |
1985 | 2.44M | if (ctxt == NULL) |
1986 | 0 | return(NULL); |
1987 | 2.44M | if (ctxt->inputNr <= 0) |
1988 | 1.18M | return (NULL); |
1989 | 1.25M | ctxt->inputNr--; |
1990 | 1.25M | if (ctxt->inputNr > 0) |
1991 | 729k | ctxt->input = ctxt->inputTab[ctxt->inputNr - 1]; |
1992 | 527k | else |
1993 | 527k | ctxt->input = NULL; |
1994 | 1.25M | ret = ctxt->inputTab[ctxt->inputNr]; |
1995 | 1.25M | ctxt->inputTab[ctxt->inputNr] = NULL; |
1996 | 1.25M | return (ret); |
1997 | 2.44M | } |
1998 | | /** |
1999 | | * nodePush: |
2000 | | * @ctxt: an XML parser context |
2001 | | * @value: the element node |
2002 | | * |
2003 | | * DEPRECATED: Internal function, do not use. |
2004 | | * |
2005 | | * Pushes a new element node on top of the node stack |
2006 | | * |
2007 | | * Returns -1 in case of error, the index in the stack otherwise |
2008 | | */ |
2009 | | int |
2010 | | nodePush(xmlParserCtxtPtr ctxt, xmlNodePtr value) |
2011 | 10.6M | { |
2012 | 10.6M | int maxDepth; |
2013 | | |
2014 | 10.6M | if (ctxt == NULL) |
2015 | 0 | return(0); |
2016 | | |
2017 | 10.6M | maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
2018 | 10.6M | if (ctxt->nodeNr > maxDepth) { |
2019 | 73 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
2020 | 73 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
2021 | 73 | ctxt->nodeNr); |
2022 | 73 | xmlHaltParser(ctxt); |
2023 | 73 | return(-1); |
2024 | 73 | } |
2025 | 10.6M | if (ctxt->nodeNr >= ctxt->nodeMax) { |
2026 | 362k | xmlNodePtr *tmp; |
2027 | | |
2028 | 362k | tmp = (xmlNodePtr *) xmlRealloc(ctxt->nodeTab, |
2029 | 362k | ctxt->nodeMax * 2 * |
2030 | 362k | sizeof(ctxt->nodeTab[0])); |
2031 | 362k | if (tmp == NULL) { |
2032 | 160 | xmlErrMemory(ctxt); |
2033 | 160 | return (-1); |
2034 | 160 | } |
2035 | 362k | ctxt->nodeTab = tmp; |
2036 | 362k | ctxt->nodeMax *= 2; |
2037 | 362k | } |
2038 | 10.6M | ctxt->nodeTab[ctxt->nodeNr] = value; |
2039 | 10.6M | ctxt->node = value; |
2040 | 10.6M | return (ctxt->nodeNr++); |
2041 | 10.6M | } |
2042 | | |
2043 | | /** |
2044 | | * nodePop: |
2045 | | * @ctxt: an XML parser context |
2046 | | * |
2047 | | * DEPRECATED: Internal function, do not use. |
2048 | | * |
2049 | | * Pops the top element node from the node stack |
2050 | | * |
2051 | | * Returns the node just removed |
2052 | | */ |
2053 | | xmlNodePtr |
2054 | | nodePop(xmlParserCtxtPtr ctxt) |
2055 | 10.4M | { |
2056 | 10.4M | xmlNodePtr ret; |
2057 | | |
2058 | 10.4M | if (ctxt == NULL) return(NULL); |
2059 | 10.4M | if (ctxt->nodeNr <= 0) |
2060 | 277k | return (NULL); |
2061 | 10.1M | ctxt->nodeNr--; |
2062 | 10.1M | if (ctxt->nodeNr > 0) |
2063 | 10.0M | ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1]; |
2064 | 149k | else |
2065 | 149k | ctxt->node = NULL; |
2066 | 10.1M | ret = ctxt->nodeTab[ctxt->nodeNr]; |
2067 | 10.1M | ctxt->nodeTab[ctxt->nodeNr] = NULL; |
2068 | 10.1M | return (ret); |
2069 | 10.4M | } |
2070 | | |
2071 | | /** |
2072 | | * nameNsPush: |
2073 | | * @ctxt: an XML parser context |
2074 | | * @value: the element name |
2075 | | * @prefix: the element prefix |
2076 | | * @URI: the element namespace name |
2077 | | * @line: the current line number for error messages |
2078 | | * @nsNr: the number of namespaces pushed on the namespace table |
2079 | | * |
2080 | | * Pushes a new element name/prefix/URL on top of the name stack |
2081 | | * |
2082 | | * Returns -1 in case of error, the index in the stack otherwise |
2083 | | */ |
2084 | | static int |
2085 | | nameNsPush(xmlParserCtxtPtr ctxt, const xmlChar * value, |
2086 | | const xmlChar *prefix, const xmlChar *URI, int line, int nsNr) |
2087 | 9.80M | { |
2088 | 9.80M | xmlStartTag *tag; |
2089 | | |
2090 | 9.80M | if (ctxt->nameNr >= ctxt->nameMax) { |
2091 | 382k | const xmlChar * *tmp; |
2092 | 382k | xmlStartTag *tmp2; |
2093 | 382k | ctxt->nameMax *= 2; |
2094 | 382k | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
2095 | 382k | ctxt->nameMax * |
2096 | 382k | sizeof(ctxt->nameTab[0])); |
2097 | 382k | if (tmp == NULL) { |
2098 | 141 | ctxt->nameMax /= 2; |
2099 | 141 | goto mem_error; |
2100 | 141 | } |
2101 | 381k | ctxt->nameTab = tmp; |
2102 | 381k | tmp2 = (xmlStartTag *) xmlRealloc((void * *)ctxt->pushTab, |
2103 | 381k | ctxt->nameMax * |
2104 | 381k | sizeof(ctxt->pushTab[0])); |
2105 | 381k | if (tmp2 == NULL) { |
2106 | 195 | ctxt->nameMax /= 2; |
2107 | 195 | goto mem_error; |
2108 | 195 | } |
2109 | 381k | ctxt->pushTab = tmp2; |
2110 | 9.42M | } else if (ctxt->pushTab == NULL) { |
2111 | 230k | ctxt->pushTab = (xmlStartTag *) xmlMalloc(ctxt->nameMax * |
2112 | 230k | sizeof(ctxt->pushTab[0])); |
2113 | 230k | if (ctxt->pushTab == NULL) |
2114 | 233 | goto mem_error; |
2115 | 230k | } |
2116 | 9.80M | ctxt->nameTab[ctxt->nameNr] = value; |
2117 | 9.80M | ctxt->name = value; |
2118 | 9.80M | tag = &ctxt->pushTab[ctxt->nameNr]; |
2119 | 9.80M | tag->prefix = prefix; |
2120 | 9.80M | tag->URI = URI; |
2121 | 9.80M | tag->line = line; |
2122 | 9.80M | tag->nsNr = nsNr; |
2123 | 9.80M | return (ctxt->nameNr++); |
2124 | 569 | mem_error: |
2125 | 569 | xmlErrMemory(ctxt); |
2126 | 569 | return (-1); |
2127 | 9.80M | } |
2128 | | #ifdef LIBXML_PUSH_ENABLED |
2129 | | /** |
2130 | | * nameNsPop: |
2131 | | * @ctxt: an XML parser context |
2132 | | * |
2133 | | * Pops the top element/prefix/URI name from the name stack |
2134 | | * |
2135 | | * Returns the name just removed |
2136 | | */ |
2137 | | static const xmlChar * |
2138 | | nameNsPop(xmlParserCtxtPtr ctxt) |
2139 | 194k | { |
2140 | 194k | const xmlChar *ret; |
2141 | | |
2142 | 194k | if (ctxt->nameNr <= 0) |
2143 | 0 | return (NULL); |
2144 | 194k | ctxt->nameNr--; |
2145 | 194k | if (ctxt->nameNr > 0) |
2146 | 193k | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
2147 | 1.08k | else |
2148 | 1.08k | ctxt->name = NULL; |
2149 | 194k | ret = ctxt->nameTab[ctxt->nameNr]; |
2150 | 194k | ctxt->nameTab[ctxt->nameNr] = NULL; |
2151 | 194k | return (ret); |
2152 | 194k | } |
2153 | | #endif /* LIBXML_PUSH_ENABLED */ |
2154 | | |
2155 | | /** |
2156 | | * namePush: |
2157 | | * @ctxt: an XML parser context |
2158 | | * @value: the element name |
2159 | | * |
2160 | | * DEPRECATED: Internal function, do not use. |
2161 | | * |
2162 | | * Pushes a new element name on top of the name stack |
2163 | | * |
2164 | | * Returns -1 in case of error, the index in the stack otherwise |
2165 | | */ |
2166 | | int |
2167 | | namePush(xmlParserCtxtPtr ctxt, const xmlChar * value) |
2168 | 0 | { |
2169 | 0 | if (ctxt == NULL) return (-1); |
2170 | | |
2171 | 0 | if (ctxt->nameNr >= ctxt->nameMax) { |
2172 | 0 | const xmlChar * *tmp; |
2173 | 0 | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
2174 | 0 | ctxt->nameMax * 2 * |
2175 | 0 | sizeof(ctxt->nameTab[0])); |
2176 | 0 | if (tmp == NULL) { |
2177 | 0 | goto mem_error; |
2178 | 0 | } |
2179 | 0 | ctxt->nameTab = tmp; |
2180 | 0 | ctxt->nameMax *= 2; |
2181 | 0 | } |
2182 | 0 | ctxt->nameTab[ctxt->nameNr] = value; |
2183 | 0 | ctxt->name = value; |
2184 | 0 | return (ctxt->nameNr++); |
2185 | 0 | mem_error: |
2186 | 0 | xmlErrMemory(ctxt); |
2187 | 0 | return (-1); |
2188 | 0 | } |
2189 | | |
2190 | | /** |
2191 | | * namePop: |
2192 | | * @ctxt: an XML parser context |
2193 | | * |
2194 | | * DEPRECATED: Internal function, do not use. |
2195 | | * |
2196 | | * Pops the top element name from the name stack |
2197 | | * |
2198 | | * Returns the name just removed |
2199 | | */ |
2200 | | const xmlChar * |
2201 | | namePop(xmlParserCtxtPtr ctxt) |
2202 | 9.22M | { |
2203 | 9.22M | const xmlChar *ret; |
2204 | | |
2205 | 9.22M | if ((ctxt == NULL) || (ctxt->nameNr <= 0)) |
2206 | 78 | return (NULL); |
2207 | 9.22M | ctxt->nameNr--; |
2208 | 9.22M | if (ctxt->nameNr > 0) |
2209 | 9.09M | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
2210 | 121k | else |
2211 | 121k | ctxt->name = NULL; |
2212 | 9.22M | ret = ctxt->nameTab[ctxt->nameNr]; |
2213 | 9.22M | ctxt->nameTab[ctxt->nameNr] = NULL; |
2214 | 9.22M | return (ret); |
2215 | 9.22M | } |
2216 | | |
2217 | 12.3M | static int spacePush(xmlParserCtxtPtr ctxt, int val) { |
2218 | 12.3M | if (ctxt->spaceNr >= ctxt->spaceMax) { |
2219 | 524k | int *tmp; |
2220 | | |
2221 | 524k | ctxt->spaceMax *= 2; |
2222 | 524k | tmp = (int *) xmlRealloc(ctxt->spaceTab, |
2223 | 524k | ctxt->spaceMax * sizeof(ctxt->spaceTab[0])); |
2224 | 524k | if (tmp == NULL) { |
2225 | 267 | xmlErrMemory(ctxt); |
2226 | 267 | ctxt->spaceMax /=2; |
2227 | 267 | return(-1); |
2228 | 267 | } |
2229 | 524k | ctxt->spaceTab = tmp; |
2230 | 524k | } |
2231 | 12.3M | ctxt->spaceTab[ctxt->spaceNr] = val; |
2232 | 12.3M | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr]; |
2233 | 12.3M | return(ctxt->spaceNr++); |
2234 | 12.3M | } |
2235 | | |
2236 | 11.9M | static int spacePop(xmlParserCtxtPtr ctxt) { |
2237 | 11.9M | int ret; |
2238 | 11.9M | if (ctxt->spaceNr <= 0) return(0); |
2239 | 11.9M | ctxt->spaceNr--; |
2240 | 11.9M | if (ctxt->spaceNr > 0) |
2241 | 11.9M | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1]; |
2242 | 28.4k | else |
2243 | 28.4k | ctxt->space = &ctxt->spaceTab[0]; |
2244 | 11.9M | ret = ctxt->spaceTab[ctxt->spaceNr]; |
2245 | 11.9M | ctxt->spaceTab[ctxt->spaceNr] = -1; |
2246 | 11.9M | return(ret); |
2247 | 11.9M | } |
2248 | | |
2249 | | /* |
2250 | | * Macros for accessing the content. Those should be used only by the parser, |
2251 | | * and not exported. |
2252 | | * |
2253 | | * Dirty macros, i.e. one often need to make assumption on the context to |
2254 | | * use them |
2255 | | * |
2256 | | * CUR_PTR return the current pointer to the xmlChar to be parsed. |
2257 | | * To be used with extreme caution since operations consuming |
2258 | | * characters may move the input buffer to a different location ! |
2259 | | * CUR returns the current xmlChar value, i.e. a 8 bit value if compiled |
2260 | | * This should be used internally by the parser |
2261 | | * only to compare to ASCII values otherwise it would break when |
2262 | | * running with UTF-8 encoding. |
2263 | | * RAW same as CUR but in the input buffer, bypass any token |
2264 | | * extraction that may have been done |
2265 | | * NXT(n) returns the n'th next xmlChar. Same as CUR is should be used only |
2266 | | * to compare on ASCII based substring. |
2267 | | * SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined |
2268 | | * strings without newlines within the parser. |
2269 | | * NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII |
2270 | | * defined char within the parser. |
2271 | | * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding |
2272 | | * |
2273 | | * NEXT Skip to the next character, this does the proper decoding |
2274 | | * in UTF-8 mode. It also pop-up unfinished entities on the fly. |
2275 | | * NEXTL(l) Skip the current unicode character of l xmlChars long. |
2276 | | * CUR_SCHAR same but operate on a string instead of the context |
2277 | | * COPY_BUF copy the current unicode char to the target buffer, increment |
2278 | | * the index |
2279 | | * GROW, SHRINK handling of input buffers |
2280 | | */ |
2281 | | |
2282 | 157M | #define RAW (*ctxt->input->cur) |
2283 | 1.83G | #define CUR (*ctxt->input->cur) |
2284 | 81.6M | #define NXT(val) ctxt->input->cur[(val)] |
2285 | 3.76G | #define CUR_PTR ctxt->input->cur |
2286 | 26.3M | #define BASE_PTR ctxt->input->base |
2287 | | |
2288 | | #define CMP4( s, c1, c2, c3, c4 ) \ |
2289 | 157M | ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \ |
2290 | 79.0M | ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 ) |
2291 | | #define CMP5( s, c1, c2, c3, c4, c5 ) \ |
2292 | 154M | ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 ) |
2293 | | #define CMP6( s, c1, c2, c3, c4, c5, c6 ) \ |
2294 | 148M | ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 ) |
2295 | | #define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \ |
2296 | 144M | ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 ) |
2297 | | #define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \ |
2298 | 141M | ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 ) |
2299 | | #define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \ |
2300 | 70.0M | ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \ |
2301 | 70.0M | ((unsigned char *) s)[ 8 ] == c9 ) |
2302 | | #define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \ |
2303 | 156k | ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \ |
2304 | 156k | ((unsigned char *) s)[ 9 ] == c10 ) |
2305 | | |
2306 | 22.6M | #define SKIP(val) do { \ |
2307 | 22.6M | ctxt->input->cur += (val),ctxt->input->col+=(val); \ |
2308 | 22.6M | if (*ctxt->input->cur == 0) \ |
2309 | 22.6M | xmlParserGrow(ctxt); \ |
2310 | 22.6M | } while (0) |
2311 | | |
2312 | 225k | #define SKIPL(val) do { \ |
2313 | 225k | int skipl; \ |
2314 | 37.6M | for(skipl=0; skipl<val; skipl++) { \ |
2315 | 37.4M | if (*(ctxt->input->cur) == '\n') { \ |
2316 | 64.9k | ctxt->input->line++; ctxt->input->col = 1; \ |
2317 | 37.3M | } else ctxt->input->col++; \ |
2318 | 37.4M | ctxt->input->cur++; \ |
2319 | 37.4M | } \ |
2320 | 225k | if (*ctxt->input->cur == 0) \ |
2321 | 225k | xmlParserGrow(ctxt); \ |
2322 | 225k | } while (0) |
2323 | | |
2324 | | #define SHRINK \ |
2325 | 129M | if ((!PARSER_PROGRESSIVE(ctxt)) && \ |
2326 | 129M | (ctxt->input->cur - ctxt->input->base > 2 * INPUT_CHUNK) && \ |
2327 | 129M | (ctxt->input->end - ctxt->input->cur < 2 * INPUT_CHUNK)) \ |
2328 | 129M | xmlParserShrink(ctxt); |
2329 | | |
2330 | | #define GROW \ |
2331 | 277M | if ((!PARSER_PROGRESSIVE(ctxt)) && \ |
2332 | 277M | (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \ |
2333 | 28.0M | xmlParserGrow(ctxt); |
2334 | | |
2335 | 31.9M | #define SKIP_BLANKS xmlSkipBlankChars(ctxt) |
2336 | | |
2337 | 12.1M | #define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt) |
2338 | | |
2339 | 97.1M | #define NEXT xmlNextChar(ctxt) |
2340 | | |
2341 | 18.9M | #define NEXT1 { \ |
2342 | 18.9M | ctxt->input->col++; \ |
2343 | 18.9M | ctxt->input->cur++; \ |
2344 | 18.9M | if (*ctxt->input->cur == 0) \ |
2345 | 18.9M | xmlParserGrow(ctxt); \ |
2346 | 18.9M | } |
2347 | | |
2348 | 2.26G | #define NEXTL(l) do { \ |
2349 | 2.26G | if (*(ctxt->input->cur) == '\n') { \ |
2350 | 68.0M | ctxt->input->line++; ctxt->input->col = 1; \ |
2351 | 2.19G | } else ctxt->input->col++; \ |
2352 | 2.26G | ctxt->input->cur += l; \ |
2353 | 2.26G | } while (0) |
2354 | | |
2355 | 32.8M | #define CUR_SCHAR(s, l) xmlStringCurrentChar(ctxt, s, &l) |
2356 | | |
2357 | | #define COPY_BUF(b, i, v) \ |
2358 | 449M | if (v < 0x80) b[i++] = v; \ |
2359 | 449M | else i += xmlCopyCharMultiByte(&b[i],v) |
2360 | | |
2361 | | static int |
2362 | 455M | xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) { |
2363 | 455M | int c = xmlCurrentChar(ctxt, len); |
2364 | | |
2365 | 455M | if (c == XML_INVALID_CHAR) |
2366 | 110M | c = 0xFFFD; /* replacement character */ |
2367 | | |
2368 | 455M | return(c); |
2369 | 455M | } |
2370 | | |
2371 | | /** |
2372 | | * xmlSkipBlankChars: |
2373 | | * @ctxt: the XML parser context |
2374 | | * |
2375 | | * DEPRECATED: Internal function, do not use. |
2376 | | * |
2377 | | * Skip whitespace in the input stream. |
2378 | | * |
2379 | | * Returns the number of space chars skipped |
2380 | | */ |
2381 | | int |
2382 | 35.6M | xmlSkipBlankChars(xmlParserCtxtPtr ctxt) { |
2383 | 35.6M | const xmlChar *cur; |
2384 | 35.6M | int res = 0; |
2385 | | |
2386 | | /* |
2387 | | * It's Okay to use CUR/NEXT here since all the blanks are on |
2388 | | * the ASCII range. |
2389 | | */ |
2390 | 35.6M | cur = ctxt->input->cur; |
2391 | 35.6M | while (IS_BLANK_CH(*cur)) { |
2392 | 24.9M | if (*cur == '\n') { |
2393 | 11.4M | ctxt->input->line++; ctxt->input->col = 1; |
2394 | 13.4M | } else { |
2395 | 13.4M | ctxt->input->col++; |
2396 | 13.4M | } |
2397 | 24.9M | cur++; |
2398 | 24.9M | if (res < INT_MAX) |
2399 | 24.9M | res++; |
2400 | 24.9M | if (*cur == 0) { |
2401 | 148k | ctxt->input->cur = cur; |
2402 | 148k | xmlParserGrow(ctxt); |
2403 | 148k | cur = ctxt->input->cur; |
2404 | 148k | } |
2405 | 24.9M | } |
2406 | 35.6M | ctxt->input->cur = cur; |
2407 | | |
2408 | 35.6M | return(res); |
2409 | 35.6M | } |
2410 | | |
2411 | | static void |
2412 | 669k | xmlPopPE(xmlParserCtxtPtr ctxt) { |
2413 | 669k | unsigned long consumed; |
2414 | 669k | xmlEntityPtr ent; |
2415 | | |
2416 | 669k | ent = ctxt->input->entity; |
2417 | | |
2418 | 669k | ent->flags &= ~XML_ENT_EXPANDING; |
2419 | | |
2420 | 669k | if ((ent->flags & XML_ENT_CHECKED) == 0) { |
2421 | 33.8k | int result; |
2422 | | |
2423 | | /* |
2424 | | * Read the rest of the stream in case of errors. We want |
2425 | | * to account for the whole entity size. |
2426 | | */ |
2427 | 45.0k | do { |
2428 | 45.0k | ctxt->input->cur = ctxt->input->end; |
2429 | 45.0k | xmlParserShrink(ctxt); |
2430 | 45.0k | result = xmlParserGrow(ctxt); |
2431 | 45.0k | } while (result > 0); |
2432 | | |
2433 | 33.8k | consumed = ctxt->input->consumed; |
2434 | 33.8k | xmlSaturatedAddSizeT(&consumed, |
2435 | 33.8k | ctxt->input->end - ctxt->input->base); |
2436 | | |
2437 | 33.8k | xmlSaturatedAdd(&ent->expandedSize, consumed); |
2438 | | |
2439 | | /* |
2440 | | * Add to sizeentities when parsing an external entity |
2441 | | * for the first time. |
2442 | | */ |
2443 | 33.8k | if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
2444 | 19.8k | xmlSaturatedAdd(&ctxt->sizeentities, consumed); |
2445 | 19.8k | } |
2446 | | |
2447 | 33.8k | ent->flags |= XML_ENT_CHECKED; |
2448 | 33.8k | } |
2449 | | |
2450 | 669k | xmlPopInput(ctxt); |
2451 | | |
2452 | 669k | xmlParserEntityCheck(ctxt, ent->expandedSize); |
2453 | 669k | } |
2454 | | |
2455 | | /** |
2456 | | * xmlSkipBlankCharsPE: |
2457 | | * @ctxt: the XML parser context |
2458 | | * |
2459 | | * Skip whitespace in the input stream, also handling parameter |
2460 | | * entities. |
2461 | | * |
2462 | | * Returns the number of space chars skipped |
2463 | | */ |
2464 | | static int |
2465 | 12.1M | xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) { |
2466 | 12.1M | int res = 0; |
2467 | 12.1M | int inParam; |
2468 | 12.1M | int expandParam; |
2469 | | |
2470 | 12.1M | inParam = PARSER_IN_PE(ctxt); |
2471 | 12.1M | expandParam = PARSER_EXTERNAL(ctxt); |
2472 | | |
2473 | 12.1M | if (!inParam && !expandParam) |
2474 | 3.69M | return(xmlSkipBlankChars(ctxt)); |
2475 | | |
2476 | 21.7M | while (PARSER_STOPPED(ctxt) == 0) { |
2477 | 21.7M | if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */ |
2478 | 10.5M | NEXT; |
2479 | 11.1M | } else if (CUR == '%') { |
2480 | 2.55M | if ((expandParam == 0) || |
2481 | 2.55M | (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0)) |
2482 | 443k | break; |
2483 | | |
2484 | | /* |
2485 | | * Expand parameter entity. We continue to consume |
2486 | | * whitespace at the start of the entity and possible |
2487 | | * even consume the whole entity and pop it. We might |
2488 | | * even pop multiple PEs in this loop. |
2489 | | */ |
2490 | 2.10M | xmlParsePEReference(ctxt); |
2491 | | |
2492 | 2.10M | inParam = PARSER_IN_PE(ctxt); |
2493 | 2.10M | expandParam = PARSER_EXTERNAL(ctxt); |
2494 | 8.64M | } else if (CUR == 0) { |
2495 | 662k | if (inParam == 0) |
2496 | 4.34k | break; |
2497 | | |
2498 | 658k | xmlPopPE(ctxt); |
2499 | | |
2500 | 658k | inParam = PARSER_IN_PE(ctxt); |
2501 | 658k | expandParam = PARSER_EXTERNAL(ctxt); |
2502 | 7.98M | } else { |
2503 | 7.98M | break; |
2504 | 7.98M | } |
2505 | | |
2506 | | /* |
2507 | | * Also increase the counter when entering or exiting a PERef. |
2508 | | * The spec says: "When a parameter-entity reference is recognized |
2509 | | * in the DTD and included, its replacement text MUST be enlarged |
2510 | | * by the attachment of one leading and one following space (#x20) |
2511 | | * character." |
2512 | | */ |
2513 | 13.2M | if (res < INT_MAX) |
2514 | 13.2M | res++; |
2515 | 13.2M | } |
2516 | | |
2517 | 8.44M | return(res); |
2518 | 12.1M | } |
2519 | | |
2520 | | /************************************************************************ |
2521 | | * * |
2522 | | * Commodity functions to handle entities * |
2523 | | * * |
2524 | | ************************************************************************/ |
2525 | | |
2526 | | /** |
2527 | | * xmlPopInput: |
2528 | | * @ctxt: an XML parser context |
2529 | | * |
2530 | | * xmlPopInput: the current input pointed by ctxt->input came to an end |
2531 | | * pop it and return the next char. |
2532 | | * |
2533 | | * Returns the current xmlChar in the parser context |
2534 | | */ |
2535 | | xmlChar |
2536 | 669k | xmlPopInput(xmlParserCtxtPtr ctxt) { |
2537 | 669k | xmlParserInputPtr input; |
2538 | | |
2539 | 669k | if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0); |
2540 | 669k | input = inputPop(ctxt); |
2541 | 669k | xmlFreeInputStream(input); |
2542 | 669k | if (*ctxt->input->cur == 0) |
2543 | 78.1k | xmlParserGrow(ctxt); |
2544 | 669k | return(CUR); |
2545 | 669k | } |
2546 | | |
2547 | | /** |
2548 | | * xmlPushInput: |
2549 | | * @ctxt: an XML parser context |
2550 | | * @input: an XML parser input fragment (entity, XML fragment ...). |
2551 | | * |
2552 | | * Push an input stream onto the stack. |
2553 | | * |
2554 | | * Returns -1 in case of error or the index in the input stack |
2555 | | */ |
2556 | | int |
2557 | 1.72M | xmlPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) { |
2558 | 1.72M | int maxDepth; |
2559 | 1.72M | int ret; |
2560 | | |
2561 | 1.72M | if ((ctxt == NULL) || (input == NULL)) |
2562 | 985k | return(-1); |
2563 | | |
2564 | 740k | maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
2565 | 740k | if (ctxt->inputNr > maxDepth) { |
2566 | 18 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
2567 | 18 | "Maximum entity nesting depth exceeded"); |
2568 | 18 | xmlHaltParser(ctxt); |
2569 | 18 | return(-1); |
2570 | 18 | } |
2571 | 740k | ret = inputPush(ctxt, input); |
2572 | 740k | if (ret >= 0) |
2573 | 740k | GROW; |
2574 | 740k | return(ret); |
2575 | 740k | } |
2576 | | |
2577 | | /** |
2578 | | * xmlParseCharRef: |
2579 | | * @ctxt: an XML parser context |
2580 | | * |
2581 | | * DEPRECATED: Internal function, don't use. |
2582 | | * |
2583 | | * Parse a numeric character reference. Always consumes '&'. |
2584 | | * |
2585 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2586 | | * '&#x' [0-9a-fA-F]+ ';' |
2587 | | * |
2588 | | * [ WFC: Legal Character ] |
2589 | | * Characters referred to using character references must match the |
2590 | | * production for Char. |
2591 | | * |
2592 | | * Returns the value parsed (as an int), 0 in case of error |
2593 | | */ |
2594 | | int |
2595 | 1.46M | xmlParseCharRef(xmlParserCtxtPtr ctxt) { |
2596 | 1.46M | int val = 0; |
2597 | 1.46M | int count = 0; |
2598 | | |
2599 | | /* |
2600 | | * Using RAW/CUR/NEXT is okay since we are working on ASCII range here |
2601 | | */ |
2602 | 1.46M | if ((RAW == '&') && (NXT(1) == '#') && |
2603 | 1.46M | (NXT(2) == 'x')) { |
2604 | 735k | SKIP(3); |
2605 | 735k | GROW; |
2606 | 2.58M | while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) { |
2607 | 1.99M | if (count++ > 20) { |
2608 | 21.7k | count = 0; |
2609 | 21.7k | GROW; |
2610 | 21.7k | } |
2611 | 1.99M | if ((RAW >= '0') && (RAW <= '9')) |
2612 | 836k | val = val * 16 + (CUR - '0'); |
2613 | 1.16M | else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20)) |
2614 | 361k | val = val * 16 + (CUR - 'a') + 10; |
2615 | 800k | else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20)) |
2616 | 653k | val = val * 16 + (CUR - 'A') + 10; |
2617 | 146k | else { |
2618 | 146k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2619 | 146k | val = 0; |
2620 | 146k | break; |
2621 | 146k | } |
2622 | 1.85M | if (val > 0x110000) |
2623 | 304k | val = 0x110000; |
2624 | | |
2625 | 1.85M | NEXT; |
2626 | 1.85M | count++; |
2627 | 1.85M | } |
2628 | 735k | if (RAW == ';') { |
2629 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2630 | 588k | ctxt->input->col++; |
2631 | 588k | ctxt->input->cur++; |
2632 | 588k | } |
2633 | 735k | } else if ((RAW == '&') && (NXT(1) == '#')) { |
2634 | 731k | SKIP(2); |
2635 | 731k | GROW; |
2636 | 2.40M | while (RAW != ';') { /* loop blocked by count */ |
2637 | 1.79M | if (count++ > 20) { |
2638 | 23.4k | count = 0; |
2639 | 23.4k | GROW; |
2640 | 23.4k | } |
2641 | 1.79M | if ((RAW >= '0') && (RAW <= '9')) |
2642 | 1.67M | val = val * 10 + (CUR - '0'); |
2643 | 126k | else { |
2644 | 126k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2645 | 126k | val = 0; |
2646 | 126k | break; |
2647 | 126k | } |
2648 | 1.67M | if (val > 0x110000) |
2649 | 234k | val = 0x110000; |
2650 | | |
2651 | 1.67M | NEXT; |
2652 | 1.67M | count++; |
2653 | 1.67M | } |
2654 | 731k | if (RAW == ';') { |
2655 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2656 | 604k | ctxt->input->col++; |
2657 | 604k | ctxt->input->cur++; |
2658 | 604k | } |
2659 | 731k | } else { |
2660 | 0 | if (RAW == '&') |
2661 | 0 | SKIP(1); |
2662 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2663 | 0 | } |
2664 | | |
2665 | | /* |
2666 | | * [ WFC: Legal Character ] |
2667 | | * Characters referred to using character references must match the |
2668 | | * production for Char. |
2669 | | */ |
2670 | 1.46M | if (val >= 0x110000) { |
2671 | 23.7k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2672 | 23.7k | "xmlParseCharRef: character reference out of bounds\n", |
2673 | 23.7k | val); |
2674 | 1.44M | } else if (IS_CHAR(val)) { |
2675 | 1.13M | return(val); |
2676 | 1.13M | } else { |
2677 | 312k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2678 | 312k | "xmlParseCharRef: invalid xmlChar value %d\n", |
2679 | 312k | val); |
2680 | 312k | } |
2681 | 336k | return(0); |
2682 | 1.46M | } |
2683 | | |
2684 | | /** |
2685 | | * xmlParseStringCharRef: |
2686 | | * @ctxt: an XML parser context |
2687 | | * @str: a pointer to an index in the string |
2688 | | * |
2689 | | * parse Reference declarations, variant parsing from a string rather |
2690 | | * than an an input flow. |
2691 | | * |
2692 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2693 | | * '&#x' [0-9a-fA-F]+ ';' |
2694 | | * |
2695 | | * [ WFC: Legal Character ] |
2696 | | * Characters referred to using character references must match the |
2697 | | * production for Char. |
2698 | | * |
2699 | | * Returns the value parsed (as an int), 0 in case of error, str will be |
2700 | | * updated to the current value of the index |
2701 | | */ |
2702 | | static int |
2703 | 3.47M | xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
2704 | 3.47M | const xmlChar *ptr; |
2705 | 3.47M | xmlChar cur; |
2706 | 3.47M | int val = 0; |
2707 | | |
2708 | 3.47M | if ((str == NULL) || (*str == NULL)) return(0); |
2709 | 3.47M | ptr = *str; |
2710 | 3.47M | cur = *ptr; |
2711 | 3.47M | if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) { |
2712 | 146k | ptr += 3; |
2713 | 146k | cur = *ptr; |
2714 | 532k | while (cur != ';') { /* Non input consuming loop */ |
2715 | 393k | if ((cur >= '0') && (cur <= '9')) |
2716 | 157k | val = val * 16 + (cur - '0'); |
2717 | 235k | else if ((cur >= 'a') && (cur <= 'f')) |
2718 | 114k | val = val * 16 + (cur - 'a') + 10; |
2719 | 120k | else if ((cur >= 'A') && (cur <= 'F')) |
2720 | 113k | val = val * 16 + (cur - 'A') + 10; |
2721 | 7.16k | else { |
2722 | 7.16k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2723 | 7.16k | val = 0; |
2724 | 7.16k | break; |
2725 | 7.16k | } |
2726 | 386k | if (val > 0x110000) |
2727 | 37.6k | val = 0x110000; |
2728 | | |
2729 | 386k | ptr++; |
2730 | 386k | cur = *ptr; |
2731 | 386k | } |
2732 | 146k | if (cur == ';') |
2733 | 139k | ptr++; |
2734 | 3.32M | } else if ((cur == '&') && (ptr[1] == '#')){ |
2735 | 3.32M | ptr += 2; |
2736 | 3.32M | cur = *ptr; |
2737 | 10.0M | while (cur != ';') { /* Non input consuming loops */ |
2738 | 6.77M | if ((cur >= '0') && (cur <= '9')) |
2739 | 6.76M | val = val * 10 + (cur - '0'); |
2740 | 10.5k | else { |
2741 | 10.5k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2742 | 10.5k | val = 0; |
2743 | 10.5k | break; |
2744 | 10.5k | } |
2745 | 6.76M | if (val > 0x110000) |
2746 | 20.0k | val = 0x110000; |
2747 | | |
2748 | 6.76M | ptr++; |
2749 | 6.76M | cur = *ptr; |
2750 | 6.76M | } |
2751 | 3.32M | if (cur == ';') |
2752 | 3.31M | ptr++; |
2753 | 3.32M | } else { |
2754 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2755 | 0 | return(0); |
2756 | 0 | } |
2757 | 3.47M | *str = ptr; |
2758 | | |
2759 | | /* |
2760 | | * [ WFC: Legal Character ] |
2761 | | * Characters referred to using character references must match the |
2762 | | * production for Char. |
2763 | | */ |
2764 | 3.47M | if (val >= 0x110000) { |
2765 | 2.07k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2766 | 2.07k | "xmlParseStringCharRef: character reference out of bounds\n", |
2767 | 2.07k | val); |
2768 | 3.46M | } else if (IS_CHAR(val)) { |
2769 | 3.44M | return(val); |
2770 | 3.44M | } else { |
2771 | 26.8k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2772 | 26.8k | "xmlParseStringCharRef: invalid xmlChar value %d\n", |
2773 | 26.8k | val); |
2774 | 26.8k | } |
2775 | 28.9k | return(0); |
2776 | 3.47M | } |
2777 | | |
2778 | | /** |
2779 | | * xmlParserHandlePEReference: |
2780 | | * @ctxt: the parser context |
2781 | | * |
2782 | | * DEPRECATED: Internal function, do not use. |
2783 | | * |
2784 | | * [69] PEReference ::= '%' Name ';' |
2785 | | * |
2786 | | * [ WFC: No Recursion ] |
2787 | | * A parsed entity must not contain a recursive |
2788 | | * reference to itself, either directly or indirectly. |
2789 | | * |
2790 | | * [ WFC: Entity Declared ] |
2791 | | * In a document without any DTD, a document with only an internal DTD |
2792 | | * subset which contains no parameter entity references, or a document |
2793 | | * with "standalone='yes'", ... ... The declaration of a parameter |
2794 | | * entity must precede any reference to it... |
2795 | | * |
2796 | | * [ VC: Entity Declared ] |
2797 | | * In a document with an external subset or external parameter entities |
2798 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
2799 | | * must precede any reference to it... |
2800 | | * |
2801 | | * [ WFC: In DTD ] |
2802 | | * Parameter-entity references may only appear in the DTD. |
2803 | | * NOTE: misleading but this is handled. |
2804 | | * |
2805 | | * A PEReference may have been detected in the current input stream |
2806 | | * the handling is done accordingly to |
2807 | | * http://www.w3.org/TR/REC-xml#entproc |
2808 | | * i.e. |
2809 | | * - Included in literal in entity values |
2810 | | * - Included as Parameter Entity reference within DTDs |
2811 | | */ |
2812 | | void |
2813 | 0 | xmlParserHandlePEReference(xmlParserCtxtPtr ctxt) { |
2814 | 0 | xmlParsePEReference(ctxt); |
2815 | 0 | } |
2816 | | |
2817 | | /** |
2818 | | * xmlStringLenDecodeEntities: |
2819 | | * @ctxt: the parser context |
2820 | | * @str: the input string |
2821 | | * @len: the string length |
2822 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2823 | | * @end: an end marker xmlChar, 0 if none |
2824 | | * @end2: an end marker xmlChar, 0 if none |
2825 | | * @end3: an end marker xmlChar, 0 if none |
2826 | | * |
2827 | | * DEPRECATED: Internal function, don't use. |
2828 | | * |
2829 | | * Returns A newly allocated string with the substitution done. The caller |
2830 | | * must deallocate it ! |
2831 | | */ |
2832 | | xmlChar * |
2833 | | xmlStringLenDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2834 | | int what ATTRIBUTE_UNUSED, |
2835 | 0 | xmlChar end, xmlChar end2, xmlChar end3) { |
2836 | 0 | if ((ctxt == NULL) || (str == NULL) || (len < 0)) |
2837 | 0 | return(NULL); |
2838 | | |
2839 | 0 | if ((str[len] != 0) || |
2840 | 0 | (end != 0) || (end2 != 0) || (end3 != 0)) |
2841 | 0 | return(NULL); |
2842 | | |
2843 | 0 | return(xmlExpandEntitiesInAttValue(ctxt, str, 0)); |
2844 | 0 | } |
2845 | | |
2846 | | /** |
2847 | | * xmlStringDecodeEntities: |
2848 | | * @ctxt: the parser context |
2849 | | * @str: the input string |
2850 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2851 | | * @end: an end marker xmlChar, 0 if none |
2852 | | * @end2: an end marker xmlChar, 0 if none |
2853 | | * @end3: an end marker xmlChar, 0 if none |
2854 | | * |
2855 | | * DEPRECATED: Internal function, don't use. |
2856 | | * |
2857 | | * Returns A newly allocated string with the substitution done. The caller |
2858 | | * must deallocate it ! |
2859 | | */ |
2860 | | xmlChar * |
2861 | | xmlStringDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, |
2862 | | int what ATTRIBUTE_UNUSED, |
2863 | 0 | xmlChar end, xmlChar end2, xmlChar end3) { |
2864 | 0 | if ((ctxt == NULL) || (str == NULL)) |
2865 | 0 | return(NULL); |
2866 | | |
2867 | 0 | if ((end != 0) || (end2 != 0) || (end3 != 0)) |
2868 | 0 | return(NULL); |
2869 | | |
2870 | 0 | return(xmlExpandEntitiesInAttValue(ctxt, str, 0)); |
2871 | 0 | } |
2872 | | |
2873 | | /************************************************************************ |
2874 | | * * |
2875 | | * Commodity functions, cleanup needed ? * |
2876 | | * * |
2877 | | ************************************************************************/ |
2878 | | |
2879 | | /** |
2880 | | * areBlanks: |
2881 | | * @ctxt: an XML parser context |
2882 | | * @str: a xmlChar * |
2883 | | * @len: the size of @str |
2884 | | * @blank_chars: we know the chars are blanks |
2885 | | * |
2886 | | * Is this a sequence of blank chars that one can ignore ? |
2887 | | * |
2888 | | * Returns 1 if ignorable 0 otherwise. |
2889 | | */ |
2890 | | |
2891 | | static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2892 | 6.65M | int blank_chars) { |
2893 | 6.65M | int i; |
2894 | 6.65M | xmlNodePtr lastChild; |
2895 | | |
2896 | | /* |
2897 | | * Don't spend time trying to differentiate them, the same callback is |
2898 | | * used ! |
2899 | | */ |
2900 | 6.65M | if (ctxt->sax->ignorableWhitespace == ctxt->sax->characters) |
2901 | 2.63M | return(0); |
2902 | | |
2903 | | /* |
2904 | | * Check for xml:space value. |
2905 | | */ |
2906 | 4.02M | if ((ctxt->space == NULL) || (*(ctxt->space) == 1) || |
2907 | 4.02M | (*(ctxt->space) == -2)) |
2908 | 2.98M | return(0); |
2909 | | |
2910 | | /* |
2911 | | * Check that the string is made of blanks |
2912 | | */ |
2913 | 1.03M | if (blank_chars == 0) { |
2914 | 2.60M | for (i = 0;i < len;i++) |
2915 | 2.51M | if (!(IS_BLANK_CH(str[i]))) return(0); |
2916 | 584k | } |
2917 | | |
2918 | | /* |
2919 | | * Look if the element is mixed content in the DTD if available |
2920 | | */ |
2921 | 537k | if (ctxt->node == NULL) return(0); |
2922 | 531k | if (ctxt->myDoc != NULL) { |
2923 | 531k | xmlElementPtr elemDecl = NULL; |
2924 | 531k | xmlDocPtr doc = ctxt->myDoc; |
2925 | 531k | const xmlChar *prefix = NULL; |
2926 | | |
2927 | 531k | if (ctxt->node->ns) |
2928 | 122k | prefix = ctxt->node->ns->prefix; |
2929 | 531k | if (doc->intSubset != NULL) |
2930 | 428k | elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name, |
2931 | 428k | prefix); |
2932 | 531k | if ((elemDecl == NULL) && (doc->extSubset != NULL)) |
2933 | 19.9k | elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name, |
2934 | 19.9k | prefix); |
2935 | 531k | if (elemDecl != NULL) { |
2936 | 175k | if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT) |
2937 | 158k | return(1); |
2938 | 17.0k | if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) || |
2939 | 17.0k | (elemDecl->etype == XML_ELEMENT_TYPE_MIXED)) |
2940 | 3.71k | return(0); |
2941 | 17.0k | } |
2942 | 531k | } |
2943 | | |
2944 | | /* |
2945 | | * Otherwise, heuristic :-\ |
2946 | | */ |
2947 | 369k | if ((RAW != '<') && (RAW != 0xD)) return(0); |
2948 | 324k | if ((ctxt->node->children == NULL) && |
2949 | 324k | (RAW == '<') && (NXT(1) == '/')) return(0); |
2950 | | |
2951 | 317k | lastChild = xmlGetLastChild(ctxt->node); |
2952 | 317k | if (lastChild == NULL) { |
2953 | 198k | if ((ctxt->node->type != XML_ELEMENT_NODE) && |
2954 | 198k | (ctxt->node->content != NULL)) return(0); |
2955 | 198k | } else if (xmlNodeIsText(lastChild)) |
2956 | 15.3k | return(0); |
2957 | 104k | else if ((ctxt->node->children != NULL) && |
2958 | 104k | (xmlNodeIsText(ctxt->node->children))) |
2959 | 12.5k | return(0); |
2960 | 289k | return(1); |
2961 | 317k | } |
2962 | | |
2963 | | /************************************************************************ |
2964 | | * * |
2965 | | * Extra stuff for namespace support * |
2966 | | * Relates to http://www.w3.org/TR/WD-xml-names * |
2967 | | * * |
2968 | | ************************************************************************/ |
2969 | | |
2970 | | /** |
2971 | | * xmlSplitQName: |
2972 | | * @ctxt: an XML parser context |
2973 | | * @name: an XML parser context |
2974 | | * @prefixOut: a xmlChar ** |
2975 | | * |
2976 | | * DEPRECATED: Don't use. |
2977 | | * |
2978 | | * parse an UTF8 encoded XML qualified name string |
2979 | | * |
2980 | | * [NS 5] QName ::= (Prefix ':')? LocalPart |
2981 | | * |
2982 | | * [NS 6] Prefix ::= NCName |
2983 | | * |
2984 | | * [NS 7] LocalPart ::= NCName |
2985 | | * |
2986 | | * Returns the local part, and prefix is updated |
2987 | | * to get the Prefix if any. |
2988 | | */ |
2989 | | |
2990 | | xmlChar * |
2991 | 0 | xmlSplitQName(xmlParserCtxtPtr ctxt, const xmlChar *name, xmlChar **prefixOut) { |
2992 | 0 | xmlChar buf[XML_MAX_NAMELEN + 5]; |
2993 | 0 | xmlChar *buffer = NULL; |
2994 | 0 | int len = 0; |
2995 | 0 | int max = XML_MAX_NAMELEN; |
2996 | 0 | xmlChar *ret = NULL; |
2997 | 0 | xmlChar *prefix; |
2998 | 0 | const xmlChar *cur = name; |
2999 | 0 | int c; |
3000 | |
|
3001 | 0 | if (prefixOut == NULL) return(NULL); |
3002 | 0 | *prefixOut = NULL; |
3003 | |
|
3004 | 0 | if (cur == NULL) return(NULL); |
3005 | | |
3006 | | /* nasty but well=formed */ |
3007 | 0 | if (cur[0] == ':') |
3008 | 0 | return(xmlStrdup(name)); |
3009 | | |
3010 | 0 | c = *cur++; |
3011 | 0 | while ((c != 0) && (c != ':') && (len < max)) { /* tested bigname.xml */ |
3012 | 0 | buf[len++] = c; |
3013 | 0 | c = *cur++; |
3014 | 0 | } |
3015 | 0 | if (len >= max) { |
3016 | | /* |
3017 | | * Okay someone managed to make a huge name, so he's ready to pay |
3018 | | * for the processing speed. |
3019 | | */ |
3020 | 0 | max = len * 2; |
3021 | |
|
3022 | 0 | buffer = xmlMalloc(max); |
3023 | 0 | if (buffer == NULL) { |
3024 | 0 | xmlErrMemory(ctxt); |
3025 | 0 | return(NULL); |
3026 | 0 | } |
3027 | 0 | memcpy(buffer, buf, len); |
3028 | 0 | while ((c != 0) && (c != ':')) { /* tested bigname.xml */ |
3029 | 0 | if (len + 10 > max) { |
3030 | 0 | xmlChar *tmp; |
3031 | |
|
3032 | 0 | max *= 2; |
3033 | 0 | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3034 | 0 | if (tmp == NULL) { |
3035 | 0 | xmlFree(buffer); |
3036 | 0 | xmlErrMemory(ctxt); |
3037 | 0 | return(NULL); |
3038 | 0 | } |
3039 | 0 | buffer = tmp; |
3040 | 0 | } |
3041 | 0 | buffer[len++] = c; |
3042 | 0 | c = *cur++; |
3043 | 0 | } |
3044 | 0 | buffer[len] = 0; |
3045 | 0 | } |
3046 | | |
3047 | 0 | if ((c == ':') && (*cur == 0)) { |
3048 | 0 | if (buffer != NULL) |
3049 | 0 | xmlFree(buffer); |
3050 | 0 | return(xmlStrdup(name)); |
3051 | 0 | } |
3052 | | |
3053 | 0 | if (buffer == NULL) { |
3054 | 0 | ret = xmlStrndup(buf, len); |
3055 | 0 | if (ret == NULL) { |
3056 | 0 | xmlErrMemory(ctxt); |
3057 | 0 | return(NULL); |
3058 | 0 | } |
3059 | 0 | } else { |
3060 | 0 | ret = buffer; |
3061 | 0 | buffer = NULL; |
3062 | 0 | max = XML_MAX_NAMELEN; |
3063 | 0 | } |
3064 | | |
3065 | | |
3066 | 0 | if (c == ':') { |
3067 | 0 | c = *cur; |
3068 | 0 | prefix = ret; |
3069 | 0 | if (c == 0) { |
3070 | 0 | ret = xmlStrndup(BAD_CAST "", 0); |
3071 | 0 | if (ret == NULL) { |
3072 | 0 | xmlFree(prefix); |
3073 | 0 | return(NULL); |
3074 | 0 | } |
3075 | 0 | *prefixOut = prefix; |
3076 | 0 | return(ret); |
3077 | 0 | } |
3078 | 0 | len = 0; |
3079 | | |
3080 | | /* |
3081 | | * Check that the first character is proper to start |
3082 | | * a new name |
3083 | | */ |
3084 | 0 | if (!(((c >= 0x61) && (c <= 0x7A)) || |
3085 | 0 | ((c >= 0x41) && (c <= 0x5A)) || |
3086 | 0 | (c == '_') || (c == ':'))) { |
3087 | 0 | int l; |
3088 | 0 | int first = CUR_SCHAR(cur, l); |
3089 | |
|
3090 | 0 | if (!IS_LETTER(first) && (first != '_')) { |
3091 | 0 | xmlFatalErrMsgStr(ctxt, XML_NS_ERR_QNAME, |
3092 | 0 | "Name %s is not XML Namespace compliant\n", |
3093 | 0 | name); |
3094 | 0 | } |
3095 | 0 | } |
3096 | 0 | cur++; |
3097 | |
|
3098 | 0 | while ((c != 0) && (len < max)) { /* tested bigname2.xml */ |
3099 | 0 | buf[len++] = c; |
3100 | 0 | c = *cur++; |
3101 | 0 | } |
3102 | 0 | if (len >= max) { |
3103 | | /* |
3104 | | * Okay someone managed to make a huge name, so he's ready to pay |
3105 | | * for the processing speed. |
3106 | | */ |
3107 | 0 | max = len * 2; |
3108 | |
|
3109 | 0 | buffer = xmlMalloc(max); |
3110 | 0 | if (buffer == NULL) { |
3111 | 0 | xmlErrMemory(ctxt); |
3112 | 0 | xmlFree(prefix); |
3113 | 0 | return(NULL); |
3114 | 0 | } |
3115 | 0 | memcpy(buffer, buf, len); |
3116 | 0 | while (c != 0) { /* tested bigname2.xml */ |
3117 | 0 | if (len + 10 > max) { |
3118 | 0 | xmlChar *tmp; |
3119 | |
|
3120 | 0 | max *= 2; |
3121 | 0 | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3122 | 0 | if (tmp == NULL) { |
3123 | 0 | xmlErrMemory(ctxt); |
3124 | 0 | xmlFree(prefix); |
3125 | 0 | xmlFree(buffer); |
3126 | 0 | return(NULL); |
3127 | 0 | } |
3128 | 0 | buffer = tmp; |
3129 | 0 | } |
3130 | 0 | buffer[len++] = c; |
3131 | 0 | c = *cur++; |
3132 | 0 | } |
3133 | 0 | buffer[len] = 0; |
3134 | 0 | } |
3135 | | |
3136 | 0 | if (buffer == NULL) { |
3137 | 0 | ret = xmlStrndup(buf, len); |
3138 | 0 | if (ret == NULL) { |
3139 | 0 | xmlFree(prefix); |
3140 | 0 | return(NULL); |
3141 | 0 | } |
3142 | 0 | } else { |
3143 | 0 | ret = buffer; |
3144 | 0 | } |
3145 | | |
3146 | 0 | *prefixOut = prefix; |
3147 | 0 | } |
3148 | | |
3149 | 0 | return(ret); |
3150 | 0 | } |
3151 | | |
3152 | | /************************************************************************ |
3153 | | * * |
3154 | | * The parser itself * |
3155 | | * Relates to http://www.w3.org/TR/REC-xml * |
3156 | | * * |
3157 | | ************************************************************************/ |
3158 | | |
3159 | | /************************************************************************ |
3160 | | * * |
3161 | | * Routines to parse Name, NCName and NmToken * |
3162 | | * * |
3163 | | ************************************************************************/ |
3164 | | |
3165 | | /* |
3166 | | * The two following functions are related to the change of accepted |
3167 | | * characters for Name and NmToken in the Revision 5 of XML-1.0 |
3168 | | * They correspond to the modified production [4] and the new production [4a] |
3169 | | * changes in that revision. Also note that the macros used for the |
3170 | | * productions Letter, Digit, CombiningChar and Extender are not needed |
3171 | | * anymore. |
3172 | | * We still keep compatibility to pre-revision5 parsing semantic if the |
3173 | | * new XML_PARSE_OLD10 option is given to the parser. |
3174 | | */ |
3175 | | static int |
3176 | 12.9M | xmlIsNameStartChar(xmlParserCtxtPtr ctxt, int c) { |
3177 | 12.9M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3178 | | /* |
3179 | | * Use the new checks of production [4] [4a] amd [5] of the |
3180 | | * Update 5 of XML-1.0 |
3181 | | */ |
3182 | 8.36M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3183 | 8.36M | (((c >= 'a') && (c <= 'z')) || |
3184 | 8.35M | ((c >= 'A') && (c <= 'Z')) || |
3185 | 8.35M | (c == '_') || (c == ':') || |
3186 | 8.35M | ((c >= 0xC0) && (c <= 0xD6)) || |
3187 | 8.35M | ((c >= 0xD8) && (c <= 0xF6)) || |
3188 | 8.35M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3189 | 8.35M | ((c >= 0x370) && (c <= 0x37D)) || |
3190 | 8.35M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3191 | 8.35M | ((c >= 0x200C) && (c <= 0x200D)) || |
3192 | 8.35M | ((c >= 0x2070) && (c <= 0x218F)) || |
3193 | 8.35M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3194 | 8.35M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3195 | 8.35M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3196 | 8.35M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3197 | 8.35M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3198 | 6.07M | return(1); |
3199 | 8.36M | } else { |
3200 | 4.57M | if (IS_LETTER(c) || (c == '_') || (c == ':')) |
3201 | 3.85M | return(1); |
3202 | 4.57M | } |
3203 | 3.00M | return(0); |
3204 | 12.9M | } |
3205 | | |
3206 | | static int |
3207 | 110M | xmlIsNameChar(xmlParserCtxtPtr ctxt, int c) { |
3208 | 110M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3209 | | /* |
3210 | | * Use the new checks of production [4] [4a] amd [5] of the |
3211 | | * Update 5 of XML-1.0 |
3212 | | */ |
3213 | 87.8M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3214 | 87.8M | (((c >= 'a') && (c <= 'z')) || |
3215 | 87.7M | ((c >= 'A') && (c <= 'Z')) || |
3216 | 87.7M | ((c >= '0') && (c <= '9')) || /* !start */ |
3217 | 87.7M | (c == '_') || (c == ':') || |
3218 | 87.7M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3219 | 87.7M | ((c >= 0xC0) && (c <= 0xD6)) || |
3220 | 87.7M | ((c >= 0xD8) && (c <= 0xF6)) || |
3221 | 87.7M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3222 | 87.7M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3223 | 87.7M | ((c >= 0x370) && (c <= 0x37D)) || |
3224 | 87.7M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3225 | 87.7M | ((c >= 0x200C) && (c <= 0x200D)) || |
3226 | 87.7M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3227 | 87.7M | ((c >= 0x2070) && (c <= 0x218F)) || |
3228 | 87.7M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3229 | 87.7M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3230 | 87.7M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3231 | 87.7M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3232 | 87.7M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3233 | 81.5M | return(1); |
3234 | 87.8M | } else { |
3235 | 22.5M | if ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3236 | 22.5M | (c == '.') || (c == '-') || |
3237 | 22.5M | (c == '_') || (c == ':') || |
3238 | 22.5M | (IS_COMBINING(c)) || |
3239 | 22.5M | (IS_EXTENDER(c))) |
3240 | 18.7M | return(1); |
3241 | 22.5M | } |
3242 | 10.1M | return(0); |
3243 | 110M | } |
3244 | | |
3245 | | static const xmlChar * |
3246 | 3.61M | xmlParseNameComplex(xmlParserCtxtPtr ctxt) { |
3247 | 3.61M | const xmlChar *ret; |
3248 | 3.61M | int len = 0, l; |
3249 | 3.61M | int c; |
3250 | 3.61M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3251 | 2.04M | XML_MAX_TEXT_LENGTH : |
3252 | 3.61M | XML_MAX_NAME_LENGTH; |
3253 | | |
3254 | | /* |
3255 | | * Handler for more complex cases |
3256 | | */ |
3257 | 3.61M | c = xmlCurrentChar(ctxt, &l); |
3258 | 3.61M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3259 | | /* |
3260 | | * Use the new checks of production [4] [4a] amd [5] of the |
3261 | | * Update 5 of XML-1.0 |
3262 | | */ |
3263 | 2.60M | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3264 | 2.60M | (!(((c >= 'a') && (c <= 'z')) || |
3265 | 2.40M | ((c >= 'A') && (c <= 'Z')) || |
3266 | 2.40M | (c == '_') || (c == ':') || |
3267 | 2.40M | ((c >= 0xC0) && (c <= 0xD6)) || |
3268 | 2.40M | ((c >= 0xD8) && (c <= 0xF6)) || |
3269 | 2.40M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3270 | 2.40M | ((c >= 0x370) && (c <= 0x37D)) || |
3271 | 2.40M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3272 | 2.40M | ((c >= 0x200C) && (c <= 0x200D)) || |
3273 | 2.40M | ((c >= 0x2070) && (c <= 0x218F)) || |
3274 | 2.40M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3275 | 2.40M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3276 | 2.40M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3277 | 2.40M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3278 | 2.40M | ((c >= 0x10000) && (c <= 0xEFFFF))))) { |
3279 | 2.03M | return(NULL); |
3280 | 2.03M | } |
3281 | 571k | len += l; |
3282 | 571k | NEXTL(l); |
3283 | 571k | c = xmlCurrentChar(ctxt, &l); |
3284 | 25.4M | while ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3285 | 25.4M | (((c >= 'a') && (c <= 'z')) || |
3286 | 25.3M | ((c >= 'A') && (c <= 'Z')) || |
3287 | 25.3M | ((c >= '0') && (c <= '9')) || /* !start */ |
3288 | 25.3M | (c == '_') || (c == ':') || |
3289 | 25.3M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3290 | 25.3M | ((c >= 0xC0) && (c <= 0xD6)) || |
3291 | 25.3M | ((c >= 0xD8) && (c <= 0xF6)) || |
3292 | 25.3M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3293 | 25.3M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3294 | 25.3M | ((c >= 0x370) && (c <= 0x37D)) || |
3295 | 25.3M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3296 | 25.3M | ((c >= 0x200C) && (c <= 0x200D)) || |
3297 | 25.3M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3298 | 25.3M | ((c >= 0x2070) && (c <= 0x218F)) || |
3299 | 25.3M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3300 | 25.3M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3301 | 25.3M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3302 | 25.3M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3303 | 25.3M | ((c >= 0x10000) && (c <= 0xEFFFF)) |
3304 | 25.3M | )) { |
3305 | 24.8M | if (len <= INT_MAX - l) |
3306 | 24.8M | len += l; |
3307 | 24.8M | NEXTL(l); |
3308 | 24.8M | c = xmlCurrentChar(ctxt, &l); |
3309 | 24.8M | } |
3310 | 1.01M | } else { |
3311 | 1.01M | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3312 | 1.01M | (!IS_LETTER(c) && (c != '_') && |
3313 | 950k | (c != ':'))) { |
3314 | 686k | return(NULL); |
3315 | 686k | } |
3316 | 325k | len += l; |
3317 | 325k | NEXTL(l); |
3318 | 325k | c = xmlCurrentChar(ctxt, &l); |
3319 | | |
3320 | 8.28M | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3321 | 8.28M | ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3322 | 8.22M | (c == '.') || (c == '-') || |
3323 | 8.22M | (c == '_') || (c == ':') || |
3324 | 8.22M | (IS_COMBINING(c)) || |
3325 | 8.22M | (IS_EXTENDER(c)))) { |
3326 | 7.95M | if (len <= INT_MAX - l) |
3327 | 7.95M | len += l; |
3328 | 7.95M | NEXTL(l); |
3329 | 7.95M | c = xmlCurrentChar(ctxt, &l); |
3330 | 7.95M | } |
3331 | 325k | } |
3332 | 896k | if (len > maxLength) { |
3333 | 409 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3334 | 409 | return(NULL); |
3335 | 409 | } |
3336 | 895k | if (ctxt->input->cur - ctxt->input->base < len) { |
3337 | | /* |
3338 | | * There were a couple of bugs where PERefs lead to to a change |
3339 | | * of the buffer. Check the buffer size to avoid passing an invalid |
3340 | | * pointer to xmlDictLookup. |
3341 | | */ |
3342 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
3343 | 0 | "unexpected change of input buffer"); |
3344 | 0 | return (NULL); |
3345 | 0 | } |
3346 | 895k | if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r')) |
3347 | 2.41k | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len); |
3348 | 893k | else |
3349 | 893k | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len); |
3350 | 895k | if (ret == NULL) |
3351 | 28 | xmlErrMemory(ctxt); |
3352 | 895k | return(ret); |
3353 | 895k | } |
3354 | | |
3355 | | /** |
3356 | | * xmlParseName: |
3357 | | * @ctxt: an XML parser context |
3358 | | * |
3359 | | * DEPRECATED: Internal function, don't use. |
3360 | | * |
3361 | | * parse an XML name. |
3362 | | * |
3363 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3364 | | * CombiningChar | Extender |
3365 | | * |
3366 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3367 | | * |
3368 | | * [6] Names ::= Name (#x20 Name)* |
3369 | | * |
3370 | | * Returns the Name parsed or NULL |
3371 | | */ |
3372 | | |
3373 | | const xmlChar * |
3374 | 19.3M | xmlParseName(xmlParserCtxtPtr ctxt) { |
3375 | 19.3M | const xmlChar *in; |
3376 | 19.3M | const xmlChar *ret; |
3377 | 19.3M | size_t count = 0; |
3378 | 19.3M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3379 | 8.57M | XML_MAX_TEXT_LENGTH : |
3380 | 19.3M | XML_MAX_NAME_LENGTH; |
3381 | | |
3382 | 19.3M | GROW; |
3383 | | |
3384 | | /* |
3385 | | * Accelerator for simple ASCII names |
3386 | | */ |
3387 | 19.3M | in = ctxt->input->cur; |
3388 | 19.3M | if (((*in >= 0x61) && (*in <= 0x7A)) || |
3389 | 19.3M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3390 | 19.3M | (*in == '_') || (*in == ':')) { |
3391 | 16.3M | in++; |
3392 | 77.3M | while (((*in >= 0x61) && (*in <= 0x7A)) || |
3393 | 77.3M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3394 | 77.3M | ((*in >= 0x30) && (*in <= 0x39)) || |
3395 | 77.3M | (*in == '_') || (*in == '-') || |
3396 | 77.3M | (*in == ':') || (*in == '.')) |
3397 | 60.9M | in++; |
3398 | 16.3M | if ((*in > 0) && (*in < 0x80)) { |
3399 | 15.7M | count = in - ctxt->input->cur; |
3400 | 15.7M | if (count > maxLength) { |
3401 | 132 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3402 | 132 | return(NULL); |
3403 | 132 | } |
3404 | 15.7M | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3405 | 15.7M | ctxt->input->cur = in; |
3406 | 15.7M | ctxt->input->col += count; |
3407 | 15.7M | if (ret == NULL) |
3408 | 72 | xmlErrMemory(ctxt); |
3409 | 15.7M | return(ret); |
3410 | 15.7M | } |
3411 | 16.3M | } |
3412 | | /* accelerator for special cases */ |
3413 | 3.61M | return(xmlParseNameComplex(ctxt)); |
3414 | 19.3M | } |
3415 | | |
3416 | | static xmlHashedString |
3417 | 4.23M | xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) { |
3418 | 4.23M | xmlHashedString ret; |
3419 | 4.23M | int len = 0, l; |
3420 | 4.23M | int c; |
3421 | 4.23M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3422 | 1.97M | XML_MAX_TEXT_LENGTH : |
3423 | 4.23M | XML_MAX_NAME_LENGTH; |
3424 | 4.23M | size_t startPosition = 0; |
3425 | | |
3426 | 4.23M | ret.name = NULL; |
3427 | 4.23M | ret.hashValue = 0; |
3428 | | |
3429 | | /* |
3430 | | * Handler for more complex cases |
3431 | | */ |
3432 | 4.23M | startPosition = CUR_PTR - BASE_PTR; |
3433 | 4.23M | c = xmlCurrentChar(ctxt, &l); |
3434 | 4.23M | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3435 | 4.23M | (!xmlIsNameStartChar(ctxt, c) || (c == ':'))) { |
3436 | 3.33M | return(ret); |
3437 | 3.33M | } |
3438 | | |
3439 | 64.9M | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3440 | 64.9M | (xmlIsNameChar(ctxt, c) && (c != ':'))) { |
3441 | 64.0M | if (len <= INT_MAX - l) |
3442 | 64.0M | len += l; |
3443 | 64.0M | NEXTL(l); |
3444 | 64.0M | c = xmlCurrentChar(ctxt, &l); |
3445 | 64.0M | } |
3446 | 895k | if (len > maxLength) { |
3447 | 648 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3448 | 648 | return(ret); |
3449 | 648 | } |
3450 | 894k | ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len); |
3451 | 894k | if (ret.name == NULL) |
3452 | 19 | xmlErrMemory(ctxt); |
3453 | 894k | return(ret); |
3454 | 895k | } |
3455 | | |
3456 | | /** |
3457 | | * xmlParseNCName: |
3458 | | * @ctxt: an XML parser context |
3459 | | * @len: length of the string parsed |
3460 | | * |
3461 | | * parse an XML name. |
3462 | | * |
3463 | | * [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' | |
3464 | | * CombiningChar | Extender |
3465 | | * |
3466 | | * [5NS] NCName ::= (Letter | '_') (NCNameChar)* |
3467 | | * |
3468 | | * Returns the Name parsed or NULL |
3469 | | */ |
3470 | | |
3471 | | static xmlHashedString |
3472 | 16.3M | xmlParseNCName(xmlParserCtxtPtr ctxt) { |
3473 | 16.3M | const xmlChar *in, *e; |
3474 | 16.3M | xmlHashedString ret; |
3475 | 16.3M | size_t count = 0; |
3476 | 16.3M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3477 | 7.54M | XML_MAX_TEXT_LENGTH : |
3478 | 16.3M | XML_MAX_NAME_LENGTH; |
3479 | | |
3480 | 16.3M | ret.name = NULL; |
3481 | | |
3482 | | /* |
3483 | | * Accelerator for simple ASCII names |
3484 | | */ |
3485 | 16.3M | in = ctxt->input->cur; |
3486 | 16.3M | e = ctxt->input->end; |
3487 | 16.3M | if ((((*in >= 0x61) && (*in <= 0x7A)) || |
3488 | 16.3M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3489 | 16.3M | (*in == '_')) && (in < e)) { |
3490 | 12.8M | in++; |
3491 | 68.7M | while ((((*in >= 0x61) && (*in <= 0x7A)) || |
3492 | 68.7M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3493 | 68.7M | ((*in >= 0x30) && (*in <= 0x39)) || |
3494 | 68.7M | (*in == '_') || (*in == '-') || |
3495 | 68.7M | (*in == '.')) && (in < e)) |
3496 | 55.9M | in++; |
3497 | 12.8M | if (in >= e) |
3498 | 9.47k | goto complex; |
3499 | 12.8M | if ((*in > 0) && (*in < 0x80)) { |
3500 | 12.1M | count = in - ctxt->input->cur; |
3501 | 12.1M | if (count > maxLength) { |
3502 | 227 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3503 | 227 | return(ret); |
3504 | 227 | } |
3505 | 12.1M | ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count); |
3506 | 12.1M | ctxt->input->cur = in; |
3507 | 12.1M | ctxt->input->col += count; |
3508 | 12.1M | if (ret.name == NULL) { |
3509 | 39 | xmlErrMemory(ctxt); |
3510 | 39 | } |
3511 | 12.1M | return(ret); |
3512 | 12.1M | } |
3513 | 12.8M | } |
3514 | 4.23M | complex: |
3515 | 4.23M | return(xmlParseNCNameComplex(ctxt)); |
3516 | 16.3M | } |
3517 | | |
3518 | | /** |
3519 | | * xmlParseNameAndCompare: |
3520 | | * @ctxt: an XML parser context |
3521 | | * |
3522 | | * parse an XML name and compares for match |
3523 | | * (specialized for endtag parsing) |
3524 | | * |
3525 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
3526 | | * and the name for mismatch |
3527 | | */ |
3528 | | |
3529 | | static const xmlChar * |
3530 | 2.37M | xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) { |
3531 | 2.37M | register const xmlChar *cmp = other; |
3532 | 2.37M | register const xmlChar *in; |
3533 | 2.37M | const xmlChar *ret; |
3534 | | |
3535 | 2.37M | GROW; |
3536 | | |
3537 | 2.37M | in = ctxt->input->cur; |
3538 | 6.66M | while (*in != 0 && *in == *cmp) { |
3539 | 4.28M | ++in; |
3540 | 4.28M | ++cmp; |
3541 | 4.28M | } |
3542 | 2.37M | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
3543 | | /* success */ |
3544 | 1.90M | ctxt->input->col += in - ctxt->input->cur; |
3545 | 1.90M | ctxt->input->cur = in; |
3546 | 1.90M | return (const xmlChar*) 1; |
3547 | 1.90M | } |
3548 | | /* failure (or end of input buffer), check with full function */ |
3549 | 475k | ret = xmlParseName (ctxt); |
3550 | | /* strings coming from the dictionary direct compare possible */ |
3551 | 475k | if (ret == other) { |
3552 | 37.8k | return (const xmlChar*) 1; |
3553 | 37.8k | } |
3554 | 437k | return ret; |
3555 | 475k | } |
3556 | | |
3557 | | /** |
3558 | | * xmlParseStringName: |
3559 | | * @ctxt: an XML parser context |
3560 | | * @str: a pointer to the string pointer (IN/OUT) |
3561 | | * |
3562 | | * parse an XML name. |
3563 | | * |
3564 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3565 | | * CombiningChar | Extender |
3566 | | * |
3567 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3568 | | * |
3569 | | * [6] Names ::= Name (#x20 Name)* |
3570 | | * |
3571 | | * Returns the Name parsed or NULL. The @str pointer |
3572 | | * is updated to the current location in the string. |
3573 | | */ |
3574 | | |
3575 | | static xmlChar * |
3576 | 8.92M | xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) { |
3577 | 8.92M | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3578 | 8.92M | xmlChar *ret; |
3579 | 8.92M | const xmlChar *cur = *str; |
3580 | 8.92M | int len = 0, l; |
3581 | 8.92M | int c; |
3582 | 8.92M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3583 | 3.49M | XML_MAX_TEXT_LENGTH : |
3584 | 8.92M | XML_MAX_NAME_LENGTH; |
3585 | | |
3586 | 8.92M | c = CUR_SCHAR(cur, l); |
3587 | 8.92M | if (!xmlIsNameStartChar(ctxt, c)) { |
3588 | 18.3k | return(NULL); |
3589 | 18.3k | } |
3590 | | |
3591 | 8.90M | COPY_BUF(buf, len, c); |
3592 | 8.90M | cur += l; |
3593 | 8.90M | c = CUR_SCHAR(cur, l); |
3594 | 20.1M | while (xmlIsNameChar(ctxt, c)) { |
3595 | 11.2M | COPY_BUF(buf, len, c); |
3596 | 11.2M | cur += l; |
3597 | 11.2M | c = CUR_SCHAR(cur, l); |
3598 | 11.2M | if (len >= XML_MAX_NAMELEN) { /* test bigentname.xml */ |
3599 | | /* |
3600 | | * Okay someone managed to make a huge name, so he's ready to pay |
3601 | | * for the processing speed. |
3602 | | */ |
3603 | 7.18k | xmlChar *buffer; |
3604 | 7.18k | int max = len * 2; |
3605 | | |
3606 | 7.18k | buffer = xmlMalloc(max); |
3607 | 7.18k | if (buffer == NULL) { |
3608 | 20 | xmlErrMemory(ctxt); |
3609 | 20 | return(NULL); |
3610 | 20 | } |
3611 | 7.16k | memcpy(buffer, buf, len); |
3612 | 3.81M | while (xmlIsNameChar(ctxt, c)) { |
3613 | 3.80M | if (len + 10 > max) { |
3614 | 10.7k | xmlChar *tmp; |
3615 | | |
3616 | 10.7k | max *= 2; |
3617 | 10.7k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3618 | 10.7k | if (tmp == NULL) { |
3619 | 15 | xmlErrMemory(ctxt); |
3620 | 15 | xmlFree(buffer); |
3621 | 15 | return(NULL); |
3622 | 15 | } |
3623 | 10.7k | buffer = tmp; |
3624 | 10.7k | } |
3625 | 3.80M | COPY_BUF(buffer, len, c); |
3626 | 3.80M | cur += l; |
3627 | 3.80M | c = CUR_SCHAR(cur, l); |
3628 | 3.80M | if (len > maxLength) { |
3629 | 104 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3630 | 104 | xmlFree(buffer); |
3631 | 104 | return(NULL); |
3632 | 104 | } |
3633 | 3.80M | } |
3634 | 7.04k | buffer[len] = 0; |
3635 | 7.04k | *str = cur; |
3636 | 7.04k | return(buffer); |
3637 | 7.16k | } |
3638 | 11.2M | } |
3639 | 8.89M | if (len > maxLength) { |
3640 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3641 | 0 | return(NULL); |
3642 | 0 | } |
3643 | 8.89M | *str = cur; |
3644 | 8.89M | ret = xmlStrndup(buf, len); |
3645 | 8.89M | if (ret == NULL) |
3646 | 276 | xmlErrMemory(ctxt); |
3647 | 8.89M | return(ret); |
3648 | 8.89M | } |
3649 | | |
3650 | | /** |
3651 | | * xmlParseNmtoken: |
3652 | | * @ctxt: an XML parser context |
3653 | | * |
3654 | | * DEPRECATED: Internal function, don't use. |
3655 | | * |
3656 | | * parse an XML Nmtoken. |
3657 | | * |
3658 | | * [7] Nmtoken ::= (NameChar)+ |
3659 | | * |
3660 | | * [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)* |
3661 | | * |
3662 | | * Returns the Nmtoken parsed or NULL |
3663 | | */ |
3664 | | |
3665 | | xmlChar * |
3666 | 495k | xmlParseNmtoken(xmlParserCtxtPtr ctxt) { |
3667 | 495k | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3668 | 495k | xmlChar *ret; |
3669 | 495k | int len = 0, l; |
3670 | 495k | int c; |
3671 | 495k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3672 | 160k | XML_MAX_TEXT_LENGTH : |
3673 | 495k | XML_MAX_NAME_LENGTH; |
3674 | | |
3675 | 495k | c = xmlCurrentChar(ctxt, &l); |
3676 | | |
3677 | 2.21M | while (xmlIsNameChar(ctxt, c)) { |
3678 | 1.72M | COPY_BUF(buf, len, c); |
3679 | 1.72M | NEXTL(l); |
3680 | 1.72M | c = xmlCurrentChar(ctxt, &l); |
3681 | 1.72M | if (len >= XML_MAX_NAMELEN) { |
3682 | | /* |
3683 | | * Okay someone managed to make a huge token, so he's ready to pay |
3684 | | * for the processing speed. |
3685 | | */ |
3686 | 8.05k | xmlChar *buffer; |
3687 | 8.05k | int max = len * 2; |
3688 | | |
3689 | 8.05k | buffer = xmlMalloc(max); |
3690 | 8.05k | if (buffer == NULL) { |
3691 | 17 | xmlErrMemory(ctxt); |
3692 | 17 | return(NULL); |
3693 | 17 | } |
3694 | 8.03k | memcpy(buffer, buf, len); |
3695 | 19.4M | while (xmlIsNameChar(ctxt, c)) { |
3696 | 19.4M | if (len + 10 > max) { |
3697 | 12.7k | xmlChar *tmp; |
3698 | | |
3699 | 12.7k | max *= 2; |
3700 | 12.7k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3701 | 12.7k | if (tmp == NULL) { |
3702 | 14 | xmlErrMemory(ctxt); |
3703 | 14 | xmlFree(buffer); |
3704 | 14 | return(NULL); |
3705 | 14 | } |
3706 | 12.6k | buffer = tmp; |
3707 | 12.6k | } |
3708 | 19.4M | COPY_BUF(buffer, len, c); |
3709 | 19.4M | if (len > maxLength) { |
3710 | 561 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3711 | 561 | xmlFree(buffer); |
3712 | 561 | return(NULL); |
3713 | 561 | } |
3714 | 19.4M | NEXTL(l); |
3715 | 19.4M | c = xmlCurrentChar(ctxt, &l); |
3716 | 19.4M | } |
3717 | 7.46k | buffer[len] = 0; |
3718 | 7.46k | return(buffer); |
3719 | 8.03k | } |
3720 | 1.72M | } |
3721 | 487k | if (len == 0) |
3722 | 126k | return(NULL); |
3723 | 360k | if (len > maxLength) { |
3724 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3725 | 0 | return(NULL); |
3726 | 0 | } |
3727 | 360k | ret = xmlStrndup(buf, len); |
3728 | 360k | if (ret == NULL) |
3729 | 61 | xmlErrMemory(ctxt); |
3730 | 360k | return(ret); |
3731 | 360k | } |
3732 | | |
3733 | | /** |
3734 | | * xmlExpandPEsInEntityValue: |
3735 | | * @ctxt: parser context |
3736 | | * @buf: string buffer |
3737 | | * @str: entity value |
3738 | | * @length: size of entity value |
3739 | | * @depth: nesting depth |
3740 | | * |
3741 | | * Validate an entity value and expand parameter entities. |
3742 | | */ |
3743 | | static void |
3744 | | xmlExpandPEsInEntityValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf, |
3745 | 406k | const xmlChar *str, int length, int depth) { |
3746 | 406k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
3747 | 406k | const xmlChar *end, *chunk; |
3748 | 406k | int c, l; |
3749 | | |
3750 | 406k | if (str == NULL) |
3751 | 41.0k | return; |
3752 | | |
3753 | 365k | depth += 1; |
3754 | 365k | if (depth > maxDepth) { |
3755 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
3756 | 0 | "Maximum entity nesting depth exceeded"); |
3757 | 0 | return; |
3758 | 0 | } |
3759 | | |
3760 | 365k | end = str + length; |
3761 | 365k | chunk = str; |
3762 | | |
3763 | 564M | while ((str < end) && (!PARSER_STOPPED(ctxt))) { |
3764 | 564M | c = *str; |
3765 | | |
3766 | 564M | if (c >= 0x80) { |
3767 | 327M | l = xmlUTF8MultibyteLen(ctxt, str, |
3768 | 327M | "invalid character in entity value\n"); |
3769 | 327M | if (l == 0) { |
3770 | 59.0M | if (chunk < str) |
3771 | 683k | xmlSBufAddString(buf, chunk, str - chunk); |
3772 | 59.0M | xmlSBufAddReplChar(buf); |
3773 | 59.0M | str += 1; |
3774 | 59.0M | chunk = str; |
3775 | 268M | } else { |
3776 | 268M | str += l; |
3777 | 268M | } |
3778 | 327M | } else if (c == '&') { |
3779 | 466k | if (str[1] == '#') { |
3780 | 237k | if (chunk < str) |
3781 | 68.2k | xmlSBufAddString(buf, chunk, str - chunk); |
3782 | | |
3783 | 237k | c = xmlParseStringCharRef(ctxt, &str); |
3784 | 237k | if (c == 0) |
3785 | 28.8k | return; |
3786 | | |
3787 | 208k | xmlSBufAddChar(buf, c); |
3788 | | |
3789 | 208k | chunk = str; |
3790 | 229k | } else { |
3791 | 229k | xmlChar *name; |
3792 | | |
3793 | | /* |
3794 | | * General entity references are checked for |
3795 | | * syntactic validity. |
3796 | | */ |
3797 | 229k | str++; |
3798 | 229k | name = xmlParseStringName(ctxt, &str); |
3799 | | |
3800 | 229k | if ((name == NULL) || (*str++ != ';')) { |
3801 | 18.1k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR, |
3802 | 18.1k | "EntityValue: '&' forbidden except for entities " |
3803 | 18.1k | "references\n"); |
3804 | 18.1k | xmlFree(name); |
3805 | 18.1k | return; |
3806 | 18.1k | } |
3807 | | |
3808 | 210k | xmlFree(name); |
3809 | 210k | } |
3810 | 236M | } else if (c == '%') { |
3811 | 146k | xmlEntityPtr ent; |
3812 | | |
3813 | 146k | if (chunk < str) |
3814 | 88.5k | xmlSBufAddString(buf, chunk, str - chunk); |
3815 | | |
3816 | 146k | ent = xmlParseStringPEReference(ctxt, &str); |
3817 | 146k | if (ent == NULL) |
3818 | 35.4k | return; |
3819 | | |
3820 | 110k | if (!PARSER_EXTERNAL(ctxt)) { |
3821 | 1.65k | xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL); |
3822 | 1.65k | return; |
3823 | 1.65k | } |
3824 | | |
3825 | 109k | if (ent->content == NULL) { |
3826 | | /* |
3827 | | * Note: external parsed entities will not be loaded, |
3828 | | * it is not required for a non-validating parser to |
3829 | | * complete external PEReferences coming from the |
3830 | | * internal subset |
3831 | | */ |
3832 | 41.7k | if (((ctxt->options & XML_PARSE_NO_XXE) == 0) && |
3833 | 41.7k | ((ctxt->replaceEntities) || |
3834 | 41.7k | (ctxt->validate))) { |
3835 | 36.8k | xmlLoadEntityContent(ctxt, ent); |
3836 | 36.8k | } else { |
3837 | 4.94k | xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING, |
3838 | 4.94k | "not validating will not read content for " |
3839 | 4.94k | "PE entity %s\n", ent->name, NULL); |
3840 | 4.94k | } |
3841 | 41.7k | } |
3842 | | |
3843 | | /* |
3844 | | * TODO: Skip if ent->content is still NULL. |
3845 | | */ |
3846 | | |
3847 | 109k | if (xmlParserEntityCheck(ctxt, ent->length)) |
3848 | 89 | return; |
3849 | | |
3850 | 108k | if (ent->flags & XML_ENT_EXPANDING) { |
3851 | 152 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
3852 | 152 | xmlHaltParser(ctxt); |
3853 | 152 | return; |
3854 | 152 | } |
3855 | | |
3856 | 108k | ent->flags |= XML_ENT_EXPANDING; |
3857 | 108k | xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length, |
3858 | 108k | depth); |
3859 | 108k | ent->flags &= ~XML_ENT_EXPANDING; |
3860 | | |
3861 | 108k | chunk = str; |
3862 | 236M | } else { |
3863 | | /* Normal ASCII char */ |
3864 | 236M | if (!IS_BYTE_CHAR(c)) { |
3865 | 15.8M | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
3866 | 15.8M | "invalid character in entity value\n"); |
3867 | 15.8M | if (chunk < str) |
3868 | 111k | xmlSBufAddString(buf, chunk, str - chunk); |
3869 | 15.8M | xmlSBufAddReplChar(buf); |
3870 | 15.8M | str += 1; |
3871 | 15.8M | chunk = str; |
3872 | 220M | } else { |
3873 | 220M | str += 1; |
3874 | 220M | } |
3875 | 236M | } |
3876 | 564M | } |
3877 | | |
3878 | 281k | if (chunk < str) |
3879 | 190k | xmlSBufAddString(buf, chunk, str - chunk); |
3880 | 281k | } |
3881 | | |
3882 | | /** |
3883 | | * xmlParseEntityValue: |
3884 | | * @ctxt: an XML parser context |
3885 | | * @orig: if non-NULL store a copy of the original entity value |
3886 | | * |
3887 | | * DEPRECATED: Internal function, don't use. |
3888 | | * |
3889 | | * parse a value for ENTITY declarations |
3890 | | * |
3891 | | * [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' | |
3892 | | * "'" ([^%&'] | PEReference | Reference)* "'" |
3893 | | * |
3894 | | * Returns the EntityValue parsed with reference substituted or NULL |
3895 | | */ |
3896 | | xmlChar * |
3897 | 299k | xmlParseEntityValue(xmlParserCtxtPtr ctxt, xmlChar **orig) { |
3898 | 299k | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3899 | 92.0k | XML_MAX_HUGE_LENGTH : |
3900 | 299k | XML_MAX_TEXT_LENGTH; |
3901 | 299k | xmlSBuf buf; |
3902 | 299k | const xmlChar *start; |
3903 | 299k | int quote, length; |
3904 | | |
3905 | 299k | xmlSBufInit(&buf, maxLength); |
3906 | | |
3907 | 299k | GROW; |
3908 | | |
3909 | 299k | quote = CUR; |
3910 | 299k | if ((quote != '"') && (quote != '\'')) { |
3911 | 0 | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
3912 | 0 | return(NULL); |
3913 | 0 | } |
3914 | 299k | CUR_PTR++; |
3915 | | |
3916 | 299k | length = 0; |
3917 | | |
3918 | | /* |
3919 | | * Copy raw content of the entity into a buffer |
3920 | | */ |
3921 | 902M | while (1) { |
3922 | 902M | int c; |
3923 | | |
3924 | 902M | if (PARSER_STOPPED(ctxt)) |
3925 | 13 | goto error; |
3926 | | |
3927 | 902M | if (CUR_PTR >= ctxt->input->end) { |
3928 | 1.30k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL); |
3929 | 1.30k | goto error; |
3930 | 1.30k | } |
3931 | | |
3932 | 902M | c = CUR; |
3933 | | |
3934 | 902M | if (c == 0) { |
3935 | 311 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
3936 | 311 | "invalid character in entity value\n"); |
3937 | 311 | goto error; |
3938 | 311 | } |
3939 | 902M | if (c == quote) |
3940 | 297k | break; |
3941 | 902M | NEXTL(1); |
3942 | 902M | length += 1; |
3943 | | |
3944 | | /* |
3945 | | * TODO: Check growth threshold |
3946 | | */ |
3947 | 902M | if (ctxt->input->end - CUR_PTR < 10) |
3948 | 142k | GROW; |
3949 | 902M | } |
3950 | | |
3951 | 297k | start = CUR_PTR - length; |
3952 | | |
3953 | 297k | if (orig != NULL) { |
3954 | 297k | *orig = xmlStrndup(start, length); |
3955 | 297k | if (*orig == NULL) |
3956 | 107 | xmlErrMemory(ctxt); |
3957 | 297k | } |
3958 | | |
3959 | 297k | xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr); |
3960 | | |
3961 | 297k | NEXTL(1); |
3962 | | |
3963 | 297k | return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long")); |
3964 | | |
3965 | 1.63k | error: |
3966 | 1.63k | xmlSBufCleanup(&buf, ctxt, "entity length too long"); |
3967 | 1.63k | return(NULL); |
3968 | 299k | } |
3969 | | |
3970 | | /** |
3971 | | * xmlCheckEntityInAttValue: |
3972 | | * @ctxt: parser context |
3973 | | * @pent: entity |
3974 | | * @depth: nesting depth |
3975 | | * |
3976 | | * Check an entity reference in an attribute value for validity |
3977 | | * without expanding it. |
3978 | | */ |
3979 | | static void |
3980 | 20.9k | xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) { |
3981 | 20.9k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
3982 | 20.9k | const xmlChar *str; |
3983 | 20.9k | unsigned long expandedSize = pent->length; |
3984 | 20.9k | int c, flags; |
3985 | | |
3986 | 20.9k | depth += 1; |
3987 | 20.9k | if (depth > maxDepth) { |
3988 | 15 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
3989 | 15 | "Maximum entity nesting depth exceeded"); |
3990 | 15 | return; |
3991 | 15 | } |
3992 | | |
3993 | 20.9k | if (pent->flags & XML_ENT_EXPANDING) { |
3994 | 1.15k | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
3995 | 1.15k | xmlHaltParser(ctxt); |
3996 | 1.15k | return; |
3997 | 1.15k | } |
3998 | | |
3999 | | /* |
4000 | | * If we're parsing a default attribute value in DTD content, |
4001 | | * the entity might reference other entities which weren't |
4002 | | * defined yet, so the check isn't reliable. |
4003 | | */ |
4004 | 19.7k | if (ctxt->inSubset == 0) |
4005 | 17.6k | flags = XML_ENT_CHECKED | XML_ENT_VALIDATED; |
4006 | 2.13k | else |
4007 | 2.13k | flags = XML_ENT_VALIDATED; |
4008 | | |
4009 | 19.7k | str = pent->content; |
4010 | 19.7k | if (str == NULL) |
4011 | 35 | goto done; |
4012 | | |
4013 | | /* |
4014 | | * Note that entity values are already validated. We only check |
4015 | | * for illegal less-than signs and compute the expanded size |
4016 | | * of the entity. No special handling for multi-byte characters |
4017 | | * is needed. |
4018 | | */ |
4019 | 65.0M | while (!PARSER_STOPPED(ctxt)) { |
4020 | 65.0M | c = *str; |
4021 | | |
4022 | 65.0M | if (c != '&') { |
4023 | 65.0M | if (c == 0) |
4024 | 18.3k | break; |
4025 | | |
4026 | 65.0M | if (c == '<') |
4027 | 8.32k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
4028 | 8.32k | "'<' in entity '%s' is not allowed in attributes " |
4029 | 8.32k | "values\n", pent->name); |
4030 | | |
4031 | 65.0M | str += 1; |
4032 | 65.0M | } else if (str[1] == '#') { |
4033 | 3.86k | int val; |
4034 | | |
4035 | 3.86k | val = xmlParseStringCharRef(ctxt, &str); |
4036 | 3.86k | if (val == 0) { |
4037 | 76 | pent->content[0] = 0; |
4038 | 76 | break; |
4039 | 76 | } |
4040 | 38.6k | } else { |
4041 | 38.6k | xmlChar *name; |
4042 | 38.6k | xmlEntityPtr ent; |
4043 | | |
4044 | 38.6k | name = xmlParseStringEntityRef(ctxt, &str); |
4045 | 38.6k | if (name == NULL) { |
4046 | 99 | pent->content[0] = 0; |
4047 | 99 | break; |
4048 | 99 | } |
4049 | | |
4050 | 38.5k | ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1); |
4051 | 38.5k | xmlFree(name); |
4052 | | |
4053 | 38.5k | if ((ent != NULL) && |
4054 | 38.5k | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) { |
4055 | 28.6k | if ((ent->flags & flags) != flags) { |
4056 | 13.9k | pent->flags |= XML_ENT_EXPANDING; |
4057 | 13.9k | xmlCheckEntityInAttValue(ctxt, ent, depth); |
4058 | 13.9k | pent->flags &= ~XML_ENT_EXPANDING; |
4059 | 13.9k | } |
4060 | | |
4061 | 28.6k | xmlSaturatedAdd(&expandedSize, ent->expandedSize); |
4062 | 28.6k | xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST); |
4063 | 28.6k | } |
4064 | 38.5k | } |
4065 | 65.0M | } |
4066 | | |
4067 | 19.7k | done: |
4068 | 19.7k | if (ctxt->inSubset == 0) |
4069 | 17.6k | pent->expandedSize = expandedSize; |
4070 | | |
4071 | 19.7k | pent->flags |= flags; |
4072 | 19.7k | } |
4073 | | |
4074 | | /** |
4075 | | * xmlExpandEntityInAttValue: |
4076 | | * @ctxt: parser context |
4077 | | * @buf: string buffer |
4078 | | * @str: entity or attribute value |
4079 | | * @pent: entity for entity value, NULL for attribute values |
4080 | | * @normalize: whether to collapse whitespace |
4081 | | * @inSpace: whitespace state |
4082 | | * @depth: nesting depth |
4083 | | * @check: whether to check for amplification |
4084 | | * |
4085 | | * Expand general entity references in an entity or attribute value. |
4086 | | * Perform attribute value normalization. |
4087 | | */ |
4088 | | static void |
4089 | | xmlExpandEntityInAttValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf, |
4090 | | const xmlChar *str, xmlEntityPtr pent, int normalize, |
4091 | 6.76M | int *inSpace, int depth, int check) { |
4092 | 6.76M | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
4093 | 6.76M | int c, chunkSize; |
4094 | | |
4095 | 6.76M | if (str == NULL) |
4096 | 201 | return; |
4097 | | |
4098 | 6.76M | depth += 1; |
4099 | 6.76M | if (depth > maxDepth) { |
4100 | 72.1k | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
4101 | 72.1k | "Maximum entity nesting depth exceeded"); |
4102 | 72.1k | return; |
4103 | 72.1k | } |
4104 | | |
4105 | 6.69M | if (pent != NULL) { |
4106 | 6.32M | if (pent->flags & XML_ENT_EXPANDING) { |
4107 | 467 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
4108 | 467 | xmlHaltParser(ctxt); |
4109 | 467 | return; |
4110 | 467 | } |
4111 | | |
4112 | 6.32M | if (check) { |
4113 | 5.98M | if (xmlParserEntityCheck(ctxt, pent->length)) |
4114 | 488 | return; |
4115 | 5.98M | } |
4116 | 6.32M | } |
4117 | | |
4118 | 6.69M | chunkSize = 0; |
4119 | | |
4120 | | /* |
4121 | | * Note that entity values are already validated. No special |
4122 | | * handling for multi-byte characters is needed. |
4123 | | */ |
4124 | 4.25G | while (!PARSER_STOPPED(ctxt)) { |
4125 | 4.25G | c = *str; |
4126 | | |
4127 | 4.25G | if (c != '&') { |
4128 | 4.24G | if (c == 0) |
4129 | 6.47M | break; |
4130 | | |
4131 | | /* |
4132 | | * If this function is called without an entity, it is used to |
4133 | | * expand entities in an attribute content where less-than was |
4134 | | * already unscaped and is allowed. |
4135 | | */ |
4136 | 4.23G | if ((pent != NULL) && (c == '<')) { |
4137 | 208k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
4138 | 208k | "'<' in entity '%s' is not allowed in attributes " |
4139 | 208k | "values\n", pent->name); |
4140 | 208k | break; |
4141 | 208k | } |
4142 | | |
4143 | 4.23G | if (c <= 0x20) { |
4144 | 63.7M | if ((normalize) && (*inSpace)) { |
4145 | | /* Skip char */ |
4146 | 1.14M | if (chunkSize > 0) { |
4147 | 229k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4148 | 229k | chunkSize = 0; |
4149 | 229k | } |
4150 | 62.6M | } else if (c < 0x20) { |
4151 | 32.0M | if (chunkSize > 0) { |
4152 | 3.38M | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4153 | 3.38M | chunkSize = 0; |
4154 | 3.38M | } |
4155 | | |
4156 | 32.0M | xmlSBufAddCString(buf, " ", 1); |
4157 | 32.0M | } else { |
4158 | 30.5M | chunkSize += 1; |
4159 | 30.5M | } |
4160 | | |
4161 | 63.7M | *inSpace = 1; |
4162 | 4.16G | } else { |
4163 | 4.16G | chunkSize += 1; |
4164 | 4.16G | *inSpace = 0; |
4165 | 4.16G | } |
4166 | | |
4167 | 4.23G | str += 1; |
4168 | 4.23G | } else if (str[1] == '#') { |
4169 | 3.23M | int val; |
4170 | | |
4171 | 3.23M | if (chunkSize > 0) { |
4172 | 3.16M | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4173 | 3.16M | chunkSize = 0; |
4174 | 3.16M | } |
4175 | | |
4176 | 3.23M | val = xmlParseStringCharRef(ctxt, &str); |
4177 | 3.23M | if (val == 0) { |
4178 | 48 | if (pent != NULL) |
4179 | 48 | pent->content[0] = 0; |
4180 | 48 | break; |
4181 | 48 | } |
4182 | | |
4183 | 3.23M | if (val == ' ') { |
4184 | 61.3k | if ((!normalize) || (!*inSpace)) |
4185 | 58.8k | xmlSBufAddCString(buf, " ", 1); |
4186 | 61.3k | *inSpace = 1; |
4187 | 3.16M | } else { |
4188 | 3.16M | xmlSBufAddChar(buf, val); |
4189 | 3.16M | *inSpace = 0; |
4190 | 3.16M | } |
4191 | 8.50M | } else { |
4192 | 8.50M | xmlChar *name; |
4193 | 8.50M | xmlEntityPtr ent; |
4194 | | |
4195 | 8.50M | if (chunkSize > 0) { |
4196 | 3.53M | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4197 | 3.53M | chunkSize = 0; |
4198 | 3.53M | } |
4199 | | |
4200 | 8.50M | name = xmlParseStringEntityRef(ctxt, &str); |
4201 | 8.50M | if (name == NULL) { |
4202 | 210 | if (pent != NULL) |
4203 | 198 | pent->content[0] = 0; |
4204 | 210 | break; |
4205 | 210 | } |
4206 | | |
4207 | 8.50M | ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1); |
4208 | 8.50M | xmlFree(name); |
4209 | | |
4210 | 8.50M | if ((ent != NULL) && |
4211 | 8.50M | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
4212 | 3.25M | if (ent->content == NULL) { |
4213 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
4214 | 0 | "predefined entity has no content\n"); |
4215 | 0 | break; |
4216 | 0 | } |
4217 | | |
4218 | 3.25M | xmlSBufAddString(buf, ent->content, ent->length); |
4219 | | |
4220 | 3.25M | *inSpace = 0; |
4221 | 5.25M | } else if ((ent != NULL) && (ent->content != NULL)) { |
4222 | 4.86M | if (pent != NULL) |
4223 | 4.76M | pent->flags |= XML_ENT_EXPANDING; |
4224 | 4.86M | xmlExpandEntityInAttValue(ctxt, buf, ent->content, ent, |
4225 | 4.86M | normalize, inSpace, depth, check); |
4226 | 4.86M | if (pent != NULL) |
4227 | 4.76M | pent->flags &= ~XML_ENT_EXPANDING; |
4228 | 4.86M | } |
4229 | 8.50M | } |
4230 | 4.25G | } |
4231 | | |
4232 | 6.69M | if (chunkSize > 0) |
4233 | 4.05M | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4234 | 6.69M | } |
4235 | | |
4236 | | /** |
4237 | | * xmlExpandEntitiesInAttValue: |
4238 | | * @ctxt: parser context |
4239 | | * @str: entity or attribute value |
4240 | | * @normalize: whether to collapse whitespace |
4241 | | * |
4242 | | * Expand general entity references in an entity or attribute value. |
4243 | | * Perform attribute value normalization. |
4244 | | * |
4245 | | * Returns the expanded attribtue value. |
4246 | | */ |
4247 | | xmlChar * |
4248 | | xmlExpandEntitiesInAttValue(xmlParserCtxtPtr ctxt, const xmlChar *str, |
4249 | 361k | int normalize) { |
4250 | 361k | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4251 | 129k | XML_MAX_HUGE_LENGTH : |
4252 | 361k | XML_MAX_TEXT_LENGTH; |
4253 | 361k | xmlSBuf buf; |
4254 | 361k | int inSpace = 1; |
4255 | | |
4256 | 361k | xmlSBufInit(&buf, maxLength); |
4257 | | |
4258 | 361k | xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace, |
4259 | 361k | ctxt->inputNr, /* check */ 0); |
4260 | | |
4261 | 361k | if ((normalize) && (inSpace) && (buf.size > 0)) |
4262 | 0 | buf.size--; |
4263 | | |
4264 | 361k | return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long")); |
4265 | 361k | } |
4266 | | |
4267 | | /** |
4268 | | * xmlParseAttValueInternal: |
4269 | | * @ctxt: an XML parser context |
4270 | | * @len: attribute len result |
4271 | | * @alloc: whether the attribute was reallocated as a new string |
4272 | | * @normalize: if 1 then further non-CDATA normalization must be done |
4273 | | * |
4274 | | * parse a value for an attribute. |
4275 | | * NOTE: if no normalization is needed, the routine will return pointers |
4276 | | * directly from the data buffer. |
4277 | | * |
4278 | | * 3.3.3 Attribute-Value Normalization: |
4279 | | * Before the value of an attribute is passed to the application or |
4280 | | * checked for validity, the XML processor must normalize it as follows: |
4281 | | * - a character reference is processed by appending the referenced |
4282 | | * character to the attribute value |
4283 | | * - an entity reference is processed by recursively processing the |
4284 | | * replacement text of the entity |
4285 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
4286 | | * appending #x20 to the normalized value, except that only a single |
4287 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
4288 | | * parsed entity or the literal entity value of an internal parsed entity |
4289 | | * - other characters are processed by appending them to the normalized value |
4290 | | * If the declared value is not CDATA, then the XML processor must further |
4291 | | * process the normalized attribute value by discarding any leading and |
4292 | | * trailing space (#x20) characters, and by replacing sequences of space |
4293 | | * (#x20) characters by a single space (#x20) character. |
4294 | | * All attributes for which no declaration has been read should be treated |
4295 | | * by a non-validating parser as if declared CDATA. |
4296 | | * |
4297 | | * Returns the AttValue parsed or NULL. The value has to be freed by the |
4298 | | * caller if it was copied, this can be detected by val[*len] == 0. |
4299 | | */ |
4300 | | static xmlChar * |
4301 | | xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *attlen, int *alloc, |
4302 | 4.91M | int normalize, int isNamespace) { |
4303 | 4.91M | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4304 | 2.24M | XML_MAX_HUGE_LENGTH : |
4305 | 4.91M | XML_MAX_TEXT_LENGTH; |
4306 | 4.91M | xmlSBuf buf; |
4307 | 4.91M | xmlChar *ret; |
4308 | 4.91M | int c, l, quote, flags, chunkSize; |
4309 | 4.91M | int inSpace = 1; |
4310 | 4.91M | int replaceEntities; |
4311 | | |
4312 | | /* Always expand namespace URIs */ |
4313 | 4.91M | replaceEntities = (ctxt->replaceEntities) || (isNamespace); |
4314 | | |
4315 | 4.91M | xmlSBufInit(&buf, maxLength); |
4316 | | |
4317 | 4.91M | GROW; |
4318 | | |
4319 | 4.91M | quote = CUR; |
4320 | 4.91M | if ((quote != '"') && (quote != '\'')) { |
4321 | 142k | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
4322 | 142k | return(NULL); |
4323 | 142k | } |
4324 | 4.77M | NEXTL(1); |
4325 | | |
4326 | 4.77M | if (ctxt->inSubset == 0) |
4327 | 4.41M | flags = XML_ENT_CHECKED | XML_ENT_VALIDATED; |
4328 | 363k | else |
4329 | 363k | flags = XML_ENT_VALIDATED; |
4330 | | |
4331 | 4.77M | inSpace = 1; |
4332 | 4.77M | chunkSize = 0; |
4333 | | |
4334 | 790M | while (1) { |
4335 | 790M | if (PARSER_STOPPED(ctxt)) |
4336 | 3.12k | goto error; |
4337 | | |
4338 | 790M | if (CUR_PTR >= ctxt->input->end) { |
4339 | 41.1k | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4340 | 41.1k | "AttValue: ' expected\n"); |
4341 | 41.1k | goto error; |
4342 | 41.1k | } |
4343 | | |
4344 | | /* |
4345 | | * TODO: Check growth threshold |
4346 | | */ |
4347 | 790M | if (ctxt->input->end - CUR_PTR < 10) |
4348 | 647k | GROW; |
4349 | | |
4350 | 790M | c = CUR; |
4351 | | |
4352 | 790M | if (c >= 0x80) { |
4353 | 345M | l = xmlUTF8MultibyteLen(ctxt, CUR_PTR, |
4354 | 345M | "invalid character in attribute value\n"); |
4355 | 345M | if (l == 0) { |
4356 | 251M | if (chunkSize > 0) { |
4357 | 1.91M | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4358 | 1.91M | chunkSize = 0; |
4359 | 1.91M | } |
4360 | 251M | xmlSBufAddReplChar(&buf); |
4361 | 251M | NEXTL(1); |
4362 | 251M | } else { |
4363 | 93.6M | chunkSize += l; |
4364 | 93.6M | NEXTL(l); |
4365 | 93.6M | } |
4366 | | |
4367 | 345M | inSpace = 0; |
4368 | 445M | } else if (c != '&') { |
4369 | 440M | if (c > 0x20) { |
4370 | 147M | if (c == quote) |
4371 | 4.68M | break; |
4372 | | |
4373 | 142M | if (c == '<') |
4374 | 4.03M | xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL); |
4375 | | |
4376 | 142M | chunkSize += 1; |
4377 | 142M | inSpace = 0; |
4378 | 293M | } else if (!IS_BYTE_CHAR(c)) { |
4379 | 247M | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
4380 | 247M | "invalid character in attribute value\n"); |
4381 | 247M | if (chunkSize > 0) { |
4382 | 730k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4383 | 730k | chunkSize = 0; |
4384 | 730k | } |
4385 | 247M | xmlSBufAddReplChar(&buf); |
4386 | 247M | inSpace = 0; |
4387 | 247M | } else { |
4388 | | /* Whitespace */ |
4389 | 45.0M | if ((normalize) && (inSpace)) { |
4390 | | /* Skip char */ |
4391 | 1.21M | if (chunkSize > 0) { |
4392 | 45.2k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4393 | 45.2k | chunkSize = 0; |
4394 | 45.2k | } |
4395 | 43.8M | } else if (c < 0x20) { |
4396 | | /* Convert to space */ |
4397 | 36.6M | if (chunkSize > 0) { |
4398 | 1.45M | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4399 | 1.45M | chunkSize = 0; |
4400 | 1.45M | } |
4401 | | |
4402 | 36.6M | xmlSBufAddCString(&buf, " ", 1); |
4403 | 36.6M | } else { |
4404 | 7.16M | chunkSize += 1; |
4405 | 7.16M | } |
4406 | | |
4407 | 45.0M | inSpace = 1; |
4408 | | |
4409 | 45.0M | if ((c == 0xD) && (NXT(1) == 0xA)) |
4410 | 67.7k | CUR_PTR++; |
4411 | 45.0M | } |
4412 | | |
4413 | 435M | NEXTL(1); |
4414 | 435M | } else if (NXT(1) == '#') { |
4415 | 714k | int val; |
4416 | | |
4417 | 714k | if (chunkSize > 0) { |
4418 | 337k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4419 | 337k | chunkSize = 0; |
4420 | 337k | } |
4421 | | |
4422 | 714k | val = xmlParseCharRef(ctxt); |
4423 | 714k | if (val == 0) |
4424 | 45.0k | goto error; |
4425 | | |
4426 | 669k | if ((val == '&') && (!replaceEntities)) { |
4427 | | /* |
4428 | | * The reparsing will be done in xmlNodeParseContent() |
4429 | | * called from SAX2.c |
4430 | | */ |
4431 | 34.8k | xmlSBufAddCString(&buf, "&", 5); |
4432 | 34.8k | inSpace = 0; |
4433 | 634k | } else if (val == ' ') { |
4434 | 205k | if ((!normalize) || (!inSpace)) |
4435 | 201k | xmlSBufAddCString(&buf, " ", 1); |
4436 | 205k | inSpace = 1; |
4437 | 429k | } else { |
4438 | 429k | xmlSBufAddChar(&buf, val); |
4439 | 429k | inSpace = 0; |
4440 | 429k | } |
4441 | 4.51M | } else { |
4442 | 4.51M | const xmlChar *name; |
4443 | 4.51M | xmlEntityPtr ent; |
4444 | | |
4445 | 4.51M | if (chunkSize > 0) { |
4446 | 2.10M | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4447 | 2.10M | chunkSize = 0; |
4448 | 2.10M | } |
4449 | | |
4450 | 4.51M | name = xmlParseEntityRefInternal(ctxt); |
4451 | 4.51M | if (name == NULL) { |
4452 | | /* |
4453 | | * Probably a literal '&' which wasn't escaped. |
4454 | | * TODO: Handle gracefully in recovery mode. |
4455 | | */ |
4456 | 508k | continue; |
4457 | 508k | } |
4458 | | |
4459 | 4.00M | ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1); |
4460 | 4.00M | if (ent == NULL) |
4461 | 547k | continue; |
4462 | | |
4463 | 3.45M | if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) { |
4464 | 424k | if ((ent->content[0] == '&') && (!replaceEntities)) |
4465 | 63.4k | xmlSBufAddCString(&buf, "&", 5); |
4466 | 361k | else |
4467 | 361k | xmlSBufAddString(&buf, ent->content, ent->length); |
4468 | 424k | inSpace = 0; |
4469 | 3.02M | } else if (replaceEntities) { |
4470 | 1.54M | xmlExpandEntityInAttValue(ctxt, &buf, ent->content, ent, |
4471 | 1.54M | normalize, &inSpace, ctxt->inputNr, |
4472 | 1.54M | /* check */ 1); |
4473 | 1.54M | } else { |
4474 | 1.48M | if ((ent->flags & flags) != flags) |
4475 | 7.00k | xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr); |
4476 | | |
4477 | 1.48M | if (xmlParserEntityCheck(ctxt, ent->expandedSize)) { |
4478 | 397 | ent->content[0] = 0; |
4479 | 397 | goto error; |
4480 | 397 | } |
4481 | | |
4482 | | /* |
4483 | | * Just output the reference |
4484 | | */ |
4485 | 1.48M | xmlSBufAddCString(&buf, "&", 1); |
4486 | 1.48M | xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name)); |
4487 | 1.48M | xmlSBufAddCString(&buf, ";", 1); |
4488 | | |
4489 | 1.48M | inSpace = 0; |
4490 | 1.48M | } |
4491 | 3.45M | } |
4492 | 790M | } |
4493 | | |
4494 | 4.68M | if ((buf.mem == NULL) && (alloc != NULL)) { |
4495 | 2.50M | ret = (xmlChar *) CUR_PTR - chunkSize; |
4496 | | |
4497 | 2.50M | if (attlen != NULL) |
4498 | 2.50M | *attlen = chunkSize; |
4499 | 2.50M | if ((normalize) && (inSpace) && (chunkSize > 0)) |
4500 | 4.07k | *attlen -= 1; |
4501 | 2.50M | *alloc = 0; |
4502 | | |
4503 | | /* Report potential error */ |
4504 | 2.50M | xmlSBufCleanup(&buf, ctxt, "AttValue length too long"); |
4505 | 2.50M | } else { |
4506 | 2.18M | if (chunkSize > 0) |
4507 | 1.54M | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4508 | | |
4509 | 2.18M | if ((normalize) && (inSpace) && (buf.size > 0)) |
4510 | 10.8k | buf.size--; |
4511 | | |
4512 | 2.18M | ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long"); |
4513 | | |
4514 | 2.18M | if (ret != NULL) { |
4515 | 2.18M | if (attlen != NULL) |
4516 | 654k | *attlen = buf.size; |
4517 | 2.18M | if (alloc != NULL) |
4518 | 654k | *alloc = 1; |
4519 | 2.18M | } |
4520 | 2.18M | } |
4521 | | |
4522 | 4.68M | NEXTL(1); |
4523 | | |
4524 | 4.68M | return(ret); |
4525 | | |
4526 | 89.7k | error: |
4527 | 89.7k | xmlSBufCleanup(&buf, ctxt, "AttValue length too long"); |
4528 | 89.7k | return(NULL); |
4529 | 4.77M | } |
4530 | | |
4531 | | /** |
4532 | | * xmlParseAttValue: |
4533 | | * @ctxt: an XML parser context |
4534 | | * |
4535 | | * DEPRECATED: Internal function, don't use. |
4536 | | * |
4537 | | * parse a value for an attribute |
4538 | | * Note: the parser won't do substitution of entities here, this |
4539 | | * will be handled later in xmlStringGetNodeList |
4540 | | * |
4541 | | * [10] AttValue ::= '"' ([^<&"] | Reference)* '"' | |
4542 | | * "'" ([^<&'] | Reference)* "'" |
4543 | | * |
4544 | | * 3.3.3 Attribute-Value Normalization: |
4545 | | * Before the value of an attribute is passed to the application or |
4546 | | * checked for validity, the XML processor must normalize it as follows: |
4547 | | * - a character reference is processed by appending the referenced |
4548 | | * character to the attribute value |
4549 | | * - an entity reference is processed by recursively processing the |
4550 | | * replacement text of the entity |
4551 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
4552 | | * appending #x20 to the normalized value, except that only a single |
4553 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
4554 | | * parsed entity or the literal entity value of an internal parsed entity |
4555 | | * - other characters are processed by appending them to the normalized value |
4556 | | * If the declared value is not CDATA, then the XML processor must further |
4557 | | * process the normalized attribute value by discarding any leading and |
4558 | | * trailing space (#x20) characters, and by replacing sequences of space |
4559 | | * (#x20) characters by a single space (#x20) character. |
4560 | | * All attributes for which no declaration has been read should be treated |
4561 | | * by a non-validating parser as if declared CDATA. |
4562 | | * |
4563 | | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
4564 | | */ |
4565 | | |
4566 | | |
4567 | | xmlChar * |
4568 | 1.64M | xmlParseAttValue(xmlParserCtxtPtr ctxt) { |
4569 | 1.64M | if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL); |
4570 | 1.64M | return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0)); |
4571 | 1.64M | } |
4572 | | |
4573 | | /** |
4574 | | * xmlParseSystemLiteral: |
4575 | | * @ctxt: an XML parser context |
4576 | | * |
4577 | | * DEPRECATED: Internal function, don't use. |
4578 | | * |
4579 | | * parse an XML Literal |
4580 | | * |
4581 | | * [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") |
4582 | | * |
4583 | | * Returns the SystemLiteral parsed or NULL |
4584 | | */ |
4585 | | |
4586 | | xmlChar * |
4587 | 259k | xmlParseSystemLiteral(xmlParserCtxtPtr ctxt) { |
4588 | 259k | xmlChar *buf = NULL; |
4589 | 259k | int len = 0; |
4590 | 259k | int size = XML_PARSER_BUFFER_SIZE; |
4591 | 259k | int cur, l; |
4592 | 259k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4593 | 66.7k | XML_MAX_TEXT_LENGTH : |
4594 | 259k | XML_MAX_NAME_LENGTH; |
4595 | 259k | xmlChar stop; |
4596 | | |
4597 | 259k | if (RAW == '"') { |
4598 | 219k | NEXT; |
4599 | 219k | stop = '"'; |
4600 | 219k | } else if (RAW == '\'') { |
4601 | 29.3k | NEXT; |
4602 | 29.3k | stop = '\''; |
4603 | 29.3k | } else { |
4604 | 9.94k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4605 | 9.94k | return(NULL); |
4606 | 9.94k | } |
4607 | | |
4608 | 249k | buf = xmlMalloc(size); |
4609 | 249k | if (buf == NULL) { |
4610 | 103 | xmlErrMemory(ctxt); |
4611 | 103 | return(NULL); |
4612 | 103 | } |
4613 | 249k | cur = xmlCurrentCharRecover(ctxt, &l); |
4614 | 15.0M | while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */ |
4615 | 14.8M | if (len + 5 >= size) { |
4616 | 14.6k | xmlChar *tmp; |
4617 | | |
4618 | 14.6k | size *= 2; |
4619 | 14.6k | tmp = (xmlChar *) xmlRealloc(buf, size); |
4620 | 14.6k | if (tmp == NULL) { |
4621 | 19 | xmlFree(buf); |
4622 | 19 | xmlErrMemory(ctxt); |
4623 | 19 | return(NULL); |
4624 | 19 | } |
4625 | 14.6k | buf = tmp; |
4626 | 14.6k | } |
4627 | 14.8M | COPY_BUF(buf, len, cur); |
4628 | 14.8M | if (len > maxLength) { |
4629 | 35 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral"); |
4630 | 35 | xmlFree(buf); |
4631 | 35 | return(NULL); |
4632 | 35 | } |
4633 | 14.8M | NEXTL(l); |
4634 | 14.8M | cur = xmlCurrentCharRecover(ctxt, &l); |
4635 | 14.8M | } |
4636 | 249k | buf[len] = 0; |
4637 | 249k | if (!IS_CHAR(cur)) { |
4638 | 8.04k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4639 | 240k | } else { |
4640 | 240k | NEXT; |
4641 | 240k | } |
4642 | 249k | return(buf); |
4643 | 249k | } |
4644 | | |
4645 | | /** |
4646 | | * xmlParsePubidLiteral: |
4647 | | * @ctxt: an XML parser context |
4648 | | * |
4649 | | * DEPRECATED: Internal function, don't use. |
4650 | | * |
4651 | | * parse an XML public literal |
4652 | | * |
4653 | | * [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'" |
4654 | | * |
4655 | | * Returns the PubidLiteral parsed or NULL. |
4656 | | */ |
4657 | | |
4658 | | xmlChar * |
4659 | 143k | xmlParsePubidLiteral(xmlParserCtxtPtr ctxt) { |
4660 | 143k | xmlChar *buf = NULL; |
4661 | 143k | int len = 0; |
4662 | 143k | int size = XML_PARSER_BUFFER_SIZE; |
4663 | 143k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4664 | 34.4k | XML_MAX_TEXT_LENGTH : |
4665 | 143k | XML_MAX_NAME_LENGTH; |
4666 | 143k | xmlChar cur; |
4667 | 143k | xmlChar stop; |
4668 | | |
4669 | 143k | if (RAW == '"') { |
4670 | 105k | NEXT; |
4671 | 105k | stop = '"'; |
4672 | 105k | } else if (RAW == '\'') { |
4673 | 35.5k | NEXT; |
4674 | 35.5k | stop = '\''; |
4675 | 35.5k | } else { |
4676 | 2.94k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4677 | 2.94k | return(NULL); |
4678 | 2.94k | } |
4679 | 140k | buf = xmlMalloc(size); |
4680 | 140k | if (buf == NULL) { |
4681 | 62 | xmlErrMemory(ctxt); |
4682 | 62 | return(NULL); |
4683 | 62 | } |
4684 | 140k | cur = CUR; |
4685 | 1.14M | while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) && |
4686 | 1.14M | (PARSER_STOPPED(ctxt) == 0)) { /* checked */ |
4687 | 1.00M | if (len + 1 >= size) { |
4688 | 1.57k | xmlChar *tmp; |
4689 | | |
4690 | 1.57k | size *= 2; |
4691 | 1.57k | tmp = (xmlChar *) xmlRealloc(buf, size); |
4692 | 1.57k | if (tmp == NULL) { |
4693 | 13 | xmlErrMemory(ctxt); |
4694 | 13 | xmlFree(buf); |
4695 | 13 | return(NULL); |
4696 | 13 | } |
4697 | 1.56k | buf = tmp; |
4698 | 1.56k | } |
4699 | 1.00M | buf[len++] = cur; |
4700 | 1.00M | if (len > maxLength) { |
4701 | 3 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID"); |
4702 | 3 | xmlFree(buf); |
4703 | 3 | return(NULL); |
4704 | 3 | } |
4705 | 1.00M | NEXT; |
4706 | 1.00M | cur = CUR; |
4707 | 1.00M | } |
4708 | 140k | buf[len] = 0; |
4709 | 140k | if (cur != stop) { |
4710 | 18.5k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4711 | 122k | } else { |
4712 | 122k | NEXTL(1); |
4713 | 122k | } |
4714 | 140k | return(buf); |
4715 | 140k | } |
4716 | | |
4717 | | static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial); |
4718 | | |
4719 | | /* |
4720 | | * used for the test in the inner loop of the char data testing |
4721 | | */ |
4722 | | static const unsigned char test_char_data[256] = { |
4723 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4724 | | 0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */ |
4725 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4726 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4727 | | 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */ |
4728 | | 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F, |
4729 | | 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, |
4730 | | 0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */ |
4731 | | 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, |
4732 | | 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F, |
4733 | | 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, |
4734 | | 0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */ |
4735 | | 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, |
4736 | | 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, |
4737 | | 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, |
4738 | | 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F, |
4739 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */ |
4740 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4741 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4742 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4743 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4744 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4745 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4746 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4747 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4748 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4749 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4750 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4751 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4752 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4753 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4754 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 |
4755 | | }; |
4756 | | |
4757 | | /** |
4758 | | * xmlParseCharDataInternal: |
4759 | | * @ctxt: an XML parser context |
4760 | | * @partial: buffer may contain partial UTF-8 sequences |
4761 | | * |
4762 | | * Parse character data. Always makes progress if the first char isn't |
4763 | | * '<' or '&'. |
4764 | | * |
4765 | | * The right angle bracket (>) may be represented using the string ">", |
4766 | | * and must, for compatibility, be escaped using ">" or a character |
4767 | | * reference when it appears in the string "]]>" in content, when that |
4768 | | * string is not marking the end of a CDATA section. |
4769 | | * |
4770 | | * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) |
4771 | | */ |
4772 | | static void |
4773 | 53.7M | xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) { |
4774 | 53.7M | const xmlChar *in; |
4775 | 53.7M | int nbchar = 0; |
4776 | 53.7M | int line = ctxt->input->line; |
4777 | 53.7M | int col = ctxt->input->col; |
4778 | 53.7M | int ccol; |
4779 | | |
4780 | 53.7M | GROW; |
4781 | | /* |
4782 | | * Accelerated common case where input don't need to be |
4783 | | * modified before passing it to the handler. |
4784 | | */ |
4785 | 53.7M | in = ctxt->input->cur; |
4786 | 54.0M | do { |
4787 | 55.6M | get_more_space: |
4788 | 60.2M | while (*in == 0x20) { in++; ctxt->input->col++; } |
4789 | 55.6M | if (*in == 0xA) { |
4790 | 25.0M | do { |
4791 | 25.0M | ctxt->input->line++; ctxt->input->col = 1; |
4792 | 25.0M | in++; |
4793 | 25.0M | } while (*in == 0xA); |
4794 | 1.62M | goto get_more_space; |
4795 | 1.62M | } |
4796 | 54.0M | if (*in == '<') { |
4797 | 1.35M | nbchar = in - ctxt->input->cur; |
4798 | 1.35M | if (nbchar > 0) { |
4799 | 1.35M | const xmlChar *tmp = ctxt->input->cur; |
4800 | 1.35M | ctxt->input->cur = in; |
4801 | | |
4802 | 1.35M | if ((ctxt->sax != NULL) && |
4803 | 1.35M | (ctxt->disableSAX == 0) && |
4804 | 1.35M | (ctxt->sax->ignorableWhitespace != |
4805 | 1.28M | ctxt->sax->characters)) { |
4806 | 635k | if (areBlanks(ctxt, tmp, nbchar, 1)) { |
4807 | 413k | if (ctxt->sax->ignorableWhitespace != NULL) |
4808 | 413k | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4809 | 413k | tmp, nbchar); |
4810 | 413k | } else { |
4811 | 222k | if (ctxt->sax->characters != NULL) |
4812 | 216k | ctxt->sax->characters(ctxt->userData, |
4813 | 216k | tmp, nbchar); |
4814 | 222k | if (*ctxt->space == -1) |
4815 | 34.4k | *ctxt->space = -2; |
4816 | 222k | } |
4817 | 724k | } else if ((ctxt->sax != NULL) && |
4818 | 724k | (ctxt->disableSAX == 0) && |
4819 | 724k | (ctxt->sax->characters != NULL)) { |
4820 | 644k | ctxt->sax->characters(ctxt->userData, |
4821 | 644k | tmp, nbchar); |
4822 | 644k | } |
4823 | 1.35M | } |
4824 | 1.35M | return; |
4825 | 1.35M | } |
4826 | | |
4827 | 54.6M | get_more: |
4828 | 54.6M | ccol = ctxt->input->col; |
4829 | 175M | while (test_char_data[*in]) { |
4830 | 120M | in++; |
4831 | 120M | ccol++; |
4832 | 120M | } |
4833 | 54.6M | ctxt->input->col = ccol; |
4834 | 54.6M | if (*in == 0xA) { |
4835 | 16.5M | do { |
4836 | 16.5M | ctxt->input->line++; ctxt->input->col = 1; |
4837 | 16.5M | in++; |
4838 | 16.5M | } while (*in == 0xA); |
4839 | 1.48M | goto get_more; |
4840 | 1.48M | } |
4841 | 53.1M | if (*in == ']') { |
4842 | 543k | if ((in[1] == ']') && (in[2] == '>')) { |
4843 | 79.8k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4844 | 79.8k | ctxt->input->cur = in + 1; |
4845 | 79.8k | return; |
4846 | 79.8k | } |
4847 | 463k | in++; |
4848 | 463k | ctxt->input->col++; |
4849 | 463k | goto get_more; |
4850 | 543k | } |
4851 | 52.5M | nbchar = in - ctxt->input->cur; |
4852 | 52.5M | if (nbchar > 0) { |
4853 | 9.25M | if ((ctxt->sax != NULL) && |
4854 | 9.25M | (ctxt->disableSAX == 0) && |
4855 | 9.25M | (ctxt->sax->ignorableWhitespace != |
4856 | 6.85M | ctxt->sax->characters) && |
4857 | 9.25M | (IS_BLANK_CH(*ctxt->input->cur))) { |
4858 | 484k | const xmlChar *tmp = ctxt->input->cur; |
4859 | 484k | ctxt->input->cur = in; |
4860 | | |
4861 | 484k | if (areBlanks(ctxt, tmp, nbchar, 0)) { |
4862 | 30.5k | if (ctxt->sax->ignorableWhitespace != NULL) |
4863 | 30.5k | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4864 | 30.5k | tmp, nbchar); |
4865 | 454k | } else { |
4866 | 454k | if (ctxt->sax->characters != NULL) |
4867 | 450k | ctxt->sax->characters(ctxt->userData, |
4868 | 450k | tmp, nbchar); |
4869 | 454k | if (*ctxt->space == -1) |
4870 | 176k | *ctxt->space = -2; |
4871 | 454k | } |
4872 | 484k | line = ctxt->input->line; |
4873 | 484k | col = ctxt->input->col; |
4874 | 8.76M | } else if ((ctxt->sax != NULL) && |
4875 | 8.76M | (ctxt->disableSAX == 0)) { |
4876 | 6.37M | if (ctxt->sax->characters != NULL) |
4877 | 6.34M | ctxt->sax->characters(ctxt->userData, |
4878 | 6.34M | ctxt->input->cur, nbchar); |
4879 | 6.37M | line = ctxt->input->line; |
4880 | 6.37M | col = ctxt->input->col; |
4881 | 6.37M | } |
4882 | 9.25M | } |
4883 | 52.5M | ctxt->input->cur = in; |
4884 | 52.5M | if (*in == 0xD) { |
4885 | 366k | in++; |
4886 | 366k | if (*in == 0xA) { |
4887 | 263k | ctxt->input->cur = in; |
4888 | 263k | in++; |
4889 | 263k | ctxt->input->line++; ctxt->input->col = 1; |
4890 | 263k | continue; /* while */ |
4891 | 263k | } |
4892 | 102k | in--; |
4893 | 102k | } |
4894 | 52.3M | if (*in == '<') { |
4895 | 3.78M | return; |
4896 | 3.78M | } |
4897 | 48.5M | if (*in == '&') { |
4898 | 1.02M | return; |
4899 | 1.02M | } |
4900 | 47.5M | SHRINK; |
4901 | 47.5M | GROW; |
4902 | 47.5M | in = ctxt->input->cur; |
4903 | 47.7M | } while (((*in >= 0x20) && (*in <= 0x7F)) || |
4904 | 47.7M | (*in == 0x09) || (*in == 0x0a)); |
4905 | 47.5M | ctxt->input->line = line; |
4906 | 47.5M | ctxt->input->col = col; |
4907 | 47.5M | xmlParseCharDataComplex(ctxt, partial); |
4908 | 47.5M | } |
4909 | | |
4910 | | /** |
4911 | | * xmlParseCharDataComplex: |
4912 | | * @ctxt: an XML parser context |
4913 | | * @cdata: int indicating whether we are within a CDATA section |
4914 | | * |
4915 | | * Always makes progress if the first char isn't '<' or '&'. |
4916 | | * |
4917 | | * parse a CharData section.this is the fallback function |
4918 | | * of xmlParseCharData() when the parsing requires handling |
4919 | | * of non-ASCII characters. |
4920 | | */ |
4921 | | static void |
4922 | 47.5M | xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) { |
4923 | 47.5M | xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5]; |
4924 | 47.5M | int nbchar = 0; |
4925 | 47.5M | int cur, l; |
4926 | | |
4927 | 47.5M | cur = xmlCurrentCharRecover(ctxt, &l); |
4928 | 316M | while ((cur != '<') && /* checked */ |
4929 | 316M | (cur != '&') && |
4930 | 316M | (IS_CHAR(cur))) { |
4931 | 268M | if ((cur == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
4932 | 37.9k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4933 | 37.9k | } |
4934 | 268M | COPY_BUF(buf, nbchar, cur); |
4935 | | /* move current position before possible calling of ctxt->sax->characters */ |
4936 | 268M | NEXTL(l); |
4937 | 268M | if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) { |
4938 | 1.62M | buf[nbchar] = 0; |
4939 | | |
4940 | | /* |
4941 | | * OK the segment is to be consumed as chars. |
4942 | | */ |
4943 | 1.62M | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4944 | 1.59M | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4945 | 738 | if (ctxt->sax->ignorableWhitespace != NULL) |
4946 | 738 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4947 | 738 | buf, nbchar); |
4948 | 1.59M | } else { |
4949 | 1.59M | if (ctxt->sax->characters != NULL) |
4950 | 1.59M | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4951 | 1.59M | if ((ctxt->sax->characters != |
4952 | 1.59M | ctxt->sax->ignorableWhitespace) && |
4953 | 1.59M | (*ctxt->space == -1)) |
4954 | 7.52k | *ctxt->space = -2; |
4955 | 1.59M | } |
4956 | 1.59M | } |
4957 | 1.62M | nbchar = 0; |
4958 | 1.62M | SHRINK; |
4959 | 1.62M | } |
4960 | 268M | cur = xmlCurrentCharRecover(ctxt, &l); |
4961 | 268M | } |
4962 | 47.5M | if (nbchar != 0) { |
4963 | 6.58M | buf[nbchar] = 0; |
4964 | | /* |
4965 | | * OK the segment is to be consumed as chars. |
4966 | | */ |
4967 | 6.58M | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4968 | 3.93M | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4969 | 4.00k | if (ctxt->sax->ignorableWhitespace != NULL) |
4970 | 4.00k | ctxt->sax->ignorableWhitespace(ctxt->userData, buf, nbchar); |
4971 | 3.93M | } else { |
4972 | 3.93M | if (ctxt->sax->characters != NULL) |
4973 | 3.91M | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4974 | 3.93M | if ((ctxt->sax->characters != ctxt->sax->ignorableWhitespace) && |
4975 | 3.93M | (*ctxt->space == -1)) |
4976 | 357k | *ctxt->space = -2; |
4977 | 3.93M | } |
4978 | 3.93M | } |
4979 | 6.58M | } |
4980 | | /* |
4981 | | * cur == 0 can mean |
4982 | | * |
4983 | | * - End of buffer. |
4984 | | * - An actual 0 character. |
4985 | | * - An incomplete UTF-8 sequence. This is allowed if partial is set. |
4986 | | */ |
4987 | 47.5M | if (ctxt->input->cur < ctxt->input->end) { |
4988 | 47.4M | if ((cur == 0) && (CUR != 0)) { |
4989 | 11.3k | if (partial == 0) { |
4990 | 9.63k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4991 | 9.63k | "Incomplete UTF-8 sequence starting with %02X\n", CUR); |
4992 | 9.63k | NEXTL(1); |
4993 | 9.63k | } |
4994 | 47.4M | } else if ((cur != '<') && (cur != '&')) { |
4995 | | /* Generate the error and skip the offending character */ |
4996 | 44.5M | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4997 | 44.5M | "PCDATA invalid Char value %d\n", cur); |
4998 | 44.5M | NEXTL(l); |
4999 | 44.5M | } |
5000 | 47.4M | } |
5001 | 47.5M | } |
5002 | | |
5003 | | /** |
5004 | | * xmlParseCharData: |
5005 | | * @ctxt: an XML parser context |
5006 | | * @cdata: unused |
5007 | | * |
5008 | | * DEPRECATED: Internal function, don't use. |
5009 | | */ |
5010 | | void |
5011 | 0 | xmlParseCharData(xmlParserCtxtPtr ctxt, ATTRIBUTE_UNUSED int cdata) { |
5012 | 0 | xmlParseCharDataInternal(ctxt, 0); |
5013 | 0 | } |
5014 | | |
5015 | | /** |
5016 | | * xmlParseExternalID: |
5017 | | * @ctxt: an XML parser context |
5018 | | * @publicID: a xmlChar** receiving PubidLiteral |
5019 | | * @strict: indicate whether we should restrict parsing to only |
5020 | | * production [75], see NOTE below |
5021 | | * |
5022 | | * DEPRECATED: Internal function, don't use. |
5023 | | * |
5024 | | * Parse an External ID or a Public ID |
5025 | | * |
5026 | | * NOTE: Productions [75] and [83] interact badly since [75] can generate |
5027 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
5028 | | * |
5029 | | * [75] ExternalID ::= 'SYSTEM' S SystemLiteral |
5030 | | * | 'PUBLIC' S PubidLiteral S SystemLiteral |
5031 | | * |
5032 | | * [83] PublicID ::= 'PUBLIC' S PubidLiteral |
5033 | | * |
5034 | | * Returns the function returns SystemLiteral and in the second |
5035 | | * case publicID receives PubidLiteral, is strict is off |
5036 | | * it is possible to return NULL and have publicID set. |
5037 | | */ |
5038 | | |
5039 | | xmlChar * |
5040 | 388k | xmlParseExternalID(xmlParserCtxtPtr ctxt, xmlChar **publicID, int strict) { |
5041 | 388k | xmlChar *URI = NULL; |
5042 | | |
5043 | 388k | *publicID = NULL; |
5044 | 388k | if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) { |
5045 | 137k | SKIP(6); |
5046 | 137k | if (SKIP_BLANKS == 0) { |
5047 | 2.14k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5048 | 2.14k | "Space required after 'SYSTEM'\n"); |
5049 | 2.14k | } |
5050 | 137k | URI = xmlParseSystemLiteral(ctxt); |
5051 | 137k | if (URI == NULL) { |
5052 | 1.12k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
5053 | 1.12k | } |
5054 | 251k | } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) { |
5055 | 143k | SKIP(6); |
5056 | 143k | if (SKIP_BLANKS == 0) { |
5057 | 6.95k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5058 | 6.95k | "Space required after 'PUBLIC'\n"); |
5059 | 6.95k | } |
5060 | 143k | *publicID = xmlParsePubidLiteral(ctxt); |
5061 | 143k | if (*publicID == NULL) { |
5062 | 3.01k | xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL); |
5063 | 3.01k | } |
5064 | 143k | if (strict) { |
5065 | | /* |
5066 | | * We don't handle [83] so "S SystemLiteral" is required. |
5067 | | */ |
5068 | 117k | if (SKIP_BLANKS == 0) { |
5069 | 13.8k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5070 | 13.8k | "Space required after the Public Identifier\n"); |
5071 | 13.8k | } |
5072 | 117k | } else { |
5073 | | /* |
5074 | | * We handle [83] so we return immediately, if |
5075 | | * "S SystemLiteral" is not detected. We skip blanks if no |
5076 | | * system literal was found, but this is harmless since we must |
5077 | | * be at the end of a NotationDecl. |
5078 | | */ |
5079 | 26.5k | if (SKIP_BLANKS == 0) return(NULL); |
5080 | 6.73k | if ((CUR != '\'') && (CUR != '"')) return(NULL); |
5081 | 6.73k | } |
5082 | 122k | URI = xmlParseSystemLiteral(ctxt); |
5083 | 122k | if (URI == NULL) { |
5084 | 8.98k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
5085 | 8.98k | } |
5086 | 122k | } |
5087 | 366k | return(URI); |
5088 | 388k | } |
5089 | | |
5090 | | /** |
5091 | | * xmlParseCommentComplex: |
5092 | | * @ctxt: an XML parser context |
5093 | | * @buf: the already parsed part of the buffer |
5094 | | * @len: number of bytes in the buffer |
5095 | | * @size: allocated size of the buffer |
5096 | | * |
5097 | | * Skip an XML (SGML) comment <!-- .... --> |
5098 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
5099 | | * must not occur within comments. " |
5100 | | * This is the slow routine in case the accelerator for ascii didn't work |
5101 | | * |
5102 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
5103 | | */ |
5104 | | static void |
5105 | | xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf, |
5106 | 347k | size_t len, size_t size) { |
5107 | 347k | int q, ql; |
5108 | 347k | int r, rl; |
5109 | 347k | int cur, l; |
5110 | 347k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5111 | 70.6k | XML_MAX_HUGE_LENGTH : |
5112 | 347k | XML_MAX_TEXT_LENGTH; |
5113 | | |
5114 | 347k | if (buf == NULL) { |
5115 | 53.2k | len = 0; |
5116 | 53.2k | size = XML_PARSER_BUFFER_SIZE; |
5117 | 53.2k | buf = xmlMalloc(size); |
5118 | 53.2k | if (buf == NULL) { |
5119 | 40 | xmlErrMemory(ctxt); |
5120 | 40 | return; |
5121 | 40 | } |
5122 | 53.2k | } |
5123 | 347k | q = xmlCurrentCharRecover(ctxt, &ql); |
5124 | 347k | if (q == 0) |
5125 | 69.6k | goto not_terminated; |
5126 | 278k | if (!IS_CHAR(q)) { |
5127 | 18.3k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5128 | 18.3k | "xmlParseComment: invalid xmlChar value %d\n", |
5129 | 18.3k | q); |
5130 | 18.3k | xmlFree (buf); |
5131 | 18.3k | return; |
5132 | 18.3k | } |
5133 | 259k | NEXTL(ql); |
5134 | 259k | r = xmlCurrentCharRecover(ctxt, &rl); |
5135 | 259k | if (r == 0) |
5136 | 6.06k | goto not_terminated; |
5137 | 253k | if (!IS_CHAR(r)) { |
5138 | 9.87k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5139 | 9.87k | "xmlParseComment: invalid xmlChar value %d\n", |
5140 | 9.87k | r); |
5141 | 9.87k | xmlFree (buf); |
5142 | 9.87k | return; |
5143 | 9.87k | } |
5144 | 243k | NEXTL(rl); |
5145 | 243k | cur = xmlCurrentCharRecover(ctxt, &l); |
5146 | 243k | if (cur == 0) |
5147 | 4.00k | goto not_terminated; |
5148 | 48.0M | while (IS_CHAR(cur) && /* checked */ |
5149 | 48.0M | ((cur != '>') || |
5150 | 47.9M | (r != '-') || (q != '-'))) { |
5151 | 47.8M | if ((r == '-') && (q == '-')) { |
5152 | 1.27M | xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL); |
5153 | 1.27M | } |
5154 | 47.8M | if (len + 5 >= size) { |
5155 | 175k | xmlChar *new_buf; |
5156 | 175k | size_t new_size; |
5157 | | |
5158 | 175k | new_size = size * 2; |
5159 | 175k | new_buf = (xmlChar *) xmlRealloc(buf, new_size); |
5160 | 175k | if (new_buf == NULL) { |
5161 | 50 | xmlFree (buf); |
5162 | 50 | xmlErrMemory(ctxt); |
5163 | 50 | return; |
5164 | 50 | } |
5165 | 175k | buf = new_buf; |
5166 | 175k | size = new_size; |
5167 | 175k | } |
5168 | 47.8M | COPY_BUF(buf, len, q); |
5169 | 47.8M | if (len > maxLength) { |
5170 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5171 | 0 | "Comment too big found", NULL); |
5172 | 0 | xmlFree (buf); |
5173 | 0 | return; |
5174 | 0 | } |
5175 | | |
5176 | 47.8M | q = r; |
5177 | 47.8M | ql = rl; |
5178 | 47.8M | r = cur; |
5179 | 47.8M | rl = l; |
5180 | | |
5181 | 47.8M | NEXTL(l); |
5182 | 47.8M | cur = xmlCurrentCharRecover(ctxt, &l); |
5183 | | |
5184 | 47.8M | } |
5185 | 239k | buf[len] = 0; |
5186 | 239k | if (cur == 0) { |
5187 | 58.4k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5188 | 58.4k | "Comment not terminated \n<!--%.50s\n", buf); |
5189 | 181k | } else if (!IS_CHAR(cur)) { |
5190 | 31.1k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5191 | 31.1k | "xmlParseComment: invalid xmlChar value %d\n", |
5192 | 31.1k | cur); |
5193 | 150k | } else { |
5194 | 150k | NEXT; |
5195 | 150k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5196 | 150k | (!ctxt->disableSAX)) |
5197 | 99.6k | ctxt->sax->comment(ctxt->userData, buf); |
5198 | 150k | } |
5199 | 239k | xmlFree(buf); |
5200 | 239k | return; |
5201 | 79.7k | not_terminated: |
5202 | 79.7k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5203 | 79.7k | "Comment not terminated\n", NULL); |
5204 | 79.7k | xmlFree(buf); |
5205 | 79.7k | } |
5206 | | |
5207 | | /** |
5208 | | * xmlParseComment: |
5209 | | * @ctxt: an XML parser context |
5210 | | * |
5211 | | * DEPRECATED: Internal function, don't use. |
5212 | | * |
5213 | | * Parse an XML (SGML) comment. Always consumes '<!'. |
5214 | | * |
5215 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
5216 | | * must not occur within comments. " |
5217 | | * |
5218 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
5219 | | */ |
5220 | | void |
5221 | 3.35M | xmlParseComment(xmlParserCtxtPtr ctxt) { |
5222 | 3.35M | xmlChar *buf = NULL; |
5223 | 3.35M | size_t size = XML_PARSER_BUFFER_SIZE; |
5224 | 3.35M | size_t len = 0; |
5225 | 3.35M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5226 | 246k | XML_MAX_HUGE_LENGTH : |
5227 | 3.35M | XML_MAX_TEXT_LENGTH; |
5228 | 3.35M | const xmlChar *in; |
5229 | 3.35M | size_t nbchar = 0; |
5230 | 3.35M | int ccol; |
5231 | | |
5232 | | /* |
5233 | | * Check that there is a comment right here. |
5234 | | */ |
5235 | 3.35M | if ((RAW != '<') || (NXT(1) != '!')) |
5236 | 0 | return; |
5237 | 3.35M | SKIP(2); |
5238 | 3.35M | if ((RAW != '-') || (NXT(1) != '-')) |
5239 | 270 | return; |
5240 | 3.35M | SKIP(2); |
5241 | 3.35M | GROW; |
5242 | | |
5243 | | /* |
5244 | | * Accelerated common case where input don't need to be |
5245 | | * modified before passing it to the handler. |
5246 | | */ |
5247 | 3.35M | in = ctxt->input->cur; |
5248 | 3.35M | do { |
5249 | 3.35M | if (*in == 0xA) { |
5250 | 184k | do { |
5251 | 184k | ctxt->input->line++; ctxt->input->col = 1; |
5252 | 184k | in++; |
5253 | 184k | } while (*in == 0xA); |
5254 | 37.0k | } |
5255 | 6.03M | get_more: |
5256 | 6.03M | ccol = ctxt->input->col; |
5257 | 54.6M | while (((*in > '-') && (*in <= 0x7F)) || |
5258 | 54.6M | ((*in >= 0x20) && (*in < '-')) || |
5259 | 54.6M | (*in == 0x09)) { |
5260 | 48.6M | in++; |
5261 | 48.6M | ccol++; |
5262 | 48.6M | } |
5263 | 6.03M | ctxt->input->col = ccol; |
5264 | 6.03M | if (*in == 0xA) { |
5265 | 2.81M | do { |
5266 | 2.81M | ctxt->input->line++; ctxt->input->col = 1; |
5267 | 2.81M | in++; |
5268 | 2.81M | } while (*in == 0xA); |
5269 | 287k | goto get_more; |
5270 | 287k | } |
5271 | 5.74M | nbchar = in - ctxt->input->cur; |
5272 | | /* |
5273 | | * save current set of data |
5274 | | */ |
5275 | 5.74M | if (nbchar > 0) { |
5276 | 2.78M | if (buf == NULL) { |
5277 | 628k | if ((*in == '-') && (in[1] == '-')) |
5278 | 270k | size = nbchar + 1; |
5279 | 358k | else |
5280 | 358k | size = XML_PARSER_BUFFER_SIZE + nbchar; |
5281 | 628k | buf = xmlMalloc(size); |
5282 | 628k | if (buf == NULL) { |
5283 | 103 | xmlErrMemory(ctxt); |
5284 | 103 | return; |
5285 | 103 | } |
5286 | 628k | len = 0; |
5287 | 2.15M | } else if (len + nbchar + 1 >= size) { |
5288 | 212k | xmlChar *new_buf; |
5289 | 212k | size += len + nbchar + XML_PARSER_BUFFER_SIZE; |
5290 | 212k | new_buf = (xmlChar *) xmlRealloc(buf, size); |
5291 | 212k | if (new_buf == NULL) { |
5292 | 25 | xmlFree (buf); |
5293 | 25 | xmlErrMemory(ctxt); |
5294 | 25 | return; |
5295 | 25 | } |
5296 | 212k | buf = new_buf; |
5297 | 212k | } |
5298 | 2.78M | memcpy(&buf[len], ctxt->input->cur, nbchar); |
5299 | 2.78M | len += nbchar; |
5300 | 2.78M | buf[len] = 0; |
5301 | 2.78M | } |
5302 | 5.74M | if (len > maxLength) { |
5303 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5304 | 0 | "Comment too big found", NULL); |
5305 | 0 | xmlFree (buf); |
5306 | 0 | return; |
5307 | 0 | } |
5308 | 5.74M | ctxt->input->cur = in; |
5309 | 5.74M | if (*in == 0xA) { |
5310 | 0 | in++; |
5311 | 0 | ctxt->input->line++; ctxt->input->col = 1; |
5312 | 0 | } |
5313 | 5.74M | if (*in == 0xD) { |
5314 | 300k | in++; |
5315 | 300k | if (*in == 0xA) { |
5316 | 289k | ctxt->input->cur = in; |
5317 | 289k | in++; |
5318 | 289k | ctxt->input->line++; ctxt->input->col = 1; |
5319 | 289k | goto get_more; |
5320 | 289k | } |
5321 | 11.0k | in--; |
5322 | 11.0k | } |
5323 | 5.45M | SHRINK; |
5324 | 5.45M | GROW; |
5325 | 5.45M | in = ctxt->input->cur; |
5326 | 5.45M | if (*in == '-') { |
5327 | 5.10M | if (in[1] == '-') { |
5328 | 4.26M | if (in[2] == '>') { |
5329 | 3.00M | SKIP(3); |
5330 | 3.00M | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5331 | 3.00M | (!ctxt->disableSAX)) { |
5332 | 1.90M | if (buf != NULL) |
5333 | 253k | ctxt->sax->comment(ctxt->userData, buf); |
5334 | 1.64M | else |
5335 | 1.64M | ctxt->sax->comment(ctxt->userData, BAD_CAST ""); |
5336 | 1.90M | } |
5337 | 3.00M | if (buf != NULL) |
5338 | 333k | xmlFree(buf); |
5339 | 3.00M | return; |
5340 | 3.00M | } |
5341 | 1.26M | if (buf != NULL) { |
5342 | 1.13M | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5343 | 1.13M | "Double hyphen within comment: " |
5344 | 1.13M | "<!--%.50s\n", |
5345 | 1.13M | buf); |
5346 | 1.13M | } else |
5347 | 127k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5348 | 127k | "Double hyphen within comment\n", NULL); |
5349 | 1.26M | in++; |
5350 | 1.26M | ctxt->input->col++; |
5351 | 1.26M | } |
5352 | 2.10M | in++; |
5353 | 2.10M | ctxt->input->col++; |
5354 | 2.10M | goto get_more; |
5355 | 5.10M | } |
5356 | 5.45M | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
5357 | 347k | xmlParseCommentComplex(ctxt, buf, len, size); |
5358 | 347k | } |
5359 | | |
5360 | | |
5361 | | /** |
5362 | | * xmlParsePITarget: |
5363 | | * @ctxt: an XML parser context |
5364 | | * |
5365 | | * DEPRECATED: Internal function, don't use. |
5366 | | * |
5367 | | * parse the name of a PI |
5368 | | * |
5369 | | * [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) |
5370 | | * |
5371 | | * Returns the PITarget name or NULL |
5372 | | */ |
5373 | | |
5374 | | const xmlChar * |
5375 | 376k | xmlParsePITarget(xmlParserCtxtPtr ctxt) { |
5376 | 376k | const xmlChar *name; |
5377 | | |
5378 | 376k | name = xmlParseName(ctxt); |
5379 | 376k | if ((name != NULL) && |
5380 | 376k | ((name[0] == 'x') || (name[0] == 'X')) && |
5381 | 376k | ((name[1] == 'm') || (name[1] == 'M')) && |
5382 | 376k | ((name[2] == 'l') || (name[2] == 'L'))) { |
5383 | 108k | int i; |
5384 | 108k | if ((name[0] == 'x') && (name[1] == 'm') && |
5385 | 108k | (name[2] == 'l') && (name[3] == 0)) { |
5386 | 80.7k | xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5387 | 80.7k | "XML declaration allowed only at the start of the document\n"); |
5388 | 80.7k | return(name); |
5389 | 80.7k | } else if (name[3] == 0) { |
5390 | 14.3k | xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL); |
5391 | 14.3k | return(name); |
5392 | 14.3k | } |
5393 | 36.5k | for (i = 0;;i++) { |
5394 | 36.5k | if (xmlW3CPIs[i] == NULL) break; |
5395 | 25.3k | if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i])) |
5396 | 2.43k | return(name); |
5397 | 25.3k | } |
5398 | 11.1k | xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5399 | 11.1k | "xmlParsePITarget: invalid name prefix 'xml'\n", |
5400 | 11.1k | NULL, NULL); |
5401 | 11.1k | } |
5402 | 278k | if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) { |
5403 | 4.92k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5404 | 4.92k | "colons are forbidden from PI names '%s'\n", name, NULL, NULL); |
5405 | 4.92k | } |
5406 | 278k | return(name); |
5407 | 376k | } |
5408 | | |
5409 | | #ifdef LIBXML_CATALOG_ENABLED |
5410 | | /** |
5411 | | * xmlParseCatalogPI: |
5412 | | * @ctxt: an XML parser context |
5413 | | * @catalog: the PI value string |
5414 | | * |
5415 | | * parse an XML Catalog Processing Instruction. |
5416 | | * |
5417 | | * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?> |
5418 | | * |
5419 | | * Occurs only if allowed by the user and if happening in the Misc |
5420 | | * part of the document before any doctype information |
5421 | | * This will add the given catalog to the parsing context in order |
5422 | | * to be used if there is a resolution need further down in the document |
5423 | | */ |
5424 | | |
5425 | | static void |
5426 | 0 | xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) { |
5427 | 0 | xmlChar *URL = NULL; |
5428 | 0 | const xmlChar *tmp, *base; |
5429 | 0 | xmlChar marker; |
5430 | |
|
5431 | 0 | tmp = catalog; |
5432 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5433 | 0 | if (xmlStrncmp(tmp, BAD_CAST"catalog", 7)) |
5434 | 0 | goto error; |
5435 | 0 | tmp += 7; |
5436 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5437 | 0 | if (*tmp != '=') { |
5438 | 0 | return; |
5439 | 0 | } |
5440 | 0 | tmp++; |
5441 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5442 | 0 | marker = *tmp; |
5443 | 0 | if ((marker != '\'') && (marker != '"')) |
5444 | 0 | goto error; |
5445 | 0 | tmp++; |
5446 | 0 | base = tmp; |
5447 | 0 | while ((*tmp != 0) && (*tmp != marker)) tmp++; |
5448 | 0 | if (*tmp == 0) |
5449 | 0 | goto error; |
5450 | 0 | URL = xmlStrndup(base, tmp - base); |
5451 | 0 | tmp++; |
5452 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5453 | 0 | if (*tmp != 0) |
5454 | 0 | goto error; |
5455 | | |
5456 | 0 | if (URL != NULL) { |
5457 | | /* |
5458 | | * Unfortunately, the catalog API doesn't report OOM errors. |
5459 | | * xmlGetLastError isn't very helpful since we don't know |
5460 | | * where the last error came from. We'd have to reset it |
5461 | | * before this call and restore it afterwards. |
5462 | | */ |
5463 | 0 | ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL); |
5464 | 0 | xmlFree(URL); |
5465 | 0 | } |
5466 | 0 | return; |
5467 | | |
5468 | 0 | error: |
5469 | 0 | xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI, |
5470 | 0 | "Catalog PI syntax error: %s\n", |
5471 | 0 | catalog, NULL); |
5472 | 0 | if (URL != NULL) |
5473 | 0 | xmlFree(URL); |
5474 | 0 | } |
5475 | | #endif |
5476 | | |
5477 | | /** |
5478 | | * xmlParsePI: |
5479 | | * @ctxt: an XML parser context |
5480 | | * |
5481 | | * DEPRECATED: Internal function, don't use. |
5482 | | * |
5483 | | * parse an XML Processing Instruction. |
5484 | | * |
5485 | | * [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' |
5486 | | * |
5487 | | * The processing is transferred to SAX once parsed. |
5488 | | */ |
5489 | | |
5490 | | void |
5491 | 376k | xmlParsePI(xmlParserCtxtPtr ctxt) { |
5492 | 376k | xmlChar *buf = NULL; |
5493 | 376k | size_t len = 0; |
5494 | 376k | size_t size = XML_PARSER_BUFFER_SIZE; |
5495 | 376k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5496 | 140k | XML_MAX_HUGE_LENGTH : |
5497 | 376k | XML_MAX_TEXT_LENGTH; |
5498 | 376k | int cur, l; |
5499 | 376k | const xmlChar *target; |
5500 | | |
5501 | 376k | if ((RAW == '<') && (NXT(1) == '?')) { |
5502 | | /* |
5503 | | * this is a Processing Instruction. |
5504 | | */ |
5505 | 376k | SKIP(2); |
5506 | | |
5507 | | /* |
5508 | | * Parse the target name and check for special support like |
5509 | | * namespace. |
5510 | | */ |
5511 | 376k | target = xmlParsePITarget(ctxt); |
5512 | 376k | if (target != NULL) { |
5513 | 313k | if ((RAW == '?') && (NXT(1) == '>')) { |
5514 | 82.6k | SKIP(2); |
5515 | | |
5516 | | /* |
5517 | | * SAX: PI detected. |
5518 | | */ |
5519 | 82.6k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5520 | 82.6k | (ctxt->sax->processingInstruction != NULL)) |
5521 | 76.6k | ctxt->sax->processingInstruction(ctxt->userData, |
5522 | 76.6k | target, NULL); |
5523 | 82.6k | return; |
5524 | 82.6k | } |
5525 | 231k | buf = xmlMalloc(size); |
5526 | 231k | if (buf == NULL) { |
5527 | 125 | xmlErrMemory(ctxt); |
5528 | 125 | return; |
5529 | 125 | } |
5530 | 230k | if (SKIP_BLANKS == 0) { |
5531 | 90.7k | xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED, |
5532 | 90.7k | "ParsePI: PI %s space expected\n", target); |
5533 | 90.7k | } |
5534 | 230k | cur = xmlCurrentCharRecover(ctxt, &l); |
5535 | 27.1M | while (IS_CHAR(cur) && /* checked */ |
5536 | 27.1M | ((cur != '?') || (NXT(1) != '>'))) { |
5537 | 26.9M | if (len + 5 >= size) { |
5538 | 38.2k | xmlChar *tmp; |
5539 | 38.2k | size_t new_size = size * 2; |
5540 | 38.2k | tmp = (xmlChar *) xmlRealloc(buf, new_size); |
5541 | 38.2k | if (tmp == NULL) { |
5542 | 26 | xmlErrMemory(ctxt); |
5543 | 26 | xmlFree(buf); |
5544 | 26 | return; |
5545 | 26 | } |
5546 | 38.2k | buf = tmp; |
5547 | 38.2k | size = new_size; |
5548 | 38.2k | } |
5549 | 26.9M | COPY_BUF(buf, len, cur); |
5550 | 26.9M | if (len > maxLength) { |
5551 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5552 | 0 | "PI %s too big found", target); |
5553 | 0 | xmlFree(buf); |
5554 | 0 | return; |
5555 | 0 | } |
5556 | 26.9M | NEXTL(l); |
5557 | 26.9M | cur = xmlCurrentCharRecover(ctxt, &l); |
5558 | 26.9M | } |
5559 | 230k | buf[len] = 0; |
5560 | 230k | if (cur != '?') { |
5561 | 90.4k | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5562 | 90.4k | "ParsePI: PI %s never end ...\n", target); |
5563 | 140k | } else { |
5564 | 140k | SKIP(2); |
5565 | | |
5566 | 140k | #ifdef LIBXML_CATALOG_ENABLED |
5567 | 140k | if ((ctxt->inSubset == 0) && |
5568 | 140k | (xmlStrEqual(target, XML_CATALOG_PI))) { |
5569 | 3.84k | xmlCatalogAllow allow = xmlCatalogGetDefaults(); |
5570 | | |
5571 | 3.84k | if (((ctxt->options & XML_PARSE_NO_CATALOG_PI) == 0) && |
5572 | 3.84k | ((allow == XML_CATA_ALLOW_DOCUMENT) || |
5573 | 3.84k | (allow == XML_CATA_ALLOW_ALL))) |
5574 | 0 | xmlParseCatalogPI(ctxt, buf); |
5575 | 3.84k | } |
5576 | 140k | #endif |
5577 | | |
5578 | | /* |
5579 | | * SAX: PI detected. |
5580 | | */ |
5581 | 140k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5582 | 140k | (ctxt->sax->processingInstruction != NULL)) |
5583 | 113k | ctxt->sax->processingInstruction(ctxt->userData, |
5584 | 113k | target, buf); |
5585 | 140k | } |
5586 | 230k | xmlFree(buf); |
5587 | 230k | } else { |
5588 | 62.2k | xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL); |
5589 | 62.2k | } |
5590 | 376k | } |
5591 | 376k | } |
5592 | | |
5593 | | /** |
5594 | | * xmlParseNotationDecl: |
5595 | | * @ctxt: an XML parser context |
5596 | | * |
5597 | | * DEPRECATED: Internal function, don't use. |
5598 | | * |
5599 | | * Parse a notation declaration. Always consumes '<!'. |
5600 | | * |
5601 | | * [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>' |
5602 | | * |
5603 | | * Hence there is actually 3 choices: |
5604 | | * 'PUBLIC' S PubidLiteral |
5605 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
5606 | | * and 'SYSTEM' S SystemLiteral |
5607 | | * |
5608 | | * See the NOTE on xmlParseExternalID(). |
5609 | | */ |
5610 | | |
5611 | | void |
5612 | 43.6k | xmlParseNotationDecl(xmlParserCtxtPtr ctxt) { |
5613 | 43.6k | const xmlChar *name; |
5614 | 43.6k | xmlChar *Pubid; |
5615 | 43.6k | xmlChar *Systemid; |
5616 | | |
5617 | 43.6k | if ((CUR != '<') || (NXT(1) != '!')) |
5618 | 0 | return; |
5619 | 43.6k | SKIP(2); |
5620 | | |
5621 | 43.6k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5622 | 41.8k | int inputid = ctxt->input->id; |
5623 | 41.8k | SKIP(8); |
5624 | 41.8k | if (SKIP_BLANKS_PE == 0) { |
5625 | 1.24k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5626 | 1.24k | "Space required after '<!NOTATION'\n"); |
5627 | 1.24k | return; |
5628 | 1.24k | } |
5629 | | |
5630 | 40.5k | name = xmlParseName(ctxt); |
5631 | 40.5k | if (name == NULL) { |
5632 | 2.36k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5633 | 2.36k | return; |
5634 | 2.36k | } |
5635 | 38.2k | if (xmlStrchr(name, ':') != NULL) { |
5636 | 2.66k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5637 | 2.66k | "colons are forbidden from notation names '%s'\n", |
5638 | 2.66k | name, NULL, NULL); |
5639 | 2.66k | } |
5640 | 38.2k | if (SKIP_BLANKS_PE == 0) { |
5641 | 1.70k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5642 | 1.70k | "Space required after the NOTATION name'\n"); |
5643 | 1.70k | return; |
5644 | 1.70k | } |
5645 | | |
5646 | | /* |
5647 | | * Parse the IDs. |
5648 | | */ |
5649 | 36.5k | Systemid = xmlParseExternalID(ctxt, &Pubid, 0); |
5650 | 36.5k | SKIP_BLANKS_PE; |
5651 | | |
5652 | 36.5k | if (RAW == '>') { |
5653 | 19.6k | if (inputid != ctxt->input->id) { |
5654 | 424 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5655 | 424 | "Notation declaration doesn't start and stop" |
5656 | 424 | " in the same entity\n"); |
5657 | 424 | } |
5658 | 19.6k | NEXT; |
5659 | 19.6k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5660 | 19.6k | (ctxt->sax->notationDecl != NULL)) |
5661 | 15.1k | ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid); |
5662 | 19.6k | } else { |
5663 | 16.8k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5664 | 16.8k | } |
5665 | 36.5k | if (Systemid != NULL) xmlFree(Systemid); |
5666 | 36.5k | if (Pubid != NULL) xmlFree(Pubid); |
5667 | 36.5k | } |
5668 | 43.6k | } |
5669 | | |
5670 | | /** |
5671 | | * xmlParseEntityDecl: |
5672 | | * @ctxt: an XML parser context |
5673 | | * |
5674 | | * DEPRECATED: Internal function, don't use. |
5675 | | * |
5676 | | * Parse an entity declaration. Always consumes '<!'. |
5677 | | * |
5678 | | * [70] EntityDecl ::= GEDecl | PEDecl |
5679 | | * |
5680 | | * [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>' |
5681 | | * |
5682 | | * [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>' |
5683 | | * |
5684 | | * [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?) |
5685 | | * |
5686 | | * [74] PEDef ::= EntityValue | ExternalID |
5687 | | * |
5688 | | * [76] NDataDecl ::= S 'NDATA' S Name |
5689 | | * |
5690 | | * [ VC: Notation Declared ] |
5691 | | * The Name must match the declared name of a notation. |
5692 | | */ |
5693 | | |
5694 | | void |
5695 | 413k | xmlParseEntityDecl(xmlParserCtxtPtr ctxt) { |
5696 | 413k | const xmlChar *name = NULL; |
5697 | 413k | xmlChar *value = NULL; |
5698 | 413k | xmlChar *URI = NULL, *literal = NULL; |
5699 | 413k | const xmlChar *ndata = NULL; |
5700 | 413k | int isParameter = 0; |
5701 | 413k | xmlChar *orig = NULL; |
5702 | | |
5703 | 413k | if ((CUR != '<') || (NXT(1) != '!')) |
5704 | 0 | return; |
5705 | 413k | SKIP(2); |
5706 | | |
5707 | | /* GROW; done in the caller */ |
5708 | 413k | if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5709 | 411k | int inputid = ctxt->input->id; |
5710 | 411k | SKIP(6); |
5711 | 411k | if (SKIP_BLANKS_PE == 0) { |
5712 | 23.1k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5713 | 23.1k | "Space required after '<!ENTITY'\n"); |
5714 | 23.1k | } |
5715 | | |
5716 | 411k | if (RAW == '%') { |
5717 | 164k | NEXT; |
5718 | 164k | if (SKIP_BLANKS_PE == 0) { |
5719 | 2.83k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5720 | 2.83k | "Space required after '%%'\n"); |
5721 | 2.83k | } |
5722 | 164k | isParameter = 1; |
5723 | 164k | } |
5724 | | |
5725 | 411k | name = xmlParseName(ctxt); |
5726 | 411k | if (name == NULL) { |
5727 | 19.2k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5728 | 19.2k | "xmlParseEntityDecl: no name\n"); |
5729 | 19.2k | return; |
5730 | 19.2k | } |
5731 | 392k | if (xmlStrchr(name, ':') != NULL) { |
5732 | 3.68k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5733 | 3.68k | "colons are forbidden from entities names '%s'\n", |
5734 | 3.68k | name, NULL, NULL); |
5735 | 3.68k | } |
5736 | 392k | if (SKIP_BLANKS_PE == 0) { |
5737 | 20.5k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5738 | 20.5k | "Space required after the entity name\n"); |
5739 | 20.5k | } |
5740 | | |
5741 | | /* |
5742 | | * handle the various case of definitions... |
5743 | | */ |
5744 | 392k | if (isParameter) { |
5745 | 154k | if ((RAW == '"') || (RAW == '\'')) { |
5746 | 111k | value = xmlParseEntityValue(ctxt, &orig); |
5747 | 111k | if (value) { |
5748 | 110k | if ((ctxt->sax != NULL) && |
5749 | 110k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5750 | 85.3k | ctxt->sax->entityDecl(ctxt->userData, name, |
5751 | 85.3k | XML_INTERNAL_PARAMETER_ENTITY, |
5752 | 85.3k | NULL, NULL, value); |
5753 | 110k | } |
5754 | 111k | } else { |
5755 | 43.0k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5756 | 43.0k | if ((URI == NULL) && (literal == NULL)) { |
5757 | 2.05k | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5758 | 2.05k | } |
5759 | 43.0k | if (URI) { |
5760 | 40.3k | if (xmlStrchr(URI, '#')) { |
5761 | 1.16k | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5762 | 39.2k | } else { |
5763 | 39.2k | if ((ctxt->sax != NULL) && |
5764 | 39.2k | (!ctxt->disableSAX) && |
5765 | 39.2k | (ctxt->sax->entityDecl != NULL)) |
5766 | 36.0k | ctxt->sax->entityDecl(ctxt->userData, name, |
5767 | 36.0k | XML_EXTERNAL_PARAMETER_ENTITY, |
5768 | 36.0k | literal, URI, NULL); |
5769 | 39.2k | } |
5770 | 40.3k | } |
5771 | 43.0k | } |
5772 | 238k | } else { |
5773 | 238k | if ((RAW == '"') || (RAW == '\'')) { |
5774 | 188k | value = xmlParseEntityValue(ctxt, &orig); |
5775 | 188k | if ((ctxt->sax != NULL) && |
5776 | 188k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5777 | 153k | ctxt->sax->entityDecl(ctxt->userData, name, |
5778 | 153k | XML_INTERNAL_GENERAL_ENTITY, |
5779 | 153k | NULL, NULL, value); |
5780 | | /* |
5781 | | * For expat compatibility in SAX mode. |
5782 | | */ |
5783 | 188k | if ((ctxt->myDoc == NULL) || |
5784 | 188k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
5785 | 10.7k | if (ctxt->myDoc == NULL) { |
5786 | 2.84k | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5787 | 2.84k | if (ctxt->myDoc == NULL) { |
5788 | 7 | xmlErrMemory(ctxt); |
5789 | 7 | goto done; |
5790 | 7 | } |
5791 | 2.83k | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5792 | 2.83k | } |
5793 | 10.7k | if (ctxt->myDoc->intSubset == NULL) { |
5794 | 2.83k | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5795 | 2.83k | BAD_CAST "fake", NULL, NULL); |
5796 | 2.83k | if (ctxt->myDoc->intSubset == NULL) { |
5797 | 8 | xmlErrMemory(ctxt); |
5798 | 8 | goto done; |
5799 | 8 | } |
5800 | 2.83k | } |
5801 | | |
5802 | 10.7k | xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY, |
5803 | 10.7k | NULL, NULL, value); |
5804 | 10.7k | } |
5805 | 188k | } else { |
5806 | 49.6k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5807 | 49.6k | if ((URI == NULL) && (literal == NULL)) { |
5808 | 4.96k | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5809 | 4.96k | } |
5810 | 49.6k | if (URI) { |
5811 | 40.5k | if (xmlStrchr(URI, '#')) { |
5812 | 3.23k | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5813 | 3.23k | } |
5814 | 40.5k | } |
5815 | 49.6k | if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) { |
5816 | 5.89k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5817 | 5.89k | "Space required before 'NDATA'\n"); |
5818 | 5.89k | } |
5819 | 49.6k | if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) { |
5820 | 12.5k | SKIP(5); |
5821 | 12.5k | if (SKIP_BLANKS_PE == 0) { |
5822 | 1.77k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5823 | 1.77k | "Space required after 'NDATA'\n"); |
5824 | 1.77k | } |
5825 | 12.5k | ndata = xmlParseName(ctxt); |
5826 | 12.5k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5827 | 12.5k | (ctxt->sax->unparsedEntityDecl != NULL)) |
5828 | 10.7k | ctxt->sax->unparsedEntityDecl(ctxt->userData, name, |
5829 | 10.7k | literal, URI, ndata); |
5830 | 37.1k | } else { |
5831 | 37.1k | if ((ctxt->sax != NULL) && |
5832 | 37.1k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5833 | 30.0k | ctxt->sax->entityDecl(ctxt->userData, name, |
5834 | 30.0k | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5835 | 30.0k | literal, URI, NULL); |
5836 | | /* |
5837 | | * For expat compatibility in SAX mode. |
5838 | | * assuming the entity replacement was asked for |
5839 | | */ |
5840 | 37.1k | if ((ctxt->replaceEntities != 0) && |
5841 | 37.1k | ((ctxt->myDoc == NULL) || |
5842 | 25.5k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) { |
5843 | 2.17k | if (ctxt->myDoc == NULL) { |
5844 | 1.05k | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5845 | 1.05k | if (ctxt->myDoc == NULL) { |
5846 | 6 | xmlErrMemory(ctxt); |
5847 | 6 | goto done; |
5848 | 6 | } |
5849 | 1.04k | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5850 | 1.04k | } |
5851 | | |
5852 | 2.16k | if (ctxt->myDoc->intSubset == NULL) { |
5853 | 1.04k | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5854 | 1.04k | BAD_CAST "fake", NULL, NULL); |
5855 | 1.04k | if (ctxt->myDoc->intSubset == NULL) { |
5856 | 6 | xmlErrMemory(ctxt); |
5857 | 6 | goto done; |
5858 | 6 | } |
5859 | 1.04k | } |
5860 | 2.16k | xmlSAX2EntityDecl(ctxt, name, |
5861 | 2.16k | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5862 | 2.16k | literal, URI, NULL); |
5863 | 2.16k | } |
5864 | 37.1k | } |
5865 | 49.6k | } |
5866 | 238k | } |
5867 | 392k | SKIP_BLANKS_PE; |
5868 | 392k | if (RAW != '>') { |
5869 | 13.7k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
5870 | 13.7k | "xmlParseEntityDecl: entity %s not terminated\n", name); |
5871 | 13.7k | xmlHaltParser(ctxt); |
5872 | 378k | } else { |
5873 | 378k | if (inputid != ctxt->input->id) { |
5874 | 612 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5875 | 612 | "Entity declaration doesn't start and stop in" |
5876 | 612 | " the same entity\n"); |
5877 | 612 | } |
5878 | 378k | NEXT; |
5879 | 378k | } |
5880 | 392k | if (orig != NULL) { |
5881 | | /* |
5882 | | * Ugly mechanism to save the raw entity value. |
5883 | | */ |
5884 | 297k | xmlEntityPtr cur = NULL; |
5885 | | |
5886 | 297k | if (isParameter) { |
5887 | 110k | if ((ctxt->sax != NULL) && |
5888 | 110k | (ctxt->sax->getParameterEntity != NULL)) |
5889 | 110k | cur = ctxt->sax->getParameterEntity(ctxt->userData, name); |
5890 | 187k | } else { |
5891 | 187k | if ((ctxt->sax != NULL) && |
5892 | 187k | (ctxt->sax->getEntity != NULL)) |
5893 | 186k | cur = ctxt->sax->getEntity(ctxt->userData, name); |
5894 | 187k | if ((cur == NULL) && (ctxt->userData==ctxt)) { |
5895 | 20.6k | cur = xmlSAX2GetEntity(ctxt, name); |
5896 | 20.6k | } |
5897 | 187k | } |
5898 | 297k | if ((cur != NULL) && (cur->orig == NULL)) { |
5899 | 118k | cur->orig = orig; |
5900 | 118k | orig = NULL; |
5901 | 118k | } |
5902 | 297k | } |
5903 | | |
5904 | 392k | done: |
5905 | 392k | if (value != NULL) xmlFree(value); |
5906 | 392k | if (URI != NULL) xmlFree(URI); |
5907 | 392k | if (literal != NULL) xmlFree(literal); |
5908 | 392k | if (orig != NULL) xmlFree(orig); |
5909 | 392k | } |
5910 | 413k | } |
5911 | | |
5912 | | /** |
5913 | | * xmlParseDefaultDecl: |
5914 | | * @ctxt: an XML parser context |
5915 | | * @value: Receive a possible fixed default value for the attribute |
5916 | | * |
5917 | | * DEPRECATED: Internal function, don't use. |
5918 | | * |
5919 | | * Parse an attribute default declaration |
5920 | | * |
5921 | | * [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue) |
5922 | | * |
5923 | | * [ VC: Required Attribute ] |
5924 | | * if the default declaration is the keyword #REQUIRED, then the |
5925 | | * attribute must be specified for all elements of the type in the |
5926 | | * attribute-list declaration. |
5927 | | * |
5928 | | * [ VC: Attribute Default Legal ] |
5929 | | * The declared default value must meet the lexical constraints of |
5930 | | * the declared attribute type c.f. xmlValidateAttributeDecl() |
5931 | | * |
5932 | | * [ VC: Fixed Attribute Default ] |
5933 | | * if an attribute has a default value declared with the #FIXED |
5934 | | * keyword, instances of that attribute must match the default value. |
5935 | | * |
5936 | | * [ WFC: No < in Attribute Values ] |
5937 | | * handled in xmlParseAttValue() |
5938 | | * |
5939 | | * returns: XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED |
5940 | | * or XML_ATTRIBUTE_FIXED. |
5941 | | */ |
5942 | | |
5943 | | int |
5944 | 606k | xmlParseDefaultDecl(xmlParserCtxtPtr ctxt, xmlChar **value) { |
5945 | 606k | int val; |
5946 | 606k | xmlChar *ret; |
5947 | | |
5948 | 606k | *value = NULL; |
5949 | 606k | if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) { |
5950 | 30.5k | SKIP(9); |
5951 | 30.5k | return(XML_ATTRIBUTE_REQUIRED); |
5952 | 30.5k | } |
5953 | 576k | if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) { |
5954 | 194k | SKIP(8); |
5955 | 194k | return(XML_ATTRIBUTE_IMPLIED); |
5956 | 194k | } |
5957 | 381k | val = XML_ATTRIBUTE_NONE; |
5958 | 381k | if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) { |
5959 | 43.4k | SKIP(6); |
5960 | 43.4k | val = XML_ATTRIBUTE_FIXED; |
5961 | 43.4k | if (SKIP_BLANKS_PE == 0) { |
5962 | 5.14k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5963 | 5.14k | "Space required after '#FIXED'\n"); |
5964 | 5.14k | } |
5965 | 43.4k | } |
5966 | 381k | ret = xmlParseAttValue(ctxt); |
5967 | 381k | if (ret == NULL) { |
5968 | 52.3k | xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo, |
5969 | 52.3k | "Attribute default value declaration error\n"); |
5970 | 52.3k | } else |
5971 | 328k | *value = ret; |
5972 | 381k | return(val); |
5973 | 576k | } |
5974 | | |
5975 | | /** |
5976 | | * xmlParseNotationType: |
5977 | | * @ctxt: an XML parser context |
5978 | | * |
5979 | | * DEPRECATED: Internal function, don't use. |
5980 | | * |
5981 | | * parse an Notation attribute type. |
5982 | | * |
5983 | | * Note: the leading 'NOTATION' S part has already being parsed... |
5984 | | * |
5985 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5986 | | * |
5987 | | * [ VC: Notation Attributes ] |
5988 | | * Values of this type must match one of the notation names included |
5989 | | * in the declaration; all notation names in the declaration must be declared. |
5990 | | * |
5991 | | * Returns: the notation attribute tree built while parsing |
5992 | | */ |
5993 | | |
5994 | | xmlEnumerationPtr |
5995 | 12.3k | xmlParseNotationType(xmlParserCtxtPtr ctxt) { |
5996 | 12.3k | const xmlChar *name; |
5997 | 12.3k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5998 | | |
5999 | 12.3k | if (RAW != '(') { |
6000 | 3.07k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
6001 | 3.07k | return(NULL); |
6002 | 3.07k | } |
6003 | 15.5k | do { |
6004 | 15.5k | NEXT; |
6005 | 15.5k | SKIP_BLANKS_PE; |
6006 | 15.5k | name = xmlParseName(ctxt); |
6007 | 15.5k | if (name == NULL) { |
6008 | 1.96k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6009 | 1.96k | "Name expected in NOTATION declaration\n"); |
6010 | 1.96k | xmlFreeEnumeration(ret); |
6011 | 1.96k | return(NULL); |
6012 | 1.96k | } |
6013 | 13.5k | tmp = ret; |
6014 | 30.6k | while (tmp != NULL) { |
6015 | 20.0k | if (xmlStrEqual(name, tmp->name)) { |
6016 | 2.95k | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
6017 | 2.95k | "standalone: attribute notation value token %s duplicated\n", |
6018 | 2.95k | name, NULL); |
6019 | 2.95k | if (!xmlDictOwns(ctxt->dict, name)) |
6020 | 0 | xmlFree((xmlChar *) name); |
6021 | 2.95k | break; |
6022 | 2.95k | } |
6023 | 17.1k | tmp = tmp->next; |
6024 | 17.1k | } |
6025 | 13.5k | if (tmp == NULL) { |
6026 | 10.6k | cur = xmlCreateEnumeration(name); |
6027 | 10.6k | if (cur == NULL) { |
6028 | 36 | xmlErrMemory(ctxt); |
6029 | 36 | xmlFreeEnumeration(ret); |
6030 | 36 | return(NULL); |
6031 | 36 | } |
6032 | 10.5k | if (last == NULL) ret = last = cur; |
6033 | 2.96k | else { |
6034 | 2.96k | last->next = cur; |
6035 | 2.96k | last = cur; |
6036 | 2.96k | } |
6037 | 10.5k | } |
6038 | 13.5k | SKIP_BLANKS_PE; |
6039 | 13.5k | } while (RAW == '|'); |
6040 | 7.30k | if (RAW != ')') { |
6041 | 1.35k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
6042 | 1.35k | xmlFreeEnumeration(ret); |
6043 | 1.35k | return(NULL); |
6044 | 1.35k | } |
6045 | 5.94k | NEXT; |
6046 | 5.94k | return(ret); |
6047 | 7.30k | } |
6048 | | |
6049 | | /** |
6050 | | * xmlParseEnumerationType: |
6051 | | * @ctxt: an XML parser context |
6052 | | * |
6053 | | * DEPRECATED: Internal function, don't use. |
6054 | | * |
6055 | | * parse an Enumeration attribute type. |
6056 | | * |
6057 | | * [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' |
6058 | | * |
6059 | | * [ VC: Enumeration ] |
6060 | | * Values of this type must match one of the Nmtoken tokens in |
6061 | | * the declaration |
6062 | | * |
6063 | | * Returns: the enumeration attribute tree built while parsing |
6064 | | */ |
6065 | | |
6066 | | xmlEnumerationPtr |
6067 | 148k | xmlParseEnumerationType(xmlParserCtxtPtr ctxt) { |
6068 | 148k | xmlChar *name; |
6069 | 148k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
6070 | | |
6071 | 148k | if (RAW != '(') { |
6072 | 14.9k | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL); |
6073 | 14.9k | return(NULL); |
6074 | 14.9k | } |
6075 | 187k | do { |
6076 | 187k | NEXT; |
6077 | 187k | SKIP_BLANKS_PE; |
6078 | 187k | name = xmlParseNmtoken(ctxt); |
6079 | 187k | if (name == NULL) { |
6080 | 2.01k | xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL); |
6081 | 2.01k | return(ret); |
6082 | 2.01k | } |
6083 | 185k | tmp = ret; |
6084 | 318k | while (tmp != NULL) { |
6085 | 136k | if (xmlStrEqual(name, tmp->name)) { |
6086 | 3.28k | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
6087 | 3.28k | "standalone: attribute enumeration value token %s duplicated\n", |
6088 | 3.28k | name, NULL); |
6089 | 3.28k | if (!xmlDictOwns(ctxt->dict, name)) |
6090 | 3.28k | xmlFree(name); |
6091 | 3.28k | break; |
6092 | 3.28k | } |
6093 | 133k | tmp = tmp->next; |
6094 | 133k | } |
6095 | 185k | if (tmp == NULL) { |
6096 | 181k | cur = xmlCreateEnumeration(name); |
6097 | 181k | if (!xmlDictOwns(ctxt->dict, name)) |
6098 | 181k | xmlFree(name); |
6099 | 181k | if (cur == NULL) { |
6100 | 85 | xmlErrMemory(ctxt); |
6101 | 85 | xmlFreeEnumeration(ret); |
6102 | 85 | return(NULL); |
6103 | 85 | } |
6104 | 181k | if (last == NULL) ret = last = cur; |
6105 | 50.2k | else { |
6106 | 50.2k | last->next = cur; |
6107 | 50.2k | last = cur; |
6108 | 50.2k | } |
6109 | 181k | } |
6110 | 185k | SKIP_BLANKS_PE; |
6111 | 185k | } while (RAW == '|'); |
6112 | 131k | if (RAW != ')') { |
6113 | 3.18k | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL); |
6114 | 3.18k | return(ret); |
6115 | 3.18k | } |
6116 | 128k | NEXT; |
6117 | 128k | return(ret); |
6118 | 131k | } |
6119 | | |
6120 | | /** |
6121 | | * xmlParseEnumeratedType: |
6122 | | * @ctxt: an XML parser context |
6123 | | * @tree: the enumeration tree built while parsing |
6124 | | * |
6125 | | * DEPRECATED: Internal function, don't use. |
6126 | | * |
6127 | | * parse an Enumerated attribute type. |
6128 | | * |
6129 | | * [57] EnumeratedType ::= NotationType | Enumeration |
6130 | | * |
6131 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
6132 | | * |
6133 | | * |
6134 | | * Returns: XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION |
6135 | | */ |
6136 | | |
6137 | | int |
6138 | 161k | xmlParseEnumeratedType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
6139 | 161k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
6140 | 13.4k | SKIP(8); |
6141 | 13.4k | if (SKIP_BLANKS_PE == 0) { |
6142 | 1.12k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6143 | 1.12k | "Space required after 'NOTATION'\n"); |
6144 | 1.12k | return(0); |
6145 | 1.12k | } |
6146 | 12.3k | *tree = xmlParseNotationType(ctxt); |
6147 | 12.3k | if (*tree == NULL) return(0); |
6148 | 5.94k | return(XML_ATTRIBUTE_NOTATION); |
6149 | 12.3k | } |
6150 | 148k | *tree = xmlParseEnumerationType(ctxt); |
6151 | 148k | if (*tree == NULL) return(0); |
6152 | 131k | return(XML_ATTRIBUTE_ENUMERATION); |
6153 | 148k | } |
6154 | | |
6155 | | /** |
6156 | | * xmlParseAttributeType: |
6157 | | * @ctxt: an XML parser context |
6158 | | * @tree: the enumeration tree built while parsing |
6159 | | * |
6160 | | * DEPRECATED: Internal function, don't use. |
6161 | | * |
6162 | | * parse the Attribute list def for an element |
6163 | | * |
6164 | | * [54] AttType ::= StringType | TokenizedType | EnumeratedType |
6165 | | * |
6166 | | * [55] StringType ::= 'CDATA' |
6167 | | * |
6168 | | * [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' | |
6169 | | * 'ENTITIES' | 'NMTOKEN' | 'NMTOKENS' |
6170 | | * |
6171 | | * Validity constraints for attribute values syntax are checked in |
6172 | | * xmlValidateAttributeValue() |
6173 | | * |
6174 | | * [ VC: ID ] |
6175 | | * Values of type ID must match the Name production. A name must not |
6176 | | * appear more than once in an XML document as a value of this type; |
6177 | | * i.e., ID values must uniquely identify the elements which bear them. |
6178 | | * |
6179 | | * [ VC: One ID per Element Type ] |
6180 | | * No element type may have more than one ID attribute specified. |
6181 | | * |
6182 | | * [ VC: ID Attribute Default ] |
6183 | | * An ID attribute must have a declared default of #IMPLIED or #REQUIRED. |
6184 | | * |
6185 | | * [ VC: IDREF ] |
6186 | | * Values of type IDREF must match the Name production, and values |
6187 | | * of type IDREFS must match Names; each IDREF Name must match the value |
6188 | | * of an ID attribute on some element in the XML document; i.e. IDREF |
6189 | | * values must match the value of some ID attribute. |
6190 | | * |
6191 | | * [ VC: Entity Name ] |
6192 | | * Values of type ENTITY must match the Name production, values |
6193 | | * of type ENTITIES must match Names; each Entity Name must match the |
6194 | | * name of an unparsed entity declared in the DTD. |
6195 | | * |
6196 | | * [ VC: Name Token ] |
6197 | | * Values of type NMTOKEN must match the Nmtoken production; values |
6198 | | * of type NMTOKENS must match Nmtokens. |
6199 | | * |
6200 | | * Returns the attribute type |
6201 | | */ |
6202 | | int |
6203 | 640k | xmlParseAttributeType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
6204 | 640k | if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) { |
6205 | 168k | SKIP(5); |
6206 | 168k | return(XML_ATTRIBUTE_CDATA); |
6207 | 471k | } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) { |
6208 | 31.6k | SKIP(6); |
6209 | 31.6k | return(XML_ATTRIBUTE_IDREFS); |
6210 | 439k | } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) { |
6211 | 25.7k | SKIP(5); |
6212 | 25.7k | return(XML_ATTRIBUTE_IDREF); |
6213 | 414k | } else if ((RAW == 'I') && (NXT(1) == 'D')) { |
6214 | 132k | SKIP(2); |
6215 | 132k | return(XML_ATTRIBUTE_ID); |
6216 | 281k | } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
6217 | 32.3k | SKIP(6); |
6218 | 32.3k | return(XML_ATTRIBUTE_ENTITY); |
6219 | 249k | } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) { |
6220 | 21.5k | SKIP(8); |
6221 | 21.5k | return(XML_ATTRIBUTE_ENTITIES); |
6222 | 227k | } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) { |
6223 | 17.3k | SKIP(8); |
6224 | 17.3k | return(XML_ATTRIBUTE_NMTOKENS); |
6225 | 210k | } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) { |
6226 | 48.3k | SKIP(7); |
6227 | 48.3k | return(XML_ATTRIBUTE_NMTOKEN); |
6228 | 48.3k | } |
6229 | 161k | return(xmlParseEnumeratedType(ctxt, tree)); |
6230 | 640k | } |
6231 | | |
6232 | | /** |
6233 | | * xmlParseAttributeListDecl: |
6234 | | * @ctxt: an XML parser context |
6235 | | * |
6236 | | * DEPRECATED: Internal function, don't use. |
6237 | | * |
6238 | | * Parse an attribute list declaration for an element. Always consumes '<!'. |
6239 | | * |
6240 | | * [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>' |
6241 | | * |
6242 | | * [53] AttDef ::= S Name S AttType S DefaultDecl |
6243 | | * |
6244 | | */ |
6245 | | void |
6246 | 405k | xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt) { |
6247 | 405k | const xmlChar *elemName; |
6248 | 405k | const xmlChar *attrName; |
6249 | 405k | xmlEnumerationPtr tree; |
6250 | | |
6251 | 405k | if ((CUR != '<') || (NXT(1) != '!')) |
6252 | 0 | return; |
6253 | 405k | SKIP(2); |
6254 | | |
6255 | 405k | if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) { |
6256 | 403k | int inputid = ctxt->input->id; |
6257 | | |
6258 | 403k | SKIP(7); |
6259 | 403k | if (SKIP_BLANKS_PE == 0) { |
6260 | 14.7k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6261 | 14.7k | "Space required after '<!ATTLIST'\n"); |
6262 | 14.7k | } |
6263 | 403k | elemName = xmlParseName(ctxt); |
6264 | 403k | if (elemName == NULL) { |
6265 | 8.46k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6266 | 8.46k | "ATTLIST: no name for Element\n"); |
6267 | 8.46k | return; |
6268 | 8.46k | } |
6269 | 394k | SKIP_BLANKS_PE; |
6270 | 394k | GROW; |
6271 | 978k | while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) { |
6272 | 712k | int type; |
6273 | 712k | int def; |
6274 | 712k | xmlChar *defaultValue = NULL; |
6275 | | |
6276 | 712k | GROW; |
6277 | 712k | tree = NULL; |
6278 | 712k | attrName = xmlParseName(ctxt); |
6279 | 712k | if (attrName == NULL) { |
6280 | 63.2k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6281 | 63.2k | "ATTLIST: no name for Attribute\n"); |
6282 | 63.2k | break; |
6283 | 63.2k | } |
6284 | 648k | GROW; |
6285 | 648k | if (SKIP_BLANKS_PE == 0) { |
6286 | 8.53k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6287 | 8.53k | "Space required after the attribute name\n"); |
6288 | 8.53k | break; |
6289 | 8.53k | } |
6290 | | |
6291 | 640k | type = xmlParseAttributeType(ctxt, &tree); |
6292 | 640k | if (type <= 0) { |
6293 | 24.3k | break; |
6294 | 24.3k | } |
6295 | | |
6296 | 615k | GROW; |
6297 | 615k | if (SKIP_BLANKS_PE == 0) { |
6298 | 9.27k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6299 | 9.27k | "Space required after the attribute type\n"); |
6300 | 9.27k | if (tree != NULL) |
6301 | 3.97k | xmlFreeEnumeration(tree); |
6302 | 9.27k | break; |
6303 | 9.27k | } |
6304 | | |
6305 | 606k | def = xmlParseDefaultDecl(ctxt, &defaultValue); |
6306 | 606k | if (def <= 0) { |
6307 | 0 | if (defaultValue != NULL) |
6308 | 0 | xmlFree(defaultValue); |
6309 | 0 | if (tree != NULL) |
6310 | 0 | xmlFreeEnumeration(tree); |
6311 | 0 | break; |
6312 | 0 | } |
6313 | 606k | if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL)) |
6314 | 264k | xmlAttrNormalizeSpace(defaultValue, defaultValue); |
6315 | | |
6316 | 606k | GROW; |
6317 | 606k | if (RAW != '>') { |
6318 | 419k | if (SKIP_BLANKS_PE == 0) { |
6319 | 22.6k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6320 | 22.6k | "Space required after the attribute default value\n"); |
6321 | 22.6k | if (defaultValue != NULL) |
6322 | 2.15k | xmlFree(defaultValue); |
6323 | 22.6k | if (tree != NULL) |
6324 | 2.42k | xmlFreeEnumeration(tree); |
6325 | 22.6k | break; |
6326 | 22.6k | } |
6327 | 419k | } |
6328 | 583k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6329 | 583k | (ctxt->sax->attributeDecl != NULL)) |
6330 | 522k | ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName, |
6331 | 522k | type, def, defaultValue, tree); |
6332 | 61.1k | else if (tree != NULL) |
6333 | 17.9k | xmlFreeEnumeration(tree); |
6334 | | |
6335 | 583k | if ((ctxt->sax2) && (defaultValue != NULL) && |
6336 | 583k | (def != XML_ATTRIBUTE_IMPLIED) && |
6337 | 583k | (def != XML_ATTRIBUTE_REQUIRED)) { |
6338 | 282k | xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue); |
6339 | 282k | } |
6340 | 583k | if (ctxt->sax2) { |
6341 | 511k | xmlAddSpecialAttr(ctxt, elemName, attrName, type); |
6342 | 511k | } |
6343 | 583k | if (defaultValue != NULL) |
6344 | 326k | xmlFree(defaultValue); |
6345 | 583k | GROW; |
6346 | 583k | } |
6347 | 394k | if (RAW == '>') { |
6348 | 273k | if (inputid != ctxt->input->id) { |
6349 | 2.68k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6350 | 2.68k | "Attribute list declaration doesn't start and" |
6351 | 2.68k | " stop in the same entity\n"); |
6352 | 2.68k | } |
6353 | 273k | NEXT; |
6354 | 273k | } |
6355 | 394k | } |
6356 | 405k | } |
6357 | | |
6358 | | /** |
6359 | | * xmlParseElementMixedContentDecl: |
6360 | | * @ctxt: an XML parser context |
6361 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6362 | | * |
6363 | | * DEPRECATED: Internal function, don't use. |
6364 | | * |
6365 | | * parse the declaration for a Mixed Element content |
6366 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6367 | | * |
6368 | | * [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' | |
6369 | | * '(' S? '#PCDATA' S? ')' |
6370 | | * |
6371 | | * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49]) |
6372 | | * |
6373 | | * [ VC: No Duplicate Types ] |
6374 | | * The same name must not appear more than once in a single |
6375 | | * mixed-content declaration. |
6376 | | * |
6377 | | * returns: the list of the xmlElementContentPtr describing the element choices |
6378 | | */ |
6379 | | xmlElementContentPtr |
6380 | 61.2k | xmlParseElementMixedContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6381 | 61.2k | xmlElementContentPtr ret = NULL, cur = NULL, n; |
6382 | 61.2k | const xmlChar *elem = NULL; |
6383 | | |
6384 | 61.2k | GROW; |
6385 | 61.2k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6386 | 61.2k | SKIP(7); |
6387 | 61.2k | SKIP_BLANKS_PE; |
6388 | 61.2k | if (RAW == ')') { |
6389 | 36.9k | if (ctxt->input->id != inputchk) { |
6390 | 276 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6391 | 276 | "Element content declaration doesn't start and" |
6392 | 276 | " stop in the same entity\n"); |
6393 | 276 | } |
6394 | 36.9k | NEXT; |
6395 | 36.9k | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6396 | 36.9k | if (ret == NULL) |
6397 | 25 | goto mem_error; |
6398 | 36.9k | if (RAW == '*') { |
6399 | 832 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6400 | 832 | NEXT; |
6401 | 832 | } |
6402 | 36.9k | return(ret); |
6403 | 36.9k | } |
6404 | 24.2k | if ((RAW == '(') || (RAW == '|')) { |
6405 | 20.3k | ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6406 | 20.3k | if (ret == NULL) |
6407 | 9 | goto mem_error; |
6408 | 20.3k | } |
6409 | 182k | while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) { |
6410 | 158k | NEXT; |
6411 | 158k | n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6412 | 158k | if (n == NULL) |
6413 | 24 | goto mem_error; |
6414 | 158k | if (elem == NULL) { |
6415 | 20.1k | n->c1 = cur; |
6416 | 20.1k | if (cur != NULL) |
6417 | 20.1k | cur->parent = n; |
6418 | 20.1k | ret = cur = n; |
6419 | 138k | } else { |
6420 | 138k | cur->c2 = n; |
6421 | 138k | n->parent = cur; |
6422 | 138k | n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6423 | 138k | if (n->c1 == NULL) |
6424 | 25 | goto mem_error; |
6425 | 138k | n->c1->parent = n; |
6426 | 138k | cur = n; |
6427 | 138k | } |
6428 | 158k | SKIP_BLANKS_PE; |
6429 | 158k | elem = xmlParseName(ctxt); |
6430 | 158k | if (elem == NULL) { |
6431 | 743 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6432 | 743 | "xmlParseElementMixedContentDecl : Name expected\n"); |
6433 | 743 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6434 | 743 | return(NULL); |
6435 | 743 | } |
6436 | 157k | SKIP_BLANKS_PE; |
6437 | 157k | GROW; |
6438 | 157k | } |
6439 | 23.4k | if ((RAW == ')') && (NXT(1) == '*')) { |
6440 | 17.6k | if (elem != NULL) { |
6441 | 17.6k | cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem, |
6442 | 17.6k | XML_ELEMENT_CONTENT_ELEMENT); |
6443 | 17.6k | if (cur->c2 == NULL) |
6444 | 17 | goto mem_error; |
6445 | 17.5k | cur->c2->parent = cur; |
6446 | 17.5k | } |
6447 | 17.5k | if (ret != NULL) |
6448 | 17.5k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6449 | 17.5k | if (ctxt->input->id != inputchk) { |
6450 | 7 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6451 | 7 | "Element content declaration doesn't start and" |
6452 | 7 | " stop in the same entity\n"); |
6453 | 7 | } |
6454 | 17.5k | SKIP(2); |
6455 | 17.5k | } else { |
6456 | 5.87k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6457 | 5.87k | xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL); |
6458 | 5.87k | return(NULL); |
6459 | 5.87k | } |
6460 | | |
6461 | 23.4k | } else { |
6462 | 0 | xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL); |
6463 | 0 | } |
6464 | 17.5k | return(ret); |
6465 | | |
6466 | 100 | mem_error: |
6467 | 100 | xmlErrMemory(ctxt); |
6468 | 100 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6469 | 100 | return(NULL); |
6470 | 61.2k | } |
6471 | | |
6472 | | /** |
6473 | | * xmlParseElementChildrenContentDeclPriv: |
6474 | | * @ctxt: an XML parser context |
6475 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6476 | | * @depth: the level of recursion |
6477 | | * |
6478 | | * parse the declaration for a Mixed Element content |
6479 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6480 | | * |
6481 | | * |
6482 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6483 | | * |
6484 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6485 | | * |
6486 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6487 | | * |
6488 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6489 | | * |
6490 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6491 | | * TODO Parameter-entity replacement text must be properly nested |
6492 | | * with parenthesized groups. That is to say, if either of the |
6493 | | * opening or closing parentheses in a choice, seq, or Mixed |
6494 | | * construct is contained in the replacement text for a parameter |
6495 | | * entity, both must be contained in the same replacement text. For |
6496 | | * interoperability, if a parameter-entity reference appears in a |
6497 | | * choice, seq, or Mixed construct, its replacement text should not |
6498 | | * be empty, and neither the first nor last non-blank character of |
6499 | | * the replacement text should be a connector (| or ,). |
6500 | | * |
6501 | | * Returns the tree of xmlElementContentPtr describing the element |
6502 | | * hierarchy. |
6503 | | */ |
6504 | | static xmlElementContentPtr |
6505 | | xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int inputchk, |
6506 | 427k | int depth) { |
6507 | 427k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
6508 | 427k | xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL; |
6509 | 427k | const xmlChar *elem; |
6510 | 427k | xmlChar type = 0; |
6511 | | |
6512 | 427k | if (depth > maxDepth) { |
6513 | 22 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
6514 | 22 | "xmlParseElementChildrenContentDecl : depth %d too deep, " |
6515 | 22 | "use XML_PARSE_HUGE\n", depth); |
6516 | 22 | return(NULL); |
6517 | 22 | } |
6518 | 427k | SKIP_BLANKS_PE; |
6519 | 427k | GROW; |
6520 | 427k | if (RAW == '(') { |
6521 | 253k | int inputid = ctxt->input->id; |
6522 | | |
6523 | | /* Recurse on first child */ |
6524 | 253k | NEXT; |
6525 | 253k | SKIP_BLANKS_PE; |
6526 | 253k | cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6527 | 253k | depth + 1); |
6528 | 253k | if (cur == NULL) |
6529 | 214k | return(NULL); |
6530 | 39.6k | SKIP_BLANKS_PE; |
6531 | 39.6k | GROW; |
6532 | 173k | } else { |
6533 | 173k | elem = xmlParseName(ctxt); |
6534 | 173k | if (elem == NULL) { |
6535 | 7.17k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6536 | 7.17k | return(NULL); |
6537 | 7.17k | } |
6538 | 165k | cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6539 | 165k | if (cur == NULL) { |
6540 | 70 | xmlErrMemory(ctxt); |
6541 | 70 | return(NULL); |
6542 | 70 | } |
6543 | 165k | GROW; |
6544 | 165k | if (RAW == '?') { |
6545 | 13.3k | cur->ocur = XML_ELEMENT_CONTENT_OPT; |
6546 | 13.3k | NEXT; |
6547 | 152k | } else if (RAW == '*') { |
6548 | 30.2k | cur->ocur = XML_ELEMENT_CONTENT_MULT; |
6549 | 30.2k | NEXT; |
6550 | 122k | } else if (RAW == '+') { |
6551 | 14.1k | cur->ocur = XML_ELEMENT_CONTENT_PLUS; |
6552 | 14.1k | NEXT; |
6553 | 108k | } else { |
6554 | 108k | cur->ocur = XML_ELEMENT_CONTENT_ONCE; |
6555 | 108k | } |
6556 | 165k | GROW; |
6557 | 165k | } |
6558 | 205k | SKIP_BLANKS_PE; |
6559 | 651k | while ((RAW != ')') && (PARSER_STOPPED(ctxt) == 0)) { |
6560 | | /* |
6561 | | * Each loop we parse one separator and one element. |
6562 | | */ |
6563 | 477k | if (RAW == ',') { |
6564 | 148k | if (type == 0) type = CUR; |
6565 | | |
6566 | | /* |
6567 | | * Detect "Name | Name , Name" error |
6568 | | */ |
6569 | 100k | else if (type != CUR) { |
6570 | 235 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6571 | 235 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6572 | 235 | type); |
6573 | 235 | if ((last != NULL) && (last != ret)) |
6574 | 235 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6575 | 235 | if (ret != NULL) |
6576 | 235 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6577 | 235 | return(NULL); |
6578 | 235 | } |
6579 | 148k | NEXT; |
6580 | | |
6581 | 148k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ); |
6582 | 148k | if (op == NULL) { |
6583 | 40 | xmlErrMemory(ctxt); |
6584 | 40 | if ((last != NULL) && (last != ret)) |
6585 | 16 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6586 | 40 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6587 | 40 | return(NULL); |
6588 | 40 | } |
6589 | 148k | if (last == NULL) { |
6590 | 48.1k | op->c1 = ret; |
6591 | 48.1k | if (ret != NULL) |
6592 | 48.1k | ret->parent = op; |
6593 | 48.1k | ret = cur = op; |
6594 | 99.9k | } else { |
6595 | 99.9k | cur->c2 = op; |
6596 | 99.9k | if (op != NULL) |
6597 | 99.9k | op->parent = cur; |
6598 | 99.9k | op->c1 = last; |
6599 | 99.9k | if (last != NULL) |
6600 | 99.9k | last->parent = op; |
6601 | 99.9k | cur =op; |
6602 | 99.9k | last = NULL; |
6603 | 99.9k | } |
6604 | 329k | } else if (RAW == '|') { |
6605 | 318k | if (type == 0) type = CUR; |
6606 | | |
6607 | | /* |
6608 | | * Detect "Name , Name | Name" error |
6609 | | */ |
6610 | 235k | else if (type != CUR) { |
6611 | 256 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6612 | 256 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6613 | 256 | type); |
6614 | 256 | if ((last != NULL) && (last != ret)) |
6615 | 256 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6616 | 256 | if (ret != NULL) |
6617 | 256 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6618 | 256 | return(NULL); |
6619 | 256 | } |
6620 | 318k | NEXT; |
6621 | | |
6622 | 318k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6623 | 318k | if (op == NULL) { |
6624 | 48 | xmlErrMemory(ctxt); |
6625 | 48 | if ((last != NULL) && (last != ret)) |
6626 | 24 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6627 | 48 | if (ret != NULL) |
6628 | 48 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6629 | 48 | return(NULL); |
6630 | 48 | } |
6631 | 318k | if (last == NULL) { |
6632 | 82.7k | op->c1 = ret; |
6633 | 82.7k | if (ret != NULL) |
6634 | 82.7k | ret->parent = op; |
6635 | 82.7k | ret = cur = op; |
6636 | 235k | } else { |
6637 | 235k | cur->c2 = op; |
6638 | 235k | if (op != NULL) |
6639 | 235k | op->parent = cur; |
6640 | 235k | op->c1 = last; |
6641 | 235k | if (last != NULL) |
6642 | 235k | last->parent = op; |
6643 | 235k | cur =op; |
6644 | 235k | last = NULL; |
6645 | 235k | } |
6646 | 318k | } else { |
6647 | 10.7k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL); |
6648 | 10.7k | if ((last != NULL) && (last != ret)) |
6649 | 4.94k | xmlFreeDocElementContent(ctxt->myDoc, last); |
6650 | 10.7k | if (ret != NULL) |
6651 | 10.7k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6652 | 10.7k | return(NULL); |
6653 | 10.7k | } |
6654 | 466k | GROW; |
6655 | 466k | SKIP_BLANKS_PE; |
6656 | 466k | GROW; |
6657 | 466k | if (RAW == '(') { |
6658 | 87.2k | int inputid = ctxt->input->id; |
6659 | | /* Recurse on second child */ |
6660 | 87.2k | NEXT; |
6661 | 87.2k | SKIP_BLANKS_PE; |
6662 | 87.2k | last = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6663 | 87.2k | depth + 1); |
6664 | 87.2k | if (last == NULL) { |
6665 | 17.6k | if (ret != NULL) |
6666 | 17.6k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6667 | 17.6k | return(NULL); |
6668 | 17.6k | } |
6669 | 69.6k | SKIP_BLANKS_PE; |
6670 | 378k | } else { |
6671 | 378k | elem = xmlParseName(ctxt); |
6672 | 378k | if (elem == NULL) { |
6673 | 2.20k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6674 | 2.20k | if (ret != NULL) |
6675 | 2.20k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6676 | 2.20k | return(NULL); |
6677 | 2.20k | } |
6678 | 376k | last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6679 | 376k | if (last == NULL) { |
6680 | 80 | xmlErrMemory(ctxt); |
6681 | 80 | if (ret != NULL) |
6682 | 80 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6683 | 80 | return(NULL); |
6684 | 80 | } |
6685 | 376k | if (RAW == '?') { |
6686 | 58.3k | last->ocur = XML_ELEMENT_CONTENT_OPT; |
6687 | 58.3k | NEXT; |
6688 | 318k | } else if (RAW == '*') { |
6689 | 20.9k | last->ocur = XML_ELEMENT_CONTENT_MULT; |
6690 | 20.9k | NEXT; |
6691 | 297k | } else if (RAW == '+') { |
6692 | 19.6k | last->ocur = XML_ELEMENT_CONTENT_PLUS; |
6693 | 19.6k | NEXT; |
6694 | 277k | } else { |
6695 | 277k | last->ocur = XML_ELEMENT_CONTENT_ONCE; |
6696 | 277k | } |
6697 | 376k | } |
6698 | 446k | SKIP_BLANKS_PE; |
6699 | 446k | GROW; |
6700 | 446k | } |
6701 | 174k | if ((cur != NULL) && (last != NULL)) { |
6702 | 105k | cur->c2 = last; |
6703 | 105k | if (last != NULL) |
6704 | 105k | last->parent = cur; |
6705 | 105k | } |
6706 | 174k | if (ctxt->input->id != inputchk) { |
6707 | 179 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6708 | 179 | "Element content declaration doesn't start and stop in" |
6709 | 179 | " the same entity\n"); |
6710 | 179 | } |
6711 | 174k | NEXT; |
6712 | 174k | if (RAW == '?') { |
6713 | 17.8k | if (ret != NULL) { |
6714 | 17.8k | if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) || |
6715 | 17.8k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6716 | 4.05k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6717 | 13.7k | else |
6718 | 13.7k | ret->ocur = XML_ELEMENT_CONTENT_OPT; |
6719 | 17.8k | } |
6720 | 17.8k | NEXT; |
6721 | 156k | } else if (RAW == '*') { |
6722 | 40.3k | if (ret != NULL) { |
6723 | 40.3k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6724 | 40.3k | cur = ret; |
6725 | | /* |
6726 | | * Some normalization: |
6727 | | * (a | b* | c?)* == (a | b | c)* |
6728 | | */ |
6729 | 186k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6730 | 145k | if ((cur->c1 != NULL) && |
6731 | 145k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6732 | 145k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) |
6733 | 17.7k | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6734 | 145k | if ((cur->c2 != NULL) && |
6735 | 145k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6736 | 145k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) |
6737 | 14.1k | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6738 | 145k | cur = cur->c2; |
6739 | 145k | } |
6740 | 40.3k | } |
6741 | 40.3k | NEXT; |
6742 | 116k | } else if (RAW == '+') { |
6743 | 44.9k | if (ret != NULL) { |
6744 | 44.9k | int found = 0; |
6745 | | |
6746 | 44.9k | if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) || |
6747 | 44.9k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6748 | 8.13k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6749 | 36.8k | else |
6750 | 36.8k | ret->ocur = XML_ELEMENT_CONTENT_PLUS; |
6751 | | /* |
6752 | | * Some normalization: |
6753 | | * (a | b*)+ == (a | b)* |
6754 | | * (a | b?)+ == (a | b)* |
6755 | | */ |
6756 | 93.4k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6757 | 48.5k | if ((cur->c1 != NULL) && |
6758 | 48.5k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6759 | 48.5k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6760 | 7.69k | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6761 | 7.69k | found = 1; |
6762 | 7.69k | } |
6763 | 48.5k | if ((cur->c2 != NULL) && |
6764 | 48.5k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6765 | 48.5k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6766 | 14.9k | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6767 | 14.9k | found = 1; |
6768 | 14.9k | } |
6769 | 48.5k | cur = cur->c2; |
6770 | 48.5k | } |
6771 | 44.9k | if (found) |
6772 | 16.3k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6773 | 44.9k | } |
6774 | 44.9k | NEXT; |
6775 | 44.9k | } |
6776 | 174k | return(ret); |
6777 | 205k | } |
6778 | | |
6779 | | /** |
6780 | | * xmlParseElementChildrenContentDecl: |
6781 | | * @ctxt: an XML parser context |
6782 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6783 | | * |
6784 | | * DEPRECATED: Internal function, don't use. |
6785 | | * |
6786 | | * parse the declaration for a Mixed Element content |
6787 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6788 | | * |
6789 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6790 | | * |
6791 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6792 | | * |
6793 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6794 | | * |
6795 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6796 | | * |
6797 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6798 | | * TODO Parameter-entity replacement text must be properly nested |
6799 | | * with parenthesized groups. That is to say, if either of the |
6800 | | * opening or closing parentheses in a choice, seq, or Mixed |
6801 | | * construct is contained in the replacement text for a parameter |
6802 | | * entity, both must be contained in the same replacement text. For |
6803 | | * interoperability, if a parameter-entity reference appears in a |
6804 | | * choice, seq, or Mixed construct, its replacement text should not |
6805 | | * be empty, and neither the first nor last non-blank character of |
6806 | | * the replacement text should be a connector (| or ,). |
6807 | | * |
6808 | | * Returns the tree of xmlElementContentPtr describing the element |
6809 | | * hierarchy. |
6810 | | */ |
6811 | | xmlElementContentPtr |
6812 | 0 | xmlParseElementChildrenContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6813 | | /* stub left for API/ABI compat */ |
6814 | 0 | return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1)); |
6815 | 0 | } |
6816 | | |
6817 | | /** |
6818 | | * xmlParseElementContentDecl: |
6819 | | * @ctxt: an XML parser context |
6820 | | * @name: the name of the element being defined. |
6821 | | * @result: the Element Content pointer will be stored here if any |
6822 | | * |
6823 | | * DEPRECATED: Internal function, don't use. |
6824 | | * |
6825 | | * parse the declaration for an Element content either Mixed or Children, |
6826 | | * the cases EMPTY and ANY are handled directly in xmlParseElementDecl |
6827 | | * |
6828 | | * [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children |
6829 | | * |
6830 | | * returns: the type of element content XML_ELEMENT_TYPE_xxx |
6831 | | */ |
6832 | | |
6833 | | int |
6834 | | xmlParseElementContentDecl(xmlParserCtxtPtr ctxt, const xmlChar *name, |
6835 | 147k | xmlElementContentPtr *result) { |
6836 | | |
6837 | 147k | xmlElementContentPtr tree = NULL; |
6838 | 147k | int inputid = ctxt->input->id; |
6839 | 147k | int res; |
6840 | | |
6841 | 147k | *result = NULL; |
6842 | | |
6843 | 147k | if (RAW != '(') { |
6844 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6845 | 0 | "xmlParseElementContentDecl : %s '(' expected\n", name); |
6846 | 0 | return(-1); |
6847 | 0 | } |
6848 | 147k | NEXT; |
6849 | 147k | GROW; |
6850 | 147k | SKIP_BLANKS_PE; |
6851 | 147k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6852 | 61.2k | tree = xmlParseElementMixedContentDecl(ctxt, inputid); |
6853 | 61.2k | res = XML_ELEMENT_TYPE_MIXED; |
6854 | 85.8k | } else { |
6855 | 85.8k | tree = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, 1); |
6856 | 85.8k | res = XML_ELEMENT_TYPE_ELEMENT; |
6857 | 85.8k | } |
6858 | 147k | SKIP_BLANKS_PE; |
6859 | 147k | *result = tree; |
6860 | 147k | return(res); |
6861 | 147k | } |
6862 | | |
6863 | | /** |
6864 | | * xmlParseElementDecl: |
6865 | | * @ctxt: an XML parser context |
6866 | | * |
6867 | | * DEPRECATED: Internal function, don't use. |
6868 | | * |
6869 | | * Parse an element declaration. Always consumes '<!'. |
6870 | | * |
6871 | | * [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>' |
6872 | | * |
6873 | | * [ VC: Unique Element Type Declaration ] |
6874 | | * No element type may be declared more than once |
6875 | | * |
6876 | | * Returns the type of the element, or -1 in case of error |
6877 | | */ |
6878 | | int |
6879 | 216k | xmlParseElementDecl(xmlParserCtxtPtr ctxt) { |
6880 | 216k | const xmlChar *name; |
6881 | 216k | int ret = -1; |
6882 | 216k | xmlElementContentPtr content = NULL; |
6883 | | |
6884 | 216k | if ((CUR != '<') || (NXT(1) != '!')) |
6885 | 0 | return(ret); |
6886 | 216k | SKIP(2); |
6887 | | |
6888 | | /* GROW; done in the caller */ |
6889 | 216k | if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) { |
6890 | 214k | int inputid = ctxt->input->id; |
6891 | | |
6892 | 214k | SKIP(7); |
6893 | 214k | if (SKIP_BLANKS_PE == 0) { |
6894 | 2.40k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6895 | 2.40k | "Space required after 'ELEMENT'\n"); |
6896 | 2.40k | return(-1); |
6897 | 2.40k | } |
6898 | 212k | name = xmlParseName(ctxt); |
6899 | 212k | if (name == NULL) { |
6900 | 4.20k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6901 | 4.20k | "xmlParseElementDecl: no name for Element\n"); |
6902 | 4.20k | return(-1); |
6903 | 4.20k | } |
6904 | 208k | if (SKIP_BLANKS_PE == 0) { |
6905 | 19.7k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6906 | 19.7k | "Space required after the element name\n"); |
6907 | 19.7k | } |
6908 | 208k | if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) { |
6909 | 43.0k | SKIP(5); |
6910 | | /* |
6911 | | * Element must always be empty. |
6912 | | */ |
6913 | 43.0k | ret = XML_ELEMENT_TYPE_EMPTY; |
6914 | 165k | } else if ((RAW == 'A') && (NXT(1) == 'N') && |
6915 | 165k | (NXT(2) == 'Y')) { |
6916 | 9.57k | SKIP(3); |
6917 | | /* |
6918 | | * Element is a generic container. |
6919 | | */ |
6920 | 9.57k | ret = XML_ELEMENT_TYPE_ANY; |
6921 | 155k | } else if (RAW == '(') { |
6922 | 147k | ret = xmlParseElementContentDecl(ctxt, name, &content); |
6923 | 147k | } else { |
6924 | | /* |
6925 | | * [ WFC: PEs in Internal Subset ] error handling. |
6926 | | */ |
6927 | 8.35k | xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6928 | 8.35k | "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n"); |
6929 | 8.35k | return(-1); |
6930 | 8.35k | } |
6931 | | |
6932 | 199k | SKIP_BLANKS_PE; |
6933 | | |
6934 | 199k | if (RAW != '>') { |
6935 | 29.0k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
6936 | 29.0k | if (content != NULL) { |
6937 | 3.80k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6938 | 3.80k | } |
6939 | 170k | } else { |
6940 | 170k | if (inputid != ctxt->input->id) { |
6941 | 453 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6942 | 453 | "Element declaration doesn't start and stop in" |
6943 | 453 | " the same entity\n"); |
6944 | 453 | } |
6945 | | |
6946 | 170k | NEXT; |
6947 | 170k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6948 | 170k | (ctxt->sax->elementDecl != NULL)) { |
6949 | 152k | if (content != NULL) |
6950 | 103k | content->parent = NULL; |
6951 | 152k | ctxt->sax->elementDecl(ctxt->userData, name, ret, |
6952 | 152k | content); |
6953 | 152k | if ((content != NULL) && (content->parent == NULL)) { |
6954 | | /* |
6955 | | * this is a trick: if xmlAddElementDecl is called, |
6956 | | * instead of copying the full tree it is plugged directly |
6957 | | * if called from the parser. Avoid duplicating the |
6958 | | * interfaces or change the API/ABI |
6959 | | */ |
6960 | 26.0k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6961 | 26.0k | } |
6962 | 152k | } else if (content != NULL) { |
6963 | 12.5k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6964 | 12.5k | } |
6965 | 170k | } |
6966 | 199k | } |
6967 | 201k | return(ret); |
6968 | 216k | } |
6969 | | |
6970 | | /** |
6971 | | * xmlParseConditionalSections |
6972 | | * @ctxt: an XML parser context |
6973 | | * |
6974 | | * Parse a conditional section. Always consumes '<!['. |
6975 | | * |
6976 | | * [61] conditionalSect ::= includeSect | ignoreSect |
6977 | | * [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>' |
6978 | | * [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>' |
6979 | | * [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' Ignore)* |
6980 | | * [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*) |
6981 | | */ |
6982 | | |
6983 | | static void |
6984 | 19.1k | xmlParseConditionalSections(xmlParserCtxtPtr ctxt) { |
6985 | 19.1k | int *inputIds = NULL; |
6986 | 19.1k | size_t inputIdsSize = 0; |
6987 | 19.1k | size_t depth = 0; |
6988 | | |
6989 | 53.6k | while (PARSER_STOPPED(ctxt) == 0) { |
6990 | 53.5k | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6991 | 30.2k | int id = ctxt->input->id; |
6992 | | |
6993 | 30.2k | SKIP(3); |
6994 | 30.2k | SKIP_BLANKS_PE; |
6995 | | |
6996 | 30.2k | if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) { |
6997 | 18.7k | SKIP(7); |
6998 | 18.7k | SKIP_BLANKS_PE; |
6999 | 18.7k | if (RAW != '[') { |
7000 | 18 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
7001 | 18 | xmlHaltParser(ctxt); |
7002 | 18 | goto error; |
7003 | 18 | } |
7004 | 18.7k | if (ctxt->input->id != id) { |
7005 | 5 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
7006 | 5 | "All markup of the conditional section is" |
7007 | 5 | " not in the same entity\n"); |
7008 | 5 | } |
7009 | 18.7k | NEXT; |
7010 | | |
7011 | 18.7k | if (inputIdsSize <= depth) { |
7012 | 11.4k | int *tmp; |
7013 | | |
7014 | 11.4k | inputIdsSize = (inputIdsSize == 0 ? 4 : inputIdsSize * 2); |
7015 | 11.4k | tmp = (int *) xmlRealloc(inputIds, |
7016 | 11.4k | inputIdsSize * sizeof(int)); |
7017 | 11.4k | if (tmp == NULL) { |
7018 | 28 | xmlErrMemory(ctxt); |
7019 | 28 | goto error; |
7020 | 28 | } |
7021 | 11.3k | inputIds = tmp; |
7022 | 11.3k | } |
7023 | 18.7k | inputIds[depth] = id; |
7024 | 18.7k | depth++; |
7025 | 18.7k | } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) { |
7026 | 11.2k | size_t ignoreDepth = 0; |
7027 | | |
7028 | 11.2k | SKIP(6); |
7029 | 11.2k | SKIP_BLANKS_PE; |
7030 | 11.2k | if (RAW != '[') { |
7031 | 32 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
7032 | 32 | xmlHaltParser(ctxt); |
7033 | 32 | goto error; |
7034 | 32 | } |
7035 | 11.2k | if (ctxt->input->id != id) { |
7036 | 151 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
7037 | 151 | "All markup of the conditional section is" |
7038 | 151 | " not in the same entity\n"); |
7039 | 151 | } |
7040 | 11.2k | NEXT; |
7041 | | |
7042 | 3.35M | while (PARSER_STOPPED(ctxt) == 0) { |
7043 | 3.35M | if (RAW == 0) { |
7044 | 5.62k | xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL); |
7045 | 5.62k | goto error; |
7046 | 5.62k | } |
7047 | 3.35M | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7048 | 22.4k | SKIP(3); |
7049 | 22.4k | ignoreDepth++; |
7050 | | /* Check for integer overflow */ |
7051 | 22.4k | if (ignoreDepth == 0) { |
7052 | 0 | xmlErrMemory(ctxt); |
7053 | 0 | goto error; |
7054 | 0 | } |
7055 | 3.33M | } else if ((RAW == ']') && (NXT(1) == ']') && |
7056 | 3.33M | (NXT(2) == '>')) { |
7057 | 17.5k | SKIP(3); |
7058 | 17.5k | if (ignoreDepth == 0) |
7059 | 5.55k | break; |
7060 | 11.9k | ignoreDepth--; |
7061 | 3.31M | } else { |
7062 | 3.31M | NEXT; |
7063 | 3.31M | } |
7064 | 3.35M | } |
7065 | | |
7066 | 5.58k | if (ctxt->input->id != id) { |
7067 | 138 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
7068 | 138 | "All markup of the conditional section is" |
7069 | 138 | " not in the same entity\n"); |
7070 | 138 | } |
7071 | 5.58k | } else { |
7072 | 224 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL); |
7073 | 224 | xmlHaltParser(ctxt); |
7074 | 224 | goto error; |
7075 | 224 | } |
7076 | 30.2k | } else if ((depth > 0) && |
7077 | 23.3k | (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
7078 | 12.4k | depth--; |
7079 | 12.4k | if (ctxt->input->id != inputIds[depth]) { |
7080 | 49 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
7081 | 49 | "All markup of the conditional section is not" |
7082 | 49 | " in the same entity\n"); |
7083 | 49 | } |
7084 | 12.4k | SKIP(3); |
7085 | 12.4k | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
7086 | 10.5k | xmlParseMarkupDecl(ctxt); |
7087 | 10.5k | } else { |
7088 | 341 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7089 | 341 | xmlHaltParser(ctxt); |
7090 | 341 | goto error; |
7091 | 341 | } |
7092 | | |
7093 | 47.2k | if (depth == 0) |
7094 | 12.8k | break; |
7095 | | |
7096 | 34.4k | SKIP_BLANKS_PE; |
7097 | 34.4k | SHRINK; |
7098 | 34.4k | GROW; |
7099 | 34.4k | } |
7100 | | |
7101 | 19.1k | error: |
7102 | 19.1k | xmlFree(inputIds); |
7103 | 19.1k | } |
7104 | | |
7105 | | /** |
7106 | | * xmlParseMarkupDecl: |
7107 | | * @ctxt: an XML parser context |
7108 | | * |
7109 | | * DEPRECATED: Internal function, don't use. |
7110 | | * |
7111 | | * Parse markup declarations. Always consumes '<!' or '<?'. |
7112 | | * |
7113 | | * [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl | |
7114 | | * NotationDecl | PI | Comment |
7115 | | * |
7116 | | * [ VC: Proper Declaration/PE Nesting ] |
7117 | | * Parameter-entity replacement text must be properly nested with |
7118 | | * markup declarations. That is to say, if either the first character |
7119 | | * or the last character of a markup declaration (markupdecl above) is |
7120 | | * contained in the replacement text for a parameter-entity reference, |
7121 | | * both must be contained in the same replacement text. |
7122 | | * |
7123 | | * [ WFC: PEs in Internal Subset ] |
7124 | | * In the internal DTD subset, parameter-entity references can occur |
7125 | | * only where markup declarations can occur, not within markup declarations. |
7126 | | * (This does not apply to references that occur in external parameter |
7127 | | * entities or to the external subset.) |
7128 | | */ |
7129 | | void |
7130 | 3.89M | xmlParseMarkupDecl(xmlParserCtxtPtr ctxt) { |
7131 | 3.89M | GROW; |
7132 | 3.89M | if (CUR == '<') { |
7133 | 3.89M | if (NXT(1) == '!') { |
7134 | 3.84M | switch (NXT(2)) { |
7135 | 629k | case 'E': |
7136 | 629k | if (NXT(3) == 'L') |
7137 | 216k | xmlParseElementDecl(ctxt); |
7138 | 413k | else if (NXT(3) == 'N') |
7139 | 413k | xmlParseEntityDecl(ctxt); |
7140 | 454 | else |
7141 | 454 | SKIP(2); |
7142 | 629k | break; |
7143 | 405k | case 'A': |
7144 | 405k | xmlParseAttributeListDecl(ctxt); |
7145 | 405k | break; |
7146 | 43.6k | case 'N': |
7147 | 43.6k | xmlParseNotationDecl(ctxt); |
7148 | 43.6k | break; |
7149 | 2.71M | case '-': |
7150 | 2.71M | xmlParseComment(ctxt); |
7151 | 2.71M | break; |
7152 | 47.7k | default: |
7153 | | /* there is an error but it will be detected later */ |
7154 | 47.7k | SKIP(2); |
7155 | 47.7k | break; |
7156 | 3.84M | } |
7157 | 3.84M | } else if (NXT(1) == '?') { |
7158 | 55.9k | xmlParsePI(ctxt); |
7159 | 55.9k | } |
7160 | 3.89M | } |
7161 | 3.89M | } |
7162 | | |
7163 | | /** |
7164 | | * xmlParseTextDecl: |
7165 | | * @ctxt: an XML parser context |
7166 | | * |
7167 | | * DEPRECATED: Internal function, don't use. |
7168 | | * |
7169 | | * parse an XML declaration header for external entities |
7170 | | * |
7171 | | * [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>' |
7172 | | */ |
7173 | | |
7174 | | void |
7175 | 94.7k | xmlParseTextDecl(xmlParserCtxtPtr ctxt) { |
7176 | 94.7k | xmlChar *version; |
7177 | | |
7178 | | /* |
7179 | | * We know that '<?xml' is here. |
7180 | | */ |
7181 | 94.7k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
7182 | 94.7k | SKIP(5); |
7183 | 94.7k | } else { |
7184 | 34 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL); |
7185 | 34 | return; |
7186 | 34 | } |
7187 | | |
7188 | 94.7k | if (SKIP_BLANKS == 0) { |
7189 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
7190 | 0 | "Space needed after '<?xml'\n"); |
7191 | 0 | } |
7192 | | |
7193 | | /* |
7194 | | * We may have the VersionInfo here. |
7195 | | */ |
7196 | 94.7k | version = xmlParseVersionInfo(ctxt); |
7197 | 94.7k | if (version == NULL) { |
7198 | 69.7k | version = xmlCharStrdup(XML_DEFAULT_VERSION); |
7199 | 69.7k | if (version == NULL) { |
7200 | 68 | xmlErrMemory(ctxt); |
7201 | 68 | return; |
7202 | 68 | } |
7203 | 69.7k | } else { |
7204 | 24.9k | if (SKIP_BLANKS == 0) { |
7205 | 5.00k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
7206 | 5.00k | "Space needed here\n"); |
7207 | 5.00k | } |
7208 | 24.9k | } |
7209 | 94.6k | ctxt->input->version = version; |
7210 | | |
7211 | | /* |
7212 | | * We must have the encoding declaration |
7213 | | */ |
7214 | 94.6k | xmlParseEncodingDecl(ctxt); |
7215 | | |
7216 | 94.6k | SKIP_BLANKS; |
7217 | 94.6k | if ((RAW == '?') && (NXT(1) == '>')) { |
7218 | 6.66k | SKIP(2); |
7219 | 88.0k | } else if (RAW == '>') { |
7220 | | /* Deprecated old WD ... */ |
7221 | 782 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7222 | 782 | NEXT; |
7223 | 87.2k | } else { |
7224 | 87.2k | int c; |
7225 | | |
7226 | 87.2k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7227 | 40.0M | while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) { |
7228 | 40.0M | NEXT; |
7229 | 40.0M | if (c == '>') |
7230 | 46.1k | break; |
7231 | 40.0M | } |
7232 | 87.2k | } |
7233 | 94.6k | } |
7234 | | |
7235 | | /** |
7236 | | * xmlParseExternalSubset: |
7237 | | * @ctxt: an XML parser context |
7238 | | * @ExternalID: the external identifier |
7239 | | * @SystemID: the system identifier (or URL) |
7240 | | * |
7241 | | * parse Markup declarations from an external subset |
7242 | | * |
7243 | | * [30] extSubset ::= textDecl? extSubsetDecl |
7244 | | * |
7245 | | * [31] extSubsetDecl ::= (markupdecl | conditionalSect | PEReference | S) * |
7246 | | */ |
7247 | | void |
7248 | | xmlParseExternalSubset(xmlParserCtxtPtr ctxt, const xmlChar *ExternalID, |
7249 | 10.6k | const xmlChar *SystemID) { |
7250 | 10.6k | int oldInputNr; |
7251 | | |
7252 | 10.6k | xmlCtxtInitializeLate(ctxt); |
7253 | | |
7254 | 10.6k | xmlDetectEncoding(ctxt); |
7255 | | |
7256 | 10.6k | if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) { |
7257 | 555 | xmlParseTextDecl(ctxt); |
7258 | 555 | } |
7259 | 10.6k | if (ctxt->myDoc == NULL) { |
7260 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
7261 | 0 | if (ctxt->myDoc == NULL) { |
7262 | 0 | xmlErrMemory(ctxt); |
7263 | 0 | return; |
7264 | 0 | } |
7265 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
7266 | 0 | } |
7267 | 10.6k | if ((ctxt->myDoc != NULL) && (ctxt->myDoc->intSubset == NULL) && |
7268 | 10.6k | (xmlCreateIntSubset(ctxt->myDoc, NULL, ExternalID, SystemID) == NULL)) { |
7269 | 12 | xmlErrMemory(ctxt); |
7270 | 12 | } |
7271 | | |
7272 | 10.6k | ctxt->inSubset = 2; |
7273 | 10.6k | oldInputNr = ctxt->inputNr; |
7274 | | |
7275 | 10.6k | SKIP_BLANKS_PE; |
7276 | 889k | while (((RAW != 0) || (ctxt->inputNr > oldInputNr)) && |
7277 | 889k | (!PARSER_STOPPED(ctxt))) { |
7278 | 883k | GROW; |
7279 | 883k | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7280 | 4.72k | xmlParseConditionalSections(ctxt); |
7281 | 878k | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
7282 | 874k | xmlParseMarkupDecl(ctxt); |
7283 | 874k | } else { |
7284 | 3.98k | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7285 | 3.98k | xmlHaltParser(ctxt); |
7286 | 3.98k | return; |
7287 | 3.98k | } |
7288 | 879k | SKIP_BLANKS_PE; |
7289 | 879k | SHRINK; |
7290 | 879k | } |
7291 | | |
7292 | 7.23k | while (ctxt->inputNr > oldInputNr) |
7293 | 599 | xmlPopPE(ctxt); |
7294 | | |
7295 | 6.63k | if (RAW != 0) { |
7296 | 1.39k | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7297 | 1.39k | } |
7298 | 6.63k | } |
7299 | | |
7300 | | /** |
7301 | | * xmlParseReference: |
7302 | | * @ctxt: an XML parser context |
7303 | | * |
7304 | | * DEPRECATED: Internal function, don't use. |
7305 | | * |
7306 | | * parse and handle entity references in content, depending on the SAX |
7307 | | * interface, this may end-up in a call to character() if this is a |
7308 | | * CharRef, a predefined entity, if there is no reference() callback. |
7309 | | * or if the parser was asked to switch to that mode. |
7310 | | * |
7311 | | * Always consumes '&'. |
7312 | | * |
7313 | | * [67] Reference ::= EntityRef | CharRef |
7314 | | */ |
7315 | | void |
7316 | 2.74M | xmlParseReference(xmlParserCtxtPtr ctxt) { |
7317 | 2.74M | xmlEntityPtr ent = NULL; |
7318 | 2.74M | const xmlChar *name; |
7319 | 2.74M | xmlChar *val; |
7320 | | |
7321 | 2.74M | if (RAW != '&') |
7322 | 0 | return; |
7323 | | |
7324 | | /* |
7325 | | * Simple case of a CharRef |
7326 | | */ |
7327 | 2.74M | if (NXT(1) == '#') { |
7328 | 752k | int i = 0; |
7329 | 752k | xmlChar out[16]; |
7330 | 752k | int value = xmlParseCharRef(ctxt); |
7331 | | |
7332 | 752k | if (value == 0) |
7333 | 291k | return; |
7334 | | |
7335 | | /* |
7336 | | * Just encode the value in UTF-8 |
7337 | | */ |
7338 | 460k | COPY_BUF(out, i, value); |
7339 | 460k | out[i] = 0; |
7340 | 460k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7341 | 460k | (!ctxt->disableSAX)) |
7342 | 413k | ctxt->sax->characters(ctxt->userData, out, i); |
7343 | 460k | return; |
7344 | 752k | } |
7345 | | |
7346 | | /* |
7347 | | * We are seeing an entity reference |
7348 | | */ |
7349 | 1.99M | name = xmlParseEntityRefInternal(ctxt); |
7350 | 1.99M | if (name == NULL) |
7351 | 908k | return; |
7352 | 1.08M | ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0); |
7353 | 1.08M | if (ent == NULL) { |
7354 | | /* |
7355 | | * Create a reference for undeclared entities. |
7356 | | */ |
7357 | 583k | if ((ctxt->replaceEntities == 0) && |
7358 | 583k | (ctxt->sax != NULL) && |
7359 | 583k | (ctxt->disableSAX == 0) && |
7360 | 583k | (ctxt->sax->reference != NULL)) { |
7361 | 306k | ctxt->sax->reference(ctxt->userData, name); |
7362 | 306k | } |
7363 | 583k | return; |
7364 | 583k | } |
7365 | 499k | if (!ctxt->wellFormed) |
7366 | 344k | return; |
7367 | | |
7368 | | /* special case of predefined entities */ |
7369 | 155k | if ((ent->name == NULL) || |
7370 | 155k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
7371 | 40.0k | val = ent->content; |
7372 | 40.0k | if (val == NULL) return; |
7373 | | /* |
7374 | | * inline the entity. |
7375 | | */ |
7376 | 40.0k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7377 | 40.0k | (!ctxt->disableSAX)) |
7378 | 40.0k | ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val)); |
7379 | 40.0k | return; |
7380 | 40.0k | } |
7381 | | |
7382 | | /* |
7383 | | * The first reference to the entity trigger a parsing phase |
7384 | | * where the ent->children is filled with the result from |
7385 | | * the parsing. |
7386 | | * Note: external parsed entities will not be loaded, it is not |
7387 | | * required for a non-validating parser, unless the parsing option |
7388 | | * of validating, or substituting entities were given. Doing so is |
7389 | | * far more secure as the parser will only process data coming from |
7390 | | * the document entity by default. |
7391 | | * |
7392 | | * FIXME: This doesn't work correctly since entities can be |
7393 | | * expanded with different namespace declarations in scope. |
7394 | | * For example: |
7395 | | * |
7396 | | * <!DOCTYPE doc [ |
7397 | | * <!ENTITY ent "<ns:elem/>"> |
7398 | | * ]> |
7399 | | * <doc> |
7400 | | * <decl1 xmlns:ns="urn:ns1"> |
7401 | | * &ent; |
7402 | | * </decl1> |
7403 | | * <decl2 xmlns:ns="urn:ns2"> |
7404 | | * &ent; |
7405 | | * </decl2> |
7406 | | * </doc> |
7407 | | * |
7408 | | * Proposed fix: |
7409 | | * |
7410 | | * - Ignore current namespace declarations when parsing the |
7411 | | * entity. If a prefix can't be resolved, don't report an error |
7412 | | * but mark it as unresolved. |
7413 | | * - Try to resolve these prefixes when expanding the entity. |
7414 | | * This will require a specialized version of xmlStaticCopyNode |
7415 | | * which can also make use of the namespace hash table to avoid |
7416 | | * quadratic behavior. |
7417 | | * |
7418 | | * Alternatively, we could simply reparse the entity on each |
7419 | | * expansion like we already do with custom SAX callbacks. |
7420 | | * External entity content should be cached in this case. |
7421 | | */ |
7422 | 115k | if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) || |
7423 | 115k | (((ctxt->options & XML_PARSE_NO_XXE) == 0) && |
7424 | 34.1k | ((ctxt->replaceEntities) || |
7425 | 112k | (ctxt->validate)))) { |
7426 | 112k | if ((ent->flags & XML_ENT_PARSED) == 0) { |
7427 | 23.4k | xmlCtxtParseEntity(ctxt, ent); |
7428 | 89.0k | } else if (ent->children == NULL) { |
7429 | | /* |
7430 | | * Probably running in SAX mode and the callbacks don't |
7431 | | * build the entity content. Parse the entity again. |
7432 | | * |
7433 | | * This will also be triggered in normal tree builder mode |
7434 | | * if an entity happens to be empty, causing unnecessary |
7435 | | * reloads. It's hard to come up with a reliable check in |
7436 | | * which mode we're running. |
7437 | | */ |
7438 | 46.6k | xmlCtxtParseEntity(ctxt, ent); |
7439 | 46.6k | } |
7440 | 112k | } |
7441 | | |
7442 | | /* |
7443 | | * We also check for amplification if entities aren't substituted. |
7444 | | * They might be expanded later. |
7445 | | */ |
7446 | 115k | if (xmlParserEntityCheck(ctxt, ent->expandedSize)) |
7447 | 435 | return; |
7448 | | |
7449 | 114k | if ((ctxt->sax == NULL) || (ctxt->disableSAX)) |
7450 | 7.76k | return; |
7451 | | |
7452 | 107k | if (ctxt->replaceEntities == 0) { |
7453 | | /* |
7454 | | * Create a reference |
7455 | | */ |
7456 | 21.6k | if (ctxt->sax->reference != NULL) |
7457 | 21.6k | ctxt->sax->reference(ctxt->userData, ent->name); |
7458 | 85.4k | } else if ((ent->children != NULL) && (ctxt->node != NULL)) { |
7459 | 39.4k | xmlNodePtr copy, cur; |
7460 | | |
7461 | | /* |
7462 | | * Seems we are generating the DOM content, copy the tree |
7463 | | */ |
7464 | 39.4k | cur = ent->children; |
7465 | | |
7466 | | /* |
7467 | | * Handle first text node with SAX to coalesce text efficiently |
7468 | | */ |
7469 | 39.4k | if ((cur->type == XML_TEXT_NODE) || |
7470 | 39.4k | (cur->type == XML_CDATA_SECTION_NODE)) { |
7471 | 30.8k | int len = xmlStrlen(cur->content); |
7472 | | |
7473 | 30.8k | if ((cur->type == XML_TEXT_NODE) || |
7474 | 30.8k | (ctxt->sax->cdataBlock == NULL)) { |
7475 | 29.9k | if (ctxt->sax->characters != NULL) |
7476 | 29.9k | ctxt->sax->characters(ctxt, cur->content, len); |
7477 | 29.9k | } else { |
7478 | 892 | if (ctxt->sax->cdataBlock != NULL) |
7479 | 892 | ctxt->sax->cdataBlock(ctxt, cur->content, len); |
7480 | 892 | } |
7481 | | |
7482 | 30.8k | cur = cur->next; |
7483 | 30.8k | } |
7484 | | |
7485 | 149k | while (cur != NULL) { |
7486 | 115k | xmlNodePtr last; |
7487 | | |
7488 | | /* |
7489 | | * Handle last text node with SAX to coalesce text efficiently |
7490 | | */ |
7491 | 115k | if ((cur->next == NULL) && |
7492 | 115k | ((cur->type == XML_TEXT_NODE) || |
7493 | 21.4k | (cur->type == XML_CDATA_SECTION_NODE))) { |
7494 | 5.23k | int len = xmlStrlen(cur->content); |
7495 | | |
7496 | 5.23k | if ((cur->type == XML_TEXT_NODE) || |
7497 | 5.23k | (ctxt->sax->cdataBlock == NULL)) { |
7498 | 4.62k | if (ctxt->sax->characters != NULL) |
7499 | 4.62k | ctxt->sax->characters(ctxt, cur->content, len); |
7500 | 4.62k | } else { |
7501 | 609 | if (ctxt->sax->cdataBlock != NULL) |
7502 | 609 | ctxt->sax->cdataBlock(ctxt, cur->content, len); |
7503 | 609 | } |
7504 | | |
7505 | 5.23k | break; |
7506 | 5.23k | } |
7507 | | |
7508 | | /* |
7509 | | * Reset coalesce buffer stats only for non-text nodes. |
7510 | | */ |
7511 | 110k | ctxt->nodemem = 0; |
7512 | 110k | ctxt->nodelen = 0; |
7513 | | |
7514 | 110k | copy = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7515 | | |
7516 | 110k | if (copy == NULL) { |
7517 | 739 | xmlErrMemory(ctxt); |
7518 | 739 | break; |
7519 | 739 | } |
7520 | | |
7521 | 109k | if (ctxt->parseMode == XML_PARSE_READER) { |
7522 | | /* Needed for reader */ |
7523 | 22.7k | copy->extra = cur->extra; |
7524 | | /* Maybe needed for reader */ |
7525 | 22.7k | copy->_private = cur->_private; |
7526 | 22.7k | } |
7527 | | |
7528 | 109k | copy->parent = ctxt->node; |
7529 | 109k | last = ctxt->node->last; |
7530 | 109k | if (last == NULL) { |
7531 | 2.50k | ctxt->node->children = copy; |
7532 | 107k | } else { |
7533 | 107k | last->next = copy; |
7534 | 107k | copy->prev = last; |
7535 | 107k | } |
7536 | 109k | ctxt->node->last = copy; |
7537 | | |
7538 | 109k | cur = cur->next; |
7539 | 109k | } |
7540 | 39.4k | } |
7541 | 107k | } |
7542 | | |
7543 | | static void |
7544 | 2.19M | xmlHandleUndeclaredEntity(xmlParserCtxtPtr ctxt, const xmlChar *name) { |
7545 | | /* |
7546 | | * [ WFC: Entity Declared ] |
7547 | | * In a document without any DTD, a document with only an |
7548 | | * internal DTD subset which contains no parameter entity |
7549 | | * references, or a document with "standalone='yes'", the |
7550 | | * Name given in the entity reference must match that in an |
7551 | | * entity declaration, except that well-formed documents |
7552 | | * need not declare any of the following entities: amp, lt, |
7553 | | * gt, apos, quot. |
7554 | | * The declaration of a parameter entity must precede any |
7555 | | * reference to it. |
7556 | | * Similarly, the declaration of a general entity must |
7557 | | * precede any reference to it which appears in a default |
7558 | | * value in an attribute-list declaration. Note that if |
7559 | | * entities are declared in the external subset or in |
7560 | | * external parameter entities, a non-validating processor |
7561 | | * is not obligated to read and process their declarations; |
7562 | | * for such documents, the rule that an entity must be |
7563 | | * declared is a well-formedness constraint only if |
7564 | | * standalone='yes'. |
7565 | | */ |
7566 | 2.19M | if ((ctxt->standalone == 1) || |
7567 | 2.19M | ((ctxt->hasExternalSubset == 0) && |
7568 | 2.16M | (ctxt->hasPErefs == 0))) { |
7569 | 1.27M | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7570 | 1.27M | "Entity '%s' not defined\n", name); |
7571 | 1.27M | } else if (ctxt->validate) { |
7572 | | /* |
7573 | | * [ VC: Entity Declared ] |
7574 | | * In a document with an external subset or external |
7575 | | * parameter entities with "standalone='no'", ... |
7576 | | * ... The declaration of a parameter entity must |
7577 | | * precede any reference to it... |
7578 | | */ |
7579 | 158k | xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7580 | 158k | "Entity '%s' not defined\n", name, NULL); |
7581 | 758k | } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) || |
7582 | 758k | ((ctxt->replaceEntities) && |
7583 | 741k | ((ctxt->options & XML_PARSE_NO_XXE) == 0))) { |
7584 | | /* |
7585 | | * Also raise a non-fatal error |
7586 | | * |
7587 | | * - if the external subset is loaded and all entity declarations |
7588 | | * should be available, or |
7589 | | * - entity substition was requested without restricting |
7590 | | * external entity access. |
7591 | | */ |
7592 | 741k | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7593 | 741k | "Entity '%s' not defined\n", name); |
7594 | 741k | } else { |
7595 | 16.8k | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7596 | 16.8k | "Entity '%s' not defined\n", name, NULL); |
7597 | 16.8k | } |
7598 | | |
7599 | 2.19M | ctxt->valid = 0; |
7600 | 2.19M | } |
7601 | | |
7602 | | static xmlEntityPtr |
7603 | 13.6M | xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) { |
7604 | 13.6M | xmlEntityPtr ent; |
7605 | | |
7606 | | /* |
7607 | | * Predefined entities override any extra definition |
7608 | | */ |
7609 | 13.6M | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7610 | 11.0M | ent = xmlGetPredefinedEntity(name); |
7611 | 11.0M | if (ent != NULL) |
7612 | 3.04M | return(ent); |
7613 | 11.0M | } |
7614 | | |
7615 | | /* |
7616 | | * Ask first SAX for entity resolution, otherwise try the |
7617 | | * entities which may have stored in the parser context. |
7618 | | */ |
7619 | 10.5M | if (ctxt->sax != NULL) { |
7620 | 10.5M | if (ctxt->sax->getEntity != NULL) |
7621 | 10.5M | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7622 | 10.5M | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7623 | 10.5M | (ctxt->options & XML_PARSE_OLDSAX)) |
7624 | 67.5k | ent = xmlGetPredefinedEntity(name); |
7625 | 10.5M | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7626 | 10.5M | (ctxt->userData==ctxt)) { |
7627 | 314k | ent = xmlSAX2GetEntity(ctxt, name); |
7628 | 314k | } |
7629 | 10.5M | } |
7630 | | |
7631 | 10.5M | if (ent == NULL) { |
7632 | 1.45M | xmlHandleUndeclaredEntity(ctxt, name); |
7633 | 1.45M | } |
7634 | | |
7635 | | /* |
7636 | | * [ WFC: Parsed Entity ] |
7637 | | * An entity reference must not contain the name of an |
7638 | | * unparsed entity |
7639 | | */ |
7640 | 9.12M | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7641 | 1.44k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7642 | 1.44k | "Entity reference to unparsed entity %s\n", name); |
7643 | 1.44k | ent = NULL; |
7644 | 1.44k | } |
7645 | | |
7646 | | /* |
7647 | | * [ WFC: No External Entity References ] |
7648 | | * Attribute values cannot contain direct or indirect |
7649 | | * entity references to external entities. |
7650 | | */ |
7651 | 9.12M | else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7652 | 213k | if (inAttr) { |
7653 | 67.5k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7654 | 67.5k | "Attribute references external entity '%s'\n", name); |
7655 | 67.5k | ent = NULL; |
7656 | 67.5k | } |
7657 | 213k | } |
7658 | | |
7659 | 10.5M | return(ent); |
7660 | 13.6M | } |
7661 | | |
7662 | | /** |
7663 | | * xmlParseEntityRefInternal: |
7664 | | * @ctxt: an XML parser context |
7665 | | * @inAttr: whether we are in an attribute value |
7666 | | * |
7667 | | * Parse an entity reference. Always consumes '&'. |
7668 | | * |
7669 | | * [68] EntityRef ::= '&' Name ';' |
7670 | | * |
7671 | | * Returns the name, or NULL in case of error. |
7672 | | */ |
7673 | | static const xmlChar * |
7674 | 6.50M | xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) { |
7675 | 6.50M | const xmlChar *name; |
7676 | | |
7677 | 6.50M | GROW; |
7678 | | |
7679 | 6.50M | if (RAW != '&') |
7680 | 0 | return(NULL); |
7681 | 6.50M | NEXT; |
7682 | 6.50M | name = xmlParseName(ctxt); |
7683 | 6.50M | if (name == NULL) { |
7684 | 902k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7685 | 902k | "xmlParseEntityRef: no name\n"); |
7686 | 902k | return(NULL); |
7687 | 902k | } |
7688 | 5.59M | if (RAW != ';') { |
7689 | 514k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7690 | 514k | return(NULL); |
7691 | 514k | } |
7692 | 5.08M | NEXT; |
7693 | | |
7694 | 5.08M | return(name); |
7695 | 5.59M | } |
7696 | | |
7697 | | /** |
7698 | | * xmlParseEntityRef: |
7699 | | * @ctxt: an XML parser context |
7700 | | * |
7701 | | * DEPRECATED: Internal function, don't use. |
7702 | | * |
7703 | | * Returns the xmlEntityPtr if found, or NULL otherwise. |
7704 | | */ |
7705 | | xmlEntityPtr |
7706 | 0 | xmlParseEntityRef(xmlParserCtxtPtr ctxt) { |
7707 | 0 | const xmlChar *name; |
7708 | |
|
7709 | 0 | if (ctxt == NULL) |
7710 | 0 | return(NULL); |
7711 | | |
7712 | 0 | name = xmlParseEntityRefInternal(ctxt); |
7713 | 0 | if (name == NULL) |
7714 | 0 | return(NULL); |
7715 | | |
7716 | 0 | return(xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 0)); |
7717 | 0 | } |
7718 | | |
7719 | | /** |
7720 | | * xmlParseStringEntityRef: |
7721 | | * @ctxt: an XML parser context |
7722 | | * @str: a pointer to an index in the string |
7723 | | * |
7724 | | * parse ENTITY references declarations, but this version parses it from |
7725 | | * a string value. |
7726 | | * |
7727 | | * [68] EntityRef ::= '&' Name ';' |
7728 | | * |
7729 | | * [ WFC: Entity Declared ] |
7730 | | * In a document without any DTD, a document with only an internal DTD |
7731 | | * subset which contains no parameter entity references, or a document |
7732 | | * with "standalone='yes'", the Name given in the entity reference |
7733 | | * must match that in an entity declaration, except that well-formed |
7734 | | * documents need not declare any of the following entities: amp, lt, |
7735 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7736 | | * any reference to it. Similarly, the declaration of a general entity |
7737 | | * must precede any reference to it which appears in a default value in an |
7738 | | * attribute-list declaration. Note that if entities are declared in the |
7739 | | * external subset or in external parameter entities, a non-validating |
7740 | | * processor is not obligated to read and process their declarations; |
7741 | | * for such documents, the rule that an entity must be declared is a |
7742 | | * well-formedness constraint only if standalone='yes'. |
7743 | | * |
7744 | | * [ WFC: Parsed Entity ] |
7745 | | * An entity reference must not contain the name of an unparsed entity |
7746 | | * |
7747 | | * Returns the xmlEntityPtr if found, or NULL otherwise. The str pointer |
7748 | | * is updated to the current location in the string. |
7749 | | */ |
7750 | | static xmlChar * |
7751 | 8.54M | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) { |
7752 | 8.54M | xmlChar *name; |
7753 | 8.54M | const xmlChar *ptr; |
7754 | 8.54M | xmlChar cur; |
7755 | | |
7756 | 8.54M | if ((str == NULL) || (*str == NULL)) |
7757 | 0 | return(NULL); |
7758 | 8.54M | ptr = *str; |
7759 | 8.54M | cur = *ptr; |
7760 | 8.54M | if (cur != '&') |
7761 | 0 | return(NULL); |
7762 | | |
7763 | 8.54M | ptr++; |
7764 | 8.54M | name = xmlParseStringName(ctxt, &ptr); |
7765 | 8.54M | if (name == NULL) { |
7766 | 207 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7767 | 207 | "xmlParseStringEntityRef: no name\n"); |
7768 | 207 | *str = ptr; |
7769 | 207 | return(NULL); |
7770 | 207 | } |
7771 | 8.54M | if (*ptr != ';') { |
7772 | 102 | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7773 | 102 | xmlFree(name); |
7774 | 102 | *str = ptr; |
7775 | 102 | return(NULL); |
7776 | 102 | } |
7777 | 8.54M | ptr++; |
7778 | | |
7779 | 8.54M | *str = ptr; |
7780 | 8.54M | return(name); |
7781 | 8.54M | } |
7782 | | |
7783 | | /** |
7784 | | * xmlParsePEReference: |
7785 | | * @ctxt: an XML parser context |
7786 | | * |
7787 | | * DEPRECATED: Internal function, don't use. |
7788 | | * |
7789 | | * Parse a parameter entity reference. Always consumes '%'. |
7790 | | * |
7791 | | * The entity content is handled directly by pushing it's content as |
7792 | | * a new input stream. |
7793 | | * |
7794 | | * [69] PEReference ::= '%' Name ';' |
7795 | | * |
7796 | | * [ WFC: No Recursion ] |
7797 | | * A parsed entity must not contain a recursive |
7798 | | * reference to itself, either directly or indirectly. |
7799 | | * |
7800 | | * [ WFC: Entity Declared ] |
7801 | | * In a document without any DTD, a document with only an internal DTD |
7802 | | * subset which contains no parameter entity references, or a document |
7803 | | * with "standalone='yes'", ... ... The declaration of a parameter |
7804 | | * entity must precede any reference to it... |
7805 | | * |
7806 | | * [ VC: Entity Declared ] |
7807 | | * In a document with an external subset or external parameter entities |
7808 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
7809 | | * must precede any reference to it... |
7810 | | * |
7811 | | * [ WFC: In DTD ] |
7812 | | * Parameter-entity references may only appear in the DTD. |
7813 | | * NOTE: misleading but this is handled. |
7814 | | */ |
7815 | | void |
7816 | | xmlParsePEReference(xmlParserCtxtPtr ctxt) |
7817 | 2.49M | { |
7818 | 2.49M | const xmlChar *name; |
7819 | 2.49M | xmlEntityPtr entity = NULL; |
7820 | 2.49M | xmlParserInputPtr input; |
7821 | | |
7822 | 2.49M | if (RAW != '%') |
7823 | 0 | return; |
7824 | 2.49M | NEXT; |
7825 | 2.49M | name = xmlParseName(ctxt); |
7826 | 2.49M | if (name == NULL) { |
7827 | 44.1k | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n"); |
7828 | 44.1k | return; |
7829 | 44.1k | } |
7830 | 2.45M | if (RAW != ';') { |
7831 | 77.3k | xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL); |
7832 | 77.3k | return; |
7833 | 77.3k | } |
7834 | | |
7835 | 2.37M | NEXT; |
7836 | | |
7837 | | /* Must be set before xmlHandleUndeclaredEntity */ |
7838 | 2.37M | ctxt->hasPErefs = 1; |
7839 | | |
7840 | | /* |
7841 | | * Request the entity from SAX |
7842 | | */ |
7843 | 2.37M | if ((ctxt->sax != NULL) && |
7844 | 2.37M | (ctxt->sax->getParameterEntity != NULL)) |
7845 | 2.37M | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
7846 | | |
7847 | 2.37M | if (entity == NULL) { |
7848 | 720k | xmlHandleUndeclaredEntity(ctxt, name); |
7849 | 1.65M | } else { |
7850 | | /* |
7851 | | * Internal checking in case the entity quest barfed |
7852 | | */ |
7853 | 1.65M | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
7854 | 1.65M | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
7855 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7856 | 0 | "Internal: %%%s; is not a parameter entity\n", |
7857 | 0 | name, NULL); |
7858 | 1.65M | } else { |
7859 | 1.65M | if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
7860 | 1.65M | ((ctxt->options & XML_PARSE_NO_XXE) || |
7861 | 1.38M | ((ctxt->loadsubset == 0) && |
7862 | 1.38M | (ctxt->replaceEntities == 0) && |
7863 | 1.38M | (ctxt->validate == 0)))) |
7864 | 1.88k | return; |
7865 | | |
7866 | 1.65M | if (entity->flags & XML_ENT_EXPANDING) { |
7867 | 110 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7868 | 110 | xmlHaltParser(ctxt); |
7869 | 110 | return; |
7870 | 110 | } |
7871 | | |
7872 | 1.65M | if (ctxt->input_id >= INT_MAX) { |
7873 | 0 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, |
7874 | 0 | "Input ID overflow\n"); |
7875 | 0 | return; |
7876 | 0 | } |
7877 | | |
7878 | 1.65M | input = xmlNewEntityInputStream(ctxt, entity); |
7879 | 1.65M | if (xmlPushInput(ctxt, input) < 0) { |
7880 | 985k | xmlFreeInputStream(input); |
7881 | 985k | return; |
7882 | 985k | } |
7883 | | |
7884 | 669k | input->id = ++ctxt->input_id; |
7885 | | |
7886 | 669k | entity->flags |= XML_ENT_EXPANDING; |
7887 | | |
7888 | 669k | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
7889 | 398k | xmlDetectEncoding(ctxt); |
7890 | | |
7891 | 398k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
7892 | 398k | (IS_BLANK_CH(NXT(5)))) { |
7893 | 77.1k | xmlParseTextDecl(ctxt); |
7894 | 77.1k | } |
7895 | 398k | } |
7896 | 669k | } |
7897 | 1.65M | } |
7898 | 2.37M | } |
7899 | | |
7900 | | /** |
7901 | | * xmlLoadEntityContent: |
7902 | | * @ctxt: an XML parser context |
7903 | | * @entity: an unloaded system entity |
7904 | | * |
7905 | | * Load the content of an entity. |
7906 | | * |
7907 | | * Returns 0 in case of success and -1 in case of failure |
7908 | | */ |
7909 | | static int |
7910 | 36.8k | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) { |
7911 | 36.8k | xmlParserInputPtr oldinput, input = NULL; |
7912 | 36.8k | xmlParserInputPtr *oldinputTab; |
7913 | 36.8k | const xmlChar *oldencoding; |
7914 | 36.8k | xmlChar *content = NULL; |
7915 | 36.8k | xmlResourceType rtype; |
7916 | 36.8k | size_t length, i; |
7917 | 36.8k | int oldinputNr, oldinputMax; |
7918 | 36.8k | int ret = -1; |
7919 | 36.8k | int res; |
7920 | | |
7921 | 36.8k | if ((ctxt == NULL) || (entity == NULL) || |
7922 | 36.8k | ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) && |
7923 | 36.8k | (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) || |
7924 | 36.8k | (entity->content != NULL)) { |
7925 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, |
7926 | 0 | "xmlLoadEntityContent parameter error"); |
7927 | 0 | return(-1); |
7928 | 0 | } |
7929 | | |
7930 | 36.8k | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) |
7931 | 36.8k | rtype = XML_RESOURCE_PARAMETER_ENTITY; |
7932 | 0 | else |
7933 | 0 | rtype = XML_RESOURCE_GENERAL_ENTITY; |
7934 | | |
7935 | 36.8k | input = xmlLoadResource(ctxt, (char *) entity->URI, |
7936 | 36.8k | (char *) entity->ExternalID, rtype); |
7937 | 36.8k | if (input == NULL) |
7938 | 4.23k | return(-1); |
7939 | | |
7940 | 32.5k | oldinput = ctxt->input; |
7941 | 32.5k | oldinputNr = ctxt->inputNr; |
7942 | 32.5k | oldinputMax = ctxt->inputMax; |
7943 | 32.5k | oldinputTab = ctxt->inputTab; |
7944 | 32.5k | oldencoding = ctxt->encoding; |
7945 | | |
7946 | 32.5k | ctxt->input = NULL; |
7947 | 32.5k | ctxt->inputNr = 0; |
7948 | 32.5k | ctxt->inputMax = 1; |
7949 | 32.5k | ctxt->encoding = NULL; |
7950 | 32.5k | ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr)); |
7951 | 32.5k | if (ctxt->inputTab == NULL) { |
7952 | 15 | xmlErrMemory(ctxt); |
7953 | 15 | xmlFreeInputStream(input); |
7954 | 15 | goto error; |
7955 | 15 | } |
7956 | | |
7957 | 32.5k | xmlBufResetInput(input->buf->buffer, input); |
7958 | | |
7959 | 32.5k | if (inputPush(ctxt, input) < 0) { |
7960 | 37 | xmlFreeInputStream(input); |
7961 | 37 | goto error; |
7962 | 37 | } |
7963 | | |
7964 | 32.5k | xmlDetectEncoding(ctxt); |
7965 | | |
7966 | | /* |
7967 | | * Parse a possible text declaration first |
7968 | | */ |
7969 | 32.5k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
7970 | 15.0k | xmlParseTextDecl(ctxt); |
7971 | | /* |
7972 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
7973 | | */ |
7974 | 15.0k | if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) && |
7975 | 15.0k | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
7976 | 2.35k | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
7977 | 2.35k | "Version mismatch between document and entity\n"); |
7978 | 2.35k | } |
7979 | 15.0k | } |
7980 | | |
7981 | 32.5k | length = input->cur - input->base; |
7982 | 32.5k | xmlBufShrink(input->buf->buffer, length); |
7983 | 32.5k | xmlSaturatedAdd(&ctxt->sizeentities, length); |
7984 | | |
7985 | 58.9k | while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0) |
7986 | 26.4k | ; |
7987 | | |
7988 | 32.5k | xmlBufResetInput(input->buf->buffer, input); |
7989 | | |
7990 | 32.5k | if (res < 0) { |
7991 | 3.63k | xmlCtxtErrIO(ctxt, input->buf->error, NULL); |
7992 | 3.63k | goto error; |
7993 | 3.63k | } |
7994 | | |
7995 | 28.8k | length = xmlBufUse(input->buf->buffer); |
7996 | 28.8k | if (length > INT_MAX) { |
7997 | 0 | xmlErrMemory(ctxt); |
7998 | 0 | goto error; |
7999 | 0 | } |
8000 | | |
8001 | 28.8k | content = xmlStrndup(xmlBufContent(input->buf->buffer), length); |
8002 | 28.8k | if (content == NULL) { |
8003 | 46 | xmlErrMemory(ctxt); |
8004 | 46 | goto error; |
8005 | 46 | } |
8006 | | |
8007 | 13.7M | for (i = 0; i < length; ) { |
8008 | 13.7M | int clen = length - i; |
8009 | 13.7M | int c = xmlGetUTF8Char(content + i, &clen); |
8010 | | |
8011 | 13.7M | if ((c < 0) || (!IS_CHAR(c))) { |
8012 | 28.2k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
8013 | 28.2k | "xmlLoadEntityContent: invalid char value %d\n", |
8014 | 28.2k | content[i]); |
8015 | 28.2k | goto error; |
8016 | 28.2k | } |
8017 | 13.7M | i += clen; |
8018 | 13.7M | } |
8019 | | |
8020 | 576 | xmlSaturatedAdd(&ctxt->sizeentities, length); |
8021 | 576 | entity->content = content; |
8022 | 576 | entity->length = length; |
8023 | 576 | content = NULL; |
8024 | 576 | ret = 0; |
8025 | | |
8026 | 32.5k | error: |
8027 | 65.1k | while (ctxt->inputNr > 0) |
8028 | 32.5k | xmlFreeInputStream(inputPop(ctxt)); |
8029 | 32.5k | xmlFree(ctxt->inputTab); |
8030 | 32.5k | xmlFree((xmlChar *) ctxt->encoding); |
8031 | | |
8032 | 32.5k | ctxt->input = oldinput; |
8033 | 32.5k | ctxt->inputNr = oldinputNr; |
8034 | 32.5k | ctxt->inputMax = oldinputMax; |
8035 | 32.5k | ctxt->inputTab = oldinputTab; |
8036 | 32.5k | ctxt->encoding = oldencoding; |
8037 | | |
8038 | 32.5k | xmlFree(content); |
8039 | | |
8040 | 32.5k | return(ret); |
8041 | 576 | } |
8042 | | |
8043 | | /** |
8044 | | * xmlParseStringPEReference: |
8045 | | * @ctxt: an XML parser context |
8046 | | * @str: a pointer to an index in the string |
8047 | | * |
8048 | | * parse PEReference declarations |
8049 | | * |
8050 | | * [69] PEReference ::= '%' Name ';' |
8051 | | * |
8052 | | * [ WFC: No Recursion ] |
8053 | | * A parsed entity must not contain a recursive |
8054 | | * reference to itself, either directly or indirectly. |
8055 | | * |
8056 | | * [ WFC: Entity Declared ] |
8057 | | * In a document without any DTD, a document with only an internal DTD |
8058 | | * subset which contains no parameter entity references, or a document |
8059 | | * with "standalone='yes'", ... ... The declaration of a parameter |
8060 | | * entity must precede any reference to it... |
8061 | | * |
8062 | | * [ VC: Entity Declared ] |
8063 | | * In a document with an external subset or external parameter entities |
8064 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
8065 | | * must precede any reference to it... |
8066 | | * |
8067 | | * [ WFC: In DTD ] |
8068 | | * Parameter-entity references may only appear in the DTD. |
8069 | | * NOTE: misleading but this is handled. |
8070 | | * |
8071 | | * Returns the string of the entity content. |
8072 | | * str is updated to the current value of the index |
8073 | | */ |
8074 | | static xmlEntityPtr |
8075 | 146k | xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
8076 | 146k | const xmlChar *ptr; |
8077 | 146k | xmlChar cur; |
8078 | 146k | xmlChar *name; |
8079 | 146k | xmlEntityPtr entity = NULL; |
8080 | | |
8081 | 146k | if ((str == NULL) || (*str == NULL)) return(NULL); |
8082 | 146k | ptr = *str; |
8083 | 146k | cur = *ptr; |
8084 | 146k | if (cur != '%') |
8085 | 0 | return(NULL); |
8086 | 146k | ptr++; |
8087 | 146k | name = xmlParseStringName(ctxt, &ptr); |
8088 | 146k | if (name == NULL) { |
8089 | 7.86k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8090 | 7.86k | "xmlParseStringPEReference: no name\n"); |
8091 | 7.86k | *str = ptr; |
8092 | 7.86k | return(NULL); |
8093 | 7.86k | } |
8094 | 138k | cur = *ptr; |
8095 | 138k | if (cur != ';') { |
8096 | 16.2k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
8097 | 16.2k | xmlFree(name); |
8098 | 16.2k | *str = ptr; |
8099 | 16.2k | return(NULL); |
8100 | 16.2k | } |
8101 | 122k | ptr++; |
8102 | | |
8103 | | /* Must be set before xmlHandleUndeclaredEntity */ |
8104 | 122k | ctxt->hasPErefs = 1; |
8105 | | |
8106 | | /* |
8107 | | * Request the entity from SAX |
8108 | | */ |
8109 | 122k | if ((ctxt->sax != NULL) && |
8110 | 122k | (ctxt->sax->getParameterEntity != NULL)) |
8111 | 122k | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
8112 | | |
8113 | 122k | if (entity == NULL) { |
8114 | 11.3k | xmlHandleUndeclaredEntity(ctxt, name); |
8115 | 110k | } else { |
8116 | | /* |
8117 | | * Internal checking in case the entity quest barfed |
8118 | | */ |
8119 | 110k | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
8120 | 110k | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
8121 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8122 | 0 | "%%%s; is not a parameter entity\n", |
8123 | 0 | name, NULL); |
8124 | 0 | } |
8125 | 110k | } |
8126 | | |
8127 | 122k | xmlFree(name); |
8128 | 122k | *str = ptr; |
8129 | 122k | return(entity); |
8130 | 138k | } |
8131 | | |
8132 | | /** |
8133 | | * xmlParseDocTypeDecl: |
8134 | | * @ctxt: an XML parser context |
8135 | | * |
8136 | | * DEPRECATED: Internal function, don't use. |
8137 | | * |
8138 | | * parse a DOCTYPE declaration |
8139 | | * |
8140 | | * [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? |
8141 | | * ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8142 | | * |
8143 | | * [ VC: Root Element Type ] |
8144 | | * The Name in the document type declaration must match the element |
8145 | | * type of the root element. |
8146 | | */ |
8147 | | |
8148 | | void |
8149 | 258k | xmlParseDocTypeDecl(xmlParserCtxtPtr ctxt) { |
8150 | 258k | const xmlChar *name = NULL; |
8151 | 258k | xmlChar *ExternalID = NULL; |
8152 | 258k | xmlChar *URI = NULL; |
8153 | | |
8154 | | /* |
8155 | | * We know that '<!DOCTYPE' has been detected. |
8156 | | */ |
8157 | 258k | SKIP(9); |
8158 | | |
8159 | 258k | SKIP_BLANKS; |
8160 | | |
8161 | | /* |
8162 | | * Parse the DOCTYPE name. |
8163 | | */ |
8164 | 258k | name = xmlParseName(ctxt); |
8165 | 258k | if (name == NULL) { |
8166 | 9.63k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8167 | 9.63k | "xmlParseDocTypeDecl : no DOCTYPE name !\n"); |
8168 | 9.63k | } |
8169 | 258k | ctxt->intSubName = name; |
8170 | | |
8171 | 258k | SKIP_BLANKS; |
8172 | | |
8173 | | /* |
8174 | | * Check for SystemID and ExternalID |
8175 | | */ |
8176 | 258k | URI = xmlParseExternalID(ctxt, &ExternalID, 1); |
8177 | | |
8178 | 258k | if ((URI != NULL) || (ExternalID != NULL)) { |
8179 | 160k | ctxt->hasExternalSubset = 1; |
8180 | 160k | } |
8181 | 258k | ctxt->extSubURI = URI; |
8182 | 258k | ctxt->extSubSystem = ExternalID; |
8183 | | |
8184 | 258k | SKIP_BLANKS; |
8185 | | |
8186 | | /* |
8187 | | * Create and update the internal subset. |
8188 | | */ |
8189 | 258k | if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) && |
8190 | 258k | (!ctxt->disableSAX)) |
8191 | 226k | ctxt->sax->internalSubset(ctxt->userData, name, ExternalID, URI); |
8192 | | |
8193 | | /* |
8194 | | * Is there any internal subset declarations ? |
8195 | | * they are handled separately in xmlParseInternalSubset() |
8196 | | */ |
8197 | 258k | if (RAW == '[') |
8198 | 197k | return; |
8199 | | |
8200 | | /* |
8201 | | * We should be at the end of the DOCTYPE declaration. |
8202 | | */ |
8203 | 61.5k | if (RAW != '>') { |
8204 | 18.8k | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8205 | 18.8k | } |
8206 | 61.5k | NEXT; |
8207 | 61.5k | } |
8208 | | |
8209 | | /** |
8210 | | * xmlParseInternalSubset: |
8211 | | * @ctxt: an XML parser context |
8212 | | * |
8213 | | * parse the internal subset declaration |
8214 | | * |
8215 | | * [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8216 | | */ |
8217 | | |
8218 | | static void |
8219 | 201k | xmlParseInternalSubset(xmlParserCtxtPtr ctxt) { |
8220 | | /* |
8221 | | * Is there any DTD definition ? |
8222 | | */ |
8223 | 201k | if (RAW == '[') { |
8224 | 201k | int oldInputNr = ctxt->inputNr; |
8225 | | |
8226 | 201k | NEXT; |
8227 | | /* |
8228 | | * Parse the succession of Markup declarations and |
8229 | | * PEReferences. |
8230 | | * Subsequence (markupdecl | PEReference | S)* |
8231 | | */ |
8232 | 201k | SKIP_BLANKS; |
8233 | 3.62M | while (((RAW != ']') || (ctxt->inputNr > oldInputNr)) && |
8234 | 3.62M | (PARSER_STOPPED(ctxt) == 0)) { |
8235 | | |
8236 | | /* |
8237 | | * Conditional sections are allowed from external entities included |
8238 | | * by PE References in the internal subset. |
8239 | | */ |
8240 | 3.49M | if ((PARSER_EXTERNAL(ctxt)) && |
8241 | 3.49M | (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
8242 | 14.4k | xmlParseConditionalSections(ctxt); |
8243 | 3.47M | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
8244 | 3.01M | xmlParseMarkupDecl(ctxt); |
8245 | 3.01M | } else if (RAW == '%') { |
8246 | 390k | xmlParsePEReference(ctxt); |
8247 | 390k | } else { |
8248 | 72.8k | xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL); |
8249 | 72.8k | break; |
8250 | 72.8k | } |
8251 | 3.41M | SKIP_BLANKS_PE; |
8252 | 3.41M | SHRINK; |
8253 | 3.41M | GROW; |
8254 | 3.41M | } |
8255 | | |
8256 | 212k | while (ctxt->inputNr > oldInputNr) |
8257 | 10.5k | xmlPopPE(ctxt); |
8258 | | |
8259 | 201k | if (RAW == ']') { |
8260 | 107k | NEXT; |
8261 | 107k | SKIP_BLANKS; |
8262 | 107k | } |
8263 | 201k | } |
8264 | | |
8265 | | /* |
8266 | | * We should be at the end of the DOCTYPE declaration. |
8267 | | */ |
8268 | 201k | if ((ctxt->wellFormed) && (RAW != '>')) { |
8269 | 2.90k | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8270 | 2.90k | return; |
8271 | 2.90k | } |
8272 | 198k | NEXT; |
8273 | 198k | } |
8274 | | |
8275 | | #ifdef LIBXML_SAX1_ENABLED |
8276 | | /** |
8277 | | * xmlParseAttribute: |
8278 | | * @ctxt: an XML parser context |
8279 | | * @value: a xmlChar ** used to store the value of the attribute |
8280 | | * |
8281 | | * DEPRECATED: Internal function, don't use. |
8282 | | * |
8283 | | * parse an attribute |
8284 | | * |
8285 | | * [41] Attribute ::= Name Eq AttValue |
8286 | | * |
8287 | | * [ WFC: No External Entity References ] |
8288 | | * Attribute values cannot contain direct or indirect entity references |
8289 | | * to external entities. |
8290 | | * |
8291 | | * [ WFC: No < in Attribute Values ] |
8292 | | * The replacement text of any entity referred to directly or indirectly in |
8293 | | * an attribute value (other than "<") must not contain a <. |
8294 | | * |
8295 | | * [ VC: Attribute Value Type ] |
8296 | | * The attribute must have been declared; the value must be of the type |
8297 | | * declared for it. |
8298 | | * |
8299 | | * [25] Eq ::= S? '=' S? |
8300 | | * |
8301 | | * With namespace: |
8302 | | * |
8303 | | * [NS 11] Attribute ::= QName Eq AttValue |
8304 | | * |
8305 | | * Also the case QName == xmlns:??? is handled independently as a namespace |
8306 | | * definition. |
8307 | | * |
8308 | | * Returns the attribute name, and the value in *value. |
8309 | | */ |
8310 | | |
8311 | | const xmlChar * |
8312 | 2.66M | xmlParseAttribute(xmlParserCtxtPtr ctxt, xmlChar **value) { |
8313 | 2.66M | const xmlChar *name; |
8314 | 2.66M | xmlChar *val; |
8315 | | |
8316 | 2.66M | *value = NULL; |
8317 | 2.66M | GROW; |
8318 | 2.66M | name = xmlParseName(ctxt); |
8319 | 2.66M | if (name == NULL) { |
8320 | 956k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8321 | 956k | "error parsing attribute name\n"); |
8322 | 956k | return(NULL); |
8323 | 956k | } |
8324 | | |
8325 | | /* |
8326 | | * read the value |
8327 | | */ |
8328 | 1.71M | SKIP_BLANKS; |
8329 | 1.71M | if (RAW == '=') { |
8330 | 1.25M | NEXT; |
8331 | 1.25M | SKIP_BLANKS; |
8332 | 1.25M | val = xmlParseAttValue(ctxt); |
8333 | 1.25M | } else { |
8334 | 452k | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8335 | 452k | "Specification mandates value for attribute %s\n", name); |
8336 | 452k | return(name); |
8337 | 452k | } |
8338 | | |
8339 | | /* |
8340 | | * Check that xml:lang conforms to the specification |
8341 | | * No more registered as an error, just generate a warning now |
8342 | | * since this was deprecated in XML second edition |
8343 | | */ |
8344 | 1.25M | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) { |
8345 | 173k | if (!xmlCheckLanguageID(val)) { |
8346 | 154k | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8347 | 154k | "Malformed value for xml:lang : %s\n", |
8348 | 154k | val, NULL); |
8349 | 154k | } |
8350 | 173k | } |
8351 | | |
8352 | | /* |
8353 | | * Check that xml:space conforms to the specification |
8354 | | */ |
8355 | 1.25M | if (xmlStrEqual(name, BAD_CAST "xml:space")) { |
8356 | 8.75k | if (xmlStrEqual(val, BAD_CAST "default")) |
8357 | 1.50k | *(ctxt->space) = 0; |
8358 | 7.25k | else if (xmlStrEqual(val, BAD_CAST "preserve")) |
8359 | 2.93k | *(ctxt->space) = 1; |
8360 | 4.31k | else { |
8361 | 4.31k | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8362 | 4.31k | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8363 | 4.31k | val, NULL); |
8364 | 4.31k | } |
8365 | 8.75k | } |
8366 | | |
8367 | 1.25M | *value = val; |
8368 | 1.25M | return(name); |
8369 | 1.71M | } |
8370 | | |
8371 | | /** |
8372 | | * xmlParseStartTag: |
8373 | | * @ctxt: an XML parser context |
8374 | | * |
8375 | | * DEPRECATED: Internal function, don't use. |
8376 | | * |
8377 | | * Parse a start tag. Always consumes '<'. |
8378 | | * |
8379 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8380 | | * |
8381 | | * [ WFC: Unique Att Spec ] |
8382 | | * No attribute name may appear more than once in the same start-tag or |
8383 | | * empty-element tag. |
8384 | | * |
8385 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8386 | | * |
8387 | | * [ WFC: Unique Att Spec ] |
8388 | | * No attribute name may appear more than once in the same start-tag or |
8389 | | * empty-element tag. |
8390 | | * |
8391 | | * With namespace: |
8392 | | * |
8393 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8394 | | * |
8395 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8396 | | * |
8397 | | * Returns the element name parsed |
8398 | | */ |
8399 | | |
8400 | | const xmlChar * |
8401 | 4.05M | xmlParseStartTag(xmlParserCtxtPtr ctxt) { |
8402 | 4.05M | const xmlChar *name; |
8403 | 4.05M | const xmlChar *attname; |
8404 | 4.05M | xmlChar *attvalue; |
8405 | 4.05M | const xmlChar **atts = ctxt->atts; |
8406 | 4.05M | int nbatts = 0; |
8407 | 4.05M | int maxatts = ctxt->maxatts; |
8408 | 4.05M | int i; |
8409 | | |
8410 | 4.05M | if (RAW != '<') return(NULL); |
8411 | 4.05M | NEXT1; |
8412 | | |
8413 | 4.05M | name = xmlParseName(ctxt); |
8414 | 4.05M | if (name == NULL) { |
8415 | 565k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8416 | 565k | "xmlParseStartTag: invalid element name\n"); |
8417 | 565k | return(NULL); |
8418 | 565k | } |
8419 | | |
8420 | | /* |
8421 | | * Now parse the attributes, it ends up with the ending |
8422 | | * |
8423 | | * (S Attribute)* S? |
8424 | | */ |
8425 | 3.48M | SKIP_BLANKS; |
8426 | 3.48M | GROW; |
8427 | | |
8428 | 4.84M | while (((RAW != '>') && |
8429 | 4.84M | ((RAW != '/') || (NXT(1) != '>')) && |
8430 | 4.84M | (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) { |
8431 | 2.66M | attname = xmlParseAttribute(ctxt, &attvalue); |
8432 | 2.66M | if (attname == NULL) |
8433 | 956k | break; |
8434 | 1.71M | if (attvalue != NULL) { |
8435 | | /* |
8436 | | * [ WFC: Unique Att Spec ] |
8437 | | * No attribute name may appear more than once in the same |
8438 | | * start-tag or empty-element tag. |
8439 | | */ |
8440 | 8.08M | for (i = 0; i < nbatts;i += 2) { |
8441 | 6.91M | if (xmlStrEqual(atts[i], attname)) { |
8442 | 38.5k | xmlErrAttributeDup(ctxt, NULL, attname); |
8443 | 38.5k | xmlFree(attvalue); |
8444 | 38.5k | goto failed; |
8445 | 38.5k | } |
8446 | 6.91M | } |
8447 | | /* |
8448 | | * Add the pair to atts |
8449 | | */ |
8450 | 1.16M | if (atts == NULL) { |
8451 | 49.1k | maxatts = 22; /* allow for 10 attrs by default */ |
8452 | 49.1k | atts = (const xmlChar **) |
8453 | 49.1k | xmlMalloc(maxatts * sizeof(xmlChar *)); |
8454 | 49.1k | if (atts == NULL) { |
8455 | 27 | xmlErrMemory(ctxt); |
8456 | 27 | if (attvalue != NULL) |
8457 | 27 | xmlFree(attvalue); |
8458 | 27 | goto failed; |
8459 | 27 | } |
8460 | 49.0k | ctxt->atts = atts; |
8461 | 49.0k | ctxt->maxatts = maxatts; |
8462 | 1.11M | } else if (nbatts + 4 > maxatts) { |
8463 | 4.59k | const xmlChar **n; |
8464 | | |
8465 | 4.59k | maxatts *= 2; |
8466 | 4.59k | n = (const xmlChar **) xmlRealloc((void *) atts, |
8467 | 4.59k | maxatts * sizeof(const xmlChar *)); |
8468 | 4.59k | if (n == NULL) { |
8469 | 10 | xmlErrMemory(ctxt); |
8470 | 10 | if (attvalue != NULL) |
8471 | 10 | xmlFree(attvalue); |
8472 | 10 | goto failed; |
8473 | 10 | } |
8474 | 4.58k | atts = n; |
8475 | 4.58k | ctxt->atts = atts; |
8476 | 4.58k | ctxt->maxatts = maxatts; |
8477 | 4.58k | } |
8478 | 1.16M | atts[nbatts++] = attname; |
8479 | 1.16M | atts[nbatts++] = attvalue; |
8480 | 1.16M | atts[nbatts] = NULL; |
8481 | 1.16M | atts[nbatts + 1] = NULL; |
8482 | 1.16M | } else { |
8483 | 510k | if (attvalue != NULL) |
8484 | 0 | xmlFree(attvalue); |
8485 | 510k | } |
8486 | | |
8487 | 1.71M | failed: |
8488 | | |
8489 | 1.71M | GROW |
8490 | 1.71M | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
8491 | 350k | break; |
8492 | 1.36M | if (SKIP_BLANKS == 0) { |
8493 | 780k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
8494 | 780k | "attributes construct error\n"); |
8495 | 780k | } |
8496 | 1.36M | SHRINK; |
8497 | 1.36M | GROW; |
8498 | 1.36M | } |
8499 | | |
8500 | | /* |
8501 | | * SAX: Start of Element ! |
8502 | | */ |
8503 | 3.48M | if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) && |
8504 | 3.48M | (!ctxt->disableSAX)) { |
8505 | 3.36M | if (nbatts > 0) |
8506 | 740k | ctxt->sax->startElement(ctxt->userData, name, atts); |
8507 | 2.62M | else |
8508 | 2.62M | ctxt->sax->startElement(ctxt->userData, name, NULL); |
8509 | 3.36M | } |
8510 | | |
8511 | 3.48M | if (atts != NULL) { |
8512 | | /* Free only the content strings */ |
8513 | 4.33M | for (i = 1;i < nbatts;i+=2) |
8514 | 1.16M | if (atts[i] != NULL) |
8515 | 1.16M | xmlFree((xmlChar *) atts[i]); |
8516 | 3.16M | } |
8517 | 3.48M | return(name); |
8518 | 3.48M | } |
8519 | | |
8520 | | /** |
8521 | | * xmlParseEndTag1: |
8522 | | * @ctxt: an XML parser context |
8523 | | * @line: line of the start tag |
8524 | | * @nsNr: number of namespaces on the start tag |
8525 | | * |
8526 | | * Parse an end tag. Always consumes '</'. |
8527 | | * |
8528 | | * [42] ETag ::= '</' Name S? '>' |
8529 | | * |
8530 | | * With namespace |
8531 | | * |
8532 | | * [NS 9] ETag ::= '</' QName S? '>' |
8533 | | */ |
8534 | | |
8535 | | static void |
8536 | 1.16M | xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) { |
8537 | 1.16M | const xmlChar *name; |
8538 | | |
8539 | 1.16M | GROW; |
8540 | 1.16M | if ((RAW != '<') || (NXT(1) != '/')) { |
8541 | 3.92k | xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED, |
8542 | 3.92k | "xmlParseEndTag: '</' not found\n"); |
8543 | 3.92k | return; |
8544 | 3.92k | } |
8545 | 1.15M | SKIP(2); |
8546 | | |
8547 | 1.15M | name = xmlParseNameAndCompare(ctxt,ctxt->name); |
8548 | | |
8549 | | /* |
8550 | | * We should definitely be at the ending "S? '>'" part |
8551 | | */ |
8552 | 1.15M | GROW; |
8553 | 1.15M | SKIP_BLANKS; |
8554 | 1.15M | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
8555 | 104k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
8556 | 104k | } else |
8557 | 1.05M | NEXT1; |
8558 | | |
8559 | | /* |
8560 | | * [ WFC: Element Type Match ] |
8561 | | * The Name in an element's end-tag must match the element type in the |
8562 | | * start-tag. |
8563 | | * |
8564 | | */ |
8565 | 1.15M | if (name != (xmlChar*)1) { |
8566 | 182k | if (name == NULL) name = BAD_CAST "unparsable"; |
8567 | 182k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
8568 | 182k | "Opening and ending tag mismatch: %s line %d and %s\n", |
8569 | 182k | ctxt->name, line, name); |
8570 | 182k | } |
8571 | | |
8572 | | /* |
8573 | | * SAX: End of Tag |
8574 | | */ |
8575 | 1.15M | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
8576 | 1.15M | (!ctxt->disableSAX)) |
8577 | 1.12M | ctxt->sax->endElement(ctxt->userData, ctxt->name); |
8578 | | |
8579 | 1.15M | namePop(ctxt); |
8580 | 1.15M | spacePop(ctxt); |
8581 | 1.15M | } |
8582 | | |
8583 | | /** |
8584 | | * xmlParseEndTag: |
8585 | | * @ctxt: an XML parser context |
8586 | | * |
8587 | | * DEPRECATED: Internal function, don't use. |
8588 | | * |
8589 | | * parse an end of tag |
8590 | | * |
8591 | | * [42] ETag ::= '</' Name S? '>' |
8592 | | * |
8593 | | * With namespace |
8594 | | * |
8595 | | * [NS 9] ETag ::= '</' QName S? '>' |
8596 | | */ |
8597 | | |
8598 | | void |
8599 | 0 | xmlParseEndTag(xmlParserCtxtPtr ctxt) { |
8600 | 0 | xmlParseEndTag1(ctxt, 0); |
8601 | 0 | } |
8602 | | #endif /* LIBXML_SAX1_ENABLED */ |
8603 | | |
8604 | | /************************************************************************ |
8605 | | * * |
8606 | | * SAX 2 specific operations * |
8607 | | * * |
8608 | | ************************************************************************/ |
8609 | | |
8610 | | /** |
8611 | | * xmlParseQNameHashed: |
8612 | | * @ctxt: an XML parser context |
8613 | | * @prefix: pointer to store the prefix part |
8614 | | * |
8615 | | * parse an XML Namespace QName |
8616 | | * |
8617 | | * [6] QName ::= (Prefix ':')? LocalPart |
8618 | | * [7] Prefix ::= NCName |
8619 | | * [8] LocalPart ::= NCName |
8620 | | * |
8621 | | * Returns the Name parsed or NULL |
8622 | | */ |
8623 | | |
8624 | | static xmlHashedString |
8625 | 13.4M | xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) { |
8626 | 13.4M | xmlHashedString l, p; |
8627 | 13.4M | int start, isNCName = 0; |
8628 | | |
8629 | 13.4M | l.name = NULL; |
8630 | 13.4M | p.name = NULL; |
8631 | | |
8632 | 13.4M | GROW; |
8633 | 13.4M | start = CUR_PTR - BASE_PTR; |
8634 | | |
8635 | 13.4M | l = xmlParseNCName(ctxt); |
8636 | 13.4M | if (l.name != NULL) { |
8637 | 10.2M | isNCName = 1; |
8638 | 10.2M | if (CUR == ':') { |
8639 | 2.90M | NEXT; |
8640 | 2.90M | p = l; |
8641 | 2.90M | l = xmlParseNCName(ctxt); |
8642 | 2.90M | } |
8643 | 10.2M | } |
8644 | 13.4M | if ((l.name == NULL) || (CUR == ':')) { |
8645 | 3.37M | xmlChar *tmp; |
8646 | | |
8647 | 3.37M | l.name = NULL; |
8648 | 3.37M | p.name = NULL; |
8649 | 3.37M | if ((isNCName == 0) && (CUR != ':')) |
8650 | 3.07M | return(l); |
8651 | 308k | tmp = xmlParseNmtoken(ctxt); |
8652 | 308k | if (tmp != NULL) |
8653 | 183k | xmlFree(tmp); |
8654 | 308k | l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start, |
8655 | 308k | CUR_PTR - (BASE_PTR + start)); |
8656 | 308k | if (l.name == NULL) { |
8657 | 21 | xmlErrMemory(ctxt); |
8658 | 21 | return(l); |
8659 | 21 | } |
8660 | 308k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8661 | 308k | "Failed to parse QName '%s'\n", l.name, NULL, NULL); |
8662 | 308k | } |
8663 | | |
8664 | 10.3M | *prefix = p; |
8665 | 10.3M | return(l); |
8666 | 13.4M | } |
8667 | | |
8668 | | /** |
8669 | | * xmlParseQName: |
8670 | | * @ctxt: an XML parser context |
8671 | | * @prefix: pointer to store the prefix part |
8672 | | * |
8673 | | * parse an XML Namespace QName |
8674 | | * |
8675 | | * [6] QName ::= (Prefix ':')? LocalPart |
8676 | | * [7] Prefix ::= NCName |
8677 | | * [8] LocalPart ::= NCName |
8678 | | * |
8679 | | * Returns the Name parsed or NULL |
8680 | | */ |
8681 | | |
8682 | | static const xmlChar * |
8683 | 113k | xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) { |
8684 | 113k | xmlHashedString n, p; |
8685 | | |
8686 | 113k | n = xmlParseQNameHashed(ctxt, &p); |
8687 | 113k | if (n.name == NULL) |
8688 | 12.3k | return(NULL); |
8689 | 101k | *prefix = p.name; |
8690 | 101k | return(n.name); |
8691 | 113k | } |
8692 | | |
8693 | | /** |
8694 | | * xmlParseQNameAndCompare: |
8695 | | * @ctxt: an XML parser context |
8696 | | * @name: the localname |
8697 | | * @prefix: the prefix, if any. |
8698 | | * |
8699 | | * parse an XML name and compares for match |
8700 | | * (specialized for endtag parsing) |
8701 | | * |
8702 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
8703 | | * and the name for mismatch |
8704 | | */ |
8705 | | |
8706 | | static const xmlChar * |
8707 | | xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name, |
8708 | 261k | xmlChar const *prefix) { |
8709 | 261k | const xmlChar *cmp; |
8710 | 261k | const xmlChar *in; |
8711 | 261k | const xmlChar *ret; |
8712 | 261k | const xmlChar *prefix2; |
8713 | | |
8714 | 261k | if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name)); |
8715 | | |
8716 | 261k | GROW; |
8717 | 261k | in = ctxt->input->cur; |
8718 | | |
8719 | 261k | cmp = prefix; |
8720 | 727k | while (*in != 0 && *in == *cmp) { |
8721 | 465k | ++in; |
8722 | 465k | ++cmp; |
8723 | 465k | } |
8724 | 261k | if ((*cmp == 0) && (*in == ':')) { |
8725 | 192k | in++; |
8726 | 192k | cmp = name; |
8727 | 1.17M | while (*in != 0 && *in == *cmp) { |
8728 | 978k | ++in; |
8729 | 978k | ++cmp; |
8730 | 978k | } |
8731 | 192k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
8732 | | /* success */ |
8733 | 148k | ctxt->input->col += in - ctxt->input->cur; |
8734 | 148k | ctxt->input->cur = in; |
8735 | 148k | return((const xmlChar*) 1); |
8736 | 148k | } |
8737 | 192k | } |
8738 | | /* |
8739 | | * all strings coms from the dictionary, equality can be done directly |
8740 | | */ |
8741 | 113k | ret = xmlParseQName (ctxt, &prefix2); |
8742 | 113k | if (ret == NULL) |
8743 | 12.3k | return(NULL); |
8744 | 101k | if ((ret == name) && (prefix == prefix2)) |
8745 | 10.3k | return((const xmlChar*) 1); |
8746 | 90.7k | return ret; |
8747 | 101k | } |
8748 | | |
8749 | | /** |
8750 | | * xmlParseAttribute2: |
8751 | | * @ctxt: an XML parser context |
8752 | | * @pref: the element prefix |
8753 | | * @elem: the element name |
8754 | | * @prefix: a xmlChar ** used to store the value of the attribute prefix |
8755 | | * @value: a xmlChar ** used to store the value of the attribute |
8756 | | * @len: an int * to save the length of the attribute |
8757 | | * @alloc: an int * to indicate if the attribute was allocated |
8758 | | * |
8759 | | * parse an attribute in the new SAX2 framework. |
8760 | | * |
8761 | | * Returns the attribute name, and the value in *value, . |
8762 | | */ |
8763 | | |
8764 | | static xmlHashedString |
8765 | | xmlParseAttribute2(xmlParserCtxtPtr ctxt, |
8766 | | const xmlChar * pref, const xmlChar * elem, |
8767 | | xmlHashedString * hprefix, xmlChar ** value, |
8768 | | int *len, int *alloc) |
8769 | 5.11M | { |
8770 | 5.11M | xmlHashedString hname; |
8771 | 5.11M | const xmlChar *prefix, *name; |
8772 | 5.11M | xmlChar *val = NULL, *internal_val = NULL; |
8773 | 5.11M | int normalize = 0; |
8774 | 5.11M | int isNamespace; |
8775 | | |
8776 | 5.11M | *value = NULL; |
8777 | 5.11M | GROW; |
8778 | 5.11M | hname = xmlParseQNameHashed(ctxt, hprefix); |
8779 | 5.11M | if (hname.name == NULL) { |
8780 | 1.42M | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8781 | 1.42M | "error parsing attribute name\n"); |
8782 | 1.42M | return(hname); |
8783 | 1.42M | } |
8784 | 3.69M | name = hname.name; |
8785 | 3.69M | if (hprefix->name != NULL) |
8786 | 1.59M | prefix = hprefix->name; |
8787 | 2.09M | else |
8788 | 2.09M | prefix = NULL; |
8789 | | |
8790 | | /* |
8791 | | * get the type if needed |
8792 | | */ |
8793 | 3.69M | if (ctxt->attsSpecial != NULL) { |
8794 | 1.45M | int type; |
8795 | | |
8796 | 1.45M | type = (int) (ptrdiff_t) xmlHashQLookup2(ctxt->attsSpecial, |
8797 | 1.45M | pref, elem, |
8798 | 1.45M | prefix, name); |
8799 | 1.45M | if (type != 0) |
8800 | 211k | normalize = 1; |
8801 | 1.45M | } |
8802 | | |
8803 | | /* |
8804 | | * read the value |
8805 | | */ |
8806 | 3.69M | SKIP_BLANKS; |
8807 | 3.69M | if (RAW == '=') { |
8808 | 3.27M | NEXT; |
8809 | 3.27M | SKIP_BLANKS; |
8810 | 3.27M | isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) || |
8811 | 3.27M | (prefix == ctxt->str_xmlns)); |
8812 | 3.27M | val = xmlParseAttValueInternal(ctxt, len, alloc, normalize, |
8813 | 3.27M | isNamespace); |
8814 | 3.27M | if (val == NULL) |
8815 | 123k | goto error; |
8816 | 3.27M | } else { |
8817 | 413k | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8818 | 413k | "Specification mandates value for attribute %s\n", |
8819 | 413k | name); |
8820 | 413k | goto error; |
8821 | 413k | } |
8822 | | |
8823 | 3.15M | if (prefix == ctxt->str_xml) { |
8824 | | /* |
8825 | | * Check that xml:lang conforms to the specification |
8826 | | * No more registered as an error, just generate a warning now |
8827 | | * since this was deprecated in XML second edition |
8828 | | */ |
8829 | 331k | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) { |
8830 | 105k | internal_val = xmlStrndup(val, *len); |
8831 | 105k | if (internal_val == NULL) |
8832 | 46 | goto mem_error; |
8833 | 105k | if (!xmlCheckLanguageID(internal_val)) { |
8834 | 86.6k | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8835 | 86.6k | "Malformed value for xml:lang : %s\n", |
8836 | 86.6k | internal_val, NULL); |
8837 | 86.6k | } |
8838 | 105k | } |
8839 | | |
8840 | | /* |
8841 | | * Check that xml:space conforms to the specification |
8842 | | */ |
8843 | 331k | if (xmlStrEqual(name, BAD_CAST "space")) { |
8844 | 10.7k | internal_val = xmlStrndup(val, *len); |
8845 | 10.7k | if (internal_val == NULL) |
8846 | 15 | goto mem_error; |
8847 | 10.7k | if (xmlStrEqual(internal_val, BAD_CAST "default")) |
8848 | 2.71k | *(ctxt->space) = 0; |
8849 | 8.05k | else if (xmlStrEqual(internal_val, BAD_CAST "preserve")) |
8850 | 4.88k | *(ctxt->space) = 1; |
8851 | 3.17k | else { |
8852 | 3.17k | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8853 | 3.17k | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8854 | 3.17k | internal_val, NULL); |
8855 | 3.17k | } |
8856 | 10.7k | } |
8857 | 331k | if (internal_val) { |
8858 | 116k | xmlFree(internal_val); |
8859 | 116k | } |
8860 | 331k | } |
8861 | | |
8862 | 3.15M | *value = val; |
8863 | 3.15M | return (hname); |
8864 | | |
8865 | 61 | mem_error: |
8866 | 61 | xmlErrMemory(ctxt); |
8867 | 537k | error: |
8868 | 537k | if ((val != NULL) && (*alloc != 0)) |
8869 | 31 | xmlFree(val); |
8870 | 537k | return(hname); |
8871 | 61 | } |
8872 | | |
8873 | | /** |
8874 | | * xmlAttrHashInsert: |
8875 | | * @ctxt: parser context |
8876 | | * @size: size of the hash table |
8877 | | * @name: attribute name |
8878 | | * @uri: namespace uri |
8879 | | * @hashValue: combined hash value of name and uri |
8880 | | * @aindex: attribute index (this is a multiple of 5) |
8881 | | * |
8882 | | * Inserts a new attribute into the hash table. |
8883 | | * |
8884 | | * Returns INT_MAX if no existing attribute was found, the attribute |
8885 | | * index if an attribute was found, -1 if a memory allocation failed. |
8886 | | */ |
8887 | | static int |
8888 | | xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name, |
8889 | 2.69M | const xmlChar *uri, unsigned hashValue, int aindex) { |
8890 | 2.69M | xmlAttrHashBucket *table = ctxt->attrHash; |
8891 | 2.69M | xmlAttrHashBucket *bucket; |
8892 | 2.69M | unsigned hindex; |
8893 | | |
8894 | 2.69M | hindex = hashValue & (size - 1); |
8895 | 2.69M | bucket = &table[hindex]; |
8896 | | |
8897 | 3.32M | while (bucket->index >= 0) { |
8898 | 823k | const xmlChar **atts = &ctxt->atts[bucket->index]; |
8899 | | |
8900 | 823k | if (name == atts[0]) { |
8901 | 213k | int nsIndex = (int) (ptrdiff_t) atts[2]; |
8902 | | |
8903 | 213k | if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) : |
8904 | 213k | (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) : |
8905 | 69.4k | (uri == ctxt->nsTab[nsIndex * 2 + 1])) |
8906 | 194k | return(bucket->index); |
8907 | 213k | } |
8908 | | |
8909 | 629k | hindex++; |
8910 | 629k | bucket++; |
8911 | 629k | if (hindex >= size) { |
8912 | 39.6k | hindex = 0; |
8913 | 39.6k | bucket = table; |
8914 | 39.6k | } |
8915 | 629k | } |
8916 | | |
8917 | 2.49M | bucket->index = aindex; |
8918 | | |
8919 | 2.49M | return(INT_MAX); |
8920 | 2.69M | } |
8921 | | |
8922 | | /** |
8923 | | * xmlParseStartTag2: |
8924 | | * @ctxt: an XML parser context |
8925 | | * |
8926 | | * Parse a start tag. Always consumes '<'. |
8927 | | * |
8928 | | * This routine is called when running SAX2 parsing |
8929 | | * |
8930 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8931 | | * |
8932 | | * [ WFC: Unique Att Spec ] |
8933 | | * No attribute name may appear more than once in the same start-tag or |
8934 | | * empty-element tag. |
8935 | | * |
8936 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8937 | | * |
8938 | | * [ WFC: Unique Att Spec ] |
8939 | | * No attribute name may appear more than once in the same start-tag or |
8940 | | * empty-element tag. |
8941 | | * |
8942 | | * With namespace: |
8943 | | * |
8944 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8945 | | * |
8946 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8947 | | * |
8948 | | * Returns the element name parsed |
8949 | | */ |
8950 | | |
8951 | | static const xmlChar * |
8952 | | xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref, |
8953 | 8.23M | const xmlChar **URI, int *nbNsPtr) { |
8954 | 8.23M | xmlHashedString hlocalname; |
8955 | 8.23M | xmlHashedString hprefix; |
8956 | 8.23M | xmlHashedString hattname; |
8957 | 8.23M | xmlHashedString haprefix; |
8958 | 8.23M | const xmlChar *localname; |
8959 | 8.23M | const xmlChar *prefix; |
8960 | 8.23M | const xmlChar *attname; |
8961 | 8.23M | const xmlChar *aprefix; |
8962 | 8.23M | const xmlChar *uri; |
8963 | 8.23M | xmlChar *attvalue = NULL; |
8964 | 8.23M | const xmlChar **atts = ctxt->atts; |
8965 | 8.23M | unsigned attrHashSize = 0; |
8966 | 8.23M | int maxatts = ctxt->maxatts; |
8967 | 8.23M | int nratts, nbatts, nbdef; |
8968 | 8.23M | int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts; |
8969 | 8.23M | int alloc = 0; |
8970 | | |
8971 | 8.23M | if (RAW != '<') return(NULL); |
8972 | 8.23M | NEXT1; |
8973 | | |
8974 | 8.23M | nbatts = 0; |
8975 | 8.23M | nratts = 0; |
8976 | 8.23M | nbdef = 0; |
8977 | 8.23M | nbNs = 0; |
8978 | 8.23M | nbTotalDef = 0; |
8979 | 8.23M | attval = 0; |
8980 | | |
8981 | 8.23M | if (xmlParserNsStartElement(ctxt->nsdb) < 0) { |
8982 | 0 | xmlErrMemory(ctxt); |
8983 | 0 | return(NULL); |
8984 | 0 | } |
8985 | | |
8986 | 8.23M | hlocalname = xmlParseQNameHashed(ctxt, &hprefix); |
8987 | 8.23M | if (hlocalname.name == NULL) { |
8988 | 1.63M | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8989 | 1.63M | "StartTag: invalid element name\n"); |
8990 | 1.63M | return(NULL); |
8991 | 1.63M | } |
8992 | 6.60M | localname = hlocalname.name; |
8993 | 6.60M | prefix = hprefix.name; |
8994 | | |
8995 | | /* |
8996 | | * Now parse the attributes, it ends up with the ending |
8997 | | * |
8998 | | * (S Attribute)* S? |
8999 | | */ |
9000 | 6.60M | SKIP_BLANKS; |
9001 | 6.60M | GROW; |
9002 | | |
9003 | | /* |
9004 | | * The ctxt->atts array will be ultimately passed to the SAX callback |
9005 | | * containing five xmlChar pointers for each attribute: |
9006 | | * |
9007 | | * [0] attribute name |
9008 | | * [1] attribute prefix |
9009 | | * [2] namespace URI |
9010 | | * [3] attribute value |
9011 | | * [4] end of attribute value |
9012 | | * |
9013 | | * To save memory, we reuse this array temporarily and store integers |
9014 | | * in these pointer variables. |
9015 | | * |
9016 | | * [0] attribute name |
9017 | | * [1] attribute prefix |
9018 | | * [2] hash value of attribute prefix, and later namespace index |
9019 | | * [3] for non-allocated values: ptrdiff_t offset into input buffer |
9020 | | * [4] for non-allocated values: ptrdiff_t offset into input buffer |
9021 | | * |
9022 | | * The ctxt->attallocs array contains an additional unsigned int for |
9023 | | * each attribute, containing the hash value of the attribute name |
9024 | | * and the alloc flag in bit 31. |
9025 | | */ |
9026 | | |
9027 | 8.12M | while (((RAW != '>') && |
9028 | 8.12M | ((RAW != '/') || (NXT(1) != '>')) && |
9029 | 8.12M | (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) { |
9030 | 5.11M | int len = -1; |
9031 | | |
9032 | 5.11M | hattname = xmlParseAttribute2(ctxt, prefix, localname, |
9033 | 5.11M | &haprefix, &attvalue, &len, |
9034 | 5.11M | &alloc); |
9035 | 5.11M | if (hattname.name == NULL) |
9036 | 1.42M | break; |
9037 | 3.69M | if (attvalue == NULL) |
9038 | 537k | goto next_attr; |
9039 | 3.15M | attname = hattname.name; |
9040 | 3.15M | aprefix = haprefix.name; |
9041 | 3.15M | if (len < 0) len = xmlStrlen(attvalue); |
9042 | | |
9043 | 3.15M | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9044 | 375k | xmlHashedString huri; |
9045 | 375k | xmlURIPtr parsedUri; |
9046 | | |
9047 | 375k | huri = xmlDictLookupHashed(ctxt->dict, attvalue, len); |
9048 | 375k | uri = huri.name; |
9049 | 375k | if (uri == NULL) { |
9050 | 15 | xmlErrMemory(ctxt); |
9051 | 15 | goto next_attr; |
9052 | 15 | } |
9053 | 375k | if (*uri != 0) { |
9054 | 361k | if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) { |
9055 | 155 | xmlErrMemory(ctxt); |
9056 | 155 | goto next_attr; |
9057 | 155 | } |
9058 | 361k | if (parsedUri == NULL) { |
9059 | 246k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9060 | 246k | "xmlns: '%s' is not a valid URI\n", |
9061 | 246k | uri, NULL, NULL); |
9062 | 246k | } else { |
9063 | 114k | if (parsedUri->scheme == NULL) { |
9064 | 73.3k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9065 | 73.3k | "xmlns: URI %s is not absolute\n", |
9066 | 73.3k | uri, NULL, NULL); |
9067 | 73.3k | } |
9068 | 114k | xmlFreeURI(parsedUri); |
9069 | 114k | } |
9070 | 361k | if (uri == ctxt->str_xml_ns) { |
9071 | 477 | if (attname != ctxt->str_xml) { |
9072 | 477 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9073 | 477 | "xml namespace URI cannot be the default namespace\n", |
9074 | 477 | NULL, NULL, NULL); |
9075 | 477 | } |
9076 | 477 | goto next_attr; |
9077 | 477 | } |
9078 | 360k | if ((len == 29) && |
9079 | 360k | (xmlStrEqual(uri, |
9080 | 9.04k | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9081 | 2.68k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9082 | 2.68k | "reuse of the xmlns namespace name is forbidden\n", |
9083 | 2.68k | NULL, NULL, NULL); |
9084 | 2.68k | goto next_attr; |
9085 | 2.68k | } |
9086 | 360k | } |
9087 | | |
9088 | 372k | if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0) |
9089 | 344k | nbNs++; |
9090 | 2.77M | } else if (aprefix == ctxt->str_xmlns) { |
9091 | 546k | xmlHashedString huri; |
9092 | 546k | xmlURIPtr parsedUri; |
9093 | | |
9094 | 546k | huri = xmlDictLookupHashed(ctxt->dict, attvalue, len); |
9095 | 546k | uri = huri.name; |
9096 | 546k | if (uri == NULL) { |
9097 | 25 | xmlErrMemory(ctxt); |
9098 | 25 | goto next_attr; |
9099 | 25 | } |
9100 | | |
9101 | 546k | if (attname == ctxt->str_xml) { |
9102 | 2.56k | if (uri != ctxt->str_xml_ns) { |
9103 | 2.28k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9104 | 2.28k | "xml namespace prefix mapped to wrong URI\n", |
9105 | 2.28k | NULL, NULL, NULL); |
9106 | 2.28k | } |
9107 | | /* |
9108 | | * Do not keep a namespace definition node |
9109 | | */ |
9110 | 2.56k | goto next_attr; |
9111 | 2.56k | } |
9112 | 544k | if (uri == ctxt->str_xml_ns) { |
9113 | 567 | if (attname != ctxt->str_xml) { |
9114 | 567 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9115 | 567 | "xml namespace URI mapped to wrong prefix\n", |
9116 | 567 | NULL, NULL, NULL); |
9117 | 567 | } |
9118 | 567 | goto next_attr; |
9119 | 567 | } |
9120 | 543k | if (attname == ctxt->str_xmlns) { |
9121 | 3.35k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9122 | 3.35k | "redefinition of the xmlns prefix is forbidden\n", |
9123 | 3.35k | NULL, NULL, NULL); |
9124 | 3.35k | goto next_attr; |
9125 | 3.35k | } |
9126 | 540k | if ((len == 29) && |
9127 | 540k | (xmlStrEqual(uri, |
9128 | 11.3k | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9129 | 1.24k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9130 | 1.24k | "reuse of the xmlns namespace name is forbidden\n", |
9131 | 1.24k | NULL, NULL, NULL); |
9132 | 1.24k | goto next_attr; |
9133 | 1.24k | } |
9134 | 538k | if ((uri == NULL) || (uri[0] == 0)) { |
9135 | 19.8k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9136 | 19.8k | "xmlns:%s: Empty XML namespace is not allowed\n", |
9137 | 19.8k | attname, NULL, NULL); |
9138 | 19.8k | goto next_attr; |
9139 | 518k | } else { |
9140 | 518k | if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) { |
9141 | 106 | xmlErrMemory(ctxt); |
9142 | 106 | goto next_attr; |
9143 | 106 | } |
9144 | 518k | if (parsedUri == NULL) { |
9145 | 108k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9146 | 108k | "xmlns:%s: '%s' is not a valid URI\n", |
9147 | 108k | attname, uri, NULL); |
9148 | 410k | } else { |
9149 | 410k | if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) { |
9150 | 102k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9151 | 102k | "xmlns:%s: URI %s is not absolute\n", |
9152 | 102k | attname, uri, NULL); |
9153 | 102k | } |
9154 | 410k | xmlFreeURI(parsedUri); |
9155 | 410k | } |
9156 | 518k | } |
9157 | | |
9158 | 518k | if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0) |
9159 | 486k | nbNs++; |
9160 | 2.23M | } else { |
9161 | | /* |
9162 | | * Populate attributes array, see above for repurposing |
9163 | | * of xmlChar pointers. |
9164 | | */ |
9165 | 2.23M | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9166 | 83.5k | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9167 | 85 | goto next_attr; |
9168 | 85 | } |
9169 | 83.4k | maxatts = ctxt->maxatts; |
9170 | 83.4k | atts = ctxt->atts; |
9171 | 83.4k | } |
9172 | 2.23M | ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) | |
9173 | 2.23M | ((unsigned) alloc << 31); |
9174 | 2.23M | atts[nbatts++] = attname; |
9175 | 2.23M | atts[nbatts++] = aprefix; |
9176 | 2.23M | atts[nbatts++] = (const xmlChar *) (size_t) haprefix.hashValue; |
9177 | 2.23M | if (alloc) { |
9178 | 459k | atts[nbatts++] = attvalue; |
9179 | 459k | attvalue += len; |
9180 | 459k | atts[nbatts++] = attvalue; |
9181 | 1.77M | } else { |
9182 | | /* |
9183 | | * attvalue points into the input buffer which can be |
9184 | | * reallocated. Store differences to input->base instead. |
9185 | | * The pointers will be reconstructed later. |
9186 | | */ |
9187 | 1.77M | atts[nbatts++] = (void *) (attvalue - BASE_PTR); |
9188 | 1.77M | attvalue += len; |
9189 | 1.77M | atts[nbatts++] = (void *) (attvalue - BASE_PTR); |
9190 | 1.77M | } |
9191 | | /* |
9192 | | * tag if some deallocation is needed |
9193 | | */ |
9194 | 2.23M | if (alloc != 0) attval = 1; |
9195 | 2.23M | attvalue = NULL; /* moved into atts */ |
9196 | 2.23M | } |
9197 | | |
9198 | 3.69M | next_attr: |
9199 | 3.69M | if ((attvalue != NULL) && (alloc != 0)) { |
9200 | 194k | xmlFree(attvalue); |
9201 | 194k | attvalue = NULL; |
9202 | 194k | } |
9203 | | |
9204 | 3.69M | GROW |
9205 | 3.69M | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
9206 | 1.17M | break; |
9207 | 2.51M | if (SKIP_BLANKS == 0) { |
9208 | 989k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
9209 | 989k | "attributes construct error\n"); |
9210 | 989k | break; |
9211 | 989k | } |
9212 | 1.52M | GROW; |
9213 | 1.52M | } |
9214 | | |
9215 | | /* |
9216 | | * Namespaces from default attributes |
9217 | | */ |
9218 | 6.60M | if (ctxt->attsDefault != NULL) { |
9219 | 1.61M | xmlDefAttrsPtr defaults; |
9220 | | |
9221 | 1.61M | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9222 | 1.61M | if (defaults != NULL) { |
9223 | 2.80M | for (i = 0; i < defaults->nbAttrs; i++) { |
9224 | 2.23M | xmlDefAttr *attr = &defaults->attrs[i]; |
9225 | | |
9226 | 2.23M | attname = attr->name.name; |
9227 | 2.23M | aprefix = attr->prefix.name; |
9228 | | |
9229 | 2.23M | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9230 | 127k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9231 | | |
9232 | 127k | if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0) |
9233 | 119k | nbNs++; |
9234 | 2.10M | } else if (aprefix == ctxt->str_xmlns) { |
9235 | 127k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9236 | | |
9237 | 127k | if (xmlParserNsPush(ctxt, &attr->name, &attr->value, |
9238 | 127k | NULL, 1) > 0) |
9239 | 119k | nbNs++; |
9240 | 1.97M | } else { |
9241 | 1.97M | nbTotalDef += 1; |
9242 | 1.97M | } |
9243 | 2.23M | } |
9244 | 570k | } |
9245 | 1.61M | } |
9246 | | |
9247 | | /* |
9248 | | * Resolve attribute namespaces |
9249 | | */ |
9250 | 8.83M | for (i = 0; i < nbatts; i += 5) { |
9251 | 2.23M | attname = atts[i]; |
9252 | 2.23M | aprefix = atts[i+1]; |
9253 | | |
9254 | | /* |
9255 | | * The default namespace does not apply to attribute names. |
9256 | | */ |
9257 | 2.23M | if (aprefix == NULL) { |
9258 | 1.30M | nsIndex = NS_INDEX_EMPTY; |
9259 | 1.30M | } else if (aprefix == ctxt->str_xml) { |
9260 | 331k | nsIndex = NS_INDEX_XML; |
9261 | 597k | } else { |
9262 | 597k | haprefix.name = aprefix; |
9263 | 597k | haprefix.hashValue = (size_t) atts[i+2]; |
9264 | 597k | nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL); |
9265 | | |
9266 | 597k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) { |
9267 | 289k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9268 | 289k | "Namespace prefix %s for %s on %s is not defined\n", |
9269 | 289k | aprefix, attname, localname); |
9270 | 289k | nsIndex = NS_INDEX_EMPTY; |
9271 | 289k | } |
9272 | 597k | } |
9273 | | |
9274 | 2.23M | atts[i+2] = (const xmlChar *) (ptrdiff_t) nsIndex; |
9275 | 2.23M | } |
9276 | | |
9277 | | /* |
9278 | | * Maximum number of attributes including default attributes. |
9279 | | */ |
9280 | 6.60M | maxAtts = nratts + nbTotalDef; |
9281 | | |
9282 | | /* |
9283 | | * Verify that attribute names are unique. |
9284 | | */ |
9285 | 6.60M | if (maxAtts > 1) { |
9286 | 481k | attrHashSize = 4; |
9287 | 766k | while (attrHashSize / 2 < (unsigned) maxAtts) |
9288 | 284k | attrHashSize *= 2; |
9289 | | |
9290 | 481k | if (attrHashSize > ctxt->attrHashMax) { |
9291 | 37.5k | xmlAttrHashBucket *tmp; |
9292 | | |
9293 | 37.5k | tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0])); |
9294 | 37.5k | if (tmp == NULL) { |
9295 | 21 | xmlErrMemory(ctxt); |
9296 | 21 | goto done; |
9297 | 21 | } |
9298 | | |
9299 | 37.4k | ctxt->attrHash = tmp; |
9300 | 37.4k | ctxt->attrHashMax = attrHashSize; |
9301 | 37.4k | } |
9302 | | |
9303 | 481k | memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0])); |
9304 | | |
9305 | 1.69M | for (i = 0, j = 0; j < nratts; i += 5, j++) { |
9306 | 1.21M | const xmlChar *nsuri; |
9307 | 1.21M | unsigned hashValue, nameHashValue, uriHashValue; |
9308 | 1.21M | int res; |
9309 | | |
9310 | 1.21M | attname = atts[i]; |
9311 | 1.21M | aprefix = atts[i+1]; |
9312 | 1.21M | nsIndex = (ptrdiff_t) atts[i+2]; |
9313 | | /* Hash values always have bit 31 set, see dict.c */ |
9314 | 1.21M | nameHashValue = ctxt->attallocs[j] | 0x80000000; |
9315 | | |
9316 | 1.21M | if (nsIndex == NS_INDEX_EMPTY) { |
9317 | | /* |
9318 | | * Prefix with empty namespace means an undeclared |
9319 | | * prefix which was already reported above. |
9320 | | */ |
9321 | 892k | if (aprefix != NULL) |
9322 | 246k | continue; |
9323 | 645k | nsuri = NULL; |
9324 | 645k | uriHashValue = URI_HASH_EMPTY; |
9325 | 645k | } else if (nsIndex == NS_INDEX_XML) { |
9326 | 95.3k | nsuri = ctxt->str_xml_ns; |
9327 | 95.3k | uriHashValue = URI_HASH_XML; |
9328 | 222k | } else { |
9329 | 222k | nsuri = ctxt->nsTab[nsIndex * 2 + 1]; |
9330 | 222k | uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue; |
9331 | 222k | } |
9332 | | |
9333 | 963k | hashValue = xmlDictCombineHash(nameHashValue, uriHashValue); |
9334 | 963k | res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri, |
9335 | 963k | hashValue, i); |
9336 | 963k | if (res < 0) |
9337 | 0 | continue; |
9338 | | |
9339 | | /* |
9340 | | * [ WFC: Unique Att Spec ] |
9341 | | * No attribute name may appear more than once in the same |
9342 | | * start-tag or empty-element tag. |
9343 | | * As extended by the Namespace in XML REC. |
9344 | | */ |
9345 | 963k | if (res < INT_MAX) { |
9346 | 90.2k | if (aprefix == atts[res+1]) { |
9347 | 84.0k | xmlErrAttributeDup(ctxt, aprefix, attname); |
9348 | 84.0k | } else { |
9349 | 6.28k | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9350 | 6.28k | "Namespaced Attribute %s in '%s' redefined\n", |
9351 | 6.28k | attname, nsuri, NULL); |
9352 | 6.28k | } |
9353 | 90.2k | } |
9354 | 963k | } |
9355 | 481k | } |
9356 | | |
9357 | | /* |
9358 | | * Default attributes |
9359 | | */ |
9360 | 6.60M | if (ctxt->attsDefault != NULL) { |
9361 | 1.61M | xmlDefAttrsPtr defaults; |
9362 | | |
9363 | 1.61M | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9364 | 1.61M | if (defaults != NULL) { |
9365 | 2.80M | for (i = 0; i < defaults->nbAttrs; i++) { |
9366 | 2.23M | xmlDefAttr *attr = &defaults->attrs[i]; |
9367 | 2.23M | const xmlChar *nsuri; |
9368 | 2.23M | unsigned hashValue, uriHashValue; |
9369 | 2.23M | int res; |
9370 | | |
9371 | 2.23M | attname = attr->name.name; |
9372 | 2.23M | aprefix = attr->prefix.name; |
9373 | | |
9374 | 2.23M | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) |
9375 | 127k | continue; |
9376 | 2.10M | if (aprefix == ctxt->str_xmlns) |
9377 | 127k | continue; |
9378 | | |
9379 | 1.97M | if (aprefix == NULL) { |
9380 | 1.66M | nsIndex = NS_INDEX_EMPTY; |
9381 | 1.66M | nsuri = NULL; |
9382 | 1.66M | uriHashValue = URI_HASH_EMPTY; |
9383 | 1.97M | } if (aprefix == ctxt->str_xml) { |
9384 | 36.9k | nsIndex = NS_INDEX_XML; |
9385 | 36.9k | nsuri = ctxt->str_xml_ns; |
9386 | 36.9k | uriHashValue = URI_HASH_XML; |
9387 | 1.94M | } else if (aprefix != NULL) { |
9388 | 282k | nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL); |
9389 | 282k | if ((nsIndex == INT_MAX) || |
9390 | 282k | (nsIndex < ctxt->nsdb->minNsIndex)) { |
9391 | 163k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9392 | 163k | "Namespace prefix %s for %s on %s is not " |
9393 | 163k | "defined\n", |
9394 | 163k | aprefix, attname, localname); |
9395 | 163k | nsIndex = NS_INDEX_EMPTY; |
9396 | 163k | nsuri = NULL; |
9397 | 163k | uriHashValue = URI_HASH_EMPTY; |
9398 | 163k | } else { |
9399 | 118k | nsuri = ctxt->nsTab[nsIndex * 2 + 1]; |
9400 | 118k | uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue; |
9401 | 118k | } |
9402 | 282k | } |
9403 | | |
9404 | | /* |
9405 | | * Check whether the attribute exists |
9406 | | */ |
9407 | 1.97M | if (maxAtts > 1) { |
9408 | 1.72M | hashValue = xmlDictCombineHash(attr->name.hashValue, |
9409 | 1.72M | uriHashValue); |
9410 | 1.72M | res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri, |
9411 | 1.72M | hashValue, nbatts); |
9412 | 1.72M | if (res < 0) |
9413 | 0 | continue; |
9414 | 1.72M | if (res < INT_MAX) { |
9415 | 103k | if (aprefix == atts[res+1]) |
9416 | 24.8k | continue; |
9417 | 78.9k | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9418 | 78.9k | "Namespaced Attribute %s in '%s' redefined\n", |
9419 | 78.9k | attname, nsuri, NULL); |
9420 | 78.9k | } |
9421 | 1.72M | } |
9422 | | |
9423 | 1.95M | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9424 | | |
9425 | 1.95M | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9426 | 12.8k | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9427 | 34 | localname = NULL; |
9428 | 34 | goto done; |
9429 | 34 | } |
9430 | 12.8k | maxatts = ctxt->maxatts; |
9431 | 12.8k | atts = ctxt->atts; |
9432 | 12.8k | } |
9433 | | |
9434 | 1.95M | atts[nbatts++] = attname; |
9435 | 1.95M | atts[nbatts++] = aprefix; |
9436 | 1.95M | atts[nbatts++] = (const xmlChar *) (ptrdiff_t) nsIndex; |
9437 | 1.95M | atts[nbatts++] = attr->value.name; |
9438 | 1.95M | atts[nbatts++] = attr->valueEnd; |
9439 | 1.95M | if ((ctxt->standalone == 1) && (attr->external != 0)) { |
9440 | 784 | xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED, |
9441 | 784 | "standalone: attribute %s on %s defaulted " |
9442 | 784 | "from external subset\n", |
9443 | 784 | attname, localname); |
9444 | 784 | } |
9445 | 1.95M | nbdef++; |
9446 | 1.95M | } |
9447 | 570k | } |
9448 | 1.61M | } |
9449 | | |
9450 | | /* |
9451 | | * Reconstruct attribute pointers |
9452 | | */ |
9453 | 10.7M | for (i = 0, j = 0; i < nbatts; i += 5, j++) { |
9454 | | /* namespace URI */ |
9455 | 4.18M | nsIndex = (ptrdiff_t) atts[i+2]; |
9456 | 4.18M | if (nsIndex == INT_MAX) |
9457 | 3.40M | atts[i+2] = NULL; |
9458 | 786k | else if (nsIndex == INT_MAX - 1) |
9459 | 365k | atts[i+2] = ctxt->str_xml_ns; |
9460 | 421k | else |
9461 | 421k | atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1]; |
9462 | | |
9463 | 4.18M | if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) { |
9464 | 1.77M | atts[i+3] = BASE_PTR + (ptrdiff_t) atts[i+3]; /* value */ |
9465 | 1.77M | atts[i+4] = BASE_PTR + (ptrdiff_t) atts[i+4]; /* valuend */ |
9466 | 1.77M | } |
9467 | 4.18M | } |
9468 | | |
9469 | 6.60M | uri = xmlParserNsLookupUri(ctxt, &hprefix); |
9470 | 6.60M | if ((prefix != NULL) && (uri == NULL)) { |
9471 | 368k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9472 | 368k | "Namespace prefix %s on %s is not defined\n", |
9473 | 368k | prefix, localname, NULL); |
9474 | 368k | } |
9475 | 6.60M | *pref = prefix; |
9476 | 6.60M | *URI = uri; |
9477 | | |
9478 | | /* |
9479 | | * SAX callback |
9480 | | */ |
9481 | 6.60M | if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) && |
9482 | 6.60M | (!ctxt->disableSAX)) { |
9483 | 6.01M | if (nbNs > 0) |
9484 | 748k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri, |
9485 | 748k | nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs), |
9486 | 748k | nbatts / 5, nbdef, atts); |
9487 | 5.26M | else |
9488 | 5.26M | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri, |
9489 | 5.26M | 0, NULL, nbatts / 5, nbdef, atts); |
9490 | 6.01M | } |
9491 | | |
9492 | 6.60M | done: |
9493 | | /* |
9494 | | * Free allocated attribute values |
9495 | | */ |
9496 | 6.60M | if (attval != 0) { |
9497 | 950k | for (i = 0, j = 0; j < nratts; i += 5, j++) |
9498 | 566k | if (ctxt->attallocs[j] & 0x80000000) |
9499 | 459k | xmlFree((xmlChar *) atts[i+3]); |
9500 | 384k | } |
9501 | | |
9502 | 6.60M | *nbNsPtr = nbNs; |
9503 | 6.60M | return(localname); |
9504 | 6.60M | } |
9505 | | |
9506 | | /** |
9507 | | * xmlParseEndTag2: |
9508 | | * @ctxt: an XML parser context |
9509 | | * @line: line of the start tag |
9510 | | * @nsNr: number of namespaces on the start tag |
9511 | | * |
9512 | | * Parse an end tag. Always consumes '</'. |
9513 | | * |
9514 | | * [42] ETag ::= '</' Name S? '>' |
9515 | | * |
9516 | | * With namespace |
9517 | | * |
9518 | | * [NS 9] ETag ::= '</' QName S? '>' |
9519 | | */ |
9520 | | |
9521 | | static void |
9522 | 1.48M | xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) { |
9523 | 1.48M | const xmlChar *name; |
9524 | | |
9525 | 1.48M | GROW; |
9526 | 1.48M | if ((RAW != '<') || (NXT(1) != '/')) { |
9527 | 6.51k | xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL); |
9528 | 6.51k | return; |
9529 | 6.51k | } |
9530 | 1.48M | SKIP(2); |
9531 | | |
9532 | 1.48M | if (tag->prefix == NULL) |
9533 | 1.22M | name = xmlParseNameAndCompare(ctxt, ctxt->name); |
9534 | 261k | else |
9535 | 261k | name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix); |
9536 | | |
9537 | | /* |
9538 | | * We should definitely be at the ending "S? '>'" part |
9539 | | */ |
9540 | 1.48M | GROW; |
9541 | 1.48M | SKIP_BLANKS; |
9542 | 1.48M | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
9543 | 230k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
9544 | 230k | } else |
9545 | 1.25M | NEXT1; |
9546 | | |
9547 | | /* |
9548 | | * [ WFC: Element Type Match ] |
9549 | | * The Name in an element's end-tag must match the element type in the |
9550 | | * start-tag. |
9551 | | * |
9552 | | */ |
9553 | 1.48M | if (name != (xmlChar*)1) { |
9554 | 358k | if (name == NULL) name = BAD_CAST "unparsable"; |
9555 | 358k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
9556 | 358k | "Opening and ending tag mismatch: %s line %d and %s\n", |
9557 | 358k | ctxt->name, tag->line, name); |
9558 | 358k | } |
9559 | | |
9560 | | /* |
9561 | | * SAX: End of Tag |
9562 | | */ |
9563 | 1.48M | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9564 | 1.48M | (!ctxt->disableSAX)) |
9565 | 1.35M | ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix, |
9566 | 1.35M | tag->URI); |
9567 | | |
9568 | 1.48M | spacePop(ctxt); |
9569 | 1.48M | if (tag->nsNr != 0) |
9570 | 141k | xmlParserNsPop(ctxt, tag->nsNr); |
9571 | 1.48M | } |
9572 | | |
9573 | | /** |
9574 | | * xmlParseCDSect: |
9575 | | * @ctxt: an XML parser context |
9576 | | * |
9577 | | * DEPRECATED: Internal function, don't use. |
9578 | | * |
9579 | | * Parse escaped pure raw content. Always consumes '<!['. |
9580 | | * |
9581 | | * [18] CDSect ::= CDStart CData CDEnd |
9582 | | * |
9583 | | * [19] CDStart ::= '<![CDATA[' |
9584 | | * |
9585 | | * [20] Data ::= (Char* - (Char* ']]>' Char*)) |
9586 | | * |
9587 | | * [21] CDEnd ::= ']]>' |
9588 | | */ |
9589 | | void |
9590 | 965k | xmlParseCDSect(xmlParserCtxtPtr ctxt) { |
9591 | 965k | xmlChar *buf = NULL; |
9592 | 965k | int len = 0; |
9593 | 965k | int size = XML_PARSER_BUFFER_SIZE; |
9594 | 965k | int r, rl; |
9595 | 965k | int s, sl; |
9596 | 965k | int cur, l; |
9597 | 965k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9598 | 127k | XML_MAX_HUGE_LENGTH : |
9599 | 965k | XML_MAX_TEXT_LENGTH; |
9600 | | |
9601 | 965k | if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '[')) |
9602 | 0 | return; |
9603 | 965k | SKIP(3); |
9604 | | |
9605 | 965k | if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '[')) |
9606 | 0 | return; |
9607 | 965k | SKIP(6); |
9608 | | |
9609 | 965k | r = xmlCurrentCharRecover(ctxt, &rl); |
9610 | 965k | if (!IS_CHAR(r)) { |
9611 | 7.36k | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9612 | 7.36k | goto out; |
9613 | 7.36k | } |
9614 | 957k | NEXTL(rl); |
9615 | 957k | s = xmlCurrentCharRecover(ctxt, &sl); |
9616 | 957k | if (!IS_CHAR(s)) { |
9617 | 17.0k | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9618 | 17.0k | goto out; |
9619 | 17.0k | } |
9620 | 940k | NEXTL(sl); |
9621 | 940k | cur = xmlCurrentCharRecover(ctxt, &l); |
9622 | 940k | buf = xmlMalloc(size); |
9623 | 940k | if (buf == NULL) { |
9624 | 40 | xmlErrMemory(ctxt); |
9625 | 40 | goto out; |
9626 | 40 | } |
9627 | 46.5M | while (IS_CHAR(cur) && |
9628 | 46.5M | ((r != ']') || (s != ']') || (cur != '>'))) { |
9629 | 45.6M | if (len + 5 >= size) { |
9630 | 21.3k | xmlChar *tmp; |
9631 | | |
9632 | 21.3k | tmp = (xmlChar *) xmlRealloc(buf, size * 2); |
9633 | 21.3k | if (tmp == NULL) { |
9634 | 16 | xmlErrMemory(ctxt); |
9635 | 16 | goto out; |
9636 | 16 | } |
9637 | 21.3k | buf = tmp; |
9638 | 21.3k | size *= 2; |
9639 | 21.3k | } |
9640 | 45.6M | COPY_BUF(buf, len, r); |
9641 | 45.6M | if (len > maxLength) { |
9642 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9643 | 0 | "CData section too big found\n"); |
9644 | 0 | goto out; |
9645 | 0 | } |
9646 | 45.6M | r = s; |
9647 | 45.6M | rl = sl; |
9648 | 45.6M | s = cur; |
9649 | 45.6M | sl = l; |
9650 | 45.6M | NEXTL(l); |
9651 | 45.6M | cur = xmlCurrentCharRecover(ctxt, &l); |
9652 | 45.6M | } |
9653 | 940k | buf[len] = 0; |
9654 | 940k | if (cur != '>') { |
9655 | 41.7k | xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9656 | 41.7k | "CData section not finished\n%.50s\n", buf); |
9657 | 41.7k | goto out; |
9658 | 41.7k | } |
9659 | 898k | NEXTL(l); |
9660 | | |
9661 | | /* |
9662 | | * OK the buffer is to be consumed as cdata. |
9663 | | */ |
9664 | 898k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
9665 | 893k | if (ctxt->sax->cdataBlock != NULL) |
9666 | 367k | ctxt->sax->cdataBlock(ctxt->userData, buf, len); |
9667 | 526k | else if (ctxt->sax->characters != NULL) |
9668 | 525k | ctxt->sax->characters(ctxt->userData, buf, len); |
9669 | 893k | } |
9670 | | |
9671 | 965k | out: |
9672 | 965k | xmlFree(buf); |
9673 | 965k | } |
9674 | | |
9675 | | /** |
9676 | | * xmlParseContentInternal: |
9677 | | * @ctxt: an XML parser context |
9678 | | * |
9679 | | * Parse a content sequence. Stops at EOF or '</'. Leaves checking of |
9680 | | * unexpected EOF to the caller. |
9681 | | */ |
9682 | | |
9683 | | static void |
9684 | 210k | xmlParseContentInternal(xmlParserCtxtPtr ctxt) { |
9685 | 210k | int oldNameNr = ctxt->nameNr; |
9686 | 210k | int oldSpaceNr = ctxt->spaceNr; |
9687 | 210k | int oldNodeNr = ctxt->nodeNr; |
9688 | | |
9689 | 210k | GROW; |
9690 | 69.0M | while ((ctxt->input->cur < ctxt->input->end) && |
9691 | 69.0M | (PARSER_STOPPED(ctxt) == 0)) { |
9692 | 68.9M | const xmlChar *cur = ctxt->input->cur; |
9693 | | |
9694 | | /* |
9695 | | * First case : a Processing Instruction. |
9696 | | */ |
9697 | 68.9M | if ((*cur == '<') && (cur[1] == '?')) { |
9698 | 207k | xmlParsePI(ctxt); |
9699 | 207k | } |
9700 | | |
9701 | | /* |
9702 | | * Second case : a CDSection |
9703 | | */ |
9704 | | /* 2.6.0 test was *cur not RAW */ |
9705 | 68.7M | else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9706 | 965k | xmlParseCDSect(ctxt); |
9707 | 965k | } |
9708 | | |
9709 | | /* |
9710 | | * Third case : a comment |
9711 | | */ |
9712 | 67.7M | else if ((*cur == '<') && (NXT(1) == '!') && |
9713 | 67.7M | (NXT(2) == '-') && (NXT(3) == '-')) { |
9714 | 462k | xmlParseComment(ctxt); |
9715 | 462k | } |
9716 | | |
9717 | | /* |
9718 | | * Fourth case : a sub-element. |
9719 | | */ |
9720 | 67.2M | else if (*cur == '<') { |
9721 | 13.5M | if (NXT(1) == '/') { |
9722 | 2.40M | if (ctxt->nameNr <= oldNameNr) |
9723 | 67.8k | break; |
9724 | 2.33M | xmlParseElementEnd(ctxt); |
9725 | 11.1M | } else { |
9726 | 11.1M | xmlParseElementStart(ctxt); |
9727 | 11.1M | } |
9728 | 13.5M | } |
9729 | | |
9730 | | /* |
9731 | | * Fifth case : a reference. If if has not been resolved, |
9732 | | * parsing returns it's Name, create the node |
9733 | | */ |
9734 | | |
9735 | 53.7M | else if (*cur == '&') { |
9736 | 2.59M | xmlParseReference(ctxt); |
9737 | 2.59M | } |
9738 | | |
9739 | | /* |
9740 | | * Last case, text. Note that References are handled directly. |
9741 | | */ |
9742 | 51.1M | else { |
9743 | 51.1M | xmlParseCharDataInternal(ctxt, 0); |
9744 | 51.1M | } |
9745 | | |
9746 | 68.8M | SHRINK; |
9747 | 68.8M | GROW; |
9748 | 68.8M | } |
9749 | | |
9750 | 210k | if ((ctxt->nameNr > oldNameNr) && |
9751 | 210k | (ctxt->input->cur >= ctxt->input->end) && |
9752 | 210k | (ctxt->wellFormed)) { |
9753 | 4.71k | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9754 | 4.71k | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9755 | 4.71k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9756 | 4.71k | "Premature end of data in tag %s line %d\n", |
9757 | 4.71k | name, line, NULL); |
9758 | 4.71k | } |
9759 | | |
9760 | | /* |
9761 | | * Clean up in error case |
9762 | | */ |
9763 | | |
9764 | 1.90M | while (ctxt->nodeNr > oldNodeNr) |
9765 | 1.69M | nodePop(ctxt); |
9766 | | |
9767 | 2.12M | while (ctxt->nameNr > oldNameNr) { |
9768 | 1.91M | xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1]; |
9769 | | |
9770 | 1.91M | if (tag->nsNr != 0) |
9771 | 323k | xmlParserNsPop(ctxt, tag->nsNr); |
9772 | | |
9773 | 1.91M | namePop(ctxt); |
9774 | 1.91M | } |
9775 | | |
9776 | 2.12M | while (ctxt->spaceNr > oldSpaceNr) |
9777 | 1.91M | spacePop(ctxt); |
9778 | 210k | } |
9779 | | |
9780 | | /** |
9781 | | * xmlParseContent: |
9782 | | * @ctxt: an XML parser context |
9783 | | * |
9784 | | * Parse XML element content. This is useful if you're only interested |
9785 | | * in custom SAX callbacks. If you want a node list, use |
9786 | | * xmlCtxtParseContent. |
9787 | | */ |
9788 | | void |
9789 | 0 | xmlParseContent(xmlParserCtxtPtr ctxt) { |
9790 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
9791 | 0 | return; |
9792 | | |
9793 | 0 | xmlCtxtInitializeLate(ctxt); |
9794 | |
|
9795 | 0 | xmlParseContentInternal(ctxt); |
9796 | |
|
9797 | 0 | if (ctxt->input->cur < ctxt->input->end) |
9798 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
9799 | 0 | } |
9800 | | |
9801 | | /** |
9802 | | * xmlParseElement: |
9803 | | * @ctxt: an XML parser context |
9804 | | * |
9805 | | * DEPRECATED: Internal function, don't use. |
9806 | | * |
9807 | | * parse an XML element |
9808 | | * |
9809 | | * [39] element ::= EmptyElemTag | STag content ETag |
9810 | | * |
9811 | | * [ WFC: Element Type Match ] |
9812 | | * The Name in an element's end-tag must match the element type in the |
9813 | | * start-tag. |
9814 | | * |
9815 | | */ |
9816 | | |
9817 | | void |
9818 | 213k | xmlParseElement(xmlParserCtxtPtr ctxt) { |
9819 | 213k | if (xmlParseElementStart(ctxt) != 0) |
9820 | 63.4k | return; |
9821 | | |
9822 | 150k | xmlParseContentInternal(ctxt); |
9823 | | |
9824 | 150k | if (ctxt->input->cur >= ctxt->input->end) { |
9825 | 72.7k | if (ctxt->wellFormed) { |
9826 | 5.24k | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9827 | 5.24k | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9828 | 5.24k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9829 | 5.24k | "Premature end of data in tag %s line %d\n", |
9830 | 5.24k | name, line, NULL); |
9831 | 5.24k | } |
9832 | 72.7k | return; |
9833 | 72.7k | } |
9834 | | |
9835 | 77.6k | xmlParseElementEnd(ctxt); |
9836 | 77.6k | } |
9837 | | |
9838 | | /** |
9839 | | * xmlParseElementStart: |
9840 | | * @ctxt: an XML parser context |
9841 | | * |
9842 | | * Parse the start of an XML element. Returns -1 in case of error, 0 if an |
9843 | | * opening tag was parsed, 1 if an empty element was parsed. |
9844 | | * |
9845 | | * Always consumes '<'. |
9846 | | */ |
9847 | | static int |
9848 | 11.3M | xmlParseElementStart(xmlParserCtxtPtr ctxt) { |
9849 | 11.3M | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
9850 | 11.3M | const xmlChar *name; |
9851 | 11.3M | const xmlChar *prefix = NULL; |
9852 | 11.3M | const xmlChar *URI = NULL; |
9853 | 11.3M | xmlParserNodeInfo node_info; |
9854 | 11.3M | int line; |
9855 | 11.3M | xmlNodePtr cur; |
9856 | 11.3M | int nbNs = 0; |
9857 | | |
9858 | 11.3M | if (ctxt->nameNr > maxDepth) { |
9859 | 217 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
9860 | 217 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
9861 | 217 | ctxt->nameNr); |
9862 | 217 | xmlHaltParser(ctxt); |
9863 | 217 | return(-1); |
9864 | 217 | } |
9865 | | |
9866 | | /* Capture start position */ |
9867 | 11.3M | if (ctxt->record_info) { |
9868 | 0 | node_info.begin_pos = ctxt->input->consumed + |
9869 | 0 | (CUR_PTR - ctxt->input->base); |
9870 | 0 | node_info.begin_line = ctxt->input->line; |
9871 | 0 | } |
9872 | | |
9873 | 11.3M | if (ctxt->spaceNr == 0) |
9874 | 83.5k | spacePush(ctxt, -1); |
9875 | 11.2M | else if (*ctxt->space == -2) |
9876 | 2.85M | spacePush(ctxt, -1); |
9877 | 8.44M | else |
9878 | 8.44M | spacePush(ctxt, *ctxt->space); |
9879 | | |
9880 | 11.3M | line = ctxt->input->line; |
9881 | 11.3M | #ifdef LIBXML_SAX1_ENABLED |
9882 | 11.3M | if (ctxt->sax2) |
9883 | 7.58M | #endif /* LIBXML_SAX1_ENABLED */ |
9884 | 7.58M | name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs); |
9885 | 3.79M | #ifdef LIBXML_SAX1_ENABLED |
9886 | 3.79M | else |
9887 | 3.79M | name = xmlParseStartTag(ctxt); |
9888 | 11.3M | #endif /* LIBXML_SAX1_ENABLED */ |
9889 | 11.3M | if (name == NULL) { |
9890 | 2.18M | spacePop(ctxt); |
9891 | 2.18M | return(-1); |
9892 | 2.18M | } |
9893 | 9.19M | nameNsPush(ctxt, name, prefix, URI, line, nbNs); |
9894 | 9.19M | cur = ctxt->node; |
9895 | | |
9896 | 9.19M | #ifdef LIBXML_VALID_ENABLED |
9897 | | /* |
9898 | | * [ VC: Root Element Type ] |
9899 | | * The Name in the document type declaration must match the element |
9900 | | * type of the root element. |
9901 | | */ |
9902 | 9.19M | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
9903 | 9.19M | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
9904 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
9905 | 9.19M | #endif /* LIBXML_VALID_ENABLED */ |
9906 | | |
9907 | | /* |
9908 | | * Check for an Empty Element. |
9909 | | */ |
9910 | 9.19M | if ((RAW == '/') && (NXT(1) == '>')) { |
9911 | 1.36M | SKIP(2); |
9912 | 1.36M | if (ctxt->sax2) { |
9913 | 799k | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9914 | 799k | (!ctxt->disableSAX)) |
9915 | 757k | ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI); |
9916 | 799k | #ifdef LIBXML_SAX1_ENABLED |
9917 | 799k | } else { |
9918 | 564k | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
9919 | 564k | (!ctxt->disableSAX)) |
9920 | 555k | ctxt->sax->endElement(ctxt->userData, name); |
9921 | 564k | #endif /* LIBXML_SAX1_ENABLED */ |
9922 | 564k | } |
9923 | 1.36M | namePop(ctxt); |
9924 | 1.36M | spacePop(ctxt); |
9925 | 1.36M | if (nbNs > 0) |
9926 | 72.2k | xmlParserNsPop(ctxt, nbNs); |
9927 | 1.36M | if (cur != NULL && ctxt->record_info) { |
9928 | 0 | node_info.node = cur; |
9929 | 0 | node_info.end_pos = ctxt->input->consumed + |
9930 | 0 | (CUR_PTR - ctxt->input->base); |
9931 | 0 | node_info.end_line = ctxt->input->line; |
9932 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
9933 | 0 | } |
9934 | 1.36M | return(1); |
9935 | 1.36M | } |
9936 | 7.82M | if (RAW == '>') { |
9937 | 4.39M | NEXT1; |
9938 | 4.39M | if (cur != NULL && ctxt->record_info) { |
9939 | 0 | node_info.node = cur; |
9940 | 0 | node_info.end_pos = 0; |
9941 | 0 | node_info.end_line = 0; |
9942 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
9943 | 0 | } |
9944 | 4.39M | } else { |
9945 | 3.43M | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED, |
9946 | 3.43M | "Couldn't find end of Start Tag %s line %d\n", |
9947 | 3.43M | name, line, NULL); |
9948 | | |
9949 | | /* |
9950 | | * end of parsing of this node. |
9951 | | */ |
9952 | 3.43M | nodePop(ctxt); |
9953 | 3.43M | namePop(ctxt); |
9954 | 3.43M | spacePop(ctxt); |
9955 | 3.43M | if (nbNs > 0) |
9956 | 245k | xmlParserNsPop(ctxt, nbNs); |
9957 | 3.43M | return(-1); |
9958 | 3.43M | } |
9959 | | |
9960 | 4.39M | return(0); |
9961 | 7.82M | } |
9962 | | |
9963 | | /** |
9964 | | * xmlParseElementEnd: |
9965 | | * @ctxt: an XML parser context |
9966 | | * |
9967 | | * Parse the end of an XML element. Always consumes '</'. |
9968 | | */ |
9969 | | static void |
9970 | 2.41M | xmlParseElementEnd(xmlParserCtxtPtr ctxt) { |
9971 | 2.41M | xmlNodePtr cur = ctxt->node; |
9972 | | |
9973 | 2.41M | if (ctxt->nameNr <= 0) { |
9974 | 139 | if ((RAW == '<') && (NXT(1) == '/')) |
9975 | 23 | SKIP(2); |
9976 | 139 | return; |
9977 | 139 | } |
9978 | | |
9979 | | /* |
9980 | | * parse the end of tag: '</' should be here. |
9981 | | */ |
9982 | 2.41M | if (ctxt->sax2) { |
9983 | 1.29M | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
9984 | 1.29M | namePop(ctxt); |
9985 | 1.29M | } |
9986 | 1.11M | #ifdef LIBXML_SAX1_ENABLED |
9987 | 1.11M | else |
9988 | 1.11M | xmlParseEndTag1(ctxt, 0); |
9989 | 2.41M | #endif /* LIBXML_SAX1_ENABLED */ |
9990 | | |
9991 | | /* |
9992 | | * Capture end position |
9993 | | */ |
9994 | 2.41M | if (cur != NULL && ctxt->record_info) { |
9995 | 0 | xmlParserNodeInfoPtr node_info; |
9996 | |
|
9997 | 0 | node_info = (xmlParserNodeInfoPtr) xmlParserFindNodeInfo(ctxt, cur); |
9998 | 0 | if (node_info != NULL) { |
9999 | 0 | node_info->end_pos = ctxt->input->consumed + |
10000 | 0 | (CUR_PTR - ctxt->input->base); |
10001 | 0 | node_info->end_line = ctxt->input->line; |
10002 | 0 | } |
10003 | 0 | } |
10004 | 2.41M | } |
10005 | | |
10006 | | /** |
10007 | | * xmlParseVersionNum: |
10008 | | * @ctxt: an XML parser context |
10009 | | * |
10010 | | * DEPRECATED: Internal function, don't use. |
10011 | | * |
10012 | | * parse the XML version value. |
10013 | | * |
10014 | | * [26] VersionNum ::= '1.' [0-9]+ |
10015 | | * |
10016 | | * In practice allow [0-9].[0-9]+ at that level |
10017 | | * |
10018 | | * Returns the string giving the XML version number, or NULL |
10019 | | */ |
10020 | | xmlChar * |
10021 | 200k | xmlParseVersionNum(xmlParserCtxtPtr ctxt) { |
10022 | 200k | xmlChar *buf = NULL; |
10023 | 200k | int len = 0; |
10024 | 200k | int size = 10; |
10025 | 200k | xmlChar cur; |
10026 | | |
10027 | 200k | buf = xmlMalloc(size); |
10028 | 200k | if (buf == NULL) { |
10029 | 185 | xmlErrMemory(ctxt); |
10030 | 185 | return(NULL); |
10031 | 185 | } |
10032 | 200k | cur = CUR; |
10033 | 200k | if (!((cur >= '0') && (cur <= '9'))) { |
10034 | 13.3k | xmlFree(buf); |
10035 | 13.3k | return(NULL); |
10036 | 13.3k | } |
10037 | 187k | buf[len++] = cur; |
10038 | 187k | NEXT; |
10039 | 187k | cur=CUR; |
10040 | 187k | if (cur != '.') { |
10041 | 5.71k | xmlFree(buf); |
10042 | 5.71k | return(NULL); |
10043 | 5.71k | } |
10044 | 181k | buf[len++] = cur; |
10045 | 181k | NEXT; |
10046 | 181k | cur=CUR; |
10047 | 476k | while ((cur >= '0') && (cur <= '9')) { |
10048 | 294k | if (len + 1 >= size) { |
10049 | 5.40k | xmlChar *tmp; |
10050 | | |
10051 | 5.40k | size *= 2; |
10052 | 5.40k | tmp = (xmlChar *) xmlRealloc(buf, size); |
10053 | 5.40k | if (tmp == NULL) { |
10054 | 16 | xmlFree(buf); |
10055 | 16 | xmlErrMemory(ctxt); |
10056 | 16 | return(NULL); |
10057 | 16 | } |
10058 | 5.39k | buf = tmp; |
10059 | 5.39k | } |
10060 | 294k | buf[len++] = cur; |
10061 | 294k | NEXT; |
10062 | 294k | cur=CUR; |
10063 | 294k | } |
10064 | 181k | buf[len] = 0; |
10065 | 181k | return(buf); |
10066 | 181k | } |
10067 | | |
10068 | | /** |
10069 | | * xmlParseVersionInfo: |
10070 | | * @ctxt: an XML parser context |
10071 | | * |
10072 | | * DEPRECATED: Internal function, don't use. |
10073 | | * |
10074 | | * parse the XML version. |
10075 | | * |
10076 | | * [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ") |
10077 | | * |
10078 | | * [25] Eq ::= S? '=' S? |
10079 | | * |
10080 | | * Returns the version string, e.g. "1.0" |
10081 | | */ |
10082 | | |
10083 | | xmlChar * |
10084 | 279k | xmlParseVersionInfo(xmlParserCtxtPtr ctxt) { |
10085 | 279k | xmlChar *version = NULL; |
10086 | | |
10087 | 279k | if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) { |
10088 | 209k | SKIP(7); |
10089 | 209k | SKIP_BLANKS; |
10090 | 209k | if (RAW != '=') { |
10091 | 4.91k | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10092 | 4.91k | return(NULL); |
10093 | 4.91k | } |
10094 | 204k | NEXT; |
10095 | 204k | SKIP_BLANKS; |
10096 | 204k | if (RAW == '"') { |
10097 | 166k | NEXT; |
10098 | 166k | version = xmlParseVersionNum(ctxt); |
10099 | 166k | if (RAW != '"') { |
10100 | 11.0k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10101 | 11.0k | } else |
10102 | 155k | NEXT; |
10103 | 166k | } else if (RAW == '\''){ |
10104 | 34.1k | NEXT; |
10105 | 34.1k | version = xmlParseVersionNum(ctxt); |
10106 | 34.1k | if (RAW != '\'') { |
10107 | 11.0k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10108 | 11.0k | } else |
10109 | 23.1k | NEXT; |
10110 | 34.1k | } else { |
10111 | 3.97k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10112 | 3.97k | } |
10113 | 204k | } |
10114 | 274k | return(version); |
10115 | 279k | } |
10116 | | |
10117 | | /** |
10118 | | * xmlParseEncName: |
10119 | | * @ctxt: an XML parser context |
10120 | | * |
10121 | | * DEPRECATED: Internal function, don't use. |
10122 | | * |
10123 | | * parse the XML encoding name |
10124 | | * |
10125 | | * [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* |
10126 | | * |
10127 | | * Returns the encoding name value or NULL |
10128 | | */ |
10129 | | xmlChar * |
10130 | 69.6k | xmlParseEncName(xmlParserCtxtPtr ctxt) { |
10131 | 69.6k | xmlChar *buf = NULL; |
10132 | 69.6k | int len = 0; |
10133 | 69.6k | int size = 10; |
10134 | 69.6k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
10135 | 14.3k | XML_MAX_TEXT_LENGTH : |
10136 | 69.6k | XML_MAX_NAME_LENGTH; |
10137 | 69.6k | xmlChar cur; |
10138 | | |
10139 | 69.6k | cur = CUR; |
10140 | 69.6k | if (((cur >= 'a') && (cur <= 'z')) || |
10141 | 69.6k | ((cur >= 'A') && (cur <= 'Z'))) { |
10142 | 48.0k | buf = xmlMalloc(size); |
10143 | 48.0k | if (buf == NULL) { |
10144 | 80 | xmlErrMemory(ctxt); |
10145 | 80 | return(NULL); |
10146 | 80 | } |
10147 | | |
10148 | 48.0k | buf[len++] = cur; |
10149 | 48.0k | NEXT; |
10150 | 48.0k | cur = CUR; |
10151 | 6.79M | while (((cur >= 'a') && (cur <= 'z')) || |
10152 | 6.79M | ((cur >= 'A') && (cur <= 'Z')) || |
10153 | 6.79M | ((cur >= '0') && (cur <= '9')) || |
10154 | 6.79M | (cur == '.') || (cur == '_') || |
10155 | 6.79M | (cur == '-')) { |
10156 | 6.74M | if (len + 1 >= size) { |
10157 | 32.0k | xmlChar *tmp; |
10158 | | |
10159 | 32.0k | size *= 2; |
10160 | 32.0k | tmp = (xmlChar *) xmlRealloc(buf, size); |
10161 | 32.0k | if (tmp == NULL) { |
10162 | 41 | xmlErrMemory(ctxt); |
10163 | 41 | xmlFree(buf); |
10164 | 41 | return(NULL); |
10165 | 41 | } |
10166 | 32.0k | buf = tmp; |
10167 | 32.0k | } |
10168 | 6.74M | buf[len++] = cur; |
10169 | 6.74M | if (len > maxLength) { |
10170 | 83 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName"); |
10171 | 83 | xmlFree(buf); |
10172 | 83 | return(NULL); |
10173 | 83 | } |
10174 | 6.74M | NEXT; |
10175 | 6.74M | cur = CUR; |
10176 | 6.74M | } |
10177 | 47.8k | buf[len] = 0; |
10178 | 47.8k | } else { |
10179 | 21.5k | xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL); |
10180 | 21.5k | } |
10181 | 69.4k | return(buf); |
10182 | 69.6k | } |
10183 | | |
10184 | | /** |
10185 | | * xmlParseEncodingDecl: |
10186 | | * @ctxt: an XML parser context |
10187 | | * |
10188 | | * DEPRECATED: Internal function, don't use. |
10189 | | * |
10190 | | * parse the XML encoding declaration |
10191 | | * |
10192 | | * [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'") |
10193 | | * |
10194 | | * this setups the conversion filters. |
10195 | | * |
10196 | | * Returns the encoding value or NULL |
10197 | | */ |
10198 | | |
10199 | | const xmlChar * |
10200 | 266k | xmlParseEncodingDecl(xmlParserCtxtPtr ctxt) { |
10201 | 266k | xmlChar *encoding = NULL; |
10202 | | |
10203 | 266k | SKIP_BLANKS; |
10204 | 266k | if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0) |
10205 | 190k | return(NULL); |
10206 | | |
10207 | 75.6k | SKIP(8); |
10208 | 75.6k | SKIP_BLANKS; |
10209 | 75.6k | if (RAW != '=') { |
10210 | 3.04k | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10211 | 3.04k | return(NULL); |
10212 | 3.04k | } |
10213 | 72.5k | NEXT; |
10214 | 72.5k | SKIP_BLANKS; |
10215 | 72.5k | if (RAW == '"') { |
10216 | 46.1k | NEXT; |
10217 | 46.1k | encoding = xmlParseEncName(ctxt); |
10218 | 46.1k | if (RAW != '"') { |
10219 | 6.91k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10220 | 6.91k | xmlFree((xmlChar *) encoding); |
10221 | 6.91k | return(NULL); |
10222 | 6.91k | } else |
10223 | 39.2k | NEXT; |
10224 | 46.1k | } else if (RAW == '\''){ |
10225 | 23.5k | NEXT; |
10226 | 23.5k | encoding = xmlParseEncName(ctxt); |
10227 | 23.5k | if (RAW != '\'') { |
10228 | 9.11k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10229 | 9.11k | xmlFree((xmlChar *) encoding); |
10230 | 9.11k | return(NULL); |
10231 | 9.11k | } else |
10232 | 14.4k | NEXT; |
10233 | 23.5k | } else { |
10234 | 2.86k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10235 | 2.86k | } |
10236 | | |
10237 | 56.5k | if (encoding == NULL) |
10238 | 15.4k | return(NULL); |
10239 | | |
10240 | 41.0k | xmlSetDeclaredEncoding(ctxt, encoding); |
10241 | | |
10242 | 41.0k | return(ctxt->encoding); |
10243 | 56.5k | } |
10244 | | |
10245 | | /** |
10246 | | * xmlParseSDDecl: |
10247 | | * @ctxt: an XML parser context |
10248 | | * |
10249 | | * DEPRECATED: Internal function, don't use. |
10250 | | * |
10251 | | * parse the XML standalone declaration |
10252 | | * |
10253 | | * [32] SDDecl ::= S 'standalone' Eq |
10254 | | * (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"')) |
10255 | | * |
10256 | | * [ VC: Standalone Document Declaration ] |
10257 | | * TODO The standalone document declaration must have the value "no" |
10258 | | * if any external markup declarations contain declarations of: |
10259 | | * - attributes with default values, if elements to which these |
10260 | | * attributes apply appear in the document without specifications |
10261 | | * of values for these attributes, or |
10262 | | * - entities (other than amp, lt, gt, apos, quot), if references |
10263 | | * to those entities appear in the document, or |
10264 | | * - attributes with values subject to normalization, where the |
10265 | | * attribute appears in the document with a value which will change |
10266 | | * as a result of normalization, or |
10267 | | * - element types with element content, if white space occurs directly |
10268 | | * within any instance of those types. |
10269 | | * |
10270 | | * Returns: |
10271 | | * 1 if standalone="yes" |
10272 | | * 0 if standalone="no" |
10273 | | * -2 if standalone attribute is missing or invalid |
10274 | | * (A standalone value of -2 means that the XML declaration was found, |
10275 | | * but no value was specified for the standalone attribute). |
10276 | | */ |
10277 | | |
10278 | | int |
10279 | 156k | xmlParseSDDecl(xmlParserCtxtPtr ctxt) { |
10280 | 156k | int standalone = -2; |
10281 | | |
10282 | 156k | SKIP_BLANKS; |
10283 | 156k | if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) { |
10284 | 113k | SKIP(10); |
10285 | 113k | SKIP_BLANKS; |
10286 | 113k | if (RAW != '=') { |
10287 | 297 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10288 | 297 | return(standalone); |
10289 | 297 | } |
10290 | 113k | NEXT; |
10291 | 113k | SKIP_BLANKS; |
10292 | 113k | if (RAW == '\''){ |
10293 | 2.35k | NEXT; |
10294 | 2.35k | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10295 | 558 | standalone = 0; |
10296 | 558 | SKIP(2); |
10297 | 1.79k | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10298 | 1.79k | (NXT(2) == 's')) { |
10299 | 583 | standalone = 1; |
10300 | 583 | SKIP(3); |
10301 | 1.21k | } else { |
10302 | 1.21k | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10303 | 1.21k | } |
10304 | 2.35k | if (RAW != '\'') { |
10305 | 1.53k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10306 | 1.53k | } else |
10307 | 826 | NEXT; |
10308 | 111k | } else if (RAW == '"'){ |
10309 | 110k | NEXT; |
10310 | 110k | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10311 | 912 | standalone = 0; |
10312 | 912 | SKIP(2); |
10313 | 109k | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10314 | 109k | (NXT(2) == 's')) { |
10315 | 108k | standalone = 1; |
10316 | 108k | SKIP(3); |
10317 | 108k | } else { |
10318 | 1.10k | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10319 | 1.10k | } |
10320 | 110k | if (RAW != '"') { |
10321 | 1.74k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10322 | 1.74k | } else |
10323 | 109k | NEXT; |
10324 | 110k | } else { |
10325 | 307 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10326 | 307 | } |
10327 | 113k | } |
10328 | 156k | return(standalone); |
10329 | 156k | } |
10330 | | |
10331 | | /** |
10332 | | * xmlParseXMLDecl: |
10333 | | * @ctxt: an XML parser context |
10334 | | * |
10335 | | * DEPRECATED: Internal function, don't use. |
10336 | | * |
10337 | | * parse an XML declaration header |
10338 | | * |
10339 | | * [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' |
10340 | | */ |
10341 | | |
10342 | | void |
10343 | 184k | xmlParseXMLDecl(xmlParserCtxtPtr ctxt) { |
10344 | 184k | xmlChar *version; |
10345 | | |
10346 | | /* |
10347 | | * This value for standalone indicates that the document has an |
10348 | | * XML declaration but it does not have a standalone attribute. |
10349 | | * It will be overwritten later if a standalone attribute is found. |
10350 | | */ |
10351 | | |
10352 | 184k | ctxt->standalone = -2; |
10353 | | |
10354 | | /* |
10355 | | * We know that '<?xml' is here. |
10356 | | */ |
10357 | 184k | SKIP(5); |
10358 | | |
10359 | 184k | if (!IS_BLANK_CH(RAW)) { |
10360 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
10361 | 0 | "Blank needed after '<?xml'\n"); |
10362 | 0 | } |
10363 | 184k | SKIP_BLANKS; |
10364 | | |
10365 | | /* |
10366 | | * We must have the VersionInfo here. |
10367 | | */ |
10368 | 184k | version = xmlParseVersionInfo(ctxt); |
10369 | 184k | if (version == NULL) { |
10370 | 28.4k | xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL); |
10371 | 156k | } else { |
10372 | 156k | if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) { |
10373 | | /* |
10374 | | * Changed here for XML-1.0 5th edition |
10375 | | */ |
10376 | 12.6k | if (ctxt->options & XML_PARSE_OLD10) { |
10377 | 2.56k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10378 | 2.56k | "Unsupported version '%s'\n", |
10379 | 2.56k | version); |
10380 | 10.1k | } else { |
10381 | 10.1k | if ((version[0] == '1') && ((version[1] == '.'))) { |
10382 | 5.11k | xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION, |
10383 | 5.11k | "Unsupported version '%s'\n", |
10384 | 5.11k | version, NULL); |
10385 | 5.11k | } else { |
10386 | 5.01k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10387 | 5.01k | "Unsupported version '%s'\n", |
10388 | 5.01k | version); |
10389 | 5.01k | } |
10390 | 10.1k | } |
10391 | 12.6k | } |
10392 | 156k | if (ctxt->version != NULL) |
10393 | 0 | xmlFree((void *) ctxt->version); |
10394 | 156k | ctxt->version = version; |
10395 | 156k | } |
10396 | | |
10397 | | /* |
10398 | | * We may have the encoding declaration |
10399 | | */ |
10400 | 184k | if (!IS_BLANK_CH(RAW)) { |
10401 | 38.7k | if ((RAW == '?') && (NXT(1) == '>')) { |
10402 | 13.3k | SKIP(2); |
10403 | 13.3k | return; |
10404 | 13.3k | } |
10405 | 25.4k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10406 | 25.4k | } |
10407 | 171k | xmlParseEncodingDecl(ctxt); |
10408 | | |
10409 | | /* |
10410 | | * We may have the standalone status. |
10411 | | */ |
10412 | 171k | if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) { |
10413 | 18.7k | if ((RAW == '?') && (NXT(1) == '>')) { |
10414 | 14.8k | SKIP(2); |
10415 | 14.8k | return; |
10416 | 14.8k | } |
10417 | 3.86k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10418 | 3.86k | } |
10419 | | |
10420 | | /* |
10421 | | * We can grow the input buffer freely at that point |
10422 | | */ |
10423 | 156k | GROW; |
10424 | | |
10425 | 156k | SKIP_BLANKS; |
10426 | 156k | ctxt->standalone = xmlParseSDDecl(ctxt); |
10427 | | |
10428 | 156k | SKIP_BLANKS; |
10429 | 156k | if ((RAW == '?') && (NXT(1) == '>')) { |
10430 | 111k | SKIP(2); |
10431 | 111k | } else if (RAW == '>') { |
10432 | | /* Deprecated old WD ... */ |
10433 | 1.06k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10434 | 1.06k | NEXT; |
10435 | 44.2k | } else { |
10436 | 44.2k | int c; |
10437 | | |
10438 | 44.2k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10439 | 1.39M | while ((PARSER_STOPPED(ctxt) == 0) && |
10440 | 1.39M | ((c = CUR) != 0)) { |
10441 | 1.37M | NEXT; |
10442 | 1.37M | if (c == '>') |
10443 | 26.8k | break; |
10444 | 1.37M | } |
10445 | 44.2k | } |
10446 | 156k | } |
10447 | | |
10448 | | /** |
10449 | | * xmlCtxtGetVersion: |
10450 | | * ctxt: parser context |
10451 | | * |
10452 | | * Available since 2.14.0. |
10453 | | * |
10454 | | * Returns the version from the XML declaration. |
10455 | | */ |
10456 | | const xmlChar * |
10457 | 0 | xmlCtxtGetVersion(xmlParserCtxtPtr ctxt) { |
10458 | 0 | if (ctxt == NULL) |
10459 | 0 | return(NULL); |
10460 | | |
10461 | 0 | return(ctxt->version); |
10462 | 0 | } |
10463 | | |
10464 | | /** |
10465 | | * xmlCtxtGetStandalone: |
10466 | | * ctxt: parser context |
10467 | | * |
10468 | | * Available since 2.14.0. |
10469 | | * |
10470 | | * Returns the value from the standalone document declaration. |
10471 | | */ |
10472 | | int |
10473 | 0 | xmlCtxtGetStandalone(xmlParserCtxtPtr ctxt) { |
10474 | 0 | if (ctxt == NULL) |
10475 | 0 | return(0); |
10476 | | |
10477 | 0 | return(ctxt->standalone); |
10478 | 0 | } |
10479 | | |
10480 | | /** |
10481 | | * xmlParseMisc: |
10482 | | * @ctxt: an XML parser context |
10483 | | * |
10484 | | * DEPRECATED: Internal function, don't use. |
10485 | | * |
10486 | | * parse an XML Misc* optional field. |
10487 | | * |
10488 | | * [27] Misc ::= Comment | PI | S |
10489 | | */ |
10490 | | |
10491 | | void |
10492 | 828k | xmlParseMisc(xmlParserCtxtPtr ctxt) { |
10493 | 961k | while (PARSER_STOPPED(ctxt) == 0) { |
10494 | 918k | SKIP_BLANKS; |
10495 | 918k | GROW; |
10496 | 918k | if ((RAW == '<') && (NXT(1) == '?')) { |
10497 | 78.8k | xmlParsePI(ctxt); |
10498 | 839k | } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) { |
10499 | 54.1k | xmlParseComment(ctxt); |
10500 | 785k | } else { |
10501 | 785k | break; |
10502 | 785k | } |
10503 | 918k | } |
10504 | 828k | } |
10505 | | |
10506 | | static void |
10507 | 436k | xmlFinishDocument(xmlParserCtxtPtr ctxt) { |
10508 | 436k | xmlDocPtr doc; |
10509 | | |
10510 | | /* |
10511 | | * SAX: end of the document processing. |
10512 | | */ |
10513 | 436k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10514 | 435k | ctxt->sax->endDocument(ctxt->userData); |
10515 | | |
10516 | 436k | doc = ctxt->myDoc; |
10517 | 436k | if (doc != NULL) { |
10518 | 405k | if (ctxt->wellFormed) { |
10519 | 61.3k | doc->properties |= XML_DOC_WELLFORMED; |
10520 | 61.3k | if (ctxt->valid) |
10521 | 30.0k | doc->properties |= XML_DOC_DTDVALID; |
10522 | 61.3k | if (ctxt->nsWellFormed) |
10523 | 46.3k | doc->properties |= XML_DOC_NSVALID; |
10524 | 61.3k | } |
10525 | | |
10526 | 405k | if (ctxt->options & XML_PARSE_OLD10) |
10527 | 132k | doc->properties |= XML_DOC_OLD10; |
10528 | | |
10529 | | /* |
10530 | | * Remove locally kept entity definitions if the tree was not built |
10531 | | */ |
10532 | 405k | if (xmlStrEqual(doc->version, SAX_COMPAT_MODE)) { |
10533 | 3.88k | xmlFreeDoc(doc); |
10534 | 3.88k | ctxt->myDoc = NULL; |
10535 | 3.88k | } |
10536 | 405k | } |
10537 | 436k | } |
10538 | | |
10539 | | /** |
10540 | | * xmlParseDocument: |
10541 | | * @ctxt: an XML parser context |
10542 | | * |
10543 | | * Parse an XML document and invoke the SAX handlers. This is useful |
10544 | | * if you're only interested in custom SAX callbacks. If you want a |
10545 | | * document tree, use xmlCtxtParseDocument. |
10546 | | * |
10547 | | * Returns 0, -1 in case of error. |
10548 | | */ |
10549 | | |
10550 | | int |
10551 | 404k | xmlParseDocument(xmlParserCtxtPtr ctxt) { |
10552 | 404k | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10553 | 0 | return(-1); |
10554 | | |
10555 | 404k | GROW; |
10556 | | |
10557 | | /* |
10558 | | * SAX: detecting the level. |
10559 | | */ |
10560 | 404k | xmlCtxtInitializeLate(ctxt); |
10561 | | |
10562 | 404k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
10563 | 404k | ctxt->sax->setDocumentLocator(ctxt->userData, |
10564 | 404k | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
10565 | 404k | } |
10566 | | |
10567 | 404k | xmlDetectEncoding(ctxt); |
10568 | | |
10569 | 404k | if (CUR == 0) { |
10570 | 10.2k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10571 | 10.2k | return(-1); |
10572 | 10.2k | } |
10573 | | |
10574 | 394k | GROW; |
10575 | 394k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10576 | | |
10577 | | /* |
10578 | | * Note that we will switch encoding on the fly. |
10579 | | */ |
10580 | 172k | xmlParseXMLDecl(ctxt); |
10581 | 172k | SKIP_BLANKS; |
10582 | 221k | } else { |
10583 | 221k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10584 | 221k | if (ctxt->version == NULL) { |
10585 | 64 | xmlErrMemory(ctxt); |
10586 | 64 | return(-1); |
10587 | 64 | } |
10588 | 221k | } |
10589 | 394k | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10590 | 360k | ctxt->sax->startDocument(ctxt->userData); |
10591 | 394k | if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) && |
10592 | 394k | (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) { |
10593 | 0 | ctxt->myDoc->compression = ctxt->input->buf->compressed; |
10594 | 0 | } |
10595 | | |
10596 | | /* |
10597 | | * The Misc part of the Prolog |
10598 | | */ |
10599 | 394k | xmlParseMisc(ctxt); |
10600 | | |
10601 | | /* |
10602 | | * Then possibly doc type declaration(s) and more Misc |
10603 | | * (doctypedecl Misc*)? |
10604 | | */ |
10605 | 394k | GROW; |
10606 | 394k | if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) { |
10607 | | |
10608 | 220k | ctxt->inSubset = 1; |
10609 | 220k | xmlParseDocTypeDecl(ctxt); |
10610 | 220k | if (RAW == '[') { |
10611 | 170k | xmlParseInternalSubset(ctxt); |
10612 | 170k | } |
10613 | | |
10614 | | /* |
10615 | | * Create and update the external subset. |
10616 | | */ |
10617 | 220k | ctxt->inSubset = 2; |
10618 | 220k | if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) && |
10619 | 220k | (!ctxt->disableSAX)) |
10620 | 158k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
10621 | 158k | ctxt->extSubSystem, ctxt->extSubURI); |
10622 | 220k | ctxt->inSubset = 0; |
10623 | | |
10624 | 220k | xmlCleanSpecialAttr(ctxt); |
10625 | | |
10626 | 220k | xmlParseMisc(ctxt); |
10627 | 220k | } |
10628 | | |
10629 | | /* |
10630 | | * Time to start parsing the tree itself |
10631 | | */ |
10632 | 394k | GROW; |
10633 | 394k | if (RAW != '<') { |
10634 | 180k | if (ctxt->wellFormed) |
10635 | 35.1k | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
10636 | 35.1k | "Start tag expected, '<' not found\n"); |
10637 | 213k | } else { |
10638 | 213k | xmlParseElement(ctxt); |
10639 | | |
10640 | | /* |
10641 | | * The Misc part at the end |
10642 | | */ |
10643 | 213k | xmlParseMisc(ctxt); |
10644 | | |
10645 | 213k | if (ctxt->input->cur < ctxt->input->end) { |
10646 | 54.8k | if (ctxt->wellFormed) |
10647 | 1.90k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
10648 | 158k | } else if ((ctxt->input->buf != NULL) && |
10649 | 158k | (ctxt->input->buf->encoder != NULL) && |
10650 | 158k | (ctxt->input->buf->error == 0) && |
10651 | 158k | (!xmlBufIsEmpty(ctxt->input->buf->raw))) { |
10652 | 810 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
10653 | 810 | "Truncated multi-byte sequence at EOF\n"); |
10654 | 810 | } |
10655 | 213k | } |
10656 | | |
10657 | 394k | ctxt->instate = XML_PARSER_EOF; |
10658 | 394k | xmlFinishDocument(ctxt); |
10659 | | |
10660 | 394k | if (! ctxt->wellFormed) { |
10661 | 335k | ctxt->valid = 0; |
10662 | 335k | return(-1); |
10663 | 335k | } |
10664 | | |
10665 | 58.8k | return(0); |
10666 | 394k | } |
10667 | | |
10668 | | /** |
10669 | | * xmlParseExtParsedEnt: |
10670 | | * @ctxt: an XML parser context |
10671 | | * |
10672 | | * parse a general parsed entity |
10673 | | * An external general parsed entity is well-formed if it matches the |
10674 | | * production labeled extParsedEnt. |
10675 | | * |
10676 | | * [78] extParsedEnt ::= TextDecl? content |
10677 | | * |
10678 | | * Returns 0, -1 in case of error. the parser context is augmented |
10679 | | * as a result of the parsing. |
10680 | | */ |
10681 | | |
10682 | | int |
10683 | 0 | xmlParseExtParsedEnt(xmlParserCtxtPtr ctxt) { |
10684 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10685 | 0 | return(-1); |
10686 | | |
10687 | 0 | xmlCtxtInitializeLate(ctxt); |
10688 | |
|
10689 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
10690 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
10691 | 0 | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
10692 | 0 | } |
10693 | |
|
10694 | 0 | xmlDetectEncoding(ctxt); |
10695 | |
|
10696 | 0 | if (CUR == 0) { |
10697 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10698 | 0 | } |
10699 | | |
10700 | | /* |
10701 | | * Check for the XMLDecl in the Prolog. |
10702 | | */ |
10703 | 0 | GROW; |
10704 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10705 | | |
10706 | | /* |
10707 | | * Note that we will switch encoding on the fly. |
10708 | | */ |
10709 | 0 | xmlParseXMLDecl(ctxt); |
10710 | 0 | SKIP_BLANKS; |
10711 | 0 | } else { |
10712 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10713 | 0 | } |
10714 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10715 | 0 | ctxt->sax->startDocument(ctxt->userData); |
10716 | | |
10717 | | /* |
10718 | | * Doing validity checking on chunk doesn't make sense |
10719 | | */ |
10720 | 0 | ctxt->options &= ~XML_PARSE_DTDVALID; |
10721 | 0 | ctxt->validate = 0; |
10722 | 0 | ctxt->depth = 0; |
10723 | |
|
10724 | 0 | xmlParseContentInternal(ctxt); |
10725 | |
|
10726 | 0 | if (ctxt->input->cur < ctxt->input->end) |
10727 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
10728 | | |
10729 | | /* |
10730 | | * SAX: end of the document processing. |
10731 | | */ |
10732 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10733 | 0 | ctxt->sax->endDocument(ctxt->userData); |
10734 | |
|
10735 | 0 | if (! ctxt->wellFormed) return(-1); |
10736 | 0 | return(0); |
10737 | 0 | } |
10738 | | |
10739 | | #ifdef LIBXML_PUSH_ENABLED |
10740 | | /************************************************************************ |
10741 | | * * |
10742 | | * Progressive parsing interfaces * |
10743 | | * * |
10744 | | ************************************************************************/ |
10745 | | |
10746 | | /** |
10747 | | * xmlParseLookupChar: |
10748 | | * @ctxt: an XML parser context |
10749 | | * @c: character |
10750 | | * |
10751 | | * Check whether the input buffer contains a character. |
10752 | | */ |
10753 | | static int |
10754 | 416k | xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) { |
10755 | 416k | const xmlChar *cur; |
10756 | | |
10757 | 416k | if (ctxt->checkIndex == 0) { |
10758 | 307k | cur = ctxt->input->cur + 1; |
10759 | 307k | } else { |
10760 | 108k | cur = ctxt->input->cur + ctxt->checkIndex; |
10761 | 108k | } |
10762 | | |
10763 | 416k | if (memchr(cur, c, ctxt->input->end - cur) == NULL) { |
10764 | 111k | size_t index = ctxt->input->end - ctxt->input->cur; |
10765 | | |
10766 | 111k | if (index > LONG_MAX) { |
10767 | 0 | ctxt->checkIndex = 0; |
10768 | 0 | return(1); |
10769 | 0 | } |
10770 | 111k | ctxt->checkIndex = index; |
10771 | 111k | return(0); |
10772 | 305k | } else { |
10773 | 305k | ctxt->checkIndex = 0; |
10774 | 305k | return(1); |
10775 | 305k | } |
10776 | 416k | } |
10777 | | |
10778 | | /** |
10779 | | * xmlParseLookupString: |
10780 | | * @ctxt: an XML parser context |
10781 | | * @startDelta: delta to apply at the start |
10782 | | * @str: string |
10783 | | * @strLen: length of string |
10784 | | * |
10785 | | * Check whether the input buffer contains a string. |
10786 | | */ |
10787 | | static const xmlChar * |
10788 | | xmlParseLookupString(xmlParserCtxtPtr ctxt, size_t startDelta, |
10789 | 1.05M | const char *str, size_t strLen) { |
10790 | 1.05M | const xmlChar *cur, *term; |
10791 | | |
10792 | 1.05M | if (ctxt->checkIndex == 0) { |
10793 | 549k | cur = ctxt->input->cur + startDelta; |
10794 | 549k | } else { |
10795 | 506k | cur = ctxt->input->cur + ctxt->checkIndex; |
10796 | 506k | } |
10797 | | |
10798 | 1.05M | term = BAD_CAST strstr((const char *) cur, str); |
10799 | 1.05M | if (term == NULL) { |
10800 | 777k | const xmlChar *end = ctxt->input->end; |
10801 | 777k | size_t index; |
10802 | | |
10803 | | /* Rescan (strLen - 1) characters. */ |
10804 | 777k | if ((size_t) (end - cur) < strLen) |
10805 | 4.06k | end = cur; |
10806 | 773k | else |
10807 | 773k | end -= strLen - 1; |
10808 | 777k | index = end - ctxt->input->cur; |
10809 | 777k | if (index > LONG_MAX) { |
10810 | 0 | ctxt->checkIndex = 0; |
10811 | 0 | return(ctxt->input->end - strLen); |
10812 | 0 | } |
10813 | 777k | ctxt->checkIndex = index; |
10814 | 777k | } else { |
10815 | 278k | ctxt->checkIndex = 0; |
10816 | 278k | } |
10817 | | |
10818 | 1.05M | return(term); |
10819 | 1.05M | } |
10820 | | |
10821 | | /** |
10822 | | * xmlParseLookupCharData: |
10823 | | * @ctxt: an XML parser context |
10824 | | * |
10825 | | * Check whether the input buffer contains terminated char data. |
10826 | | */ |
10827 | | static int |
10828 | 424k | xmlParseLookupCharData(xmlParserCtxtPtr ctxt) { |
10829 | 424k | const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex; |
10830 | 424k | const xmlChar *end = ctxt->input->end; |
10831 | 424k | size_t index; |
10832 | | |
10833 | 10.0M | while (cur < end) { |
10834 | 10.0M | if ((*cur == '<') || (*cur == '&')) { |
10835 | 395k | ctxt->checkIndex = 0; |
10836 | 395k | return(1); |
10837 | 395k | } |
10838 | 9.65M | cur++; |
10839 | 9.65M | } |
10840 | | |
10841 | 28.2k | index = cur - ctxt->input->cur; |
10842 | 28.2k | if (index > LONG_MAX) { |
10843 | 0 | ctxt->checkIndex = 0; |
10844 | 0 | return(1); |
10845 | 0 | } |
10846 | 28.2k | ctxt->checkIndex = index; |
10847 | 28.2k | return(0); |
10848 | 28.2k | } |
10849 | | |
10850 | | /** |
10851 | | * xmlParseLookupGt: |
10852 | | * @ctxt: an XML parser context |
10853 | | * |
10854 | | * Check whether there's enough data in the input buffer to finish parsing |
10855 | | * a start tag. This has to take quotes into account. |
10856 | | */ |
10857 | | static int |
10858 | 2.13M | xmlParseLookupGt(xmlParserCtxtPtr ctxt) { |
10859 | 2.13M | const xmlChar *cur; |
10860 | 2.13M | const xmlChar *end = ctxt->input->end; |
10861 | 2.13M | int state = ctxt->endCheckState; |
10862 | 2.13M | size_t index; |
10863 | | |
10864 | 2.13M | if (ctxt->checkIndex == 0) |
10865 | 741k | cur = ctxt->input->cur + 1; |
10866 | 1.39M | else |
10867 | 1.39M | cur = ctxt->input->cur + ctxt->checkIndex; |
10868 | | |
10869 | 391M | while (cur < end) { |
10870 | 390M | if (state) { |
10871 | 334M | if (*cur == state) |
10872 | 982k | state = 0; |
10873 | 334M | } else if (*cur == '\'' || *cur == '"') { |
10874 | 989k | state = *cur; |
10875 | 54.7M | } else if (*cur == '>') { |
10876 | 724k | ctxt->checkIndex = 0; |
10877 | 724k | ctxt->endCheckState = 0; |
10878 | 724k | return(1); |
10879 | 724k | } |
10880 | 389M | cur++; |
10881 | 389M | } |
10882 | | |
10883 | 1.41M | index = cur - ctxt->input->cur; |
10884 | 1.41M | if (index > LONG_MAX) { |
10885 | 0 | ctxt->checkIndex = 0; |
10886 | 0 | ctxt->endCheckState = 0; |
10887 | 0 | return(1); |
10888 | 0 | } |
10889 | 1.41M | ctxt->checkIndex = index; |
10890 | 1.41M | ctxt->endCheckState = state; |
10891 | 1.41M | return(0); |
10892 | 1.41M | } |
10893 | | |
10894 | | /** |
10895 | | * xmlParseLookupInternalSubset: |
10896 | | * @ctxt: an XML parser context |
10897 | | * |
10898 | | * Check whether there's enough data in the input buffer to finish parsing |
10899 | | * the internal subset. |
10900 | | */ |
10901 | | static int |
10902 | 1.00M | xmlParseLookupInternalSubset(xmlParserCtxtPtr ctxt) { |
10903 | | /* |
10904 | | * Sorry, but progressive parsing of the internal subset is not |
10905 | | * supported. We first check that the full content of the internal |
10906 | | * subset is available and parsing is launched only at that point. |
10907 | | * Internal subset ends with "']' S? '>'" in an unescaped section and |
10908 | | * not in a ']]>' sequence which are conditional sections. |
10909 | | */ |
10910 | 1.00M | const xmlChar *cur, *start; |
10911 | 1.00M | const xmlChar *end = ctxt->input->end; |
10912 | 1.00M | int state = ctxt->endCheckState; |
10913 | 1.00M | size_t index; |
10914 | | |
10915 | 1.00M | if (ctxt->checkIndex == 0) { |
10916 | 28.5k | cur = ctxt->input->cur + 1; |
10917 | 973k | } else { |
10918 | 973k | cur = ctxt->input->cur + ctxt->checkIndex; |
10919 | 973k | } |
10920 | 1.00M | start = cur; |
10921 | | |
10922 | 411M | while (cur < end) { |
10923 | 410M | if (state == '-') { |
10924 | 746k | if ((*cur == '-') && |
10925 | 746k | (cur[1] == '-') && |
10926 | 746k | (cur[2] == '>')) { |
10927 | 23.9k | state = 0; |
10928 | 23.9k | cur += 3; |
10929 | 23.9k | start = cur; |
10930 | 23.9k | continue; |
10931 | 23.9k | } |
10932 | 746k | } |
10933 | 410M | else if (state == ']') { |
10934 | 36.7k | if (*cur == '>') { |
10935 | 21.4k | ctxt->checkIndex = 0; |
10936 | 21.4k | ctxt->endCheckState = 0; |
10937 | 21.4k | return(1); |
10938 | 21.4k | } |
10939 | 15.2k | if (IS_BLANK_CH(*cur)) { |
10940 | 5.07k | state = ' '; |
10941 | 10.2k | } else if (*cur != ']') { |
10942 | 3.32k | state = 0; |
10943 | 3.32k | start = cur; |
10944 | 3.32k | continue; |
10945 | 3.32k | } |
10946 | 15.2k | } |
10947 | 410M | else if (state == ' ') { |
10948 | 345k | if (*cur == '>') { |
10949 | 153 | ctxt->checkIndex = 0; |
10950 | 153 | ctxt->endCheckState = 0; |
10951 | 153 | return(1); |
10952 | 153 | } |
10953 | 345k | if (!IS_BLANK_CH(*cur)) { |
10954 | 4.91k | state = 0; |
10955 | 4.91k | start = cur; |
10956 | 4.91k | continue; |
10957 | 4.91k | } |
10958 | 345k | } |
10959 | 409M | else if (state != 0) { |
10960 | 383M | if (*cur == state) { |
10961 | 100k | state = 0; |
10962 | 100k | start = cur + 1; |
10963 | 100k | } |
10964 | 383M | } |
10965 | 26.3M | else if (*cur == '<') { |
10966 | 151k | if ((cur[1] == '!') && |
10967 | 151k | (cur[2] == '-') && |
10968 | 151k | (cur[3] == '-')) { |
10969 | 24.1k | state = '-'; |
10970 | 24.1k | cur += 4; |
10971 | | /* Don't treat <!--> as comment */ |
10972 | 24.1k | start = cur; |
10973 | 24.1k | continue; |
10974 | 24.1k | } |
10975 | 151k | } |
10976 | 26.1M | else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) { |
10977 | 132k | state = *cur; |
10978 | 132k | } |
10979 | | |
10980 | 410M | cur++; |
10981 | 410M | } |
10982 | | |
10983 | | /* |
10984 | | * Rescan the three last characters to detect "<!--" and "-->" |
10985 | | * split across chunks. |
10986 | | */ |
10987 | 980k | if ((state == 0) || (state == '-')) { |
10988 | 63.3k | if (cur - start < 3) |
10989 | 2.39k | cur = start; |
10990 | 60.9k | else |
10991 | 60.9k | cur -= 3; |
10992 | 63.3k | } |
10993 | 980k | index = cur - ctxt->input->cur; |
10994 | 980k | if (index > LONG_MAX) { |
10995 | 0 | ctxt->checkIndex = 0; |
10996 | 0 | ctxt->endCheckState = 0; |
10997 | 0 | return(1); |
10998 | 0 | } |
10999 | 980k | ctxt->checkIndex = index; |
11000 | 980k | ctxt->endCheckState = state; |
11001 | 980k | return(0); |
11002 | 980k | } |
11003 | | |
11004 | | /** |
11005 | | * xmlCheckCdataPush: |
11006 | | * @cur: pointer to the block of characters |
11007 | | * @len: length of the block in bytes |
11008 | | * @complete: 1 if complete CDATA block is passed in, 0 if partial block |
11009 | | * |
11010 | | * Check that the block of characters is okay as SCdata content [20] |
11011 | | * |
11012 | | * Returns the number of bytes to pass if okay, a negative index where an |
11013 | | * UTF-8 error occurred otherwise |
11014 | | */ |
11015 | | static int |
11016 | 403k | xmlCheckCdataPush(const xmlChar *utf, int len, int complete) { |
11017 | 403k | int ix; |
11018 | 403k | unsigned char c; |
11019 | 403k | int codepoint; |
11020 | | |
11021 | 403k | if ((utf == NULL) || (len <= 0)) |
11022 | 5.27k | return(0); |
11023 | | |
11024 | 19.7M | for (ix = 0; ix < len;) { /* string is 0-terminated */ |
11025 | 19.5M | c = utf[ix]; |
11026 | 19.5M | if ((c & 0x80) == 0x00) { /* 1-byte code, starts with 10 */ |
11027 | 2.99M | if (c >= 0x20) |
11028 | 2.54M | ix++; |
11029 | 443k | else if ((c == 0xA) || (c == 0xD) || (c == 0x9)) |
11030 | 377k | ix++; |
11031 | 65.3k | else |
11032 | 65.3k | return(-ix); |
11033 | 16.5M | } else if ((c & 0xe0) == 0xc0) {/* 2-byte code, starts with 110 */ |
11034 | 336k | if (ix + 2 > len) return(complete ? -ix : ix); |
11035 | 335k | if ((utf[ix+1] & 0xc0 ) != 0x80) |
11036 | 2.29k | return(-ix); |
11037 | 332k | codepoint = (utf[ix] & 0x1f) << 6; |
11038 | 332k | codepoint |= utf[ix+1] & 0x3f; |
11039 | 332k | if (!xmlIsCharQ(codepoint)) |
11040 | 1.27k | return(-ix); |
11041 | 331k | ix += 2; |
11042 | 16.2M | } else if ((c & 0xf0) == 0xe0) {/* 3-byte code, starts with 1110 */ |
11043 | 16.2M | if (ix + 3 > len) return(complete ? -ix : ix); |
11044 | 16.2M | if (((utf[ix+1] & 0xc0) != 0x80) || |
11045 | 16.2M | ((utf[ix+2] & 0xc0) != 0x80)) |
11046 | 15.5k | return(-ix); |
11047 | 16.1M | codepoint = (utf[ix] & 0xf) << 12; |
11048 | 16.1M | codepoint |= (utf[ix+1] & 0x3f) << 6; |
11049 | 16.1M | codepoint |= utf[ix+2] & 0x3f; |
11050 | 16.1M | if (!xmlIsCharQ(codepoint)) |
11051 | 72.0k | return(-ix); |
11052 | 16.1M | ix += 3; |
11053 | 16.1M | } else if ((c & 0xf8) == 0xf0) {/* 4-byte code, starts with 11110 */ |
11054 | 17.7k | if (ix + 4 > len) return(complete ? -ix : ix); |
11055 | 16.8k | if (((utf[ix+1] & 0xc0) != 0x80) || |
11056 | 16.8k | ((utf[ix+2] & 0xc0) != 0x80) || |
11057 | 16.8k | ((utf[ix+3] & 0xc0) != 0x80)) |
11058 | 1.66k | return(-ix); |
11059 | 15.1k | codepoint = (utf[ix] & 0x7) << 18; |
11060 | 15.1k | codepoint |= (utf[ix+1] & 0x3f) << 12; |
11061 | 15.1k | codepoint |= (utf[ix+2] & 0x3f) << 6; |
11062 | 15.1k | codepoint |= utf[ix+3] & 0x3f; |
11063 | 15.1k | if (!xmlIsCharQ(codepoint)) |
11064 | 1.51k | return(-ix); |
11065 | 13.6k | ix += 4; |
11066 | 13.6k | } else /* unknown encoding */ |
11067 | 2.63k | return(-ix); |
11068 | 19.5M | } |
11069 | 213k | return(ix); |
11070 | 398k | } |
11071 | | |
11072 | | /** |
11073 | | * xmlParseTryOrFinish: |
11074 | | * @ctxt: an XML parser context |
11075 | | * @terminate: last chunk indicator |
11076 | | * |
11077 | | * Try to progress on parsing |
11078 | | * |
11079 | | * Returns zero if no parsing was possible |
11080 | | */ |
11081 | | static int |
11082 | 4.22M | xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) { |
11083 | 4.22M | int ret = 0; |
11084 | 4.22M | size_t avail; |
11085 | 4.22M | xmlChar cur, next; |
11086 | | |
11087 | 4.22M | if (ctxt->input == NULL) |
11088 | 0 | return(0); |
11089 | | |
11090 | 4.22M | if ((ctxt->input != NULL) && |
11091 | 4.22M | (ctxt->input->cur - ctxt->input->base > 4096)) { |
11092 | 11.6k | xmlParserShrink(ctxt); |
11093 | 11.6k | } |
11094 | | |
11095 | 10.0M | while (ctxt->disableSAX == 0) { |
11096 | 10.0M | avail = ctxt->input->end - ctxt->input->cur; |
11097 | 10.0M | if (avail < 1) |
11098 | 44.8k | goto done; |
11099 | 9.95M | switch (ctxt->instate) { |
11100 | 896k | case XML_PARSER_EOF: |
11101 | | /* |
11102 | | * Document parsing is done ! |
11103 | | */ |
11104 | 896k | goto done; |
11105 | 64.8k | case XML_PARSER_START: |
11106 | | /* |
11107 | | * Very first chars read from the document flow. |
11108 | | */ |
11109 | 64.8k | if ((!terminate) && (avail < 4)) |
11110 | 733 | goto done; |
11111 | | |
11112 | | /* |
11113 | | * We need more bytes to detect EBCDIC code pages. |
11114 | | * See xmlDetectEBCDIC. |
11115 | | */ |
11116 | 64.0k | if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) && |
11117 | 64.0k | (!terminate) && (avail < 200)) |
11118 | 718 | goto done; |
11119 | | |
11120 | 63.3k | xmlDetectEncoding(ctxt); |
11121 | 63.3k | ctxt->instate = XML_PARSER_XML_DECL; |
11122 | 63.3k | break; |
11123 | | |
11124 | 297k | case XML_PARSER_XML_DECL: |
11125 | 297k | if ((!terminate) && (avail < 2)) |
11126 | 31 | goto done; |
11127 | 297k | cur = ctxt->input->cur[0]; |
11128 | 297k | next = ctxt->input->cur[1]; |
11129 | 297k | if ((cur == '<') && (next == '?')) { |
11130 | | /* PI or XML decl */ |
11131 | 251k | if ((!terminate) && |
11132 | 251k | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11133 | 234k | goto done; |
11134 | 17.4k | if ((ctxt->input->cur[2] == 'x') && |
11135 | 17.4k | (ctxt->input->cur[3] == 'm') && |
11136 | 17.4k | (ctxt->input->cur[4] == 'l') && |
11137 | 17.4k | (IS_BLANK_CH(ctxt->input->cur[5]))) { |
11138 | 12.1k | ret += 5; |
11139 | 12.1k | xmlParseXMLDecl(ctxt); |
11140 | 12.1k | } else { |
11141 | 5.30k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11142 | 5.30k | if (ctxt->version == NULL) { |
11143 | 11 | xmlErrMemory(ctxt); |
11144 | 11 | break; |
11145 | 11 | } |
11146 | 5.30k | } |
11147 | 45.7k | } else { |
11148 | 45.7k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11149 | 45.7k | if (ctxt->version == NULL) { |
11150 | 44 | xmlErrMemory(ctxt); |
11151 | 44 | break; |
11152 | 44 | } |
11153 | 45.7k | } |
11154 | 63.1k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
11155 | 63.1k | ctxt->sax->setDocumentLocator(ctxt->userData, |
11156 | 63.1k | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
11157 | 63.1k | } |
11158 | 63.1k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11159 | 63.1k | (!ctxt->disableSAX)) |
11160 | 61.8k | ctxt->sax->startDocument(ctxt->userData); |
11161 | 63.1k | ctxt->instate = XML_PARSER_MISC; |
11162 | 63.1k | break; |
11163 | 1.87M | case XML_PARSER_START_TAG: { |
11164 | 1.87M | const xmlChar *name; |
11165 | 1.87M | const xmlChar *prefix = NULL; |
11166 | 1.87M | const xmlChar *URI = NULL; |
11167 | 1.87M | int line = ctxt->input->line; |
11168 | 1.87M | int nbNs = 0; |
11169 | | |
11170 | 1.87M | if ((!terminate) && (avail < 2)) |
11171 | 267 | goto done; |
11172 | 1.87M | cur = ctxt->input->cur[0]; |
11173 | 1.87M | if (cur != '<') { |
11174 | 6.29k | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
11175 | 6.29k | "Start tag expected, '<' not found"); |
11176 | 6.29k | ctxt->instate = XML_PARSER_EOF; |
11177 | 6.29k | xmlFinishDocument(ctxt); |
11178 | 6.29k | goto done; |
11179 | 6.29k | } |
11180 | 1.86M | if ((!terminate) && (!xmlParseLookupGt(ctxt))) |
11181 | 957k | goto done; |
11182 | 908k | if (ctxt->spaceNr == 0) |
11183 | 0 | spacePush(ctxt, -1); |
11184 | 908k | else if (*ctxt->space == -2) |
11185 | 150k | spacePush(ctxt, -1); |
11186 | 757k | else |
11187 | 757k | spacePush(ctxt, *ctxt->space); |
11188 | 908k | #ifdef LIBXML_SAX1_ENABLED |
11189 | 908k | if (ctxt->sax2) |
11190 | 646k | #endif /* LIBXML_SAX1_ENABLED */ |
11191 | 646k | name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs); |
11192 | 261k | #ifdef LIBXML_SAX1_ENABLED |
11193 | 261k | else |
11194 | 261k | name = xmlParseStartTag(ctxt); |
11195 | 908k | #endif /* LIBXML_SAX1_ENABLED */ |
11196 | 908k | if (name == NULL) { |
11197 | 10.6k | spacePop(ctxt); |
11198 | 10.6k | ctxt->instate = XML_PARSER_EOF; |
11199 | 10.6k | xmlFinishDocument(ctxt); |
11200 | 10.6k | goto done; |
11201 | 10.6k | } |
11202 | 897k | #ifdef LIBXML_VALID_ENABLED |
11203 | | /* |
11204 | | * [ VC: Root Element Type ] |
11205 | | * The Name in the document type declaration must match |
11206 | | * the element type of the root element. |
11207 | | */ |
11208 | 897k | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
11209 | 897k | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
11210 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
11211 | 897k | #endif /* LIBXML_VALID_ENABLED */ |
11212 | | |
11213 | | /* |
11214 | | * Check for an Empty Element. |
11215 | | */ |
11216 | 897k | if ((RAW == '/') && (NXT(1) == '>')) { |
11217 | 142k | SKIP(2); |
11218 | | |
11219 | 142k | if (ctxt->sax2) { |
11220 | 88.1k | if ((ctxt->sax != NULL) && |
11221 | 88.1k | (ctxt->sax->endElementNs != NULL) && |
11222 | 88.1k | (!ctxt->disableSAX)) |
11223 | 87.9k | ctxt->sax->endElementNs(ctxt->userData, name, |
11224 | 87.9k | prefix, URI); |
11225 | 88.1k | if (nbNs > 0) |
11226 | 11.0k | xmlParserNsPop(ctxt, nbNs); |
11227 | 88.1k | #ifdef LIBXML_SAX1_ENABLED |
11228 | 88.1k | } else { |
11229 | 54.6k | if ((ctxt->sax != NULL) && |
11230 | 54.6k | (ctxt->sax->endElement != NULL) && |
11231 | 54.6k | (!ctxt->disableSAX)) |
11232 | 54.5k | ctxt->sax->endElement(ctxt->userData, name); |
11233 | 54.6k | #endif /* LIBXML_SAX1_ENABLED */ |
11234 | 54.6k | } |
11235 | 142k | spacePop(ctxt); |
11236 | 754k | } else if (RAW == '>') { |
11237 | 550k | NEXT; |
11238 | 550k | nameNsPush(ctxt, name, prefix, URI, line, nbNs); |
11239 | 550k | } else { |
11240 | 204k | xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED, |
11241 | 204k | "Couldn't find end of Start Tag %s\n", |
11242 | 204k | name); |
11243 | 204k | nodePop(ctxt); |
11244 | 204k | spacePop(ctxt); |
11245 | 204k | if (nbNs > 0) |
11246 | 24.0k | xmlParserNsPop(ctxt, nbNs); |
11247 | 204k | } |
11248 | | |
11249 | 897k | if (ctxt->nameNr == 0) |
11250 | 7.53k | ctxt->instate = XML_PARSER_EPILOG; |
11251 | 890k | else |
11252 | 890k | ctxt->instate = XML_PARSER_CONTENT; |
11253 | 897k | break; |
11254 | 908k | } |
11255 | 4.41M | case XML_PARSER_CONTENT: { |
11256 | 4.41M | cur = ctxt->input->cur[0]; |
11257 | | |
11258 | 4.41M | if (cur == '<') { |
11259 | 1.47M | if ((!terminate) && (avail < 2)) |
11260 | 5.48k | goto done; |
11261 | 1.47M | next = ctxt->input->cur[1]; |
11262 | | |
11263 | 1.47M | if (next == '/') { |
11264 | 239k | ctxt->instate = XML_PARSER_END_TAG; |
11265 | 239k | break; |
11266 | 1.23M | } else if (next == '?') { |
11267 | 41.7k | if ((!terminate) && |
11268 | 41.7k | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11269 | 23.0k | goto done; |
11270 | 18.6k | xmlParsePI(ctxt); |
11271 | 18.6k | ctxt->instate = XML_PARSER_CONTENT; |
11272 | 18.6k | break; |
11273 | 1.19M | } else if (next == '!') { |
11274 | 328k | if ((!terminate) && (avail < 3)) |
11275 | 1.51k | goto done; |
11276 | 326k | next = ctxt->input->cur[2]; |
11277 | | |
11278 | 326k | if (next == '-') { |
11279 | 201k | if ((!terminate) && (avail < 4)) |
11280 | 563 | goto done; |
11281 | 201k | if (ctxt->input->cur[3] == '-') { |
11282 | 201k | if ((!terminate) && |
11283 | 201k | (!xmlParseLookupString(ctxt, 4, "-->", 3))) |
11284 | 106k | goto done; |
11285 | 94.4k | xmlParseComment(ctxt); |
11286 | 94.4k | ctxt->instate = XML_PARSER_CONTENT; |
11287 | 94.4k | break; |
11288 | 201k | } |
11289 | 201k | } else if (next == '[') { |
11290 | 121k | if ((!terminate) && (avail < 9)) |
11291 | 4.86k | goto done; |
11292 | 116k | if ((ctxt->input->cur[2] == '[') && |
11293 | 116k | (ctxt->input->cur[3] == 'C') && |
11294 | 116k | (ctxt->input->cur[4] == 'D') && |
11295 | 116k | (ctxt->input->cur[5] == 'A') && |
11296 | 116k | (ctxt->input->cur[6] == 'T') && |
11297 | 116k | (ctxt->input->cur[7] == 'A') && |
11298 | 116k | (ctxt->input->cur[8] == '[')) { |
11299 | 115k | SKIP(9); |
11300 | 115k | ctxt->instate = XML_PARSER_CDATA_SECTION; |
11301 | 115k | break; |
11302 | 115k | } |
11303 | 116k | } |
11304 | 326k | } |
11305 | 2.94M | } else if (cur == '&') { |
11306 | 241k | if ((!terminate) && (!xmlParseLookupChar(ctxt, ';'))) |
11307 | 97.8k | goto done; |
11308 | 143k | xmlParseReference(ctxt); |
11309 | 143k | break; |
11310 | 2.69M | } else { |
11311 | | /* TODO Avoid the extra copy, handle directly !!! */ |
11312 | | /* |
11313 | | * Goal of the following test is: |
11314 | | * - minimize calls to the SAX 'character' callback |
11315 | | * when they are mergeable |
11316 | | * - handle an problem for isBlank when we only parse |
11317 | | * a sequence of blank chars and the next one is |
11318 | | * not available to check against '<' presence. |
11319 | | * - tries to homogenize the differences in SAX |
11320 | | * callbacks between the push and pull versions |
11321 | | * of the parser. |
11322 | | */ |
11323 | 2.69M | if (avail < XML_PARSER_BIG_BUFFER_SIZE) { |
11324 | 486k | if ((!terminate) && (!xmlParseLookupCharData(ctxt))) |
11325 | 28.2k | goto done; |
11326 | 486k | } |
11327 | 2.67M | ctxt->checkIndex = 0; |
11328 | 2.67M | xmlParseCharDataInternal(ctxt, !terminate); |
11329 | 2.67M | break; |
11330 | 2.69M | } |
11331 | | |
11332 | 868k | ctxt->instate = XML_PARSER_START_TAG; |
11333 | 868k | break; |
11334 | 4.41M | } |
11335 | 252k | case XML_PARSER_END_TAG: |
11336 | 252k | if ((!terminate) && (!xmlParseLookupChar(ctxt, '>'))) |
11337 | 13.1k | goto done; |
11338 | 239k | if (ctxt->sax2) { |
11339 | 194k | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
11340 | 194k | nameNsPop(ctxt); |
11341 | 194k | } |
11342 | 45.5k | #ifdef LIBXML_SAX1_ENABLED |
11343 | 45.5k | else |
11344 | 45.5k | xmlParseEndTag1(ctxt, 0); |
11345 | 239k | #endif /* LIBXML_SAX1_ENABLED */ |
11346 | 239k | if (ctxt->nameNr == 0) { |
11347 | 1.55k | ctxt->instate = XML_PARSER_EPILOG; |
11348 | 238k | } else { |
11349 | 238k | ctxt->instate = XML_PARSER_CONTENT; |
11350 | 238k | } |
11351 | 239k | break; |
11352 | 474k | case XML_PARSER_CDATA_SECTION: { |
11353 | | /* |
11354 | | * The Push mode need to have the SAX callback for |
11355 | | * cdataBlock merge back contiguous callbacks. |
11356 | | */ |
11357 | 474k | const xmlChar *term; |
11358 | | |
11359 | 474k | if (terminate) { |
11360 | | /* |
11361 | | * Don't call xmlParseLookupString. If 'terminate' |
11362 | | * is set, checkIndex is invalid. |
11363 | | */ |
11364 | 3.55k | term = BAD_CAST strstr((const char *) ctxt->input->cur, |
11365 | 3.55k | "]]>"); |
11366 | 471k | } else { |
11367 | 471k | term = xmlParseLookupString(ctxt, 0, "]]>", 3); |
11368 | 471k | } |
11369 | | |
11370 | 474k | if (term == NULL) { |
11371 | 335k | int tmp, size; |
11372 | | |
11373 | 335k | if (terminate) { |
11374 | | /* Unfinished CDATA section */ |
11375 | 1.27k | size = ctxt->input->end - ctxt->input->cur; |
11376 | 333k | } else { |
11377 | 333k | if (avail < XML_PARSER_BIG_BUFFER_SIZE + 2) |
11378 | 71.5k | goto done; |
11379 | 262k | ctxt->checkIndex = 0; |
11380 | | /* XXX: Why don't we pass the full buffer? */ |
11381 | 262k | size = XML_PARSER_BIG_BUFFER_SIZE; |
11382 | 262k | } |
11383 | 263k | tmp = xmlCheckCdataPush(ctxt->input->cur, size, 0); |
11384 | 263k | if (tmp <= 0) { |
11385 | 152k | tmp = -tmp; |
11386 | 152k | ctxt->input->cur += tmp; |
11387 | 152k | goto encoding_error; |
11388 | 152k | } |
11389 | 111k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
11390 | 111k | if (ctxt->sax->cdataBlock != NULL) |
11391 | 5.53k | ctxt->sax->cdataBlock(ctxt->userData, |
11392 | 5.53k | ctxt->input->cur, tmp); |
11393 | 105k | else if (ctxt->sax->characters != NULL) |
11394 | 105k | ctxt->sax->characters(ctxt->userData, |
11395 | 105k | ctxt->input->cur, tmp); |
11396 | 111k | } |
11397 | 111k | SKIPL(tmp); |
11398 | 139k | } else { |
11399 | 139k | int base = term - CUR_PTR; |
11400 | 139k | int tmp; |
11401 | | |
11402 | 139k | tmp = xmlCheckCdataPush(ctxt->input->cur, base, 1); |
11403 | 139k | if ((tmp < 0) || (tmp != base)) { |
11404 | 25.4k | tmp = -tmp; |
11405 | 25.4k | ctxt->input->cur += tmp; |
11406 | 25.4k | goto encoding_error; |
11407 | 25.4k | } |
11408 | 114k | if ((ctxt->sax != NULL) && (base == 0) && |
11409 | 114k | (ctxt->sax->cdataBlock != NULL) && |
11410 | 114k | (!ctxt->disableSAX)) { |
11411 | | /* |
11412 | | * Special case to provide identical behaviour |
11413 | | * between pull and push parsers on enpty CDATA |
11414 | | * sections |
11415 | | */ |
11416 | 2.81k | if ((ctxt->input->cur - ctxt->input->base >= 9) && |
11417 | 2.81k | (!strncmp((const char *)&ctxt->input->cur[-9], |
11418 | 2.81k | "<![CDATA[", 9))) |
11419 | 2.81k | ctxt->sax->cdataBlock(ctxt->userData, |
11420 | 2.81k | BAD_CAST "", 0); |
11421 | 111k | } else if ((ctxt->sax != NULL) && (base > 0) && |
11422 | 111k | (!ctxt->disableSAX)) { |
11423 | 109k | if (ctxt->sax->cdataBlock != NULL) |
11424 | 106k | ctxt->sax->cdataBlock(ctxt->userData, |
11425 | 106k | ctxt->input->cur, base); |
11426 | 2.37k | else if (ctxt->sax->characters != NULL) |
11427 | 2.37k | ctxt->sax->characters(ctxt->userData, |
11428 | 2.37k | ctxt->input->cur, base); |
11429 | 109k | } |
11430 | 114k | SKIPL(base + 3); |
11431 | 114k | ctxt->instate = XML_PARSER_CONTENT; |
11432 | 114k | } |
11433 | 225k | break; |
11434 | 474k | } |
11435 | 572k | case XML_PARSER_MISC: |
11436 | 605k | case XML_PARSER_PROLOG: |
11437 | 663k | case XML_PARSER_EPILOG: |
11438 | 663k | SKIP_BLANKS; |
11439 | 663k | avail = ctxt->input->end - ctxt->input->cur; |
11440 | 663k | if (avail < 1) |
11441 | 1.86k | goto done; |
11442 | 661k | if (ctxt->input->cur[0] == '<') { |
11443 | 653k | if ((!terminate) && (avail < 2)) |
11444 | 305 | goto done; |
11445 | 652k | next = ctxt->input->cur[1]; |
11446 | 652k | if (next == '?') { |
11447 | 76.1k | if ((!terminate) && |
11448 | 76.1k | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11449 | 60.7k | goto done; |
11450 | 15.4k | xmlParsePI(ctxt); |
11451 | 15.4k | break; |
11452 | 576k | } else if (next == '!') { |
11453 | 535k | if ((!terminate) && (avail < 3)) |
11454 | 206 | goto done; |
11455 | | |
11456 | 535k | if (ctxt->input->cur[2] == '-') { |
11457 | 43.0k | if ((!terminate) && (avail < 4)) |
11458 | 265 | goto done; |
11459 | 42.7k | if (ctxt->input->cur[3] == '-') { |
11460 | 42.7k | if ((!terminate) && |
11461 | 42.7k | (!xmlParseLookupString(ctxt, 4, "-->", 3))) |
11462 | 18.6k | goto done; |
11463 | 24.0k | xmlParseComment(ctxt); |
11464 | 24.0k | break; |
11465 | 42.7k | } |
11466 | 491k | } else if (ctxt->instate == XML_PARSER_MISC) { |
11467 | 491k | if ((!terminate) && (avail < 9)) |
11468 | 65 | goto done; |
11469 | 491k | if ((ctxt->input->cur[2] == 'D') && |
11470 | 491k | (ctxt->input->cur[3] == 'O') && |
11471 | 491k | (ctxt->input->cur[4] == 'C') && |
11472 | 491k | (ctxt->input->cur[5] == 'T') && |
11473 | 491k | (ctxt->input->cur[6] == 'Y') && |
11474 | 491k | (ctxt->input->cur[7] == 'P') && |
11475 | 491k | (ctxt->input->cur[8] == 'E')) { |
11476 | 491k | if ((!terminate) && (!xmlParseLookupGt(ctxt))) |
11477 | 453k | goto done; |
11478 | 38.5k | ctxt->inSubset = 1; |
11479 | 38.5k | xmlParseDocTypeDecl(ctxt); |
11480 | 38.5k | if (RAW == '[') { |
11481 | 32.0k | ctxt->instate = XML_PARSER_DTD; |
11482 | 32.0k | } else { |
11483 | | /* |
11484 | | * Create and update the external subset. |
11485 | | */ |
11486 | 6.53k | ctxt->inSubset = 2; |
11487 | 6.53k | if ((ctxt->sax != NULL) && |
11488 | 6.53k | (!ctxt->disableSAX) && |
11489 | 6.53k | (ctxt->sax->externalSubset != NULL)) |
11490 | 6.27k | ctxt->sax->externalSubset( |
11491 | 6.27k | ctxt->userData, |
11492 | 6.27k | ctxt->intSubName, |
11493 | 6.27k | ctxt->extSubSystem, |
11494 | 6.27k | ctxt->extSubURI); |
11495 | 6.53k | ctxt->inSubset = 0; |
11496 | 6.53k | xmlCleanSpecialAttr(ctxt); |
11497 | 6.53k | ctxt->instate = XML_PARSER_PROLOG; |
11498 | 6.53k | } |
11499 | 38.5k | break; |
11500 | 491k | } |
11501 | 491k | } |
11502 | 535k | } |
11503 | 652k | } |
11504 | | |
11505 | 50.3k | if (ctxt->instate == XML_PARSER_EPILOG) { |
11506 | 2.77k | if (ctxt->errNo == XML_ERR_OK) |
11507 | 29 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
11508 | 2.77k | ctxt->instate = XML_PARSER_EOF; |
11509 | 2.77k | xmlFinishDocument(ctxt); |
11510 | 47.5k | } else { |
11511 | 47.5k | ctxt->instate = XML_PARSER_START_TAG; |
11512 | 47.5k | } |
11513 | 50.3k | break; |
11514 | 1.01M | case XML_PARSER_DTD: { |
11515 | 1.01M | if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt))) |
11516 | 980k | goto done; |
11517 | 31.2k | xmlParseInternalSubset(ctxt); |
11518 | 31.2k | ctxt->inSubset = 2; |
11519 | 31.2k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
11520 | 31.2k | (ctxt->sax->externalSubset != NULL)) |
11521 | 24.8k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
11522 | 24.8k | ctxt->extSubSystem, ctxt->extSubURI); |
11523 | 31.2k | ctxt->inSubset = 0; |
11524 | 31.2k | xmlCleanSpecialAttr(ctxt); |
11525 | 31.2k | ctxt->instate = XML_PARSER_PROLOG; |
11526 | 31.2k | break; |
11527 | 1.01M | } |
11528 | 0 | default: |
11529 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
11530 | 0 | "PP: internal error\n"); |
11531 | 0 | ctxt->instate = XML_PARSER_EOF; |
11532 | 0 | break; |
11533 | 9.95M | } |
11534 | 9.95M | } |
11535 | 4.04M | done: |
11536 | 4.04M | return(ret); |
11537 | 177k | encoding_error: |
11538 | | /* Only report the first error */ |
11539 | 177k | if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) { |
11540 | 928 | xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL); |
11541 | 928 | ctxt->input->flags |= XML_INPUT_ENCODING_ERROR; |
11542 | 928 | } |
11543 | 177k | return(0); |
11544 | 4.22M | } |
11545 | | |
11546 | | /** |
11547 | | * xmlParseChunk: |
11548 | | * @ctxt: an XML parser context |
11549 | | * @chunk: chunk of memory |
11550 | | * @size: size of chunk in bytes |
11551 | | * @terminate: last chunk indicator |
11552 | | * |
11553 | | * Parse a chunk of memory in push parser mode. |
11554 | | * |
11555 | | * Assumes that the parser context was initialized with |
11556 | | * xmlCreatePushParserCtxt. |
11557 | | * |
11558 | | * The last chunk, which will often be empty, must be marked with |
11559 | | * the @terminate flag. With the default SAX callbacks, the resulting |
11560 | | * document will be available in ctxt->myDoc. This pointer will not |
11561 | | * be freed by the library. |
11562 | | * |
11563 | | * If the document isn't well-formed, ctxt->myDoc is set to NULL. |
11564 | | * The push parser doesn't support recovery mode. |
11565 | | * |
11566 | | * Returns an xmlParserErrors code (0 on success). |
11567 | | */ |
11568 | | int |
11569 | | xmlParseChunk(xmlParserCtxtPtr ctxt, const char *chunk, int size, |
11570 | 5.00M | int terminate) { |
11571 | 5.00M | size_t curBase; |
11572 | 5.00M | size_t maxLength; |
11573 | 5.00M | size_t pos; |
11574 | 5.00M | int end_in_lf = 0; |
11575 | 5.00M | int res; |
11576 | | |
11577 | 5.00M | if ((ctxt == NULL) || (size < 0)) |
11578 | 0 | return(XML_ERR_ARGUMENT); |
11579 | 5.00M | if ((chunk == NULL) && (size > 0)) |
11580 | 0 | return(XML_ERR_ARGUMENT); |
11581 | 5.00M | if ((ctxt->input == NULL) || (ctxt->input->buf == NULL)) |
11582 | 0 | return(XML_ERR_ARGUMENT); |
11583 | 5.00M | if (ctxt->disableSAX != 0) |
11584 | 778k | return(ctxt->errNo); |
11585 | | |
11586 | 4.22M | ctxt->input->flags |= XML_INPUT_PROGRESSIVE; |
11587 | 4.22M | if (ctxt->instate == XML_PARSER_START) |
11588 | 65.3k | xmlCtxtInitializeLate(ctxt); |
11589 | 4.22M | if ((size > 0) && (chunk != NULL) && (!terminate) && |
11590 | 4.22M | (chunk[size - 1] == '\r')) { |
11591 | 14.7k | end_in_lf = 1; |
11592 | 14.7k | size--; |
11593 | 14.7k | } |
11594 | | |
11595 | | /* |
11596 | | * Also push an empty chunk to make sure that the raw buffer |
11597 | | * will be flushed if there is an encoder. |
11598 | | */ |
11599 | 4.22M | pos = ctxt->input->cur - ctxt->input->base; |
11600 | 4.22M | res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
11601 | 4.22M | xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos); |
11602 | 4.22M | if (res < 0) { |
11603 | 149 | xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL); |
11604 | 149 | xmlHaltParser(ctxt); |
11605 | 149 | return(ctxt->errNo); |
11606 | 149 | } |
11607 | | |
11608 | 4.22M | xmlParseTryOrFinish(ctxt, terminate); |
11609 | | |
11610 | 4.22M | curBase = ctxt->input->cur - ctxt->input->base; |
11611 | 4.22M | maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
11612 | 1.28M | XML_MAX_HUGE_LENGTH : |
11613 | 4.22M | XML_MAX_LOOKUP_LIMIT; |
11614 | 4.22M | if (curBase > maxLength) { |
11615 | 0 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, |
11616 | 0 | "Buffer size limit exceeded, try XML_PARSE_HUGE\n"); |
11617 | 0 | xmlHaltParser(ctxt); |
11618 | 0 | } |
11619 | | |
11620 | 4.22M | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
11621 | 11.3k | return(ctxt->errNo); |
11622 | | |
11623 | 4.21M | if (end_in_lf == 1) { |
11624 | 14.7k | pos = ctxt->input->cur - ctxt->input->base; |
11625 | 14.7k | res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r"); |
11626 | 14.7k | xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos); |
11627 | 14.7k | if (res < 0) { |
11628 | 6 | xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL); |
11629 | 6 | xmlHaltParser(ctxt); |
11630 | 6 | return(ctxt->errNo); |
11631 | 6 | } |
11632 | 14.7k | } |
11633 | 4.21M | if (terminate) { |
11634 | | /* |
11635 | | * Check for termination |
11636 | | */ |
11637 | 55.4k | if ((ctxt->instate != XML_PARSER_EOF) && |
11638 | 55.4k | (ctxt->instate != XML_PARSER_EPILOG)) { |
11639 | 18.6k | if (ctxt->nameNr > 0) { |
11640 | 11.3k | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
11641 | 11.3k | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
11642 | 11.3k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
11643 | 11.3k | "Premature end of data in tag %s line %d\n", |
11644 | 11.3k | name, line, NULL); |
11645 | 11.3k | } else if (ctxt->instate == XML_PARSER_START) { |
11646 | 160 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11647 | 7.13k | } else { |
11648 | 7.13k | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
11649 | 7.13k | "Start tag expected, '<' not found\n"); |
11650 | 7.13k | } |
11651 | 36.7k | } else if ((ctxt->input->buf->encoder != NULL) && |
11652 | 36.7k | (ctxt->input->buf->error == 0) && |
11653 | 36.7k | (!xmlBufIsEmpty(ctxt->input->buf->raw))) { |
11654 | 571 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
11655 | 571 | "Truncated multi-byte sequence at EOF\n"); |
11656 | 571 | } |
11657 | 55.4k | if (ctxt->instate != XML_PARSER_EOF) { |
11658 | 22.3k | ctxt->instate = XML_PARSER_EOF; |
11659 | 22.3k | xmlFinishDocument(ctxt); |
11660 | 22.3k | } |
11661 | 55.4k | } |
11662 | 4.21M | if (ctxt->wellFormed == 0) |
11663 | 2.12M | return((xmlParserErrors) ctxt->errNo); |
11664 | 2.09M | else |
11665 | 2.09M | return(0); |
11666 | 4.21M | } |
11667 | | |
11668 | | /************************************************************************ |
11669 | | * * |
11670 | | * I/O front end functions to the parser * |
11671 | | * * |
11672 | | ************************************************************************/ |
11673 | | |
11674 | | /** |
11675 | | * xmlCreatePushParserCtxt: |
11676 | | * @sax: a SAX handler (optional) |
11677 | | * @user_data: user data for SAX callbacks (optional) |
11678 | | * @chunk: initial chunk (optional, deprecated) |
11679 | | * @size: size of initial chunk in bytes |
11680 | | * @filename: file name or URI (optional) |
11681 | | * |
11682 | | * Create a parser context for using the XML parser in push mode. |
11683 | | * See xmlParseChunk. |
11684 | | * |
11685 | | * Passing an initial chunk is useless and deprecated. |
11686 | | * |
11687 | | * @filename is used as base URI to fetch external entities and for |
11688 | | * error reports. |
11689 | | * |
11690 | | * Returns the new parser context or NULL if a memory allocation |
11691 | | * failed. |
11692 | | */ |
11693 | | |
11694 | | xmlParserCtxtPtr |
11695 | | xmlCreatePushParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
11696 | 64.1k | const char *chunk, int size, const char *filename) { |
11697 | 64.1k | xmlParserCtxtPtr ctxt; |
11698 | 64.1k | xmlParserInputPtr input; |
11699 | | |
11700 | 64.1k | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
11701 | 64.1k | if (ctxt == NULL) |
11702 | 50 | return(NULL); |
11703 | | |
11704 | 64.0k | ctxt->options &= ~XML_PARSE_NODICT; |
11705 | 64.0k | ctxt->dictNames = 1; |
11706 | | |
11707 | 64.0k | input = xmlNewPushInput(filename, chunk, size); |
11708 | 64.0k | if (input == NULL) { |
11709 | 39 | xmlFreeParserCtxt(ctxt); |
11710 | 39 | return(NULL); |
11711 | 39 | } |
11712 | 64.0k | if (inputPush(ctxt, input) < 0) { |
11713 | 14 | xmlFreeInputStream(input); |
11714 | 14 | xmlFreeParserCtxt(ctxt); |
11715 | 14 | return(NULL); |
11716 | 14 | } |
11717 | | |
11718 | 63.9k | return(ctxt); |
11719 | 64.0k | } |
11720 | | #endif /* LIBXML_PUSH_ENABLED */ |
11721 | | |
11722 | | /** |
11723 | | * xmlStopParser: |
11724 | | * @ctxt: an XML parser context |
11725 | | * |
11726 | | * Blocks further parser processing |
11727 | | */ |
11728 | | void |
11729 | 27.8k | xmlStopParser(xmlParserCtxtPtr ctxt) { |
11730 | 27.8k | if (ctxt == NULL) |
11731 | 0 | return; |
11732 | 27.8k | xmlHaltParser(ctxt); |
11733 | 27.8k | if (ctxt->errNo != XML_ERR_NO_MEMORY) |
11734 | 19.0k | ctxt->errNo = XML_ERR_USER_STOP; |
11735 | 27.8k | } |
11736 | | |
11737 | | /** |
11738 | | * xmlCreateIOParserCtxt: |
11739 | | * @sax: a SAX handler (optional) |
11740 | | * @user_data: user data for SAX callbacks (optional) |
11741 | | * @ioread: an I/O read function |
11742 | | * @ioclose: an I/O close function (optional) |
11743 | | * @ioctx: an I/O handler |
11744 | | * @enc: the charset encoding if known (deprecated) |
11745 | | * |
11746 | | * Create a parser context for using the XML parser with an existing |
11747 | | * I/O stream |
11748 | | * |
11749 | | * Returns the new parser context or NULL |
11750 | | */ |
11751 | | xmlParserCtxtPtr |
11752 | | xmlCreateIOParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
11753 | | xmlInputReadCallback ioread, |
11754 | | xmlInputCloseCallback ioclose, |
11755 | 0 | void *ioctx, xmlCharEncoding enc) { |
11756 | 0 | xmlParserCtxtPtr ctxt; |
11757 | 0 | xmlParserInputPtr input; |
11758 | 0 | const char *encoding; |
11759 | |
|
11760 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
11761 | 0 | if (ctxt == NULL) |
11762 | 0 | return(NULL); |
11763 | | |
11764 | 0 | encoding = xmlGetCharEncodingName(enc); |
11765 | 0 | input = xmlCtxtNewInputFromIO(ctxt, NULL, ioread, ioclose, ioctx, |
11766 | 0 | encoding, 0); |
11767 | 0 | if (input == NULL) { |
11768 | 0 | xmlFreeParserCtxt(ctxt); |
11769 | 0 | return (NULL); |
11770 | 0 | } |
11771 | 0 | if (inputPush(ctxt, input) < 0) { |
11772 | 0 | xmlFreeInputStream(input); |
11773 | 0 | xmlFreeParserCtxt(ctxt); |
11774 | 0 | return(NULL); |
11775 | 0 | } |
11776 | | |
11777 | 0 | return(ctxt); |
11778 | 0 | } |
11779 | | |
11780 | | #ifdef LIBXML_VALID_ENABLED |
11781 | | /************************************************************************ |
11782 | | * * |
11783 | | * Front ends when parsing a DTD * |
11784 | | * * |
11785 | | ************************************************************************/ |
11786 | | |
11787 | | /** |
11788 | | * xmlIOParseDTD: |
11789 | | * @sax: the SAX handler block or NULL |
11790 | | * @input: an Input Buffer |
11791 | | * @enc: the charset encoding if known |
11792 | | * |
11793 | | * Load and parse a DTD |
11794 | | * |
11795 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
11796 | | * @input will be freed by the function in any case. |
11797 | | */ |
11798 | | |
11799 | | xmlDtdPtr |
11800 | | xmlIOParseDTD(xmlSAXHandlerPtr sax, xmlParserInputBufferPtr input, |
11801 | 0 | xmlCharEncoding enc) { |
11802 | 0 | xmlDtdPtr ret = NULL; |
11803 | 0 | xmlParserCtxtPtr ctxt; |
11804 | 0 | xmlParserInputPtr pinput = NULL; |
11805 | |
|
11806 | 0 | if (input == NULL) |
11807 | 0 | return(NULL); |
11808 | | |
11809 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
11810 | 0 | if (ctxt == NULL) { |
11811 | 0 | xmlFreeParserInputBuffer(input); |
11812 | 0 | return(NULL); |
11813 | 0 | } |
11814 | | |
11815 | | /* |
11816 | | * generate a parser input from the I/O handler |
11817 | | */ |
11818 | | |
11819 | 0 | pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
11820 | 0 | if (pinput == NULL) { |
11821 | 0 | xmlFreeParserInputBuffer(input); |
11822 | 0 | xmlFreeParserCtxt(ctxt); |
11823 | 0 | return(NULL); |
11824 | 0 | } |
11825 | | |
11826 | | /* |
11827 | | * plug some encoding conversion routines here. |
11828 | | */ |
11829 | 0 | if (xmlPushInput(ctxt, pinput) < 0) { |
11830 | 0 | xmlFreeInputStream(pinput); |
11831 | 0 | xmlFreeParserCtxt(ctxt); |
11832 | 0 | return(NULL); |
11833 | 0 | } |
11834 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
11835 | 0 | xmlSwitchEncoding(ctxt, enc); |
11836 | 0 | } |
11837 | | |
11838 | | /* |
11839 | | * let's parse that entity knowing it's an external subset. |
11840 | | */ |
11841 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
11842 | 0 | if (ctxt->myDoc == NULL) { |
11843 | 0 | xmlErrMemory(ctxt); |
11844 | 0 | return(NULL); |
11845 | 0 | } |
11846 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
11847 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
11848 | 0 | BAD_CAST "none", BAD_CAST "none"); |
11849 | |
|
11850 | 0 | xmlParseExternalSubset(ctxt, BAD_CAST "none", BAD_CAST "none"); |
11851 | |
|
11852 | 0 | if (ctxt->myDoc != NULL) { |
11853 | 0 | if (ctxt->wellFormed) { |
11854 | 0 | ret = ctxt->myDoc->extSubset; |
11855 | 0 | ctxt->myDoc->extSubset = NULL; |
11856 | 0 | if (ret != NULL) { |
11857 | 0 | xmlNodePtr tmp; |
11858 | |
|
11859 | 0 | ret->doc = NULL; |
11860 | 0 | tmp = ret->children; |
11861 | 0 | while (tmp != NULL) { |
11862 | 0 | tmp->doc = NULL; |
11863 | 0 | tmp = tmp->next; |
11864 | 0 | } |
11865 | 0 | } |
11866 | 0 | } else { |
11867 | 0 | ret = NULL; |
11868 | 0 | } |
11869 | 0 | xmlFreeDoc(ctxt->myDoc); |
11870 | 0 | ctxt->myDoc = NULL; |
11871 | 0 | } |
11872 | 0 | xmlFreeParserCtxt(ctxt); |
11873 | |
|
11874 | 0 | return(ret); |
11875 | 0 | } |
11876 | | |
11877 | | /** |
11878 | | * xmlSAXParseDTD: |
11879 | | * @sax: the SAX handler block |
11880 | | * @ExternalID: a NAME* containing the External ID of the DTD |
11881 | | * @SystemID: a NAME* containing the URL to the DTD |
11882 | | * |
11883 | | * DEPRECATED: Don't use. |
11884 | | * |
11885 | | * Load and parse an external subset. |
11886 | | * |
11887 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
11888 | | */ |
11889 | | |
11890 | | xmlDtdPtr |
11891 | | xmlSAXParseDTD(xmlSAXHandlerPtr sax, const xmlChar *ExternalID, |
11892 | 16.1k | const xmlChar *SystemID) { |
11893 | 16.1k | xmlDtdPtr ret = NULL; |
11894 | 16.1k | xmlParserCtxtPtr ctxt; |
11895 | 16.1k | xmlParserInputPtr input = NULL; |
11896 | 16.1k | xmlChar* systemIdCanonic; |
11897 | | |
11898 | 16.1k | if ((ExternalID == NULL) && (SystemID == NULL)) return(NULL); |
11899 | | |
11900 | 16.1k | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
11901 | 16.1k | if (ctxt == NULL) { |
11902 | 92 | return(NULL); |
11903 | 92 | } |
11904 | | |
11905 | | /* |
11906 | | * Canonicalise the system ID |
11907 | | */ |
11908 | 16.0k | systemIdCanonic = xmlCanonicPath(SystemID); |
11909 | 16.0k | if ((SystemID != NULL) && (systemIdCanonic == NULL)) { |
11910 | 11 | xmlFreeParserCtxt(ctxt); |
11911 | 11 | return(NULL); |
11912 | 11 | } |
11913 | | |
11914 | | /* |
11915 | | * Ask the Entity resolver to load the damn thing |
11916 | | */ |
11917 | | |
11918 | 16.0k | if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) |
11919 | 16.0k | input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID, |
11920 | 16.0k | systemIdCanonic); |
11921 | 16.0k | if (input == NULL) { |
11922 | 14.4k | xmlFreeParserCtxt(ctxt); |
11923 | 14.4k | if (systemIdCanonic != NULL) |
11924 | 13.9k | xmlFree(systemIdCanonic); |
11925 | 14.4k | return(NULL); |
11926 | 14.4k | } |
11927 | | |
11928 | | /* |
11929 | | * plug some encoding conversion routines here. |
11930 | | */ |
11931 | 1.62k | if (xmlPushInput(ctxt, input) < 0) { |
11932 | 2 | xmlFreeInputStream(input); |
11933 | 2 | xmlFreeParserCtxt(ctxt); |
11934 | 2 | if (systemIdCanonic != NULL) |
11935 | 2 | xmlFree(systemIdCanonic); |
11936 | 2 | return(NULL); |
11937 | 2 | } |
11938 | | |
11939 | 1.62k | xmlDetectEncoding(ctxt); |
11940 | | |
11941 | 1.62k | if (input->filename == NULL) |
11942 | 772 | input->filename = (char *) systemIdCanonic; |
11943 | 850 | else |
11944 | 850 | xmlFree(systemIdCanonic); |
11945 | | |
11946 | | /* |
11947 | | * let's parse that entity knowing it's an external subset. |
11948 | | */ |
11949 | 1.62k | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
11950 | 1.62k | if (ctxt->myDoc == NULL) { |
11951 | 12 | xmlErrMemory(ctxt); |
11952 | 12 | xmlFreeParserCtxt(ctxt); |
11953 | 12 | return(NULL); |
11954 | 12 | } |
11955 | 1.61k | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
11956 | 1.61k | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
11957 | 1.61k | ExternalID, SystemID); |
11958 | 1.61k | if (ctxt->myDoc->extSubset == NULL) { |
11959 | 19 | xmlFreeDoc(ctxt->myDoc); |
11960 | 19 | xmlFreeParserCtxt(ctxt); |
11961 | 19 | return(NULL); |
11962 | 19 | } |
11963 | 1.59k | xmlParseExternalSubset(ctxt, ExternalID, SystemID); |
11964 | | |
11965 | 1.59k | if (ctxt->myDoc != NULL) { |
11966 | 1.59k | if (ctxt->wellFormed) { |
11967 | 92 | ret = ctxt->myDoc->extSubset; |
11968 | 92 | ctxt->myDoc->extSubset = NULL; |
11969 | 92 | if (ret != NULL) { |
11970 | 92 | xmlNodePtr tmp; |
11971 | | |
11972 | 92 | ret->doc = NULL; |
11973 | 92 | tmp = ret->children; |
11974 | 4.63k | while (tmp != NULL) { |
11975 | 4.54k | tmp->doc = NULL; |
11976 | 4.54k | tmp = tmp->next; |
11977 | 4.54k | } |
11978 | 92 | } |
11979 | 1.49k | } else { |
11980 | 1.49k | ret = NULL; |
11981 | 1.49k | } |
11982 | 1.59k | xmlFreeDoc(ctxt->myDoc); |
11983 | 1.59k | ctxt->myDoc = NULL; |
11984 | 1.59k | } |
11985 | 1.59k | xmlFreeParserCtxt(ctxt); |
11986 | | |
11987 | 1.59k | return(ret); |
11988 | 1.61k | } |
11989 | | |
11990 | | |
11991 | | /** |
11992 | | * xmlParseDTD: |
11993 | | * @ExternalID: a NAME* containing the External ID of the DTD |
11994 | | * @SystemID: a NAME* containing the URL to the DTD |
11995 | | * |
11996 | | * Load and parse an external subset. |
11997 | | * |
11998 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
11999 | | */ |
12000 | | |
12001 | | xmlDtdPtr |
12002 | 16.1k | xmlParseDTD(const xmlChar *ExternalID, const xmlChar *SystemID) { |
12003 | 16.1k | return(xmlSAXParseDTD(NULL, ExternalID, SystemID)); |
12004 | 16.1k | } |
12005 | | #endif /* LIBXML_VALID_ENABLED */ |
12006 | | |
12007 | | /************************************************************************ |
12008 | | * * |
12009 | | * Front ends when parsing an Entity * |
12010 | | * * |
12011 | | ************************************************************************/ |
12012 | | |
12013 | | static xmlNodePtr |
12014 | | xmlCtxtParseContentInternal(xmlParserCtxtPtr ctxt, xmlParserInputPtr input, |
12015 | 60.0k | int hasTextDecl, int buildTree) { |
12016 | 60.0k | xmlNodePtr root = NULL; |
12017 | 60.0k | xmlNodePtr list = NULL; |
12018 | 60.0k | xmlChar *rootName = BAD_CAST "#root"; |
12019 | 60.0k | int result; |
12020 | | |
12021 | 60.0k | if (buildTree) { |
12022 | 60.0k | root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL); |
12023 | 60.0k | if (root == NULL) { |
12024 | 55 | xmlErrMemory(ctxt); |
12025 | 55 | goto error; |
12026 | 55 | } |
12027 | 60.0k | } |
12028 | | |
12029 | 59.9k | if (xmlPushInput(ctxt, input) < 0) |
12030 | 46 | goto error; |
12031 | | |
12032 | 59.9k | nameNsPush(ctxt, rootName, NULL, NULL, 0, 0); |
12033 | 59.9k | spacePush(ctxt, -1); |
12034 | | |
12035 | 59.9k | if (buildTree) |
12036 | 59.9k | nodePush(ctxt, root); |
12037 | | |
12038 | 59.9k | if (hasTextDecl) { |
12039 | 14.2k | xmlDetectEncoding(ctxt); |
12040 | | |
12041 | | /* |
12042 | | * Parse a possible text declaration first |
12043 | | */ |
12044 | 14.2k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
12045 | 14.2k | (IS_BLANK_CH(NXT(5)))) { |
12046 | 1.99k | xmlParseTextDecl(ctxt); |
12047 | | /* |
12048 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
12049 | | */ |
12050 | 1.99k | if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) && |
12051 | 1.99k | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
12052 | 46 | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
12053 | 46 | "Version mismatch between document and " |
12054 | 46 | "entity\n"); |
12055 | 46 | } |
12056 | 1.99k | } |
12057 | 14.2k | } |
12058 | | |
12059 | 59.9k | xmlParseContentInternal(ctxt); |
12060 | | |
12061 | 59.9k | if (ctxt->input->cur < ctxt->input->end) |
12062 | 3.97k | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
12063 | | |
12064 | 59.9k | if ((ctxt->wellFormed) || |
12065 | 59.9k | ((ctxt->recovery) && (ctxt->errNo != XML_ERR_NO_MEMORY))) { |
12066 | 53.2k | if (root != NULL) { |
12067 | 53.2k | xmlNodePtr cur; |
12068 | | |
12069 | | /* |
12070 | | * Unlink newly created node list. |
12071 | | */ |
12072 | 53.2k | list = root->children; |
12073 | 53.2k | root->children = NULL; |
12074 | 53.2k | root->last = NULL; |
12075 | 107k | for (cur = list; cur != NULL; cur = cur->next) |
12076 | 54.6k | cur->parent = NULL; |
12077 | 53.2k | } |
12078 | 53.2k | } |
12079 | | |
12080 | | /* |
12081 | | * Read the rest of the stream in case of errors. We want |
12082 | | * to account for the whole entity size. |
12083 | | */ |
12084 | 188k | do { |
12085 | 188k | ctxt->input->cur = ctxt->input->end; |
12086 | 188k | xmlParserShrink(ctxt); |
12087 | 188k | result = xmlParserGrow(ctxt); |
12088 | 188k | } while (result > 0); |
12089 | | |
12090 | 59.9k | if (buildTree) |
12091 | 59.9k | nodePop(ctxt); |
12092 | | |
12093 | 59.9k | namePop(ctxt); |
12094 | 59.9k | spacePop(ctxt); |
12095 | | |
12096 | | /* xmlPopInput would free the stream */ |
12097 | 59.9k | inputPop(ctxt); |
12098 | | |
12099 | 60.0k | error: |
12100 | 60.0k | xmlFreeNode(root); |
12101 | | |
12102 | 60.0k | return(list); |
12103 | 59.9k | } |
12104 | | |
12105 | | static void |
12106 | 70.1k | xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) { |
12107 | 70.1k | xmlParserInputPtr input; |
12108 | 70.1k | xmlNodePtr list; |
12109 | 70.1k | unsigned long consumed; |
12110 | 70.1k | int isExternal; |
12111 | 70.1k | int buildTree; |
12112 | 70.1k | int oldMinNsIndex; |
12113 | 70.1k | int oldNodelen, oldNodemem; |
12114 | | |
12115 | 70.1k | isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY); |
12116 | 70.1k | buildTree = (ctxt->node != NULL); |
12117 | | |
12118 | | /* |
12119 | | * Recursion check |
12120 | | */ |
12121 | 70.1k | if (ent->flags & XML_ENT_EXPANDING) { |
12122 | 750 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
12123 | 750 | xmlHaltParser(ctxt); |
12124 | 750 | goto error; |
12125 | 750 | } |
12126 | | |
12127 | | /* |
12128 | | * Load entity |
12129 | | */ |
12130 | 69.3k | input = xmlNewEntityInputStream(ctxt, ent); |
12131 | 69.3k | if (input == NULL) |
12132 | 9.33k | goto error; |
12133 | | |
12134 | | /* |
12135 | | * When building a tree, we need to limit the scope of namespace |
12136 | | * declarations, so that entities don't reference xmlNs structs |
12137 | | * from the parent of a reference. |
12138 | | */ |
12139 | 60.0k | oldMinNsIndex = ctxt->nsdb->minNsIndex; |
12140 | 60.0k | if (buildTree) |
12141 | 60.0k | ctxt->nsdb->minNsIndex = ctxt->nsNr; |
12142 | | |
12143 | 60.0k | oldNodelen = ctxt->nodelen; |
12144 | 60.0k | oldNodemem = ctxt->nodemem; |
12145 | 60.0k | ctxt->nodelen = 0; |
12146 | 60.0k | ctxt->nodemem = 0; |
12147 | | |
12148 | | /* |
12149 | | * Parse content |
12150 | | * |
12151 | | * This initiates a recursive call chain: |
12152 | | * |
12153 | | * - xmlCtxtParseContentInternal |
12154 | | * - xmlParseContentInternal |
12155 | | * - xmlParseReference |
12156 | | * - xmlCtxtParseEntity |
12157 | | * |
12158 | | * The nesting depth is limited by the maximum number of inputs, |
12159 | | * see xmlPushInput. |
12160 | | * |
12161 | | * It's possible to make this non-recursive (minNsIndex must be |
12162 | | * stored in the input struct) at the expense of code readability. |
12163 | | */ |
12164 | | |
12165 | 60.0k | ent->flags |= XML_ENT_EXPANDING; |
12166 | | |
12167 | 60.0k | list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree); |
12168 | | |
12169 | 60.0k | ent->flags &= ~XML_ENT_EXPANDING; |
12170 | | |
12171 | 60.0k | ctxt->nsdb->minNsIndex = oldMinNsIndex; |
12172 | 60.0k | ctxt->nodelen = oldNodelen; |
12173 | 60.0k | ctxt->nodemem = oldNodemem; |
12174 | | |
12175 | | /* |
12176 | | * Entity size accounting |
12177 | | */ |
12178 | 60.0k | consumed = input->consumed; |
12179 | 60.0k | xmlSaturatedAddSizeT(&consumed, input->end - input->base); |
12180 | | |
12181 | 60.0k | if ((ent->flags & XML_ENT_CHECKED) == 0) |
12182 | 21.2k | xmlSaturatedAdd(&ent->expandedSize, consumed); |
12183 | | |
12184 | 60.0k | if ((ent->flags & XML_ENT_PARSED) == 0) { |
12185 | 21.3k | if (isExternal) |
12186 | 12.9k | xmlSaturatedAdd(&ctxt->sizeentities, consumed); |
12187 | | |
12188 | 21.3k | ent->children = list; |
12189 | | |
12190 | 75.8k | while (list != NULL) { |
12191 | 54.5k | list->parent = (xmlNodePtr) ent; |
12192 | 54.5k | if (list->next == NULL) |
12193 | 14.4k | ent->last = list; |
12194 | 54.5k | list = list->next; |
12195 | 54.5k | } |
12196 | 38.7k | } else { |
12197 | 38.7k | xmlFreeNodeList(list); |
12198 | 38.7k | } |
12199 | | |
12200 | 60.0k | xmlFreeInputStream(input); |
12201 | | |
12202 | 70.1k | error: |
12203 | 70.1k | ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED; |
12204 | 70.1k | } |
12205 | | |
12206 | | /** |
12207 | | * xmlParseCtxtExternalEntity: |
12208 | | * @ctxt: the existing parsing context |
12209 | | * @URL: the URL for the entity to load |
12210 | | * @ID: the System ID for the entity to load |
12211 | | * @listOut: the return value for the set of parsed nodes |
12212 | | * |
12213 | | * Parse an external general entity within an existing parsing context |
12214 | | * An external general parsed entity is well-formed if it matches the |
12215 | | * production labeled extParsedEnt. |
12216 | | * |
12217 | | * [78] extParsedEnt ::= TextDecl? content |
12218 | | * |
12219 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
12220 | | * the parser error code otherwise |
12221 | | */ |
12222 | | |
12223 | | int |
12224 | | xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctxt, const xmlChar *URL, |
12225 | 0 | const xmlChar *ID, xmlNodePtr *listOut) { |
12226 | 0 | xmlParserInputPtr input; |
12227 | 0 | xmlNodePtr list; |
12228 | |
|
12229 | 0 | if (listOut != NULL) |
12230 | 0 | *listOut = NULL; |
12231 | |
|
12232 | 0 | if (ctxt == NULL) |
12233 | 0 | return(XML_ERR_ARGUMENT); |
12234 | | |
12235 | 0 | input = xmlLoadResource(ctxt, (char *) URL, (char *) ID, |
12236 | 0 | XML_RESOURCE_GENERAL_ENTITY); |
12237 | 0 | if (input == NULL) |
12238 | 0 | return(ctxt->errNo); |
12239 | | |
12240 | 0 | xmlCtxtInitializeLate(ctxt); |
12241 | |
|
12242 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1); |
12243 | 0 | if (listOut != NULL) |
12244 | 0 | *listOut = list; |
12245 | 0 | else |
12246 | 0 | xmlFreeNodeList(list); |
12247 | |
|
12248 | 0 | xmlFreeInputStream(input); |
12249 | 0 | return(ctxt->errNo); |
12250 | 0 | } |
12251 | | |
12252 | | #ifdef LIBXML_SAX1_ENABLED |
12253 | | /** |
12254 | | * xmlParseExternalEntity: |
12255 | | * @doc: the document the chunk pertains to |
12256 | | * @sax: the SAX handler block (possibly NULL) |
12257 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
12258 | | * @depth: Used for loop detection, use 0 |
12259 | | * @URL: the URL for the entity to load |
12260 | | * @ID: the System ID for the entity to load |
12261 | | * @list: the return value for the set of parsed nodes |
12262 | | * |
12263 | | * DEPRECATED: Use xmlParseCtxtExternalEntity. |
12264 | | * |
12265 | | * Parse an external general entity |
12266 | | * An external general parsed entity is well-formed if it matches the |
12267 | | * production labeled extParsedEnt. |
12268 | | * |
12269 | | * [78] extParsedEnt ::= TextDecl? content |
12270 | | * |
12271 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
12272 | | * the parser error code otherwise |
12273 | | */ |
12274 | | |
12275 | | int |
12276 | | xmlParseExternalEntity(xmlDocPtr doc, xmlSAXHandlerPtr sax, void *user_data, |
12277 | 0 | int depth, const xmlChar *URL, const xmlChar *ID, xmlNodePtr *list) { |
12278 | 0 | xmlParserCtxtPtr ctxt; |
12279 | 0 | int ret; |
12280 | |
|
12281 | 0 | if (list != NULL) |
12282 | 0 | *list = NULL; |
12283 | |
|
12284 | 0 | if (doc == NULL) |
12285 | 0 | return(XML_ERR_ARGUMENT); |
12286 | | |
12287 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
12288 | 0 | if (ctxt == NULL) |
12289 | 0 | return(XML_ERR_NO_MEMORY); |
12290 | | |
12291 | 0 | ctxt->depth = depth; |
12292 | 0 | ctxt->myDoc = doc; |
12293 | 0 | ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list); |
12294 | |
|
12295 | 0 | xmlFreeParserCtxt(ctxt); |
12296 | 0 | return(ret); |
12297 | 0 | } |
12298 | | |
12299 | | /** |
12300 | | * xmlParseBalancedChunkMemory: |
12301 | | * @doc: the document the chunk pertains to (must not be NULL) |
12302 | | * @sax: the SAX handler block (possibly NULL) |
12303 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
12304 | | * @depth: Used for loop detection, use 0 |
12305 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
12306 | | * @lst: the return value for the set of parsed nodes |
12307 | | * |
12308 | | * Parse a well-balanced chunk of an XML document |
12309 | | * called by the parser |
12310 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
12311 | | * the content production in the XML grammar: |
12312 | | * |
12313 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
12314 | | * |
12315 | | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
12316 | | * the parser error code otherwise |
12317 | | */ |
12318 | | |
12319 | | int |
12320 | | xmlParseBalancedChunkMemory(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
12321 | 0 | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst) { |
12322 | 0 | return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data, |
12323 | 0 | depth, string, lst, 0 ); |
12324 | 0 | } |
12325 | | #endif /* LIBXML_SAX1_ENABLED */ |
12326 | | |
12327 | | /** |
12328 | | * xmlCtxtParseContent: |
12329 | | * @ctxt: parser context |
12330 | | * @input: parser input |
12331 | | * @node: target node or document |
12332 | | * @hasTextDecl: whether to parse text declaration |
12333 | | * |
12334 | | * Parse a well-balanced chunk of XML matching the 'content' production. |
12335 | | * |
12336 | | * Namespaces in scope of @node and entities of @node's document are |
12337 | | * recognized. When validating, the DTD of @node's document is used. |
12338 | | * |
12339 | | * Always consumes @input even in error case. |
12340 | | * |
12341 | | * Available since 2.14.0. |
12342 | | * |
12343 | | * Returns a node list or NULL in case of error. |
12344 | | */ |
12345 | | xmlNodePtr |
12346 | | xmlCtxtParseContent(xmlParserCtxtPtr ctxt, xmlParserInputPtr input, |
12347 | 0 | xmlNodePtr node, int hasTextDecl) { |
12348 | 0 | xmlDocPtr doc; |
12349 | 0 | xmlNodePtr cur, list = NULL; |
12350 | 0 | int nsnr = 0; |
12351 | 0 | xmlDictPtr oldDict; |
12352 | 0 | int oldOptions, oldDictNames, oldLoadSubset; |
12353 | |
|
12354 | 0 | if ((ctxt == NULL) || (input == NULL) || (node == NULL)) { |
12355 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12356 | 0 | goto exit; |
12357 | 0 | } |
12358 | | |
12359 | 0 | doc = node->doc; |
12360 | 0 | if (doc == NULL) { |
12361 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12362 | 0 | goto exit; |
12363 | 0 | } |
12364 | | |
12365 | 0 | switch (node->type) { |
12366 | 0 | case XML_ELEMENT_NODE: |
12367 | 0 | case XML_DOCUMENT_NODE: |
12368 | 0 | case XML_HTML_DOCUMENT_NODE: |
12369 | 0 | break; |
12370 | | |
12371 | 0 | case XML_ATTRIBUTE_NODE: |
12372 | 0 | case XML_TEXT_NODE: |
12373 | 0 | case XML_CDATA_SECTION_NODE: |
12374 | 0 | case XML_ENTITY_REF_NODE: |
12375 | 0 | case XML_PI_NODE: |
12376 | 0 | case XML_COMMENT_NODE: |
12377 | 0 | for (cur = node->parent; cur != NULL; cur = node->parent) { |
12378 | 0 | if ((cur->type == XML_ELEMENT_NODE) || |
12379 | 0 | (cur->type == XML_DOCUMENT_NODE) || |
12380 | 0 | (cur->type == XML_HTML_DOCUMENT_NODE)) { |
12381 | 0 | node = cur; |
12382 | 0 | break; |
12383 | 0 | } |
12384 | 0 | } |
12385 | 0 | break; |
12386 | | |
12387 | 0 | default: |
12388 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12389 | 0 | goto exit; |
12390 | 0 | } |
12391 | | |
12392 | 0 | #ifdef LIBXML_HTML_ENABLED |
12393 | 0 | if (ctxt->html) |
12394 | 0 | htmlCtxtReset(ctxt); |
12395 | 0 | else |
12396 | 0 | #endif |
12397 | 0 | xmlCtxtReset(ctxt); |
12398 | |
|
12399 | 0 | oldDict = ctxt->dict; |
12400 | 0 | oldOptions = ctxt->options; |
12401 | 0 | oldDictNames = ctxt->dictNames; |
12402 | 0 | oldLoadSubset = ctxt->loadsubset; |
12403 | | |
12404 | | /* |
12405 | | * Use input doc's dict if present, else assure XML_PARSE_NODICT is set. |
12406 | | */ |
12407 | 0 | if (doc->dict != NULL) { |
12408 | 0 | ctxt->dict = doc->dict; |
12409 | 0 | } else { |
12410 | 0 | ctxt->options |= XML_PARSE_NODICT; |
12411 | 0 | ctxt->dictNames = 0; |
12412 | 0 | } |
12413 | | |
12414 | | /* |
12415 | | * Disable IDs |
12416 | | */ |
12417 | 0 | ctxt->loadsubset |= XML_SKIP_IDS; |
12418 | |
|
12419 | 0 | ctxt->myDoc = doc; |
12420 | |
|
12421 | 0 | #ifdef LIBXML_HTML_ENABLED |
12422 | 0 | if (ctxt->html) { |
12423 | | /* |
12424 | | * When parsing in context, it makes no sense to add implied |
12425 | | * elements like html/body/etc... |
12426 | | */ |
12427 | 0 | ctxt->options |= HTML_PARSE_NOIMPLIED; |
12428 | |
|
12429 | 0 | list = htmlCtxtParseContentInternal(ctxt, input); |
12430 | 0 | } else |
12431 | 0 | #endif |
12432 | 0 | { |
12433 | 0 | xmlCtxtInitializeLate(ctxt); |
12434 | | |
12435 | | /* |
12436 | | * This hack lowers the error level of undeclared entities |
12437 | | * from XML_ERR_FATAL (well-formedness error) to XML_ERR_ERROR |
12438 | | * or XML_ERR_WARNING. |
12439 | | */ |
12440 | 0 | ctxt->hasExternalSubset = 1; |
12441 | | |
12442 | | /* |
12443 | | * initialize the SAX2 namespaces stack |
12444 | | */ |
12445 | 0 | cur = node; |
12446 | 0 | while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) { |
12447 | 0 | xmlNsPtr ns = cur->nsDef; |
12448 | 0 | xmlHashedString hprefix, huri; |
12449 | |
|
12450 | 0 | while (ns != NULL) { |
12451 | 0 | hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1); |
12452 | 0 | huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1); |
12453 | 0 | if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0) |
12454 | 0 | nsnr++; |
12455 | 0 | ns = ns->next; |
12456 | 0 | } |
12457 | 0 | cur = cur->parent; |
12458 | 0 | } |
12459 | |
|
12460 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1); |
12461 | |
|
12462 | 0 | if (nsnr > 0) |
12463 | 0 | xmlParserNsPop(ctxt, nsnr); |
12464 | 0 | } |
12465 | |
|
12466 | 0 | ctxt->dict = oldDict; |
12467 | 0 | ctxt->options = oldOptions; |
12468 | 0 | ctxt->dictNames = oldDictNames; |
12469 | 0 | ctxt->loadsubset = oldLoadSubset; |
12470 | 0 | ctxt->myDoc = NULL; |
12471 | 0 | ctxt->node = NULL; |
12472 | |
|
12473 | 0 | exit: |
12474 | 0 | xmlFreeInputStream(input); |
12475 | 0 | return(list); |
12476 | 0 | } |
12477 | | |
12478 | | /** |
12479 | | * xmlParseInNodeContext: |
12480 | | * @node: the context node |
12481 | | * @data: the input string |
12482 | | * @datalen: the input string length in bytes |
12483 | | * @options: a combination of xmlParserOption |
12484 | | * @listOut: the return value for the set of parsed nodes |
12485 | | * |
12486 | | * Parse a well-balanced chunk of an XML document |
12487 | | * within the context (DTD, namespaces, etc ...) of the given node. |
12488 | | * |
12489 | | * The allowed sequence for the data is a Well Balanced Chunk defined by |
12490 | | * the content production in the XML grammar: |
12491 | | * |
12492 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
12493 | | * |
12494 | | * This function assumes the encoding of @node's document which is |
12495 | | * typically not what you want. A better alternative is |
12496 | | * xmlCtxtParseContent. |
12497 | | * |
12498 | | * Returns XML_ERR_OK if the chunk is well balanced, and the parser |
12499 | | * error code otherwise |
12500 | | */ |
12501 | | xmlParserErrors |
12502 | | xmlParseInNodeContext(xmlNodePtr node, const char *data, int datalen, |
12503 | 0 | int options, xmlNodePtr *listOut) { |
12504 | 0 | xmlParserCtxtPtr ctxt; |
12505 | 0 | xmlParserInputPtr input; |
12506 | 0 | xmlDocPtr doc; |
12507 | 0 | xmlNodePtr list; |
12508 | 0 | xmlParserErrors ret; |
12509 | |
|
12510 | 0 | if (listOut == NULL) |
12511 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12512 | 0 | *listOut = NULL; |
12513 | |
|
12514 | 0 | if ((node == NULL) || (data == NULL) || (datalen < 0)) |
12515 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12516 | | |
12517 | 0 | doc = node->doc; |
12518 | 0 | if (doc == NULL) |
12519 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12520 | | |
12521 | 0 | #ifdef LIBXML_HTML_ENABLED |
12522 | 0 | if (doc->type == XML_HTML_DOCUMENT_NODE) { |
12523 | 0 | ctxt = htmlNewParserCtxt(); |
12524 | 0 | } |
12525 | 0 | else |
12526 | 0 | #endif |
12527 | 0 | ctxt = xmlNewParserCtxt(); |
12528 | |
|
12529 | 0 | if (ctxt == NULL) |
12530 | 0 | return(XML_ERR_NO_MEMORY); |
12531 | | |
12532 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen, |
12533 | 0 | (const char *) doc->encoding, |
12534 | 0 | XML_INPUT_BUF_STATIC); |
12535 | 0 | if (input == NULL) { |
12536 | 0 | xmlFreeParserCtxt(ctxt); |
12537 | 0 | return(XML_ERR_NO_MEMORY); |
12538 | 0 | } |
12539 | | |
12540 | 0 | xmlCtxtUseOptions(ctxt, options); |
12541 | |
|
12542 | 0 | list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0); |
12543 | |
|
12544 | 0 | if (list == NULL) { |
12545 | 0 | ret = ctxt->errNo; |
12546 | 0 | if (ret == XML_ERR_ARGUMENT) |
12547 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
12548 | 0 | } else { |
12549 | 0 | ret = XML_ERR_OK; |
12550 | 0 | *listOut = list; |
12551 | 0 | } |
12552 | |
|
12553 | 0 | xmlFreeParserCtxt(ctxt); |
12554 | |
|
12555 | 0 | return(ret); |
12556 | 0 | } |
12557 | | |
12558 | | #ifdef LIBXML_SAX1_ENABLED |
12559 | | /** |
12560 | | * xmlParseBalancedChunkMemoryRecover: |
12561 | | * @doc: the document the chunk pertains to (must not be NULL) |
12562 | | * @sax: the SAX handler block (possibly NULL) |
12563 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
12564 | | * @depth: Used for loop detection, use 0 |
12565 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
12566 | | * @listOut: the return value for the set of parsed nodes |
12567 | | * @recover: return nodes even if the data is broken (use 0) |
12568 | | * |
12569 | | * Parse a well-balanced chunk of an XML document |
12570 | | * |
12571 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
12572 | | * the content production in the XML grammar: |
12573 | | * |
12574 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
12575 | | * |
12576 | | * Returns 0 if the chunk is well balanced, or thehe parser error code |
12577 | | * otherwise. |
12578 | | * |
12579 | | * In case recover is set to 1, the nodelist will not be empty even if |
12580 | | * the parsed chunk is not well balanced, assuming the parsing succeeded to |
12581 | | * some extent. |
12582 | | */ |
12583 | | int |
12584 | | xmlParseBalancedChunkMemoryRecover(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
12585 | | void *user_data, int depth, const xmlChar *string, xmlNodePtr *listOut, |
12586 | 0 | int recover) { |
12587 | 0 | xmlParserCtxtPtr ctxt; |
12588 | 0 | xmlParserInputPtr input; |
12589 | 0 | xmlNodePtr list; |
12590 | 0 | int ret; |
12591 | |
|
12592 | 0 | if (listOut != NULL) |
12593 | 0 | *listOut = NULL; |
12594 | |
|
12595 | 0 | if (string == NULL) |
12596 | 0 | return(XML_ERR_ARGUMENT); |
12597 | | |
12598 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
12599 | 0 | if (ctxt == NULL) |
12600 | 0 | return(XML_ERR_NO_MEMORY); |
12601 | | |
12602 | 0 | xmlCtxtInitializeLate(ctxt); |
12603 | |
|
12604 | 0 | ctxt->depth = depth; |
12605 | 0 | ctxt->myDoc = doc; |
12606 | 0 | if (recover) { |
12607 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12608 | 0 | ctxt->recovery = 1; |
12609 | 0 | } |
12610 | |
|
12611 | 0 | input = xmlNewStringInputStream(ctxt, string); |
12612 | 0 | if (input == NULL) { |
12613 | 0 | ret = ctxt->errNo; |
12614 | 0 | goto error; |
12615 | 0 | } |
12616 | | |
12617 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1); |
12618 | 0 | if (listOut != NULL) |
12619 | 0 | *listOut = list; |
12620 | 0 | else |
12621 | 0 | xmlFreeNodeList(list); |
12622 | |
|
12623 | 0 | if (!ctxt->wellFormed) |
12624 | 0 | ret = ctxt->errNo; |
12625 | 0 | else |
12626 | 0 | ret = XML_ERR_OK; |
12627 | |
|
12628 | 0 | error: |
12629 | 0 | xmlFreeInputStream(input); |
12630 | 0 | xmlFreeParserCtxt(ctxt); |
12631 | 0 | return(ret); |
12632 | 0 | } |
12633 | | |
12634 | | /** |
12635 | | * xmlSAXParseEntity: |
12636 | | * @sax: the SAX handler block |
12637 | | * @filename: the filename |
12638 | | * |
12639 | | * DEPRECATED: Don't use. |
12640 | | * |
12641 | | * parse an XML external entity out of context and build a tree. |
12642 | | * It use the given SAX function block to handle the parsing callback. |
12643 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12644 | | * |
12645 | | * [78] extParsedEnt ::= TextDecl? content |
12646 | | * |
12647 | | * This correspond to a "Well Balanced" chunk |
12648 | | * |
12649 | | * Returns the resulting document tree |
12650 | | */ |
12651 | | |
12652 | | xmlDocPtr |
12653 | 0 | xmlSAXParseEntity(xmlSAXHandlerPtr sax, const char *filename) { |
12654 | 0 | xmlDocPtr ret; |
12655 | 0 | xmlParserCtxtPtr ctxt; |
12656 | |
|
12657 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
12658 | 0 | if (ctxt == NULL) { |
12659 | 0 | return(NULL); |
12660 | 0 | } |
12661 | 0 | if (sax != NULL) { |
12662 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
12663 | 0 | *ctxt->sax = *sax; |
12664 | 0 | } else { |
12665 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
12666 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12667 | 0 | } |
12668 | 0 | ctxt->userData = NULL; |
12669 | 0 | } |
12670 | |
|
12671 | 0 | xmlParseExtParsedEnt(ctxt); |
12672 | |
|
12673 | 0 | if (ctxt->wellFormed) { |
12674 | 0 | ret = ctxt->myDoc; |
12675 | 0 | } else { |
12676 | 0 | ret = NULL; |
12677 | 0 | xmlFreeDoc(ctxt->myDoc); |
12678 | 0 | } |
12679 | |
|
12680 | 0 | xmlFreeParserCtxt(ctxt); |
12681 | |
|
12682 | 0 | return(ret); |
12683 | 0 | } |
12684 | | |
12685 | | /** |
12686 | | * xmlParseEntity: |
12687 | | * @filename: the filename |
12688 | | * |
12689 | | * parse an XML external entity out of context and build a tree. |
12690 | | * |
12691 | | * [78] extParsedEnt ::= TextDecl? content |
12692 | | * |
12693 | | * This correspond to a "Well Balanced" chunk |
12694 | | * |
12695 | | * Returns the resulting document tree |
12696 | | */ |
12697 | | |
12698 | | xmlDocPtr |
12699 | 0 | xmlParseEntity(const char *filename) { |
12700 | 0 | return(xmlSAXParseEntity(NULL, filename)); |
12701 | 0 | } |
12702 | | #endif /* LIBXML_SAX1_ENABLED */ |
12703 | | |
12704 | | /** |
12705 | | * xmlCreateEntityParserCtxt: |
12706 | | * @URL: the entity URL |
12707 | | * @ID: the entity PUBLIC ID |
12708 | | * @base: a possible base for the target URI |
12709 | | * |
12710 | | * DEPRECATED: Don't use. |
12711 | | * |
12712 | | * Create a parser context for an external entity |
12713 | | * Automatic support for ZLIB/Compress compressed document is provided |
12714 | | * by default if found at compile-time. |
12715 | | * |
12716 | | * Returns the new parser context or NULL |
12717 | | */ |
12718 | | xmlParserCtxtPtr |
12719 | | xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID, |
12720 | 0 | const xmlChar *base) { |
12721 | 0 | xmlParserCtxtPtr ctxt; |
12722 | 0 | xmlParserInputPtr input; |
12723 | 0 | xmlChar *uri = NULL; |
12724 | |
|
12725 | 0 | ctxt = xmlNewParserCtxt(); |
12726 | 0 | if (ctxt == NULL) |
12727 | 0 | return(NULL); |
12728 | | |
12729 | 0 | if (base != NULL) { |
12730 | 0 | if (xmlBuildURISafe(URL, base, &uri) < 0) |
12731 | 0 | goto error; |
12732 | 0 | if (uri != NULL) |
12733 | 0 | URL = uri; |
12734 | 0 | } |
12735 | | |
12736 | 0 | input = xmlLoadResource(ctxt, (char *) URL, (char *) ID, |
12737 | 0 | XML_RESOURCE_UNKNOWN); |
12738 | 0 | if (input == NULL) |
12739 | 0 | goto error; |
12740 | | |
12741 | 0 | if (inputPush(ctxt, input) < 0) { |
12742 | 0 | xmlFreeInputStream(input); |
12743 | 0 | goto error; |
12744 | 0 | } |
12745 | | |
12746 | 0 | xmlFree(uri); |
12747 | 0 | return(ctxt); |
12748 | | |
12749 | 0 | error: |
12750 | 0 | xmlFree(uri); |
12751 | 0 | xmlFreeParserCtxt(ctxt); |
12752 | 0 | return(NULL); |
12753 | 0 | } |
12754 | | |
12755 | | /************************************************************************ |
12756 | | * * |
12757 | | * Front ends when parsing from a file * |
12758 | | * * |
12759 | | ************************************************************************/ |
12760 | | |
12761 | | /** |
12762 | | * xmlCreateURLParserCtxt: |
12763 | | * @filename: the filename or URL |
12764 | | * @options: a combination of xmlParserOption |
12765 | | * |
12766 | | * DEPRECATED: Use xmlNewParserCtxt and xmlCtxtReadFile. |
12767 | | * |
12768 | | * Create a parser context for a file or URL content. |
12769 | | * Automatic support for ZLIB/Compress compressed document is provided |
12770 | | * by default if found at compile-time and for file accesses |
12771 | | * |
12772 | | * Returns the new parser context or NULL |
12773 | | */ |
12774 | | xmlParserCtxtPtr |
12775 | | xmlCreateURLParserCtxt(const char *filename, int options) |
12776 | 0 | { |
12777 | 0 | xmlParserCtxtPtr ctxt; |
12778 | 0 | xmlParserInputPtr input; |
12779 | |
|
12780 | 0 | ctxt = xmlNewParserCtxt(); |
12781 | 0 | if (ctxt == NULL) |
12782 | 0 | return(NULL); |
12783 | | |
12784 | 0 | xmlCtxtUseOptions(ctxt, options); |
12785 | 0 | ctxt->linenumbers = 1; |
12786 | |
|
12787 | 0 | input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT); |
12788 | 0 | if (input == NULL) { |
12789 | 0 | xmlFreeParserCtxt(ctxt); |
12790 | 0 | return(NULL); |
12791 | 0 | } |
12792 | 0 | if (inputPush(ctxt, input) < 0) { |
12793 | 0 | xmlFreeInputStream(input); |
12794 | 0 | xmlFreeParserCtxt(ctxt); |
12795 | 0 | return(NULL); |
12796 | 0 | } |
12797 | | |
12798 | 0 | return(ctxt); |
12799 | 0 | } |
12800 | | |
12801 | | /** |
12802 | | * xmlCreateFileParserCtxt: |
12803 | | * @filename: the filename |
12804 | | * |
12805 | | * DEPRECATED: Use xmlNewParserCtxt and xmlCtxtReadFile. |
12806 | | * |
12807 | | * Create a parser context for a file content. |
12808 | | * Automatic support for ZLIB/Compress compressed document is provided |
12809 | | * by default if found at compile-time. |
12810 | | * |
12811 | | * Returns the new parser context or NULL |
12812 | | */ |
12813 | | xmlParserCtxtPtr |
12814 | | xmlCreateFileParserCtxt(const char *filename) |
12815 | 0 | { |
12816 | 0 | return(xmlCreateURLParserCtxt(filename, 0)); |
12817 | 0 | } |
12818 | | |
12819 | | #ifdef LIBXML_SAX1_ENABLED |
12820 | | /** |
12821 | | * xmlSAXParseFileWithData: |
12822 | | * @sax: the SAX handler block |
12823 | | * @filename: the filename |
12824 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
12825 | | * documents |
12826 | | * @data: the userdata |
12827 | | * |
12828 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
12829 | | * |
12830 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12831 | | * compressed document is provided by default if found at compile-time. |
12832 | | * It use the given SAX function block to handle the parsing callback. |
12833 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12834 | | * |
12835 | | * User data (void *) is stored within the parser context in the |
12836 | | * context's _private member, so it is available nearly everywhere in libxml |
12837 | | * |
12838 | | * Returns the resulting document tree |
12839 | | */ |
12840 | | |
12841 | | xmlDocPtr |
12842 | | xmlSAXParseFileWithData(xmlSAXHandlerPtr sax, const char *filename, |
12843 | 0 | int recovery, void *data) { |
12844 | 0 | xmlDocPtr ret; |
12845 | 0 | xmlParserCtxtPtr ctxt; |
12846 | 0 | xmlParserInputPtr input; |
12847 | |
|
12848 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12849 | 0 | if (ctxt == NULL) |
12850 | 0 | return(NULL); |
12851 | | |
12852 | 0 | if (data != NULL) |
12853 | 0 | ctxt->_private = data; |
12854 | |
|
12855 | 0 | if (recovery) { |
12856 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12857 | 0 | ctxt->recovery = 1; |
12858 | 0 | } |
12859 | |
|
12860 | 0 | if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0)) |
12861 | 0 | input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0); |
12862 | 0 | else |
12863 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0); |
12864 | |
|
12865 | 0 | ret = xmlCtxtParseDocument(ctxt, input); |
12866 | |
|
12867 | 0 | xmlFreeParserCtxt(ctxt); |
12868 | 0 | return(ret); |
12869 | 0 | } |
12870 | | |
12871 | | /** |
12872 | | * xmlSAXParseFile: |
12873 | | * @sax: the SAX handler block |
12874 | | * @filename: the filename |
12875 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
12876 | | * documents |
12877 | | * |
12878 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
12879 | | * |
12880 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12881 | | * compressed document is provided by default if found at compile-time. |
12882 | | * It use the given SAX function block to handle the parsing callback. |
12883 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12884 | | * |
12885 | | * Returns the resulting document tree |
12886 | | */ |
12887 | | |
12888 | | xmlDocPtr |
12889 | | xmlSAXParseFile(xmlSAXHandlerPtr sax, const char *filename, |
12890 | 0 | int recovery) { |
12891 | 0 | return(xmlSAXParseFileWithData(sax,filename,recovery,NULL)); |
12892 | 0 | } |
12893 | | |
12894 | | /** |
12895 | | * xmlRecoverDoc: |
12896 | | * @cur: a pointer to an array of xmlChar |
12897 | | * |
12898 | | * DEPRECATED: Use xmlReadDoc with XML_PARSE_RECOVER. |
12899 | | * |
12900 | | * parse an XML in-memory document and build a tree. |
12901 | | * In the case the document is not Well Formed, a attempt to build a |
12902 | | * tree is tried anyway |
12903 | | * |
12904 | | * Returns the resulting document tree or NULL in case of failure |
12905 | | */ |
12906 | | |
12907 | | xmlDocPtr |
12908 | 0 | xmlRecoverDoc(const xmlChar *cur) { |
12909 | 0 | return(xmlSAXParseDoc(NULL, cur, 1)); |
12910 | 0 | } |
12911 | | |
12912 | | /** |
12913 | | * xmlParseFile: |
12914 | | * @filename: the filename |
12915 | | * |
12916 | | * DEPRECATED: Use xmlReadFile. |
12917 | | * |
12918 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12919 | | * compressed document is provided by default if found at compile-time. |
12920 | | * |
12921 | | * Returns the resulting document tree if the file was wellformed, |
12922 | | * NULL otherwise. |
12923 | | */ |
12924 | | |
12925 | | xmlDocPtr |
12926 | 0 | xmlParseFile(const char *filename) { |
12927 | 0 | return(xmlSAXParseFile(NULL, filename, 0)); |
12928 | 0 | } |
12929 | | |
12930 | | /** |
12931 | | * xmlRecoverFile: |
12932 | | * @filename: the filename |
12933 | | * |
12934 | | * DEPRECATED: Use xmlReadFile with XML_PARSE_RECOVER. |
12935 | | * |
12936 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12937 | | * compressed document is provided by default if found at compile-time. |
12938 | | * In the case the document is not Well Formed, it attempts to build |
12939 | | * a tree anyway |
12940 | | * |
12941 | | * Returns the resulting document tree or NULL in case of failure |
12942 | | */ |
12943 | | |
12944 | | xmlDocPtr |
12945 | 0 | xmlRecoverFile(const char *filename) { |
12946 | 0 | return(xmlSAXParseFile(NULL, filename, 1)); |
12947 | 0 | } |
12948 | | |
12949 | | |
12950 | | /** |
12951 | | * xmlSetupParserForBuffer: |
12952 | | * @ctxt: an XML parser context |
12953 | | * @buffer: a xmlChar * buffer |
12954 | | * @filename: a file name |
12955 | | * |
12956 | | * DEPRECATED: Don't use. |
12957 | | * |
12958 | | * Setup the parser context to parse a new buffer; Clears any prior |
12959 | | * contents from the parser context. The buffer parameter must not be |
12960 | | * NULL, but the filename parameter can be |
12961 | | */ |
12962 | | void |
12963 | | xmlSetupParserForBuffer(xmlParserCtxtPtr ctxt, const xmlChar* buffer, |
12964 | | const char* filename) |
12965 | 0 | { |
12966 | 0 | xmlParserInputPtr input; |
12967 | |
|
12968 | 0 | if ((ctxt == NULL) || (buffer == NULL)) |
12969 | 0 | return; |
12970 | | |
12971 | 0 | xmlClearParserCtxt(ctxt); |
12972 | |
|
12973 | 0 | input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer, |
12974 | 0 | NULL, 0); |
12975 | 0 | if (input == NULL) |
12976 | 0 | return; |
12977 | 0 | if (inputPush(ctxt, input) < 0) |
12978 | 0 | xmlFreeInputStream(input); |
12979 | 0 | } |
12980 | | |
12981 | | /** |
12982 | | * xmlSAXUserParseFile: |
12983 | | * @sax: a SAX handler |
12984 | | * @user_data: The user data returned on SAX callbacks |
12985 | | * @filename: a file name |
12986 | | * |
12987 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
12988 | | * |
12989 | | * parse an XML file and call the given SAX handler routines. |
12990 | | * Automatic support for ZLIB/Compress compressed document is provided |
12991 | | * |
12992 | | * Returns 0 in case of success or a error number otherwise |
12993 | | */ |
12994 | | int |
12995 | | xmlSAXUserParseFile(xmlSAXHandlerPtr sax, void *user_data, |
12996 | 0 | const char *filename) { |
12997 | 0 | int ret = 0; |
12998 | 0 | xmlParserCtxtPtr ctxt; |
12999 | |
|
13000 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
13001 | 0 | if (ctxt == NULL) return -1; |
13002 | 0 | if (sax != NULL) { |
13003 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
13004 | 0 | *ctxt->sax = *sax; |
13005 | 0 | } else { |
13006 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
13007 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
13008 | 0 | } |
13009 | 0 | ctxt->userData = user_data; |
13010 | 0 | } |
13011 | |
|
13012 | 0 | xmlParseDocument(ctxt); |
13013 | |
|
13014 | 0 | if (ctxt->wellFormed) |
13015 | 0 | ret = 0; |
13016 | 0 | else { |
13017 | 0 | if (ctxt->errNo != 0) |
13018 | 0 | ret = ctxt->errNo; |
13019 | 0 | else |
13020 | 0 | ret = -1; |
13021 | 0 | } |
13022 | 0 | if (ctxt->myDoc != NULL) { |
13023 | 0 | xmlFreeDoc(ctxt->myDoc); |
13024 | 0 | ctxt->myDoc = NULL; |
13025 | 0 | } |
13026 | 0 | xmlFreeParserCtxt(ctxt); |
13027 | |
|
13028 | 0 | return ret; |
13029 | 0 | } |
13030 | | #endif /* LIBXML_SAX1_ENABLED */ |
13031 | | |
13032 | | /************************************************************************ |
13033 | | * * |
13034 | | * Front ends when parsing from memory * |
13035 | | * * |
13036 | | ************************************************************************/ |
13037 | | |
13038 | | /** |
13039 | | * xmlCreateMemoryParserCtxt: |
13040 | | * @buffer: a pointer to a char array |
13041 | | * @size: the size of the array |
13042 | | * |
13043 | | * Create a parser context for an XML in-memory document. The input buffer |
13044 | | * must not contain a terminating null byte. |
13045 | | * |
13046 | | * Returns the new parser context or NULL |
13047 | | */ |
13048 | | xmlParserCtxtPtr |
13049 | 0 | xmlCreateMemoryParserCtxt(const char *buffer, int size) { |
13050 | 0 | xmlParserCtxtPtr ctxt; |
13051 | 0 | xmlParserInputPtr input; |
13052 | |
|
13053 | 0 | if (size < 0) |
13054 | 0 | return(NULL); |
13055 | | |
13056 | 0 | ctxt = xmlNewParserCtxt(); |
13057 | 0 | if (ctxt == NULL) |
13058 | 0 | return(NULL); |
13059 | | |
13060 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0); |
13061 | 0 | if (input == NULL) { |
13062 | 0 | xmlFreeParserCtxt(ctxt); |
13063 | 0 | return(NULL); |
13064 | 0 | } |
13065 | 0 | if (inputPush(ctxt, input) < 0) { |
13066 | 0 | xmlFreeInputStream(input); |
13067 | 0 | xmlFreeParserCtxt(ctxt); |
13068 | 0 | return(NULL); |
13069 | 0 | } |
13070 | | |
13071 | 0 | return(ctxt); |
13072 | 0 | } |
13073 | | |
13074 | | #ifdef LIBXML_SAX1_ENABLED |
13075 | | /** |
13076 | | * xmlSAXParseMemoryWithData: |
13077 | | * @sax: the SAX handler block |
13078 | | * @buffer: an pointer to a char array |
13079 | | * @size: the size of the array |
13080 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
13081 | | * documents |
13082 | | * @data: the userdata |
13083 | | * |
13084 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
13085 | | * |
13086 | | * parse an XML in-memory block and use the given SAX function block |
13087 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
13088 | | * DOM tree building routines. |
13089 | | * |
13090 | | * User data (void *) is stored within the parser context in the |
13091 | | * context's _private member, so it is available nearly everywhere in libxml |
13092 | | * |
13093 | | * Returns the resulting document tree |
13094 | | */ |
13095 | | |
13096 | | xmlDocPtr |
13097 | | xmlSAXParseMemoryWithData(xmlSAXHandlerPtr sax, const char *buffer, |
13098 | 0 | int size, int recovery, void *data) { |
13099 | 0 | xmlDocPtr ret; |
13100 | 0 | xmlParserCtxtPtr ctxt; |
13101 | 0 | xmlParserInputPtr input; |
13102 | |
|
13103 | 0 | if (size < 0) |
13104 | 0 | return(NULL); |
13105 | | |
13106 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
13107 | 0 | if (ctxt == NULL) |
13108 | 0 | return(NULL); |
13109 | | |
13110 | 0 | if (data != NULL) |
13111 | 0 | ctxt->_private=data; |
13112 | |
|
13113 | 0 | if (recovery) { |
13114 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
13115 | 0 | ctxt->recovery = 1; |
13116 | 0 | } |
13117 | |
|
13118 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, |
13119 | 0 | XML_INPUT_BUF_STATIC); |
13120 | |
|
13121 | 0 | ret = xmlCtxtParseDocument(ctxt, input); |
13122 | |
|
13123 | 0 | xmlFreeParserCtxt(ctxt); |
13124 | 0 | return(ret); |
13125 | 0 | } |
13126 | | |
13127 | | /** |
13128 | | * xmlSAXParseMemory: |
13129 | | * @sax: the SAX handler block |
13130 | | * @buffer: an pointer to a char array |
13131 | | * @size: the size of the array |
13132 | | * @recovery: work in recovery mode, i.e. tries to read not Well Formed |
13133 | | * documents |
13134 | | * |
13135 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
13136 | | * |
13137 | | * parse an XML in-memory block and use the given SAX function block |
13138 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
13139 | | * DOM tree building routines. |
13140 | | * |
13141 | | * Returns the resulting document tree |
13142 | | */ |
13143 | | xmlDocPtr |
13144 | | xmlSAXParseMemory(xmlSAXHandlerPtr sax, const char *buffer, |
13145 | 0 | int size, int recovery) { |
13146 | 0 | return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL); |
13147 | 0 | } |
13148 | | |
13149 | | /** |
13150 | | * xmlParseMemory: |
13151 | | * @buffer: an pointer to a char array |
13152 | | * @size: the size of the array |
13153 | | * |
13154 | | * DEPRECATED: Use xmlReadMemory. |
13155 | | * |
13156 | | * parse an XML in-memory block and build a tree. |
13157 | | * |
13158 | | * Returns the resulting document tree |
13159 | | */ |
13160 | | |
13161 | 0 | xmlDocPtr xmlParseMemory(const char *buffer, int size) { |
13162 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 0)); |
13163 | 0 | } |
13164 | | |
13165 | | /** |
13166 | | * xmlRecoverMemory: |
13167 | | * @buffer: an pointer to a char array |
13168 | | * @size: the size of the array |
13169 | | * |
13170 | | * DEPRECATED: Use xmlReadMemory with XML_PARSE_RECOVER. |
13171 | | * |
13172 | | * parse an XML in-memory block and build a tree. |
13173 | | * In the case the document is not Well Formed, an attempt to |
13174 | | * build a tree is tried anyway |
13175 | | * |
13176 | | * Returns the resulting document tree or NULL in case of error |
13177 | | */ |
13178 | | |
13179 | 0 | xmlDocPtr xmlRecoverMemory(const char *buffer, int size) { |
13180 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 1)); |
13181 | 0 | } |
13182 | | |
13183 | | /** |
13184 | | * xmlSAXUserParseMemory: |
13185 | | * @sax: a SAX handler |
13186 | | * @user_data: The user data returned on SAX callbacks |
13187 | | * @buffer: an in-memory XML document input |
13188 | | * @size: the length of the XML document in bytes |
13189 | | * |
13190 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
13191 | | * |
13192 | | * parse an XML in-memory buffer and call the given SAX handler routines. |
13193 | | * |
13194 | | * Returns 0 in case of success or a error number otherwise |
13195 | | */ |
13196 | | int xmlSAXUserParseMemory(xmlSAXHandlerPtr sax, void *user_data, |
13197 | 0 | const char *buffer, int size) { |
13198 | 0 | int ret = 0; |
13199 | 0 | xmlParserCtxtPtr ctxt; |
13200 | |
|
13201 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
13202 | 0 | if (ctxt == NULL) return -1; |
13203 | 0 | if (sax != NULL) { |
13204 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
13205 | 0 | *ctxt->sax = *sax; |
13206 | 0 | } else { |
13207 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
13208 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
13209 | 0 | } |
13210 | 0 | ctxt->userData = user_data; |
13211 | 0 | } |
13212 | |
|
13213 | 0 | xmlParseDocument(ctxt); |
13214 | |
|
13215 | 0 | if (ctxt->wellFormed) |
13216 | 0 | ret = 0; |
13217 | 0 | else { |
13218 | 0 | if (ctxt->errNo != 0) |
13219 | 0 | ret = ctxt->errNo; |
13220 | 0 | else |
13221 | 0 | ret = -1; |
13222 | 0 | } |
13223 | 0 | if (ctxt->myDoc != NULL) { |
13224 | 0 | xmlFreeDoc(ctxt->myDoc); |
13225 | 0 | ctxt->myDoc = NULL; |
13226 | 0 | } |
13227 | 0 | xmlFreeParserCtxt(ctxt); |
13228 | |
|
13229 | 0 | return ret; |
13230 | 0 | } |
13231 | | #endif /* LIBXML_SAX1_ENABLED */ |
13232 | | |
13233 | | /** |
13234 | | * xmlCreateDocParserCtxt: |
13235 | | * @str: a pointer to an array of xmlChar |
13236 | | * |
13237 | | * Creates a parser context for an XML in-memory document. |
13238 | | * |
13239 | | * Returns the new parser context or NULL |
13240 | | */ |
13241 | | xmlParserCtxtPtr |
13242 | 0 | xmlCreateDocParserCtxt(const xmlChar *str) { |
13243 | 0 | xmlParserCtxtPtr ctxt; |
13244 | 0 | xmlParserInputPtr input; |
13245 | |
|
13246 | 0 | ctxt = xmlNewParserCtxt(); |
13247 | 0 | if (ctxt == NULL) |
13248 | 0 | return(NULL); |
13249 | | |
13250 | 0 | input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0); |
13251 | 0 | if (input == NULL) { |
13252 | 0 | xmlFreeParserCtxt(ctxt); |
13253 | 0 | return(NULL); |
13254 | 0 | } |
13255 | 0 | if (inputPush(ctxt, input) < 0) { |
13256 | 0 | xmlFreeInputStream(input); |
13257 | 0 | xmlFreeParserCtxt(ctxt); |
13258 | 0 | return(NULL); |
13259 | 0 | } |
13260 | | |
13261 | 0 | return(ctxt); |
13262 | 0 | } |
13263 | | |
13264 | | #ifdef LIBXML_SAX1_ENABLED |
13265 | | /** |
13266 | | * xmlSAXParseDoc: |
13267 | | * @sax: the SAX handler block |
13268 | | * @cur: a pointer to an array of xmlChar |
13269 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
13270 | | * documents |
13271 | | * |
13272 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadDoc. |
13273 | | * |
13274 | | * parse an XML in-memory document and build a tree. |
13275 | | * It use the given SAX function block to handle the parsing callback. |
13276 | | * If sax is NULL, fallback to the default DOM tree building routines. |
13277 | | * |
13278 | | * Returns the resulting document tree |
13279 | | */ |
13280 | | |
13281 | | xmlDocPtr |
13282 | 0 | xmlSAXParseDoc(xmlSAXHandlerPtr sax, const xmlChar *cur, int recovery) { |
13283 | 0 | xmlDocPtr ret; |
13284 | 0 | xmlParserCtxtPtr ctxt; |
13285 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
13286 | |
|
13287 | 0 | if (cur == NULL) return(NULL); |
13288 | | |
13289 | | |
13290 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
13291 | 0 | if (ctxt == NULL) return(NULL); |
13292 | 0 | if (sax != NULL) { |
13293 | 0 | oldsax = ctxt->sax; |
13294 | 0 | ctxt->sax = sax; |
13295 | 0 | ctxt->userData = NULL; |
13296 | 0 | } |
13297 | |
|
13298 | 0 | xmlParseDocument(ctxt); |
13299 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
13300 | 0 | else { |
13301 | 0 | ret = NULL; |
13302 | 0 | xmlFreeDoc(ctxt->myDoc); |
13303 | 0 | ctxt->myDoc = NULL; |
13304 | 0 | } |
13305 | 0 | if (sax != NULL) |
13306 | 0 | ctxt->sax = oldsax; |
13307 | 0 | xmlFreeParserCtxt(ctxt); |
13308 | |
|
13309 | 0 | return(ret); |
13310 | 0 | } |
13311 | | |
13312 | | /** |
13313 | | * xmlParseDoc: |
13314 | | * @cur: a pointer to an array of xmlChar |
13315 | | * |
13316 | | * DEPRECATED: Use xmlReadDoc. |
13317 | | * |
13318 | | * parse an XML in-memory document and build a tree. |
13319 | | * |
13320 | | * Returns the resulting document tree |
13321 | | */ |
13322 | | |
13323 | | xmlDocPtr |
13324 | 0 | xmlParseDoc(const xmlChar *cur) { |
13325 | 0 | return(xmlSAXParseDoc(NULL, cur, 0)); |
13326 | 0 | } |
13327 | | #endif /* LIBXML_SAX1_ENABLED */ |
13328 | | |
13329 | | /************************************************************************ |
13330 | | * * |
13331 | | * New set (2.6.0) of simpler and more flexible APIs * |
13332 | | * * |
13333 | | ************************************************************************/ |
13334 | | |
13335 | | /** |
13336 | | * DICT_FREE: |
13337 | | * @str: a string |
13338 | | * |
13339 | | * Free a string if it is not owned by the "dict" dictionary in the |
13340 | | * current scope |
13341 | | */ |
13342 | | #define DICT_FREE(str) \ |
13343 | 487k | if ((str) && ((!dict) || \ |
13344 | 0 | (xmlDictOwns(dict, (const xmlChar *)(str)) == 0))) \ |
13345 | 487k | xmlFree((char *)(str)); |
13346 | | |
13347 | | /** |
13348 | | * xmlCtxtReset: |
13349 | | * @ctxt: an XML parser context |
13350 | | * |
13351 | | * Reset a parser context |
13352 | | */ |
13353 | | void |
13354 | | xmlCtxtReset(xmlParserCtxtPtr ctxt) |
13355 | 121k | { |
13356 | 121k | xmlParserInputPtr input; |
13357 | 121k | xmlDictPtr dict; |
13358 | | |
13359 | 121k | if (ctxt == NULL) |
13360 | 0 | return; |
13361 | | |
13362 | 121k | dict = ctxt->dict; |
13363 | | |
13364 | 121k | while ((input = inputPop(ctxt)) != NULL) { /* Non consuming */ |
13365 | 0 | xmlFreeInputStream(input); |
13366 | 0 | } |
13367 | 121k | ctxt->inputNr = 0; |
13368 | 121k | ctxt->input = NULL; |
13369 | | |
13370 | 121k | ctxt->spaceNr = 0; |
13371 | 121k | if (ctxt->spaceTab != NULL) { |
13372 | 121k | ctxt->spaceTab[0] = -1; |
13373 | 121k | ctxt->space = &ctxt->spaceTab[0]; |
13374 | 121k | } else { |
13375 | 0 | ctxt->space = NULL; |
13376 | 0 | } |
13377 | | |
13378 | | |
13379 | 121k | ctxt->nodeNr = 0; |
13380 | 121k | ctxt->node = NULL; |
13381 | | |
13382 | 121k | ctxt->nameNr = 0; |
13383 | 121k | ctxt->name = NULL; |
13384 | | |
13385 | 121k | ctxt->nsNr = 0; |
13386 | 121k | xmlParserNsReset(ctxt->nsdb); |
13387 | | |
13388 | 121k | DICT_FREE(ctxt->version); |
13389 | 121k | ctxt->version = NULL; |
13390 | 121k | DICT_FREE(ctxt->encoding); |
13391 | 121k | ctxt->encoding = NULL; |
13392 | 121k | DICT_FREE(ctxt->extSubURI); |
13393 | 121k | ctxt->extSubURI = NULL; |
13394 | 121k | DICT_FREE(ctxt->extSubSystem); |
13395 | 121k | ctxt->extSubSystem = NULL; |
13396 | | |
13397 | 121k | if (ctxt->directory != NULL) { |
13398 | 0 | xmlFree(ctxt->directory); |
13399 | 0 | ctxt->directory = NULL; |
13400 | 0 | } |
13401 | | |
13402 | 121k | if (ctxt->myDoc != NULL) |
13403 | 0 | xmlFreeDoc(ctxt->myDoc); |
13404 | 121k | ctxt->myDoc = NULL; |
13405 | | |
13406 | 121k | ctxt->standalone = -1; |
13407 | 121k | ctxt->hasExternalSubset = 0; |
13408 | 121k | ctxt->hasPErefs = 0; |
13409 | 121k | ctxt->html = 0; |
13410 | 121k | ctxt->instate = XML_PARSER_START; |
13411 | | |
13412 | 121k | ctxt->wellFormed = 1; |
13413 | 121k | ctxt->nsWellFormed = 1; |
13414 | 121k | ctxt->disableSAX = 0; |
13415 | 121k | ctxt->valid = 1; |
13416 | 121k | ctxt->record_info = 0; |
13417 | 121k | ctxt->checkIndex = 0; |
13418 | 121k | ctxt->endCheckState = 0; |
13419 | 121k | ctxt->inSubset = 0; |
13420 | 121k | ctxt->errNo = XML_ERR_OK; |
13421 | 121k | ctxt->depth = 0; |
13422 | 121k | ctxt->catalogs = NULL; |
13423 | 121k | ctxt->sizeentities = 0; |
13424 | 121k | ctxt->sizeentcopy = 0; |
13425 | 121k | xmlInitNodeInfoSeq(&ctxt->node_seq); |
13426 | | |
13427 | 121k | if (ctxt->attsDefault != NULL) { |
13428 | 0 | xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator); |
13429 | 0 | ctxt->attsDefault = NULL; |
13430 | 0 | } |
13431 | 121k | if (ctxt->attsSpecial != NULL) { |
13432 | 0 | xmlHashFree(ctxt->attsSpecial, NULL); |
13433 | 0 | ctxt->attsSpecial = NULL; |
13434 | 0 | } |
13435 | | |
13436 | 121k | #ifdef LIBXML_CATALOG_ENABLED |
13437 | 121k | if (ctxt->catalogs != NULL) |
13438 | 0 | xmlCatalogFreeLocal(ctxt->catalogs); |
13439 | 121k | #endif |
13440 | 121k | ctxt->nbErrors = 0; |
13441 | 121k | ctxt->nbWarnings = 0; |
13442 | 121k | if (ctxt->lastError.code != XML_ERR_OK) |
13443 | 0 | xmlResetError(&ctxt->lastError); |
13444 | 121k | } |
13445 | | |
13446 | | /** |
13447 | | * xmlCtxtResetPush: |
13448 | | * @ctxt: an XML parser context |
13449 | | * @chunk: a pointer to an array of chars |
13450 | | * @size: number of chars in the array |
13451 | | * @filename: an optional file name or URI |
13452 | | * @encoding: the document encoding, or NULL |
13453 | | * |
13454 | | * Reset a push parser context |
13455 | | * |
13456 | | * Returns 0 in case of success and 1 in case of error |
13457 | | */ |
13458 | | int |
13459 | | xmlCtxtResetPush(xmlParserCtxtPtr ctxt, const char *chunk, |
13460 | | int size, const char *filename, const char *encoding) |
13461 | 0 | { |
13462 | 0 | xmlParserInputPtr input; |
13463 | |
|
13464 | 0 | if (ctxt == NULL) |
13465 | 0 | return(1); |
13466 | | |
13467 | 0 | xmlCtxtReset(ctxt); |
13468 | |
|
13469 | 0 | input = xmlNewPushInput(filename, chunk, size); |
13470 | 0 | if (input == NULL) |
13471 | 0 | return(1); |
13472 | | |
13473 | 0 | if (inputPush(ctxt, input) < 0) { |
13474 | 0 | xmlFreeInputStream(input); |
13475 | 0 | return(1); |
13476 | 0 | } |
13477 | | |
13478 | 0 | if (encoding != NULL) |
13479 | 0 | xmlSwitchEncodingName(ctxt, encoding); |
13480 | |
|
13481 | 0 | return(0); |
13482 | 0 | } |
13483 | | |
13484 | | static int |
13485 | | xmlCtxtSetOptionsInternal(xmlParserCtxtPtr ctxt, int options, int keepMask) |
13486 | 496k | { |
13487 | 496k | int allMask; |
13488 | | |
13489 | 496k | if (ctxt == NULL) |
13490 | 0 | return(-1); |
13491 | | |
13492 | | /* |
13493 | | * XInclude options aren't handled by the parser. |
13494 | | * |
13495 | | * XML_PARSE_XINCLUDE |
13496 | | * XML_PARSE_NOXINCNODE |
13497 | | * XML_PARSE_NOBASEFIX |
13498 | | */ |
13499 | 496k | allMask = XML_PARSE_RECOVER | |
13500 | 496k | XML_PARSE_NOENT | |
13501 | 496k | XML_PARSE_DTDLOAD | |
13502 | 496k | XML_PARSE_DTDATTR | |
13503 | 496k | XML_PARSE_DTDVALID | |
13504 | 496k | XML_PARSE_NOERROR | |
13505 | 496k | XML_PARSE_NOWARNING | |
13506 | 496k | XML_PARSE_PEDANTIC | |
13507 | 496k | XML_PARSE_NOBLANKS | |
13508 | 496k | #ifdef LIBXML_SAX1_ENABLED |
13509 | 496k | XML_PARSE_SAX1 | |
13510 | 496k | #endif |
13511 | 496k | XML_PARSE_NONET | |
13512 | 496k | XML_PARSE_NODICT | |
13513 | 496k | XML_PARSE_NSCLEAN | |
13514 | 496k | XML_PARSE_NOCDATA | |
13515 | 496k | XML_PARSE_COMPACT | |
13516 | 496k | XML_PARSE_OLD10 | |
13517 | 496k | XML_PARSE_HUGE | |
13518 | 496k | XML_PARSE_OLDSAX | |
13519 | 496k | XML_PARSE_IGNORE_ENC | |
13520 | 496k | XML_PARSE_BIG_LINES | |
13521 | 496k | XML_PARSE_NO_XXE; |
13522 | | |
13523 | 496k | ctxt->options = (ctxt->options & keepMask) | (options & allMask); |
13524 | | |
13525 | | /* |
13526 | | * For some options, struct members are historically the source |
13527 | | * of truth. The values are initalized from global variables and |
13528 | | * old code could also modify them directly. Several older API |
13529 | | * functions that don't take an options argument rely on these |
13530 | | * deprecated mechanisms. |
13531 | | * |
13532 | | * Once public access to struct members and the globals are |
13533 | | * disabled, we can use the options bitmask as source of |
13534 | | * truth, making all these struct members obsolete. |
13535 | | * |
13536 | | * The XML_DETECT_IDS flags is misnamed. It simply enables |
13537 | | * loading of the external subset. |
13538 | | */ |
13539 | 496k | ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0; |
13540 | 496k | ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0; |
13541 | 496k | ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0; |
13542 | 496k | ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0; |
13543 | 496k | ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0; |
13544 | 496k | ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0; |
13545 | 496k | ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1; |
13546 | 496k | ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1; |
13547 | | |
13548 | | /* |
13549 | | * Changing SAX callbacks is a bad idea. This should be fixed. |
13550 | | */ |
13551 | 496k | if (options & XML_PARSE_NOBLANKS) { |
13552 | 168k | ctxt->sax->ignorableWhitespace = xmlSAX2IgnorableWhitespace; |
13553 | 168k | } |
13554 | 496k | if (options & XML_PARSE_NOCDATA) { |
13555 | 143k | ctxt->sax->cdataBlock = NULL; |
13556 | 143k | } |
13557 | 496k | if (options & XML_PARSE_HUGE) { |
13558 | 140k | if (ctxt->dict != NULL) |
13559 | 140k | xmlDictSetLimit(ctxt->dict, 0); |
13560 | 140k | } |
13561 | | |
13562 | 496k | ctxt->linenumbers = 1; |
13563 | | |
13564 | 496k | return(options & ~allMask); |
13565 | 496k | } |
13566 | | |
13567 | | /** |
13568 | | * xmlCtxtSetOptions: |
13569 | | * @ctxt: an XML parser context |
13570 | | * @options: a bitmask of xmlParserOption values |
13571 | | * |
13572 | | * Applies the options to the parser context. Unset options are |
13573 | | * cleared. |
13574 | | * |
13575 | | * Available since 2.13.0. With older versions, you can use |
13576 | | * xmlCtxtUseOptions. |
13577 | | * |
13578 | | * XML_PARSE_RECOVER |
13579 | | * |
13580 | | * Enable "recovery" mode which allows non-wellformed documents. |
13581 | | * How this mode behaves exactly is unspecified and may change |
13582 | | * without further notice. Use of this feature is DISCOURAGED. |
13583 | | * |
13584 | | * XML_PARSE_NOENT |
13585 | | * |
13586 | | * Despite the confusing name, this option enables substitution |
13587 | | * of entities. The resulting tree won't contain any entity |
13588 | | * reference nodes. |
13589 | | * |
13590 | | * This option also enables loading of external entities (both |
13591 | | * general and parameter entities) which is dangerous. If you |
13592 | | * process untrusted data, it's recommended to set the |
13593 | | * XML_PARSE_NO_XXE option to disable loading of external |
13594 | | * entities. |
13595 | | * |
13596 | | * XML_PARSE_DTDLOAD |
13597 | | * |
13598 | | * Enables loading of an external DTD and the loading and |
13599 | | * substitution of external parameter entities. Has no effect |
13600 | | * if XML_PARSE_NO_XXE is set. |
13601 | | * |
13602 | | * XML_PARSE_DTDATTR |
13603 | | * |
13604 | | * Adds default attributes from the DTD to the result document. |
13605 | | * |
13606 | | * Implies XML_PARSE_DTDLOAD, but loading of external content |
13607 | | * can be disabled with XML_PARSE_NO_XXE. |
13608 | | * |
13609 | | * XML_PARSE_DTDVALID |
13610 | | * |
13611 | | * This option enables DTD validation which requires to load |
13612 | | * external DTDs and external entities (both general and |
13613 | | * parameter entities) unless XML_PARSE_NO_XXE was set. |
13614 | | * |
13615 | | * XML_PARSE_NO_XXE |
13616 | | * |
13617 | | * Disables loading of external DTDs or entities. |
13618 | | * |
13619 | | * XML_PARSE_NOERROR |
13620 | | * |
13621 | | * Disable error and warning reports to the error handlers. |
13622 | | * Errors are still accessible with xmlCtxtGetLastError. |
13623 | | * |
13624 | | * XML_PARSE_NOWARNING |
13625 | | * |
13626 | | * Disable warning reports. |
13627 | | * |
13628 | | * XML_PARSE_PEDANTIC |
13629 | | * |
13630 | | * Enable some pedantic warnings. |
13631 | | * |
13632 | | * XML_PARSE_NOBLANKS |
13633 | | * |
13634 | | * Remove some text nodes containing only whitespace from the |
13635 | | * result document. Which nodes are removed depends on DTD |
13636 | | * element declarations or a conservative heuristic. The |
13637 | | * reindenting feature of the serialization code relies on this |
13638 | | * option to be set when parsing. Use of this option is |
13639 | | * DISCOURAGED. |
13640 | | * |
13641 | | * XML_PARSE_SAX1 |
13642 | | * |
13643 | | * Always invoke the deprecated SAX1 startElement and endElement |
13644 | | * handlers. This option is DEPRECATED. |
13645 | | * |
13646 | | * XML_PARSE_NONET |
13647 | | * |
13648 | | * Disable network access with the builtin HTTP client. |
13649 | | * |
13650 | | * XML_PARSE_NODICT |
13651 | | * |
13652 | | * Create a document without interned strings, making all |
13653 | | * strings separate memory allocations. |
13654 | | * |
13655 | | * XML_PARSE_NSCLEAN |
13656 | | * |
13657 | | * Remove redundant namespace declarations from the result |
13658 | | * document. |
13659 | | * |
13660 | | * XML_PARSE_NOCDATA |
13661 | | * |
13662 | | * Output normal text nodes instead of CDATA nodes. |
13663 | | * |
13664 | | * XML_PARSE_COMPACT |
13665 | | * |
13666 | | * Store small strings directly in the node struct to save |
13667 | | * memory. |
13668 | | * |
13669 | | * XML_PARSE_OLD10 |
13670 | | * |
13671 | | * Use old Name productions from before XML 1.0 Fifth Edition. |
13672 | | * This options is DEPRECATED. |
13673 | | * |
13674 | | * XML_PARSE_HUGE |
13675 | | * |
13676 | | * Relax some internal limits. |
13677 | | * |
13678 | | * Maximum size of text nodes, tags, comments, processing instructions, |
13679 | | * CDATA sections, entity values |
13680 | | * |
13681 | | * normal: 10M |
13682 | | * huge: 1B |
13683 | | * |
13684 | | * Maximum size of names, system literals, pubid literals |
13685 | | * |
13686 | | * normal: 50K |
13687 | | * huge: 10M |
13688 | | * |
13689 | | * Maximum nesting depth of elements |
13690 | | * |
13691 | | * normal: 256 |
13692 | | * huge: 2048 |
13693 | | * |
13694 | | * Maximum nesting depth of entities |
13695 | | * |
13696 | | * normal: 20 |
13697 | | * huge: 40 |
13698 | | * |
13699 | | * XML_PARSE_OLDSAX |
13700 | | * |
13701 | | * Enable an unspecified legacy mode for SAX parsers. This |
13702 | | * option is DEPRECATED. |
13703 | | * |
13704 | | * XML_PARSE_IGNORE_ENC |
13705 | | * |
13706 | | * Ignore the encoding in the XML declaration. This option is |
13707 | | * mostly unneeded these days. The only effect is to enforce |
13708 | | * UTF-8 decoding of ASCII-like data. |
13709 | | * |
13710 | | * XML_PARSE_BIG_LINES |
13711 | | * |
13712 | | * Enable reporting of line numbers larger than 65535. |
13713 | | * |
13714 | | * XML_PARSE_NO_UNZIP |
13715 | | * |
13716 | | * Disables input decompression. Setting this option is recommended |
13717 | | * to avoid zip bombs. |
13718 | | * |
13719 | | * Available since 2.14.0. |
13720 | | * |
13721 | | * XML_PARSE_NO_SYS_CATALOG |
13722 | | * |
13723 | | * Disables the global system XML catalog. |
13724 | | * |
13725 | | * Available since 2.14.0. |
13726 | | * |
13727 | | * XML_PARSE_NO_CATALOG_PI |
13728 | | * |
13729 | | * Ignore XML catalog processing instructions. |
13730 | | * |
13731 | | * Available since 2.14.0. |
13732 | | * |
13733 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
13734 | | * in case of error. |
13735 | | */ |
13736 | | int |
13737 | | xmlCtxtSetOptions(xmlParserCtxtPtr ctxt, int options) |
13738 | 0 | { |
13739 | 0 | return(xmlCtxtSetOptionsInternal(ctxt, options, 0)); |
13740 | 0 | } |
13741 | | |
13742 | | /** |
13743 | | * xmlCtxtGetOptions: |
13744 | | * @ctxt: an XML parser context |
13745 | | * |
13746 | | * Get the current options of the parser context. |
13747 | | * |
13748 | | * Available since 2.14.0. |
13749 | | * |
13750 | | * Returns the current options set in the parser context, or -1 if ctxt is NULL. |
13751 | | */ |
13752 | | int |
13753 | | xmlCtxtGetOptions(xmlParserCtxtPtr ctxt) |
13754 | 0 | { |
13755 | 0 | if (ctxt == NULL) |
13756 | 0 | return(-1); |
13757 | | |
13758 | 0 | return(ctxt->options); |
13759 | 0 | } |
13760 | | |
13761 | | /** |
13762 | | * xmlCtxtUseOptions: |
13763 | | * @ctxt: an XML parser context |
13764 | | * @options: a combination of xmlParserOption |
13765 | | * |
13766 | | * DEPRECATED: Use xmlCtxtSetOptions. |
13767 | | * |
13768 | | * Applies the options to the parser context. The following options |
13769 | | * are never cleared and can only be enabled: |
13770 | | * |
13771 | | * XML_PARSE_NOERROR |
13772 | | * XML_PARSE_NOWARNING |
13773 | | * XML_PARSE_NONET |
13774 | | * XML_PARSE_NSCLEAN |
13775 | | * XML_PARSE_NOCDATA |
13776 | | * XML_PARSE_COMPACT |
13777 | | * XML_PARSE_OLD10 |
13778 | | * XML_PARSE_HUGE |
13779 | | * XML_PARSE_OLDSAX |
13780 | | * XML_PARSE_IGNORE_ENC |
13781 | | * XML_PARSE_BIG_LINES |
13782 | | * |
13783 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
13784 | | * in case of error. |
13785 | | */ |
13786 | | int |
13787 | | xmlCtxtUseOptions(xmlParserCtxtPtr ctxt, int options) |
13788 | 496k | { |
13789 | 496k | int keepMask; |
13790 | | |
13791 | | /* |
13792 | | * For historic reasons, some options can only be enabled. |
13793 | | */ |
13794 | 496k | keepMask = XML_PARSE_NOERROR | |
13795 | 496k | XML_PARSE_NOWARNING | |
13796 | 496k | XML_PARSE_NONET | |
13797 | 496k | XML_PARSE_NSCLEAN | |
13798 | 496k | XML_PARSE_NOCDATA | |
13799 | 496k | XML_PARSE_COMPACT | |
13800 | 496k | XML_PARSE_OLD10 | |
13801 | 496k | XML_PARSE_HUGE | |
13802 | 496k | XML_PARSE_OLDSAX | |
13803 | 496k | XML_PARSE_IGNORE_ENC | |
13804 | 496k | XML_PARSE_BIG_LINES; |
13805 | | |
13806 | 496k | return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask)); |
13807 | 496k | } |
13808 | | |
13809 | | /** |
13810 | | * xmlCtxtSetMaxAmplification: |
13811 | | * @ctxt: an XML parser context |
13812 | | * @maxAmpl: maximum amplification factor |
13813 | | * |
13814 | | * To protect against exponential entity expansion ("billion laughs"), the |
13815 | | * size of serialized output is (roughly) limited to the input size |
13816 | | * multiplied by this factor. The default value is 5. |
13817 | | * |
13818 | | * When working with documents making heavy use of entity expansion, it can |
13819 | | * be necessary to increase the value. For security reasons, this should only |
13820 | | * be considered when processing trusted input. |
13821 | | */ |
13822 | | void |
13823 | | xmlCtxtSetMaxAmplification(xmlParserCtxtPtr ctxt, unsigned maxAmpl) |
13824 | 244 | { |
13825 | 244 | ctxt->maxAmpl = maxAmpl; |
13826 | 244 | } |
13827 | | |
13828 | | /** |
13829 | | * xmlCtxtParseDocument: |
13830 | | * @ctxt: an XML parser context |
13831 | | * @input: parser input |
13832 | | * |
13833 | | * Parse an XML document and return the resulting document tree. |
13834 | | * Takes ownership of the input object. |
13835 | | * |
13836 | | * Available since 2.13.0. |
13837 | | * |
13838 | | * Returns the resulting document tree or NULL |
13839 | | */ |
13840 | | xmlDocPtr |
13841 | | xmlCtxtParseDocument(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) |
13842 | 422k | { |
13843 | 422k | xmlDocPtr ret = NULL; |
13844 | | |
13845 | 422k | if ((ctxt == NULL) || (input == NULL)) |
13846 | 21.0k | return(NULL); |
13847 | | |
13848 | | /* assert(ctxt->inputNr == 0); */ |
13849 | 401k | while (ctxt->inputNr > 0) |
13850 | 0 | xmlFreeInputStream(inputPop(ctxt)); |
13851 | | |
13852 | 401k | if (inputPush(ctxt, input) < 0) { |
13853 | 32 | xmlFreeInputStream(input); |
13854 | 32 | return(NULL); |
13855 | 32 | } |
13856 | | |
13857 | 401k | xmlParseDocument(ctxt); |
13858 | | |
13859 | 401k | if ((ctxt->wellFormed) || |
13860 | 401k | ((ctxt->recovery) && (ctxt->errNo != XML_ERR_NO_MEMORY))) { |
13861 | 238k | ret = ctxt->myDoc; |
13862 | 238k | } else { |
13863 | 162k | if (ctxt->errNo == XML_ERR_OK) |
13864 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, "unknown error\n"); |
13865 | | |
13866 | 162k | ret = NULL; |
13867 | 162k | xmlFreeDoc(ctxt->myDoc); |
13868 | 162k | } |
13869 | 401k | ctxt->myDoc = NULL; |
13870 | | |
13871 | | /* assert(ctxt->inputNr == 1); */ |
13872 | 802k | while (ctxt->inputNr > 0) |
13873 | 401k | xmlFreeInputStream(inputPop(ctxt)); |
13874 | | |
13875 | 401k | return(ret); |
13876 | 401k | } |
13877 | | |
13878 | | /** |
13879 | | * xmlReadDoc: |
13880 | | * @cur: a pointer to a zero terminated string |
13881 | | * @URL: base URL (optional) |
13882 | | * @encoding: the document encoding (optional) |
13883 | | * @options: a combination of xmlParserOption |
13884 | | * |
13885 | | * Convenience function to parse an XML document from a |
13886 | | * zero-terminated string. |
13887 | | * |
13888 | | * See xmlCtxtReadDoc for details. |
13889 | | * |
13890 | | * Returns the resulting document tree |
13891 | | */ |
13892 | | xmlDocPtr |
13893 | | xmlReadDoc(const xmlChar *cur, const char *URL, const char *encoding, |
13894 | | int options) |
13895 | 276k | { |
13896 | 276k | xmlParserCtxtPtr ctxt; |
13897 | 276k | xmlParserInputPtr input; |
13898 | 276k | xmlDocPtr doc; |
13899 | | |
13900 | 276k | ctxt = xmlNewParserCtxt(); |
13901 | 276k | if (ctxt == NULL) |
13902 | 294 | return(NULL); |
13903 | | |
13904 | 276k | xmlCtxtUseOptions(ctxt, options); |
13905 | | |
13906 | 276k | input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding, |
13907 | 276k | XML_INPUT_BUF_STATIC); |
13908 | | |
13909 | 276k | doc = xmlCtxtParseDocument(ctxt, input); |
13910 | | |
13911 | 276k | xmlFreeParserCtxt(ctxt); |
13912 | 276k | return(doc); |
13913 | 276k | } |
13914 | | |
13915 | | /** |
13916 | | * xmlReadFile: |
13917 | | * @filename: a file or URL |
13918 | | * @encoding: the document encoding (optional) |
13919 | | * @options: a combination of xmlParserOption |
13920 | | * |
13921 | | * Convenience function to parse an XML file from the filesystem, |
13922 | | * the network or a global user-define resource loader. |
13923 | | * |
13924 | | * See xmlCtxtReadFile for details. |
13925 | | * |
13926 | | * Returns the resulting document tree |
13927 | | */ |
13928 | | xmlDocPtr |
13929 | | xmlReadFile(const char *filename, const char *encoding, int options) |
13930 | 0 | { |
13931 | 0 | xmlParserCtxtPtr ctxt; |
13932 | 0 | xmlParserInputPtr input; |
13933 | 0 | xmlDocPtr doc; |
13934 | |
|
13935 | 0 | ctxt = xmlNewParserCtxt(); |
13936 | 0 | if (ctxt == NULL) |
13937 | 0 | return(NULL); |
13938 | | |
13939 | 0 | xmlCtxtUseOptions(ctxt, options); |
13940 | | |
13941 | | /* |
13942 | | * Backward compatibility for users of command line utilities like |
13943 | | * xmlstarlet expecting "-" to mean stdin. This is dangerous and |
13944 | | * should be removed at some point. |
13945 | | */ |
13946 | 0 | if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0)) |
13947 | 0 | input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, |
13948 | 0 | encoding, 0); |
13949 | 0 | else |
13950 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0); |
13951 | |
|
13952 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13953 | |
|
13954 | 0 | xmlFreeParserCtxt(ctxt); |
13955 | 0 | return(doc); |
13956 | 0 | } |
13957 | | |
13958 | | /** |
13959 | | * xmlReadMemory: |
13960 | | * @buffer: a pointer to a char array |
13961 | | * @size: the size of the array |
13962 | | * @url: base URL (optional) |
13963 | | * @encoding: the document encoding (optional) |
13964 | | * @options: a combination of xmlParserOption |
13965 | | * |
13966 | | * Parse an XML in-memory document and build a tree. The input buffer must |
13967 | | * not contain a terminating null byte. |
13968 | | * |
13969 | | * See xmlCtxtReadMemory for details. |
13970 | | * |
13971 | | * Returns the resulting document tree |
13972 | | */ |
13973 | | xmlDocPtr |
13974 | | xmlReadMemory(const char *buffer, int size, const char *url, |
13975 | | const char *encoding, int options) |
13976 | 23.8k | { |
13977 | 23.8k | xmlParserCtxtPtr ctxt; |
13978 | 23.8k | xmlParserInputPtr input; |
13979 | 23.8k | xmlDocPtr doc; |
13980 | | |
13981 | 23.8k | if (size < 0) |
13982 | 0 | return(NULL); |
13983 | | |
13984 | 23.8k | ctxt = xmlNewParserCtxt(); |
13985 | 23.8k | if (ctxt == NULL) |
13986 | 0 | return(NULL); |
13987 | | |
13988 | 23.8k | xmlCtxtUseOptions(ctxt, options); |
13989 | | |
13990 | 23.8k | input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding, |
13991 | 23.8k | XML_INPUT_BUF_STATIC); |
13992 | | |
13993 | 23.8k | doc = xmlCtxtParseDocument(ctxt, input); |
13994 | | |
13995 | 23.8k | xmlFreeParserCtxt(ctxt); |
13996 | 23.8k | return(doc); |
13997 | 23.8k | } |
13998 | | |
13999 | | /** |
14000 | | * xmlReadFd: |
14001 | | * @fd: an open file descriptor |
14002 | | * @URL: base URL (optional) |
14003 | | * @encoding: the document encoding (optional) |
14004 | | * @options: a combination of xmlParserOption |
14005 | | * |
14006 | | * Parse an XML from a file descriptor and build a tree. |
14007 | | * |
14008 | | * See xmlCtxtReadFd for details. |
14009 | | * |
14010 | | * NOTE that the file descriptor will not be closed when the |
14011 | | * context is freed or reset. |
14012 | | * |
14013 | | * Returns the resulting document tree |
14014 | | */ |
14015 | | xmlDocPtr |
14016 | | xmlReadFd(int fd, const char *URL, const char *encoding, int options) |
14017 | 0 | { |
14018 | 0 | xmlParserCtxtPtr ctxt; |
14019 | 0 | xmlParserInputPtr input; |
14020 | 0 | xmlDocPtr doc; |
14021 | |
|
14022 | 0 | ctxt = xmlNewParserCtxt(); |
14023 | 0 | if (ctxt == NULL) |
14024 | 0 | return(NULL); |
14025 | | |
14026 | 0 | xmlCtxtUseOptions(ctxt, options); |
14027 | |
|
14028 | 0 | input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0); |
14029 | |
|
14030 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
14031 | |
|
14032 | 0 | xmlFreeParserCtxt(ctxt); |
14033 | 0 | return(doc); |
14034 | 0 | } |
14035 | | |
14036 | | /** |
14037 | | * xmlReadIO: |
14038 | | * @ioread: an I/O read function |
14039 | | * @ioclose: an I/O close function (optional) |
14040 | | * @ioctx: an I/O handler |
14041 | | * @URL: base URL (optional) |
14042 | | * @encoding: the document encoding (optional) |
14043 | | * @options: a combination of xmlParserOption |
14044 | | * |
14045 | | * Parse an XML document from I/O functions and context and build a tree. |
14046 | | * |
14047 | | * See xmlCtxtReadIO for details. |
14048 | | * |
14049 | | * Returns the resulting document tree |
14050 | | */ |
14051 | | xmlDocPtr |
14052 | | xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
14053 | | void *ioctx, const char *URL, const char *encoding, int options) |
14054 | 0 | { |
14055 | 0 | xmlParserCtxtPtr ctxt; |
14056 | 0 | xmlParserInputPtr input; |
14057 | 0 | xmlDocPtr doc; |
14058 | |
|
14059 | 0 | ctxt = xmlNewParserCtxt(); |
14060 | 0 | if (ctxt == NULL) |
14061 | 0 | return(NULL); |
14062 | | |
14063 | 0 | xmlCtxtUseOptions(ctxt, options); |
14064 | |
|
14065 | 0 | input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx, |
14066 | 0 | encoding, 0); |
14067 | |
|
14068 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
14069 | |
|
14070 | 0 | xmlFreeParserCtxt(ctxt); |
14071 | 0 | return(doc); |
14072 | 0 | } |
14073 | | |
14074 | | /** |
14075 | | * xmlCtxtReadDoc: |
14076 | | * @ctxt: an XML parser context |
14077 | | * @str: a pointer to a zero terminated string |
14078 | | * @URL: base URL (optional) |
14079 | | * @encoding: the document encoding (optional) |
14080 | | * @options: a combination of xmlParserOption |
14081 | | * |
14082 | | * Parse an XML in-memory document and build a tree. |
14083 | | * |
14084 | | * @URL is used as base to resolve external entities and for error |
14085 | | * reporting. |
14086 | | * |
14087 | | * See xmlCtxtUseOptions for details. |
14088 | | * |
14089 | | * Returns the resulting document tree |
14090 | | */ |
14091 | | xmlDocPtr |
14092 | | xmlCtxtReadDoc(xmlParserCtxtPtr ctxt, const xmlChar *str, |
14093 | | const char *URL, const char *encoding, int options) |
14094 | 0 | { |
14095 | 0 | xmlParserInputPtr input; |
14096 | |
|
14097 | 0 | if (ctxt == NULL) |
14098 | 0 | return(NULL); |
14099 | | |
14100 | 0 | xmlCtxtReset(ctxt); |
14101 | 0 | xmlCtxtUseOptions(ctxt, options); |
14102 | |
|
14103 | 0 | input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) str, encoding, |
14104 | 0 | XML_INPUT_BUF_STATIC); |
14105 | |
|
14106 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
14107 | 0 | } |
14108 | | |
14109 | | /** |
14110 | | * xmlCtxtReadFile: |
14111 | | * @ctxt: an XML parser context |
14112 | | * @filename: a file or URL |
14113 | | * @encoding: the document encoding (optional) |
14114 | | * @options: a combination of xmlParserOption |
14115 | | * |
14116 | | * Parse an XML file from the filesystem, the network or a user-defined |
14117 | | * resource loader. |
14118 | | * |
14119 | | * Returns the resulting document tree |
14120 | | */ |
14121 | | xmlDocPtr |
14122 | | xmlCtxtReadFile(xmlParserCtxtPtr ctxt, const char *filename, |
14123 | | const char *encoding, int options) |
14124 | 32.7k | { |
14125 | 32.7k | xmlParserInputPtr input; |
14126 | | |
14127 | 32.7k | if (ctxt == NULL) |
14128 | 0 | return(NULL); |
14129 | | |
14130 | 32.7k | xmlCtxtReset(ctxt); |
14131 | 32.7k | xmlCtxtUseOptions(ctxt, options); |
14132 | | |
14133 | 32.7k | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0); |
14134 | | |
14135 | 32.7k | return(xmlCtxtParseDocument(ctxt, input)); |
14136 | 32.7k | } |
14137 | | |
14138 | | /** |
14139 | | * xmlCtxtReadMemory: |
14140 | | * @ctxt: an XML parser context |
14141 | | * @buffer: a pointer to a char array |
14142 | | * @size: the size of the array |
14143 | | * @URL: base URL (optional) |
14144 | | * @encoding: the document encoding (optional) |
14145 | | * @options: a combination of xmlParserOption |
14146 | | * |
14147 | | * Parse an XML in-memory document and build a tree. The input buffer must |
14148 | | * not contain a terminating null byte. |
14149 | | * |
14150 | | * @URL is used as base to resolve external entities and for error |
14151 | | * reporting. |
14152 | | * |
14153 | | * See xmlCtxtUseOptions for details. |
14154 | | * |
14155 | | * Returns the resulting document tree |
14156 | | */ |
14157 | | xmlDocPtr |
14158 | | xmlCtxtReadMemory(xmlParserCtxtPtr ctxt, const char *buffer, int size, |
14159 | | const char *URL, const char *encoding, int options) |
14160 | 89.1k | { |
14161 | 89.1k | xmlParserInputPtr input; |
14162 | | |
14163 | 89.1k | if ((ctxt == NULL) || (size < 0)) |
14164 | 0 | return(NULL); |
14165 | | |
14166 | 89.1k | xmlCtxtReset(ctxt); |
14167 | 89.1k | xmlCtxtUseOptions(ctxt, options); |
14168 | | |
14169 | 89.1k | input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding, |
14170 | 89.1k | XML_INPUT_BUF_STATIC); |
14171 | | |
14172 | 89.1k | return(xmlCtxtParseDocument(ctxt, input)); |
14173 | 89.1k | } |
14174 | | |
14175 | | /** |
14176 | | * xmlCtxtReadFd: |
14177 | | * @ctxt: an XML parser context |
14178 | | * @fd: an open file descriptor |
14179 | | * @URL: base URL (optional) |
14180 | | * @encoding: the document encoding (optional) |
14181 | | * @options: a combination of xmlParserOption |
14182 | | * |
14183 | | * Parse an XML document from a file descriptor and build a tree. |
14184 | | * |
14185 | | * NOTE that the file descriptor will not be closed when the |
14186 | | * context is freed or reset. |
14187 | | * |
14188 | | * @URL is used as base to resolve external entities and for error |
14189 | | * reporting. |
14190 | | * |
14191 | | * See xmlCtxtUseOptions for details. |
14192 | | * |
14193 | | * Returns the resulting document tree |
14194 | | */ |
14195 | | xmlDocPtr |
14196 | | xmlCtxtReadFd(xmlParserCtxtPtr ctxt, int fd, |
14197 | | const char *URL, const char *encoding, int options) |
14198 | 0 | { |
14199 | 0 | xmlParserInputPtr input; |
14200 | |
|
14201 | 0 | if (ctxt == NULL) |
14202 | 0 | return(NULL); |
14203 | | |
14204 | 0 | xmlCtxtReset(ctxt); |
14205 | 0 | xmlCtxtUseOptions(ctxt, options); |
14206 | |
|
14207 | 0 | input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0); |
14208 | |
|
14209 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
14210 | 0 | } |
14211 | | |
14212 | | /** |
14213 | | * xmlCtxtReadIO: |
14214 | | * @ctxt: an XML parser context |
14215 | | * @ioread: an I/O read function |
14216 | | * @ioclose: an I/O close function |
14217 | | * @ioctx: an I/O handler |
14218 | | * @URL: the base URL to use for the document |
14219 | | * @encoding: the document encoding, or NULL |
14220 | | * @options: a combination of xmlParserOption |
14221 | | * |
14222 | | * parse an XML document from I/O functions and source and build a tree. |
14223 | | * This reuses the existing @ctxt parser context |
14224 | | * |
14225 | | * @URL is used as base to resolve external entities and for error |
14226 | | * reporting. |
14227 | | * |
14228 | | * See xmlCtxtUseOptions for details. |
14229 | | * |
14230 | | * Returns the resulting document tree |
14231 | | */ |
14232 | | xmlDocPtr |
14233 | | xmlCtxtReadIO(xmlParserCtxtPtr ctxt, xmlInputReadCallback ioread, |
14234 | | xmlInputCloseCallback ioclose, void *ioctx, |
14235 | | const char *URL, |
14236 | | const char *encoding, int options) |
14237 | 2 | { |
14238 | 2 | xmlParserInputPtr input; |
14239 | | |
14240 | 2 | if (ctxt == NULL) |
14241 | 0 | return(NULL); |
14242 | | |
14243 | 2 | xmlCtxtReset(ctxt); |
14244 | 2 | xmlCtxtUseOptions(ctxt, options); |
14245 | | |
14246 | 2 | input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx, |
14247 | 2 | encoding, 0); |
14248 | | |
14249 | 2 | return(xmlCtxtParseDocument(ctxt, input)); |
14250 | 2 | } |
14251 | | |