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 | 1.36M | #define NS_INDEX_EMPTY INT_MAX |
78 | 71.0k | #define NS_INDEX_XML (INT_MAX - 1) |
79 | 497k | #define URI_HASH_EMPTY 0xD943A04E |
80 | 9.89k | #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 | 607k | #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 | 609k | #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 | 64.7M | #define XML_PARSER_BIG_BUFFER_SIZE 300 |
173 | 140k | #define XML_PARSER_BUFFER_SIZE 100 |
174 | 56.0k | #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 | 0 | xmlErrMemory(xmlParserCtxtPtr ctxt) { |
224 | 0 | xmlCtxtErrMemory(ctxt); |
225 | 0 | } |
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 | 371k | { |
239 | 371k | if (prefix == NULL) |
240 | 366k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED, |
241 | 366k | XML_ERR_FATAL, localname, NULL, NULL, 0, |
242 | 366k | "Attribute %s redefined\n", localname); |
243 | 5.16k | else |
244 | 5.16k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED, |
245 | 5.16k | XML_ERR_FATAL, prefix, localname, NULL, 0, |
246 | 5.16k | "Attribute %s:%s redefined\n", prefix, localname); |
247 | 371k | } |
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 | 3.51M | { |
261 | 3.51M | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
262 | 3.51M | NULL, NULL, NULL, 0, "%s", msg); |
263 | 3.51M | } |
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 | 27.0k | { |
279 | 27.0k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING, |
280 | 27.0k | str1, str2, NULL, 0, msg, str1, str2); |
281 | 27.0k | } |
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 | 792 | { |
296 | 792 | ctxt->valid = 0; |
297 | | |
298 | 792 | xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR, |
299 | 792 | str1, str2, NULL, 0, msg, str1, str2); |
300 | 792 | } |
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 | 8.84M | { |
315 | 8.84M | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
316 | 8.84M | NULL, NULL, NULL, val, msg, val); |
317 | 8.84M | } |
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 | 249k | { |
335 | 249k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
336 | 249k | str1, str2, NULL, val, msg, str1, val, str2); |
337 | 249k | } |
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 | 899k | { |
352 | 899k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
353 | 899k | val, NULL, NULL, 0, msg, val); |
354 | 899k | } |
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 | 0 | { |
369 | 0 | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR, |
370 | 0 | val, NULL, NULL, 0, msg, val); |
371 | 0 | } |
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 | 245k | { |
389 | 245k | ctxt->nsWellFormed = 0; |
390 | | |
391 | 245k | xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR, |
392 | 245k | info1, info2, info3, 0, msg, info1, info2, info3); |
393 | 245k | } |
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 | 7.58k | { |
411 | 7.58k | xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING, |
412 | 7.58k | info1, info2, info3, 0, msg, info1, info2, info3); |
413 | 7.58k | } |
414 | | |
415 | | static void |
416 | 1.82M | xmlSaturatedAdd(unsigned long *dst, unsigned long val) { |
417 | 1.82M | if (val > ULONG_MAX - *dst) |
418 | 0 | *dst = ULONG_MAX; |
419 | 1.82M | else |
420 | 1.82M | *dst += val; |
421 | 1.82M | } |
422 | | |
423 | | static void |
424 | 609k | xmlSaturatedAddSizeT(unsigned long *dst, unsigned long val) { |
425 | 609k | if (val > ULONG_MAX - *dst) |
426 | 0 | *dst = ULONG_MAX; |
427 | 609k | else |
428 | 609k | *dst += val; |
429 | 609k | } |
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 | 609k | { |
458 | 609k | unsigned long consumed; |
459 | 609k | unsigned long *expandedSize; |
460 | 609k | xmlParserInputPtr input = ctxt->input; |
461 | 609k | xmlEntityPtr entity = input->entity; |
462 | | |
463 | 609k | if ((entity) && (entity->flags & XML_ENT_CHECKED)) |
464 | 2.26k | return(0); |
465 | | |
466 | | /* |
467 | | * Compute total consumed bytes so far, including input streams of |
468 | | * external entities. |
469 | | */ |
470 | 607k | consumed = input->consumed; |
471 | 607k | xmlSaturatedAddSizeT(&consumed, input->cur - input->base); |
472 | 607k | xmlSaturatedAdd(&consumed, ctxt->sizeentities); |
473 | | |
474 | 607k | if (entity) |
475 | 1.22k | expandedSize = &entity->expandedSize; |
476 | 605k | else |
477 | 605k | expandedSize = &ctxt->sizeentcopy; |
478 | | |
479 | | /* |
480 | | * Add extra cost and some fixed cost. |
481 | | */ |
482 | 607k | xmlSaturatedAdd(expandedSize, extra); |
483 | 607k | 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 | 607k | if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) && |
491 | 607k | ((*expandedSize >= ULONG_MAX) || |
492 | 235k | (*expandedSize / ctxt->maxAmpl > consumed))) { |
493 | 114 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
494 | 114 | "Maximum entity amplification factor exceeded, see " |
495 | 114 | "xmlCtxtSetMaxAmplification.\n"); |
496 | 114 | xmlHaltParser(ctxt); |
497 | 114 | return(1); |
498 | 114 | } |
499 | | |
500 | 607k | return(0); |
501 | 607k | } |
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 | 0 | { |
522 | 0 | switch (feature) { |
523 | 0 | case XML_WITH_THREAD: |
524 | 0 | #ifdef LIBXML_THREAD_ENABLED |
525 | 0 | return(1); |
526 | | #else |
527 | | return(0); |
528 | | #endif |
529 | 0 | case XML_WITH_TREE: |
530 | 0 | return(1); |
531 | 0 | case XML_WITH_OUTPUT: |
532 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
533 | 0 | return(1); |
534 | | #else |
535 | | return(0); |
536 | | #endif |
537 | 0 | case XML_WITH_PUSH: |
538 | 0 | #ifdef LIBXML_PUSH_ENABLED |
539 | 0 | return(1); |
540 | | #else |
541 | | return(0); |
542 | | #endif |
543 | 0 | case XML_WITH_READER: |
544 | 0 | #ifdef LIBXML_READER_ENABLED |
545 | 0 | return(1); |
546 | | #else |
547 | | return(0); |
548 | | #endif |
549 | 0 | case XML_WITH_PATTERN: |
550 | 0 | #ifdef LIBXML_PATTERN_ENABLED |
551 | 0 | return(1); |
552 | | #else |
553 | | return(0); |
554 | | #endif |
555 | 0 | case XML_WITH_WRITER: |
556 | 0 | #ifdef LIBXML_WRITER_ENABLED |
557 | 0 | return(1); |
558 | | #else |
559 | | return(0); |
560 | | #endif |
561 | 0 | case XML_WITH_SAX1: |
562 | 0 | #ifdef LIBXML_SAX1_ENABLED |
563 | 0 | return(1); |
564 | | #else |
565 | | return(0); |
566 | | #endif |
567 | 0 | case XML_WITH_HTTP: |
568 | | #ifdef LIBXML_HTTP_ENABLED |
569 | | return(1); |
570 | | #else |
571 | 0 | return(0); |
572 | 0 | #endif |
573 | 0 | case XML_WITH_VALID: |
574 | 0 | #ifdef LIBXML_VALID_ENABLED |
575 | 0 | return(1); |
576 | | #else |
577 | | return(0); |
578 | | #endif |
579 | 0 | case XML_WITH_HTML: |
580 | 0 | #ifdef LIBXML_HTML_ENABLED |
581 | 0 | return(1); |
582 | | #else |
583 | | return(0); |
584 | | #endif |
585 | 0 | case XML_WITH_LEGACY: |
586 | | #ifdef LIBXML_LEGACY_ENABLED |
587 | | return(1); |
588 | | #else |
589 | 0 | return(0); |
590 | 0 | #endif |
591 | 0 | case XML_WITH_C14N: |
592 | 0 | #ifdef LIBXML_C14N_ENABLED |
593 | 0 | return(1); |
594 | | #else |
595 | | return(0); |
596 | | #endif |
597 | 0 | case XML_WITH_CATALOG: |
598 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
599 | 0 | return(1); |
600 | | #else |
601 | | return(0); |
602 | | #endif |
603 | 0 | case XML_WITH_XPATH: |
604 | 0 | #ifdef LIBXML_XPATH_ENABLED |
605 | 0 | return(1); |
606 | | #else |
607 | | return(0); |
608 | | #endif |
609 | 0 | case XML_WITH_XPTR: |
610 | 0 | #ifdef LIBXML_XPTR_ENABLED |
611 | 0 | return(1); |
612 | | #else |
613 | | return(0); |
614 | | #endif |
615 | 0 | case XML_WITH_XINCLUDE: |
616 | 0 | #ifdef LIBXML_XINCLUDE_ENABLED |
617 | 0 | return(1); |
618 | | #else |
619 | | return(0); |
620 | | #endif |
621 | 0 | case XML_WITH_ICONV: |
622 | 0 | #ifdef LIBXML_ICONV_ENABLED |
623 | 0 | return(1); |
624 | | #else |
625 | | return(0); |
626 | | #endif |
627 | 0 | case XML_WITH_ISO8859X: |
628 | 0 | #ifdef LIBXML_ISO8859X_ENABLED |
629 | 0 | return(1); |
630 | | #else |
631 | | return(0); |
632 | | #endif |
633 | 0 | case XML_WITH_UNICODE: |
634 | 0 | #ifdef LIBXML_UNICODE_ENABLED |
635 | 0 | return(1); |
636 | | #else |
637 | | return(0); |
638 | | #endif |
639 | 0 | case XML_WITH_REGEXP: |
640 | 0 | #ifdef LIBXML_REGEXP_ENABLED |
641 | 0 | return(1); |
642 | | #else |
643 | | return(0); |
644 | | #endif |
645 | 0 | case XML_WITH_AUTOMATA: |
646 | 0 | #ifdef LIBXML_REGEXP_ENABLED |
647 | 0 | return(1); |
648 | | #else |
649 | | return(0); |
650 | | #endif |
651 | 0 | case XML_WITH_EXPR: |
652 | | #ifdef LIBXML_EXPR_ENABLED |
653 | | return(1); |
654 | | #else |
655 | 0 | return(0); |
656 | 0 | #endif |
657 | 0 | case XML_WITH_SCHEMAS: |
658 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
659 | 0 | return(1); |
660 | | #else |
661 | | return(0); |
662 | | #endif |
663 | 0 | case XML_WITH_SCHEMATRON: |
664 | 0 | #ifdef LIBXML_SCHEMATRON_ENABLED |
665 | 0 | return(1); |
666 | | #else |
667 | | return(0); |
668 | | #endif |
669 | 0 | case XML_WITH_MODULES: |
670 | 0 | #ifdef LIBXML_MODULES_ENABLED |
671 | 0 | return(1); |
672 | | #else |
673 | | return(0); |
674 | | #endif |
675 | 0 | case XML_WITH_DEBUG: |
676 | | #ifdef LIBXML_DEBUG_ENABLED |
677 | | return(1); |
678 | | #else |
679 | 0 | return(0); |
680 | 0 | #endif |
681 | 0 | case XML_WITH_DEBUG_MEM: |
682 | 0 | return(0); |
683 | 0 | case XML_WITH_ZLIB: |
684 | | #ifdef LIBXML_ZLIB_ENABLED |
685 | | return(1); |
686 | | #else |
687 | 0 | return(0); |
688 | 0 | #endif |
689 | 0 | case XML_WITH_LZMA: |
690 | | #ifdef LIBXML_LZMA_ENABLED |
691 | | return(1); |
692 | | #else |
693 | 0 | return(0); |
694 | 0 | #endif |
695 | 0 | case XML_WITH_ICU: |
696 | | #ifdef LIBXML_ICU_ENABLED |
697 | | return(1); |
698 | | #else |
699 | 0 | return(0); |
700 | 0 | #endif |
701 | 0 | default: |
702 | 0 | break; |
703 | 0 | } |
704 | 0 | return(0); |
705 | 0 | } |
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 | 732k | xmlSBufInit(xmlSBuf *buf, unsigned max) { |
723 | 732k | buf->mem = NULL; |
724 | 732k | buf->size = 0; |
725 | 732k | buf->cap = 0; |
726 | 732k | buf->max = max; |
727 | 732k | buf->code = XML_ERR_OK; |
728 | 732k | } |
729 | | |
730 | | static int |
731 | 174k | xmlSBufGrow(xmlSBuf *buf, unsigned len) { |
732 | 174k | xmlChar *mem; |
733 | 174k | unsigned cap; |
734 | | |
735 | 174k | 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 | 174k | cap = (buf->size + len) * 2; |
742 | 174k | if (cap < 240) |
743 | 131k | cap = 240; |
744 | | |
745 | 174k | mem = xmlRealloc(buf->mem, cap); |
746 | 174k | if (mem == NULL) { |
747 | 0 | buf->code = XML_ERR_NO_MEMORY; |
748 | 0 | return(-1); |
749 | 0 | } |
750 | | |
751 | 174k | buf->mem = mem; |
752 | 174k | buf->cap = cap; |
753 | | |
754 | 174k | return(0); |
755 | 174k | } |
756 | | |
757 | | static void |
758 | 23.7M | xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) { |
759 | 23.7M | if (buf->max - buf->size < len) { |
760 | 735k | if (buf->code == XML_ERR_OK) |
761 | 108 | buf->code = XML_ERR_RESOURCE_LIMIT; |
762 | 735k | return; |
763 | 735k | } |
764 | | |
765 | 23.0M | if (buf->cap - buf->size <= len) { |
766 | 165k | if (xmlSBufGrow(buf, len) < 0) |
767 | 0 | return; |
768 | 165k | } |
769 | | |
770 | 23.0M | if (len > 0) |
771 | 23.0M | memcpy(buf->mem + buf->size, str, len); |
772 | 23.0M | buf->size += len; |
773 | 23.0M | } |
774 | | |
775 | | static void |
776 | 18.7M | xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) { |
777 | 18.7M | xmlSBufAddString(buf, (const xmlChar *) str, len); |
778 | 18.7M | } |
779 | | |
780 | | static void |
781 | 141k | xmlSBufAddChar(xmlSBuf *buf, int c) { |
782 | 141k | xmlChar *end; |
783 | | |
784 | 141k | if (buf->max - buf->size < 4) { |
785 | 593 | if (buf->code == XML_ERR_OK) |
786 | 0 | buf->code = XML_ERR_RESOURCE_LIMIT; |
787 | 593 | return; |
788 | 593 | } |
789 | | |
790 | 140k | if (buf->cap - buf->size <= 4) { |
791 | 9.32k | if (xmlSBufGrow(buf, 4) < 0) |
792 | 0 | return; |
793 | 9.32k | } |
794 | | |
795 | 140k | end = buf->mem + buf->size; |
796 | | |
797 | 140k | if (c < 0x80) { |
798 | 45.2k | *end = (xmlChar) c; |
799 | 45.2k | buf->size += 1; |
800 | 95.7k | } else { |
801 | 95.7k | buf->size += xmlCopyCharMultiByte(end, c); |
802 | 95.7k | } |
803 | 140k | } |
804 | | |
805 | | static void |
806 | 12.4M | xmlSBufAddReplChar(xmlSBuf *buf) { |
807 | 12.4M | xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3); |
808 | 12.4M | } |
809 | | |
810 | | static void |
811 | 108 | xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) { |
812 | 108 | if (buf->code == XML_ERR_NO_MEMORY) |
813 | 0 | xmlCtxtErrMemory(ctxt); |
814 | 108 | else |
815 | 108 | xmlFatalErr(ctxt, buf->code, errMsg); |
816 | 108 | } |
817 | | |
818 | | static xmlChar * |
819 | | xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt, |
820 | 159k | const char *errMsg) { |
821 | 159k | if (buf->mem == NULL) { |
822 | 26.3k | buf->mem = xmlMalloc(1); |
823 | 26.3k | if (buf->mem == NULL) { |
824 | 0 | buf->code = XML_ERR_NO_MEMORY; |
825 | 26.3k | } else { |
826 | 26.3k | buf->mem[0] = 0; |
827 | 26.3k | } |
828 | 133k | } else { |
829 | 133k | buf->mem[buf->size] = 0; |
830 | 133k | } |
831 | | |
832 | 159k | if (buf->code == XML_ERR_OK) { |
833 | 159k | if (sizeOut != NULL) |
834 | 68.5k | *sizeOut = buf->size; |
835 | 159k | return(buf->mem); |
836 | 159k | } |
837 | | |
838 | 35 | xmlSBufReportError(buf, ctxt, errMsg); |
839 | | |
840 | 35 | xmlFree(buf->mem); |
841 | | |
842 | 35 | if (sizeOut != NULL) |
843 | 34 | *sizeOut = 0; |
844 | 35 | return(NULL); |
845 | 159k | } |
846 | | |
847 | | static void |
848 | 561k | xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) { |
849 | 561k | if (buf->code != XML_ERR_OK) |
850 | 73 | xmlSBufReportError(buf, ctxt, errMsg); |
851 | | |
852 | 561k | xmlFree(buf->mem); |
853 | 561k | } |
854 | | |
855 | | static int |
856 | | xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str, |
857 | 409M | const char *errMsg) { |
858 | 409M | int c = str[0]; |
859 | 409M | int c1 = str[1]; |
860 | | |
861 | 409M | if ((c1 & 0xC0) != 0x80) |
862 | 1.15M | goto encoding_error; |
863 | | |
864 | 408M | if (c < 0xE0) { |
865 | | /* 2-byte sequence */ |
866 | 10.0M | if (c < 0xC2) |
867 | 8.41M | goto encoding_error; |
868 | | |
869 | 1.64M | return(2); |
870 | 398M | } else { |
871 | 398M | int c2 = str[2]; |
872 | | |
873 | 398M | if ((c2 & 0xC0) != 0x80) |
874 | 8.36k | goto encoding_error; |
875 | | |
876 | 398M | if (c < 0xF0) { |
877 | | /* 3-byte sequence */ |
878 | 397M | if (c == 0xE0) { |
879 | | /* overlong */ |
880 | 397M | if (c1 < 0xA0) |
881 | 5.87k | goto encoding_error; |
882 | 397M | } else if (c == 0xED) { |
883 | | /* surrogate */ |
884 | 2.09k | if (c1 >= 0xA0) |
885 | 567 | goto encoding_error; |
886 | 244k | } else if (c == 0xEF) { |
887 | | /* U+FFFE and U+FFFF are invalid Chars */ |
888 | 8.07k | if ((c1 == 0xBF) && (c2 >= 0xBE)) |
889 | 2.51k | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg); |
890 | 8.07k | } |
891 | | |
892 | 397M | return(3); |
893 | 397M | } else { |
894 | | /* 4-byte sequence */ |
895 | 14.4k | if ((str[3] & 0xC0) != 0x80) |
896 | 1.43k | goto encoding_error; |
897 | 13.0k | if (c == 0xF0) { |
898 | | /* overlong */ |
899 | 2.10k | if (c1 < 0x90) |
900 | 1.17k | goto encoding_error; |
901 | 10.9k | } else if (c >= 0xF4) { |
902 | | /* greater than 0x10FFFF */ |
903 | 3.31k | if ((c > 0xF4) || (c1 >= 0x90)) |
904 | 2.05k | goto encoding_error; |
905 | 3.31k | } |
906 | | |
907 | 9.80k | return(4); |
908 | 13.0k | } |
909 | 398M | } |
910 | | |
911 | 9.59M | encoding_error: |
912 | | /* Only report the first error */ |
913 | 9.59M | if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) { |
914 | 1.33k | xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL); |
915 | 1.33k | ctxt->input->flags |= XML_INPUT_ENCODING_ERROR; |
916 | 1.33k | } |
917 | | |
918 | 9.59M | return(0); |
919 | 408M | } |
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 | 20.2k | xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) { |
938 | 20.2k | xmlSAXHandlerPtr sax; |
939 | | |
940 | | /* Avoid unused variable warning if features are disabled. */ |
941 | 20.2k | (void) sax; |
942 | | |
943 | | /* |
944 | | * Changing the SAX struct directly is still widespread practice |
945 | | * in internal and external code. |
946 | | */ |
947 | 20.2k | if (ctxt == NULL) return; |
948 | 20.2k | sax = ctxt->sax; |
949 | 20.2k | #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 | 20.2k | if (((ctxt->options & XML_PARSE_SAX1) == 0) && |
955 | 20.2k | (sax) && |
956 | 20.2k | (sax->initialized == XML_SAX2_MAGIC) && |
957 | 20.2k | ((sax->startElementNs != NULL) || |
958 | 20.2k | (sax->endElementNs != NULL) || |
959 | 20.2k | ((sax->startElement == NULL) && (sax->endElement == NULL)))) |
960 | 20.2k | 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 | 20.2k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
970 | 20.2k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
971 | 20.2k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
972 | 20.2k | if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) || |
973 | 20.2k | (ctxt->str_xml_ns == NULL)) { |
974 | 0 | xmlErrMemory(ctxt); |
975 | 0 | } |
976 | | |
977 | 20.2k | xmlDictSetLimit(ctxt->dict, |
978 | 20.2k | (ctxt->options & XML_PARSE_HUGE) ? |
979 | 0 | 0 : |
980 | 20.2k | XML_MAX_DICTIONARY_LIMIT); |
981 | 20.2k | } |
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 | 39.8k | { |
1025 | 39.8k | if ((src == NULL) || (dst == NULL)) |
1026 | 0 | return(NULL); |
1027 | | |
1028 | 42.8k | while (*src == 0x20) src++; |
1029 | 634k | while (*src != 0) { |
1030 | 594k | if (*src == 0x20) { |
1031 | 88.8k | while (*src == 0x20) src++; |
1032 | 27.2k | if (*src != 0) |
1033 | 21.8k | *dst++ = 0x20; |
1034 | 567k | } else { |
1035 | 567k | *dst++ = *src++; |
1036 | 567k | } |
1037 | 594k | } |
1038 | 39.8k | *dst = 0; |
1039 | 39.8k | if (dst == src) |
1040 | 30.9k | return(NULL); |
1041 | 8.91k | return(dst); |
1042 | 39.8k | } |
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 | 45.5k | const xmlChar *value) { |
1058 | 45.5k | xmlDefAttrsPtr defaults; |
1059 | 45.5k | xmlDefAttr *attr; |
1060 | 45.5k | int len, expandedSize; |
1061 | 45.5k | xmlHashedString name; |
1062 | 45.5k | xmlHashedString prefix; |
1063 | 45.5k | xmlHashedString hvalue; |
1064 | 45.5k | const xmlChar *localname; |
1065 | | |
1066 | | /* |
1067 | | * Allows to detect attribute redefinitions |
1068 | | */ |
1069 | 45.5k | if (ctxt->attsSpecial != NULL) { |
1070 | 43.4k | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1071 | 24.2k | return; |
1072 | 43.4k | } |
1073 | | |
1074 | 21.2k | if (ctxt->attsDefault == NULL) { |
1075 | 2.06k | ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict); |
1076 | 2.06k | if (ctxt->attsDefault == NULL) |
1077 | 0 | goto mem_error; |
1078 | 2.06k | } |
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 | 21.2k | localname = xmlSplitQName3(fullname, &len); |
1085 | 21.2k | if (localname == NULL) { |
1086 | 16.5k | name = xmlDictLookupHashed(ctxt->dict, fullname, -1); |
1087 | 16.5k | prefix.name = NULL; |
1088 | 16.5k | } else { |
1089 | 4.74k | name = xmlDictLookupHashed(ctxt->dict, localname, -1); |
1090 | 4.74k | prefix = xmlDictLookupHashed(ctxt->dict, fullname, len); |
1091 | 4.74k | if (prefix.name == NULL) |
1092 | 0 | goto mem_error; |
1093 | 4.74k | } |
1094 | 21.2k | if (name.name == NULL) |
1095 | 0 | goto mem_error; |
1096 | | |
1097 | | /* |
1098 | | * make sure there is some storage |
1099 | | */ |
1100 | 21.2k | defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name); |
1101 | 21.2k | if ((defaults == NULL) || |
1102 | 21.2k | (defaults->nbAttrs >= defaults->maxAttrs)) { |
1103 | 6.22k | xmlDefAttrsPtr temp; |
1104 | 6.22k | int newSize; |
1105 | | |
1106 | 6.22k | newSize = (defaults != NULL) ? 2 * defaults->maxAttrs : 4; |
1107 | 6.22k | temp = xmlRealloc(defaults, |
1108 | 6.22k | sizeof(*defaults) + newSize * sizeof(xmlDefAttr)); |
1109 | 6.22k | if (temp == NULL) |
1110 | 0 | goto mem_error; |
1111 | 6.22k | if (defaults == NULL) |
1112 | 3.32k | temp->nbAttrs = 0; |
1113 | 6.22k | temp->maxAttrs = newSize; |
1114 | 6.22k | defaults = temp; |
1115 | 6.22k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name, |
1116 | 6.22k | defaults, NULL) < 0) { |
1117 | 0 | xmlFree(defaults); |
1118 | 0 | goto mem_error; |
1119 | 0 | } |
1120 | 6.22k | } |
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 | 21.2k | localname = xmlSplitQName3(fullattr, &len); |
1127 | 21.2k | if (localname == NULL) { |
1128 | 12.2k | name = xmlDictLookupHashed(ctxt->dict, fullattr, -1); |
1129 | 12.2k | prefix.name = NULL; |
1130 | 12.2k | } else { |
1131 | 9.04k | name = xmlDictLookupHashed(ctxt->dict, localname, -1); |
1132 | 9.04k | prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len); |
1133 | 9.04k | if (prefix.name == NULL) |
1134 | 0 | goto mem_error; |
1135 | 9.04k | } |
1136 | 21.2k | if (name.name == NULL) |
1137 | 0 | goto mem_error; |
1138 | | |
1139 | | /* intern the string and precompute the end */ |
1140 | 21.2k | len = strlen((const char *) value); |
1141 | 21.2k | hvalue = xmlDictLookupHashed(ctxt->dict, value, len); |
1142 | 21.2k | if (hvalue.name == NULL) |
1143 | 0 | goto mem_error; |
1144 | | |
1145 | 21.2k | expandedSize = strlen((const char *) name.name); |
1146 | 21.2k | if (prefix.name != NULL) |
1147 | 9.04k | expandedSize += strlen((const char *) prefix.name); |
1148 | 21.2k | expandedSize += len; |
1149 | | |
1150 | 21.2k | attr = &defaults->attrs[defaults->nbAttrs++]; |
1151 | 21.2k | attr->name = name; |
1152 | 21.2k | attr->prefix = prefix; |
1153 | 21.2k | attr->value = hvalue; |
1154 | 21.2k | attr->valueEnd = hvalue.name + len; |
1155 | 21.2k | attr->external = PARSER_EXTERNAL(ctxt); |
1156 | 21.2k | attr->expandedSize = expandedSize; |
1157 | | |
1158 | 21.2k | return; |
1159 | | |
1160 | 0 | mem_error: |
1161 | 0 | xmlErrMemory(ctxt); |
1162 | 0 | } |
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 | 47.6k | { |
1179 | 47.6k | if (ctxt->attsSpecial == NULL) { |
1180 | 2.19k | ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict); |
1181 | 2.19k | if (ctxt->attsSpecial == NULL) |
1182 | 0 | goto mem_error; |
1183 | 2.19k | } |
1184 | | |
1185 | 47.6k | if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr, |
1186 | 47.6k | (void *) (ptrdiff_t) type) < 0) |
1187 | 0 | goto mem_error; |
1188 | 47.6k | return; |
1189 | | |
1190 | 47.6k | mem_error: |
1191 | 0 | xmlErrMemory(ctxt); |
1192 | 0 | } |
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 | 21.9k | const xmlChar *unused ATTRIBUTE_UNUSED) { |
1203 | 21.9k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data; |
1204 | | |
1205 | 21.9k | if (((ptrdiff_t) payload) == XML_ATTRIBUTE_CDATA) { |
1206 | 4.41k | xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL); |
1207 | 4.41k | } |
1208 | 21.9k | } |
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 | 6.97k | { |
1221 | 6.97k | if (ctxt->attsSpecial == NULL) |
1222 | 4.77k | return; |
1223 | | |
1224 | 2.19k | xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt); |
1225 | | |
1226 | 2.19k | if (xmlHashSize(ctxt->attsSpecial) == 0) { |
1227 | 29 | xmlHashFree(ctxt->attsSpecial, NULL); |
1228 | 29 | ctxt->attsSpecial = NULL; |
1229 | 29 | } |
1230 | 2.19k | } |
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 | 0 | { |
1295 | 0 | const xmlChar *cur = lang, *nxt; |
1296 | |
|
1297 | 0 | if (cur == NULL) |
1298 | 0 | return (0); |
1299 | 0 | if (((cur[0] == 'i') && (cur[1] == '-')) || |
1300 | 0 | ((cur[0] == 'I') && (cur[1] == '-')) || |
1301 | 0 | ((cur[0] == 'x') && (cur[1] == '-')) || |
1302 | 0 | ((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 | 0 | cur += 2; |
1309 | 0 | while (((cur[0] >= 'A') && (cur[0] <= 'Z')) || |
1310 | 0 | ((cur[0] >= 'a') && (cur[0] <= 'z'))) |
1311 | 0 | cur++; |
1312 | 0 | return(cur[0] == 0); |
1313 | 0 | } |
1314 | 0 | nxt = cur; |
1315 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1316 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1317 | 0 | nxt++; |
1318 | 0 | if (nxt - cur >= 4) { |
1319 | | /* |
1320 | | * Reserved |
1321 | | */ |
1322 | 0 | if ((nxt - cur > 8) || (nxt[0] != 0)) |
1323 | 0 | return(0); |
1324 | 0 | return(1); |
1325 | 0 | } |
1326 | 0 | if (nxt - cur < 2) |
1327 | 0 | return(0); |
1328 | | /* we got an ISO 639 code */ |
1329 | 0 | if (nxt[0] == 0) |
1330 | 0 | return(1); |
1331 | 0 | if (nxt[0] != '-') |
1332 | 0 | return(0); |
1333 | | |
1334 | 0 | nxt++; |
1335 | 0 | cur = nxt; |
1336 | | /* now we can have extlang or script or region or variant */ |
1337 | 0 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1338 | 0 | goto region_m49; |
1339 | | |
1340 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1341 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1342 | 0 | nxt++; |
1343 | 0 | if (nxt - cur == 4) |
1344 | 0 | goto script; |
1345 | 0 | if (nxt - cur == 2) |
1346 | 0 | goto region; |
1347 | 0 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1348 | 0 | goto variant; |
1349 | 0 | if (nxt - cur != 3) |
1350 | 0 | return(0); |
1351 | | /* we parsed an extlang */ |
1352 | 0 | if (nxt[0] == 0) |
1353 | 0 | return(1); |
1354 | 0 | if (nxt[0] != '-') |
1355 | 0 | return(0); |
1356 | | |
1357 | 0 | nxt++; |
1358 | 0 | cur = nxt; |
1359 | | /* now we can have script or region or variant */ |
1360 | 0 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1361 | 0 | goto region_m49; |
1362 | | |
1363 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1364 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1365 | 0 | nxt++; |
1366 | 0 | if (nxt - cur == 2) |
1367 | 0 | goto region; |
1368 | 0 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1369 | 0 | goto variant; |
1370 | 0 | if (nxt - cur != 4) |
1371 | 0 | return(0); |
1372 | | /* we parsed a script */ |
1373 | 0 | script: |
1374 | 0 | if (nxt[0] == 0) |
1375 | 0 | return(1); |
1376 | 0 | if (nxt[0] != '-') |
1377 | 0 | return(0); |
1378 | | |
1379 | 0 | nxt++; |
1380 | 0 | cur = nxt; |
1381 | | /* now we can have region or variant */ |
1382 | 0 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1383 | 0 | goto region_m49; |
1384 | | |
1385 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1386 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1387 | 0 | nxt++; |
1388 | |
|
1389 | 0 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1390 | 0 | goto variant; |
1391 | 0 | if (nxt - cur != 2) |
1392 | 0 | return(0); |
1393 | | /* we parsed a region */ |
1394 | 0 | region: |
1395 | 0 | if (nxt[0] == 0) |
1396 | 0 | return(1); |
1397 | 0 | if (nxt[0] != '-') |
1398 | 0 | return(0); |
1399 | | |
1400 | 0 | nxt++; |
1401 | 0 | cur = nxt; |
1402 | | /* now we can just have a variant */ |
1403 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1404 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1405 | 0 | nxt++; |
1406 | |
|
1407 | 0 | if ((nxt - cur < 5) || (nxt - cur > 8)) |
1408 | 0 | return(0); |
1409 | | |
1410 | | /* we parsed a variant */ |
1411 | 0 | variant: |
1412 | 0 | if (nxt[0] == 0) |
1413 | 0 | return(1); |
1414 | 0 | if (nxt[0] != '-') |
1415 | 0 | return(0); |
1416 | | /* extensions and private use subtags not checked */ |
1417 | 0 | return (1); |
1418 | | |
1419 | 0 | region_m49: |
1420 | 0 | if (((nxt[1] >= '0') && (nxt[1] <= '9')) && |
1421 | 0 | ((nxt[2] >= '0') && (nxt[2] <= '9'))) { |
1422 | 0 | nxt += 3; |
1423 | 0 | goto region; |
1424 | 0 | } |
1425 | 0 | return(0); |
1426 | 0 | } |
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 | 20.2k | xmlParserNsCreate(void) { |
1446 | 20.2k | xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb)); |
1447 | | |
1448 | 20.2k | if (nsdb == NULL) |
1449 | 0 | return(NULL); |
1450 | 20.2k | memset(nsdb, 0, sizeof(*nsdb)); |
1451 | 20.2k | nsdb->defaultNsIndex = INT_MAX; |
1452 | | |
1453 | 20.2k | return(nsdb); |
1454 | 20.2k | } |
1455 | | |
1456 | | /** |
1457 | | * xmlParserNsFree: |
1458 | | * @nsdb: namespace database |
1459 | | * |
1460 | | * Free a namespace database. |
1461 | | */ |
1462 | | void |
1463 | 20.2k | xmlParserNsFree(xmlParserNsData *nsdb) { |
1464 | 20.2k | if (nsdb == NULL) |
1465 | 0 | return; |
1466 | | |
1467 | 20.2k | xmlFree(nsdb->extra); |
1468 | 20.2k | xmlFree(nsdb->hash); |
1469 | 20.2k | xmlFree(nsdb); |
1470 | 20.2k | } |
1471 | | |
1472 | | /** |
1473 | | * xmlParserNsReset: |
1474 | | * @nsdb: namespace database |
1475 | | * |
1476 | | * Reset a namespace database. |
1477 | | */ |
1478 | | static void |
1479 | 0 | xmlParserNsReset(xmlParserNsData *nsdb) { |
1480 | 0 | if (nsdb == NULL) |
1481 | 0 | return; |
1482 | | |
1483 | 0 | nsdb->hashElems = 0; |
1484 | 0 | nsdb->elementId = 0; |
1485 | 0 | nsdb->defaultNsIndex = INT_MAX; |
1486 | |
|
1487 | 0 | if (nsdb->hash) |
1488 | 0 | memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0])); |
1489 | 0 | } |
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 | 665k | xmlParserNsStartElement(xmlParserNsData *nsdb) { |
1501 | 665k | if (nsdb->elementId == UINT_MAX) |
1502 | 0 | return(-1); |
1503 | 665k | nsdb->elementId++; |
1504 | | |
1505 | 665k | return(0); |
1506 | 665k | } |
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 | 792k | xmlParserNsBucket **bucketPtr) { |
1524 | 792k | xmlParserNsBucket *bucket, *tombstone; |
1525 | 792k | unsigned index, hashValue; |
1526 | | |
1527 | 792k | if (prefix->name == NULL) |
1528 | 398k | return(ctxt->nsdb->defaultNsIndex); |
1529 | | |
1530 | 393k | if (ctxt->nsdb->hashSize == 0) |
1531 | 14.3k | return(INT_MAX); |
1532 | | |
1533 | 379k | hashValue = prefix->hashValue; |
1534 | 379k | index = hashValue & (ctxt->nsdb->hashSize - 1); |
1535 | 379k | bucket = &ctxt->nsdb->hash[index]; |
1536 | 379k | tombstone = NULL; |
1537 | | |
1538 | 516k | while (bucket->hashValue) { |
1539 | 375k | if (bucket->index == INT_MAX) { |
1540 | 58.9k | if (tombstone == NULL) |
1541 | 56.1k | tombstone = bucket; |
1542 | 316k | } else if (bucket->hashValue == hashValue) { |
1543 | 238k | if (ctxt->nsTab[bucket->index * 2] == prefix->name) { |
1544 | 238k | if (bucketPtr != NULL) |
1545 | 168k | *bucketPtr = bucket; |
1546 | 238k | return(bucket->index); |
1547 | 238k | } |
1548 | 238k | } |
1549 | | |
1550 | 136k | index++; |
1551 | 136k | bucket++; |
1552 | 136k | if (index == ctxt->nsdb->hashSize) { |
1553 | 14.4k | index = 0; |
1554 | 14.4k | bucket = ctxt->nsdb->hash; |
1555 | 14.4k | } |
1556 | 136k | } |
1557 | | |
1558 | 141k | if (bucketPtr != NULL) |
1559 | 41.5k | *bucketPtr = tombstone ? tombstone : bucket; |
1560 | 141k | return(INT_MAX); |
1561 | 379k | } |
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 | 448k | xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) { |
1574 | 448k | const xmlChar *ret; |
1575 | 448k | int nsIndex; |
1576 | | |
1577 | 448k | if (prefix->name == ctxt->str_xml) |
1578 | 4.25k | 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 | 444k | nsIndex = xmlParserNsLookup(ctxt, prefix, NULL); |
1585 | 444k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1586 | 212k | return(NULL); |
1587 | | |
1588 | 231k | ret = ctxt->nsTab[nsIndex * 2 + 1]; |
1589 | 231k | if (ret[0] == 0) |
1590 | 6.40k | ret = NULL; |
1591 | 231k | return(ret); |
1592 | 444k | } |
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 | 13.5k | xmlParserNsLookupSax(xmlParserCtxtPtr ctxt, const xmlChar *prefix) { |
1606 | 13.5k | xmlHashedString hprefix; |
1607 | 13.5k | int nsIndex; |
1608 | | |
1609 | 13.5k | if (prefix == ctxt->str_xml) |
1610 | 8.13k | return(NULL); |
1611 | | |
1612 | 5.41k | hprefix.name = prefix; |
1613 | 5.41k | if (prefix != NULL) |
1614 | 1.54k | hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix); |
1615 | 3.87k | else |
1616 | 3.87k | hprefix.hashValue = 0; |
1617 | 5.41k | nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL); |
1618 | 5.41k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1619 | 0 | return(NULL); |
1620 | | |
1621 | 5.41k | return(ctxt->nsdb->extra[nsIndex].saxData); |
1622 | 5.41k | } |
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 | 8.37k | void *saxData) { |
1639 | 8.37k | xmlHashedString hprefix; |
1640 | 8.37k | int nsIndex; |
1641 | | |
1642 | 8.37k | if (prefix == ctxt->str_xml) |
1643 | 0 | return(-1); |
1644 | | |
1645 | 8.37k | hprefix.name = prefix; |
1646 | 8.37k | if (prefix != NULL) |
1647 | 5.82k | hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix); |
1648 | 2.54k | else |
1649 | 2.54k | hprefix.hashValue = 0; |
1650 | 8.37k | nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL); |
1651 | 8.37k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1652 | 0 | return(-1); |
1653 | | |
1654 | 8.37k | ctxt->nsdb->extra[nsIndex].saxData = saxData; |
1655 | 8.37k | return(0); |
1656 | 8.37k | } |
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 | 5.02k | xmlParserNsGrow(xmlParserCtxtPtr ctxt) { |
1668 | 5.02k | const xmlChar **table; |
1669 | 5.02k | xmlParserNsExtra *extra; |
1670 | 5.02k | int newSize; |
1671 | | |
1672 | 5.02k | if (ctxt->nsMax > INT_MAX / 2) |
1673 | 0 | goto error; |
1674 | 5.02k | newSize = ctxt->nsMax ? ctxt->nsMax * 2 : 16; |
1675 | | |
1676 | 5.02k | table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0])); |
1677 | 5.02k | if (table == NULL) |
1678 | 0 | goto error; |
1679 | 5.02k | ctxt->nsTab = table; |
1680 | | |
1681 | 5.02k | extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0])); |
1682 | 5.02k | if (extra == NULL) |
1683 | 0 | goto error; |
1684 | 5.02k | ctxt->nsdb->extra = extra; |
1685 | | |
1686 | 5.02k | ctxt->nsMax = newSize; |
1687 | 5.02k | return(0); |
1688 | | |
1689 | 0 | error: |
1690 | 0 | xmlErrMemory(ctxt); |
1691 | 0 | return(-1); |
1692 | 5.02k | } |
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 | 196k | const xmlHashedString *uri, void *saxData, int defAttr) { |
1710 | 196k | xmlParserNsBucket *bucket = NULL; |
1711 | 196k | xmlParserNsExtra *extra; |
1712 | 196k | const xmlChar **ns; |
1713 | 196k | unsigned hashValue, nsIndex, oldIndex; |
1714 | | |
1715 | 196k | if ((prefix != NULL) && (prefix->name == ctxt->str_xml)) |
1716 | 444 | return(0); |
1717 | | |
1718 | 196k | if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) { |
1719 | 0 | xmlErrMemory(ctxt); |
1720 | 0 | return(-1); |
1721 | 0 | } |
1722 | | |
1723 | | /* |
1724 | | * Default namespace and 'xml' namespace |
1725 | | */ |
1726 | 196k | if ((prefix == NULL) || (prefix->name == NULL)) { |
1727 | 88.2k | oldIndex = ctxt->nsdb->defaultNsIndex; |
1728 | | |
1729 | 88.2k | if (oldIndex != INT_MAX) { |
1730 | 62.3k | extra = &ctxt->nsdb->extra[oldIndex]; |
1731 | | |
1732 | 62.3k | if (extra->elementId == ctxt->nsdb->elementId) { |
1733 | 7.51k | if (defAttr == 0) |
1734 | 6.59k | xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns"); |
1735 | 7.51k | return(0); |
1736 | 7.51k | } |
1737 | | |
1738 | 54.8k | if ((ctxt->options & XML_PARSE_NSCLEAN) && |
1739 | 54.8k | (uri->name == ctxt->nsTab[oldIndex * 2 + 1])) |
1740 | 0 | return(0); |
1741 | 54.8k | } |
1742 | | |
1743 | 80.7k | ctxt->nsdb->defaultNsIndex = ctxt->nsNr; |
1744 | 80.7k | goto populate_entry; |
1745 | 88.2k | } |
1746 | | |
1747 | | /* |
1748 | | * Hash table lookup |
1749 | | */ |
1750 | 108k | oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket); |
1751 | 108k | if (oldIndex != INT_MAX) { |
1752 | 65.0k | extra = &ctxt->nsdb->extra[oldIndex]; |
1753 | | |
1754 | | /* |
1755 | | * Check for duplicate definitions on the same element. |
1756 | | */ |
1757 | 65.0k | if (extra->elementId == ctxt->nsdb->elementId) { |
1758 | 3.54k | if (defAttr == 0) |
1759 | 2.13k | xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name); |
1760 | 3.54k | return(0); |
1761 | 3.54k | } |
1762 | | |
1763 | 61.5k | if ((ctxt->options & XML_PARSE_NSCLEAN) && |
1764 | 61.5k | (uri->name == ctxt->nsTab[bucket->index * 2 + 1])) |
1765 | 0 | return(0); |
1766 | | |
1767 | 61.5k | bucket->index = ctxt->nsNr; |
1768 | 61.5k | goto populate_entry; |
1769 | 61.5k | } |
1770 | | |
1771 | | /* |
1772 | | * Insert new bucket |
1773 | | */ |
1774 | | |
1775 | 42.9k | hashValue = prefix->hashValue; |
1776 | | |
1777 | | /* |
1778 | | * Grow hash table, 50% fill factor |
1779 | | */ |
1780 | 42.9k | if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) { |
1781 | 2.57k | xmlParserNsBucket *newHash; |
1782 | 2.57k | unsigned newSize, i, index; |
1783 | | |
1784 | 2.57k | if (ctxt->nsdb->hashSize > UINT_MAX / 2) { |
1785 | 0 | xmlErrMemory(ctxt); |
1786 | 0 | return(-1); |
1787 | 0 | } |
1788 | 2.57k | newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16; |
1789 | 2.57k | newHash = xmlMalloc(newSize * sizeof(newHash[0])); |
1790 | 2.57k | if (newHash == NULL) { |
1791 | 0 | xmlErrMemory(ctxt); |
1792 | 0 | return(-1); |
1793 | 0 | } |
1794 | 2.57k | memset(newHash, 0, newSize * sizeof(newHash[0])); |
1795 | | |
1796 | 118k | for (i = 0; i < ctxt->nsdb->hashSize; i++) { |
1797 | 116k | unsigned hv = ctxt->nsdb->hash[i].hashValue; |
1798 | 116k | unsigned newIndex; |
1799 | | |
1800 | 116k | if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX)) |
1801 | 113k | continue; |
1802 | 2.66k | newIndex = hv & (newSize - 1); |
1803 | | |
1804 | 3.07k | while (newHash[newIndex].hashValue != 0) { |
1805 | 407 | newIndex++; |
1806 | 407 | if (newIndex == newSize) |
1807 | 58 | newIndex = 0; |
1808 | 407 | } |
1809 | | |
1810 | 2.66k | newHash[newIndex] = ctxt->nsdb->hash[i]; |
1811 | 2.66k | } |
1812 | | |
1813 | 2.57k | xmlFree(ctxt->nsdb->hash); |
1814 | 2.57k | ctxt->nsdb->hash = newHash; |
1815 | 2.57k | ctxt->nsdb->hashSize = newSize; |
1816 | | |
1817 | | /* |
1818 | | * Relookup |
1819 | | */ |
1820 | 2.57k | index = hashValue & (newSize - 1); |
1821 | | |
1822 | 2.79k | while (newHash[index].hashValue != 0) { |
1823 | 217 | index++; |
1824 | 217 | if (index == newSize) |
1825 | 31 | index = 0; |
1826 | 217 | } |
1827 | | |
1828 | 2.57k | bucket = &newHash[index]; |
1829 | 2.57k | } |
1830 | | |
1831 | 42.9k | bucket->hashValue = hashValue; |
1832 | 42.9k | bucket->index = ctxt->nsNr; |
1833 | 42.9k | ctxt->nsdb->hashElems++; |
1834 | 42.9k | oldIndex = INT_MAX; |
1835 | | |
1836 | 185k | populate_entry: |
1837 | 185k | nsIndex = ctxt->nsNr; |
1838 | | |
1839 | 185k | ns = &ctxt->nsTab[nsIndex * 2]; |
1840 | 185k | ns[0] = prefix ? prefix->name : NULL; |
1841 | 185k | ns[1] = uri->name; |
1842 | | |
1843 | 185k | extra = &ctxt->nsdb->extra[nsIndex]; |
1844 | 185k | extra->saxData = saxData; |
1845 | 185k | extra->prefixHashValue = prefix ? prefix->hashValue : 0; |
1846 | 185k | extra->uriHashValue = uri->hashValue; |
1847 | 185k | extra->elementId = ctxt->nsdb->elementId; |
1848 | 185k | extra->oldIndex = oldIndex; |
1849 | | |
1850 | 185k | ctxt->nsNr++; |
1851 | | |
1852 | 185k | return(1); |
1853 | 42.9k | } |
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 | 123k | { |
1867 | 123k | int i; |
1868 | | |
1869 | | /* assert(nr <= ctxt->nsNr); */ |
1870 | | |
1871 | 306k | for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) { |
1872 | 182k | const xmlChar *prefix = ctxt->nsTab[i * 2]; |
1873 | 182k | xmlParserNsExtra *extra = &ctxt->nsdb->extra[i]; |
1874 | | |
1875 | 182k | if (prefix == NULL) { |
1876 | 79.8k | ctxt->nsdb->defaultNsIndex = extra->oldIndex; |
1877 | 103k | } else { |
1878 | 103k | xmlHashedString hprefix; |
1879 | 103k | xmlParserNsBucket *bucket = NULL; |
1880 | | |
1881 | 103k | hprefix.name = prefix; |
1882 | 103k | hprefix.hashValue = extra->prefixHashValue; |
1883 | 103k | xmlParserNsLookup(ctxt, &hprefix, &bucket); |
1884 | | /* assert(bucket && bucket->hashValue); */ |
1885 | 103k | bucket->index = extra->oldIndex; |
1886 | 103k | } |
1887 | 182k | } |
1888 | | |
1889 | 123k | ctxt->nsNr -= nr; |
1890 | 123k | return(nr); |
1891 | 123k | } |
1892 | | |
1893 | | static int |
1894 | 3.80k | xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt, int nr) { |
1895 | 3.80k | const xmlChar **atts; |
1896 | 3.80k | unsigned *attallocs; |
1897 | 3.80k | int maxatts; |
1898 | | |
1899 | 3.80k | if (nr + 5 > ctxt->maxatts) { |
1900 | 3.80k | maxatts = ctxt->maxatts == 0 ? 55 : (nr + 5) * 2; |
1901 | 3.80k | atts = (const xmlChar **) xmlMalloc( |
1902 | 3.80k | maxatts * sizeof(const xmlChar *)); |
1903 | 3.80k | if (atts == NULL) goto mem_error; |
1904 | 3.80k | attallocs = xmlRealloc(ctxt->attallocs, |
1905 | 3.80k | (maxatts / 5) * sizeof(attallocs[0])); |
1906 | 3.80k | if (attallocs == NULL) { |
1907 | 0 | xmlFree(atts); |
1908 | 0 | goto mem_error; |
1909 | 0 | } |
1910 | 3.80k | if (ctxt->maxatts > 0) |
1911 | 203 | memcpy(atts, ctxt->atts, ctxt->maxatts * sizeof(const xmlChar *)); |
1912 | 3.80k | xmlFree(ctxt->atts); |
1913 | 3.80k | ctxt->atts = atts; |
1914 | 3.80k | ctxt->attallocs = attallocs; |
1915 | 3.80k | ctxt->maxatts = maxatts; |
1916 | 3.80k | } |
1917 | 3.80k | return(ctxt->maxatts); |
1918 | 0 | mem_error: |
1919 | 0 | xmlErrMemory(ctxt); |
1920 | 0 | return(-1); |
1921 | 3.80k | } |
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 | 38.7k | { |
1935 | 38.7k | char *directory = NULL; |
1936 | | |
1937 | 38.7k | if ((ctxt == NULL) || (value == NULL)) |
1938 | 0 | return(-1); |
1939 | | |
1940 | 38.7k | if (ctxt->inputNr >= ctxt->inputMax) { |
1941 | 1.85k | size_t newSize = ctxt->inputMax * 2; |
1942 | 1.85k | xmlParserInputPtr *tmp; |
1943 | | |
1944 | 1.85k | tmp = (xmlParserInputPtr *) xmlRealloc(ctxt->inputTab, |
1945 | 1.85k | newSize * sizeof(*tmp)); |
1946 | 1.85k | if (tmp == NULL) { |
1947 | 0 | xmlErrMemory(ctxt); |
1948 | 0 | return (-1); |
1949 | 0 | } |
1950 | 1.85k | ctxt->inputTab = tmp; |
1951 | 1.85k | ctxt->inputMax = newSize; |
1952 | 1.85k | } |
1953 | | |
1954 | 38.7k | if ((ctxt->inputNr == 0) && (value->filename != NULL)) { |
1955 | 9.89k | directory = xmlParserGetDirectory(value->filename); |
1956 | 9.89k | if (directory == NULL) { |
1957 | 0 | xmlErrMemory(ctxt); |
1958 | 0 | return(-1); |
1959 | 0 | } |
1960 | 9.89k | } |
1961 | | |
1962 | 38.7k | if (ctxt->input_id >= INT_MAX) { |
1963 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, "Input ID overflow\n"); |
1964 | 0 | return(-1); |
1965 | 0 | } |
1966 | | |
1967 | 38.7k | ctxt->inputTab[ctxt->inputNr] = value; |
1968 | 38.7k | ctxt->input = value; |
1969 | | |
1970 | 38.7k | if (ctxt->inputNr == 0) { |
1971 | 20.2k | xmlFree(ctxt->directory); |
1972 | 20.2k | ctxt->directory = directory; |
1973 | 20.2k | } |
1974 | | |
1975 | | /* |
1976 | | * Internally, the input ID is only used to detect parameter entity |
1977 | | * boundaries. But there are entity loaders in downstream code that |
1978 | | * detect the main document by checking for "input_id == 1". |
1979 | | */ |
1980 | 38.7k | value->id = ctxt->input_id++; |
1981 | | |
1982 | 38.7k | return(ctxt->inputNr++); |
1983 | 38.7k | } |
1984 | | /** |
1985 | | * inputPop: |
1986 | | * @ctxt: an XML parser context |
1987 | | * |
1988 | | * Pops the top parser input from the input stack |
1989 | | * |
1990 | | * Returns the input just removed |
1991 | | */ |
1992 | | xmlParserInputPtr |
1993 | | inputPop(xmlParserCtxtPtr ctxt) |
1994 | 79.1k | { |
1995 | 79.1k | xmlParserInputPtr ret; |
1996 | | |
1997 | 79.1k | if (ctxt == NULL) |
1998 | 0 | return(NULL); |
1999 | 79.1k | if (ctxt->inputNr <= 0) |
2000 | 40.4k | return (NULL); |
2001 | 38.7k | ctxt->inputNr--; |
2002 | 38.7k | if (ctxt->inputNr > 0) |
2003 | 18.5k | ctxt->input = ctxt->inputTab[ctxt->inputNr - 1]; |
2004 | 20.2k | else |
2005 | 20.2k | ctxt->input = NULL; |
2006 | 38.7k | ret = ctxt->inputTab[ctxt->inputNr]; |
2007 | 38.7k | ctxt->inputTab[ctxt->inputNr] = NULL; |
2008 | 38.7k | return (ret); |
2009 | 79.1k | } |
2010 | | /** |
2011 | | * nodePush: |
2012 | | * @ctxt: an XML parser context |
2013 | | * @value: the element node |
2014 | | * |
2015 | | * DEPRECATED: Internal function, do not use. |
2016 | | * |
2017 | | * Pushes a new element node on top of the node stack |
2018 | | * |
2019 | | * Returns -1 in case of error, the index in the stack otherwise |
2020 | | */ |
2021 | | int |
2022 | | nodePush(xmlParserCtxtPtr ctxt, xmlNodePtr value) |
2023 | 43.8k | { |
2024 | 43.8k | int maxDepth; |
2025 | | |
2026 | 43.8k | if (ctxt == NULL) |
2027 | 0 | return(0); |
2028 | | |
2029 | 43.8k | maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
2030 | 43.8k | if (ctxt->nodeNr > maxDepth) { |
2031 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
2032 | 0 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
2033 | 0 | ctxt->nodeNr); |
2034 | 0 | xmlHaltParser(ctxt); |
2035 | 0 | return(-1); |
2036 | 0 | } |
2037 | 43.8k | if (ctxt->nodeNr >= ctxt->nodeMax) { |
2038 | 7.19k | xmlNodePtr *tmp; |
2039 | | |
2040 | 7.19k | tmp = (xmlNodePtr *) xmlRealloc(ctxt->nodeTab, |
2041 | 7.19k | ctxt->nodeMax * 2 * |
2042 | 7.19k | sizeof(ctxt->nodeTab[0])); |
2043 | 7.19k | if (tmp == NULL) { |
2044 | 0 | xmlErrMemory(ctxt); |
2045 | 0 | return (-1); |
2046 | 0 | } |
2047 | 7.19k | ctxt->nodeTab = tmp; |
2048 | 7.19k | ctxt->nodeMax *= 2; |
2049 | 7.19k | } |
2050 | 43.8k | ctxt->nodeTab[ctxt->nodeNr] = value; |
2051 | 43.8k | ctxt->node = value; |
2052 | 43.8k | return (ctxt->nodeNr++); |
2053 | 43.8k | } |
2054 | | |
2055 | | /** |
2056 | | * nodePop: |
2057 | | * @ctxt: an XML parser context |
2058 | | * |
2059 | | * DEPRECATED: Internal function, do not use. |
2060 | | * |
2061 | | * Pops the top element node from the node stack |
2062 | | * |
2063 | | * Returns the node just removed |
2064 | | */ |
2065 | | xmlNodePtr |
2066 | | nodePop(xmlParserCtxtPtr ctxt) |
2067 | 254k | { |
2068 | 254k | xmlNodePtr ret; |
2069 | | |
2070 | 254k | if (ctxt == NULL) return(NULL); |
2071 | 254k | if (ctxt->nodeNr <= 0) |
2072 | 213k | return (NULL); |
2073 | 40.4k | ctxt->nodeNr--; |
2074 | 40.4k | if (ctxt->nodeNr > 0) |
2075 | 37.3k | ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1]; |
2076 | 3.03k | else |
2077 | 3.03k | ctxt->node = NULL; |
2078 | 40.4k | ret = ctxt->nodeTab[ctxt->nodeNr]; |
2079 | 40.4k | ctxt->nodeTab[ctxt->nodeNr] = NULL; |
2080 | 40.4k | return (ret); |
2081 | 254k | } |
2082 | | |
2083 | | /** |
2084 | | * nameNsPush: |
2085 | | * @ctxt: an XML parser context |
2086 | | * @value: the element name |
2087 | | * @prefix: the element prefix |
2088 | | * @URI: the element namespace name |
2089 | | * @line: the current line number for error messages |
2090 | | * @nsNr: the number of namespaces pushed on the namespace table |
2091 | | * |
2092 | | * Pushes a new element name/prefix/URL on top of the name stack |
2093 | | * |
2094 | | * Returns -1 in case of error, the index in the stack otherwise |
2095 | | */ |
2096 | | static int |
2097 | | nameNsPush(xmlParserCtxtPtr ctxt, const xmlChar * value, |
2098 | | const xmlChar *prefix, const xmlChar *URI, int line, int nsNr) |
2099 | 449k | { |
2100 | 449k | xmlStartTag *tag; |
2101 | | |
2102 | 449k | if (ctxt->nameNr >= ctxt->nameMax) { |
2103 | 19.6k | const xmlChar * *tmp; |
2104 | 19.6k | xmlStartTag *tmp2; |
2105 | 19.6k | ctxt->nameMax *= 2; |
2106 | 19.6k | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
2107 | 19.6k | ctxt->nameMax * |
2108 | 19.6k | sizeof(ctxt->nameTab[0])); |
2109 | 19.6k | if (tmp == NULL) { |
2110 | 0 | ctxt->nameMax /= 2; |
2111 | 0 | goto mem_error; |
2112 | 0 | } |
2113 | 19.6k | ctxt->nameTab = tmp; |
2114 | 19.6k | tmp2 = (xmlStartTag *) xmlRealloc((void * *)ctxt->pushTab, |
2115 | 19.6k | ctxt->nameMax * |
2116 | 19.6k | sizeof(ctxt->pushTab[0])); |
2117 | 19.6k | if (tmp2 == NULL) { |
2118 | 0 | ctxt->nameMax /= 2; |
2119 | 0 | goto mem_error; |
2120 | 0 | } |
2121 | 19.6k | ctxt->pushTab = tmp2; |
2122 | 429k | } else if (ctxt->pushTab == NULL) { |
2123 | 9.46k | ctxt->pushTab = (xmlStartTag *) xmlMalloc(ctxt->nameMax * |
2124 | 9.46k | sizeof(ctxt->pushTab[0])); |
2125 | 9.46k | if (ctxt->pushTab == NULL) |
2126 | 0 | goto mem_error; |
2127 | 9.46k | } |
2128 | 449k | ctxt->nameTab[ctxt->nameNr] = value; |
2129 | 449k | ctxt->name = value; |
2130 | 449k | tag = &ctxt->pushTab[ctxt->nameNr]; |
2131 | 449k | tag->prefix = prefix; |
2132 | 449k | tag->URI = URI; |
2133 | 449k | tag->line = line; |
2134 | 449k | tag->nsNr = nsNr; |
2135 | 449k | return (ctxt->nameNr++); |
2136 | 0 | mem_error: |
2137 | 0 | xmlErrMemory(ctxt); |
2138 | 0 | return (-1); |
2139 | 449k | } |
2140 | | #ifdef LIBXML_PUSH_ENABLED |
2141 | | /** |
2142 | | * nameNsPop: |
2143 | | * @ctxt: an XML parser context |
2144 | | * |
2145 | | * Pops the top element/prefix/URI name from the name stack |
2146 | | * |
2147 | | * Returns the name just removed |
2148 | | */ |
2149 | | static const xmlChar * |
2150 | | nameNsPop(xmlParserCtxtPtr ctxt) |
2151 | 0 | { |
2152 | 0 | const xmlChar *ret; |
2153 | |
|
2154 | 0 | if (ctxt->nameNr <= 0) |
2155 | 0 | return (NULL); |
2156 | 0 | ctxt->nameNr--; |
2157 | 0 | if (ctxt->nameNr > 0) |
2158 | 0 | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
2159 | 0 | else |
2160 | 0 | ctxt->name = NULL; |
2161 | 0 | ret = ctxt->nameTab[ctxt->nameNr]; |
2162 | 0 | ctxt->nameTab[ctxt->nameNr] = NULL; |
2163 | 0 | return (ret); |
2164 | 0 | } |
2165 | | #endif /* LIBXML_PUSH_ENABLED */ |
2166 | | |
2167 | | /** |
2168 | | * namePush: |
2169 | | * @ctxt: an XML parser context |
2170 | | * @value: the element name |
2171 | | * |
2172 | | * DEPRECATED: Internal function, do not use. |
2173 | | * |
2174 | | * Pushes a new element name on top of the name stack |
2175 | | * |
2176 | | * Returns -1 in case of error, the index in the stack otherwise |
2177 | | */ |
2178 | | int |
2179 | | namePush(xmlParserCtxtPtr ctxt, const xmlChar * value) |
2180 | 0 | { |
2181 | 0 | if (ctxt == NULL) return (-1); |
2182 | | |
2183 | 0 | if (ctxt->nameNr >= ctxt->nameMax) { |
2184 | 0 | const xmlChar * *tmp; |
2185 | 0 | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
2186 | 0 | ctxt->nameMax * 2 * |
2187 | 0 | sizeof(ctxt->nameTab[0])); |
2188 | 0 | if (tmp == NULL) { |
2189 | 0 | goto mem_error; |
2190 | 0 | } |
2191 | 0 | ctxt->nameTab = tmp; |
2192 | 0 | ctxt->nameMax *= 2; |
2193 | 0 | } |
2194 | 0 | ctxt->nameTab[ctxt->nameNr] = value; |
2195 | 0 | ctxt->name = value; |
2196 | 0 | return (ctxt->nameNr++); |
2197 | 0 | mem_error: |
2198 | 0 | xmlErrMemory(ctxt); |
2199 | 0 | return (-1); |
2200 | 0 | } |
2201 | | |
2202 | | /** |
2203 | | * namePop: |
2204 | | * @ctxt: an XML parser context |
2205 | | * |
2206 | | * DEPRECATED: Internal function, do not use. |
2207 | | * |
2208 | | * Pops the top element name from the name stack |
2209 | | * |
2210 | | * Returns the name just removed |
2211 | | */ |
2212 | | const xmlChar * |
2213 | | namePop(xmlParserCtxtPtr ctxt) |
2214 | 441k | { |
2215 | 441k | const xmlChar *ret; |
2216 | | |
2217 | 441k | if ((ctxt == NULL) || (ctxt->nameNr <= 0)) |
2218 | 0 | return (NULL); |
2219 | 441k | ctxt->nameNr--; |
2220 | 441k | if (ctxt->nameNr > 0) |
2221 | 439k | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
2222 | 2.23k | else |
2223 | 2.23k | ctxt->name = NULL; |
2224 | 441k | ret = ctxt->nameTab[ctxt->nameNr]; |
2225 | 441k | ctxt->nameTab[ctxt->nameNr] = NULL; |
2226 | 441k | return (ret); |
2227 | 441k | } |
2228 | | |
2229 | 666k | static int spacePush(xmlParserCtxtPtr ctxt, int val) { |
2230 | 666k | if (ctxt->spaceNr >= ctxt->spaceMax) { |
2231 | 26.0k | int *tmp; |
2232 | | |
2233 | 26.0k | ctxt->spaceMax *= 2; |
2234 | 26.0k | tmp = (int *) xmlRealloc(ctxt->spaceTab, |
2235 | 26.0k | ctxt->spaceMax * sizeof(ctxt->spaceTab[0])); |
2236 | 26.0k | if (tmp == NULL) { |
2237 | 0 | xmlErrMemory(ctxt); |
2238 | 0 | ctxt->spaceMax /=2; |
2239 | 0 | return(-1); |
2240 | 0 | } |
2241 | 26.0k | ctxt->spaceTab = tmp; |
2242 | 26.0k | } |
2243 | 666k | ctxt->spaceTab[ctxt->spaceNr] = val; |
2244 | 666k | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr]; |
2245 | 666k | return(ctxt->spaceNr++); |
2246 | 666k | } |
2247 | | |
2248 | 659k | static int spacePop(xmlParserCtxtPtr ctxt) { |
2249 | 659k | int ret; |
2250 | 659k | if (ctxt->spaceNr <= 0) return(0); |
2251 | 659k | ctxt->spaceNr--; |
2252 | 659k | if (ctxt->spaceNr > 0) |
2253 | 659k | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1]; |
2254 | 0 | else |
2255 | 0 | ctxt->space = &ctxt->spaceTab[0]; |
2256 | 659k | ret = ctxt->spaceTab[ctxt->spaceNr]; |
2257 | 659k | ctxt->spaceTab[ctxt->spaceNr] = -1; |
2258 | 659k | return(ret); |
2259 | 659k | } |
2260 | | |
2261 | | /* |
2262 | | * Macros for accessing the content. Those should be used only by the parser, |
2263 | | * and not exported. |
2264 | | * |
2265 | | * Dirty macros, i.e. one often need to make assumption on the context to |
2266 | | * use them |
2267 | | * |
2268 | | * CUR_PTR return the current pointer to the xmlChar to be parsed. |
2269 | | * To be used with extreme caution since operations consuming |
2270 | | * characters may move the input buffer to a different location ! |
2271 | | * CUR returns the current xmlChar value, i.e. a 8 bit value if compiled |
2272 | | * This should be used internally by the parser |
2273 | | * only to compare to ASCII values otherwise it would break when |
2274 | | * running with UTF-8 encoding. |
2275 | | * RAW same as CUR but in the input buffer, bypass any token |
2276 | | * extraction that may have been done |
2277 | | * NXT(n) returns the n'th next xmlChar. Same as CUR is should be used only |
2278 | | * to compare on ASCII based substring. |
2279 | | * SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined |
2280 | | * strings without newlines within the parser. |
2281 | | * NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII |
2282 | | * defined char within the parser. |
2283 | | * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding |
2284 | | * |
2285 | | * NEXT Skip to the next character, this does the proper decoding |
2286 | | * in UTF-8 mode. It also pop-up unfinished entities on the fly. |
2287 | | * NEXTL(l) Skip the current unicode character of l xmlChars long. |
2288 | | * CUR_SCHAR same but operate on a string instead of the context |
2289 | | * COPY_BUF copy the current unicode char to the target buffer, increment |
2290 | | * the index |
2291 | | * GROW, SHRINK handling of input buffers |
2292 | | */ |
2293 | | |
2294 | 11.2M | #define RAW (*ctxt->input->cur) |
2295 | 744M | #define CUR (*ctxt->input->cur) |
2296 | 3.95M | #define NXT(val) ctxt->input->cur[(val)] |
2297 | 1.73G | #define CUR_PTR ctxt->input->cur |
2298 | 3.92M | #define BASE_PTR ctxt->input->base |
2299 | | |
2300 | | #define CMP4( s, c1, c2, c3, c4 ) \ |
2301 | 22.1M | ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \ |
2302 | 11.0M | ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 ) |
2303 | | #define CMP5( s, c1, c2, c3, c4, c5 ) \ |
2304 | 21.9M | ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 ) |
2305 | | #define CMP6( s, c1, c2, c3, c4, c5, c6 ) \ |
2306 | 21.6M | ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 ) |
2307 | | #define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \ |
2308 | 21.3M | ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 ) |
2309 | | #define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \ |
2310 | 21.2M | ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 ) |
2311 | | #define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \ |
2312 | 10.5M | ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \ |
2313 | 10.5M | ((unsigned char *) s)[ 8 ] == c9 ) |
2314 | | #define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \ |
2315 | 1.81k | ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \ |
2316 | 1.81k | ((unsigned char *) s)[ 9 ] == c10 ) |
2317 | | |
2318 | 639k | #define SKIP(val) do { \ |
2319 | 639k | ctxt->input->cur += (val),ctxt->input->col+=(val); \ |
2320 | 639k | if (*ctxt->input->cur == 0) \ |
2321 | 639k | xmlParserGrow(ctxt); \ |
2322 | 639k | } while (0) |
2323 | | |
2324 | | #define SKIPL(val) do { \ |
2325 | | int skipl; \ |
2326 | | for(skipl=0; skipl<val; skipl++) { \ |
2327 | | if (*(ctxt->input->cur) == '\n') { \ |
2328 | | ctxt->input->line++; ctxt->input->col = 1; \ |
2329 | | } else ctxt->input->col++; \ |
2330 | | ctxt->input->cur++; \ |
2331 | | } \ |
2332 | | if (*ctxt->input->cur == 0) \ |
2333 | | xmlParserGrow(ctxt); \ |
2334 | | } while (0) |
2335 | | |
2336 | | #define SHRINK \ |
2337 | 20.8M | if ((!PARSER_PROGRESSIVE(ctxt)) && \ |
2338 | 20.8M | (ctxt->input->cur - ctxt->input->base > 2 * INPUT_CHUNK) && \ |
2339 | 20.8M | (ctxt->input->end - ctxt->input->cur < 2 * INPUT_CHUNK)) \ |
2340 | 20.8M | xmlParserShrink(ctxt); |
2341 | | |
2342 | | #define GROW \ |
2343 | 37.3M | if ((!PARSER_PROGRESSIVE(ctxt)) && \ |
2344 | 37.3M | (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \ |
2345 | 1.09M | xmlParserGrow(ctxt); |
2346 | | |
2347 | 2.58M | #define SKIP_BLANKS xmlSkipBlankChars(ctxt) |
2348 | | |
2349 | 636k | #define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt) |
2350 | | |
2351 | 5.26M | #define NEXT xmlNextChar(ctxt) |
2352 | | |
2353 | 909k | #define NEXT1 { \ |
2354 | 909k | ctxt->input->col++; \ |
2355 | 909k | ctxt->input->cur++; \ |
2356 | 909k | if (*ctxt->input->cur == 0) \ |
2357 | 909k | xmlParserGrow(ctxt); \ |
2358 | 909k | } |
2359 | | |
2360 | 950M | #define NEXTL(l) do { \ |
2361 | 950M | if (*(ctxt->input->cur) == '\n') { \ |
2362 | 11.1M | ctxt->input->line++; ctxt->input->col = 1; \ |
2363 | 939M | } else ctxt->input->col++; \ |
2364 | 950M | ctxt->input->cur += l; \ |
2365 | 950M | } while (0) |
2366 | | |
2367 | 6.91M | #define CUR_SCHAR(s, l) xmlStringCurrentChar(ctxt, s, &l) |
2368 | | |
2369 | | #define COPY_BUF(b, i, v) \ |
2370 | 171M | if (v < 0x80) b[i++] = v; \ |
2371 | 171M | else i += xmlCopyCharMultiByte(&b[i],v) |
2372 | | |
2373 | | static int |
2374 | 171M | xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) { |
2375 | 171M | int c = xmlCurrentChar(ctxt, len); |
2376 | | |
2377 | 171M | if (c == XML_INVALID_CHAR) |
2378 | 8.48M | c = 0xFFFD; /* replacement character */ |
2379 | | |
2380 | 171M | return(c); |
2381 | 171M | } |
2382 | | |
2383 | | /** |
2384 | | * xmlSkipBlankChars: |
2385 | | * @ctxt: the XML parser context |
2386 | | * |
2387 | | * DEPRECATED: Internal function, do not use. |
2388 | | * |
2389 | | * Skip whitespace in the input stream. |
2390 | | * |
2391 | | * Returns the number of space chars skipped |
2392 | | */ |
2393 | | int |
2394 | 2.97M | xmlSkipBlankChars(xmlParserCtxtPtr ctxt) { |
2395 | 2.97M | const xmlChar *cur; |
2396 | 2.97M | int res = 0; |
2397 | | |
2398 | | /* |
2399 | | * It's Okay to use CUR/NEXT here since all the blanks are on |
2400 | | * the ASCII range. |
2401 | | */ |
2402 | 2.97M | cur = ctxt->input->cur; |
2403 | 10.0M | while (IS_BLANK_CH(*cur)) { |
2404 | 10.0M | if (*cur == '\n') { |
2405 | 2.21M | ctxt->input->line++; ctxt->input->col = 1; |
2406 | 7.82M | } else { |
2407 | 7.82M | ctxt->input->col++; |
2408 | 7.82M | } |
2409 | 10.0M | cur++; |
2410 | 10.0M | if (res < INT_MAX) |
2411 | 10.0M | res++; |
2412 | 10.0M | if (*cur == 0) { |
2413 | 5.22k | ctxt->input->cur = cur; |
2414 | 5.22k | xmlParserGrow(ctxt); |
2415 | 5.22k | cur = ctxt->input->cur; |
2416 | 5.22k | } |
2417 | 10.0M | } |
2418 | 2.97M | ctxt->input->cur = cur; |
2419 | | |
2420 | 2.97M | return(res); |
2421 | 2.97M | } |
2422 | | |
2423 | | static void |
2424 | 17.6k | xmlPopPE(xmlParserCtxtPtr ctxt) { |
2425 | 17.6k | unsigned long consumed; |
2426 | 17.6k | xmlEntityPtr ent; |
2427 | | |
2428 | 17.6k | ent = ctxt->input->entity; |
2429 | | |
2430 | 17.6k | ent->flags &= ~XML_ENT_EXPANDING; |
2431 | | |
2432 | 17.6k | if ((ent->flags & XML_ENT_CHECKED) == 0) { |
2433 | 1.45k | int result; |
2434 | | |
2435 | | /* |
2436 | | * Read the rest of the stream in case of errors. We want |
2437 | | * to account for the whole entity size. |
2438 | | */ |
2439 | 1.45k | do { |
2440 | 1.45k | ctxt->input->cur = ctxt->input->end; |
2441 | 1.45k | xmlParserShrink(ctxt); |
2442 | 1.45k | result = xmlParserGrow(ctxt); |
2443 | 1.45k | } while (result > 0); |
2444 | | |
2445 | 1.45k | consumed = ctxt->input->consumed; |
2446 | 1.45k | xmlSaturatedAddSizeT(&consumed, |
2447 | 1.45k | ctxt->input->end - ctxt->input->base); |
2448 | | |
2449 | 1.45k | xmlSaturatedAdd(&ent->expandedSize, consumed); |
2450 | | |
2451 | | /* |
2452 | | * Add to sizeentities when parsing an external entity |
2453 | | * for the first time. |
2454 | | */ |
2455 | 1.45k | if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
2456 | 0 | xmlSaturatedAdd(&ctxt->sizeentities, consumed); |
2457 | 0 | } |
2458 | | |
2459 | 1.45k | ent->flags |= XML_ENT_CHECKED; |
2460 | 1.45k | } |
2461 | | |
2462 | 17.6k | xmlPopInput(ctxt); |
2463 | | |
2464 | 17.6k | xmlParserEntityCheck(ctxt, ent->expandedSize); |
2465 | 17.6k | } |
2466 | | |
2467 | | /** |
2468 | | * xmlSkipBlankCharsPE: |
2469 | | * @ctxt: the XML parser context |
2470 | | * |
2471 | | * Skip whitespace in the input stream, also handling parameter |
2472 | | * entities. |
2473 | | * |
2474 | | * Returns the number of space chars skipped |
2475 | | */ |
2476 | | static int |
2477 | 636k | xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) { |
2478 | 636k | int res = 0; |
2479 | 636k | int inParam; |
2480 | 636k | int expandParam; |
2481 | | |
2482 | 636k | inParam = PARSER_IN_PE(ctxt); |
2483 | 636k | expandParam = PARSER_EXTERNAL(ctxt); |
2484 | | |
2485 | 636k | if (!inParam && !expandParam) |
2486 | 395k | return(xmlSkipBlankChars(ctxt)); |
2487 | | |
2488 | 450k | while (PARSER_STOPPED(ctxt) == 0) { |
2489 | 450k | if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */ |
2490 | 192k | NEXT; |
2491 | 258k | } else if (CUR == '%') { |
2492 | 16.2k | if ((expandParam == 0) || |
2493 | 16.2k | (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0)) |
2494 | 16.2k | break; |
2495 | | |
2496 | | /* |
2497 | | * Expand parameter entity. We continue to consume |
2498 | | * whitespace at the start of the entity and possible |
2499 | | * even consume the whole entity and pop it. We might |
2500 | | * even pop multiple PEs in this loop. |
2501 | | */ |
2502 | 0 | xmlParsePEReference(ctxt); |
2503 | |
|
2504 | 0 | inParam = PARSER_IN_PE(ctxt); |
2505 | 0 | expandParam = PARSER_EXTERNAL(ctxt); |
2506 | 242k | } else if (CUR == 0) { |
2507 | 17.9k | if (inParam == 0) |
2508 | 306 | break; |
2509 | | |
2510 | 17.6k | xmlPopPE(ctxt); |
2511 | | |
2512 | 17.6k | inParam = PARSER_IN_PE(ctxt); |
2513 | 17.6k | expandParam = PARSER_EXTERNAL(ctxt); |
2514 | 224k | } else { |
2515 | 224k | break; |
2516 | 224k | } |
2517 | | |
2518 | | /* |
2519 | | * Also increase the counter when entering or exiting a PERef. |
2520 | | * The spec says: "When a parameter-entity reference is recognized |
2521 | | * in the DTD and included, its replacement text MUST be enlarged |
2522 | | * by the attachment of one leading and one following space (#x20) |
2523 | | * character." |
2524 | | */ |
2525 | 209k | if (res < INT_MAX) |
2526 | 209k | res++; |
2527 | 209k | } |
2528 | | |
2529 | 240k | return(res); |
2530 | 636k | } |
2531 | | |
2532 | | /************************************************************************ |
2533 | | * * |
2534 | | * Commodity functions to handle entities * |
2535 | | * * |
2536 | | ************************************************************************/ |
2537 | | |
2538 | | /** |
2539 | | * xmlPopInput: |
2540 | | * @ctxt: an XML parser context |
2541 | | * |
2542 | | * xmlPopInput: the current input pointed by ctxt->input came to an end |
2543 | | * pop it and return the next char. |
2544 | | * |
2545 | | * Returns the current xmlChar in the parser context |
2546 | | */ |
2547 | | xmlChar |
2548 | 17.6k | xmlPopInput(xmlParserCtxtPtr ctxt) { |
2549 | 17.6k | xmlParserInputPtr input; |
2550 | | |
2551 | 17.6k | if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0); |
2552 | 17.6k | input = inputPop(ctxt); |
2553 | 17.6k | xmlFreeInputStream(input); |
2554 | 17.6k | if (*ctxt->input->cur == 0) |
2555 | 310 | xmlParserGrow(ctxt); |
2556 | 17.6k | return(CUR); |
2557 | 17.6k | } |
2558 | | |
2559 | | /** |
2560 | | * xmlPushInput: |
2561 | | * @ctxt: an XML parser context |
2562 | | * @input: an XML parser input fragment (entity, XML fragment ...). |
2563 | | * |
2564 | | * Push an input stream onto the stack. |
2565 | | * |
2566 | | * Returns -1 in case of error or the index in the input stack |
2567 | | */ |
2568 | | int |
2569 | 18.5k | xmlPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) { |
2570 | 18.5k | int maxDepth; |
2571 | 18.5k | int ret; |
2572 | | |
2573 | 18.5k | if ((ctxt == NULL) || (input == NULL)) |
2574 | 0 | return(-1); |
2575 | | |
2576 | 18.5k | maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
2577 | 18.5k | if (ctxt->inputNr > maxDepth) { |
2578 | 2 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
2579 | 2 | "Maximum entity nesting depth exceeded"); |
2580 | 2 | xmlHaltParser(ctxt); |
2581 | 2 | return(-1); |
2582 | 2 | } |
2583 | 18.5k | ret = inputPush(ctxt, input); |
2584 | 18.5k | if (ret >= 0) |
2585 | 18.5k | GROW; |
2586 | 18.5k | return(ret); |
2587 | 18.5k | } |
2588 | | |
2589 | | /** |
2590 | | * xmlParseCharRef: |
2591 | | * @ctxt: an XML parser context |
2592 | | * |
2593 | | * DEPRECATED: Internal function, don't use. |
2594 | | * |
2595 | | * Parse a numeric character reference. Always consumes '&'. |
2596 | | * |
2597 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2598 | | * '&#x' [0-9a-fA-F]+ ';' |
2599 | | * |
2600 | | * [ WFC: Legal Character ] |
2601 | | * Characters referred to using character references must match the |
2602 | | * production for Char. |
2603 | | * |
2604 | | * Returns the value parsed (as an int), 0 in case of error |
2605 | | */ |
2606 | | int |
2607 | 134k | xmlParseCharRef(xmlParserCtxtPtr ctxt) { |
2608 | 134k | int val = 0; |
2609 | 134k | int count = 0; |
2610 | | |
2611 | | /* |
2612 | | * Using RAW/CUR/NEXT is okay since we are working on ASCII range here |
2613 | | */ |
2614 | 134k | if ((RAW == '&') && (NXT(1) == '#') && |
2615 | 134k | (NXT(2) == 'x')) { |
2616 | 74.5k | SKIP(3); |
2617 | 74.5k | GROW; |
2618 | 364k | while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) { |
2619 | 329k | if (count++ > 20) { |
2620 | 1.77k | count = 0; |
2621 | 1.77k | GROW; |
2622 | 1.77k | } |
2623 | 329k | if ((RAW >= '0') && (RAW <= '9')) |
2624 | 64.2k | val = val * 16 + (CUR - '0'); |
2625 | 264k | else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20)) |
2626 | 18.8k | val = val * 16 + (CUR - 'a') + 10; |
2627 | 245k | else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20)) |
2628 | 207k | val = val * 16 + (CUR - 'A') + 10; |
2629 | 38.8k | else { |
2630 | 38.8k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2631 | 38.8k | val = 0; |
2632 | 38.8k | break; |
2633 | 38.8k | } |
2634 | 290k | if (val > 0x110000) |
2635 | 103k | val = 0x110000; |
2636 | | |
2637 | 290k | NEXT; |
2638 | 290k | count++; |
2639 | 290k | } |
2640 | 74.5k | if (RAW == ';') { |
2641 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2642 | 35.6k | ctxt->input->col++; |
2643 | 35.6k | ctxt->input->cur++; |
2644 | 35.6k | } |
2645 | 74.5k | } else if ((RAW == '&') && (NXT(1) == '#')) { |
2646 | 60.0k | SKIP(2); |
2647 | 60.0k | GROW; |
2648 | 215k | while (RAW != ';') { /* loop blocked by count */ |
2649 | 170k | if (count++ > 20) { |
2650 | 1.34k | count = 0; |
2651 | 1.34k | GROW; |
2652 | 1.34k | } |
2653 | 170k | if ((RAW >= '0') && (RAW <= '9')) |
2654 | 155k | val = val * 10 + (CUR - '0'); |
2655 | 15.6k | else { |
2656 | 15.6k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2657 | 15.6k | val = 0; |
2658 | 15.6k | break; |
2659 | 15.6k | } |
2660 | 155k | if (val > 0x110000) |
2661 | 16.0k | val = 0x110000; |
2662 | | |
2663 | 155k | NEXT; |
2664 | 155k | count++; |
2665 | 155k | } |
2666 | 60.0k | if (RAW == ';') { |
2667 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2668 | 44.3k | ctxt->input->col++; |
2669 | 44.3k | ctxt->input->cur++; |
2670 | 44.3k | } |
2671 | 60.0k | } else { |
2672 | 0 | if (RAW == '&') |
2673 | 0 | SKIP(1); |
2674 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2675 | 0 | } |
2676 | | |
2677 | | /* |
2678 | | * [ WFC: Legal Character ] |
2679 | | * Characters referred to using character references must match the |
2680 | | * production for Char. |
2681 | | */ |
2682 | 134k | if (val >= 0x110000) { |
2683 | 4.46k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2684 | 4.46k | "xmlParseCharRef: character reference out of bounds\n", |
2685 | 4.46k | val); |
2686 | 130k | } else if (IS_CHAR(val)) { |
2687 | 70.7k | return(val); |
2688 | 70.7k | } else { |
2689 | 59.3k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2690 | 59.3k | "xmlParseCharRef: invalid xmlChar value %d\n", |
2691 | 59.3k | val); |
2692 | 59.3k | } |
2693 | 63.8k | return(0); |
2694 | 134k | } |
2695 | | |
2696 | | /** |
2697 | | * xmlParseStringCharRef: |
2698 | | * @ctxt: an XML parser context |
2699 | | * @str: a pointer to an index in the string |
2700 | | * |
2701 | | * parse Reference declarations, variant parsing from a string rather |
2702 | | * than an an input flow. |
2703 | | * |
2704 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2705 | | * '&#x' [0-9a-fA-F]+ ';' |
2706 | | * |
2707 | | * [ WFC: Legal Character ] |
2708 | | * Characters referred to using character references must match the |
2709 | | * production for Char. |
2710 | | * |
2711 | | * Returns the value parsed (as an int), 0 in case of error, str will be |
2712 | | * updated to the current value of the index |
2713 | | */ |
2714 | | static int |
2715 | 131k | xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
2716 | 131k | const xmlChar *ptr; |
2717 | 131k | xmlChar cur; |
2718 | 131k | int val = 0; |
2719 | | |
2720 | 131k | if ((str == NULL) || (*str == NULL)) return(0); |
2721 | 131k | ptr = *str; |
2722 | 131k | cur = *ptr; |
2723 | 131k | if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) { |
2724 | 69.7k | ptr += 3; |
2725 | 69.7k | cur = *ptr; |
2726 | 252k | while (cur != ';') { /* Non input consuming loop */ |
2727 | 190k | if ((cur >= '0') && (cur <= '9')) |
2728 | 68.8k | val = val * 16 + (cur - '0'); |
2729 | 121k | else if ((cur >= 'a') && (cur <= 'f')) |
2730 | 47.1k | val = val * 16 + (cur - 'a') + 10; |
2731 | 74.7k | else if ((cur >= 'A') && (cur <= 'F')) |
2732 | 66.5k | val = val * 16 + (cur - 'A') + 10; |
2733 | 8.18k | else { |
2734 | 8.18k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2735 | 8.18k | val = 0; |
2736 | 8.18k | break; |
2737 | 8.18k | } |
2738 | 182k | if (val > 0x110000) |
2739 | 5.10k | val = 0x110000; |
2740 | | |
2741 | 182k | ptr++; |
2742 | 182k | cur = *ptr; |
2743 | 182k | } |
2744 | 69.7k | if (cur == ';') |
2745 | 61.5k | ptr++; |
2746 | 69.7k | } else if ((cur == '&') && (ptr[1] == '#')){ |
2747 | 61.4k | ptr += 2; |
2748 | 61.4k | cur = *ptr; |
2749 | 301k | while (cur != ';') { /* Non input consuming loops */ |
2750 | 246k | if ((cur >= '0') && (cur <= '9')) |
2751 | 240k | val = val * 10 + (cur - '0'); |
2752 | 6.53k | else { |
2753 | 6.53k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2754 | 6.53k | val = 0; |
2755 | 6.53k | break; |
2756 | 6.53k | } |
2757 | 240k | if (val > 0x110000) |
2758 | 13.8k | val = 0x110000; |
2759 | | |
2760 | 240k | ptr++; |
2761 | 240k | cur = *ptr; |
2762 | 240k | } |
2763 | 61.4k | if (cur == ';') |
2764 | 54.8k | ptr++; |
2765 | 61.4k | } else { |
2766 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2767 | 0 | return(0); |
2768 | 0 | } |
2769 | 131k | *str = ptr; |
2770 | | |
2771 | | /* |
2772 | | * [ WFC: Legal Character ] |
2773 | | * Characters referred to using character references must match the |
2774 | | * production for Char. |
2775 | | */ |
2776 | 131k | if (val >= 0x110000) { |
2777 | 2.67k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2778 | 2.67k | "xmlParseStringCharRef: character reference out of bounds\n", |
2779 | 2.67k | val); |
2780 | 128k | } else if (IS_CHAR(val)) { |
2781 | 111k | return(val); |
2782 | 111k | } else { |
2783 | 16.9k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2784 | 16.9k | "xmlParseStringCharRef: invalid xmlChar value %d\n", |
2785 | 16.9k | val); |
2786 | 16.9k | } |
2787 | 19.6k | return(0); |
2788 | 131k | } |
2789 | | |
2790 | | /** |
2791 | | * xmlParserHandlePEReference: |
2792 | | * @ctxt: the parser context |
2793 | | * |
2794 | | * DEPRECATED: Internal function, do not use. |
2795 | | * |
2796 | | * [69] PEReference ::= '%' Name ';' |
2797 | | * |
2798 | | * [ WFC: No Recursion ] |
2799 | | * A parsed entity must not contain a recursive |
2800 | | * reference to itself, either directly or indirectly. |
2801 | | * |
2802 | | * [ WFC: Entity Declared ] |
2803 | | * In a document without any DTD, a document with only an internal DTD |
2804 | | * subset which contains no parameter entity references, or a document |
2805 | | * with "standalone='yes'", ... ... The declaration of a parameter |
2806 | | * entity must precede any reference to it... |
2807 | | * |
2808 | | * [ VC: Entity Declared ] |
2809 | | * In a document with an external subset or external parameter entities |
2810 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
2811 | | * must precede any reference to it... |
2812 | | * |
2813 | | * [ WFC: In DTD ] |
2814 | | * Parameter-entity references may only appear in the DTD. |
2815 | | * NOTE: misleading but this is handled. |
2816 | | * |
2817 | | * A PEReference may have been detected in the current input stream |
2818 | | * the handling is done accordingly to |
2819 | | * http://www.w3.org/TR/REC-xml#entproc |
2820 | | * i.e. |
2821 | | * - Included in literal in entity values |
2822 | | * - Included as Parameter Entity reference within DTDs |
2823 | | */ |
2824 | | void |
2825 | 0 | xmlParserHandlePEReference(xmlParserCtxtPtr ctxt) { |
2826 | 0 | xmlParsePEReference(ctxt); |
2827 | 0 | } |
2828 | | |
2829 | | /** |
2830 | | * xmlStringLenDecodeEntities: |
2831 | | * @ctxt: the parser context |
2832 | | * @str: the input string |
2833 | | * @len: the string length |
2834 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2835 | | * @end: an end marker xmlChar, 0 if none |
2836 | | * @end2: an end marker xmlChar, 0 if none |
2837 | | * @end3: an end marker xmlChar, 0 if none |
2838 | | * |
2839 | | * DEPRECATED: Internal function, don't use. |
2840 | | * |
2841 | | * Returns A newly allocated string with the substitution done. The caller |
2842 | | * must deallocate it ! |
2843 | | */ |
2844 | | xmlChar * |
2845 | | xmlStringLenDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2846 | | int what ATTRIBUTE_UNUSED, |
2847 | 0 | xmlChar end, xmlChar end2, xmlChar end3) { |
2848 | 0 | if ((ctxt == NULL) || (str == NULL) || (len < 0)) |
2849 | 0 | return(NULL); |
2850 | | |
2851 | 0 | if ((str[len] != 0) || |
2852 | 0 | (end != 0) || (end2 != 0) || (end3 != 0)) |
2853 | 0 | return(NULL); |
2854 | | |
2855 | 0 | return(xmlExpandEntitiesInAttValue(ctxt, str, 0)); |
2856 | 0 | } |
2857 | | |
2858 | | /** |
2859 | | * xmlStringDecodeEntities: |
2860 | | * @ctxt: the parser context |
2861 | | * @str: the input string |
2862 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2863 | | * @end: an end marker xmlChar, 0 if none |
2864 | | * @end2: an end marker xmlChar, 0 if none |
2865 | | * @end3: an end marker xmlChar, 0 if none |
2866 | | * |
2867 | | * DEPRECATED: Internal function, don't use. |
2868 | | * |
2869 | | * Returns A newly allocated string with the substitution done. The caller |
2870 | | * must deallocate it ! |
2871 | | */ |
2872 | | xmlChar * |
2873 | | xmlStringDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, |
2874 | | int what ATTRIBUTE_UNUSED, |
2875 | 0 | xmlChar end, xmlChar end2, xmlChar end3) { |
2876 | 0 | if ((ctxt == NULL) || (str == NULL)) |
2877 | 0 | return(NULL); |
2878 | | |
2879 | 0 | if ((end != 0) || (end2 != 0) || (end3 != 0)) |
2880 | 0 | return(NULL); |
2881 | | |
2882 | 0 | return(xmlExpandEntitiesInAttValue(ctxt, str, 0)); |
2883 | 0 | } |
2884 | | |
2885 | | /************************************************************************ |
2886 | | * * |
2887 | | * Commodity functions, cleanup needed ? * |
2888 | | * * |
2889 | | ************************************************************************/ |
2890 | | |
2891 | | /** |
2892 | | * areBlanks: |
2893 | | * @ctxt: an XML parser context |
2894 | | * @str: a xmlChar * |
2895 | | * @len: the size of @str |
2896 | | * @blank_chars: we know the chars are blanks |
2897 | | * |
2898 | | * Is this a sequence of blank chars that one can ignore ? |
2899 | | * |
2900 | | * Returns 1 if ignorable 0 otherwise. |
2901 | | */ |
2902 | | |
2903 | | static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2904 | 87.4k | int blank_chars) { |
2905 | 87.4k | int i; |
2906 | 87.4k | xmlNodePtr lastChild; |
2907 | | |
2908 | | /* |
2909 | | * Don't spend time trying to differentiate them, the same callback is |
2910 | | * used ! |
2911 | | */ |
2912 | 87.4k | if (ctxt->sax->ignorableWhitespace == ctxt->sax->characters) |
2913 | 87.4k | return(0); |
2914 | | |
2915 | | /* |
2916 | | * Check for xml:space value. |
2917 | | */ |
2918 | 0 | if ((ctxt->space == NULL) || (*(ctxt->space) == 1) || |
2919 | 0 | (*(ctxt->space) == -2)) |
2920 | 0 | return(0); |
2921 | | |
2922 | | /* |
2923 | | * Check that the string is made of blanks |
2924 | | */ |
2925 | 0 | if (blank_chars == 0) { |
2926 | 0 | for (i = 0;i < len;i++) |
2927 | 0 | if (!(IS_BLANK_CH(str[i]))) return(0); |
2928 | 0 | } |
2929 | | |
2930 | | /* |
2931 | | * Look if the element is mixed content in the DTD if available |
2932 | | */ |
2933 | 0 | if (ctxt->node == NULL) return(0); |
2934 | 0 | if (ctxt->myDoc != NULL) { |
2935 | 0 | xmlElementPtr elemDecl = NULL; |
2936 | 0 | xmlDocPtr doc = ctxt->myDoc; |
2937 | 0 | const xmlChar *prefix = NULL; |
2938 | |
|
2939 | 0 | if (ctxt->node->ns) |
2940 | 0 | prefix = ctxt->node->ns->prefix; |
2941 | 0 | if (doc->intSubset != NULL) |
2942 | 0 | elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name, |
2943 | 0 | prefix); |
2944 | 0 | if ((elemDecl == NULL) && (doc->extSubset != NULL)) |
2945 | 0 | elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name, |
2946 | 0 | prefix); |
2947 | 0 | if (elemDecl != NULL) { |
2948 | 0 | if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT) |
2949 | 0 | return(1); |
2950 | 0 | if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) || |
2951 | 0 | (elemDecl->etype == XML_ELEMENT_TYPE_MIXED)) |
2952 | 0 | return(0); |
2953 | 0 | } |
2954 | 0 | } |
2955 | | |
2956 | | /* |
2957 | | * Otherwise, heuristic :-\ |
2958 | | */ |
2959 | 0 | if ((RAW != '<') && (RAW != 0xD)) return(0); |
2960 | 0 | if ((ctxt->node->children == NULL) && |
2961 | 0 | (RAW == '<') && (NXT(1) == '/')) return(0); |
2962 | | |
2963 | 0 | lastChild = xmlGetLastChild(ctxt->node); |
2964 | 0 | if (lastChild == NULL) { |
2965 | 0 | if ((ctxt->node->type != XML_ELEMENT_NODE) && |
2966 | 0 | (ctxt->node->content != NULL)) return(0); |
2967 | 0 | } else if (xmlNodeIsText(lastChild)) |
2968 | 0 | return(0); |
2969 | 0 | else if ((ctxt->node->children != NULL) && |
2970 | 0 | (xmlNodeIsText(ctxt->node->children))) |
2971 | 0 | return(0); |
2972 | 0 | return(1); |
2973 | 0 | } |
2974 | | |
2975 | | /************************************************************************ |
2976 | | * * |
2977 | | * Extra stuff for namespace support * |
2978 | | * Relates to http://www.w3.org/TR/WD-xml-names * |
2979 | | * * |
2980 | | ************************************************************************/ |
2981 | | |
2982 | | /** |
2983 | | * xmlSplitQName: |
2984 | | * @ctxt: an XML parser context |
2985 | | * @name: an XML parser context |
2986 | | * @prefixOut: a xmlChar ** |
2987 | | * |
2988 | | * DEPRECATED: Don't use. |
2989 | | * |
2990 | | * parse an UTF8 encoded XML qualified name string |
2991 | | * |
2992 | | * [NS 5] QName ::= (Prefix ':')? LocalPart |
2993 | | * |
2994 | | * [NS 6] Prefix ::= NCName |
2995 | | * |
2996 | | * [NS 7] LocalPart ::= NCName |
2997 | | * |
2998 | | * Returns the local part, and prefix is updated |
2999 | | * to get the Prefix if any. |
3000 | | */ |
3001 | | |
3002 | | xmlChar * |
3003 | 0 | xmlSplitQName(xmlParserCtxtPtr ctxt, const xmlChar *name, xmlChar **prefixOut) { |
3004 | 0 | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3005 | 0 | xmlChar *buffer = NULL; |
3006 | 0 | int len = 0; |
3007 | 0 | int max = XML_MAX_NAMELEN; |
3008 | 0 | xmlChar *ret = NULL; |
3009 | 0 | xmlChar *prefix; |
3010 | 0 | const xmlChar *cur = name; |
3011 | 0 | int c; |
3012 | |
|
3013 | 0 | if (prefixOut == NULL) return(NULL); |
3014 | 0 | *prefixOut = NULL; |
3015 | |
|
3016 | 0 | if (cur == NULL) return(NULL); |
3017 | | |
3018 | | /* nasty but well=formed */ |
3019 | 0 | if (cur[0] == ':') |
3020 | 0 | return(xmlStrdup(name)); |
3021 | | |
3022 | 0 | c = *cur++; |
3023 | 0 | while ((c != 0) && (c != ':') && (len < max)) { /* tested bigname.xml */ |
3024 | 0 | buf[len++] = c; |
3025 | 0 | c = *cur++; |
3026 | 0 | } |
3027 | 0 | if (len >= max) { |
3028 | | /* |
3029 | | * Okay someone managed to make a huge name, so he's ready to pay |
3030 | | * for the processing speed. |
3031 | | */ |
3032 | 0 | max = len * 2; |
3033 | |
|
3034 | 0 | buffer = xmlMalloc(max); |
3035 | 0 | if (buffer == NULL) { |
3036 | 0 | xmlErrMemory(ctxt); |
3037 | 0 | return(NULL); |
3038 | 0 | } |
3039 | 0 | memcpy(buffer, buf, len); |
3040 | 0 | while ((c != 0) && (c != ':')) { /* tested bigname.xml */ |
3041 | 0 | if (len + 10 > max) { |
3042 | 0 | xmlChar *tmp; |
3043 | |
|
3044 | 0 | max *= 2; |
3045 | 0 | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3046 | 0 | if (tmp == NULL) { |
3047 | 0 | xmlFree(buffer); |
3048 | 0 | xmlErrMemory(ctxt); |
3049 | 0 | return(NULL); |
3050 | 0 | } |
3051 | 0 | buffer = tmp; |
3052 | 0 | } |
3053 | 0 | buffer[len++] = c; |
3054 | 0 | c = *cur++; |
3055 | 0 | } |
3056 | 0 | buffer[len] = 0; |
3057 | 0 | } |
3058 | | |
3059 | 0 | if ((c == ':') && (*cur == 0)) { |
3060 | 0 | if (buffer != NULL) |
3061 | 0 | xmlFree(buffer); |
3062 | 0 | return(xmlStrdup(name)); |
3063 | 0 | } |
3064 | | |
3065 | 0 | if (buffer == NULL) { |
3066 | 0 | ret = xmlStrndup(buf, len); |
3067 | 0 | if (ret == NULL) { |
3068 | 0 | xmlErrMemory(ctxt); |
3069 | 0 | return(NULL); |
3070 | 0 | } |
3071 | 0 | } else { |
3072 | 0 | ret = buffer; |
3073 | 0 | buffer = NULL; |
3074 | 0 | max = XML_MAX_NAMELEN; |
3075 | 0 | } |
3076 | | |
3077 | | |
3078 | 0 | if (c == ':') { |
3079 | 0 | c = *cur; |
3080 | 0 | prefix = ret; |
3081 | 0 | if (c == 0) { |
3082 | 0 | ret = xmlStrndup(BAD_CAST "", 0); |
3083 | 0 | if (ret == NULL) { |
3084 | 0 | xmlFree(prefix); |
3085 | 0 | return(NULL); |
3086 | 0 | } |
3087 | 0 | *prefixOut = prefix; |
3088 | 0 | return(ret); |
3089 | 0 | } |
3090 | 0 | len = 0; |
3091 | | |
3092 | | /* |
3093 | | * Check that the first character is proper to start |
3094 | | * a new name |
3095 | | */ |
3096 | 0 | if (!(((c >= 0x61) && (c <= 0x7A)) || |
3097 | 0 | ((c >= 0x41) && (c <= 0x5A)) || |
3098 | 0 | (c == '_') || (c == ':'))) { |
3099 | 0 | int l; |
3100 | 0 | int first = CUR_SCHAR(cur, l); |
3101 | |
|
3102 | 0 | if (!IS_LETTER(first) && (first != '_')) { |
3103 | 0 | xmlFatalErrMsgStr(ctxt, XML_NS_ERR_QNAME, |
3104 | 0 | "Name %s is not XML Namespace compliant\n", |
3105 | 0 | name); |
3106 | 0 | } |
3107 | 0 | } |
3108 | 0 | cur++; |
3109 | |
|
3110 | 0 | while ((c != 0) && (len < max)) { /* tested bigname2.xml */ |
3111 | 0 | buf[len++] = c; |
3112 | 0 | c = *cur++; |
3113 | 0 | } |
3114 | 0 | if (len >= max) { |
3115 | | /* |
3116 | | * Okay someone managed to make a huge name, so he's ready to pay |
3117 | | * for the processing speed. |
3118 | | */ |
3119 | 0 | max = len * 2; |
3120 | |
|
3121 | 0 | buffer = xmlMalloc(max); |
3122 | 0 | if (buffer == NULL) { |
3123 | 0 | xmlErrMemory(ctxt); |
3124 | 0 | xmlFree(prefix); |
3125 | 0 | return(NULL); |
3126 | 0 | } |
3127 | 0 | memcpy(buffer, buf, len); |
3128 | 0 | while (c != 0) { /* tested bigname2.xml */ |
3129 | 0 | if (len + 10 > max) { |
3130 | 0 | xmlChar *tmp; |
3131 | |
|
3132 | 0 | max *= 2; |
3133 | 0 | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3134 | 0 | if (tmp == NULL) { |
3135 | 0 | xmlErrMemory(ctxt); |
3136 | 0 | xmlFree(prefix); |
3137 | 0 | xmlFree(buffer); |
3138 | 0 | return(NULL); |
3139 | 0 | } |
3140 | 0 | buffer = tmp; |
3141 | 0 | } |
3142 | 0 | buffer[len++] = c; |
3143 | 0 | c = *cur++; |
3144 | 0 | } |
3145 | 0 | buffer[len] = 0; |
3146 | 0 | } |
3147 | | |
3148 | 0 | if (buffer == NULL) { |
3149 | 0 | ret = xmlStrndup(buf, len); |
3150 | 0 | if (ret == NULL) { |
3151 | 0 | xmlFree(prefix); |
3152 | 0 | return(NULL); |
3153 | 0 | } |
3154 | 0 | } else { |
3155 | 0 | ret = buffer; |
3156 | 0 | } |
3157 | | |
3158 | 0 | *prefixOut = prefix; |
3159 | 0 | } |
3160 | | |
3161 | 0 | return(ret); |
3162 | 0 | } |
3163 | | |
3164 | | /************************************************************************ |
3165 | | * * |
3166 | | * The parser itself * |
3167 | | * Relates to http://www.w3.org/TR/REC-xml * |
3168 | | * * |
3169 | | ************************************************************************/ |
3170 | | |
3171 | | /************************************************************************ |
3172 | | * * |
3173 | | * Routines to parse Name, NCName and NmToken * |
3174 | | * * |
3175 | | ************************************************************************/ |
3176 | | |
3177 | | /* |
3178 | | * The two following functions are related to the change of accepted |
3179 | | * characters for Name and NmToken in the Revision 5 of XML-1.0 |
3180 | | * They correspond to the modified production [4] and the new production [4a] |
3181 | | * changes in that revision. Also note that the macros used for the |
3182 | | * productions Letter, Digit, CombiningChar and Extender are not needed |
3183 | | * anymore. |
3184 | | * We still keep compatibility to pre-revision5 parsing semantic if the |
3185 | | * new XML_PARSE_OLD10 option is given to the parser. |
3186 | | */ |
3187 | | static int |
3188 | 2.10M | xmlIsNameStartChar(xmlParserCtxtPtr ctxt, int c) { |
3189 | 2.10M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3190 | | /* |
3191 | | * Use the new checks of production [4] [4a] amd [5] of the |
3192 | | * Update 5 of XML-1.0 |
3193 | | */ |
3194 | 2.10M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3195 | 2.10M | (((c >= 'a') && (c <= 'z')) || |
3196 | 2.10M | ((c >= 'A') && (c <= 'Z')) || |
3197 | 2.10M | (c == '_') || (c == ':') || |
3198 | 2.10M | ((c >= 0xC0) && (c <= 0xD6)) || |
3199 | 2.10M | ((c >= 0xD8) && (c <= 0xF6)) || |
3200 | 2.10M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3201 | 2.10M | ((c >= 0x370) && (c <= 0x37D)) || |
3202 | 2.10M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3203 | 2.10M | ((c >= 0x200C) && (c <= 0x200D)) || |
3204 | 2.10M | ((c >= 0x2070) && (c <= 0x218F)) || |
3205 | 2.10M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3206 | 2.10M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3207 | 2.10M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3208 | 2.10M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3209 | 2.10M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3210 | 1.80M | return(1); |
3211 | 2.10M | } else { |
3212 | 0 | if (IS_LETTER(c) || (c == '_') || (c == ':')) |
3213 | 0 | return(1); |
3214 | 0 | } |
3215 | 296k | return(0); |
3216 | 2.10M | } |
3217 | | |
3218 | | static int |
3219 | 48.2M | xmlIsNameChar(xmlParserCtxtPtr ctxt, int c) { |
3220 | 48.2M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3221 | | /* |
3222 | | * Use the new checks of production [4] [4a] amd [5] of the |
3223 | | * Update 5 of XML-1.0 |
3224 | | */ |
3225 | 48.2M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3226 | 48.2M | (((c >= 'a') && (c <= 'z')) || |
3227 | 48.2M | ((c >= 'A') && (c <= 'Z')) || |
3228 | 48.2M | ((c >= '0') && (c <= '9')) || /* !start */ |
3229 | 48.2M | (c == '_') || (c == ':') || |
3230 | 48.2M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3231 | 48.2M | ((c >= 0xC0) && (c <= 0xD6)) || |
3232 | 48.2M | ((c >= 0xD8) && (c <= 0xF6)) || |
3233 | 48.2M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3234 | 48.2M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3235 | 48.2M | ((c >= 0x370) && (c <= 0x37D)) || |
3236 | 48.2M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3237 | 48.2M | ((c >= 0x200C) && (c <= 0x200D)) || |
3238 | 48.2M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3239 | 48.2M | ((c >= 0x2070) && (c <= 0x218F)) || |
3240 | 48.2M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3241 | 48.2M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3242 | 48.2M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3243 | 48.2M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3244 | 48.2M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3245 | 46.4M | return(1); |
3246 | 48.2M | } else { |
3247 | 0 | if ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3248 | 0 | (c == '.') || (c == '-') || |
3249 | 0 | (c == '_') || (c == ':') || |
3250 | 0 | (IS_COMBINING(c)) || |
3251 | 0 | (IS_EXTENDER(c))) |
3252 | 0 | return(1); |
3253 | 0 | } |
3254 | 1.82M | return(0); |
3255 | 48.2M | } |
3256 | | |
3257 | | static const xmlChar * |
3258 | 209k | xmlParseNameComplex(xmlParserCtxtPtr ctxt) { |
3259 | 209k | const xmlChar *ret; |
3260 | 209k | int len = 0, l; |
3261 | 209k | int c; |
3262 | 209k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3263 | 0 | XML_MAX_TEXT_LENGTH : |
3264 | 209k | XML_MAX_NAME_LENGTH; |
3265 | | |
3266 | | /* |
3267 | | * Handler for more complex cases |
3268 | | */ |
3269 | 209k | c = xmlCurrentChar(ctxt, &l); |
3270 | 209k | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3271 | | /* |
3272 | | * Use the new checks of production [4] [4a] amd [5] of the |
3273 | | * Update 5 of XML-1.0 |
3274 | | */ |
3275 | 209k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3276 | 209k | (!(((c >= 'a') && (c <= 'z')) || |
3277 | 203k | ((c >= 'A') && (c <= 'Z')) || |
3278 | 203k | (c == '_') || (c == ':') || |
3279 | 203k | ((c >= 0xC0) && (c <= 0xD6)) || |
3280 | 203k | ((c >= 0xD8) && (c <= 0xF6)) || |
3281 | 203k | ((c >= 0xF8) && (c <= 0x2FF)) || |
3282 | 203k | ((c >= 0x370) && (c <= 0x37D)) || |
3283 | 203k | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3284 | 203k | ((c >= 0x200C) && (c <= 0x200D)) || |
3285 | 203k | ((c >= 0x2070) && (c <= 0x218F)) || |
3286 | 203k | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3287 | 203k | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3288 | 203k | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3289 | 203k | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3290 | 203k | ((c >= 0x10000) && (c <= 0xEFFFF))))) { |
3291 | 135k | return(NULL); |
3292 | 135k | } |
3293 | 74.6k | len += l; |
3294 | 74.6k | NEXTL(l); |
3295 | 74.6k | c = xmlCurrentChar(ctxt, &l); |
3296 | 5.37M | while ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3297 | 5.37M | (((c >= 'a') && (c <= 'z')) || |
3298 | 5.36M | ((c >= 'A') && (c <= 'Z')) || |
3299 | 5.36M | ((c >= '0') && (c <= '9')) || /* !start */ |
3300 | 5.36M | (c == '_') || (c == ':') || |
3301 | 5.36M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3302 | 5.36M | ((c >= 0xC0) && (c <= 0xD6)) || |
3303 | 5.36M | ((c >= 0xD8) && (c <= 0xF6)) || |
3304 | 5.36M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3305 | 5.36M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3306 | 5.36M | ((c >= 0x370) && (c <= 0x37D)) || |
3307 | 5.36M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3308 | 5.36M | ((c >= 0x200C) && (c <= 0x200D)) || |
3309 | 5.36M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3310 | 5.36M | ((c >= 0x2070) && (c <= 0x218F)) || |
3311 | 5.36M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3312 | 5.36M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3313 | 5.36M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3314 | 5.36M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3315 | 5.36M | ((c >= 0x10000) && (c <= 0xEFFFF)) |
3316 | 5.36M | )) { |
3317 | 5.30M | if (len <= INT_MAX - l) |
3318 | 5.30M | len += l; |
3319 | 5.30M | NEXTL(l); |
3320 | 5.30M | c = xmlCurrentChar(ctxt, &l); |
3321 | 5.30M | } |
3322 | 74.6k | } else { |
3323 | 0 | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3324 | 0 | (!IS_LETTER(c) && (c != '_') && |
3325 | 0 | (c != ':'))) { |
3326 | 0 | return(NULL); |
3327 | 0 | } |
3328 | 0 | len += l; |
3329 | 0 | NEXTL(l); |
3330 | 0 | c = xmlCurrentChar(ctxt, &l); |
3331 | |
|
3332 | 0 | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3333 | 0 | ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3334 | 0 | (c == '.') || (c == '-') || |
3335 | 0 | (c == '_') || (c == ':') || |
3336 | 0 | (IS_COMBINING(c)) || |
3337 | 0 | (IS_EXTENDER(c)))) { |
3338 | 0 | if (len <= INT_MAX - l) |
3339 | 0 | len += l; |
3340 | 0 | NEXTL(l); |
3341 | 0 | c = xmlCurrentChar(ctxt, &l); |
3342 | 0 | } |
3343 | 0 | } |
3344 | 74.6k | if (len > maxLength) { |
3345 | 48 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3346 | 48 | return(NULL); |
3347 | 48 | } |
3348 | 74.5k | if (ctxt->input->cur - ctxt->input->base < len) { |
3349 | | /* |
3350 | | * There were a couple of bugs where PERefs lead to to a change |
3351 | | * of the buffer. Check the buffer size to avoid passing an invalid |
3352 | | * pointer to xmlDictLookup. |
3353 | | */ |
3354 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
3355 | 0 | "unexpected change of input buffer"); |
3356 | 0 | return (NULL); |
3357 | 0 | } |
3358 | 74.5k | if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r')) |
3359 | 837 | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len); |
3360 | 73.7k | else |
3361 | 73.7k | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len); |
3362 | 74.5k | if (ret == NULL) |
3363 | 0 | xmlErrMemory(ctxt); |
3364 | 74.5k | return(ret); |
3365 | 74.5k | } |
3366 | | |
3367 | | /** |
3368 | | * xmlParseName: |
3369 | | * @ctxt: an XML parser context |
3370 | | * |
3371 | | * DEPRECATED: Internal function, don't use. |
3372 | | * |
3373 | | * parse an XML name. |
3374 | | * |
3375 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3376 | | * CombiningChar | Extender |
3377 | | * |
3378 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3379 | | * |
3380 | | * [6] Names ::= Name (#x20 Name)* |
3381 | | * |
3382 | | * Returns the Name parsed or NULL |
3383 | | */ |
3384 | | |
3385 | | const xmlChar * |
3386 | 1.11M | xmlParseName(xmlParserCtxtPtr ctxt) { |
3387 | 1.11M | const xmlChar *in; |
3388 | 1.11M | const xmlChar *ret; |
3389 | 1.11M | size_t count = 0; |
3390 | 1.11M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3391 | 0 | XML_MAX_TEXT_LENGTH : |
3392 | 1.11M | XML_MAX_NAME_LENGTH; |
3393 | | |
3394 | 1.11M | GROW; |
3395 | | |
3396 | | /* |
3397 | | * Accelerator for simple ASCII names |
3398 | | */ |
3399 | 1.11M | in = ctxt->input->cur; |
3400 | 1.11M | if (((*in >= 0x61) && (*in <= 0x7A)) || |
3401 | 1.11M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3402 | 1.11M | (*in == '_') || (*in == ':')) { |
3403 | 942k | in++; |
3404 | 2.21M | while (((*in >= 0x61) && (*in <= 0x7A)) || |
3405 | 2.21M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3406 | 2.21M | ((*in >= 0x30) && (*in <= 0x39)) || |
3407 | 2.21M | (*in == '_') || (*in == '-') || |
3408 | 2.21M | (*in == ':') || (*in == '.')) |
3409 | 1.27M | in++; |
3410 | 942k | if ((*in > 0) && (*in < 0x80)) { |
3411 | 905k | count = in - ctxt->input->cur; |
3412 | 905k | if (count > maxLength) { |
3413 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3414 | 0 | return(NULL); |
3415 | 0 | } |
3416 | 905k | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3417 | 905k | ctxt->input->cur = in; |
3418 | 905k | ctxt->input->col += count; |
3419 | 905k | if (ret == NULL) |
3420 | 0 | xmlErrMemory(ctxt); |
3421 | 905k | return(ret); |
3422 | 905k | } |
3423 | 942k | } |
3424 | | /* accelerator for special cases */ |
3425 | 209k | return(xmlParseNameComplex(ctxt)); |
3426 | 1.11M | } |
3427 | | |
3428 | | static xmlHashedString |
3429 | 446k | xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) { |
3430 | 446k | xmlHashedString ret; |
3431 | 446k | int len = 0, l; |
3432 | 446k | int c; |
3433 | 446k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3434 | 0 | XML_MAX_TEXT_LENGTH : |
3435 | 446k | XML_MAX_NAME_LENGTH; |
3436 | 446k | size_t startPosition = 0; |
3437 | | |
3438 | 446k | ret.name = NULL; |
3439 | 446k | ret.hashValue = 0; |
3440 | | |
3441 | | /* |
3442 | | * Handler for more complex cases |
3443 | | */ |
3444 | 446k | startPosition = CUR_PTR - BASE_PTR; |
3445 | 446k | c = xmlCurrentChar(ctxt, &l); |
3446 | 446k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3447 | 446k | (!xmlIsNameStartChar(ctxt, c) || (c == ':'))) { |
3448 | 329k | return(ret); |
3449 | 329k | } |
3450 | | |
3451 | 38.3M | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3452 | 38.3M | (xmlIsNameChar(ctxt, c) && (c != ':'))) { |
3453 | 38.2M | if (len <= INT_MAX - l) |
3454 | 38.2M | len += l; |
3455 | 38.2M | NEXTL(l); |
3456 | 38.2M | c = xmlCurrentChar(ctxt, &l); |
3457 | 38.2M | } |
3458 | 117k | if (len > maxLength) { |
3459 | 1.14k | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3460 | 1.14k | return(ret); |
3461 | 1.14k | } |
3462 | 116k | ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len); |
3463 | 116k | if (ret.name == NULL) |
3464 | 0 | xmlErrMemory(ctxt); |
3465 | 116k | return(ret); |
3466 | 117k | } |
3467 | | |
3468 | | /** |
3469 | | * xmlParseNCName: |
3470 | | * @ctxt: an XML parser context |
3471 | | * @len: length of the string parsed |
3472 | | * |
3473 | | * parse an XML name. |
3474 | | * |
3475 | | * [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' | |
3476 | | * CombiningChar | Extender |
3477 | | * |
3478 | | * [5NS] NCName ::= (Letter | '_') (NCNameChar)* |
3479 | | * |
3480 | | * Returns the Name parsed or NULL |
3481 | | */ |
3482 | | |
3483 | | static xmlHashedString |
3484 | 1.65M | xmlParseNCName(xmlParserCtxtPtr ctxt) { |
3485 | 1.65M | const xmlChar *in, *e; |
3486 | 1.65M | xmlHashedString ret; |
3487 | 1.65M | size_t count = 0; |
3488 | 1.65M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3489 | 0 | XML_MAX_TEXT_LENGTH : |
3490 | 1.65M | XML_MAX_NAME_LENGTH; |
3491 | | |
3492 | 1.65M | ret.name = NULL; |
3493 | | |
3494 | | /* |
3495 | | * Accelerator for simple ASCII names |
3496 | | */ |
3497 | 1.65M | in = ctxt->input->cur; |
3498 | 1.65M | e = ctxt->input->end; |
3499 | 1.65M | if ((((*in >= 0x61) && (*in <= 0x7A)) || |
3500 | 1.65M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3501 | 1.65M | (*in == '_')) && (in < e)) { |
3502 | 1.25M | in++; |
3503 | 3.53M | while ((((*in >= 0x61) && (*in <= 0x7A)) || |
3504 | 3.53M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3505 | 3.53M | ((*in >= 0x30) && (*in <= 0x39)) || |
3506 | 3.53M | (*in == '_') || (*in == '-') || |
3507 | 3.53M | (*in == '.')) && (in < e)) |
3508 | 2.28M | in++; |
3509 | 1.25M | if (in >= e) |
3510 | 759 | goto complex; |
3511 | 1.25M | if ((*in > 0) && (*in < 0x80)) { |
3512 | 1.21M | count = in - ctxt->input->cur; |
3513 | 1.21M | if (count > maxLength) { |
3514 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3515 | 0 | return(ret); |
3516 | 0 | } |
3517 | 1.21M | ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count); |
3518 | 1.21M | ctxt->input->cur = in; |
3519 | 1.21M | ctxt->input->col += count; |
3520 | 1.21M | if (ret.name == NULL) { |
3521 | 0 | xmlErrMemory(ctxt); |
3522 | 0 | } |
3523 | 1.21M | return(ret); |
3524 | 1.21M | } |
3525 | 1.25M | } |
3526 | 446k | complex: |
3527 | 446k | return(xmlParseNCNameComplex(ctxt)); |
3528 | 1.65M | } |
3529 | | |
3530 | | /** |
3531 | | * xmlParseNameAndCompare: |
3532 | | * @ctxt: an XML parser context |
3533 | | * |
3534 | | * parse an XML name and compares for match |
3535 | | * (specialized for endtag parsing) |
3536 | | * |
3537 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
3538 | | * and the name for mismatch |
3539 | | */ |
3540 | | |
3541 | | static const xmlChar * |
3542 | 43.8k | xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) { |
3543 | 43.8k | register const xmlChar *cmp = other; |
3544 | 43.8k | register const xmlChar *in; |
3545 | 43.8k | const xmlChar *ret; |
3546 | | |
3547 | 43.8k | GROW; |
3548 | | |
3549 | 43.8k | in = ctxt->input->cur; |
3550 | 185k | while (*in != 0 && *in == *cmp) { |
3551 | 141k | ++in; |
3552 | 141k | ++cmp; |
3553 | 141k | } |
3554 | 43.8k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
3555 | | /* success */ |
3556 | 26.2k | ctxt->input->col += in - ctxt->input->cur; |
3557 | 26.2k | ctxt->input->cur = in; |
3558 | 26.2k | return (const xmlChar*) 1; |
3559 | 26.2k | } |
3560 | | /* failure (or end of input buffer), check with full function */ |
3561 | 17.5k | ret = xmlParseName (ctxt); |
3562 | | /* strings coming from the dictionary direct compare possible */ |
3563 | 17.5k | if (ret == other) { |
3564 | 2.28k | return (const xmlChar*) 1; |
3565 | 2.28k | } |
3566 | 15.3k | return ret; |
3567 | 17.5k | } |
3568 | | |
3569 | | /** |
3570 | | * xmlParseStringName: |
3571 | | * @ctxt: an XML parser context |
3572 | | * @str: a pointer to the string pointer (IN/OUT) |
3573 | | * |
3574 | | * parse an XML name. |
3575 | | * |
3576 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3577 | | * CombiningChar | Extender |
3578 | | * |
3579 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3580 | | * |
3581 | | * [6] Names ::= Name (#x20 Name)* |
3582 | | * |
3583 | | * Returns the Name parsed or NULL. The @str pointer |
3584 | | * is updated to the current location in the string. |
3585 | | */ |
3586 | | |
3587 | | static xmlChar * |
3588 | 1.67M | xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) { |
3589 | 1.67M | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3590 | 1.67M | xmlChar *ret; |
3591 | 1.67M | const xmlChar *cur = *str; |
3592 | 1.67M | int len = 0, l; |
3593 | 1.67M | int c; |
3594 | 1.67M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3595 | 0 | XML_MAX_TEXT_LENGTH : |
3596 | 1.67M | XML_MAX_NAME_LENGTH; |
3597 | | |
3598 | 1.67M | c = CUR_SCHAR(cur, l); |
3599 | 1.67M | if (!xmlIsNameStartChar(ctxt, c)) { |
3600 | 2.49k | return(NULL); |
3601 | 2.49k | } |
3602 | | |
3603 | 1.67M | COPY_BUF(buf, len, c); |
3604 | 1.67M | cur += l; |
3605 | 1.67M | c = CUR_SCHAR(cur, l); |
3606 | 3.66M | while (xmlIsNameChar(ctxt, c)) { |
3607 | 2.00M | COPY_BUF(buf, len, c); |
3608 | 2.00M | cur += l; |
3609 | 2.00M | c = CUR_SCHAR(cur, l); |
3610 | 2.00M | if (len >= XML_MAX_NAMELEN) { /* test bigentname.xml */ |
3611 | | /* |
3612 | | * Okay someone managed to make a huge name, so he's ready to pay |
3613 | | * for the processing speed. |
3614 | | */ |
3615 | 6.61k | xmlChar *buffer; |
3616 | 6.61k | int max = len * 2; |
3617 | | |
3618 | 6.61k | buffer = xmlMalloc(max); |
3619 | 6.61k | if (buffer == NULL) { |
3620 | 0 | xmlErrMemory(ctxt); |
3621 | 0 | return(NULL); |
3622 | 0 | } |
3623 | 6.61k | memcpy(buffer, buf, len); |
3624 | 1.56M | while (xmlIsNameChar(ctxt, c)) { |
3625 | 1.55M | if (len + 10 > max) { |
3626 | 3.90k | xmlChar *tmp; |
3627 | | |
3628 | 3.90k | max *= 2; |
3629 | 3.90k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3630 | 3.90k | if (tmp == NULL) { |
3631 | 0 | xmlErrMemory(ctxt); |
3632 | 0 | xmlFree(buffer); |
3633 | 0 | return(NULL); |
3634 | 0 | } |
3635 | 3.90k | buffer = tmp; |
3636 | 3.90k | } |
3637 | 1.55M | COPY_BUF(buffer, len, c); |
3638 | 1.55M | cur += l; |
3639 | 1.55M | c = CUR_SCHAR(cur, l); |
3640 | 1.55M | if (len > maxLength) { |
3641 | 74 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3642 | 74 | xmlFree(buffer); |
3643 | 74 | return(NULL); |
3644 | 74 | } |
3645 | 1.55M | } |
3646 | 6.54k | buffer[len] = 0; |
3647 | 6.54k | *str = cur; |
3648 | 6.54k | return(buffer); |
3649 | 6.61k | } |
3650 | 2.00M | } |
3651 | 1.66M | if (len > maxLength) { |
3652 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3653 | 0 | return(NULL); |
3654 | 0 | } |
3655 | 1.66M | *str = cur; |
3656 | 1.66M | ret = xmlStrndup(buf, len); |
3657 | 1.66M | if (ret == NULL) |
3658 | 0 | xmlErrMemory(ctxt); |
3659 | 1.66M | return(ret); |
3660 | 1.66M | } |
3661 | | |
3662 | | /** |
3663 | | * xmlParseNmtoken: |
3664 | | * @ctxt: an XML parser context |
3665 | | * |
3666 | | * DEPRECATED: Internal function, don't use. |
3667 | | * |
3668 | | * parse an XML Nmtoken. |
3669 | | * |
3670 | | * [7] Nmtoken ::= (NameChar)+ |
3671 | | * |
3672 | | * [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)* |
3673 | | * |
3674 | | * Returns the Nmtoken parsed or NULL |
3675 | | */ |
3676 | | |
3677 | | xmlChar * |
3678 | 48.3k | xmlParseNmtoken(xmlParserCtxtPtr ctxt) { |
3679 | 48.3k | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3680 | 48.3k | xmlChar *ret; |
3681 | 48.3k | int len = 0, l; |
3682 | 48.3k | int c; |
3683 | 48.3k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3684 | 0 | XML_MAX_TEXT_LENGTH : |
3685 | 48.3k | XML_MAX_NAME_LENGTH; |
3686 | | |
3687 | 48.3k | c = xmlCurrentChar(ctxt, &l); |
3688 | | |
3689 | 251k | while (xmlIsNameChar(ctxt, c)) { |
3690 | 206k | COPY_BUF(buf, len, c); |
3691 | 206k | NEXTL(l); |
3692 | 206k | c = xmlCurrentChar(ctxt, &l); |
3693 | 206k | if (len >= XML_MAX_NAMELEN) { |
3694 | | /* |
3695 | | * Okay someone managed to make a huge token, so he's ready to pay |
3696 | | * for the processing speed. |
3697 | | */ |
3698 | 2.69k | xmlChar *buffer; |
3699 | 2.69k | int max = len * 2; |
3700 | | |
3701 | 2.69k | buffer = xmlMalloc(max); |
3702 | 2.69k | if (buffer == NULL) { |
3703 | 0 | xmlErrMemory(ctxt); |
3704 | 0 | return(NULL); |
3705 | 0 | } |
3706 | 2.69k | memcpy(buffer, buf, len); |
3707 | 4.43M | while (xmlIsNameChar(ctxt, c)) { |
3708 | 4.43M | if (len + 10 > max) { |
3709 | 6.26k | xmlChar *tmp; |
3710 | | |
3711 | 6.26k | max *= 2; |
3712 | 6.26k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3713 | 6.26k | if (tmp == NULL) { |
3714 | 0 | xmlErrMemory(ctxt); |
3715 | 0 | xmlFree(buffer); |
3716 | 0 | return(NULL); |
3717 | 0 | } |
3718 | 6.26k | buffer = tmp; |
3719 | 6.26k | } |
3720 | 4.43M | COPY_BUF(buffer, len, c); |
3721 | 4.43M | if (len > maxLength) { |
3722 | 203 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3723 | 203 | xmlFree(buffer); |
3724 | 203 | return(NULL); |
3725 | 203 | } |
3726 | 4.43M | NEXTL(l); |
3727 | 4.43M | c = xmlCurrentChar(ctxt, &l); |
3728 | 4.43M | } |
3729 | 2.49k | buffer[len] = 0; |
3730 | 2.49k | return(buffer); |
3731 | 2.69k | } |
3732 | 206k | } |
3733 | 45.6k | if (len == 0) |
3734 | 8.61k | return(NULL); |
3735 | 37.0k | if (len > maxLength) { |
3736 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3737 | 0 | return(NULL); |
3738 | 0 | } |
3739 | 37.0k | ret = xmlStrndup(buf, len); |
3740 | 37.0k | if (ret == NULL) |
3741 | 0 | xmlErrMemory(ctxt); |
3742 | 37.0k | return(ret); |
3743 | 37.0k | } |
3744 | | |
3745 | | /** |
3746 | | * xmlExpandPEsInEntityValue: |
3747 | | * @ctxt: parser context |
3748 | | * @buf: string buffer |
3749 | | * @str: entity value |
3750 | | * @length: size of entity value |
3751 | | * @depth: nesting depth |
3752 | | * |
3753 | | * Validate an entity value and expand parameter entities. |
3754 | | */ |
3755 | | static void |
3756 | | xmlExpandPEsInEntityValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf, |
3757 | 45.4k | const xmlChar *str, int length, int depth) { |
3758 | 45.4k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
3759 | 45.4k | const xmlChar *end, *chunk; |
3760 | 45.4k | int c, l; |
3761 | | |
3762 | 45.4k | if (str == NULL) |
3763 | 0 | return; |
3764 | | |
3765 | 45.4k | depth += 1; |
3766 | 45.4k | if (depth > maxDepth) { |
3767 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
3768 | 0 | "Maximum entity nesting depth exceeded"); |
3769 | 0 | return; |
3770 | 0 | } |
3771 | | |
3772 | 45.4k | end = str + length; |
3773 | 45.4k | chunk = str; |
3774 | | |
3775 | 146M | while ((str < end) && (!PARSER_STOPPED(ctxt))) { |
3776 | 146M | c = *str; |
3777 | | |
3778 | 146M | if (c >= 0x80) { |
3779 | 142M | l = xmlUTF8MultibyteLen(ctxt, str, |
3780 | 142M | "invalid character in entity value\n"); |
3781 | 142M | if (l == 0) { |
3782 | 4.24M | if (chunk < str) |
3783 | 71.8k | xmlSBufAddString(buf, chunk, str - chunk); |
3784 | 4.24M | xmlSBufAddReplChar(buf); |
3785 | 4.24M | str += 1; |
3786 | 4.24M | chunk = str; |
3787 | 138M | } else { |
3788 | 138M | str += l; |
3789 | 138M | } |
3790 | 142M | } else if (c == '&') { |
3791 | 152k | if (str[1] == '#') { |
3792 | 79.6k | if (chunk < str) |
3793 | 61.3k | xmlSBufAddString(buf, chunk, str - chunk); |
3794 | | |
3795 | 79.6k | c = xmlParseStringCharRef(ctxt, &str); |
3796 | 79.6k | if (c == 0) |
3797 | 19.5k | return; |
3798 | | |
3799 | 60.0k | xmlSBufAddChar(buf, c); |
3800 | | |
3801 | 60.0k | chunk = str; |
3802 | 72.4k | } else { |
3803 | 72.4k | xmlChar *name; |
3804 | | |
3805 | | /* |
3806 | | * General entity references are checked for |
3807 | | * syntactic validity. |
3808 | | */ |
3809 | 72.4k | str++; |
3810 | 72.4k | name = xmlParseStringName(ctxt, &str); |
3811 | | |
3812 | 72.4k | if ((name == NULL) || (*str++ != ';')) { |
3813 | 5.71k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR, |
3814 | 5.71k | "EntityValue: '&' forbidden except for entities " |
3815 | 5.71k | "references\n"); |
3816 | 5.71k | xmlFree(name); |
3817 | 5.71k | return; |
3818 | 5.71k | } |
3819 | | |
3820 | 66.6k | xmlFree(name); |
3821 | 66.6k | } |
3822 | 3.94M | } else if (c == '%') { |
3823 | 4.15k | xmlEntityPtr ent; |
3824 | | |
3825 | 4.15k | if (chunk < str) |
3826 | 3.51k | xmlSBufAddString(buf, chunk, str - chunk); |
3827 | | |
3828 | 4.15k | ent = xmlParseStringPEReference(ctxt, &str); |
3829 | 4.15k | if (ent == NULL) |
3830 | 3.50k | return; |
3831 | | |
3832 | 642 | if (!PARSER_EXTERNAL(ctxt)) { |
3833 | 642 | xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL); |
3834 | 642 | return; |
3835 | 642 | } |
3836 | | |
3837 | 0 | if (ent->content == NULL) { |
3838 | | /* |
3839 | | * Note: external parsed entities will not be loaded, |
3840 | | * it is not required for a non-validating parser to |
3841 | | * complete external PEReferences coming from the |
3842 | | * internal subset |
3843 | | */ |
3844 | 0 | if (((ctxt->options & XML_PARSE_NO_XXE) == 0) && |
3845 | 0 | ((ctxt->replaceEntities) || |
3846 | 0 | (ctxt->validate))) { |
3847 | 0 | xmlLoadEntityContent(ctxt, ent); |
3848 | 0 | } else { |
3849 | 0 | xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING, |
3850 | 0 | "not validating will not read content for " |
3851 | 0 | "PE entity %s\n", ent->name, NULL); |
3852 | 0 | } |
3853 | 0 | } |
3854 | | |
3855 | | /* |
3856 | | * TODO: Skip if ent->content is still NULL. |
3857 | | */ |
3858 | |
|
3859 | 0 | if (xmlParserEntityCheck(ctxt, ent->length)) |
3860 | 0 | return; |
3861 | | |
3862 | 0 | if (ent->flags & XML_ENT_EXPANDING) { |
3863 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
3864 | 0 | xmlHaltParser(ctxt); |
3865 | 0 | return; |
3866 | 0 | } |
3867 | | |
3868 | 0 | ent->flags |= XML_ENT_EXPANDING; |
3869 | 0 | xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length, |
3870 | 0 | depth); |
3871 | 0 | ent->flags &= ~XML_ENT_EXPANDING; |
3872 | |
|
3873 | 0 | chunk = str; |
3874 | 3.93M | } else { |
3875 | | /* Normal ASCII char */ |
3876 | 3.93M | if (!IS_BYTE_CHAR(c)) { |
3877 | 158k | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
3878 | 158k | "invalid character in entity value\n"); |
3879 | 158k | if (chunk < str) |
3880 | 5.22k | xmlSBufAddString(buf, chunk, str - chunk); |
3881 | 158k | xmlSBufAddReplChar(buf); |
3882 | 158k | str += 1; |
3883 | 158k | chunk = str; |
3884 | 3.77M | } else { |
3885 | 3.77M | str += 1; |
3886 | 3.77M | } |
3887 | 3.93M | } |
3888 | 146M | } |
3889 | | |
3890 | 15.9k | if (chunk < str) |
3891 | 11.6k | xmlSBufAddString(buf, chunk, str - chunk); |
3892 | 15.9k | } |
3893 | | |
3894 | | /** |
3895 | | * xmlParseEntityValue: |
3896 | | * @ctxt: an XML parser context |
3897 | | * @orig: if non-NULL store a copy of the original entity value |
3898 | | * |
3899 | | * DEPRECATED: Internal function, don't use. |
3900 | | * |
3901 | | * parse a value for ENTITY declarations |
3902 | | * |
3903 | | * [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' | |
3904 | | * "'" ([^%&'] | PEReference | Reference)* "'" |
3905 | | * |
3906 | | * Returns the EntityValue parsed with reference substituted or NULL |
3907 | | */ |
3908 | | xmlChar * |
3909 | 45.8k | xmlParseEntityValue(xmlParserCtxtPtr ctxt, xmlChar **orig) { |
3910 | 45.8k | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3911 | 0 | XML_MAX_HUGE_LENGTH : |
3912 | 45.8k | XML_MAX_TEXT_LENGTH; |
3913 | 45.8k | xmlSBuf buf; |
3914 | 45.8k | const xmlChar *start; |
3915 | 45.8k | int quote, length; |
3916 | | |
3917 | 45.8k | xmlSBufInit(&buf, maxLength); |
3918 | | |
3919 | 45.8k | GROW; |
3920 | | |
3921 | 45.8k | quote = CUR; |
3922 | 45.8k | if ((quote != '"') && (quote != '\'')) { |
3923 | 0 | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
3924 | 0 | return(NULL); |
3925 | 0 | } |
3926 | 45.8k | CUR_PTR++; |
3927 | | |
3928 | 45.8k | length = 0; |
3929 | | |
3930 | | /* |
3931 | | * Copy raw content of the entity into a buffer |
3932 | | */ |
3933 | 446M | while (1) { |
3934 | 446M | int c; |
3935 | | |
3936 | 446M | if (PARSER_STOPPED(ctxt)) |
3937 | 1 | goto error; |
3938 | | |
3939 | 446M | if (CUR_PTR >= ctxt->input->end) { |
3940 | 393 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL); |
3941 | 393 | goto error; |
3942 | 393 | } |
3943 | | |
3944 | 446M | c = CUR; |
3945 | | |
3946 | 446M | if (c == 0) { |
3947 | 36 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
3948 | 36 | "invalid character in entity value\n"); |
3949 | 36 | goto error; |
3950 | 36 | } |
3951 | 446M | if (c == quote) |
3952 | 45.4k | break; |
3953 | 446M | NEXTL(1); |
3954 | 446M | length += 1; |
3955 | | |
3956 | | /* |
3957 | | * TODO: Check growth threshold |
3958 | | */ |
3959 | 446M | if (ctxt->input->end - CUR_PTR < 10) |
3960 | 18.2k | GROW; |
3961 | 446M | } |
3962 | | |
3963 | 45.4k | start = CUR_PTR - length; |
3964 | | |
3965 | 45.4k | if (orig != NULL) { |
3966 | 45.4k | *orig = xmlStrndup(start, length); |
3967 | 45.4k | if (*orig == NULL) |
3968 | 0 | xmlErrMemory(ctxt); |
3969 | 45.4k | } |
3970 | | |
3971 | 45.4k | xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr); |
3972 | | |
3973 | 45.4k | NEXTL(1); |
3974 | | |
3975 | 45.4k | return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long")); |
3976 | | |
3977 | 430 | error: |
3978 | 430 | xmlSBufCleanup(&buf, ctxt, "entity length too long"); |
3979 | 430 | return(NULL); |
3980 | 45.8k | } |
3981 | | |
3982 | | /** |
3983 | | * xmlCheckEntityInAttValue: |
3984 | | * @ctxt: parser context |
3985 | | * @pent: entity |
3986 | | * @depth: nesting depth |
3987 | | * |
3988 | | * Check an entity reference in an attribute value for validity |
3989 | | * without expanding it. |
3990 | | */ |
3991 | | static void |
3992 | 1.23k | xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) { |
3993 | 1.23k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
3994 | 1.23k | const xmlChar *str; |
3995 | 1.23k | unsigned long expandedSize = pent->length; |
3996 | 1.23k | int c, flags; |
3997 | | |
3998 | 1.23k | depth += 1; |
3999 | 1.23k | if (depth > maxDepth) { |
4000 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
4001 | 0 | "Maximum entity nesting depth exceeded"); |
4002 | 0 | return; |
4003 | 0 | } |
4004 | | |
4005 | 1.23k | if (pent->flags & XML_ENT_EXPANDING) { |
4006 | 29 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
4007 | 29 | xmlHaltParser(ctxt); |
4008 | 29 | return; |
4009 | 29 | } |
4010 | | |
4011 | | /* |
4012 | | * If we're parsing a default attribute value in DTD content, |
4013 | | * the entity might reference other entities which weren't |
4014 | | * defined yet, so the check isn't reliable. |
4015 | | */ |
4016 | 1.21k | if (ctxt->inSubset == 0) |
4017 | 1.18k | flags = XML_ENT_CHECKED | XML_ENT_VALIDATED; |
4018 | 30 | else |
4019 | 30 | flags = XML_ENT_VALIDATED; |
4020 | | |
4021 | 1.21k | str = pent->content; |
4022 | 1.21k | if (str == NULL) |
4023 | 0 | goto done; |
4024 | | |
4025 | | /* |
4026 | | * Note that entity values are already validated. We only check |
4027 | | * for illegal less-than signs and compute the expanded size |
4028 | | * of the entity. No special handling for multi-byte characters |
4029 | | * is needed. |
4030 | | */ |
4031 | 33.7M | while (!PARSER_STOPPED(ctxt)) { |
4032 | 33.7M | c = *str; |
4033 | | |
4034 | 33.7M | if (c != '&') { |
4035 | 33.7M | if (c == 0) |
4036 | 1.14k | break; |
4037 | | |
4038 | 33.7M | if (c == '<') |
4039 | 620 | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
4040 | 620 | "'<' in entity '%s' is not allowed in attributes " |
4041 | 620 | "values\n", pent->name); |
4042 | | |
4043 | 33.7M | str += 1; |
4044 | 33.7M | } else if (str[1] == '#') { |
4045 | 935 | int val; |
4046 | | |
4047 | 935 | val = xmlParseStringCharRef(ctxt, &str); |
4048 | 935 | if (val == 0) { |
4049 | 6 | pent->content[0] = 0; |
4050 | 6 | break; |
4051 | 6 | } |
4052 | 9.57k | } else { |
4053 | 9.57k | xmlChar *name; |
4054 | 9.57k | xmlEntityPtr ent; |
4055 | | |
4056 | 9.57k | name = xmlParseStringEntityRef(ctxt, &str); |
4057 | 9.57k | if (name == NULL) { |
4058 | 7 | pent->content[0] = 0; |
4059 | 7 | break; |
4060 | 7 | } |
4061 | | |
4062 | 9.57k | ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1); |
4063 | 9.57k | xmlFree(name); |
4064 | | |
4065 | 9.57k | if ((ent != NULL) && |
4066 | 9.57k | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) { |
4067 | 2.60k | if ((ent->flags & flags) != flags) { |
4068 | 598 | pent->flags |= XML_ENT_EXPANDING; |
4069 | 598 | xmlCheckEntityInAttValue(ctxt, ent, depth); |
4070 | 598 | pent->flags &= ~XML_ENT_EXPANDING; |
4071 | 598 | } |
4072 | | |
4073 | 2.60k | xmlSaturatedAdd(&expandedSize, ent->expandedSize); |
4074 | 2.60k | xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST); |
4075 | 2.60k | } |
4076 | 9.57k | } |
4077 | 33.7M | } |
4078 | | |
4079 | 1.21k | done: |
4080 | 1.21k | if (ctxt->inSubset == 0) |
4081 | 1.18k | pent->expandedSize = expandedSize; |
4082 | | |
4083 | 1.21k | pent->flags |= flags; |
4084 | 1.21k | } |
4085 | | |
4086 | | /** |
4087 | | * xmlExpandEntityInAttValue: |
4088 | | * @ctxt: parser context |
4089 | | * @buf: string buffer |
4090 | | * @str: entity or attribute value |
4091 | | * @pent: entity for entity value, NULL for attribute values |
4092 | | * @normalize: whether to collapse whitespace |
4093 | | * @inSpace: whitespace state |
4094 | | * @depth: nesting depth |
4095 | | * @check: whether to check for amplification |
4096 | | * |
4097 | | * Expand general entity references in an entity or attribute value. |
4098 | | * Perform attribute value normalization. |
4099 | | */ |
4100 | | static void |
4101 | | xmlExpandEntityInAttValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf, |
4102 | | const xmlChar *str, xmlEntityPtr pent, int normalize, |
4103 | 357k | int *inSpace, int depth, int check) { |
4104 | 357k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
4105 | 357k | int c, chunkSize; |
4106 | | |
4107 | 357k | if (str == NULL) |
4108 | 0 | return; |
4109 | | |
4110 | 357k | depth += 1; |
4111 | 357k | if (depth > maxDepth) { |
4112 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
4113 | 0 | "Maximum entity nesting depth exceeded"); |
4114 | 0 | return; |
4115 | 0 | } |
4116 | | |
4117 | 357k | if (pent != NULL) { |
4118 | 357k | if (pent->flags & XML_ENT_EXPANDING) { |
4119 | 6 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
4120 | 6 | xmlHaltParser(ctxt); |
4121 | 6 | return; |
4122 | 6 | } |
4123 | | |
4124 | 357k | if (check) { |
4125 | 357k | if (xmlParserEntityCheck(ctxt, pent->length)) |
4126 | 33 | return; |
4127 | 357k | } |
4128 | 357k | } |
4129 | | |
4130 | 357k | chunkSize = 0; |
4131 | | |
4132 | | /* |
4133 | | * Note that entity values are already validated. No special |
4134 | | * handling for multi-byte characters is needed. |
4135 | | */ |
4136 | 1.25G | while (!PARSER_STOPPED(ctxt)) { |
4137 | 1.25G | c = *str; |
4138 | | |
4139 | 1.25G | if (c != '&') { |
4140 | 1.24G | if (c == 0) |
4141 | 325k | break; |
4142 | | |
4143 | | /* |
4144 | | * If this function is called without an entity, it is used to |
4145 | | * expand entities in an attribute content where less-than was |
4146 | | * already unscaped and is allowed. |
4147 | | */ |
4148 | 1.24G | if ((pent != NULL) && (c == '<')) { |
4149 | 32.0k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
4150 | 32.0k | "'<' in entity '%s' is not allowed in attributes " |
4151 | 32.0k | "values\n", pent->name); |
4152 | 32.0k | break; |
4153 | 32.0k | } |
4154 | | |
4155 | 1.24G | if (c <= 0x20) { |
4156 | 1.34M | if ((normalize) && (*inSpace)) { |
4157 | | /* Skip char */ |
4158 | 0 | if (chunkSize > 0) { |
4159 | 0 | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4160 | 0 | chunkSize = 0; |
4161 | 0 | } |
4162 | 1.34M | } else if (c < 0x20) { |
4163 | 1.33M | if (chunkSize > 0) { |
4164 | 163k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4165 | 163k | chunkSize = 0; |
4166 | 163k | } |
4167 | | |
4168 | 1.33M | xmlSBufAddCString(buf, " ", 1); |
4169 | 1.33M | } else { |
4170 | 14.2k | chunkSize += 1; |
4171 | 14.2k | } |
4172 | | |
4173 | 1.34M | *inSpace = 1; |
4174 | 1.24G | } else { |
4175 | 1.24G | chunkSize += 1; |
4176 | 1.24G | *inSpace = 0; |
4177 | 1.24G | } |
4178 | | |
4179 | 1.24G | str += 1; |
4180 | 1.24G | } else if (str[1] == '#') { |
4181 | 50.5k | int val; |
4182 | | |
4183 | 50.5k | if (chunkSize > 0) { |
4184 | 22.0k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4185 | 22.0k | chunkSize = 0; |
4186 | 22.0k | } |
4187 | | |
4188 | 50.5k | val = xmlParseStringCharRef(ctxt, &str); |
4189 | 50.5k | if (val == 0) { |
4190 | 8 | if (pent != NULL) |
4191 | 8 | pent->content[0] = 0; |
4192 | 8 | break; |
4193 | 8 | } |
4194 | | |
4195 | 50.5k | if (val == ' ') { |
4196 | 2.88k | if ((!normalize) || (!*inSpace)) |
4197 | 2.88k | xmlSBufAddCString(buf, " ", 1); |
4198 | 2.88k | *inSpace = 1; |
4199 | 47.6k | } else { |
4200 | 47.6k | xmlSBufAddChar(buf, val); |
4201 | 47.6k | *inSpace = 0; |
4202 | 47.6k | } |
4203 | 1.58M | } else { |
4204 | 1.58M | xmlChar *name; |
4205 | 1.58M | xmlEntityPtr ent; |
4206 | | |
4207 | 1.58M | if (chunkSize > 0) { |
4208 | 1.54M | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4209 | 1.54M | chunkSize = 0; |
4210 | 1.54M | } |
4211 | | |
4212 | 1.58M | name = xmlParseStringEntityRef(ctxt, &str); |
4213 | 1.58M | if (name == NULL) { |
4214 | 5 | if (pent != NULL) |
4215 | 5 | pent->content[0] = 0; |
4216 | 5 | break; |
4217 | 5 | } |
4218 | | |
4219 | 1.58M | ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1); |
4220 | 1.58M | xmlFree(name); |
4221 | | |
4222 | 1.58M | if ((ent != NULL) && |
4223 | 1.58M | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
4224 | 1.46M | if (ent->content == NULL) { |
4225 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
4226 | 0 | "predefined entity has no content\n"); |
4227 | 0 | break; |
4228 | 0 | } |
4229 | | |
4230 | 1.46M | xmlSBufAddString(buf, ent->content, ent->length); |
4231 | | |
4232 | 1.46M | *inSpace = 0; |
4233 | 1.46M | } else if ((ent != NULL) && (ent->content != NULL)) { |
4234 | 104k | if (pent != NULL) |
4235 | 104k | pent->flags |= XML_ENT_EXPANDING; |
4236 | 104k | xmlExpandEntityInAttValue(ctxt, buf, ent->content, ent, |
4237 | 104k | normalize, inSpace, depth, check); |
4238 | 104k | if (pent != NULL) |
4239 | 104k | pent->flags &= ~XML_ENT_EXPANDING; |
4240 | 104k | } |
4241 | 1.58M | } |
4242 | 1.25G | } |
4243 | | |
4244 | 357k | if (chunkSize > 0) |
4245 | 344k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4246 | 357k | } |
4247 | | |
4248 | | /** |
4249 | | * xmlExpandEntitiesInAttValue: |
4250 | | * @ctxt: parser context |
4251 | | * @str: entity or attribute value |
4252 | | * @normalize: whether to collapse whitespace |
4253 | | * |
4254 | | * Expand general entity references in an entity or attribute value. |
4255 | | * Perform attribute value normalization. |
4256 | | * |
4257 | | * Returns the expanded attribtue value. |
4258 | | */ |
4259 | | xmlChar * |
4260 | | xmlExpandEntitiesInAttValue(xmlParserCtxtPtr ctxt, const xmlChar *str, |
4261 | 0 | int normalize) { |
4262 | 0 | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4263 | 0 | XML_MAX_HUGE_LENGTH : |
4264 | 0 | XML_MAX_TEXT_LENGTH; |
4265 | 0 | xmlSBuf buf; |
4266 | 0 | int inSpace = 1; |
4267 | |
|
4268 | 0 | xmlSBufInit(&buf, maxLength); |
4269 | |
|
4270 | 0 | xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace, |
4271 | 0 | ctxt->inputNr, /* check */ 0); |
4272 | |
|
4273 | 0 | if ((normalize) && (inSpace) && (buf.size > 0)) |
4274 | 0 | buf.size--; |
4275 | |
|
4276 | 0 | return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long")); |
4277 | 0 | } |
4278 | | |
4279 | | /** |
4280 | | * xmlParseAttValueInternal: |
4281 | | * @ctxt: an XML parser context |
4282 | | * @len: attribute len result |
4283 | | * @alloc: whether the attribute was reallocated as a new string |
4284 | | * @normalize: if 1 then further non-CDATA normalization must be done |
4285 | | * |
4286 | | * parse a value for an attribute. |
4287 | | * NOTE: if no normalization is needed, the routine will return pointers |
4288 | | * directly from the data buffer. |
4289 | | * |
4290 | | * 3.3.3 Attribute-Value Normalization: |
4291 | | * Before the value of an attribute is passed to the application or |
4292 | | * checked for validity, the XML processor must normalize it as follows: |
4293 | | * - a character reference is processed by appending the referenced |
4294 | | * character to the attribute value |
4295 | | * - an entity reference is processed by recursively processing the |
4296 | | * replacement text of the entity |
4297 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
4298 | | * appending #x20 to the normalized value, except that only a single |
4299 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
4300 | | * parsed entity or the literal entity value of an internal parsed entity |
4301 | | * - other characters are processed by appending them to the normalized value |
4302 | | * If the declared value is not CDATA, then the XML processor must further |
4303 | | * process the normalized attribute value by discarding any leading and |
4304 | | * trailing space (#x20) characters, and by replacing sequences of space |
4305 | | * (#x20) characters by a single space (#x20) character. |
4306 | | * All attributes for which no declaration has been read should be treated |
4307 | | * by a non-validating parser as if declared CDATA. |
4308 | | * |
4309 | | * Returns the AttValue parsed or NULL. The value has to be freed by the |
4310 | | * caller if it was copied, this can be detected by val[*len] == 0. |
4311 | | */ |
4312 | | static xmlChar * |
4313 | | xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *attlen, int *alloc, |
4314 | 686k | int normalize, int isNamespace) { |
4315 | 686k | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4316 | 0 | XML_MAX_HUGE_LENGTH : |
4317 | 686k | XML_MAX_TEXT_LENGTH; |
4318 | 686k | xmlSBuf buf; |
4319 | 686k | xmlChar *ret; |
4320 | 686k | int c, l, quote, flags, chunkSize; |
4321 | 686k | int inSpace = 1; |
4322 | 686k | int replaceEntities; |
4323 | | |
4324 | | /* Always expand namespace URIs */ |
4325 | 686k | replaceEntities = (ctxt->replaceEntities) || (isNamespace); |
4326 | | |
4327 | 686k | xmlSBufInit(&buf, maxLength); |
4328 | | |
4329 | 686k | GROW; |
4330 | | |
4331 | 686k | quote = CUR; |
4332 | 686k | if ((quote != '"') && (quote != '\'')) { |
4333 | 10.5k | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
4334 | 10.5k | return(NULL); |
4335 | 10.5k | } |
4336 | 675k | NEXTL(1); |
4337 | | |
4338 | 675k | if (ctxt->inSubset == 0) |
4339 | 629k | flags = XML_ENT_CHECKED | XML_ENT_VALIDATED; |
4340 | 46.4k | else |
4341 | 46.4k | flags = XML_ENT_VALIDATED; |
4342 | | |
4343 | 675k | inSpace = 1; |
4344 | 675k | chunkSize = 0; |
4345 | | |
4346 | 284M | while (1) { |
4347 | 284M | if (PARSER_STOPPED(ctxt)) |
4348 | 76 | goto error; |
4349 | | |
4350 | 284M | if (CUR_PTR >= ctxt->input->end) { |
4351 | 1.80k | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4352 | 1.80k | "AttValue: ' expected\n"); |
4353 | 1.80k | goto error; |
4354 | 1.80k | } |
4355 | | |
4356 | | /* |
4357 | | * TODO: Check growth threshold |
4358 | | */ |
4359 | 284M | if (ctxt->input->end - CUR_PTR < 10) |
4360 | 53.8k | GROW; |
4361 | | |
4362 | 284M | c = CUR; |
4363 | | |
4364 | 284M | if (c >= 0x80) { |
4365 | 266M | l = xmlUTF8MultibyteLen(ctxt, CUR_PTR, |
4366 | 266M | "invalid character in attribute value\n"); |
4367 | 266M | if (l == 0) { |
4368 | 5.34M | if (chunkSize > 0) { |
4369 | 194k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4370 | 194k | chunkSize = 0; |
4371 | 194k | } |
4372 | 5.34M | xmlSBufAddReplChar(&buf); |
4373 | 5.34M | NEXTL(1); |
4374 | 261M | } else { |
4375 | 261M | chunkSize += l; |
4376 | 261M | NEXTL(l); |
4377 | 261M | } |
4378 | | |
4379 | 266M | inSpace = 0; |
4380 | 266M | } else if (c != '&') { |
4381 | 17.3M | if (c > 0x20) { |
4382 | 9.09M | if (c == quote) |
4383 | 669k | break; |
4384 | | |
4385 | 8.42M | if (c == '<') |
4386 | 164k | xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL); |
4387 | | |
4388 | 8.42M | chunkSize += 1; |
4389 | 8.42M | inSpace = 0; |
4390 | 8.42M | } else if (!IS_BYTE_CHAR(c)) { |
4391 | 2.72M | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
4392 | 2.72M | "invalid character in attribute value\n"); |
4393 | 2.72M | if (chunkSize > 0) { |
4394 | 121k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4395 | 121k | chunkSize = 0; |
4396 | 121k | } |
4397 | 2.72M | xmlSBufAddReplChar(&buf); |
4398 | 2.72M | inSpace = 0; |
4399 | 5.49M | } else { |
4400 | | /* Whitespace */ |
4401 | 5.49M | if ((normalize) && (inSpace)) { |
4402 | | /* Skip char */ |
4403 | 281k | if (chunkSize > 0) { |
4404 | 23.4k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4405 | 23.4k | chunkSize = 0; |
4406 | 23.4k | } |
4407 | 5.21M | } else if (c < 0x20) { |
4408 | | /* Convert to space */ |
4409 | 4.81M | if (chunkSize > 0) { |
4410 | 162k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4411 | 162k | chunkSize = 0; |
4412 | 162k | } |
4413 | | |
4414 | 4.81M | xmlSBufAddCString(&buf, " ", 1); |
4415 | 4.81M | } else { |
4416 | 400k | chunkSize += 1; |
4417 | 400k | } |
4418 | | |
4419 | 5.49M | inSpace = 1; |
4420 | | |
4421 | 5.49M | if ((c == 0xD) && (NXT(1) == 0xA)) |
4422 | 4.88k | CUR_PTR++; |
4423 | 5.49M | } |
4424 | | |
4425 | 16.6M | NEXTL(1); |
4426 | 16.6M | } else if (NXT(1) == '#') { |
4427 | 43.9k | int val; |
4428 | | |
4429 | 43.9k | if (chunkSize > 0) { |
4430 | 25.3k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4431 | 25.3k | chunkSize = 0; |
4432 | 25.3k | } |
4433 | | |
4434 | 43.9k | val = xmlParseCharRef(ctxt); |
4435 | 43.9k | if (val == 0) |
4436 | 3.81k | goto error; |
4437 | | |
4438 | 40.0k | if ((val == '&') && (!replaceEntities)) { |
4439 | | /* |
4440 | | * The reparsing will be done in xmlNodeParseContent() |
4441 | | * called from SAX2.c |
4442 | | */ |
4443 | 3.63k | xmlSBufAddCString(&buf, "&", 5); |
4444 | 3.63k | inSpace = 0; |
4445 | 36.4k | } else if (val == ' ') { |
4446 | 2.59k | if ((!normalize) || (!inSpace)) |
4447 | 2.17k | xmlSBufAddCString(&buf, " ", 1); |
4448 | 2.59k | inSpace = 1; |
4449 | 33.8k | } else { |
4450 | 33.8k | xmlSBufAddChar(&buf, val); |
4451 | 33.8k | inSpace = 0; |
4452 | 33.8k | } |
4453 | 646k | } else { |
4454 | 646k | const xmlChar *name; |
4455 | 646k | xmlEntityPtr ent; |
4456 | | |
4457 | 646k | if (chunkSize > 0) { |
4458 | 498k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4459 | 498k | chunkSize = 0; |
4460 | 498k | } |
4461 | | |
4462 | 646k | name = xmlParseEntityRefInternal(ctxt); |
4463 | 646k | if (name == NULL) { |
4464 | | /* |
4465 | | * Probably a literal '&' which wasn't escaped. |
4466 | | * TODO: Handle gracefully in recovery mode. |
4467 | | */ |
4468 | 51.7k | continue; |
4469 | 51.7k | } |
4470 | | |
4471 | 594k | ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1); |
4472 | 594k | if (ent == NULL) |
4473 | 76.5k | continue; |
4474 | | |
4475 | 518k | if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) { |
4476 | 215k | if ((ent->content[0] == '&') && (!replaceEntities)) |
4477 | 10.0k | xmlSBufAddCString(&buf, "&", 5); |
4478 | 205k | else |
4479 | 205k | xmlSBufAddString(&buf, ent->content, ent->length); |
4480 | 215k | inSpace = 0; |
4481 | 302k | } else if (replaceEntities) { |
4482 | 252k | xmlExpandEntityInAttValue(ctxt, &buf, ent->content, ent, |
4483 | 252k | normalize, &inSpace, ctxt->inputNr, |
4484 | 252k | /* check */ 1); |
4485 | 252k | } else { |
4486 | 50.0k | if ((ent->flags & flags) != flags) |
4487 | 641 | xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr); |
4488 | | |
4489 | 50.0k | if (xmlParserEntityCheck(ctxt, ent->expandedSize)) { |
4490 | 27 | ent->content[0] = 0; |
4491 | 27 | goto error; |
4492 | 27 | } |
4493 | | |
4494 | | /* |
4495 | | * Just output the reference |
4496 | | */ |
4497 | 49.9k | xmlSBufAddCString(&buf, "&", 1); |
4498 | 49.9k | xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name)); |
4499 | 49.9k | xmlSBufAddCString(&buf, ";", 1); |
4500 | | |
4501 | 49.9k | inSpace = 0; |
4502 | 49.9k | } |
4503 | 518k | } |
4504 | 284M | } |
4505 | | |
4506 | 669k | if ((buf.mem == NULL) && (alloc != NULL)) { |
4507 | 555k | ret = (xmlChar *) CUR_PTR - chunkSize; |
4508 | | |
4509 | 555k | if (attlen != NULL) |
4510 | 555k | *attlen = chunkSize; |
4511 | 555k | if ((normalize) && (inSpace) && (chunkSize > 0)) |
4512 | 336 | *attlen -= 1; |
4513 | 555k | *alloc = 0; |
4514 | | |
4515 | | /* Report potential error */ |
4516 | 555k | xmlSBufCleanup(&buf, ctxt, "AttValue length too long"); |
4517 | 555k | } else { |
4518 | 114k | if (chunkSize > 0) |
4519 | 74.0k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4520 | | |
4521 | 114k | if ((normalize) && (inSpace) && (buf.size > 0)) |
4522 | 2.10k | buf.size--; |
4523 | | |
4524 | 114k | ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long"); |
4525 | | |
4526 | 114k | if (ret != NULL) { |
4527 | 114k | if (attlen != NULL) |
4528 | 68.5k | *attlen = buf.size; |
4529 | 114k | if (alloc != NULL) |
4530 | 68.5k | *alloc = 1; |
4531 | 114k | } |
4532 | 114k | } |
4533 | | |
4534 | 669k | NEXTL(1); |
4535 | | |
4536 | 669k | return(ret); |
4537 | | |
4538 | 5.71k | error: |
4539 | 5.71k | xmlSBufCleanup(&buf, ctxt, "AttValue length too long"); |
4540 | 5.71k | return(NULL); |
4541 | 675k | } |
4542 | | |
4543 | | /** |
4544 | | * xmlParseAttValue: |
4545 | | * @ctxt: an XML parser context |
4546 | | * |
4547 | | * DEPRECATED: Internal function, don't use. |
4548 | | * |
4549 | | * parse a value for an attribute |
4550 | | * Note: the parser won't do substitution of entities here, this |
4551 | | * will be handled later in xmlStringGetNodeList |
4552 | | * |
4553 | | * [10] AttValue ::= '"' ([^<&"] | Reference)* '"' | |
4554 | | * "'" ([^<&'] | Reference)* "'" |
4555 | | * |
4556 | | * 3.3.3 Attribute-Value Normalization: |
4557 | | * Before the value of an attribute is passed to the application or |
4558 | | * checked for validity, the XML processor must normalize it as follows: |
4559 | | * - a character reference is processed by appending the referenced |
4560 | | * character to the attribute value |
4561 | | * - an entity reference is processed by recursively processing the |
4562 | | * replacement text of the entity |
4563 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
4564 | | * appending #x20 to the normalized value, except that only a single |
4565 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
4566 | | * parsed entity or the literal entity value of an internal parsed entity |
4567 | | * - other characters are processed by appending them to the normalized value |
4568 | | * If the declared value is not CDATA, then the XML processor must further |
4569 | | * process the normalized attribute value by discarding any leading and |
4570 | | * trailing space (#x20) characters, and by replacing sequences of space |
4571 | | * (#x20) characters by a single space (#x20) character. |
4572 | | * All attributes for which no declaration has been read should be treated |
4573 | | * by a non-validating parser as if declared CDATA. |
4574 | | * |
4575 | | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
4576 | | */ |
4577 | | |
4578 | | |
4579 | | xmlChar * |
4580 | 47.8k | xmlParseAttValue(xmlParserCtxtPtr ctxt) { |
4581 | 47.8k | if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL); |
4582 | 47.8k | return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0)); |
4583 | 47.8k | } |
4584 | | |
4585 | | /** |
4586 | | * xmlParseSystemLiteral: |
4587 | | * @ctxt: an XML parser context |
4588 | | * |
4589 | | * DEPRECATED: Internal function, don't use. |
4590 | | * |
4591 | | * parse an XML Literal |
4592 | | * |
4593 | | * [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") |
4594 | | * |
4595 | | * Returns the SystemLiteral parsed or NULL |
4596 | | */ |
4597 | | |
4598 | | xmlChar * |
4599 | 9.07k | xmlParseSystemLiteral(xmlParserCtxtPtr ctxt) { |
4600 | 9.07k | xmlChar *buf = NULL; |
4601 | 9.07k | int len = 0; |
4602 | 9.07k | int size = XML_PARSER_BUFFER_SIZE; |
4603 | 9.07k | int cur, l; |
4604 | 9.07k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4605 | 0 | XML_MAX_TEXT_LENGTH : |
4606 | 9.07k | XML_MAX_NAME_LENGTH; |
4607 | 9.07k | xmlChar stop; |
4608 | | |
4609 | 9.07k | if (RAW == '"') { |
4610 | 5.15k | NEXT; |
4611 | 5.15k | stop = '"'; |
4612 | 5.15k | } else if (RAW == '\'') { |
4613 | 3.04k | NEXT; |
4614 | 3.04k | stop = '\''; |
4615 | 3.04k | } else { |
4616 | 877 | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4617 | 877 | return(NULL); |
4618 | 877 | } |
4619 | | |
4620 | 8.19k | buf = xmlMalloc(size); |
4621 | 8.19k | if (buf == NULL) { |
4622 | 0 | xmlErrMemory(ctxt); |
4623 | 0 | return(NULL); |
4624 | 0 | } |
4625 | 8.19k | cur = xmlCurrentCharRecover(ctxt, &l); |
4626 | 1.13M | while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */ |
4627 | 1.12M | if (len + 5 >= size) { |
4628 | 1.48k | xmlChar *tmp; |
4629 | | |
4630 | 1.48k | size *= 2; |
4631 | 1.48k | tmp = (xmlChar *) xmlRealloc(buf, size); |
4632 | 1.48k | if (tmp == NULL) { |
4633 | 0 | xmlFree(buf); |
4634 | 0 | xmlErrMemory(ctxt); |
4635 | 0 | return(NULL); |
4636 | 0 | } |
4637 | 1.48k | buf = tmp; |
4638 | 1.48k | } |
4639 | 1.12M | COPY_BUF(buf, len, cur); |
4640 | 1.12M | if (len > maxLength) { |
4641 | 5 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral"); |
4642 | 5 | xmlFree(buf); |
4643 | 5 | return(NULL); |
4644 | 5 | } |
4645 | 1.12M | NEXTL(l); |
4646 | 1.12M | cur = xmlCurrentCharRecover(ctxt, &l); |
4647 | 1.12M | } |
4648 | 8.19k | buf[len] = 0; |
4649 | 8.19k | if (!IS_CHAR(cur)) { |
4650 | 403 | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4651 | 7.79k | } else { |
4652 | 7.79k | NEXT; |
4653 | 7.79k | } |
4654 | 8.19k | return(buf); |
4655 | 8.19k | } |
4656 | | |
4657 | | /** |
4658 | | * xmlParsePubidLiteral: |
4659 | | * @ctxt: an XML parser context |
4660 | | * |
4661 | | * DEPRECATED: Internal function, don't use. |
4662 | | * |
4663 | | * parse an XML public literal |
4664 | | * |
4665 | | * [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'" |
4666 | | * |
4667 | | * Returns the PubidLiteral parsed or NULL. |
4668 | | */ |
4669 | | |
4670 | | xmlChar * |
4671 | 4.68k | xmlParsePubidLiteral(xmlParserCtxtPtr ctxt) { |
4672 | 4.68k | xmlChar *buf = NULL; |
4673 | 4.68k | int len = 0; |
4674 | 4.68k | int size = XML_PARSER_BUFFER_SIZE; |
4675 | 4.68k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4676 | 0 | XML_MAX_TEXT_LENGTH : |
4677 | 4.68k | XML_MAX_NAME_LENGTH; |
4678 | 4.68k | xmlChar cur; |
4679 | 4.68k | xmlChar stop; |
4680 | | |
4681 | 4.68k | if (RAW == '"') { |
4682 | 3.51k | NEXT; |
4683 | 3.51k | stop = '"'; |
4684 | 3.51k | } else if (RAW == '\'') { |
4685 | 804 | NEXT; |
4686 | 804 | stop = '\''; |
4687 | 804 | } else { |
4688 | 371 | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4689 | 371 | return(NULL); |
4690 | 371 | } |
4691 | 4.31k | buf = xmlMalloc(size); |
4692 | 4.31k | if (buf == NULL) { |
4693 | 0 | xmlErrMemory(ctxt); |
4694 | 0 | return(NULL); |
4695 | 0 | } |
4696 | 4.31k | cur = CUR; |
4697 | 299k | while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) && |
4698 | 299k | (PARSER_STOPPED(ctxt) == 0)) { /* checked */ |
4699 | 294k | if (len + 1 >= size) { |
4700 | 340 | xmlChar *tmp; |
4701 | | |
4702 | 340 | size *= 2; |
4703 | 340 | tmp = (xmlChar *) xmlRealloc(buf, size); |
4704 | 340 | if (tmp == NULL) { |
4705 | 0 | xmlErrMemory(ctxt); |
4706 | 0 | xmlFree(buf); |
4707 | 0 | return(NULL); |
4708 | 0 | } |
4709 | 340 | buf = tmp; |
4710 | 340 | } |
4711 | 294k | buf[len++] = cur; |
4712 | 294k | if (len > maxLength) { |
4713 | 1 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID"); |
4714 | 1 | xmlFree(buf); |
4715 | 1 | return(NULL); |
4716 | 1 | } |
4717 | 294k | NEXT; |
4718 | 294k | cur = CUR; |
4719 | 294k | } |
4720 | 4.31k | buf[len] = 0; |
4721 | 4.31k | if (cur != stop) { |
4722 | 781 | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4723 | 3.53k | } else { |
4724 | 3.53k | NEXTL(1); |
4725 | 3.53k | } |
4726 | 4.31k | return(buf); |
4727 | 4.31k | } |
4728 | | |
4729 | | static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial); |
4730 | | |
4731 | | /* |
4732 | | * used for the test in the inner loop of the char data testing |
4733 | | */ |
4734 | | static const unsigned char test_char_data[256] = { |
4735 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4736 | | 0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */ |
4737 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4738 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4739 | | 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */ |
4740 | | 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F, |
4741 | | 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, |
4742 | | 0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */ |
4743 | | 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, |
4744 | | 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F, |
4745 | | 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, |
4746 | | 0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */ |
4747 | | 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, |
4748 | | 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, |
4749 | | 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, |
4750 | | 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F, |
4751 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */ |
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 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4756 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4757 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4758 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4759 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4760 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4761 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4762 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4763 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4764 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4765 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4766 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 |
4767 | | }; |
4768 | | |
4769 | | /** |
4770 | | * xmlParseCharDataInternal: |
4771 | | * @ctxt: an XML parser context |
4772 | | * @partial: buffer may contain partial UTF-8 sequences |
4773 | | * |
4774 | | * Parse character data. Always makes progress if the first char isn't |
4775 | | * '<' or '&'. |
4776 | | * |
4777 | | * The right angle bracket (>) may be represented using the string ">", |
4778 | | * and must, for compatibility, be escaped using ">" or a character |
4779 | | * reference when it appears in the string "]]>" in content, when that |
4780 | | * string is not marking the end of a CDATA section. |
4781 | | * |
4782 | | * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) |
4783 | | */ |
4784 | | static void |
4785 | 9.40M | xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) { |
4786 | 9.40M | const xmlChar *in; |
4787 | 9.40M | int nbchar = 0; |
4788 | 9.40M | int line = ctxt->input->line; |
4789 | 9.40M | int col = ctxt->input->col; |
4790 | 9.40M | int ccol; |
4791 | | |
4792 | 9.40M | GROW; |
4793 | | /* |
4794 | | * Accelerated common case where input don't need to be |
4795 | | * modified before passing it to the handler. |
4796 | | */ |
4797 | 9.40M | in = ctxt->input->cur; |
4798 | 9.43M | do { |
4799 | 9.47M | get_more_space: |
4800 | 10.5M | while (*in == 0x20) { in++; ctxt->input->col++; } |
4801 | 9.47M | if (*in == 0xA) { |
4802 | 797k | do { |
4803 | 797k | ctxt->input->line++; ctxt->input->col = 1; |
4804 | 797k | in++; |
4805 | 797k | } while (*in == 0xA); |
4806 | 40.9k | goto get_more_space; |
4807 | 40.9k | } |
4808 | 9.43M | if (*in == '<') { |
4809 | 42.3k | nbchar = in - ctxt->input->cur; |
4810 | 42.3k | if (nbchar > 0) { |
4811 | 42.3k | const xmlChar *tmp = ctxt->input->cur; |
4812 | 42.3k | ctxt->input->cur = in; |
4813 | | |
4814 | 42.3k | if ((ctxt->sax != NULL) && |
4815 | 42.3k | (ctxt->disableSAX == 0) && |
4816 | 42.3k | (ctxt->sax->ignorableWhitespace != |
4817 | 5.75k | ctxt->sax->characters)) { |
4818 | 0 | if (areBlanks(ctxt, tmp, nbchar, 1)) { |
4819 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4820 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4821 | 0 | tmp, nbchar); |
4822 | 0 | } else { |
4823 | 0 | if (ctxt->sax->characters != NULL) |
4824 | 0 | ctxt->sax->characters(ctxt->userData, |
4825 | 0 | tmp, nbchar); |
4826 | 0 | if (*ctxt->space == -1) |
4827 | 0 | *ctxt->space = -2; |
4828 | 0 | } |
4829 | 42.3k | } else if ((ctxt->sax != NULL) && |
4830 | 42.3k | (ctxt->disableSAX == 0) && |
4831 | 42.3k | (ctxt->sax->characters != NULL)) { |
4832 | 5.75k | ctxt->sax->characters(ctxt->userData, |
4833 | 5.75k | tmp, nbchar); |
4834 | 5.75k | } |
4835 | 42.3k | } |
4836 | 42.3k | return; |
4837 | 42.3k | } |
4838 | | |
4839 | 9.49M | get_more: |
4840 | 9.49M | ccol = ctxt->input->col; |
4841 | 19.2M | while (test_char_data[*in]) { |
4842 | 9.74M | in++; |
4843 | 9.74M | ccol++; |
4844 | 9.74M | } |
4845 | 9.49M | ctxt->input->col = ccol; |
4846 | 9.49M | if (*in == 0xA) { |
4847 | 362k | do { |
4848 | 362k | ctxt->input->line++; ctxt->input->col = 1; |
4849 | 362k | in++; |
4850 | 362k | } while (*in == 0xA); |
4851 | 68.9k | goto get_more; |
4852 | 68.9k | } |
4853 | 9.42M | if (*in == ']') { |
4854 | 37.4k | if ((in[1] == ']') && (in[2] == '>')) { |
4855 | 3.50k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4856 | 3.50k | ctxt->input->cur = in + 1; |
4857 | 3.50k | return; |
4858 | 3.50k | } |
4859 | 33.8k | in++; |
4860 | 33.8k | ctxt->input->col++; |
4861 | 33.8k | goto get_more; |
4862 | 37.4k | } |
4863 | 9.38M | nbchar = in - ctxt->input->cur; |
4864 | 9.38M | if (nbchar > 0) { |
4865 | 1.52M | if ((ctxt->sax != NULL) && |
4866 | 1.52M | (ctxt->disableSAX == 0) && |
4867 | 1.52M | (ctxt->sax->ignorableWhitespace != |
4868 | 13.2k | ctxt->sax->characters) && |
4869 | 1.52M | (IS_BLANK_CH(*ctxt->input->cur))) { |
4870 | 0 | const xmlChar *tmp = ctxt->input->cur; |
4871 | 0 | ctxt->input->cur = in; |
4872 | |
|
4873 | 0 | if (areBlanks(ctxt, tmp, nbchar, 0)) { |
4874 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4875 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4876 | 0 | tmp, nbchar); |
4877 | 0 | } else { |
4878 | 0 | if (ctxt->sax->characters != NULL) |
4879 | 0 | ctxt->sax->characters(ctxt->userData, |
4880 | 0 | tmp, nbchar); |
4881 | 0 | if (*ctxt->space == -1) |
4882 | 0 | *ctxt->space = -2; |
4883 | 0 | } |
4884 | 0 | line = ctxt->input->line; |
4885 | 0 | col = ctxt->input->col; |
4886 | 1.52M | } else if ((ctxt->sax != NULL) && |
4887 | 1.52M | (ctxt->disableSAX == 0)) { |
4888 | 13.2k | if (ctxt->sax->characters != NULL) |
4889 | 13.2k | ctxt->sax->characters(ctxt->userData, |
4890 | 13.2k | ctxt->input->cur, nbchar); |
4891 | 13.2k | line = ctxt->input->line; |
4892 | 13.2k | col = ctxt->input->col; |
4893 | 13.2k | } |
4894 | 1.52M | } |
4895 | 9.38M | ctxt->input->cur = in; |
4896 | 9.38M | if (*in == 0xD) { |
4897 | 76.3k | in++; |
4898 | 76.3k | if (*in == 0xA) { |
4899 | 30.6k | ctxt->input->cur = in; |
4900 | 30.6k | in++; |
4901 | 30.6k | ctxt->input->line++; ctxt->input->col = 1; |
4902 | 30.6k | continue; /* while */ |
4903 | 30.6k | } |
4904 | 45.7k | in--; |
4905 | 45.7k | } |
4906 | 9.35M | if (*in == '<') { |
4907 | 212k | return; |
4908 | 212k | } |
4909 | 9.14M | if (*in == '&') { |
4910 | 113k | return; |
4911 | 113k | } |
4912 | 9.03M | SHRINK; |
4913 | 9.03M | GROW; |
4914 | 9.03M | in = ctxt->input->cur; |
4915 | 9.06M | } while (((*in >= 0x20) && (*in <= 0x7F)) || |
4916 | 9.06M | (*in == 0x09) || (*in == 0x0a)); |
4917 | 9.03M | ctxt->input->line = line; |
4918 | 9.03M | ctxt->input->col = col; |
4919 | 9.03M | xmlParseCharDataComplex(ctxt, partial); |
4920 | 9.03M | } |
4921 | | |
4922 | | /** |
4923 | | * xmlParseCharDataComplex: |
4924 | | * @ctxt: an XML parser context |
4925 | | * @cdata: int indicating whether we are within a CDATA section |
4926 | | * |
4927 | | * Always makes progress if the first char isn't '<' or '&'. |
4928 | | * |
4929 | | * parse a CharData section.this is the fallback function |
4930 | | * of xmlParseCharData() when the parsing requires handling |
4931 | | * of non-ASCII characters. |
4932 | | */ |
4933 | | static void |
4934 | 9.03M | xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) { |
4935 | 9.03M | xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5]; |
4936 | 9.03M | int nbchar = 0; |
4937 | 9.03M | int cur, l; |
4938 | | |
4939 | 9.03M | cur = xmlCurrentCharRecover(ctxt, &l); |
4940 | 73.7M | while ((cur != '<') && /* checked */ |
4941 | 73.7M | (cur != '&') && |
4942 | 73.7M | (IS_CHAR(cur))) { |
4943 | 64.7M | if ((cur == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
4944 | 6.27k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4945 | 6.27k | } |
4946 | 64.7M | COPY_BUF(buf, nbchar, cur); |
4947 | | /* move current position before possible calling of ctxt->sax->characters */ |
4948 | 64.7M | NEXTL(l); |
4949 | 64.7M | if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) { |
4950 | 458k | buf[nbchar] = 0; |
4951 | | |
4952 | | /* |
4953 | | * OK the segment is to be consumed as chars. |
4954 | | */ |
4955 | 458k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4956 | 84.3k | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4957 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4958 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4959 | 0 | buf, nbchar); |
4960 | 84.3k | } else { |
4961 | 84.3k | if (ctxt->sax->characters != NULL) |
4962 | 84.3k | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4963 | 84.3k | if ((ctxt->sax->characters != |
4964 | 84.3k | ctxt->sax->ignorableWhitespace) && |
4965 | 84.3k | (*ctxt->space == -1)) |
4966 | 0 | *ctxt->space = -2; |
4967 | 84.3k | } |
4968 | 84.3k | } |
4969 | 458k | nbchar = 0; |
4970 | 458k | SHRINK; |
4971 | 458k | } |
4972 | 64.7M | cur = xmlCurrentCharRecover(ctxt, &l); |
4973 | 64.7M | } |
4974 | 9.03M | if (nbchar != 0) { |
4975 | 1.89M | buf[nbchar] = 0; |
4976 | | /* |
4977 | | * OK the segment is to be consumed as chars. |
4978 | | */ |
4979 | 1.89M | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4980 | 3.09k | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4981 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4982 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, buf, nbchar); |
4983 | 3.09k | } else { |
4984 | 3.09k | if (ctxt->sax->characters != NULL) |
4985 | 3.09k | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4986 | 3.09k | if ((ctxt->sax->characters != ctxt->sax->ignorableWhitespace) && |
4987 | 3.09k | (*ctxt->space == -1)) |
4988 | 0 | *ctxt->space = -2; |
4989 | 3.09k | } |
4990 | 3.09k | } |
4991 | 1.89M | } |
4992 | | /* |
4993 | | * cur == 0 can mean |
4994 | | * |
4995 | | * - End of buffer. |
4996 | | * - An actual 0 character. |
4997 | | * - An incomplete UTF-8 sequence. This is allowed if partial is set. |
4998 | | */ |
4999 | 9.03M | if (ctxt->input->cur < ctxt->input->end) { |
5000 | 9.02M | if ((cur == 0) && (CUR != 0)) { |
5001 | 679 | if (partial == 0) { |
5002 | 679 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5003 | 679 | "Incomplete UTF-8 sequence starting with %02X\n", CUR); |
5004 | 679 | NEXTL(1); |
5005 | 679 | } |
5006 | 9.02M | } else if ((cur != '<') && (cur != '&')) { |
5007 | | /* Generate the error and skip the offending character */ |
5008 | 8.74M | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5009 | 8.74M | "PCDATA invalid Char value %d\n", cur); |
5010 | 8.74M | NEXTL(l); |
5011 | 8.74M | } |
5012 | 9.02M | } |
5013 | 9.03M | } |
5014 | | |
5015 | | /** |
5016 | | * xmlParseCharData: |
5017 | | * @ctxt: an XML parser context |
5018 | | * @cdata: unused |
5019 | | * |
5020 | | * DEPRECATED: Internal function, don't use. |
5021 | | */ |
5022 | | void |
5023 | 0 | xmlParseCharData(xmlParserCtxtPtr ctxt, ATTRIBUTE_UNUSED int cdata) { |
5024 | 0 | xmlParseCharDataInternal(ctxt, 0); |
5025 | 0 | } |
5026 | | |
5027 | | /** |
5028 | | * xmlParseExternalID: |
5029 | | * @ctxt: an XML parser context |
5030 | | * @publicID: a xmlChar** receiving PubidLiteral |
5031 | | * @strict: indicate whether we should restrict parsing to only |
5032 | | * production [75], see NOTE below |
5033 | | * |
5034 | | * DEPRECATED: Internal function, don't use. |
5035 | | * |
5036 | | * Parse an External ID or a Public ID |
5037 | | * |
5038 | | * NOTE: Productions [75] and [83] interact badly since [75] can generate |
5039 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
5040 | | * |
5041 | | * [75] ExternalID ::= 'SYSTEM' S SystemLiteral |
5042 | | * | 'PUBLIC' S PubidLiteral S SystemLiteral |
5043 | | * |
5044 | | * [83] PublicID ::= 'PUBLIC' S PubidLiteral |
5045 | | * |
5046 | | * Returns the function returns SystemLiteral and in the second |
5047 | | * case publicID receives PubidLiteral, is strict is off |
5048 | | * it is possible to return NULL and have publicID set. |
5049 | | */ |
5050 | | |
5051 | | xmlChar * |
5052 | 17.9k | xmlParseExternalID(xmlParserCtxtPtr ctxt, xmlChar **publicID, int strict) { |
5053 | 17.9k | xmlChar *URI = NULL; |
5054 | | |
5055 | 17.9k | *publicID = NULL; |
5056 | 17.9k | if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) { |
5057 | 6.05k | SKIP(6); |
5058 | 6.05k | if (SKIP_BLANKS == 0) { |
5059 | 370 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5060 | 370 | "Space required after 'SYSTEM'\n"); |
5061 | 370 | } |
5062 | 6.05k | URI = xmlParseSystemLiteral(ctxt); |
5063 | 6.05k | if (URI == NULL) { |
5064 | 313 | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
5065 | 313 | } |
5066 | 11.8k | } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) { |
5067 | 4.68k | SKIP(6); |
5068 | 4.68k | if (SKIP_BLANKS == 0) { |
5069 | 756 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5070 | 756 | "Space required after 'PUBLIC'\n"); |
5071 | 756 | } |
5072 | 4.68k | *publicID = xmlParsePubidLiteral(ctxt); |
5073 | 4.68k | if (*publicID == NULL) { |
5074 | 372 | xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL); |
5075 | 372 | } |
5076 | 4.68k | if (strict) { |
5077 | | /* |
5078 | | * We don't handle [83] so "S SystemLiteral" is required. |
5079 | | */ |
5080 | 2.71k | if (SKIP_BLANKS == 0) { |
5081 | 496 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5082 | 496 | "Space required after the Public Identifier\n"); |
5083 | 496 | } |
5084 | 2.71k | } else { |
5085 | | /* |
5086 | | * We handle [83] so we return immediately, if |
5087 | | * "S SystemLiteral" is not detected. We skip blanks if no |
5088 | | * system literal was found, but this is harmless since we must |
5089 | | * be at the end of a NotationDecl. |
5090 | | */ |
5091 | 1.97k | if (SKIP_BLANKS == 0) return(NULL); |
5092 | 548 | if ((CUR != '\'') && (CUR != '"')) return(NULL); |
5093 | 548 | } |
5094 | 3.02k | URI = xmlParseSystemLiteral(ctxt); |
5095 | 3.02k | if (URI == NULL) { |
5096 | 569 | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
5097 | 569 | } |
5098 | 3.02k | } |
5099 | 16.2k | return(URI); |
5100 | 17.9k | } |
5101 | | |
5102 | | /** |
5103 | | * xmlParseCommentComplex: |
5104 | | * @ctxt: an XML parser context |
5105 | | * @buf: the already parsed part of the buffer |
5106 | | * @len: number of bytes in the buffer |
5107 | | * @size: allocated size of the buffer |
5108 | | * |
5109 | | * Skip an XML (SGML) comment <!-- .... --> |
5110 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
5111 | | * must not occur within comments. " |
5112 | | * This is the slow routine in case the accelerator for ascii didn't work |
5113 | | * |
5114 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
5115 | | */ |
5116 | | static void |
5117 | | xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf, |
5118 | 28.2k | size_t len, size_t size) { |
5119 | 28.2k | int q, ql; |
5120 | 28.2k | int r, rl; |
5121 | 28.2k | int cur, l; |
5122 | 28.2k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5123 | 0 | XML_MAX_HUGE_LENGTH : |
5124 | 28.2k | XML_MAX_TEXT_LENGTH; |
5125 | | |
5126 | 28.2k | if (buf == NULL) { |
5127 | 6.93k | len = 0; |
5128 | 6.93k | size = XML_PARSER_BUFFER_SIZE; |
5129 | 6.93k | buf = xmlMalloc(size); |
5130 | 6.93k | if (buf == NULL) { |
5131 | 0 | xmlErrMemory(ctxt); |
5132 | 0 | return; |
5133 | 0 | } |
5134 | 6.93k | } |
5135 | 28.2k | q = xmlCurrentCharRecover(ctxt, &ql); |
5136 | 28.2k | if (q == 0) |
5137 | 1.16k | goto not_terminated; |
5138 | 27.0k | if (!IS_CHAR(q)) { |
5139 | 6.47k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5140 | 6.47k | "xmlParseComment: invalid xmlChar value %d\n", |
5141 | 6.47k | q); |
5142 | 6.47k | xmlFree (buf); |
5143 | 6.47k | return; |
5144 | 6.47k | } |
5145 | 20.6k | NEXTL(ql); |
5146 | 20.6k | r = xmlCurrentCharRecover(ctxt, &rl); |
5147 | 20.6k | if (r == 0) |
5148 | 558 | goto not_terminated; |
5149 | 20.0k | if (!IS_CHAR(r)) { |
5150 | 1.70k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5151 | 1.70k | "xmlParseComment: invalid xmlChar value %d\n", |
5152 | 1.70k | r); |
5153 | 1.70k | xmlFree (buf); |
5154 | 1.70k | return; |
5155 | 1.70k | } |
5156 | 18.3k | NEXTL(rl); |
5157 | 18.3k | cur = xmlCurrentCharRecover(ctxt, &l); |
5158 | 18.3k | if (cur == 0) |
5159 | 297 | goto not_terminated; |
5160 | 33.0M | while (IS_CHAR(cur) && /* checked */ |
5161 | 33.0M | ((cur != '>') || |
5162 | 33.0M | (r != '-') || (q != '-'))) { |
5163 | 33.0M | if ((r == '-') && (q == '-')) { |
5164 | 107k | xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL); |
5165 | 107k | } |
5166 | 33.0M | if (len + 5 >= size) { |
5167 | 3.19k | xmlChar *new_buf; |
5168 | 3.19k | size_t new_size; |
5169 | | |
5170 | 3.19k | new_size = size * 2; |
5171 | 3.19k | new_buf = (xmlChar *) xmlRealloc(buf, new_size); |
5172 | 3.19k | if (new_buf == NULL) { |
5173 | 0 | xmlFree (buf); |
5174 | 0 | xmlErrMemory(ctxt); |
5175 | 0 | return; |
5176 | 0 | } |
5177 | 3.19k | buf = new_buf; |
5178 | 3.19k | size = new_size; |
5179 | 3.19k | } |
5180 | 33.0M | COPY_BUF(buf, len, q); |
5181 | 33.0M | if (len > maxLength) { |
5182 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5183 | 0 | "Comment too big found", NULL); |
5184 | 0 | xmlFree (buf); |
5185 | 0 | return; |
5186 | 0 | } |
5187 | | |
5188 | 33.0M | q = r; |
5189 | 33.0M | ql = rl; |
5190 | 33.0M | r = cur; |
5191 | 33.0M | rl = l; |
5192 | | |
5193 | 33.0M | NEXTL(l); |
5194 | 33.0M | cur = xmlCurrentCharRecover(ctxt, &l); |
5195 | | |
5196 | 33.0M | } |
5197 | 18.0k | buf[len] = 0; |
5198 | 18.0k | if (cur == 0) { |
5199 | 2.48k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5200 | 2.48k | "Comment not terminated \n<!--%.50s\n", buf); |
5201 | 15.5k | } else if (!IS_CHAR(cur)) { |
5202 | 9.94k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5203 | 9.94k | "xmlParseComment: invalid xmlChar value %d\n", |
5204 | 9.94k | cur); |
5205 | 9.94k | } else { |
5206 | 5.62k | NEXT; |
5207 | 5.62k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5208 | 5.62k | (!ctxt->disableSAX)) |
5209 | 657 | ctxt->sax->comment(ctxt->userData, buf); |
5210 | 5.62k | } |
5211 | 18.0k | xmlFree(buf); |
5212 | 18.0k | return; |
5213 | 2.01k | not_terminated: |
5214 | 2.01k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5215 | 2.01k | "Comment not terminated\n", NULL); |
5216 | 2.01k | xmlFree(buf); |
5217 | 2.01k | } |
5218 | | |
5219 | | /** |
5220 | | * xmlParseComment: |
5221 | | * @ctxt: an XML parser context |
5222 | | * |
5223 | | * DEPRECATED: Internal function, don't use. |
5224 | | * |
5225 | | * Parse an XML (SGML) comment. Always consumes '<!'. |
5226 | | * |
5227 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
5228 | | * must not occur within comments. " |
5229 | | * |
5230 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
5231 | | */ |
5232 | | void |
5233 | 39.6k | xmlParseComment(xmlParserCtxtPtr ctxt) { |
5234 | 39.6k | xmlChar *buf = NULL; |
5235 | 39.6k | size_t size = XML_PARSER_BUFFER_SIZE; |
5236 | 39.6k | size_t len = 0; |
5237 | 39.6k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5238 | 0 | XML_MAX_HUGE_LENGTH : |
5239 | 39.6k | XML_MAX_TEXT_LENGTH; |
5240 | 39.6k | const xmlChar *in; |
5241 | 39.6k | size_t nbchar = 0; |
5242 | 39.6k | int ccol; |
5243 | | |
5244 | | /* |
5245 | | * Check that there is a comment right here. |
5246 | | */ |
5247 | 39.6k | if ((RAW != '<') || (NXT(1) != '!')) |
5248 | 0 | return; |
5249 | 39.6k | SKIP(2); |
5250 | 39.6k | if ((RAW != '-') || (NXT(1) != '-')) |
5251 | 6 | return; |
5252 | 39.6k | SKIP(2); |
5253 | 39.6k | GROW; |
5254 | | |
5255 | | /* |
5256 | | * Accelerated common case where input don't need to be |
5257 | | * modified before passing it to the handler. |
5258 | | */ |
5259 | 39.6k | in = ctxt->input->cur; |
5260 | 39.9k | do { |
5261 | 39.9k | if (*in == 0xA) { |
5262 | 271k | do { |
5263 | 271k | ctxt->input->line++; ctxt->input->col = 1; |
5264 | 271k | in++; |
5265 | 271k | } while (*in == 0xA); |
5266 | 2.14k | } |
5267 | 757k | get_more: |
5268 | 757k | ccol = ctxt->input->col; |
5269 | 3.28M | while (((*in > '-') && (*in <= 0x7F)) || |
5270 | 3.28M | ((*in >= 0x20) && (*in < '-')) || |
5271 | 3.28M | (*in == 0x09)) { |
5272 | 2.52M | in++; |
5273 | 2.52M | ccol++; |
5274 | 2.52M | } |
5275 | 757k | ctxt->input->col = ccol; |
5276 | 757k | if (*in == 0xA) { |
5277 | 270k | do { |
5278 | 270k | ctxt->input->line++; ctxt->input->col = 1; |
5279 | 270k | in++; |
5280 | 270k | } while (*in == 0xA); |
5281 | 7.91k | goto get_more; |
5282 | 7.91k | } |
5283 | 749k | nbchar = in - ctxt->input->cur; |
5284 | | /* |
5285 | | * save current set of data |
5286 | | */ |
5287 | 749k | if (nbchar > 0) { |
5288 | 728k | if (buf == NULL) { |
5289 | 24.4k | if ((*in == '-') && (in[1] == '-')) |
5290 | 4.21k | size = nbchar + 1; |
5291 | 20.1k | else |
5292 | 20.1k | size = XML_PARSER_BUFFER_SIZE + nbchar; |
5293 | 24.4k | buf = xmlMalloc(size); |
5294 | 24.4k | if (buf == NULL) { |
5295 | 0 | xmlErrMemory(ctxt); |
5296 | 0 | return; |
5297 | 0 | } |
5298 | 24.4k | len = 0; |
5299 | 704k | } else if (len + nbchar + 1 >= size) { |
5300 | 4.37k | xmlChar *new_buf; |
5301 | 4.37k | size += len + nbchar + XML_PARSER_BUFFER_SIZE; |
5302 | 4.37k | new_buf = (xmlChar *) xmlRealloc(buf, size); |
5303 | 4.37k | if (new_buf == NULL) { |
5304 | 0 | xmlFree (buf); |
5305 | 0 | xmlErrMemory(ctxt); |
5306 | 0 | return; |
5307 | 0 | } |
5308 | 4.37k | buf = new_buf; |
5309 | 4.37k | } |
5310 | 728k | memcpy(&buf[len], ctxt->input->cur, nbchar); |
5311 | 728k | len += nbchar; |
5312 | 728k | buf[len] = 0; |
5313 | 728k | } |
5314 | 749k | if (len > maxLength) { |
5315 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5316 | 0 | "Comment too big found", NULL); |
5317 | 0 | xmlFree (buf); |
5318 | 0 | return; |
5319 | 0 | } |
5320 | 749k | ctxt->input->cur = in; |
5321 | 749k | if (*in == 0xA) { |
5322 | 0 | in++; |
5323 | 0 | ctxt->input->line++; ctxt->input->col = 1; |
5324 | 0 | } |
5325 | 749k | if (*in == 0xD) { |
5326 | 9.01k | in++; |
5327 | 9.01k | if (*in == 0xA) { |
5328 | 2.68k | ctxt->input->cur = in; |
5329 | 2.68k | in++; |
5330 | 2.68k | ctxt->input->line++; ctxt->input->col = 1; |
5331 | 2.68k | goto get_more; |
5332 | 2.68k | } |
5333 | 6.33k | in--; |
5334 | 6.33k | } |
5335 | 746k | SHRINK; |
5336 | 746k | GROW; |
5337 | 746k | in = ctxt->input->cur; |
5338 | 746k | if (*in == '-') { |
5339 | 718k | if (in[1] == '-') { |
5340 | 664k | if (in[2] == '>') { |
5341 | 11.3k | SKIP(3); |
5342 | 11.3k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5343 | 11.3k | (!ctxt->disableSAX)) { |
5344 | 2.05k | if (buf != NULL) |
5345 | 545 | ctxt->sax->comment(ctxt->userData, buf); |
5346 | 1.50k | else |
5347 | 1.50k | ctxt->sax->comment(ctxt->userData, BAD_CAST ""); |
5348 | 2.05k | } |
5349 | 11.3k | if (buf != NULL) |
5350 | 3.10k | xmlFree(buf); |
5351 | 11.3k | return; |
5352 | 11.3k | } |
5353 | 652k | if (buf != NULL) { |
5354 | 650k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5355 | 650k | "Double hyphen within comment: " |
5356 | 650k | "<!--%.50s\n", |
5357 | 650k | buf); |
5358 | 650k | } else |
5359 | 1.90k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5360 | 1.90k | "Double hyphen within comment\n", NULL); |
5361 | 652k | in++; |
5362 | 652k | ctxt->input->col++; |
5363 | 652k | } |
5364 | 706k | in++; |
5365 | 706k | ctxt->input->col++; |
5366 | 706k | goto get_more; |
5367 | 718k | } |
5368 | 746k | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
5369 | 28.2k | xmlParseCommentComplex(ctxt, buf, len, size); |
5370 | 28.2k | } |
5371 | | |
5372 | | |
5373 | | /** |
5374 | | * xmlParsePITarget: |
5375 | | * @ctxt: an XML parser context |
5376 | | * |
5377 | | * DEPRECATED: Internal function, don't use. |
5378 | | * |
5379 | | * parse the name of a PI |
5380 | | * |
5381 | | * [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) |
5382 | | * |
5383 | | * Returns the PITarget name or NULL |
5384 | | */ |
5385 | | |
5386 | | const xmlChar * |
5387 | 37.1k | xmlParsePITarget(xmlParserCtxtPtr ctxt) { |
5388 | 37.1k | const xmlChar *name; |
5389 | | |
5390 | 37.1k | name = xmlParseName(ctxt); |
5391 | 37.1k | if ((name != NULL) && |
5392 | 37.1k | ((name[0] == 'x') || (name[0] == 'X')) && |
5393 | 37.1k | ((name[1] == 'm') || (name[1] == 'M')) && |
5394 | 37.1k | ((name[2] == 'l') || (name[2] == 'L'))) { |
5395 | 8.59k | int i; |
5396 | 8.59k | if ((name[0] == 'x') && (name[1] == 'm') && |
5397 | 8.59k | (name[2] == 'l') && (name[3] == 0)) { |
5398 | 4.42k | xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5399 | 4.42k | "XML declaration allowed only at the start of the document\n"); |
5400 | 4.42k | return(name); |
5401 | 4.42k | } else if (name[3] == 0) { |
5402 | 1.58k | xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL); |
5403 | 1.58k | return(name); |
5404 | 1.58k | } |
5405 | 7.15k | for (i = 0;;i++) { |
5406 | 7.15k | if (xmlW3CPIs[i] == NULL) break; |
5407 | 4.94k | if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i])) |
5408 | 372 | return(name); |
5409 | 4.94k | } |
5410 | 2.20k | xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5411 | 2.20k | "xmlParsePITarget: invalid name prefix 'xml'\n", |
5412 | 2.20k | NULL, NULL); |
5413 | 2.20k | } |
5414 | 30.7k | if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) { |
5415 | 822 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5416 | 822 | "colons are forbidden from PI names '%s'\n", name, NULL, NULL); |
5417 | 822 | } |
5418 | 30.7k | return(name); |
5419 | 37.1k | } |
5420 | | |
5421 | | #ifdef LIBXML_CATALOG_ENABLED |
5422 | | /** |
5423 | | * xmlParseCatalogPI: |
5424 | | * @ctxt: an XML parser context |
5425 | | * @catalog: the PI value string |
5426 | | * |
5427 | | * parse an XML Catalog Processing Instruction. |
5428 | | * |
5429 | | * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?> |
5430 | | * |
5431 | | * Occurs only if allowed by the user and if happening in the Misc |
5432 | | * part of the document before any doctype information |
5433 | | * This will add the given catalog to the parsing context in order |
5434 | | * to be used if there is a resolution need further down in the document |
5435 | | */ |
5436 | | |
5437 | | static void |
5438 | 4.04k | xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) { |
5439 | 4.04k | xmlChar *URL = NULL; |
5440 | 4.04k | const xmlChar *tmp, *base; |
5441 | 4.04k | xmlChar marker; |
5442 | | |
5443 | 4.04k | tmp = catalog; |
5444 | 4.04k | while (IS_BLANK_CH(*tmp)) tmp++; |
5445 | 4.04k | if (xmlStrncmp(tmp, BAD_CAST"catalog", 7)) |
5446 | 839 | goto error; |
5447 | 3.20k | tmp += 7; |
5448 | 3.20k | while (IS_BLANK_CH(*tmp)) tmp++; |
5449 | 3.20k | if (*tmp != '=') { |
5450 | 406 | return; |
5451 | 406 | } |
5452 | 2.79k | tmp++; |
5453 | 8.49k | while (IS_BLANK_CH(*tmp)) tmp++; |
5454 | 2.79k | marker = *tmp; |
5455 | 2.79k | if ((marker != '\'') && (marker != '"')) |
5456 | 677 | goto error; |
5457 | 2.11k | tmp++; |
5458 | 2.11k | base = tmp; |
5459 | 13.3k | while ((*tmp != 0) && (*tmp != marker)) tmp++; |
5460 | 2.11k | if (*tmp == 0) |
5461 | 294 | goto error; |
5462 | 1.82k | URL = xmlStrndup(base, tmp - base); |
5463 | 1.82k | tmp++; |
5464 | 14.0k | while (IS_BLANK_CH(*tmp)) tmp++; |
5465 | 1.82k | if (*tmp != 0) |
5466 | 395 | goto error; |
5467 | | |
5468 | 1.42k | if (URL != NULL) { |
5469 | | /* |
5470 | | * Unfortunately, the catalog API doesn't report OOM errors. |
5471 | | * xmlGetLastError isn't very helpful since we don't know |
5472 | | * where the last error came from. We'd have to reset it |
5473 | | * before this call and restore it afterwards. |
5474 | | */ |
5475 | 1.42k | ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL); |
5476 | 1.42k | xmlFree(URL); |
5477 | 1.42k | } |
5478 | 1.42k | return; |
5479 | | |
5480 | 2.20k | error: |
5481 | 2.20k | xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI, |
5482 | 2.20k | "Catalog PI syntax error: %s\n", |
5483 | 2.20k | catalog, NULL); |
5484 | 2.20k | if (URL != NULL) |
5485 | 395 | xmlFree(URL); |
5486 | 2.20k | } |
5487 | | #endif |
5488 | | |
5489 | | /** |
5490 | | * xmlParsePI: |
5491 | | * @ctxt: an XML parser context |
5492 | | * |
5493 | | * DEPRECATED: Internal function, don't use. |
5494 | | * |
5495 | | * parse an XML Processing Instruction. |
5496 | | * |
5497 | | * [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' |
5498 | | * |
5499 | | * The processing is transferred to SAX once parsed. |
5500 | | */ |
5501 | | |
5502 | | void |
5503 | 37.1k | xmlParsePI(xmlParserCtxtPtr ctxt) { |
5504 | 37.1k | xmlChar *buf = NULL; |
5505 | 37.1k | size_t len = 0; |
5506 | 37.1k | size_t size = XML_PARSER_BUFFER_SIZE; |
5507 | 37.1k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5508 | 0 | XML_MAX_HUGE_LENGTH : |
5509 | 37.1k | XML_MAX_TEXT_LENGTH; |
5510 | 37.1k | int cur, l; |
5511 | 37.1k | const xmlChar *target; |
5512 | | |
5513 | 37.1k | if ((RAW == '<') && (NXT(1) == '?')) { |
5514 | | /* |
5515 | | * this is a Processing Instruction. |
5516 | | */ |
5517 | 37.1k | SKIP(2); |
5518 | | |
5519 | | /* |
5520 | | * Parse the target name and check for special support like |
5521 | | * namespace. |
5522 | | */ |
5523 | 37.1k | target = xmlParsePITarget(ctxt); |
5524 | 37.1k | if (target != NULL) { |
5525 | 32.8k | if ((RAW == '?') && (NXT(1) == '>')) { |
5526 | 7.22k | SKIP(2); |
5527 | | |
5528 | | /* |
5529 | | * SAX: PI detected. |
5530 | | */ |
5531 | 7.22k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5532 | 7.22k | (ctxt->sax->processingInstruction != NULL)) |
5533 | 3.32k | ctxt->sax->processingInstruction(ctxt->userData, |
5534 | 3.32k | target, NULL); |
5535 | 7.22k | return; |
5536 | 7.22k | } |
5537 | 25.6k | buf = xmlMalloc(size); |
5538 | 25.6k | if (buf == NULL) { |
5539 | 0 | xmlErrMemory(ctxt); |
5540 | 0 | return; |
5541 | 0 | } |
5542 | 25.6k | if (SKIP_BLANKS == 0) { |
5543 | 12.8k | xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED, |
5544 | 12.8k | "ParsePI: PI %s space expected\n", target); |
5545 | 12.8k | } |
5546 | 25.6k | cur = xmlCurrentCharRecover(ctxt, &l); |
5547 | 15.0M | while (IS_CHAR(cur) && /* checked */ |
5548 | 15.0M | ((cur != '?') || (NXT(1) != '>'))) { |
5549 | 15.0M | if (len + 5 >= size) { |
5550 | 8.13k | xmlChar *tmp; |
5551 | 8.13k | size_t new_size = size * 2; |
5552 | 8.13k | tmp = (xmlChar *) xmlRealloc(buf, new_size); |
5553 | 8.13k | if (tmp == NULL) { |
5554 | 0 | xmlErrMemory(ctxt); |
5555 | 0 | xmlFree(buf); |
5556 | 0 | return; |
5557 | 0 | } |
5558 | 8.13k | buf = tmp; |
5559 | 8.13k | size = new_size; |
5560 | 8.13k | } |
5561 | 15.0M | COPY_BUF(buf, len, cur); |
5562 | 15.0M | if (len > maxLength) { |
5563 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5564 | 0 | "PI %s too big found", target); |
5565 | 0 | xmlFree(buf); |
5566 | 0 | return; |
5567 | 0 | } |
5568 | 15.0M | NEXTL(l); |
5569 | 15.0M | cur = xmlCurrentCharRecover(ctxt, &l); |
5570 | 15.0M | } |
5571 | 25.6k | buf[len] = 0; |
5572 | 25.6k | if (cur != '?') { |
5573 | 12.2k | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5574 | 12.2k | "ParsePI: PI %s never end ...\n", target); |
5575 | 13.4k | } else { |
5576 | 13.4k | SKIP(2); |
5577 | | |
5578 | 13.4k | #ifdef LIBXML_CATALOG_ENABLED |
5579 | 13.4k | if ((ctxt->inSubset == 0) && |
5580 | 13.4k | (xmlStrEqual(target, XML_CATALOG_PI))) { |
5581 | 4.04k | xmlCatalogAllow allow = xmlCatalogGetDefaults(); |
5582 | | |
5583 | 4.04k | if (((ctxt->options & XML_PARSE_NO_CATALOG_PI) == 0) && |
5584 | 4.04k | ((allow == XML_CATA_ALLOW_DOCUMENT) || |
5585 | 4.04k | (allow == XML_CATA_ALLOW_ALL))) |
5586 | 4.04k | xmlParseCatalogPI(ctxt, buf); |
5587 | 4.04k | } |
5588 | 13.4k | #endif |
5589 | | |
5590 | | /* |
5591 | | * SAX: PI detected. |
5592 | | */ |
5593 | 13.4k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5594 | 13.4k | (ctxt->sax->processingInstruction != NULL)) |
5595 | 2.28k | ctxt->sax->processingInstruction(ctxt->userData, |
5596 | 2.28k | target, buf); |
5597 | 13.4k | } |
5598 | 25.6k | xmlFree(buf); |
5599 | 25.6k | } else { |
5600 | 4.32k | xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL); |
5601 | 4.32k | } |
5602 | 37.1k | } |
5603 | 37.1k | } |
5604 | | |
5605 | | /** |
5606 | | * xmlParseNotationDecl: |
5607 | | * @ctxt: an XML parser context |
5608 | | * |
5609 | | * DEPRECATED: Internal function, don't use. |
5610 | | * |
5611 | | * Parse a notation declaration. Always consumes '<!'. |
5612 | | * |
5613 | | * [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>' |
5614 | | * |
5615 | | * Hence there is actually 3 choices: |
5616 | | * 'PUBLIC' S PubidLiteral |
5617 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
5618 | | * and 'SYSTEM' S SystemLiteral |
5619 | | * |
5620 | | * See the NOTE on xmlParseExternalID(). |
5621 | | */ |
5622 | | |
5623 | | void |
5624 | 3.30k | xmlParseNotationDecl(xmlParserCtxtPtr ctxt) { |
5625 | 3.30k | const xmlChar *name; |
5626 | 3.30k | xmlChar *Pubid; |
5627 | 3.30k | xmlChar *Systemid; |
5628 | | |
5629 | 3.30k | if ((CUR != '<') || (NXT(1) != '!')) |
5630 | 0 | return; |
5631 | 3.30k | SKIP(2); |
5632 | | |
5633 | 3.30k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5634 | 3.28k | int inputid = ctxt->input->id; |
5635 | 3.28k | SKIP(8); |
5636 | 3.28k | if (SKIP_BLANKS_PE == 0) { |
5637 | 28 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5638 | 28 | "Space required after '<!NOTATION'\n"); |
5639 | 28 | return; |
5640 | 28 | } |
5641 | | |
5642 | 3.25k | name = xmlParseName(ctxt); |
5643 | 3.25k | if (name == NULL) { |
5644 | 69 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5645 | 69 | return; |
5646 | 69 | } |
5647 | 3.18k | if (xmlStrchr(name, ':') != NULL) { |
5648 | 135 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5649 | 135 | "colons are forbidden from notation names '%s'\n", |
5650 | 135 | name, NULL, NULL); |
5651 | 135 | } |
5652 | 3.18k | if (SKIP_BLANKS_PE == 0) { |
5653 | 145 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5654 | 145 | "Space required after the NOTATION name'\n"); |
5655 | 145 | return; |
5656 | 145 | } |
5657 | | |
5658 | | /* |
5659 | | * Parse the IDs. |
5660 | | */ |
5661 | 3.04k | Systemid = xmlParseExternalID(ctxt, &Pubid, 0); |
5662 | 3.04k | SKIP_BLANKS_PE; |
5663 | | |
5664 | 3.04k | if (RAW == '>') { |
5665 | 2.02k | if (inputid != ctxt->input->id) { |
5666 | 200 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5667 | 200 | "Notation declaration doesn't start and stop" |
5668 | 200 | " in the same entity\n"); |
5669 | 200 | } |
5670 | 2.02k | NEXT; |
5671 | 2.02k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5672 | 2.02k | (ctxt->sax->notationDecl != NULL)) |
5673 | 1.41k | ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid); |
5674 | 2.02k | } else { |
5675 | 1.01k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5676 | 1.01k | } |
5677 | 3.04k | if (Systemid != NULL) xmlFree(Systemid); |
5678 | 3.04k | if (Pubid != NULL) xmlFree(Pubid); |
5679 | 3.04k | } |
5680 | 3.30k | } |
5681 | | |
5682 | | /** |
5683 | | * xmlParseEntityDecl: |
5684 | | * @ctxt: an XML parser context |
5685 | | * |
5686 | | * DEPRECATED: Internal function, don't use. |
5687 | | * |
5688 | | * Parse an entity declaration. Always consumes '<!'. |
5689 | | * |
5690 | | * [70] EntityDecl ::= GEDecl | PEDecl |
5691 | | * |
5692 | | * [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>' |
5693 | | * |
5694 | | * [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>' |
5695 | | * |
5696 | | * [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?) |
5697 | | * |
5698 | | * [74] PEDef ::= EntityValue | ExternalID |
5699 | | * |
5700 | | * [76] NDataDecl ::= S 'NDATA' S Name |
5701 | | * |
5702 | | * [ VC: Notation Declared ] |
5703 | | * The Name must match the declared name of a notation. |
5704 | | */ |
5705 | | |
5706 | | void |
5707 | 54.1k | xmlParseEntityDecl(xmlParserCtxtPtr ctxt) { |
5708 | 54.1k | const xmlChar *name = NULL; |
5709 | 54.1k | xmlChar *value = NULL; |
5710 | 54.1k | xmlChar *URI = NULL, *literal = NULL; |
5711 | 54.1k | const xmlChar *ndata = NULL; |
5712 | 54.1k | int isParameter = 0; |
5713 | 54.1k | xmlChar *orig = NULL; |
5714 | | |
5715 | 54.1k | if ((CUR != '<') || (NXT(1) != '!')) |
5716 | 0 | return; |
5717 | 54.1k | SKIP(2); |
5718 | | |
5719 | | /* GROW; done in the caller */ |
5720 | 54.1k | if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5721 | 54.0k | int inputid = ctxt->input->id; |
5722 | 54.0k | SKIP(6); |
5723 | 54.0k | if (SKIP_BLANKS_PE == 0) { |
5724 | 20.7k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5725 | 20.7k | "Space required after '<!ENTITY'\n"); |
5726 | 20.7k | } |
5727 | | |
5728 | 54.0k | if (RAW == '%') { |
5729 | 8.29k | NEXT; |
5730 | 8.29k | if (SKIP_BLANKS_PE == 0) { |
5731 | 2.18k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5732 | 2.18k | "Space required after '%%'\n"); |
5733 | 2.18k | } |
5734 | 8.29k | isParameter = 1; |
5735 | 8.29k | } |
5736 | | |
5737 | 54.0k | name = xmlParseName(ctxt); |
5738 | 54.0k | if (name == NULL) { |
5739 | 304 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5740 | 304 | "xmlParseEntityDecl: no name\n"); |
5741 | 304 | return; |
5742 | 304 | } |
5743 | 53.7k | if (xmlStrchr(name, ':') != NULL) { |
5744 | 2.69k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5745 | 2.69k | "colons are forbidden from entities names '%s'\n", |
5746 | 2.69k | name, NULL, NULL); |
5747 | 2.69k | } |
5748 | 53.7k | if (SKIP_BLANKS_PE == 0) { |
5749 | 25.6k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5750 | 25.6k | "Space required after the entity name\n"); |
5751 | 25.6k | } |
5752 | | |
5753 | | /* |
5754 | | * handle the various case of definitions... |
5755 | | */ |
5756 | 53.7k | if (isParameter) { |
5757 | 8.26k | if ((RAW == '"') || (RAW == '\'')) { |
5758 | 5.31k | value = xmlParseEntityValue(ctxt, &orig); |
5759 | 5.31k | if (value) { |
5760 | 5.28k | if ((ctxt->sax != NULL) && |
5761 | 5.28k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5762 | 2.06k | ctxt->sax->entityDecl(ctxt->userData, name, |
5763 | 2.06k | XML_INTERNAL_PARAMETER_ENTITY, |
5764 | 2.06k | NULL, NULL, value); |
5765 | 5.28k | } |
5766 | 5.31k | } else { |
5767 | 2.95k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5768 | 2.95k | if ((URI == NULL) && (literal == NULL)) { |
5769 | 113 | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5770 | 113 | } |
5771 | 2.95k | if (URI) { |
5772 | 2.64k | if (xmlStrchr(URI, '#')) { |
5773 | 693 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5774 | 1.95k | } else { |
5775 | 1.95k | if ((ctxt->sax != NULL) && |
5776 | 1.95k | (!ctxt->disableSAX) && |
5777 | 1.95k | (ctxt->sax->entityDecl != NULL)) |
5778 | 442 | ctxt->sax->entityDecl(ctxt->userData, name, |
5779 | 442 | XML_EXTERNAL_PARAMETER_ENTITY, |
5780 | 442 | literal, URI, NULL); |
5781 | 1.95k | } |
5782 | 2.64k | } |
5783 | 2.95k | } |
5784 | 45.5k | } else { |
5785 | 45.5k | if ((RAW == '"') || (RAW == '\'')) { |
5786 | 40.5k | value = xmlParseEntityValue(ctxt, &orig); |
5787 | 40.5k | if ((ctxt->sax != NULL) && |
5788 | 40.5k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5789 | 4.91k | ctxt->sax->entityDecl(ctxt->userData, name, |
5790 | 4.91k | XML_INTERNAL_GENERAL_ENTITY, |
5791 | 4.91k | NULL, NULL, value); |
5792 | | /* |
5793 | | * For expat compatibility in SAX mode. |
5794 | | */ |
5795 | 40.5k | if ((ctxt->myDoc == NULL) || |
5796 | 40.5k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
5797 | 3.48k | if (ctxt->myDoc == NULL) { |
5798 | 469 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5799 | 469 | if (ctxt->myDoc == NULL) { |
5800 | 0 | xmlErrMemory(ctxt); |
5801 | 0 | goto done; |
5802 | 0 | } |
5803 | 469 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5804 | 469 | } |
5805 | 3.48k | if (ctxt->myDoc->intSubset == NULL) { |
5806 | 469 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5807 | 469 | BAD_CAST "fake", NULL, NULL); |
5808 | 469 | if (ctxt->myDoc->intSubset == NULL) { |
5809 | 0 | xmlErrMemory(ctxt); |
5810 | 0 | goto done; |
5811 | 0 | } |
5812 | 469 | } |
5813 | | |
5814 | 3.48k | xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY, |
5815 | 3.48k | NULL, NULL, value); |
5816 | 3.48k | } |
5817 | 40.5k | } else { |
5818 | 4.98k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5819 | 4.98k | if ((URI == NULL) && (literal == NULL)) { |
5820 | 569 | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5821 | 569 | } |
5822 | 4.98k | if (URI) { |
5823 | 4.27k | if (xmlStrchr(URI, '#')) { |
5824 | 707 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5825 | 707 | } |
5826 | 4.27k | } |
5827 | 4.98k | if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) { |
5828 | 255 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5829 | 255 | "Space required before 'NDATA'\n"); |
5830 | 255 | } |
5831 | 4.98k | if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) { |
5832 | 749 | SKIP(5); |
5833 | 749 | if (SKIP_BLANKS_PE == 0) { |
5834 | 188 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5835 | 188 | "Space required after 'NDATA'\n"); |
5836 | 188 | } |
5837 | 749 | ndata = xmlParseName(ctxt); |
5838 | 749 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5839 | 749 | (ctxt->sax->unparsedEntityDecl != NULL)) |
5840 | 189 | ctxt->sax->unparsedEntityDecl(ctxt->userData, name, |
5841 | 189 | literal, URI, ndata); |
5842 | 4.23k | } else { |
5843 | 4.23k | if ((ctxt->sax != NULL) && |
5844 | 4.23k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5845 | 2.07k | ctxt->sax->entityDecl(ctxt->userData, name, |
5846 | 2.07k | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5847 | 2.07k | literal, URI, NULL); |
5848 | | /* |
5849 | | * For expat compatibility in SAX mode. |
5850 | | * assuming the entity replacement was asked for |
5851 | | */ |
5852 | 4.23k | if ((ctxt->replaceEntities != 0) && |
5853 | 4.23k | ((ctxt->myDoc == NULL) || |
5854 | 0 | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) { |
5855 | 0 | if (ctxt->myDoc == NULL) { |
5856 | 0 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5857 | 0 | if (ctxt->myDoc == NULL) { |
5858 | 0 | xmlErrMemory(ctxt); |
5859 | 0 | goto done; |
5860 | 0 | } |
5861 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5862 | 0 | } |
5863 | | |
5864 | 0 | if (ctxt->myDoc->intSubset == NULL) { |
5865 | 0 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5866 | 0 | BAD_CAST "fake", NULL, NULL); |
5867 | 0 | if (ctxt->myDoc->intSubset == NULL) { |
5868 | 0 | xmlErrMemory(ctxt); |
5869 | 0 | goto done; |
5870 | 0 | } |
5871 | 0 | } |
5872 | 0 | xmlSAX2EntityDecl(ctxt, name, |
5873 | 0 | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5874 | 0 | literal, URI, NULL); |
5875 | 0 | } |
5876 | 4.23k | } |
5877 | 4.98k | } |
5878 | 45.5k | } |
5879 | 53.7k | SKIP_BLANKS_PE; |
5880 | 53.7k | if (RAW != '>') { |
5881 | 725 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
5882 | 725 | "xmlParseEntityDecl: entity %s not terminated\n", name); |
5883 | 725 | xmlHaltParser(ctxt); |
5884 | 53.0k | } else { |
5885 | 53.0k | if (inputid != ctxt->input->id) { |
5886 | 432 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5887 | 432 | "Entity declaration doesn't start and stop in" |
5888 | 432 | " the same entity\n"); |
5889 | 432 | } |
5890 | 53.0k | NEXT; |
5891 | 53.0k | } |
5892 | 53.7k | if (orig != NULL) { |
5893 | | /* |
5894 | | * Ugly mechanism to save the raw entity value. |
5895 | | */ |
5896 | 45.4k | xmlEntityPtr cur = NULL; |
5897 | | |
5898 | 45.4k | if (isParameter) { |
5899 | 5.28k | if ((ctxt->sax != NULL) && |
5900 | 5.28k | (ctxt->sax->getParameterEntity != NULL)) |
5901 | 5.28k | cur = ctxt->sax->getParameterEntity(ctxt->userData, name); |
5902 | 40.1k | } else { |
5903 | 40.1k | if ((ctxt->sax != NULL) && |
5904 | 40.1k | (ctxt->sax->getEntity != NULL)) |
5905 | 40.1k | cur = ctxt->sax->getEntity(ctxt->userData, name); |
5906 | 40.1k | if ((cur == NULL) && (ctxt->userData==ctxt)) { |
5907 | 28.0k | cur = xmlSAX2GetEntity(ctxt, name); |
5908 | 28.0k | } |
5909 | 40.1k | } |
5910 | 45.4k | if ((cur != NULL) && (cur->orig == NULL)) { |
5911 | 4.77k | cur->orig = orig; |
5912 | 4.77k | orig = NULL; |
5913 | 4.77k | } |
5914 | 45.4k | } |
5915 | | |
5916 | 53.7k | done: |
5917 | 53.7k | if (value != NULL) xmlFree(value); |
5918 | 53.7k | if (URI != NULL) xmlFree(URI); |
5919 | 53.7k | if (literal != NULL) xmlFree(literal); |
5920 | 53.7k | if (orig != NULL) xmlFree(orig); |
5921 | 53.7k | } |
5922 | 54.1k | } |
5923 | | |
5924 | | /** |
5925 | | * xmlParseDefaultDecl: |
5926 | | * @ctxt: an XML parser context |
5927 | | * @value: Receive a possible fixed default value for the attribute |
5928 | | * |
5929 | | * DEPRECATED: Internal function, don't use. |
5930 | | * |
5931 | | * Parse an attribute default declaration |
5932 | | * |
5933 | | * [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue) |
5934 | | * |
5935 | | * [ VC: Required Attribute ] |
5936 | | * if the default declaration is the keyword #REQUIRED, then the |
5937 | | * attribute must be specified for all elements of the type in the |
5938 | | * attribute-list declaration. |
5939 | | * |
5940 | | * [ VC: Attribute Default Legal ] |
5941 | | * The declared default value must meet the lexical constraints of |
5942 | | * the declared attribute type c.f. xmlValidateAttributeDecl() |
5943 | | * |
5944 | | * [ VC: Fixed Attribute Default ] |
5945 | | * if an attribute has a default value declared with the #FIXED |
5946 | | * keyword, instances of that attribute must match the default value. |
5947 | | * |
5948 | | * [ WFC: No < in Attribute Values ] |
5949 | | * handled in xmlParseAttValue() |
5950 | | * |
5951 | | * returns: XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED |
5952 | | * or XML_ATTRIBUTE_FIXED. |
5953 | | */ |
5954 | | |
5955 | | int |
5956 | 49.1k | xmlParseDefaultDecl(xmlParserCtxtPtr ctxt, xmlChar **value) { |
5957 | 49.1k | int val; |
5958 | 49.1k | xmlChar *ret; |
5959 | | |
5960 | 49.1k | *value = NULL; |
5961 | 49.1k | if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) { |
5962 | 698 | SKIP(9); |
5963 | 698 | return(XML_ATTRIBUTE_REQUIRED); |
5964 | 698 | } |
5965 | 48.4k | if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) { |
5966 | 621 | SKIP(8); |
5967 | 621 | return(XML_ATTRIBUTE_IMPLIED); |
5968 | 621 | } |
5969 | 47.8k | val = XML_ATTRIBUTE_NONE; |
5970 | 47.8k | if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) { |
5971 | 545 | SKIP(6); |
5972 | 545 | val = XML_ATTRIBUTE_FIXED; |
5973 | 545 | if (SKIP_BLANKS_PE == 0) { |
5974 | 248 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5975 | 248 | "Space required after '#FIXED'\n"); |
5976 | 248 | } |
5977 | 545 | } |
5978 | 47.8k | ret = xmlParseAttValue(ctxt); |
5979 | 47.8k | if (ret == NULL) { |
5980 | 2.19k | xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo, |
5981 | 2.19k | "Attribute default value declaration error\n"); |
5982 | 2.19k | } else |
5983 | 45.6k | *value = ret; |
5984 | 47.8k | return(val); |
5985 | 48.4k | } |
5986 | | |
5987 | | /** |
5988 | | * xmlParseNotationType: |
5989 | | * @ctxt: an XML parser context |
5990 | | * |
5991 | | * DEPRECATED: Internal function, don't use. |
5992 | | * |
5993 | | * parse an Notation attribute type. |
5994 | | * |
5995 | | * Note: the leading 'NOTATION' S part has already being parsed... |
5996 | | * |
5997 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5998 | | * |
5999 | | * [ VC: Notation Attributes ] |
6000 | | * Values of this type must match one of the notation names included |
6001 | | * in the declaration; all notation names in the declaration must be declared. |
6002 | | * |
6003 | | * Returns: the notation attribute tree built while parsing |
6004 | | */ |
6005 | | |
6006 | | xmlEnumerationPtr |
6007 | 2.88k | xmlParseNotationType(xmlParserCtxtPtr ctxt) { |
6008 | 2.88k | const xmlChar *name; |
6009 | 2.88k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
6010 | | |
6011 | 2.88k | if (RAW != '(') { |
6012 | 76 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
6013 | 76 | return(NULL); |
6014 | 76 | } |
6015 | 3.27k | do { |
6016 | 3.27k | NEXT; |
6017 | 3.27k | SKIP_BLANKS_PE; |
6018 | 3.27k | name = xmlParseName(ctxt); |
6019 | 3.27k | if (name == NULL) { |
6020 | 53 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6021 | 53 | "Name expected in NOTATION declaration\n"); |
6022 | 53 | xmlFreeEnumeration(ret); |
6023 | 53 | return(NULL); |
6024 | 53 | } |
6025 | 3.22k | tmp = ret; |
6026 | 4.20k | while (tmp != NULL) { |
6027 | 1.34k | if (xmlStrEqual(name, tmp->name)) { |
6028 | 373 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
6029 | 373 | "standalone: attribute notation value token %s duplicated\n", |
6030 | 373 | name, NULL); |
6031 | 373 | if (!xmlDictOwns(ctxt->dict, name)) |
6032 | 0 | xmlFree((xmlChar *) name); |
6033 | 373 | break; |
6034 | 373 | } |
6035 | 976 | tmp = tmp->next; |
6036 | 976 | } |
6037 | 3.22k | if (tmp == NULL) { |
6038 | 2.85k | cur = xmlCreateEnumeration(name); |
6039 | 2.85k | if (cur == NULL) { |
6040 | 0 | xmlErrMemory(ctxt); |
6041 | 0 | xmlFreeEnumeration(ret); |
6042 | 0 | return(NULL); |
6043 | 0 | } |
6044 | 2.85k | if (last == NULL) ret = last = cur; |
6045 | 84 | else { |
6046 | 84 | last->next = cur; |
6047 | 84 | last = cur; |
6048 | 84 | } |
6049 | 2.85k | } |
6050 | 3.22k | SKIP_BLANKS_PE; |
6051 | 3.22k | } while (RAW == '|'); |
6052 | 2.75k | if (RAW != ')') { |
6053 | 38 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
6054 | 38 | xmlFreeEnumeration(ret); |
6055 | 38 | return(NULL); |
6056 | 38 | } |
6057 | 2.71k | NEXT; |
6058 | 2.71k | return(ret); |
6059 | 2.75k | } |
6060 | | |
6061 | | /** |
6062 | | * xmlParseEnumerationType: |
6063 | | * @ctxt: an XML parser context |
6064 | | * |
6065 | | * DEPRECATED: Internal function, don't use. |
6066 | | * |
6067 | | * parse an Enumeration attribute type. |
6068 | | * |
6069 | | * [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' |
6070 | | * |
6071 | | * [ VC: Enumeration ] |
6072 | | * Values of this type must match one of the Nmtoken tokens in |
6073 | | * the declaration |
6074 | | * |
6075 | | * Returns: the enumeration attribute tree built while parsing |
6076 | | */ |
6077 | | |
6078 | | xmlEnumerationPtr |
6079 | 12.6k | xmlParseEnumerationType(xmlParserCtxtPtr ctxt) { |
6080 | 12.6k | xmlChar *name; |
6081 | 12.6k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
6082 | | |
6083 | 12.6k | if (RAW != '(') { |
6084 | 507 | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL); |
6085 | 507 | return(NULL); |
6086 | 507 | } |
6087 | 14.5k | do { |
6088 | 14.5k | NEXT; |
6089 | 14.5k | SKIP_BLANKS_PE; |
6090 | 14.5k | name = xmlParseNmtoken(ctxt); |
6091 | 14.5k | if (name == NULL) { |
6092 | 89 | xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL); |
6093 | 89 | return(ret); |
6094 | 89 | } |
6095 | 14.4k | tmp = ret; |
6096 | 23.8k | while (tmp != NULL) { |
6097 | 9.80k | if (xmlStrEqual(name, tmp->name)) { |
6098 | 419 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
6099 | 419 | "standalone: attribute enumeration value token %s duplicated\n", |
6100 | 419 | name, NULL); |
6101 | 419 | if (!xmlDictOwns(ctxt->dict, name)) |
6102 | 419 | xmlFree(name); |
6103 | 419 | break; |
6104 | 419 | } |
6105 | 9.38k | tmp = tmp->next; |
6106 | 9.38k | } |
6107 | 14.4k | if (tmp == NULL) { |
6108 | 14.0k | cur = xmlCreateEnumeration(name); |
6109 | 14.0k | if (!xmlDictOwns(ctxt->dict, name)) |
6110 | 14.0k | xmlFree(name); |
6111 | 14.0k | if (cur == NULL) { |
6112 | 0 | xmlErrMemory(ctxt); |
6113 | 0 | xmlFreeEnumeration(ret); |
6114 | 0 | return(NULL); |
6115 | 0 | } |
6116 | 14.0k | if (last == NULL) ret = last = cur; |
6117 | 2.00k | else { |
6118 | 2.00k | last->next = cur; |
6119 | 2.00k | last = cur; |
6120 | 2.00k | } |
6121 | 14.0k | } |
6122 | 14.4k | SKIP_BLANKS_PE; |
6123 | 14.4k | } while (RAW == '|'); |
6124 | 12.0k | if (RAW != ')') { |
6125 | 338 | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL); |
6126 | 338 | return(ret); |
6127 | 338 | } |
6128 | 11.6k | NEXT; |
6129 | 11.6k | return(ret); |
6130 | 12.0k | } |
6131 | | |
6132 | | /** |
6133 | | * xmlParseEnumeratedType: |
6134 | | * @ctxt: an XML parser context |
6135 | | * @tree: the enumeration tree built while parsing |
6136 | | * |
6137 | | * DEPRECATED: Internal function, don't use. |
6138 | | * |
6139 | | * parse an Enumerated attribute type. |
6140 | | * |
6141 | | * [57] EnumeratedType ::= NotationType | Enumeration |
6142 | | * |
6143 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
6144 | | * |
6145 | | * |
6146 | | * Returns: XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION |
6147 | | */ |
6148 | | |
6149 | | int |
6150 | 15.5k | xmlParseEnumeratedType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
6151 | 15.5k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
6152 | 2.88k | SKIP(8); |
6153 | 2.88k | if (SKIP_BLANKS_PE == 0) { |
6154 | 3 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6155 | 3 | "Space required after 'NOTATION'\n"); |
6156 | 3 | return(0); |
6157 | 3 | } |
6158 | 2.88k | *tree = xmlParseNotationType(ctxt); |
6159 | 2.88k | if (*tree == NULL) return(0); |
6160 | 2.71k | return(XML_ATTRIBUTE_NOTATION); |
6161 | 2.88k | } |
6162 | 12.6k | *tree = xmlParseEnumerationType(ctxt); |
6163 | 12.6k | if (*tree == NULL) return(0); |
6164 | 12.0k | return(XML_ATTRIBUTE_ENUMERATION); |
6165 | 12.6k | } |
6166 | | |
6167 | | /** |
6168 | | * xmlParseAttributeType: |
6169 | | * @ctxt: an XML parser context |
6170 | | * @tree: the enumeration tree built while parsing |
6171 | | * |
6172 | | * DEPRECATED: Internal function, don't use. |
6173 | | * |
6174 | | * parse the Attribute list def for an element |
6175 | | * |
6176 | | * [54] AttType ::= StringType | TokenizedType | EnumeratedType |
6177 | | * |
6178 | | * [55] StringType ::= 'CDATA' |
6179 | | * |
6180 | | * [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' | |
6181 | | * 'ENTITIES' | 'NMTOKEN' | 'NMTOKENS' |
6182 | | * |
6183 | | * Validity constraints for attribute values syntax are checked in |
6184 | | * xmlValidateAttributeValue() |
6185 | | * |
6186 | | * [ VC: ID ] |
6187 | | * Values of type ID must match the Name production. A name must not |
6188 | | * appear more than once in an XML document as a value of this type; |
6189 | | * i.e., ID values must uniquely identify the elements which bear them. |
6190 | | * |
6191 | | * [ VC: One ID per Element Type ] |
6192 | | * No element type may have more than one ID attribute specified. |
6193 | | * |
6194 | | * [ VC: ID Attribute Default ] |
6195 | | * An ID attribute must have a declared default of #IMPLIED or #REQUIRED. |
6196 | | * |
6197 | | * [ VC: IDREF ] |
6198 | | * Values of type IDREF must match the Name production, and values |
6199 | | * of type IDREFS must match Names; each IDREF Name must match the value |
6200 | | * of an ID attribute on some element in the XML document; i.e. IDREF |
6201 | | * values must match the value of some ID attribute. |
6202 | | * |
6203 | | * [ VC: Entity Name ] |
6204 | | * Values of type ENTITY must match the Name production, values |
6205 | | * of type ENTITIES must match Names; each Entity Name must match the |
6206 | | * name of an unparsed entity declared in the DTD. |
6207 | | * |
6208 | | * [ VC: Name Token ] |
6209 | | * Values of type NMTOKEN must match the Nmtoken production; values |
6210 | | * of type NMTOKENS must match Nmtokens. |
6211 | | * |
6212 | | * Returns the attribute type |
6213 | | */ |
6214 | | int |
6215 | 50.3k | xmlParseAttributeType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
6216 | 50.3k | if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) { |
6217 | 5.81k | SKIP(5); |
6218 | 5.81k | return(XML_ATTRIBUTE_CDATA); |
6219 | 44.4k | } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) { |
6220 | 5.47k | SKIP(6); |
6221 | 5.47k | return(XML_ATTRIBUTE_IDREFS); |
6222 | 39.0k | } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) { |
6223 | 1.01k | SKIP(5); |
6224 | 1.01k | return(XML_ATTRIBUTE_IDREF); |
6225 | 38.0k | } else if ((RAW == 'I') && (NXT(1) == 'D')) { |
6226 | 20.1k | SKIP(2); |
6227 | 20.1k | return(XML_ATTRIBUTE_ID); |
6228 | 20.1k | } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
6229 | 527 | SKIP(6); |
6230 | 527 | return(XML_ATTRIBUTE_ENTITY); |
6231 | 17.3k | } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) { |
6232 | 579 | SKIP(8); |
6233 | 579 | return(XML_ATTRIBUTE_ENTITIES); |
6234 | 16.7k | } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) { |
6235 | 157 | SKIP(8); |
6236 | 157 | return(XML_ATTRIBUTE_NMTOKENS); |
6237 | 16.5k | } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) { |
6238 | 1.06k | SKIP(7); |
6239 | 1.06k | return(XML_ATTRIBUTE_NMTOKEN); |
6240 | 1.06k | } |
6241 | 15.5k | return(xmlParseEnumeratedType(ctxt, tree)); |
6242 | 50.3k | } |
6243 | | |
6244 | | /** |
6245 | | * xmlParseAttributeListDecl: |
6246 | | * @ctxt: an XML parser context |
6247 | | * |
6248 | | * DEPRECATED: Internal function, don't use. |
6249 | | * |
6250 | | * Parse an attribute list declaration for an element. Always consumes '<!'. |
6251 | | * |
6252 | | * [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>' |
6253 | | * |
6254 | | * [53] AttDef ::= S Name S AttType S DefaultDecl |
6255 | | * |
6256 | | */ |
6257 | | void |
6258 | 11.9k | xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt) { |
6259 | 11.9k | const xmlChar *elemName; |
6260 | 11.9k | const xmlChar *attrName; |
6261 | 11.9k | xmlEnumerationPtr tree; |
6262 | | |
6263 | 11.9k | if ((CUR != '<') || (NXT(1) != '!')) |
6264 | 0 | return; |
6265 | 11.9k | SKIP(2); |
6266 | | |
6267 | 11.9k | if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) { |
6268 | 11.9k | int inputid = ctxt->input->id; |
6269 | | |
6270 | 11.9k | SKIP(7); |
6271 | 11.9k | if (SKIP_BLANKS_PE == 0) { |
6272 | 901 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6273 | 901 | "Space required after '<!ATTLIST'\n"); |
6274 | 901 | } |
6275 | 11.9k | elemName = xmlParseName(ctxt); |
6276 | 11.9k | if (elemName == NULL) { |
6277 | 165 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6278 | 165 | "ATTLIST: no name for Element\n"); |
6279 | 165 | return; |
6280 | 165 | } |
6281 | 11.7k | SKIP_BLANKS_PE; |
6282 | 11.7k | GROW; |
6283 | 59.4k | while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) { |
6284 | 52.0k | int type; |
6285 | 52.0k | int def; |
6286 | 52.0k | xmlChar *defaultValue = NULL; |
6287 | | |
6288 | 52.0k | GROW; |
6289 | 52.0k | tree = NULL; |
6290 | 52.0k | attrName = xmlParseName(ctxt); |
6291 | 52.0k | if (attrName == NULL) { |
6292 | 1.19k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6293 | 1.19k | "ATTLIST: no name for Attribute\n"); |
6294 | 1.19k | break; |
6295 | 1.19k | } |
6296 | 50.8k | GROW; |
6297 | 50.8k | if (SKIP_BLANKS_PE == 0) { |
6298 | 558 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6299 | 558 | "Space required after the attribute name\n"); |
6300 | 558 | break; |
6301 | 558 | } |
6302 | | |
6303 | 50.3k | type = xmlParseAttributeType(ctxt, &tree); |
6304 | 50.3k | if (type <= 0) { |
6305 | 747 | break; |
6306 | 747 | } |
6307 | | |
6308 | 49.5k | GROW; |
6309 | 49.5k | if (SKIP_BLANKS_PE == 0) { |
6310 | 431 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6311 | 431 | "Space required after the attribute type\n"); |
6312 | 431 | if (tree != NULL) |
6313 | 362 | xmlFreeEnumeration(tree); |
6314 | 431 | break; |
6315 | 431 | } |
6316 | | |
6317 | 49.1k | def = xmlParseDefaultDecl(ctxt, &defaultValue); |
6318 | 49.1k | if (def <= 0) { |
6319 | 0 | if (defaultValue != NULL) |
6320 | 0 | xmlFree(defaultValue); |
6321 | 0 | if (tree != NULL) |
6322 | 0 | xmlFreeEnumeration(tree); |
6323 | 0 | break; |
6324 | 0 | } |
6325 | 49.1k | if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL)) |
6326 | 39.8k | xmlAttrNormalizeSpace(defaultValue, defaultValue); |
6327 | | |
6328 | 49.1k | GROW; |
6329 | 49.1k | if (RAW != '>') { |
6330 | 41.9k | if (SKIP_BLANKS_PE == 0) { |
6331 | 1.49k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6332 | 1.49k | "Space required after the attribute default value\n"); |
6333 | 1.49k | if (defaultValue != NULL) |
6334 | 79 | xmlFree(defaultValue); |
6335 | 1.49k | if (tree != NULL) |
6336 | 195 | xmlFreeEnumeration(tree); |
6337 | 1.49k | break; |
6338 | 1.49k | } |
6339 | 41.9k | } |
6340 | 47.6k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6341 | 47.6k | (ctxt->sax->attributeDecl != NULL)) |
6342 | 38.4k | ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName, |
6343 | 38.4k | type, def, defaultValue, tree); |
6344 | 9.14k | else if (tree != NULL) |
6345 | 1.96k | xmlFreeEnumeration(tree); |
6346 | | |
6347 | 47.6k | if ((ctxt->sax2) && (defaultValue != NULL) && |
6348 | 47.6k | (def != XML_ATTRIBUTE_IMPLIED) && |
6349 | 47.6k | (def != XML_ATTRIBUTE_REQUIRED)) { |
6350 | 45.5k | xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue); |
6351 | 45.5k | } |
6352 | 47.6k | if (ctxt->sax2) { |
6353 | 47.6k | xmlAddSpecialAttr(ctxt, elemName, attrName, type); |
6354 | 47.6k | } |
6355 | 47.6k | if (defaultValue != NULL) |
6356 | 45.5k | xmlFree(defaultValue); |
6357 | 47.6k | GROW; |
6358 | 47.6k | } |
6359 | 11.7k | if (RAW == '>') { |
6360 | 7.41k | if (inputid != ctxt->input->id) { |
6361 | 20 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6362 | 20 | "Attribute list declaration doesn't start and" |
6363 | 20 | " stop in the same entity\n"); |
6364 | 20 | } |
6365 | 7.41k | NEXT; |
6366 | 7.41k | } |
6367 | 11.7k | } |
6368 | 11.9k | } |
6369 | | |
6370 | | /** |
6371 | | * xmlParseElementMixedContentDecl: |
6372 | | * @ctxt: an XML parser context |
6373 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6374 | | * |
6375 | | * DEPRECATED: Internal function, don't use. |
6376 | | * |
6377 | | * parse the declaration for a Mixed Element content |
6378 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6379 | | * |
6380 | | * [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' | |
6381 | | * '(' S? '#PCDATA' S? ')' |
6382 | | * |
6383 | | * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49]) |
6384 | | * |
6385 | | * [ VC: No Duplicate Types ] |
6386 | | * The same name must not appear more than once in a single |
6387 | | * mixed-content declaration. |
6388 | | * |
6389 | | * returns: the list of the xmlElementContentPtr describing the element choices |
6390 | | */ |
6391 | | xmlElementContentPtr |
6392 | 1.37k | xmlParseElementMixedContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6393 | 1.37k | xmlElementContentPtr ret = NULL, cur = NULL, n; |
6394 | 1.37k | const xmlChar *elem = NULL; |
6395 | | |
6396 | 1.37k | GROW; |
6397 | 1.37k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6398 | 1.37k | SKIP(7); |
6399 | 1.37k | SKIP_BLANKS_PE; |
6400 | 1.37k | if (RAW == ')') { |
6401 | 410 | if (ctxt->input->id != inputchk) { |
6402 | 68 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6403 | 68 | "Element content declaration doesn't start and" |
6404 | 68 | " stop in the same entity\n"); |
6405 | 68 | } |
6406 | 410 | NEXT; |
6407 | 410 | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6408 | 410 | if (ret == NULL) |
6409 | 0 | goto mem_error; |
6410 | 410 | if (RAW == '*') { |
6411 | 156 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6412 | 156 | NEXT; |
6413 | 156 | } |
6414 | 410 | return(ret); |
6415 | 410 | } |
6416 | 969 | if ((RAW == '(') || (RAW == '|')) { |
6417 | 787 | ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6418 | 787 | if (ret == NULL) |
6419 | 0 | goto mem_error; |
6420 | 787 | } |
6421 | 2.38k | while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) { |
6422 | 1.48k | NEXT; |
6423 | 1.48k | n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6424 | 1.48k | if (n == NULL) |
6425 | 0 | goto mem_error; |
6426 | 1.48k | if (elem == NULL) { |
6427 | 785 | n->c1 = cur; |
6428 | 785 | if (cur != NULL) |
6429 | 785 | cur->parent = n; |
6430 | 785 | ret = cur = n; |
6431 | 785 | } else { |
6432 | 702 | cur->c2 = n; |
6433 | 702 | n->parent = cur; |
6434 | 702 | n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6435 | 702 | if (n->c1 == NULL) |
6436 | 0 | goto mem_error; |
6437 | 702 | n->c1->parent = n; |
6438 | 702 | cur = n; |
6439 | 702 | } |
6440 | 1.48k | SKIP_BLANKS_PE; |
6441 | 1.48k | elem = xmlParseName(ctxt); |
6442 | 1.48k | if (elem == NULL) { |
6443 | 67 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6444 | 67 | "xmlParseElementMixedContentDecl : Name expected\n"); |
6445 | 67 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6446 | 67 | return(NULL); |
6447 | 67 | } |
6448 | 1.42k | SKIP_BLANKS_PE; |
6449 | 1.42k | GROW; |
6450 | 1.42k | } |
6451 | 902 | if ((RAW == ')') && (NXT(1) == '*')) { |
6452 | 704 | if (elem != NULL) { |
6453 | 704 | cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem, |
6454 | 704 | XML_ELEMENT_CONTENT_ELEMENT); |
6455 | 704 | if (cur->c2 == NULL) |
6456 | 0 | goto mem_error; |
6457 | 704 | cur->c2->parent = cur; |
6458 | 704 | } |
6459 | 704 | if (ret != NULL) |
6460 | 704 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6461 | 704 | if (ctxt->input->id != inputchk) { |
6462 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6463 | 0 | "Element content declaration doesn't start and" |
6464 | 0 | " stop in the same entity\n"); |
6465 | 0 | } |
6466 | 704 | SKIP(2); |
6467 | 704 | } else { |
6468 | 198 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6469 | 198 | xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL); |
6470 | 198 | return(NULL); |
6471 | 198 | } |
6472 | | |
6473 | 902 | } else { |
6474 | 0 | xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL); |
6475 | 0 | } |
6476 | 704 | return(ret); |
6477 | | |
6478 | 0 | mem_error: |
6479 | 0 | xmlErrMemory(ctxt); |
6480 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6481 | 0 | return(NULL); |
6482 | 1.37k | } |
6483 | | |
6484 | | /** |
6485 | | * xmlParseElementChildrenContentDeclPriv: |
6486 | | * @ctxt: an XML parser context |
6487 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6488 | | * @depth: the level of recursion |
6489 | | * |
6490 | | * parse the declaration for a Mixed Element content |
6491 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6492 | | * |
6493 | | * |
6494 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6495 | | * |
6496 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6497 | | * |
6498 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6499 | | * |
6500 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6501 | | * |
6502 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6503 | | * TODO Parameter-entity replacement text must be properly nested |
6504 | | * with parenthesized groups. That is to say, if either of the |
6505 | | * opening or closing parentheses in a choice, seq, or Mixed |
6506 | | * construct is contained in the replacement text for a parameter |
6507 | | * entity, both must be contained in the same replacement text. For |
6508 | | * interoperability, if a parameter-entity reference appears in a |
6509 | | * choice, seq, or Mixed construct, its replacement text should not |
6510 | | * be empty, and neither the first nor last non-blank character of |
6511 | | * the replacement text should be a connector (| or ,). |
6512 | | * |
6513 | | * Returns the tree of xmlElementContentPtr describing the element |
6514 | | * hierarchy. |
6515 | | */ |
6516 | | static xmlElementContentPtr |
6517 | | xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int inputchk, |
6518 | 19.0k | int depth) { |
6519 | 19.0k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
6520 | 19.0k | xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL; |
6521 | 19.0k | const xmlChar *elem; |
6522 | 19.0k | xmlChar type = 0; |
6523 | | |
6524 | 19.0k | if (depth > maxDepth) { |
6525 | 2 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
6526 | 2 | "xmlParseElementChildrenContentDecl : depth %d too deep, " |
6527 | 2 | "use XML_PARSE_HUGE\n", depth); |
6528 | 2 | return(NULL); |
6529 | 2 | } |
6530 | 19.0k | SKIP_BLANKS_PE; |
6531 | 19.0k | GROW; |
6532 | 19.0k | if (RAW == '(') { |
6533 | 8.07k | int inputid = ctxt->input->id; |
6534 | | |
6535 | | /* Recurse on first child */ |
6536 | 8.07k | NEXT; |
6537 | 8.07k | SKIP_BLANKS_PE; |
6538 | 8.07k | cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6539 | 8.07k | depth + 1); |
6540 | 8.07k | if (cur == NULL) |
6541 | 5.60k | return(NULL); |
6542 | 2.46k | SKIP_BLANKS_PE; |
6543 | 2.46k | GROW; |
6544 | 10.9k | } else { |
6545 | 10.9k | elem = xmlParseName(ctxt); |
6546 | 10.9k | if (elem == NULL) { |
6547 | 378 | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6548 | 378 | return(NULL); |
6549 | 378 | } |
6550 | 10.5k | cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6551 | 10.5k | if (cur == NULL) { |
6552 | 0 | xmlErrMemory(ctxt); |
6553 | 0 | return(NULL); |
6554 | 0 | } |
6555 | 10.5k | GROW; |
6556 | 10.5k | if (RAW == '?') { |
6557 | 1.19k | cur->ocur = XML_ELEMENT_CONTENT_OPT; |
6558 | 1.19k | NEXT; |
6559 | 9.35k | } else if (RAW == '*') { |
6560 | 399 | cur->ocur = XML_ELEMENT_CONTENT_MULT; |
6561 | 399 | NEXT; |
6562 | 8.95k | } else if (RAW == '+') { |
6563 | 438 | cur->ocur = XML_ELEMENT_CONTENT_PLUS; |
6564 | 438 | NEXT; |
6565 | 8.51k | } else { |
6566 | 8.51k | cur->ocur = XML_ELEMENT_CONTENT_ONCE; |
6567 | 8.51k | } |
6568 | 10.5k | GROW; |
6569 | 10.5k | } |
6570 | 13.0k | SKIP_BLANKS_PE; |
6571 | 25.4k | while ((RAW != ')') && (PARSER_STOPPED(ctxt) == 0)) { |
6572 | | /* |
6573 | | * Each loop we parse one separator and one element. |
6574 | | */ |
6575 | 16.0k | if (RAW == ',') { |
6576 | 1.88k | if (type == 0) type = CUR; |
6577 | | |
6578 | | /* |
6579 | | * Detect "Name | Name , Name" error |
6580 | | */ |
6581 | 1.28k | else if (type != CUR) { |
6582 | 3 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6583 | 3 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6584 | 3 | type); |
6585 | 3 | if ((last != NULL) && (last != ret)) |
6586 | 3 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6587 | 3 | if (ret != NULL) |
6588 | 3 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6589 | 3 | return(NULL); |
6590 | 3 | } |
6591 | 1.88k | NEXT; |
6592 | | |
6593 | 1.88k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ); |
6594 | 1.88k | if (op == NULL) { |
6595 | 0 | xmlErrMemory(ctxt); |
6596 | 0 | if ((last != NULL) && (last != ret)) |
6597 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6598 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6599 | 0 | return(NULL); |
6600 | 0 | } |
6601 | 1.88k | if (last == NULL) { |
6602 | 598 | op->c1 = ret; |
6603 | 598 | if (ret != NULL) |
6604 | 598 | ret->parent = op; |
6605 | 598 | ret = cur = op; |
6606 | 1.28k | } else { |
6607 | 1.28k | cur->c2 = op; |
6608 | 1.28k | if (op != NULL) |
6609 | 1.28k | op->parent = cur; |
6610 | 1.28k | op->c1 = last; |
6611 | 1.28k | if (last != NULL) |
6612 | 1.28k | last->parent = op; |
6613 | 1.28k | cur =op; |
6614 | 1.28k | last = NULL; |
6615 | 1.28k | } |
6616 | 14.1k | } else if (RAW == '|') { |
6617 | 12.7k | if (type == 0) type = CUR; |
6618 | | |
6619 | | /* |
6620 | | * Detect "Name , Name | Name" error |
6621 | | */ |
6622 | 5.08k | else if (type != CUR) { |
6623 | 3 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6624 | 3 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6625 | 3 | type); |
6626 | 3 | if ((last != NULL) && (last != ret)) |
6627 | 3 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6628 | 3 | if (ret != NULL) |
6629 | 3 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6630 | 3 | return(NULL); |
6631 | 3 | } |
6632 | 12.7k | NEXT; |
6633 | | |
6634 | 12.7k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6635 | 12.7k | if (op == NULL) { |
6636 | 0 | xmlErrMemory(ctxt); |
6637 | 0 | if ((last != NULL) && (last != ret)) |
6638 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6639 | 0 | if (ret != NULL) |
6640 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6641 | 0 | return(NULL); |
6642 | 0 | } |
6643 | 12.7k | if (last == NULL) { |
6644 | 7.65k | op->c1 = ret; |
6645 | 7.65k | if (ret != NULL) |
6646 | 7.65k | ret->parent = op; |
6647 | 7.65k | ret = cur = op; |
6648 | 7.65k | } else { |
6649 | 5.07k | cur->c2 = op; |
6650 | 5.07k | if (op != NULL) |
6651 | 5.07k | op->parent = cur; |
6652 | 5.07k | op->c1 = last; |
6653 | 5.07k | if (last != NULL) |
6654 | 5.07k | last->parent = op; |
6655 | 5.07k | cur =op; |
6656 | 5.07k | last = NULL; |
6657 | 5.07k | } |
6658 | 12.7k | } else { |
6659 | 1.44k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL); |
6660 | 1.44k | if ((last != NULL) && (last != ret)) |
6661 | 818 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6662 | 1.44k | if (ret != NULL) |
6663 | 1.44k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6664 | 1.44k | return(NULL); |
6665 | 1.44k | } |
6666 | 14.6k | GROW; |
6667 | 14.6k | SKIP_BLANKS_PE; |
6668 | 14.6k | GROW; |
6669 | 14.6k | if (RAW == '(') { |
6670 | 4.21k | int inputid = ctxt->input->id; |
6671 | | /* Recurse on second child */ |
6672 | 4.21k | NEXT; |
6673 | 4.21k | SKIP_BLANKS_PE; |
6674 | 4.21k | last = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6675 | 4.21k | depth + 1); |
6676 | 4.21k | if (last == NULL) { |
6677 | 2.14k | if (ret != NULL) |
6678 | 2.14k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6679 | 2.14k | return(NULL); |
6680 | 2.14k | } |
6681 | 2.06k | SKIP_BLANKS_PE; |
6682 | 10.4k | } else { |
6683 | 10.4k | elem = xmlParseName(ctxt); |
6684 | 10.4k | if (elem == NULL) { |
6685 | 68 | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6686 | 68 | if (ret != NULL) |
6687 | 68 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6688 | 68 | return(NULL); |
6689 | 68 | } |
6690 | 10.3k | last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6691 | 10.3k | if (last == NULL) { |
6692 | 0 | xmlErrMemory(ctxt); |
6693 | 0 | if (ret != NULL) |
6694 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6695 | 0 | return(NULL); |
6696 | 0 | } |
6697 | 10.3k | if (RAW == '?') { |
6698 | 434 | last->ocur = XML_ELEMENT_CONTENT_OPT; |
6699 | 434 | NEXT; |
6700 | 9.89k | } else if (RAW == '*') { |
6701 | 301 | last->ocur = XML_ELEMENT_CONTENT_MULT; |
6702 | 301 | NEXT; |
6703 | 9.59k | } else if (RAW == '+') { |
6704 | 371 | last->ocur = XML_ELEMENT_CONTENT_PLUS; |
6705 | 371 | NEXT; |
6706 | 9.22k | } else { |
6707 | 9.22k | last->ocur = XML_ELEMENT_CONTENT_ONCE; |
6708 | 9.22k | } |
6709 | 10.3k | } |
6710 | 12.3k | SKIP_BLANKS_PE; |
6711 | 12.3k | GROW; |
6712 | 12.3k | } |
6713 | 9.34k | if ((cur != NULL) && (last != NULL)) { |
6714 | 5.21k | cur->c2 = last; |
6715 | 5.21k | if (last != NULL) |
6716 | 5.21k | last->parent = cur; |
6717 | 5.21k | } |
6718 | 9.34k | if (ctxt->input->id != inputchk) { |
6719 | 23 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6720 | 23 | "Element content declaration doesn't start and stop in" |
6721 | 23 | " the same entity\n"); |
6722 | 23 | } |
6723 | 9.34k | NEXT; |
6724 | 9.34k | if (RAW == '?') { |
6725 | 1.99k | if (ret != NULL) { |
6726 | 1.99k | if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) || |
6727 | 1.99k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6728 | 659 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6729 | 1.33k | else |
6730 | 1.33k | ret->ocur = XML_ELEMENT_CONTENT_OPT; |
6731 | 1.99k | } |
6732 | 1.99k | NEXT; |
6733 | 7.34k | } else if (RAW == '*') { |
6734 | 3.91k | if (ret != NULL) { |
6735 | 3.91k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6736 | 3.91k | cur = ret; |
6737 | | /* |
6738 | | * Some normalization: |
6739 | | * (a | b* | c?)* == (a | b | c)* |
6740 | | */ |
6741 | 15.6k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6742 | 11.7k | if ((cur->c1 != NULL) && |
6743 | 11.7k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6744 | 11.7k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) |
6745 | 647 | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6746 | 11.7k | if ((cur->c2 != NULL) && |
6747 | 11.7k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6748 | 11.7k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) |
6749 | 908 | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6750 | 11.7k | cur = cur->c2; |
6751 | 11.7k | } |
6752 | 3.91k | } |
6753 | 3.91k | NEXT; |
6754 | 3.91k | } else if (RAW == '+') { |
6755 | 1.44k | if (ret != NULL) { |
6756 | 1.44k | int found = 0; |
6757 | | |
6758 | 1.44k | if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) || |
6759 | 1.44k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6760 | 371 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6761 | 1.07k | else |
6762 | 1.07k | ret->ocur = XML_ELEMENT_CONTENT_PLUS; |
6763 | | /* |
6764 | | * Some normalization: |
6765 | | * (a | b*)+ == (a | b)* |
6766 | | * (a | b?)+ == (a | b)* |
6767 | | */ |
6768 | 6.48k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6769 | 5.04k | if ((cur->c1 != NULL) && |
6770 | 5.04k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6771 | 5.04k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6772 | 458 | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6773 | 458 | found = 1; |
6774 | 458 | } |
6775 | 5.04k | if ((cur->c2 != NULL) && |
6776 | 5.04k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6777 | 5.04k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6778 | 768 | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6779 | 768 | found = 1; |
6780 | 768 | } |
6781 | 5.04k | cur = cur->c2; |
6782 | 5.04k | } |
6783 | 1.44k | if (found) |
6784 | 740 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6785 | 1.44k | } |
6786 | 1.44k | NEXT; |
6787 | 1.44k | } |
6788 | 9.34k | return(ret); |
6789 | 13.0k | } |
6790 | | |
6791 | | /** |
6792 | | * xmlParseElementChildrenContentDecl: |
6793 | | * @ctxt: an XML parser context |
6794 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6795 | | * |
6796 | | * DEPRECATED: Internal function, don't use. |
6797 | | * |
6798 | | * parse the declaration for a Mixed Element content |
6799 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6800 | | * |
6801 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6802 | | * |
6803 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6804 | | * |
6805 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6806 | | * |
6807 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6808 | | * |
6809 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6810 | | * TODO Parameter-entity replacement text must be properly nested |
6811 | | * with parenthesized groups. That is to say, if either of the |
6812 | | * opening or closing parentheses in a choice, seq, or Mixed |
6813 | | * construct is contained in the replacement text for a parameter |
6814 | | * entity, both must be contained in the same replacement text. For |
6815 | | * interoperability, if a parameter-entity reference appears in a |
6816 | | * choice, seq, or Mixed construct, its replacement text should not |
6817 | | * be empty, and neither the first nor last non-blank character of |
6818 | | * the replacement text should be a connector (| or ,). |
6819 | | * |
6820 | | * Returns the tree of xmlElementContentPtr describing the element |
6821 | | * hierarchy. |
6822 | | */ |
6823 | | xmlElementContentPtr |
6824 | 0 | xmlParseElementChildrenContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6825 | | /* stub left for API/ABI compat */ |
6826 | 0 | return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1)); |
6827 | 0 | } |
6828 | | |
6829 | | /** |
6830 | | * xmlParseElementContentDecl: |
6831 | | * @ctxt: an XML parser context |
6832 | | * @name: the name of the element being defined. |
6833 | | * @result: the Element Content pointer will be stored here if any |
6834 | | * |
6835 | | * DEPRECATED: Internal function, don't use. |
6836 | | * |
6837 | | * parse the declaration for an Element content either Mixed or Children, |
6838 | | * the cases EMPTY and ANY are handled directly in xmlParseElementDecl |
6839 | | * |
6840 | | * [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children |
6841 | | * |
6842 | | * returns: the type of element content XML_ELEMENT_TYPE_xxx |
6843 | | */ |
6844 | | |
6845 | | int |
6846 | | xmlParseElementContentDecl(xmlParserCtxtPtr ctxt, const xmlChar *name, |
6847 | 8.10k | xmlElementContentPtr *result) { |
6848 | | |
6849 | 8.10k | xmlElementContentPtr tree = NULL; |
6850 | 8.10k | int inputid = ctxt->input->id; |
6851 | 8.10k | int res; |
6852 | | |
6853 | 8.10k | *result = NULL; |
6854 | | |
6855 | 8.10k | if (RAW != '(') { |
6856 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6857 | 0 | "xmlParseElementContentDecl : %s '(' expected\n", name); |
6858 | 0 | return(-1); |
6859 | 0 | } |
6860 | 8.10k | NEXT; |
6861 | 8.10k | GROW; |
6862 | 8.10k | SKIP_BLANKS_PE; |
6863 | 8.10k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6864 | 1.37k | tree = xmlParseElementMixedContentDecl(ctxt, inputid); |
6865 | 1.37k | res = XML_ELEMENT_TYPE_MIXED; |
6866 | 6.72k | } else { |
6867 | 6.72k | tree = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, 1); |
6868 | 6.72k | res = XML_ELEMENT_TYPE_ELEMENT; |
6869 | 6.72k | } |
6870 | 8.10k | SKIP_BLANKS_PE; |
6871 | 8.10k | *result = tree; |
6872 | 8.10k | return(res); |
6873 | 8.10k | } |
6874 | | |
6875 | | /** |
6876 | | * xmlParseElementDecl: |
6877 | | * @ctxt: an XML parser context |
6878 | | * |
6879 | | * DEPRECATED: Internal function, don't use. |
6880 | | * |
6881 | | * Parse an element declaration. Always consumes '<!'. |
6882 | | * |
6883 | | * [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>' |
6884 | | * |
6885 | | * [ VC: Unique Element Type Declaration ] |
6886 | | * No element type may be declared more than once |
6887 | | * |
6888 | | * Returns the type of the element, or -1 in case of error |
6889 | | */ |
6890 | | int |
6891 | 10.0k | xmlParseElementDecl(xmlParserCtxtPtr ctxt) { |
6892 | 10.0k | const xmlChar *name; |
6893 | 10.0k | int ret = -1; |
6894 | 10.0k | xmlElementContentPtr content = NULL; |
6895 | | |
6896 | 10.0k | if ((CUR != '<') || (NXT(1) != '!')) |
6897 | 0 | return(ret); |
6898 | 10.0k | SKIP(2); |
6899 | | |
6900 | | /* GROW; done in the caller */ |
6901 | 10.0k | if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) { |
6902 | 9.99k | int inputid = ctxt->input->id; |
6903 | | |
6904 | 9.99k | SKIP(7); |
6905 | 9.99k | if (SKIP_BLANKS_PE == 0) { |
6906 | 550 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6907 | 550 | "Space required after 'ELEMENT'\n"); |
6908 | 550 | return(-1); |
6909 | 550 | } |
6910 | 9.44k | name = xmlParseName(ctxt); |
6911 | 9.44k | if (name == NULL) { |
6912 | 510 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6913 | 510 | "xmlParseElementDecl: no name for Element\n"); |
6914 | 510 | return(-1); |
6915 | 510 | } |
6916 | 8.93k | if (SKIP_BLANKS_PE == 0) { |
6917 | 1.23k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6918 | 1.23k | "Space required after the element name\n"); |
6919 | 1.23k | } |
6920 | 8.93k | if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) { |
6921 | 457 | SKIP(5); |
6922 | | /* |
6923 | | * Element must always be empty. |
6924 | | */ |
6925 | 457 | ret = XML_ELEMENT_TYPE_EMPTY; |
6926 | 8.47k | } else if ((RAW == 'A') && (NXT(1) == 'N') && |
6927 | 8.47k | (NXT(2) == 'Y')) { |
6928 | 226 | SKIP(3); |
6929 | | /* |
6930 | | * Element is a generic container. |
6931 | | */ |
6932 | 226 | ret = XML_ELEMENT_TYPE_ANY; |
6933 | 8.25k | } else if (RAW == '(') { |
6934 | 8.10k | ret = xmlParseElementContentDecl(ctxt, name, &content); |
6935 | 8.10k | } else { |
6936 | | /* |
6937 | | * [ WFC: PEs in Internal Subset ] error handling. |
6938 | | */ |
6939 | 151 | xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6940 | 151 | "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n"); |
6941 | 151 | return(-1); |
6942 | 151 | } |
6943 | | |
6944 | 8.78k | SKIP_BLANKS_PE; |
6945 | | |
6946 | 8.78k | if (RAW != '>') { |
6947 | 2.20k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
6948 | 2.20k | if (content != NULL) { |
6949 | 226 | xmlFreeDocElementContent(ctxt->myDoc, content); |
6950 | 226 | } |
6951 | 6.58k | } else { |
6952 | 6.58k | if (inputid != ctxt->input->id) { |
6953 | 64 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6954 | 64 | "Element declaration doesn't start and stop in" |
6955 | 64 | " the same entity\n"); |
6956 | 64 | } |
6957 | | |
6958 | 6.58k | NEXT; |
6959 | 6.58k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6960 | 6.58k | (ctxt->sax->elementDecl != NULL)) { |
6961 | 2.12k | if (content != NULL) |
6962 | 1.81k | content->parent = NULL; |
6963 | 2.12k | ctxt->sax->elementDecl(ctxt->userData, name, ret, |
6964 | 2.12k | content); |
6965 | 2.12k | if ((content != NULL) && (content->parent == NULL)) { |
6966 | | /* |
6967 | | * this is a trick: if xmlAddElementDecl is called, |
6968 | | * instead of copying the full tree it is plugged directly |
6969 | | * if called from the parser. Avoid duplicating the |
6970 | | * interfaces or change the API/ABI |
6971 | | */ |
6972 | 1.51k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6973 | 1.51k | } |
6974 | 4.45k | } else if (content != NULL) { |
6975 | 3.89k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6976 | 3.89k | } |
6977 | 6.58k | } |
6978 | 8.78k | } |
6979 | 8.81k | return(ret); |
6980 | 10.0k | } |
6981 | | |
6982 | | /** |
6983 | | * xmlParseConditionalSections |
6984 | | * @ctxt: an XML parser context |
6985 | | * |
6986 | | * Parse a conditional section. Always consumes '<!['. |
6987 | | * |
6988 | | * [61] conditionalSect ::= includeSect | ignoreSect |
6989 | | * [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>' |
6990 | | * [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>' |
6991 | | * [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' Ignore)* |
6992 | | * [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*) |
6993 | | */ |
6994 | | |
6995 | | static void |
6996 | 0 | xmlParseConditionalSections(xmlParserCtxtPtr ctxt) { |
6997 | 0 | int *inputIds = NULL; |
6998 | 0 | size_t inputIdsSize = 0; |
6999 | 0 | size_t depth = 0; |
7000 | |
|
7001 | 0 | while (PARSER_STOPPED(ctxt) == 0) { |
7002 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7003 | 0 | int id = ctxt->input->id; |
7004 | |
|
7005 | 0 | SKIP(3); |
7006 | 0 | SKIP_BLANKS_PE; |
7007 | |
|
7008 | 0 | if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) { |
7009 | 0 | SKIP(7); |
7010 | 0 | SKIP_BLANKS_PE; |
7011 | 0 | if (RAW != '[') { |
7012 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
7013 | 0 | xmlHaltParser(ctxt); |
7014 | 0 | goto error; |
7015 | 0 | } |
7016 | 0 | if (ctxt->input->id != id) { |
7017 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
7018 | 0 | "All markup of the conditional section is" |
7019 | 0 | " not in the same entity\n"); |
7020 | 0 | } |
7021 | 0 | NEXT; |
7022 | |
|
7023 | 0 | if (inputIdsSize <= depth) { |
7024 | 0 | int *tmp; |
7025 | |
|
7026 | 0 | inputIdsSize = (inputIdsSize == 0 ? 4 : inputIdsSize * 2); |
7027 | 0 | tmp = (int *) xmlRealloc(inputIds, |
7028 | 0 | inputIdsSize * sizeof(int)); |
7029 | 0 | if (tmp == NULL) { |
7030 | 0 | xmlErrMemory(ctxt); |
7031 | 0 | goto error; |
7032 | 0 | } |
7033 | 0 | inputIds = tmp; |
7034 | 0 | } |
7035 | 0 | inputIds[depth] = id; |
7036 | 0 | depth++; |
7037 | 0 | } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) { |
7038 | 0 | size_t ignoreDepth = 0; |
7039 | |
|
7040 | 0 | SKIP(6); |
7041 | 0 | SKIP_BLANKS_PE; |
7042 | 0 | if (RAW != '[') { |
7043 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
7044 | 0 | xmlHaltParser(ctxt); |
7045 | 0 | goto error; |
7046 | 0 | } |
7047 | 0 | if (ctxt->input->id != id) { |
7048 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
7049 | 0 | "All markup of the conditional section is" |
7050 | 0 | " not in the same entity\n"); |
7051 | 0 | } |
7052 | 0 | NEXT; |
7053 | |
|
7054 | 0 | while (PARSER_STOPPED(ctxt) == 0) { |
7055 | 0 | if (RAW == 0) { |
7056 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL); |
7057 | 0 | goto error; |
7058 | 0 | } |
7059 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7060 | 0 | SKIP(3); |
7061 | 0 | ignoreDepth++; |
7062 | | /* Check for integer overflow */ |
7063 | 0 | if (ignoreDepth == 0) { |
7064 | 0 | xmlErrMemory(ctxt); |
7065 | 0 | goto error; |
7066 | 0 | } |
7067 | 0 | } else if ((RAW == ']') && (NXT(1) == ']') && |
7068 | 0 | (NXT(2) == '>')) { |
7069 | 0 | SKIP(3); |
7070 | 0 | if (ignoreDepth == 0) |
7071 | 0 | break; |
7072 | 0 | ignoreDepth--; |
7073 | 0 | } else { |
7074 | 0 | NEXT; |
7075 | 0 | } |
7076 | 0 | } |
7077 | | |
7078 | 0 | if (ctxt->input->id != id) { |
7079 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
7080 | 0 | "All markup of the conditional section is" |
7081 | 0 | " not in the same entity\n"); |
7082 | 0 | } |
7083 | 0 | } else { |
7084 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL); |
7085 | 0 | xmlHaltParser(ctxt); |
7086 | 0 | goto error; |
7087 | 0 | } |
7088 | 0 | } else if ((depth > 0) && |
7089 | 0 | (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
7090 | 0 | depth--; |
7091 | 0 | if (ctxt->input->id != inputIds[depth]) { |
7092 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
7093 | 0 | "All markup of the conditional section is not" |
7094 | 0 | " in the same entity\n"); |
7095 | 0 | } |
7096 | 0 | SKIP(3); |
7097 | 0 | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
7098 | 0 | xmlParseMarkupDecl(ctxt); |
7099 | 0 | } else { |
7100 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7101 | 0 | xmlHaltParser(ctxt); |
7102 | 0 | goto error; |
7103 | 0 | } |
7104 | | |
7105 | 0 | if (depth == 0) |
7106 | 0 | break; |
7107 | | |
7108 | 0 | SKIP_BLANKS_PE; |
7109 | 0 | SHRINK; |
7110 | 0 | GROW; |
7111 | 0 | } |
7112 | | |
7113 | 0 | error: |
7114 | 0 | xmlFree(inputIds); |
7115 | 0 | } |
7116 | | |
7117 | | /** |
7118 | | * xmlParseMarkupDecl: |
7119 | | * @ctxt: an XML parser context |
7120 | | * |
7121 | | * DEPRECATED: Internal function, don't use. |
7122 | | * |
7123 | | * Parse markup declarations. Always consumes '<!' or '<?'. |
7124 | | * |
7125 | | * [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl | |
7126 | | * NotationDecl | PI | Comment |
7127 | | * |
7128 | | * [ VC: Proper Declaration/PE Nesting ] |
7129 | | * Parameter-entity replacement text must be properly nested with |
7130 | | * markup declarations. That is to say, if either the first character |
7131 | | * or the last character of a markup declaration (markupdecl above) is |
7132 | | * contained in the replacement text for a parameter-entity reference, |
7133 | | * both must be contained in the same replacement text. |
7134 | | * |
7135 | | * [ WFC: PEs in Internal Subset ] |
7136 | | * In the internal DTD subset, parameter-entity references can occur |
7137 | | * only where markup declarations can occur, not within markup declarations. |
7138 | | * (This does not apply to references that occur in external parameter |
7139 | | * entities or to the external subset.) |
7140 | | */ |
7141 | | void |
7142 | 93.1k | xmlParseMarkupDecl(xmlParserCtxtPtr ctxt) { |
7143 | 93.1k | GROW; |
7144 | 93.1k | if (CUR == '<') { |
7145 | 93.1k | if (NXT(1) == '!') { |
7146 | 84.5k | switch (NXT(2)) { |
7147 | 64.2k | case 'E': |
7148 | 64.2k | if (NXT(3) == 'L') |
7149 | 10.0k | xmlParseElementDecl(ctxt); |
7150 | 54.2k | else if (NXT(3) == 'N') |
7151 | 54.1k | xmlParseEntityDecl(ctxt); |
7152 | 63 | else |
7153 | 63 | SKIP(2); |
7154 | 64.2k | break; |
7155 | 11.9k | case 'A': |
7156 | 11.9k | xmlParseAttributeListDecl(ctxt); |
7157 | 11.9k | break; |
7158 | 3.30k | case 'N': |
7159 | 3.30k | xmlParseNotationDecl(ctxt); |
7160 | 3.30k | break; |
7161 | 4.39k | case '-': |
7162 | 4.39k | xmlParseComment(ctxt); |
7163 | 4.39k | break; |
7164 | 595 | default: |
7165 | | /* there is an error but it will be detected later */ |
7166 | 595 | SKIP(2); |
7167 | 595 | break; |
7168 | 84.5k | } |
7169 | 84.5k | } else if (NXT(1) == '?') { |
7170 | 8.68k | xmlParsePI(ctxt); |
7171 | 8.68k | } |
7172 | 93.1k | } |
7173 | 93.1k | } |
7174 | | |
7175 | | /** |
7176 | | * xmlParseTextDecl: |
7177 | | * @ctxt: an XML parser context |
7178 | | * |
7179 | | * DEPRECATED: Internal function, don't use. |
7180 | | * |
7181 | | * parse an XML declaration header for external entities |
7182 | | * |
7183 | | * [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>' |
7184 | | */ |
7185 | | |
7186 | | void |
7187 | 0 | xmlParseTextDecl(xmlParserCtxtPtr ctxt) { |
7188 | 0 | xmlChar *version; |
7189 | | |
7190 | | /* |
7191 | | * We know that '<?xml' is here. |
7192 | | */ |
7193 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
7194 | 0 | SKIP(5); |
7195 | 0 | } else { |
7196 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL); |
7197 | 0 | return; |
7198 | 0 | } |
7199 | | |
7200 | 0 | if (SKIP_BLANKS == 0) { |
7201 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
7202 | 0 | "Space needed after '<?xml'\n"); |
7203 | 0 | } |
7204 | | |
7205 | | /* |
7206 | | * We may have the VersionInfo here. |
7207 | | */ |
7208 | 0 | version = xmlParseVersionInfo(ctxt); |
7209 | 0 | if (version == NULL) { |
7210 | 0 | version = xmlCharStrdup(XML_DEFAULT_VERSION); |
7211 | 0 | if (version == NULL) { |
7212 | 0 | xmlErrMemory(ctxt); |
7213 | 0 | return; |
7214 | 0 | } |
7215 | 0 | } else { |
7216 | 0 | if (SKIP_BLANKS == 0) { |
7217 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
7218 | 0 | "Space needed here\n"); |
7219 | 0 | } |
7220 | 0 | } |
7221 | 0 | ctxt->input->version = version; |
7222 | | |
7223 | | /* |
7224 | | * We must have the encoding declaration |
7225 | | */ |
7226 | 0 | xmlParseEncodingDecl(ctxt); |
7227 | |
|
7228 | 0 | SKIP_BLANKS; |
7229 | 0 | if ((RAW == '?') && (NXT(1) == '>')) { |
7230 | 0 | SKIP(2); |
7231 | 0 | } else if (RAW == '>') { |
7232 | | /* Deprecated old WD ... */ |
7233 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7234 | 0 | NEXT; |
7235 | 0 | } else { |
7236 | 0 | int c; |
7237 | |
|
7238 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7239 | 0 | while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) { |
7240 | 0 | NEXT; |
7241 | 0 | if (c == '>') |
7242 | 0 | break; |
7243 | 0 | } |
7244 | 0 | } |
7245 | 0 | } |
7246 | | |
7247 | | /** |
7248 | | * xmlParseExternalSubset: |
7249 | | * @ctxt: an XML parser context |
7250 | | * @ExternalID: the external identifier |
7251 | | * @SystemID: the system identifier (or URL) |
7252 | | * |
7253 | | * parse Markup declarations from an external subset |
7254 | | * |
7255 | | * [30] extSubset ::= textDecl? extSubsetDecl |
7256 | | * |
7257 | | * [31] extSubsetDecl ::= (markupdecl | conditionalSect | PEReference | S) * |
7258 | | */ |
7259 | | void |
7260 | | xmlParseExternalSubset(xmlParserCtxtPtr ctxt, const xmlChar *ExternalID, |
7261 | 0 | const xmlChar *SystemID) { |
7262 | 0 | int oldInputNr; |
7263 | |
|
7264 | 0 | xmlCtxtInitializeLate(ctxt); |
7265 | |
|
7266 | 0 | xmlDetectEncoding(ctxt); |
7267 | |
|
7268 | 0 | if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) { |
7269 | 0 | xmlParseTextDecl(ctxt); |
7270 | 0 | } |
7271 | 0 | if (ctxt->myDoc == NULL) { |
7272 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
7273 | 0 | if (ctxt->myDoc == NULL) { |
7274 | 0 | xmlErrMemory(ctxt); |
7275 | 0 | return; |
7276 | 0 | } |
7277 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
7278 | 0 | } |
7279 | 0 | if ((ctxt->myDoc != NULL) && (ctxt->myDoc->intSubset == NULL) && |
7280 | 0 | (xmlCreateIntSubset(ctxt->myDoc, NULL, ExternalID, SystemID) == NULL)) { |
7281 | 0 | xmlErrMemory(ctxt); |
7282 | 0 | } |
7283 | |
|
7284 | 0 | ctxt->inSubset = 2; |
7285 | 0 | oldInputNr = ctxt->inputNr; |
7286 | |
|
7287 | 0 | SKIP_BLANKS_PE; |
7288 | 0 | while (((RAW != 0) || (ctxt->inputNr > oldInputNr)) && |
7289 | 0 | (!PARSER_STOPPED(ctxt))) { |
7290 | 0 | GROW; |
7291 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7292 | 0 | xmlParseConditionalSections(ctxt); |
7293 | 0 | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
7294 | 0 | xmlParseMarkupDecl(ctxt); |
7295 | 0 | } else { |
7296 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7297 | 0 | xmlHaltParser(ctxt); |
7298 | 0 | return; |
7299 | 0 | } |
7300 | 0 | SKIP_BLANKS_PE; |
7301 | 0 | SHRINK; |
7302 | 0 | } |
7303 | | |
7304 | 0 | while (ctxt->inputNr > oldInputNr) |
7305 | 0 | xmlPopPE(ctxt); |
7306 | |
|
7307 | 0 | if (RAW != 0) { |
7308 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7309 | 0 | } |
7310 | 0 | } |
7311 | | |
7312 | | /** |
7313 | | * xmlParseReference: |
7314 | | * @ctxt: an XML parser context |
7315 | | * |
7316 | | * DEPRECATED: Internal function, don't use. |
7317 | | * |
7318 | | * parse and handle entity references in content, depending on the SAX |
7319 | | * interface, this may end-up in a call to character() if this is a |
7320 | | * CharRef, a predefined entity, if there is no reference() callback. |
7321 | | * or if the parser was asked to switch to that mode. |
7322 | | * |
7323 | | * Always consumes '&'. |
7324 | | * |
7325 | | * [67] Reference ::= EntityRef | CharRef |
7326 | | */ |
7327 | | void |
7328 | 307k | xmlParseReference(xmlParserCtxtPtr ctxt) { |
7329 | 307k | xmlEntityPtr ent = NULL; |
7330 | 307k | const xmlChar *name; |
7331 | 307k | xmlChar *val; |
7332 | | |
7333 | 307k | if (RAW != '&') |
7334 | 0 | return; |
7335 | | |
7336 | | /* |
7337 | | * Simple case of a CharRef |
7338 | | */ |
7339 | 307k | if (NXT(1) == '#') { |
7340 | 90.6k | int i = 0; |
7341 | 90.6k | xmlChar out[16]; |
7342 | 90.6k | int value = xmlParseCharRef(ctxt); |
7343 | | |
7344 | 90.6k | if (value == 0) |
7345 | 59.9k | return; |
7346 | | |
7347 | | /* |
7348 | | * Just encode the value in UTF-8 |
7349 | | */ |
7350 | 30.6k | COPY_BUF(out, i, value); |
7351 | 30.6k | out[i] = 0; |
7352 | 30.6k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7353 | 30.6k | (!ctxt->disableSAX)) |
7354 | 1.07k | ctxt->sax->characters(ctxt->userData, out, i); |
7355 | 30.6k | return; |
7356 | 90.6k | } |
7357 | | |
7358 | | /* |
7359 | | * We are seeing an entity reference |
7360 | | */ |
7361 | 216k | name = xmlParseEntityRefInternal(ctxt); |
7362 | 216k | if (name == NULL) |
7363 | 124k | return; |
7364 | 92.5k | ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0); |
7365 | 92.5k | if (ent == NULL) { |
7366 | | /* |
7367 | | * Create a reference for undeclared entities. |
7368 | | */ |
7369 | 47.0k | if ((ctxt->replaceEntities == 0) && |
7370 | 47.0k | (ctxt->sax != NULL) && |
7371 | 47.0k | (ctxt->disableSAX == 0) && |
7372 | 47.0k | (ctxt->sax->reference != NULL)) { |
7373 | 448 | ctxt->sax->reference(ctxt->userData, name); |
7374 | 448 | } |
7375 | 47.0k | return; |
7376 | 47.0k | } |
7377 | 45.5k | if (!ctxt->wellFormed) |
7378 | 42.9k | return; |
7379 | | |
7380 | | /* special case of predefined entities */ |
7381 | 2.58k | if ((ent->name == NULL) || |
7382 | 2.58k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
7383 | 316 | val = ent->content; |
7384 | 316 | if (val == NULL) return; |
7385 | | /* |
7386 | | * inline the entity. |
7387 | | */ |
7388 | 316 | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7389 | 316 | (!ctxt->disableSAX)) |
7390 | 316 | ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val)); |
7391 | 316 | return; |
7392 | 316 | } |
7393 | | |
7394 | | /* |
7395 | | * Some users try to parse entities on their own and used to set |
7396 | | * the renamed "checked" member. Fix the flags to cover this |
7397 | | * case. |
7398 | | */ |
7399 | 2.26k | if (((ent->flags & XML_ENT_PARSED) == 0) && (ent->children != NULL)) |
7400 | 0 | ent->flags |= XML_ENT_PARSED; |
7401 | | |
7402 | | /* |
7403 | | * The first reference to the entity trigger a parsing phase |
7404 | | * where the ent->children is filled with the result from |
7405 | | * the parsing. |
7406 | | * Note: external parsed entities will not be loaded, it is not |
7407 | | * required for a non-validating parser, unless the parsing option |
7408 | | * of validating, or substituting entities were given. Doing so is |
7409 | | * far more secure as the parser will only process data coming from |
7410 | | * the document entity by default. |
7411 | | * |
7412 | | * FIXME: This doesn't work correctly since entities can be |
7413 | | * expanded with different namespace declarations in scope. |
7414 | | * For example: |
7415 | | * |
7416 | | * <!DOCTYPE doc [ |
7417 | | * <!ENTITY ent "<ns:elem/>"> |
7418 | | * ]> |
7419 | | * <doc> |
7420 | | * <decl1 xmlns:ns="urn:ns1"> |
7421 | | * &ent; |
7422 | | * </decl1> |
7423 | | * <decl2 xmlns:ns="urn:ns2"> |
7424 | | * &ent; |
7425 | | * </decl2> |
7426 | | * </doc> |
7427 | | * |
7428 | | * Proposed fix: |
7429 | | * |
7430 | | * - Ignore current namespace declarations when parsing the |
7431 | | * entity. If a prefix can't be resolved, don't report an error |
7432 | | * but mark it as unresolved. |
7433 | | * - Try to resolve these prefixes when expanding the entity. |
7434 | | * This will require a specialized version of xmlStaticCopyNode |
7435 | | * which can also make use of the namespace hash table to avoid |
7436 | | * quadratic behavior. |
7437 | | * |
7438 | | * Alternatively, we could simply reparse the entity on each |
7439 | | * expansion like we already do with custom SAX callbacks. |
7440 | | * External entity content should be cached in this case. |
7441 | | */ |
7442 | 2.26k | if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) || |
7443 | 2.26k | (((ctxt->options & XML_PARSE_NO_XXE) == 0) && |
7444 | 448 | ((ctxt->replaceEntities) || |
7445 | 1.81k | (ctxt->validate)))) { |
7446 | 1.81k | if ((ent->flags & XML_ENT_PARSED) == 0) { |
7447 | 556 | xmlCtxtParseEntity(ctxt, ent); |
7448 | 1.26k | } else if (ent->children == NULL) { |
7449 | | /* |
7450 | | * Probably running in SAX mode and the callbacks don't |
7451 | | * build the entity content. Parse the entity again. |
7452 | | * |
7453 | | * This will also be triggered in normal tree builder mode |
7454 | | * if an entity happens to be empty, causing unnecessary |
7455 | | * reloads. It's hard to come up with a reliable check in |
7456 | | * which mode we're running. |
7457 | | */ |
7458 | 328 | xmlCtxtParseEntity(ctxt, ent); |
7459 | 328 | } |
7460 | 1.81k | } |
7461 | | |
7462 | | /* |
7463 | | * We also check for amplification if entities aren't substituted. |
7464 | | * They might be expanded later. |
7465 | | */ |
7466 | 2.26k | if (xmlParserEntityCheck(ctxt, ent->expandedSize)) |
7467 | 10 | return; |
7468 | | |
7469 | 2.25k | if ((ctxt->sax == NULL) || (ctxt->disableSAX)) |
7470 | 425 | return; |
7471 | | |
7472 | 1.83k | if (ctxt->replaceEntities == 0) { |
7473 | | /* |
7474 | | * Create a reference |
7475 | | */ |
7476 | 1.83k | if (ctxt->sax->reference != NULL) |
7477 | 1.83k | ctxt->sax->reference(ctxt->userData, ent->name); |
7478 | 1.83k | } else if ((ent->children != NULL) && (ctxt->node != NULL)) { |
7479 | 0 | xmlNodePtr copy, cur; |
7480 | | |
7481 | | /* |
7482 | | * Seems we are generating the DOM content, copy the tree |
7483 | | */ |
7484 | 0 | cur = ent->children; |
7485 | | |
7486 | | /* |
7487 | | * Handle first text node with SAX to coalesce text efficiently |
7488 | | */ |
7489 | 0 | if ((cur->type == XML_TEXT_NODE) || |
7490 | 0 | (cur->type == XML_CDATA_SECTION_NODE)) { |
7491 | 0 | int len = xmlStrlen(cur->content); |
7492 | |
|
7493 | 0 | if ((cur->type == XML_TEXT_NODE) || |
7494 | 0 | (ctxt->sax->cdataBlock == NULL)) { |
7495 | 0 | if (ctxt->sax->characters != NULL) |
7496 | 0 | ctxt->sax->characters(ctxt, cur->content, len); |
7497 | 0 | } else { |
7498 | 0 | if (ctxt->sax->cdataBlock != NULL) |
7499 | 0 | ctxt->sax->cdataBlock(ctxt, cur->content, len); |
7500 | 0 | } |
7501 | |
|
7502 | 0 | cur = cur->next; |
7503 | 0 | } |
7504 | |
|
7505 | 0 | while (cur != NULL) { |
7506 | 0 | xmlNodePtr last; |
7507 | | |
7508 | | /* |
7509 | | * Handle last text node with SAX to coalesce text efficiently |
7510 | | */ |
7511 | 0 | if ((cur->next == NULL) && |
7512 | 0 | ((cur->type == XML_TEXT_NODE) || |
7513 | 0 | (cur->type == XML_CDATA_SECTION_NODE))) { |
7514 | 0 | int len = xmlStrlen(cur->content); |
7515 | |
|
7516 | 0 | if ((cur->type == XML_TEXT_NODE) || |
7517 | 0 | (ctxt->sax->cdataBlock == NULL)) { |
7518 | 0 | if (ctxt->sax->characters != NULL) |
7519 | 0 | ctxt->sax->characters(ctxt, cur->content, len); |
7520 | 0 | } else { |
7521 | 0 | if (ctxt->sax->cdataBlock != NULL) |
7522 | 0 | ctxt->sax->cdataBlock(ctxt, cur->content, len); |
7523 | 0 | } |
7524 | |
|
7525 | 0 | break; |
7526 | 0 | } |
7527 | | |
7528 | | /* |
7529 | | * Reset coalesce buffer stats only for non-text nodes. |
7530 | | */ |
7531 | 0 | ctxt->nodemem = 0; |
7532 | 0 | ctxt->nodelen = 0; |
7533 | |
|
7534 | 0 | copy = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7535 | |
|
7536 | 0 | if (copy == NULL) { |
7537 | 0 | xmlErrMemory(ctxt); |
7538 | 0 | break; |
7539 | 0 | } |
7540 | | |
7541 | 0 | if (ctxt->parseMode == XML_PARSE_READER) { |
7542 | | /* Needed for reader */ |
7543 | 0 | copy->extra = cur->extra; |
7544 | | /* Maybe needed for reader */ |
7545 | 0 | copy->_private = cur->_private; |
7546 | 0 | } |
7547 | |
|
7548 | 0 | copy->parent = ctxt->node; |
7549 | 0 | last = ctxt->node->last; |
7550 | 0 | if (last == NULL) { |
7551 | 0 | ctxt->node->children = copy; |
7552 | 0 | } else { |
7553 | 0 | last->next = copy; |
7554 | 0 | copy->prev = last; |
7555 | 0 | } |
7556 | 0 | ctxt->node->last = copy; |
7557 | |
|
7558 | 0 | cur = cur->next; |
7559 | 0 | } |
7560 | 0 | } |
7561 | 1.83k | } |
7562 | | |
7563 | | static void |
7564 | 156k | xmlHandleUndeclaredEntity(xmlParserCtxtPtr ctxt, const xmlChar *name) { |
7565 | | /* |
7566 | | * [ WFC: Entity Declared ] |
7567 | | * In a document without any DTD, a document with only an |
7568 | | * internal DTD subset which contains no parameter entity |
7569 | | * references, or a document with "standalone='yes'", the |
7570 | | * Name given in the entity reference must match that in an |
7571 | | * entity declaration, except that well-formed documents |
7572 | | * need not declare any of the following entities: amp, lt, |
7573 | | * gt, apos, quot. |
7574 | | * The declaration of a parameter entity must precede any |
7575 | | * reference to it. |
7576 | | * Similarly, the declaration of a general entity must |
7577 | | * precede any reference to it which appears in a default |
7578 | | * value in an attribute-list declaration. Note that if |
7579 | | * entities are declared in the external subset or in |
7580 | | * external parameter entities, a non-validating processor |
7581 | | * is not obligated to read and process their declarations; |
7582 | | * for such documents, the rule that an entity must be |
7583 | | * declared is a well-formedness constraint only if |
7584 | | * standalone='yes'. |
7585 | | */ |
7586 | 156k | if ((ctxt->standalone == 1) || |
7587 | 156k | ((ctxt->hasExternalSubset == 0) && |
7588 | 155k | (ctxt->hasPErefs == 0))) { |
7589 | 135k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7590 | 135k | "Entity '%s' not defined\n", name); |
7591 | 135k | } else if (ctxt->validate) { |
7592 | | /* |
7593 | | * [ VC: Entity Declared ] |
7594 | | * In a document with an external subset or external |
7595 | | * parameter entities with "standalone='no'", ... |
7596 | | * ... The declaration of a parameter entity must |
7597 | | * precede any reference to it... |
7598 | | */ |
7599 | 0 | xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7600 | 0 | "Entity '%s' not defined\n", name, NULL); |
7601 | 20.7k | } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) || |
7602 | 20.7k | ((ctxt->replaceEntities) && |
7603 | 20.7k | ((ctxt->options & XML_PARSE_NO_XXE) == 0))) { |
7604 | | /* |
7605 | | * Also raise a non-fatal error |
7606 | | * |
7607 | | * - if the external subset is loaded and all entity declarations |
7608 | | * should be available, or |
7609 | | * - entity substition was requested without restricting |
7610 | | * external entity access. |
7611 | | */ |
7612 | 0 | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7613 | 0 | "Entity '%s' not defined\n", name); |
7614 | 20.7k | } else { |
7615 | 20.7k | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7616 | 20.7k | "Entity '%s' not defined\n", name, NULL); |
7617 | 20.7k | } |
7618 | | |
7619 | 156k | ctxt->valid = 0; |
7620 | 156k | } |
7621 | | |
7622 | | static xmlEntityPtr |
7623 | 2.28M | xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) { |
7624 | 2.28M | xmlEntityPtr ent = NULL; |
7625 | | |
7626 | | /* |
7627 | | * Predefined entities override any extra definition |
7628 | | */ |
7629 | 2.28M | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7630 | 2.28M | ent = xmlGetPredefinedEntity(name); |
7631 | 2.28M | if (ent != NULL) |
7632 | 1.70M | return(ent); |
7633 | 2.28M | } |
7634 | | |
7635 | | /* |
7636 | | * Ask first SAX for entity resolution, otherwise try the |
7637 | | * entities which may have stored in the parser context. |
7638 | | */ |
7639 | 577k | if (ctxt->sax != NULL) { |
7640 | 577k | if (ctxt->sax->getEntity != NULL) |
7641 | 577k | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7642 | 577k | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7643 | 577k | (ctxt->options & XML_PARSE_OLDSAX)) |
7644 | 0 | ent = xmlGetPredefinedEntity(name); |
7645 | 577k | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7646 | 577k | (ctxt->userData==ctxt)) { |
7647 | 3.29k | ent = xmlSAX2GetEntity(ctxt, name); |
7648 | 3.29k | } |
7649 | 577k | } |
7650 | | |
7651 | 577k | if (ent == NULL) { |
7652 | 146k | xmlHandleUndeclaredEntity(ctxt, name); |
7653 | 146k | } |
7654 | | |
7655 | | /* |
7656 | | * [ WFC: Parsed Entity ] |
7657 | | * An entity reference must not contain the name of an |
7658 | | * unparsed entity |
7659 | | */ |
7660 | 430k | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7661 | 276 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7662 | 276 | "Entity reference to unparsed entity %s\n", name); |
7663 | 276 | ent = NULL; |
7664 | 276 | } |
7665 | | |
7666 | | /* |
7667 | | * [ WFC: No External Entity References ] |
7668 | | * Attribute values cannot contain direct or indirect |
7669 | | * entity references to external entities. |
7670 | | */ |
7671 | 430k | else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7672 | 1.57k | if (inAttr) { |
7673 | 417 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7674 | 417 | "Attribute references external entity '%s'\n", name); |
7675 | 417 | ent = NULL; |
7676 | 417 | } |
7677 | 1.57k | } |
7678 | | |
7679 | 577k | return(ent); |
7680 | 2.28M | } |
7681 | | |
7682 | | /** |
7683 | | * xmlParseEntityRefInternal: |
7684 | | * @ctxt: an XML parser context |
7685 | | * @inAttr: whether we are in an attribute value |
7686 | | * |
7687 | | * Parse an entity reference. Always consumes '&'. |
7688 | | * |
7689 | | * [68] EntityRef ::= '&' Name ';' |
7690 | | * |
7691 | | * Returns the name, or NULL in case of error. |
7692 | | */ |
7693 | | static const xmlChar * |
7694 | 863k | xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) { |
7695 | 863k | const xmlChar *name; |
7696 | | |
7697 | 863k | GROW; |
7698 | | |
7699 | 863k | if (RAW != '&') |
7700 | 0 | return(NULL); |
7701 | 863k | NEXT; |
7702 | 863k | name = xmlParseName(ctxt); |
7703 | 863k | if (name == NULL) { |
7704 | 116k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7705 | 116k | "xmlParseEntityRef: no name\n"); |
7706 | 116k | return(NULL); |
7707 | 116k | } |
7708 | 746k | if (RAW != ';') { |
7709 | 59.4k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7710 | 59.4k | return(NULL); |
7711 | 59.4k | } |
7712 | 687k | NEXT; |
7713 | | |
7714 | 687k | return(name); |
7715 | 746k | } |
7716 | | |
7717 | | /** |
7718 | | * xmlParseEntityRef: |
7719 | | * @ctxt: an XML parser context |
7720 | | * |
7721 | | * DEPRECATED: Internal function, don't use. |
7722 | | * |
7723 | | * Returns the xmlEntityPtr if found, or NULL otherwise. |
7724 | | */ |
7725 | | xmlEntityPtr |
7726 | 0 | xmlParseEntityRef(xmlParserCtxtPtr ctxt) { |
7727 | 0 | const xmlChar *name; |
7728 | |
|
7729 | 0 | if (ctxt == NULL) |
7730 | 0 | return(NULL); |
7731 | | |
7732 | 0 | name = xmlParseEntityRefInternal(ctxt); |
7733 | 0 | if (name == NULL) |
7734 | 0 | return(NULL); |
7735 | | |
7736 | 0 | return(xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 0)); |
7737 | 0 | } |
7738 | | |
7739 | | /** |
7740 | | * xmlParseStringEntityRef: |
7741 | | * @ctxt: an XML parser context |
7742 | | * @str: a pointer to an index in the string |
7743 | | * |
7744 | | * parse ENTITY references declarations, but this version parses it from |
7745 | | * a string value. |
7746 | | * |
7747 | | * [68] EntityRef ::= '&' Name ';' |
7748 | | * |
7749 | | * [ WFC: Entity Declared ] |
7750 | | * In a document without any DTD, a document with only an internal DTD |
7751 | | * subset which contains no parameter entity references, or a document |
7752 | | * with "standalone='yes'", the Name given in the entity reference |
7753 | | * must match that in an entity declaration, except that well-formed |
7754 | | * documents need not declare any of the following entities: amp, lt, |
7755 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7756 | | * any reference to it. Similarly, the declaration of a general entity |
7757 | | * must precede any reference to it which appears in a default value in an |
7758 | | * attribute-list declaration. Note that if entities are declared in the |
7759 | | * external subset or in external parameter entities, a non-validating |
7760 | | * processor is not obligated to read and process their declarations; |
7761 | | * for such documents, the rule that an entity must be declared is a |
7762 | | * well-formedness constraint only if standalone='yes'. |
7763 | | * |
7764 | | * [ WFC: Parsed Entity ] |
7765 | | * An entity reference must not contain the name of an unparsed entity |
7766 | | * |
7767 | | * Returns the xmlEntityPtr if found, or NULL otherwise. The str pointer |
7768 | | * is updated to the current location in the string. |
7769 | | */ |
7770 | | static xmlChar * |
7771 | 1.59M | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) { |
7772 | 1.59M | xmlChar *name; |
7773 | 1.59M | const xmlChar *ptr; |
7774 | 1.59M | xmlChar cur; |
7775 | | |
7776 | 1.59M | if ((str == NULL) || (*str == NULL)) |
7777 | 0 | return(NULL); |
7778 | 1.59M | ptr = *str; |
7779 | 1.59M | cur = *ptr; |
7780 | 1.59M | if (cur != '&') |
7781 | 0 | return(NULL); |
7782 | | |
7783 | 1.59M | ptr++; |
7784 | 1.59M | name = xmlParseStringName(ctxt, &ptr); |
7785 | 1.59M | if (name == NULL) { |
7786 | 7 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7787 | 7 | "xmlParseStringEntityRef: no name\n"); |
7788 | 7 | *str = ptr; |
7789 | 7 | return(NULL); |
7790 | 7 | } |
7791 | 1.59M | if (*ptr != ';') { |
7792 | 5 | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7793 | 5 | xmlFree(name); |
7794 | 5 | *str = ptr; |
7795 | 5 | return(NULL); |
7796 | 5 | } |
7797 | 1.59M | ptr++; |
7798 | | |
7799 | 1.59M | *str = ptr; |
7800 | 1.59M | return(name); |
7801 | 1.59M | } |
7802 | | |
7803 | | /** |
7804 | | * xmlParsePEReference: |
7805 | | * @ctxt: an XML parser context |
7806 | | * |
7807 | | * DEPRECATED: Internal function, don't use. |
7808 | | * |
7809 | | * Parse a parameter entity reference. Always consumes '%'. |
7810 | | * |
7811 | | * The entity content is handled directly by pushing it's content as |
7812 | | * a new input stream. |
7813 | | * |
7814 | | * [69] PEReference ::= '%' Name ';' |
7815 | | * |
7816 | | * [ WFC: No Recursion ] |
7817 | | * A parsed entity must not contain a recursive |
7818 | | * reference to itself, either directly or indirectly. |
7819 | | * |
7820 | | * [ WFC: Entity Declared ] |
7821 | | * In a document without any DTD, a document with only an internal DTD |
7822 | | * subset which contains no parameter entity references, or a document |
7823 | | * with "standalone='yes'", ... ... The declaration of a parameter |
7824 | | * entity must precede any reference to it... |
7825 | | * |
7826 | | * [ VC: Entity Declared ] |
7827 | | * In a document with an external subset or external parameter entities |
7828 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
7829 | | * must precede any reference to it... |
7830 | | * |
7831 | | * [ WFC: In DTD ] |
7832 | | * Parameter-entity references may only appear in the DTD. |
7833 | | * NOTE: misleading but this is handled. |
7834 | | */ |
7835 | | void |
7836 | | xmlParsePEReference(xmlParserCtxtPtr ctxt) |
7837 | 32.8k | { |
7838 | 32.8k | const xmlChar *name; |
7839 | 32.8k | xmlEntityPtr entity = NULL; |
7840 | 32.8k | xmlParserInputPtr input; |
7841 | | |
7842 | 32.8k | if (RAW != '%') |
7843 | 0 | return; |
7844 | 32.8k | NEXT; |
7845 | 32.8k | name = xmlParseName(ctxt); |
7846 | 32.8k | if (name == NULL) { |
7847 | 4.61k | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n"); |
7848 | 4.61k | return; |
7849 | 4.61k | } |
7850 | 28.2k | if (RAW != ';') { |
7851 | 1.93k | xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL); |
7852 | 1.93k | return; |
7853 | 1.93k | } |
7854 | | |
7855 | 26.3k | NEXT; |
7856 | | |
7857 | | /* Must be set before xmlHandleUndeclaredEntity */ |
7858 | 26.3k | ctxt->hasPErefs = 1; |
7859 | | |
7860 | | /* |
7861 | | * Request the entity from SAX |
7862 | | */ |
7863 | 26.3k | if ((ctxt->sax != NULL) && |
7864 | 26.3k | (ctxt->sax->getParameterEntity != NULL)) |
7865 | 26.3k | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
7866 | | |
7867 | 26.3k | if (entity == NULL) { |
7868 | 7.95k | xmlHandleUndeclaredEntity(ctxt, name); |
7869 | 18.3k | } else { |
7870 | | /* |
7871 | | * Internal checking in case the entity quest barfed |
7872 | | */ |
7873 | 18.3k | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
7874 | 18.3k | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
7875 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7876 | 0 | "Internal: %%%s; is not a parameter entity\n", |
7877 | 0 | name, NULL); |
7878 | 18.3k | } else { |
7879 | 18.3k | if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
7880 | 18.3k | ((ctxt->options & XML_PARSE_NO_XXE) || |
7881 | 681 | ((ctxt->loadsubset == 0) && |
7882 | 681 | (ctxt->replaceEntities == 0) && |
7883 | 681 | (ctxt->validate == 0)))) |
7884 | 681 | return; |
7885 | | |
7886 | 17.6k | if (entity->flags & XML_ENT_EXPANDING) { |
7887 | 2 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7888 | 2 | xmlHaltParser(ctxt); |
7889 | 2 | return; |
7890 | 2 | } |
7891 | | |
7892 | 17.6k | input = xmlNewEntityInputStream(ctxt, entity); |
7893 | 17.6k | if (xmlPushInput(ctxt, input) < 0) { |
7894 | 0 | xmlFreeInputStream(input); |
7895 | 0 | return; |
7896 | 0 | } |
7897 | | |
7898 | 17.6k | entity->flags |= XML_ENT_EXPANDING; |
7899 | | |
7900 | 17.6k | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
7901 | 0 | xmlDetectEncoding(ctxt); |
7902 | |
|
7903 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
7904 | 0 | (IS_BLANK_CH(NXT(5)))) { |
7905 | 0 | xmlParseTextDecl(ctxt); |
7906 | 0 | } |
7907 | 0 | } |
7908 | 17.6k | } |
7909 | 18.3k | } |
7910 | 26.3k | } |
7911 | | |
7912 | | /** |
7913 | | * xmlLoadEntityContent: |
7914 | | * @ctxt: an XML parser context |
7915 | | * @entity: an unloaded system entity |
7916 | | * |
7917 | | * Load the content of an entity. |
7918 | | * |
7919 | | * Returns 0 in case of success and -1 in case of failure |
7920 | | */ |
7921 | | static int |
7922 | 0 | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) { |
7923 | 0 | xmlParserInputPtr oldinput, input = NULL; |
7924 | 0 | xmlParserInputPtr *oldinputTab; |
7925 | 0 | const xmlChar *oldencoding; |
7926 | 0 | xmlChar *content = NULL; |
7927 | 0 | xmlResourceType rtype; |
7928 | 0 | size_t length, i; |
7929 | 0 | int oldinputNr, oldinputMax; |
7930 | 0 | int ret = -1; |
7931 | 0 | int res; |
7932 | |
|
7933 | 0 | if ((ctxt == NULL) || (entity == NULL) || |
7934 | 0 | ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) && |
7935 | 0 | (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) || |
7936 | 0 | (entity->content != NULL)) { |
7937 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, |
7938 | 0 | "xmlLoadEntityContent parameter error"); |
7939 | 0 | return(-1); |
7940 | 0 | } |
7941 | | |
7942 | 0 | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) |
7943 | 0 | rtype = XML_RESOURCE_PARAMETER_ENTITY; |
7944 | 0 | else |
7945 | 0 | rtype = XML_RESOURCE_GENERAL_ENTITY; |
7946 | |
|
7947 | 0 | input = xmlLoadResource(ctxt, (char *) entity->URI, |
7948 | 0 | (char *) entity->ExternalID, rtype); |
7949 | 0 | if (input == NULL) |
7950 | 0 | return(-1); |
7951 | | |
7952 | 0 | oldinput = ctxt->input; |
7953 | 0 | oldinputNr = ctxt->inputNr; |
7954 | 0 | oldinputMax = ctxt->inputMax; |
7955 | 0 | oldinputTab = ctxt->inputTab; |
7956 | 0 | oldencoding = ctxt->encoding; |
7957 | |
|
7958 | 0 | ctxt->input = NULL; |
7959 | 0 | ctxt->inputNr = 0; |
7960 | 0 | ctxt->inputMax = 1; |
7961 | 0 | ctxt->encoding = NULL; |
7962 | 0 | ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr)); |
7963 | 0 | if (ctxt->inputTab == NULL) { |
7964 | 0 | xmlErrMemory(ctxt); |
7965 | 0 | xmlFreeInputStream(input); |
7966 | 0 | goto error; |
7967 | 0 | } |
7968 | | |
7969 | 0 | xmlBufResetInput(input->buf->buffer, input); |
7970 | |
|
7971 | 0 | if (inputPush(ctxt, input) < 0) { |
7972 | 0 | xmlFreeInputStream(input); |
7973 | 0 | goto error; |
7974 | 0 | } |
7975 | | |
7976 | 0 | xmlDetectEncoding(ctxt); |
7977 | | |
7978 | | /* |
7979 | | * Parse a possible text declaration first |
7980 | | */ |
7981 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
7982 | 0 | xmlParseTextDecl(ctxt); |
7983 | | /* |
7984 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
7985 | | */ |
7986 | 0 | if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) && |
7987 | 0 | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
7988 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
7989 | 0 | "Version mismatch between document and entity\n"); |
7990 | 0 | } |
7991 | 0 | } |
7992 | |
|
7993 | 0 | length = input->cur - input->base; |
7994 | 0 | xmlBufShrink(input->buf->buffer, length); |
7995 | 0 | xmlSaturatedAdd(&ctxt->sizeentities, length); |
7996 | |
|
7997 | 0 | while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0) |
7998 | 0 | ; |
7999 | |
|
8000 | 0 | xmlBufResetInput(input->buf->buffer, input); |
8001 | |
|
8002 | 0 | if (res < 0) { |
8003 | 0 | xmlCtxtErrIO(ctxt, input->buf->error, NULL); |
8004 | 0 | goto error; |
8005 | 0 | } |
8006 | | |
8007 | 0 | length = xmlBufUse(input->buf->buffer); |
8008 | 0 | if (length > INT_MAX) { |
8009 | 0 | xmlErrMemory(ctxt); |
8010 | 0 | goto error; |
8011 | 0 | } |
8012 | | |
8013 | 0 | content = xmlStrndup(xmlBufContent(input->buf->buffer), length); |
8014 | 0 | if (content == NULL) { |
8015 | 0 | xmlErrMemory(ctxt); |
8016 | 0 | goto error; |
8017 | 0 | } |
8018 | | |
8019 | 0 | for (i = 0; i < length; ) { |
8020 | 0 | int clen = length - i; |
8021 | 0 | int c = xmlGetUTF8Char(content + i, &clen); |
8022 | |
|
8023 | 0 | if ((c < 0) || (!IS_CHAR(c))) { |
8024 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
8025 | 0 | "xmlLoadEntityContent: invalid char value %d\n", |
8026 | 0 | content[i]); |
8027 | 0 | goto error; |
8028 | 0 | } |
8029 | 0 | i += clen; |
8030 | 0 | } |
8031 | | |
8032 | 0 | xmlSaturatedAdd(&ctxt->sizeentities, length); |
8033 | 0 | entity->content = content; |
8034 | 0 | entity->length = length; |
8035 | 0 | content = NULL; |
8036 | 0 | ret = 0; |
8037 | |
|
8038 | 0 | error: |
8039 | 0 | while (ctxt->inputNr > 0) |
8040 | 0 | xmlFreeInputStream(inputPop(ctxt)); |
8041 | 0 | xmlFree(ctxt->inputTab); |
8042 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
8043 | |
|
8044 | 0 | ctxt->input = oldinput; |
8045 | 0 | ctxt->inputNr = oldinputNr; |
8046 | 0 | ctxt->inputMax = oldinputMax; |
8047 | 0 | ctxt->inputTab = oldinputTab; |
8048 | 0 | ctxt->encoding = oldencoding; |
8049 | |
|
8050 | 0 | xmlFree(content); |
8051 | |
|
8052 | 0 | return(ret); |
8053 | 0 | } |
8054 | | |
8055 | | /** |
8056 | | * xmlParseStringPEReference: |
8057 | | * @ctxt: an XML parser context |
8058 | | * @str: a pointer to an index in the string |
8059 | | * |
8060 | | * parse PEReference declarations |
8061 | | * |
8062 | | * [69] PEReference ::= '%' Name ';' |
8063 | | * |
8064 | | * [ WFC: No Recursion ] |
8065 | | * A parsed entity must not contain a recursive |
8066 | | * reference to itself, either directly or indirectly. |
8067 | | * |
8068 | | * [ WFC: Entity Declared ] |
8069 | | * In a document without any DTD, a document with only an internal DTD |
8070 | | * subset which contains no parameter entity references, or a document |
8071 | | * with "standalone='yes'", ... ... The declaration of a parameter |
8072 | | * entity must precede any reference to it... |
8073 | | * |
8074 | | * [ VC: Entity Declared ] |
8075 | | * In a document with an external subset or external parameter entities |
8076 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
8077 | | * must precede any reference to it... |
8078 | | * |
8079 | | * [ WFC: In DTD ] |
8080 | | * Parameter-entity references may only appear in the DTD. |
8081 | | * NOTE: misleading but this is handled. |
8082 | | * |
8083 | | * Returns the string of the entity content. |
8084 | | * str is updated to the current value of the index |
8085 | | */ |
8086 | | static xmlEntityPtr |
8087 | 4.15k | xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
8088 | 4.15k | const xmlChar *ptr; |
8089 | 4.15k | xmlChar cur; |
8090 | 4.15k | xmlChar *name; |
8091 | 4.15k | xmlEntityPtr entity = NULL; |
8092 | | |
8093 | 4.15k | if ((str == NULL) || (*str == NULL)) return(NULL); |
8094 | 4.15k | ptr = *str; |
8095 | 4.15k | cur = *ptr; |
8096 | 4.15k | if (cur != '%') |
8097 | 0 | return(NULL); |
8098 | 4.15k | ptr++; |
8099 | 4.15k | name = xmlParseStringName(ctxt, &ptr); |
8100 | 4.15k | if (name == NULL) { |
8101 | 958 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8102 | 958 | "xmlParseStringPEReference: no name\n"); |
8103 | 958 | *str = ptr; |
8104 | 958 | return(NULL); |
8105 | 958 | } |
8106 | 3.19k | cur = *ptr; |
8107 | 3.19k | if (cur != ';') { |
8108 | 716 | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
8109 | 716 | xmlFree(name); |
8110 | 716 | *str = ptr; |
8111 | 716 | return(NULL); |
8112 | 716 | } |
8113 | 2.47k | ptr++; |
8114 | | |
8115 | | /* Must be set before xmlHandleUndeclaredEntity */ |
8116 | 2.47k | ctxt->hasPErefs = 1; |
8117 | | |
8118 | | /* |
8119 | | * Request the entity from SAX |
8120 | | */ |
8121 | 2.47k | if ((ctxt->sax != NULL) && |
8122 | 2.47k | (ctxt->sax->getParameterEntity != NULL)) |
8123 | 2.47k | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
8124 | | |
8125 | 2.47k | if (entity == NULL) { |
8126 | 1.83k | xmlHandleUndeclaredEntity(ctxt, name); |
8127 | 1.83k | } else { |
8128 | | /* |
8129 | | * Internal checking in case the entity quest barfed |
8130 | | */ |
8131 | 642 | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
8132 | 642 | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
8133 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8134 | 0 | "%%%s; is not a parameter entity\n", |
8135 | 0 | name, NULL); |
8136 | 0 | } |
8137 | 642 | } |
8138 | | |
8139 | 2.47k | xmlFree(name); |
8140 | 2.47k | *str = ptr; |
8141 | 2.47k | return(entity); |
8142 | 3.19k | } |
8143 | | |
8144 | | /** |
8145 | | * xmlParseDocTypeDecl: |
8146 | | * @ctxt: an XML parser context |
8147 | | * |
8148 | | * DEPRECATED: Internal function, don't use. |
8149 | | * |
8150 | | * parse a DOCTYPE declaration |
8151 | | * |
8152 | | * [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? |
8153 | | * ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8154 | | * |
8155 | | * [ VC: Root Element Type ] |
8156 | | * The Name in the document type declaration must match the element |
8157 | | * type of the root element. |
8158 | | */ |
8159 | | |
8160 | | void |
8161 | 6.97k | xmlParseDocTypeDecl(xmlParserCtxtPtr ctxt) { |
8162 | 6.97k | const xmlChar *name = NULL; |
8163 | 6.97k | xmlChar *ExternalID = NULL; |
8164 | 6.97k | xmlChar *URI = NULL; |
8165 | | |
8166 | | /* |
8167 | | * We know that '<!DOCTYPE' has been detected. |
8168 | | */ |
8169 | 6.97k | SKIP(9); |
8170 | | |
8171 | 6.97k | SKIP_BLANKS; |
8172 | | |
8173 | | /* |
8174 | | * Parse the DOCTYPE name. |
8175 | | */ |
8176 | 6.97k | name = xmlParseName(ctxt); |
8177 | 6.97k | if (name == NULL) { |
8178 | 527 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8179 | 527 | "xmlParseDocTypeDecl : no DOCTYPE name !\n"); |
8180 | 527 | } |
8181 | 6.97k | ctxt->intSubName = name; |
8182 | | |
8183 | 6.97k | SKIP_BLANKS; |
8184 | | |
8185 | | /* |
8186 | | * Check for SystemID and ExternalID |
8187 | | */ |
8188 | 6.97k | URI = xmlParseExternalID(ctxt, &ExternalID, 1); |
8189 | | |
8190 | 6.97k | if ((URI != NULL) || (ExternalID != NULL)) { |
8191 | 277 | ctxt->hasExternalSubset = 1; |
8192 | 277 | } |
8193 | 6.97k | ctxt->extSubURI = URI; |
8194 | 6.97k | ctxt->extSubSystem = ExternalID; |
8195 | | |
8196 | 6.97k | SKIP_BLANKS; |
8197 | | |
8198 | | /* |
8199 | | * Create and update the internal subset. |
8200 | | */ |
8201 | 6.97k | if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) && |
8202 | 6.97k | (!ctxt->disableSAX)) |
8203 | 5.77k | ctxt->sax->internalSubset(ctxt->userData, name, ExternalID, URI); |
8204 | | |
8205 | | /* |
8206 | | * Is there any internal subset declarations ? |
8207 | | * they are handled separately in xmlParseInternalSubset() |
8208 | | */ |
8209 | 6.97k | if (RAW == '[') |
8210 | 6.50k | return; |
8211 | | |
8212 | | /* |
8213 | | * We should be at the end of the DOCTYPE declaration. |
8214 | | */ |
8215 | 466 | if (RAW != '>') { |
8216 | 306 | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8217 | 306 | } |
8218 | 466 | NEXT; |
8219 | 466 | } |
8220 | | |
8221 | | /** |
8222 | | * xmlParseInternalSubset: |
8223 | | * @ctxt: an XML parser context |
8224 | | * |
8225 | | * parse the internal subset declaration |
8226 | | * |
8227 | | * [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8228 | | */ |
8229 | | |
8230 | | static void |
8231 | 6.59k | xmlParseInternalSubset(xmlParserCtxtPtr ctxt) { |
8232 | | /* |
8233 | | * Is there any DTD definition ? |
8234 | | */ |
8235 | 6.59k | if (RAW == '[') { |
8236 | 6.59k | int oldInputNr = ctxt->inputNr; |
8237 | | |
8238 | 6.59k | NEXT; |
8239 | | /* |
8240 | | * Parse the succession of Markup declarations and |
8241 | | * PEReferences. |
8242 | | * Subsequence (markupdecl | PEReference | S)* |
8243 | | */ |
8244 | 6.59k | SKIP_BLANKS; |
8245 | 132k | while (((RAW != ']') || (ctxt->inputNr > oldInputNr)) && |
8246 | 132k | (PARSER_STOPPED(ctxt) == 0)) { |
8247 | | |
8248 | | /* |
8249 | | * Conditional sections are allowed from external entities included |
8250 | | * by PE References in the internal subset. |
8251 | | */ |
8252 | 129k | if ((PARSER_EXTERNAL(ctxt)) && |
8253 | 129k | (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
8254 | 0 | xmlParseConditionalSections(ctxt); |
8255 | 129k | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
8256 | 93.1k | xmlParseMarkupDecl(ctxt); |
8257 | 93.1k | } else if (RAW == '%') { |
8258 | 32.8k | xmlParsePEReference(ctxt); |
8259 | 32.8k | } else { |
8260 | 3.60k | xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL); |
8261 | 3.60k | break; |
8262 | 3.60k | } |
8263 | 126k | SKIP_BLANKS_PE; |
8264 | 126k | SHRINK; |
8265 | 126k | GROW; |
8266 | 126k | } |
8267 | | |
8268 | 6.65k | while (ctxt->inputNr > oldInputNr) |
8269 | 60 | xmlPopPE(ctxt); |
8270 | | |
8271 | 6.59k | if (RAW == ']') { |
8272 | 2.27k | NEXT; |
8273 | 2.27k | SKIP_BLANKS; |
8274 | 2.27k | } |
8275 | 6.59k | } |
8276 | | |
8277 | | /* |
8278 | | * We should be at the end of the DOCTYPE declaration. |
8279 | | */ |
8280 | 6.59k | if ((ctxt->wellFormed) && (RAW != '>')) { |
8281 | 108 | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8282 | 108 | return; |
8283 | 108 | } |
8284 | 6.49k | NEXT; |
8285 | 6.49k | } |
8286 | | |
8287 | | #ifdef LIBXML_SAX1_ENABLED |
8288 | | /** |
8289 | | * xmlParseAttribute: |
8290 | | * @ctxt: an XML parser context |
8291 | | * @value: a xmlChar ** used to store the value of the attribute |
8292 | | * |
8293 | | * DEPRECATED: Internal function, don't use. |
8294 | | * |
8295 | | * parse an attribute |
8296 | | * |
8297 | | * [41] Attribute ::= Name Eq AttValue |
8298 | | * |
8299 | | * [ WFC: No External Entity References ] |
8300 | | * Attribute values cannot contain direct or indirect entity references |
8301 | | * to external entities. |
8302 | | * |
8303 | | * [ WFC: No < in Attribute Values ] |
8304 | | * The replacement text of any entity referred to directly or indirectly in |
8305 | | * an attribute value (other than "<") must not contain a <. |
8306 | | * |
8307 | | * [ VC: Attribute Value Type ] |
8308 | | * The attribute must have been declared; the value must be of the type |
8309 | | * declared for it. |
8310 | | * |
8311 | | * [25] Eq ::= S? '=' S? |
8312 | | * |
8313 | | * With namespace: |
8314 | | * |
8315 | | * [NS 11] Attribute ::= QName Eq AttValue |
8316 | | * |
8317 | | * Also the case QName == xmlns:??? is handled independently as a namespace |
8318 | | * definition. |
8319 | | * |
8320 | | * Returns the attribute name, and the value in *value. |
8321 | | */ |
8322 | | |
8323 | | const xmlChar * |
8324 | 0 | xmlParseAttribute(xmlParserCtxtPtr ctxt, xmlChar **value) { |
8325 | 0 | const xmlChar *name; |
8326 | 0 | xmlChar *val; |
8327 | |
|
8328 | 0 | *value = NULL; |
8329 | 0 | GROW; |
8330 | 0 | name = xmlParseName(ctxt); |
8331 | 0 | if (name == NULL) { |
8332 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8333 | 0 | "error parsing attribute name\n"); |
8334 | 0 | return(NULL); |
8335 | 0 | } |
8336 | | |
8337 | | /* |
8338 | | * read the value |
8339 | | */ |
8340 | 0 | SKIP_BLANKS; |
8341 | 0 | if (RAW == '=') { |
8342 | 0 | NEXT; |
8343 | 0 | SKIP_BLANKS; |
8344 | 0 | val = xmlParseAttValue(ctxt); |
8345 | 0 | } else { |
8346 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8347 | 0 | "Specification mandates value for attribute %s\n", name); |
8348 | 0 | return(name); |
8349 | 0 | } |
8350 | | |
8351 | | /* |
8352 | | * Check that xml:lang conforms to the specification |
8353 | | * No more registered as an error, just generate a warning now |
8354 | | * since this was deprecated in XML second edition |
8355 | | */ |
8356 | 0 | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) { |
8357 | 0 | if (!xmlCheckLanguageID(val)) { |
8358 | 0 | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8359 | 0 | "Malformed value for xml:lang : %s\n", |
8360 | 0 | val, NULL); |
8361 | 0 | } |
8362 | 0 | } |
8363 | | |
8364 | | /* |
8365 | | * Check that xml:space conforms to the specification |
8366 | | */ |
8367 | 0 | if (xmlStrEqual(name, BAD_CAST "xml:space")) { |
8368 | 0 | if (xmlStrEqual(val, BAD_CAST "default")) |
8369 | 0 | *(ctxt->space) = 0; |
8370 | 0 | else if (xmlStrEqual(val, BAD_CAST "preserve")) |
8371 | 0 | *(ctxt->space) = 1; |
8372 | 0 | else { |
8373 | 0 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8374 | 0 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8375 | 0 | val, NULL); |
8376 | 0 | } |
8377 | 0 | } |
8378 | |
|
8379 | 0 | *value = val; |
8380 | 0 | return(name); |
8381 | 0 | } |
8382 | | |
8383 | | /** |
8384 | | * xmlParseStartTag: |
8385 | | * @ctxt: an XML parser context |
8386 | | * |
8387 | | * DEPRECATED: Internal function, don't use. |
8388 | | * |
8389 | | * Parse a start tag. Always consumes '<'. |
8390 | | * |
8391 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8392 | | * |
8393 | | * [ WFC: Unique Att Spec ] |
8394 | | * No attribute name may appear more than once in the same start-tag or |
8395 | | * empty-element tag. |
8396 | | * |
8397 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8398 | | * |
8399 | | * [ WFC: Unique Att Spec ] |
8400 | | * No attribute name may appear more than once in the same start-tag or |
8401 | | * empty-element tag. |
8402 | | * |
8403 | | * With namespace: |
8404 | | * |
8405 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8406 | | * |
8407 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8408 | | * |
8409 | | * Returns the element name parsed |
8410 | | */ |
8411 | | |
8412 | | const xmlChar * |
8413 | 0 | xmlParseStartTag(xmlParserCtxtPtr ctxt) { |
8414 | 0 | const xmlChar *name; |
8415 | 0 | const xmlChar *attname; |
8416 | 0 | xmlChar *attvalue; |
8417 | 0 | const xmlChar **atts = ctxt->atts; |
8418 | 0 | int nbatts = 0; |
8419 | 0 | int maxatts = ctxt->maxatts; |
8420 | 0 | int i; |
8421 | |
|
8422 | 0 | if (RAW != '<') return(NULL); |
8423 | 0 | NEXT1; |
8424 | |
|
8425 | 0 | name = xmlParseName(ctxt); |
8426 | 0 | if (name == NULL) { |
8427 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8428 | 0 | "xmlParseStartTag: invalid element name\n"); |
8429 | 0 | return(NULL); |
8430 | 0 | } |
8431 | | |
8432 | | /* |
8433 | | * Now parse the attributes, it ends up with the ending |
8434 | | * |
8435 | | * (S Attribute)* S? |
8436 | | */ |
8437 | 0 | SKIP_BLANKS; |
8438 | 0 | GROW; |
8439 | |
|
8440 | 0 | while (((RAW != '>') && |
8441 | 0 | ((RAW != '/') || (NXT(1) != '>')) && |
8442 | 0 | (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) { |
8443 | 0 | attname = xmlParseAttribute(ctxt, &attvalue); |
8444 | 0 | if (attname == NULL) |
8445 | 0 | break; |
8446 | 0 | if (attvalue != NULL) { |
8447 | | /* |
8448 | | * [ WFC: Unique Att Spec ] |
8449 | | * No attribute name may appear more than once in the same |
8450 | | * start-tag or empty-element tag. |
8451 | | */ |
8452 | 0 | for (i = 0; i < nbatts;i += 2) { |
8453 | 0 | if (xmlStrEqual(atts[i], attname)) { |
8454 | 0 | xmlErrAttributeDup(ctxt, NULL, attname); |
8455 | 0 | xmlFree(attvalue); |
8456 | 0 | goto failed; |
8457 | 0 | } |
8458 | 0 | } |
8459 | | /* |
8460 | | * Add the pair to atts |
8461 | | */ |
8462 | 0 | if (atts == NULL) { |
8463 | 0 | maxatts = 22; /* allow for 10 attrs by default */ |
8464 | 0 | atts = (const xmlChar **) |
8465 | 0 | xmlMalloc(maxatts * sizeof(xmlChar *)); |
8466 | 0 | if (atts == NULL) { |
8467 | 0 | xmlErrMemory(ctxt); |
8468 | 0 | if (attvalue != NULL) |
8469 | 0 | xmlFree(attvalue); |
8470 | 0 | goto failed; |
8471 | 0 | } |
8472 | 0 | ctxt->atts = atts; |
8473 | 0 | ctxt->maxatts = maxatts; |
8474 | 0 | } else if (nbatts + 4 > maxatts) { |
8475 | 0 | const xmlChar **n; |
8476 | |
|
8477 | 0 | maxatts *= 2; |
8478 | 0 | n = (const xmlChar **) xmlRealloc((void *) atts, |
8479 | 0 | maxatts * sizeof(const xmlChar *)); |
8480 | 0 | if (n == NULL) { |
8481 | 0 | xmlErrMemory(ctxt); |
8482 | 0 | if (attvalue != NULL) |
8483 | 0 | xmlFree(attvalue); |
8484 | 0 | goto failed; |
8485 | 0 | } |
8486 | 0 | atts = n; |
8487 | 0 | ctxt->atts = atts; |
8488 | 0 | ctxt->maxatts = maxatts; |
8489 | 0 | } |
8490 | 0 | atts[nbatts++] = attname; |
8491 | 0 | atts[nbatts++] = attvalue; |
8492 | 0 | atts[nbatts] = NULL; |
8493 | 0 | atts[nbatts + 1] = NULL; |
8494 | 0 | } else { |
8495 | 0 | if (attvalue != NULL) |
8496 | 0 | xmlFree(attvalue); |
8497 | 0 | } |
8498 | | |
8499 | 0 | failed: |
8500 | |
|
8501 | 0 | GROW |
8502 | 0 | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
8503 | 0 | break; |
8504 | 0 | if (SKIP_BLANKS == 0) { |
8505 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
8506 | 0 | "attributes construct error\n"); |
8507 | 0 | } |
8508 | 0 | SHRINK; |
8509 | 0 | GROW; |
8510 | 0 | } |
8511 | | |
8512 | | /* |
8513 | | * SAX: Start of Element ! |
8514 | | */ |
8515 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) && |
8516 | 0 | (!ctxt->disableSAX)) { |
8517 | 0 | if (nbatts > 0) |
8518 | 0 | ctxt->sax->startElement(ctxt->userData, name, atts); |
8519 | 0 | else |
8520 | 0 | ctxt->sax->startElement(ctxt->userData, name, NULL); |
8521 | 0 | } |
8522 | |
|
8523 | 0 | if (atts != NULL) { |
8524 | | /* Free only the content strings */ |
8525 | 0 | for (i = 1;i < nbatts;i+=2) |
8526 | 0 | if (atts[i] != NULL) |
8527 | 0 | xmlFree((xmlChar *) atts[i]); |
8528 | 0 | } |
8529 | 0 | return(name); |
8530 | 0 | } |
8531 | | |
8532 | | /** |
8533 | | * xmlParseEndTag1: |
8534 | | * @ctxt: an XML parser context |
8535 | | * @line: line of the start tag |
8536 | | * @nsNr: number of namespaces on the start tag |
8537 | | * |
8538 | | * Parse an end tag. Always consumes '</'. |
8539 | | * |
8540 | | * [42] ETag ::= '</' Name S? '>' |
8541 | | * |
8542 | | * With namespace |
8543 | | * |
8544 | | * [NS 9] ETag ::= '</' QName S? '>' |
8545 | | */ |
8546 | | |
8547 | | static void |
8548 | 0 | xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) { |
8549 | 0 | const xmlChar *name; |
8550 | |
|
8551 | 0 | GROW; |
8552 | 0 | if ((RAW != '<') || (NXT(1) != '/')) { |
8553 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED, |
8554 | 0 | "xmlParseEndTag: '</' not found\n"); |
8555 | 0 | return; |
8556 | 0 | } |
8557 | 0 | SKIP(2); |
8558 | |
|
8559 | 0 | name = xmlParseNameAndCompare(ctxt,ctxt->name); |
8560 | | |
8561 | | /* |
8562 | | * We should definitely be at the ending "S? '>'" part |
8563 | | */ |
8564 | 0 | GROW; |
8565 | 0 | SKIP_BLANKS; |
8566 | 0 | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
8567 | 0 | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
8568 | 0 | } else |
8569 | 0 | NEXT1; |
8570 | | |
8571 | | /* |
8572 | | * [ WFC: Element Type Match ] |
8573 | | * The Name in an element's end-tag must match the element type in the |
8574 | | * start-tag. |
8575 | | * |
8576 | | */ |
8577 | 0 | if (name != (xmlChar*)1) { |
8578 | 0 | if (name == NULL) name = BAD_CAST "unparsable"; |
8579 | 0 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
8580 | 0 | "Opening and ending tag mismatch: %s line %d and %s\n", |
8581 | 0 | ctxt->name, line, name); |
8582 | 0 | } |
8583 | | |
8584 | | /* |
8585 | | * SAX: End of Tag |
8586 | | */ |
8587 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
8588 | 0 | (!ctxt->disableSAX)) |
8589 | 0 | ctxt->sax->endElement(ctxt->userData, ctxt->name); |
8590 | |
|
8591 | 0 | namePop(ctxt); |
8592 | 0 | spacePop(ctxt); |
8593 | 0 | } |
8594 | | |
8595 | | /** |
8596 | | * xmlParseEndTag: |
8597 | | * @ctxt: an XML parser context |
8598 | | * |
8599 | | * DEPRECATED: Internal function, don't use. |
8600 | | * |
8601 | | * parse an end of tag |
8602 | | * |
8603 | | * [42] ETag ::= '</' Name S? '>' |
8604 | | * |
8605 | | * With namespace |
8606 | | * |
8607 | | * [NS 9] ETag ::= '</' QName S? '>' |
8608 | | */ |
8609 | | |
8610 | | void |
8611 | 0 | xmlParseEndTag(xmlParserCtxtPtr ctxt) { |
8612 | 0 | xmlParseEndTag1(ctxt, 0); |
8613 | 0 | } |
8614 | | #endif /* LIBXML_SAX1_ENABLED */ |
8615 | | |
8616 | | /************************************************************************ |
8617 | | * * |
8618 | | * SAX 2 specific operations * |
8619 | | * * |
8620 | | ************************************************************************/ |
8621 | | |
8622 | | /** |
8623 | | * xmlParseQNameHashed: |
8624 | | * @ctxt: an XML parser context |
8625 | | * @prefix: pointer to store the prefix part |
8626 | | * |
8627 | | * parse an XML Namespace QName |
8628 | | * |
8629 | | * [6] QName ::= (Prefix ':')? LocalPart |
8630 | | * [7] Prefix ::= NCName |
8631 | | * [8] LocalPart ::= NCName |
8632 | | * |
8633 | | * Returns the Name parsed or NULL |
8634 | | */ |
8635 | | |
8636 | | static xmlHashedString |
8637 | 1.44M | xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) { |
8638 | 1.44M | xmlHashedString l, p; |
8639 | 1.44M | int start, isNCName = 0; |
8640 | | |
8641 | 1.44M | l.name = NULL; |
8642 | 1.44M | p.name = NULL; |
8643 | | |
8644 | 1.44M | GROW; |
8645 | 1.44M | start = CUR_PTR - BASE_PTR; |
8646 | | |
8647 | 1.44M | l = xmlParseNCName(ctxt); |
8648 | 1.44M | if (l.name != NULL) { |
8649 | 1.11M | isNCName = 1; |
8650 | 1.11M | if (CUR == ':') { |
8651 | 217k | NEXT; |
8652 | 217k | p = l; |
8653 | 217k | l = xmlParseNCName(ctxt); |
8654 | 217k | } |
8655 | 1.11M | } |
8656 | 1.44M | if ((l.name == NULL) || (CUR == ':')) { |
8657 | 336k | xmlChar *tmp; |
8658 | | |
8659 | 336k | l.name = NULL; |
8660 | 336k | p.name = NULL; |
8661 | 336k | if ((isNCName == 0) && (CUR != ':')) |
8662 | 302k | return(l); |
8663 | 33.7k | tmp = xmlParseNmtoken(ctxt); |
8664 | 33.7k | if (tmp != NULL) |
8665 | 25.0k | xmlFree(tmp); |
8666 | 33.7k | l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start, |
8667 | 33.7k | CUR_PTR - (BASE_PTR + start)); |
8668 | 33.7k | if (l.name == NULL) { |
8669 | 0 | xmlErrMemory(ctxt); |
8670 | 0 | return(l); |
8671 | 0 | } |
8672 | 33.7k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8673 | 33.7k | "Failed to parse QName '%s'\n", l.name, NULL, NULL); |
8674 | 33.7k | } |
8675 | | |
8676 | 1.13M | *prefix = p; |
8677 | 1.13M | return(l); |
8678 | 1.44M | } |
8679 | | |
8680 | | /** |
8681 | | * xmlParseQName: |
8682 | | * @ctxt: an XML parser context |
8683 | | * @prefix: pointer to store the prefix part |
8684 | | * |
8685 | | * parse an XML Namespace QName |
8686 | | * |
8687 | | * [6] QName ::= (Prefix ':')? LocalPart |
8688 | | * [7] Prefix ::= NCName |
8689 | | * [8] LocalPart ::= NCName |
8690 | | * |
8691 | | * Returns the Name parsed or NULL |
8692 | | */ |
8693 | | |
8694 | | static const xmlChar * |
8695 | 11.6k | xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) { |
8696 | 11.6k | xmlHashedString n, p; |
8697 | | |
8698 | 11.6k | n = xmlParseQNameHashed(ctxt, &p); |
8699 | 11.6k | if (n.name == NULL) |
8700 | 5.19k | return(NULL); |
8701 | 6.40k | *prefix = p.name; |
8702 | 6.40k | return(n.name); |
8703 | 11.6k | } |
8704 | | |
8705 | | /** |
8706 | | * xmlParseQNameAndCompare: |
8707 | | * @ctxt: an XML parser context |
8708 | | * @name: the localname |
8709 | | * @prefix: the prefix, if any. |
8710 | | * |
8711 | | * parse an XML name and compares for match |
8712 | | * (specialized for endtag parsing) |
8713 | | * |
8714 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
8715 | | * and the name for mismatch |
8716 | | */ |
8717 | | |
8718 | | static const xmlChar * |
8719 | | xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name, |
8720 | 20.6k | xmlChar const *prefix) { |
8721 | 20.6k | const xmlChar *cmp; |
8722 | 20.6k | const xmlChar *in; |
8723 | 20.6k | const xmlChar *ret; |
8724 | 20.6k | const xmlChar *prefix2; |
8725 | | |
8726 | 20.6k | if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name)); |
8727 | | |
8728 | 20.6k | GROW; |
8729 | 20.6k | in = ctxt->input->cur; |
8730 | | |
8731 | 20.6k | cmp = prefix; |
8732 | 43.8k | while (*in != 0 && *in == *cmp) { |
8733 | 23.2k | ++in; |
8734 | 23.2k | ++cmp; |
8735 | 23.2k | } |
8736 | 20.6k | if ((*cmp == 0) && (*in == ':')) { |
8737 | 10.8k | in++; |
8738 | 10.8k | cmp = name; |
8739 | 84.8k | while (*in != 0 && *in == *cmp) { |
8740 | 73.9k | ++in; |
8741 | 73.9k | ++cmp; |
8742 | 73.9k | } |
8743 | 10.8k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
8744 | | /* success */ |
8745 | 9.01k | ctxt->input->col += in - ctxt->input->cur; |
8746 | 9.01k | ctxt->input->cur = in; |
8747 | 9.01k | return((const xmlChar*) 1); |
8748 | 9.01k | } |
8749 | 10.8k | } |
8750 | | /* |
8751 | | * all strings coms from the dictionary, equality can be done directly |
8752 | | */ |
8753 | 11.6k | ret = xmlParseQName (ctxt, &prefix2); |
8754 | 11.6k | if (ret == NULL) |
8755 | 5.19k | return(NULL); |
8756 | 6.40k | if ((ret == name) && (prefix == prefix2)) |
8757 | 655 | return((const xmlChar*) 1); |
8758 | 5.74k | return ret; |
8759 | 6.40k | } |
8760 | | |
8761 | | /** |
8762 | | * xmlParseAttribute2: |
8763 | | * @ctxt: an XML parser context |
8764 | | * @pref: the element prefix |
8765 | | * @elem: the element name |
8766 | | * @prefix: a xmlChar ** used to store the value of the attribute prefix |
8767 | | * @value: a xmlChar ** used to store the value of the attribute |
8768 | | * @len: an int * to save the length of the attribute |
8769 | | * @alloc: an int * to indicate if the attribute was allocated |
8770 | | * |
8771 | | * parse an attribute in the new SAX2 framework. |
8772 | | * |
8773 | | * Returns the attribute name, and the value in *value, . |
8774 | | */ |
8775 | | |
8776 | | static xmlHashedString |
8777 | | xmlParseAttribute2(xmlParserCtxtPtr ctxt, |
8778 | | const xmlChar * pref, const xmlChar * elem, |
8779 | | xmlHashedString * hprefix, xmlChar ** value, |
8780 | | int *len, int *alloc) |
8781 | 762k | { |
8782 | 762k | xmlHashedString hname; |
8783 | 762k | const xmlChar *prefix, *name; |
8784 | 762k | xmlChar *val = NULL, *internal_val = NULL; |
8785 | 762k | int normalize = 0; |
8786 | 762k | int isNamespace; |
8787 | | |
8788 | 762k | *value = NULL; |
8789 | 762k | GROW; |
8790 | 762k | hname = xmlParseQNameHashed(ctxt, hprefix); |
8791 | 762k | if (hname.name == NULL) { |
8792 | 79.4k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8793 | 79.4k | "error parsing attribute name\n"); |
8794 | 79.4k | return(hname); |
8795 | 79.4k | } |
8796 | 683k | name = hname.name; |
8797 | 683k | if (hprefix->name != NULL) |
8798 | 143k | prefix = hprefix->name; |
8799 | 539k | else |
8800 | 539k | prefix = NULL; |
8801 | | |
8802 | | /* |
8803 | | * get the type if needed |
8804 | | */ |
8805 | 683k | if (ctxt->attsSpecial != NULL) { |
8806 | 71.4k | int type; |
8807 | | |
8808 | 71.4k | type = (int) (ptrdiff_t) xmlHashQLookup2(ctxt->attsSpecial, |
8809 | 71.4k | pref, elem, |
8810 | 71.4k | prefix, name); |
8811 | 71.4k | if (type != 0) |
8812 | 8.60k | normalize = 1; |
8813 | 71.4k | } |
8814 | | |
8815 | | /* |
8816 | | * read the value |
8817 | | */ |
8818 | 683k | SKIP_BLANKS; |
8819 | 683k | if (RAW == '=') { |
8820 | 638k | NEXT; |
8821 | 638k | SKIP_BLANKS; |
8822 | 638k | isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) || |
8823 | 638k | (prefix == ctxt->str_xmlns)); |
8824 | 638k | val = xmlParseAttValueInternal(ctxt, len, alloc, normalize, |
8825 | 638k | isNamespace); |
8826 | 638k | if (val == NULL) |
8827 | 14.0k | goto error; |
8828 | 638k | } else { |
8829 | 44.6k | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8830 | 44.6k | "Specification mandates value for attribute %s\n", |
8831 | 44.6k | name); |
8832 | 44.6k | goto error; |
8833 | 44.6k | } |
8834 | | |
8835 | 624k | if (prefix == ctxt->str_xml) { |
8836 | | /* |
8837 | | * Check that xml:lang conforms to the specification |
8838 | | * No more registered as an error, just generate a warning now |
8839 | | * since this was deprecated in XML second edition |
8840 | | */ |
8841 | 19.0k | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) { |
8842 | 0 | internal_val = xmlStrndup(val, *len); |
8843 | 0 | if (internal_val == NULL) |
8844 | 0 | goto mem_error; |
8845 | 0 | if (!xmlCheckLanguageID(internal_val)) { |
8846 | 0 | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8847 | 0 | "Malformed value for xml:lang : %s\n", |
8848 | 0 | internal_val, NULL); |
8849 | 0 | } |
8850 | 0 | } |
8851 | | |
8852 | | /* |
8853 | | * Check that xml:space conforms to the specification |
8854 | | */ |
8855 | 19.0k | if (xmlStrEqual(name, BAD_CAST "space")) { |
8856 | 1.96k | internal_val = xmlStrndup(val, *len); |
8857 | 1.96k | if (internal_val == NULL) |
8858 | 0 | goto mem_error; |
8859 | 1.96k | if (xmlStrEqual(internal_val, BAD_CAST "default")) |
8860 | 480 | *(ctxt->space) = 0; |
8861 | 1.48k | else if (xmlStrEqual(internal_val, BAD_CAST "preserve")) |
8862 | 749 | *(ctxt->space) = 1; |
8863 | 734 | else { |
8864 | 734 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8865 | 734 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8866 | 734 | internal_val, NULL); |
8867 | 734 | } |
8868 | 1.96k | } |
8869 | 19.0k | if (internal_val) { |
8870 | 1.96k | xmlFree(internal_val); |
8871 | 1.96k | } |
8872 | 19.0k | } |
8873 | | |
8874 | 624k | *value = val; |
8875 | 624k | return (hname); |
8876 | | |
8877 | 0 | mem_error: |
8878 | 0 | xmlErrMemory(ctxt); |
8879 | 58.7k | error: |
8880 | 58.7k | if ((val != NULL) && (*alloc != 0)) |
8881 | 0 | xmlFree(val); |
8882 | 58.7k | return(hname); |
8883 | 0 | } |
8884 | | |
8885 | | /** |
8886 | | * xmlAttrHashInsert: |
8887 | | * @ctxt: parser context |
8888 | | * @size: size of the hash table |
8889 | | * @name: attribute name |
8890 | | * @uri: namespace uri |
8891 | | * @hashValue: combined hash value of name and uri |
8892 | | * @aindex: attribute index (this is a multiple of 5) |
8893 | | * |
8894 | | * Inserts a new attribute into the hash table. |
8895 | | * |
8896 | | * Returns INT_MAX if no existing attribute was found, the attribute |
8897 | | * index if an attribute was found, -1 if a memory allocation failed. |
8898 | | */ |
8899 | | static int |
8900 | | xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name, |
8901 | 542k | const xmlChar *uri, unsigned hashValue, int aindex) { |
8902 | 542k | xmlAttrHashBucket *table = ctxt->attrHash; |
8903 | 542k | xmlAttrHashBucket *bucket; |
8904 | 542k | unsigned hindex; |
8905 | | |
8906 | 542k | hindex = hashValue & (size - 1); |
8907 | 542k | bucket = &table[hindex]; |
8908 | | |
8909 | 581k | while (bucket->index >= 0) { |
8910 | 417k | const xmlChar **atts = &ctxt->atts[bucket->index]; |
8911 | | |
8912 | 417k | if (name == atts[0]) { |
8913 | 382k | int nsIndex = (int) (ptrdiff_t) atts[2]; |
8914 | | |
8915 | 382k | if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) : |
8916 | 382k | (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) : |
8917 | 5.96k | (uri == ctxt->nsTab[nsIndex * 2 + 1])) |
8918 | 379k | return(bucket->index); |
8919 | 382k | } |
8920 | | |
8921 | 38.8k | hindex++; |
8922 | 38.8k | bucket++; |
8923 | 38.8k | if (hindex >= size) { |
8924 | 9.78k | hindex = 0; |
8925 | 9.78k | bucket = table; |
8926 | 9.78k | } |
8927 | 38.8k | } |
8928 | | |
8929 | 163k | bucket->index = aindex; |
8930 | | |
8931 | 163k | return(INT_MAX); |
8932 | 542k | } |
8933 | | |
8934 | | /** |
8935 | | * xmlParseStartTag2: |
8936 | | * @ctxt: an XML parser context |
8937 | | * |
8938 | | * Parse a start tag. Always consumes '<'. |
8939 | | * |
8940 | | * This routine is called when running SAX2 parsing |
8941 | | * |
8942 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8943 | | * |
8944 | | * [ WFC: Unique Att Spec ] |
8945 | | * No attribute name may appear more than once in the same start-tag or |
8946 | | * empty-element tag. |
8947 | | * |
8948 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8949 | | * |
8950 | | * [ WFC: Unique Att Spec ] |
8951 | | * No attribute name may appear more than once in the same start-tag or |
8952 | | * empty-element tag. |
8953 | | * |
8954 | | * With namespace: |
8955 | | * |
8956 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8957 | | * |
8958 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8959 | | * |
8960 | | * Returns the element name parsed |
8961 | | */ |
8962 | | |
8963 | | static const xmlChar * |
8964 | | xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref, |
8965 | 665k | const xmlChar **URI, int *nbNsPtr) { |
8966 | 665k | xmlHashedString hlocalname; |
8967 | 665k | xmlHashedString hprefix; |
8968 | 665k | xmlHashedString hattname; |
8969 | 665k | xmlHashedString haprefix; |
8970 | 665k | const xmlChar *localname; |
8971 | 665k | const xmlChar *prefix; |
8972 | 665k | const xmlChar *attname; |
8973 | 665k | const xmlChar *aprefix; |
8974 | 665k | const xmlChar *uri; |
8975 | 665k | xmlChar *attvalue = NULL; |
8976 | 665k | const xmlChar **atts = ctxt->atts; |
8977 | 665k | unsigned attrHashSize = 0; |
8978 | 665k | int maxatts = ctxt->maxatts; |
8979 | 665k | int nratts, nbatts, nbdef; |
8980 | 665k | int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts; |
8981 | 665k | int alloc = 0; |
8982 | | |
8983 | 665k | if (RAW != '<') return(NULL); |
8984 | 665k | NEXT1; |
8985 | | |
8986 | 665k | nbatts = 0; |
8987 | 665k | nratts = 0; |
8988 | 665k | nbdef = 0; |
8989 | 665k | nbNs = 0; |
8990 | 665k | nbTotalDef = 0; |
8991 | 665k | attval = 0; |
8992 | | |
8993 | 665k | if (xmlParserNsStartElement(ctxt->nsdb) < 0) { |
8994 | 0 | xmlErrMemory(ctxt); |
8995 | 0 | return(NULL); |
8996 | 0 | } |
8997 | | |
8998 | 665k | hlocalname = xmlParseQNameHashed(ctxt, &hprefix); |
8999 | 665k | if (hlocalname.name == NULL) { |
9000 | 217k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
9001 | 217k | "StartTag: invalid element name\n"); |
9002 | 217k | return(NULL); |
9003 | 217k | } |
9004 | 448k | localname = hlocalname.name; |
9005 | 448k | prefix = hprefix.name; |
9006 | | |
9007 | | /* |
9008 | | * Now parse the attributes, it ends up with the ending |
9009 | | * |
9010 | | * (S Attribute)* S? |
9011 | | */ |
9012 | 448k | SKIP_BLANKS; |
9013 | 448k | GROW; |
9014 | | |
9015 | | /* |
9016 | | * The ctxt->atts array will be ultimately passed to the SAX callback |
9017 | | * containing five xmlChar pointers for each attribute: |
9018 | | * |
9019 | | * [0] attribute name |
9020 | | * [1] attribute prefix |
9021 | | * [2] namespace URI |
9022 | | * [3] attribute value |
9023 | | * [4] end of attribute value |
9024 | | * |
9025 | | * To save memory, we reuse this array temporarily and store integers |
9026 | | * in these pointer variables. |
9027 | | * |
9028 | | * [0] attribute name |
9029 | | * [1] attribute prefix |
9030 | | * [2] hash value of attribute prefix, and later namespace index |
9031 | | * [3] for non-allocated values: ptrdiff_t offset into input buffer |
9032 | | * [4] for non-allocated values: ptrdiff_t offset into input buffer |
9033 | | * |
9034 | | * The ctxt->attallocs array contains an additional unsigned int for |
9035 | | * each attribute, containing the hash value of the attribute name |
9036 | | * and the alloc flag in bit 31. |
9037 | | */ |
9038 | | |
9039 | 927k | while (((RAW != '>') && |
9040 | 927k | ((RAW != '/') || (NXT(1) != '>')) && |
9041 | 927k | (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) { |
9042 | 762k | int len = -1; |
9043 | | |
9044 | 762k | hattname = xmlParseAttribute2(ctxt, prefix, localname, |
9045 | 762k | &haprefix, &attvalue, &len, |
9046 | 762k | &alloc); |
9047 | 762k | if (hattname.name == NULL) |
9048 | 79.4k | break; |
9049 | 683k | if (attvalue == NULL) |
9050 | 58.7k | goto next_attr; |
9051 | 624k | attname = hattname.name; |
9052 | 624k | aprefix = haprefix.name; |
9053 | 624k | if (len < 0) len = xmlStrlen(attvalue); |
9054 | | |
9055 | 624k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9056 | 77.8k | xmlHashedString huri; |
9057 | 77.8k | xmlURIPtr parsedUri; |
9058 | | |
9059 | 77.8k | huri = xmlDictLookupHashed(ctxt->dict, attvalue, len); |
9060 | 77.8k | uri = huri.name; |
9061 | 77.8k | if (uri == NULL) { |
9062 | 0 | xmlErrMemory(ctxt); |
9063 | 0 | goto next_attr; |
9064 | 0 | } |
9065 | 77.8k | if (*uri != 0) { |
9066 | 74.7k | if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) { |
9067 | 0 | xmlErrMemory(ctxt); |
9068 | 0 | goto next_attr; |
9069 | 0 | } |
9070 | 74.7k | if (parsedUri == NULL) { |
9071 | 59.2k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9072 | 59.2k | "xmlns: '%s' is not a valid URI\n", |
9073 | 59.2k | uri, NULL, NULL); |
9074 | 59.2k | } else { |
9075 | 15.4k | if (parsedUri->scheme == NULL) { |
9076 | 7.58k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9077 | 7.58k | "xmlns: URI %s is not absolute\n", |
9078 | 7.58k | uri, NULL, NULL); |
9079 | 7.58k | } |
9080 | 15.4k | xmlFreeURI(parsedUri); |
9081 | 15.4k | } |
9082 | 74.7k | if (uri == ctxt->str_xml_ns) { |
9083 | 487 | if (attname != ctxt->str_xml) { |
9084 | 487 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9085 | 487 | "xml namespace URI cannot be the default namespace\n", |
9086 | 487 | NULL, NULL, NULL); |
9087 | 487 | } |
9088 | 487 | goto next_attr; |
9089 | 487 | } |
9090 | 74.2k | if ((len == 29) && |
9091 | 74.2k | (xmlStrEqual(uri, |
9092 | 2.14k | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9093 | 1.15k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9094 | 1.15k | "reuse of the xmlns namespace name is forbidden\n", |
9095 | 1.15k | NULL, NULL, NULL); |
9096 | 1.15k | goto next_attr; |
9097 | 1.15k | } |
9098 | 74.2k | } |
9099 | | |
9100 | 76.1k | if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0) |
9101 | 69.6k | nbNs++; |
9102 | 546k | } else if (aprefix == ctxt->str_xmlns) { |
9103 | 54.1k | xmlHashedString huri; |
9104 | 54.1k | xmlURIPtr parsedUri; |
9105 | | |
9106 | 54.1k | huri = xmlDictLookupHashed(ctxt->dict, attvalue, len); |
9107 | 54.1k | uri = huri.name; |
9108 | 54.1k | if (uri == NULL) { |
9109 | 0 | xmlErrMemory(ctxt); |
9110 | 0 | goto next_attr; |
9111 | 0 | } |
9112 | | |
9113 | 54.1k | if (attname == ctxt->str_xml) { |
9114 | 1.88k | if (uri != ctxt->str_xml_ns) { |
9115 | 1.82k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9116 | 1.82k | "xml namespace prefix mapped to wrong URI\n", |
9117 | 1.82k | NULL, NULL, NULL); |
9118 | 1.82k | } |
9119 | | /* |
9120 | | * Do not keep a namespace definition node |
9121 | | */ |
9122 | 1.88k | goto next_attr; |
9123 | 1.88k | } |
9124 | 52.2k | if (uri == ctxt->str_xml_ns) { |
9125 | 272 | if (attname != ctxt->str_xml) { |
9126 | 272 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9127 | 272 | "xml namespace URI mapped to wrong prefix\n", |
9128 | 272 | NULL, NULL, NULL); |
9129 | 272 | } |
9130 | 272 | goto next_attr; |
9131 | 272 | } |
9132 | 51.9k | if (attname == ctxt->str_xmlns) { |
9133 | 303 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9134 | 303 | "redefinition of the xmlns prefix is forbidden\n", |
9135 | 303 | NULL, NULL, NULL); |
9136 | 303 | goto next_attr; |
9137 | 303 | } |
9138 | 51.6k | if ((len == 29) && |
9139 | 51.6k | (xmlStrEqual(uri, |
9140 | 758 | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9141 | 208 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9142 | 208 | "reuse of the xmlns namespace name is forbidden\n", |
9143 | 208 | NULL, NULL, NULL); |
9144 | 208 | goto next_attr; |
9145 | 208 | } |
9146 | 51.4k | if ((uri == NULL) || (uri[0] == 0)) { |
9147 | 1.28k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9148 | 1.28k | "xmlns:%s: Empty XML namespace is not allowed\n", |
9149 | 1.28k | attname, NULL, NULL); |
9150 | 1.28k | goto next_attr; |
9151 | 50.1k | } else { |
9152 | 50.1k | if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) { |
9153 | 0 | xmlErrMemory(ctxt); |
9154 | 0 | goto next_attr; |
9155 | 0 | } |
9156 | 50.1k | if (parsedUri == NULL) { |
9157 | 18.9k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9158 | 18.9k | "xmlns:%s: '%s' is not a valid URI\n", |
9159 | 18.9k | attname, uri, NULL); |
9160 | 31.1k | } else { |
9161 | 31.1k | if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) { |
9162 | 0 | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9163 | 0 | "xmlns:%s: URI %s is not absolute\n", |
9164 | 0 | attname, uri, NULL); |
9165 | 0 | } |
9166 | 31.1k | xmlFreeURI(parsedUri); |
9167 | 31.1k | } |
9168 | 50.1k | } |
9169 | | |
9170 | 50.1k | if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0) |
9171 | 48.0k | nbNs++; |
9172 | 492k | } else { |
9173 | | /* |
9174 | | * Populate attributes array, see above for repurposing |
9175 | | * of xmlChar pointers. |
9176 | | */ |
9177 | 492k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9178 | 3.30k | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9179 | 0 | goto next_attr; |
9180 | 0 | } |
9181 | 3.30k | maxatts = ctxt->maxatts; |
9182 | 3.30k | atts = ctxt->atts; |
9183 | 3.30k | } |
9184 | 492k | ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) | |
9185 | 492k | ((unsigned) alloc << 31); |
9186 | 492k | atts[nbatts++] = attname; |
9187 | 492k | atts[nbatts++] = aprefix; |
9188 | 492k | atts[nbatts++] = (const xmlChar *) (size_t) haprefix.hashValue; |
9189 | 492k | if (alloc) { |
9190 | 28.2k | atts[nbatts++] = attvalue; |
9191 | 28.2k | attvalue += len; |
9192 | 28.2k | atts[nbatts++] = attvalue; |
9193 | 464k | } else { |
9194 | | /* |
9195 | | * attvalue points into the input buffer which can be |
9196 | | * reallocated. Store differences to input->base instead. |
9197 | | * The pointers will be reconstructed later. |
9198 | | */ |
9199 | 464k | atts[nbatts++] = (void *) (attvalue - BASE_PTR); |
9200 | 464k | attvalue += len; |
9201 | 464k | atts[nbatts++] = (void *) (attvalue - BASE_PTR); |
9202 | 464k | } |
9203 | | /* |
9204 | | * tag if some deallocation is needed |
9205 | | */ |
9206 | 492k | if (alloc != 0) attval = 1; |
9207 | 492k | attvalue = NULL; /* moved into atts */ |
9208 | 492k | } |
9209 | | |
9210 | 683k | next_attr: |
9211 | 683k | if ((attvalue != NULL) && (alloc != 0)) { |
9212 | 40.3k | xmlFree(attvalue); |
9213 | 40.3k | attvalue = NULL; |
9214 | 40.3k | } |
9215 | | |
9216 | 683k | GROW |
9217 | 683k | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
9218 | 74.7k | break; |
9219 | 608k | if (SKIP_BLANKS == 0) { |
9220 | 129k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
9221 | 129k | "attributes construct error\n"); |
9222 | 129k | break; |
9223 | 129k | } |
9224 | 479k | GROW; |
9225 | 479k | } |
9226 | | |
9227 | | /* |
9228 | | * Namespaces from default attributes |
9229 | | */ |
9230 | 448k | if (ctxt->attsDefault != NULL) { |
9231 | 84.4k | xmlDefAttrsPtr defaults; |
9232 | | |
9233 | 84.4k | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9234 | 84.4k | if (defaults != NULL) { |
9235 | 214k | for (i = 0; i < defaults->nbAttrs; i++) { |
9236 | 187k | xmlDefAttr *attr = &defaults->attrs[i]; |
9237 | | |
9238 | 187k | attname = attr->name.name; |
9239 | 187k | aprefix = attr->prefix.name; |
9240 | | |
9241 | 187k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9242 | 12.0k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9243 | | |
9244 | 12.0k | if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0) |
9245 | 11.1k | nbNs++; |
9246 | 174k | } else if (aprefix == ctxt->str_xmlns) { |
9247 | 58.3k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9248 | | |
9249 | 58.3k | if (xmlParserNsPush(ctxt, &attr->name, &attr->value, |
9250 | 58.3k | NULL, 1) > 0) |
9251 | 56.4k | nbNs++; |
9252 | 116k | } else { |
9253 | 116k | nbTotalDef += 1; |
9254 | 116k | } |
9255 | 187k | } |
9256 | 27.2k | } |
9257 | 84.4k | } |
9258 | | |
9259 | | /* |
9260 | | * Resolve attribute namespaces |
9261 | | */ |
9262 | 940k | for (i = 0; i < nbatts; i += 5) { |
9263 | 492k | attname = atts[i]; |
9264 | 492k | aprefix = atts[i+1]; |
9265 | | |
9266 | | /* |
9267 | | * The default namespace does not apply to attribute names. |
9268 | | */ |
9269 | 492k | if (aprefix == NULL) { |
9270 | 417k | nsIndex = NS_INDEX_EMPTY; |
9271 | 417k | } else if (aprefix == ctxt->str_xml) { |
9272 | 19.0k | nsIndex = NS_INDEX_XML; |
9273 | 55.9k | } else { |
9274 | 55.9k | haprefix.name = aprefix; |
9275 | 55.9k | haprefix.hashValue = (size_t) atts[i+2]; |
9276 | 55.9k | nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL); |
9277 | | |
9278 | 55.9k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) { |
9279 | 16.7k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9280 | 16.7k | "Namespace prefix %s for %s on %s is not defined\n", |
9281 | 16.7k | aprefix, attname, localname); |
9282 | 16.7k | nsIndex = NS_INDEX_EMPTY; |
9283 | 16.7k | } |
9284 | 55.9k | } |
9285 | | |
9286 | 492k | atts[i+2] = (const xmlChar *) (ptrdiff_t) nsIndex; |
9287 | 492k | } |
9288 | | |
9289 | | /* |
9290 | | * Maximum number of attributes including default attributes. |
9291 | | */ |
9292 | 448k | maxAtts = nratts + nbTotalDef; |
9293 | | |
9294 | | /* |
9295 | | * Verify that attribute names are unique. |
9296 | | */ |
9297 | 448k | if (maxAtts > 1) { |
9298 | 39.1k | attrHashSize = 4; |
9299 | 88.3k | while (attrHashSize / 2 < (unsigned) maxAtts) |
9300 | 49.2k | attrHashSize *= 2; |
9301 | | |
9302 | 39.1k | if (attrHashSize > ctxt->attrHashMax) { |
9303 | 1.35k | xmlAttrHashBucket *tmp; |
9304 | | |
9305 | 1.35k | tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0])); |
9306 | 1.35k | if (tmp == NULL) { |
9307 | 0 | xmlErrMemory(ctxt); |
9308 | 0 | goto done; |
9309 | 0 | } |
9310 | | |
9311 | 1.35k | ctxt->attrHash = tmp; |
9312 | 1.35k | ctxt->attrHashMax = attrHashSize; |
9313 | 1.35k | } |
9314 | | |
9315 | 39.1k | memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0])); |
9316 | | |
9317 | 480k | for (i = 0, j = 0; j < nratts; i += 5, j++) { |
9318 | 441k | const xmlChar *nsuri; |
9319 | 441k | unsigned hashValue, nameHashValue, uriHashValue; |
9320 | 441k | int res; |
9321 | | |
9322 | 441k | attname = atts[i]; |
9323 | 441k | aprefix = atts[i+1]; |
9324 | 441k | nsIndex = (ptrdiff_t) atts[i+2]; |
9325 | | /* Hash values always have bit 31 set, see dict.c */ |
9326 | 441k | nameHashValue = ctxt->attallocs[j] | 0x80000000; |
9327 | | |
9328 | 441k | if (nsIndex == NS_INDEX_EMPTY) { |
9329 | | /* |
9330 | | * Prefix with empty namespace means an undeclared |
9331 | | * prefix which was already reported above. |
9332 | | */ |
9333 | 401k | if (aprefix != NULL) |
9334 | 10.1k | continue; |
9335 | 391k | nsuri = NULL; |
9336 | 391k | uriHashValue = URI_HASH_EMPTY; |
9337 | 391k | } else if (nsIndex == NS_INDEX_XML) { |
9338 | 2.86k | nsuri = ctxt->str_xml_ns; |
9339 | 2.86k | uriHashValue = URI_HASH_XML; |
9340 | 36.1k | } else { |
9341 | 36.1k | nsuri = ctxt->nsTab[nsIndex * 2 + 1]; |
9342 | 36.1k | uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue; |
9343 | 36.1k | } |
9344 | | |
9345 | 430k | hashValue = xmlDictCombineHash(nameHashValue, uriHashValue); |
9346 | 430k | res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri, |
9347 | 430k | hashValue, i); |
9348 | 430k | if (res < 0) |
9349 | 0 | continue; |
9350 | | |
9351 | | /* |
9352 | | * [ WFC: Unique Att Spec ] |
9353 | | * No attribute name may appear more than once in the same |
9354 | | * start-tag or empty-element tag. |
9355 | | * As extended by the Namespace in XML REC. |
9356 | | */ |
9357 | 430k | if (res < INT_MAX) { |
9358 | 363k | if (aprefix == atts[res+1]) { |
9359 | 362k | xmlErrAttributeDup(ctxt, aprefix, attname); |
9360 | 362k | } else { |
9361 | 939 | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9362 | 939 | "Namespaced Attribute %s in '%s' redefined\n", |
9363 | 939 | attname, nsuri, NULL); |
9364 | 939 | } |
9365 | 363k | } |
9366 | 430k | } |
9367 | 39.1k | } |
9368 | | |
9369 | | /* |
9370 | | * Default attributes |
9371 | | */ |
9372 | 448k | if (ctxt->attsDefault != NULL) { |
9373 | 84.4k | xmlDefAttrsPtr defaults; |
9374 | | |
9375 | 84.4k | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9376 | 84.4k | if (defaults != NULL) { |
9377 | 214k | for (i = 0; i < defaults->nbAttrs; i++) { |
9378 | 187k | xmlDefAttr *attr = &defaults->attrs[i]; |
9379 | 187k | const xmlChar *nsuri = NULL; |
9380 | 187k | unsigned hashValue, uriHashValue = 0; |
9381 | 187k | int res; |
9382 | | |
9383 | 187k | attname = attr->name.name; |
9384 | 187k | aprefix = attr->prefix.name; |
9385 | | |
9386 | 187k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) |
9387 | 12.0k | continue; |
9388 | 174k | if (aprefix == ctxt->str_xmlns) |
9389 | 58.3k | continue; |
9390 | | |
9391 | 116k | if (aprefix == NULL) { |
9392 | 42.0k | nsIndex = NS_INDEX_EMPTY; |
9393 | 42.0k | nsuri = NULL; |
9394 | 42.0k | uriHashValue = URI_HASH_EMPTY; |
9395 | 116k | } if (aprefix == ctxt->str_xml) { |
9396 | 7.02k | nsIndex = NS_INDEX_XML; |
9397 | 7.02k | nsuri = ctxt->str_xml_ns; |
9398 | 7.02k | uriHashValue = URI_HASH_XML; |
9399 | 109k | } else if (aprefix != NULL) { |
9400 | 67.5k | nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL); |
9401 | 67.5k | if ((nsIndex == INT_MAX) || |
9402 | 67.5k | (nsIndex < ctxt->nsdb->minNsIndex)) { |
9403 | 63.9k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9404 | 63.9k | "Namespace prefix %s for %s on %s is not " |
9405 | 63.9k | "defined\n", |
9406 | 63.9k | aprefix, attname, localname); |
9407 | 63.9k | nsIndex = NS_INDEX_EMPTY; |
9408 | 63.9k | nsuri = NULL; |
9409 | 63.9k | uriHashValue = URI_HASH_EMPTY; |
9410 | 63.9k | } else { |
9411 | 3.53k | nsuri = ctxt->nsTab[nsIndex * 2 + 1]; |
9412 | 3.53k | uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue; |
9413 | 3.53k | } |
9414 | 67.5k | } |
9415 | | |
9416 | | /* |
9417 | | * Check whether the attribute exists |
9418 | | */ |
9419 | 116k | if (maxAtts > 1) { |
9420 | 111k | hashValue = xmlDictCombineHash(attr->name.hashValue, |
9421 | 111k | uriHashValue); |
9422 | 111k | res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri, |
9423 | 111k | hashValue, nbatts); |
9424 | 111k | if (res < 0) |
9425 | 0 | continue; |
9426 | 111k | if (res < INT_MAX) { |
9427 | 15.4k | if (aprefix == atts[res+1]) |
9428 | 4.64k | continue; |
9429 | 10.7k | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9430 | 10.7k | "Namespaced Attribute %s in '%s' redefined\n", |
9431 | 10.7k | attname, nsuri, NULL); |
9432 | 10.7k | } |
9433 | 111k | } |
9434 | | |
9435 | 112k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9436 | | |
9437 | 112k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9438 | 506 | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9439 | 0 | localname = NULL; |
9440 | 0 | goto done; |
9441 | 0 | } |
9442 | 506 | maxatts = ctxt->maxatts; |
9443 | 506 | atts = ctxt->atts; |
9444 | 506 | } |
9445 | | |
9446 | 112k | atts[nbatts++] = attname; |
9447 | 112k | atts[nbatts++] = aprefix; |
9448 | 112k | atts[nbatts++] = (const xmlChar *) (ptrdiff_t) nsIndex; |
9449 | 112k | atts[nbatts++] = attr->value.name; |
9450 | 112k | atts[nbatts++] = attr->valueEnd; |
9451 | 112k | if ((ctxt->standalone == 1) && (attr->external != 0)) { |
9452 | 0 | xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED, |
9453 | 0 | "standalone: attribute %s on %s defaulted " |
9454 | 0 | "from external subset\n", |
9455 | 0 | attname, localname); |
9456 | 0 | } |
9457 | 112k | nbdef++; |
9458 | 112k | } |
9459 | 27.2k | } |
9460 | 84.4k | } |
9461 | | |
9462 | | /* |
9463 | | * Reconstruct attribute pointers |
9464 | | */ |
9465 | 1.05M | for (i = 0, j = 0; i < nbatts; i += 5, j++) { |
9466 | | /* namespace URI */ |
9467 | 604k | nsIndex = (ptrdiff_t) atts[i+2]; |
9468 | 604k | if (nsIndex == INT_MAX) |
9469 | 535k | atts[i+2] = NULL; |
9470 | 68.4k | else if (nsIndex == INT_MAX - 1) |
9471 | 25.7k | atts[i+2] = ctxt->str_xml_ns; |
9472 | 42.6k | else |
9473 | 42.6k | atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1]; |
9474 | | |
9475 | 604k | if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) { |
9476 | 464k | atts[i+3] = BASE_PTR + (ptrdiff_t) atts[i+3]; /* value */ |
9477 | 464k | atts[i+4] = BASE_PTR + (ptrdiff_t) atts[i+4]; /* valuend */ |
9478 | 464k | } |
9479 | 604k | } |
9480 | | |
9481 | 448k | uri = xmlParserNsLookupUri(ctxt, &hprefix); |
9482 | 448k | if ((prefix != NULL) && (uri == NULL)) { |
9483 | 32.0k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9484 | 32.0k | "Namespace prefix %s on %s is not defined\n", |
9485 | 32.0k | prefix, localname, NULL); |
9486 | 32.0k | } |
9487 | 448k | *pref = prefix; |
9488 | 448k | *URI = uri; |
9489 | | |
9490 | | /* |
9491 | | * SAX callback |
9492 | | */ |
9493 | 448k | if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) && |
9494 | 448k | (!ctxt->disableSAX)) { |
9495 | 43.0k | if (nbNs > 0) |
9496 | 4.08k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri, |
9497 | 4.08k | nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs), |
9498 | 4.08k | nbatts / 5, nbdef, atts); |
9499 | 38.9k | else |
9500 | 38.9k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri, |
9501 | 38.9k | 0, NULL, nbatts / 5, nbdef, atts); |
9502 | 43.0k | } |
9503 | | |
9504 | 448k | done: |
9505 | | /* |
9506 | | * Free allocated attribute values |
9507 | | */ |
9508 | 448k | if (attval != 0) { |
9509 | 344k | for (i = 0, j = 0; j < nratts; i += 5, j++) |
9510 | 317k | if (ctxt->attallocs[j] & 0x80000000) |
9511 | 28.2k | xmlFree((xmlChar *) atts[i+3]); |
9512 | 26.6k | } |
9513 | | |
9514 | 448k | *nbNsPtr = nbNs; |
9515 | 448k | return(localname); |
9516 | 448k | } |
9517 | | |
9518 | | /** |
9519 | | * xmlParseEndTag2: |
9520 | | * @ctxt: an XML parser context |
9521 | | * @line: line of the start tag |
9522 | | * @nsNr: number of namespaces on the start tag |
9523 | | * |
9524 | | * Parse an end tag. Always consumes '</'. |
9525 | | * |
9526 | | * [42] ETag ::= '</' Name S? '>' |
9527 | | * |
9528 | | * With namespace |
9529 | | * |
9530 | | * [NS 9] ETag ::= '</' QName S? '>' |
9531 | | */ |
9532 | | |
9533 | | static void |
9534 | 64.5k | xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) { |
9535 | 64.5k | const xmlChar *name; |
9536 | | |
9537 | 64.5k | GROW; |
9538 | 64.5k | if ((RAW != '<') || (NXT(1) != '/')) { |
9539 | 131 | xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL); |
9540 | 131 | return; |
9541 | 131 | } |
9542 | 64.4k | SKIP(2); |
9543 | | |
9544 | 64.4k | if (tag->prefix == NULL) |
9545 | 43.8k | name = xmlParseNameAndCompare(ctxt, ctxt->name); |
9546 | 20.6k | else |
9547 | 20.6k | name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix); |
9548 | | |
9549 | | /* |
9550 | | * We should definitely be at the ending "S? '>'" part |
9551 | | */ |
9552 | 64.4k | GROW; |
9553 | 64.4k | SKIP_BLANKS; |
9554 | 64.4k | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
9555 | 23.1k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
9556 | 23.1k | } else |
9557 | 41.2k | NEXT1; |
9558 | | |
9559 | | /* |
9560 | | * [ WFC: Element Type Match ] |
9561 | | * The Name in an element's end-tag must match the element type in the |
9562 | | * start-tag. |
9563 | | * |
9564 | | */ |
9565 | 64.4k | if (name != (xmlChar*)1) { |
9566 | 26.2k | if (name == NULL) name = BAD_CAST "unparsable"; |
9567 | 26.2k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
9568 | 26.2k | "Opening and ending tag mismatch: %s line %d and %s\n", |
9569 | 26.2k | ctxt->name, tag->line, name); |
9570 | 26.2k | } |
9571 | | |
9572 | | /* |
9573 | | * SAX: End of Tag |
9574 | | */ |
9575 | 64.4k | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9576 | 64.4k | (!ctxt->disableSAX)) |
9577 | 4.43k | ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix, |
9578 | 4.43k | tag->URI); |
9579 | | |
9580 | 64.4k | spacePop(ctxt); |
9581 | 64.4k | if (tag->nsNr != 0) |
9582 | 4.21k | xmlParserNsPop(ctxt, tag->nsNr); |
9583 | 64.4k | } |
9584 | | |
9585 | | /** |
9586 | | * xmlParseCDSect: |
9587 | | * @ctxt: an XML parser context |
9588 | | * |
9589 | | * DEPRECATED: Internal function, don't use. |
9590 | | * |
9591 | | * Parse escaped pure raw content. Always consumes '<!['. |
9592 | | * |
9593 | | * [18] CDSect ::= CDStart CData CDEnd |
9594 | | * |
9595 | | * [19] CDStart ::= '<![CDATA[' |
9596 | | * |
9597 | | * [20] Data ::= (Char* - (Char* ']]>' Char*)) |
9598 | | * |
9599 | | * [21] CDEnd ::= ']]>' |
9600 | | */ |
9601 | | void |
9602 | 18.4k | xmlParseCDSect(xmlParserCtxtPtr ctxt) { |
9603 | 18.4k | xmlChar *buf = NULL; |
9604 | 18.4k | int len = 0; |
9605 | 18.4k | int size = XML_PARSER_BUFFER_SIZE; |
9606 | 18.4k | int r, rl; |
9607 | 18.4k | int s, sl; |
9608 | 18.4k | int cur, l; |
9609 | 18.4k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9610 | 0 | XML_MAX_HUGE_LENGTH : |
9611 | 18.4k | XML_MAX_TEXT_LENGTH; |
9612 | | |
9613 | 18.4k | if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '[')) |
9614 | 0 | return; |
9615 | 18.4k | SKIP(3); |
9616 | | |
9617 | 18.4k | if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '[')) |
9618 | 0 | return; |
9619 | 18.4k | SKIP(6); |
9620 | | |
9621 | 18.4k | r = xmlCurrentCharRecover(ctxt, &rl); |
9622 | 18.4k | if (!IS_CHAR(r)) { |
9623 | 2.86k | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9624 | 2.86k | goto out; |
9625 | 2.86k | } |
9626 | 15.6k | NEXTL(rl); |
9627 | 15.6k | s = xmlCurrentCharRecover(ctxt, &sl); |
9628 | 15.6k | if (!IS_CHAR(s)) { |
9629 | 1.57k | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9630 | 1.57k | goto out; |
9631 | 1.57k | } |
9632 | 14.0k | NEXTL(sl); |
9633 | 14.0k | cur = xmlCurrentCharRecover(ctxt, &l); |
9634 | 14.0k | buf = xmlMalloc(size); |
9635 | 14.0k | if (buf == NULL) { |
9636 | 0 | xmlErrMemory(ctxt); |
9637 | 0 | goto out; |
9638 | 0 | } |
9639 | 48.0M | while (IS_CHAR(cur) && |
9640 | 48.0M | ((r != ']') || (s != ']') || (cur != '>'))) { |
9641 | 47.9M | if (len + 5 >= size) { |
9642 | 3.50k | xmlChar *tmp; |
9643 | | |
9644 | 3.50k | tmp = (xmlChar *) xmlRealloc(buf, size * 2); |
9645 | 3.50k | if (tmp == NULL) { |
9646 | 0 | xmlErrMemory(ctxt); |
9647 | 0 | goto out; |
9648 | 0 | } |
9649 | 3.50k | buf = tmp; |
9650 | 3.50k | size *= 2; |
9651 | 3.50k | } |
9652 | 47.9M | COPY_BUF(buf, len, r); |
9653 | 47.9M | if (len > maxLength) { |
9654 | 3 | xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9655 | 3 | "CData section too big found\n"); |
9656 | 3 | goto out; |
9657 | 3 | } |
9658 | 47.9M | r = s; |
9659 | 47.9M | rl = sl; |
9660 | 47.9M | s = cur; |
9661 | 47.9M | sl = l; |
9662 | 47.9M | NEXTL(l); |
9663 | 47.9M | cur = xmlCurrentCharRecover(ctxt, &l); |
9664 | 47.9M | } |
9665 | 14.0k | buf[len] = 0; |
9666 | 14.0k | if (cur != '>') { |
9667 | 2.95k | xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9668 | 2.95k | "CData section not finished\n%.50s\n", buf); |
9669 | 2.95k | goto out; |
9670 | 2.95k | } |
9671 | 11.0k | NEXTL(l); |
9672 | | |
9673 | | /* |
9674 | | * OK the buffer is to be consumed as cdata. |
9675 | | */ |
9676 | 11.0k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
9677 | 2.36k | if (ctxt->sax->cdataBlock != NULL) |
9678 | 2.36k | ctxt->sax->cdataBlock(ctxt->userData, buf, len); |
9679 | 0 | else if (ctxt->sax->characters != NULL) |
9680 | 0 | ctxt->sax->characters(ctxt->userData, buf, len); |
9681 | 2.36k | } |
9682 | | |
9683 | 18.4k | out: |
9684 | 18.4k | xmlFree(buf); |
9685 | 18.4k | } |
9686 | | |
9687 | | /** |
9688 | | * xmlParseContentInternal: |
9689 | | * @ctxt: an XML parser context |
9690 | | * |
9691 | | * Parse a content sequence. Stops at EOF or '</'. Leaves checking of |
9692 | | * unexpected EOF to the caller. |
9693 | | */ |
9694 | | |
9695 | | static void |
9696 | 8.39k | xmlParseContentInternal(xmlParserCtxtPtr ctxt) { |
9697 | 8.39k | int oldNameNr = ctxt->nameNr; |
9698 | 8.39k | int oldSpaceNr = ctxt->spaceNr; |
9699 | 8.39k | int oldNodeNr = ctxt->nodeNr; |
9700 | | |
9701 | 8.39k | GROW; |
9702 | 10.5M | while ((ctxt->input->cur < ctxt->input->end) && |
9703 | 10.5M | (PARSER_STOPPED(ctxt) == 0)) { |
9704 | 10.5M | const xmlChar *cur = ctxt->input->cur; |
9705 | | |
9706 | | /* |
9707 | | * First case : a Processing Instruction. |
9708 | | */ |
9709 | 10.5M | if ((*cur == '<') && (cur[1] == '?')) { |
9710 | 22.6k | xmlParsePI(ctxt); |
9711 | 22.6k | } |
9712 | | |
9713 | | /* |
9714 | | * Second case : a CDSection |
9715 | | */ |
9716 | | /* 2.6.0 test was *cur not RAW */ |
9717 | 10.4M | else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9718 | 18.4k | xmlParseCDSect(ctxt); |
9719 | 18.4k | } |
9720 | | |
9721 | | /* |
9722 | | * Third case : a comment |
9723 | | */ |
9724 | 10.4M | else if ((*cur == '<') && (NXT(1) == '!') && |
9725 | 10.4M | (NXT(2) == '-') && (NXT(3) == '-')) { |
9726 | 32.4k | xmlParseComment(ctxt); |
9727 | 32.4k | } |
9728 | | |
9729 | | /* |
9730 | | * Fourth case : a sub-element. |
9731 | | */ |
9732 | 10.4M | else if (*cur == '<') { |
9733 | 720k | if (NXT(1) == '/') { |
9734 | 64.4k | if (ctxt->nameNr <= oldNameNr) |
9735 | 190 | break; |
9736 | 64.2k | xmlParseElementEnd(ctxt); |
9737 | 656k | } else { |
9738 | 656k | xmlParseElementStart(ctxt); |
9739 | 656k | } |
9740 | 720k | } |
9741 | | |
9742 | | /* |
9743 | | * Fifth case : a reference. If if has not been resolved, |
9744 | | * parsing returns it's Name, create the node |
9745 | | */ |
9746 | | |
9747 | 9.71M | else if (*cur == '&') { |
9748 | 307k | xmlParseReference(ctxt); |
9749 | 307k | } |
9750 | | |
9751 | | /* |
9752 | | * Last case, text. Note that References are handled directly. |
9753 | | */ |
9754 | 9.40M | else { |
9755 | 9.40M | xmlParseCharDataInternal(ctxt, 0); |
9756 | 9.40M | } |
9757 | | |
9758 | 10.5M | SHRINK; |
9759 | 10.5M | GROW; |
9760 | 10.5M | } |
9761 | | |
9762 | 8.39k | if ((ctxt->nameNr > oldNameNr) && |
9763 | 8.39k | (ctxt->input->cur >= ctxt->input->end) && |
9764 | 8.39k | (ctxt->wellFormed)) { |
9765 | 333 | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9766 | 333 | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9767 | 333 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9768 | 333 | "Premature end of data in tag %s line %d\n", |
9769 | 333 | name, line, NULL); |
9770 | 333 | } |
9771 | | |
9772 | | /* |
9773 | | * Clean up in error case |
9774 | | */ |
9775 | | |
9776 | 30.3k | while (ctxt->nodeNr > oldNodeNr) |
9777 | 21.9k | nodePop(ctxt); |
9778 | | |
9779 | 138k | while (ctxt->nameNr > oldNameNr) { |
9780 | 130k | xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1]; |
9781 | | |
9782 | 130k | if (tag->nsNr != 0) |
9783 | 34.1k | xmlParserNsPop(ctxt, tag->nsNr); |
9784 | | |
9785 | 130k | namePop(ctxt); |
9786 | 130k | } |
9787 | | |
9788 | 138k | while (ctxt->spaceNr > oldSpaceNr) |
9789 | 130k | spacePop(ctxt); |
9790 | 8.39k | } |
9791 | | |
9792 | | /** |
9793 | | * xmlParseContent: |
9794 | | * @ctxt: an XML parser context |
9795 | | * |
9796 | | * Parse XML element content. This is useful if you're only interested |
9797 | | * in custom SAX callbacks. If you want a node list, use |
9798 | | * xmlCtxtParseContent. |
9799 | | */ |
9800 | | void |
9801 | 0 | xmlParseContent(xmlParserCtxtPtr ctxt) { |
9802 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
9803 | 0 | return; |
9804 | | |
9805 | 0 | xmlCtxtInitializeLate(ctxt); |
9806 | |
|
9807 | 0 | xmlParseContentInternal(ctxt); |
9808 | |
|
9809 | 0 | if (ctxt->input->cur < ctxt->input->end) |
9810 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
9811 | 0 | } |
9812 | | |
9813 | | /** |
9814 | | * xmlParseElement: |
9815 | | * @ctxt: an XML parser context |
9816 | | * |
9817 | | * DEPRECATED: Internal function, don't use. |
9818 | | * |
9819 | | * parse an XML element |
9820 | | * |
9821 | | * [39] element ::= EmptyElemTag | STag content ETag |
9822 | | * |
9823 | | * [ WFC: Element Type Match ] |
9824 | | * The Name in an element's end-tag must match the element type in the |
9825 | | * start-tag. |
9826 | | * |
9827 | | */ |
9828 | | |
9829 | | void |
9830 | 9.76k | xmlParseElement(xmlParserCtxtPtr ctxt) { |
9831 | 9.76k | if (xmlParseElementStart(ctxt) != 0) |
9832 | 2.22k | return; |
9833 | | |
9834 | 7.53k | xmlParseContentInternal(ctxt); |
9835 | | |
9836 | 7.53k | if (ctxt->input->cur >= ctxt->input->end) { |
9837 | 7.23k | if (ctxt->wellFormed) { |
9838 | 616 | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9839 | 616 | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9840 | 616 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9841 | 616 | "Premature end of data in tag %s line %d\n", |
9842 | 616 | name, line, NULL); |
9843 | 616 | } |
9844 | 7.23k | return; |
9845 | 7.23k | } |
9846 | | |
9847 | 303 | xmlParseElementEnd(ctxt); |
9848 | 303 | } |
9849 | | |
9850 | | /** |
9851 | | * xmlParseElementStart: |
9852 | | * @ctxt: an XML parser context |
9853 | | * |
9854 | | * Parse the start of an XML element. Returns -1 in case of error, 0 if an |
9855 | | * opening tag was parsed, 1 if an empty element was parsed. |
9856 | | * |
9857 | | * Always consumes '<'. |
9858 | | */ |
9859 | | static int |
9860 | 665k | xmlParseElementStart(xmlParserCtxtPtr ctxt) { |
9861 | 665k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
9862 | 665k | const xmlChar *name; |
9863 | 665k | const xmlChar *prefix = NULL; |
9864 | 665k | const xmlChar *URI = NULL; |
9865 | 665k | xmlParserNodeInfo node_info; |
9866 | 665k | int line; |
9867 | 665k | xmlNodePtr cur; |
9868 | 665k | int nbNs = 0; |
9869 | | |
9870 | 665k | if (ctxt->nameNr > maxDepth) { |
9871 | 48 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
9872 | 48 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
9873 | 48 | ctxt->nameNr); |
9874 | 48 | xmlHaltParser(ctxt); |
9875 | 48 | return(-1); |
9876 | 48 | } |
9877 | | |
9878 | | /* Capture start position */ |
9879 | 665k | if (ctxt->record_info) { |
9880 | 0 | node_info.begin_pos = ctxt->input->consumed + |
9881 | 0 | (CUR_PTR - ctxt->input->base); |
9882 | 0 | node_info.begin_line = ctxt->input->line; |
9883 | 0 | } |
9884 | | |
9885 | 665k | if (ctxt->spaceNr == 0) |
9886 | 0 | spacePush(ctxt, -1); |
9887 | 665k | else if (*ctxt->space == -2) |
9888 | 0 | spacePush(ctxt, -1); |
9889 | 665k | else |
9890 | 665k | spacePush(ctxt, *ctxt->space); |
9891 | | |
9892 | 665k | line = ctxt->input->line; |
9893 | 665k | #ifdef LIBXML_SAX1_ENABLED |
9894 | 665k | if (ctxt->sax2) |
9895 | 665k | #endif /* LIBXML_SAX1_ENABLED */ |
9896 | 665k | name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs); |
9897 | 0 | #ifdef LIBXML_SAX1_ENABLED |
9898 | 0 | else |
9899 | 0 | name = xmlParseStartTag(ctxt); |
9900 | 665k | #endif /* LIBXML_SAX1_ENABLED */ |
9901 | 665k | if (name == NULL) { |
9902 | 217k | spacePop(ctxt); |
9903 | 217k | return(-1); |
9904 | 217k | } |
9905 | 448k | nameNsPush(ctxt, name, prefix, URI, line, nbNs); |
9906 | 448k | cur = ctxt->node; |
9907 | | |
9908 | 448k | #ifdef LIBXML_VALID_ENABLED |
9909 | | /* |
9910 | | * [ VC: Root Element Type ] |
9911 | | * The Name in the document type declaration must match the element |
9912 | | * type of the root element. |
9913 | | */ |
9914 | 448k | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
9915 | 448k | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
9916 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
9917 | 448k | #endif /* LIBXML_VALID_ENABLED */ |
9918 | | |
9919 | | /* |
9920 | | * Check for an Empty Element. |
9921 | | */ |
9922 | 448k | if ((RAW == '/') && (NXT(1) == '>')) { |
9923 | 23.9k | SKIP(2); |
9924 | 23.9k | if (ctxt->sax2) { |
9925 | 23.9k | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9926 | 23.9k | (!ctxt->disableSAX)) |
9927 | 4.94k | ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI); |
9928 | 23.9k | #ifdef LIBXML_SAX1_ENABLED |
9929 | 23.9k | } else { |
9930 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
9931 | 0 | (!ctxt->disableSAX)) |
9932 | 0 | ctxt->sax->endElement(ctxt->userData, name); |
9933 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
9934 | 0 | } |
9935 | 23.9k | namePop(ctxt); |
9936 | 23.9k | spacePop(ctxt); |
9937 | 23.9k | if (nbNs > 0) |
9938 | 7.24k | xmlParserNsPop(ctxt, nbNs); |
9939 | 23.9k | if (cur != NULL && ctxt->record_info) { |
9940 | 0 | node_info.node = cur; |
9941 | 0 | node_info.end_pos = ctxt->input->consumed + |
9942 | 0 | (CUR_PTR - ctxt->input->base); |
9943 | 0 | node_info.end_line = ctxt->input->line; |
9944 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
9945 | 0 | } |
9946 | 23.9k | return(1); |
9947 | 23.9k | } |
9948 | 424k | if (RAW == '>') { |
9949 | 202k | NEXT1; |
9950 | 202k | if (cur != NULL && ctxt->record_info) { |
9951 | 0 | node_info.node = cur; |
9952 | 0 | node_info.end_pos = 0; |
9953 | 0 | node_info.end_line = 0; |
9954 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
9955 | 0 | } |
9956 | 222k | } else { |
9957 | 222k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED, |
9958 | 222k | "Couldn't find end of Start Tag %s line %d\n", |
9959 | 222k | name, line, NULL); |
9960 | | |
9961 | | /* |
9962 | | * end of parsing of this node. |
9963 | | */ |
9964 | 222k | nodePop(ctxt); |
9965 | 222k | namePop(ctxt); |
9966 | 222k | spacePop(ctxt); |
9967 | 222k | if (nbNs > 0) |
9968 | 78.1k | xmlParserNsPop(ctxt, nbNs); |
9969 | 222k | return(-1); |
9970 | 222k | } |
9971 | | |
9972 | 202k | return(0); |
9973 | 424k | } |
9974 | | |
9975 | | /** |
9976 | | * xmlParseElementEnd: |
9977 | | * @ctxt: an XML parser context |
9978 | | * |
9979 | | * Parse the end of an XML element. Always consumes '</'. |
9980 | | */ |
9981 | | static void |
9982 | 64.5k | xmlParseElementEnd(xmlParserCtxtPtr ctxt) { |
9983 | 64.5k | xmlNodePtr cur = ctxt->node; |
9984 | | |
9985 | 64.5k | if (ctxt->nameNr <= 0) { |
9986 | 0 | if ((RAW == '<') && (NXT(1) == '/')) |
9987 | 0 | SKIP(2); |
9988 | 0 | return; |
9989 | 0 | } |
9990 | | |
9991 | | /* |
9992 | | * parse the end of tag: '</' should be here. |
9993 | | */ |
9994 | 64.5k | if (ctxt->sax2) { |
9995 | 64.5k | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
9996 | 64.5k | namePop(ctxt); |
9997 | 64.5k | } |
9998 | 0 | #ifdef LIBXML_SAX1_ENABLED |
9999 | 0 | else |
10000 | 0 | xmlParseEndTag1(ctxt, 0); |
10001 | 64.5k | #endif /* LIBXML_SAX1_ENABLED */ |
10002 | | |
10003 | | /* |
10004 | | * Capture end position |
10005 | | */ |
10006 | 64.5k | if (cur != NULL && ctxt->record_info) { |
10007 | 0 | xmlParserNodeInfoPtr node_info; |
10008 | |
|
10009 | 0 | node_info = (xmlParserNodeInfoPtr) xmlParserFindNodeInfo(ctxt, cur); |
10010 | 0 | if (node_info != NULL) { |
10011 | 0 | node_info->end_pos = ctxt->input->consumed + |
10012 | 0 | (CUR_PTR - ctxt->input->base); |
10013 | 0 | node_info->end_line = ctxt->input->line; |
10014 | 0 | } |
10015 | 0 | } |
10016 | 64.5k | } |
10017 | | |
10018 | | /** |
10019 | | * xmlParseVersionNum: |
10020 | | * @ctxt: an XML parser context |
10021 | | * |
10022 | | * DEPRECATED: Internal function, don't use. |
10023 | | * |
10024 | | * parse the XML version value. |
10025 | | * |
10026 | | * [26] VersionNum ::= '1.' [0-9]+ |
10027 | | * |
10028 | | * In practice allow [0-9].[0-9]+ at that level |
10029 | | * |
10030 | | * Returns the string giving the XML version number, or NULL |
10031 | | */ |
10032 | | xmlChar * |
10033 | 2.03k | xmlParseVersionNum(xmlParserCtxtPtr ctxt) { |
10034 | 2.03k | xmlChar *buf = NULL; |
10035 | 2.03k | int len = 0; |
10036 | 2.03k | int size = 10; |
10037 | 2.03k | xmlChar cur; |
10038 | | |
10039 | 2.03k | buf = xmlMalloc(size); |
10040 | 2.03k | if (buf == NULL) { |
10041 | 0 | xmlErrMemory(ctxt); |
10042 | 0 | return(NULL); |
10043 | 0 | } |
10044 | 2.03k | cur = CUR; |
10045 | 2.03k | if (!((cur >= '0') && (cur <= '9'))) { |
10046 | 82 | xmlFree(buf); |
10047 | 82 | return(NULL); |
10048 | 82 | } |
10049 | 1.95k | buf[len++] = cur; |
10050 | 1.95k | NEXT; |
10051 | 1.95k | cur=CUR; |
10052 | 1.95k | if (cur != '.') { |
10053 | 23 | xmlFree(buf); |
10054 | 23 | return(NULL); |
10055 | 23 | } |
10056 | 1.93k | buf[len++] = cur; |
10057 | 1.93k | NEXT; |
10058 | 1.93k | cur=CUR; |
10059 | 398k | while ((cur >= '0') && (cur <= '9')) { |
10060 | 396k | if (len + 1 >= size) { |
10061 | 149 | xmlChar *tmp; |
10062 | | |
10063 | 149 | size *= 2; |
10064 | 149 | tmp = (xmlChar *) xmlRealloc(buf, size); |
10065 | 149 | if (tmp == NULL) { |
10066 | 0 | xmlFree(buf); |
10067 | 0 | xmlErrMemory(ctxt); |
10068 | 0 | return(NULL); |
10069 | 0 | } |
10070 | 149 | buf = tmp; |
10071 | 149 | } |
10072 | 396k | buf[len++] = cur; |
10073 | 396k | NEXT; |
10074 | 396k | cur=CUR; |
10075 | 396k | } |
10076 | 1.93k | buf[len] = 0; |
10077 | 1.93k | return(buf); |
10078 | 1.93k | } |
10079 | | |
10080 | | /** |
10081 | | * xmlParseVersionInfo: |
10082 | | * @ctxt: an XML parser context |
10083 | | * |
10084 | | * DEPRECATED: Internal function, don't use. |
10085 | | * |
10086 | | * parse the XML version. |
10087 | | * |
10088 | | * [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ") |
10089 | | * |
10090 | | * [25] Eq ::= S? '=' S? |
10091 | | * |
10092 | | * Returns the version string, e.g. "1.0" |
10093 | | */ |
10094 | | |
10095 | | xmlChar * |
10096 | 3.44k | xmlParseVersionInfo(xmlParserCtxtPtr ctxt) { |
10097 | 3.44k | xmlChar *version = NULL; |
10098 | | |
10099 | 3.44k | if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) { |
10100 | 2.12k | SKIP(7); |
10101 | 2.12k | SKIP_BLANKS; |
10102 | 2.12k | if (RAW != '=') { |
10103 | 25 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10104 | 25 | return(NULL); |
10105 | 25 | } |
10106 | 2.09k | NEXT; |
10107 | 2.09k | SKIP_BLANKS; |
10108 | 2.09k | if (RAW == '"') { |
10109 | 1.77k | NEXT; |
10110 | 1.77k | version = xmlParseVersionNum(ctxt); |
10111 | 1.77k | if (RAW != '"') { |
10112 | 88 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10113 | 88 | } else |
10114 | 1.68k | NEXT; |
10115 | 1.77k | } else if (RAW == '\''){ |
10116 | 263 | NEXT; |
10117 | 263 | version = xmlParseVersionNum(ctxt); |
10118 | 263 | if (RAW != '\'') { |
10119 | 21 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10120 | 21 | } else |
10121 | 242 | NEXT; |
10122 | 263 | } else { |
10123 | 62 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10124 | 62 | } |
10125 | 2.09k | } |
10126 | 3.41k | return(version); |
10127 | 3.44k | } |
10128 | | |
10129 | | /** |
10130 | | * xmlParseEncName: |
10131 | | * @ctxt: an XML parser context |
10132 | | * |
10133 | | * DEPRECATED: Internal function, don't use. |
10134 | | * |
10135 | | * parse the XML encoding name |
10136 | | * |
10137 | | * [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* |
10138 | | * |
10139 | | * Returns the encoding name value or NULL |
10140 | | */ |
10141 | | xmlChar * |
10142 | 2.40k | xmlParseEncName(xmlParserCtxtPtr ctxt) { |
10143 | 2.40k | xmlChar *buf = NULL; |
10144 | 2.40k | int len = 0; |
10145 | 2.40k | int size = 10; |
10146 | 2.40k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
10147 | 0 | XML_MAX_TEXT_LENGTH : |
10148 | 2.40k | XML_MAX_NAME_LENGTH; |
10149 | 2.40k | xmlChar cur; |
10150 | | |
10151 | 2.40k | cur = CUR; |
10152 | 2.40k | if (((cur >= 'a') && (cur <= 'z')) || |
10153 | 2.40k | ((cur >= 'A') && (cur <= 'Z'))) { |
10154 | 2.38k | buf = xmlMalloc(size); |
10155 | 2.38k | if (buf == NULL) { |
10156 | 0 | xmlErrMemory(ctxt); |
10157 | 0 | return(NULL); |
10158 | 0 | } |
10159 | | |
10160 | 2.38k | buf[len++] = cur; |
10161 | 2.38k | NEXT; |
10162 | 2.38k | cur = CUR; |
10163 | 68.5k | while (((cur >= 'a') && (cur <= 'z')) || |
10164 | 68.5k | ((cur >= 'A') && (cur <= 'Z')) || |
10165 | 68.5k | ((cur >= '0') && (cur <= '9')) || |
10166 | 68.5k | (cur == '.') || (cur == '_') || |
10167 | 68.5k | (cur == '-')) { |
10168 | 66.1k | if (len + 1 >= size) { |
10169 | 431 | xmlChar *tmp; |
10170 | | |
10171 | 431 | size *= 2; |
10172 | 431 | tmp = (xmlChar *) xmlRealloc(buf, size); |
10173 | 431 | if (tmp == NULL) { |
10174 | 0 | xmlErrMemory(ctxt); |
10175 | 0 | xmlFree(buf); |
10176 | 0 | return(NULL); |
10177 | 0 | } |
10178 | 431 | buf = tmp; |
10179 | 431 | } |
10180 | 66.1k | buf[len++] = cur; |
10181 | 66.1k | if (len > maxLength) { |
10182 | 1 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName"); |
10183 | 1 | xmlFree(buf); |
10184 | 1 | return(NULL); |
10185 | 1 | } |
10186 | 66.1k | NEXT; |
10187 | 66.1k | cur = CUR; |
10188 | 66.1k | } |
10189 | 2.38k | buf[len] = 0; |
10190 | 2.38k | } else { |
10191 | 17 | xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL); |
10192 | 17 | } |
10193 | 2.40k | return(buf); |
10194 | 2.40k | } |
10195 | | |
10196 | | /** |
10197 | | * xmlParseEncodingDecl: |
10198 | | * @ctxt: an XML parser context |
10199 | | * |
10200 | | * DEPRECATED: Internal function, don't use. |
10201 | | * |
10202 | | * parse the XML encoding declaration |
10203 | | * |
10204 | | * [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'") |
10205 | | * |
10206 | | * this setups the conversion filters. |
10207 | | * |
10208 | | * Returns the encoding value or NULL |
10209 | | */ |
10210 | | |
10211 | | const xmlChar * |
10212 | 3.42k | xmlParseEncodingDecl(xmlParserCtxtPtr ctxt) { |
10213 | 3.42k | xmlChar *encoding = NULL; |
10214 | | |
10215 | 3.42k | SKIP_BLANKS; |
10216 | 3.42k | if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0) |
10217 | 916 | return(NULL); |
10218 | | |
10219 | 2.50k | SKIP(8); |
10220 | 2.50k | SKIP_BLANKS; |
10221 | 2.50k | if (RAW != '=') { |
10222 | 47 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10223 | 47 | return(NULL); |
10224 | 47 | } |
10225 | 2.45k | NEXT; |
10226 | 2.45k | SKIP_BLANKS; |
10227 | 2.45k | if (RAW == '"') { |
10228 | 2.22k | NEXT; |
10229 | 2.22k | encoding = xmlParseEncName(ctxt); |
10230 | 2.22k | if (RAW != '"') { |
10231 | 144 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10232 | 144 | xmlFree((xmlChar *) encoding); |
10233 | 144 | return(NULL); |
10234 | 144 | } else |
10235 | 2.08k | NEXT; |
10236 | 2.22k | } else if (RAW == '\''){ |
10237 | 178 | NEXT; |
10238 | 178 | encoding = xmlParseEncName(ctxt); |
10239 | 178 | if (RAW != '\'') { |
10240 | 20 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10241 | 20 | xmlFree((xmlChar *) encoding); |
10242 | 20 | return(NULL); |
10243 | 20 | } else |
10244 | 158 | NEXT; |
10245 | 178 | } else { |
10246 | 54 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10247 | 54 | } |
10248 | | |
10249 | 2.29k | if (encoding == NULL) |
10250 | 54 | return(NULL); |
10251 | | |
10252 | 2.24k | xmlSetDeclaredEncoding(ctxt, encoding); |
10253 | | |
10254 | 2.24k | return(ctxt->encoding); |
10255 | 2.29k | } |
10256 | | |
10257 | | /** |
10258 | | * xmlParseSDDecl: |
10259 | | * @ctxt: an XML parser context |
10260 | | * |
10261 | | * DEPRECATED: Internal function, don't use. |
10262 | | * |
10263 | | * parse the XML standalone declaration |
10264 | | * |
10265 | | * [32] SDDecl ::= S 'standalone' Eq |
10266 | | * (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"')) |
10267 | | * |
10268 | | * [ VC: Standalone Document Declaration ] |
10269 | | * TODO The standalone document declaration must have the value "no" |
10270 | | * if any external markup declarations contain declarations of: |
10271 | | * - attributes with default values, if elements to which these |
10272 | | * attributes apply appear in the document without specifications |
10273 | | * of values for these attributes, or |
10274 | | * - entities (other than amp, lt, gt, apos, quot), if references |
10275 | | * to those entities appear in the document, or |
10276 | | * - attributes with values subject to normalization, where the |
10277 | | * attribute appears in the document with a value which will change |
10278 | | * as a result of normalization, or |
10279 | | * - element types with element content, if white space occurs directly |
10280 | | * within any instance of those types. |
10281 | | * |
10282 | | * Returns: |
10283 | | * 1 if standalone="yes" |
10284 | | * 0 if standalone="no" |
10285 | | * -2 if standalone attribute is missing or invalid |
10286 | | * (A standalone value of -2 means that the XML declaration was found, |
10287 | | * but no value was specified for the standalone attribute). |
10288 | | */ |
10289 | | |
10290 | | int |
10291 | 1.81k | xmlParseSDDecl(xmlParserCtxtPtr ctxt) { |
10292 | 1.81k | int standalone = -2; |
10293 | | |
10294 | 1.81k | SKIP_BLANKS; |
10295 | 1.81k | if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) { |
10296 | 311 | SKIP(10); |
10297 | 311 | SKIP_BLANKS; |
10298 | 311 | if (RAW != '=') { |
10299 | 21 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10300 | 21 | return(standalone); |
10301 | 21 | } |
10302 | 290 | NEXT; |
10303 | 290 | SKIP_BLANKS; |
10304 | 290 | if (RAW == '\''){ |
10305 | 146 | NEXT; |
10306 | 146 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10307 | 126 | standalone = 0; |
10308 | 126 | SKIP(2); |
10309 | 126 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10310 | 20 | (NXT(2) == 's')) { |
10311 | 7 | standalone = 1; |
10312 | 7 | SKIP(3); |
10313 | 13 | } else { |
10314 | 13 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10315 | 13 | } |
10316 | 146 | if (RAW != '\'') { |
10317 | 23 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10318 | 23 | } else |
10319 | 123 | NEXT; |
10320 | 146 | } else if (RAW == '"'){ |
10321 | 137 | NEXT; |
10322 | 137 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10323 | 69 | standalone = 0; |
10324 | 69 | SKIP(2); |
10325 | 69 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10326 | 68 | (NXT(2) == 's')) { |
10327 | 55 | standalone = 1; |
10328 | 55 | SKIP(3); |
10329 | 55 | } else { |
10330 | 13 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10331 | 13 | } |
10332 | 137 | if (RAW != '"') { |
10333 | 19 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10334 | 19 | } else |
10335 | 118 | NEXT; |
10336 | 137 | } else { |
10337 | 7 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10338 | 7 | } |
10339 | 290 | } |
10340 | 1.79k | return(standalone); |
10341 | 1.81k | } |
10342 | | |
10343 | | /** |
10344 | | * xmlParseXMLDecl: |
10345 | | * @ctxt: an XML parser context |
10346 | | * |
10347 | | * DEPRECATED: Internal function, don't use. |
10348 | | * |
10349 | | * parse an XML declaration header |
10350 | | * |
10351 | | * [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' |
10352 | | */ |
10353 | | |
10354 | | void |
10355 | 3.44k | xmlParseXMLDecl(xmlParserCtxtPtr ctxt) { |
10356 | 3.44k | xmlChar *version; |
10357 | | |
10358 | | /* |
10359 | | * This value for standalone indicates that the document has an |
10360 | | * XML declaration but it does not have a standalone attribute. |
10361 | | * It will be overwritten later if a standalone attribute is found. |
10362 | | */ |
10363 | | |
10364 | 3.44k | ctxt->standalone = -2; |
10365 | | |
10366 | | /* |
10367 | | * We know that '<?xml' is here. |
10368 | | */ |
10369 | 3.44k | SKIP(5); |
10370 | | |
10371 | 3.44k | if (!IS_BLANK_CH(RAW)) { |
10372 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
10373 | 0 | "Blank needed after '<?xml'\n"); |
10374 | 0 | } |
10375 | 3.44k | SKIP_BLANKS; |
10376 | | |
10377 | | /* |
10378 | | * We must have the VersionInfo here. |
10379 | | */ |
10380 | 3.44k | version = xmlParseVersionInfo(ctxt); |
10381 | 3.44k | if (version == NULL) { |
10382 | 1.51k | xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL); |
10383 | 1.93k | } else { |
10384 | 1.93k | if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) { |
10385 | | /* |
10386 | | * Changed here for XML-1.0 5th edition |
10387 | | */ |
10388 | 1.28k | if (ctxt->options & XML_PARSE_OLD10) { |
10389 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10390 | 0 | "Unsupported version '%s'\n", |
10391 | 0 | version); |
10392 | 1.28k | } else { |
10393 | 1.28k | if ((version[0] == '1') && ((version[1] == '.'))) { |
10394 | 1.12k | xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION, |
10395 | 1.12k | "Unsupported version '%s'\n", |
10396 | 1.12k | version, NULL); |
10397 | 1.12k | } else { |
10398 | 158 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10399 | 158 | "Unsupported version '%s'\n", |
10400 | 158 | version); |
10401 | 158 | } |
10402 | 1.28k | } |
10403 | 1.28k | } |
10404 | 1.93k | if (ctxt->version != NULL) |
10405 | 0 | xmlFree((void *) ctxt->version); |
10406 | 1.93k | ctxt->version = version; |
10407 | 1.93k | } |
10408 | | |
10409 | | /* |
10410 | | * We may have the encoding declaration |
10411 | | */ |
10412 | 3.44k | if (!IS_BLANK_CH(RAW)) { |
10413 | 1.51k | if ((RAW == '?') && (NXT(1) == '>')) { |
10414 | 23 | SKIP(2); |
10415 | 23 | return; |
10416 | 23 | } |
10417 | 1.48k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10418 | 1.48k | } |
10419 | 3.42k | xmlParseEncodingDecl(ctxt); |
10420 | | |
10421 | | /* |
10422 | | * We may have the standalone status. |
10423 | | */ |
10424 | 3.42k | if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) { |
10425 | 1.91k | if ((RAW == '?') && (NXT(1) == '>')) { |
10426 | 1.60k | SKIP(2); |
10427 | 1.60k | return; |
10428 | 1.60k | } |
10429 | 311 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10430 | 311 | } |
10431 | | |
10432 | | /* |
10433 | | * We can grow the input buffer freely at that point |
10434 | | */ |
10435 | 1.81k | GROW; |
10436 | | |
10437 | 1.81k | SKIP_BLANKS; |
10438 | 1.81k | ctxt->standalone = xmlParseSDDecl(ctxt); |
10439 | | |
10440 | 1.81k | SKIP_BLANKS; |
10441 | 1.81k | if ((RAW == '?') && (NXT(1) == '>')) { |
10442 | 258 | SKIP(2); |
10443 | 1.55k | } else if (RAW == '>') { |
10444 | | /* Deprecated old WD ... */ |
10445 | 173 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10446 | 173 | NEXT; |
10447 | 1.38k | } else { |
10448 | 1.38k | int c; |
10449 | | |
10450 | 1.38k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10451 | 1.17M | while ((PARSER_STOPPED(ctxt) == 0) && |
10452 | 1.17M | ((c = CUR) != 0)) { |
10453 | 1.17M | NEXT; |
10454 | 1.17M | if (c == '>') |
10455 | 691 | break; |
10456 | 1.17M | } |
10457 | 1.38k | } |
10458 | 1.81k | } |
10459 | | |
10460 | | /** |
10461 | | * xmlCtxtGetVersion: |
10462 | | * ctxt: parser context |
10463 | | * |
10464 | | * Available since 2.14.0. |
10465 | | * |
10466 | | * Returns the version from the XML declaration. |
10467 | | */ |
10468 | | const xmlChar * |
10469 | 0 | xmlCtxtGetVersion(xmlParserCtxtPtr ctxt) { |
10470 | 0 | if (ctxt == NULL) |
10471 | 0 | return(NULL); |
10472 | | |
10473 | 0 | return(ctxt->version); |
10474 | 0 | } |
10475 | | |
10476 | | /** |
10477 | | * xmlCtxtGetStandalone: |
10478 | | * ctxt: parser context |
10479 | | * |
10480 | | * Available since 2.14.0. |
10481 | | * |
10482 | | * Returns the value from the standalone document declaration. |
10483 | | */ |
10484 | | int |
10485 | 0 | xmlCtxtGetStandalone(xmlParserCtxtPtr ctxt) { |
10486 | 0 | if (ctxt == NULL) |
10487 | 0 | return(0); |
10488 | | |
10489 | 0 | return(ctxt->standalone); |
10490 | 0 | } |
10491 | | |
10492 | | /** |
10493 | | * xmlParseMisc: |
10494 | | * @ctxt: an XML parser context |
10495 | | * |
10496 | | * DEPRECATED: Internal function, don't use. |
10497 | | * |
10498 | | * parse an XML Misc* optional field. |
10499 | | * |
10500 | | * [27] Misc ::= Comment | PI | S |
10501 | | */ |
10502 | | |
10503 | | void |
10504 | 33.7k | xmlParseMisc(xmlParserCtxtPtr ctxt) { |
10505 | 42.3k | while (PARSER_STOPPED(ctxt) == 0) { |
10506 | 41.4k | SKIP_BLANKS; |
10507 | 41.4k | GROW; |
10508 | 41.4k | if ((RAW == '<') && (NXT(1) == '?')) { |
10509 | 5.81k | xmlParsePI(ctxt); |
10510 | 35.5k | } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) { |
10511 | 2.83k | xmlParseComment(ctxt); |
10512 | 32.7k | } else { |
10513 | 32.7k | break; |
10514 | 32.7k | } |
10515 | 41.4k | } |
10516 | 33.7k | } |
10517 | | |
10518 | | static void |
10519 | 16.9k | xmlFinishDocument(xmlParserCtxtPtr ctxt) { |
10520 | 16.9k | xmlDocPtr doc; |
10521 | | |
10522 | | /* |
10523 | | * SAX: end of the document processing. |
10524 | | */ |
10525 | 16.9k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10526 | 16.9k | ctxt->sax->endDocument(ctxt->userData); |
10527 | | |
10528 | 16.9k | doc = ctxt->myDoc; |
10529 | 16.9k | if (doc != NULL) { |
10530 | 15.5k | if (ctxt->wellFormed) { |
10531 | 36 | doc->properties |= XML_DOC_WELLFORMED; |
10532 | 36 | if (ctxt->valid) |
10533 | 35 | doc->properties |= XML_DOC_DTDVALID; |
10534 | 36 | if (ctxt->nsWellFormed) |
10535 | 32 | doc->properties |= XML_DOC_NSVALID; |
10536 | 36 | } |
10537 | | |
10538 | 15.5k | if (ctxt->options & XML_PARSE_OLD10) |
10539 | 0 | doc->properties |= XML_DOC_OLD10; |
10540 | | |
10541 | | /* |
10542 | | * Remove locally kept entity definitions if the tree was not built |
10543 | | */ |
10544 | 15.5k | if (xmlStrEqual(doc->version, SAX_COMPAT_MODE)) { |
10545 | 469 | xmlFreeDoc(doc); |
10546 | 469 | ctxt->myDoc = NULL; |
10547 | 469 | } |
10548 | 15.5k | } |
10549 | 16.9k | } |
10550 | | |
10551 | | /** |
10552 | | * xmlParseDocument: |
10553 | | * @ctxt: an XML parser context |
10554 | | * |
10555 | | * Parse an XML document and invoke the SAX handlers. This is useful |
10556 | | * if you're only interested in custom SAX callbacks. If you want a |
10557 | | * document tree, use xmlCtxtParseDocument. |
10558 | | * |
10559 | | * Returns 0, -1 in case of error. |
10560 | | */ |
10561 | | |
10562 | | int |
10563 | 20.2k | xmlParseDocument(xmlParserCtxtPtr ctxt) { |
10564 | 20.2k | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10565 | 0 | return(-1); |
10566 | | |
10567 | 20.2k | GROW; |
10568 | | |
10569 | | /* |
10570 | | * SAX: detecting the level. |
10571 | | */ |
10572 | 20.2k | xmlCtxtInitializeLate(ctxt); |
10573 | | |
10574 | 20.2k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
10575 | 20.2k | ctxt->sax->setDocumentLocator(ctxt->userData, |
10576 | 20.2k | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
10577 | 20.2k | } |
10578 | | |
10579 | 20.2k | xmlDetectEncoding(ctxt); |
10580 | | |
10581 | 20.2k | if (CUR == 0) { |
10582 | 3.25k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10583 | 3.25k | return(-1); |
10584 | 3.25k | } |
10585 | | |
10586 | 16.9k | GROW; |
10587 | 16.9k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10588 | | |
10589 | | /* |
10590 | | * Note that we will switch encoding on the fly. |
10591 | | */ |
10592 | 3.44k | xmlParseXMLDecl(ctxt); |
10593 | 3.44k | SKIP_BLANKS; |
10594 | 13.5k | } else { |
10595 | 13.5k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10596 | 13.5k | if (ctxt->version == NULL) { |
10597 | 0 | xmlErrMemory(ctxt); |
10598 | 0 | return(-1); |
10599 | 0 | } |
10600 | 13.5k | } |
10601 | 16.9k | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10602 | 15.0k | ctxt->sax->startDocument(ctxt->userData); |
10603 | 16.9k | if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) && |
10604 | 16.9k | (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) { |
10605 | 0 | ctxt->myDoc->compression = ctxt->input->buf->compressed; |
10606 | 0 | } |
10607 | | |
10608 | | /* |
10609 | | * The Misc part of the Prolog |
10610 | | */ |
10611 | 16.9k | xmlParseMisc(ctxt); |
10612 | | |
10613 | | /* |
10614 | | * Then possibly doc type declaration(s) and more Misc |
10615 | | * (doctypedecl Misc*)? |
10616 | | */ |
10617 | 16.9k | GROW; |
10618 | 16.9k | if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) { |
10619 | | |
10620 | 6.97k | ctxt->inSubset = 1; |
10621 | 6.97k | xmlParseDocTypeDecl(ctxt); |
10622 | 6.97k | if (RAW == '[') { |
10623 | 6.59k | xmlParseInternalSubset(ctxt); |
10624 | 6.59k | } |
10625 | | |
10626 | | /* |
10627 | | * Create and update the external subset. |
10628 | | */ |
10629 | 6.97k | ctxt->inSubset = 2; |
10630 | 6.97k | if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) && |
10631 | 6.97k | (!ctxt->disableSAX)) |
10632 | 1.72k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
10633 | 1.72k | ctxt->extSubSystem, ctxt->extSubURI); |
10634 | 6.97k | ctxt->inSubset = 0; |
10635 | | |
10636 | 6.97k | xmlCleanSpecialAttr(ctxt); |
10637 | | |
10638 | 6.97k | xmlParseMisc(ctxt); |
10639 | 6.97k | } |
10640 | | |
10641 | | /* |
10642 | | * Time to start parsing the tree itself |
10643 | | */ |
10644 | 16.9k | GROW; |
10645 | 16.9k | if (RAW != '<') { |
10646 | 7.20k | if (ctxt->wellFormed) |
10647 | 630 | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
10648 | 630 | "Start tag expected, '<' not found\n"); |
10649 | 9.76k | } else { |
10650 | 9.76k | xmlParseElement(ctxt); |
10651 | | |
10652 | | /* |
10653 | | * The Misc part at the end |
10654 | | */ |
10655 | 9.76k | xmlParseMisc(ctxt); |
10656 | | |
10657 | 9.76k | if (ctxt->input->cur < ctxt->input->end) { |
10658 | 1.20k | if (ctxt->wellFormed) |
10659 | 12 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
10660 | 8.56k | } else if ((ctxt->input->buf != NULL) && |
10661 | 8.56k | (ctxt->input->buf->encoder != NULL) && |
10662 | 8.56k | (ctxt->input->buf->error == 0) && |
10663 | 8.56k | (!xmlBufIsEmpty(ctxt->input->buf->raw))) { |
10664 | 90 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
10665 | 90 | "Truncated multi-byte sequence at EOF\n"); |
10666 | 90 | } |
10667 | 9.76k | } |
10668 | | |
10669 | 16.9k | ctxt->instate = XML_PARSER_EOF; |
10670 | 16.9k | xmlFinishDocument(ctxt); |
10671 | | |
10672 | 16.9k | if (! ctxt->wellFormed) { |
10673 | 16.9k | ctxt->valid = 0; |
10674 | 16.9k | return(-1); |
10675 | 16.9k | } |
10676 | | |
10677 | 36 | return(0); |
10678 | 16.9k | } |
10679 | | |
10680 | | /** |
10681 | | * xmlParseExtParsedEnt: |
10682 | | * @ctxt: an XML parser context |
10683 | | * |
10684 | | * parse a general parsed entity |
10685 | | * An external general parsed entity is well-formed if it matches the |
10686 | | * production labeled extParsedEnt. |
10687 | | * |
10688 | | * [78] extParsedEnt ::= TextDecl? content |
10689 | | * |
10690 | | * Returns 0, -1 in case of error. the parser context is augmented |
10691 | | * as a result of the parsing. |
10692 | | */ |
10693 | | |
10694 | | int |
10695 | 0 | xmlParseExtParsedEnt(xmlParserCtxtPtr ctxt) { |
10696 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10697 | 0 | return(-1); |
10698 | | |
10699 | 0 | xmlCtxtInitializeLate(ctxt); |
10700 | |
|
10701 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
10702 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
10703 | 0 | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
10704 | 0 | } |
10705 | |
|
10706 | 0 | xmlDetectEncoding(ctxt); |
10707 | |
|
10708 | 0 | if (CUR == 0) { |
10709 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10710 | 0 | } |
10711 | | |
10712 | | /* |
10713 | | * Check for the XMLDecl in the Prolog. |
10714 | | */ |
10715 | 0 | GROW; |
10716 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10717 | | |
10718 | | /* |
10719 | | * Note that we will switch encoding on the fly. |
10720 | | */ |
10721 | 0 | xmlParseXMLDecl(ctxt); |
10722 | 0 | SKIP_BLANKS; |
10723 | 0 | } else { |
10724 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10725 | 0 | } |
10726 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10727 | 0 | ctxt->sax->startDocument(ctxt->userData); |
10728 | | |
10729 | | /* |
10730 | | * Doing validity checking on chunk doesn't make sense |
10731 | | */ |
10732 | 0 | ctxt->options &= ~XML_PARSE_DTDVALID; |
10733 | 0 | ctxt->validate = 0; |
10734 | 0 | ctxt->depth = 0; |
10735 | |
|
10736 | 0 | xmlParseContentInternal(ctxt); |
10737 | |
|
10738 | 0 | if (ctxt->input->cur < ctxt->input->end) |
10739 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
10740 | | |
10741 | | /* |
10742 | | * SAX: end of the document processing. |
10743 | | */ |
10744 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10745 | 0 | ctxt->sax->endDocument(ctxt->userData); |
10746 | |
|
10747 | 0 | if (! ctxt->wellFormed) return(-1); |
10748 | 0 | return(0); |
10749 | 0 | } |
10750 | | |
10751 | | #ifdef LIBXML_PUSH_ENABLED |
10752 | | /************************************************************************ |
10753 | | * * |
10754 | | * Progressive parsing interfaces * |
10755 | | * * |
10756 | | ************************************************************************/ |
10757 | | |
10758 | | /** |
10759 | | * xmlParseLookupChar: |
10760 | | * @ctxt: an XML parser context |
10761 | | * @c: character |
10762 | | * |
10763 | | * Check whether the input buffer contains a character. |
10764 | | */ |
10765 | | static int |
10766 | 0 | xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) { |
10767 | 0 | const xmlChar *cur; |
10768 | |
|
10769 | 0 | if (ctxt->checkIndex == 0) { |
10770 | 0 | cur = ctxt->input->cur + 1; |
10771 | 0 | } else { |
10772 | 0 | cur = ctxt->input->cur + ctxt->checkIndex; |
10773 | 0 | } |
10774 | |
|
10775 | 0 | if (memchr(cur, c, ctxt->input->end - cur) == NULL) { |
10776 | 0 | size_t index = ctxt->input->end - ctxt->input->cur; |
10777 | |
|
10778 | 0 | if (index > LONG_MAX) { |
10779 | 0 | ctxt->checkIndex = 0; |
10780 | 0 | return(1); |
10781 | 0 | } |
10782 | 0 | ctxt->checkIndex = index; |
10783 | 0 | return(0); |
10784 | 0 | } else { |
10785 | 0 | ctxt->checkIndex = 0; |
10786 | 0 | return(1); |
10787 | 0 | } |
10788 | 0 | } |
10789 | | |
10790 | | /** |
10791 | | * xmlParseLookupString: |
10792 | | * @ctxt: an XML parser context |
10793 | | * @startDelta: delta to apply at the start |
10794 | | * @str: string |
10795 | | * @strLen: length of string |
10796 | | * |
10797 | | * Check whether the input buffer contains a string. |
10798 | | */ |
10799 | | static const xmlChar * |
10800 | | xmlParseLookupString(xmlParserCtxtPtr ctxt, size_t startDelta, |
10801 | 0 | const char *str, size_t strLen) { |
10802 | 0 | const xmlChar *cur, *term; |
10803 | |
|
10804 | 0 | if (ctxt->checkIndex == 0) { |
10805 | 0 | cur = ctxt->input->cur + startDelta; |
10806 | 0 | } else { |
10807 | 0 | cur = ctxt->input->cur + ctxt->checkIndex; |
10808 | 0 | } |
10809 | |
|
10810 | 0 | term = BAD_CAST strstr((const char *) cur, str); |
10811 | 0 | if (term == NULL) { |
10812 | 0 | const xmlChar *end = ctxt->input->end; |
10813 | 0 | size_t index; |
10814 | | |
10815 | | /* Rescan (strLen - 1) characters. */ |
10816 | 0 | if ((size_t) (end - cur) < strLen) |
10817 | 0 | end = cur; |
10818 | 0 | else |
10819 | 0 | end -= strLen - 1; |
10820 | 0 | index = end - ctxt->input->cur; |
10821 | 0 | if (index > LONG_MAX) { |
10822 | 0 | ctxt->checkIndex = 0; |
10823 | 0 | return(ctxt->input->end - strLen); |
10824 | 0 | } |
10825 | 0 | ctxt->checkIndex = index; |
10826 | 0 | } else { |
10827 | 0 | ctxt->checkIndex = 0; |
10828 | 0 | } |
10829 | | |
10830 | 0 | return(term); |
10831 | 0 | } |
10832 | | |
10833 | | /** |
10834 | | * xmlParseLookupCharData: |
10835 | | * @ctxt: an XML parser context |
10836 | | * |
10837 | | * Check whether the input buffer contains terminated char data. |
10838 | | */ |
10839 | | static int |
10840 | 0 | xmlParseLookupCharData(xmlParserCtxtPtr ctxt) { |
10841 | 0 | const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex; |
10842 | 0 | const xmlChar *end = ctxt->input->end; |
10843 | 0 | size_t index; |
10844 | |
|
10845 | 0 | while (cur < end) { |
10846 | 0 | if ((*cur == '<') || (*cur == '&')) { |
10847 | 0 | ctxt->checkIndex = 0; |
10848 | 0 | return(1); |
10849 | 0 | } |
10850 | 0 | cur++; |
10851 | 0 | } |
10852 | | |
10853 | 0 | index = cur - ctxt->input->cur; |
10854 | 0 | if (index > LONG_MAX) { |
10855 | 0 | ctxt->checkIndex = 0; |
10856 | 0 | return(1); |
10857 | 0 | } |
10858 | 0 | ctxt->checkIndex = index; |
10859 | 0 | return(0); |
10860 | 0 | } |
10861 | | |
10862 | | /** |
10863 | | * xmlParseLookupGt: |
10864 | | * @ctxt: an XML parser context |
10865 | | * |
10866 | | * Check whether there's enough data in the input buffer to finish parsing |
10867 | | * a start tag. This has to take quotes into account. |
10868 | | */ |
10869 | | static int |
10870 | 0 | xmlParseLookupGt(xmlParserCtxtPtr ctxt) { |
10871 | 0 | const xmlChar *cur; |
10872 | 0 | const xmlChar *end = ctxt->input->end; |
10873 | 0 | int state = ctxt->endCheckState; |
10874 | 0 | size_t index; |
10875 | |
|
10876 | 0 | if (ctxt->checkIndex == 0) |
10877 | 0 | cur = ctxt->input->cur + 1; |
10878 | 0 | else |
10879 | 0 | cur = ctxt->input->cur + ctxt->checkIndex; |
10880 | |
|
10881 | 0 | while (cur < end) { |
10882 | 0 | if (state) { |
10883 | 0 | if (*cur == state) |
10884 | 0 | state = 0; |
10885 | 0 | } else if (*cur == '\'' || *cur == '"') { |
10886 | 0 | state = *cur; |
10887 | 0 | } else if (*cur == '>') { |
10888 | 0 | ctxt->checkIndex = 0; |
10889 | 0 | ctxt->endCheckState = 0; |
10890 | 0 | return(1); |
10891 | 0 | } |
10892 | 0 | cur++; |
10893 | 0 | } |
10894 | | |
10895 | 0 | index = cur - ctxt->input->cur; |
10896 | 0 | if (index > LONG_MAX) { |
10897 | 0 | ctxt->checkIndex = 0; |
10898 | 0 | ctxt->endCheckState = 0; |
10899 | 0 | return(1); |
10900 | 0 | } |
10901 | 0 | ctxt->checkIndex = index; |
10902 | 0 | ctxt->endCheckState = state; |
10903 | 0 | return(0); |
10904 | 0 | } |
10905 | | |
10906 | | /** |
10907 | | * xmlParseLookupInternalSubset: |
10908 | | * @ctxt: an XML parser context |
10909 | | * |
10910 | | * Check whether there's enough data in the input buffer to finish parsing |
10911 | | * the internal subset. |
10912 | | */ |
10913 | | static int |
10914 | 0 | xmlParseLookupInternalSubset(xmlParserCtxtPtr ctxt) { |
10915 | | /* |
10916 | | * Sorry, but progressive parsing of the internal subset is not |
10917 | | * supported. We first check that the full content of the internal |
10918 | | * subset is available and parsing is launched only at that point. |
10919 | | * Internal subset ends with "']' S? '>'" in an unescaped section and |
10920 | | * not in a ']]>' sequence which are conditional sections. |
10921 | | */ |
10922 | 0 | const xmlChar *cur, *start; |
10923 | 0 | const xmlChar *end = ctxt->input->end; |
10924 | 0 | int state = ctxt->endCheckState; |
10925 | 0 | size_t index; |
10926 | |
|
10927 | 0 | if (ctxt->checkIndex == 0) { |
10928 | 0 | cur = ctxt->input->cur + 1; |
10929 | 0 | } else { |
10930 | 0 | cur = ctxt->input->cur + ctxt->checkIndex; |
10931 | 0 | } |
10932 | 0 | start = cur; |
10933 | |
|
10934 | 0 | while (cur < end) { |
10935 | 0 | if (state == '-') { |
10936 | 0 | if ((*cur == '-') && |
10937 | 0 | (cur[1] == '-') && |
10938 | 0 | (cur[2] == '>')) { |
10939 | 0 | state = 0; |
10940 | 0 | cur += 3; |
10941 | 0 | start = cur; |
10942 | 0 | continue; |
10943 | 0 | } |
10944 | 0 | } |
10945 | 0 | else if (state == ']') { |
10946 | 0 | if (*cur == '>') { |
10947 | 0 | ctxt->checkIndex = 0; |
10948 | 0 | ctxt->endCheckState = 0; |
10949 | 0 | return(1); |
10950 | 0 | } |
10951 | 0 | if (IS_BLANK_CH(*cur)) { |
10952 | 0 | state = ' '; |
10953 | 0 | } else if (*cur != ']') { |
10954 | 0 | state = 0; |
10955 | 0 | start = cur; |
10956 | 0 | continue; |
10957 | 0 | } |
10958 | 0 | } |
10959 | 0 | else if (state == ' ') { |
10960 | 0 | if (*cur == '>') { |
10961 | 0 | ctxt->checkIndex = 0; |
10962 | 0 | ctxt->endCheckState = 0; |
10963 | 0 | return(1); |
10964 | 0 | } |
10965 | 0 | if (!IS_BLANK_CH(*cur)) { |
10966 | 0 | state = 0; |
10967 | 0 | start = cur; |
10968 | 0 | continue; |
10969 | 0 | } |
10970 | 0 | } |
10971 | 0 | else if (state != 0) { |
10972 | 0 | if (*cur == state) { |
10973 | 0 | state = 0; |
10974 | 0 | start = cur + 1; |
10975 | 0 | } |
10976 | 0 | } |
10977 | 0 | else if (*cur == '<') { |
10978 | 0 | if ((cur[1] == '!') && |
10979 | 0 | (cur[2] == '-') && |
10980 | 0 | (cur[3] == '-')) { |
10981 | 0 | state = '-'; |
10982 | 0 | cur += 4; |
10983 | | /* Don't treat <!--> as comment */ |
10984 | 0 | start = cur; |
10985 | 0 | continue; |
10986 | 0 | } |
10987 | 0 | } |
10988 | 0 | else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) { |
10989 | 0 | state = *cur; |
10990 | 0 | } |
10991 | | |
10992 | 0 | cur++; |
10993 | 0 | } |
10994 | | |
10995 | | /* |
10996 | | * Rescan the three last characters to detect "<!--" and "-->" |
10997 | | * split across chunks. |
10998 | | */ |
10999 | 0 | if ((state == 0) || (state == '-')) { |
11000 | 0 | if (cur - start < 3) |
11001 | 0 | cur = start; |
11002 | 0 | else |
11003 | 0 | cur -= 3; |
11004 | 0 | } |
11005 | 0 | index = cur - ctxt->input->cur; |
11006 | 0 | if (index > LONG_MAX) { |
11007 | 0 | ctxt->checkIndex = 0; |
11008 | 0 | ctxt->endCheckState = 0; |
11009 | 0 | return(1); |
11010 | 0 | } |
11011 | 0 | ctxt->checkIndex = index; |
11012 | 0 | ctxt->endCheckState = state; |
11013 | 0 | return(0); |
11014 | 0 | } |
11015 | | |
11016 | | /** |
11017 | | * xmlParseTryOrFinish: |
11018 | | * @ctxt: an XML parser context |
11019 | | * @terminate: last chunk indicator |
11020 | | * |
11021 | | * Try to progress on parsing |
11022 | | * |
11023 | | * Returns zero if no parsing was possible |
11024 | | */ |
11025 | | static int |
11026 | 0 | xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) { |
11027 | 0 | int ret = 0; |
11028 | 0 | size_t avail; |
11029 | 0 | xmlChar cur, next; |
11030 | |
|
11031 | 0 | if (ctxt->input == NULL) |
11032 | 0 | return(0); |
11033 | | |
11034 | 0 | if ((ctxt->input != NULL) && |
11035 | 0 | (ctxt->input->cur - ctxt->input->base > 4096)) { |
11036 | 0 | xmlParserShrink(ctxt); |
11037 | 0 | } |
11038 | |
|
11039 | 0 | while (ctxt->disableSAX == 0) { |
11040 | 0 | avail = ctxt->input->end - ctxt->input->cur; |
11041 | 0 | if (avail < 1) |
11042 | 0 | goto done; |
11043 | 0 | switch (ctxt->instate) { |
11044 | 0 | case XML_PARSER_EOF: |
11045 | | /* |
11046 | | * Document parsing is done ! |
11047 | | */ |
11048 | 0 | goto done; |
11049 | 0 | case XML_PARSER_START: |
11050 | | /* |
11051 | | * Very first chars read from the document flow. |
11052 | | */ |
11053 | 0 | if ((!terminate) && (avail < 4)) |
11054 | 0 | goto done; |
11055 | | |
11056 | | /* |
11057 | | * We need more bytes to detect EBCDIC code pages. |
11058 | | * See xmlDetectEBCDIC. |
11059 | | */ |
11060 | 0 | if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) && |
11061 | 0 | (!terminate) && (avail < 200)) |
11062 | 0 | goto done; |
11063 | | |
11064 | 0 | xmlDetectEncoding(ctxt); |
11065 | 0 | ctxt->instate = XML_PARSER_XML_DECL; |
11066 | 0 | break; |
11067 | | |
11068 | 0 | case XML_PARSER_XML_DECL: |
11069 | 0 | if ((!terminate) && (avail < 2)) |
11070 | 0 | goto done; |
11071 | 0 | cur = ctxt->input->cur[0]; |
11072 | 0 | next = ctxt->input->cur[1]; |
11073 | 0 | if ((cur == '<') && (next == '?')) { |
11074 | | /* PI or XML decl */ |
11075 | 0 | if ((!terminate) && |
11076 | 0 | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11077 | 0 | goto done; |
11078 | 0 | if ((ctxt->input->cur[2] == 'x') && |
11079 | 0 | (ctxt->input->cur[3] == 'm') && |
11080 | 0 | (ctxt->input->cur[4] == 'l') && |
11081 | 0 | (IS_BLANK_CH(ctxt->input->cur[5]))) { |
11082 | 0 | ret += 5; |
11083 | 0 | xmlParseXMLDecl(ctxt); |
11084 | 0 | } else { |
11085 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11086 | 0 | if (ctxt->version == NULL) { |
11087 | 0 | xmlErrMemory(ctxt); |
11088 | 0 | break; |
11089 | 0 | } |
11090 | 0 | } |
11091 | 0 | } else { |
11092 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11093 | 0 | if (ctxt->version == NULL) { |
11094 | 0 | xmlErrMemory(ctxt); |
11095 | 0 | break; |
11096 | 0 | } |
11097 | 0 | } |
11098 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
11099 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
11100 | 0 | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
11101 | 0 | } |
11102 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11103 | 0 | (!ctxt->disableSAX)) |
11104 | 0 | ctxt->sax->startDocument(ctxt->userData); |
11105 | 0 | ctxt->instate = XML_PARSER_MISC; |
11106 | 0 | break; |
11107 | 0 | case XML_PARSER_START_TAG: { |
11108 | 0 | const xmlChar *name; |
11109 | 0 | const xmlChar *prefix = NULL; |
11110 | 0 | const xmlChar *URI = NULL; |
11111 | 0 | int line = ctxt->input->line; |
11112 | 0 | int nbNs = 0; |
11113 | |
|
11114 | 0 | if ((!terminate) && (avail < 2)) |
11115 | 0 | goto done; |
11116 | 0 | cur = ctxt->input->cur[0]; |
11117 | 0 | if (cur != '<') { |
11118 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
11119 | 0 | "Start tag expected, '<' not found"); |
11120 | 0 | ctxt->instate = XML_PARSER_EOF; |
11121 | 0 | xmlFinishDocument(ctxt); |
11122 | 0 | goto done; |
11123 | 0 | } |
11124 | 0 | if ((!terminate) && (!xmlParseLookupGt(ctxt))) |
11125 | 0 | goto done; |
11126 | 0 | if (ctxt->spaceNr == 0) |
11127 | 0 | spacePush(ctxt, -1); |
11128 | 0 | else if (*ctxt->space == -2) |
11129 | 0 | spacePush(ctxt, -1); |
11130 | 0 | else |
11131 | 0 | spacePush(ctxt, *ctxt->space); |
11132 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11133 | 0 | if (ctxt->sax2) |
11134 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11135 | 0 | name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs); |
11136 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11137 | 0 | else |
11138 | 0 | name = xmlParseStartTag(ctxt); |
11139 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11140 | 0 | if (name == NULL) { |
11141 | 0 | spacePop(ctxt); |
11142 | 0 | ctxt->instate = XML_PARSER_EOF; |
11143 | 0 | xmlFinishDocument(ctxt); |
11144 | 0 | goto done; |
11145 | 0 | } |
11146 | 0 | #ifdef LIBXML_VALID_ENABLED |
11147 | | /* |
11148 | | * [ VC: Root Element Type ] |
11149 | | * The Name in the document type declaration must match |
11150 | | * the element type of the root element. |
11151 | | */ |
11152 | 0 | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
11153 | 0 | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
11154 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
11155 | 0 | #endif /* LIBXML_VALID_ENABLED */ |
11156 | | |
11157 | | /* |
11158 | | * Check for an Empty Element. |
11159 | | */ |
11160 | 0 | if ((RAW == '/') && (NXT(1) == '>')) { |
11161 | 0 | SKIP(2); |
11162 | |
|
11163 | 0 | if (ctxt->sax2) { |
11164 | 0 | if ((ctxt->sax != NULL) && |
11165 | 0 | (ctxt->sax->endElementNs != NULL) && |
11166 | 0 | (!ctxt->disableSAX)) |
11167 | 0 | ctxt->sax->endElementNs(ctxt->userData, name, |
11168 | 0 | prefix, URI); |
11169 | 0 | if (nbNs > 0) |
11170 | 0 | xmlParserNsPop(ctxt, nbNs); |
11171 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11172 | 0 | } else { |
11173 | 0 | if ((ctxt->sax != NULL) && |
11174 | 0 | (ctxt->sax->endElement != NULL) && |
11175 | 0 | (!ctxt->disableSAX)) |
11176 | 0 | ctxt->sax->endElement(ctxt->userData, name); |
11177 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11178 | 0 | } |
11179 | 0 | spacePop(ctxt); |
11180 | 0 | } else if (RAW == '>') { |
11181 | 0 | NEXT; |
11182 | 0 | nameNsPush(ctxt, name, prefix, URI, line, nbNs); |
11183 | 0 | } else { |
11184 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED, |
11185 | 0 | "Couldn't find end of Start Tag %s\n", |
11186 | 0 | name); |
11187 | 0 | nodePop(ctxt); |
11188 | 0 | spacePop(ctxt); |
11189 | 0 | if (nbNs > 0) |
11190 | 0 | xmlParserNsPop(ctxt, nbNs); |
11191 | 0 | } |
11192 | |
|
11193 | 0 | if (ctxt->nameNr == 0) |
11194 | 0 | ctxt->instate = XML_PARSER_EPILOG; |
11195 | 0 | else |
11196 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11197 | 0 | break; |
11198 | 0 | } |
11199 | 0 | case XML_PARSER_CONTENT: { |
11200 | 0 | cur = ctxt->input->cur[0]; |
11201 | |
|
11202 | 0 | if (cur == '<') { |
11203 | 0 | if ((!terminate) && (avail < 2)) |
11204 | 0 | goto done; |
11205 | 0 | next = ctxt->input->cur[1]; |
11206 | |
|
11207 | 0 | if (next == '/') { |
11208 | 0 | ctxt->instate = XML_PARSER_END_TAG; |
11209 | 0 | break; |
11210 | 0 | } else if (next == '?') { |
11211 | 0 | if ((!terminate) && |
11212 | 0 | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11213 | 0 | goto done; |
11214 | 0 | xmlParsePI(ctxt); |
11215 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11216 | 0 | break; |
11217 | 0 | } else if (next == '!') { |
11218 | 0 | if ((!terminate) && (avail < 3)) |
11219 | 0 | goto done; |
11220 | 0 | next = ctxt->input->cur[2]; |
11221 | |
|
11222 | 0 | if (next == '-') { |
11223 | 0 | if ((!terminate) && (avail < 4)) |
11224 | 0 | goto done; |
11225 | 0 | if (ctxt->input->cur[3] == '-') { |
11226 | 0 | if ((!terminate) && |
11227 | 0 | (!xmlParseLookupString(ctxt, 4, "-->", 3))) |
11228 | 0 | goto done; |
11229 | 0 | xmlParseComment(ctxt); |
11230 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11231 | 0 | break; |
11232 | 0 | } |
11233 | 0 | } else if (next == '[') { |
11234 | 0 | if ((!terminate) && (avail < 9)) |
11235 | 0 | goto done; |
11236 | 0 | if ((ctxt->input->cur[2] == '[') && |
11237 | 0 | (ctxt->input->cur[3] == 'C') && |
11238 | 0 | (ctxt->input->cur[4] == 'D') && |
11239 | 0 | (ctxt->input->cur[5] == 'A') && |
11240 | 0 | (ctxt->input->cur[6] == 'T') && |
11241 | 0 | (ctxt->input->cur[7] == 'A') && |
11242 | 0 | (ctxt->input->cur[8] == '[')) { |
11243 | 0 | if ((!terminate) && |
11244 | 0 | (!xmlParseLookupString(ctxt, 9, "]]>", 3))) |
11245 | 0 | goto done; |
11246 | 0 | ctxt->instate = XML_PARSER_CDATA_SECTION; |
11247 | 0 | xmlParseCDSect(ctxt); |
11248 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11249 | 0 | break; |
11250 | 0 | } |
11251 | 0 | } |
11252 | 0 | } |
11253 | 0 | } else if (cur == '&') { |
11254 | 0 | if ((!terminate) && (!xmlParseLookupChar(ctxt, ';'))) |
11255 | 0 | goto done; |
11256 | 0 | xmlParseReference(ctxt); |
11257 | 0 | break; |
11258 | 0 | } else { |
11259 | | /* TODO Avoid the extra copy, handle directly !!! */ |
11260 | | /* |
11261 | | * Goal of the following test is: |
11262 | | * - minimize calls to the SAX 'character' callback |
11263 | | * when they are mergeable |
11264 | | * - handle an problem for isBlank when we only parse |
11265 | | * a sequence of blank chars and the next one is |
11266 | | * not available to check against '<' presence. |
11267 | | * - tries to homogenize the differences in SAX |
11268 | | * callbacks between the push and pull versions |
11269 | | * of the parser. |
11270 | | */ |
11271 | 0 | if (avail < XML_PARSER_BIG_BUFFER_SIZE) { |
11272 | 0 | if ((!terminate) && (!xmlParseLookupCharData(ctxt))) |
11273 | 0 | goto done; |
11274 | 0 | } |
11275 | 0 | ctxt->checkIndex = 0; |
11276 | 0 | xmlParseCharDataInternal(ctxt, !terminate); |
11277 | 0 | break; |
11278 | 0 | } |
11279 | | |
11280 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
11281 | 0 | break; |
11282 | 0 | } |
11283 | 0 | case XML_PARSER_END_TAG: |
11284 | 0 | if ((!terminate) && (!xmlParseLookupChar(ctxt, '>'))) |
11285 | 0 | goto done; |
11286 | 0 | if (ctxt->sax2) { |
11287 | 0 | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
11288 | 0 | nameNsPop(ctxt); |
11289 | 0 | } |
11290 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11291 | 0 | else |
11292 | 0 | xmlParseEndTag1(ctxt, 0); |
11293 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11294 | 0 | if (ctxt->nameNr == 0) { |
11295 | 0 | ctxt->instate = XML_PARSER_EPILOG; |
11296 | 0 | } else { |
11297 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11298 | 0 | } |
11299 | 0 | break; |
11300 | 0 | case XML_PARSER_MISC: |
11301 | 0 | case XML_PARSER_PROLOG: |
11302 | 0 | case XML_PARSER_EPILOG: |
11303 | 0 | SKIP_BLANKS; |
11304 | 0 | avail = ctxt->input->end - ctxt->input->cur; |
11305 | 0 | if (avail < 1) |
11306 | 0 | goto done; |
11307 | 0 | if (ctxt->input->cur[0] == '<') { |
11308 | 0 | if ((!terminate) && (avail < 2)) |
11309 | 0 | goto done; |
11310 | 0 | next = ctxt->input->cur[1]; |
11311 | 0 | if (next == '?') { |
11312 | 0 | if ((!terminate) && |
11313 | 0 | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11314 | 0 | goto done; |
11315 | 0 | xmlParsePI(ctxt); |
11316 | 0 | break; |
11317 | 0 | } else if (next == '!') { |
11318 | 0 | if ((!terminate) && (avail < 3)) |
11319 | 0 | goto done; |
11320 | | |
11321 | 0 | if (ctxt->input->cur[2] == '-') { |
11322 | 0 | if ((!terminate) && (avail < 4)) |
11323 | 0 | goto done; |
11324 | 0 | if (ctxt->input->cur[3] == '-') { |
11325 | 0 | if ((!terminate) && |
11326 | 0 | (!xmlParseLookupString(ctxt, 4, "-->", 3))) |
11327 | 0 | goto done; |
11328 | 0 | xmlParseComment(ctxt); |
11329 | 0 | break; |
11330 | 0 | } |
11331 | 0 | } else if (ctxt->instate == XML_PARSER_MISC) { |
11332 | 0 | if ((!terminate) && (avail < 9)) |
11333 | 0 | goto done; |
11334 | 0 | if ((ctxt->input->cur[2] == 'D') && |
11335 | 0 | (ctxt->input->cur[3] == 'O') && |
11336 | 0 | (ctxt->input->cur[4] == 'C') && |
11337 | 0 | (ctxt->input->cur[5] == 'T') && |
11338 | 0 | (ctxt->input->cur[6] == 'Y') && |
11339 | 0 | (ctxt->input->cur[7] == 'P') && |
11340 | 0 | (ctxt->input->cur[8] == 'E')) { |
11341 | 0 | if ((!terminate) && (!xmlParseLookupGt(ctxt))) |
11342 | 0 | goto done; |
11343 | 0 | ctxt->inSubset = 1; |
11344 | 0 | xmlParseDocTypeDecl(ctxt); |
11345 | 0 | if (RAW == '[') { |
11346 | 0 | ctxt->instate = XML_PARSER_DTD; |
11347 | 0 | } else { |
11348 | | /* |
11349 | | * Create and update the external subset. |
11350 | | */ |
11351 | 0 | ctxt->inSubset = 2; |
11352 | 0 | if ((ctxt->sax != NULL) && |
11353 | 0 | (!ctxt->disableSAX) && |
11354 | 0 | (ctxt->sax->externalSubset != NULL)) |
11355 | 0 | ctxt->sax->externalSubset( |
11356 | 0 | ctxt->userData, |
11357 | 0 | ctxt->intSubName, |
11358 | 0 | ctxt->extSubSystem, |
11359 | 0 | ctxt->extSubURI); |
11360 | 0 | ctxt->inSubset = 0; |
11361 | 0 | xmlCleanSpecialAttr(ctxt); |
11362 | 0 | ctxt->instate = XML_PARSER_PROLOG; |
11363 | 0 | } |
11364 | 0 | break; |
11365 | 0 | } |
11366 | 0 | } |
11367 | 0 | } |
11368 | 0 | } |
11369 | | |
11370 | 0 | if (ctxt->instate == XML_PARSER_EPILOG) { |
11371 | 0 | if (ctxt->errNo == XML_ERR_OK) |
11372 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
11373 | 0 | ctxt->instate = XML_PARSER_EOF; |
11374 | 0 | xmlFinishDocument(ctxt); |
11375 | 0 | } else { |
11376 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
11377 | 0 | } |
11378 | 0 | break; |
11379 | 0 | case XML_PARSER_DTD: { |
11380 | 0 | if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt))) |
11381 | 0 | goto done; |
11382 | 0 | xmlParseInternalSubset(ctxt); |
11383 | 0 | ctxt->inSubset = 2; |
11384 | 0 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
11385 | 0 | (ctxt->sax->externalSubset != NULL)) |
11386 | 0 | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
11387 | 0 | ctxt->extSubSystem, ctxt->extSubURI); |
11388 | 0 | ctxt->inSubset = 0; |
11389 | 0 | xmlCleanSpecialAttr(ctxt); |
11390 | 0 | ctxt->instate = XML_PARSER_PROLOG; |
11391 | 0 | break; |
11392 | 0 | } |
11393 | 0 | default: |
11394 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
11395 | 0 | "PP: internal error\n"); |
11396 | 0 | ctxt->instate = XML_PARSER_EOF; |
11397 | 0 | break; |
11398 | 0 | } |
11399 | 0 | } |
11400 | 0 | done: |
11401 | 0 | return(ret); |
11402 | 0 | } |
11403 | | |
11404 | | /** |
11405 | | * xmlParseChunk: |
11406 | | * @ctxt: an XML parser context |
11407 | | * @chunk: chunk of memory |
11408 | | * @size: size of chunk in bytes |
11409 | | * @terminate: last chunk indicator |
11410 | | * |
11411 | | * Parse a chunk of memory in push parser mode. |
11412 | | * |
11413 | | * Assumes that the parser context was initialized with |
11414 | | * xmlCreatePushParserCtxt. |
11415 | | * |
11416 | | * The last chunk, which will often be empty, must be marked with |
11417 | | * the @terminate flag. With the default SAX callbacks, the resulting |
11418 | | * document will be available in ctxt->myDoc. This pointer will not |
11419 | | * be freed by the library. |
11420 | | * |
11421 | | * If the document isn't well-formed, ctxt->myDoc is set to NULL. |
11422 | | * The push parser doesn't support recovery mode. |
11423 | | * |
11424 | | * Returns an xmlParserErrors code (0 on success). |
11425 | | */ |
11426 | | int |
11427 | | xmlParseChunk(xmlParserCtxtPtr ctxt, const char *chunk, int size, |
11428 | 0 | int terminate) { |
11429 | 0 | size_t curBase; |
11430 | 0 | size_t maxLength; |
11431 | 0 | size_t pos; |
11432 | 0 | int end_in_lf = 0; |
11433 | 0 | int res; |
11434 | |
|
11435 | 0 | if ((ctxt == NULL) || (size < 0)) |
11436 | 0 | return(XML_ERR_ARGUMENT); |
11437 | 0 | if ((chunk == NULL) && (size > 0)) |
11438 | 0 | return(XML_ERR_ARGUMENT); |
11439 | 0 | if ((ctxt->input == NULL) || (ctxt->input->buf == NULL)) |
11440 | 0 | return(XML_ERR_ARGUMENT); |
11441 | 0 | if (ctxt->disableSAX != 0) |
11442 | 0 | return(ctxt->errNo); |
11443 | | |
11444 | 0 | ctxt->input->flags |= XML_INPUT_PROGRESSIVE; |
11445 | 0 | if (ctxt->instate == XML_PARSER_START) |
11446 | 0 | xmlCtxtInitializeLate(ctxt); |
11447 | 0 | if ((size > 0) && (chunk != NULL) && (!terminate) && |
11448 | 0 | (chunk[size - 1] == '\r')) { |
11449 | 0 | end_in_lf = 1; |
11450 | 0 | size--; |
11451 | 0 | } |
11452 | | |
11453 | | /* |
11454 | | * Also push an empty chunk to make sure that the raw buffer |
11455 | | * will be flushed if there is an encoder. |
11456 | | */ |
11457 | 0 | pos = ctxt->input->cur - ctxt->input->base; |
11458 | 0 | res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
11459 | 0 | xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos); |
11460 | 0 | if (res < 0) { |
11461 | 0 | xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL); |
11462 | 0 | xmlHaltParser(ctxt); |
11463 | 0 | return(ctxt->errNo); |
11464 | 0 | } |
11465 | | |
11466 | 0 | xmlParseTryOrFinish(ctxt, terminate); |
11467 | |
|
11468 | 0 | curBase = ctxt->input->cur - ctxt->input->base; |
11469 | 0 | maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
11470 | 0 | XML_MAX_HUGE_LENGTH : |
11471 | 0 | XML_MAX_LOOKUP_LIMIT; |
11472 | 0 | if (curBase > maxLength) { |
11473 | 0 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, |
11474 | 0 | "Buffer size limit exceeded, try XML_PARSE_HUGE\n"); |
11475 | 0 | xmlHaltParser(ctxt); |
11476 | 0 | } |
11477 | |
|
11478 | 0 | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0)) |
11479 | 0 | return(ctxt->errNo); |
11480 | | |
11481 | 0 | if (end_in_lf == 1) { |
11482 | 0 | pos = ctxt->input->cur - ctxt->input->base; |
11483 | 0 | res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r"); |
11484 | 0 | xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos); |
11485 | 0 | if (res < 0) { |
11486 | 0 | xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL); |
11487 | 0 | xmlHaltParser(ctxt); |
11488 | 0 | return(ctxt->errNo); |
11489 | 0 | } |
11490 | 0 | } |
11491 | 0 | if (terminate) { |
11492 | | /* |
11493 | | * Check for termination |
11494 | | */ |
11495 | 0 | if ((ctxt->instate != XML_PARSER_EOF) && |
11496 | 0 | (ctxt->instate != XML_PARSER_EPILOG)) { |
11497 | 0 | if (ctxt->nameNr > 0) { |
11498 | 0 | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
11499 | 0 | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
11500 | 0 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
11501 | 0 | "Premature end of data in tag %s line %d\n", |
11502 | 0 | name, line, NULL); |
11503 | 0 | } else if (ctxt->instate == XML_PARSER_START) { |
11504 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11505 | 0 | } else { |
11506 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
11507 | 0 | "Start tag expected, '<' not found\n"); |
11508 | 0 | } |
11509 | 0 | } else if ((ctxt->input->buf->encoder != NULL) && |
11510 | 0 | (ctxt->input->buf->error == 0) && |
11511 | 0 | (!xmlBufIsEmpty(ctxt->input->buf->raw))) { |
11512 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
11513 | 0 | "Truncated multi-byte sequence at EOF\n"); |
11514 | 0 | } |
11515 | 0 | if (ctxt->instate != XML_PARSER_EOF) { |
11516 | 0 | ctxt->instate = XML_PARSER_EOF; |
11517 | 0 | xmlFinishDocument(ctxt); |
11518 | 0 | } |
11519 | 0 | } |
11520 | 0 | if (ctxt->wellFormed == 0) |
11521 | 0 | return((xmlParserErrors) ctxt->errNo); |
11522 | 0 | else |
11523 | 0 | return(0); |
11524 | 0 | } |
11525 | | |
11526 | | /************************************************************************ |
11527 | | * * |
11528 | | * I/O front end functions to the parser * |
11529 | | * * |
11530 | | ************************************************************************/ |
11531 | | |
11532 | | /** |
11533 | | * xmlCreatePushParserCtxt: |
11534 | | * @sax: a SAX handler (optional) |
11535 | | * @user_data: user data for SAX callbacks (optional) |
11536 | | * @chunk: initial chunk (optional, deprecated) |
11537 | | * @size: size of initial chunk in bytes |
11538 | | * @filename: file name or URI (optional) |
11539 | | * |
11540 | | * Create a parser context for using the XML parser in push mode. |
11541 | | * See xmlParseChunk. |
11542 | | * |
11543 | | * Passing an initial chunk is useless and deprecated. |
11544 | | * |
11545 | | * @filename is used as base URI to fetch external entities and for |
11546 | | * error reports. |
11547 | | * |
11548 | | * Returns the new parser context or NULL if a memory allocation |
11549 | | * failed. |
11550 | | */ |
11551 | | |
11552 | | xmlParserCtxtPtr |
11553 | | xmlCreatePushParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
11554 | 0 | const char *chunk, int size, const char *filename) { |
11555 | 0 | xmlParserCtxtPtr ctxt; |
11556 | 0 | xmlParserInputPtr input; |
11557 | |
|
11558 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
11559 | 0 | if (ctxt == NULL) |
11560 | 0 | return(NULL); |
11561 | | |
11562 | 0 | ctxt->options &= ~XML_PARSE_NODICT; |
11563 | 0 | ctxt->dictNames = 1; |
11564 | |
|
11565 | 0 | input = xmlNewPushInput(filename, chunk, size); |
11566 | 0 | if (input == NULL) { |
11567 | 0 | xmlFreeParserCtxt(ctxt); |
11568 | 0 | return(NULL); |
11569 | 0 | } |
11570 | 0 | if (inputPush(ctxt, input) < 0) { |
11571 | 0 | xmlFreeInputStream(input); |
11572 | 0 | xmlFreeParserCtxt(ctxt); |
11573 | 0 | return(NULL); |
11574 | 0 | } |
11575 | | |
11576 | 0 | return(ctxt); |
11577 | 0 | } |
11578 | | #endif /* LIBXML_PUSH_ENABLED */ |
11579 | | |
11580 | | /** |
11581 | | * xmlStopParser: |
11582 | | * @ctxt: an XML parser context |
11583 | | * |
11584 | | * Blocks further parser processing |
11585 | | */ |
11586 | | void |
11587 | 0 | xmlStopParser(xmlParserCtxtPtr ctxt) { |
11588 | 0 | if (ctxt == NULL) |
11589 | 0 | return; |
11590 | 0 | xmlHaltParser(ctxt); |
11591 | 0 | if (ctxt->errNo != XML_ERR_NO_MEMORY) |
11592 | 0 | ctxt->errNo = XML_ERR_USER_STOP; |
11593 | 0 | } |
11594 | | |
11595 | | /** |
11596 | | * xmlCreateIOParserCtxt: |
11597 | | * @sax: a SAX handler (optional) |
11598 | | * @user_data: user data for SAX callbacks (optional) |
11599 | | * @ioread: an I/O read function |
11600 | | * @ioclose: an I/O close function (optional) |
11601 | | * @ioctx: an I/O handler |
11602 | | * @enc: the charset encoding if known (deprecated) |
11603 | | * |
11604 | | * Create a parser context for using the XML parser with an existing |
11605 | | * I/O stream |
11606 | | * |
11607 | | * Returns the new parser context or NULL |
11608 | | */ |
11609 | | xmlParserCtxtPtr |
11610 | | xmlCreateIOParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
11611 | | xmlInputReadCallback ioread, |
11612 | | xmlInputCloseCallback ioclose, |
11613 | 0 | void *ioctx, xmlCharEncoding enc) { |
11614 | 0 | xmlParserCtxtPtr ctxt; |
11615 | 0 | xmlParserInputPtr input; |
11616 | 0 | const char *encoding; |
11617 | |
|
11618 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
11619 | 0 | if (ctxt == NULL) |
11620 | 0 | return(NULL); |
11621 | | |
11622 | 0 | encoding = xmlGetCharEncodingName(enc); |
11623 | 0 | input = xmlCtxtNewInputFromIO(ctxt, NULL, ioread, ioclose, ioctx, |
11624 | 0 | encoding, 0); |
11625 | 0 | if (input == NULL) { |
11626 | 0 | xmlFreeParserCtxt(ctxt); |
11627 | 0 | return (NULL); |
11628 | 0 | } |
11629 | 0 | if (inputPush(ctxt, input) < 0) { |
11630 | 0 | xmlFreeInputStream(input); |
11631 | 0 | xmlFreeParserCtxt(ctxt); |
11632 | 0 | return(NULL); |
11633 | 0 | } |
11634 | | |
11635 | 0 | return(ctxt); |
11636 | 0 | } |
11637 | | |
11638 | | #ifdef LIBXML_VALID_ENABLED |
11639 | | /************************************************************************ |
11640 | | * * |
11641 | | * Front ends when parsing a DTD * |
11642 | | * * |
11643 | | ************************************************************************/ |
11644 | | |
11645 | | /** |
11646 | | * xmlIOParseDTD: |
11647 | | * @sax: the SAX handler block or NULL |
11648 | | * @input: an Input Buffer |
11649 | | * @enc: the charset encoding if known |
11650 | | * |
11651 | | * Load and parse a DTD |
11652 | | * |
11653 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
11654 | | * @input will be freed by the function in any case. |
11655 | | */ |
11656 | | |
11657 | | xmlDtdPtr |
11658 | | xmlIOParseDTD(xmlSAXHandlerPtr sax, xmlParserInputBufferPtr input, |
11659 | 0 | xmlCharEncoding enc) { |
11660 | 0 | xmlDtdPtr ret = NULL; |
11661 | 0 | xmlParserCtxtPtr ctxt; |
11662 | 0 | xmlParserInputPtr pinput = NULL; |
11663 | |
|
11664 | 0 | if (input == NULL) |
11665 | 0 | return(NULL); |
11666 | | |
11667 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
11668 | 0 | if (ctxt == NULL) { |
11669 | 0 | xmlFreeParserInputBuffer(input); |
11670 | 0 | return(NULL); |
11671 | 0 | } |
11672 | | |
11673 | | /* |
11674 | | * generate a parser input from the I/O handler |
11675 | | */ |
11676 | | |
11677 | 0 | pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
11678 | 0 | if (pinput == NULL) { |
11679 | 0 | xmlFreeParserInputBuffer(input); |
11680 | 0 | xmlFreeParserCtxt(ctxt); |
11681 | 0 | return(NULL); |
11682 | 0 | } |
11683 | | |
11684 | | /* |
11685 | | * plug some encoding conversion routines here. |
11686 | | */ |
11687 | 0 | if (xmlPushInput(ctxt, pinput) < 0) { |
11688 | 0 | xmlFreeInputStream(pinput); |
11689 | 0 | xmlFreeParserCtxt(ctxt); |
11690 | 0 | return(NULL); |
11691 | 0 | } |
11692 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
11693 | 0 | xmlSwitchEncoding(ctxt, enc); |
11694 | 0 | } |
11695 | | |
11696 | | /* |
11697 | | * let's parse that entity knowing it's an external subset. |
11698 | | */ |
11699 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
11700 | 0 | if (ctxt->myDoc == NULL) { |
11701 | 0 | xmlErrMemory(ctxt); |
11702 | 0 | return(NULL); |
11703 | 0 | } |
11704 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
11705 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
11706 | 0 | BAD_CAST "none", BAD_CAST "none"); |
11707 | |
|
11708 | 0 | xmlParseExternalSubset(ctxt, BAD_CAST "none", BAD_CAST "none"); |
11709 | |
|
11710 | 0 | if (ctxt->myDoc != NULL) { |
11711 | 0 | if (ctxt->wellFormed) { |
11712 | 0 | ret = ctxt->myDoc->extSubset; |
11713 | 0 | ctxt->myDoc->extSubset = NULL; |
11714 | 0 | if (ret != NULL) { |
11715 | 0 | xmlNodePtr tmp; |
11716 | |
|
11717 | 0 | ret->doc = NULL; |
11718 | 0 | tmp = ret->children; |
11719 | 0 | while (tmp != NULL) { |
11720 | 0 | tmp->doc = NULL; |
11721 | 0 | tmp = tmp->next; |
11722 | 0 | } |
11723 | 0 | } |
11724 | 0 | } else { |
11725 | 0 | ret = NULL; |
11726 | 0 | } |
11727 | 0 | xmlFreeDoc(ctxt->myDoc); |
11728 | 0 | ctxt->myDoc = NULL; |
11729 | 0 | } |
11730 | 0 | xmlFreeParserCtxt(ctxt); |
11731 | |
|
11732 | 0 | return(ret); |
11733 | 0 | } |
11734 | | |
11735 | | /** |
11736 | | * xmlSAXParseDTD: |
11737 | | * @sax: the SAX handler block |
11738 | | * @ExternalID: a NAME* containing the External ID of the DTD |
11739 | | * @SystemID: a NAME* containing the URL to the DTD |
11740 | | * |
11741 | | * DEPRECATED: Don't use. |
11742 | | * |
11743 | | * Load and parse an external subset. |
11744 | | * |
11745 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
11746 | | */ |
11747 | | |
11748 | | xmlDtdPtr |
11749 | | xmlSAXParseDTD(xmlSAXHandlerPtr sax, const xmlChar *ExternalID, |
11750 | 0 | const xmlChar *SystemID) { |
11751 | 0 | xmlDtdPtr ret = NULL; |
11752 | 0 | xmlParserCtxtPtr ctxt; |
11753 | 0 | xmlParserInputPtr input = NULL; |
11754 | 0 | xmlChar* systemIdCanonic; |
11755 | |
|
11756 | 0 | if ((ExternalID == NULL) && (SystemID == NULL)) return(NULL); |
11757 | | |
11758 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
11759 | 0 | if (ctxt == NULL) { |
11760 | 0 | return(NULL); |
11761 | 0 | } |
11762 | | |
11763 | | /* |
11764 | | * Canonicalise the system ID |
11765 | | */ |
11766 | 0 | systemIdCanonic = xmlCanonicPath(SystemID); |
11767 | 0 | if ((SystemID != NULL) && (systemIdCanonic == NULL)) { |
11768 | 0 | xmlFreeParserCtxt(ctxt); |
11769 | 0 | return(NULL); |
11770 | 0 | } |
11771 | | |
11772 | | /* |
11773 | | * Ask the Entity resolver to load the damn thing |
11774 | | */ |
11775 | | |
11776 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) |
11777 | 0 | input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID, |
11778 | 0 | systemIdCanonic); |
11779 | 0 | if (input == NULL) { |
11780 | 0 | xmlFreeParserCtxt(ctxt); |
11781 | 0 | if (systemIdCanonic != NULL) |
11782 | 0 | xmlFree(systemIdCanonic); |
11783 | 0 | return(NULL); |
11784 | 0 | } |
11785 | | |
11786 | | /* |
11787 | | * plug some encoding conversion routines here. |
11788 | | */ |
11789 | 0 | if (xmlPushInput(ctxt, input) < 0) { |
11790 | 0 | xmlFreeInputStream(input); |
11791 | 0 | xmlFreeParserCtxt(ctxt); |
11792 | 0 | if (systemIdCanonic != NULL) |
11793 | 0 | xmlFree(systemIdCanonic); |
11794 | 0 | return(NULL); |
11795 | 0 | } |
11796 | | |
11797 | 0 | xmlDetectEncoding(ctxt); |
11798 | |
|
11799 | 0 | if (input->filename == NULL) |
11800 | 0 | input->filename = (char *) systemIdCanonic; |
11801 | 0 | else |
11802 | 0 | xmlFree(systemIdCanonic); |
11803 | | |
11804 | | /* |
11805 | | * let's parse that entity knowing it's an external subset. |
11806 | | */ |
11807 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
11808 | 0 | if (ctxt->myDoc == NULL) { |
11809 | 0 | xmlErrMemory(ctxt); |
11810 | 0 | xmlFreeParserCtxt(ctxt); |
11811 | 0 | return(NULL); |
11812 | 0 | } |
11813 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
11814 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
11815 | 0 | ExternalID, SystemID); |
11816 | 0 | if (ctxt->myDoc->extSubset == NULL) { |
11817 | 0 | xmlFreeDoc(ctxt->myDoc); |
11818 | 0 | xmlFreeParserCtxt(ctxt); |
11819 | 0 | return(NULL); |
11820 | 0 | } |
11821 | 0 | xmlParseExternalSubset(ctxt, ExternalID, SystemID); |
11822 | |
|
11823 | 0 | if (ctxt->myDoc != NULL) { |
11824 | 0 | if (ctxt->wellFormed) { |
11825 | 0 | ret = ctxt->myDoc->extSubset; |
11826 | 0 | ctxt->myDoc->extSubset = NULL; |
11827 | 0 | if (ret != NULL) { |
11828 | 0 | xmlNodePtr tmp; |
11829 | |
|
11830 | 0 | ret->doc = NULL; |
11831 | 0 | tmp = ret->children; |
11832 | 0 | while (tmp != NULL) { |
11833 | 0 | tmp->doc = NULL; |
11834 | 0 | tmp = tmp->next; |
11835 | 0 | } |
11836 | 0 | } |
11837 | 0 | } else { |
11838 | 0 | ret = NULL; |
11839 | 0 | } |
11840 | 0 | xmlFreeDoc(ctxt->myDoc); |
11841 | 0 | ctxt->myDoc = NULL; |
11842 | 0 | } |
11843 | 0 | xmlFreeParserCtxt(ctxt); |
11844 | |
|
11845 | 0 | return(ret); |
11846 | 0 | } |
11847 | | |
11848 | | |
11849 | | /** |
11850 | | * xmlParseDTD: |
11851 | | * @ExternalID: a NAME* containing the External ID of the DTD |
11852 | | * @SystemID: a NAME* containing the URL to the DTD |
11853 | | * |
11854 | | * Load and parse an external subset. |
11855 | | * |
11856 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
11857 | | */ |
11858 | | |
11859 | | xmlDtdPtr |
11860 | 0 | xmlParseDTD(const xmlChar *ExternalID, const xmlChar *SystemID) { |
11861 | 0 | return(xmlSAXParseDTD(NULL, ExternalID, SystemID)); |
11862 | 0 | } |
11863 | | #endif /* LIBXML_VALID_ENABLED */ |
11864 | | |
11865 | | /************************************************************************ |
11866 | | * * |
11867 | | * Front ends when parsing an Entity * |
11868 | | * * |
11869 | | ************************************************************************/ |
11870 | | |
11871 | | static xmlNodePtr |
11872 | | xmlCtxtParseContentInternal(xmlParserCtxtPtr ctxt, xmlParserInputPtr input, |
11873 | 863 | int hasTextDecl, int buildTree) { |
11874 | 863 | xmlNodePtr root = NULL; |
11875 | 863 | xmlNodePtr list = NULL; |
11876 | 863 | xmlChar *rootName = BAD_CAST "#root"; |
11877 | 863 | int result; |
11878 | | |
11879 | 863 | if (buildTree) { |
11880 | 863 | root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL); |
11881 | 863 | if (root == NULL) { |
11882 | 0 | xmlErrMemory(ctxt); |
11883 | 0 | goto error; |
11884 | 0 | } |
11885 | 863 | } |
11886 | | |
11887 | 863 | if (xmlPushInput(ctxt, input) < 0) |
11888 | 2 | goto error; |
11889 | | |
11890 | 861 | nameNsPush(ctxt, rootName, NULL, NULL, 0, 0); |
11891 | 861 | spacePush(ctxt, -1); |
11892 | | |
11893 | 861 | if (buildTree) |
11894 | 861 | nodePush(ctxt, root); |
11895 | | |
11896 | 861 | if (hasTextDecl) { |
11897 | 0 | xmlDetectEncoding(ctxt); |
11898 | | |
11899 | | /* |
11900 | | * Parse a possible text declaration first |
11901 | | */ |
11902 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
11903 | 0 | (IS_BLANK_CH(NXT(5)))) { |
11904 | 0 | xmlParseTextDecl(ctxt); |
11905 | | /* |
11906 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
11907 | | */ |
11908 | 0 | if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) && |
11909 | 0 | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
11910 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
11911 | 0 | "Version mismatch between document and " |
11912 | 0 | "entity\n"); |
11913 | 0 | } |
11914 | 0 | } |
11915 | 0 | } |
11916 | | |
11917 | 861 | xmlParseContentInternal(ctxt); |
11918 | | |
11919 | 861 | if (ctxt->input->cur < ctxt->input->end) |
11920 | 101 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
11921 | | |
11922 | 861 | if ((ctxt->wellFormed) || |
11923 | 861 | ((ctxt->recovery) && (ctxt->errNo != XML_ERR_NO_MEMORY))) { |
11924 | 456 | if (root != NULL) { |
11925 | 456 | xmlNodePtr cur; |
11926 | | |
11927 | | /* |
11928 | | * Unlink newly created node list. |
11929 | | */ |
11930 | 456 | list = root->children; |
11931 | 456 | root->children = NULL; |
11932 | 456 | root->last = NULL; |
11933 | 1.41k | for (cur = list; cur != NULL; cur = cur->next) |
11934 | 956 | cur->parent = NULL; |
11935 | 456 | } |
11936 | 456 | } |
11937 | | |
11938 | | /* |
11939 | | * Read the rest of the stream in case of errors. We want |
11940 | | * to account for the whole entity size. |
11941 | | */ |
11942 | 861 | do { |
11943 | 861 | ctxt->input->cur = ctxt->input->end; |
11944 | 861 | xmlParserShrink(ctxt); |
11945 | 861 | result = xmlParserGrow(ctxt); |
11946 | 861 | } while (result > 0); |
11947 | | |
11948 | 861 | if (buildTree) |
11949 | 861 | nodePop(ctxt); |
11950 | | |
11951 | 861 | namePop(ctxt); |
11952 | 861 | spacePop(ctxt); |
11953 | | |
11954 | | /* xmlPopInput would free the stream */ |
11955 | 861 | inputPop(ctxt); |
11956 | | |
11957 | 863 | error: |
11958 | 863 | xmlFreeNode(root); |
11959 | | |
11960 | 863 | return(list); |
11961 | 861 | } |
11962 | | |
11963 | | static void |
11964 | 884 | xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) { |
11965 | 884 | xmlParserInputPtr input; |
11966 | 884 | xmlNodePtr list; |
11967 | 884 | unsigned long consumed; |
11968 | 884 | int isExternal; |
11969 | 884 | int buildTree; |
11970 | 884 | int oldMinNsIndex; |
11971 | 884 | int oldNodelen, oldNodemem; |
11972 | | |
11973 | 884 | isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY); |
11974 | 884 | buildTree = (ctxt->node != NULL); |
11975 | | |
11976 | | /* |
11977 | | * Recursion check |
11978 | | */ |
11979 | 884 | if (ent->flags & XML_ENT_EXPANDING) { |
11980 | 21 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
11981 | 21 | xmlHaltParser(ctxt); |
11982 | 21 | goto error; |
11983 | 21 | } |
11984 | | |
11985 | | /* |
11986 | | * Load entity |
11987 | | */ |
11988 | 863 | input = xmlNewEntityInputStream(ctxt, ent); |
11989 | 863 | if (input == NULL) |
11990 | 0 | goto error; |
11991 | | |
11992 | | /* |
11993 | | * When building a tree, we need to limit the scope of namespace |
11994 | | * declarations, so that entities don't reference xmlNs structs |
11995 | | * from the parent of a reference. |
11996 | | */ |
11997 | 863 | oldMinNsIndex = ctxt->nsdb->minNsIndex; |
11998 | 863 | if (buildTree) |
11999 | 863 | ctxt->nsdb->minNsIndex = ctxt->nsNr; |
12000 | | |
12001 | 863 | oldNodelen = ctxt->nodelen; |
12002 | 863 | oldNodemem = ctxt->nodemem; |
12003 | 863 | ctxt->nodelen = 0; |
12004 | 863 | ctxt->nodemem = 0; |
12005 | | |
12006 | | /* |
12007 | | * Parse content |
12008 | | * |
12009 | | * This initiates a recursive call chain: |
12010 | | * |
12011 | | * - xmlCtxtParseContentInternal |
12012 | | * - xmlParseContentInternal |
12013 | | * - xmlParseReference |
12014 | | * - xmlCtxtParseEntity |
12015 | | * |
12016 | | * The nesting depth is limited by the maximum number of inputs, |
12017 | | * see xmlPushInput. |
12018 | | * |
12019 | | * It's possible to make this non-recursive (minNsIndex must be |
12020 | | * stored in the input struct) at the expense of code readability. |
12021 | | */ |
12022 | | |
12023 | 863 | ent->flags |= XML_ENT_EXPANDING; |
12024 | | |
12025 | 863 | list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree); |
12026 | | |
12027 | 863 | ent->flags &= ~XML_ENT_EXPANDING; |
12028 | | |
12029 | 863 | ctxt->nsdb->minNsIndex = oldMinNsIndex; |
12030 | 863 | ctxt->nodelen = oldNodelen; |
12031 | 863 | ctxt->nodemem = oldNodemem; |
12032 | | |
12033 | | /* |
12034 | | * Entity size accounting |
12035 | | */ |
12036 | 863 | consumed = input->consumed; |
12037 | 863 | xmlSaturatedAddSizeT(&consumed, input->end - input->base); |
12038 | | |
12039 | 863 | if ((ent->flags & XML_ENT_CHECKED) == 0) |
12040 | 514 | xmlSaturatedAdd(&ent->expandedSize, consumed); |
12041 | | |
12042 | 863 | if ((ent->flags & XML_ENT_PARSED) == 0) { |
12043 | 514 | if (isExternal) |
12044 | 0 | xmlSaturatedAdd(&ctxt->sizeentities, consumed); |
12045 | | |
12046 | 514 | ent->children = list; |
12047 | | |
12048 | 1.47k | while (list != NULL) { |
12049 | 956 | list->parent = (xmlNodePtr) ent; |
12050 | 956 | if (list->next == NULL) |
12051 | 110 | ent->last = list; |
12052 | 956 | list = list->next; |
12053 | 956 | } |
12054 | 514 | } else { |
12055 | 349 | xmlFreeNodeList(list); |
12056 | 349 | } |
12057 | | |
12058 | 863 | xmlFreeInputStream(input); |
12059 | | |
12060 | 884 | error: |
12061 | 884 | ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED; |
12062 | 884 | } |
12063 | | |
12064 | | /** |
12065 | | * xmlParseCtxtExternalEntity: |
12066 | | * @ctxt: the existing parsing context |
12067 | | * @URL: the URL for the entity to load |
12068 | | * @ID: the System ID for the entity to load |
12069 | | * @listOut: the return value for the set of parsed nodes |
12070 | | * |
12071 | | * Parse an external general entity within an existing parsing context |
12072 | | * An external general parsed entity is well-formed if it matches the |
12073 | | * production labeled extParsedEnt. |
12074 | | * |
12075 | | * [78] extParsedEnt ::= TextDecl? content |
12076 | | * |
12077 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
12078 | | * the parser error code otherwise |
12079 | | */ |
12080 | | |
12081 | | int |
12082 | | xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctxt, const xmlChar *URL, |
12083 | 0 | const xmlChar *ID, xmlNodePtr *listOut) { |
12084 | 0 | xmlParserInputPtr input; |
12085 | 0 | xmlNodePtr list; |
12086 | |
|
12087 | 0 | if (listOut != NULL) |
12088 | 0 | *listOut = NULL; |
12089 | |
|
12090 | 0 | if (ctxt == NULL) |
12091 | 0 | return(XML_ERR_ARGUMENT); |
12092 | | |
12093 | 0 | input = xmlLoadResource(ctxt, (char *) URL, (char *) ID, |
12094 | 0 | XML_RESOURCE_GENERAL_ENTITY); |
12095 | 0 | if (input == NULL) |
12096 | 0 | return(ctxt->errNo); |
12097 | | |
12098 | 0 | xmlCtxtInitializeLate(ctxt); |
12099 | |
|
12100 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1); |
12101 | 0 | if (listOut != NULL) |
12102 | 0 | *listOut = list; |
12103 | 0 | else |
12104 | 0 | xmlFreeNodeList(list); |
12105 | |
|
12106 | 0 | xmlFreeInputStream(input); |
12107 | 0 | return(ctxt->errNo); |
12108 | 0 | } |
12109 | | |
12110 | | #ifdef LIBXML_SAX1_ENABLED |
12111 | | /** |
12112 | | * xmlParseExternalEntity: |
12113 | | * @doc: the document the chunk pertains to |
12114 | | * @sax: the SAX handler block (possibly NULL) |
12115 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
12116 | | * @depth: Used for loop detection, use 0 |
12117 | | * @URL: the URL for the entity to load |
12118 | | * @ID: the System ID for the entity to load |
12119 | | * @list: the return value for the set of parsed nodes |
12120 | | * |
12121 | | * DEPRECATED: Use xmlParseCtxtExternalEntity. |
12122 | | * |
12123 | | * Parse an external general entity |
12124 | | * An external general parsed entity is well-formed if it matches the |
12125 | | * production labeled extParsedEnt. |
12126 | | * |
12127 | | * [78] extParsedEnt ::= TextDecl? content |
12128 | | * |
12129 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
12130 | | * the parser error code otherwise |
12131 | | */ |
12132 | | |
12133 | | int |
12134 | | xmlParseExternalEntity(xmlDocPtr doc, xmlSAXHandlerPtr sax, void *user_data, |
12135 | 0 | int depth, const xmlChar *URL, const xmlChar *ID, xmlNodePtr *list) { |
12136 | 0 | xmlParserCtxtPtr ctxt; |
12137 | 0 | int ret; |
12138 | |
|
12139 | 0 | if (list != NULL) |
12140 | 0 | *list = NULL; |
12141 | |
|
12142 | 0 | if (doc == NULL) |
12143 | 0 | return(XML_ERR_ARGUMENT); |
12144 | | |
12145 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
12146 | 0 | if (ctxt == NULL) |
12147 | 0 | return(XML_ERR_NO_MEMORY); |
12148 | | |
12149 | 0 | ctxt->depth = depth; |
12150 | 0 | ctxt->myDoc = doc; |
12151 | 0 | ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list); |
12152 | |
|
12153 | 0 | xmlFreeParserCtxt(ctxt); |
12154 | 0 | return(ret); |
12155 | 0 | } |
12156 | | |
12157 | | /** |
12158 | | * xmlParseBalancedChunkMemory: |
12159 | | * @doc: the document the chunk pertains to (must not be NULL) |
12160 | | * @sax: the SAX handler block (possibly NULL) |
12161 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
12162 | | * @depth: Used for loop detection, use 0 |
12163 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
12164 | | * @lst: the return value for the set of parsed nodes |
12165 | | * |
12166 | | * Parse a well-balanced chunk of an XML document |
12167 | | * called by the parser |
12168 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
12169 | | * the content production in the XML grammar: |
12170 | | * |
12171 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
12172 | | * |
12173 | | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
12174 | | * the parser error code otherwise |
12175 | | */ |
12176 | | |
12177 | | int |
12178 | | xmlParseBalancedChunkMemory(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
12179 | 0 | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst) { |
12180 | 0 | return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data, |
12181 | 0 | depth, string, lst, 0 ); |
12182 | 0 | } |
12183 | | #endif /* LIBXML_SAX1_ENABLED */ |
12184 | | |
12185 | | /** |
12186 | | * xmlCtxtParseContent: |
12187 | | * @ctxt: parser context |
12188 | | * @input: parser input |
12189 | | * @node: target node or document |
12190 | | * @hasTextDecl: whether to parse text declaration |
12191 | | * |
12192 | | * Parse a well-balanced chunk of XML matching the 'content' production. |
12193 | | * |
12194 | | * Namespaces in scope of @node and entities of @node's document are |
12195 | | * recognized. When validating, the DTD of @node's document is used. |
12196 | | * |
12197 | | * Always consumes @input even in error case. |
12198 | | * |
12199 | | * Available since 2.14.0. |
12200 | | * |
12201 | | * Returns a node list or NULL in case of error. |
12202 | | */ |
12203 | | xmlNodePtr |
12204 | | xmlCtxtParseContent(xmlParserCtxtPtr ctxt, xmlParserInputPtr input, |
12205 | 0 | xmlNodePtr node, int hasTextDecl) { |
12206 | 0 | xmlDocPtr doc; |
12207 | 0 | xmlNodePtr cur, list = NULL; |
12208 | 0 | int nsnr = 0; |
12209 | 0 | xmlDictPtr oldDict; |
12210 | 0 | int oldOptions, oldDictNames, oldLoadSubset; |
12211 | |
|
12212 | 0 | if ((ctxt == NULL) || (input == NULL) || (node == NULL)) { |
12213 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12214 | 0 | goto exit; |
12215 | 0 | } |
12216 | | |
12217 | 0 | doc = node->doc; |
12218 | 0 | if (doc == NULL) { |
12219 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12220 | 0 | goto exit; |
12221 | 0 | } |
12222 | | |
12223 | 0 | switch (node->type) { |
12224 | 0 | case XML_ELEMENT_NODE: |
12225 | 0 | case XML_DOCUMENT_NODE: |
12226 | 0 | case XML_HTML_DOCUMENT_NODE: |
12227 | 0 | break; |
12228 | | |
12229 | 0 | case XML_ATTRIBUTE_NODE: |
12230 | 0 | case XML_TEXT_NODE: |
12231 | 0 | case XML_CDATA_SECTION_NODE: |
12232 | 0 | case XML_ENTITY_REF_NODE: |
12233 | 0 | case XML_PI_NODE: |
12234 | 0 | case XML_COMMENT_NODE: |
12235 | 0 | for (cur = node->parent; cur != NULL; cur = node->parent) { |
12236 | 0 | if ((cur->type == XML_ELEMENT_NODE) || |
12237 | 0 | (cur->type == XML_DOCUMENT_NODE) || |
12238 | 0 | (cur->type == XML_HTML_DOCUMENT_NODE)) { |
12239 | 0 | node = cur; |
12240 | 0 | break; |
12241 | 0 | } |
12242 | 0 | } |
12243 | 0 | break; |
12244 | | |
12245 | 0 | default: |
12246 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12247 | 0 | goto exit; |
12248 | 0 | } |
12249 | | |
12250 | 0 | #ifdef LIBXML_HTML_ENABLED |
12251 | 0 | if (ctxt->html) |
12252 | 0 | htmlCtxtReset(ctxt); |
12253 | 0 | else |
12254 | 0 | #endif |
12255 | 0 | xmlCtxtReset(ctxt); |
12256 | |
|
12257 | 0 | oldDict = ctxt->dict; |
12258 | 0 | oldOptions = ctxt->options; |
12259 | 0 | oldDictNames = ctxt->dictNames; |
12260 | 0 | oldLoadSubset = ctxt->loadsubset; |
12261 | | |
12262 | | /* |
12263 | | * Use input doc's dict if present, else assure XML_PARSE_NODICT is set. |
12264 | | */ |
12265 | 0 | if (doc->dict != NULL) { |
12266 | 0 | ctxt->dict = doc->dict; |
12267 | 0 | } else { |
12268 | 0 | ctxt->options |= XML_PARSE_NODICT; |
12269 | 0 | ctxt->dictNames = 0; |
12270 | 0 | } |
12271 | | |
12272 | | /* |
12273 | | * Disable IDs |
12274 | | */ |
12275 | 0 | ctxt->loadsubset |= XML_SKIP_IDS; |
12276 | |
|
12277 | 0 | ctxt->myDoc = doc; |
12278 | |
|
12279 | 0 | #ifdef LIBXML_HTML_ENABLED |
12280 | 0 | if (ctxt->html) { |
12281 | | /* |
12282 | | * When parsing in context, it makes no sense to add implied |
12283 | | * elements like html/body/etc... |
12284 | | */ |
12285 | 0 | ctxt->options |= HTML_PARSE_NOIMPLIED; |
12286 | |
|
12287 | 0 | list = htmlCtxtParseContentInternal(ctxt, input); |
12288 | 0 | } else |
12289 | 0 | #endif |
12290 | 0 | { |
12291 | 0 | xmlCtxtInitializeLate(ctxt); |
12292 | | |
12293 | | /* |
12294 | | * This hack lowers the error level of undeclared entities |
12295 | | * from XML_ERR_FATAL (well-formedness error) to XML_ERR_ERROR |
12296 | | * or XML_ERR_WARNING. |
12297 | | */ |
12298 | 0 | ctxt->hasExternalSubset = 1; |
12299 | | |
12300 | | /* |
12301 | | * initialize the SAX2 namespaces stack |
12302 | | */ |
12303 | 0 | cur = node; |
12304 | 0 | while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) { |
12305 | 0 | xmlNsPtr ns = cur->nsDef; |
12306 | 0 | xmlHashedString hprefix, huri; |
12307 | |
|
12308 | 0 | while (ns != NULL) { |
12309 | 0 | hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1); |
12310 | 0 | huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1); |
12311 | 0 | if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0) |
12312 | 0 | nsnr++; |
12313 | 0 | ns = ns->next; |
12314 | 0 | } |
12315 | 0 | cur = cur->parent; |
12316 | 0 | } |
12317 | |
|
12318 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1); |
12319 | |
|
12320 | 0 | if (nsnr > 0) |
12321 | 0 | xmlParserNsPop(ctxt, nsnr); |
12322 | 0 | } |
12323 | |
|
12324 | 0 | ctxt->dict = oldDict; |
12325 | 0 | ctxt->options = oldOptions; |
12326 | 0 | ctxt->dictNames = oldDictNames; |
12327 | 0 | ctxt->loadsubset = oldLoadSubset; |
12328 | 0 | ctxt->myDoc = NULL; |
12329 | 0 | ctxt->node = NULL; |
12330 | |
|
12331 | 0 | exit: |
12332 | 0 | xmlFreeInputStream(input); |
12333 | 0 | return(list); |
12334 | 0 | } |
12335 | | |
12336 | | /** |
12337 | | * xmlParseInNodeContext: |
12338 | | * @node: the context node |
12339 | | * @data: the input string |
12340 | | * @datalen: the input string length in bytes |
12341 | | * @options: a combination of xmlParserOption |
12342 | | * @listOut: the return value for the set of parsed nodes |
12343 | | * |
12344 | | * Parse a well-balanced chunk of an XML document |
12345 | | * within the context (DTD, namespaces, etc ...) of the given node. |
12346 | | * |
12347 | | * The allowed sequence for the data is a Well Balanced Chunk defined by |
12348 | | * the content production in the XML grammar: |
12349 | | * |
12350 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
12351 | | * |
12352 | | * This function assumes the encoding of @node's document which is |
12353 | | * typically not what you want. A better alternative is |
12354 | | * xmlCtxtParseContent. |
12355 | | * |
12356 | | * Returns XML_ERR_OK if the chunk is well balanced, and the parser |
12357 | | * error code otherwise |
12358 | | */ |
12359 | | xmlParserErrors |
12360 | | xmlParseInNodeContext(xmlNodePtr node, const char *data, int datalen, |
12361 | 0 | int options, xmlNodePtr *listOut) { |
12362 | 0 | xmlParserCtxtPtr ctxt; |
12363 | 0 | xmlParserInputPtr input; |
12364 | 0 | xmlDocPtr doc; |
12365 | 0 | xmlNodePtr list; |
12366 | 0 | xmlParserErrors ret; |
12367 | |
|
12368 | 0 | if (listOut == NULL) |
12369 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12370 | 0 | *listOut = NULL; |
12371 | |
|
12372 | 0 | if ((node == NULL) || (data == NULL) || (datalen < 0)) |
12373 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12374 | | |
12375 | 0 | doc = node->doc; |
12376 | 0 | if (doc == NULL) |
12377 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12378 | | |
12379 | 0 | #ifdef LIBXML_HTML_ENABLED |
12380 | 0 | if (doc->type == XML_HTML_DOCUMENT_NODE) { |
12381 | 0 | ctxt = htmlNewParserCtxt(); |
12382 | 0 | } |
12383 | 0 | else |
12384 | 0 | #endif |
12385 | 0 | ctxt = xmlNewParserCtxt(); |
12386 | |
|
12387 | 0 | if (ctxt == NULL) |
12388 | 0 | return(XML_ERR_NO_MEMORY); |
12389 | | |
12390 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen, |
12391 | 0 | (const char *) doc->encoding, |
12392 | 0 | XML_INPUT_BUF_STATIC); |
12393 | 0 | if (input == NULL) { |
12394 | 0 | xmlFreeParserCtxt(ctxt); |
12395 | 0 | return(XML_ERR_NO_MEMORY); |
12396 | 0 | } |
12397 | | |
12398 | 0 | xmlCtxtUseOptions(ctxt, options); |
12399 | |
|
12400 | 0 | list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0); |
12401 | |
|
12402 | 0 | if (list == NULL) { |
12403 | 0 | ret = ctxt->errNo; |
12404 | 0 | if (ret == XML_ERR_ARGUMENT) |
12405 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
12406 | 0 | } else { |
12407 | 0 | ret = XML_ERR_OK; |
12408 | 0 | *listOut = list; |
12409 | 0 | } |
12410 | |
|
12411 | 0 | xmlFreeParserCtxt(ctxt); |
12412 | |
|
12413 | 0 | return(ret); |
12414 | 0 | } |
12415 | | |
12416 | | #ifdef LIBXML_SAX1_ENABLED |
12417 | | /** |
12418 | | * xmlParseBalancedChunkMemoryRecover: |
12419 | | * @doc: the document the chunk pertains to (must not be NULL) |
12420 | | * @sax: the SAX handler block (possibly NULL) |
12421 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
12422 | | * @depth: Used for loop detection, use 0 |
12423 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
12424 | | * @listOut: the return value for the set of parsed nodes |
12425 | | * @recover: return nodes even if the data is broken (use 0) |
12426 | | * |
12427 | | * Parse a well-balanced chunk of an XML document |
12428 | | * |
12429 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
12430 | | * the content production in the XML grammar: |
12431 | | * |
12432 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
12433 | | * |
12434 | | * Returns 0 if the chunk is well balanced, or thehe parser error code |
12435 | | * otherwise. |
12436 | | * |
12437 | | * In case recover is set to 1, the nodelist will not be empty even if |
12438 | | * the parsed chunk is not well balanced, assuming the parsing succeeded to |
12439 | | * some extent. |
12440 | | */ |
12441 | | int |
12442 | | xmlParseBalancedChunkMemoryRecover(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
12443 | | void *user_data, int depth, const xmlChar *string, xmlNodePtr *listOut, |
12444 | 0 | int recover) { |
12445 | 0 | xmlParserCtxtPtr ctxt; |
12446 | 0 | xmlParserInputPtr input; |
12447 | 0 | xmlNodePtr list; |
12448 | 0 | int ret; |
12449 | |
|
12450 | 0 | if (listOut != NULL) |
12451 | 0 | *listOut = NULL; |
12452 | |
|
12453 | 0 | if (string == NULL) |
12454 | 0 | return(XML_ERR_ARGUMENT); |
12455 | | |
12456 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
12457 | 0 | if (ctxt == NULL) |
12458 | 0 | return(XML_ERR_NO_MEMORY); |
12459 | | |
12460 | 0 | xmlCtxtInitializeLate(ctxt); |
12461 | |
|
12462 | 0 | ctxt->depth = depth; |
12463 | 0 | ctxt->myDoc = doc; |
12464 | 0 | if (recover) { |
12465 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12466 | 0 | ctxt->recovery = 1; |
12467 | 0 | } |
12468 | |
|
12469 | 0 | input = xmlNewStringInputStream(ctxt, string); |
12470 | 0 | if (input == NULL) { |
12471 | 0 | ret = ctxt->errNo; |
12472 | 0 | goto error; |
12473 | 0 | } |
12474 | | |
12475 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1); |
12476 | 0 | if (listOut != NULL) |
12477 | 0 | *listOut = list; |
12478 | 0 | else |
12479 | 0 | xmlFreeNodeList(list); |
12480 | |
|
12481 | 0 | if (!ctxt->wellFormed) |
12482 | 0 | ret = ctxt->errNo; |
12483 | 0 | else |
12484 | 0 | ret = XML_ERR_OK; |
12485 | |
|
12486 | 0 | error: |
12487 | 0 | xmlFreeInputStream(input); |
12488 | 0 | xmlFreeParserCtxt(ctxt); |
12489 | 0 | return(ret); |
12490 | 0 | } |
12491 | | |
12492 | | /** |
12493 | | * xmlSAXParseEntity: |
12494 | | * @sax: the SAX handler block |
12495 | | * @filename: the filename |
12496 | | * |
12497 | | * DEPRECATED: Don't use. |
12498 | | * |
12499 | | * parse an XML external entity out of context and build a tree. |
12500 | | * It use the given SAX function block to handle the parsing callback. |
12501 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12502 | | * |
12503 | | * [78] extParsedEnt ::= TextDecl? content |
12504 | | * |
12505 | | * This correspond to a "Well Balanced" chunk |
12506 | | * |
12507 | | * Returns the resulting document tree |
12508 | | */ |
12509 | | |
12510 | | xmlDocPtr |
12511 | 0 | xmlSAXParseEntity(xmlSAXHandlerPtr sax, const char *filename) { |
12512 | 0 | xmlDocPtr ret; |
12513 | 0 | xmlParserCtxtPtr ctxt; |
12514 | |
|
12515 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
12516 | 0 | if (ctxt == NULL) { |
12517 | 0 | return(NULL); |
12518 | 0 | } |
12519 | 0 | if (sax != NULL) { |
12520 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
12521 | 0 | *ctxt->sax = *sax; |
12522 | 0 | } else { |
12523 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
12524 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12525 | 0 | } |
12526 | 0 | ctxt->userData = NULL; |
12527 | 0 | } |
12528 | |
|
12529 | 0 | xmlParseExtParsedEnt(ctxt); |
12530 | |
|
12531 | 0 | if (ctxt->wellFormed) { |
12532 | 0 | ret = ctxt->myDoc; |
12533 | 0 | } else { |
12534 | 0 | ret = NULL; |
12535 | 0 | xmlFreeDoc(ctxt->myDoc); |
12536 | 0 | } |
12537 | |
|
12538 | 0 | xmlFreeParserCtxt(ctxt); |
12539 | |
|
12540 | 0 | return(ret); |
12541 | 0 | } |
12542 | | |
12543 | | /** |
12544 | | * xmlParseEntity: |
12545 | | * @filename: the filename |
12546 | | * |
12547 | | * parse an XML external entity out of context and build a tree. |
12548 | | * |
12549 | | * [78] extParsedEnt ::= TextDecl? content |
12550 | | * |
12551 | | * This correspond to a "Well Balanced" chunk |
12552 | | * |
12553 | | * Returns the resulting document tree |
12554 | | */ |
12555 | | |
12556 | | xmlDocPtr |
12557 | 0 | xmlParseEntity(const char *filename) { |
12558 | 0 | return(xmlSAXParseEntity(NULL, filename)); |
12559 | 0 | } |
12560 | | #endif /* LIBXML_SAX1_ENABLED */ |
12561 | | |
12562 | | /** |
12563 | | * xmlCreateEntityParserCtxt: |
12564 | | * @URL: the entity URL |
12565 | | * @ID: the entity PUBLIC ID |
12566 | | * @base: a possible base for the target URI |
12567 | | * |
12568 | | * DEPRECATED: Don't use. |
12569 | | * |
12570 | | * Create a parser context for an external entity |
12571 | | * Automatic support for ZLIB/Compress compressed document is provided |
12572 | | * by default if found at compile-time. |
12573 | | * |
12574 | | * Returns the new parser context or NULL |
12575 | | */ |
12576 | | xmlParserCtxtPtr |
12577 | | xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID, |
12578 | 0 | const xmlChar *base) { |
12579 | 0 | xmlParserCtxtPtr ctxt; |
12580 | 0 | xmlParserInputPtr input; |
12581 | 0 | xmlChar *uri = NULL; |
12582 | |
|
12583 | 0 | ctxt = xmlNewParserCtxt(); |
12584 | 0 | if (ctxt == NULL) |
12585 | 0 | return(NULL); |
12586 | | |
12587 | 0 | if (base != NULL) { |
12588 | 0 | if (xmlBuildURISafe(URL, base, &uri) < 0) |
12589 | 0 | goto error; |
12590 | 0 | if (uri != NULL) |
12591 | 0 | URL = uri; |
12592 | 0 | } |
12593 | | |
12594 | 0 | input = xmlLoadResource(ctxt, (char *) URL, (char *) ID, |
12595 | 0 | XML_RESOURCE_UNKNOWN); |
12596 | 0 | if (input == NULL) |
12597 | 0 | goto error; |
12598 | | |
12599 | 0 | if (inputPush(ctxt, input) < 0) { |
12600 | 0 | xmlFreeInputStream(input); |
12601 | 0 | goto error; |
12602 | 0 | } |
12603 | | |
12604 | 0 | xmlFree(uri); |
12605 | 0 | return(ctxt); |
12606 | | |
12607 | 0 | error: |
12608 | 0 | xmlFree(uri); |
12609 | 0 | xmlFreeParserCtxt(ctxt); |
12610 | 0 | return(NULL); |
12611 | 0 | } |
12612 | | |
12613 | | /************************************************************************ |
12614 | | * * |
12615 | | * Front ends when parsing from a file * |
12616 | | * * |
12617 | | ************************************************************************/ |
12618 | | |
12619 | | /** |
12620 | | * xmlCreateURLParserCtxt: |
12621 | | * @filename: the filename or URL |
12622 | | * @options: a combination of xmlParserOption |
12623 | | * |
12624 | | * DEPRECATED: Use xmlNewParserCtxt and xmlCtxtReadFile. |
12625 | | * |
12626 | | * Create a parser context for a file or URL content. |
12627 | | * Automatic support for ZLIB/Compress compressed document is provided |
12628 | | * by default if found at compile-time and for file accesses |
12629 | | * |
12630 | | * Returns the new parser context or NULL |
12631 | | */ |
12632 | | xmlParserCtxtPtr |
12633 | | xmlCreateURLParserCtxt(const char *filename, int options) |
12634 | 0 | { |
12635 | 0 | xmlParserCtxtPtr ctxt; |
12636 | 0 | xmlParserInputPtr input; |
12637 | |
|
12638 | 0 | ctxt = xmlNewParserCtxt(); |
12639 | 0 | if (ctxt == NULL) |
12640 | 0 | return(NULL); |
12641 | | |
12642 | 0 | xmlCtxtUseOptions(ctxt, options); |
12643 | 0 | ctxt->linenumbers = 1; |
12644 | |
|
12645 | 0 | input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT); |
12646 | 0 | if (input == NULL) { |
12647 | 0 | xmlFreeParserCtxt(ctxt); |
12648 | 0 | return(NULL); |
12649 | 0 | } |
12650 | 0 | if (inputPush(ctxt, input) < 0) { |
12651 | 0 | xmlFreeInputStream(input); |
12652 | 0 | xmlFreeParserCtxt(ctxt); |
12653 | 0 | return(NULL); |
12654 | 0 | } |
12655 | | |
12656 | 0 | return(ctxt); |
12657 | 0 | } |
12658 | | |
12659 | | /** |
12660 | | * xmlCreateFileParserCtxt: |
12661 | | * @filename: the filename |
12662 | | * |
12663 | | * DEPRECATED: Use xmlNewParserCtxt and xmlCtxtReadFile. |
12664 | | * |
12665 | | * Create a parser context for a file content. |
12666 | | * Automatic support for ZLIB/Compress compressed document is provided |
12667 | | * by default if found at compile-time. |
12668 | | * |
12669 | | * Returns the new parser context or NULL |
12670 | | */ |
12671 | | xmlParserCtxtPtr |
12672 | | xmlCreateFileParserCtxt(const char *filename) |
12673 | 0 | { |
12674 | 0 | return(xmlCreateURLParserCtxt(filename, 0)); |
12675 | 0 | } |
12676 | | |
12677 | | #ifdef LIBXML_SAX1_ENABLED |
12678 | | /** |
12679 | | * xmlSAXParseFileWithData: |
12680 | | * @sax: the SAX handler block |
12681 | | * @filename: the filename |
12682 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
12683 | | * documents |
12684 | | * @data: the userdata |
12685 | | * |
12686 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
12687 | | * |
12688 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12689 | | * compressed document is provided by default if found at compile-time. |
12690 | | * It use the given SAX function block to handle the parsing callback. |
12691 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12692 | | * |
12693 | | * User data (void *) is stored within the parser context in the |
12694 | | * context's _private member, so it is available nearly everywhere in libxml |
12695 | | * |
12696 | | * Returns the resulting document tree |
12697 | | */ |
12698 | | |
12699 | | xmlDocPtr |
12700 | | xmlSAXParseFileWithData(xmlSAXHandlerPtr sax, const char *filename, |
12701 | 0 | int recovery, void *data) { |
12702 | 0 | xmlDocPtr ret; |
12703 | 0 | xmlParserCtxtPtr ctxt; |
12704 | 0 | xmlParserInputPtr input; |
12705 | |
|
12706 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12707 | 0 | if (ctxt == NULL) |
12708 | 0 | return(NULL); |
12709 | | |
12710 | 0 | if (data != NULL) |
12711 | 0 | ctxt->_private = data; |
12712 | |
|
12713 | 0 | if (recovery) { |
12714 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12715 | 0 | ctxt->recovery = 1; |
12716 | 0 | } |
12717 | |
|
12718 | 0 | if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0)) |
12719 | 0 | input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0); |
12720 | 0 | else |
12721 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0); |
12722 | |
|
12723 | 0 | ret = xmlCtxtParseDocument(ctxt, input); |
12724 | |
|
12725 | 0 | xmlFreeParserCtxt(ctxt); |
12726 | 0 | return(ret); |
12727 | 0 | } |
12728 | | |
12729 | | /** |
12730 | | * xmlSAXParseFile: |
12731 | | * @sax: the SAX handler block |
12732 | | * @filename: the filename |
12733 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
12734 | | * documents |
12735 | | * |
12736 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
12737 | | * |
12738 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12739 | | * compressed document is provided by default if found at compile-time. |
12740 | | * It use the given SAX function block to handle the parsing callback. |
12741 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12742 | | * |
12743 | | * Returns the resulting document tree |
12744 | | */ |
12745 | | |
12746 | | xmlDocPtr |
12747 | | xmlSAXParseFile(xmlSAXHandlerPtr sax, const char *filename, |
12748 | 0 | int recovery) { |
12749 | 0 | return(xmlSAXParseFileWithData(sax,filename,recovery,NULL)); |
12750 | 0 | } |
12751 | | |
12752 | | /** |
12753 | | * xmlRecoverDoc: |
12754 | | * @cur: a pointer to an array of xmlChar |
12755 | | * |
12756 | | * DEPRECATED: Use xmlReadDoc with XML_PARSE_RECOVER. |
12757 | | * |
12758 | | * parse an XML in-memory document and build a tree. |
12759 | | * In the case the document is not Well Formed, a attempt to build a |
12760 | | * tree is tried anyway |
12761 | | * |
12762 | | * Returns the resulting document tree or NULL in case of failure |
12763 | | */ |
12764 | | |
12765 | | xmlDocPtr |
12766 | 0 | xmlRecoverDoc(const xmlChar *cur) { |
12767 | 0 | return(xmlSAXParseDoc(NULL, cur, 1)); |
12768 | 0 | } |
12769 | | |
12770 | | /** |
12771 | | * xmlParseFile: |
12772 | | * @filename: the filename |
12773 | | * |
12774 | | * DEPRECATED: Use xmlReadFile. |
12775 | | * |
12776 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12777 | | * compressed document is provided by default if found at compile-time. |
12778 | | * |
12779 | | * Returns the resulting document tree if the file was wellformed, |
12780 | | * NULL otherwise. |
12781 | | */ |
12782 | | |
12783 | | xmlDocPtr |
12784 | 0 | xmlParseFile(const char *filename) { |
12785 | 0 | return(xmlSAXParseFile(NULL, filename, 0)); |
12786 | 0 | } |
12787 | | |
12788 | | /** |
12789 | | * xmlRecoverFile: |
12790 | | * @filename: the filename |
12791 | | * |
12792 | | * DEPRECATED: Use xmlReadFile with XML_PARSE_RECOVER. |
12793 | | * |
12794 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12795 | | * compressed document is provided by default if found at compile-time. |
12796 | | * In the case the document is not Well Formed, it attempts to build |
12797 | | * a tree anyway |
12798 | | * |
12799 | | * Returns the resulting document tree or NULL in case of failure |
12800 | | */ |
12801 | | |
12802 | | xmlDocPtr |
12803 | 0 | xmlRecoverFile(const char *filename) { |
12804 | 0 | return(xmlSAXParseFile(NULL, filename, 1)); |
12805 | 0 | } |
12806 | | |
12807 | | |
12808 | | /** |
12809 | | * xmlSetupParserForBuffer: |
12810 | | * @ctxt: an XML parser context |
12811 | | * @buffer: a xmlChar * buffer |
12812 | | * @filename: a file name |
12813 | | * |
12814 | | * DEPRECATED: Don't use. |
12815 | | * |
12816 | | * Setup the parser context to parse a new buffer; Clears any prior |
12817 | | * contents from the parser context. The buffer parameter must not be |
12818 | | * NULL, but the filename parameter can be |
12819 | | */ |
12820 | | void |
12821 | | xmlSetupParserForBuffer(xmlParserCtxtPtr ctxt, const xmlChar* buffer, |
12822 | | const char* filename) |
12823 | 0 | { |
12824 | 0 | xmlParserInputPtr input; |
12825 | |
|
12826 | 0 | if ((ctxt == NULL) || (buffer == NULL)) |
12827 | 0 | return; |
12828 | | |
12829 | 0 | xmlClearParserCtxt(ctxt); |
12830 | |
|
12831 | 0 | input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer, |
12832 | 0 | NULL, 0); |
12833 | 0 | if (input == NULL) |
12834 | 0 | return; |
12835 | 0 | if (inputPush(ctxt, input) < 0) |
12836 | 0 | xmlFreeInputStream(input); |
12837 | 0 | } |
12838 | | |
12839 | | /** |
12840 | | * xmlSAXUserParseFile: |
12841 | | * @sax: a SAX handler |
12842 | | * @user_data: The user data returned on SAX callbacks |
12843 | | * @filename: a file name |
12844 | | * |
12845 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
12846 | | * |
12847 | | * parse an XML file and call the given SAX handler routines. |
12848 | | * Automatic support for ZLIB/Compress compressed document is provided |
12849 | | * |
12850 | | * Returns 0 in case of success or a error number otherwise |
12851 | | */ |
12852 | | int |
12853 | | xmlSAXUserParseFile(xmlSAXHandlerPtr sax, void *user_data, |
12854 | 0 | const char *filename) { |
12855 | 0 | int ret = 0; |
12856 | 0 | xmlParserCtxtPtr ctxt; |
12857 | |
|
12858 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
12859 | 0 | if (ctxt == NULL) return -1; |
12860 | 0 | if (sax != NULL) { |
12861 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
12862 | 0 | *ctxt->sax = *sax; |
12863 | 0 | } else { |
12864 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
12865 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12866 | 0 | } |
12867 | 0 | ctxt->userData = user_data; |
12868 | 0 | } |
12869 | |
|
12870 | 0 | xmlParseDocument(ctxt); |
12871 | |
|
12872 | 0 | if (ctxt->wellFormed) |
12873 | 0 | ret = 0; |
12874 | 0 | else { |
12875 | 0 | if (ctxt->errNo != 0) |
12876 | 0 | ret = ctxt->errNo; |
12877 | 0 | else |
12878 | 0 | ret = -1; |
12879 | 0 | } |
12880 | 0 | if (ctxt->myDoc != NULL) { |
12881 | 0 | xmlFreeDoc(ctxt->myDoc); |
12882 | 0 | ctxt->myDoc = NULL; |
12883 | 0 | } |
12884 | 0 | xmlFreeParserCtxt(ctxt); |
12885 | |
|
12886 | 0 | return ret; |
12887 | 0 | } |
12888 | | #endif /* LIBXML_SAX1_ENABLED */ |
12889 | | |
12890 | | /************************************************************************ |
12891 | | * * |
12892 | | * Front ends when parsing from memory * |
12893 | | * * |
12894 | | ************************************************************************/ |
12895 | | |
12896 | | /** |
12897 | | * xmlCreateMemoryParserCtxt: |
12898 | | * @buffer: a pointer to a char array |
12899 | | * @size: the size of the array |
12900 | | * |
12901 | | * Create a parser context for an XML in-memory document. The input buffer |
12902 | | * must not contain a terminating null byte. |
12903 | | * |
12904 | | * Returns the new parser context or NULL |
12905 | | */ |
12906 | | xmlParserCtxtPtr |
12907 | 0 | xmlCreateMemoryParserCtxt(const char *buffer, int size) { |
12908 | 0 | xmlParserCtxtPtr ctxt; |
12909 | 0 | xmlParserInputPtr input; |
12910 | |
|
12911 | 0 | if (size < 0) |
12912 | 0 | return(NULL); |
12913 | | |
12914 | 0 | ctxt = xmlNewParserCtxt(); |
12915 | 0 | if (ctxt == NULL) |
12916 | 0 | return(NULL); |
12917 | | |
12918 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0); |
12919 | 0 | if (input == NULL) { |
12920 | 0 | xmlFreeParserCtxt(ctxt); |
12921 | 0 | return(NULL); |
12922 | 0 | } |
12923 | 0 | if (inputPush(ctxt, input) < 0) { |
12924 | 0 | xmlFreeInputStream(input); |
12925 | 0 | xmlFreeParserCtxt(ctxt); |
12926 | 0 | return(NULL); |
12927 | 0 | } |
12928 | | |
12929 | 0 | return(ctxt); |
12930 | 0 | } |
12931 | | |
12932 | | #ifdef LIBXML_SAX1_ENABLED |
12933 | | /** |
12934 | | * xmlSAXParseMemoryWithData: |
12935 | | * @sax: the SAX handler block |
12936 | | * @buffer: an pointer to a char array |
12937 | | * @size: the size of the array |
12938 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
12939 | | * documents |
12940 | | * @data: the userdata |
12941 | | * |
12942 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
12943 | | * |
12944 | | * parse an XML in-memory block and use the given SAX function block |
12945 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
12946 | | * DOM tree building routines. |
12947 | | * |
12948 | | * User data (void *) is stored within the parser context in the |
12949 | | * context's _private member, so it is available nearly everywhere in libxml |
12950 | | * |
12951 | | * Returns the resulting document tree |
12952 | | */ |
12953 | | |
12954 | | xmlDocPtr |
12955 | | xmlSAXParseMemoryWithData(xmlSAXHandlerPtr sax, const char *buffer, |
12956 | 0 | int size, int recovery, void *data) { |
12957 | 0 | xmlDocPtr ret; |
12958 | 0 | xmlParserCtxtPtr ctxt; |
12959 | 0 | xmlParserInputPtr input; |
12960 | |
|
12961 | 0 | if (size < 0) |
12962 | 0 | return(NULL); |
12963 | | |
12964 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12965 | 0 | if (ctxt == NULL) |
12966 | 0 | return(NULL); |
12967 | | |
12968 | 0 | if (data != NULL) |
12969 | 0 | ctxt->_private=data; |
12970 | |
|
12971 | 0 | if (recovery) { |
12972 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12973 | 0 | ctxt->recovery = 1; |
12974 | 0 | } |
12975 | |
|
12976 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, |
12977 | 0 | XML_INPUT_BUF_STATIC); |
12978 | |
|
12979 | 0 | ret = xmlCtxtParseDocument(ctxt, input); |
12980 | |
|
12981 | 0 | xmlFreeParserCtxt(ctxt); |
12982 | 0 | return(ret); |
12983 | 0 | } |
12984 | | |
12985 | | /** |
12986 | | * xmlSAXParseMemory: |
12987 | | * @sax: the SAX handler block |
12988 | | * @buffer: an pointer to a char array |
12989 | | * @size: the size of the array |
12990 | | * @recovery: work in recovery mode, i.e. tries to read not Well Formed |
12991 | | * documents |
12992 | | * |
12993 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
12994 | | * |
12995 | | * parse an XML in-memory block and use the given SAX function block |
12996 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
12997 | | * DOM tree building routines. |
12998 | | * |
12999 | | * Returns the resulting document tree |
13000 | | */ |
13001 | | xmlDocPtr |
13002 | | xmlSAXParseMemory(xmlSAXHandlerPtr sax, const char *buffer, |
13003 | 0 | int size, int recovery) { |
13004 | 0 | return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL); |
13005 | 0 | } |
13006 | | |
13007 | | /** |
13008 | | * xmlParseMemory: |
13009 | | * @buffer: an pointer to a char array |
13010 | | * @size: the size of the array |
13011 | | * |
13012 | | * DEPRECATED: Use xmlReadMemory. |
13013 | | * |
13014 | | * parse an XML in-memory block and build a tree. |
13015 | | * |
13016 | | * Returns the resulting document tree |
13017 | | */ |
13018 | | |
13019 | 0 | xmlDocPtr xmlParseMemory(const char *buffer, int size) { |
13020 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 0)); |
13021 | 0 | } |
13022 | | |
13023 | | /** |
13024 | | * xmlRecoverMemory: |
13025 | | * @buffer: an pointer to a char array |
13026 | | * @size: the size of the array |
13027 | | * |
13028 | | * DEPRECATED: Use xmlReadMemory with XML_PARSE_RECOVER. |
13029 | | * |
13030 | | * parse an XML in-memory block and build a tree. |
13031 | | * In the case the document is not Well Formed, an attempt to |
13032 | | * build a tree is tried anyway |
13033 | | * |
13034 | | * Returns the resulting document tree or NULL in case of error |
13035 | | */ |
13036 | | |
13037 | 0 | xmlDocPtr xmlRecoverMemory(const char *buffer, int size) { |
13038 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 1)); |
13039 | 0 | } |
13040 | | |
13041 | | /** |
13042 | | * xmlSAXUserParseMemory: |
13043 | | * @sax: a SAX handler |
13044 | | * @user_data: The user data returned on SAX callbacks |
13045 | | * @buffer: an in-memory XML document input |
13046 | | * @size: the length of the XML document in bytes |
13047 | | * |
13048 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
13049 | | * |
13050 | | * parse an XML in-memory buffer and call the given SAX handler routines. |
13051 | | * |
13052 | | * Returns 0 in case of success or a error number otherwise |
13053 | | */ |
13054 | | int xmlSAXUserParseMemory(xmlSAXHandlerPtr sax, void *user_data, |
13055 | 0 | const char *buffer, int size) { |
13056 | 0 | int ret = 0; |
13057 | 0 | xmlParserCtxtPtr ctxt; |
13058 | |
|
13059 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
13060 | 0 | if (ctxt == NULL) return -1; |
13061 | 0 | if (sax != NULL) { |
13062 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
13063 | 0 | *ctxt->sax = *sax; |
13064 | 0 | } else { |
13065 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
13066 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
13067 | 0 | } |
13068 | 0 | ctxt->userData = user_data; |
13069 | 0 | } |
13070 | |
|
13071 | 0 | xmlParseDocument(ctxt); |
13072 | |
|
13073 | 0 | if (ctxt->wellFormed) |
13074 | 0 | ret = 0; |
13075 | 0 | else { |
13076 | 0 | if (ctxt->errNo != 0) |
13077 | 0 | ret = ctxt->errNo; |
13078 | 0 | else |
13079 | 0 | ret = -1; |
13080 | 0 | } |
13081 | 0 | if (ctxt->myDoc != NULL) { |
13082 | 0 | xmlFreeDoc(ctxt->myDoc); |
13083 | 0 | ctxt->myDoc = NULL; |
13084 | 0 | } |
13085 | 0 | xmlFreeParserCtxt(ctxt); |
13086 | |
|
13087 | 0 | return ret; |
13088 | 0 | } |
13089 | | #endif /* LIBXML_SAX1_ENABLED */ |
13090 | | |
13091 | | /** |
13092 | | * xmlCreateDocParserCtxt: |
13093 | | * @str: a pointer to an array of xmlChar |
13094 | | * |
13095 | | * Creates a parser context for an XML in-memory document. |
13096 | | * |
13097 | | * Returns the new parser context or NULL |
13098 | | */ |
13099 | | xmlParserCtxtPtr |
13100 | 0 | xmlCreateDocParserCtxt(const xmlChar *str) { |
13101 | 0 | xmlParserCtxtPtr ctxt; |
13102 | 0 | xmlParserInputPtr input; |
13103 | |
|
13104 | 0 | ctxt = xmlNewParserCtxt(); |
13105 | 0 | if (ctxt == NULL) |
13106 | 0 | return(NULL); |
13107 | | |
13108 | 0 | input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0); |
13109 | 0 | if (input == NULL) { |
13110 | 0 | xmlFreeParserCtxt(ctxt); |
13111 | 0 | return(NULL); |
13112 | 0 | } |
13113 | 0 | if (inputPush(ctxt, input) < 0) { |
13114 | 0 | xmlFreeInputStream(input); |
13115 | 0 | xmlFreeParserCtxt(ctxt); |
13116 | 0 | return(NULL); |
13117 | 0 | } |
13118 | | |
13119 | 0 | return(ctxt); |
13120 | 0 | } |
13121 | | |
13122 | | #ifdef LIBXML_SAX1_ENABLED |
13123 | | /** |
13124 | | * xmlSAXParseDoc: |
13125 | | * @sax: the SAX handler block |
13126 | | * @cur: a pointer to an array of xmlChar |
13127 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
13128 | | * documents |
13129 | | * |
13130 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadDoc. |
13131 | | * |
13132 | | * parse an XML in-memory document and build a tree. |
13133 | | * It use the given SAX function block to handle the parsing callback. |
13134 | | * If sax is NULL, fallback to the default DOM tree building routines. |
13135 | | * |
13136 | | * Returns the resulting document tree |
13137 | | */ |
13138 | | |
13139 | | xmlDocPtr |
13140 | 0 | xmlSAXParseDoc(xmlSAXHandlerPtr sax, const xmlChar *cur, int recovery) { |
13141 | 0 | xmlDocPtr ret; |
13142 | 0 | xmlParserCtxtPtr ctxt; |
13143 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
13144 | |
|
13145 | 0 | if (cur == NULL) return(NULL); |
13146 | | |
13147 | | |
13148 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
13149 | 0 | if (ctxt == NULL) return(NULL); |
13150 | 0 | if (sax != NULL) { |
13151 | 0 | oldsax = ctxt->sax; |
13152 | 0 | ctxt->sax = sax; |
13153 | 0 | ctxt->userData = NULL; |
13154 | 0 | } |
13155 | |
|
13156 | 0 | xmlParseDocument(ctxt); |
13157 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
13158 | 0 | else { |
13159 | 0 | ret = NULL; |
13160 | 0 | xmlFreeDoc(ctxt->myDoc); |
13161 | 0 | ctxt->myDoc = NULL; |
13162 | 0 | } |
13163 | 0 | if (sax != NULL) |
13164 | 0 | ctxt->sax = oldsax; |
13165 | 0 | xmlFreeParserCtxt(ctxt); |
13166 | |
|
13167 | 0 | return(ret); |
13168 | 0 | } |
13169 | | |
13170 | | /** |
13171 | | * xmlParseDoc: |
13172 | | * @cur: a pointer to an array of xmlChar |
13173 | | * |
13174 | | * DEPRECATED: Use xmlReadDoc. |
13175 | | * |
13176 | | * parse an XML in-memory document and build a tree. |
13177 | | * |
13178 | | * Returns the resulting document tree |
13179 | | */ |
13180 | | |
13181 | | xmlDocPtr |
13182 | 0 | xmlParseDoc(const xmlChar *cur) { |
13183 | 0 | return(xmlSAXParseDoc(NULL, cur, 0)); |
13184 | 0 | } |
13185 | | #endif /* LIBXML_SAX1_ENABLED */ |
13186 | | |
13187 | | /************************************************************************ |
13188 | | * * |
13189 | | * New set (2.6.0) of simpler and more flexible APIs * |
13190 | | * * |
13191 | | ************************************************************************/ |
13192 | | |
13193 | | /** |
13194 | | * DICT_FREE: |
13195 | | * @str: a string |
13196 | | * |
13197 | | * Free a string if it is not owned by the "dict" dictionary in the |
13198 | | * current scope |
13199 | | */ |
13200 | | #define DICT_FREE(str) \ |
13201 | 0 | if ((str) && ((!dict) || \ |
13202 | 0 | (xmlDictOwns(dict, (const xmlChar *)(str)) == 0))) \ |
13203 | 0 | xmlFree((char *)(str)); |
13204 | | |
13205 | | /** |
13206 | | * xmlCtxtReset: |
13207 | | * @ctxt: an XML parser context |
13208 | | * |
13209 | | * Reset a parser context |
13210 | | */ |
13211 | | void |
13212 | | xmlCtxtReset(xmlParserCtxtPtr ctxt) |
13213 | 0 | { |
13214 | 0 | xmlParserInputPtr input; |
13215 | 0 | xmlDictPtr dict; |
13216 | |
|
13217 | 0 | if (ctxt == NULL) |
13218 | 0 | return; |
13219 | | |
13220 | 0 | dict = ctxt->dict; |
13221 | |
|
13222 | 0 | while ((input = inputPop(ctxt)) != NULL) { /* Non consuming */ |
13223 | 0 | xmlFreeInputStream(input); |
13224 | 0 | } |
13225 | 0 | ctxt->inputNr = 0; |
13226 | 0 | ctxt->input = NULL; |
13227 | |
|
13228 | 0 | ctxt->spaceNr = 0; |
13229 | 0 | if (ctxt->spaceTab != NULL) { |
13230 | 0 | ctxt->spaceTab[0] = -1; |
13231 | 0 | ctxt->space = &ctxt->spaceTab[0]; |
13232 | 0 | } else { |
13233 | 0 | ctxt->space = NULL; |
13234 | 0 | } |
13235 | | |
13236 | |
|
13237 | 0 | ctxt->nodeNr = 0; |
13238 | 0 | ctxt->node = NULL; |
13239 | |
|
13240 | 0 | ctxt->nameNr = 0; |
13241 | 0 | ctxt->name = NULL; |
13242 | |
|
13243 | 0 | ctxt->nsNr = 0; |
13244 | 0 | xmlParserNsReset(ctxt->nsdb); |
13245 | |
|
13246 | 0 | DICT_FREE(ctxt->version); |
13247 | 0 | ctxt->version = NULL; |
13248 | 0 | DICT_FREE(ctxt->encoding); |
13249 | 0 | ctxt->encoding = NULL; |
13250 | 0 | DICT_FREE(ctxt->extSubURI); |
13251 | 0 | ctxt->extSubURI = NULL; |
13252 | 0 | DICT_FREE(ctxt->extSubSystem); |
13253 | 0 | ctxt->extSubSystem = NULL; |
13254 | |
|
13255 | 0 | if (ctxt->directory != NULL) { |
13256 | 0 | xmlFree(ctxt->directory); |
13257 | 0 | ctxt->directory = NULL; |
13258 | 0 | } |
13259 | |
|
13260 | 0 | if (ctxt->myDoc != NULL) |
13261 | 0 | xmlFreeDoc(ctxt->myDoc); |
13262 | 0 | ctxt->myDoc = NULL; |
13263 | |
|
13264 | 0 | ctxt->standalone = -1; |
13265 | 0 | ctxt->hasExternalSubset = 0; |
13266 | 0 | ctxt->hasPErefs = 0; |
13267 | 0 | ctxt->html = 0; |
13268 | 0 | ctxt->instate = XML_PARSER_START; |
13269 | |
|
13270 | 0 | ctxt->wellFormed = 1; |
13271 | 0 | ctxt->nsWellFormed = 1; |
13272 | 0 | ctxt->disableSAX = 0; |
13273 | 0 | ctxt->valid = 1; |
13274 | 0 | ctxt->record_info = 0; |
13275 | 0 | ctxt->checkIndex = 0; |
13276 | 0 | ctxt->endCheckState = 0; |
13277 | 0 | ctxt->inSubset = 0; |
13278 | 0 | ctxt->errNo = XML_ERR_OK; |
13279 | 0 | ctxt->depth = 0; |
13280 | 0 | ctxt->catalogs = NULL; |
13281 | 0 | ctxt->sizeentities = 0; |
13282 | 0 | ctxt->sizeentcopy = 0; |
13283 | 0 | xmlInitNodeInfoSeq(&ctxt->node_seq); |
13284 | |
|
13285 | 0 | if (ctxt->attsDefault != NULL) { |
13286 | 0 | xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator); |
13287 | 0 | ctxt->attsDefault = NULL; |
13288 | 0 | } |
13289 | 0 | if (ctxt->attsSpecial != NULL) { |
13290 | 0 | xmlHashFree(ctxt->attsSpecial, NULL); |
13291 | 0 | ctxt->attsSpecial = NULL; |
13292 | 0 | } |
13293 | |
|
13294 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
13295 | 0 | if (ctxt->catalogs != NULL) |
13296 | 0 | xmlCatalogFreeLocal(ctxt->catalogs); |
13297 | 0 | #endif |
13298 | 0 | ctxt->nbErrors = 0; |
13299 | 0 | ctxt->nbWarnings = 0; |
13300 | 0 | if (ctxt->lastError.code != XML_ERR_OK) |
13301 | 0 | xmlResetError(&ctxt->lastError); |
13302 | 0 | } |
13303 | | |
13304 | | /** |
13305 | | * xmlCtxtResetPush: |
13306 | | * @ctxt: an XML parser context |
13307 | | * @chunk: a pointer to an array of chars |
13308 | | * @size: number of chars in the array |
13309 | | * @filename: an optional file name or URI |
13310 | | * @encoding: the document encoding, or NULL |
13311 | | * |
13312 | | * Reset a push parser context |
13313 | | * |
13314 | | * Returns 0 in case of success and 1 in case of error |
13315 | | */ |
13316 | | int |
13317 | | xmlCtxtResetPush(xmlParserCtxtPtr ctxt, const char *chunk, |
13318 | | int size, const char *filename, const char *encoding) |
13319 | 0 | { |
13320 | 0 | xmlParserInputPtr input; |
13321 | |
|
13322 | 0 | if (ctxt == NULL) |
13323 | 0 | return(1); |
13324 | | |
13325 | 0 | xmlCtxtReset(ctxt); |
13326 | |
|
13327 | 0 | input = xmlNewPushInput(filename, chunk, size); |
13328 | 0 | if (input == NULL) |
13329 | 0 | return(1); |
13330 | | |
13331 | 0 | if (inputPush(ctxt, input) < 0) { |
13332 | 0 | xmlFreeInputStream(input); |
13333 | 0 | return(1); |
13334 | 0 | } |
13335 | | |
13336 | 0 | if (encoding != NULL) |
13337 | 0 | xmlSwitchEncodingName(ctxt, encoding); |
13338 | |
|
13339 | 0 | return(0); |
13340 | 0 | } |
13341 | | |
13342 | | static int |
13343 | | xmlCtxtSetOptionsInternal(xmlParserCtxtPtr ctxt, int options, int keepMask) |
13344 | 20.2k | { |
13345 | 20.2k | int allMask; |
13346 | | |
13347 | 20.2k | if (ctxt == NULL) |
13348 | 0 | return(-1); |
13349 | | |
13350 | | /* |
13351 | | * XInclude options aren't handled by the parser. |
13352 | | * |
13353 | | * XML_PARSE_XINCLUDE |
13354 | | * XML_PARSE_NOXINCNODE |
13355 | | * XML_PARSE_NOBASEFIX |
13356 | | */ |
13357 | 20.2k | allMask = XML_PARSE_RECOVER | |
13358 | 20.2k | XML_PARSE_NOENT | |
13359 | 20.2k | XML_PARSE_DTDLOAD | |
13360 | 20.2k | XML_PARSE_DTDATTR | |
13361 | 20.2k | XML_PARSE_DTDVALID | |
13362 | 20.2k | XML_PARSE_NOERROR | |
13363 | 20.2k | XML_PARSE_NOWARNING | |
13364 | 20.2k | XML_PARSE_PEDANTIC | |
13365 | 20.2k | XML_PARSE_NOBLANKS | |
13366 | 20.2k | #ifdef LIBXML_SAX1_ENABLED |
13367 | 20.2k | XML_PARSE_SAX1 | |
13368 | 20.2k | #endif |
13369 | 20.2k | XML_PARSE_NONET | |
13370 | 20.2k | XML_PARSE_NODICT | |
13371 | 20.2k | XML_PARSE_NSCLEAN | |
13372 | 20.2k | XML_PARSE_NOCDATA | |
13373 | 20.2k | XML_PARSE_COMPACT | |
13374 | 20.2k | XML_PARSE_OLD10 | |
13375 | 20.2k | XML_PARSE_HUGE | |
13376 | 20.2k | XML_PARSE_OLDSAX | |
13377 | 20.2k | XML_PARSE_IGNORE_ENC | |
13378 | 20.2k | XML_PARSE_BIG_LINES | |
13379 | 20.2k | XML_PARSE_NO_XXE; |
13380 | | |
13381 | 20.2k | ctxt->options = (ctxt->options & keepMask) | (options & allMask); |
13382 | | |
13383 | | /* |
13384 | | * For some options, struct members are historically the source |
13385 | | * of truth. The values are initalized from global variables and |
13386 | | * old code could also modify them directly. Several older API |
13387 | | * functions that don't take an options argument rely on these |
13388 | | * deprecated mechanisms. |
13389 | | * |
13390 | | * Once public access to struct members and the globals are |
13391 | | * disabled, we can use the options bitmask as source of |
13392 | | * truth, making all these struct members obsolete. |
13393 | | * |
13394 | | * The XML_DETECT_IDS flags is misnamed. It simply enables |
13395 | | * loading of the external subset. |
13396 | | */ |
13397 | 20.2k | ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0; |
13398 | 20.2k | ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0; |
13399 | 20.2k | ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0; |
13400 | 20.2k | ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0; |
13401 | 20.2k | ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0; |
13402 | 20.2k | ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0; |
13403 | 20.2k | ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1; |
13404 | 20.2k | ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1; |
13405 | | |
13406 | | /* |
13407 | | * Changing SAX callbacks is a bad idea. This should be fixed. |
13408 | | */ |
13409 | 20.2k | if (options & XML_PARSE_NOBLANKS) { |
13410 | 0 | ctxt->sax->ignorableWhitespace = xmlSAX2IgnorableWhitespace; |
13411 | 0 | } |
13412 | 20.2k | if (options & XML_PARSE_NOCDATA) { |
13413 | 0 | ctxt->sax->cdataBlock = NULL; |
13414 | 0 | } |
13415 | 20.2k | if (options & XML_PARSE_HUGE) { |
13416 | 0 | if (ctxt->dict != NULL) |
13417 | 0 | xmlDictSetLimit(ctxt->dict, 0); |
13418 | 0 | } |
13419 | | |
13420 | 20.2k | ctxt->linenumbers = 1; |
13421 | | |
13422 | 20.2k | return(options & ~allMask); |
13423 | 20.2k | } |
13424 | | |
13425 | | /** |
13426 | | * xmlCtxtSetOptions: |
13427 | | * @ctxt: an XML parser context |
13428 | | * @options: a bitmask of xmlParserOption values |
13429 | | * |
13430 | | * Applies the options to the parser context. Unset options are |
13431 | | * cleared. |
13432 | | * |
13433 | | * Available since 2.13.0. With older versions, you can use |
13434 | | * xmlCtxtUseOptions. |
13435 | | * |
13436 | | * XML_PARSE_RECOVER |
13437 | | * |
13438 | | * Enable "recovery" mode which allows non-wellformed documents. |
13439 | | * How this mode behaves exactly is unspecified and may change |
13440 | | * without further notice. Use of this feature is DISCOURAGED. |
13441 | | * |
13442 | | * XML_PARSE_NOENT |
13443 | | * |
13444 | | * Despite the confusing name, this option enables substitution |
13445 | | * of entities. The resulting tree won't contain any entity |
13446 | | * reference nodes. |
13447 | | * |
13448 | | * This option also enables loading of external entities (both |
13449 | | * general and parameter entities) which is dangerous. If you |
13450 | | * process untrusted data, it's recommended to set the |
13451 | | * XML_PARSE_NO_XXE option to disable loading of external |
13452 | | * entities. |
13453 | | * |
13454 | | * XML_PARSE_DTDLOAD |
13455 | | * |
13456 | | * Enables loading of an external DTD and the loading and |
13457 | | * substitution of external parameter entities. Has no effect |
13458 | | * if XML_PARSE_NO_XXE is set. |
13459 | | * |
13460 | | * XML_PARSE_DTDATTR |
13461 | | * |
13462 | | * Adds default attributes from the DTD to the result document. |
13463 | | * |
13464 | | * Implies XML_PARSE_DTDLOAD, but loading of external content |
13465 | | * can be disabled with XML_PARSE_NO_XXE. |
13466 | | * |
13467 | | * XML_PARSE_DTDVALID |
13468 | | * |
13469 | | * This option enables DTD validation which requires to load |
13470 | | * external DTDs and external entities (both general and |
13471 | | * parameter entities) unless XML_PARSE_NO_XXE was set. |
13472 | | * |
13473 | | * XML_PARSE_NO_XXE |
13474 | | * |
13475 | | * Disables loading of external DTDs or entities. |
13476 | | * |
13477 | | * Available since 2.13.0. |
13478 | | * |
13479 | | * XML_PARSE_NOERROR |
13480 | | * |
13481 | | * Disable error and warning reports to the error handlers. |
13482 | | * Errors are still accessible with xmlCtxtGetLastError. |
13483 | | * |
13484 | | * XML_PARSE_NOWARNING |
13485 | | * |
13486 | | * Disable warning reports. |
13487 | | * |
13488 | | * XML_PARSE_PEDANTIC |
13489 | | * |
13490 | | * Enable some pedantic warnings. |
13491 | | * |
13492 | | * XML_PARSE_NOBLANKS |
13493 | | * |
13494 | | * Remove some text nodes containing only whitespace from the |
13495 | | * result document. Which nodes are removed depends on DTD |
13496 | | * element declarations or a conservative heuristic. The |
13497 | | * reindenting feature of the serialization code relies on this |
13498 | | * option to be set when parsing. Use of this option is |
13499 | | * DISCOURAGED. |
13500 | | * |
13501 | | * XML_PARSE_SAX1 |
13502 | | * |
13503 | | * Always invoke the deprecated SAX1 startElement and endElement |
13504 | | * handlers. This option is DEPRECATED. |
13505 | | * |
13506 | | * XML_PARSE_NONET |
13507 | | * |
13508 | | * Disable network access with the builtin HTTP client. |
13509 | | * |
13510 | | * XML_PARSE_NODICT |
13511 | | * |
13512 | | * Create a document without interned strings, making all |
13513 | | * strings separate memory allocations. |
13514 | | * |
13515 | | * XML_PARSE_NSCLEAN |
13516 | | * |
13517 | | * Remove redundant namespace declarations from the result |
13518 | | * document. |
13519 | | * |
13520 | | * XML_PARSE_NOCDATA |
13521 | | * |
13522 | | * Output normal text nodes instead of CDATA nodes. |
13523 | | * |
13524 | | * XML_PARSE_COMPACT |
13525 | | * |
13526 | | * Store small strings directly in the node struct to save |
13527 | | * memory. |
13528 | | * |
13529 | | * XML_PARSE_OLD10 |
13530 | | * |
13531 | | * Use old Name productions from before XML 1.0 Fifth Edition. |
13532 | | * This options is DEPRECATED. |
13533 | | * |
13534 | | * XML_PARSE_HUGE |
13535 | | * |
13536 | | * Relax some internal limits. |
13537 | | * |
13538 | | * Maximum size of text nodes, tags, comments, processing instructions, |
13539 | | * CDATA sections, entity values |
13540 | | * |
13541 | | * normal: 10M |
13542 | | * huge: 1B |
13543 | | * |
13544 | | * Maximum size of names, system literals, pubid literals |
13545 | | * |
13546 | | * normal: 50K |
13547 | | * huge: 10M |
13548 | | * |
13549 | | * Maximum nesting depth of elements |
13550 | | * |
13551 | | * normal: 256 |
13552 | | * huge: 2048 |
13553 | | * |
13554 | | * Maximum nesting depth of entities |
13555 | | * |
13556 | | * normal: 20 |
13557 | | * huge: 40 |
13558 | | * |
13559 | | * XML_PARSE_OLDSAX |
13560 | | * |
13561 | | * Enable an unspecified legacy mode for SAX parsers. This |
13562 | | * option is DEPRECATED. |
13563 | | * |
13564 | | * XML_PARSE_IGNORE_ENC |
13565 | | * |
13566 | | * Ignore the encoding in the XML declaration. This option is |
13567 | | * mostly unneeded these days. The only effect is to enforce |
13568 | | * UTF-8 decoding of ASCII-like data. |
13569 | | * |
13570 | | * XML_PARSE_BIG_LINES |
13571 | | * |
13572 | | * Enable reporting of line numbers larger than 65535. |
13573 | | * |
13574 | | * XML_PARSE_NO_UNZIP |
13575 | | * |
13576 | | * Disables input decompression. Setting this option is recommended |
13577 | | * to avoid zip bombs. |
13578 | | * |
13579 | | * Available since 2.14.0. |
13580 | | * |
13581 | | * XML_PARSE_NO_SYS_CATALOG |
13582 | | * |
13583 | | * Disables the global system XML catalog. |
13584 | | * |
13585 | | * Available since 2.14.0. |
13586 | | * |
13587 | | * XML_PARSE_NO_CATALOG_PI |
13588 | | * |
13589 | | * Ignore XML catalog processing instructions. |
13590 | | * |
13591 | | * Available since 2.14.0. |
13592 | | * |
13593 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
13594 | | * in case of error. |
13595 | | */ |
13596 | | int |
13597 | | xmlCtxtSetOptions(xmlParserCtxtPtr ctxt, int options) |
13598 | 0 | { |
13599 | 0 | return(xmlCtxtSetOptionsInternal(ctxt, options, 0)); |
13600 | 0 | } |
13601 | | |
13602 | | /** |
13603 | | * xmlCtxtGetOptions: |
13604 | | * @ctxt: an XML parser context |
13605 | | * |
13606 | | * Get the current options of the parser context. |
13607 | | * |
13608 | | * Available since 2.14.0. |
13609 | | * |
13610 | | * Returns the current options set in the parser context, or -1 if ctxt is NULL. |
13611 | | */ |
13612 | | int |
13613 | | xmlCtxtGetOptions(xmlParserCtxtPtr ctxt) |
13614 | 0 | { |
13615 | 0 | if (ctxt == NULL) |
13616 | 0 | return(-1); |
13617 | | |
13618 | 0 | return(ctxt->options); |
13619 | 0 | } |
13620 | | |
13621 | | /** |
13622 | | * xmlCtxtUseOptions: |
13623 | | * @ctxt: an XML parser context |
13624 | | * @options: a combination of xmlParserOption |
13625 | | * |
13626 | | * DEPRECATED: Use xmlCtxtSetOptions. |
13627 | | * |
13628 | | * Applies the options to the parser context. The following options |
13629 | | * are never cleared and can only be enabled: |
13630 | | * |
13631 | | * XML_PARSE_NOERROR |
13632 | | * XML_PARSE_NOWARNING |
13633 | | * XML_PARSE_NONET |
13634 | | * XML_PARSE_NSCLEAN |
13635 | | * XML_PARSE_NOCDATA |
13636 | | * XML_PARSE_COMPACT |
13637 | | * XML_PARSE_OLD10 |
13638 | | * XML_PARSE_HUGE |
13639 | | * XML_PARSE_OLDSAX |
13640 | | * XML_PARSE_IGNORE_ENC |
13641 | | * XML_PARSE_BIG_LINES |
13642 | | * |
13643 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
13644 | | * in case of error. |
13645 | | */ |
13646 | | int |
13647 | | xmlCtxtUseOptions(xmlParserCtxtPtr ctxt, int options) |
13648 | 20.2k | { |
13649 | 20.2k | int keepMask; |
13650 | | |
13651 | | /* |
13652 | | * For historic reasons, some options can only be enabled. |
13653 | | */ |
13654 | 20.2k | keepMask = XML_PARSE_NOERROR | |
13655 | 20.2k | XML_PARSE_NOWARNING | |
13656 | 20.2k | XML_PARSE_NONET | |
13657 | 20.2k | XML_PARSE_NSCLEAN | |
13658 | 20.2k | XML_PARSE_NOCDATA | |
13659 | 20.2k | XML_PARSE_COMPACT | |
13660 | 20.2k | XML_PARSE_OLD10 | |
13661 | 20.2k | XML_PARSE_HUGE | |
13662 | 20.2k | XML_PARSE_OLDSAX | |
13663 | 20.2k | XML_PARSE_IGNORE_ENC | |
13664 | 20.2k | XML_PARSE_BIG_LINES; |
13665 | | |
13666 | 20.2k | return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask)); |
13667 | 20.2k | } |
13668 | | |
13669 | | /** |
13670 | | * xmlCtxtSetMaxAmplification: |
13671 | | * @ctxt: an XML parser context |
13672 | | * @maxAmpl: maximum amplification factor |
13673 | | * |
13674 | | * To protect against exponential entity expansion ("billion laughs"), the |
13675 | | * size of serialized output is (roughly) limited to the input size |
13676 | | * multiplied by this factor. The default value is 5. |
13677 | | * |
13678 | | * When working with documents making heavy use of entity expansion, it can |
13679 | | * be necessary to increase the value. For security reasons, this should only |
13680 | | * be considered when processing trusted input. |
13681 | | */ |
13682 | | void |
13683 | | xmlCtxtSetMaxAmplification(xmlParserCtxtPtr ctxt, unsigned maxAmpl) |
13684 | 0 | { |
13685 | 0 | ctxt->maxAmpl = maxAmpl; |
13686 | 0 | } |
13687 | | |
13688 | | /** |
13689 | | * xmlCtxtParseDocument: |
13690 | | * @ctxt: an XML parser context |
13691 | | * @input: parser input |
13692 | | * |
13693 | | * Parse an XML document and return the resulting document tree. |
13694 | | * Takes ownership of the input object. |
13695 | | * |
13696 | | * Available since 2.13.0. |
13697 | | * |
13698 | | * Returns the resulting document tree or NULL |
13699 | | */ |
13700 | | xmlDocPtr |
13701 | | xmlCtxtParseDocument(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) |
13702 | 20.2k | { |
13703 | 20.2k | xmlDocPtr ret = NULL; |
13704 | | |
13705 | 20.2k | if ((ctxt == NULL) || (input == NULL)) |
13706 | 0 | return(NULL); |
13707 | | |
13708 | | /* assert(ctxt->inputNr == 0); */ |
13709 | 20.2k | while (ctxt->inputNr > 0) |
13710 | 0 | xmlFreeInputStream(inputPop(ctxt)); |
13711 | | |
13712 | 20.2k | if (inputPush(ctxt, input) < 0) { |
13713 | 0 | xmlFreeInputStream(input); |
13714 | 0 | return(NULL); |
13715 | 0 | } |
13716 | | |
13717 | 20.2k | xmlParseDocument(ctxt); |
13718 | | |
13719 | 20.2k | if ((ctxt->wellFormed) || |
13720 | 20.2k | ((ctxt->recovery) && (ctxt->errNo != XML_ERR_NO_MEMORY))) { |
13721 | 36 | ret = ctxt->myDoc; |
13722 | 20.1k | } else { |
13723 | 20.1k | if (ctxt->errNo == XML_ERR_OK) |
13724 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, "unknown error\n"); |
13725 | | |
13726 | 20.1k | ret = NULL; |
13727 | 20.1k | xmlFreeDoc(ctxt->myDoc); |
13728 | 20.1k | } |
13729 | 20.2k | ctxt->myDoc = NULL; |
13730 | | |
13731 | | /* assert(ctxt->inputNr == 1); */ |
13732 | 40.4k | while (ctxt->inputNr > 0) |
13733 | 20.2k | xmlFreeInputStream(inputPop(ctxt)); |
13734 | | |
13735 | 20.2k | return(ret); |
13736 | 20.2k | } |
13737 | | |
13738 | | /** |
13739 | | * xmlReadDoc: |
13740 | | * @cur: a pointer to a zero terminated string |
13741 | | * @URL: base URL (optional) |
13742 | | * @encoding: the document encoding (optional) |
13743 | | * @options: a combination of xmlParserOption |
13744 | | * |
13745 | | * Convenience function to parse an XML document from a |
13746 | | * zero-terminated string. |
13747 | | * |
13748 | | * See xmlCtxtReadDoc for details. |
13749 | | * |
13750 | | * Returns the resulting document tree |
13751 | | */ |
13752 | | xmlDocPtr |
13753 | | xmlReadDoc(const xmlChar *cur, const char *URL, const char *encoding, |
13754 | | int options) |
13755 | 0 | { |
13756 | 0 | xmlParserCtxtPtr ctxt; |
13757 | 0 | xmlParserInputPtr input; |
13758 | 0 | xmlDocPtr doc; |
13759 | |
|
13760 | 0 | ctxt = xmlNewParserCtxt(); |
13761 | 0 | if (ctxt == NULL) |
13762 | 0 | return(NULL); |
13763 | | |
13764 | 0 | xmlCtxtUseOptions(ctxt, options); |
13765 | |
|
13766 | 0 | input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding, |
13767 | 0 | XML_INPUT_BUF_STATIC); |
13768 | |
|
13769 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13770 | |
|
13771 | 0 | xmlFreeParserCtxt(ctxt); |
13772 | 0 | return(doc); |
13773 | 0 | } |
13774 | | |
13775 | | /** |
13776 | | * xmlReadFile: |
13777 | | * @filename: a file or URL |
13778 | | * @encoding: the document encoding (optional) |
13779 | | * @options: a combination of xmlParserOption |
13780 | | * |
13781 | | * Convenience function to parse an XML file from the filesystem, |
13782 | | * the network or a global user-define resource loader. |
13783 | | * |
13784 | | * See xmlCtxtReadFile for details. |
13785 | | * |
13786 | | * Returns the resulting document tree |
13787 | | */ |
13788 | | xmlDocPtr |
13789 | | xmlReadFile(const char *filename, const char *encoding, int options) |
13790 | 0 | { |
13791 | 0 | xmlParserCtxtPtr ctxt; |
13792 | 0 | xmlParserInputPtr input; |
13793 | 0 | xmlDocPtr doc; |
13794 | |
|
13795 | 0 | ctxt = xmlNewParserCtxt(); |
13796 | 0 | if (ctxt == NULL) |
13797 | 0 | return(NULL); |
13798 | | |
13799 | 0 | xmlCtxtUseOptions(ctxt, options); |
13800 | | |
13801 | | /* |
13802 | | * Backward compatibility for users of command line utilities like |
13803 | | * xmlstarlet expecting "-" to mean stdin. This is dangerous and |
13804 | | * should be removed at some point. |
13805 | | */ |
13806 | 0 | if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0)) |
13807 | 0 | input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, |
13808 | 0 | encoding, 0); |
13809 | 0 | else |
13810 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0); |
13811 | |
|
13812 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13813 | |
|
13814 | 0 | xmlFreeParserCtxt(ctxt); |
13815 | 0 | return(doc); |
13816 | 0 | } |
13817 | | |
13818 | | /** |
13819 | | * xmlReadMemory: |
13820 | | * @buffer: a pointer to a char array |
13821 | | * @size: the size of the array |
13822 | | * @url: base URL (optional) |
13823 | | * @encoding: the document encoding (optional) |
13824 | | * @options: a combination of xmlParserOption |
13825 | | * |
13826 | | * Parse an XML in-memory document and build a tree. The input buffer must |
13827 | | * not contain a terminating null byte. |
13828 | | * |
13829 | | * See xmlCtxtReadMemory for details. |
13830 | | * |
13831 | | * Returns the resulting document tree |
13832 | | */ |
13833 | | xmlDocPtr |
13834 | | xmlReadMemory(const char *buffer, int size, const char *url, |
13835 | | const char *encoding, int options) |
13836 | 20.2k | { |
13837 | 20.2k | xmlParserCtxtPtr ctxt; |
13838 | 20.2k | xmlParserInputPtr input; |
13839 | 20.2k | xmlDocPtr doc; |
13840 | | |
13841 | 20.2k | if (size < 0) |
13842 | 0 | return(NULL); |
13843 | | |
13844 | 20.2k | ctxt = xmlNewParserCtxt(); |
13845 | 20.2k | if (ctxt == NULL) |
13846 | 0 | return(NULL); |
13847 | | |
13848 | 20.2k | xmlCtxtUseOptions(ctxt, options); |
13849 | | |
13850 | 20.2k | input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding, |
13851 | 20.2k | XML_INPUT_BUF_STATIC); |
13852 | | |
13853 | 20.2k | doc = xmlCtxtParseDocument(ctxt, input); |
13854 | | |
13855 | 20.2k | xmlFreeParserCtxt(ctxt); |
13856 | 20.2k | return(doc); |
13857 | 20.2k | } |
13858 | | |
13859 | | /** |
13860 | | * xmlReadFd: |
13861 | | * @fd: an open file descriptor |
13862 | | * @URL: base URL (optional) |
13863 | | * @encoding: the document encoding (optional) |
13864 | | * @options: a combination of xmlParserOption |
13865 | | * |
13866 | | * Parse an XML from a file descriptor and build a tree. |
13867 | | * |
13868 | | * See xmlCtxtReadFd for details. |
13869 | | * |
13870 | | * NOTE that the file descriptor will not be closed when the |
13871 | | * context is freed or reset. |
13872 | | * |
13873 | | * Returns the resulting document tree |
13874 | | */ |
13875 | | xmlDocPtr |
13876 | | xmlReadFd(int fd, const char *URL, const char *encoding, int options) |
13877 | 0 | { |
13878 | 0 | xmlParserCtxtPtr ctxt; |
13879 | 0 | xmlParserInputPtr input; |
13880 | 0 | xmlDocPtr doc; |
13881 | |
|
13882 | 0 | ctxt = xmlNewParserCtxt(); |
13883 | 0 | if (ctxt == NULL) |
13884 | 0 | return(NULL); |
13885 | | |
13886 | 0 | xmlCtxtUseOptions(ctxt, options); |
13887 | |
|
13888 | 0 | input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0); |
13889 | |
|
13890 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13891 | |
|
13892 | 0 | xmlFreeParserCtxt(ctxt); |
13893 | 0 | return(doc); |
13894 | 0 | } |
13895 | | |
13896 | | /** |
13897 | | * xmlReadIO: |
13898 | | * @ioread: an I/O read function |
13899 | | * @ioclose: an I/O close function (optional) |
13900 | | * @ioctx: an I/O handler |
13901 | | * @URL: base URL (optional) |
13902 | | * @encoding: the document encoding (optional) |
13903 | | * @options: a combination of xmlParserOption |
13904 | | * |
13905 | | * Parse an XML document from I/O functions and context and build a tree. |
13906 | | * |
13907 | | * See xmlCtxtReadIO for details. |
13908 | | * |
13909 | | * Returns the resulting document tree |
13910 | | */ |
13911 | | xmlDocPtr |
13912 | | xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
13913 | | void *ioctx, const char *URL, const char *encoding, int options) |
13914 | 0 | { |
13915 | 0 | xmlParserCtxtPtr ctxt; |
13916 | 0 | xmlParserInputPtr input; |
13917 | 0 | xmlDocPtr doc; |
13918 | |
|
13919 | 0 | ctxt = xmlNewParserCtxt(); |
13920 | 0 | if (ctxt == NULL) |
13921 | 0 | return(NULL); |
13922 | | |
13923 | 0 | xmlCtxtUseOptions(ctxt, options); |
13924 | |
|
13925 | 0 | input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx, |
13926 | 0 | encoding, 0); |
13927 | |
|
13928 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13929 | |
|
13930 | 0 | xmlFreeParserCtxt(ctxt); |
13931 | 0 | return(doc); |
13932 | 0 | } |
13933 | | |
13934 | | /** |
13935 | | * xmlCtxtReadDoc: |
13936 | | * @ctxt: an XML parser context |
13937 | | * @str: a pointer to a zero terminated string |
13938 | | * @URL: base URL (optional) |
13939 | | * @encoding: the document encoding (optional) |
13940 | | * @options: a combination of xmlParserOption |
13941 | | * |
13942 | | * Parse an XML in-memory document and build a tree. |
13943 | | * |
13944 | | * @URL is used as base to resolve external entities and for error |
13945 | | * reporting. |
13946 | | * |
13947 | | * See xmlCtxtUseOptions for details. |
13948 | | * |
13949 | | * Returns the resulting document tree |
13950 | | */ |
13951 | | xmlDocPtr |
13952 | | xmlCtxtReadDoc(xmlParserCtxtPtr ctxt, const xmlChar *str, |
13953 | | const char *URL, const char *encoding, int options) |
13954 | 0 | { |
13955 | 0 | xmlParserInputPtr input; |
13956 | |
|
13957 | 0 | if (ctxt == NULL) |
13958 | 0 | return(NULL); |
13959 | | |
13960 | 0 | xmlCtxtReset(ctxt); |
13961 | 0 | xmlCtxtUseOptions(ctxt, options); |
13962 | |
|
13963 | 0 | input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) str, encoding, |
13964 | 0 | XML_INPUT_BUF_STATIC); |
13965 | |
|
13966 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13967 | 0 | } |
13968 | | |
13969 | | /** |
13970 | | * xmlCtxtReadFile: |
13971 | | * @ctxt: an XML parser context |
13972 | | * @filename: a file or URL |
13973 | | * @encoding: the document encoding (optional) |
13974 | | * @options: a combination of xmlParserOption |
13975 | | * |
13976 | | * Parse an XML file from the filesystem, the network or a user-defined |
13977 | | * resource loader. |
13978 | | * |
13979 | | * Returns the resulting document tree |
13980 | | */ |
13981 | | xmlDocPtr |
13982 | | xmlCtxtReadFile(xmlParserCtxtPtr ctxt, const char *filename, |
13983 | | const char *encoding, int options) |
13984 | 0 | { |
13985 | 0 | xmlParserInputPtr input; |
13986 | |
|
13987 | 0 | if (ctxt == NULL) |
13988 | 0 | return(NULL); |
13989 | | |
13990 | 0 | xmlCtxtReset(ctxt); |
13991 | 0 | xmlCtxtUseOptions(ctxt, options); |
13992 | |
|
13993 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0); |
13994 | |
|
13995 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13996 | 0 | } |
13997 | | |
13998 | | /** |
13999 | | * xmlCtxtReadMemory: |
14000 | | * @ctxt: an XML parser context |
14001 | | * @buffer: a pointer to a char array |
14002 | | * @size: the size of the array |
14003 | | * @URL: base URL (optional) |
14004 | | * @encoding: the document encoding (optional) |
14005 | | * @options: a combination of xmlParserOption |
14006 | | * |
14007 | | * Parse an XML in-memory document and build a tree. The input buffer must |
14008 | | * not contain a terminating null byte. |
14009 | | * |
14010 | | * @URL is used as base to resolve external entities and for error |
14011 | | * reporting. |
14012 | | * |
14013 | | * See xmlCtxtUseOptions for details. |
14014 | | * |
14015 | | * Returns the resulting document tree |
14016 | | */ |
14017 | | xmlDocPtr |
14018 | | xmlCtxtReadMemory(xmlParserCtxtPtr ctxt, const char *buffer, int size, |
14019 | | const char *URL, const char *encoding, int options) |
14020 | 0 | { |
14021 | 0 | xmlParserInputPtr input; |
14022 | |
|
14023 | 0 | if ((ctxt == NULL) || (size < 0)) |
14024 | 0 | return(NULL); |
14025 | | |
14026 | 0 | xmlCtxtReset(ctxt); |
14027 | 0 | xmlCtxtUseOptions(ctxt, options); |
14028 | |
|
14029 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding, |
14030 | 0 | XML_INPUT_BUF_STATIC); |
14031 | |
|
14032 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
14033 | 0 | } |
14034 | | |
14035 | | /** |
14036 | | * xmlCtxtReadFd: |
14037 | | * @ctxt: an XML parser context |
14038 | | * @fd: an open file descriptor |
14039 | | * @URL: base URL (optional) |
14040 | | * @encoding: the document encoding (optional) |
14041 | | * @options: a combination of xmlParserOption |
14042 | | * |
14043 | | * Parse an XML document from a file descriptor and build a tree. |
14044 | | * |
14045 | | * NOTE that the file descriptor will not be closed when the |
14046 | | * context is freed or reset. |
14047 | | * |
14048 | | * @URL is used as base to resolve external entities and for error |
14049 | | * reporting. |
14050 | | * |
14051 | | * See xmlCtxtUseOptions for details. |
14052 | | * |
14053 | | * Returns the resulting document tree |
14054 | | */ |
14055 | | xmlDocPtr |
14056 | | xmlCtxtReadFd(xmlParserCtxtPtr ctxt, int fd, |
14057 | | const char *URL, const char *encoding, int options) |
14058 | 0 | { |
14059 | 0 | xmlParserInputPtr input; |
14060 | |
|
14061 | 0 | if (ctxt == NULL) |
14062 | 0 | return(NULL); |
14063 | | |
14064 | 0 | xmlCtxtReset(ctxt); |
14065 | 0 | xmlCtxtUseOptions(ctxt, options); |
14066 | |
|
14067 | 0 | input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0); |
14068 | |
|
14069 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
14070 | 0 | } |
14071 | | |
14072 | | /** |
14073 | | * xmlCtxtReadIO: |
14074 | | * @ctxt: an XML parser context |
14075 | | * @ioread: an I/O read function |
14076 | | * @ioclose: an I/O close function |
14077 | | * @ioctx: an I/O handler |
14078 | | * @URL: the base URL to use for the document |
14079 | | * @encoding: the document encoding, or NULL |
14080 | | * @options: a combination of xmlParserOption |
14081 | | * |
14082 | | * parse an XML document from I/O functions and source and build a tree. |
14083 | | * This reuses the existing @ctxt parser context |
14084 | | * |
14085 | | * @URL is used as base to resolve external entities and for error |
14086 | | * reporting. |
14087 | | * |
14088 | | * See xmlCtxtUseOptions for details. |
14089 | | * |
14090 | | * Returns the resulting document tree |
14091 | | */ |
14092 | | xmlDocPtr |
14093 | | xmlCtxtReadIO(xmlParserCtxtPtr ctxt, xmlInputReadCallback ioread, |
14094 | | xmlInputCloseCallback ioclose, void *ioctx, |
14095 | | const char *URL, |
14096 | | const char *encoding, int options) |
14097 | 0 | { |
14098 | 0 | xmlParserInputPtr input; |
14099 | |
|
14100 | 0 | if (ctxt == NULL) |
14101 | 0 | return(NULL); |
14102 | | |
14103 | 0 | xmlCtxtReset(ctxt); |
14104 | 0 | xmlCtxtUseOptions(ctxt, options); |
14105 | |
|
14106 | 0 | input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx, |
14107 | 0 | encoding, 0); |
14108 | |
|
14109 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
14110 | 0 | } |
14111 | | |