/src/libprotobuf-mutator/build/examples/libxml2/external.libxml2/src/external.libxml2/parser.c
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 | | * Author: Daniel Veillard |
31 | | */ |
32 | | |
33 | | /* To avoid EBCDIC trouble when parsing on zOS */ |
34 | | #if defined(__MVS__) |
35 | | #pragma convert("ISO8859-1") |
36 | | #endif |
37 | | |
38 | | #define IN_LIBXML |
39 | | #include "libxml.h" |
40 | | |
41 | | #if defined(_WIN32) |
42 | | #define XML_DIR_SEP '\\' |
43 | | #else |
44 | | #define XML_DIR_SEP '/' |
45 | | #endif |
46 | | |
47 | | #include <stdlib.h> |
48 | | #include <limits.h> |
49 | | #include <string.h> |
50 | | #include <stdarg.h> |
51 | | #include <stddef.h> |
52 | | #include <ctype.h> |
53 | | #include <stdlib.h> |
54 | | #include <libxml/parser.h> |
55 | | #include <libxml/xmlmemory.h> |
56 | | #include <libxml/tree.h> |
57 | | #include <libxml/parserInternals.h> |
58 | | #include <libxml/valid.h> |
59 | | #include <libxml/entities.h> |
60 | | #include <libxml/xmlerror.h> |
61 | | #include <libxml/encoding.h> |
62 | | #include <libxml/xmlIO.h> |
63 | | #include <libxml/uri.h> |
64 | | #include <libxml/SAX2.h> |
65 | | #include <libxml/HTMLparser.h> |
66 | | #ifdef LIBXML_CATALOG_ENABLED |
67 | | #include <libxml/catalog.h> |
68 | | #endif |
69 | | |
70 | | #include "private/buf.h" |
71 | | #include "private/dict.h" |
72 | | #include "private/entities.h" |
73 | | #include "private/error.h" |
74 | | #include "private/html.h" |
75 | | #include "private/io.h" |
76 | | #include "private/memory.h" |
77 | | #include "private/parser.h" |
78 | | |
79 | 48.6k | #define NS_INDEX_EMPTY INT_MAX |
80 | 12.0k | #define NS_INDEX_XML (INT_MAX - 1) |
81 | 21.1k | #define URI_HASH_EMPTY 0xD943A04E |
82 | 2.05k | #define URI_HASH_XML 0xF0451F02 |
83 | | |
84 | | #ifndef STDIN_FILENO |
85 | 0 | #define STDIN_FILENO 0 |
86 | | #endif |
87 | | |
88 | | #ifndef SIZE_MAX |
89 | | #define SIZE_MAX ((size_t) -1) |
90 | | #endif |
91 | | |
92 | 25.9k | #define XML_MAX_ATTRS 100000000 /* 100 million */ |
93 | | |
94 | | struct _xmlStartTag { |
95 | | const xmlChar *prefix; |
96 | | const xmlChar *URI; |
97 | | int line; |
98 | | int nsNr; |
99 | | }; |
100 | | |
101 | | typedef struct { |
102 | | void *saxData; |
103 | | unsigned prefixHashValue; |
104 | | unsigned uriHashValue; |
105 | | unsigned elementId; |
106 | | int oldIndex; |
107 | | } xmlParserNsExtra; |
108 | | |
109 | | typedef struct { |
110 | | unsigned hashValue; |
111 | | int index; |
112 | | } xmlParserNsBucket; |
113 | | |
114 | | struct _xmlParserNsData { |
115 | | xmlParserNsExtra *extra; |
116 | | |
117 | | unsigned hashSize; |
118 | | unsigned hashElems; |
119 | | xmlParserNsBucket *hash; |
120 | | |
121 | | unsigned elementId; |
122 | | int defaultNsIndex; |
123 | | int minNsIndex; |
124 | | }; |
125 | | |
126 | | static int |
127 | | xmlParseElementStart(xmlParserCtxtPtr ctxt); |
128 | | |
129 | | static void |
130 | | xmlParseElementEnd(xmlParserCtxtPtr ctxt); |
131 | | |
132 | | static xmlEntityPtr |
133 | | xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr); |
134 | | |
135 | | static const xmlChar * |
136 | | xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt); |
137 | | |
138 | | /************************************************************************ |
139 | | * * |
140 | | * Arbitrary limits set in the parser. See XML_PARSE_HUGE * |
141 | | * * |
142 | | ************************************************************************/ |
143 | | |
144 | | #define XML_PARSER_BIG_ENTITY 1000 |
145 | | #define XML_PARSER_LOT_ENTITY 5000 |
146 | | |
147 | | /* |
148 | | * Constants for protection against abusive entity expansion |
149 | | * ("billion laughs"). |
150 | | */ |
151 | | |
152 | | /* |
153 | | * A certain amount of entity expansion which is always allowed. |
154 | | */ |
155 | 675k | #define XML_PARSER_ALLOWED_EXPANSION 1000000 |
156 | | |
157 | | /* |
158 | | * Fixed cost for each entity reference. This crudely models processing time |
159 | | * as well to protect, for example, against exponential expansion of empty |
160 | | * or very short entities. |
161 | | */ |
162 | 678k | #define XML_ENT_FIXED_COST 20 |
163 | | |
164 | 39.3k | #define XML_PARSER_BIG_BUFFER_SIZE 300 |
165 | 58.8k | #define XML_PARSER_BUFFER_SIZE 100 |
166 | 26.5k | #define SAX_COMPAT_MODE BAD_CAST "SAX compatibility mode document" |
167 | | |
168 | | /** |
169 | | * XML_PARSER_CHUNK_SIZE |
170 | | * |
171 | | * When calling GROW that's the minimal amount of data |
172 | | * the parser expected to have received. It is not a hard |
173 | | * limit but an optimization when reading strings like Names |
174 | | * It is not strictly needed as long as inputs available characters |
175 | | * are followed by 0, which should be provided by the I/O level |
176 | | */ |
177 | | #define XML_PARSER_CHUNK_SIZE 100 |
178 | | |
179 | | /** |
180 | | * Constant string describing the internal version of the library |
181 | | */ |
182 | | const char *const |
183 | | xmlParserVersion = LIBXML_VERSION_STRING LIBXML_VERSION_EXTRA; |
184 | | |
185 | | /* |
186 | | * List of XML prefixed PI allowed by W3C specs |
187 | | */ |
188 | | |
189 | | static const char* const xmlW3CPIs[] = { |
190 | | "xml-stylesheet", |
191 | | "xml-model", |
192 | | NULL |
193 | | }; |
194 | | |
195 | | |
196 | | /* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */ |
197 | | static xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt, |
198 | | const xmlChar **str); |
199 | | |
200 | | static void |
201 | | xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent); |
202 | | |
203 | | static int |
204 | | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity); |
205 | | |
206 | | /************************************************************************ |
207 | | * * |
208 | | * Some factorized error routines * |
209 | | * * |
210 | | ************************************************************************/ |
211 | | |
212 | | static void |
213 | 0 | xmlErrMemory(xmlParserCtxtPtr ctxt) { |
214 | 0 | xmlCtxtErrMemory(ctxt); |
215 | 0 | } |
216 | | |
217 | | /** |
218 | | * Handle a redefinition of attribute error |
219 | | * |
220 | | * @param ctxt an XML parser context |
221 | | * @param prefix the attribute prefix |
222 | | * @param localname the attribute localname |
223 | | */ |
224 | | static void |
225 | | xmlErrAttributeDup(xmlParserCtxtPtr ctxt, const xmlChar * prefix, |
226 | | const xmlChar * localname) |
227 | 10.5k | { |
228 | 10.5k | if (prefix == NULL) |
229 | 8.19k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED, |
230 | 8.19k | XML_ERR_FATAL, localname, NULL, NULL, 0, |
231 | 8.19k | "Attribute %s redefined\n", localname); |
232 | 2.36k | else |
233 | 2.36k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED, |
234 | 2.36k | XML_ERR_FATAL, prefix, localname, NULL, 0, |
235 | 2.36k | "Attribute %s:%s redefined\n", prefix, localname); |
236 | 10.5k | } |
237 | | |
238 | | /** |
239 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
240 | | * |
241 | | * @param ctxt an XML parser context |
242 | | * @param error the error number |
243 | | * @param msg the error message |
244 | | */ |
245 | | static void LIBXML_ATTR_FORMAT(3,0) |
246 | | xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
247 | | const char *msg) |
248 | 273k | { |
249 | 273k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
250 | 273k | NULL, NULL, NULL, 0, "%s", msg); |
251 | 273k | } |
252 | | |
253 | | /** |
254 | | * Handle a warning. |
255 | | * |
256 | | * @param ctxt an XML parser context |
257 | | * @param error the error number |
258 | | * @param msg the error message |
259 | | * @param str1 extra data |
260 | | * @param str2 extra data |
261 | | */ |
262 | | void LIBXML_ATTR_FORMAT(3,0) |
263 | | xmlWarningMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
264 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
265 | 16.6k | { |
266 | 16.6k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING, |
267 | 16.6k | str1, str2, NULL, 0, msg, str1, str2); |
268 | 16.6k | } |
269 | | |
270 | | /** |
271 | | * Handle a validity error. |
272 | | * |
273 | | * @param ctxt an XML parser context |
274 | | * @param error the error number |
275 | | * @param msg the error message |
276 | | * @param str1 extra data |
277 | | * @param str2 extra data |
278 | | */ |
279 | | static void LIBXML_ATTR_FORMAT(3,0) |
280 | | xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
281 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
282 | 1.33k | { |
283 | 1.33k | ctxt->valid = 0; |
284 | | |
285 | 1.33k | xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR, |
286 | 1.33k | str1, str2, NULL, 0, msg, str1, str2); |
287 | 1.33k | } |
288 | | |
289 | | /** |
290 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
291 | | * |
292 | | * @param ctxt an XML parser context |
293 | | * @param error the error number |
294 | | * @param msg the error message |
295 | | * @param val an integer value |
296 | | */ |
297 | | static void LIBXML_ATTR_FORMAT(3,0) |
298 | | xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
299 | | const char *msg, int val) |
300 | 12.0k | { |
301 | 12.0k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
302 | 12.0k | NULL, NULL, NULL, val, msg, val); |
303 | 12.0k | } |
304 | | |
305 | | /** |
306 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
307 | | * |
308 | | * @param ctxt an XML parser context |
309 | | * @param error the error number |
310 | | * @param msg the error message |
311 | | * @param str1 an string info |
312 | | * @param val an integer value |
313 | | * @param str2 an string info |
314 | | */ |
315 | | static void LIBXML_ATTR_FORMAT(3,0) |
316 | | xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
317 | | const char *msg, const xmlChar *str1, int val, |
318 | | const xmlChar *str2) |
319 | 94.4k | { |
320 | 94.4k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
321 | 94.4k | str1, str2, NULL, val, msg, str1, val, str2); |
322 | 94.4k | } |
323 | | |
324 | | /** |
325 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
326 | | * |
327 | | * @param ctxt an XML parser context |
328 | | * @param error the error number |
329 | | * @param msg the error message |
330 | | * @param val a string value |
331 | | */ |
332 | | static void LIBXML_ATTR_FORMAT(3,0) |
333 | | xmlFatalErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
334 | | const char *msg, const xmlChar * val) |
335 | 61.8k | { |
336 | 61.8k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
337 | 61.8k | val, NULL, NULL, 0, msg, val); |
338 | 61.8k | } |
339 | | |
340 | | /** |
341 | | * Handle a non fatal parser error |
342 | | * |
343 | | * @param ctxt an XML parser context |
344 | | * @param error the error number |
345 | | * @param msg the error message |
346 | | * @param val a string value |
347 | | */ |
348 | | static void LIBXML_ATTR_FORMAT(3,0) |
349 | | xmlErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
350 | | const char *msg, const xmlChar * val) |
351 | 0 | { |
352 | 0 | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR, |
353 | 0 | val, NULL, NULL, 0, msg, val); |
354 | 0 | } |
355 | | |
356 | | /** |
357 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
358 | | * |
359 | | * @param ctxt an XML parser context |
360 | | * @param error the error number |
361 | | * @param msg the message |
362 | | * @param info1 extra information string |
363 | | * @param info2 extra information string |
364 | | * @param info3 extra information string |
365 | | */ |
366 | | static void LIBXML_ATTR_FORMAT(3,0) |
367 | | xmlNsErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
368 | | const char *msg, |
369 | | const xmlChar * info1, const xmlChar * info2, |
370 | | const xmlChar * info3) |
371 | 35.9k | { |
372 | 35.9k | ctxt->nsWellFormed = 0; |
373 | | |
374 | 35.9k | xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR, |
375 | 35.9k | info1, info2, info3, 0, msg, info1, info2, info3); |
376 | 35.9k | } |
377 | | |
378 | | /** |
379 | | * Handle a namespace warning error |
380 | | * |
381 | | * @param ctxt an XML parser context |
382 | | * @param error the error number |
383 | | * @param msg the message |
384 | | * @param info1 extra information string |
385 | | * @param info2 extra information string |
386 | | * @param info3 extra information string |
387 | | */ |
388 | | static void LIBXML_ATTR_FORMAT(3,0) |
389 | | xmlNsWarn(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
390 | | const char *msg, |
391 | | const xmlChar * info1, const xmlChar * info2, |
392 | | const xmlChar * info3) |
393 | 1.14k | { |
394 | 1.14k | xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING, |
395 | 1.14k | info1, info2, info3, 0, msg, info1, info2, info3); |
396 | 1.14k | } |
397 | | |
398 | | static void |
399 | 2.03M | xmlSaturatedAdd(unsigned long *dst, unsigned long val) { |
400 | 2.03M | if (val > ULONG_MAX - *dst) |
401 | 0 | *dst = ULONG_MAX; |
402 | 2.03M | else |
403 | 2.03M | *dst += val; |
404 | 2.03M | } |
405 | | |
406 | | static void |
407 | 679k | xmlSaturatedAddSizeT(unsigned long *dst, unsigned long val) { |
408 | 679k | if (val > ULONG_MAX - *dst) |
409 | 0 | *dst = ULONG_MAX; |
410 | 679k | else |
411 | 679k | *dst += val; |
412 | 679k | } |
413 | | |
414 | | /** |
415 | | * Check for non-linear entity expansion behaviour. |
416 | | * |
417 | | * In some cases like xmlExpandEntityInAttValue(), this function is called |
418 | | * for each, possibly nested entity and its unexpanded content length. |
419 | | * |
420 | | * In other cases like xmlParseReference(), it's only called for each |
421 | | * top-level entity with its unexpanded content length plus the sum of |
422 | | * the unexpanded content lengths (plus fixed cost) of all nested |
423 | | * entities. |
424 | | * |
425 | | * Summing the unexpanded lengths also adds the length of the reference. |
426 | | * This is by design. Taking the length of the entity name into account |
427 | | * discourages attacks that try to waste CPU time with abusively long |
428 | | * entity names. See test/recurse/lol6.xml for example. Each call also |
429 | | * adds some fixed cost XML_ENT_FIXED_COST to discourage attacks with |
430 | | * short entities. |
431 | | * |
432 | | * @param ctxt parser context |
433 | | * @param extra sum of unexpanded entity sizes |
434 | | * @returns 1 on error, 0 on success. |
435 | | */ |
436 | | static int |
437 | | xmlParserEntityCheck(xmlParserCtxtPtr ctxt, unsigned long extra) |
438 | 702k | { |
439 | 702k | unsigned long consumed; |
440 | 702k | unsigned long *expandedSize; |
441 | 702k | xmlParserInputPtr input = ctxt->input; |
442 | 702k | xmlEntityPtr entity = input->entity; |
443 | | |
444 | 702k | if ((entity) && (entity->flags & XML_ENT_CHECKED)) |
445 | 26.7k | return(0); |
446 | | |
447 | | /* |
448 | | * Compute total consumed bytes so far, including input streams of |
449 | | * external entities. |
450 | | */ |
451 | 675k | consumed = input->consumed; |
452 | 675k | xmlSaturatedAddSizeT(&consumed, input->cur - input->base); |
453 | 675k | xmlSaturatedAdd(&consumed, ctxt->sizeentities); |
454 | | |
455 | 675k | if (entity) |
456 | 9.12k | expandedSize = &entity->expandedSize; |
457 | 666k | else |
458 | 666k | expandedSize = &ctxt->sizeentcopy; |
459 | | |
460 | | /* |
461 | | * Add extra cost and some fixed cost. |
462 | | */ |
463 | 675k | xmlSaturatedAdd(expandedSize, extra); |
464 | 675k | xmlSaturatedAdd(expandedSize, XML_ENT_FIXED_COST); |
465 | | |
466 | | /* |
467 | | * It's important to always use saturation arithmetic when tracking |
468 | | * entity sizes to make the size checks reliable. If "sizeentcopy" |
469 | | * overflows, we have to abort. |
470 | | */ |
471 | 675k | if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) && |
472 | 675k | ((*expandedSize >= ULONG_MAX) || |
473 | 13 | (*expandedSize / ctxt->maxAmpl > consumed))) { |
474 | 13 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
475 | 13 | "Maximum entity amplification factor exceeded, see " |
476 | 13 | "xmlCtxtSetMaxAmplification.\n"); |
477 | 13 | xmlHaltParser(ctxt); |
478 | 13 | return(1); |
479 | 13 | } |
480 | | |
481 | 675k | return(0); |
482 | 675k | } |
483 | | |
484 | | /************************************************************************ |
485 | | * * |
486 | | * Library wide options * |
487 | | * * |
488 | | ************************************************************************/ |
489 | | |
490 | | /** |
491 | | * Examines if the library has been compiled with a given feature. |
492 | | * |
493 | | * @param feature the feature to be examined |
494 | | * @returns zero (0) if the feature does not exist or an unknown |
495 | | * feature is requested, non-zero otherwise. |
496 | | */ |
497 | | int |
498 | | xmlHasFeature(xmlFeature feature) |
499 | 0 | { |
500 | 0 | switch (feature) { |
501 | 0 | case XML_WITH_THREAD: |
502 | 0 | #ifdef LIBXML_THREAD_ENABLED |
503 | 0 | return(1); |
504 | | #else |
505 | | return(0); |
506 | | #endif |
507 | 0 | case XML_WITH_TREE: |
508 | 0 | return(1); |
509 | 0 | case XML_WITH_OUTPUT: |
510 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
511 | 0 | return(1); |
512 | | #else |
513 | | return(0); |
514 | | #endif |
515 | 0 | case XML_WITH_PUSH: |
516 | 0 | #ifdef LIBXML_PUSH_ENABLED |
517 | 0 | return(1); |
518 | | #else |
519 | | return(0); |
520 | | #endif |
521 | 0 | case XML_WITH_READER: |
522 | 0 | #ifdef LIBXML_READER_ENABLED |
523 | 0 | return(1); |
524 | | #else |
525 | | return(0); |
526 | | #endif |
527 | 0 | case XML_WITH_PATTERN: |
528 | 0 | #ifdef LIBXML_PATTERN_ENABLED |
529 | 0 | return(1); |
530 | | #else |
531 | | return(0); |
532 | | #endif |
533 | 0 | case XML_WITH_WRITER: |
534 | 0 | #ifdef LIBXML_WRITER_ENABLED |
535 | 0 | return(1); |
536 | | #else |
537 | | return(0); |
538 | | #endif |
539 | 0 | case XML_WITH_SAX1: |
540 | 0 | #ifdef LIBXML_SAX1_ENABLED |
541 | 0 | return(1); |
542 | | #else |
543 | | return(0); |
544 | | #endif |
545 | 0 | case XML_WITH_HTTP: |
546 | 0 | return(0); |
547 | 0 | case XML_WITH_VALID: |
548 | 0 | #ifdef LIBXML_VALID_ENABLED |
549 | 0 | return(1); |
550 | | #else |
551 | | return(0); |
552 | | #endif |
553 | 0 | case XML_WITH_HTML: |
554 | 0 | #ifdef LIBXML_HTML_ENABLED |
555 | 0 | return(1); |
556 | | #else |
557 | | return(0); |
558 | | #endif |
559 | 0 | case XML_WITH_LEGACY: |
560 | 0 | return(0); |
561 | 0 | case XML_WITH_C14N: |
562 | 0 | #ifdef LIBXML_C14N_ENABLED |
563 | 0 | return(1); |
564 | | #else |
565 | | return(0); |
566 | | #endif |
567 | 0 | case XML_WITH_CATALOG: |
568 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
569 | 0 | return(1); |
570 | | #else |
571 | | return(0); |
572 | | #endif |
573 | 0 | case XML_WITH_XPATH: |
574 | 0 | #ifdef LIBXML_XPATH_ENABLED |
575 | 0 | return(1); |
576 | | #else |
577 | | return(0); |
578 | | #endif |
579 | 0 | case XML_WITH_XPTR: |
580 | 0 | #ifdef LIBXML_XPTR_ENABLED |
581 | 0 | return(1); |
582 | | #else |
583 | | return(0); |
584 | | #endif |
585 | 0 | case XML_WITH_XINCLUDE: |
586 | 0 | #ifdef LIBXML_XINCLUDE_ENABLED |
587 | 0 | return(1); |
588 | | #else |
589 | | return(0); |
590 | | #endif |
591 | 0 | case XML_WITH_ICONV: |
592 | 0 | #ifdef LIBXML_ICONV_ENABLED |
593 | 0 | return(1); |
594 | | #else |
595 | | return(0); |
596 | | #endif |
597 | 0 | case XML_WITH_ISO8859X: |
598 | 0 | #ifdef LIBXML_ISO8859X_ENABLED |
599 | 0 | return(1); |
600 | | #else |
601 | | return(0); |
602 | | #endif |
603 | 0 | case XML_WITH_UNICODE: |
604 | 0 | return(0); |
605 | 0 | case XML_WITH_REGEXP: |
606 | 0 | #ifdef LIBXML_REGEXP_ENABLED |
607 | 0 | return(1); |
608 | | #else |
609 | | return(0); |
610 | | #endif |
611 | 0 | case XML_WITH_AUTOMATA: |
612 | 0 | #ifdef LIBXML_REGEXP_ENABLED |
613 | 0 | return(1); |
614 | | #else |
615 | | return(0); |
616 | | #endif |
617 | 0 | case XML_WITH_EXPR: |
618 | | #ifdef LIBXML_EXPR_ENABLED |
619 | | return(1); |
620 | | #else |
621 | 0 | return(0); |
622 | 0 | #endif |
623 | 0 | case XML_WITH_RELAXNG: |
624 | 0 | #ifdef LIBXML_RELAXNG_ENABLED |
625 | 0 | return(1); |
626 | | #else |
627 | | return(0); |
628 | | #endif |
629 | 0 | case XML_WITH_SCHEMAS: |
630 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
631 | 0 | return(1); |
632 | | #else |
633 | | return(0); |
634 | | #endif |
635 | 0 | case XML_WITH_SCHEMATRON: |
636 | 0 | #ifdef LIBXML_SCHEMATRON_ENABLED |
637 | 0 | return(1); |
638 | | #else |
639 | | return(0); |
640 | | #endif |
641 | 0 | case XML_WITH_MODULES: |
642 | 0 | #ifdef LIBXML_MODULES_ENABLED |
643 | 0 | return(1); |
644 | | #else |
645 | | return(0); |
646 | | #endif |
647 | 0 | case XML_WITH_DEBUG: |
648 | 0 | #ifdef LIBXML_DEBUG_ENABLED |
649 | 0 | return(1); |
650 | | #else |
651 | | return(0); |
652 | | #endif |
653 | 0 | case XML_WITH_DEBUG_MEM: |
654 | 0 | return(0); |
655 | 0 | case XML_WITH_ZLIB: |
656 | | #ifdef LIBXML_ZLIB_ENABLED |
657 | | return(1); |
658 | | #else |
659 | 0 | return(0); |
660 | 0 | #endif |
661 | 0 | case XML_WITH_LZMA: |
662 | | #ifdef LIBXML_LZMA_ENABLED |
663 | | return(1); |
664 | | #else |
665 | 0 | return(0); |
666 | 0 | #endif |
667 | 0 | case XML_WITH_ICU: |
668 | | #ifdef LIBXML_ICU_ENABLED |
669 | | return(1); |
670 | | #else |
671 | 0 | return(0); |
672 | 0 | #endif |
673 | 0 | default: |
674 | 0 | break; |
675 | 0 | } |
676 | 0 | return(0); |
677 | 0 | } |
678 | | |
679 | | /************************************************************************ |
680 | | * * |
681 | | * Simple string buffer * |
682 | | * * |
683 | | ************************************************************************/ |
684 | | |
685 | | typedef struct { |
686 | | xmlChar *mem; |
687 | | unsigned size; |
688 | | unsigned cap; /* size < cap */ |
689 | | unsigned max; /* size <= max */ |
690 | | xmlParserErrors code; |
691 | | } xmlSBuf; |
692 | | |
693 | | static void |
694 | 114k | xmlSBufInit(xmlSBuf *buf, unsigned max) { |
695 | 114k | buf->mem = NULL; |
696 | 114k | buf->size = 0; |
697 | 114k | buf->cap = 0; |
698 | 114k | buf->max = max; |
699 | 114k | buf->code = XML_ERR_OK; |
700 | 114k | } |
701 | | |
702 | | static int |
703 | 78.7k | xmlSBufGrow(xmlSBuf *buf, unsigned len) { |
704 | 78.7k | xmlChar *mem; |
705 | 78.7k | unsigned cap; |
706 | | |
707 | 78.7k | if (len >= UINT_MAX / 2 - buf->size) { |
708 | 0 | if (buf->code == XML_ERR_OK) |
709 | 0 | buf->code = XML_ERR_RESOURCE_LIMIT; |
710 | 0 | return(-1); |
711 | 0 | } |
712 | | |
713 | 78.7k | cap = (buf->size + len) * 2; |
714 | 78.7k | if (cap < 240) |
715 | 67.3k | cap = 240; |
716 | | |
717 | 78.7k | mem = xmlRealloc(buf->mem, cap); |
718 | 78.7k | if (mem == NULL) { |
719 | 0 | buf->code = XML_ERR_NO_MEMORY; |
720 | 0 | return(-1); |
721 | 0 | } |
722 | | |
723 | 78.7k | buf->mem = mem; |
724 | 78.7k | buf->cap = cap; |
725 | | |
726 | 78.7k | return(0); |
727 | 78.7k | } |
728 | | |
729 | | static void |
730 | 1.76M | xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) { |
731 | 1.76M | if (buf->max - buf->size < len) { |
732 | 0 | if (buf->code == XML_ERR_OK) |
733 | 0 | buf->code = XML_ERR_RESOURCE_LIMIT; |
734 | 0 | return; |
735 | 0 | } |
736 | | |
737 | 1.76M | if (buf->cap - buf->size <= len) { |
738 | 76.4k | if (xmlSBufGrow(buf, len) < 0) |
739 | 0 | return; |
740 | 76.4k | } |
741 | | |
742 | 1.76M | if (len > 0) |
743 | 1.76M | memcpy(buf->mem + buf->size, str, len); |
744 | 1.76M | buf->size += len; |
745 | 1.76M | } |
746 | | |
747 | | static void |
748 | 532k | xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) { |
749 | 532k | xmlSBufAddString(buf, (const xmlChar *) str, len); |
750 | 532k | } |
751 | | |
752 | | static void |
753 | 24.1k | xmlSBufAddChar(xmlSBuf *buf, int c) { |
754 | 24.1k | xmlChar *end; |
755 | | |
756 | 24.1k | if (buf->max - buf->size < 4) { |
757 | 0 | if (buf->code == XML_ERR_OK) |
758 | 0 | buf->code = XML_ERR_RESOURCE_LIMIT; |
759 | 0 | return; |
760 | 0 | } |
761 | | |
762 | 24.1k | if (buf->cap - buf->size <= 4) { |
763 | 2.36k | if (xmlSBufGrow(buf, 4) < 0) |
764 | 0 | return; |
765 | 2.36k | } |
766 | | |
767 | 24.1k | end = buf->mem + buf->size; |
768 | | |
769 | 24.1k | if (c < 0x80) { |
770 | 13.7k | *end = (xmlChar) c; |
771 | 13.7k | buf->size += 1; |
772 | 13.7k | } else { |
773 | 10.3k | buf->size += xmlCopyCharMultiByte(end, c); |
774 | 10.3k | } |
775 | 24.1k | } |
776 | | |
777 | | static void |
778 | 18.1k | xmlSBufAddReplChar(xmlSBuf *buf) { |
779 | 18.1k | xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3); |
780 | 18.1k | } |
781 | | |
782 | | static void |
783 | 0 | xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) { |
784 | 0 | if (buf->code == XML_ERR_NO_MEMORY) |
785 | 0 | xmlCtxtErrMemory(ctxt); |
786 | 0 | else |
787 | 0 | xmlFatalErr(ctxt, buf->code, errMsg); |
788 | 0 | } |
789 | | |
790 | | static xmlChar * |
791 | | xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt, |
792 | 78.1k | const char *errMsg) { |
793 | 78.1k | if (buf->mem == NULL) { |
794 | 15.2k | buf->mem = xmlMalloc(1); |
795 | 15.2k | if (buf->mem == NULL) { |
796 | 0 | buf->code = XML_ERR_NO_MEMORY; |
797 | 15.2k | } else { |
798 | 15.2k | buf->mem[0] = 0; |
799 | 15.2k | } |
800 | 62.9k | } else { |
801 | 62.9k | buf->mem[buf->size] = 0; |
802 | 62.9k | } |
803 | | |
804 | 78.1k | if (buf->code == XML_ERR_OK) { |
805 | 78.1k | if (sizeOut != NULL) |
806 | 2.93k | *sizeOut = buf->size; |
807 | 78.1k | return(buf->mem); |
808 | 78.1k | } |
809 | | |
810 | 0 | xmlSBufReportError(buf, ctxt, errMsg); |
811 | |
|
812 | 0 | xmlFree(buf->mem); |
813 | |
|
814 | 0 | if (sizeOut != NULL) |
815 | 0 | *sizeOut = 0; |
816 | 0 | return(NULL); |
817 | 78.1k | } |
818 | | |
819 | | static void |
820 | 32.0k | xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) { |
821 | 32.0k | if (buf->code != XML_ERR_OK) |
822 | 0 | xmlSBufReportError(buf, ctxt, errMsg); |
823 | | |
824 | 32.0k | xmlFree(buf->mem); |
825 | 32.0k | } |
826 | | |
827 | | static int |
828 | | xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str, |
829 | 95.5k | const char *errMsg) { |
830 | 95.5k | int c = str[0]; |
831 | 95.5k | int c1 = str[1]; |
832 | | |
833 | 95.5k | if ((c1 & 0xC0) != 0x80) |
834 | 6.90k | goto encoding_error; |
835 | | |
836 | 88.6k | if (c < 0xE0) { |
837 | | /* 2-byte sequence */ |
838 | 14.2k | if (c < 0xC2) |
839 | 4.73k | goto encoding_error; |
840 | | |
841 | 9.52k | return(2); |
842 | 74.3k | } else { |
843 | 74.3k | int c2 = str[2]; |
844 | | |
845 | 74.3k | if ((c2 & 0xC0) != 0x80) |
846 | 256 | goto encoding_error; |
847 | | |
848 | 74.1k | if (c < 0xF0) { |
849 | | /* 3-byte sequence */ |
850 | 71.8k | if (c == 0xE0) { |
851 | | /* overlong */ |
852 | 1.44k | if (c1 < 0xA0) |
853 | 66 | goto encoding_error; |
854 | 70.4k | } else if (c == 0xED) { |
855 | | /* surrogate */ |
856 | 283 | if (c1 >= 0xA0) |
857 | 67 | goto encoding_error; |
858 | 70.1k | } else if (c == 0xEF) { |
859 | | /* U+FFFE and U+FFFF are invalid Chars */ |
860 | 25.0k | if ((c1 == 0xBF) && (c2 >= 0xBE)) |
861 | 273 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg); |
862 | 25.0k | } |
863 | | |
864 | 71.7k | return(3); |
865 | 71.8k | } else { |
866 | | /* 4-byte sequence */ |
867 | 2.25k | if ((str[3] & 0xC0) != 0x80) |
868 | 67 | goto encoding_error; |
869 | 2.19k | if (c == 0xF0) { |
870 | | /* overlong */ |
871 | 530 | if (c1 < 0x90) |
872 | 66 | goto encoding_error; |
873 | 1.66k | } else if (c >= 0xF4) { |
874 | | /* greater than 0x10FFFF */ |
875 | 906 | if ((c > 0xF4) || (c1 >= 0x90)) |
876 | 696 | goto encoding_error; |
877 | 906 | } |
878 | | |
879 | 1.42k | return(4); |
880 | 2.19k | } |
881 | 74.1k | } |
882 | | |
883 | 12.8k | encoding_error: |
884 | | /* Only report the first error */ |
885 | 12.8k | if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) { |
886 | 665 | xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL); |
887 | 665 | ctxt->input->flags |= XML_INPUT_ENCODING_ERROR; |
888 | 665 | } |
889 | | |
890 | 12.8k | return(0); |
891 | 88.6k | } |
892 | | |
893 | | /************************************************************************ |
894 | | * * |
895 | | * SAX2 defaulted attributes handling * |
896 | | * * |
897 | | ************************************************************************/ |
898 | | |
899 | | /** |
900 | | * Final initialization of the parser context before starting to parse. |
901 | | * |
902 | | * This accounts for users modifying struct members of parser context |
903 | | * directly. |
904 | | * |
905 | | * @param ctxt an XML parser context |
906 | | */ |
907 | | static void |
908 | 21.9k | xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) { |
909 | 21.9k | xmlSAXHandlerPtr sax; |
910 | | |
911 | | /* Avoid unused variable warning if features are disabled. */ |
912 | 21.9k | (void) sax; |
913 | | |
914 | | /* |
915 | | * Changing the SAX struct directly is still widespread practice |
916 | | * in internal and external code. |
917 | | */ |
918 | 21.9k | if (ctxt == NULL) return; |
919 | 21.9k | sax = ctxt->sax; |
920 | 21.9k | #ifdef LIBXML_SAX1_ENABLED |
921 | | /* |
922 | | * Only enable SAX2 if there SAX2 element handlers, except when there |
923 | | * are no element handlers at all. |
924 | | */ |
925 | 21.9k | if (((ctxt->options & XML_PARSE_SAX1) == 0) && |
926 | 21.9k | (sax) && |
927 | 21.9k | (sax->initialized == XML_SAX2_MAGIC) && |
928 | 21.9k | ((sax->startElementNs != NULL) || |
929 | 18.5k | (sax->endElementNs != NULL) || |
930 | 18.5k | ((sax->startElement == NULL) && (sax->endElement == NULL)))) |
931 | 18.5k | ctxt->sax2 = 1; |
932 | | #else |
933 | | ctxt->sax2 = 1; |
934 | | #endif /* LIBXML_SAX1_ENABLED */ |
935 | | |
936 | | /* |
937 | | * Some users replace the dictionary directly in the context struct. |
938 | | * We really need an API function to do that cleanly. |
939 | | */ |
940 | 21.9k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
941 | 21.9k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
942 | 21.9k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
943 | 21.9k | if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) || |
944 | 21.9k | (ctxt->str_xml_ns == NULL)) { |
945 | 0 | xmlErrMemory(ctxt); |
946 | 0 | } |
947 | | |
948 | 21.9k | xmlDictSetLimit(ctxt->dict, |
949 | 21.9k | (ctxt->options & XML_PARSE_HUGE) ? |
950 | 0 | 0 : |
951 | 21.9k | XML_MAX_DICTIONARY_LIMIT); |
952 | 21.9k | } |
953 | | |
954 | | typedef struct { |
955 | | xmlHashedString prefix; |
956 | | xmlHashedString name; |
957 | | xmlHashedString value; |
958 | | const xmlChar *valueEnd; |
959 | | int external; |
960 | | int expandedSize; |
961 | | } xmlDefAttr; |
962 | | |
963 | | typedef struct _xmlDefAttrs xmlDefAttrs; |
964 | | typedef xmlDefAttrs *xmlDefAttrsPtr; |
965 | | struct _xmlDefAttrs { |
966 | | int nbAttrs; /* number of defaulted attributes on that element */ |
967 | | int maxAttrs; /* the size of the array */ |
968 | | #if __STDC_VERSION__ >= 199901L |
969 | | /* Using a C99 flexible array member avoids UBSan errors. */ |
970 | | xmlDefAttr attrs[] ATTRIBUTE_COUNTED_BY(maxAttrs); |
971 | | #else |
972 | | xmlDefAttr attrs[1]; |
973 | | #endif |
974 | | }; |
975 | | |
976 | | /** |
977 | | * Normalize the space in non CDATA attribute values: |
978 | | * If the attribute type is not CDATA, then the XML processor MUST further |
979 | | * process the normalized attribute value by discarding any leading and |
980 | | * trailing space (\#x20) characters, and by replacing sequences of space |
981 | | * (\#x20) characters by a single space (\#x20) character. |
982 | | * Note that the size of dst need to be at least src, and if one doesn't need |
983 | | * to preserve dst (and it doesn't come from a dictionary or read-only) then |
984 | | * passing src as dst is just fine. |
985 | | * |
986 | | * @param src the source string |
987 | | * @param dst the target string |
988 | | * @returns a pointer to the normalized value (dst) or NULL if no conversion |
989 | | * is needed. |
990 | | */ |
991 | | static xmlChar * |
992 | | xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst) |
993 | 22.3k | { |
994 | 22.3k | if ((src == NULL) || (dst == NULL)) |
995 | 0 | return(NULL); |
996 | | |
997 | 28.4k | while (*src == 0x20) src++; |
998 | 119k | while (*src != 0) { |
999 | 96.8k | if (*src == 0x20) { |
1000 | 20.7k | while (*src == 0x20) src++; |
1001 | 10.2k | if (*src != 0) |
1002 | 9.55k | *dst++ = 0x20; |
1003 | 86.6k | } else { |
1004 | 86.6k | *dst++ = *src++; |
1005 | 86.6k | } |
1006 | 96.8k | } |
1007 | 22.3k | *dst = 0; |
1008 | 22.3k | if (dst == src) |
1009 | 16.0k | return(NULL); |
1010 | 6.34k | return(dst); |
1011 | 22.3k | } |
1012 | | |
1013 | | /** |
1014 | | * Add a defaulted attribute for an element |
1015 | | * |
1016 | | * @param ctxt an XML parser context |
1017 | | * @param fullname the element fullname |
1018 | | * @param fullattr the attribute fullname |
1019 | | * @param value the attribute value |
1020 | | */ |
1021 | | static void |
1022 | | xmlAddDefAttrs(xmlParserCtxtPtr ctxt, |
1023 | | const xmlChar *fullname, |
1024 | | const xmlChar *fullattr, |
1025 | 21.1k | const xmlChar *value) { |
1026 | 21.1k | xmlDefAttrsPtr defaults; |
1027 | 21.1k | xmlDefAttr *attr; |
1028 | 21.1k | int len, expandedSize; |
1029 | 21.1k | xmlHashedString name; |
1030 | 21.1k | xmlHashedString prefix; |
1031 | 21.1k | xmlHashedString hvalue; |
1032 | 21.1k | const xmlChar *localname; |
1033 | | |
1034 | | /* |
1035 | | * Allows to detect attribute redefinitions |
1036 | | */ |
1037 | 21.1k | if (ctxt->attsSpecial != NULL) { |
1038 | 18.5k | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1039 | 13.9k | return; |
1040 | 18.5k | } |
1041 | | |
1042 | 7.17k | if (ctxt->attsDefault == NULL) { |
1043 | 2.60k | ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict); |
1044 | 2.60k | if (ctxt->attsDefault == NULL) |
1045 | 0 | goto mem_error; |
1046 | 2.60k | } |
1047 | | |
1048 | | /* |
1049 | | * split the element name into prefix:localname , the string found |
1050 | | * are within the DTD and then not associated to namespace names. |
1051 | | */ |
1052 | 7.17k | localname = xmlSplitQName3(fullname, &len); |
1053 | 7.17k | if (localname == NULL) { |
1054 | 6.80k | name = xmlDictLookupHashed(ctxt->dict, fullname, -1); |
1055 | 6.80k | prefix.name = NULL; |
1056 | 6.80k | } else { |
1057 | 370 | name = xmlDictLookupHashed(ctxt->dict, localname, -1); |
1058 | 370 | prefix = xmlDictLookupHashed(ctxt->dict, fullname, len); |
1059 | 370 | if (prefix.name == NULL) |
1060 | 0 | goto mem_error; |
1061 | 370 | } |
1062 | 7.17k | if (name.name == NULL) |
1063 | 0 | goto mem_error; |
1064 | | |
1065 | | /* |
1066 | | * make sure there is some storage |
1067 | | */ |
1068 | 7.17k | defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name); |
1069 | 7.17k | if ((defaults == NULL) || |
1070 | 7.17k | (defaults->nbAttrs >= defaults->maxAttrs)) { |
1071 | 3.31k | xmlDefAttrsPtr temp; |
1072 | 3.31k | int newSize; |
1073 | | |
1074 | 3.31k | if (defaults == NULL) { |
1075 | 2.73k | newSize = 4; |
1076 | 2.73k | } else { |
1077 | 575 | if ((defaults->maxAttrs >= XML_MAX_ATTRS) || |
1078 | 575 | ((size_t) defaults->maxAttrs > |
1079 | 575 | SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults))) |
1080 | 0 | goto mem_error; |
1081 | | |
1082 | 575 | if (defaults->maxAttrs > XML_MAX_ATTRS / 2) |
1083 | 0 | newSize = XML_MAX_ATTRS; |
1084 | 575 | else |
1085 | 575 | newSize = defaults->maxAttrs * 2; |
1086 | 575 | } |
1087 | 3.31k | temp = xmlRealloc(defaults, |
1088 | 3.31k | sizeof(*defaults) + newSize * sizeof(xmlDefAttr)); |
1089 | 3.31k | if (temp == NULL) |
1090 | 0 | goto mem_error; |
1091 | 3.31k | if (defaults == NULL) |
1092 | 2.73k | temp->nbAttrs = 0; |
1093 | 3.31k | temp->maxAttrs = newSize; |
1094 | 3.31k | defaults = temp; |
1095 | 3.31k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name, |
1096 | 3.31k | defaults, NULL) < 0) { |
1097 | 0 | xmlFree(defaults); |
1098 | 0 | goto mem_error; |
1099 | 0 | } |
1100 | 3.31k | } |
1101 | | |
1102 | | /* |
1103 | | * Split the attribute name into prefix:localname , the string found |
1104 | | * are within the DTD and hen not associated to namespace names. |
1105 | | */ |
1106 | 7.17k | localname = xmlSplitQName3(fullattr, &len); |
1107 | 7.17k | if (localname == NULL) { |
1108 | 4.84k | name = xmlDictLookupHashed(ctxt->dict, fullattr, -1); |
1109 | 4.84k | prefix.name = NULL; |
1110 | 4.84k | } else { |
1111 | 2.32k | name = xmlDictLookupHashed(ctxt->dict, localname, -1); |
1112 | 2.32k | prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len); |
1113 | 2.32k | if (prefix.name == NULL) |
1114 | 0 | goto mem_error; |
1115 | 2.32k | } |
1116 | 7.17k | if (name.name == NULL) |
1117 | 0 | goto mem_error; |
1118 | | |
1119 | | /* intern the string and precompute the end */ |
1120 | 7.17k | len = strlen((const char *) value); |
1121 | 7.17k | hvalue = xmlDictLookupHashed(ctxt->dict, value, len); |
1122 | 7.17k | if (hvalue.name == NULL) |
1123 | 0 | goto mem_error; |
1124 | | |
1125 | 7.17k | expandedSize = strlen((const char *) name.name); |
1126 | 7.17k | if (prefix.name != NULL) |
1127 | 2.32k | expandedSize += strlen((const char *) prefix.name); |
1128 | 7.17k | expandedSize += len; |
1129 | | |
1130 | 7.17k | attr = &defaults->attrs[defaults->nbAttrs++]; |
1131 | 7.17k | attr->name = name; |
1132 | 7.17k | attr->prefix = prefix; |
1133 | 7.17k | attr->value = hvalue; |
1134 | 7.17k | attr->valueEnd = hvalue.name + len; |
1135 | 7.17k | attr->external = PARSER_EXTERNAL(ctxt); |
1136 | 7.17k | attr->expandedSize = expandedSize; |
1137 | | |
1138 | 7.17k | return; |
1139 | | |
1140 | 0 | mem_error: |
1141 | 0 | xmlErrMemory(ctxt); |
1142 | 0 | } |
1143 | | |
1144 | | /** |
1145 | | * Register this attribute type |
1146 | | * |
1147 | | * @param ctxt an XML parser context |
1148 | | * @param fullname the element fullname |
1149 | | * @param fullattr the attribute fullname |
1150 | | * @param type the attribute type |
1151 | | */ |
1152 | | static void |
1153 | | xmlAddSpecialAttr(xmlParserCtxtPtr ctxt, |
1154 | | const xmlChar *fullname, |
1155 | | const xmlChar *fullattr, |
1156 | | int type) |
1157 | 31.8k | { |
1158 | 31.8k | if (ctxt->attsSpecial == NULL) { |
1159 | 3.14k | ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict); |
1160 | 3.14k | if (ctxt->attsSpecial == NULL) |
1161 | 0 | goto mem_error; |
1162 | 3.14k | } |
1163 | | |
1164 | 31.8k | if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr, |
1165 | 31.8k | XML_INT_TO_PTR(type)) < 0) |
1166 | 0 | goto mem_error; |
1167 | 31.8k | return; |
1168 | | |
1169 | 31.8k | mem_error: |
1170 | 0 | xmlErrMemory(ctxt); |
1171 | 0 | } |
1172 | | |
1173 | | /** |
1174 | | * Removes CDATA attributes from the special attribute table |
1175 | | */ |
1176 | | static void |
1177 | | xmlCleanSpecialAttrCallback(void *payload, void *data, |
1178 | | const xmlChar *fullname, const xmlChar *fullattr, |
1179 | 7.96k | const xmlChar *unused ATTRIBUTE_UNUSED) { |
1180 | 7.96k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data; |
1181 | | |
1182 | 7.96k | if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) { |
1183 | 992 | xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL); |
1184 | 992 | } |
1185 | 7.96k | } |
1186 | | |
1187 | | /** |
1188 | | * Trim the list of attributes defined to remove all those of type |
1189 | | * CDATA as they are not special. This call should be done when finishing |
1190 | | * to parse the DTD and before starting to parse the document root. |
1191 | | * |
1192 | | * @param ctxt an XML parser context |
1193 | | */ |
1194 | | static void |
1195 | | xmlCleanSpecialAttr(xmlParserCtxtPtr ctxt) |
1196 | 10.7k | { |
1197 | 10.7k | if (ctxt->attsSpecial == NULL) |
1198 | 7.65k | return; |
1199 | | |
1200 | 3.14k | xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt); |
1201 | | |
1202 | 3.14k | if (xmlHashSize(ctxt->attsSpecial) == 0) { |
1203 | 308 | xmlHashFree(ctxt->attsSpecial, NULL); |
1204 | 308 | ctxt->attsSpecial = NULL; |
1205 | 308 | } |
1206 | 3.14k | } |
1207 | | |
1208 | | /** |
1209 | | * Checks that the value conforms to the LanguageID production: |
1210 | | * |
1211 | | * @deprecated Internal function, do not use. |
1212 | | * |
1213 | | * NOTE: this is somewhat deprecated, those productions were removed from |
1214 | | * the XML Second edition. |
1215 | | * |
1216 | | * [33] LanguageID ::= Langcode ('-' Subcode)* |
1217 | | * [34] Langcode ::= ISO639Code | IanaCode | UserCode |
1218 | | * [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z]) |
1219 | | * [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+ |
1220 | | * [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+ |
1221 | | * [38] Subcode ::= ([a-z] | [A-Z])+ |
1222 | | * |
1223 | | * The current REC reference the successors of RFC 1766, currently 5646 |
1224 | | * |
1225 | | * http://www.rfc-editor.org/rfc/rfc5646.txt |
1226 | | * |
1227 | | * langtag = language |
1228 | | * ["-" script] |
1229 | | * ["-" region] |
1230 | | * *("-" variant) |
1231 | | * *("-" extension) |
1232 | | * ["-" privateuse] |
1233 | | * language = 2*3ALPHA ; shortest ISO 639 code |
1234 | | * ["-" extlang] ; sometimes followed by |
1235 | | * ; extended language subtags |
1236 | | * / 4ALPHA ; or reserved for future use |
1237 | | * / 5*8ALPHA ; or registered language subtag |
1238 | | * |
1239 | | * extlang = 3ALPHA ; selected ISO 639 codes |
1240 | | * *2("-" 3ALPHA) ; permanently reserved |
1241 | | * |
1242 | | * script = 4ALPHA ; ISO 15924 code |
1243 | | * |
1244 | | * region = 2ALPHA ; ISO 3166-1 code |
1245 | | * / 3DIGIT ; UN M.49 code |
1246 | | * |
1247 | | * variant = 5*8alphanum ; registered variants |
1248 | | * / (DIGIT 3alphanum) |
1249 | | * |
1250 | | * extension = singleton 1*("-" (2*8alphanum)) |
1251 | | * |
1252 | | * ; Single alphanumerics |
1253 | | * ; "x" reserved for private use |
1254 | | * singleton = DIGIT ; 0 - 9 |
1255 | | * / %x41-57 ; A - W |
1256 | | * / %x59-5A ; Y - Z |
1257 | | * / %x61-77 ; a - w |
1258 | | * / %x79-7A ; y - z |
1259 | | * |
1260 | | * it sounds right to still allow Irregular i-xxx IANA and user codes too |
1261 | | * The parser below doesn't try to cope with extension or privateuse |
1262 | | * that could be added but that's not interoperable anyway |
1263 | | * |
1264 | | * @param lang pointer to the string value |
1265 | | * @returns 1 if correct 0 otherwise |
1266 | | **/ |
1267 | | int |
1268 | | xmlCheckLanguageID(const xmlChar * lang) |
1269 | 2.89k | { |
1270 | 2.89k | const xmlChar *cur = lang, *nxt; |
1271 | | |
1272 | 2.89k | if (cur == NULL) |
1273 | 117 | return (0); |
1274 | 2.77k | if (((cur[0] == 'i') && (cur[1] == '-')) || |
1275 | 2.77k | ((cur[0] == 'I') && (cur[1] == '-')) || |
1276 | 2.77k | ((cur[0] == 'x') && (cur[1] == '-')) || |
1277 | 2.77k | ((cur[0] == 'X') && (cur[1] == '-'))) { |
1278 | | /* |
1279 | | * Still allow IANA code and user code which were coming |
1280 | | * from the previous version of the XML-1.0 specification |
1281 | | * it's deprecated but we should not fail |
1282 | | */ |
1283 | 289 | cur += 2; |
1284 | 967 | while (((cur[0] >= 'A') && (cur[0] <= 'Z')) || |
1285 | 967 | ((cur[0] >= 'a') && (cur[0] <= 'z'))) |
1286 | 678 | cur++; |
1287 | 289 | return(cur[0] == 0); |
1288 | 289 | } |
1289 | 2.48k | nxt = cur; |
1290 | 8.79k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1291 | 8.79k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1292 | 6.30k | nxt++; |
1293 | 2.48k | if (nxt - cur >= 4) { |
1294 | | /* |
1295 | | * Reserved |
1296 | | */ |
1297 | 218 | if ((nxt - cur > 8) || (nxt[0] != 0)) |
1298 | 149 | return(0); |
1299 | 69 | return(1); |
1300 | 218 | } |
1301 | 2.27k | if (nxt - cur < 2) |
1302 | 133 | return(0); |
1303 | | /* we got an ISO 639 code */ |
1304 | 2.13k | if (nxt[0] == 0) |
1305 | 123 | return(1); |
1306 | 2.01k | if (nxt[0] != '-') |
1307 | 80 | return(0); |
1308 | | |
1309 | 1.93k | nxt++; |
1310 | 1.93k | cur = nxt; |
1311 | | /* now we can have extlang or script or region or variant */ |
1312 | 1.93k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1313 | 203 | goto region_m49; |
1314 | | |
1315 | 8.26k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1316 | 8.26k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1317 | 6.53k | nxt++; |
1318 | 1.73k | if (nxt - cur == 4) |
1319 | 593 | goto script; |
1320 | 1.13k | if (nxt - cur == 2) |
1321 | 243 | goto region; |
1322 | 896 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1323 | 164 | goto variant; |
1324 | 732 | if (nxt - cur != 3) |
1325 | 151 | return(0); |
1326 | | /* we parsed an extlang */ |
1327 | 581 | if (nxt[0] == 0) |
1328 | 68 | return(1); |
1329 | 513 | if (nxt[0] != '-') |
1330 | 67 | return(0); |
1331 | | |
1332 | 446 | nxt++; |
1333 | 446 | cur = nxt; |
1334 | | /* now we can have script or region or variant */ |
1335 | 446 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1336 | 82 | goto region_m49; |
1337 | | |
1338 | 2.25k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1339 | 2.25k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1340 | 1.89k | nxt++; |
1341 | 364 | if (nxt - cur == 2) |
1342 | 74 | goto region; |
1343 | 290 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1344 | 69 | goto variant; |
1345 | 221 | if (nxt - cur != 4) |
1346 | 152 | return(0); |
1347 | | /* we parsed a script */ |
1348 | 662 | script: |
1349 | 662 | if (nxt[0] == 0) |
1350 | 159 | return(1); |
1351 | 503 | if (nxt[0] != '-') |
1352 | 70 | return(0); |
1353 | | |
1354 | 433 | nxt++; |
1355 | 433 | cur = nxt; |
1356 | | /* now we can have region or variant */ |
1357 | 433 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1358 | 66 | goto region_m49; |
1359 | | |
1360 | 2.00k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1361 | 2.00k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1362 | 1.63k | nxt++; |
1363 | | |
1364 | 367 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1365 | 72 | goto variant; |
1366 | 295 | if (nxt - cur != 2) |
1367 | 151 | return(0); |
1368 | | /* we parsed a region */ |
1369 | 532 | region: |
1370 | 532 | if (nxt[0] == 0) |
1371 | 125 | return(1); |
1372 | 407 | if (nxt[0] != '-') |
1373 | 124 | return(0); |
1374 | | |
1375 | 283 | nxt++; |
1376 | 283 | cur = nxt; |
1377 | | /* now we can just have a variant */ |
1378 | 1.74k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1379 | 1.74k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1380 | 1.45k | nxt++; |
1381 | | |
1382 | 283 | if ((nxt - cur < 5) || (nxt - cur > 8)) |
1383 | 215 | return(0); |
1384 | | |
1385 | | /* we parsed a variant */ |
1386 | 373 | variant: |
1387 | 373 | if (nxt[0] == 0) |
1388 | 253 | return(1); |
1389 | 120 | if (nxt[0] != '-') |
1390 | 47 | return(0); |
1391 | | /* extensions and private use subtags not checked */ |
1392 | 73 | return (1); |
1393 | | |
1394 | 351 | region_m49: |
1395 | 351 | if (((nxt[1] >= '0') && (nxt[1] <= '9')) && |
1396 | 351 | ((nxt[2] >= '0') && (nxt[2] <= '9'))) { |
1397 | 71 | nxt += 3; |
1398 | 71 | goto region; |
1399 | 71 | } |
1400 | 280 | return(0); |
1401 | 351 | } |
1402 | | |
1403 | | /************************************************************************ |
1404 | | * * |
1405 | | * Parser stacks related functions and macros * |
1406 | | * * |
1407 | | ************************************************************************/ |
1408 | | |
1409 | | static xmlChar * |
1410 | | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar **str); |
1411 | | |
1412 | | /** |
1413 | | * Create a new namespace database. |
1414 | | * |
1415 | | * @returns the new obejct. |
1416 | | */ |
1417 | | xmlParserNsData * |
1418 | 21.9k | xmlParserNsCreate(void) { |
1419 | 21.9k | xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb)); |
1420 | | |
1421 | 21.9k | if (nsdb == NULL) |
1422 | 0 | return(NULL); |
1423 | 21.9k | memset(nsdb, 0, sizeof(*nsdb)); |
1424 | 21.9k | nsdb->defaultNsIndex = INT_MAX; |
1425 | | |
1426 | 21.9k | return(nsdb); |
1427 | 21.9k | } |
1428 | | |
1429 | | /** |
1430 | | * Free a namespace database. |
1431 | | * |
1432 | | * @param nsdb namespace database |
1433 | | */ |
1434 | | void |
1435 | 21.9k | xmlParserNsFree(xmlParserNsData *nsdb) { |
1436 | 21.9k | if (nsdb == NULL) |
1437 | 0 | return; |
1438 | | |
1439 | 21.9k | xmlFree(nsdb->extra); |
1440 | 21.9k | xmlFree(nsdb->hash); |
1441 | 21.9k | xmlFree(nsdb); |
1442 | 21.9k | } |
1443 | | |
1444 | | /** |
1445 | | * Reset a namespace database. |
1446 | | * |
1447 | | * @param nsdb namespace database |
1448 | | */ |
1449 | | static void |
1450 | 0 | xmlParserNsReset(xmlParserNsData *nsdb) { |
1451 | 0 | if (nsdb == NULL) |
1452 | 0 | return; |
1453 | | |
1454 | 0 | nsdb->hashElems = 0; |
1455 | 0 | nsdb->elementId = 0; |
1456 | 0 | nsdb->defaultNsIndex = INT_MAX; |
1457 | |
|
1458 | 0 | if (nsdb->hash) |
1459 | 0 | memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0])); |
1460 | 0 | } |
1461 | | |
1462 | | /** |
1463 | | * Signal that a new element has started. |
1464 | | * |
1465 | | * @param nsdb namespace database |
1466 | | * @returns 0 on success, -1 if the element counter overflowed. |
1467 | | */ |
1468 | | static int |
1469 | 98.7k | xmlParserNsStartElement(xmlParserNsData *nsdb) { |
1470 | 98.7k | if (nsdb->elementId == UINT_MAX) |
1471 | 0 | return(-1); |
1472 | 98.7k | nsdb->elementId++; |
1473 | | |
1474 | 98.7k | return(0); |
1475 | 98.7k | } |
1476 | | |
1477 | | /** |
1478 | | * Lookup namespace with given prefix. If `bucketPtr` is non-NULL, it will |
1479 | | * be set to the matching bucket, or the first empty bucket if no match |
1480 | | * was found. |
1481 | | * |
1482 | | * @param ctxt parser context |
1483 | | * @param prefix namespace prefix |
1484 | | * @param bucketPtr optional bucket (return value) |
1485 | | * @returns the namespace index on success, INT_MAX if no namespace was |
1486 | | * found. |
1487 | | */ |
1488 | | static int |
1489 | | xmlParserNsLookup(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix, |
1490 | 273k | xmlParserNsBucket **bucketPtr) { |
1491 | 273k | xmlParserNsBucket *bucket, *tombstone; |
1492 | 273k | unsigned index, hashValue; |
1493 | | |
1494 | 273k | if (prefix->name == NULL) |
1495 | 82.1k | return(ctxt->nsdb->defaultNsIndex); |
1496 | | |
1497 | 191k | if (ctxt->nsdb->hashSize == 0) |
1498 | 6.52k | return(INT_MAX); |
1499 | | |
1500 | 185k | hashValue = prefix->hashValue; |
1501 | 185k | index = hashValue & (ctxt->nsdb->hashSize - 1); |
1502 | 185k | bucket = &ctxt->nsdb->hash[index]; |
1503 | 185k | tombstone = NULL; |
1504 | | |
1505 | 283k | while (bucket->hashValue) { |
1506 | 256k | if (bucket->index == INT_MAX) { |
1507 | 31.6k | if (tombstone == NULL) |
1508 | 20.1k | tombstone = bucket; |
1509 | 225k | } else if (bucket->hashValue == hashValue) { |
1510 | 158k | if (ctxt->nsTab[bucket->index * 2] == prefix->name) { |
1511 | 158k | if (bucketPtr != NULL) |
1512 | 129k | *bucketPtr = bucket; |
1513 | 158k | return(bucket->index); |
1514 | 158k | } |
1515 | 158k | } |
1516 | | |
1517 | 98.5k | index++; |
1518 | 98.5k | bucket++; |
1519 | 98.5k | if (index == ctxt->nsdb->hashSize) { |
1520 | 13.3k | index = 0; |
1521 | 13.3k | bucket = ctxt->nsdb->hash; |
1522 | 13.3k | } |
1523 | 98.5k | } |
1524 | | |
1525 | 26.9k | if (bucketPtr != NULL) |
1526 | 19.4k | *bucketPtr = tombstone ? tombstone : bucket; |
1527 | 26.9k | return(INT_MAX); |
1528 | 185k | } |
1529 | | |
1530 | | /** |
1531 | | * Lookup namespace URI with given prefix. |
1532 | | * |
1533 | | * @param ctxt parser context |
1534 | | * @param prefix namespace prefix |
1535 | | * @returns the namespace URI on success, NULL if no namespace was found. |
1536 | | */ |
1537 | | static const xmlChar * |
1538 | 81.7k | xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) { |
1539 | 81.7k | const xmlChar *ret; |
1540 | 81.7k | int nsIndex; |
1541 | | |
1542 | 81.7k | if (prefix->name == ctxt->str_xml) |
1543 | 939 | return(ctxt->str_xml_ns); |
1544 | | |
1545 | | /* |
1546 | | * minNsIndex is used when building an entity tree. We must |
1547 | | * ignore namespaces declared outside the entity. |
1548 | | */ |
1549 | 80.7k | nsIndex = xmlParserNsLookup(ctxt, prefix, NULL); |
1550 | 80.7k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1551 | 67.6k | return(NULL); |
1552 | | |
1553 | 13.0k | ret = ctxt->nsTab[nsIndex * 2 + 1]; |
1554 | 13.0k | if (ret[0] == 0) |
1555 | 4.58k | ret = NULL; |
1556 | 13.0k | return(ret); |
1557 | 80.7k | } |
1558 | | |
1559 | | /** |
1560 | | * Lookup extra data for the given prefix. This returns data stored |
1561 | | * with xmlParserNsUdpateSax(). |
1562 | | * |
1563 | | * @param ctxt parser context |
1564 | | * @param prefix namespace prefix |
1565 | | * @returns the data on success, NULL if no namespace was found. |
1566 | | */ |
1567 | | void * |
1568 | 6.58k | xmlParserNsLookupSax(xmlParserCtxtPtr ctxt, const xmlChar *prefix) { |
1569 | 6.58k | xmlHashedString hprefix; |
1570 | 6.58k | int nsIndex; |
1571 | | |
1572 | 6.58k | if (prefix == ctxt->str_xml) |
1573 | 3.99k | return(NULL); |
1574 | | |
1575 | 2.58k | hprefix.name = prefix; |
1576 | 2.58k | if (prefix != NULL) |
1577 | 1.25k | hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix); |
1578 | 1.32k | else |
1579 | 1.32k | hprefix.hashValue = 0; |
1580 | 2.58k | nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL); |
1581 | 2.58k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1582 | 0 | return(NULL); |
1583 | | |
1584 | 2.58k | return(ctxt->nsdb->extra[nsIndex].saxData); |
1585 | 2.58k | } |
1586 | | |
1587 | | /** |
1588 | | * Sets or updates extra data for the given prefix. This value will be |
1589 | | * returned by xmlParserNsLookupSax() as long as the namespace with the |
1590 | | * given prefix is in scope. |
1591 | | * |
1592 | | * @param ctxt parser context |
1593 | | * @param prefix namespace prefix |
1594 | | * @param saxData extra data for SAX handler |
1595 | | * @returns the data on success, NULL if no namespace was found. |
1596 | | */ |
1597 | | int |
1598 | | xmlParserNsUpdateSax(xmlParserCtxtPtr ctxt, const xmlChar *prefix, |
1599 | 25.0k | void *saxData) { |
1600 | 25.0k | xmlHashedString hprefix; |
1601 | 25.0k | int nsIndex; |
1602 | | |
1603 | 25.0k | if (prefix == ctxt->str_xml) |
1604 | 0 | return(-1); |
1605 | | |
1606 | 25.0k | hprefix.name = prefix; |
1607 | 25.0k | if (prefix != NULL) |
1608 | 20.1k | hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix); |
1609 | 4.87k | else |
1610 | 4.87k | hprefix.hashValue = 0; |
1611 | 25.0k | nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL); |
1612 | 25.0k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1613 | 0 | return(-1); |
1614 | | |
1615 | 25.0k | ctxt->nsdb->extra[nsIndex].saxData = saxData; |
1616 | 25.0k | return(0); |
1617 | 25.0k | } |
1618 | | |
1619 | | /** |
1620 | | * Grows the namespace tables. |
1621 | | * |
1622 | | * @param ctxt parser context |
1623 | | * @returns 0 on success, -1 if a memory allocation failed. |
1624 | | */ |
1625 | | static int |
1626 | 7.37k | xmlParserNsGrow(xmlParserCtxtPtr ctxt) { |
1627 | 7.37k | const xmlChar **table; |
1628 | 7.37k | xmlParserNsExtra *extra; |
1629 | 7.37k | int newSize; |
1630 | | |
1631 | 7.37k | newSize = xmlGrowCapacity(ctxt->nsMax, |
1632 | 7.37k | sizeof(table[0]) + sizeof(extra[0]), |
1633 | 7.37k | 16, XML_MAX_ITEMS); |
1634 | 7.37k | if (newSize < 0) |
1635 | 0 | goto error; |
1636 | | |
1637 | 7.37k | table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0])); |
1638 | 7.37k | if (table == NULL) |
1639 | 0 | goto error; |
1640 | 7.37k | ctxt->nsTab = table; |
1641 | | |
1642 | 7.37k | extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0])); |
1643 | 7.37k | if (extra == NULL) |
1644 | 0 | goto error; |
1645 | 7.37k | ctxt->nsdb->extra = extra; |
1646 | | |
1647 | 7.37k | ctxt->nsMax = newSize; |
1648 | 7.37k | return(0); |
1649 | | |
1650 | 0 | error: |
1651 | 0 | xmlErrMemory(ctxt); |
1652 | 0 | return(-1); |
1653 | 7.37k | } |
1654 | | |
1655 | | /** |
1656 | | * Push a new namespace on the table. |
1657 | | * |
1658 | | * @param ctxt parser context |
1659 | | * @param prefix prefix with hash value |
1660 | | * @param uri uri with hash value |
1661 | | * @param saxData extra data for SAX handler |
1662 | | * @param defAttr whether the namespace comes from a default attribute |
1663 | | * @returns 1 if the namespace was pushed, 0 if the namespace was ignored, |
1664 | | * -1 if a memory allocation failed. |
1665 | | */ |
1666 | | static int |
1667 | | xmlParserNsPush(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix, |
1668 | 87.2k | const xmlHashedString *uri, void *saxData, int defAttr) { |
1669 | 87.2k | xmlParserNsBucket *bucket = NULL; |
1670 | 87.2k | xmlParserNsExtra *extra; |
1671 | 87.2k | const xmlChar **ns; |
1672 | 87.2k | unsigned hashValue, nsIndex, oldIndex; |
1673 | | |
1674 | 87.2k | if ((prefix != NULL) && (prefix->name == ctxt->str_xml)) |
1675 | 194 | return(0); |
1676 | | |
1677 | 87.0k | if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) { |
1678 | 0 | xmlErrMemory(ctxt); |
1679 | 0 | return(-1); |
1680 | 0 | } |
1681 | | |
1682 | | /* |
1683 | | * Default namespace and 'xml' namespace |
1684 | | */ |
1685 | 87.0k | if ((prefix == NULL) || (prefix->name == NULL)) { |
1686 | 11.2k | oldIndex = ctxt->nsdb->defaultNsIndex; |
1687 | | |
1688 | 11.2k | if (oldIndex != INT_MAX) { |
1689 | 8.54k | extra = &ctxt->nsdb->extra[oldIndex]; |
1690 | | |
1691 | 8.54k | if (extra->elementId == ctxt->nsdb->elementId) { |
1692 | 558 | if (defAttr == 0) |
1693 | 491 | xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns"); |
1694 | 558 | return(0); |
1695 | 558 | } |
1696 | | |
1697 | 7.98k | if ((ctxt->options & XML_PARSE_NSCLEAN) && |
1698 | 7.98k | (uri->name == ctxt->nsTab[oldIndex * 2 + 1])) |
1699 | 440 | return(0); |
1700 | 7.98k | } |
1701 | | |
1702 | 10.2k | ctxt->nsdb->defaultNsIndex = ctxt->nsNr; |
1703 | 10.2k | goto populate_entry; |
1704 | 11.2k | } |
1705 | | |
1706 | | /* |
1707 | | * Hash table lookup |
1708 | | */ |
1709 | 75.8k | oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket); |
1710 | 75.8k | if (oldIndex != INT_MAX) { |
1711 | 55.0k | extra = &ctxt->nsdb->extra[oldIndex]; |
1712 | | |
1713 | | /* |
1714 | | * Check for duplicate definitions on the same element. |
1715 | | */ |
1716 | 55.0k | if (extra->elementId == ctxt->nsdb->elementId) { |
1717 | 202 | if (defAttr == 0) |
1718 | 132 | xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name); |
1719 | 202 | return(0); |
1720 | 202 | } |
1721 | | |
1722 | 54.8k | if ((ctxt->options & XML_PARSE_NSCLEAN) && |
1723 | 54.8k | (uri->name == ctxt->nsTab[bucket->index * 2 + 1])) |
1724 | 398 | return(0); |
1725 | | |
1726 | 54.4k | bucket->index = ctxt->nsNr; |
1727 | 54.4k | goto populate_entry; |
1728 | 54.8k | } |
1729 | | |
1730 | | /* |
1731 | | * Insert new bucket |
1732 | | */ |
1733 | | |
1734 | 20.7k | hashValue = prefix->hashValue; |
1735 | | |
1736 | | /* |
1737 | | * Grow hash table, 50% fill factor |
1738 | | */ |
1739 | 20.7k | if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) { |
1740 | 1.75k | xmlParserNsBucket *newHash; |
1741 | 1.75k | unsigned newSize, i, index; |
1742 | | |
1743 | 1.75k | if (ctxt->nsdb->hashSize > UINT_MAX / 2) { |
1744 | 0 | xmlErrMemory(ctxt); |
1745 | 0 | return(-1); |
1746 | 0 | } |
1747 | 1.75k | newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16; |
1748 | 1.75k | newHash = xmlMalloc(newSize * sizeof(newHash[0])); |
1749 | 1.75k | if (newHash == NULL) { |
1750 | 0 | xmlErrMemory(ctxt); |
1751 | 0 | return(-1); |
1752 | 0 | } |
1753 | 1.75k | memset(newHash, 0, newSize * sizeof(newHash[0])); |
1754 | | |
1755 | 65.8k | for (i = 0; i < ctxt->nsdb->hashSize; i++) { |
1756 | 64.1k | unsigned hv = ctxt->nsdb->hash[i].hashValue; |
1757 | 64.1k | unsigned newIndex; |
1758 | | |
1759 | 64.1k | if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX)) |
1760 | 62.1k | continue; |
1761 | 1.96k | newIndex = hv & (newSize - 1); |
1762 | | |
1763 | 3.77k | while (newHash[newIndex].hashValue != 0) { |
1764 | 1.80k | newIndex++; |
1765 | 1.80k | if (newIndex == newSize) |
1766 | 433 | newIndex = 0; |
1767 | 1.80k | } |
1768 | | |
1769 | 1.96k | newHash[newIndex] = ctxt->nsdb->hash[i]; |
1770 | 1.96k | } |
1771 | | |
1772 | 1.75k | xmlFree(ctxt->nsdb->hash); |
1773 | 1.75k | ctxt->nsdb->hash = newHash; |
1774 | 1.75k | ctxt->nsdb->hashSize = newSize; |
1775 | | |
1776 | | /* |
1777 | | * Relookup |
1778 | | */ |
1779 | 1.75k | index = hashValue & (newSize - 1); |
1780 | | |
1781 | 2.21k | while (newHash[index].hashValue != 0) { |
1782 | 459 | index++; |
1783 | 459 | if (index == newSize) |
1784 | 51 | index = 0; |
1785 | 459 | } |
1786 | | |
1787 | 1.75k | bucket = &newHash[index]; |
1788 | 1.75k | } |
1789 | | |
1790 | 20.7k | bucket->hashValue = hashValue; |
1791 | 20.7k | bucket->index = ctxt->nsNr; |
1792 | 20.7k | ctxt->nsdb->hashElems++; |
1793 | 20.7k | oldIndex = INT_MAX; |
1794 | | |
1795 | 85.4k | populate_entry: |
1796 | 85.4k | nsIndex = ctxt->nsNr; |
1797 | | |
1798 | 85.4k | ns = &ctxt->nsTab[nsIndex * 2]; |
1799 | 85.4k | ns[0] = prefix ? prefix->name : NULL; |
1800 | 85.4k | ns[1] = uri->name; |
1801 | | |
1802 | 85.4k | extra = &ctxt->nsdb->extra[nsIndex]; |
1803 | 85.4k | extra->saxData = saxData; |
1804 | 85.4k | extra->prefixHashValue = prefix ? prefix->hashValue : 0; |
1805 | 85.4k | extra->uriHashValue = uri->hashValue; |
1806 | 85.4k | extra->elementId = ctxt->nsdb->elementId; |
1807 | 85.4k | extra->oldIndex = oldIndex; |
1808 | | |
1809 | 85.4k | ctxt->nsNr++; |
1810 | | |
1811 | 85.4k | return(1); |
1812 | 20.7k | } |
1813 | | |
1814 | | /** |
1815 | | * Pops the top `nr` namespaces and restores the hash table. |
1816 | | * |
1817 | | * @param ctxt an XML parser context |
1818 | | * @param nr the number to pop |
1819 | | * @returns the number of namespaces popped. |
1820 | | */ |
1821 | | static int |
1822 | | xmlParserNsPop(xmlParserCtxtPtr ctxt, int nr) |
1823 | 22.9k | { |
1824 | 22.9k | int i; |
1825 | | |
1826 | | /* assert(nr <= ctxt->nsNr); */ |
1827 | | |
1828 | 107k | for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) { |
1829 | 84.4k | const xmlChar *prefix = ctxt->nsTab[i * 2]; |
1830 | 84.4k | xmlParserNsExtra *extra = &ctxt->nsdb->extra[i]; |
1831 | | |
1832 | 84.4k | if (prefix == NULL) { |
1833 | 10.0k | ctxt->nsdb->defaultNsIndex = extra->oldIndex; |
1834 | 74.4k | } else { |
1835 | 74.4k | xmlHashedString hprefix; |
1836 | 74.4k | xmlParserNsBucket *bucket = NULL; |
1837 | | |
1838 | 74.4k | hprefix.name = prefix; |
1839 | 74.4k | hprefix.hashValue = extra->prefixHashValue; |
1840 | 74.4k | xmlParserNsLookup(ctxt, &hprefix, &bucket); |
1841 | | /* assert(bucket && bucket->hashValue); */ |
1842 | 74.4k | bucket->index = extra->oldIndex; |
1843 | 74.4k | } |
1844 | 84.4k | } |
1845 | | |
1846 | 22.9k | ctxt->nsNr -= nr; |
1847 | 22.9k | return(nr); |
1848 | 22.9k | } |
1849 | | |
1850 | | static int |
1851 | 4.90k | xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) { |
1852 | 4.90k | const xmlChar **atts; |
1853 | 4.90k | unsigned *attallocs; |
1854 | 4.90k | int newSize; |
1855 | | |
1856 | 4.90k | newSize = xmlGrowCapacity(ctxt->maxatts / 5, |
1857 | 4.90k | sizeof(atts[0]) * 5 + sizeof(attallocs[0]), |
1858 | 4.90k | 10, XML_MAX_ATTRS); |
1859 | 4.90k | if (newSize < 0) { |
1860 | 0 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, |
1861 | 0 | "Maximum number of attributes exceeded"); |
1862 | 0 | return(-1); |
1863 | 0 | } |
1864 | | |
1865 | 4.90k | atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5); |
1866 | 4.90k | if (atts == NULL) |
1867 | 0 | goto mem_error; |
1868 | 4.90k | ctxt->atts = atts; |
1869 | | |
1870 | 4.90k | attallocs = xmlRealloc(ctxt->attallocs, |
1871 | 4.90k | newSize * sizeof(attallocs[0])); |
1872 | 4.90k | if (attallocs == NULL) |
1873 | 0 | goto mem_error; |
1874 | 4.90k | ctxt->attallocs = attallocs; |
1875 | | |
1876 | 4.90k | ctxt->maxatts = newSize * 5; |
1877 | | |
1878 | 4.90k | return(0); |
1879 | | |
1880 | 0 | mem_error: |
1881 | 0 | xmlErrMemory(ctxt); |
1882 | 0 | return(-1); |
1883 | 4.90k | } |
1884 | | |
1885 | | /** |
1886 | | * Pushes a new parser input on top of the input stack |
1887 | | * |
1888 | | * @param ctxt an XML parser context |
1889 | | * @param value the parser input |
1890 | | * @returns -1 in case of error, the index in the stack otherwise |
1891 | | */ |
1892 | | int |
1893 | | xmlCtxtPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr value) |
1894 | 76.4k | { |
1895 | 76.4k | char *directory = NULL; |
1896 | 76.4k | int maxDepth; |
1897 | | |
1898 | 76.4k | if ((ctxt == NULL) || (value == NULL)) |
1899 | 0 | return(-1); |
1900 | | |
1901 | 76.4k | maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
1902 | | |
1903 | 76.4k | if (ctxt->inputNr >= ctxt->inputMax) { |
1904 | 2.90k | xmlParserInputPtr *tmp; |
1905 | 2.90k | int newSize; |
1906 | | |
1907 | 2.90k | newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]), |
1908 | 2.90k | 5, maxDepth); |
1909 | 2.90k | if (newSize < 0) { |
1910 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
1911 | 0 | "Maximum entity nesting depth exceeded"); |
1912 | 0 | xmlHaltParser(ctxt); |
1913 | 0 | return(-1); |
1914 | 0 | } |
1915 | 2.90k | tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0])); |
1916 | 2.90k | if (tmp == NULL) { |
1917 | 0 | xmlErrMemory(ctxt); |
1918 | 0 | return(-1); |
1919 | 0 | } |
1920 | 2.90k | ctxt->inputTab = tmp; |
1921 | 2.90k | ctxt->inputMax = newSize; |
1922 | 2.90k | } |
1923 | | |
1924 | 76.4k | if ((ctxt->inputNr == 0) && (value->filename != NULL)) { |
1925 | 21.9k | directory = xmlParserGetDirectory(value->filename); |
1926 | 21.9k | if (directory == NULL) { |
1927 | 0 | xmlErrMemory(ctxt); |
1928 | 0 | return(-1); |
1929 | 0 | } |
1930 | 21.9k | } |
1931 | | |
1932 | 76.4k | if (ctxt->input_id >= INT_MAX) { |
1933 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, "Input ID overflow\n"); |
1934 | 0 | return(-1); |
1935 | 0 | } |
1936 | | |
1937 | 76.4k | ctxt->inputTab[ctxt->inputNr] = value; |
1938 | 76.4k | ctxt->input = value; |
1939 | | |
1940 | 76.4k | if (ctxt->inputNr == 0) { |
1941 | 21.9k | xmlFree(ctxt->directory); |
1942 | 21.9k | ctxt->directory = directory; |
1943 | 21.9k | } |
1944 | | |
1945 | | /* |
1946 | | * Internally, the input ID is only used to detect parameter entity |
1947 | | * boundaries. But there are entity loaders in downstream code that |
1948 | | * detect the main document by checking for "input_id == 1". |
1949 | | */ |
1950 | 76.4k | value->id = ctxt->input_id++; |
1951 | | |
1952 | 76.4k | return(ctxt->inputNr++); |
1953 | 76.4k | } |
1954 | | |
1955 | | /** |
1956 | | * Pops the top parser input from the input stack |
1957 | | * |
1958 | | * @param ctxt an XML parser context |
1959 | | * @returns the input just removed |
1960 | | */ |
1961 | | xmlParserInputPtr |
1962 | | xmlCtxtPopInput(xmlParserCtxtPtr ctxt) |
1963 | 120k | { |
1964 | 120k | xmlParserInputPtr ret; |
1965 | | |
1966 | 120k | if (ctxt == NULL) |
1967 | 0 | return(NULL); |
1968 | 120k | if (ctxt->inputNr <= 0) |
1969 | 43.9k | return (NULL); |
1970 | 76.4k | ctxt->inputNr--; |
1971 | 76.4k | if (ctxt->inputNr > 0) |
1972 | 54.4k | ctxt->input = ctxt->inputTab[ctxt->inputNr - 1]; |
1973 | 21.9k | else |
1974 | 21.9k | ctxt->input = NULL; |
1975 | 76.4k | ret = ctxt->inputTab[ctxt->inputNr]; |
1976 | 76.4k | ctxt->inputTab[ctxt->inputNr] = NULL; |
1977 | 76.4k | return (ret); |
1978 | 120k | } |
1979 | | |
1980 | | /** |
1981 | | * Pushes a new element node on top of the node stack |
1982 | | * |
1983 | | * @deprecated Internal function, do not use. |
1984 | | * |
1985 | | * @param ctxt an XML parser context |
1986 | | * @param value the element node |
1987 | | * @returns -1 in case of error, the index in the stack otherwise |
1988 | | */ |
1989 | | int |
1990 | | nodePush(xmlParserCtxtPtr ctxt, xmlNodePtr value) |
1991 | 85.8k | { |
1992 | 85.8k | if (ctxt == NULL) |
1993 | 0 | return(0); |
1994 | | |
1995 | 85.8k | if (ctxt->nodeNr >= ctxt->nodeMax) { |
1996 | 7.92k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
1997 | 7.92k | xmlNodePtr *tmp; |
1998 | 7.92k | int newSize; |
1999 | | |
2000 | 7.92k | newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]), |
2001 | 7.92k | 10, maxDepth); |
2002 | 7.92k | if (newSize < 0) { |
2003 | 14 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
2004 | 14 | "Excessive depth in document: %d," |
2005 | 14 | " use XML_PARSE_HUGE option\n", |
2006 | 14 | ctxt->nodeNr); |
2007 | 14 | xmlHaltParser(ctxt); |
2008 | 14 | return(-1); |
2009 | 14 | } |
2010 | | |
2011 | 7.91k | tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0])); |
2012 | 7.91k | if (tmp == NULL) { |
2013 | 0 | xmlErrMemory(ctxt); |
2014 | 0 | return (-1); |
2015 | 0 | } |
2016 | 7.91k | ctxt->nodeTab = tmp; |
2017 | 7.91k | ctxt->nodeMax = newSize; |
2018 | 7.91k | } |
2019 | | |
2020 | 85.8k | ctxt->nodeTab[ctxt->nodeNr] = value; |
2021 | 85.8k | ctxt->node = value; |
2022 | 85.8k | return (ctxt->nodeNr++); |
2023 | 85.8k | } |
2024 | | |
2025 | | /** |
2026 | | * Pops the top element node from the node stack |
2027 | | * |
2028 | | * @deprecated Internal function, do not use. |
2029 | | * |
2030 | | * @param ctxt an XML parser context |
2031 | | * @returns the node just removed |
2032 | | */ |
2033 | | xmlNodePtr |
2034 | | nodePop(xmlParserCtxtPtr ctxt) |
2035 | 105k | { |
2036 | 105k | xmlNodePtr ret; |
2037 | | |
2038 | 105k | if (ctxt == NULL) return(NULL); |
2039 | 105k | if (ctxt->nodeNr <= 0) |
2040 | 23.2k | return (NULL); |
2041 | 82.6k | ctxt->nodeNr--; |
2042 | 82.6k | if (ctxt->nodeNr > 0) |
2043 | 78.3k | ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1]; |
2044 | 4.28k | else |
2045 | 4.28k | ctxt->node = NULL; |
2046 | 82.6k | ret = ctxt->nodeTab[ctxt->nodeNr]; |
2047 | 82.6k | ctxt->nodeTab[ctxt->nodeNr] = NULL; |
2048 | 82.6k | return (ret); |
2049 | 105k | } |
2050 | | |
2051 | | /** |
2052 | | * Pushes a new element name/prefix/URL on top of the name stack |
2053 | | * |
2054 | | * @param ctxt an XML parser context |
2055 | | * @param value the element name |
2056 | | * @param prefix the element prefix |
2057 | | * @param URI the element namespace name |
2058 | | * @param line the current line number for error messages |
2059 | | * @param nsNr the number of namespaces pushed on the namespace table |
2060 | | * @returns -1 in case of error, the index in the stack otherwise |
2061 | | */ |
2062 | | static int |
2063 | | nameNsPush(xmlParserCtxtPtr ctxt, const xmlChar * value, |
2064 | | const xmlChar *prefix, const xmlChar *URI, int line, int nsNr) |
2065 | 132k | { |
2066 | 132k | xmlStartTag *tag; |
2067 | | |
2068 | 132k | if (ctxt->nameNr >= ctxt->nameMax) { |
2069 | 12.5k | const xmlChar **tmp; |
2070 | 12.5k | xmlStartTag *tmp2; |
2071 | 12.5k | int newSize; |
2072 | | |
2073 | 12.5k | newSize = xmlGrowCapacity(ctxt->nameMax, |
2074 | 12.5k | sizeof(tmp[0]) + sizeof(tmp2[0]), |
2075 | 12.5k | 10, XML_MAX_ITEMS); |
2076 | 12.5k | if (newSize < 0) |
2077 | 0 | goto mem_error; |
2078 | | |
2079 | 12.5k | tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0])); |
2080 | 12.5k | if (tmp == NULL) |
2081 | 0 | goto mem_error; |
2082 | 12.5k | ctxt->nameTab = tmp; |
2083 | | |
2084 | 12.5k | tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0])); |
2085 | 12.5k | if (tmp2 == NULL) |
2086 | 0 | goto mem_error; |
2087 | 12.5k | ctxt->pushTab = tmp2; |
2088 | | |
2089 | 12.5k | ctxt->nameMax = newSize; |
2090 | 120k | } else if (ctxt->pushTab == NULL) { |
2091 | 13.9k | ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0])); |
2092 | 13.9k | if (ctxt->pushTab == NULL) |
2093 | 0 | goto mem_error; |
2094 | 13.9k | } |
2095 | 132k | ctxt->nameTab[ctxt->nameNr] = value; |
2096 | 132k | ctxt->name = value; |
2097 | 132k | tag = &ctxt->pushTab[ctxt->nameNr]; |
2098 | 132k | tag->prefix = prefix; |
2099 | 132k | tag->URI = URI; |
2100 | 132k | tag->line = line; |
2101 | 132k | tag->nsNr = nsNr; |
2102 | 132k | return (ctxt->nameNr++); |
2103 | 0 | mem_error: |
2104 | 0 | xmlErrMemory(ctxt); |
2105 | 0 | return (-1); |
2106 | 132k | } |
2107 | | #ifdef LIBXML_PUSH_ENABLED |
2108 | | /** |
2109 | | * Pops the top element/prefix/URI name from the name stack |
2110 | | * |
2111 | | * @param ctxt an XML parser context |
2112 | | * @returns the name just removed |
2113 | | */ |
2114 | | static const xmlChar * |
2115 | | nameNsPop(xmlParserCtxtPtr ctxt) |
2116 | 0 | { |
2117 | 0 | const xmlChar *ret; |
2118 | |
|
2119 | 0 | if (ctxt->nameNr <= 0) |
2120 | 0 | return (NULL); |
2121 | 0 | ctxt->nameNr--; |
2122 | 0 | if (ctxt->nameNr > 0) |
2123 | 0 | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
2124 | 0 | else |
2125 | 0 | ctxt->name = NULL; |
2126 | 0 | ret = ctxt->nameTab[ctxt->nameNr]; |
2127 | 0 | ctxt->nameTab[ctxt->nameNr] = NULL; |
2128 | 0 | return (ret); |
2129 | 0 | } |
2130 | | #endif /* LIBXML_PUSH_ENABLED */ |
2131 | | |
2132 | | /** |
2133 | | * Pops the top element name from the name stack |
2134 | | * |
2135 | | * @deprecated Internal function, do not use. |
2136 | | * |
2137 | | * @param ctxt an XML parser context |
2138 | | * @returns the name just removed |
2139 | | */ |
2140 | | static const xmlChar * |
2141 | | namePop(xmlParserCtxtPtr ctxt) |
2142 | 127k | { |
2143 | 127k | const xmlChar *ret; |
2144 | | |
2145 | 127k | if ((ctxt == NULL) || (ctxt->nameNr <= 0)) |
2146 | 0 | return (NULL); |
2147 | 127k | ctxt->nameNr--; |
2148 | 127k | if (ctxt->nameNr > 0) |
2149 | 118k | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
2150 | 8.97k | else |
2151 | 8.97k | ctxt->name = NULL; |
2152 | 127k | ret = ctxt->nameTab[ctxt->nameNr]; |
2153 | 127k | ctxt->nameTab[ctxt->nameNr] = NULL; |
2154 | 127k | return (ret); |
2155 | 127k | } |
2156 | | |
2157 | 156k | static int spacePush(xmlParserCtxtPtr ctxt, int val) { |
2158 | 156k | if (ctxt->spaceNr >= ctxt->spaceMax) { |
2159 | 29.6k | int *tmp; |
2160 | 29.6k | int newSize; |
2161 | | |
2162 | 29.6k | newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]), |
2163 | 29.6k | 10, XML_MAX_ITEMS); |
2164 | 29.6k | if (newSize < 0) { |
2165 | 0 | xmlErrMemory(ctxt); |
2166 | 0 | return(-1); |
2167 | 0 | } |
2168 | | |
2169 | 29.6k | tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0])); |
2170 | 29.6k | if (tmp == NULL) { |
2171 | 0 | xmlErrMemory(ctxt); |
2172 | 0 | return(-1); |
2173 | 0 | } |
2174 | 29.6k | ctxt->spaceTab = tmp; |
2175 | | |
2176 | 29.6k | ctxt->spaceMax = newSize; |
2177 | 29.6k | } |
2178 | 156k | ctxt->spaceTab[ctxt->spaceNr] = val; |
2179 | 156k | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr]; |
2180 | 156k | return(ctxt->spaceNr++); |
2181 | 156k | } |
2182 | | |
2183 | 151k | static int spacePop(xmlParserCtxtPtr ctxt) { |
2184 | 151k | int ret; |
2185 | 151k | if (ctxt->spaceNr <= 0) return(0); |
2186 | 151k | ctxt->spaceNr--; |
2187 | 151k | if (ctxt->spaceNr > 0) |
2188 | 151k | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1]; |
2189 | 0 | else |
2190 | 0 | ctxt->space = &ctxt->spaceTab[0]; |
2191 | 151k | ret = ctxt->spaceTab[ctxt->spaceNr]; |
2192 | 151k | ctxt->spaceTab[ctxt->spaceNr] = -1; |
2193 | 151k | return(ret); |
2194 | 151k | } |
2195 | | |
2196 | | /* |
2197 | | * Macros for accessing the content. Those should be used only by the parser, |
2198 | | * and not exported. |
2199 | | * |
2200 | | * Dirty macros, i.e. one often need to make assumption on the context to |
2201 | | * use them |
2202 | | * |
2203 | | * CUR_PTR return the current pointer to the xmlChar to be parsed. |
2204 | | * To be used with extreme caution since operations consuming |
2205 | | * characters may move the input buffer to a different location ! |
2206 | | * CUR returns the current xmlChar value, i.e. a 8 bit value if compiled |
2207 | | * This should be used internally by the parser |
2208 | | * only to compare to ASCII values otherwise it would break when |
2209 | | * running with UTF-8 encoding. |
2210 | | * RAW same as CUR but in the input buffer, bypass any token |
2211 | | * extraction that may have been done |
2212 | | * NXT(n) returns the n'th next xmlChar. Same as CUR is should be used only |
2213 | | * to compare on ASCII based substring. |
2214 | | * SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined |
2215 | | * strings without newlines within the parser. |
2216 | | * NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII |
2217 | | * defined char within the parser. |
2218 | | * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding |
2219 | | * |
2220 | | * NEXT Skip to the next character, this does the proper decoding |
2221 | | * in UTF-8 mode. It also pop-up unfinished entities on the fly. |
2222 | | * NEXTL(l) Skip the current unicode character of l xmlChars long. |
2223 | | * CUR_SCHAR same but operate on a string instead of the context |
2224 | | * COPY_BUF copy the current unicode char to the target buffer, increment |
2225 | | * the index |
2226 | | * GROW, SHRINK handling of input buffers |
2227 | | */ |
2228 | | |
2229 | 2.37M | #define RAW (*ctxt->input->cur) |
2230 | 2.86M | #define CUR (*ctxt->input->cur) |
2231 | 1.02M | #define NXT(val) ctxt->input->cur[(val)] |
2232 | 3.12M | #define CUR_PTR ctxt->input->cur |
2233 | 325k | #define BASE_PTR ctxt->input->base |
2234 | | |
2235 | | #define CMP4( s, c1, c2, c3, c4 ) \ |
2236 | 1.47M | ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \ |
2237 | 761k | ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 ) |
2238 | | #define CMP5( s, c1, c2, c3, c4, c5 ) \ |
2239 | 1.30M | ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 ) |
2240 | | #define CMP6( s, c1, c2, c3, c4, c5, c6 ) \ |
2241 | 1.04M | ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 ) |
2242 | | #define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \ |
2243 | 828k | ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 ) |
2244 | | #define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \ |
2245 | 665k | ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 ) |
2246 | | #define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \ |
2247 | 298k | ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \ |
2248 | 298k | ((unsigned char *) s)[ 8 ] == c9 ) |
2249 | | #define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \ |
2250 | 9.97k | ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \ |
2251 | 9.97k | ((unsigned char *) s)[ 9 ] == c10 ) |
2252 | | |
2253 | 308k | #define SKIP(val) do { \ |
2254 | 308k | ctxt->input->cur += (val),ctxt->input->col+=(val); \ |
2255 | 308k | if (*ctxt->input->cur == 0) \ |
2256 | 308k | xmlParserGrow(ctxt); \ |
2257 | 308k | } while (0) |
2258 | | |
2259 | | #define SKIPL(val) do { \ |
2260 | | int skipl; \ |
2261 | | for(skipl=0; skipl<val; skipl++) { \ |
2262 | | if (*(ctxt->input->cur) == '\n') { \ |
2263 | | ctxt->input->line++; ctxt->input->col = 1; \ |
2264 | | } else ctxt->input->col++; \ |
2265 | | ctxt->input->cur++; \ |
2266 | | } \ |
2267 | | if (*ctxt->input->cur == 0) \ |
2268 | | xmlParserGrow(ctxt); \ |
2269 | | } while (0) |
2270 | | |
2271 | | #define SHRINK \ |
2272 | 410k | if (!PARSER_PROGRESSIVE(ctxt)) \ |
2273 | 410k | xmlParserShrink(ctxt); |
2274 | | |
2275 | | #define GROW \ |
2276 | 2.39M | if ((!PARSER_PROGRESSIVE(ctxt)) && \ |
2277 | 2.39M | (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \ |
2278 | 1.99M | xmlParserGrow(ctxt); |
2279 | | |
2280 | 458k | #define SKIP_BLANKS xmlSkipBlankChars(ctxt) |
2281 | | |
2282 | 522k | #define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt) |
2283 | | |
2284 | 720k | #define NEXT xmlNextChar(ctxt) |
2285 | | |
2286 | 211k | #define NEXT1 { \ |
2287 | 211k | ctxt->input->col++; \ |
2288 | 211k | ctxt->input->cur++; \ |
2289 | 211k | if (*ctxt->input->cur == 0) \ |
2290 | 211k | xmlParserGrow(ctxt); \ |
2291 | 211k | } |
2292 | | |
2293 | 1.77M | #define NEXTL(l) do { \ |
2294 | 1.77M | if (*(ctxt->input->cur) == '\n') { \ |
2295 | 5.42k | ctxt->input->line++; ctxt->input->col = 1; \ |
2296 | 1.77M | } else ctxt->input->col++; \ |
2297 | 1.77M | ctxt->input->cur += l; \ |
2298 | 1.77M | } while (0) |
2299 | | |
2300 | 1.37M | #define CUR_SCHAR(s, l) xmlStringCurrentChar(ctxt, s, &l) |
2301 | | |
2302 | | #define COPY_BUF(b, i, v) \ |
2303 | 1.14M | if (v < 0x80) b[i++] = v; \ |
2304 | 1.14M | else i += xmlCopyCharMultiByte(&b[i],v) |
2305 | | |
2306 | | static int |
2307 | 315k | xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) { |
2308 | 315k | int c = xmlCurrentChar(ctxt, len); |
2309 | | |
2310 | 315k | if (c == XML_INVALID_CHAR) |
2311 | 17.0k | c = 0xFFFD; /* replacement character */ |
2312 | | |
2313 | 315k | return(c); |
2314 | 315k | } |
2315 | | |
2316 | | /** |
2317 | | * Skip whitespace in the input stream. |
2318 | | * |
2319 | | * @deprecated Internal function, do not use. |
2320 | | * |
2321 | | * @param ctxt the XML parser context |
2322 | | * @returns the number of space chars skipped |
2323 | | */ |
2324 | | int |
2325 | 591k | xmlSkipBlankChars(xmlParserCtxtPtr ctxt) { |
2326 | 591k | const xmlChar *cur; |
2327 | 591k | int res = 0; |
2328 | | |
2329 | 591k | cur = ctxt->input->cur; |
2330 | 591k | while (IS_BLANK_CH(*cur)) { |
2331 | 109k | if (*cur == '\n') { |
2332 | 633 | ctxt->input->line++; ctxt->input->col = 1; |
2333 | 109k | } else { |
2334 | 109k | ctxt->input->col++; |
2335 | 109k | } |
2336 | 109k | cur++; |
2337 | 109k | if (res < INT_MAX) |
2338 | 109k | res++; |
2339 | 109k | if (*cur == 0) { |
2340 | 496 | ctxt->input->cur = cur; |
2341 | 496 | xmlParserGrow(ctxt); |
2342 | 496 | cur = ctxt->input->cur; |
2343 | 496 | } |
2344 | 109k | } |
2345 | 591k | ctxt->input->cur = cur; |
2346 | | |
2347 | 591k | if (res > 4) |
2348 | 447 | GROW; |
2349 | | |
2350 | 591k | return(res); |
2351 | 591k | } |
2352 | | |
2353 | | static void |
2354 | 53.4k | xmlPopPE(xmlParserCtxtPtr ctxt) { |
2355 | 53.4k | unsigned long consumed; |
2356 | 53.4k | xmlEntityPtr ent; |
2357 | | |
2358 | 53.4k | ent = ctxt->input->entity; |
2359 | | |
2360 | 53.4k | ent->flags &= ~XML_ENT_EXPANDING; |
2361 | | |
2362 | 53.4k | if ((ent->flags & XML_ENT_CHECKED) == 0) { |
2363 | 2.35k | int result; |
2364 | | |
2365 | | /* |
2366 | | * Read the rest of the stream in case of errors. We want |
2367 | | * to account for the whole entity size. |
2368 | | */ |
2369 | 2.35k | do { |
2370 | 2.35k | ctxt->input->cur = ctxt->input->end; |
2371 | 2.35k | xmlParserShrink(ctxt); |
2372 | 2.35k | result = xmlParserGrow(ctxt); |
2373 | 2.35k | } while (result > 0); |
2374 | | |
2375 | 2.35k | consumed = ctxt->input->consumed; |
2376 | 2.35k | xmlSaturatedAddSizeT(&consumed, |
2377 | 2.35k | ctxt->input->end - ctxt->input->base); |
2378 | | |
2379 | 2.35k | xmlSaturatedAdd(&ent->expandedSize, consumed); |
2380 | | |
2381 | | /* |
2382 | | * Add to sizeentities when parsing an external entity |
2383 | | * for the first time. |
2384 | | */ |
2385 | 2.35k | if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
2386 | 0 | xmlSaturatedAdd(&ctxt->sizeentities, consumed); |
2387 | 0 | } |
2388 | | |
2389 | 2.35k | ent->flags |= XML_ENT_CHECKED; |
2390 | 2.35k | } |
2391 | | |
2392 | 53.4k | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
2393 | | |
2394 | 53.4k | xmlParserEntityCheck(ctxt, ent->expandedSize); |
2395 | | |
2396 | 53.4k | GROW; |
2397 | 53.4k | } |
2398 | | |
2399 | | /** |
2400 | | * Skip whitespace in the input stream, also handling parameter |
2401 | | * entities. |
2402 | | * |
2403 | | * @param ctxt the XML parser context |
2404 | | * @returns the number of space chars skipped |
2405 | | */ |
2406 | | static int |
2407 | 522k | xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) { |
2408 | 522k | int res = 0; |
2409 | 522k | int inParam; |
2410 | 522k | int expandParam; |
2411 | | |
2412 | 522k | inParam = PARSER_IN_PE(ctxt); |
2413 | 522k | expandParam = PARSER_EXTERNAL(ctxt); |
2414 | | |
2415 | 522k | if (!inParam && !expandParam) |
2416 | 132k | return(xmlSkipBlankChars(ctxt)); |
2417 | | |
2418 | | /* |
2419 | | * It's Okay to use CUR/NEXT here since all the blanks are on |
2420 | | * the ASCII range. |
2421 | | */ |
2422 | 568k | while (PARSER_STOPPED(ctxt) == 0) { |
2423 | 568k | if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */ |
2424 | 125k | NEXT; |
2425 | 442k | } else if (CUR == '%') { |
2426 | 51.2k | if ((expandParam == 0) || |
2427 | 51.2k | (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0)) |
2428 | 51.2k | break; |
2429 | | |
2430 | | /* |
2431 | | * Expand parameter entity. We continue to consume |
2432 | | * whitespace at the start of the entity and possible |
2433 | | * even consume the whole entity and pop it. We might |
2434 | | * even pop multiple PEs in this loop. |
2435 | | */ |
2436 | 0 | xmlParsePEReference(ctxt); |
2437 | |
|
2438 | 0 | inParam = PARSER_IN_PE(ctxt); |
2439 | 0 | expandParam = PARSER_EXTERNAL(ctxt); |
2440 | 391k | } else if (CUR == 0) { |
2441 | 53.3k | if (inParam == 0) |
2442 | 0 | break; |
2443 | | |
2444 | 53.3k | xmlPopPE(ctxt); |
2445 | | |
2446 | 53.3k | inParam = PARSER_IN_PE(ctxt); |
2447 | 53.3k | expandParam = PARSER_EXTERNAL(ctxt); |
2448 | 338k | } else { |
2449 | 338k | break; |
2450 | 338k | } |
2451 | | |
2452 | | /* |
2453 | | * Also increase the counter when entering or exiting a PERef. |
2454 | | * The spec says: "When a parameter-entity reference is recognized |
2455 | | * in the DTD and included, its replacement text MUST be enlarged |
2456 | | * by the attachment of one leading and one following space (#x20) |
2457 | | * character." |
2458 | | */ |
2459 | 179k | if (res < INT_MAX) |
2460 | 179k | res++; |
2461 | 179k | } |
2462 | | |
2463 | 389k | return(res); |
2464 | 522k | } |
2465 | | |
2466 | | /************************************************************************ |
2467 | | * * |
2468 | | * Commodity functions to handle entities * |
2469 | | * * |
2470 | | ************************************************************************/ |
2471 | | |
2472 | | /** |
2473 | | * @deprecated Internal function, don't use. |
2474 | | * |
2475 | | * @param ctxt an XML parser context |
2476 | | * @returns the current xmlChar in the parser context |
2477 | | */ |
2478 | | xmlChar |
2479 | 0 | xmlPopInput(xmlParserCtxtPtr ctxt) { |
2480 | 0 | xmlParserInputPtr input; |
2481 | |
|
2482 | 0 | if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0); |
2483 | 0 | input = xmlCtxtPopInput(ctxt); |
2484 | 0 | xmlFreeInputStream(input); |
2485 | 0 | if (*ctxt->input->cur == 0) |
2486 | 0 | xmlParserGrow(ctxt); |
2487 | 0 | return(CUR); |
2488 | 0 | } |
2489 | | |
2490 | | /** |
2491 | | * Push an input stream onto the stack. |
2492 | | * |
2493 | | * @deprecated Internal function, don't use. |
2494 | | * |
2495 | | * @param ctxt an XML parser context |
2496 | | * @param input an XML parser input fragment (entity, XML fragment ...). |
2497 | | * @returns -1 in case of error or the index in the input stack |
2498 | | */ |
2499 | | int |
2500 | 0 | xmlPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) { |
2501 | 0 | int ret; |
2502 | |
|
2503 | 0 | if ((ctxt == NULL) || (input == NULL)) |
2504 | 0 | return(-1); |
2505 | | |
2506 | 0 | ret = xmlCtxtPushInput(ctxt, input); |
2507 | 0 | if (ret >= 0) |
2508 | 0 | GROW; |
2509 | 0 | return(ret); |
2510 | 0 | } |
2511 | | |
2512 | | /** |
2513 | | * Parse a numeric character reference. Always consumes '&'. |
2514 | | * |
2515 | | * @deprecated Internal function, don't use. |
2516 | | * |
2517 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2518 | | * '&#x' [0-9a-fA-F]+ ';' |
2519 | | * |
2520 | | * [ WFC: Legal Character ] |
2521 | | * Characters referred to using character references must match the |
2522 | | * production for Char. |
2523 | | * |
2524 | | * @param ctxt an XML parser context |
2525 | | * @returns the value parsed (as an int), 0 in case of error |
2526 | | */ |
2527 | | int |
2528 | 12.1k | xmlParseCharRef(xmlParserCtxtPtr ctxt) { |
2529 | 12.1k | int val = 0; |
2530 | 12.1k | int count = 0; |
2531 | | |
2532 | | /* |
2533 | | * Using RAW/CUR/NEXT is okay since we are working on ASCII range here |
2534 | | */ |
2535 | 12.1k | if ((RAW == '&') && (NXT(1) == '#') && |
2536 | 12.1k | (NXT(2) == 'x')) { |
2537 | 4.28k | SKIP(3); |
2538 | 4.28k | GROW; |
2539 | 20.0k | while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) { |
2540 | 17.2k | if (count++ > 20) { |
2541 | 588 | count = 0; |
2542 | 588 | GROW; |
2543 | 588 | } |
2544 | 17.2k | if ((RAW >= '0') && (RAW <= '9')) |
2545 | 8.47k | val = val * 16 + (CUR - '0'); |
2546 | 8.73k | else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20)) |
2547 | 3.07k | val = val * 16 + (CUR - 'a') + 10; |
2548 | 5.66k | else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20)) |
2549 | 4.17k | val = val * 16 + (CUR - 'A') + 10; |
2550 | 1.48k | else { |
2551 | 1.48k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2552 | 1.48k | val = 0; |
2553 | 1.48k | break; |
2554 | 1.48k | } |
2555 | 15.7k | if (val > 0x110000) |
2556 | 7.09k | val = 0x110000; |
2557 | | |
2558 | 15.7k | NEXT; |
2559 | 15.7k | count++; |
2560 | 15.7k | } |
2561 | 4.28k | if (RAW == ';') { |
2562 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2563 | 2.79k | ctxt->input->col++; |
2564 | 2.79k | ctxt->input->cur++; |
2565 | 2.79k | } |
2566 | 7.81k | } else if ((RAW == '&') && (NXT(1) == '#')) { |
2567 | 7.81k | SKIP(2); |
2568 | 7.81k | GROW; |
2569 | 28.1k | while (RAW != ';') { /* loop blocked by count */ |
2570 | 21.8k | if (count++ > 20) { |
2571 | 566 | count = 0; |
2572 | 566 | GROW; |
2573 | 566 | } |
2574 | 21.8k | if ((RAW >= '0') && (RAW <= '9')) |
2575 | 20.2k | val = val * 10 + (CUR - '0'); |
2576 | 1.56k | else { |
2577 | 1.56k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2578 | 1.56k | val = 0; |
2579 | 1.56k | break; |
2580 | 1.56k | } |
2581 | 20.2k | if (val > 0x110000) |
2582 | 5.65k | val = 0x110000; |
2583 | | |
2584 | 20.2k | NEXT; |
2585 | 20.2k | count++; |
2586 | 20.2k | } |
2587 | 7.81k | if (RAW == ';') { |
2588 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2589 | 6.25k | ctxt->input->col++; |
2590 | 6.25k | ctxt->input->cur++; |
2591 | 6.25k | } |
2592 | 7.81k | } else { |
2593 | 0 | if (RAW == '&') |
2594 | 0 | SKIP(1); |
2595 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2596 | 0 | } |
2597 | | |
2598 | | /* |
2599 | | * [ WFC: Legal Character ] |
2600 | | * Characters referred to using character references must match the |
2601 | | * production for Char. |
2602 | | */ |
2603 | 12.1k | if (val >= 0x110000) { |
2604 | 244 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2605 | 244 | "xmlParseCharRef: character reference out of bounds\n", |
2606 | 244 | val); |
2607 | 11.8k | } else if (IS_CHAR(val)) { |
2608 | 8.14k | return(val); |
2609 | 8.14k | } else { |
2610 | 3.71k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2611 | 3.71k | "xmlParseCharRef: invalid xmlChar value %d\n", |
2612 | 3.71k | val); |
2613 | 3.71k | } |
2614 | 3.95k | return(0); |
2615 | 12.1k | } |
2616 | | |
2617 | | /** |
2618 | | * parse Reference declarations, variant parsing from a string rather |
2619 | | * than an an input flow. |
2620 | | * |
2621 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2622 | | * '&#x' [0-9a-fA-F]+ ';' |
2623 | | * |
2624 | | * [ WFC: Legal Character ] |
2625 | | * Characters referred to using character references must match the |
2626 | | * production for Char. |
2627 | | * |
2628 | | * @param ctxt an XML parser context |
2629 | | * @param str a pointer to an index in the string |
2630 | | * @returns the value parsed (as an int), 0 in case of error, str will be |
2631 | | * updated to the current value of the index |
2632 | | */ |
2633 | | static int |
2634 | 113k | xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
2635 | 113k | const xmlChar *ptr; |
2636 | 113k | xmlChar cur; |
2637 | 113k | int val = 0; |
2638 | | |
2639 | 113k | if ((str == NULL) || (*str == NULL)) return(0); |
2640 | 113k | ptr = *str; |
2641 | 113k | cur = *ptr; |
2642 | 113k | if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) { |
2643 | 17.0k | ptr += 3; |
2644 | 17.0k | cur = *ptr; |
2645 | 47.4k | while (cur != ';') { /* Non input consuming loop */ |
2646 | 30.8k | if ((cur >= '0') && (cur <= '9')) |
2647 | 5.37k | val = val * 16 + (cur - '0'); |
2648 | 25.4k | else if ((cur >= 'a') && (cur <= 'f')) |
2649 | 3.17k | val = val * 16 + (cur - 'a') + 10; |
2650 | 22.2k | else if ((cur >= 'A') && (cur <= 'F')) |
2651 | 21.8k | val = val * 16 + (cur - 'A') + 10; |
2652 | 440 | else { |
2653 | 440 | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2654 | 440 | val = 0; |
2655 | 440 | break; |
2656 | 440 | } |
2657 | 30.3k | if (val > 0x110000) |
2658 | 281 | val = 0x110000; |
2659 | | |
2660 | 30.3k | ptr++; |
2661 | 30.3k | cur = *ptr; |
2662 | 30.3k | } |
2663 | 17.0k | if (cur == ';') |
2664 | 16.6k | ptr++; |
2665 | 96.0k | } else if ((cur == '&') && (ptr[1] == '#')){ |
2666 | 96.0k | ptr += 2; |
2667 | 96.0k | cur = *ptr; |
2668 | 288k | while (cur != ';') { /* Non input consuming loops */ |
2669 | 193k | if ((cur >= '0') && (cur <= '9')) |
2670 | 192k | val = val * 10 + (cur - '0'); |
2671 | 1.25k | else { |
2672 | 1.25k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2673 | 1.25k | val = 0; |
2674 | 1.25k | break; |
2675 | 1.25k | } |
2676 | 192k | if (val > 0x110000) |
2677 | 517 | val = 0x110000; |
2678 | | |
2679 | 192k | ptr++; |
2680 | 192k | cur = *ptr; |
2681 | 192k | } |
2682 | 96.0k | if (cur == ';') |
2683 | 94.7k | ptr++; |
2684 | 96.0k | } else { |
2685 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2686 | 0 | return(0); |
2687 | 0 | } |
2688 | 113k | *str = ptr; |
2689 | | |
2690 | | /* |
2691 | | * [ WFC: Legal Character ] |
2692 | | * Characters referred to using character references must match the |
2693 | | * production for Char. |
2694 | | */ |
2695 | 113k | if (val >= 0x110000) { |
2696 | 207 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2697 | 207 | "xmlParseStringCharRef: character reference out of bounds\n", |
2698 | 207 | val); |
2699 | 112k | } else if (IS_CHAR(val)) { |
2700 | 110k | return(val); |
2701 | 110k | } else { |
2702 | 2.31k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2703 | 2.31k | "xmlParseStringCharRef: invalid xmlChar value %d\n", |
2704 | 2.31k | val); |
2705 | 2.31k | } |
2706 | 2.51k | return(0); |
2707 | 113k | } |
2708 | | |
2709 | | /** |
2710 | | * [69] PEReference ::= '%' Name ';' |
2711 | | * |
2712 | | * @deprecated Internal function, do not use. |
2713 | | * |
2714 | | * [ WFC: No Recursion ] |
2715 | | * A parsed entity must not contain a recursive |
2716 | | * reference to itself, either directly or indirectly. |
2717 | | * |
2718 | | * [ WFC: Entity Declared ] |
2719 | | * In a document without any DTD, a document with only an internal DTD |
2720 | | * subset which contains no parameter entity references, or a document |
2721 | | * with "standalone='yes'", ... ... The declaration of a parameter |
2722 | | * entity must precede any reference to it... |
2723 | | * |
2724 | | * [ VC: Entity Declared ] |
2725 | | * In a document with an external subset or external parameter entities |
2726 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
2727 | | * must precede any reference to it... |
2728 | | * |
2729 | | * [ WFC: In DTD ] |
2730 | | * Parameter-entity references may only appear in the DTD. |
2731 | | * NOTE: misleading but this is handled. |
2732 | | * |
2733 | | * A PEReference may have been detected in the current input stream |
2734 | | * the handling is done accordingly to |
2735 | | * http://www.w3.org/TR/REC-xml#entproc |
2736 | | * i.e. |
2737 | | * - Included in literal in entity values |
2738 | | * - Included as Parameter Entity reference within DTDs |
2739 | | * @param ctxt the parser context |
2740 | | */ |
2741 | | void |
2742 | 0 | xmlParserHandlePEReference(xmlParserCtxtPtr ctxt) { |
2743 | 0 | xmlParsePEReference(ctxt); |
2744 | 0 | } |
2745 | | |
2746 | | /** |
2747 | | * @deprecated Internal function, don't use. |
2748 | | * |
2749 | | * @param ctxt the parser context |
2750 | | * @param str the input string |
2751 | | * @param len the string length |
2752 | | * @param what combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2753 | | * @param end an end marker xmlChar, 0 if none |
2754 | | * @param end2 an end marker xmlChar, 0 if none |
2755 | | * @param end3 an end marker xmlChar, 0 if none |
2756 | | * @returns A newly allocated string with the substitution done. The caller |
2757 | | * must deallocate it ! |
2758 | | */ |
2759 | | xmlChar * |
2760 | | xmlStringLenDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2761 | | int what ATTRIBUTE_UNUSED, |
2762 | 0 | xmlChar end, xmlChar end2, xmlChar end3) { |
2763 | 0 | if ((ctxt == NULL) || (str == NULL) || (len < 0)) |
2764 | 0 | return(NULL); |
2765 | | |
2766 | 0 | if ((str[len] != 0) || |
2767 | 0 | (end != 0) || (end2 != 0) || (end3 != 0)) |
2768 | 0 | return(NULL); |
2769 | | |
2770 | 0 | return(xmlExpandEntitiesInAttValue(ctxt, str, 0)); |
2771 | 0 | } |
2772 | | |
2773 | | /** |
2774 | | * @deprecated Internal function, don't use. |
2775 | | * |
2776 | | * @param ctxt the parser context |
2777 | | * @param str the input string |
2778 | | * @param what combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2779 | | * @param end an end marker xmlChar, 0 if none |
2780 | | * @param end2 an end marker xmlChar, 0 if none |
2781 | | * @param end3 an end marker xmlChar, 0 if none |
2782 | | * @returns A newly allocated string with the substitution done. The caller |
2783 | | * must deallocate it ! |
2784 | | */ |
2785 | | xmlChar * |
2786 | | xmlStringDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, |
2787 | | int what ATTRIBUTE_UNUSED, |
2788 | 0 | xmlChar end, xmlChar end2, xmlChar end3) { |
2789 | 0 | if ((ctxt == NULL) || (str == NULL)) |
2790 | 0 | return(NULL); |
2791 | | |
2792 | 0 | if ((end != 0) || (end2 != 0) || (end3 != 0)) |
2793 | 0 | return(NULL); |
2794 | | |
2795 | 0 | return(xmlExpandEntitiesInAttValue(ctxt, str, 0)); |
2796 | 0 | } |
2797 | | |
2798 | | /************************************************************************ |
2799 | | * * |
2800 | | * Commodity functions, cleanup needed ? * |
2801 | | * * |
2802 | | ************************************************************************/ |
2803 | | |
2804 | | /** |
2805 | | * Is this a sequence of blank chars that one can ignore ? |
2806 | | * |
2807 | | * @param ctxt an XML parser context |
2808 | | * @param str a xmlChar * |
2809 | | * @param len the size of `str` |
2810 | | * @param blank_chars we know the chars are blanks |
2811 | | * @returns 1 if ignorable 0 otherwise. |
2812 | | */ |
2813 | | |
2814 | | static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2815 | 10.7k | int blank_chars) { |
2816 | 10.7k | int i; |
2817 | 10.7k | xmlNodePtr lastChild; |
2818 | | |
2819 | | /* |
2820 | | * Check for xml:space value. |
2821 | | */ |
2822 | 10.7k | if ((ctxt->space == NULL) || (*(ctxt->space) == 1) || |
2823 | 10.7k | (*(ctxt->space) == -2)) |
2824 | 4.10k | return(0); |
2825 | | |
2826 | | /* |
2827 | | * Check that the string is made of blanks |
2828 | | */ |
2829 | 6.63k | if (blank_chars == 0) { |
2830 | 6.36k | for (i = 0;i < len;i++) |
2831 | 4.88k | if (!(IS_BLANK_CH(str[i]))) return(0); |
2832 | 4.22k | } |
2833 | | |
2834 | | /* |
2835 | | * Look if the element is mixed content in the DTD if available |
2836 | | */ |
2837 | 3.89k | if (ctxt->node == NULL) return(0); |
2838 | 3.89k | if (ctxt->myDoc != NULL) { |
2839 | 3.89k | xmlElementPtr elemDecl = NULL; |
2840 | 3.89k | xmlDocPtr doc = ctxt->myDoc; |
2841 | 3.89k | const xmlChar *prefix = NULL; |
2842 | | |
2843 | 3.89k | if (ctxt->node->ns) |
2844 | 265 | prefix = ctxt->node->ns->prefix; |
2845 | 3.89k | if (doc->intSubset != NULL) |
2846 | 1.40k | elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name, |
2847 | 1.40k | prefix); |
2848 | 3.89k | if ((elemDecl == NULL) && (doc->extSubset != NULL)) |
2849 | 0 | elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name, |
2850 | 0 | prefix); |
2851 | 3.89k | if (elemDecl != NULL) { |
2852 | 985 | if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT) |
2853 | 204 | return(1); |
2854 | 781 | if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) || |
2855 | 781 | (elemDecl->etype == XML_ELEMENT_TYPE_MIXED)) |
2856 | 504 | return(0); |
2857 | 781 | } |
2858 | 3.89k | } |
2859 | | |
2860 | | /* |
2861 | | * Otherwise, heuristic :-\ |
2862 | | * |
2863 | | * When push parsing, we could be at the end of a chunk. |
2864 | | * This makes the look-ahead and consequently the NOBLANKS |
2865 | | * option unreliable. |
2866 | | */ |
2867 | 3.18k | if ((RAW != '<') && (RAW != 0xD)) return(0); |
2868 | 2.96k | if ((ctxt->node->children == NULL) && |
2869 | 2.96k | (RAW == '<') && (NXT(1) == '/')) return(0); |
2870 | | |
2871 | 2.51k | lastChild = xmlGetLastChild(ctxt->node); |
2872 | 2.51k | if (lastChild == NULL) { |
2873 | 1.66k | if ((ctxt->node->type != XML_ELEMENT_NODE) && |
2874 | 1.66k | (ctxt->node->content != NULL)) return(0); |
2875 | 1.66k | } else if (xmlNodeIsText(lastChild)) |
2876 | 195 | return(0); |
2877 | 654 | else if ((ctxt->node->children != NULL) && |
2878 | 654 | (xmlNodeIsText(ctxt->node->children))) |
2879 | 67 | return(0); |
2880 | 2.25k | return(1); |
2881 | 2.51k | } |
2882 | | |
2883 | | /************************************************************************ |
2884 | | * * |
2885 | | * Extra stuff for namespace support * |
2886 | | * Relates to http://www.w3.org/TR/WD-xml-names * |
2887 | | * * |
2888 | | ************************************************************************/ |
2889 | | |
2890 | | /** |
2891 | | * parse an UTF8 encoded XML qualified name string |
2892 | | * |
2893 | | * @deprecated Don't use. |
2894 | | * |
2895 | | * [NS 5] QName ::= (Prefix ':')? LocalPart |
2896 | | * |
2897 | | * [NS 6] Prefix ::= NCName |
2898 | | * |
2899 | | * [NS 7] LocalPart ::= NCName |
2900 | | * |
2901 | | * @param ctxt an XML parser context |
2902 | | * @param name an XML parser context |
2903 | | * @param prefixOut a xmlChar ** |
2904 | | * @returns the local part, and prefix is updated |
2905 | | * to get the Prefix if any. |
2906 | | */ |
2907 | | |
2908 | | xmlChar * |
2909 | 0 | xmlSplitQName(xmlParserCtxtPtr ctxt, const xmlChar *name, xmlChar **prefixOut) { |
2910 | 0 | xmlChar buf[XML_MAX_NAMELEN + 5]; |
2911 | 0 | xmlChar *buffer = NULL; |
2912 | 0 | int len = 0; |
2913 | 0 | int max = XML_MAX_NAMELEN; |
2914 | 0 | xmlChar *ret = NULL; |
2915 | 0 | xmlChar *prefix; |
2916 | 0 | const xmlChar *cur = name; |
2917 | 0 | int c; |
2918 | |
|
2919 | 0 | if (prefixOut == NULL) return(NULL); |
2920 | 0 | *prefixOut = NULL; |
2921 | |
|
2922 | 0 | if (cur == NULL) return(NULL); |
2923 | | |
2924 | | /* nasty but well=formed */ |
2925 | 0 | if (cur[0] == ':') |
2926 | 0 | return(xmlStrdup(name)); |
2927 | | |
2928 | 0 | c = *cur++; |
2929 | 0 | while ((c != 0) && (c != ':') && (len < max)) { /* tested bigname.xml */ |
2930 | 0 | buf[len++] = c; |
2931 | 0 | c = *cur++; |
2932 | 0 | } |
2933 | 0 | if (len >= max) { |
2934 | | /* |
2935 | | * Okay someone managed to make a huge name, so he's ready to pay |
2936 | | * for the processing speed. |
2937 | | */ |
2938 | 0 | max = len * 2; |
2939 | |
|
2940 | 0 | buffer = xmlMalloc(max); |
2941 | 0 | if (buffer == NULL) { |
2942 | 0 | xmlErrMemory(ctxt); |
2943 | 0 | return(NULL); |
2944 | 0 | } |
2945 | 0 | memcpy(buffer, buf, len); |
2946 | 0 | while ((c != 0) && (c != ':')) { /* tested bigname.xml */ |
2947 | 0 | if (len + 10 > max) { |
2948 | 0 | xmlChar *tmp; |
2949 | 0 | int newSize; |
2950 | |
|
2951 | 0 | newSize = xmlGrowCapacity(max, 1, 1, XML_MAX_ITEMS); |
2952 | 0 | if (newSize < 0) { |
2953 | 0 | xmlErrMemory(ctxt); |
2954 | 0 | xmlFree(buffer); |
2955 | 0 | return(NULL); |
2956 | 0 | } |
2957 | 0 | tmp = xmlRealloc(buffer, newSize); |
2958 | 0 | if (tmp == NULL) { |
2959 | 0 | xmlErrMemory(ctxt); |
2960 | 0 | xmlFree(buffer); |
2961 | 0 | return(NULL); |
2962 | 0 | } |
2963 | 0 | buffer = tmp; |
2964 | 0 | max = newSize; |
2965 | 0 | } |
2966 | 0 | buffer[len++] = c; |
2967 | 0 | c = *cur++; |
2968 | 0 | } |
2969 | 0 | buffer[len] = 0; |
2970 | 0 | } |
2971 | | |
2972 | 0 | if ((c == ':') && (*cur == 0)) { |
2973 | 0 | if (buffer != NULL) |
2974 | 0 | xmlFree(buffer); |
2975 | 0 | return(xmlStrdup(name)); |
2976 | 0 | } |
2977 | | |
2978 | 0 | if (buffer == NULL) { |
2979 | 0 | ret = xmlStrndup(buf, len); |
2980 | 0 | if (ret == NULL) { |
2981 | 0 | xmlErrMemory(ctxt); |
2982 | 0 | return(NULL); |
2983 | 0 | } |
2984 | 0 | } else { |
2985 | 0 | ret = buffer; |
2986 | 0 | buffer = NULL; |
2987 | 0 | max = XML_MAX_NAMELEN; |
2988 | 0 | } |
2989 | | |
2990 | | |
2991 | 0 | if (c == ':') { |
2992 | 0 | c = *cur; |
2993 | 0 | prefix = ret; |
2994 | 0 | if (c == 0) { |
2995 | 0 | ret = xmlStrndup(BAD_CAST "", 0); |
2996 | 0 | if (ret == NULL) { |
2997 | 0 | xmlFree(prefix); |
2998 | 0 | return(NULL); |
2999 | 0 | } |
3000 | 0 | *prefixOut = prefix; |
3001 | 0 | return(ret); |
3002 | 0 | } |
3003 | 0 | len = 0; |
3004 | | |
3005 | | /* |
3006 | | * Check that the first character is proper to start |
3007 | | * a new name |
3008 | | */ |
3009 | 0 | if (!(((c >= 0x61) && (c <= 0x7A)) || |
3010 | 0 | ((c >= 0x41) && (c <= 0x5A)) || |
3011 | 0 | (c == '_') || (c == ':'))) { |
3012 | 0 | int l; |
3013 | 0 | int first = CUR_SCHAR(cur, l); |
3014 | |
|
3015 | 0 | if (!IS_LETTER(first) && (first != '_')) { |
3016 | 0 | xmlFatalErrMsgStr(ctxt, XML_NS_ERR_QNAME, |
3017 | 0 | "Name %s is not XML Namespace compliant\n", |
3018 | 0 | name); |
3019 | 0 | } |
3020 | 0 | } |
3021 | 0 | cur++; |
3022 | |
|
3023 | 0 | while ((c != 0) && (len < max)) { /* tested bigname2.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 | xmlFree(prefix); |
3038 | 0 | return(NULL); |
3039 | 0 | } |
3040 | 0 | memcpy(buffer, buf, len); |
3041 | 0 | while (c != 0) { /* tested bigname2.xml */ |
3042 | 0 | if (len + 10 > max) { |
3043 | 0 | xmlChar *tmp; |
3044 | 0 | int newSize; |
3045 | |
|
3046 | 0 | newSize = xmlGrowCapacity(max, 1, 1, XML_MAX_ITEMS); |
3047 | 0 | if (newSize < 0) { |
3048 | 0 | xmlErrMemory(ctxt); |
3049 | 0 | xmlFree(buffer); |
3050 | 0 | return(NULL); |
3051 | 0 | } |
3052 | 0 | tmp = xmlRealloc(buffer, newSize); |
3053 | 0 | if (tmp == NULL) { |
3054 | 0 | xmlErrMemory(ctxt); |
3055 | 0 | xmlFree(prefix); |
3056 | 0 | xmlFree(buffer); |
3057 | 0 | return(NULL); |
3058 | 0 | } |
3059 | 0 | buffer = tmp; |
3060 | 0 | max = newSize; |
3061 | 0 | } |
3062 | 0 | buffer[len++] = c; |
3063 | 0 | c = *cur++; |
3064 | 0 | } |
3065 | 0 | buffer[len] = 0; |
3066 | 0 | } |
3067 | | |
3068 | 0 | if (buffer == NULL) { |
3069 | 0 | ret = xmlStrndup(buf, len); |
3070 | 0 | if (ret == NULL) { |
3071 | 0 | xmlFree(prefix); |
3072 | 0 | return(NULL); |
3073 | 0 | } |
3074 | 0 | } else { |
3075 | 0 | ret = buffer; |
3076 | 0 | } |
3077 | | |
3078 | 0 | *prefixOut = prefix; |
3079 | 0 | } |
3080 | | |
3081 | 0 | return(ret); |
3082 | 0 | } |
3083 | | |
3084 | | /************************************************************************ |
3085 | | * * |
3086 | | * The parser itself * |
3087 | | * Relates to http://www.w3.org/TR/REC-xml * |
3088 | | * * |
3089 | | ************************************************************************/ |
3090 | | |
3091 | | /************************************************************************ |
3092 | | * * |
3093 | | * Routines to parse Name, NCName and NmToken * |
3094 | | * * |
3095 | | ************************************************************************/ |
3096 | | |
3097 | | /* |
3098 | | * The two following functions are related to the change of accepted |
3099 | | * characters for Name and NmToken in the Revision 5 of XML-1.0 |
3100 | | * They correspond to the modified production [4] and the new production [4a] |
3101 | | * changes in that revision. Also note that the macros used for the |
3102 | | * productions Letter, Digit, CombiningChar and Extender are not needed |
3103 | | * anymore. |
3104 | | * We still keep compatibility to pre-revision5 parsing semantic if the |
3105 | | * new XML_PARSE_OLD10 option is given to the parser. |
3106 | | */ |
3107 | | static int |
3108 | 669k | xmlIsNameStartChar(xmlParserCtxtPtr ctxt, int c) { |
3109 | 669k | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3110 | | /* |
3111 | | * Use the new checks of production [4] [4a] amd [5] of the |
3112 | | * Update 5 of XML-1.0 |
3113 | | */ |
3114 | 661k | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3115 | 661k | (((c >= 'a') && (c <= 'z')) || |
3116 | 660k | ((c >= 'A') && (c <= 'Z')) || |
3117 | 660k | (c == '_') || (c == ':') || |
3118 | 660k | ((c >= 0xC0) && (c <= 0xD6)) || |
3119 | 660k | ((c >= 0xD8) && (c <= 0xF6)) || |
3120 | 660k | ((c >= 0xF8) && (c <= 0x2FF)) || |
3121 | 660k | ((c >= 0x370) && (c <= 0x37D)) || |
3122 | 660k | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3123 | 660k | ((c >= 0x200C) && (c <= 0x200D)) || |
3124 | 660k | ((c >= 0x2070) && (c <= 0x218F)) || |
3125 | 660k | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3126 | 660k | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3127 | 660k | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3128 | 660k | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3129 | 660k | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3130 | 616k | return(1); |
3131 | 661k | } else { |
3132 | 8.01k | if (IS_LETTER(c) || (c == '_') || (c == ':')) |
3133 | 6.17k | return(1); |
3134 | 8.01k | } |
3135 | 47.0k | return(0); |
3136 | 669k | } |
3137 | | |
3138 | | static int |
3139 | 920k | xmlIsNameChar(xmlParserCtxtPtr ctxt, int c) { |
3140 | 920k | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3141 | | /* |
3142 | | * Use the new checks of production [4] [4a] amd [5] of the |
3143 | | * Update 5 of XML-1.0 |
3144 | | */ |
3145 | 907k | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3146 | 907k | (((c >= 'a') && (c <= 'z')) || |
3147 | 896k | ((c >= 'A') && (c <= 'Z')) || |
3148 | 896k | ((c >= '0') && (c <= '9')) || /* !start */ |
3149 | 896k | (c == '_') || (c == ':') || |
3150 | 896k | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3151 | 896k | ((c >= 0xC0) && (c <= 0xD6)) || |
3152 | 896k | ((c >= 0xD8) && (c <= 0xF6)) || |
3153 | 896k | ((c >= 0xF8) && (c <= 0x2FF)) || |
3154 | 896k | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3155 | 896k | ((c >= 0x370) && (c <= 0x37D)) || |
3156 | 896k | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3157 | 896k | ((c >= 0x200C) && (c <= 0x200D)) || |
3158 | 896k | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3159 | 896k | ((c >= 0x2070) && (c <= 0x218F)) || |
3160 | 896k | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3161 | 896k | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3162 | 896k | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3163 | 896k | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3164 | 896k | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3165 | 282k | return(1); |
3166 | 907k | } else { |
3167 | 12.8k | if ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3168 | 12.8k | (c == '.') || (c == '-') || |
3169 | 12.8k | (c == '_') || (c == ':') || |
3170 | 12.8k | (IS_COMBINING(c)) || |
3171 | 12.8k | (IS_EXTENDER(c))) |
3172 | 6.24k | return(1); |
3173 | 12.8k | } |
3174 | 632k | return(0); |
3175 | 920k | } |
3176 | | |
3177 | | static const xmlChar * |
3178 | 111k | xmlParseNameComplex(xmlParserCtxtPtr ctxt) { |
3179 | 111k | const xmlChar *ret; |
3180 | 111k | int len = 0, l; |
3181 | 111k | int c; |
3182 | 111k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3183 | 0 | XML_MAX_TEXT_LENGTH : |
3184 | 111k | XML_MAX_NAME_LENGTH; |
3185 | | |
3186 | | /* |
3187 | | * Handler for more complex cases |
3188 | | */ |
3189 | 111k | c = xmlCurrentChar(ctxt, &l); |
3190 | 111k | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3191 | | /* |
3192 | | * Use the new checks of production [4] [4a] amd [5] of the |
3193 | | * Update 5 of XML-1.0 |
3194 | | */ |
3195 | 101k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3196 | 101k | (!(((c >= 'a') && (c <= 'z')) || |
3197 | 90.4k | ((c >= 'A') && (c <= 'Z')) || |
3198 | 90.4k | (c == '_') || (c == ':') || |
3199 | 90.4k | ((c >= 0xC0) && (c <= 0xD6)) || |
3200 | 90.4k | ((c >= 0xD8) && (c <= 0xF6)) || |
3201 | 90.4k | ((c >= 0xF8) && (c <= 0x2FF)) || |
3202 | 90.4k | ((c >= 0x370) && (c <= 0x37D)) || |
3203 | 90.4k | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3204 | 90.4k | ((c >= 0x200C) && (c <= 0x200D)) || |
3205 | 90.4k | ((c >= 0x2070) && (c <= 0x218F)) || |
3206 | 90.4k | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3207 | 90.4k | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3208 | 90.4k | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3209 | 90.4k | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3210 | 92.7k | ((c >= 0x10000) && (c <= 0xEFFFF))))) { |
3211 | 92.7k | return(NULL); |
3212 | 92.7k | } |
3213 | 8.96k | len += l; |
3214 | 8.96k | NEXTL(l); |
3215 | 8.96k | c = xmlCurrentChar(ctxt, &l); |
3216 | 22.2k | while ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3217 | 22.2k | (((c >= 'a') && (c <= 'z')) || |
3218 | 20.4k | ((c >= 'A') && (c <= 'Z')) || |
3219 | 20.4k | ((c >= '0') && (c <= '9')) || /* !start */ |
3220 | 20.4k | (c == '_') || (c == ':') || |
3221 | 20.4k | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3222 | 20.4k | ((c >= 0xC0) && (c <= 0xD6)) || |
3223 | 20.4k | ((c >= 0xD8) && (c <= 0xF6)) || |
3224 | 20.4k | ((c >= 0xF8) && (c <= 0x2FF)) || |
3225 | 20.4k | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3226 | 20.4k | ((c >= 0x370) && (c <= 0x37D)) || |
3227 | 20.4k | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3228 | 20.4k | ((c >= 0x200C) && (c <= 0x200D)) || |
3229 | 20.4k | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3230 | 20.4k | ((c >= 0x2070) && (c <= 0x218F)) || |
3231 | 20.4k | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3232 | 20.4k | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3233 | 20.4k | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3234 | 20.4k | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3235 | 20.4k | ((c >= 0x10000) && (c <= 0xEFFFF)) |
3236 | 20.4k | )) { |
3237 | 13.3k | if (len <= INT_MAX - l) |
3238 | 13.3k | len += l; |
3239 | 13.3k | NEXTL(l); |
3240 | 13.3k | c = xmlCurrentChar(ctxt, &l); |
3241 | 13.3k | } |
3242 | 9.29k | } else { |
3243 | 9.29k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3244 | 9.29k | (!IS_LETTER(c) && (c != '_') && |
3245 | 8.04k | (c != ':'))) { |
3246 | 5.96k | return(NULL); |
3247 | 5.96k | } |
3248 | 3.33k | len += l; |
3249 | 3.33k | NEXTL(l); |
3250 | 3.33k | c = xmlCurrentChar(ctxt, &l); |
3251 | | |
3252 | 7.18k | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3253 | 7.18k | ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3254 | 6.15k | (c == '.') || (c == '-') || |
3255 | 6.15k | (c == '_') || (c == ':') || |
3256 | 6.15k | (IS_COMBINING(c)) || |
3257 | 6.15k | (IS_EXTENDER(c)))) { |
3258 | 3.84k | if (len <= INT_MAX - l) |
3259 | 3.84k | len += l; |
3260 | 3.84k | NEXTL(l); |
3261 | 3.84k | c = xmlCurrentChar(ctxt, &l); |
3262 | 3.84k | } |
3263 | 3.33k | } |
3264 | 12.2k | if (len > maxLength) { |
3265 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3266 | 0 | return(NULL); |
3267 | 0 | } |
3268 | 12.2k | if (ctxt->input->cur - ctxt->input->base < len) { |
3269 | | /* |
3270 | | * There were a couple of bugs where PERefs lead to to a change |
3271 | | * of the buffer. Check the buffer size to avoid passing an invalid |
3272 | | * pointer to xmlDictLookup. |
3273 | | */ |
3274 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
3275 | 0 | "unexpected change of input buffer"); |
3276 | 0 | return (NULL); |
3277 | 0 | } |
3278 | 12.2k | if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r')) |
3279 | 194 | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len); |
3280 | 12.1k | else |
3281 | 12.1k | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len); |
3282 | 12.2k | if (ret == NULL) |
3283 | 0 | xmlErrMemory(ctxt); |
3284 | 12.2k | return(ret); |
3285 | 12.2k | } |
3286 | | |
3287 | | /** |
3288 | | * parse an XML name. |
3289 | | * |
3290 | | * @deprecated Internal function, don't use. |
3291 | | * |
3292 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3293 | | * CombiningChar | Extender |
3294 | | * |
3295 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3296 | | * |
3297 | | * [6] Names ::= Name (#x20 Name)* |
3298 | | * |
3299 | | * @param ctxt an XML parser context |
3300 | | * @returns the Name parsed or NULL |
3301 | | */ |
3302 | | |
3303 | | const xmlChar * |
3304 | 433k | xmlParseName(xmlParserCtxtPtr ctxt) { |
3305 | 433k | const xmlChar *in; |
3306 | 433k | const xmlChar *ret; |
3307 | 433k | size_t count = 0; |
3308 | 433k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3309 | 0 | XML_MAX_TEXT_LENGTH : |
3310 | 433k | XML_MAX_NAME_LENGTH; |
3311 | | |
3312 | 433k | GROW; |
3313 | | |
3314 | | /* |
3315 | | * Accelerator for simple ASCII names |
3316 | | */ |
3317 | 433k | in = ctxt->input->cur; |
3318 | 433k | if (((*in >= 0x61) && (*in <= 0x7A)) || |
3319 | 433k | ((*in >= 0x41) && (*in <= 0x5A)) || |
3320 | 433k | (*in == '_') || (*in == ':')) { |
3321 | 327k | in++; |
3322 | 629k | while (((*in >= 0x61) && (*in <= 0x7A)) || |
3323 | 629k | ((*in >= 0x41) && (*in <= 0x5A)) || |
3324 | 629k | ((*in >= 0x30) && (*in <= 0x39)) || |
3325 | 629k | (*in == '_') || (*in == '-') || |
3326 | 629k | (*in == ':') || (*in == '.')) |
3327 | 302k | in++; |
3328 | 327k | if ((*in > 0) && (*in < 0x80)) { |
3329 | 322k | count = in - ctxt->input->cur; |
3330 | 322k | if (count > maxLength) { |
3331 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3332 | 0 | return(NULL); |
3333 | 0 | } |
3334 | 322k | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3335 | 322k | ctxt->input->cur = in; |
3336 | 322k | ctxt->input->col += count; |
3337 | 322k | if (ret == NULL) |
3338 | 0 | xmlErrMemory(ctxt); |
3339 | 322k | return(ret); |
3340 | 322k | } |
3341 | 327k | } |
3342 | | /* accelerator for special cases */ |
3343 | 111k | return(xmlParseNameComplex(ctxt)); |
3344 | 433k | } |
3345 | | |
3346 | | static xmlHashedString |
3347 | 70.0k | xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) { |
3348 | 70.0k | xmlHashedString ret; |
3349 | 70.0k | int len = 0, l; |
3350 | 70.0k | int c; |
3351 | 70.0k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3352 | 0 | XML_MAX_TEXT_LENGTH : |
3353 | 70.0k | XML_MAX_NAME_LENGTH; |
3354 | 70.0k | size_t startPosition = 0; |
3355 | | |
3356 | 70.0k | ret.name = NULL; |
3357 | 70.0k | ret.hashValue = 0; |
3358 | | |
3359 | | /* |
3360 | | * Handler for more complex cases |
3361 | | */ |
3362 | 70.0k | startPosition = CUR_PTR - BASE_PTR; |
3363 | 70.0k | c = xmlCurrentChar(ctxt, &l); |
3364 | 70.0k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3365 | 70.0k | (!xmlIsNameStartChar(ctxt, c) || (c == ':'))) { |
3366 | 63.7k | return(ret); |
3367 | 63.7k | } |
3368 | | |
3369 | 27.3k | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3370 | 27.3k | (xmlIsNameChar(ctxt, c) && (c != ':'))) { |
3371 | 21.0k | if (len <= INT_MAX - l) |
3372 | 21.0k | len += l; |
3373 | 21.0k | NEXTL(l); |
3374 | 21.0k | c = xmlCurrentChar(ctxt, &l); |
3375 | 21.0k | } |
3376 | 6.33k | if (len > maxLength) { |
3377 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3378 | 0 | return(ret); |
3379 | 0 | } |
3380 | 6.33k | ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len); |
3381 | 6.33k | if (ret.name == NULL) |
3382 | 0 | xmlErrMemory(ctxt); |
3383 | 6.33k | return(ret); |
3384 | 6.33k | } |
3385 | | |
3386 | | /** |
3387 | | * parse an XML name. |
3388 | | * |
3389 | | * [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' | |
3390 | | * CombiningChar | Extender |
3391 | | * |
3392 | | * [5NS] NCName ::= (Letter | '_') (NCNameChar)* |
3393 | | * |
3394 | | * @param ctxt an XML parser context |
3395 | | * @returns the Name parsed or NULL |
3396 | | */ |
3397 | | |
3398 | | static xmlHashedString |
3399 | 180k | xmlParseNCName(xmlParserCtxtPtr ctxt) { |
3400 | 180k | const xmlChar *in, *e; |
3401 | 180k | xmlHashedString ret; |
3402 | 180k | size_t count = 0; |
3403 | 180k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3404 | 0 | XML_MAX_TEXT_LENGTH : |
3405 | 180k | XML_MAX_NAME_LENGTH; |
3406 | | |
3407 | 180k | ret.name = NULL; |
3408 | | |
3409 | | /* |
3410 | | * Accelerator for simple ASCII names |
3411 | | */ |
3412 | 180k | in = ctxt->input->cur; |
3413 | 180k | e = ctxt->input->end; |
3414 | 180k | if ((((*in >= 0x61) && (*in <= 0x7A)) || |
3415 | 180k | ((*in >= 0x41) && (*in <= 0x5A)) || |
3416 | 180k | (*in == '_')) && (in < e)) { |
3417 | 112k | in++; |
3418 | 201k | while ((((*in >= 0x61) && (*in <= 0x7A)) || |
3419 | 201k | ((*in >= 0x41) && (*in <= 0x5A)) || |
3420 | 201k | ((*in >= 0x30) && (*in <= 0x39)) || |
3421 | 201k | (*in == '_') || (*in == '-') || |
3422 | 201k | (*in == '.')) && (in < e)) |
3423 | 89.2k | in++; |
3424 | 112k | if (in >= e) |
3425 | 19 | goto complex; |
3426 | 112k | if ((*in > 0) && (*in < 0x80)) { |
3427 | 110k | count = in - ctxt->input->cur; |
3428 | 110k | if (count > maxLength) { |
3429 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3430 | 0 | return(ret); |
3431 | 0 | } |
3432 | 110k | ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count); |
3433 | 110k | ctxt->input->cur = in; |
3434 | 110k | ctxt->input->col += count; |
3435 | 110k | if (ret.name == NULL) { |
3436 | 0 | xmlErrMemory(ctxt); |
3437 | 0 | } |
3438 | 110k | return(ret); |
3439 | 110k | } |
3440 | 112k | } |
3441 | 70.0k | complex: |
3442 | 70.0k | return(xmlParseNCNameComplex(ctxt)); |
3443 | 180k | } |
3444 | | |
3445 | | /** |
3446 | | * parse an XML name and compares for match |
3447 | | * (specialized for endtag parsing) |
3448 | | * |
3449 | | * @param ctxt an XML parser context |
3450 | | * @param other the name to compare with |
3451 | | * @returns NULL for an illegal name, (xmlChar*) 1 for success |
3452 | | * and the name for mismatch |
3453 | | */ |
3454 | | |
3455 | | static const xmlChar * |
3456 | 13.8k | xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) { |
3457 | 13.8k | register const xmlChar *cmp = other; |
3458 | 13.8k | register const xmlChar *in; |
3459 | 13.8k | const xmlChar *ret; |
3460 | | |
3461 | 13.8k | GROW; |
3462 | | |
3463 | 13.8k | in = ctxt->input->cur; |
3464 | 19.2k | while (*in != 0 && *in == *cmp) { |
3465 | 5.46k | ++in; |
3466 | 5.46k | ++cmp; |
3467 | 5.46k | } |
3468 | 13.8k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
3469 | | /* success */ |
3470 | 2.54k | ctxt->input->col += in - ctxt->input->cur; |
3471 | 2.54k | ctxt->input->cur = in; |
3472 | 2.54k | return (const xmlChar*) 1; |
3473 | 2.54k | } |
3474 | | /* failure (or end of input buffer), check with full function */ |
3475 | 11.2k | ret = xmlParseName (ctxt); |
3476 | | /* strings coming from the dictionary direct compare possible */ |
3477 | 11.2k | if (ret == other) { |
3478 | 915 | return (const xmlChar*) 1; |
3479 | 915 | } |
3480 | 10.3k | return ret; |
3481 | 11.2k | } |
3482 | | |
3483 | | /** |
3484 | | * parse an XML name. |
3485 | | * |
3486 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3487 | | * CombiningChar | Extender |
3488 | | * |
3489 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3490 | | * |
3491 | | * [6] Names ::= Name (#x20 Name)* |
3492 | | * |
3493 | | * @param ctxt an XML parser context |
3494 | | * @param str a pointer to the string pointer (IN/OUT) |
3495 | | * @returns the Name parsed or NULL. The `str` pointer |
3496 | | * is updated to the current location in the string. |
3497 | | */ |
3498 | | |
3499 | | static xmlChar * |
3500 | 606k | xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) { |
3501 | 606k | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3502 | 606k | xmlChar *ret; |
3503 | 606k | const xmlChar *cur = *str; |
3504 | 606k | int len = 0, l; |
3505 | 606k | int c; |
3506 | 606k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3507 | 0 | XML_MAX_TEXT_LENGTH : |
3508 | 606k | XML_MAX_NAME_LENGTH; |
3509 | | |
3510 | 606k | c = CUR_SCHAR(cur, l); |
3511 | 606k | if (!xmlIsNameStartChar(ctxt, c)) { |
3512 | 1.52k | return(NULL); |
3513 | 1.52k | } |
3514 | | |
3515 | 605k | COPY_BUF(buf, len, c); |
3516 | 605k | cur += l; |
3517 | 605k | c = CUR_SCHAR(cur, l); |
3518 | 748k | while (xmlIsNameChar(ctxt, c)) { |
3519 | 144k | COPY_BUF(buf, len, c); |
3520 | 144k | cur += l; |
3521 | 144k | c = CUR_SCHAR(cur, l); |
3522 | 144k | if (len >= XML_MAX_NAMELEN) { /* test bigentname.xml */ |
3523 | | /* |
3524 | | * Okay someone managed to make a huge name, so he's ready to pay |
3525 | | * for the processing speed. |
3526 | | */ |
3527 | 654 | xmlChar *buffer; |
3528 | 654 | int max = len * 2; |
3529 | | |
3530 | 654 | buffer = xmlMalloc(max); |
3531 | 654 | if (buffer == NULL) { |
3532 | 0 | xmlErrMemory(ctxt); |
3533 | 0 | return(NULL); |
3534 | 0 | } |
3535 | 654 | memcpy(buffer, buf, len); |
3536 | 20.7k | while (xmlIsNameChar(ctxt, c)) { |
3537 | 20.1k | if (len + 10 > max) { |
3538 | 206 | xmlChar *tmp; |
3539 | 206 | int newSize; |
3540 | | |
3541 | 206 | newSize = xmlGrowCapacity(max, 1, 1, maxLength); |
3542 | 206 | if (newSize < 0) { |
3543 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3544 | 0 | xmlFree(buffer); |
3545 | 0 | return(NULL); |
3546 | 0 | } |
3547 | 206 | tmp = xmlRealloc(buffer, newSize); |
3548 | 206 | if (tmp == NULL) { |
3549 | 0 | xmlErrMemory(ctxt); |
3550 | 0 | xmlFree(buffer); |
3551 | 0 | return(NULL); |
3552 | 0 | } |
3553 | 206 | buffer = tmp; |
3554 | 206 | max = newSize; |
3555 | 206 | } |
3556 | 20.1k | COPY_BUF(buffer, len, c); |
3557 | 20.1k | cur += l; |
3558 | 20.1k | c = CUR_SCHAR(cur, l); |
3559 | 20.1k | } |
3560 | 654 | buffer[len] = 0; |
3561 | 654 | *str = cur; |
3562 | 654 | return(buffer); |
3563 | 654 | } |
3564 | 144k | } |
3565 | 604k | if (len > maxLength) { |
3566 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3567 | 0 | return(NULL); |
3568 | 0 | } |
3569 | 604k | *str = cur; |
3570 | 604k | ret = xmlStrndup(buf, len); |
3571 | 604k | if (ret == NULL) |
3572 | 0 | xmlErrMemory(ctxt); |
3573 | 604k | return(ret); |
3574 | 604k | } |
3575 | | |
3576 | | /** |
3577 | | * parse an XML Nmtoken. |
3578 | | * |
3579 | | * @deprecated Internal function, don't use. |
3580 | | * |
3581 | | * [7] Nmtoken ::= (NameChar)+ |
3582 | | * |
3583 | | * [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)* |
3584 | | * |
3585 | | * @param ctxt an XML parser context |
3586 | | * @returns the Nmtoken parsed or NULL |
3587 | | */ |
3588 | | |
3589 | | xmlChar * |
3590 | 22.1k | xmlParseNmtoken(xmlParserCtxtPtr ctxt) { |
3591 | 22.1k | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3592 | 22.1k | xmlChar *ret; |
3593 | 22.1k | int len = 0, l; |
3594 | 22.1k | int c; |
3595 | 22.1k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3596 | 0 | XML_MAX_TEXT_LENGTH : |
3597 | 22.1k | XML_MAX_NAME_LENGTH; |
3598 | | |
3599 | 22.1k | c = xmlCurrentChar(ctxt, &l); |
3600 | | |
3601 | 105k | while (xmlIsNameChar(ctxt, c)) { |
3602 | 84.3k | COPY_BUF(buf, len, c); |
3603 | 84.3k | NEXTL(l); |
3604 | 84.3k | c = xmlCurrentChar(ctxt, &l); |
3605 | 84.3k | if (len >= XML_MAX_NAMELEN) { |
3606 | | /* |
3607 | | * Okay someone managed to make a huge token, so he's ready to pay |
3608 | | * for the processing speed. |
3609 | | */ |
3610 | 610 | xmlChar *buffer; |
3611 | 610 | int max = len * 2; |
3612 | | |
3613 | 610 | buffer = xmlMalloc(max); |
3614 | 610 | if (buffer == NULL) { |
3615 | 0 | xmlErrMemory(ctxt); |
3616 | 0 | return(NULL); |
3617 | 0 | } |
3618 | 610 | memcpy(buffer, buf, len); |
3619 | 18.8k | while (xmlIsNameChar(ctxt, c)) { |
3620 | 18.2k | if (len + 10 > max) { |
3621 | 196 | xmlChar *tmp; |
3622 | 196 | int newSize; |
3623 | | |
3624 | 196 | newSize = xmlGrowCapacity(max, 1, 1, maxLength); |
3625 | 196 | if (newSize < 0) { |
3626 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3627 | 0 | xmlFree(buffer); |
3628 | 0 | return(NULL); |
3629 | 0 | } |
3630 | 196 | tmp = xmlRealloc(buffer, newSize); |
3631 | 196 | if (tmp == NULL) { |
3632 | 0 | xmlErrMemory(ctxt); |
3633 | 0 | xmlFree(buffer); |
3634 | 0 | return(NULL); |
3635 | 0 | } |
3636 | 196 | buffer = tmp; |
3637 | 196 | max = newSize; |
3638 | 196 | } |
3639 | 18.2k | COPY_BUF(buffer, len, c); |
3640 | 18.2k | NEXTL(l); |
3641 | 18.2k | c = xmlCurrentChar(ctxt, &l); |
3642 | 18.2k | } |
3643 | 610 | buffer[len] = 0; |
3644 | 610 | return(buffer); |
3645 | 610 | } |
3646 | 84.3k | } |
3647 | 21.5k | if (len == 0) |
3648 | 1.67k | return(NULL); |
3649 | 19.8k | if (len > maxLength) { |
3650 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3651 | 0 | return(NULL); |
3652 | 0 | } |
3653 | 19.8k | ret = xmlStrndup(buf, len); |
3654 | 19.8k | if (ret == NULL) |
3655 | 0 | xmlErrMemory(ctxt); |
3656 | 19.8k | return(ret); |
3657 | 19.8k | } |
3658 | | |
3659 | | /** |
3660 | | * Validate an entity value and expand parameter entities. |
3661 | | * |
3662 | | * @param ctxt parser context |
3663 | | * @param buf string buffer |
3664 | | * @param str entity value |
3665 | | * @param length size of entity value |
3666 | | * @param depth nesting depth |
3667 | | */ |
3668 | | static void |
3669 | | xmlExpandPEsInEntityValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf, |
3670 | 13.8k | const xmlChar *str, int length, int depth) { |
3671 | 13.8k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
3672 | 13.8k | const xmlChar *end, *chunk; |
3673 | 13.8k | int c, l; |
3674 | | |
3675 | 13.8k | if (str == NULL) |
3676 | 0 | return; |
3677 | | |
3678 | 13.8k | depth += 1; |
3679 | 13.8k | if (depth > maxDepth) { |
3680 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
3681 | 0 | "Maximum entity nesting depth exceeded"); |
3682 | 0 | return; |
3683 | 0 | } |
3684 | | |
3685 | 13.8k | end = str + length; |
3686 | 13.8k | chunk = str; |
3687 | | |
3688 | 363k | while ((str < end) && (!PARSER_STOPPED(ctxt))) { |
3689 | 355k | c = *str; |
3690 | | |
3691 | 355k | if (c >= 0x80) { |
3692 | 24.3k | l = xmlUTF8MultibyteLen(ctxt, str, |
3693 | 24.3k | "invalid character in entity value\n"); |
3694 | 24.3k | if (l == 0) { |
3695 | 5.68k | if (chunk < str) |
3696 | 1.34k | xmlSBufAddString(buf, chunk, str - chunk); |
3697 | 5.68k | xmlSBufAddReplChar(buf); |
3698 | 5.68k | str += 1; |
3699 | 5.68k | chunk = str; |
3700 | 18.6k | } else { |
3701 | 18.6k | str += l; |
3702 | 18.6k | } |
3703 | 330k | } else if (c == '&') { |
3704 | 27.6k | if (str[1] == '#') { |
3705 | 5.93k | if (chunk < str) |
3706 | 3.16k | xmlSBufAddString(buf, chunk, str - chunk); |
3707 | | |
3708 | 5.93k | c = xmlParseStringCharRef(ctxt, &str); |
3709 | 5.93k | if (c == 0) |
3710 | 2.51k | return; |
3711 | | |
3712 | 3.42k | xmlSBufAddChar(buf, c); |
3713 | | |
3714 | 3.42k | chunk = str; |
3715 | 21.6k | } else { |
3716 | 21.6k | xmlChar *name; |
3717 | | |
3718 | | /* |
3719 | | * General entity references are checked for |
3720 | | * syntactic validity. |
3721 | | */ |
3722 | 21.6k | str++; |
3723 | 21.6k | name = xmlParseStringName(ctxt, &str); |
3724 | | |
3725 | 21.6k | if ((name == NULL) || (*str++ != ';')) { |
3726 | 636 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR, |
3727 | 636 | "EntityValue: '&' forbidden except for entities " |
3728 | 636 | "references\n"); |
3729 | 636 | xmlFree(name); |
3730 | 636 | return; |
3731 | 636 | } |
3732 | | |
3733 | 21.0k | xmlFree(name); |
3734 | 21.0k | } |
3735 | 303k | } else if (c == '%') { |
3736 | 2.13k | xmlEntityPtr ent; |
3737 | | |
3738 | 2.13k | if (chunk < str) |
3739 | 1.24k | xmlSBufAddString(buf, chunk, str - chunk); |
3740 | | |
3741 | 2.13k | ent = xmlParseStringPEReference(ctxt, &str); |
3742 | 2.13k | if (ent == NULL) |
3743 | 1.85k | return; |
3744 | | |
3745 | 277 | if (!PARSER_EXTERNAL(ctxt)) { |
3746 | 277 | xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL); |
3747 | 277 | return; |
3748 | 277 | } |
3749 | | |
3750 | 0 | if (ent->content == NULL) { |
3751 | | /* |
3752 | | * Note: external parsed entities will not be loaded, |
3753 | | * it is not required for a non-validating parser to |
3754 | | * complete external PEReferences coming from the |
3755 | | * internal subset |
3756 | | */ |
3757 | 0 | if (((ctxt->options & XML_PARSE_NO_XXE) == 0) && |
3758 | 0 | ((ctxt->replaceEntities) || |
3759 | 0 | (ctxt->validate))) { |
3760 | 0 | xmlLoadEntityContent(ctxt, ent); |
3761 | 0 | } else { |
3762 | 0 | xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING, |
3763 | 0 | "not validating will not read content for " |
3764 | 0 | "PE entity %s\n", ent->name, NULL); |
3765 | 0 | } |
3766 | 0 | } |
3767 | | |
3768 | | /* |
3769 | | * TODO: Skip if ent->content is still NULL. |
3770 | | */ |
3771 | |
|
3772 | 0 | if (xmlParserEntityCheck(ctxt, ent->length)) |
3773 | 0 | return; |
3774 | | |
3775 | 0 | if (ent->flags & XML_ENT_EXPANDING) { |
3776 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
3777 | 0 | xmlHaltParser(ctxt); |
3778 | 0 | return; |
3779 | 0 | } |
3780 | | |
3781 | 0 | ent->flags |= XML_ENT_EXPANDING; |
3782 | 0 | xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length, |
3783 | 0 | depth); |
3784 | 0 | ent->flags &= ~XML_ENT_EXPANDING; |
3785 | |
|
3786 | 0 | chunk = str; |
3787 | 300k | } else { |
3788 | | /* Normal ASCII char */ |
3789 | 300k | if (!IS_BYTE_CHAR(c)) { |
3790 | 2.82k | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
3791 | 2.82k | "invalid character in entity value\n"); |
3792 | 2.82k | if (chunk < str) |
3793 | 890 | xmlSBufAddString(buf, chunk, str - chunk); |
3794 | 2.82k | xmlSBufAddReplChar(buf); |
3795 | 2.82k | str += 1; |
3796 | 2.82k | chunk = str; |
3797 | 298k | } else { |
3798 | 298k | str += 1; |
3799 | 298k | } |
3800 | 300k | } |
3801 | 355k | } |
3802 | | |
3803 | 8.55k | if (chunk < str) |
3804 | 7.33k | xmlSBufAddString(buf, chunk, str - chunk); |
3805 | 8.55k | } |
3806 | | |
3807 | | /** |
3808 | | * parse a value for ENTITY declarations |
3809 | | * |
3810 | | * @deprecated Internal function, don't use. |
3811 | | * |
3812 | | * [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' | |
3813 | | * "'" ([^%&'] | PEReference | Reference)* "'" |
3814 | | * |
3815 | | * @param ctxt an XML parser context |
3816 | | * @param orig if non-NULL store a copy of the original entity value |
3817 | | * @returns the EntityValue parsed with reference substituted or NULL |
3818 | | */ |
3819 | | xmlChar * |
3820 | 14.7k | xmlParseEntityValue(xmlParserCtxtPtr ctxt, xmlChar **orig) { |
3821 | 14.7k | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3822 | 0 | XML_MAX_HUGE_LENGTH : |
3823 | 14.7k | XML_MAX_TEXT_LENGTH; |
3824 | 14.7k | xmlSBuf buf; |
3825 | 14.7k | const xmlChar *start; |
3826 | 14.7k | int quote, length; |
3827 | | |
3828 | 14.7k | xmlSBufInit(&buf, maxLength); |
3829 | | |
3830 | 14.7k | GROW; |
3831 | | |
3832 | 14.7k | quote = CUR; |
3833 | 14.7k | if ((quote != '"') && (quote != '\'')) { |
3834 | 0 | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
3835 | 0 | return(NULL); |
3836 | 0 | } |
3837 | 14.7k | CUR_PTR++; |
3838 | | |
3839 | 14.7k | length = 0; |
3840 | | |
3841 | | /* |
3842 | | * Copy raw content of the entity into a buffer |
3843 | | */ |
3844 | 492k | while (1) { |
3845 | 492k | int c; |
3846 | | |
3847 | 492k | if (PARSER_STOPPED(ctxt)) |
3848 | 0 | goto error; |
3849 | | |
3850 | 492k | if (CUR_PTR >= ctxt->input->end) { |
3851 | 916 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL); |
3852 | 916 | goto error; |
3853 | 916 | } |
3854 | | |
3855 | 491k | c = CUR; |
3856 | | |
3857 | 491k | if (c == 0) { |
3858 | 2 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
3859 | 2 | "invalid character in entity value\n"); |
3860 | 2 | goto error; |
3861 | 2 | } |
3862 | 491k | if (c == quote) |
3863 | 13.8k | break; |
3864 | 477k | NEXTL(1); |
3865 | 477k | length += 1; |
3866 | | |
3867 | | /* |
3868 | | * TODO: Check growth threshold |
3869 | | */ |
3870 | 477k | if (ctxt->input->end - CUR_PTR < 10) |
3871 | 27.0k | GROW; |
3872 | 477k | } |
3873 | | |
3874 | 13.8k | start = CUR_PTR - length; |
3875 | | |
3876 | 13.8k | if (orig != NULL) { |
3877 | 13.8k | *orig = xmlStrndup(start, length); |
3878 | 13.8k | if (*orig == NULL) |
3879 | 0 | xmlErrMemory(ctxt); |
3880 | 13.8k | } |
3881 | | |
3882 | 13.8k | xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr); |
3883 | | |
3884 | 13.8k | NEXTL(1); |
3885 | | |
3886 | 13.8k | return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long")); |
3887 | | |
3888 | 918 | error: |
3889 | 918 | xmlSBufCleanup(&buf, ctxt, "entity length too long"); |
3890 | 918 | return(NULL); |
3891 | 14.7k | } |
3892 | | |
3893 | | /** |
3894 | | * Check an entity reference in an attribute value for validity |
3895 | | * without expanding it. |
3896 | | * |
3897 | | * @param ctxt parser context |
3898 | | * @param pent entity |
3899 | | * @param depth nesting depth |
3900 | | */ |
3901 | | static void |
3902 | 449 | xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) { |
3903 | 449 | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
3904 | 449 | const xmlChar *str; |
3905 | 449 | unsigned long expandedSize = pent->length; |
3906 | 449 | int c, flags; |
3907 | | |
3908 | 449 | depth += 1; |
3909 | 449 | if (depth > maxDepth) { |
3910 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
3911 | 0 | "Maximum entity nesting depth exceeded"); |
3912 | 0 | return; |
3913 | 0 | } |
3914 | | |
3915 | 449 | if (pent->flags & XML_ENT_EXPANDING) { |
3916 | 12 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
3917 | 12 | xmlHaltParser(ctxt); |
3918 | 12 | return; |
3919 | 12 | } |
3920 | | |
3921 | | /* |
3922 | | * If we're parsing a default attribute value in DTD content, |
3923 | | * the entity might reference other entities which weren't |
3924 | | * defined yet, so the check isn't reliable. |
3925 | | */ |
3926 | 437 | if (ctxt->inSubset == 0) |
3927 | 396 | flags = XML_ENT_CHECKED | XML_ENT_VALIDATED; |
3928 | 41 | else |
3929 | 41 | flags = XML_ENT_VALIDATED; |
3930 | | |
3931 | 437 | str = pent->content; |
3932 | 437 | if (str == NULL) |
3933 | 8 | goto done; |
3934 | | |
3935 | | /* |
3936 | | * Note that entity values are already validated. We only check |
3937 | | * for illegal less-than signs and compute the expanded size |
3938 | | * of the entity. No special handling for multi-byte characters |
3939 | | * is needed. |
3940 | | */ |
3941 | 15.3k | while (!PARSER_STOPPED(ctxt)) { |
3942 | 15.3k | c = *str; |
3943 | | |
3944 | 15.3k | if (c != '&') { |
3945 | 11.2k | if (c == 0) |
3946 | 403 | break; |
3947 | | |
3948 | 10.8k | if (c == '<') |
3949 | 370 | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
3950 | 370 | "'<' in entity '%s' is not allowed in attributes " |
3951 | 370 | "values\n", pent->name); |
3952 | | |
3953 | 10.8k | str += 1; |
3954 | 10.8k | } else if (str[1] == '#') { |
3955 | 386 | int val; |
3956 | | |
3957 | 386 | val = xmlParseStringCharRef(ctxt, &str); |
3958 | 386 | if (val == 0) { |
3959 | 4 | pent->content[0] = 0; |
3960 | 4 | break; |
3961 | 4 | } |
3962 | 3.74k | } else { |
3963 | 3.74k | xmlChar *name; |
3964 | 3.74k | xmlEntityPtr ent; |
3965 | | |
3966 | 3.74k | name = xmlParseStringEntityRef(ctxt, &str); |
3967 | 3.74k | if (name == NULL) { |
3968 | 5 | pent->content[0] = 0; |
3969 | 5 | break; |
3970 | 5 | } |
3971 | | |
3972 | 3.73k | ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1); |
3973 | 3.73k | xmlFree(name); |
3974 | | |
3975 | 3.73k | if ((ent != NULL) && |
3976 | 3.73k | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) { |
3977 | 2.27k | if ((ent->flags & flags) != flags) { |
3978 | 92 | pent->flags |= XML_ENT_EXPANDING; |
3979 | 92 | xmlCheckEntityInAttValue(ctxt, ent, depth); |
3980 | 92 | pent->flags &= ~XML_ENT_EXPANDING; |
3981 | 92 | } |
3982 | | |
3983 | 2.27k | xmlSaturatedAdd(&expandedSize, ent->expandedSize); |
3984 | 2.27k | xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST); |
3985 | 2.27k | } |
3986 | 3.73k | } |
3987 | 15.3k | } |
3988 | | |
3989 | 437 | done: |
3990 | 437 | if (ctxt->inSubset == 0) |
3991 | 396 | pent->expandedSize = expandedSize; |
3992 | | |
3993 | 437 | pent->flags |= flags; |
3994 | 437 | } |
3995 | | |
3996 | | /** |
3997 | | * Expand general entity references in an entity or attribute value. |
3998 | | * Perform attribute value normalization. |
3999 | | * |
4000 | | * @param ctxt parser context |
4001 | | * @param buf string buffer |
4002 | | * @param str entity or attribute value |
4003 | | * @param pent entity for entity value, NULL for attribute values |
4004 | | * @param normalize whether to collapse whitespace |
4005 | | * @param inSpace whitespace state |
4006 | | * @param depth nesting depth |
4007 | | * @param check whether to check for amplification |
4008 | | */ |
4009 | | static void |
4010 | | xmlExpandEntityInAttValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf, |
4011 | | const xmlChar *str, xmlEntityPtr pent, int normalize, |
4012 | 565k | int *inSpace, int depth, int check) { |
4013 | 565k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
4014 | 565k | int c, chunkSize; |
4015 | | |
4016 | 565k | if (str == NULL) |
4017 | 66 | return; |
4018 | | |
4019 | 565k | depth += 1; |
4020 | 565k | if (depth > maxDepth) { |
4021 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
4022 | 0 | "Maximum entity nesting depth exceeded"); |
4023 | 0 | return; |
4024 | 0 | } |
4025 | | |
4026 | 565k | if (pent != NULL) { |
4027 | 535k | if (pent->flags & XML_ENT_EXPANDING) { |
4028 | 4 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
4029 | 4 | xmlHaltParser(ctxt); |
4030 | 4 | return; |
4031 | 4 | } |
4032 | | |
4033 | 535k | if (check) { |
4034 | 515k | if (xmlParserEntityCheck(ctxt, pent->length)) |
4035 | 5 | return; |
4036 | 515k | } |
4037 | 535k | } |
4038 | | |
4039 | 565k | chunkSize = 0; |
4040 | | |
4041 | | /* |
4042 | | * Note that entity values are already validated. No special |
4043 | | * handling for multi-byte characters is needed. |
4044 | | */ |
4045 | 75.6M | while (!PARSER_STOPPED(ctxt)) { |
4046 | 75.6M | c = *str; |
4047 | | |
4048 | 75.6M | if (c != '&') { |
4049 | 74.9M | if (c == 0) |
4050 | 562k | break; |
4051 | | |
4052 | | /* |
4053 | | * If this function is called without an entity, it is used to |
4054 | | * expand entities in an attribute content where less-than was |
4055 | | * already unscaped and is allowed. |
4056 | | */ |
4057 | 74.3M | if ((pent != NULL) && (c == '<')) { |
4058 | 2.62k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
4059 | 2.62k | "'<' in entity '%s' is not allowed in attributes " |
4060 | 2.62k | "values\n", pent->name); |
4061 | 2.62k | break; |
4062 | 2.62k | } |
4063 | | |
4064 | 74.3M | if (c <= 0x20) { |
4065 | 764k | if ((normalize) && (*inSpace)) { |
4066 | | /* Skip char */ |
4067 | 6.66k | if (chunkSize > 0) { |
4068 | 1.91k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4069 | 1.91k | chunkSize = 0; |
4070 | 1.91k | } |
4071 | 757k | } else if (c < 0x20) { |
4072 | 353k | if (chunkSize > 0) { |
4073 | 201k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4074 | 201k | chunkSize = 0; |
4075 | 201k | } |
4076 | | |
4077 | 353k | xmlSBufAddCString(buf, " ", 1); |
4078 | 404k | } else { |
4079 | 404k | chunkSize += 1; |
4080 | 404k | } |
4081 | | |
4082 | 764k | *inSpace = 1; |
4083 | 73.6M | } else { |
4084 | 73.6M | chunkSize += 1; |
4085 | 73.6M | *inSpace = 0; |
4086 | 73.6M | } |
4087 | | |
4088 | 74.3M | str += 1; |
4089 | 74.3M | } else if (str[1] == '#') { |
4090 | 106k | int val; |
4091 | | |
4092 | 106k | if (chunkSize > 0) { |
4093 | 105k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4094 | 105k | chunkSize = 0; |
4095 | 105k | } |
4096 | | |
4097 | 106k | val = xmlParseStringCharRef(ctxt, &str); |
4098 | 106k | if (val == 0) { |
4099 | 4 | if (pent != NULL) |
4100 | 4 | pent->content[0] = 0; |
4101 | 4 | break; |
4102 | 4 | } |
4103 | | |
4104 | 106k | if (val == ' ') { |
4105 | 89.6k | if ((!normalize) || (!*inSpace)) |
4106 | 89.4k | xmlSBufAddCString(buf, " ", 1); |
4107 | 89.6k | *inSpace = 1; |
4108 | 89.6k | } else { |
4109 | 17.0k | xmlSBufAddChar(buf, val); |
4110 | 17.0k | *inSpace = 0; |
4111 | 17.0k | } |
4112 | 579k | } else { |
4113 | 579k | xmlChar *name; |
4114 | 579k | xmlEntityPtr ent; |
4115 | | |
4116 | 579k | if (chunkSize > 0) { |
4117 | 232k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4118 | 232k | chunkSize = 0; |
4119 | 232k | } |
4120 | | |
4121 | 579k | name = xmlParseStringEntityRef(ctxt, &str); |
4122 | 579k | if (name == NULL) { |
4123 | 4 | if (pent != NULL) |
4124 | 4 | pent->content[0] = 0; |
4125 | 4 | break; |
4126 | 4 | } |
4127 | | |
4128 | 579k | ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1); |
4129 | 579k | xmlFree(name); |
4130 | | |
4131 | 579k | if ((ent != NULL) && |
4132 | 579k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
4133 | 35.4k | if (ent->content == NULL) { |
4134 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
4135 | 0 | "predefined entity has no content\n"); |
4136 | 0 | break; |
4137 | 0 | } |
4138 | | |
4139 | 35.4k | xmlSBufAddString(buf, ent->content, ent->length); |
4140 | | |
4141 | 35.4k | *inSpace = 0; |
4142 | 543k | } else if ((ent != NULL) && (ent->content != NULL)) { |
4143 | 513k | if (pent != NULL) |
4144 | 512k | pent->flags |= XML_ENT_EXPANDING; |
4145 | 513k | xmlExpandEntityInAttValue(ctxt, buf, ent->content, ent, |
4146 | 513k | normalize, inSpace, depth, check); |
4147 | 513k | if (pent != NULL) |
4148 | 512k | pent->flags &= ~XML_ENT_EXPANDING; |
4149 | 513k | } |
4150 | 579k | } |
4151 | 75.6M | } |
4152 | | |
4153 | 565k | if (chunkSize > 0) |
4154 | 551k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
4155 | 565k | } |
4156 | | |
4157 | | /** |
4158 | | * Expand general entity references in an entity or attribute value. |
4159 | | * Perform attribute value normalization. |
4160 | | * |
4161 | | * @param ctxt parser context |
4162 | | * @param str entity or attribute value |
4163 | | * @param normalize whether to collapse whitespace |
4164 | | * @returns the expanded attribtue value. |
4165 | | */ |
4166 | | xmlChar * |
4167 | | xmlExpandEntitiesInAttValue(xmlParserCtxtPtr ctxt, const xmlChar *str, |
4168 | 29.2k | int normalize) { |
4169 | 29.2k | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4170 | 0 | XML_MAX_HUGE_LENGTH : |
4171 | 29.2k | XML_MAX_TEXT_LENGTH; |
4172 | 29.2k | xmlSBuf buf; |
4173 | 29.2k | int inSpace = 1; |
4174 | | |
4175 | 29.2k | xmlSBufInit(&buf, maxLength); |
4176 | | |
4177 | 29.2k | xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace, |
4178 | 29.2k | ctxt->inputNr, /* check */ 0); |
4179 | | |
4180 | 29.2k | if ((normalize) && (inSpace) && (buf.size > 0)) |
4181 | 0 | buf.size--; |
4182 | | |
4183 | 29.2k | return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long")); |
4184 | 29.2k | } |
4185 | | |
4186 | | /** |
4187 | | * parse a value for an attribute. |
4188 | | * |
4189 | | * NOTE: if no normalization is needed, the routine will return pointers |
4190 | | * directly from the data buffer. |
4191 | | * |
4192 | | * 3.3.3 Attribute-Value Normalization: |
4193 | | * |
4194 | | * Before the value of an attribute is passed to the application or |
4195 | | * checked for validity, the XML processor must normalize it as follows: |
4196 | | * |
4197 | | * - a character reference is processed by appending the referenced |
4198 | | * character to the attribute value |
4199 | | * - an entity reference is processed by recursively processing the |
4200 | | * replacement text of the entity |
4201 | | * - a whitespace character (\#x20, \#xD, \#xA, \#x9) is processed by |
4202 | | * appending \#x20 to the normalized value, except that only a single |
4203 | | * \#x20 is appended for a "#xD#xA" sequence that is part of an external |
4204 | | * parsed entity or the literal entity value of an internal parsed entity |
4205 | | * - other characters are processed by appending them to the normalized value |
4206 | | * |
4207 | | * If the declared value is not CDATA, then the XML processor must further |
4208 | | * process the normalized attribute value by discarding any leading and |
4209 | | * trailing space (\#x20) characters, and by replacing sequences of space |
4210 | | * (\#x20) characters by a single space (\#x20) character. |
4211 | | * All attributes for which no declaration has been read should be treated |
4212 | | * by a non-validating parser as if declared CDATA. |
4213 | | * |
4214 | | * @param ctxt an XML parser context |
4215 | | * @param attlen attribute len result |
4216 | | * @param alloc whether the attribute was reallocated as a new string |
4217 | | * @param normalize if 1 then further non-CDATA normalization must be done |
4218 | | * @param isNamespace whether this is a namespace declaration |
4219 | | * @returns the AttValue parsed or NULL. The value has to be freed by the |
4220 | | * caller if it was copied, this can be detected by val[*len] == 0. |
4221 | | */ |
4222 | | static xmlChar * |
4223 | | xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *attlen, int *alloc, |
4224 | 70.7k | int normalize, int isNamespace) { |
4225 | 70.7k | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4226 | 0 | XML_MAX_HUGE_LENGTH : |
4227 | 70.7k | XML_MAX_TEXT_LENGTH; |
4228 | 70.7k | xmlSBuf buf; |
4229 | 70.7k | xmlChar *ret; |
4230 | 70.7k | int c, l, quote, flags, chunkSize; |
4231 | 70.7k | int inSpace = 1; |
4232 | 70.7k | int replaceEntities; |
4233 | | |
4234 | | /* Always expand namespace URIs */ |
4235 | 70.7k | replaceEntities = (ctxt->replaceEntities) || (isNamespace); |
4236 | | |
4237 | 70.7k | xmlSBufInit(&buf, maxLength); |
4238 | | |
4239 | 70.7k | GROW; |
4240 | | |
4241 | 70.7k | quote = CUR; |
4242 | 70.7k | if ((quote != '"') && (quote != '\'')) { |
4243 | 4.49k | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
4244 | 4.49k | return(NULL); |
4245 | 4.49k | } |
4246 | 66.2k | NEXTL(1); |
4247 | | |
4248 | 66.2k | if (ctxt->inSubset == 0) |
4249 | 32.7k | flags = XML_ENT_CHECKED | XML_ENT_VALIDATED; |
4250 | 33.4k | else |
4251 | 33.4k | flags = XML_ENT_VALIDATED; |
4252 | | |
4253 | 66.2k | inSpace = 1; |
4254 | 66.2k | chunkSize = 0; |
4255 | | |
4256 | 871k | while (1) { |
4257 | 871k | if (PARSER_STOPPED(ctxt)) |
4258 | 21 | goto error; |
4259 | | |
4260 | 871k | if (CUR_PTR >= ctxt->input->end) { |
4261 | 9.17k | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4262 | 9.17k | "AttValue: ' expected\n"); |
4263 | 9.17k | goto error; |
4264 | 9.17k | } |
4265 | | |
4266 | | /* |
4267 | | * TODO: Check growth threshold |
4268 | | */ |
4269 | 862k | if (ctxt->input->end - CUR_PTR < 10) |
4270 | 98.5k | GROW; |
4271 | | |
4272 | 862k | c = CUR; |
4273 | | |
4274 | 862k | if (c >= 0x80) { |
4275 | 71.2k | l = xmlUTF8MultibyteLen(ctxt, CUR_PTR, |
4276 | 71.2k | "invalid character in attribute value\n"); |
4277 | 71.2k | if (l == 0) { |
4278 | 7.17k | if (chunkSize > 0) { |
4279 | 991 | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4280 | 991 | chunkSize = 0; |
4281 | 991 | } |
4282 | 7.17k | xmlSBufAddReplChar(&buf); |
4283 | 7.17k | NEXTL(1); |
4284 | 64.0k | } else { |
4285 | 64.0k | chunkSize += l; |
4286 | 64.0k | NEXTL(l); |
4287 | 64.0k | } |
4288 | | |
4289 | 71.2k | inSpace = 0; |
4290 | 791k | } else if (c != '&') { |
4291 | 710k | if (c > 0x20) { |
4292 | 509k | if (c == quote) |
4293 | 55.6k | break; |
4294 | | |
4295 | 453k | if (c == '<') |
4296 | 35.4k | xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL); |
4297 | | |
4298 | 453k | chunkSize += 1; |
4299 | 453k | inSpace = 0; |
4300 | 453k | } else if (!IS_BYTE_CHAR(c)) { |
4301 | 2.43k | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
4302 | 2.43k | "invalid character in attribute value\n"); |
4303 | 2.43k | if (chunkSize > 0) { |
4304 | 761 | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4305 | 761 | chunkSize = 0; |
4306 | 761 | } |
4307 | 2.43k | xmlSBufAddReplChar(&buf); |
4308 | 2.43k | inSpace = 0; |
4309 | 198k | } else { |
4310 | | /* Whitespace */ |
4311 | 198k | if ((normalize) && (inSpace)) { |
4312 | | /* Skip char */ |
4313 | 704 | if (chunkSize > 0) { |
4314 | 362 | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4315 | 362 | chunkSize = 0; |
4316 | 362 | } |
4317 | 197k | } else if (c < 0x20) { |
4318 | | /* Convert to space */ |
4319 | 4.19k | if (chunkSize > 0) { |
4320 | 2.37k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4321 | 2.37k | chunkSize = 0; |
4322 | 2.37k | } |
4323 | | |
4324 | 4.19k | xmlSBufAddCString(&buf, " ", 1); |
4325 | 193k | } else { |
4326 | 193k | chunkSize += 1; |
4327 | 193k | } |
4328 | | |
4329 | 198k | inSpace = 1; |
4330 | | |
4331 | 198k | if ((c == 0xD) && (NXT(1) == 0xA)) |
4332 | 194 | CUR_PTR++; |
4333 | 198k | } |
4334 | | |
4335 | 654k | NEXTL(1); |
4336 | 654k | } else if (NXT(1) == '#') { |
4337 | 7.93k | int val; |
4338 | | |
4339 | 7.93k | if (chunkSize > 0) { |
4340 | 3.68k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4341 | 3.68k | chunkSize = 0; |
4342 | 3.68k | } |
4343 | | |
4344 | 7.93k | val = xmlParseCharRef(ctxt); |
4345 | 7.93k | if (val == 0) |
4346 | 1.36k | goto error; |
4347 | | |
4348 | 6.56k | if ((val == '&') && (!replaceEntities)) { |
4349 | | /* |
4350 | | * The reparsing will be done in xmlNodeParseContent() |
4351 | | * called from SAX2.c |
4352 | | */ |
4353 | 1.69k | xmlSBufAddCString(&buf, "&", 5); |
4354 | 1.69k | inSpace = 0; |
4355 | 4.86k | } else if (val == ' ') { |
4356 | 1.20k | if ((!normalize) || (!inSpace)) |
4357 | 1.01k | xmlSBufAddCString(&buf, " ", 1); |
4358 | 1.20k | inSpace = 1; |
4359 | 3.65k | } else { |
4360 | 3.65k | xmlSBufAddChar(&buf, val); |
4361 | 3.65k | inSpace = 0; |
4362 | 3.65k | } |
4363 | 72.7k | } else { |
4364 | 72.7k | const xmlChar *name; |
4365 | 72.7k | xmlEntityPtr ent; |
4366 | | |
4367 | 72.7k | if (chunkSize > 0) { |
4368 | 22.0k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4369 | 22.0k | chunkSize = 0; |
4370 | 22.0k | } |
4371 | | |
4372 | 72.7k | name = xmlParseEntityRefInternal(ctxt); |
4373 | 72.7k | if (name == NULL) { |
4374 | | /* |
4375 | | * Probably a literal '&' which wasn't escaped. |
4376 | | * TODO: Handle gracefully in recovery mode. |
4377 | | */ |
4378 | 4.86k | continue; |
4379 | 4.86k | } |
4380 | | |
4381 | 67.9k | ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1); |
4382 | 67.9k | if (ent == NULL) |
4383 | 10.8k | continue; |
4384 | | |
4385 | 57.1k | if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) { |
4386 | 3.03k | if ((ent->content[0] == '&') && (!replaceEntities)) |
4387 | 1.56k | xmlSBufAddCString(&buf, "&", 5); |
4388 | 1.46k | else |
4389 | 1.46k | xmlSBufAddString(&buf, ent->content, ent->length); |
4390 | 3.03k | inSpace = 0; |
4391 | 54.0k | } else if (replaceEntities) { |
4392 | 22.3k | xmlExpandEntityInAttValue(ctxt, &buf, ent->content, ent, |
4393 | 22.3k | normalize, &inSpace, ctxt->inputNr, |
4394 | 22.3k | /* check */ 1); |
4395 | 31.7k | } else { |
4396 | 31.7k | if ((ent->flags & flags) != flags) |
4397 | 357 | xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr); |
4398 | | |
4399 | 31.7k | if (xmlParserEntityCheck(ctxt, ent->expandedSize)) { |
4400 | 8 | ent->content[0] = 0; |
4401 | 8 | goto error; |
4402 | 8 | } |
4403 | | |
4404 | | /* |
4405 | | * Just output the reference |
4406 | | */ |
4407 | 31.7k | xmlSBufAddCString(&buf, "&", 1); |
4408 | 31.7k | xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name)); |
4409 | 31.7k | xmlSBufAddCString(&buf, ";", 1); |
4410 | | |
4411 | 31.7k | inSpace = 0; |
4412 | 31.7k | } |
4413 | 57.1k | } |
4414 | 862k | } |
4415 | | |
4416 | 55.6k | if ((buf.mem == NULL) && (alloc != NULL)) { |
4417 | 20.6k | ret = (xmlChar *) CUR_PTR - chunkSize; |
4418 | | |
4419 | 20.6k | if (attlen != NULL) |
4420 | 20.6k | *attlen = chunkSize; |
4421 | 20.6k | if ((normalize) && (inSpace) && (chunkSize > 0)) |
4422 | 81 | *attlen -= 1; |
4423 | 20.6k | *alloc = 0; |
4424 | | |
4425 | | /* Report potential error */ |
4426 | 20.6k | xmlSBufCleanup(&buf, ctxt, "AttValue length too long"); |
4427 | 35.0k | } else { |
4428 | 35.0k | if (chunkSize > 0) |
4429 | 21.9k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4430 | | |
4431 | 35.0k | if ((normalize) && (inSpace) && (buf.size > 0)) |
4432 | 108 | buf.size--; |
4433 | | |
4434 | 35.0k | ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long"); |
4435 | | |
4436 | 35.0k | if (ret != NULL) { |
4437 | 35.0k | if (attlen != NULL) |
4438 | 2.93k | *attlen = buf.size; |
4439 | 35.0k | if (alloc != NULL) |
4440 | 2.93k | *alloc = 1; |
4441 | 35.0k | } |
4442 | 35.0k | } |
4443 | | |
4444 | 55.6k | NEXTL(1); |
4445 | | |
4446 | 55.6k | return(ret); |
4447 | | |
4448 | 10.5k | error: |
4449 | 10.5k | xmlSBufCleanup(&buf, ctxt, "AttValue length too long"); |
4450 | 10.5k | return(NULL); |
4451 | 66.2k | } |
4452 | | |
4453 | | /** |
4454 | | * parse a value for an attribute |
4455 | | * Note: the parser won't do substitution of entities here, this |
4456 | | * will be handled later in xmlStringGetNodeList() |
4457 | | * |
4458 | | * @deprecated Internal function, don't use. |
4459 | | * |
4460 | | * [10] AttValue ::= '"' ([^<&"] | Reference)* '"' | |
4461 | | * "'" ([^<&'] | Reference)* "'" |
4462 | | * |
4463 | | * 3.3.3 Attribute-Value Normalization: |
4464 | | * |
4465 | | * Before the value of an attribute is passed to the application or |
4466 | | * checked for validity, the XML processor must normalize it as follows: |
4467 | | * |
4468 | | * - a character reference is processed by appending the referenced |
4469 | | * character to the attribute value |
4470 | | * - an entity reference is processed by recursively processing the |
4471 | | * replacement text of the entity |
4472 | | * - a whitespace character (\#x20, \#xD, \#xA, \#x9) is processed by |
4473 | | * appending \#x20 to the normalized value, except that only a single |
4474 | | * \#x20 is appended for a "#xD#xA" sequence that is part of an external |
4475 | | * parsed entity or the literal entity value of an internal parsed entity |
4476 | | * - other characters are processed by appending them to the normalized value |
4477 | | * |
4478 | | * If the declared value is not CDATA, then the XML processor must further |
4479 | | * process the normalized attribute value by discarding any leading and |
4480 | | * trailing space (\#x20) characters, and by replacing sequences of space |
4481 | | * (\#x20) characters by a single space (\#x20) character. |
4482 | | * All attributes for which no declaration has been read should be treated |
4483 | | * by a non-validating parser as if declared CDATA. |
4484 | | * |
4485 | | * @param ctxt an XML parser context |
4486 | | * @returns the AttValue parsed or NULL. The value has to be freed by the |
4487 | | * caller. |
4488 | | */ |
4489 | | xmlChar * |
4490 | 45.3k | xmlParseAttValue(xmlParserCtxtPtr ctxt) { |
4491 | 45.3k | if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL); |
4492 | 45.3k | return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0)); |
4493 | 45.3k | } |
4494 | | |
4495 | | /** |
4496 | | * parse an XML Literal |
4497 | | * |
4498 | | * @deprecated Internal function, don't use. |
4499 | | * |
4500 | | * [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") |
4501 | | * |
4502 | | * @param ctxt an XML parser context |
4503 | | * @returns the SystemLiteral parsed or NULL |
4504 | | */ |
4505 | | |
4506 | | xmlChar * |
4507 | 4.58k | xmlParseSystemLiteral(xmlParserCtxtPtr ctxt) { |
4508 | 4.58k | xmlChar *buf = NULL; |
4509 | 4.58k | int len = 0; |
4510 | 4.58k | int size = XML_PARSER_BUFFER_SIZE; |
4511 | 4.58k | int cur, l; |
4512 | 4.58k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4513 | 0 | XML_MAX_TEXT_LENGTH : |
4514 | 4.58k | XML_MAX_NAME_LENGTH; |
4515 | 4.58k | xmlChar stop; |
4516 | | |
4517 | 4.58k | if (RAW == '"') { |
4518 | 2.28k | NEXT; |
4519 | 2.28k | stop = '"'; |
4520 | 2.29k | } else if (RAW == '\'') { |
4521 | 465 | NEXT; |
4522 | 465 | stop = '\''; |
4523 | 1.83k | } else { |
4524 | 1.83k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4525 | 1.83k | return(NULL); |
4526 | 1.83k | } |
4527 | | |
4528 | 2.75k | buf = xmlMalloc(size); |
4529 | 2.75k | if (buf == NULL) { |
4530 | 0 | xmlErrMemory(ctxt); |
4531 | 0 | return(NULL); |
4532 | 0 | } |
4533 | 2.75k | cur = xmlCurrentCharRecover(ctxt, &l); |
4534 | 26.6k | while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */ |
4535 | 23.9k | if (len + 5 >= size) { |
4536 | 227 | xmlChar *tmp; |
4537 | 227 | int newSize; |
4538 | | |
4539 | 227 | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
4540 | 227 | if (newSize < 0) { |
4541 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral"); |
4542 | 0 | xmlFree(buf); |
4543 | 0 | return(NULL); |
4544 | 0 | } |
4545 | 227 | tmp = xmlRealloc(buf, newSize); |
4546 | 227 | if (tmp == NULL) { |
4547 | 0 | xmlFree(buf); |
4548 | 0 | xmlErrMemory(ctxt); |
4549 | 0 | return(NULL); |
4550 | 0 | } |
4551 | 227 | buf = tmp; |
4552 | 227 | size = newSize; |
4553 | 227 | } |
4554 | 23.9k | COPY_BUF(buf, len, cur); |
4555 | 23.9k | NEXTL(l); |
4556 | 23.9k | cur = xmlCurrentCharRecover(ctxt, &l); |
4557 | 23.9k | } |
4558 | 2.75k | buf[len] = 0; |
4559 | 2.75k | if (!IS_CHAR(cur)) { |
4560 | 1.13k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4561 | 1.62k | } else { |
4562 | 1.62k | NEXT; |
4563 | 1.62k | } |
4564 | 2.75k | return(buf); |
4565 | 2.75k | } |
4566 | | |
4567 | | /** |
4568 | | * parse an XML public literal |
4569 | | * |
4570 | | * @deprecated Internal function, don't use. |
4571 | | * |
4572 | | * [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'" |
4573 | | * |
4574 | | * @param ctxt an XML parser context |
4575 | | * @returns the PubidLiteral parsed or NULL. |
4576 | | */ |
4577 | | |
4578 | | xmlChar * |
4579 | 3.96k | xmlParsePubidLiteral(xmlParserCtxtPtr ctxt) { |
4580 | 3.96k | xmlChar *buf = NULL; |
4581 | 3.96k | int len = 0; |
4582 | 3.96k | int size = XML_PARSER_BUFFER_SIZE; |
4583 | 3.96k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4584 | 0 | XML_MAX_TEXT_LENGTH : |
4585 | 3.96k | XML_MAX_NAME_LENGTH; |
4586 | 3.96k | xmlChar cur; |
4587 | 3.96k | xmlChar stop; |
4588 | | |
4589 | 3.96k | if (RAW == '"') { |
4590 | 2.27k | NEXT; |
4591 | 2.27k | stop = '"'; |
4592 | 2.27k | } else if (RAW == '\'') { |
4593 | 1.12k | NEXT; |
4594 | 1.12k | stop = '\''; |
4595 | 1.12k | } else { |
4596 | 568 | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4597 | 568 | return(NULL); |
4598 | 568 | } |
4599 | 3.40k | buf = xmlMalloc(size); |
4600 | 3.40k | if (buf == NULL) { |
4601 | 0 | xmlErrMemory(ctxt); |
4602 | 0 | return(NULL); |
4603 | 0 | } |
4604 | 3.40k | cur = CUR; |
4605 | 46.8k | while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) && |
4606 | 46.8k | (PARSER_STOPPED(ctxt) == 0)) { /* checked */ |
4607 | 43.6k | if (len + 1 >= size) { |
4608 | 196 | xmlChar *tmp; |
4609 | 196 | int newSize; |
4610 | | |
4611 | 196 | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
4612 | 196 | if (newSize) { |
4613 | 196 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID"); |
4614 | 196 | xmlFree(buf); |
4615 | 196 | return(NULL); |
4616 | 196 | } |
4617 | 0 | tmp = xmlRealloc(buf, size); |
4618 | 0 | if (tmp == NULL) { |
4619 | 0 | xmlErrMemory(ctxt); |
4620 | 0 | xmlFree(buf); |
4621 | 0 | return(NULL); |
4622 | 0 | } |
4623 | 0 | buf = tmp; |
4624 | 0 | size = newSize; |
4625 | 0 | } |
4626 | 43.4k | buf[len++] = cur; |
4627 | 43.4k | NEXT; |
4628 | 43.4k | cur = CUR; |
4629 | 43.4k | } |
4630 | 3.20k | buf[len] = 0; |
4631 | 3.20k | if (cur != stop) { |
4632 | 2.52k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4633 | 2.52k | } else { |
4634 | 680 | NEXTL(1); |
4635 | 680 | } |
4636 | 3.20k | return(buf); |
4637 | 3.40k | } |
4638 | | |
4639 | | static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial); |
4640 | | |
4641 | | /* |
4642 | | * used for the test in the inner loop of the char data testing |
4643 | | */ |
4644 | | static const unsigned char test_char_data[256] = { |
4645 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4646 | | 0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */ |
4647 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4648 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4649 | | 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */ |
4650 | | 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F, |
4651 | | 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, |
4652 | | 0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */ |
4653 | | 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, |
4654 | | 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F, |
4655 | | 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, |
4656 | | 0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */ |
4657 | | 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, |
4658 | | 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, |
4659 | | 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, |
4660 | | 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F, |
4661 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */ |
4662 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4663 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4664 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4665 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4666 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4667 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4668 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4669 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4670 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4671 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4672 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4673 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4674 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4675 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4676 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 |
4677 | | }; |
4678 | | |
4679 | | static void |
4680 | | xmlCharacters(xmlParserCtxtPtr ctxt, const xmlChar *buf, int size, |
4681 | 45.7k | int isBlank) { |
4682 | 45.7k | int checkBlanks; |
4683 | | |
4684 | 45.7k | if ((ctxt->sax == NULL) || (ctxt->disableSAX)) |
4685 | 19.9k | return; |
4686 | | |
4687 | 25.7k | checkBlanks = (!ctxt->keepBlanks) || |
4688 | 25.7k | (ctxt->sax->ignorableWhitespace != ctxt->sax->characters); |
4689 | | |
4690 | | /* |
4691 | | * Calling areBlanks with only parts of a text node |
4692 | | * is fundamentally broken, making the NOBLANKS option |
4693 | | * essentially unusable. |
4694 | | */ |
4695 | 25.7k | if ((checkBlanks) && |
4696 | 25.7k | (areBlanks(ctxt, buf, size, isBlank))) { |
4697 | 2.45k | if ((ctxt->sax->ignorableWhitespace != NULL) && |
4698 | 2.45k | (ctxt->keepBlanks)) |
4699 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size); |
4700 | 23.3k | } else { |
4701 | 23.3k | if (ctxt->sax->characters != NULL) |
4702 | 23.3k | ctxt->sax->characters(ctxt->userData, buf, size); |
4703 | | |
4704 | | /* |
4705 | | * The old code used to update this value for "complex" data |
4706 | | * even if checkBlanks was false. This was probably a bug. |
4707 | | */ |
4708 | 23.3k | if ((checkBlanks) && (*ctxt->space == -1)) |
4709 | 3.57k | *ctxt->space = -2; |
4710 | 23.3k | } |
4711 | 25.7k | } |
4712 | | |
4713 | | /** |
4714 | | * Parse character data. Always makes progress if the first char isn't |
4715 | | * '<' or '&'. |
4716 | | * |
4717 | | * The right angle bracket (>) may be represented using the string ">", |
4718 | | * and must, for compatibility, be escaped using ">" or a character |
4719 | | * reference when it appears in the string "]]>" in content, when that |
4720 | | * string is not marking the end of a CDATA section. |
4721 | | * |
4722 | | * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) |
4723 | | * @param ctxt an XML parser context |
4724 | | * @param partial buffer may contain partial UTF-8 sequences |
4725 | | */ |
4726 | | static void |
4727 | 49.1k | xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) { |
4728 | 49.1k | const xmlChar *in; |
4729 | 49.1k | int nbchar = 0; |
4730 | 49.1k | int line = ctxt->input->line; |
4731 | 49.1k | int col = ctxt->input->col; |
4732 | 49.1k | int ccol; |
4733 | | |
4734 | 49.1k | GROW; |
4735 | | /* |
4736 | | * Accelerated common case where input don't need to be |
4737 | | * modified before passing it to the handler. |
4738 | | */ |
4739 | 49.1k | in = ctxt->input->cur; |
4740 | 49.5k | do { |
4741 | 49.9k | get_more_space: |
4742 | 59.8k | while (*in == 0x20) { in++; ctxt->input->col++; } |
4743 | 49.9k | if (*in == 0xA) { |
4744 | 716 | do { |
4745 | 716 | ctxt->input->line++; ctxt->input->col = 1; |
4746 | 716 | in++; |
4747 | 716 | } while (*in == 0xA); |
4748 | 325 | goto get_more_space; |
4749 | 325 | } |
4750 | 49.5k | if (*in == '<') { |
4751 | 5.45k | nbchar = in - ctxt->input->cur; |
4752 | 5.45k | if (nbchar > 0) { |
4753 | 5.45k | const xmlChar *tmp = ctxt->input->cur; |
4754 | 5.45k | ctxt->input->cur = in; |
4755 | | |
4756 | 5.45k | xmlCharacters(ctxt, tmp, nbchar, 1); |
4757 | 5.45k | } |
4758 | 5.45k | return; |
4759 | 5.45k | } |
4760 | | |
4761 | 46.9k | get_more: |
4762 | 46.9k | ccol = ctxt->input->col; |
4763 | 130k | while (test_char_data[*in]) { |
4764 | 83.5k | in++; |
4765 | 83.5k | ccol++; |
4766 | 83.5k | } |
4767 | 46.9k | ctxt->input->col = ccol; |
4768 | 46.9k | if (*in == 0xA) { |
4769 | 504 | do { |
4770 | 504 | ctxt->input->line++; ctxt->input->col = 1; |
4771 | 504 | in++; |
4772 | 504 | } while (*in == 0xA); |
4773 | 310 | goto get_more; |
4774 | 310 | } |
4775 | 46.6k | if (*in == ']') { |
4776 | 2.77k | if ((in[1] == ']') && (in[2] == '>')) { |
4777 | 270 | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4778 | 270 | ctxt->input->cur = in + 1; |
4779 | 270 | return; |
4780 | 270 | } |
4781 | 2.50k | if ((!partial) || (ctxt->input->end - in >= 2)) { |
4782 | 2.50k | in++; |
4783 | 2.50k | ctxt->input->col++; |
4784 | 2.50k | goto get_more; |
4785 | 2.50k | } |
4786 | 2.50k | } |
4787 | 43.8k | nbchar = in - ctxt->input->cur; |
4788 | 43.8k | if (nbchar > 0) { |
4789 | 33.3k | const xmlChar *tmp = ctxt->input->cur; |
4790 | 33.3k | ctxt->input->cur = in; |
4791 | | |
4792 | 33.3k | xmlCharacters(ctxt, tmp, nbchar, 0); |
4793 | | |
4794 | 33.3k | line = ctxt->input->line; |
4795 | 33.3k | col = ctxt->input->col; |
4796 | 33.3k | } |
4797 | 43.8k | ctxt->input->cur = in; |
4798 | 43.8k | if (*in == 0xD) { |
4799 | 1.57k | in++; |
4800 | 1.57k | if (*in == 0xA) { |
4801 | 388 | ctxt->input->cur = in; |
4802 | 388 | in++; |
4803 | 388 | ctxt->input->line++; ctxt->input->col = 1; |
4804 | 388 | continue; /* while */ |
4805 | 388 | } |
4806 | 1.18k | in--; |
4807 | 1.18k | } |
4808 | 43.4k | if (*in == '<') { |
4809 | 26.7k | return; |
4810 | 26.7k | } |
4811 | 16.7k | if (*in == '&') { |
4812 | 3.80k | return; |
4813 | 3.80k | } |
4814 | 12.9k | if ((partial) && (*in == ']') && (ctxt->input->end - in < 2)) { |
4815 | 0 | return; |
4816 | 0 | } |
4817 | 12.9k | SHRINK; |
4818 | 12.9k | GROW; |
4819 | 12.9k | in = ctxt->input->cur; |
4820 | 13.2k | } while (((*in >= 0x20) && (*in <= 0x7F)) || |
4821 | 13.2k | (*in == 0x09) || (*in == 0x0a)); |
4822 | 12.9k | ctxt->input->line = line; |
4823 | 12.9k | ctxt->input->col = col; |
4824 | 12.9k | xmlParseCharDataComplex(ctxt, partial); |
4825 | 12.9k | } |
4826 | | |
4827 | | /** |
4828 | | * Always makes progress if the first char isn't '<' or '&'. |
4829 | | * |
4830 | | * parse a CharData section.this is the fallback function |
4831 | | * of xmlParseCharData() when the parsing requires handling |
4832 | | * of non-ASCII characters. |
4833 | | * |
4834 | | * @param ctxt an XML parser context |
4835 | | * @param partial whether the input can end with truncated UTF-8 |
4836 | | */ |
4837 | | static void |
4838 | 12.9k | xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) { |
4839 | 12.9k | xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5]; |
4840 | 12.9k | int nbchar = 0; |
4841 | 12.9k | int cur, l; |
4842 | | |
4843 | 12.9k | cur = xmlCurrentCharRecover(ctxt, &l); |
4844 | 52.2k | while ((cur != '<') && /* checked */ |
4845 | 52.2k | (cur != '&') && |
4846 | 52.2k | ((!partial) || (cur != ']') || |
4847 | 45.7k | (ctxt->input->end - ctxt->input->cur >= 2)) && |
4848 | 52.2k | (IS_CHAR(cur))) { |
4849 | 39.3k | if ((cur == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
4850 | 313 | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4851 | 313 | } |
4852 | 39.3k | COPY_BUF(buf, nbchar, cur); |
4853 | | /* move current position before possible calling of ctxt->sax->characters */ |
4854 | 39.3k | NEXTL(l); |
4855 | 39.3k | if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) { |
4856 | 181 | buf[nbchar] = 0; |
4857 | | |
4858 | 181 | xmlCharacters(ctxt, buf, nbchar, 0); |
4859 | 181 | nbchar = 0; |
4860 | 181 | SHRINK; |
4861 | 181 | } |
4862 | 39.3k | cur = xmlCurrentCharRecover(ctxt, &l); |
4863 | 39.3k | } |
4864 | 12.9k | if (nbchar != 0) { |
4865 | 6.79k | buf[nbchar] = 0; |
4866 | | |
4867 | 6.79k | xmlCharacters(ctxt, buf, nbchar, 0); |
4868 | 6.79k | } |
4869 | | /* |
4870 | | * cur == 0 can mean |
4871 | | * |
4872 | | * - End of buffer. |
4873 | | * - An actual 0 character. |
4874 | | * - An incomplete UTF-8 sequence. This is allowed if partial is set. |
4875 | | */ |
4876 | 12.9k | if (ctxt->input->cur < ctxt->input->end) { |
4877 | 11.5k | if ((cur == 0) && (CUR != 0)) { |
4878 | 21 | if (partial == 0) { |
4879 | 21 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4880 | 21 | "Incomplete UTF-8 sequence starting with %02X\n", CUR); |
4881 | 21 | NEXTL(1); |
4882 | 21 | } |
4883 | 11.5k | } else if ((cur != '<') && (cur != '&') && (cur != ']')) { |
4884 | | /* Generate the error and skip the offending character */ |
4885 | 4.98k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4886 | 4.98k | "PCDATA invalid Char value %d\n", cur); |
4887 | 4.98k | NEXTL(l); |
4888 | 4.98k | } |
4889 | 11.5k | } |
4890 | 12.9k | } |
4891 | | |
4892 | | /** |
4893 | | * @deprecated Internal function, don't use. |
4894 | | * @param ctxt an XML parser context |
4895 | | * @param cdata unused |
4896 | | */ |
4897 | | void |
4898 | 0 | xmlParseCharData(xmlParserCtxtPtr ctxt, ATTRIBUTE_UNUSED int cdata) { |
4899 | 0 | xmlParseCharDataInternal(ctxt, 0); |
4900 | 0 | } |
4901 | | |
4902 | | /** |
4903 | | * Parse an External ID or a Public ID |
4904 | | * |
4905 | | * @deprecated Internal function, don't use. |
4906 | | * |
4907 | | * NOTE: Productions [75] and [83] interact badly since [75] can generate |
4908 | | * `'PUBLIC' S PubidLiteral S SystemLiteral` |
4909 | | * |
4910 | | * [75] ExternalID ::= 'SYSTEM' S SystemLiteral |
4911 | | * | 'PUBLIC' S PubidLiteral S SystemLiteral |
4912 | | * |
4913 | | * [83] PublicID ::= 'PUBLIC' S PubidLiteral |
4914 | | * |
4915 | | * @param ctxt an XML parser context |
4916 | | * @param publicID a xmlChar** receiving PubidLiteral |
4917 | | * @param strict indicate whether we should restrict parsing to only |
4918 | | * production [75], see NOTE below |
4919 | | * @returns the function returns SystemLiteral and in the second |
4920 | | * case publicID receives PubidLiteral, is strict is off |
4921 | | * it is possible to return NULL and have publicID set. |
4922 | | */ |
4923 | | |
4924 | | xmlChar * |
4925 | 18.7k | xmlParseExternalID(xmlParserCtxtPtr ctxt, xmlChar **publicID, int strict) { |
4926 | 18.7k | xmlChar *URI = NULL; |
4927 | | |
4928 | 18.7k | *publicID = NULL; |
4929 | 18.7k | if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) { |
4930 | 2.19k | SKIP(6); |
4931 | 2.19k | if (SKIP_BLANKS == 0) { |
4932 | 1.89k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4933 | 1.89k | "Space required after 'SYSTEM'\n"); |
4934 | 1.89k | } |
4935 | 2.19k | URI = xmlParseSystemLiteral(ctxt); |
4936 | 2.19k | if (URI == NULL) { |
4937 | 397 | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4938 | 397 | } |
4939 | 16.5k | } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) { |
4940 | 3.96k | SKIP(6); |
4941 | 3.96k | if (SKIP_BLANKS == 0) { |
4942 | 3.39k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4943 | 3.39k | "Space required after 'PUBLIC'\n"); |
4944 | 3.39k | } |
4945 | 3.96k | *publicID = xmlParsePubidLiteral(ctxt); |
4946 | 3.96k | if (*publicID == NULL) { |
4947 | 764 | xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL); |
4948 | 764 | } |
4949 | 3.96k | if (strict) { |
4950 | | /* |
4951 | | * We don't handle [83] so "S SystemLiteral" is required. |
4952 | | */ |
4953 | 1.79k | if (SKIP_BLANKS == 0) { |
4954 | 1.41k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4955 | 1.41k | "Space required after the Public Identifier\n"); |
4956 | 1.41k | } |
4957 | 2.17k | } else { |
4958 | | /* |
4959 | | * We handle [83] so we return immediately, if |
4960 | | * "S SystemLiteral" is not detected. We skip blanks if no |
4961 | | * system literal was found, but this is harmless since we must |
4962 | | * be at the end of a NotationDecl. |
4963 | | */ |
4964 | 2.17k | if (SKIP_BLANKS == 0) return(NULL); |
4965 | 992 | if ((CUR != '\'') && (CUR != '"')) return(NULL); |
4966 | 992 | } |
4967 | 2.39k | URI = xmlParseSystemLiteral(ctxt); |
4968 | 2.39k | if (URI == NULL) { |
4969 | 1.43k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4970 | 1.43k | } |
4971 | 2.39k | } |
4972 | 17.1k | return(URI); |
4973 | 18.7k | } |
4974 | | |
4975 | | /** |
4976 | | * Skip an XML (SGML) comment <!-- .... --> |
4977 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4978 | | * must not occur within comments. " |
4979 | | * This is the slow routine in case the accelerator for ascii didn't work |
4980 | | * |
4981 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4982 | | * @param ctxt an XML parser context |
4983 | | * @param buf the already parsed part of the buffer |
4984 | | * @param len number of bytes in the buffer |
4985 | | * @param size allocated size of the buffer |
4986 | | */ |
4987 | | static void |
4988 | | xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf, |
4989 | 8.71k | size_t len, size_t size) { |
4990 | 8.71k | int q, ql; |
4991 | 8.71k | int r, rl; |
4992 | 8.71k | int cur, l; |
4993 | 8.71k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4994 | 0 | XML_MAX_HUGE_LENGTH : |
4995 | 8.71k | XML_MAX_TEXT_LENGTH; |
4996 | | |
4997 | 8.71k | if (buf == NULL) { |
4998 | 4.66k | len = 0; |
4999 | 4.66k | size = XML_PARSER_BUFFER_SIZE; |
5000 | 4.66k | buf = xmlMalloc(size); |
5001 | 4.66k | if (buf == NULL) { |
5002 | 0 | xmlErrMemory(ctxt); |
5003 | 0 | return; |
5004 | 0 | } |
5005 | 4.66k | } |
5006 | 8.71k | q = xmlCurrentCharRecover(ctxt, &ql); |
5007 | 8.71k | if (q == 0) |
5008 | 3.66k | goto not_terminated; |
5009 | 5.04k | if (!IS_CHAR(q)) { |
5010 | 270 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5011 | 270 | "xmlParseComment: invalid xmlChar value %d\n", |
5012 | 270 | q); |
5013 | 270 | xmlFree (buf); |
5014 | 270 | return; |
5015 | 270 | } |
5016 | 4.77k | NEXTL(ql); |
5017 | 4.77k | r = xmlCurrentCharRecover(ctxt, &rl); |
5018 | 4.77k | if (r == 0) |
5019 | 418 | goto not_terminated; |
5020 | 4.36k | if (!IS_CHAR(r)) { |
5021 | 135 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5022 | 135 | "xmlParseComment: invalid xmlChar value %d\n", |
5023 | 135 | r); |
5024 | 135 | xmlFree (buf); |
5025 | 135 | return; |
5026 | 135 | } |
5027 | 4.22k | NEXTL(rl); |
5028 | 4.22k | cur = xmlCurrentCharRecover(ctxt, &l); |
5029 | 4.22k | if (cur == 0) |
5030 | 230 | goto not_terminated; |
5031 | 96.6k | while (IS_CHAR(cur) && /* checked */ |
5032 | 96.6k | ((cur != '>') || |
5033 | 95.5k | (r != '-') || (q != '-'))) { |
5034 | 92.6k | if ((r == '-') && (q == '-')) { |
5035 | 441 | xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL); |
5036 | 441 | } |
5037 | 92.6k | if (len + 5 >= size) { |
5038 | 1.35k | xmlChar *tmp; |
5039 | 1.35k | int newSize; |
5040 | | |
5041 | 1.35k | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
5042 | 1.35k | if (newSize < 0) { |
5043 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5044 | 0 | "Comment too big found", NULL); |
5045 | 0 | xmlFree (buf); |
5046 | 0 | return; |
5047 | 0 | } |
5048 | 1.35k | tmp = xmlRealloc(buf, newSize); |
5049 | 1.35k | if (tmp == NULL) { |
5050 | 0 | xmlErrMemory(ctxt); |
5051 | 0 | xmlFree(buf); |
5052 | 0 | return; |
5053 | 0 | } |
5054 | 1.35k | buf = tmp; |
5055 | 1.35k | size = newSize; |
5056 | 1.35k | } |
5057 | 92.6k | COPY_BUF(buf, len, q); |
5058 | | |
5059 | 92.6k | q = r; |
5060 | 92.6k | ql = rl; |
5061 | 92.6k | r = cur; |
5062 | 92.6k | rl = l; |
5063 | | |
5064 | 92.6k | NEXTL(l); |
5065 | 92.6k | cur = xmlCurrentCharRecover(ctxt, &l); |
5066 | | |
5067 | 92.6k | } |
5068 | 3.99k | buf[len] = 0; |
5069 | 3.99k | if (cur == 0) { |
5070 | 930 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5071 | 930 | "Comment not terminated \n<!--%.50s\n", buf); |
5072 | 3.06k | } else if (!IS_CHAR(cur)) { |
5073 | 138 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
5074 | 138 | "xmlParseComment: invalid xmlChar value %d\n", |
5075 | 138 | cur); |
5076 | 2.92k | } else { |
5077 | 2.92k | NEXT; |
5078 | 2.92k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5079 | 2.92k | (!ctxt->disableSAX)) |
5080 | 2.72k | ctxt->sax->comment(ctxt->userData, buf); |
5081 | 2.92k | } |
5082 | 3.99k | xmlFree(buf); |
5083 | 3.99k | return; |
5084 | 4.31k | not_terminated: |
5085 | 4.31k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5086 | 4.31k | "Comment not terminated\n", NULL); |
5087 | 4.31k | xmlFree(buf); |
5088 | 4.31k | } |
5089 | | |
5090 | | /** |
5091 | | * Parse an XML (SGML) comment. Always consumes '<!'. |
5092 | | * |
5093 | | * @deprecated Internal function, don't use. |
5094 | | * |
5095 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
5096 | | * must not occur within comments. " |
5097 | | * |
5098 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
5099 | | * @param ctxt an XML parser context |
5100 | | */ |
5101 | | void |
5102 | 12.1k | xmlParseComment(xmlParserCtxtPtr ctxt) { |
5103 | 12.1k | xmlChar *buf = NULL; |
5104 | 12.1k | size_t size = XML_PARSER_BUFFER_SIZE; |
5105 | 12.1k | size_t len = 0; |
5106 | 12.1k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5107 | 0 | XML_MAX_HUGE_LENGTH : |
5108 | 12.1k | XML_MAX_TEXT_LENGTH; |
5109 | 12.1k | const xmlChar *in; |
5110 | 12.1k | size_t nbchar = 0; |
5111 | 12.1k | int ccol; |
5112 | | |
5113 | | /* |
5114 | | * Check that there is a comment right here. |
5115 | | */ |
5116 | 12.1k | if ((RAW != '<') || (NXT(1) != '!')) |
5117 | 0 | return; |
5118 | 12.1k | SKIP(2); |
5119 | 12.1k | if ((RAW != '-') || (NXT(1) != '-')) |
5120 | 3 | return; |
5121 | 12.1k | SKIP(2); |
5122 | 12.1k | GROW; |
5123 | | |
5124 | | /* |
5125 | | * Accelerated common case where input don't need to be |
5126 | | * modified before passing it to the handler. |
5127 | | */ |
5128 | 12.1k | in = ctxt->input->cur; |
5129 | 12.1k | do { |
5130 | 12.1k | if (*in == 0xA) { |
5131 | 411 | do { |
5132 | 411 | ctxt->input->line++; ctxt->input->col = 1; |
5133 | 411 | in++; |
5134 | 411 | } while (*in == 0xA); |
5135 | 212 | } |
5136 | 18.5k | get_more: |
5137 | 18.5k | ccol = ctxt->input->col; |
5138 | 83.8k | while (((*in > '-') && (*in <= 0x7F)) || |
5139 | 83.8k | ((*in >= 0x20) && (*in < '-')) || |
5140 | 83.8k | (*in == 0x09)) { |
5141 | 65.2k | in++; |
5142 | 65.2k | ccol++; |
5143 | 65.2k | } |
5144 | 18.5k | ctxt->input->col = ccol; |
5145 | 18.5k | if (*in == 0xA) { |
5146 | 432 | do { |
5147 | 432 | ctxt->input->line++; ctxt->input->col = 1; |
5148 | 432 | in++; |
5149 | 432 | } while (*in == 0xA); |
5150 | 238 | goto get_more; |
5151 | 238 | } |
5152 | 18.3k | nbchar = in - ctxt->input->cur; |
5153 | | /* |
5154 | | * save current set of data |
5155 | | */ |
5156 | 18.3k | if (nbchar > 0) { |
5157 | 9.58k | if (nbchar > maxLength - len) { |
5158 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5159 | 0 | "Comment too big found", NULL); |
5160 | 0 | xmlFree(buf); |
5161 | 0 | return; |
5162 | 0 | } |
5163 | 9.58k | if (buf == NULL) { |
5164 | 5.55k | if ((*in == '-') && (in[1] == '-')) |
5165 | 1.41k | size = nbchar + 1; |
5166 | 4.13k | else |
5167 | 4.13k | size = XML_PARSER_BUFFER_SIZE + nbchar; |
5168 | 5.55k | buf = xmlMalloc(size); |
5169 | 5.55k | if (buf == NULL) { |
5170 | 0 | xmlErrMemory(ctxt); |
5171 | 0 | return; |
5172 | 0 | } |
5173 | 5.55k | len = 0; |
5174 | 5.55k | } else if (len + nbchar + 1 >= size) { |
5175 | 719 | xmlChar *new_buf; |
5176 | 719 | size += len + nbchar + XML_PARSER_BUFFER_SIZE; |
5177 | 719 | new_buf = xmlRealloc(buf, size); |
5178 | 719 | if (new_buf == NULL) { |
5179 | 0 | xmlErrMemory(ctxt); |
5180 | 0 | xmlFree(buf); |
5181 | 0 | return; |
5182 | 0 | } |
5183 | 719 | buf = new_buf; |
5184 | 719 | } |
5185 | 9.58k | memcpy(&buf[len], ctxt->input->cur, nbchar); |
5186 | 9.58k | len += nbchar; |
5187 | 9.58k | buf[len] = 0; |
5188 | 9.58k | } |
5189 | 18.3k | ctxt->input->cur = in; |
5190 | 18.3k | if (*in == 0xA) { |
5191 | 0 | in++; |
5192 | 0 | ctxt->input->line++; ctxt->input->col = 1; |
5193 | 0 | } |
5194 | 18.3k | if (*in == 0xD) { |
5195 | 1.61k | in++; |
5196 | 1.61k | if (*in == 0xA) { |
5197 | 227 | ctxt->input->cur = in; |
5198 | 227 | in++; |
5199 | 227 | ctxt->input->line++; ctxt->input->col = 1; |
5200 | 227 | goto get_more; |
5201 | 227 | } |
5202 | 1.38k | in--; |
5203 | 1.38k | } |
5204 | 18.0k | SHRINK; |
5205 | 18.0k | GROW; |
5206 | 18.0k | in = ctxt->input->cur; |
5207 | 18.0k | if (*in == '-') { |
5208 | 9.36k | if (in[1] == '-') { |
5209 | 6.63k | if (in[2] == '>') { |
5210 | 3.46k | SKIP(3); |
5211 | 3.46k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5212 | 3.46k | (!ctxt->disableSAX)) { |
5213 | 2.91k | if (buf != NULL) |
5214 | 1.34k | ctxt->sax->comment(ctxt->userData, buf); |
5215 | 1.56k | else |
5216 | 1.56k | ctxt->sax->comment(ctxt->userData, BAD_CAST ""); |
5217 | 2.91k | } |
5218 | 3.46k | if (buf != NULL) |
5219 | 1.50k | xmlFree(buf); |
5220 | 3.46k | return; |
5221 | 3.46k | } |
5222 | 3.17k | if (buf != NULL) { |
5223 | 1.50k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5224 | 1.50k | "Double hyphen within comment: " |
5225 | 1.50k | "<!--%.50s\n", |
5226 | 1.50k | buf); |
5227 | 1.50k | } else |
5228 | 1.67k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5229 | 1.67k | "Double hyphen within comment\n", NULL); |
5230 | 3.17k | in++; |
5231 | 3.17k | ctxt->input->col++; |
5232 | 3.17k | } |
5233 | 5.89k | in++; |
5234 | 5.89k | ctxt->input->col++; |
5235 | 5.89k | goto get_more; |
5236 | 9.36k | } |
5237 | 18.0k | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
5238 | 8.71k | xmlParseCommentComplex(ctxt, buf, len, size); |
5239 | 8.71k | } |
5240 | | |
5241 | | |
5242 | | /** |
5243 | | * parse the name of a PI |
5244 | | * |
5245 | | * @deprecated Internal function, don't use. |
5246 | | * |
5247 | | * [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) |
5248 | | * |
5249 | | * @param ctxt an XML parser context |
5250 | | * @returns the PITarget name or NULL |
5251 | | */ |
5252 | | |
5253 | | const xmlChar * |
5254 | 26.8k | xmlParsePITarget(xmlParserCtxtPtr ctxt) { |
5255 | 26.8k | const xmlChar *name; |
5256 | | |
5257 | 26.8k | name = xmlParseName(ctxt); |
5258 | 26.8k | if ((name != NULL) && |
5259 | 26.8k | ((name[0] == 'x') || (name[0] == 'X')) && |
5260 | 26.8k | ((name[1] == 'm') || (name[1] == 'M')) && |
5261 | 26.8k | ((name[2] == 'l') || (name[2] == 'L'))) { |
5262 | 1.48k | int i; |
5263 | 1.48k | if ((name[0] == 'x') && (name[1] == 'm') && |
5264 | 1.48k | (name[2] == 'l') && (name[3] == 0)) { |
5265 | 292 | xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5266 | 292 | "XML declaration allowed only at the start of the document\n"); |
5267 | 292 | return(name); |
5268 | 1.19k | } else if (name[3] == 0) { |
5269 | 545 | xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL); |
5270 | 545 | return(name); |
5271 | 545 | } |
5272 | 1.75k | for (i = 0;;i++) { |
5273 | 1.75k | if (xmlW3CPIs[i] == NULL) break; |
5274 | 1.29k | if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i])) |
5275 | 194 | return(name); |
5276 | 1.29k | } |
5277 | 455 | xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5278 | 455 | "xmlParsePITarget: invalid name prefix 'xml'\n", |
5279 | 455 | NULL, NULL); |
5280 | 455 | } |
5281 | 25.7k | if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) { |
5282 | 835 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5283 | 835 | "colons are forbidden from PI names '%s'\n", name, NULL, NULL); |
5284 | 835 | } |
5285 | 25.7k | return(name); |
5286 | 26.8k | } |
5287 | | |
5288 | | #ifdef LIBXML_CATALOG_ENABLED |
5289 | | /** |
5290 | | * parse an XML Catalog Processing Instruction. |
5291 | | * |
5292 | | * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?> |
5293 | | * |
5294 | | * Occurs only if allowed by the user and if happening in the Misc |
5295 | | * part of the document before any doctype information |
5296 | | * This will add the given catalog to the parsing context in order |
5297 | | * to be used if there is a resolution need further down in the document |
5298 | | * |
5299 | | * @param ctxt an XML parser context |
5300 | | * @param catalog the PI value string |
5301 | | */ |
5302 | | |
5303 | | static void |
5304 | 608 | xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) { |
5305 | 608 | xmlChar *URL = NULL; |
5306 | 608 | const xmlChar *tmp, *base; |
5307 | 608 | xmlChar marker; |
5308 | | |
5309 | 608 | tmp = catalog; |
5310 | 608 | while (IS_BLANK_CH(*tmp)) tmp++; |
5311 | 608 | if (xmlStrncmp(tmp, BAD_CAST"catalog", 7)) |
5312 | 187 | goto error; |
5313 | 421 | tmp += 7; |
5314 | 421 | while (IS_BLANK_CH(*tmp)) tmp++; |
5315 | 421 | if (*tmp != '=') { |
5316 | 94 | return; |
5317 | 94 | } |
5318 | 327 | tmp++; |
5319 | 418 | while (IS_BLANK_CH(*tmp)) tmp++; |
5320 | 327 | marker = *tmp; |
5321 | 327 | if ((marker != '\'') && (marker != '"')) |
5322 | 45 | goto error; |
5323 | 282 | tmp++; |
5324 | 282 | base = tmp; |
5325 | 1.43k | while ((*tmp != 0) && (*tmp != marker)) tmp++; |
5326 | 282 | if (*tmp == 0) |
5327 | 130 | goto error; |
5328 | 152 | URL = xmlStrndup(base, tmp - base); |
5329 | 152 | tmp++; |
5330 | 388 | while (IS_BLANK_CH(*tmp)) tmp++; |
5331 | 152 | if (*tmp != 0) |
5332 | 50 | goto error; |
5333 | | |
5334 | 102 | if (URL != NULL) { |
5335 | | /* |
5336 | | * Unfortunately, the catalog API doesn't report OOM errors. |
5337 | | * xmlGetLastError isn't very helpful since we don't know |
5338 | | * where the last error came from. We'd have to reset it |
5339 | | * before this call and restore it afterwards. |
5340 | | */ |
5341 | 102 | ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL); |
5342 | 102 | xmlFree(URL); |
5343 | 102 | } |
5344 | 102 | return; |
5345 | | |
5346 | 412 | error: |
5347 | 412 | xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI, |
5348 | 412 | "Catalog PI syntax error: %s\n", |
5349 | 412 | catalog, NULL); |
5350 | 412 | if (URL != NULL) |
5351 | 50 | xmlFree(URL); |
5352 | 412 | } |
5353 | | #endif |
5354 | | |
5355 | | /** |
5356 | | * parse an XML Processing Instruction. |
5357 | | * |
5358 | | * @deprecated Internal function, don't use. |
5359 | | * |
5360 | | * [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' |
5361 | | * |
5362 | | * The processing is transferred to SAX once parsed. |
5363 | | * |
5364 | | * @param ctxt an XML parser context |
5365 | | */ |
5366 | | |
5367 | | void |
5368 | 26.8k | xmlParsePI(xmlParserCtxtPtr ctxt) { |
5369 | 26.8k | xmlChar *buf = NULL; |
5370 | 26.8k | size_t len = 0; |
5371 | 26.8k | size_t size = XML_PARSER_BUFFER_SIZE; |
5372 | 26.8k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5373 | 0 | XML_MAX_HUGE_LENGTH : |
5374 | 26.8k | XML_MAX_TEXT_LENGTH; |
5375 | 26.8k | int cur, l; |
5376 | 26.8k | const xmlChar *target; |
5377 | | |
5378 | 26.8k | if ((RAW == '<') && (NXT(1) == '?')) { |
5379 | | /* |
5380 | | * this is a Processing Instruction. |
5381 | | */ |
5382 | 26.8k | SKIP(2); |
5383 | | |
5384 | | /* |
5385 | | * Parse the target name and check for special support like |
5386 | | * namespace. |
5387 | | */ |
5388 | 26.8k | target = xmlParsePITarget(ctxt); |
5389 | 26.8k | if (target != NULL) { |
5390 | 15.3k | if ((RAW == '?') && (NXT(1) == '>')) { |
5391 | 5.01k | SKIP(2); |
5392 | | |
5393 | | /* |
5394 | | * SAX: PI detected. |
5395 | | */ |
5396 | 5.01k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5397 | 5.01k | (ctxt->sax->processingInstruction != NULL)) |
5398 | 4.11k | ctxt->sax->processingInstruction(ctxt->userData, |
5399 | 4.11k | target, NULL); |
5400 | 5.01k | return; |
5401 | 5.01k | } |
5402 | 10.2k | buf = xmlMalloc(size); |
5403 | 10.2k | if (buf == NULL) { |
5404 | 0 | xmlErrMemory(ctxt); |
5405 | 0 | return; |
5406 | 0 | } |
5407 | 10.2k | if (SKIP_BLANKS == 0) { |
5408 | 6.18k | xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED, |
5409 | 6.18k | "ParsePI: PI %s space expected\n", target); |
5410 | 6.18k | } |
5411 | 10.2k | cur = xmlCurrentCharRecover(ctxt, &l); |
5412 | 113k | while (IS_CHAR(cur) && /* checked */ |
5413 | 113k | ((cur != '?') || (NXT(1) != '>'))) { |
5414 | 102k | if (len + 5 >= size) { |
5415 | 1.00k | xmlChar *tmp; |
5416 | 1.00k | int newSize; |
5417 | | |
5418 | 1.00k | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
5419 | 1.00k | if (newSize < 0) { |
5420 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5421 | 0 | "PI %s too big found", target); |
5422 | 0 | xmlFree(buf); |
5423 | 0 | return; |
5424 | 0 | } |
5425 | 1.00k | tmp = xmlRealloc(buf, newSize); |
5426 | 1.00k | if (tmp == NULL) { |
5427 | 0 | xmlErrMemory(ctxt); |
5428 | 0 | xmlFree(buf); |
5429 | 0 | return; |
5430 | 0 | } |
5431 | 1.00k | buf = tmp; |
5432 | 1.00k | size = newSize; |
5433 | 1.00k | } |
5434 | 102k | COPY_BUF(buf, len, cur); |
5435 | 102k | NEXTL(l); |
5436 | 102k | cur = xmlCurrentCharRecover(ctxt, &l); |
5437 | 102k | } |
5438 | 10.2k | buf[len] = 0; |
5439 | 10.2k | if (cur != '?') { |
5440 | 4.75k | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5441 | 4.75k | "ParsePI: PI %s never end ...\n", target); |
5442 | 5.52k | } else { |
5443 | 5.52k | SKIP(2); |
5444 | | |
5445 | 5.52k | #ifdef LIBXML_CATALOG_ENABLED |
5446 | 5.52k | if ((ctxt->inSubset == 0) && |
5447 | 5.52k | (xmlStrEqual(target, XML_CATALOG_PI))) { |
5448 | 679 | xmlCatalogAllow allow = xmlCatalogGetDefaults(); |
5449 | | |
5450 | 679 | if ((ctxt->options & XML_PARSE_CATALOG_PI) && |
5451 | 679 | ((allow == XML_CATA_ALLOW_DOCUMENT) || |
5452 | 608 | (allow == XML_CATA_ALLOW_ALL))) |
5453 | 608 | xmlParseCatalogPI(ctxt, buf); |
5454 | 679 | } |
5455 | 5.52k | #endif |
5456 | | |
5457 | | /* |
5458 | | * SAX: PI detected. |
5459 | | */ |
5460 | 5.52k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5461 | 5.52k | (ctxt->sax->processingInstruction != NULL)) |
5462 | 4.16k | ctxt->sax->processingInstruction(ctxt->userData, |
5463 | 4.16k | target, buf); |
5464 | 5.52k | } |
5465 | 10.2k | xmlFree(buf); |
5466 | 11.5k | } else { |
5467 | 11.5k | xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL); |
5468 | 11.5k | } |
5469 | 26.8k | } |
5470 | 26.8k | } |
5471 | | |
5472 | | /** |
5473 | | * Parse a notation declaration. Always consumes '<!'. |
5474 | | * |
5475 | | * @deprecated Internal function, don't use. |
5476 | | * |
5477 | | * [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) |
5478 | | * S? '>' |
5479 | | * |
5480 | | * Hence there is actually 3 choices: |
5481 | | * |
5482 | | * 'PUBLIC' S PubidLiteral |
5483 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
5484 | | * 'SYSTEM' S SystemLiteral |
5485 | | * |
5486 | | * See the NOTE on xmlParseExternalID(). |
5487 | | * |
5488 | | * @param ctxt an XML parser context |
5489 | | */ |
5490 | | |
5491 | | void |
5492 | 4.74k | xmlParseNotationDecl(xmlParserCtxtPtr ctxt) { |
5493 | 4.74k | const xmlChar *name; |
5494 | 4.74k | xmlChar *Pubid; |
5495 | 4.74k | xmlChar *Systemid; |
5496 | | |
5497 | 4.74k | if ((CUR != '<') || (NXT(1) != '!')) |
5498 | 0 | return; |
5499 | 4.74k | SKIP(2); |
5500 | | |
5501 | 4.74k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5502 | 4.72k | int inputid = ctxt->input->id; |
5503 | 4.72k | SKIP(8); |
5504 | 4.72k | if (SKIP_BLANKS_PE == 0) { |
5505 | 214 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5506 | 214 | "Space required after '<!NOTATION'\n"); |
5507 | 214 | return; |
5508 | 214 | } |
5509 | | |
5510 | 4.51k | name = xmlParseName(ctxt); |
5511 | 4.51k | if (name == NULL) { |
5512 | 408 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5513 | 408 | return; |
5514 | 408 | } |
5515 | 4.10k | if (xmlStrchr(name, ':') != NULL) { |
5516 | 472 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5517 | 472 | "colons are forbidden from notation names '%s'\n", |
5518 | 472 | name, NULL, NULL); |
5519 | 472 | } |
5520 | 4.10k | if (SKIP_BLANKS_PE == 0) { |
5521 | 208 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5522 | 208 | "Space required after the NOTATION name'\n"); |
5523 | 208 | return; |
5524 | 208 | } |
5525 | | |
5526 | | /* |
5527 | | * Parse the IDs. |
5528 | | */ |
5529 | 3.89k | Systemid = xmlParseExternalID(ctxt, &Pubid, 0); |
5530 | 3.89k | SKIP_BLANKS_PE; |
5531 | | |
5532 | 3.89k | if (RAW == '>') { |
5533 | 1.46k | if (inputid != ctxt->input->id) { |
5534 | 125 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5535 | 125 | "Notation declaration doesn't start and stop" |
5536 | 125 | " in the same entity\n"); |
5537 | 125 | } |
5538 | 1.46k | NEXT; |
5539 | 1.46k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5540 | 1.46k | (ctxt->sax->notationDecl != NULL)) |
5541 | 1.36k | ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid); |
5542 | 2.43k | } else { |
5543 | 2.43k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5544 | 2.43k | } |
5545 | 3.89k | if (Systemid != NULL) xmlFree(Systemid); |
5546 | 3.89k | if (Pubid != NULL) xmlFree(Pubid); |
5547 | 3.89k | } |
5548 | 4.74k | } |
5549 | | |
5550 | | /** |
5551 | | * Parse an entity declaration. Always consumes '<!'. |
5552 | | * |
5553 | | * @deprecated Internal function, don't use. |
5554 | | * |
5555 | | * [70] EntityDecl ::= GEDecl | PEDecl |
5556 | | * |
5557 | | * [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>' |
5558 | | * |
5559 | | * [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>' |
5560 | | * |
5561 | | * [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?) |
5562 | | * |
5563 | | * [74] PEDef ::= EntityValue | ExternalID |
5564 | | * |
5565 | | * [76] NDataDecl ::= S 'NDATA' S Name |
5566 | | * |
5567 | | * [ VC: Notation Declared ] |
5568 | | * The Name must match the declared name of a notation. |
5569 | | * |
5570 | | * @param ctxt an XML parser context |
5571 | | */ |
5572 | | |
5573 | | void |
5574 | 19.3k | xmlParseEntityDecl(xmlParserCtxtPtr ctxt) { |
5575 | 19.3k | const xmlChar *name = NULL; |
5576 | 19.3k | xmlChar *value = NULL; |
5577 | 19.3k | xmlChar *URI = NULL, *literal = NULL; |
5578 | 19.3k | const xmlChar *ndata = NULL; |
5579 | 19.3k | int isParameter = 0; |
5580 | 19.3k | xmlChar *orig = NULL; |
5581 | | |
5582 | 19.3k | if ((CUR != '<') || (NXT(1) != '!')) |
5583 | 0 | return; |
5584 | 19.3k | SKIP(2); |
5585 | | |
5586 | | /* GROW; done in the caller */ |
5587 | 19.3k | if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5588 | 19.3k | int inputid = ctxt->input->id; |
5589 | 19.3k | SKIP(6); |
5590 | 19.3k | if (SKIP_BLANKS_PE == 0) { |
5591 | 16.9k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5592 | 16.9k | "Space required after '<!ENTITY'\n"); |
5593 | 16.9k | } |
5594 | | |
5595 | 19.3k | if (RAW == '%') { |
5596 | 3.97k | NEXT; |
5597 | 3.97k | if (SKIP_BLANKS_PE == 0) { |
5598 | 3.76k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5599 | 3.76k | "Space required after '%%'\n"); |
5600 | 3.76k | } |
5601 | 3.97k | isParameter = 1; |
5602 | 3.97k | } |
5603 | | |
5604 | 19.3k | name = xmlParseName(ctxt); |
5605 | 19.3k | if (name == NULL) { |
5606 | 506 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5607 | 506 | "xmlParseEntityDecl: no name\n"); |
5608 | 506 | return; |
5609 | 506 | } |
5610 | 18.7k | if (xmlStrchr(name, ':') != NULL) { |
5611 | 4.03k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5612 | 4.03k | "colons are forbidden from entities names '%s'\n", |
5613 | 4.03k | name, NULL, NULL); |
5614 | 4.03k | } |
5615 | 18.7k | if (SKIP_BLANKS_PE == 0) { |
5616 | 13.7k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5617 | 13.7k | "Space required after the entity name\n"); |
5618 | 13.7k | } |
5619 | | |
5620 | | /* |
5621 | | * handle the various case of definitions... |
5622 | | */ |
5623 | 18.7k | if (isParameter) { |
5624 | 3.76k | if ((RAW == '"') || (RAW == '\'')) { |
5625 | 2.80k | value = xmlParseEntityValue(ctxt, &orig); |
5626 | 2.80k | if (value) { |
5627 | 2.73k | if ((ctxt->sax != NULL) && |
5628 | 2.73k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5629 | 2.59k | ctxt->sax->entityDecl(ctxt->userData, name, |
5630 | 2.59k | XML_INTERNAL_PARAMETER_ENTITY, |
5631 | 2.59k | NULL, NULL, value); |
5632 | 2.73k | } |
5633 | 2.80k | } else { |
5634 | 953 | URI = xmlParseExternalID(ctxt, &literal, 1); |
5635 | 953 | if ((URI == NULL) && (literal == NULL)) { |
5636 | 200 | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5637 | 200 | } |
5638 | 953 | if (URI) { |
5639 | 558 | if (xmlStrchr(URI, '#')) { |
5640 | 194 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5641 | 364 | } else { |
5642 | 364 | if ((ctxt->sax != NULL) && |
5643 | 364 | (!ctxt->disableSAX) && |
5644 | 364 | (ctxt->sax->entityDecl != NULL)) |
5645 | 292 | ctxt->sax->entityDecl(ctxt->userData, name, |
5646 | 292 | XML_EXTERNAL_PARAMETER_ENTITY, |
5647 | 292 | literal, URI, NULL); |
5648 | 364 | } |
5649 | 558 | } |
5650 | 953 | } |
5651 | 15.0k | } else { |
5652 | 15.0k | if ((RAW == '"') || (RAW == '\'')) { |
5653 | 11.9k | value = xmlParseEntityValue(ctxt, &orig); |
5654 | 11.9k | if ((ctxt->sax != NULL) && |
5655 | 11.9k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5656 | 10.8k | ctxt->sax->entityDecl(ctxt->userData, name, |
5657 | 10.8k | XML_INTERNAL_GENERAL_ENTITY, |
5658 | 10.8k | NULL, NULL, value); |
5659 | | /* |
5660 | | * For expat compatibility in SAX mode. |
5661 | | */ |
5662 | 11.9k | if ((ctxt->myDoc == NULL) || |
5663 | 11.9k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
5664 | 267 | if (ctxt->myDoc == NULL) { |
5665 | 135 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5666 | 135 | if (ctxt->myDoc == NULL) { |
5667 | 0 | xmlErrMemory(ctxt); |
5668 | 0 | goto done; |
5669 | 0 | } |
5670 | 135 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5671 | 135 | } |
5672 | 267 | if (ctxt->myDoc->intSubset == NULL) { |
5673 | 135 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5674 | 135 | BAD_CAST "fake", NULL, NULL); |
5675 | 135 | if (ctxt->myDoc->intSubset == NULL) { |
5676 | 0 | xmlErrMemory(ctxt); |
5677 | 0 | goto done; |
5678 | 0 | } |
5679 | 135 | } |
5680 | | |
5681 | 267 | xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY, |
5682 | 267 | NULL, NULL, value); |
5683 | 267 | } |
5684 | 11.9k | } else { |
5685 | 3.09k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5686 | 3.09k | if ((URI == NULL) && (literal == NULL)) { |
5687 | 2.19k | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5688 | 2.19k | } |
5689 | 3.09k | if (URI) { |
5690 | 540 | if (xmlStrchr(URI, '#')) { |
5691 | 235 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5692 | 235 | } |
5693 | 540 | } |
5694 | 3.09k | if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) { |
5695 | 679 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5696 | 679 | "Space required before 'NDATA'\n"); |
5697 | 679 | } |
5698 | 3.09k | if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) { |
5699 | 566 | SKIP(5); |
5700 | 566 | if (SKIP_BLANKS_PE == 0) { |
5701 | 307 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5702 | 307 | "Space required after 'NDATA'\n"); |
5703 | 307 | } |
5704 | 566 | ndata = xmlParseName(ctxt); |
5705 | 566 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5706 | 566 | (ctxt->sax->unparsedEntityDecl != NULL)) |
5707 | 495 | ctxt->sax->unparsedEntityDecl(ctxt->userData, name, |
5708 | 495 | literal, URI, ndata); |
5709 | 2.52k | } else { |
5710 | 2.52k | if ((ctxt->sax != NULL) && |
5711 | 2.52k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5712 | 2.36k | ctxt->sax->entityDecl(ctxt->userData, name, |
5713 | 2.36k | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5714 | 2.36k | literal, URI, NULL); |
5715 | | /* |
5716 | | * For expat compatibility in SAX mode. |
5717 | | * assuming the entity replacement was asked for |
5718 | | */ |
5719 | 2.52k | if ((ctxt->replaceEntities != 0) && |
5720 | 2.52k | ((ctxt->myDoc == NULL) || |
5721 | 0 | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) { |
5722 | 0 | if (ctxt->myDoc == NULL) { |
5723 | 0 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5724 | 0 | if (ctxt->myDoc == NULL) { |
5725 | 0 | xmlErrMemory(ctxt); |
5726 | 0 | goto done; |
5727 | 0 | } |
5728 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5729 | 0 | } |
5730 | | |
5731 | 0 | if (ctxt->myDoc->intSubset == NULL) { |
5732 | 0 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5733 | 0 | BAD_CAST "fake", NULL, NULL); |
5734 | 0 | if (ctxt->myDoc->intSubset == NULL) { |
5735 | 0 | xmlErrMemory(ctxt); |
5736 | 0 | goto done; |
5737 | 0 | } |
5738 | 0 | } |
5739 | 0 | xmlSAX2EntityDecl(ctxt, name, |
5740 | 0 | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5741 | 0 | literal, URI, NULL); |
5742 | 0 | } |
5743 | 2.52k | } |
5744 | 3.09k | } |
5745 | 15.0k | } |
5746 | 18.7k | SKIP_BLANKS_PE; |
5747 | 18.7k | if (RAW != '>') { |
5748 | 1.13k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
5749 | 1.13k | "xmlParseEntityDecl: entity %s not terminated\n", name); |
5750 | 1.13k | xmlHaltParser(ctxt); |
5751 | 17.6k | } else { |
5752 | 17.6k | if (inputid != ctxt->input->id) { |
5753 | 314 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5754 | 314 | "Entity declaration doesn't start and stop in" |
5755 | 314 | " the same entity\n"); |
5756 | 314 | } |
5757 | 17.6k | NEXT; |
5758 | 17.6k | } |
5759 | 18.7k | if (orig != NULL) { |
5760 | | /* |
5761 | | * Ugly mechanism to save the raw entity value. |
5762 | | */ |
5763 | 13.8k | xmlEntityPtr cur = NULL; |
5764 | | |
5765 | 13.8k | if (isParameter) { |
5766 | 2.73k | if ((ctxt->sax != NULL) && |
5767 | 2.73k | (ctxt->sax->getParameterEntity != NULL)) |
5768 | 2.73k | cur = ctxt->sax->getParameterEntity(ctxt->userData, name); |
5769 | 11.0k | } else { |
5770 | 11.0k | if ((ctxt->sax != NULL) && |
5771 | 11.0k | (ctxt->sax->getEntity != NULL)) |
5772 | 11.0k | cur = ctxt->sax->getEntity(ctxt->userData, name); |
5773 | 11.0k | if ((cur == NULL) && (ctxt->userData==ctxt)) { |
5774 | 712 | cur = xmlSAX2GetEntity(ctxt, name); |
5775 | 712 | } |
5776 | 11.0k | } |
5777 | 13.8k | if ((cur != NULL) && (cur->orig == NULL)) { |
5778 | 6.06k | cur->orig = orig; |
5779 | 6.06k | orig = NULL; |
5780 | 6.06k | } |
5781 | 13.8k | } |
5782 | | |
5783 | 18.7k | done: |
5784 | 18.7k | if (value != NULL) xmlFree(value); |
5785 | 18.7k | if (URI != NULL) xmlFree(URI); |
5786 | 18.7k | if (literal != NULL) xmlFree(literal); |
5787 | 18.7k | if (orig != NULL) xmlFree(orig); |
5788 | 18.7k | } |
5789 | 19.3k | } |
5790 | | |
5791 | | /** |
5792 | | * Parse an attribute default declaration |
5793 | | * |
5794 | | * @deprecated Internal function, don't use. |
5795 | | * |
5796 | | * [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue) |
5797 | | * |
5798 | | * [ VC: Required Attribute ] |
5799 | | * if the default declaration is the keyword \#REQUIRED, then the |
5800 | | * attribute must be specified for all elements of the type in the |
5801 | | * attribute-list declaration. |
5802 | | * |
5803 | | * [ VC: Attribute Default Legal ] |
5804 | | * The declared default value must meet the lexical constraints of |
5805 | | * the declared attribute type c.f. xmlValidateAttributeDecl() |
5806 | | * |
5807 | | * [ VC: Fixed Attribute Default ] |
5808 | | * if an attribute has a default value declared with the \#FIXED |
5809 | | * keyword, instances of that attribute must match the default value. |
5810 | | * |
5811 | | * [ WFC: No < in Attribute Values ] |
5812 | | * handled in xmlParseAttValue() |
5813 | | * |
5814 | | * @param ctxt an XML parser context |
5815 | | * @param value Receive a possible fixed default value for the attribute |
5816 | | * @returns XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED |
5817 | | * or XML_ATTRIBUTE_FIXED. |
5818 | | */ |
5819 | | |
5820 | | int |
5821 | 37.9k | xmlParseDefaultDecl(xmlParserCtxtPtr ctxt, xmlChar **value) { |
5822 | 37.9k | int val; |
5823 | 37.9k | xmlChar *ret; |
5824 | | |
5825 | 37.9k | *value = NULL; |
5826 | 37.9k | if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) { |
5827 | 488 | SKIP(9); |
5828 | 488 | return(XML_ATTRIBUTE_REQUIRED); |
5829 | 488 | } |
5830 | 37.4k | if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) { |
5831 | 535 | SKIP(8); |
5832 | 535 | return(XML_ATTRIBUTE_IMPLIED); |
5833 | 535 | } |
5834 | 36.8k | val = XML_ATTRIBUTE_NONE; |
5835 | 36.8k | if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) { |
5836 | 436 | SKIP(6); |
5837 | 436 | val = XML_ATTRIBUTE_FIXED; |
5838 | 436 | if (SKIP_BLANKS_PE == 0) { |
5839 | 201 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5840 | 201 | "Space required after '#FIXED'\n"); |
5841 | 201 | } |
5842 | 436 | } |
5843 | 36.8k | ret = xmlParseAttValue(ctxt); |
5844 | 36.8k | if (ret == NULL) { |
5845 | 12.4k | xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo, |
5846 | 12.4k | "Attribute default value declaration error\n"); |
5847 | 12.4k | } else |
5848 | 24.3k | *value = ret; |
5849 | 36.8k | return(val); |
5850 | 37.4k | } |
5851 | | |
5852 | | /** |
5853 | | * parse an Notation attribute type. |
5854 | | * |
5855 | | * @deprecated Internal function, don't use. |
5856 | | * |
5857 | | * Note: the leading 'NOTATION' S part has already being parsed... |
5858 | | * |
5859 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5860 | | * |
5861 | | * [ VC: Notation Attributes ] |
5862 | | * Values of this type must match one of the notation names included |
5863 | | * in the declaration; all notation names in the declaration must be declared. |
5864 | | * |
5865 | | * @param ctxt an XML parser context |
5866 | | * @returns the notation attribute tree built while parsing |
5867 | | */ |
5868 | | |
5869 | | xmlEnumerationPtr |
5870 | 1.07k | xmlParseNotationType(xmlParserCtxtPtr ctxt) { |
5871 | 1.07k | const xmlChar *name; |
5872 | 1.07k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5873 | | |
5874 | 1.07k | if (RAW != '(') { |
5875 | 207 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5876 | 207 | return(NULL); |
5877 | 207 | } |
5878 | 1.91k | do { |
5879 | 1.91k | NEXT; |
5880 | 1.91k | SKIP_BLANKS_PE; |
5881 | 1.91k | name = xmlParseName(ctxt); |
5882 | 1.91k | if (name == NULL) { |
5883 | 200 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5884 | 200 | "Name expected in NOTATION declaration\n"); |
5885 | 200 | xmlFreeEnumeration(ret); |
5886 | 200 | return(NULL); |
5887 | 200 | } |
5888 | 1.71k | tmp = ret; |
5889 | 3.23k | while (tmp != NULL) { |
5890 | 2.21k | if (xmlStrEqual(name, tmp->name)) { |
5891 | 701 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5892 | 701 | "standalone: attribute notation value token %s duplicated\n", |
5893 | 701 | name, NULL); |
5894 | 701 | if (!xmlDictOwns(ctxt->dict, name)) |
5895 | 0 | xmlFree((xmlChar *) name); |
5896 | 701 | break; |
5897 | 701 | } |
5898 | 1.51k | tmp = tmp->next; |
5899 | 1.51k | } |
5900 | 1.71k | if (tmp == NULL) { |
5901 | 1.01k | cur = xmlCreateEnumeration(name); |
5902 | 1.01k | if (cur == NULL) { |
5903 | 0 | xmlErrMemory(ctxt); |
5904 | 0 | xmlFreeEnumeration(ret); |
5905 | 0 | return(NULL); |
5906 | 0 | } |
5907 | 1.01k | if (last == NULL) ret = last = cur; |
5908 | 346 | else { |
5909 | 346 | last->next = cur; |
5910 | 346 | last = cur; |
5911 | 346 | } |
5912 | 1.01k | } |
5913 | 1.71k | SKIP_BLANKS_PE; |
5914 | 1.71k | } while (RAW == '|'); |
5915 | 663 | if (RAW != ')') { |
5916 | 382 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5917 | 382 | xmlFreeEnumeration(ret); |
5918 | 382 | return(NULL); |
5919 | 382 | } |
5920 | 281 | NEXT; |
5921 | 281 | return(ret); |
5922 | 663 | } |
5923 | | |
5924 | | /** |
5925 | | * parse an Enumeration attribute type. |
5926 | | * |
5927 | | * @deprecated Internal function, don't use. |
5928 | | * |
5929 | | * [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' |
5930 | | * |
5931 | | * [ VC: Enumeration ] |
5932 | | * Values of this type must match one of the Nmtoken tokens in |
5933 | | * the declaration |
5934 | | * |
5935 | | * @param ctxt an XML parser context |
5936 | | * @returns the enumeration attribute tree built while parsing |
5937 | | */ |
5938 | | |
5939 | | xmlEnumerationPtr |
5940 | 8.93k | xmlParseEnumerationType(xmlParserCtxtPtr ctxt) { |
5941 | 8.93k | xmlChar *name; |
5942 | 8.93k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5943 | | |
5944 | 8.93k | if (RAW != '(') { |
5945 | 817 | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL); |
5946 | 817 | return(NULL); |
5947 | 817 | } |
5948 | 9.31k | do { |
5949 | 9.31k | NEXT; |
5950 | 9.31k | SKIP_BLANKS_PE; |
5951 | 9.31k | name = xmlParseNmtoken(ctxt); |
5952 | 9.31k | if (name == NULL) { |
5953 | 306 | xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL); |
5954 | 306 | return(ret); |
5955 | 306 | } |
5956 | 9.00k | tmp = ret; |
5957 | 11.0k | while (tmp != NULL) { |
5958 | 2.64k | if (xmlStrEqual(name, tmp->name)) { |
5959 | 635 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5960 | 635 | "standalone: attribute enumeration value token %s duplicated\n", |
5961 | 635 | name, NULL); |
5962 | 635 | if (!xmlDictOwns(ctxt->dict, name)) |
5963 | 635 | xmlFree(name); |
5964 | 635 | break; |
5965 | 635 | } |
5966 | 2.00k | tmp = tmp->next; |
5967 | 2.00k | } |
5968 | 9.00k | if (tmp == NULL) { |
5969 | 8.37k | cur = xmlCreateEnumeration(name); |
5970 | 8.37k | if (!xmlDictOwns(ctxt->dict, name)) |
5971 | 8.37k | xmlFree(name); |
5972 | 8.37k | if (cur == NULL) { |
5973 | 0 | xmlErrMemory(ctxt); |
5974 | 0 | xmlFreeEnumeration(ret); |
5975 | 0 | return(NULL); |
5976 | 0 | } |
5977 | 8.37k | if (last == NULL) ret = last = cur; |
5978 | 528 | else { |
5979 | 528 | last->next = cur; |
5980 | 528 | last = cur; |
5981 | 528 | } |
5982 | 8.37k | } |
5983 | 9.00k | SKIP_BLANKS_PE; |
5984 | 9.00k | } while (RAW == '|'); |
5985 | 7.81k | if (RAW != ')') { |
5986 | 1.01k | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL); |
5987 | 1.01k | return(ret); |
5988 | 1.01k | } |
5989 | 6.79k | NEXT; |
5990 | 6.79k | return(ret); |
5991 | 7.81k | } |
5992 | | |
5993 | | /** |
5994 | | * parse an Enumerated attribute type. |
5995 | | * |
5996 | | * @deprecated Internal function, don't use. |
5997 | | * |
5998 | | * [57] EnumeratedType ::= NotationType | Enumeration |
5999 | | * |
6000 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
6001 | | * |
6002 | | * @param ctxt an XML parser context |
6003 | | * @param tree the enumeration tree built while parsing |
6004 | | * @returns XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION |
6005 | | */ |
6006 | | |
6007 | | int |
6008 | 10.2k | xmlParseEnumeratedType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
6009 | 10.2k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
6010 | 1.26k | SKIP(8); |
6011 | 1.26k | if (SKIP_BLANKS_PE == 0) { |
6012 | 195 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6013 | 195 | "Space required after 'NOTATION'\n"); |
6014 | 195 | return(0); |
6015 | 195 | } |
6016 | 1.07k | *tree = xmlParseNotationType(ctxt); |
6017 | 1.07k | if (*tree == NULL) return(0); |
6018 | 281 | return(XML_ATTRIBUTE_NOTATION); |
6019 | 1.07k | } |
6020 | 8.93k | *tree = xmlParseEnumerationType(ctxt); |
6021 | 8.93k | if (*tree == NULL) return(0); |
6022 | 7.84k | return(XML_ATTRIBUTE_ENUMERATION); |
6023 | 8.93k | } |
6024 | | |
6025 | | /** |
6026 | | * parse the Attribute list def for an element |
6027 | | * |
6028 | | * @deprecated Internal function, don't use. |
6029 | | * |
6030 | | * [54] AttType ::= StringType | TokenizedType | EnumeratedType |
6031 | | * |
6032 | | * [55] StringType ::= 'CDATA' |
6033 | | * |
6034 | | * [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' | |
6035 | | * 'ENTITIES' | 'NMTOKEN' | 'NMTOKENS' |
6036 | | * |
6037 | | * Validity constraints for attribute values syntax are checked in |
6038 | | * xmlValidateAttributeValue() |
6039 | | * |
6040 | | * [ VC: ID ] |
6041 | | * Values of type ID must match the Name production. A name must not |
6042 | | * appear more than once in an XML document as a value of this type; |
6043 | | * i.e., ID values must uniquely identify the elements which bear them. |
6044 | | * |
6045 | | * [ VC: One ID per Element Type ] |
6046 | | * No element type may have more than one ID attribute specified. |
6047 | | * |
6048 | | * [ VC: ID Attribute Default ] |
6049 | | * An ID attribute must have a declared default of \#IMPLIED or \#REQUIRED. |
6050 | | * |
6051 | | * [ VC: IDREF ] |
6052 | | * Values of type IDREF must match the Name production, and values |
6053 | | * of type IDREFS must match Names; each IDREF Name must match the value |
6054 | | * of an ID attribute on some element in the XML document; i.e. IDREF |
6055 | | * values must match the value of some ID attribute. |
6056 | | * |
6057 | | * [ VC: Entity Name ] |
6058 | | * Values of type ENTITY must match the Name production, values |
6059 | | * of type ENTITIES must match Names; each Entity Name must match the |
6060 | | * name of an unparsed entity declared in the DTD. |
6061 | | * |
6062 | | * [ VC: Name Token ] |
6063 | | * Values of type NMTOKEN must match the Nmtoken production; values |
6064 | | * of type NMTOKENS must match Nmtokens. |
6065 | | * |
6066 | | * @param ctxt an XML parser context |
6067 | | * @param tree the enumeration tree built while parsing |
6068 | | * @returns the attribute type |
6069 | | */ |
6070 | | int |
6071 | 41.6k | xmlParseAttributeType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
6072 | 41.6k | if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) { |
6073 | 2.44k | SKIP(5); |
6074 | 2.44k | return(XML_ATTRIBUTE_CDATA); |
6075 | 39.2k | } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) { |
6076 | 2.38k | SKIP(6); |
6077 | 2.38k | return(XML_ATTRIBUTE_IDREFS); |
6078 | 36.8k | } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) { |
6079 | 707 | SKIP(5); |
6080 | 707 | return(XML_ATTRIBUTE_IDREF); |
6081 | 36.1k | } else if ((RAW == 'I') && (NXT(1) == 'D')) { |
6082 | 21.8k | SKIP(2); |
6083 | 21.8k | return(XML_ATTRIBUTE_ID); |
6084 | 21.8k | } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
6085 | 718 | SKIP(6); |
6086 | 718 | return(XML_ATTRIBUTE_ENTITY); |
6087 | 13.5k | } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) { |
6088 | 1.56k | SKIP(8); |
6089 | 1.56k | return(XML_ATTRIBUTE_ENTITIES); |
6090 | 12.0k | } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) { |
6091 | 587 | SKIP(8); |
6092 | 587 | return(XML_ATTRIBUTE_NMTOKENS); |
6093 | 11.4k | } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) { |
6094 | 1.21k | SKIP(7); |
6095 | 1.21k | return(XML_ATTRIBUTE_NMTOKEN); |
6096 | 1.21k | } |
6097 | 10.2k | return(xmlParseEnumeratedType(ctxt, tree)); |
6098 | 41.6k | } |
6099 | | |
6100 | | /** |
6101 | | * Parse an attribute list declaration for an element. Always consumes '<!'. |
6102 | | * |
6103 | | * @deprecated Internal function, don't use. |
6104 | | * |
6105 | | * [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>' |
6106 | | * |
6107 | | * [53] AttDef ::= S Name S AttType S DefaultDecl |
6108 | | * @param ctxt an XML parser context |
6109 | | */ |
6110 | | void |
6111 | 30.0k | xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt) { |
6112 | 30.0k | const xmlChar *elemName; |
6113 | 30.0k | const xmlChar *attrName; |
6114 | 30.0k | xmlEnumerationPtr tree; |
6115 | | |
6116 | 30.0k | if ((CUR != '<') || (NXT(1) != '!')) |
6117 | 0 | return; |
6118 | 30.0k | SKIP(2); |
6119 | | |
6120 | 30.0k | if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) { |
6121 | 29.9k | int inputid = ctxt->input->id; |
6122 | | |
6123 | 29.9k | SKIP(7); |
6124 | 29.9k | if (SKIP_BLANKS_PE == 0) { |
6125 | 28.4k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6126 | 28.4k | "Space required after '<!ATTLIST'\n"); |
6127 | 28.4k | } |
6128 | 29.9k | elemName = xmlParseName(ctxt); |
6129 | 29.9k | if (elemName == NULL) { |
6130 | 535 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6131 | 535 | "ATTLIST: no name for Element\n"); |
6132 | 535 | return; |
6133 | 535 | } |
6134 | 29.4k | SKIP_BLANKS_PE; |
6135 | 29.4k | GROW; |
6136 | 64.4k | while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) { |
6137 | 57.9k | int type; |
6138 | 57.9k | int def; |
6139 | 57.9k | xmlChar *defaultValue = NULL; |
6140 | | |
6141 | 57.9k | GROW; |
6142 | 57.9k | tree = NULL; |
6143 | 57.9k | attrName = xmlParseName(ctxt); |
6144 | 57.9k | if (attrName == NULL) { |
6145 | 15.6k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6146 | 15.6k | "ATTLIST: no name for Attribute\n"); |
6147 | 15.6k | break; |
6148 | 15.6k | } |
6149 | 42.3k | GROW; |
6150 | 42.3k | if (SKIP_BLANKS_PE == 0) { |
6151 | 622 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6152 | 622 | "Space required after the attribute name\n"); |
6153 | 622 | break; |
6154 | 622 | } |
6155 | | |
6156 | 41.6k | type = xmlParseAttributeType(ctxt, &tree); |
6157 | 41.6k | if (type <= 0) { |
6158 | 2.07k | break; |
6159 | 2.07k | } |
6160 | | |
6161 | 39.6k | GROW; |
6162 | 39.6k | if (SKIP_BLANKS_PE == 0) { |
6163 | 1.71k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6164 | 1.71k | "Space required after the attribute type\n"); |
6165 | 1.71k | if (tree != NULL) |
6166 | 1.05k | xmlFreeEnumeration(tree); |
6167 | 1.71k | break; |
6168 | 1.71k | } |
6169 | | |
6170 | 37.9k | def = xmlParseDefaultDecl(ctxt, &defaultValue); |
6171 | 37.9k | if (def <= 0) { |
6172 | 0 | if (defaultValue != NULL) |
6173 | 0 | xmlFree(defaultValue); |
6174 | 0 | if (tree != NULL) |
6175 | 0 | xmlFreeEnumeration(tree); |
6176 | 0 | break; |
6177 | 0 | } |
6178 | 37.9k | if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL)) |
6179 | 22.3k | xmlAttrNormalizeSpace(defaultValue, defaultValue); |
6180 | | |
6181 | 37.9k | GROW; |
6182 | 37.9k | if (RAW != '>') { |
6183 | 31.9k | if (SKIP_BLANKS_PE == 0) { |
6184 | 2.86k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6185 | 2.86k | "Space required after the attribute default value\n"); |
6186 | 2.86k | if (defaultValue != NULL) |
6187 | 230 | xmlFree(defaultValue); |
6188 | 2.86k | if (tree != NULL) |
6189 | 460 | xmlFreeEnumeration(tree); |
6190 | 2.86k | break; |
6191 | 2.86k | } |
6192 | 31.9k | } |
6193 | 35.0k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6194 | 35.0k | (ctxt->sax->attributeDecl != NULL)) |
6195 | 32.8k | ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName, |
6196 | 32.8k | type, def, defaultValue, tree); |
6197 | 2.17k | else if (tree != NULL) |
6198 | 128 | xmlFreeEnumeration(tree); |
6199 | | |
6200 | 35.0k | if ((ctxt->sax2) && (defaultValue != NULL) && |
6201 | 35.0k | (def != XML_ATTRIBUTE_IMPLIED) && |
6202 | 35.0k | (def != XML_ATTRIBUTE_REQUIRED)) { |
6203 | 21.1k | xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue); |
6204 | 21.1k | } |
6205 | 35.0k | if (ctxt->sax2) { |
6206 | 31.8k | xmlAddSpecialAttr(ctxt, elemName, attrName, type); |
6207 | 31.8k | } |
6208 | 35.0k | if (defaultValue != NULL) |
6209 | 24.1k | xmlFree(defaultValue); |
6210 | 35.0k | GROW; |
6211 | 35.0k | } |
6212 | 29.4k | if (RAW == '>') { |
6213 | 9.04k | if (inputid != ctxt->input->id) { |
6214 | 197 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6215 | 197 | "Attribute list declaration doesn't start and" |
6216 | 197 | " stop in the same entity\n"); |
6217 | 197 | } |
6218 | 9.04k | NEXT; |
6219 | 9.04k | } |
6220 | 29.4k | } |
6221 | 30.0k | } |
6222 | | |
6223 | | /** |
6224 | | * parse the declaration for a Mixed Element content |
6225 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl() |
6226 | | * |
6227 | | * @deprecated Internal function, don't use. |
6228 | | * |
6229 | | * [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' | |
6230 | | * '(' S? '#PCDATA' S? ')' |
6231 | | * |
6232 | | * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49]) |
6233 | | * |
6234 | | * [ VC: No Duplicate Types ] |
6235 | | * The same name must not appear more than once in a single |
6236 | | * mixed-content declaration. |
6237 | | * |
6238 | | * @param ctxt an XML parser context |
6239 | | * @param inputchk the input used for the current entity, needed for boundary checks |
6240 | | * @returns the list of the xmlElementContentPtr describing the element choices |
6241 | | */ |
6242 | | xmlElementContentPtr |
6243 | 2.00k | xmlParseElementMixedContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6244 | 2.00k | xmlElementContentPtr ret = NULL, cur = NULL, n; |
6245 | 2.00k | const xmlChar *elem = NULL; |
6246 | | |
6247 | 2.00k | GROW; |
6248 | 2.00k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6249 | 2.00k | SKIP(7); |
6250 | 2.00k | SKIP_BLANKS_PE; |
6251 | 2.00k | if (RAW == ')') { |
6252 | 515 | if (ctxt->input->id != inputchk) { |
6253 | 66 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6254 | 66 | "Element content declaration doesn't start and" |
6255 | 66 | " stop in the same entity\n"); |
6256 | 66 | } |
6257 | 515 | NEXT; |
6258 | 515 | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6259 | 515 | if (ret == NULL) |
6260 | 0 | goto mem_error; |
6261 | 515 | if (RAW == '*') { |
6262 | 219 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6263 | 219 | NEXT; |
6264 | 219 | } |
6265 | 515 | return(ret); |
6266 | 515 | } |
6267 | 1.49k | if ((RAW == '(') || (RAW == '|')) { |
6268 | 1.12k | ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6269 | 1.12k | if (ret == NULL) |
6270 | 0 | goto mem_error; |
6271 | 1.12k | } |
6272 | 2.57k | while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) { |
6273 | 1.41k | NEXT; |
6274 | 1.41k | n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6275 | 1.41k | if (n == NULL) |
6276 | 0 | goto mem_error; |
6277 | 1.41k | if (elem == NULL) { |
6278 | 1.11k | n->c1 = cur; |
6279 | 1.11k | if (cur != NULL) |
6280 | 1.11k | cur->parent = n; |
6281 | 1.11k | ret = cur = n; |
6282 | 1.11k | } else { |
6283 | 295 | cur->c2 = n; |
6284 | 295 | n->parent = cur; |
6285 | 295 | n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6286 | 295 | if (n->c1 == NULL) |
6287 | 0 | goto mem_error; |
6288 | 295 | n->c1->parent = n; |
6289 | 295 | cur = n; |
6290 | 295 | } |
6291 | 1.41k | SKIP_BLANKS_PE; |
6292 | 1.41k | elem = xmlParseName(ctxt); |
6293 | 1.41k | if (elem == NULL) { |
6294 | 329 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6295 | 329 | "xmlParseElementMixedContentDecl : Name expected\n"); |
6296 | 329 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6297 | 329 | return(NULL); |
6298 | 329 | } |
6299 | 1.08k | SKIP_BLANKS_PE; |
6300 | 1.08k | GROW; |
6301 | 1.08k | } |
6302 | 1.16k | if ((RAW == ')') && (NXT(1) == '*')) { |
6303 | 570 | if (elem != NULL) { |
6304 | 570 | cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem, |
6305 | 570 | XML_ELEMENT_CONTENT_ELEMENT); |
6306 | 570 | if (cur->c2 == NULL) |
6307 | 0 | goto mem_error; |
6308 | 570 | cur->c2->parent = cur; |
6309 | 570 | } |
6310 | 570 | if (ret != NULL) |
6311 | 570 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6312 | 570 | if (ctxt->input->id != inputchk) { |
6313 | 10 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6314 | 10 | "Element content declaration doesn't start and" |
6315 | 10 | " stop in the same entity\n"); |
6316 | 10 | } |
6317 | 570 | SKIP(2); |
6318 | 593 | } else { |
6319 | 593 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6320 | 593 | xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL); |
6321 | 593 | return(NULL); |
6322 | 593 | } |
6323 | | |
6324 | 1.16k | } else { |
6325 | 0 | xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL); |
6326 | 0 | } |
6327 | 570 | return(ret); |
6328 | | |
6329 | 0 | mem_error: |
6330 | 0 | xmlErrMemory(ctxt); |
6331 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6332 | 0 | return(NULL); |
6333 | 2.00k | } |
6334 | | |
6335 | | /** |
6336 | | * parse the declaration for a Mixed Element content |
6337 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl() |
6338 | | * |
6339 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6340 | | * |
6341 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6342 | | * |
6343 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6344 | | * |
6345 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6346 | | * |
6347 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6348 | | * TODO Parameter-entity replacement text must be properly nested |
6349 | | * with parenthesized groups. That is to say, if either of the |
6350 | | * opening or closing parentheses in a choice, seq, or Mixed |
6351 | | * construct is contained in the replacement text for a parameter |
6352 | | * entity, both must be contained in the same replacement text. For |
6353 | | * interoperability, if a parameter-entity reference appears in a |
6354 | | * choice, seq, or Mixed construct, its replacement text should not |
6355 | | * be empty, and neither the first nor last non-blank character of |
6356 | | * the replacement text should be a connector (| or ,). |
6357 | | * |
6358 | | * @param ctxt an XML parser context |
6359 | | * @param inputchk the input used for the current entity, needed for boundary checks |
6360 | | * @param depth the level of recursion |
6361 | | * @returns the tree of xmlElementContentPtr describing the element |
6362 | | * hierarchy. |
6363 | | */ |
6364 | | static xmlElementContentPtr |
6365 | | xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int inputchk, |
6366 | 12.8k | int depth) { |
6367 | 12.8k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
6368 | 12.8k | xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL; |
6369 | 12.8k | const xmlChar *elem; |
6370 | 12.8k | xmlChar type = 0; |
6371 | | |
6372 | 12.8k | if (depth > maxDepth) { |
6373 | 1 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
6374 | 1 | "xmlParseElementChildrenContentDecl : depth %d too deep, " |
6375 | 1 | "use XML_PARSE_HUGE\n", depth); |
6376 | 1 | return(NULL); |
6377 | 1 | } |
6378 | 12.8k | SKIP_BLANKS_PE; |
6379 | 12.8k | GROW; |
6380 | 12.8k | if (RAW == '(') { |
6381 | 4.22k | int inputid = ctxt->input->id; |
6382 | | |
6383 | | /* Recurse on first child */ |
6384 | 4.22k | NEXT; |
6385 | 4.22k | SKIP_BLANKS_PE; |
6386 | 4.22k | cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6387 | 4.22k | depth + 1); |
6388 | 4.22k | if (cur == NULL) |
6389 | 3.16k | return(NULL); |
6390 | 1.06k | SKIP_BLANKS_PE; |
6391 | 1.06k | GROW; |
6392 | 8.59k | } else { |
6393 | 8.59k | elem = xmlParseName(ctxt); |
6394 | 8.59k | if (elem == NULL) { |
6395 | 727 | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6396 | 727 | return(NULL); |
6397 | 727 | } |
6398 | 7.87k | cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6399 | 7.87k | if (cur == NULL) { |
6400 | 0 | xmlErrMemory(ctxt); |
6401 | 0 | return(NULL); |
6402 | 0 | } |
6403 | 7.87k | GROW; |
6404 | 7.87k | if (RAW == '?') { |
6405 | 1.21k | cur->ocur = XML_ELEMENT_CONTENT_OPT; |
6406 | 1.21k | NEXT; |
6407 | 6.65k | } else if (RAW == '*') { |
6408 | 1.31k | cur->ocur = XML_ELEMENT_CONTENT_MULT; |
6409 | 1.31k | NEXT; |
6410 | 5.34k | } else if (RAW == '+') { |
6411 | 504 | cur->ocur = XML_ELEMENT_CONTENT_PLUS; |
6412 | 504 | NEXT; |
6413 | 4.83k | } else { |
6414 | 4.83k | cur->ocur = XML_ELEMENT_CONTENT_ONCE; |
6415 | 4.83k | } |
6416 | 7.87k | GROW; |
6417 | 7.87k | } |
6418 | 8.93k | SKIP_BLANKS_PE; |
6419 | 16.5k | while ((RAW != ')') && (PARSER_STOPPED(ctxt) == 0)) { |
6420 | | /* |
6421 | | * Each loop we parse one separator and one element. |
6422 | | */ |
6423 | 10.9k | if (RAW == ',') { |
6424 | 2.22k | if (type == 0) type = CUR; |
6425 | | |
6426 | | /* |
6427 | | * Detect "Name | Name , Name" error |
6428 | | */ |
6429 | 1.56k | else if (type != CUR) { |
6430 | 1 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6431 | 1 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6432 | 1 | type); |
6433 | 1 | if ((last != NULL) && (last != ret)) |
6434 | 1 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6435 | 1 | if (ret != NULL) |
6436 | 1 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6437 | 1 | return(NULL); |
6438 | 1 | } |
6439 | 2.22k | NEXT; |
6440 | | |
6441 | 2.22k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ); |
6442 | 2.22k | if (op == NULL) { |
6443 | 0 | xmlErrMemory(ctxt); |
6444 | 0 | if ((last != NULL) && (last != ret)) |
6445 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6446 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6447 | 0 | return(NULL); |
6448 | 0 | } |
6449 | 2.22k | if (last == NULL) { |
6450 | 656 | op->c1 = ret; |
6451 | 656 | if (ret != NULL) |
6452 | 656 | ret->parent = op; |
6453 | 656 | ret = cur = op; |
6454 | 1.56k | } else { |
6455 | 1.56k | cur->c2 = op; |
6456 | 1.56k | if (op != NULL) |
6457 | 1.56k | op->parent = cur; |
6458 | 1.56k | op->c1 = last; |
6459 | 1.56k | if (last != NULL) |
6460 | 1.56k | last->parent = op; |
6461 | 1.56k | cur =op; |
6462 | 1.56k | last = NULL; |
6463 | 1.56k | } |
6464 | 8.75k | } else if (RAW == '|') { |
6465 | 6.77k | if (type == 0) type = CUR; |
6466 | | |
6467 | | /* |
6468 | | * Detect "Name , Name | Name" error |
6469 | | */ |
6470 | 2.64k | else if (type != CUR) { |
6471 | 1 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6472 | 1 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6473 | 1 | type); |
6474 | 1 | if ((last != NULL) && (last != ret)) |
6475 | 1 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6476 | 1 | if (ret != NULL) |
6477 | 1 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6478 | 1 | return(NULL); |
6479 | 1 | } |
6480 | 6.76k | NEXT; |
6481 | | |
6482 | 6.76k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6483 | 6.76k | if (op == NULL) { |
6484 | 0 | xmlErrMemory(ctxt); |
6485 | 0 | if ((last != NULL) && (last != ret)) |
6486 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6487 | 0 | if (ret != NULL) |
6488 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6489 | 0 | return(NULL); |
6490 | 0 | } |
6491 | 6.76k | if (last == NULL) { |
6492 | 4.12k | op->c1 = ret; |
6493 | 4.12k | if (ret != NULL) |
6494 | 4.12k | ret->parent = op; |
6495 | 4.12k | ret = cur = op; |
6496 | 4.12k | } else { |
6497 | 2.64k | cur->c2 = op; |
6498 | 2.64k | if (op != NULL) |
6499 | 2.64k | op->parent = cur; |
6500 | 2.64k | op->c1 = last; |
6501 | 2.64k | if (last != NULL) |
6502 | 2.64k | last->parent = op; |
6503 | 2.64k | cur =op; |
6504 | 2.64k | last = NULL; |
6505 | 2.64k | } |
6506 | 6.76k | } else { |
6507 | 1.98k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL); |
6508 | 1.98k | if ((last != NULL) && (last != ret)) |
6509 | 1.14k | xmlFreeDocElementContent(ctxt->myDoc, last); |
6510 | 1.98k | if (ret != NULL) |
6511 | 1.98k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6512 | 1.98k | return(NULL); |
6513 | 1.98k | } |
6514 | 8.99k | GROW; |
6515 | 8.99k | SKIP_BLANKS_PE; |
6516 | 8.99k | GROW; |
6517 | 8.99k | if (RAW == '(') { |
6518 | 2.19k | int inputid = ctxt->input->id; |
6519 | | /* Recurse on second child */ |
6520 | 2.19k | NEXT; |
6521 | 2.19k | SKIP_BLANKS_PE; |
6522 | 2.19k | last = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6523 | 2.19k | depth + 1); |
6524 | 2.19k | if (last == NULL) { |
6525 | 1.07k | if (ret != NULL) |
6526 | 1.07k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6527 | 1.07k | return(NULL); |
6528 | 1.07k | } |
6529 | 1.11k | SKIP_BLANKS_PE; |
6530 | 6.79k | } else { |
6531 | 6.79k | elem = xmlParseName(ctxt); |
6532 | 6.79k | if (elem == NULL) { |
6533 | 335 | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6534 | 335 | if (ret != NULL) |
6535 | 335 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6536 | 335 | return(NULL); |
6537 | 335 | } |
6538 | 6.46k | last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6539 | 6.46k | if (last == NULL) { |
6540 | 0 | xmlErrMemory(ctxt); |
6541 | 0 | if (ret != NULL) |
6542 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6543 | 0 | return(NULL); |
6544 | 0 | } |
6545 | 6.46k | if (RAW == '?') { |
6546 | 995 | last->ocur = XML_ELEMENT_CONTENT_OPT; |
6547 | 995 | NEXT; |
6548 | 5.46k | } else if (RAW == '*') { |
6549 | 1.56k | last->ocur = XML_ELEMENT_CONTENT_MULT; |
6550 | 1.56k | NEXT; |
6551 | 3.90k | } else if (RAW == '+') { |
6552 | 201 | last->ocur = XML_ELEMENT_CONTENT_PLUS; |
6553 | 201 | NEXT; |
6554 | 3.70k | } else { |
6555 | 3.70k | last->ocur = XML_ELEMENT_CONTENT_ONCE; |
6556 | 3.70k | } |
6557 | 6.46k | } |
6558 | 7.57k | SKIP_BLANKS_PE; |
6559 | 7.57k | GROW; |
6560 | 7.57k | } |
6561 | 5.52k | if ((cur != NULL) && (last != NULL)) { |
6562 | 2.21k | cur->c2 = last; |
6563 | 2.21k | if (last != NULL) |
6564 | 2.21k | last->parent = cur; |
6565 | 2.21k | } |
6566 | 5.52k | if (ctxt->input->id != inputchk) { |
6567 | 66 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6568 | 66 | "Element content declaration doesn't start and stop in" |
6569 | 66 | " the same entity\n"); |
6570 | 66 | } |
6571 | 5.52k | NEXT; |
6572 | 5.52k | if (RAW == '?') { |
6573 | 822 | if (ret != NULL) { |
6574 | 822 | if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) || |
6575 | 822 | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6576 | 525 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6577 | 297 | else |
6578 | 297 | ret->ocur = XML_ELEMENT_CONTENT_OPT; |
6579 | 822 | } |
6580 | 822 | NEXT; |
6581 | 4.70k | } else if (RAW == '*') { |
6582 | 1.12k | if (ret != NULL) { |
6583 | 1.12k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6584 | 1.12k | cur = ret; |
6585 | | /* |
6586 | | * Some normalization: |
6587 | | * (a | b* | c?)* == (a | b | c)* |
6588 | | */ |
6589 | 2.79k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6590 | 1.67k | if ((cur->c1 != NULL) && |
6591 | 1.67k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6592 | 1.67k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) |
6593 | 578 | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6594 | 1.67k | if ((cur->c2 != NULL) && |
6595 | 1.67k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6596 | 1.67k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) |
6597 | 811 | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6598 | 1.67k | cur = cur->c2; |
6599 | 1.67k | } |
6600 | 1.12k | } |
6601 | 1.12k | NEXT; |
6602 | 3.58k | } else if (RAW == '+') { |
6603 | 2.11k | if (ret != NULL) { |
6604 | 2.11k | int found = 0; |
6605 | | |
6606 | 2.11k | if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) || |
6607 | 2.11k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6608 | 637 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6609 | 1.47k | else |
6610 | 1.47k | ret->ocur = XML_ELEMENT_CONTENT_PLUS; |
6611 | | /* |
6612 | | * Some normalization: |
6613 | | * (a | b*)+ == (a | b)* |
6614 | | * (a | b?)+ == (a | b)* |
6615 | | */ |
6616 | 3.71k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6617 | 1.60k | if ((cur->c1 != NULL) && |
6618 | 1.60k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6619 | 1.60k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6620 | 701 | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6621 | 701 | found = 1; |
6622 | 701 | } |
6623 | 1.60k | if ((cur->c2 != NULL) && |
6624 | 1.60k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6625 | 1.60k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6626 | 755 | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6627 | 755 | found = 1; |
6628 | 755 | } |
6629 | 1.60k | cur = cur->c2; |
6630 | 1.60k | } |
6631 | 2.11k | if (found) |
6632 | 885 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6633 | 2.11k | } |
6634 | 2.11k | NEXT; |
6635 | 2.11k | } |
6636 | 5.52k | return(ret); |
6637 | 8.93k | } |
6638 | | |
6639 | | /** |
6640 | | * parse the declaration for a Mixed Element content |
6641 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl() |
6642 | | * |
6643 | | * @deprecated Internal function, don't use. |
6644 | | * |
6645 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6646 | | * |
6647 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6648 | | * |
6649 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6650 | | * |
6651 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6652 | | * |
6653 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6654 | | * TODO Parameter-entity replacement text must be properly nested |
6655 | | * with parenthesized groups. That is to say, if either of the |
6656 | | * opening or closing parentheses in a choice, seq, or Mixed |
6657 | | * construct is contained in the replacement text for a parameter |
6658 | | * entity, both must be contained in the same replacement text. For |
6659 | | * interoperability, if a parameter-entity reference appears in a |
6660 | | * choice, seq, or Mixed construct, its replacement text should not |
6661 | | * be empty, and neither the first nor last non-blank character of |
6662 | | * the replacement text should be a connector (| or ,). |
6663 | | * |
6664 | | * @param ctxt an XML parser context |
6665 | | * @param inputchk the input used for the current entity, needed for boundary checks |
6666 | | * @returns the tree of xmlElementContentPtr describing the element |
6667 | | * hierarchy. |
6668 | | */ |
6669 | | xmlElementContentPtr |
6670 | 0 | xmlParseElementChildrenContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6671 | | /* stub left for API/ABI compat */ |
6672 | 0 | return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1)); |
6673 | 0 | } |
6674 | | |
6675 | | /** |
6676 | | * parse the declaration for an Element content either Mixed or Children, |
6677 | | * the cases EMPTY and ANY are handled directly in xmlParseElementDecl() |
6678 | | * |
6679 | | * @deprecated Internal function, don't use. |
6680 | | * |
6681 | | * [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children |
6682 | | * |
6683 | | * @param ctxt an XML parser context |
6684 | | * @param name the name of the element being defined. |
6685 | | * @param result the Element Content pointer will be stored here if any |
6686 | | * @returns the type of element content XML_ELEMENT_TYPE_xxx |
6687 | | */ |
6688 | | |
6689 | | int |
6690 | | xmlParseElementContentDecl(xmlParserCtxtPtr ctxt, const xmlChar *name, |
6691 | 8.41k | xmlElementContentPtr *result) { |
6692 | | |
6693 | 8.41k | xmlElementContentPtr tree = NULL; |
6694 | 8.41k | int inputid = ctxt->input->id; |
6695 | 8.41k | int res; |
6696 | | |
6697 | 8.41k | *result = NULL; |
6698 | | |
6699 | 8.41k | if (RAW != '(') { |
6700 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6701 | 0 | "xmlParseElementContentDecl : %s '(' expected\n", name); |
6702 | 0 | return(-1); |
6703 | 0 | } |
6704 | 8.41k | NEXT; |
6705 | 8.41k | GROW; |
6706 | 8.41k | SKIP_BLANKS_PE; |
6707 | 8.41k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6708 | 2.00k | tree = xmlParseElementMixedContentDecl(ctxt, inputid); |
6709 | 2.00k | res = XML_ELEMENT_TYPE_MIXED; |
6710 | 6.40k | } else { |
6711 | 6.40k | tree = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, 1); |
6712 | 6.40k | res = XML_ELEMENT_TYPE_ELEMENT; |
6713 | 6.40k | } |
6714 | 8.41k | SKIP_BLANKS_PE; |
6715 | 8.41k | *result = tree; |
6716 | 8.41k | return(res); |
6717 | 8.41k | } |
6718 | | |
6719 | | /** |
6720 | | * Parse an element declaration. Always consumes '<!'. |
6721 | | * |
6722 | | * @deprecated Internal function, don't use. |
6723 | | * |
6724 | | * [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>' |
6725 | | * |
6726 | | * [ VC: Unique Element Type Declaration ] |
6727 | | * No element type may be declared more than once |
6728 | | * |
6729 | | * @param ctxt an XML parser context |
6730 | | * @returns the type of the element, or -1 in case of error |
6731 | | */ |
6732 | | int |
6733 | 10.0k | xmlParseElementDecl(xmlParserCtxtPtr ctxt) { |
6734 | 10.0k | const xmlChar *name; |
6735 | 10.0k | int ret = -1; |
6736 | 10.0k | xmlElementContentPtr content = NULL; |
6737 | | |
6738 | 10.0k | if ((CUR != '<') || (NXT(1) != '!')) |
6739 | 0 | return(ret); |
6740 | 10.0k | SKIP(2); |
6741 | | |
6742 | | /* GROW; done in the caller */ |
6743 | 10.0k | if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) { |
6744 | 9.99k | int inputid = ctxt->input->id; |
6745 | | |
6746 | 9.99k | SKIP(7); |
6747 | 9.99k | if (SKIP_BLANKS_PE == 0) { |
6748 | 279 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6749 | 279 | "Space required after 'ELEMENT'\n"); |
6750 | 279 | return(-1); |
6751 | 279 | } |
6752 | 9.71k | name = xmlParseName(ctxt); |
6753 | 9.71k | if (name == NULL) { |
6754 | 243 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6755 | 243 | "xmlParseElementDecl: no name for Element\n"); |
6756 | 243 | return(-1); |
6757 | 243 | } |
6758 | 9.47k | if (SKIP_BLANKS_PE == 0) { |
6759 | 8.53k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6760 | 8.53k | "Space required after the element name\n"); |
6761 | 8.53k | } |
6762 | 9.47k | if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) { |
6763 | 401 | SKIP(5); |
6764 | | /* |
6765 | | * Element must always be empty. |
6766 | | */ |
6767 | 401 | ret = XML_ELEMENT_TYPE_EMPTY; |
6768 | 9.07k | } else if ((RAW == 'A') && (NXT(1) == 'N') && |
6769 | 9.07k | (NXT(2) == 'Y')) { |
6770 | 430 | SKIP(3); |
6771 | | /* |
6772 | | * Element is a generic container. |
6773 | | */ |
6774 | 430 | ret = XML_ELEMENT_TYPE_ANY; |
6775 | 8.64k | } else if (RAW == '(') { |
6776 | 8.41k | ret = xmlParseElementContentDecl(ctxt, name, &content); |
6777 | 8.41k | } else { |
6778 | | /* |
6779 | | * [ WFC: PEs in Internal Subset ] error handling. |
6780 | | */ |
6781 | 233 | xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6782 | 233 | "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n"); |
6783 | 233 | return(-1); |
6784 | 233 | } |
6785 | | |
6786 | 9.24k | SKIP_BLANKS_PE; |
6787 | | |
6788 | 9.24k | if (RAW != '>') { |
6789 | 4.21k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
6790 | 4.21k | if (content != NULL) { |
6791 | 1.64k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6792 | 1.64k | } |
6793 | 5.02k | } else { |
6794 | 5.02k | if (inputid != ctxt->input->id) { |
6795 | 192 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6796 | 192 | "Element declaration doesn't start and stop in" |
6797 | 192 | " the same entity\n"); |
6798 | 192 | } |
6799 | | |
6800 | 5.02k | NEXT; |
6801 | 5.02k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6802 | 5.02k | (ctxt->sax->elementDecl != NULL)) { |
6803 | 4.60k | if (content != NULL) |
6804 | 2.61k | content->parent = NULL; |
6805 | 4.60k | ctxt->sax->elementDecl(ctxt->userData, name, ret, |
6806 | 4.60k | content); |
6807 | 4.60k | if ((content != NULL) && (content->parent == NULL)) { |
6808 | | /* |
6809 | | * this is a trick: if xmlAddElementDecl is called, |
6810 | | * instead of copying the full tree it is plugged directly |
6811 | | * if called from the parser. Avoid duplicating the |
6812 | | * interfaces or change the API/ABI |
6813 | | */ |
6814 | 2.31k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6815 | 2.31k | } |
6816 | 4.60k | } else if (content != NULL) { |
6817 | 174 | xmlFreeDocElementContent(ctxt->myDoc, content); |
6818 | 174 | } |
6819 | 5.02k | } |
6820 | 9.24k | } |
6821 | 9.25k | return(ret); |
6822 | 10.0k | } |
6823 | | |
6824 | | /** |
6825 | | * Parse a conditional section. Always consumes '<!['. |
6826 | | * |
6827 | | * [61] conditionalSect ::= includeSect | ignoreSect |
6828 | | * [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>' |
6829 | | * [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>' |
6830 | | * [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' |
6831 | | * Ignore)* |
6832 | | * [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*) |
6833 | | * @param ctxt an XML parser context |
6834 | | */ |
6835 | | |
6836 | | static void |
6837 | 0 | xmlParseConditionalSections(xmlParserCtxtPtr ctxt) { |
6838 | 0 | int *inputIds = NULL; |
6839 | 0 | size_t inputIdsSize = 0; |
6840 | 0 | size_t depth = 0; |
6841 | |
|
6842 | 0 | while (PARSER_STOPPED(ctxt) == 0) { |
6843 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6844 | 0 | int id = ctxt->input->id; |
6845 | |
|
6846 | 0 | SKIP(3); |
6847 | 0 | SKIP_BLANKS_PE; |
6848 | |
|
6849 | 0 | if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) { |
6850 | 0 | SKIP(7); |
6851 | 0 | SKIP_BLANKS_PE; |
6852 | 0 | if (RAW != '[') { |
6853 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6854 | 0 | xmlHaltParser(ctxt); |
6855 | 0 | goto error; |
6856 | 0 | } |
6857 | 0 | if (ctxt->input->id != id) { |
6858 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6859 | 0 | "All markup of the conditional section is" |
6860 | 0 | " not in the same entity\n"); |
6861 | 0 | } |
6862 | 0 | NEXT; |
6863 | |
|
6864 | 0 | if (inputIdsSize <= depth) { |
6865 | 0 | int *tmp; |
6866 | 0 | int newSize; |
6867 | |
|
6868 | 0 | newSize = xmlGrowCapacity(inputIdsSize, sizeof(tmp[0]), |
6869 | 0 | 4, 1000); |
6870 | 0 | if (newSize < 0) { |
6871 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
6872 | 0 | "Maximum conditional section nesting" |
6873 | 0 | " depth exceeded\n"); |
6874 | 0 | goto error; |
6875 | 0 | } |
6876 | 0 | tmp = xmlRealloc(inputIds, newSize * sizeof(tmp[0])); |
6877 | 0 | if (tmp == NULL) { |
6878 | 0 | xmlErrMemory(ctxt); |
6879 | 0 | goto error; |
6880 | 0 | } |
6881 | 0 | inputIds = tmp; |
6882 | 0 | inputIdsSize = newSize; |
6883 | 0 | } |
6884 | 0 | inputIds[depth] = id; |
6885 | 0 | depth++; |
6886 | 0 | } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) { |
6887 | 0 | size_t ignoreDepth = 0; |
6888 | |
|
6889 | 0 | SKIP(6); |
6890 | 0 | SKIP_BLANKS_PE; |
6891 | 0 | if (RAW != '[') { |
6892 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6893 | 0 | xmlHaltParser(ctxt); |
6894 | 0 | goto error; |
6895 | 0 | } |
6896 | 0 | if (ctxt->input->id != id) { |
6897 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6898 | 0 | "All markup of the conditional section is" |
6899 | 0 | " not in the same entity\n"); |
6900 | 0 | } |
6901 | 0 | NEXT; |
6902 | |
|
6903 | 0 | while (PARSER_STOPPED(ctxt) == 0) { |
6904 | 0 | if (RAW == 0) { |
6905 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL); |
6906 | 0 | goto error; |
6907 | 0 | } |
6908 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6909 | 0 | SKIP(3); |
6910 | 0 | ignoreDepth++; |
6911 | | /* Check for integer overflow */ |
6912 | 0 | if (ignoreDepth == 0) { |
6913 | 0 | xmlErrMemory(ctxt); |
6914 | 0 | goto error; |
6915 | 0 | } |
6916 | 0 | } else if ((RAW == ']') && (NXT(1) == ']') && |
6917 | 0 | (NXT(2) == '>')) { |
6918 | 0 | SKIP(3); |
6919 | 0 | if (ignoreDepth == 0) |
6920 | 0 | break; |
6921 | 0 | ignoreDepth--; |
6922 | 0 | } else { |
6923 | 0 | NEXT; |
6924 | 0 | } |
6925 | 0 | } |
6926 | | |
6927 | 0 | if (ctxt->input->id != id) { |
6928 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6929 | 0 | "All markup of the conditional section is" |
6930 | 0 | " not in the same entity\n"); |
6931 | 0 | } |
6932 | 0 | } else { |
6933 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL); |
6934 | 0 | xmlHaltParser(ctxt); |
6935 | 0 | goto error; |
6936 | 0 | } |
6937 | 0 | } else if ((depth > 0) && |
6938 | 0 | (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
6939 | 0 | depth--; |
6940 | 0 | if (ctxt->input->id != inputIds[depth]) { |
6941 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6942 | 0 | "All markup of the conditional section is not" |
6943 | 0 | " in the same entity\n"); |
6944 | 0 | } |
6945 | 0 | SKIP(3); |
6946 | 0 | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
6947 | 0 | xmlParseMarkupDecl(ctxt); |
6948 | 0 | } else { |
6949 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
6950 | 0 | xmlHaltParser(ctxt); |
6951 | 0 | goto error; |
6952 | 0 | } |
6953 | | |
6954 | 0 | if (depth == 0) |
6955 | 0 | break; |
6956 | | |
6957 | 0 | SKIP_BLANKS_PE; |
6958 | 0 | SHRINK; |
6959 | 0 | GROW; |
6960 | 0 | } |
6961 | | |
6962 | 0 | error: |
6963 | 0 | xmlFree(inputIds); |
6964 | 0 | } |
6965 | | |
6966 | | /** |
6967 | | * Parse markup declarations. Always consumes '<!' or '<?'. |
6968 | | * |
6969 | | * @deprecated Internal function, don't use. |
6970 | | * |
6971 | | * [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl | |
6972 | | * NotationDecl | PI | Comment |
6973 | | * |
6974 | | * [ VC: Proper Declaration/PE Nesting ] |
6975 | | * Parameter-entity replacement text must be properly nested with |
6976 | | * markup declarations. That is to say, if either the first character |
6977 | | * or the last character of a markup declaration (markupdecl above) is |
6978 | | * contained in the replacement text for a parameter-entity reference, |
6979 | | * both must be contained in the same replacement text. |
6980 | | * |
6981 | | * [ WFC: PEs in Internal Subset ] |
6982 | | * In the internal DTD subset, parameter-entity references can occur |
6983 | | * only where markup declarations can occur, not within markup declarations. |
6984 | | * (This does not apply to references that occur in external parameter |
6985 | | * entities or to the external subset.) |
6986 | | * |
6987 | | * @param ctxt an XML parser context |
6988 | | */ |
6989 | | void |
6990 | 96.9k | xmlParseMarkupDecl(xmlParserCtxtPtr ctxt) { |
6991 | 96.9k | GROW; |
6992 | 96.9k | if (CUR == '<') { |
6993 | 96.9k | if (NXT(1) == '!') { |
6994 | 74.9k | switch (NXT(2)) { |
6995 | 29.3k | case 'E': |
6996 | 29.3k | if (NXT(3) == 'L') |
6997 | 10.0k | xmlParseElementDecl(ctxt); |
6998 | 19.3k | else if (NXT(3) == 'N') |
6999 | 19.3k | xmlParseEntityDecl(ctxt); |
7000 | 11 | else |
7001 | 11 | SKIP(2); |
7002 | 29.3k | break; |
7003 | 30.0k | case 'A': |
7004 | 30.0k | xmlParseAttributeListDecl(ctxt); |
7005 | 30.0k | break; |
7006 | 4.74k | case 'N': |
7007 | 4.74k | xmlParseNotationDecl(ctxt); |
7008 | 4.74k | break; |
7009 | 9.69k | case '-': |
7010 | 9.69k | xmlParseComment(ctxt); |
7011 | 9.69k | break; |
7012 | 1.21k | default: |
7013 | 1.21k | xmlFatalErr(ctxt, |
7014 | 1.21k | ctxt->inSubset == 2 ? |
7015 | 0 | XML_ERR_EXT_SUBSET_NOT_FINISHED : |
7016 | 1.21k | XML_ERR_INT_SUBSET_NOT_FINISHED, |
7017 | 1.21k | NULL); |
7018 | 1.21k | SKIP(2); |
7019 | 1.21k | break; |
7020 | 74.9k | } |
7021 | 74.9k | } else if (NXT(1) == '?') { |
7022 | 21.9k | xmlParsePI(ctxt); |
7023 | 21.9k | } |
7024 | 96.9k | } |
7025 | 96.9k | } |
7026 | | |
7027 | | /** |
7028 | | * parse an XML declaration header for external entities |
7029 | | * |
7030 | | * @deprecated Internal function, don't use. |
7031 | | * |
7032 | | * [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>' |
7033 | | * @param ctxt an XML parser context |
7034 | | */ |
7035 | | |
7036 | | void |
7037 | 0 | xmlParseTextDecl(xmlParserCtxtPtr ctxt) { |
7038 | 0 | xmlChar *version; |
7039 | | |
7040 | | /* |
7041 | | * We know that '<?xml' is here. |
7042 | | */ |
7043 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
7044 | 0 | SKIP(5); |
7045 | 0 | } else { |
7046 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL); |
7047 | 0 | return; |
7048 | 0 | } |
7049 | | |
7050 | 0 | if (SKIP_BLANKS == 0) { |
7051 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
7052 | 0 | "Space needed after '<?xml'\n"); |
7053 | 0 | } |
7054 | | |
7055 | | /* |
7056 | | * We may have the VersionInfo here. |
7057 | | */ |
7058 | 0 | version = xmlParseVersionInfo(ctxt); |
7059 | 0 | if (version == NULL) { |
7060 | 0 | version = xmlCharStrdup(XML_DEFAULT_VERSION); |
7061 | 0 | if (version == NULL) { |
7062 | 0 | xmlErrMemory(ctxt); |
7063 | 0 | return; |
7064 | 0 | } |
7065 | 0 | } else { |
7066 | 0 | if (SKIP_BLANKS == 0) { |
7067 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
7068 | 0 | "Space needed here\n"); |
7069 | 0 | } |
7070 | 0 | } |
7071 | 0 | ctxt->input->version = version; |
7072 | | |
7073 | | /* |
7074 | | * We must have the encoding declaration |
7075 | | */ |
7076 | 0 | xmlParseEncodingDecl(ctxt); |
7077 | |
|
7078 | 0 | SKIP_BLANKS; |
7079 | 0 | if ((RAW == '?') && (NXT(1) == '>')) { |
7080 | 0 | SKIP(2); |
7081 | 0 | } else if (RAW == '>') { |
7082 | | /* Deprecated old WD ... */ |
7083 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7084 | 0 | NEXT; |
7085 | 0 | } else { |
7086 | 0 | int c; |
7087 | |
|
7088 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7089 | 0 | while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) { |
7090 | 0 | NEXT; |
7091 | 0 | if (c == '>') |
7092 | 0 | break; |
7093 | 0 | } |
7094 | 0 | } |
7095 | 0 | } |
7096 | | |
7097 | | /** |
7098 | | * parse Markup declarations from an external subset |
7099 | | * |
7100 | | * @deprecated Internal function, don't use. |
7101 | | * |
7102 | | * [30] extSubset ::= textDecl? extSubsetDecl |
7103 | | * |
7104 | | * [31] extSubsetDecl ::= (markupdecl | conditionalSect | |
7105 | | * PEReference | S) * |
7106 | | * @param ctxt an XML parser context |
7107 | | * @param ExternalID the external identifier |
7108 | | * @param SystemID the system identifier (or URL) |
7109 | | */ |
7110 | | void |
7111 | | xmlParseExternalSubset(xmlParserCtxtPtr ctxt, const xmlChar *ExternalID, |
7112 | 0 | const xmlChar *SystemID) { |
7113 | 0 | int oldInputNr; |
7114 | |
|
7115 | 0 | xmlCtxtInitializeLate(ctxt); |
7116 | |
|
7117 | 0 | xmlDetectEncoding(ctxt); |
7118 | |
|
7119 | 0 | if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) { |
7120 | 0 | xmlParseTextDecl(ctxt); |
7121 | 0 | } |
7122 | 0 | if (ctxt->myDoc == NULL) { |
7123 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
7124 | 0 | if (ctxt->myDoc == NULL) { |
7125 | 0 | xmlErrMemory(ctxt); |
7126 | 0 | return; |
7127 | 0 | } |
7128 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
7129 | 0 | } |
7130 | 0 | if ((ctxt->myDoc != NULL) && (ctxt->myDoc->intSubset == NULL) && |
7131 | 0 | (xmlCreateIntSubset(ctxt->myDoc, NULL, ExternalID, SystemID) == NULL)) { |
7132 | 0 | xmlErrMemory(ctxt); |
7133 | 0 | } |
7134 | |
|
7135 | 0 | ctxt->inSubset = 2; |
7136 | 0 | oldInputNr = ctxt->inputNr; |
7137 | |
|
7138 | 0 | SKIP_BLANKS_PE; |
7139 | 0 | while (((RAW != 0) || (ctxt->inputNr > oldInputNr)) && |
7140 | 0 | (!PARSER_STOPPED(ctxt))) { |
7141 | 0 | GROW; |
7142 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7143 | 0 | xmlParseConditionalSections(ctxt); |
7144 | 0 | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
7145 | 0 | xmlParseMarkupDecl(ctxt); |
7146 | 0 | } else { |
7147 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7148 | 0 | xmlHaltParser(ctxt); |
7149 | 0 | return; |
7150 | 0 | } |
7151 | 0 | SKIP_BLANKS_PE; |
7152 | 0 | SHRINK; |
7153 | 0 | } |
7154 | | |
7155 | 0 | while (ctxt->inputNr > oldInputNr) |
7156 | 0 | xmlPopPE(ctxt); |
7157 | |
|
7158 | 0 | xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED); |
7159 | 0 | } |
7160 | | |
7161 | | /** |
7162 | | * parse and handle entity references in content, depending on the SAX |
7163 | | * interface, this may end-up in a call to character() if this is a |
7164 | | * CharRef, a predefined entity, if there is no reference() callback. |
7165 | | * or if the parser was asked to switch to that mode. |
7166 | | * |
7167 | | * @deprecated Internal function, don't use. |
7168 | | * |
7169 | | * Always consumes '&'. |
7170 | | * |
7171 | | * [67] Reference ::= EntityRef | CharRef |
7172 | | * @param ctxt an XML parser context |
7173 | | */ |
7174 | | void |
7175 | 12.8k | xmlParseReference(xmlParserCtxtPtr ctxt) { |
7176 | 12.8k | xmlEntityPtr ent = NULL; |
7177 | 12.8k | const xmlChar *name; |
7178 | 12.8k | xmlChar *val; |
7179 | | |
7180 | 12.8k | if (RAW != '&') |
7181 | 0 | return; |
7182 | | |
7183 | | /* |
7184 | | * Simple case of a CharRef |
7185 | | */ |
7186 | 12.8k | if (NXT(1) == '#') { |
7187 | 4.17k | int i = 0; |
7188 | 4.17k | xmlChar out[16]; |
7189 | 4.17k | int value = xmlParseCharRef(ctxt); |
7190 | | |
7191 | 4.17k | if (value == 0) |
7192 | 2.59k | return; |
7193 | | |
7194 | | /* |
7195 | | * Just encode the value in UTF-8 |
7196 | | */ |
7197 | 1.58k | COPY_BUF(out, i, value); |
7198 | 1.58k | out[i] = 0; |
7199 | 1.58k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7200 | 1.58k | (!ctxt->disableSAX)) |
7201 | 760 | ctxt->sax->characters(ctxt->userData, out, i); |
7202 | 1.58k | return; |
7203 | 4.17k | } |
7204 | | |
7205 | | /* |
7206 | | * We are seeing an entity reference |
7207 | | */ |
7208 | 8.67k | name = xmlParseEntityRefInternal(ctxt); |
7209 | 8.67k | if (name == NULL) |
7210 | 2.71k | return; |
7211 | 5.95k | ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0); |
7212 | 5.95k | if (ent == NULL) { |
7213 | | /* |
7214 | | * Create a reference for undeclared entities. |
7215 | | */ |
7216 | 2.45k | if ((ctxt->replaceEntities == 0) && |
7217 | 2.45k | (ctxt->sax != NULL) && |
7218 | 2.45k | (ctxt->disableSAX == 0) && |
7219 | 2.45k | (ctxt->sax->reference != NULL)) { |
7220 | 1.65k | ctxt->sax->reference(ctxt->userData, name); |
7221 | 1.65k | } |
7222 | 2.45k | return; |
7223 | 2.45k | } |
7224 | 3.50k | if (!ctxt->wellFormed) |
7225 | 917 | return; |
7226 | | |
7227 | | /* special case of predefined entities */ |
7228 | 2.58k | if ((ent->name == NULL) || |
7229 | 2.58k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
7230 | 252 | val = ent->content; |
7231 | 252 | if (val == NULL) return; |
7232 | | /* |
7233 | | * inline the entity. |
7234 | | */ |
7235 | 252 | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7236 | 252 | (!ctxt->disableSAX)) |
7237 | 252 | ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val)); |
7238 | 252 | return; |
7239 | 252 | } |
7240 | | |
7241 | | /* |
7242 | | * Some users try to parse entities on their own and used to set |
7243 | | * the renamed "checked" member. Fix the flags to cover this |
7244 | | * case. |
7245 | | */ |
7246 | 2.33k | if (((ent->flags & XML_ENT_PARSED) == 0) && (ent->children != NULL)) |
7247 | 0 | ent->flags |= XML_ENT_PARSED; |
7248 | | |
7249 | | /* |
7250 | | * The first reference to the entity trigger a parsing phase |
7251 | | * where the ent->children is filled with the result from |
7252 | | * the parsing. |
7253 | | * Note: external parsed entities will not be loaded, it is not |
7254 | | * required for a non-validating parser, unless the parsing option |
7255 | | * of validating, or substituting entities were given. Doing so is |
7256 | | * far more secure as the parser will only process data coming from |
7257 | | * the document entity by default. |
7258 | | * |
7259 | | * FIXME: This doesn't work correctly since entities can be |
7260 | | * expanded with different namespace declarations in scope. |
7261 | | * For example: |
7262 | | * |
7263 | | * <!DOCTYPE doc [ |
7264 | | * <!ENTITY ent "<ns:elem/>"> |
7265 | | * ]> |
7266 | | * <doc> |
7267 | | * <decl1 xmlns:ns="urn:ns1"> |
7268 | | * &ent; |
7269 | | * </decl1> |
7270 | | * <decl2 xmlns:ns="urn:ns2"> |
7271 | | * &ent; |
7272 | | * </decl2> |
7273 | | * </doc> |
7274 | | * |
7275 | | * Proposed fix: |
7276 | | * |
7277 | | * - Ignore current namespace declarations when parsing the |
7278 | | * entity. If a prefix can't be resolved, don't report an error |
7279 | | * but mark it as unresolved. |
7280 | | * - Try to resolve these prefixes when expanding the entity. |
7281 | | * This will require a specialized version of xmlStaticCopyNode |
7282 | | * which can also make use of the namespace hash table to avoid |
7283 | | * quadratic behavior. |
7284 | | * |
7285 | | * Alternatively, we could simply reparse the entity on each |
7286 | | * expansion like we already do with custom SAX callbacks. |
7287 | | * External entity content should be cached in this case. |
7288 | | */ |
7289 | 2.33k | if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) || |
7290 | 2.33k | (((ctxt->options & XML_PARSE_NO_XXE) == 0) && |
7291 | 388 | ((ctxt->replaceEntities) || |
7292 | 1.94k | (ctxt->validate)))) { |
7293 | 1.94k | if ((ent->flags & XML_ENT_PARSED) == 0) { |
7294 | 570 | xmlCtxtParseEntity(ctxt, ent); |
7295 | 1.37k | } else if (ent->children == NULL) { |
7296 | | /* |
7297 | | * Probably running in SAX mode and the callbacks don't |
7298 | | * build the entity content. Parse the entity again. |
7299 | | * |
7300 | | * This will also be triggered in normal tree builder mode |
7301 | | * if an entity happens to be empty, causing unnecessary |
7302 | | * reloads. It's hard to come up with a reliable check in |
7303 | | * which mode we're running. |
7304 | | */ |
7305 | 472 | xmlCtxtParseEntity(ctxt, ent); |
7306 | 472 | } |
7307 | 1.94k | } |
7308 | | |
7309 | | /* |
7310 | | * We also check for amplification if entities aren't substituted. |
7311 | | * They might be expanded later. |
7312 | | */ |
7313 | 2.33k | if (xmlParserEntityCheck(ctxt, ent->expandedSize)) |
7314 | 0 | return; |
7315 | | |
7316 | 2.33k | if ((ctxt->sax == NULL) || (ctxt->disableSAX)) |
7317 | 190 | return; |
7318 | | |
7319 | 2.14k | if (ctxt->replaceEntities == 0) { |
7320 | | /* |
7321 | | * Create a reference |
7322 | | */ |
7323 | 2.14k | if (ctxt->sax->reference != NULL) |
7324 | 2.14k | ctxt->sax->reference(ctxt->userData, ent->name); |
7325 | 2.14k | } else if ((ent->children != NULL) && (ctxt->node != NULL)) { |
7326 | 0 | xmlNodePtr copy, cur; |
7327 | | |
7328 | | /* |
7329 | | * Seems we are generating the DOM content, copy the tree |
7330 | | */ |
7331 | 0 | cur = ent->children; |
7332 | | |
7333 | | /* |
7334 | | * Handle first text node with SAX to coalesce text efficiently |
7335 | | */ |
7336 | 0 | if ((cur->type == XML_TEXT_NODE) || |
7337 | 0 | (cur->type == XML_CDATA_SECTION_NODE)) { |
7338 | 0 | int len = xmlStrlen(cur->content); |
7339 | |
|
7340 | 0 | if ((cur->type == XML_TEXT_NODE) || |
7341 | 0 | (ctxt->options & XML_PARSE_NOCDATA)) { |
7342 | 0 | if (ctxt->sax->characters != NULL) |
7343 | 0 | ctxt->sax->characters(ctxt, cur->content, len); |
7344 | 0 | } else { |
7345 | 0 | if (ctxt->sax->cdataBlock != NULL) |
7346 | 0 | ctxt->sax->cdataBlock(ctxt, cur->content, len); |
7347 | 0 | } |
7348 | |
|
7349 | 0 | cur = cur->next; |
7350 | 0 | } |
7351 | |
|
7352 | 0 | while (cur != NULL) { |
7353 | 0 | xmlNodePtr last; |
7354 | | |
7355 | | /* |
7356 | | * Handle last text node with SAX to coalesce text efficiently |
7357 | | */ |
7358 | 0 | if ((cur->next == NULL) && |
7359 | 0 | ((cur->type == XML_TEXT_NODE) || |
7360 | 0 | (cur->type == XML_CDATA_SECTION_NODE))) { |
7361 | 0 | int len = xmlStrlen(cur->content); |
7362 | |
|
7363 | 0 | if ((cur->type == XML_TEXT_NODE) || |
7364 | 0 | (ctxt->options & XML_PARSE_NOCDATA)) { |
7365 | 0 | if (ctxt->sax->characters != NULL) |
7366 | 0 | ctxt->sax->characters(ctxt, cur->content, len); |
7367 | 0 | } else { |
7368 | 0 | if (ctxt->sax->cdataBlock != NULL) |
7369 | 0 | ctxt->sax->cdataBlock(ctxt, cur->content, len); |
7370 | 0 | } |
7371 | |
|
7372 | 0 | break; |
7373 | 0 | } |
7374 | | |
7375 | | /* |
7376 | | * Reset coalesce buffer stats only for non-text nodes. |
7377 | | */ |
7378 | 0 | ctxt->nodemem = 0; |
7379 | 0 | ctxt->nodelen = 0; |
7380 | |
|
7381 | 0 | copy = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7382 | |
|
7383 | 0 | if (copy == NULL) { |
7384 | 0 | xmlErrMemory(ctxt); |
7385 | 0 | break; |
7386 | 0 | } |
7387 | | |
7388 | 0 | if (ctxt->parseMode == XML_PARSE_READER) { |
7389 | | /* Needed for reader */ |
7390 | 0 | copy->extra = cur->extra; |
7391 | | /* Maybe needed for reader */ |
7392 | 0 | copy->_private = cur->_private; |
7393 | 0 | } |
7394 | |
|
7395 | 0 | copy->parent = ctxt->node; |
7396 | 0 | last = ctxt->node->last; |
7397 | 0 | if (last == NULL) { |
7398 | 0 | ctxt->node->children = copy; |
7399 | 0 | } else { |
7400 | 0 | last->next = copy; |
7401 | 0 | copy->prev = last; |
7402 | 0 | } |
7403 | 0 | ctxt->node->last = copy; |
7404 | |
|
7405 | 0 | cur = cur->next; |
7406 | 0 | } |
7407 | 0 | } |
7408 | 2.14k | } |
7409 | | |
7410 | | static void |
7411 | 45.1k | xmlHandleUndeclaredEntity(xmlParserCtxtPtr ctxt, const xmlChar *name) { |
7412 | | /* |
7413 | | * [ WFC: Entity Declared ] |
7414 | | * In a document without any DTD, a document with only an |
7415 | | * internal DTD subset which contains no parameter entity |
7416 | | * references, or a document with "standalone='yes'", the |
7417 | | * Name given in the entity reference must match that in an |
7418 | | * entity declaration, except that well-formed documents |
7419 | | * need not declare any of the following entities: amp, lt, |
7420 | | * gt, apos, quot. |
7421 | | * The declaration of a parameter entity must precede any |
7422 | | * reference to it. |
7423 | | * Similarly, the declaration of a general entity must |
7424 | | * precede any reference to it which appears in a default |
7425 | | * value in an attribute-list declaration. Note that if |
7426 | | * entities are declared in the external subset or in |
7427 | | * external parameter entities, a non-validating processor |
7428 | | * is not obligated to read and process their declarations; |
7429 | | * for such documents, the rule that an entity must be |
7430 | | * declared is a well-formedness constraint only if |
7431 | | * standalone='yes'. |
7432 | | */ |
7433 | 45.1k | if ((ctxt->standalone == 1) || |
7434 | 45.1k | ((ctxt->hasExternalSubset == 0) && |
7435 | 44.7k | (ctxt->hasPErefs == 0))) { |
7436 | 31.4k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7437 | 31.4k | "Entity '%s' not defined\n", name); |
7438 | 31.4k | } else if (ctxt->validate) { |
7439 | | /* |
7440 | | * [ VC: Entity Declared ] |
7441 | | * In a document with an external subset or external |
7442 | | * parameter entities with "standalone='no'", ... |
7443 | | * ... The declaration of a parameter entity must |
7444 | | * precede any reference to it... |
7445 | | */ |
7446 | 0 | xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7447 | 0 | "Entity '%s' not defined\n", name, NULL); |
7448 | 13.7k | } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) || |
7449 | 13.7k | ((ctxt->replaceEntities) && |
7450 | 13.7k | ((ctxt->options & XML_PARSE_NO_XXE) == 0))) { |
7451 | | /* |
7452 | | * Also raise a non-fatal error |
7453 | | * |
7454 | | * - if the external subset is loaded and all entity declarations |
7455 | | * should be available, or |
7456 | | * - entity substition was requested without restricting |
7457 | | * external entity access. |
7458 | | */ |
7459 | 0 | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7460 | 0 | "Entity '%s' not defined\n", name); |
7461 | 13.7k | } else { |
7462 | 13.7k | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7463 | 13.7k | "Entity '%s' not defined\n", name, NULL); |
7464 | 13.7k | } |
7465 | | |
7466 | 45.1k | ctxt->valid = 0; |
7467 | 45.1k | } |
7468 | | |
7469 | | static xmlEntityPtr |
7470 | 657k | xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) { |
7471 | 657k | xmlEntityPtr ent = NULL; |
7472 | | |
7473 | | /* |
7474 | | * Predefined entities override any extra definition |
7475 | | */ |
7476 | 657k | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7477 | 656k | ent = xmlGetPredefinedEntity(name); |
7478 | 656k | if (ent != NULL) |
7479 | 39.3k | return(ent); |
7480 | 656k | } |
7481 | | |
7482 | | /* |
7483 | | * Ask first SAX for entity resolution, otherwise try the |
7484 | | * entities which may have stored in the parser context. |
7485 | | */ |
7486 | 617k | if (ctxt->sax != NULL) { |
7487 | 617k | if (ctxt->sax->getEntity != NULL) |
7488 | 617k | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7489 | 617k | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7490 | 617k | (ctxt->options & XML_PARSE_OLDSAX)) |
7491 | 194 | ent = xmlGetPredefinedEntity(name); |
7492 | 617k | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7493 | 617k | (ctxt->userData==ctxt)) { |
7494 | 1.75k | ent = xmlSAX2GetEntity(ctxt, name); |
7495 | 1.75k | } |
7496 | 617k | } |
7497 | | |
7498 | 617k | if (ent == NULL) { |
7499 | 43.5k | xmlHandleUndeclaredEntity(ctxt, name); |
7500 | 43.5k | } |
7501 | | |
7502 | | /* |
7503 | | * [ WFC: Parsed Entity ] |
7504 | | * An entity reference must not contain the name of an |
7505 | | * unparsed entity |
7506 | | */ |
7507 | 574k | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7508 | 196 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7509 | 196 | "Entity reference to unparsed entity %s\n", name); |
7510 | 196 | ent = NULL; |
7511 | 196 | } |
7512 | | |
7513 | | /* |
7514 | | * [ WFC: No External Entity References ] |
7515 | | * Attribute values cannot contain direct or indirect |
7516 | | * entity references to external entities. |
7517 | | */ |
7518 | 573k | else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7519 | 1.18k | if (inAttr) { |
7520 | 782 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7521 | 782 | "Attribute references external entity '%s'\n", name); |
7522 | 782 | ent = NULL; |
7523 | 782 | } |
7524 | 1.18k | } |
7525 | | |
7526 | 617k | return(ent); |
7527 | 657k | } |
7528 | | |
7529 | | /** |
7530 | | * Parse an entity reference. Always consumes '&'. |
7531 | | * |
7532 | | * [68] EntityRef ::= '&' Name ';' |
7533 | | * |
7534 | | * @param ctxt an XML parser context |
7535 | | * @returns the name, or NULL in case of error. |
7536 | | */ |
7537 | | static const xmlChar * |
7538 | 81.4k | xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) { |
7539 | 81.4k | const xmlChar *name; |
7540 | | |
7541 | 81.4k | GROW; |
7542 | | |
7543 | 81.4k | if (RAW != '&') |
7544 | 0 | return(NULL); |
7545 | 81.4k | NEXT; |
7546 | 81.4k | name = xmlParseName(ctxt); |
7547 | 81.4k | if (name == NULL) { |
7548 | 5.40k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7549 | 5.40k | "xmlParseEntityRef: no name\n"); |
7550 | 5.40k | return(NULL); |
7551 | 5.40k | } |
7552 | 76.0k | if (RAW != ';') { |
7553 | 2.16k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7554 | 2.16k | return(NULL); |
7555 | 2.16k | } |
7556 | 73.8k | NEXT; |
7557 | | |
7558 | 73.8k | return(name); |
7559 | 76.0k | } |
7560 | | |
7561 | | /** |
7562 | | * @deprecated Internal function, don't use. |
7563 | | * |
7564 | | * @param ctxt an XML parser context |
7565 | | * @returns the xmlEntityPtr if found, or NULL otherwise. |
7566 | | */ |
7567 | | xmlEntityPtr |
7568 | 0 | xmlParseEntityRef(xmlParserCtxtPtr ctxt) { |
7569 | 0 | const xmlChar *name; |
7570 | |
|
7571 | 0 | if (ctxt == NULL) |
7572 | 0 | return(NULL); |
7573 | | |
7574 | 0 | name = xmlParseEntityRefInternal(ctxt); |
7575 | 0 | if (name == NULL) |
7576 | 0 | return(NULL); |
7577 | | |
7578 | 0 | return(xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 0)); |
7579 | 0 | } |
7580 | | |
7581 | | /** |
7582 | | * parse ENTITY references declarations, but this version parses it from |
7583 | | * a string value. |
7584 | | * |
7585 | | * [68] EntityRef ::= '&' Name ';' |
7586 | | * |
7587 | | * [ WFC: Entity Declared ] |
7588 | | * In a document without any DTD, a document with only an internal DTD |
7589 | | * subset which contains no parameter entity references, or a document |
7590 | | * with "standalone='yes'", the Name given in the entity reference |
7591 | | * must match that in an entity declaration, except that well-formed |
7592 | | * documents need not declare any of the following entities: amp, lt, |
7593 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7594 | | * any reference to it. Similarly, the declaration of a general entity |
7595 | | * must precede any reference to it which appears in a default value in an |
7596 | | * attribute-list declaration. Note that if entities are declared in the |
7597 | | * external subset or in external parameter entities, a non-validating |
7598 | | * processor is not obligated to read and process their declarations; |
7599 | | * for such documents, the rule that an entity must be declared is a |
7600 | | * well-formedness constraint only if standalone='yes'. |
7601 | | * |
7602 | | * [ WFC: Parsed Entity ] |
7603 | | * An entity reference must not contain the name of an unparsed entity |
7604 | | * |
7605 | | * @param ctxt an XML parser context |
7606 | | * @param str a pointer to an index in the string |
7607 | | * @returns the xmlEntityPtr if found, or NULL otherwise. The str pointer |
7608 | | * is updated to the current location in the string. |
7609 | | */ |
7610 | | static xmlChar * |
7611 | 583k | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) { |
7612 | 583k | xmlChar *name; |
7613 | 583k | const xmlChar *ptr; |
7614 | 583k | xmlChar cur; |
7615 | | |
7616 | 583k | if ((str == NULL) || (*str == NULL)) |
7617 | 0 | return(NULL); |
7618 | 583k | ptr = *str; |
7619 | 583k | cur = *ptr; |
7620 | 583k | if (cur != '&') |
7621 | 0 | return(NULL); |
7622 | | |
7623 | 583k | ptr++; |
7624 | 583k | name = xmlParseStringName(ctxt, &ptr); |
7625 | 583k | if (name == NULL) { |
7626 | 5 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7627 | 5 | "xmlParseStringEntityRef: no name\n"); |
7628 | 5 | *str = ptr; |
7629 | 5 | return(NULL); |
7630 | 5 | } |
7631 | 583k | if (*ptr != ';') { |
7632 | 4 | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7633 | 4 | xmlFree(name); |
7634 | 4 | *str = ptr; |
7635 | 4 | return(NULL); |
7636 | 4 | } |
7637 | 583k | ptr++; |
7638 | | |
7639 | 583k | *str = ptr; |
7640 | 583k | return(name); |
7641 | 583k | } |
7642 | | |
7643 | | /** |
7644 | | * Parse a parameter entity reference. Always consumes '%'. |
7645 | | * |
7646 | | * @deprecated Internal function, don't use. |
7647 | | * |
7648 | | * The entity content is handled directly by pushing it's content as |
7649 | | * a new input stream. |
7650 | | * |
7651 | | * [69] PEReference ::= '%' Name ';' |
7652 | | * |
7653 | | * [ WFC: No Recursion ] |
7654 | | * A parsed entity must not contain a recursive |
7655 | | * reference to itself, either directly or indirectly. |
7656 | | * |
7657 | | * [ WFC: Entity Declared ] |
7658 | | * In a document without any DTD, a document with only an internal DTD |
7659 | | * subset which contains no parameter entity references, or a document |
7660 | | * with "standalone='yes'", ... ... The declaration of a parameter |
7661 | | * entity must precede any reference to it... |
7662 | | * |
7663 | | * [ VC: Entity Declared ] |
7664 | | * In a document with an external subset or external parameter entities |
7665 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
7666 | | * must precede any reference to it... |
7667 | | * |
7668 | | * [ WFC: In DTD ] |
7669 | | * Parameter-entity references may only appear in the DTD. |
7670 | | * NOTE: misleading but this is handled. |
7671 | | * |
7672 | | * @param ctxt an XML parser context |
7673 | | */ |
7674 | | void |
7675 | | xmlParsePEReference(xmlParserCtxtPtr ctxt) |
7676 | 56.4k | { |
7677 | 56.4k | const xmlChar *name; |
7678 | 56.4k | xmlEntityPtr entity = NULL; |
7679 | 56.4k | xmlParserInputPtr input; |
7680 | | |
7681 | 56.4k | if (RAW != '%') |
7682 | 0 | return; |
7683 | 56.4k | NEXT; |
7684 | 56.4k | name = xmlParseName(ctxt); |
7685 | 56.4k | if (name == NULL) { |
7686 | 920 | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n"); |
7687 | 920 | return; |
7688 | 920 | } |
7689 | 55.5k | if (RAW != ';') { |
7690 | 466 | xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL); |
7691 | 466 | return; |
7692 | 466 | } |
7693 | | |
7694 | 55.1k | NEXT; |
7695 | | |
7696 | | /* Must be set before xmlHandleUndeclaredEntity */ |
7697 | 55.1k | ctxt->hasPErefs = 1; |
7698 | | |
7699 | | /* |
7700 | | * Request the entity from SAX |
7701 | | */ |
7702 | 55.1k | if ((ctxt->sax != NULL) && |
7703 | 55.1k | (ctxt->sax->getParameterEntity != NULL)) |
7704 | 55.1k | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
7705 | | |
7706 | 55.1k | if (entity == NULL) { |
7707 | 1.28k | xmlHandleUndeclaredEntity(ctxt, name); |
7708 | 53.8k | } else { |
7709 | | /* |
7710 | | * Internal checking in case the entity quest barfed |
7711 | | */ |
7712 | 53.8k | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
7713 | 53.8k | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
7714 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7715 | 0 | "Internal: %%%s; is not a parameter entity\n", |
7716 | 0 | name, NULL); |
7717 | 53.8k | } else { |
7718 | 53.8k | if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
7719 | 53.8k | ((ctxt->options & XML_PARSE_NO_XXE) || |
7720 | 388 | ((ctxt->loadsubset == 0) && |
7721 | 194 | (ctxt->replaceEntities == 0) && |
7722 | 194 | (ctxt->validate == 0)))) |
7723 | 388 | return; |
7724 | | |
7725 | 53.4k | if (entity->flags & XML_ENT_EXPANDING) { |
7726 | 2 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7727 | 2 | xmlHaltParser(ctxt); |
7728 | 2 | return; |
7729 | 2 | } |
7730 | | |
7731 | 53.4k | input = xmlNewEntityInputStream(ctxt, entity); |
7732 | 53.4k | if (xmlCtxtPushInput(ctxt, input) < 0) { |
7733 | 0 | xmlFreeInputStream(input); |
7734 | 0 | return; |
7735 | 0 | } |
7736 | | |
7737 | 53.4k | entity->flags |= XML_ENT_EXPANDING; |
7738 | | |
7739 | 53.4k | GROW; |
7740 | | |
7741 | 53.4k | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
7742 | 0 | xmlDetectEncoding(ctxt); |
7743 | |
|
7744 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
7745 | 0 | (IS_BLANK_CH(NXT(5)))) { |
7746 | 0 | xmlParseTextDecl(ctxt); |
7747 | 0 | } |
7748 | 0 | } |
7749 | 53.4k | } |
7750 | 53.8k | } |
7751 | 55.1k | } |
7752 | | |
7753 | | /** |
7754 | | * Load the content of an entity. |
7755 | | * |
7756 | | * @param ctxt an XML parser context |
7757 | | * @param entity an unloaded system entity |
7758 | | * @returns 0 in case of success and -1 in case of failure |
7759 | | */ |
7760 | | static int |
7761 | 0 | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) { |
7762 | 0 | xmlParserInputPtr oldinput, input = NULL; |
7763 | 0 | xmlParserInputPtr *oldinputTab; |
7764 | 0 | const xmlChar *oldencoding; |
7765 | 0 | xmlChar *content = NULL; |
7766 | 0 | xmlResourceType rtype; |
7767 | 0 | size_t length, i; |
7768 | 0 | int oldinputNr, oldinputMax; |
7769 | 0 | int ret = -1; |
7770 | 0 | int res; |
7771 | |
|
7772 | 0 | if ((ctxt == NULL) || (entity == NULL) || |
7773 | 0 | ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) && |
7774 | 0 | (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) || |
7775 | 0 | (entity->content != NULL)) { |
7776 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, |
7777 | 0 | "xmlLoadEntityContent parameter error"); |
7778 | 0 | return(-1); |
7779 | 0 | } |
7780 | | |
7781 | 0 | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) |
7782 | 0 | rtype = XML_RESOURCE_PARAMETER_ENTITY; |
7783 | 0 | else |
7784 | 0 | rtype = XML_RESOURCE_GENERAL_ENTITY; |
7785 | |
|
7786 | 0 | input = xmlLoadResource(ctxt, (char *) entity->URI, |
7787 | 0 | (char *) entity->ExternalID, rtype); |
7788 | 0 | if (input == NULL) |
7789 | 0 | return(-1); |
7790 | | |
7791 | 0 | oldinput = ctxt->input; |
7792 | 0 | oldinputNr = ctxt->inputNr; |
7793 | 0 | oldinputMax = ctxt->inputMax; |
7794 | 0 | oldinputTab = ctxt->inputTab; |
7795 | 0 | oldencoding = ctxt->encoding; |
7796 | |
|
7797 | 0 | ctxt->input = NULL; |
7798 | 0 | ctxt->inputNr = 0; |
7799 | 0 | ctxt->inputMax = 1; |
7800 | 0 | ctxt->encoding = NULL; |
7801 | 0 | ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr)); |
7802 | 0 | if (ctxt->inputTab == NULL) { |
7803 | 0 | xmlErrMemory(ctxt); |
7804 | 0 | xmlFreeInputStream(input); |
7805 | 0 | goto error; |
7806 | 0 | } |
7807 | | |
7808 | 0 | xmlBufResetInput(input->buf->buffer, input); |
7809 | |
|
7810 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
7811 | 0 | xmlFreeInputStream(input); |
7812 | 0 | goto error; |
7813 | 0 | } |
7814 | | |
7815 | 0 | xmlDetectEncoding(ctxt); |
7816 | | |
7817 | | /* |
7818 | | * Parse a possible text declaration first |
7819 | | */ |
7820 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
7821 | 0 | xmlParseTextDecl(ctxt); |
7822 | | /* |
7823 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
7824 | | */ |
7825 | 0 | if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) && |
7826 | 0 | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
7827 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
7828 | 0 | "Version mismatch between document and entity\n"); |
7829 | 0 | } |
7830 | 0 | } |
7831 | |
|
7832 | 0 | length = input->cur - input->base; |
7833 | 0 | xmlBufShrink(input->buf->buffer, length); |
7834 | 0 | xmlSaturatedAdd(&ctxt->sizeentities, length); |
7835 | |
|
7836 | 0 | while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0) |
7837 | 0 | ; |
7838 | |
|
7839 | 0 | xmlBufResetInput(input->buf->buffer, input); |
7840 | |
|
7841 | 0 | if (res < 0) { |
7842 | 0 | xmlCtxtErrIO(ctxt, input->buf->error, NULL); |
7843 | 0 | goto error; |
7844 | 0 | } |
7845 | | |
7846 | 0 | length = xmlBufUse(input->buf->buffer); |
7847 | 0 | if (length > INT_MAX) { |
7848 | 0 | xmlErrMemory(ctxt); |
7849 | 0 | goto error; |
7850 | 0 | } |
7851 | | |
7852 | 0 | content = xmlStrndup(xmlBufContent(input->buf->buffer), length); |
7853 | 0 | if (content == NULL) { |
7854 | 0 | xmlErrMemory(ctxt); |
7855 | 0 | goto error; |
7856 | 0 | } |
7857 | | |
7858 | 0 | for (i = 0; i < length; ) { |
7859 | 0 | int clen = length - i; |
7860 | 0 | int c = xmlGetUTF8Char(content + i, &clen); |
7861 | |
|
7862 | 0 | if ((c < 0) || (!IS_CHAR(c))) { |
7863 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
7864 | 0 | "xmlLoadEntityContent: invalid char value %d\n", |
7865 | 0 | content[i]); |
7866 | 0 | goto error; |
7867 | 0 | } |
7868 | 0 | i += clen; |
7869 | 0 | } |
7870 | | |
7871 | 0 | xmlSaturatedAdd(&ctxt->sizeentities, length); |
7872 | 0 | entity->content = content; |
7873 | 0 | entity->length = length; |
7874 | 0 | content = NULL; |
7875 | 0 | ret = 0; |
7876 | |
|
7877 | 0 | error: |
7878 | 0 | while (ctxt->inputNr > 0) |
7879 | 0 | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
7880 | 0 | xmlFree(ctxt->inputTab); |
7881 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
7882 | |
|
7883 | 0 | ctxt->input = oldinput; |
7884 | 0 | ctxt->inputNr = oldinputNr; |
7885 | 0 | ctxt->inputMax = oldinputMax; |
7886 | 0 | ctxt->inputTab = oldinputTab; |
7887 | 0 | ctxt->encoding = oldencoding; |
7888 | |
|
7889 | 0 | xmlFree(content); |
7890 | |
|
7891 | 0 | return(ret); |
7892 | 0 | } |
7893 | | |
7894 | | /** |
7895 | | * parse PEReference declarations |
7896 | | * |
7897 | | * [69] PEReference ::= '%' Name ';' |
7898 | | * |
7899 | | * [ WFC: No Recursion ] |
7900 | | * A parsed entity must not contain a recursive |
7901 | | * reference to itself, either directly or indirectly. |
7902 | | * |
7903 | | * [ WFC: Entity Declared ] |
7904 | | * In a document without any DTD, a document with only an internal DTD |
7905 | | * subset which contains no parameter entity references, or a document |
7906 | | * with "standalone='yes'", ... ... The declaration of a parameter |
7907 | | * entity must precede any reference to it... |
7908 | | * |
7909 | | * [ VC: Entity Declared ] |
7910 | | * In a document with an external subset or external parameter entities |
7911 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
7912 | | * must precede any reference to it... |
7913 | | * |
7914 | | * [ WFC: In DTD ] |
7915 | | * Parameter-entity references may only appear in the DTD. |
7916 | | * NOTE: misleading but this is handled. |
7917 | | * |
7918 | | * @param ctxt an XML parser context |
7919 | | * @param str a pointer to an index in the string |
7920 | | * @returns the string of the entity content. |
7921 | | * str is updated to the current value of the index |
7922 | | */ |
7923 | | static xmlEntityPtr |
7924 | 2.13k | xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
7925 | 2.13k | const xmlChar *ptr; |
7926 | 2.13k | xmlChar cur; |
7927 | 2.13k | xmlChar *name; |
7928 | 2.13k | xmlEntityPtr entity = NULL; |
7929 | | |
7930 | 2.13k | if ((str == NULL) || (*str == NULL)) return(NULL); |
7931 | 2.13k | ptr = *str; |
7932 | 2.13k | cur = *ptr; |
7933 | 2.13k | if (cur != '%') |
7934 | 0 | return(NULL); |
7935 | 2.13k | ptr++; |
7936 | 2.13k | name = xmlParseStringName(ctxt, &ptr); |
7937 | 2.13k | if (name == NULL) { |
7938 | 1.11k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7939 | 1.11k | "xmlParseStringPEReference: no name\n"); |
7940 | 1.11k | *str = ptr; |
7941 | 1.11k | return(NULL); |
7942 | 1.11k | } |
7943 | 1.01k | cur = *ptr; |
7944 | 1.01k | if (cur != ';') { |
7945 | 415 | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7946 | 415 | xmlFree(name); |
7947 | 415 | *str = ptr; |
7948 | 415 | return(NULL); |
7949 | 415 | } |
7950 | 604 | ptr++; |
7951 | | |
7952 | | /* Must be set before xmlHandleUndeclaredEntity */ |
7953 | 604 | ctxt->hasPErefs = 1; |
7954 | | |
7955 | | /* |
7956 | | * Request the entity from SAX |
7957 | | */ |
7958 | 604 | if ((ctxt->sax != NULL) && |
7959 | 604 | (ctxt->sax->getParameterEntity != NULL)) |
7960 | 604 | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
7961 | | |
7962 | 604 | if (entity == NULL) { |
7963 | 327 | xmlHandleUndeclaredEntity(ctxt, name); |
7964 | 327 | } else { |
7965 | | /* |
7966 | | * Internal checking in case the entity quest barfed |
7967 | | */ |
7968 | 277 | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
7969 | 277 | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
7970 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7971 | 0 | "%%%s; is not a parameter entity\n", |
7972 | 0 | name, NULL); |
7973 | 0 | } |
7974 | 277 | } |
7975 | | |
7976 | 604 | xmlFree(name); |
7977 | 604 | *str = ptr; |
7978 | 604 | return(entity); |
7979 | 1.01k | } |
7980 | | |
7981 | | /** |
7982 | | * parse a DOCTYPE declaration |
7983 | | * |
7984 | | * @deprecated Internal function, don't use. |
7985 | | * |
7986 | | * [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? |
7987 | | * ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
7988 | | * |
7989 | | * [ VC: Root Element Type ] |
7990 | | * The Name in the document type declaration must match the element |
7991 | | * type of the root element. |
7992 | | * |
7993 | | * @param ctxt an XML parser context |
7994 | | */ |
7995 | | |
7996 | | void |
7997 | 10.7k | xmlParseDocTypeDecl(xmlParserCtxtPtr ctxt) { |
7998 | 10.7k | const xmlChar *name = NULL; |
7999 | 10.7k | xmlChar *ExternalID = NULL; |
8000 | 10.7k | xmlChar *URI = NULL; |
8001 | | |
8002 | | /* |
8003 | | * We know that '<!DOCTYPE' has been detected. |
8004 | | */ |
8005 | 10.7k | SKIP(9); |
8006 | | |
8007 | 10.7k | if (SKIP_BLANKS == 0) { |
8008 | 18 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
8009 | 18 | "Space required after 'DOCTYPE'\n"); |
8010 | 18 | } |
8011 | | |
8012 | | /* |
8013 | | * Parse the DOCTYPE name. |
8014 | | */ |
8015 | 10.7k | name = xmlParseName(ctxt); |
8016 | 10.7k | if (name == NULL) { |
8017 | 7.59k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8018 | 7.59k | "xmlParseDocTypeDecl : no DOCTYPE name !\n"); |
8019 | 7.59k | } |
8020 | 10.7k | ctxt->intSubName = name; |
8021 | | |
8022 | 10.7k | SKIP_BLANKS; |
8023 | | |
8024 | | /* |
8025 | | * Check for SystemID and ExternalID |
8026 | | */ |
8027 | 10.7k | URI = xmlParseExternalID(ctxt, &ExternalID, 1); |
8028 | | |
8029 | 10.7k | if ((URI != NULL) || (ExternalID != NULL)) { |
8030 | 835 | ctxt->hasExternalSubset = 1; |
8031 | 835 | } |
8032 | 10.7k | ctxt->extSubURI = URI; |
8033 | 10.7k | ctxt->extSubSystem = ExternalID; |
8034 | | |
8035 | 10.7k | SKIP_BLANKS; |
8036 | | |
8037 | | /* |
8038 | | * Create and update the internal subset. |
8039 | | */ |
8040 | 10.7k | if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) && |
8041 | 10.7k | (!ctxt->disableSAX)) |
8042 | 8.18k | ctxt->sax->internalSubset(ctxt->userData, name, ExternalID, URI); |
8043 | | |
8044 | 10.7k | if ((RAW != '[') && (RAW != '>')) { |
8045 | 1.33k | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8046 | 1.33k | } |
8047 | 10.7k | } |
8048 | | |
8049 | | /** |
8050 | | * parse the internal subset declaration |
8051 | | * |
8052 | | * [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8053 | | * @param ctxt an XML parser context |
8054 | | */ |
8055 | | |
8056 | | static void |
8057 | 9.02k | xmlParseInternalSubset(xmlParserCtxtPtr ctxt) { |
8058 | | /* |
8059 | | * Is there any DTD definition ? |
8060 | | */ |
8061 | 9.02k | if (RAW == '[') { |
8062 | 9.02k | int oldInputNr = ctxt->inputNr; |
8063 | | |
8064 | 9.02k | NEXT; |
8065 | | /* |
8066 | | * Parse the succession of Markup declarations and |
8067 | | * PEReferences. |
8068 | | * Subsequence (markupdecl | PEReference | S)* |
8069 | | */ |
8070 | 9.02k | SKIP_BLANKS; |
8071 | 162k | while (((RAW != ']') || (ctxt->inputNr > oldInputNr)) && |
8072 | 162k | (PARSER_STOPPED(ctxt) == 0)) { |
8073 | | |
8074 | | /* |
8075 | | * Conditional sections are allowed from external entities included |
8076 | | * by PE References in the internal subset. |
8077 | | */ |
8078 | 159k | if ((PARSER_EXTERNAL(ctxt)) && |
8079 | 159k | (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
8080 | 0 | xmlParseConditionalSections(ctxt); |
8081 | 159k | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
8082 | 96.9k | xmlParseMarkupDecl(ctxt); |
8083 | 96.9k | } else if (RAW == '%') { |
8084 | 56.4k | xmlParsePEReference(ctxt); |
8085 | 56.4k | } else { |
8086 | 5.88k | xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL); |
8087 | 5.88k | break; |
8088 | 5.88k | } |
8089 | 153k | SKIP_BLANKS_PE; |
8090 | 153k | SHRINK; |
8091 | 153k | GROW; |
8092 | 153k | } |
8093 | | |
8094 | 9.06k | while (ctxt->inputNr > oldInputNr) |
8095 | 48 | xmlPopPE(ctxt); |
8096 | | |
8097 | 9.02k | if (RAW == ']') { |
8098 | 2.12k | NEXT; |
8099 | 2.12k | SKIP_BLANKS; |
8100 | 2.12k | } |
8101 | 9.02k | } |
8102 | | |
8103 | | /* |
8104 | | * We should be at the end of the DOCTYPE declaration. |
8105 | | */ |
8106 | 9.02k | if ((ctxt->wellFormed) && (RAW != '>')) { |
8107 | 44 | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8108 | 44 | return; |
8109 | 44 | } |
8110 | 8.97k | NEXT; |
8111 | 8.97k | } |
8112 | | |
8113 | | #ifdef LIBXML_SAX1_ENABLED |
8114 | | /** |
8115 | | * parse an attribute |
8116 | | * |
8117 | | * @deprecated Internal function, don't use. |
8118 | | * |
8119 | | * [41] Attribute ::= Name Eq AttValue |
8120 | | * |
8121 | | * [ WFC: No External Entity References ] |
8122 | | * Attribute values cannot contain direct or indirect entity references |
8123 | | * to external entities. |
8124 | | * |
8125 | | * [ WFC: No < in Attribute Values ] |
8126 | | * The replacement text of any entity referred to directly or indirectly in |
8127 | | * an attribute value (other than "<") must not contain a <. |
8128 | | * |
8129 | | * [ VC: Attribute Value Type ] |
8130 | | * The attribute must have been declared; the value must be of the type |
8131 | | * declared for it. |
8132 | | * |
8133 | | * [25] Eq ::= S? '=' S? |
8134 | | * |
8135 | | * With namespace: |
8136 | | * |
8137 | | * [NS 11] Attribute ::= QName Eq AttValue |
8138 | | * |
8139 | | * Also the case QName == xmlns:??? is handled independently as a namespace |
8140 | | * definition. |
8141 | | * |
8142 | | * @param ctxt an XML parser context |
8143 | | * @param value a xmlChar ** used to store the value of the attribute |
8144 | | * @returns the attribute name, and the value in *value. |
8145 | | */ |
8146 | | |
8147 | | const xmlChar * |
8148 | 49.5k | xmlParseAttribute(xmlParserCtxtPtr ctxt, xmlChar **value) { |
8149 | 49.5k | const xmlChar *name; |
8150 | 49.5k | xmlChar *val; |
8151 | | |
8152 | 49.5k | *value = NULL; |
8153 | 49.5k | GROW; |
8154 | 49.5k | name = xmlParseName(ctxt); |
8155 | 49.5k | if (name == NULL) { |
8156 | 38.1k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8157 | 38.1k | "error parsing attribute name\n"); |
8158 | 38.1k | return(NULL); |
8159 | 38.1k | } |
8160 | | |
8161 | | /* |
8162 | | * read the value |
8163 | | */ |
8164 | 11.4k | SKIP_BLANKS; |
8165 | 11.4k | if (RAW == '=') { |
8166 | 8.42k | NEXT; |
8167 | 8.42k | SKIP_BLANKS; |
8168 | 8.42k | val = xmlParseAttValue(ctxt); |
8169 | 8.42k | } else { |
8170 | 3.03k | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8171 | 3.03k | "Specification mandates value for attribute %s\n", name); |
8172 | 3.03k | return(name); |
8173 | 3.03k | } |
8174 | | |
8175 | | /* |
8176 | | * Check that xml:lang conforms to the specification |
8177 | | * No more registered as an error, just generate a warning now |
8178 | | * since this was deprecated in XML second edition |
8179 | | */ |
8180 | 8.42k | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) { |
8181 | 2.39k | if (!xmlCheckLanguageID(val)) { |
8182 | 1.60k | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8183 | 1.60k | "Malformed value for xml:lang : %s\n", |
8184 | 1.60k | val, NULL); |
8185 | 1.60k | } |
8186 | 2.39k | } |
8187 | | |
8188 | | /* |
8189 | | * Check that xml:space conforms to the specification |
8190 | | */ |
8191 | 8.42k | if (xmlStrEqual(name, BAD_CAST "xml:space")) { |
8192 | 218 | if (xmlStrEqual(val, BAD_CAST "default")) |
8193 | 78 | *(ctxt->space) = 0; |
8194 | 140 | else if (xmlStrEqual(val, BAD_CAST "preserve")) |
8195 | 74 | *(ctxt->space) = 1; |
8196 | 66 | else { |
8197 | 66 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8198 | 66 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8199 | 66 | val, NULL); |
8200 | 66 | } |
8201 | 218 | } |
8202 | | |
8203 | 8.42k | *value = val; |
8204 | 8.42k | return(name); |
8205 | 11.4k | } |
8206 | | |
8207 | | /** |
8208 | | * Parse a start tag. Always consumes '<'. |
8209 | | * |
8210 | | * @deprecated Internal function, don't use. |
8211 | | * |
8212 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8213 | | * |
8214 | | * [ WFC: Unique Att Spec ] |
8215 | | * No attribute name may appear more than once in the same start-tag or |
8216 | | * empty-element tag. |
8217 | | * |
8218 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8219 | | * |
8220 | | * [ WFC: Unique Att Spec ] |
8221 | | * No attribute name may appear more than once in the same start-tag or |
8222 | | * empty-element tag. |
8223 | | * |
8224 | | * With namespace: |
8225 | | * |
8226 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8227 | | * |
8228 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8229 | | * |
8230 | | * @param ctxt an XML parser context |
8231 | | * @returns the element name parsed |
8232 | | */ |
8233 | | |
8234 | | const xmlChar * |
8235 | 56.4k | xmlParseStartTag(xmlParserCtxtPtr ctxt) { |
8236 | 56.4k | const xmlChar *name; |
8237 | 56.4k | const xmlChar *attname; |
8238 | 56.4k | xmlChar *attvalue; |
8239 | 56.4k | const xmlChar **atts = ctxt->atts; |
8240 | 56.4k | int nbatts = 0; |
8241 | 56.4k | int maxatts = ctxt->maxatts; |
8242 | 56.4k | int i; |
8243 | | |
8244 | 56.4k | if (RAW != '<') return(NULL); |
8245 | 56.4k | NEXT1; |
8246 | | |
8247 | 56.4k | name = xmlParseName(ctxt); |
8248 | 56.4k | if (name == NULL) { |
8249 | 6.70k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8250 | 6.70k | "xmlParseStartTag: invalid element name\n"); |
8251 | 6.70k | return(NULL); |
8252 | 6.70k | } |
8253 | | |
8254 | | /* |
8255 | | * Now parse the attributes, it ends up with the ending |
8256 | | * |
8257 | | * (S Attribute)* S? |
8258 | | */ |
8259 | 49.7k | SKIP_BLANKS; |
8260 | 49.7k | GROW; |
8261 | | |
8262 | 59.2k | while (((RAW != '>') && |
8263 | 59.2k | ((RAW != '/') || (NXT(1) != '>')) && |
8264 | 59.2k | (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) { |
8265 | 49.5k | attname = xmlParseAttribute(ctxt, &attvalue); |
8266 | 49.5k | if (attname == NULL) |
8267 | 38.1k | break; |
8268 | 11.4k | if (attvalue != NULL) { |
8269 | | /* |
8270 | | * [ WFC: Unique Att Spec ] |
8271 | | * No attribute name may appear more than once in the same |
8272 | | * start-tag or empty-element tag. |
8273 | | */ |
8274 | 13.1k | for (i = 0; i < nbatts;i += 2) { |
8275 | 7.22k | if (xmlStrEqual(atts[i], attname)) { |
8276 | 1.83k | xmlErrAttributeDup(ctxt, NULL, attname); |
8277 | 1.83k | goto failed; |
8278 | 1.83k | } |
8279 | 7.22k | } |
8280 | | /* |
8281 | | * Add the pair to atts |
8282 | | */ |
8283 | 5.88k | if (nbatts + 4 > maxatts) { |
8284 | 1.20k | const xmlChar **n; |
8285 | 1.20k | int newSize; |
8286 | | |
8287 | 1.20k | newSize = xmlGrowCapacity(maxatts, sizeof(n[0]) * 2, |
8288 | 1.20k | 11, XML_MAX_ATTRS); |
8289 | 1.20k | if (newSize < 0) { |
8290 | 0 | xmlErrMemory(ctxt); |
8291 | 0 | goto failed; |
8292 | 0 | } |
8293 | 1.20k | #ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION |
8294 | 1.20k | if (newSize < 2) |
8295 | 937 | newSize = 2; |
8296 | 1.20k | #endif |
8297 | 1.20k | n = xmlRealloc(atts, newSize * sizeof(n[0]) * 2); |
8298 | 1.20k | if (n == NULL) { |
8299 | 0 | xmlErrMemory(ctxt); |
8300 | 0 | goto failed; |
8301 | 0 | } |
8302 | 1.20k | atts = n; |
8303 | 1.20k | maxatts = newSize * 2; |
8304 | 1.20k | ctxt->atts = atts; |
8305 | 1.20k | ctxt->maxatts = maxatts; |
8306 | 1.20k | } |
8307 | | |
8308 | 5.88k | atts[nbatts++] = attname; |
8309 | 5.88k | atts[nbatts++] = attvalue; |
8310 | 5.88k | atts[nbatts] = NULL; |
8311 | 5.88k | atts[nbatts + 1] = NULL; |
8312 | | |
8313 | 5.88k | attvalue = NULL; |
8314 | 5.88k | } |
8315 | | |
8316 | 11.4k | failed: |
8317 | | |
8318 | 11.4k | if (attvalue != NULL) |
8319 | 1.83k | xmlFree(attvalue); |
8320 | | |
8321 | 11.4k | GROW |
8322 | 11.4k | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
8323 | 2.03k | break; |
8324 | 9.42k | if (SKIP_BLANKS == 0) { |
8325 | 7.48k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
8326 | 7.48k | "attributes construct error\n"); |
8327 | 7.48k | } |
8328 | 9.42k | SHRINK; |
8329 | 9.42k | GROW; |
8330 | 9.42k | } |
8331 | | |
8332 | | /* |
8333 | | * SAX: Start of Element ! |
8334 | | */ |
8335 | 49.7k | if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) && |
8336 | 49.7k | (!ctxt->disableSAX)) { |
8337 | 47.6k | if (nbatts > 0) |
8338 | 3.93k | ctxt->sax->startElement(ctxt->userData, name, atts); |
8339 | 43.7k | else |
8340 | 43.7k | ctxt->sax->startElement(ctxt->userData, name, NULL); |
8341 | 47.6k | } |
8342 | | |
8343 | 49.7k | if (atts != NULL) { |
8344 | | /* Free only the content strings */ |
8345 | 11.6k | for (i = 1;i < nbatts;i+=2) |
8346 | 5.88k | if (atts[i] != NULL) |
8347 | 5.88k | xmlFree((xmlChar *) atts[i]); |
8348 | 5.75k | } |
8349 | 49.7k | return(name); |
8350 | 49.7k | } |
8351 | | |
8352 | | /** |
8353 | | * Parse an end tag. Always consumes '</'. |
8354 | | * |
8355 | | * [42] ETag ::= '</' Name S? '>' |
8356 | | * |
8357 | | * With namespace |
8358 | | * |
8359 | | * [NS 9] ETag ::= '</' QName S? '>' |
8360 | | * @param ctxt an XML parser context |
8361 | | * @param line line of the start tag |
8362 | | */ |
8363 | | |
8364 | | static void |
8365 | 4.90k | xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) { |
8366 | 4.90k | const xmlChar *name; |
8367 | | |
8368 | 4.90k | GROW; |
8369 | 4.90k | if ((RAW != '<') || (NXT(1) != '/')) { |
8370 | 8 | xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED, |
8371 | 8 | "xmlParseEndTag: '</' not found\n"); |
8372 | 8 | return; |
8373 | 8 | } |
8374 | 4.89k | SKIP(2); |
8375 | | |
8376 | 4.89k | name = xmlParseNameAndCompare(ctxt,ctxt->name); |
8377 | | |
8378 | | /* |
8379 | | * We should definitely be at the ending "S? '>'" part |
8380 | | */ |
8381 | 4.89k | GROW; |
8382 | 4.89k | SKIP_BLANKS; |
8383 | 4.89k | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
8384 | 2.11k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
8385 | 2.11k | } else |
8386 | 2.78k | NEXT1; |
8387 | | |
8388 | | /* |
8389 | | * [ WFC: Element Type Match ] |
8390 | | * The Name in an element's end-tag must match the element type in the |
8391 | | * start-tag. |
8392 | | * |
8393 | | */ |
8394 | 4.89k | if (name != (xmlChar*)1) { |
8395 | 3.41k | if (name == NULL) name = BAD_CAST "unparsable"; |
8396 | 3.41k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
8397 | 3.41k | "Opening and ending tag mismatch: %s line %d and %s\n", |
8398 | 3.41k | ctxt->name, line, name); |
8399 | 3.41k | } |
8400 | | |
8401 | | /* |
8402 | | * SAX: End of Tag |
8403 | | */ |
8404 | 4.89k | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
8405 | 4.89k | (!ctxt->disableSAX)) |
8406 | 4.02k | ctxt->sax->endElement(ctxt->userData, ctxt->name); |
8407 | | |
8408 | 4.89k | namePop(ctxt); |
8409 | 4.89k | spacePop(ctxt); |
8410 | 4.89k | } |
8411 | | |
8412 | | /** |
8413 | | * parse an end of tag |
8414 | | * |
8415 | | * @deprecated Internal function, don't use. |
8416 | | * |
8417 | | * [42] ETag ::= '</' Name S? '>' |
8418 | | * |
8419 | | * With namespace |
8420 | | * |
8421 | | * [NS 9] ETag ::= '</' QName S? '>' |
8422 | | * @param ctxt an XML parser context |
8423 | | */ |
8424 | | |
8425 | | void |
8426 | 0 | xmlParseEndTag(xmlParserCtxtPtr ctxt) { |
8427 | 0 | xmlParseEndTag1(ctxt, 0); |
8428 | 0 | } |
8429 | | #endif /* LIBXML_SAX1_ENABLED */ |
8430 | | |
8431 | | /************************************************************************ |
8432 | | * * |
8433 | | * SAX 2 specific operations * |
8434 | | * * |
8435 | | ************************************************************************/ |
8436 | | |
8437 | | /** |
8438 | | * parse an XML Namespace QName |
8439 | | * |
8440 | | * [6] QName ::= (Prefix ':')? LocalPart |
8441 | | * [7] Prefix ::= NCName |
8442 | | * [8] LocalPart ::= NCName |
8443 | | * |
8444 | | * @param ctxt an XML parser context |
8445 | | * @param prefix pointer to store the prefix part |
8446 | | * @returns the Name parsed or NULL |
8447 | | */ |
8448 | | |
8449 | | static xmlHashedString |
8450 | 161k | xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) { |
8451 | 161k | xmlHashedString l, p; |
8452 | 161k | int start, isNCName = 0; |
8453 | | |
8454 | 161k | l.name = NULL; |
8455 | 161k | p.name = NULL; |
8456 | | |
8457 | 161k | GROW; |
8458 | 161k | start = CUR_PTR - BASE_PTR; |
8459 | | |
8460 | 161k | l = xmlParseNCName(ctxt); |
8461 | 161k | if (l.name != NULL) { |
8462 | 99.7k | isNCName = 1; |
8463 | 99.7k | if (CUR == ':') { |
8464 | 19.0k | NEXT; |
8465 | 19.0k | p = l; |
8466 | 19.0k | l = xmlParseNCName(ctxt); |
8467 | 19.0k | } |
8468 | 99.7k | } |
8469 | 161k | if ((l.name == NULL) || (CUR == ':')) { |
8470 | 64.1k | xmlChar *tmp; |
8471 | | |
8472 | 64.1k | l.name = NULL; |
8473 | 64.1k | p.name = NULL; |
8474 | 64.1k | if ((isNCName == 0) && (CUR != ':')) |
8475 | 51.3k | return(l); |
8476 | 12.8k | tmp = xmlParseNmtoken(ctxt); |
8477 | 12.8k | if (tmp != NULL) |
8478 | 11.4k | xmlFree(tmp); |
8479 | 12.8k | l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start, |
8480 | 12.8k | CUR_PTR - (BASE_PTR + start)); |
8481 | 12.8k | if (l.name == NULL) { |
8482 | 0 | xmlErrMemory(ctxt); |
8483 | 0 | return(l); |
8484 | 0 | } |
8485 | 12.8k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8486 | 12.8k | "Failed to parse QName '%s'\n", l.name, NULL, NULL); |
8487 | 12.8k | } |
8488 | | |
8489 | 110k | *prefix = p; |
8490 | 110k | return(l); |
8491 | 161k | } |
8492 | | |
8493 | | /** |
8494 | | * parse an XML Namespace QName |
8495 | | * |
8496 | | * [6] QName ::= (Prefix ':')? LocalPart |
8497 | | * [7] Prefix ::= NCName |
8498 | | * [8] LocalPart ::= NCName |
8499 | | * |
8500 | | * @param ctxt an XML parser context |
8501 | | * @param prefix pointer to store the prefix part |
8502 | | * @returns the Name parsed or NULL |
8503 | | */ |
8504 | | |
8505 | | static const xmlChar * |
8506 | 1.57k | xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) { |
8507 | 1.57k | xmlHashedString n, p; |
8508 | | |
8509 | 1.57k | n = xmlParseQNameHashed(ctxt, &p); |
8510 | 1.57k | if (n.name == NULL) |
8511 | 464 | return(NULL); |
8512 | 1.11k | *prefix = p.name; |
8513 | 1.11k | return(n.name); |
8514 | 1.57k | } |
8515 | | |
8516 | | /** |
8517 | | * parse an XML name and compares for match |
8518 | | * (specialized for endtag parsing) |
8519 | | * |
8520 | | * @param ctxt an XML parser context |
8521 | | * @param name the localname |
8522 | | * @param prefix the prefix, if any. |
8523 | | * @returns NULL for an illegal name, (xmlChar*) 1 for success |
8524 | | * and the name for mismatch |
8525 | | */ |
8526 | | |
8527 | | static const xmlChar * |
8528 | | xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name, |
8529 | 1.86k | xmlChar const *prefix) { |
8530 | 1.86k | const xmlChar *cmp; |
8531 | 1.86k | const xmlChar *in; |
8532 | 1.86k | const xmlChar *ret; |
8533 | 1.86k | const xmlChar *prefix2; |
8534 | | |
8535 | 1.86k | if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name)); |
8536 | | |
8537 | 1.86k | GROW; |
8538 | 1.86k | in = ctxt->input->cur; |
8539 | | |
8540 | 1.86k | cmp = prefix; |
8541 | 3.01k | while (*in != 0 && *in == *cmp) { |
8542 | 1.14k | ++in; |
8543 | 1.14k | ++cmp; |
8544 | 1.14k | } |
8545 | 1.86k | if ((*cmp == 0) && (*in == ':')) { |
8546 | 546 | in++; |
8547 | 546 | cmp = name; |
8548 | 1.34k | while (*in != 0 && *in == *cmp) { |
8549 | 803 | ++in; |
8550 | 803 | ++cmp; |
8551 | 803 | } |
8552 | 546 | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
8553 | | /* success */ |
8554 | 291 | ctxt->input->col += in - ctxt->input->cur; |
8555 | 291 | ctxt->input->cur = in; |
8556 | 291 | return((const xmlChar*) 1); |
8557 | 291 | } |
8558 | 546 | } |
8559 | | /* |
8560 | | * all strings coms from the dictionary, equality can be done directly |
8561 | | */ |
8562 | 1.57k | ret = xmlParseQName (ctxt, &prefix2); |
8563 | 1.57k | if (ret == NULL) |
8564 | 464 | return(NULL); |
8565 | 1.11k | if ((ret == name) && (prefix == prefix2)) |
8566 | 126 | return((const xmlChar*) 1); |
8567 | 987 | return ret; |
8568 | 1.11k | } |
8569 | | |
8570 | | /** |
8571 | | * parse an attribute in the new SAX2 framework. |
8572 | | * |
8573 | | * @param ctxt an XML parser context |
8574 | | * @param pref the element prefix |
8575 | | * @param elem the element name |
8576 | | * @param hprefix resulting attribute prefix |
8577 | | * @param value resulting value of the attribute |
8578 | | * @param len resulting length of the attribute |
8579 | | * @param alloc resulting indicator if the attribute was allocated |
8580 | | * @returns the attribute name, and the value in *value, . |
8581 | | */ |
8582 | | |
8583 | | static xmlHashedString |
8584 | | xmlParseAttribute2(xmlParserCtxtPtr ctxt, |
8585 | | const xmlChar * pref, const xmlChar * elem, |
8586 | | xmlHashedString * hprefix, xmlChar ** value, |
8587 | | int *len, int *alloc) |
8588 | 61.5k | { |
8589 | 61.5k | xmlHashedString hname; |
8590 | 61.5k | const xmlChar *prefix, *name; |
8591 | 61.5k | xmlChar *val = NULL, *internal_val = NULL; |
8592 | 61.5k | int normalize = 0; |
8593 | 61.5k | int isNamespace; |
8594 | | |
8595 | 61.5k | *value = NULL; |
8596 | 61.5k | GROW; |
8597 | 61.5k | hname = xmlParseQNameHashed(ctxt, hprefix); |
8598 | 61.5k | if (hname.name == NULL) { |
8599 | 33.7k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8600 | 33.7k | "error parsing attribute name\n"); |
8601 | 33.7k | return(hname); |
8602 | 33.7k | } |
8603 | 27.7k | name = hname.name; |
8604 | 27.7k | prefix = hprefix->name; |
8605 | | |
8606 | | /* |
8607 | | * get the type if needed |
8608 | | */ |
8609 | 27.7k | if (ctxt->attsSpecial != NULL) { |
8610 | 4.11k | int type; |
8611 | | |
8612 | 4.11k | type = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem, |
8613 | 4.11k | prefix, name)); |
8614 | 4.11k | if (type != 0) |
8615 | 2.30k | normalize = 1; |
8616 | 4.11k | } |
8617 | | |
8618 | | /* |
8619 | | * read the value |
8620 | | */ |
8621 | 27.7k | SKIP_BLANKS; |
8622 | 27.7k | if (RAW == '=') { |
8623 | 25.4k | NEXT; |
8624 | 25.4k | SKIP_BLANKS; |
8625 | 25.4k | isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) || |
8626 | 25.4k | (prefix == ctxt->str_xmlns)); |
8627 | 25.4k | val = xmlParseAttValueInternal(ctxt, len, alloc, normalize, |
8628 | 25.4k | isNamespace); |
8629 | 25.4k | if (val == NULL) |
8630 | 1.86k | goto error; |
8631 | 25.4k | } else { |
8632 | 2.38k | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8633 | 2.38k | "Specification mandates value for attribute %s\n", |
8634 | 2.38k | name); |
8635 | 2.38k | goto error; |
8636 | 2.38k | } |
8637 | | |
8638 | 23.5k | if (prefix == ctxt->str_xml) { |
8639 | | /* |
8640 | | * Check that xml:lang conforms to the specification |
8641 | | * No more registered as an error, just generate a warning now |
8642 | | * since this was deprecated in XML second edition |
8643 | | */ |
8644 | 3.90k | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) { |
8645 | 502 | internal_val = xmlStrndup(val, *len); |
8646 | 502 | if (internal_val == NULL) |
8647 | 0 | goto mem_error; |
8648 | 502 | if (!xmlCheckLanguageID(internal_val)) { |
8649 | 275 | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8650 | 275 | "Malformed value for xml:lang : %s\n", |
8651 | 275 | internal_val, NULL); |
8652 | 275 | } |
8653 | 502 | } |
8654 | | |
8655 | | /* |
8656 | | * Check that xml:space conforms to the specification |
8657 | | */ |
8658 | 3.90k | if (xmlStrEqual(name, BAD_CAST "space")) { |
8659 | 199 | internal_val = xmlStrndup(val, *len); |
8660 | 199 | if (internal_val == NULL) |
8661 | 0 | goto mem_error; |
8662 | 199 | if (xmlStrEqual(internal_val, BAD_CAST "default")) |
8663 | 67 | *(ctxt->space) = 0; |
8664 | 132 | else if (xmlStrEqual(internal_val, BAD_CAST "preserve")) |
8665 | 66 | *(ctxt->space) = 1; |
8666 | 66 | else { |
8667 | 66 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8668 | 66 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8669 | 66 | internal_val, NULL); |
8670 | 66 | } |
8671 | 199 | } |
8672 | 3.90k | if (internal_val) { |
8673 | 701 | xmlFree(internal_val); |
8674 | 701 | } |
8675 | 3.90k | } |
8676 | | |
8677 | 23.5k | *value = val; |
8678 | 23.5k | return (hname); |
8679 | | |
8680 | 0 | mem_error: |
8681 | 0 | xmlErrMemory(ctxt); |
8682 | 4.24k | error: |
8683 | 4.24k | if ((val != NULL) && (*alloc != 0)) |
8684 | 0 | xmlFree(val); |
8685 | 4.24k | return(hname); |
8686 | 0 | } |
8687 | | |
8688 | | /** |
8689 | | * Inserts a new attribute into the hash table. |
8690 | | * |
8691 | | * @param ctxt parser context |
8692 | | * @param size size of the hash table |
8693 | | * @param name attribute name |
8694 | | * @param uri namespace uri |
8695 | | * @param hashValue combined hash value of name and uri |
8696 | | * @param aindex attribute index (this is a multiple of 5) |
8697 | | * @returns INT_MAX if no existing attribute was found, the attribute |
8698 | | * index if an attribute was found, -1 if a memory allocation failed. |
8699 | | */ |
8700 | | static int |
8701 | | xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name, |
8702 | 23.8k | const xmlChar *uri, unsigned hashValue, int aindex) { |
8703 | 23.8k | xmlAttrHashBucket *table = ctxt->attrHash; |
8704 | 23.8k | xmlAttrHashBucket *bucket; |
8705 | 23.8k | unsigned hindex; |
8706 | | |
8707 | 23.8k | hindex = hashValue & (size - 1); |
8708 | 23.8k | bucket = &table[hindex]; |
8709 | | |
8710 | 28.0k | while (bucket->index >= 0) { |
8711 | 14.2k | const xmlChar **atts = &ctxt->atts[bucket->index]; |
8712 | | |
8713 | 14.2k | if (name == atts[0]) { |
8714 | 11.3k | int nsIndex = XML_PTR_TO_INT(atts[2]); |
8715 | | |
8716 | 11.3k | if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) : |
8717 | 11.3k | (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) : |
8718 | 4.31k | (uri == ctxt->nsTab[nsIndex * 2 + 1])) |
8719 | 10.0k | return(bucket->index); |
8720 | 11.3k | } |
8721 | | |
8722 | 4.20k | hindex++; |
8723 | 4.20k | bucket++; |
8724 | 4.20k | if (hindex >= size) { |
8725 | 1.31k | hindex = 0; |
8726 | 1.31k | bucket = table; |
8727 | 1.31k | } |
8728 | 4.20k | } |
8729 | | |
8730 | 13.7k | bucket->index = aindex; |
8731 | | |
8732 | 13.7k | return(INT_MAX); |
8733 | 23.8k | } |
8734 | | |
8735 | | static int |
8736 | | xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size, |
8737 | | const xmlChar *name, const xmlChar *prefix, |
8738 | 783 | unsigned hashValue, int aindex) { |
8739 | 783 | xmlAttrHashBucket *table = ctxt->attrHash; |
8740 | 783 | xmlAttrHashBucket *bucket; |
8741 | 783 | unsigned hindex; |
8742 | | |
8743 | 783 | hindex = hashValue & (size - 1); |
8744 | 783 | bucket = &table[hindex]; |
8745 | | |
8746 | 1.42k | while (bucket->index >= 0) { |
8747 | 940 | const xmlChar **atts = &ctxt->atts[bucket->index]; |
8748 | | |
8749 | 940 | if ((name == atts[0]) && (prefix == atts[1])) |
8750 | 303 | return(bucket->index); |
8751 | | |
8752 | 637 | hindex++; |
8753 | 637 | bucket++; |
8754 | 637 | if (hindex >= size) { |
8755 | 169 | hindex = 0; |
8756 | 169 | bucket = table; |
8757 | 169 | } |
8758 | 637 | } |
8759 | | |
8760 | 480 | bucket->index = aindex; |
8761 | | |
8762 | 480 | return(INT_MAX); |
8763 | 783 | } |
8764 | | /** |
8765 | | * Parse a start tag. Always consumes '<'. |
8766 | | * |
8767 | | * This routine is called when running SAX2 parsing |
8768 | | * |
8769 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8770 | | * |
8771 | | * [ WFC: Unique Att Spec ] |
8772 | | * No attribute name may appear more than once in the same start-tag or |
8773 | | * empty-element tag. |
8774 | | * |
8775 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8776 | | * |
8777 | | * [ WFC: Unique Att Spec ] |
8778 | | * No attribute name may appear more than once in the same start-tag or |
8779 | | * empty-element tag. |
8780 | | * |
8781 | | * With namespace: |
8782 | | * |
8783 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8784 | | * |
8785 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8786 | | * |
8787 | | * @param ctxt an XML parser context |
8788 | | * @param pref resulting namespace prefix |
8789 | | * @param URI resulting namespace URI |
8790 | | * @param nbNsPtr resulting number of namespace declarations |
8791 | | * @returns the element name parsed |
8792 | | */ |
8793 | | |
8794 | | static const xmlChar * |
8795 | | xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref, |
8796 | 98.7k | const xmlChar **URI, int *nbNsPtr) { |
8797 | 98.7k | xmlHashedString hlocalname; |
8798 | 98.7k | xmlHashedString hprefix; |
8799 | 98.7k | xmlHashedString hattname; |
8800 | 98.7k | xmlHashedString haprefix; |
8801 | 98.7k | const xmlChar *localname; |
8802 | 98.7k | const xmlChar *prefix; |
8803 | 98.7k | const xmlChar *attname; |
8804 | 98.7k | const xmlChar *aprefix; |
8805 | 98.7k | const xmlChar *uri; |
8806 | 98.7k | xmlChar *attvalue = NULL; |
8807 | 98.7k | const xmlChar **atts = ctxt->atts; |
8808 | 98.7k | unsigned attrHashSize = 0; |
8809 | 98.7k | int maxatts = ctxt->maxatts; |
8810 | 98.7k | int nratts, nbatts, nbdef; |
8811 | 98.7k | int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts; |
8812 | 98.7k | int alloc = 0; |
8813 | 98.7k | int numNsErr = 0; |
8814 | 98.7k | int numDupErr = 0; |
8815 | | |
8816 | 98.7k | if (RAW != '<') return(NULL); |
8817 | 98.7k | NEXT1; |
8818 | | |
8819 | 98.7k | nbatts = 0; |
8820 | 98.7k | nratts = 0; |
8821 | 98.7k | nbdef = 0; |
8822 | 98.7k | nbNs = 0; |
8823 | 98.7k | nbTotalDef = 0; |
8824 | 98.7k | attval = 0; |
8825 | | |
8826 | 98.7k | if (xmlParserNsStartElement(ctxt->nsdb) < 0) { |
8827 | 0 | xmlErrMemory(ctxt); |
8828 | 0 | return(NULL); |
8829 | 0 | } |
8830 | | |
8831 | 98.7k | hlocalname = xmlParseQNameHashed(ctxt, &hprefix); |
8832 | 98.7k | if (hlocalname.name == NULL) { |
8833 | 17.0k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8834 | 17.0k | "StartTag: invalid element name\n"); |
8835 | 17.0k | return(NULL); |
8836 | 17.0k | } |
8837 | 81.7k | localname = hlocalname.name; |
8838 | 81.7k | prefix = hprefix.name; |
8839 | | |
8840 | | /* |
8841 | | * Now parse the attributes, it ends up with the ending |
8842 | | * |
8843 | | * (S Attribute)* S? |
8844 | | */ |
8845 | 81.7k | SKIP_BLANKS; |
8846 | 81.7k | GROW; |
8847 | | |
8848 | | /* |
8849 | | * The ctxt->atts array will be ultimately passed to the SAX callback |
8850 | | * containing five xmlChar pointers for each attribute: |
8851 | | * |
8852 | | * [0] attribute name |
8853 | | * [1] attribute prefix |
8854 | | * [2] namespace URI |
8855 | | * [3] attribute value |
8856 | | * [4] end of attribute value |
8857 | | * |
8858 | | * To save memory, we reuse this array temporarily and store integers |
8859 | | * in these pointer variables. |
8860 | | * |
8861 | | * [0] attribute name |
8862 | | * [1] attribute prefix |
8863 | | * [2] hash value of attribute prefix, and later namespace index |
8864 | | * [3] for non-allocated values: ptrdiff_t offset into input buffer |
8865 | | * [4] for non-allocated values: ptrdiff_t offset into input buffer |
8866 | | * |
8867 | | * The ctxt->attallocs array contains an additional unsigned int for |
8868 | | * each attribute, containing the hash value of the attribute name |
8869 | | * and the alloc flag in bit 31. |
8870 | | */ |
8871 | | |
8872 | 95.8k | while (((RAW != '>') && |
8873 | 95.8k | ((RAW != '/') || (NXT(1) != '>')) && |
8874 | 95.8k | (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) { |
8875 | 61.5k | int len = -1; |
8876 | | |
8877 | 61.5k | hattname = xmlParseAttribute2(ctxt, prefix, localname, |
8878 | 61.5k | &haprefix, &attvalue, &len, |
8879 | 61.5k | &alloc); |
8880 | 61.5k | if (hattname.name == NULL) |
8881 | 33.7k | break; |
8882 | 27.7k | if (attvalue == NULL) |
8883 | 4.24k | goto next_attr; |
8884 | 23.5k | attname = hattname.name; |
8885 | 23.5k | aprefix = haprefix.name; |
8886 | 23.5k | if (len < 0) len = xmlStrlen(attvalue); |
8887 | | |
8888 | 23.5k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
8889 | 3.25k | xmlHashedString huri; |
8890 | 3.25k | xmlURIPtr parsedUri; |
8891 | | |
8892 | 3.25k | huri = xmlDictLookupHashed(ctxt->dict, attvalue, len); |
8893 | 3.25k | uri = huri.name; |
8894 | 3.25k | if (uri == NULL) { |
8895 | 0 | xmlErrMemory(ctxt); |
8896 | 0 | goto next_attr; |
8897 | 0 | } |
8898 | 3.25k | if (*uri != 0) { |
8899 | 3.01k | if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) { |
8900 | 0 | xmlErrMemory(ctxt); |
8901 | 0 | goto next_attr; |
8902 | 0 | } |
8903 | 3.01k | if (parsedUri == NULL) { |
8904 | 1.67k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
8905 | 1.67k | "xmlns: '%s' is not a valid URI\n", |
8906 | 1.67k | uri, NULL, NULL); |
8907 | 1.67k | } else { |
8908 | 1.34k | if (parsedUri->scheme == NULL) { |
8909 | 1.07k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
8910 | 1.07k | "xmlns: URI %s is not absolute\n", |
8911 | 1.07k | uri, NULL, NULL); |
8912 | 1.07k | } |
8913 | 1.34k | xmlFreeURI(parsedUri); |
8914 | 1.34k | } |
8915 | 3.01k | if (uri == ctxt->str_xml_ns) { |
8916 | 34 | if (attname != ctxt->str_xml) { |
8917 | 34 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8918 | 34 | "xml namespace URI cannot be the default namespace\n", |
8919 | 34 | NULL, NULL, NULL); |
8920 | 34 | } |
8921 | 34 | goto next_attr; |
8922 | 34 | } |
8923 | 2.98k | if ((len == 29) && |
8924 | 2.98k | (xmlStrEqual(uri, |
8925 | 102 | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
8926 | 34 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8927 | 34 | "reuse of the xmlns namespace name is forbidden\n", |
8928 | 34 | NULL, NULL, NULL); |
8929 | 34 | goto next_attr; |
8930 | 34 | } |
8931 | 2.98k | } |
8932 | | |
8933 | 3.18k | if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0) |
8934 | 2.63k | nbNs++; |
8935 | 20.2k | } else if (aprefix == ctxt->str_xmlns) { |
8936 | 3.41k | xmlHashedString huri; |
8937 | 3.41k | xmlURIPtr parsedUri; |
8938 | | |
8939 | 3.41k | huri = xmlDictLookupHashed(ctxt->dict, attvalue, len); |
8940 | 3.41k | uri = huri.name; |
8941 | 3.41k | if (uri == NULL) { |
8942 | 0 | xmlErrMemory(ctxt); |
8943 | 0 | goto next_attr; |
8944 | 0 | } |
8945 | | |
8946 | 3.41k | if (attname == ctxt->str_xml) { |
8947 | 84 | if (uri != ctxt->str_xml_ns) { |
8948 | 66 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8949 | 66 | "xml namespace prefix mapped to wrong URI\n", |
8950 | 66 | NULL, NULL, NULL); |
8951 | 66 | } |
8952 | | /* |
8953 | | * Do not keep a namespace definition node |
8954 | | */ |
8955 | 84 | goto next_attr; |
8956 | 84 | } |
8957 | 3.33k | if (uri == ctxt->str_xml_ns) { |
8958 | 34 | if (attname != ctxt->str_xml) { |
8959 | 34 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8960 | 34 | "xml namespace URI mapped to wrong prefix\n", |
8961 | 34 | NULL, NULL, NULL); |
8962 | 34 | } |
8963 | 34 | goto next_attr; |
8964 | 34 | } |
8965 | 3.30k | if (attname == ctxt->str_xmlns) { |
8966 | 68 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8967 | 68 | "redefinition of the xmlns prefix is forbidden\n", |
8968 | 68 | NULL, NULL, NULL); |
8969 | 68 | goto next_attr; |
8970 | 68 | } |
8971 | 3.23k | if ((len == 29) && |
8972 | 3.23k | (xmlStrEqual(uri, |
8973 | 68 | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
8974 | 34 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8975 | 34 | "reuse of the xmlns namespace name is forbidden\n", |
8976 | 34 | NULL, NULL, NULL); |
8977 | 34 | goto next_attr; |
8978 | 34 | } |
8979 | 3.19k | if ((uri == NULL) || (uri[0] == 0)) { |
8980 | 71 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8981 | 71 | "xmlns:%s: Empty XML namespace is not allowed\n", |
8982 | 71 | attname, NULL, NULL); |
8983 | 71 | goto next_attr; |
8984 | 3.12k | } else { |
8985 | 3.12k | if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) { |
8986 | 0 | xmlErrMemory(ctxt); |
8987 | 0 | goto next_attr; |
8988 | 0 | } |
8989 | 3.12k | if (parsedUri == NULL) { |
8990 | 878 | xmlNsErr(ctxt, XML_WAR_NS_URI, |
8991 | 878 | "xmlns:%s: '%s' is not a valid URI\n", |
8992 | 878 | attname, uri, NULL); |
8993 | 2.25k | } else { |
8994 | 2.25k | if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) { |
8995 | 71 | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
8996 | 71 | "xmlns:%s: URI %s is not absolute\n", |
8997 | 71 | attname, uri, NULL); |
8998 | 71 | } |
8999 | 2.25k | xmlFreeURI(parsedUri); |
9000 | 2.25k | } |
9001 | 3.12k | } |
9002 | | |
9003 | 3.12k | if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0) |
9004 | 2.98k | nbNs++; |
9005 | 16.8k | } else { |
9006 | | /* |
9007 | | * Populate attributes array, see above for repurposing |
9008 | | * of xmlChar pointers. |
9009 | | */ |
9010 | 16.8k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9011 | 4.24k | int res = xmlCtxtGrowAttrs(ctxt); |
9012 | | |
9013 | 4.24k | maxatts = ctxt->maxatts; |
9014 | 4.24k | atts = ctxt->atts; |
9015 | | |
9016 | 4.24k | if (res < 0) |
9017 | 0 | goto next_attr; |
9018 | 4.24k | } |
9019 | 16.8k | ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) | |
9020 | 16.8k | ((unsigned) alloc << 31); |
9021 | 16.8k | atts[nbatts++] = attname; |
9022 | 16.8k | atts[nbatts++] = aprefix; |
9023 | 16.8k | atts[nbatts++] = (const xmlChar *) (size_t) haprefix.hashValue; |
9024 | 16.8k | if (alloc) { |
9025 | 1.61k | atts[nbatts++] = attvalue; |
9026 | 1.61k | attvalue += len; |
9027 | 1.61k | atts[nbatts++] = attvalue; |
9028 | 15.2k | } else { |
9029 | | /* |
9030 | | * attvalue points into the input buffer which can be |
9031 | | * reallocated. Store differences to input->base instead. |
9032 | | * The pointers will be reconstructed later. |
9033 | | */ |
9034 | 15.2k | atts[nbatts++] = (void *) (attvalue - BASE_PTR); |
9035 | 15.2k | attvalue += len; |
9036 | 15.2k | atts[nbatts++] = (void *) (attvalue - BASE_PTR); |
9037 | 15.2k | } |
9038 | | /* |
9039 | | * tag if some deallocation is needed |
9040 | | */ |
9041 | 16.8k | if (alloc != 0) attval = 1; |
9042 | 16.8k | attvalue = NULL; /* moved into atts */ |
9043 | 16.8k | } |
9044 | | |
9045 | 27.7k | next_attr: |
9046 | 27.7k | if ((attvalue != NULL) && (alloc != 0)) { |
9047 | 1.32k | xmlFree(attvalue); |
9048 | 1.32k | attvalue = NULL; |
9049 | 1.32k | } |
9050 | | |
9051 | 27.7k | GROW |
9052 | 27.7k | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
9053 | 3.86k | break; |
9054 | 23.9k | if (SKIP_BLANKS == 0) { |
9055 | 9.78k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
9056 | 9.78k | "attributes construct error\n"); |
9057 | 9.78k | break; |
9058 | 9.78k | } |
9059 | 14.1k | GROW; |
9060 | 14.1k | } |
9061 | | |
9062 | | /* |
9063 | | * Namespaces from default attributes |
9064 | | */ |
9065 | 81.7k | if (ctxt->attsDefault != NULL) { |
9066 | 25.4k | xmlDefAttrsPtr defaults; |
9067 | | |
9068 | 25.4k | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9069 | 25.4k | if (defaults != NULL) { |
9070 | 123k | for (i = 0; i < defaults->nbAttrs; i++) { |
9071 | 99.6k | xmlDefAttr *attr = &defaults->attrs[i]; |
9072 | | |
9073 | 99.6k | attname = attr->name.name; |
9074 | 99.6k | aprefix = attr->prefix.name; |
9075 | | |
9076 | 99.6k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9077 | 8.09k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9078 | | |
9079 | 8.09k | if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0) |
9080 | 7.64k | nbNs++; |
9081 | 91.5k | } else if (aprefix == ctxt->str_xmlns) { |
9082 | 72.8k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9083 | | |
9084 | 72.8k | if (xmlParserNsPush(ctxt, &attr->name, &attr->value, |
9085 | 72.8k | NULL, 1) > 0) |
9086 | 72.2k | nbNs++; |
9087 | 72.8k | } else { |
9088 | 18.6k | if (nratts + nbTotalDef >= XML_MAX_ATTRS) { |
9089 | 0 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, |
9090 | 0 | "Maximum number of attributes exceeded"); |
9091 | 0 | break; |
9092 | 0 | } |
9093 | 18.6k | nbTotalDef += 1; |
9094 | 18.6k | } |
9095 | 99.6k | } |
9096 | 23.7k | } |
9097 | 25.4k | } |
9098 | | |
9099 | | /* |
9100 | | * Resolve attribute namespaces |
9101 | | */ |
9102 | 98.6k | for (i = 0; i < nbatts; i += 5) { |
9103 | 16.8k | attname = atts[i]; |
9104 | 16.8k | aprefix = atts[i+1]; |
9105 | | |
9106 | | /* |
9107 | | * The default namespace does not apply to attribute names. |
9108 | | */ |
9109 | 16.8k | if (aprefix == NULL) { |
9110 | 9.33k | nsIndex = NS_INDEX_EMPTY; |
9111 | 9.33k | } else if (aprefix == ctxt->str_xml) { |
9112 | 3.90k | nsIndex = NS_INDEX_XML; |
9113 | 3.90k | } else { |
9114 | 3.63k | haprefix.name = aprefix; |
9115 | 3.63k | haprefix.hashValue = (size_t) atts[i+2]; |
9116 | 3.63k | nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL); |
9117 | | |
9118 | 3.63k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) { |
9119 | 1.78k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9120 | 1.78k | "Namespace prefix %s for %s on %s is not defined\n", |
9121 | 1.78k | aprefix, attname, localname); |
9122 | 1.78k | nsIndex = NS_INDEX_EMPTY; |
9123 | 1.78k | } |
9124 | 3.63k | } |
9125 | | |
9126 | 16.8k | atts[i+2] = XML_INT_TO_PTR(nsIndex); |
9127 | 16.8k | } |
9128 | | |
9129 | | /* |
9130 | | * Maximum number of attributes including default attributes. |
9131 | | */ |
9132 | 81.7k | maxAtts = nratts + nbTotalDef; |
9133 | | |
9134 | | /* |
9135 | | * Verify that attribute names are unique. |
9136 | | */ |
9137 | 81.7k | if (maxAtts > 1) { |
9138 | 6.53k | attrHashSize = 4; |
9139 | 11.0k | while (attrHashSize / 2 < (unsigned) maxAtts) |
9140 | 4.50k | attrHashSize *= 2; |
9141 | | |
9142 | 6.53k | if (attrHashSize > ctxt->attrHashMax) { |
9143 | 1.01k | xmlAttrHashBucket *tmp; |
9144 | | |
9145 | 1.01k | tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0])); |
9146 | 1.01k | if (tmp == NULL) { |
9147 | 0 | xmlErrMemory(ctxt); |
9148 | 0 | goto done; |
9149 | 0 | } |
9150 | | |
9151 | 1.01k | ctxt->attrHash = tmp; |
9152 | 1.01k | ctxt->attrHashMax = attrHashSize; |
9153 | 1.01k | } |
9154 | | |
9155 | 6.53k | memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0])); |
9156 | | |
9157 | 18.8k | for (i = 0, j = 0; j < nratts; i += 5, j++) { |
9158 | 12.3k | const xmlChar *nsuri; |
9159 | 12.3k | unsigned hashValue, nameHashValue, uriHashValue; |
9160 | 12.3k | int res; |
9161 | | |
9162 | 12.3k | attname = atts[i]; |
9163 | 12.3k | aprefix = atts[i+1]; |
9164 | 12.3k | nsIndex = XML_PTR_TO_INT(atts[i+2]); |
9165 | | /* Hash values always have bit 31 set, see dict.c */ |
9166 | 12.3k | nameHashValue = ctxt->attallocs[j] | 0x80000000; |
9167 | | |
9168 | 12.3k | if (nsIndex == NS_INDEX_EMPTY) { |
9169 | | /* |
9170 | | * Prefix with empty namespace means an undeclared |
9171 | | * prefix which was already reported above. |
9172 | | */ |
9173 | 9.01k | if (aprefix != NULL) |
9174 | 1.67k | continue; |
9175 | 7.33k | nsuri = NULL; |
9176 | 7.33k | uriHashValue = URI_HASH_EMPTY; |
9177 | 7.33k | } else if (nsIndex == NS_INDEX_XML) { |
9178 | 1.59k | nsuri = ctxt->str_xml_ns; |
9179 | 1.59k | uriHashValue = URI_HASH_XML; |
9180 | 1.75k | } else { |
9181 | 1.75k | nsuri = ctxt->nsTab[nsIndex * 2 + 1]; |
9182 | 1.75k | uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue; |
9183 | 1.75k | } |
9184 | | |
9185 | 10.6k | hashValue = xmlDictCombineHash(nameHashValue, uriHashValue); |
9186 | 10.6k | res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri, |
9187 | 10.6k | hashValue, i); |
9188 | 10.6k | if (res < 0) |
9189 | 0 | continue; |
9190 | | |
9191 | | /* |
9192 | | * [ WFC: Unique Att Spec ] |
9193 | | * No attribute name may appear more than once in the same |
9194 | | * start-tag or empty-element tag. |
9195 | | * As extended by the Namespace in XML REC. |
9196 | | */ |
9197 | 10.6k | if (res < INT_MAX) { |
9198 | 8.22k | if (aprefix == atts[res+1]) { |
9199 | 7.79k | xmlErrAttributeDup(ctxt, aprefix, attname); |
9200 | 7.79k | numDupErr += 1; |
9201 | 7.79k | } else { |
9202 | 430 | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9203 | 430 | "Namespaced Attribute %s in '%s' redefined\n", |
9204 | 430 | attname, nsuri, NULL); |
9205 | 430 | numNsErr += 1; |
9206 | 430 | } |
9207 | 8.22k | } |
9208 | 10.6k | } |
9209 | 6.53k | } |
9210 | | |
9211 | | /* |
9212 | | * Default attributes |
9213 | | */ |
9214 | 81.7k | if (ctxt->attsDefault != NULL) { |
9215 | 25.4k | xmlDefAttrsPtr defaults; |
9216 | | |
9217 | 25.4k | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9218 | 25.4k | if (defaults != NULL) { |
9219 | 123k | for (i = 0; i < defaults->nbAttrs; i++) { |
9220 | 99.6k | xmlDefAttr *attr = &defaults->attrs[i]; |
9221 | 99.6k | const xmlChar *nsuri = NULL; |
9222 | 99.6k | unsigned hashValue, uriHashValue = 0; |
9223 | 99.6k | int res; |
9224 | | |
9225 | 99.6k | attname = attr->name.name; |
9226 | 99.6k | aprefix = attr->prefix.name; |
9227 | | |
9228 | 99.6k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) |
9229 | 8.09k | continue; |
9230 | 91.5k | if (aprefix == ctxt->str_xmlns) |
9231 | 72.8k | continue; |
9232 | | |
9233 | 18.6k | if (aprefix == NULL) { |
9234 | 6.67k | nsIndex = NS_INDEX_EMPTY; |
9235 | 6.67k | nsuri = NULL; |
9236 | 6.67k | uriHashValue = URI_HASH_EMPTY; |
9237 | 11.9k | } else if (aprefix == ctxt->str_xml) { |
9238 | 467 | nsIndex = NS_INDEX_XML; |
9239 | 467 | nsuri = ctxt->str_xml_ns; |
9240 | 467 | uriHashValue = URI_HASH_XML; |
9241 | 11.5k | } else { |
9242 | 11.5k | nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL); |
9243 | 11.5k | if ((nsIndex == INT_MAX) || |
9244 | 11.5k | (nsIndex < ctxt->nsdb->minNsIndex)) { |
9245 | 7.16k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9246 | 7.16k | "Namespace prefix %s for %s on %s is not " |
9247 | 7.16k | "defined\n", |
9248 | 7.16k | aprefix, attname, localname); |
9249 | 7.16k | nsIndex = NS_INDEX_EMPTY; |
9250 | 7.16k | nsuri = NULL; |
9251 | 7.16k | uriHashValue = URI_HASH_EMPTY; |
9252 | 7.16k | } else { |
9253 | 4.35k | nsuri = ctxt->nsTab[nsIndex * 2 + 1]; |
9254 | 4.35k | uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue; |
9255 | 4.35k | } |
9256 | 11.5k | } |
9257 | | |
9258 | | /* |
9259 | | * Check whether the attribute exists |
9260 | | */ |
9261 | 18.6k | if (maxAtts > 1) { |
9262 | 13.1k | hashValue = xmlDictCombineHash(attr->name.hashValue, |
9263 | 13.1k | uriHashValue); |
9264 | 13.1k | res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri, |
9265 | 13.1k | hashValue, nbatts); |
9266 | 13.1k | if (res < 0) |
9267 | 0 | continue; |
9268 | 13.1k | if (res < INT_MAX) { |
9269 | 1.83k | if (aprefix == atts[res+1]) |
9270 | 393 | continue; |
9271 | 1.44k | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9272 | 1.44k | "Namespaced Attribute %s in '%s' redefined\n", |
9273 | 1.44k | attname, nsuri, NULL); |
9274 | 1.44k | } |
9275 | 13.1k | } |
9276 | | |
9277 | 18.2k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9278 | | |
9279 | 18.2k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9280 | 653 | res = xmlCtxtGrowAttrs(ctxt); |
9281 | | |
9282 | 653 | maxatts = ctxt->maxatts; |
9283 | 653 | atts = ctxt->atts; |
9284 | | |
9285 | 653 | if (res < 0) { |
9286 | 0 | localname = NULL; |
9287 | 0 | goto done; |
9288 | 0 | } |
9289 | 653 | } |
9290 | | |
9291 | 18.2k | atts[nbatts++] = attname; |
9292 | 18.2k | atts[nbatts++] = aprefix; |
9293 | 18.2k | atts[nbatts++] = XML_INT_TO_PTR(nsIndex); |
9294 | 18.2k | atts[nbatts++] = attr->value.name; |
9295 | 18.2k | atts[nbatts++] = attr->valueEnd; |
9296 | 18.2k | if ((ctxt->standalone == 1) && (attr->external != 0)) { |
9297 | 0 | xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED, |
9298 | 0 | "standalone: attribute %s on %s defaulted " |
9299 | 0 | "from external subset\n", |
9300 | 0 | attname, localname); |
9301 | 0 | } |
9302 | 18.2k | nbdef++; |
9303 | 18.2k | } |
9304 | 23.7k | } |
9305 | 25.4k | } |
9306 | | |
9307 | | /* |
9308 | | * Using a single hash table for nsUri/localName pairs cannot |
9309 | | * detect duplicate QNames reliably. The following example will |
9310 | | * only result in two namespace errors. |
9311 | | * |
9312 | | * <doc xmlns:a="a" xmlns:b="a"> |
9313 | | * <elem a:a="" b:a="" b:a=""/> |
9314 | | * </doc> |
9315 | | * |
9316 | | * If we saw more than one namespace error but no duplicate QNames |
9317 | | * were found, we have to scan for duplicate QNames. |
9318 | | */ |
9319 | 81.7k | if ((numDupErr == 0) && (numNsErr > 1)) { |
9320 | 138 | memset(ctxt->attrHash, -1, |
9321 | 138 | attrHashSize * sizeof(ctxt->attrHash[0])); |
9322 | | |
9323 | 1.02k | for (i = 0, j = 0; j < nratts; i += 5, j++) { |
9324 | 882 | unsigned hashValue, nameHashValue, prefixHashValue; |
9325 | 882 | int res; |
9326 | | |
9327 | 882 | aprefix = atts[i+1]; |
9328 | 882 | if (aprefix == NULL) |
9329 | 99 | continue; |
9330 | | |
9331 | 783 | attname = atts[i]; |
9332 | | /* Hash values always have bit 31 set, see dict.c */ |
9333 | 783 | nameHashValue = ctxt->attallocs[j] | 0x80000000; |
9334 | 783 | prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix); |
9335 | | |
9336 | 783 | hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue); |
9337 | 783 | res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname, |
9338 | 783 | aprefix, hashValue, i); |
9339 | 783 | if (res < INT_MAX) |
9340 | 303 | xmlErrAttributeDup(ctxt, aprefix, attname); |
9341 | 783 | } |
9342 | 138 | } |
9343 | | |
9344 | | /* |
9345 | | * Reconstruct attribute pointers |
9346 | | */ |
9347 | 116k | for (i = 0, j = 0; i < nbatts; i += 5, j++) { |
9348 | | /* namespace URI */ |
9349 | 35.1k | nsIndex = XML_PTR_TO_INT(atts[i+2]); |
9350 | 35.1k | if (nsIndex == INT_MAX) |
9351 | 24.6k | atts[i+2] = NULL; |
9352 | 10.5k | else if (nsIndex == INT_MAX - 1) |
9353 | 4.29k | atts[i+2] = ctxt->str_xml_ns; |
9354 | 6.20k | else |
9355 | 6.20k | atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1]; |
9356 | | |
9357 | 35.1k | if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) { |
9358 | 15.2k | atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]); /* value */ |
9359 | 15.2k | atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]); /* valuend */ |
9360 | 15.2k | } |
9361 | 35.1k | } |
9362 | | |
9363 | 81.7k | uri = xmlParserNsLookupUri(ctxt, &hprefix); |
9364 | 81.7k | if ((prefix != NULL) && (uri == NULL)) { |
9365 | 4.04k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9366 | 4.04k | "Namespace prefix %s on %s is not defined\n", |
9367 | 4.04k | prefix, localname, NULL); |
9368 | 4.04k | } |
9369 | 81.7k | *pref = prefix; |
9370 | 81.7k | *URI = uri; |
9371 | | |
9372 | | /* |
9373 | | * SAX callback |
9374 | | */ |
9375 | 81.7k | if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) && |
9376 | 81.7k | (!ctxt->disableSAX)) { |
9377 | 37.1k | if (nbNs > 0) |
9378 | 8.03k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri, |
9379 | 8.03k | nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs), |
9380 | 8.03k | nbatts / 5, nbdef, atts); |
9381 | 29.1k | else |
9382 | 29.1k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri, |
9383 | 29.1k | 0, NULL, nbatts / 5, nbdef, atts); |
9384 | 37.1k | } |
9385 | | |
9386 | 81.7k | done: |
9387 | | /* |
9388 | | * Free allocated attribute values |
9389 | | */ |
9390 | 81.7k | if (attval != 0) { |
9391 | 3.06k | for (i = 0, j = 0; j < nratts; i += 5, j++) |
9392 | 1.86k | if (ctxt->attallocs[j] & 0x80000000) |
9393 | 1.61k | xmlFree((xmlChar *) atts[i+3]); |
9394 | 1.19k | } |
9395 | | |
9396 | 81.7k | *nbNsPtr = nbNs; |
9397 | 81.7k | return(localname); |
9398 | 81.7k | } |
9399 | | |
9400 | | /** |
9401 | | * Parse an end tag. Always consumes '</'. |
9402 | | * |
9403 | | * [42] ETag ::= '</' Name S? '>' |
9404 | | * |
9405 | | * With namespace |
9406 | | * |
9407 | | * [NS 9] ETag ::= '</' QName S? '>' |
9408 | | * @param ctxt an XML parser context |
9409 | | * @param tag the corresponding start tag |
9410 | | */ |
9411 | | |
9412 | | static void |
9413 | 10.8k | xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) { |
9414 | 10.8k | const xmlChar *name; |
9415 | | |
9416 | 10.8k | GROW; |
9417 | 10.8k | if ((RAW != '<') || (NXT(1) != '/')) { |
9418 | 39 | xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL); |
9419 | 39 | return; |
9420 | 39 | } |
9421 | 10.7k | SKIP(2); |
9422 | | |
9423 | 10.7k | if (tag->prefix == NULL) |
9424 | 8.90k | name = xmlParseNameAndCompare(ctxt, ctxt->name); |
9425 | 1.86k | else |
9426 | 1.86k | name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix); |
9427 | | |
9428 | | /* |
9429 | | * We should definitely be at the ending "S? '>'" part |
9430 | | */ |
9431 | 10.7k | GROW; |
9432 | 10.7k | SKIP_BLANKS; |
9433 | 10.7k | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
9434 | 3.90k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
9435 | 3.90k | } else |
9436 | 6.86k | NEXT1; |
9437 | | |
9438 | | /* |
9439 | | * [ WFC: Element Type Match ] |
9440 | | * The Name in an element's end-tag must match the element type in the |
9441 | | * start-tag. |
9442 | | * |
9443 | | */ |
9444 | 10.7k | if (name != (xmlChar*)1) { |
9445 | 8.37k | if (name == NULL) name = BAD_CAST "unparsable"; |
9446 | 8.37k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
9447 | 8.37k | "Opening and ending tag mismatch: %s line %d and %s\n", |
9448 | 8.37k | ctxt->name, tag->line, name); |
9449 | 8.37k | } |
9450 | | |
9451 | | /* |
9452 | | * SAX: End of Tag |
9453 | | */ |
9454 | 10.7k | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9455 | 10.7k | (!ctxt->disableSAX)) |
9456 | 2.61k | ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix, |
9457 | 2.61k | tag->URI); |
9458 | | |
9459 | 10.7k | spacePop(ctxt); |
9460 | 10.7k | if (tag->nsNr != 0) |
9461 | 1.44k | xmlParserNsPop(ctxt, tag->nsNr); |
9462 | 10.7k | } |
9463 | | |
9464 | | /** |
9465 | | * Parse escaped pure raw content. Always consumes '<!['. |
9466 | | * |
9467 | | * @deprecated Internal function, don't use. |
9468 | | * |
9469 | | * [18] CDSect ::= CDStart CData CDEnd |
9470 | | * |
9471 | | * [19] CDStart ::= '<![CDATA[' |
9472 | | * |
9473 | | * [20] Data ::= (Char* - (Char* ']]>' Char*)) |
9474 | | * |
9475 | | * [21] CDEnd ::= ']]>' |
9476 | | * @param ctxt an XML parser context |
9477 | | */ |
9478 | | void |
9479 | 1.80k | xmlParseCDSect(xmlParserCtxtPtr ctxt) { |
9480 | 1.80k | xmlChar *buf = NULL; |
9481 | 1.80k | int len = 0; |
9482 | 1.80k | int size = XML_PARSER_BUFFER_SIZE; |
9483 | 1.80k | int r, rl; |
9484 | 1.80k | int s, sl; |
9485 | 1.80k | int cur, l; |
9486 | 1.80k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9487 | 0 | XML_MAX_HUGE_LENGTH : |
9488 | 1.80k | XML_MAX_TEXT_LENGTH; |
9489 | | |
9490 | 1.80k | if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '[')) |
9491 | 0 | return; |
9492 | 1.80k | SKIP(3); |
9493 | | |
9494 | 1.80k | if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '[')) |
9495 | 0 | return; |
9496 | 1.80k | SKIP(6); |
9497 | | |
9498 | 1.80k | r = xmlCurrentCharRecover(ctxt, &rl); |
9499 | 1.80k | if (!IS_CHAR(r)) { |
9500 | 140 | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9501 | 140 | goto out; |
9502 | 140 | } |
9503 | 1.66k | NEXTL(rl); |
9504 | 1.66k | s = xmlCurrentCharRecover(ctxt, &sl); |
9505 | 1.66k | if (!IS_CHAR(s)) { |
9506 | 249 | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9507 | 249 | goto out; |
9508 | 249 | } |
9509 | 1.41k | NEXTL(sl); |
9510 | 1.41k | cur = xmlCurrentCharRecover(ctxt, &l); |
9511 | 1.41k | buf = xmlMalloc(size); |
9512 | 1.41k | if (buf == NULL) { |
9513 | 0 | xmlErrMemory(ctxt); |
9514 | 0 | goto out; |
9515 | 0 | } |
9516 | 9.63k | while (IS_CHAR(cur) && |
9517 | 9.63k | ((r != ']') || (s != ']') || (cur != '>'))) { |
9518 | 8.21k | if (len + 5 >= size) { |
9519 | 53 | xmlChar *tmp; |
9520 | 53 | int newSize; |
9521 | | |
9522 | 53 | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
9523 | 53 | if (newSize < 0) { |
9524 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9525 | 0 | "CData section too big found\n"); |
9526 | 0 | goto out; |
9527 | 0 | } |
9528 | 53 | tmp = xmlRealloc(buf, newSize); |
9529 | 53 | if (tmp == NULL) { |
9530 | 0 | xmlErrMemory(ctxt); |
9531 | 0 | goto out; |
9532 | 0 | } |
9533 | 53 | buf = tmp; |
9534 | 53 | size = newSize; |
9535 | 53 | } |
9536 | 8.21k | COPY_BUF(buf, len, r); |
9537 | 8.21k | r = s; |
9538 | 8.21k | rl = sl; |
9539 | 8.21k | s = cur; |
9540 | 8.21k | sl = l; |
9541 | 8.21k | NEXTL(l); |
9542 | 8.21k | cur = xmlCurrentCharRecover(ctxt, &l); |
9543 | 8.21k | } |
9544 | 1.41k | buf[len] = 0; |
9545 | 1.41k | if (cur != '>') { |
9546 | 573 | xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9547 | 573 | "CData section not finished\n%.50s\n", buf); |
9548 | 573 | goto out; |
9549 | 573 | } |
9550 | 845 | NEXTL(l); |
9551 | | |
9552 | | /* |
9553 | | * OK the buffer is to be consumed as cdata. |
9554 | | */ |
9555 | 845 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
9556 | 549 | if (ctxt->options & XML_PARSE_NOCDATA) { |
9557 | 130 | if (ctxt->sax->characters != NULL) |
9558 | 130 | ctxt->sax->characters(ctxt->userData, buf, len); |
9559 | 419 | } else { |
9560 | 419 | if (ctxt->sax->cdataBlock != NULL) |
9561 | 419 | ctxt->sax->cdataBlock(ctxt->userData, buf, len); |
9562 | 419 | } |
9563 | 549 | } |
9564 | | |
9565 | 1.80k | out: |
9566 | 1.80k | xmlFree(buf); |
9567 | 1.80k | } |
9568 | | |
9569 | | /** |
9570 | | * Parse a content sequence. Stops at EOF or '</'. Leaves checking of |
9571 | | * unexpected EOF to the caller. |
9572 | | * |
9573 | | * @param ctxt an XML parser context |
9574 | | */ |
9575 | | |
9576 | | static void |
9577 | 10.4k | xmlParseContentInternal(xmlParserCtxtPtr ctxt) { |
9578 | 10.4k | int oldNameNr = ctxt->nameNr; |
9579 | 10.4k | int oldSpaceNr = ctxt->spaceNr; |
9580 | 10.4k | int oldNodeNr = ctxt->nodeNr; |
9581 | | |
9582 | 10.4k | GROW; |
9583 | 226k | while ((ctxt->input->cur < ctxt->input->end) && |
9584 | 226k | (PARSER_STOPPED(ctxt) == 0)) { |
9585 | 220k | const xmlChar *cur = ctxt->input->cur; |
9586 | | |
9587 | | /* |
9588 | | * First case : a Processing Instruction. |
9589 | | */ |
9590 | 220k | if ((*cur == '<') && (cur[1] == '?')) { |
9591 | 2.20k | xmlParsePI(ctxt); |
9592 | 2.20k | } |
9593 | | |
9594 | | /* |
9595 | | * Second case : a CDSection |
9596 | | */ |
9597 | | /* 2.6.0 test was *cur not RAW */ |
9598 | 218k | else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9599 | 1.80k | xmlParseCDSect(ctxt); |
9600 | 1.80k | } |
9601 | | |
9602 | | /* |
9603 | | * Third case : a comment |
9604 | | */ |
9605 | 216k | else if ((*cur == '<') && (NXT(1) == '!') && |
9606 | 216k | (NXT(2) == '-') && (NXT(3) == '-')) { |
9607 | 1.02k | xmlParseComment(ctxt); |
9608 | 1.02k | } |
9609 | | |
9610 | | /* |
9611 | | * Fourth case : a sub-element. |
9612 | | */ |
9613 | 215k | else if (*cur == '<') { |
9614 | 153k | if (NXT(1) == '/') { |
9615 | 15.6k | if (ctxt->nameNr <= oldNameNr) |
9616 | 4.40k | break; |
9617 | 11.2k | xmlParseElementEnd(ctxt); |
9618 | 137k | } else { |
9619 | 137k | xmlParseElementStart(ctxt); |
9620 | 137k | } |
9621 | 153k | } |
9622 | | |
9623 | | /* |
9624 | | * Fifth case : a reference. If if has not been resolved, |
9625 | | * parsing returns it's Name, create the node |
9626 | | */ |
9627 | | |
9628 | 62.0k | else if (*cur == '&') { |
9629 | 12.8k | xmlParseReference(ctxt); |
9630 | 12.8k | } |
9631 | | |
9632 | | /* |
9633 | | * Last case, text. Note that References are handled directly. |
9634 | | */ |
9635 | 49.1k | else { |
9636 | 49.1k | xmlParseCharDataInternal(ctxt, 0); |
9637 | 49.1k | } |
9638 | | |
9639 | 216k | SHRINK; |
9640 | 216k | GROW; |
9641 | 216k | } |
9642 | | |
9643 | 10.4k | if ((ctxt->nameNr > oldNameNr) && |
9644 | 10.4k | (ctxt->input->cur >= ctxt->input->end) && |
9645 | 10.4k | (ctxt->wellFormed)) { |
9646 | 29 | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9647 | 29 | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9648 | 29 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9649 | 29 | "Premature end of data in tag %s line %d\n", |
9650 | 29 | name, line, NULL); |
9651 | 29 | } |
9652 | | |
9653 | | /* |
9654 | | * Clean up in error case |
9655 | | */ |
9656 | | |
9657 | 25.1k | while (ctxt->nodeNr > oldNodeNr) |
9658 | 14.7k | nodePop(ctxt); |
9659 | | |
9660 | 36.1k | while (ctxt->nameNr > oldNameNr) { |
9661 | 25.7k | xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1]; |
9662 | | |
9663 | 25.7k | if (tag->nsNr != 0) |
9664 | 9.82k | xmlParserNsPop(ctxt, tag->nsNr); |
9665 | | |
9666 | 25.7k | namePop(ctxt); |
9667 | 25.7k | } |
9668 | | |
9669 | 36.1k | while (ctxt->spaceNr > oldSpaceNr) |
9670 | 25.7k | spacePop(ctxt); |
9671 | 10.4k | } |
9672 | | |
9673 | | /** |
9674 | | * Parse XML element content. This is useful if you're only interested |
9675 | | * in custom SAX callbacks. If you want a node list, use |
9676 | | * xmlCtxtParseContent(). |
9677 | | * |
9678 | | * @param ctxt an XML parser context |
9679 | | */ |
9680 | | void |
9681 | 0 | xmlParseContent(xmlParserCtxtPtr ctxt) { |
9682 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
9683 | 0 | return; |
9684 | | |
9685 | 0 | xmlCtxtInitializeLate(ctxt); |
9686 | |
|
9687 | 0 | xmlParseContentInternal(ctxt); |
9688 | |
|
9689 | 0 | xmlParserCheckEOF(ctxt, XML_ERR_NOT_WELL_BALANCED); |
9690 | 0 | } |
9691 | | |
9692 | | /** |
9693 | | * parse an XML element |
9694 | | * |
9695 | | * @deprecated Internal function, don't use. |
9696 | | * |
9697 | | * [39] element ::= EmptyElemTag | STag content ETag |
9698 | | * |
9699 | | * [ WFC: Element Type Match ] |
9700 | | * The Name in an element's end-tag must match the element type in the |
9701 | | * start-tag. |
9702 | | * |
9703 | | * @param ctxt an XML parser context |
9704 | | */ |
9705 | | |
9706 | | void |
9707 | 17.4k | xmlParseElement(xmlParserCtxtPtr ctxt) { |
9708 | 17.4k | if (xmlParseElementStart(ctxt) != 0) |
9709 | 8.07k | return; |
9710 | | |
9711 | 9.40k | xmlParseContentInternal(ctxt); |
9712 | | |
9713 | 9.40k | if (ctxt->input->cur >= ctxt->input->end) { |
9714 | 4.97k | if (ctxt->wellFormed) { |
9715 | 26 | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9716 | 26 | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9717 | 26 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9718 | 26 | "Premature end of data in tag %s line %d\n", |
9719 | 26 | name, line, NULL); |
9720 | 26 | } |
9721 | 4.97k | return; |
9722 | 4.97k | } |
9723 | | |
9724 | 4.42k | xmlParseElementEnd(ctxt); |
9725 | 4.42k | } |
9726 | | |
9727 | | /** |
9728 | | * Parse the start of an XML element. Returns -1 in case of error, 0 if an |
9729 | | * opening tag was parsed, 1 if an empty element was parsed. |
9730 | | * |
9731 | | * Always consumes '<'. |
9732 | | * |
9733 | | * @param ctxt an XML parser context |
9734 | | */ |
9735 | | static int |
9736 | 155k | xmlParseElementStart(xmlParserCtxtPtr ctxt) { |
9737 | 155k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
9738 | 155k | const xmlChar *name; |
9739 | 155k | const xmlChar *prefix = NULL; |
9740 | 155k | const xmlChar *URI = NULL; |
9741 | 155k | xmlParserNodeInfo node_info; |
9742 | 155k | int line; |
9743 | 155k | xmlNodePtr cur; |
9744 | 155k | int nbNs = 0; |
9745 | | |
9746 | 155k | if (ctxt->nameNr > maxDepth) { |
9747 | 1 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
9748 | 1 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
9749 | 1 | ctxt->nameNr); |
9750 | 1 | xmlHaltParser(ctxt); |
9751 | 1 | return(-1); |
9752 | 1 | } |
9753 | | |
9754 | | /* Capture start position */ |
9755 | 155k | if (ctxt->record_info) { |
9756 | 0 | node_info.begin_pos = ctxt->input->consumed + |
9757 | 0 | (CUR_PTR - ctxt->input->base); |
9758 | 0 | node_info.begin_line = ctxt->input->line; |
9759 | 0 | } |
9760 | | |
9761 | 155k | if (ctxt->spaceNr == 0) |
9762 | 0 | spacePush(ctxt, -1); |
9763 | 155k | else if (*ctxt->space == -2) |
9764 | 18.3k | spacePush(ctxt, -1); |
9765 | 136k | else |
9766 | 136k | spacePush(ctxt, *ctxt->space); |
9767 | | |
9768 | 155k | line = ctxt->input->line; |
9769 | 155k | #ifdef LIBXML_SAX1_ENABLED |
9770 | 155k | if (ctxt->sax2) |
9771 | 98.7k | #endif /* LIBXML_SAX1_ENABLED */ |
9772 | 98.7k | name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs); |
9773 | 56.4k | #ifdef LIBXML_SAX1_ENABLED |
9774 | 56.4k | else |
9775 | 56.4k | name = xmlParseStartTag(ctxt); |
9776 | 155k | #endif /* LIBXML_SAX1_ENABLED */ |
9777 | 155k | if (name == NULL) { |
9778 | 23.7k | spacePop(ctxt); |
9779 | 23.7k | return(-1); |
9780 | 23.7k | } |
9781 | 131k | nameNsPush(ctxt, name, prefix, URI, line, nbNs); |
9782 | 131k | cur = ctxt->node; |
9783 | | |
9784 | 131k | #ifdef LIBXML_VALID_ENABLED |
9785 | | /* |
9786 | | * [ VC: Root Element Type ] |
9787 | | * The Name in the document type declaration must match the element |
9788 | | * type of the root element. |
9789 | | */ |
9790 | 131k | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
9791 | 131k | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
9792 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
9793 | 131k | #endif /* LIBXML_VALID_ENABLED */ |
9794 | | |
9795 | | /* |
9796 | | * Check for an Empty Element. |
9797 | | */ |
9798 | 131k | if ((RAW == '/') && (NXT(1) == '>')) { |
9799 | 2.50k | SKIP(2); |
9800 | 2.50k | if (ctxt->sax2) { |
9801 | 1.81k | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9802 | 1.81k | (!ctxt->disableSAX)) |
9803 | 503 | ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI); |
9804 | 1.81k | #ifdef LIBXML_SAX1_ENABLED |
9805 | 1.81k | } else { |
9806 | 693 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
9807 | 693 | (!ctxt->disableSAX)) |
9808 | 365 | ctxt->sax->endElement(ctxt->userData, name); |
9809 | 693 | #endif /* LIBXML_SAX1_ENABLED */ |
9810 | 693 | } |
9811 | 2.50k | namePop(ctxt); |
9812 | 2.50k | spacePop(ctxt); |
9813 | 2.50k | if (nbNs > 0) |
9814 | 596 | xmlParserNsPop(ctxt, nbNs); |
9815 | 2.50k | if (cur != NULL && ctxt->record_info) { |
9816 | 0 | node_info.node = cur; |
9817 | 0 | node_info.end_pos = ctxt->input->consumed + |
9818 | 0 | (CUR_PTR - ctxt->input->base); |
9819 | 0 | node_info.end_line = ctxt->input->line; |
9820 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
9821 | 0 | } |
9822 | 2.50k | return(1); |
9823 | 2.50k | } |
9824 | 129k | if (RAW == '>') { |
9825 | 46.4k | NEXT1; |
9826 | 46.4k | if (cur != NULL && ctxt->record_info) { |
9827 | 0 | node_info.node = cur; |
9828 | 0 | node_info.end_pos = 0; |
9829 | 0 | node_info.end_line = 0; |
9830 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
9831 | 0 | } |
9832 | 82.6k | } else { |
9833 | 82.6k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED, |
9834 | 82.6k | "Couldn't find end of Start Tag %s line %d\n", |
9835 | 82.6k | name, line, NULL); |
9836 | | |
9837 | | /* |
9838 | | * end of parsing of this node. |
9839 | | */ |
9840 | 82.6k | nodePop(ctxt); |
9841 | 82.6k | namePop(ctxt); |
9842 | 82.6k | spacePop(ctxt); |
9843 | 82.6k | if (nbNs > 0) |
9844 | 11.0k | xmlParserNsPop(ctxt, nbNs); |
9845 | 82.6k | return(-1); |
9846 | 82.6k | } |
9847 | | |
9848 | 46.4k | return(0); |
9849 | 129k | } |
9850 | | |
9851 | | /** |
9852 | | * Parse the end of an XML element. Always consumes '</'. |
9853 | | * |
9854 | | * @param ctxt an XML parser context |
9855 | | */ |
9856 | | static void |
9857 | 15.7k | xmlParseElementEnd(xmlParserCtxtPtr ctxt) { |
9858 | 15.7k | xmlNodePtr cur = ctxt->node; |
9859 | | |
9860 | 15.7k | if (ctxt->nameNr <= 0) { |
9861 | 0 | if ((RAW == '<') && (NXT(1) == '/')) |
9862 | 0 | SKIP(2); |
9863 | 0 | return; |
9864 | 0 | } |
9865 | | |
9866 | | /* |
9867 | | * parse the end of tag: '</' should be here. |
9868 | | */ |
9869 | 15.7k | if (ctxt->sax2) { |
9870 | 10.8k | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
9871 | 10.8k | namePop(ctxt); |
9872 | 10.8k | } |
9873 | 4.90k | #ifdef LIBXML_SAX1_ENABLED |
9874 | 4.90k | else |
9875 | 4.90k | xmlParseEndTag1(ctxt, 0); |
9876 | 15.7k | #endif /* LIBXML_SAX1_ENABLED */ |
9877 | | |
9878 | | /* |
9879 | | * Capture end position |
9880 | | */ |
9881 | 15.7k | if (cur != NULL && ctxt->record_info) { |
9882 | 0 | xmlParserNodeInfoPtr node_info; |
9883 | |
|
9884 | 0 | node_info = (xmlParserNodeInfoPtr) xmlParserFindNodeInfo(ctxt, cur); |
9885 | 0 | if (node_info != NULL) { |
9886 | 0 | node_info->end_pos = ctxt->input->consumed + |
9887 | 0 | (CUR_PTR - ctxt->input->base); |
9888 | 0 | node_info->end_line = ctxt->input->line; |
9889 | 0 | } |
9890 | 0 | } |
9891 | 15.7k | } |
9892 | | |
9893 | | /** |
9894 | | * parse the XML version value. |
9895 | | * |
9896 | | * @deprecated Internal function, don't use. |
9897 | | * |
9898 | | * [26] VersionNum ::= '1.' [0-9]+ |
9899 | | * |
9900 | | * In practice allow [0-9].[0-9]+ at that level |
9901 | | * |
9902 | | * @param ctxt an XML parser context |
9903 | | * @returns the string giving the XML version number, or NULL |
9904 | | */ |
9905 | | xmlChar * |
9906 | 7.96k | xmlParseVersionNum(xmlParserCtxtPtr ctxt) { |
9907 | 7.96k | xmlChar *buf = NULL; |
9908 | 7.96k | int len = 0; |
9909 | 7.96k | int size = 10; |
9910 | 7.96k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9911 | 0 | XML_MAX_TEXT_LENGTH : |
9912 | 7.96k | XML_MAX_NAME_LENGTH; |
9913 | 7.96k | xmlChar cur; |
9914 | | |
9915 | 7.96k | buf = xmlMalloc(size); |
9916 | 7.96k | if (buf == NULL) { |
9917 | 0 | xmlErrMemory(ctxt); |
9918 | 0 | return(NULL); |
9919 | 0 | } |
9920 | 7.96k | cur = CUR; |
9921 | 7.96k | if (!((cur >= '0') && (cur <= '9'))) { |
9922 | 7.89k | xmlFree(buf); |
9923 | 7.89k | return(NULL); |
9924 | 7.89k | } |
9925 | 63 | buf[len++] = cur; |
9926 | 63 | NEXT; |
9927 | 63 | cur=CUR; |
9928 | 63 | if (cur != '.') { |
9929 | 24 | xmlFree(buf); |
9930 | 24 | return(NULL); |
9931 | 24 | } |
9932 | 39 | buf[len++] = cur; |
9933 | 39 | NEXT; |
9934 | 39 | cur=CUR; |
9935 | 84 | while ((cur >= '0') && (cur <= '9')) { |
9936 | 46 | if (len + 1 >= size) { |
9937 | 1 | xmlChar *tmp; |
9938 | 1 | int newSize; |
9939 | | |
9940 | 1 | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
9941 | 1 | if (newSize) { |
9942 | 1 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum"); |
9943 | 1 | xmlFree(buf); |
9944 | 1 | return(NULL); |
9945 | 1 | } |
9946 | 0 | tmp = xmlRealloc(buf, newSize); |
9947 | 0 | if (tmp == NULL) { |
9948 | 0 | xmlErrMemory(ctxt); |
9949 | 0 | xmlFree(buf); |
9950 | 0 | return(NULL); |
9951 | 0 | } |
9952 | 0 | buf = tmp; |
9953 | 0 | size = newSize; |
9954 | 0 | } |
9955 | 45 | buf[len++] = cur; |
9956 | 45 | NEXT; |
9957 | 45 | cur=CUR; |
9958 | 45 | } |
9959 | 38 | buf[len] = 0; |
9960 | 38 | return(buf); |
9961 | 39 | } |
9962 | | |
9963 | | /** |
9964 | | * parse the XML version. |
9965 | | * |
9966 | | * @deprecated Internal function, don't use. |
9967 | | * |
9968 | | * [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ") |
9969 | | * |
9970 | | * [25] Eq ::= S? '=' S? |
9971 | | * |
9972 | | * @param ctxt an XML parser context |
9973 | | * @returns the version string, e.g. "1.0" |
9974 | | */ |
9975 | | |
9976 | | xmlChar * |
9977 | 11.5k | xmlParseVersionInfo(xmlParserCtxtPtr ctxt) { |
9978 | 11.5k | xmlChar *version = NULL; |
9979 | | |
9980 | 11.5k | if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) { |
9981 | 7.96k | SKIP(7); |
9982 | 7.96k | SKIP_BLANKS; |
9983 | 7.96k | if (RAW != '=') { |
9984 | 2 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
9985 | 2 | return(NULL); |
9986 | 2 | } |
9987 | 7.96k | NEXT; |
9988 | 7.96k | SKIP_BLANKS; |
9989 | 7.96k | if (RAW == '"') { |
9990 | 7.96k | NEXT; |
9991 | 7.96k | version = xmlParseVersionNum(ctxt); |
9992 | 7.96k | if (RAW != '"') { |
9993 | 7.83k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
9994 | 7.83k | } else |
9995 | 122 | NEXT; |
9996 | 7.96k | } else if (RAW == '\''){ |
9997 | 2 | NEXT; |
9998 | 2 | version = xmlParseVersionNum(ctxt); |
9999 | 2 | if (RAW != '\'') { |
10000 | 1 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10001 | 1 | } else |
10002 | 1 | NEXT; |
10003 | 2 | } else { |
10004 | 1 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10005 | 1 | } |
10006 | 7.96k | } |
10007 | 11.5k | return(version); |
10008 | 11.5k | } |
10009 | | |
10010 | | /** |
10011 | | * parse the XML encoding name |
10012 | | * |
10013 | | * @deprecated Internal function, don't use. |
10014 | | * |
10015 | | * [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* |
10016 | | * |
10017 | | * @param ctxt an XML parser context |
10018 | | * @returns the encoding name value or NULL |
10019 | | */ |
10020 | | xmlChar * |
10021 | 3.59k | xmlParseEncName(xmlParserCtxtPtr ctxt) { |
10022 | 3.59k | xmlChar *buf = NULL; |
10023 | 3.59k | int len = 0; |
10024 | 3.59k | int size = 10; |
10025 | 3.59k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
10026 | 0 | XML_MAX_TEXT_LENGTH : |
10027 | 3.59k | XML_MAX_NAME_LENGTH; |
10028 | 3.59k | xmlChar cur; |
10029 | | |
10030 | 3.59k | cur = CUR; |
10031 | 3.59k | if (((cur >= 'a') && (cur <= 'z')) || |
10032 | 3.59k | ((cur >= 'A') && (cur <= 'Z'))) { |
10033 | 2.38k | buf = xmlMalloc(size); |
10034 | 2.38k | if (buf == NULL) { |
10035 | 0 | xmlErrMemory(ctxt); |
10036 | 0 | return(NULL); |
10037 | 0 | } |
10038 | | |
10039 | 2.38k | buf[len++] = cur; |
10040 | 2.38k | NEXT; |
10041 | 2.38k | cur = CUR; |
10042 | 12.5k | while (((cur >= 'a') && (cur <= 'z')) || |
10043 | 12.5k | ((cur >= 'A') && (cur <= 'Z')) || |
10044 | 12.5k | ((cur >= '0') && (cur <= '9')) || |
10045 | 12.5k | (cur == '.') || (cur == '_') || |
10046 | 12.5k | (cur == '-')) { |
10047 | 10.1k | if (len + 1 >= size) { |
10048 | 272 | xmlChar *tmp; |
10049 | 272 | int newSize; |
10050 | | |
10051 | 272 | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
10052 | 272 | if (newSize < 0) { |
10053 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName"); |
10054 | 0 | xmlFree(buf); |
10055 | 0 | return(NULL); |
10056 | 0 | } |
10057 | 272 | tmp = xmlRealloc(buf, newSize); |
10058 | 272 | if (tmp == NULL) { |
10059 | 0 | xmlErrMemory(ctxt); |
10060 | 0 | xmlFree(buf); |
10061 | 0 | return(NULL); |
10062 | 0 | } |
10063 | 272 | buf = tmp; |
10064 | 272 | size = newSize; |
10065 | 272 | } |
10066 | 10.1k | buf[len++] = cur; |
10067 | 10.1k | NEXT; |
10068 | 10.1k | cur = CUR; |
10069 | 10.1k | } |
10070 | 2.38k | buf[len] = 0; |
10071 | 2.38k | } else { |
10072 | 1.21k | xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL); |
10073 | 1.21k | } |
10074 | 3.59k | return(buf); |
10075 | 3.59k | } |
10076 | | |
10077 | | /** |
10078 | | * parse the XML encoding declaration |
10079 | | * |
10080 | | * @deprecated Internal function, don't use. |
10081 | | * |
10082 | | * [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | |
10083 | | * "'" EncName "'") |
10084 | | * |
10085 | | * this setups the conversion filters. |
10086 | | * |
10087 | | * @param ctxt an XML parser context |
10088 | | * @returns the encoding value or NULL |
10089 | | */ |
10090 | | |
10091 | | const xmlChar * |
10092 | 11.4k | xmlParseEncodingDecl(xmlParserCtxtPtr ctxt) { |
10093 | 11.4k | xmlChar *encoding = NULL; |
10094 | | |
10095 | 11.4k | SKIP_BLANKS; |
10096 | 11.4k | if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0) |
10097 | 7.80k | return(NULL); |
10098 | | |
10099 | 3.60k | SKIP(8); |
10100 | 3.60k | SKIP_BLANKS; |
10101 | 3.60k | if (RAW != '=') { |
10102 | 3 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10103 | 3 | return(NULL); |
10104 | 3 | } |
10105 | 3.60k | NEXT; |
10106 | 3.60k | SKIP_BLANKS; |
10107 | 3.60k | if (RAW == '"') { |
10108 | 2 | NEXT; |
10109 | 2 | encoding = xmlParseEncName(ctxt); |
10110 | 2 | if (RAW != '"') { |
10111 | 1 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10112 | 1 | xmlFree((xmlChar *) encoding); |
10113 | 1 | return(NULL); |
10114 | 1 | } else |
10115 | 1 | NEXT; |
10116 | 3.59k | } else if (RAW == '\''){ |
10117 | 3.59k | NEXT; |
10118 | 3.59k | encoding = xmlParseEncName(ctxt); |
10119 | 3.59k | if (RAW != '\'') { |
10120 | 1.25k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10121 | 1.25k | xmlFree((xmlChar *) encoding); |
10122 | 1.25k | return(NULL); |
10123 | 1.25k | } else |
10124 | 2.34k | NEXT; |
10125 | 3.59k | } else { |
10126 | 3 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10127 | 3 | } |
10128 | | |
10129 | 2.34k | if (encoding == NULL) |
10130 | 11 | return(NULL); |
10131 | | |
10132 | 2.33k | xmlSetDeclaredEncoding(ctxt, encoding); |
10133 | | |
10134 | 2.33k | return(ctxt->encoding); |
10135 | 2.34k | } |
10136 | | |
10137 | | /** |
10138 | | * parse the XML standalone declaration |
10139 | | * |
10140 | | * @deprecated Internal function, don't use. |
10141 | | * |
10142 | | * [32] SDDecl ::= S 'standalone' Eq |
10143 | | * (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"')) |
10144 | | * |
10145 | | * [ VC: Standalone Document Declaration ] |
10146 | | * TODO The standalone document declaration must have the value "no" |
10147 | | * if any external markup declarations contain declarations of: |
10148 | | * - attributes with default values, if elements to which these |
10149 | | * attributes apply appear in the document without specifications |
10150 | | * of values for these attributes, or |
10151 | | * - entities (other than amp, lt, gt, apos, quot), if references |
10152 | | * to those entities appear in the document, or |
10153 | | * - attributes with values subject to normalization, where the |
10154 | | * attribute appears in the document with a value which will change |
10155 | | * as a result of normalization, or |
10156 | | * - element types with element content, if white space occurs directly |
10157 | | * within any instance of those types. |
10158 | | * |
10159 | | * @param ctxt an XML parser context |
10160 | | * @returns |
10161 | | * 1 if standalone="yes" |
10162 | | * 0 if standalone="no" |
10163 | | * -2 if standalone attribute is missing or invalid |
10164 | | * (A standalone value of -2 means that the XML declaration was found, |
10165 | | * but no value was specified for the standalone attribute). |
10166 | | */ |
10167 | | |
10168 | | int |
10169 | 9.97k | xmlParseSDDecl(xmlParserCtxtPtr ctxt) { |
10170 | 9.97k | int standalone = -2; |
10171 | | |
10172 | 9.97k | SKIP_BLANKS; |
10173 | 9.97k | if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) { |
10174 | 75 | SKIP(10); |
10175 | 75 | SKIP_BLANKS; |
10176 | 75 | if (RAW != '=') { |
10177 | 6 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10178 | 6 | return(standalone); |
10179 | 6 | } |
10180 | 69 | NEXT; |
10181 | 69 | SKIP_BLANKS; |
10182 | 69 | if (RAW == '\''){ |
10183 | 54 | NEXT; |
10184 | 54 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10185 | 2 | standalone = 0; |
10186 | 2 | SKIP(2); |
10187 | 52 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10188 | 52 | (NXT(2) == 's')) { |
10189 | 44 | standalone = 1; |
10190 | 44 | SKIP(3); |
10191 | 44 | } else { |
10192 | 8 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10193 | 8 | } |
10194 | 54 | if (RAW != '\'') { |
10195 | 53 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10196 | 53 | } else |
10197 | 1 | NEXT; |
10198 | 54 | } else if (RAW == '"'){ |
10199 | 12 | NEXT; |
10200 | 12 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10201 | 3 | standalone = 0; |
10202 | 3 | SKIP(2); |
10203 | 9 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10204 | 9 | (NXT(2) == 's')) { |
10205 | 4 | standalone = 1; |
10206 | 4 | SKIP(3); |
10207 | 5 | } else { |
10208 | 5 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10209 | 5 | } |
10210 | 12 | if (RAW != '"') { |
10211 | 6 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10212 | 6 | } else |
10213 | 6 | NEXT; |
10214 | 12 | } else { |
10215 | 3 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10216 | 3 | } |
10217 | 69 | } |
10218 | 9.96k | return(standalone); |
10219 | 9.97k | } |
10220 | | |
10221 | | /** |
10222 | | * parse an XML declaration header |
10223 | | * |
10224 | | * @deprecated Internal function, don't use. |
10225 | | * |
10226 | | * [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' |
10227 | | * @param ctxt an XML parser context |
10228 | | */ |
10229 | | |
10230 | | void |
10231 | 11.5k | xmlParseXMLDecl(xmlParserCtxtPtr ctxt) { |
10232 | 11.5k | xmlChar *version; |
10233 | | |
10234 | | /* |
10235 | | * This value for standalone indicates that the document has an |
10236 | | * XML declaration but it does not have a standalone attribute. |
10237 | | * It will be overwritten later if a standalone attribute is found. |
10238 | | */ |
10239 | | |
10240 | 11.5k | ctxt->standalone = -2; |
10241 | | |
10242 | | /* |
10243 | | * We know that '<?xml' is here. |
10244 | | */ |
10245 | 11.5k | SKIP(5); |
10246 | | |
10247 | 11.5k | if (!IS_BLANK_CH(RAW)) { |
10248 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
10249 | 0 | "Blank needed after '<?xml'\n"); |
10250 | 0 | } |
10251 | 11.5k | SKIP_BLANKS; |
10252 | | |
10253 | | /* |
10254 | | * We must have the VersionInfo here. |
10255 | | */ |
10256 | 11.5k | version = xmlParseVersionInfo(ctxt); |
10257 | 11.5k | if (version == NULL) { |
10258 | 11.5k | xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL); |
10259 | 11.5k | } else { |
10260 | 38 | if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) { |
10261 | | /* |
10262 | | * Changed here for XML-1.0 5th edition |
10263 | | */ |
10264 | 31 | if (ctxt->options & XML_PARSE_OLD10) { |
10265 | 1 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10266 | 1 | "Unsupported version '%s'\n", |
10267 | 1 | version); |
10268 | 30 | } else { |
10269 | 30 | if ((version[0] == '1') && ((version[1] == '.'))) { |
10270 | 24 | xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION, |
10271 | 24 | "Unsupported version '%s'\n", |
10272 | 24 | version, NULL); |
10273 | 24 | } else { |
10274 | 6 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10275 | 6 | "Unsupported version '%s'\n", |
10276 | 6 | version); |
10277 | 6 | } |
10278 | 30 | } |
10279 | 31 | } |
10280 | 38 | if (ctxt->version != NULL) |
10281 | 0 | xmlFree((void *) ctxt->version); |
10282 | 38 | ctxt->version = version; |
10283 | 38 | } |
10284 | | |
10285 | | /* |
10286 | | * We may have the encoding declaration |
10287 | | */ |
10288 | 11.5k | if (!IS_BLANK_CH(RAW)) { |
10289 | 11.4k | if ((RAW == '?') && (NXT(1) == '>')) { |
10290 | 150 | SKIP(2); |
10291 | 150 | return; |
10292 | 150 | } |
10293 | 11.3k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10294 | 11.3k | } |
10295 | 11.4k | xmlParseEncodingDecl(ctxt); |
10296 | | |
10297 | | /* |
10298 | | * We may have the standalone status. |
10299 | | */ |
10300 | 11.4k | if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) { |
10301 | 2.06k | if ((RAW == '?') && (NXT(1) == '>')) { |
10302 | 1.43k | SKIP(2); |
10303 | 1.43k | return; |
10304 | 1.43k | } |
10305 | 626 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10306 | 626 | } |
10307 | | |
10308 | | /* |
10309 | | * We can grow the input buffer freely at that point |
10310 | | */ |
10311 | 9.97k | GROW; |
10312 | | |
10313 | 9.97k | SKIP_BLANKS; |
10314 | 9.97k | ctxt->standalone = xmlParseSDDecl(ctxt); |
10315 | | |
10316 | 9.97k | SKIP_BLANKS; |
10317 | 9.97k | if ((RAW == '?') && (NXT(1) == '>')) { |
10318 | 159 | SKIP(2); |
10319 | 9.81k | } else if (RAW == '>') { |
10320 | | /* Deprecated old WD ... */ |
10321 | 8.30k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10322 | 8.30k | NEXT; |
10323 | 8.30k | } else { |
10324 | 1.51k | int c; |
10325 | | |
10326 | 1.51k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10327 | 21.1k | while ((PARSER_STOPPED(ctxt) == 0) && |
10328 | 21.1k | ((c = CUR) != 0)) { |
10329 | 20.9k | NEXT; |
10330 | 20.9k | if (c == '>') |
10331 | 1.30k | break; |
10332 | 20.9k | } |
10333 | 1.51k | } |
10334 | 9.97k | } |
10335 | | |
10336 | | /** |
10337 | | * @since 2.14.0 |
10338 | | * |
10339 | | * @param ctxt parser context |
10340 | | * @returns the version from the XML declaration. |
10341 | | */ |
10342 | | const xmlChar * |
10343 | 0 | xmlCtxtGetVersion(xmlParserCtxtPtr ctxt) { |
10344 | 0 | if (ctxt == NULL) |
10345 | 0 | return(NULL); |
10346 | | |
10347 | 0 | return(ctxt->version); |
10348 | 0 | } |
10349 | | |
10350 | | /** |
10351 | | * @since 2.14.0 |
10352 | | * |
10353 | | * @param ctxt parser context |
10354 | | * @returns the value from the standalone document declaration. |
10355 | | */ |
10356 | | int |
10357 | 0 | xmlCtxtGetStandalone(xmlParserCtxtPtr ctxt) { |
10358 | 0 | if (ctxt == NULL) |
10359 | 0 | return(0); |
10360 | | |
10361 | 0 | return(ctxt->standalone); |
10362 | 0 | } |
10363 | | |
10364 | | /** |
10365 | | * parse an XML Misc* optional field. |
10366 | | * |
10367 | | * @deprecated Internal function, don't use. |
10368 | | * |
10369 | | * [27] Misc ::= Comment | PI | S |
10370 | | * @param ctxt an XML parser context |
10371 | | */ |
10372 | | |
10373 | | void |
10374 | 50.2k | xmlParseMisc(xmlParserCtxtPtr ctxt) { |
10375 | 54.3k | while (PARSER_STOPPED(ctxt) == 0) { |
10376 | 53.0k | SKIP_BLANKS; |
10377 | 53.0k | GROW; |
10378 | 53.0k | if ((RAW == '<') && (NXT(1) == '?')) { |
10379 | 2.61k | xmlParsePI(ctxt); |
10380 | 50.4k | } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) { |
10381 | 1.47k | xmlParseComment(ctxt); |
10382 | 48.9k | } else { |
10383 | 48.9k | break; |
10384 | 48.9k | } |
10385 | 53.0k | } |
10386 | 50.2k | } |
10387 | | |
10388 | | static void |
10389 | 21.9k | xmlFinishDocument(xmlParserCtxtPtr ctxt) { |
10390 | 21.9k | xmlDocPtr doc; |
10391 | | |
10392 | | /* |
10393 | | * SAX: end of the document processing. |
10394 | | */ |
10395 | 21.9k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10396 | 21.9k | ctxt->sax->endDocument(ctxt->userData); |
10397 | | |
10398 | | /* |
10399 | | * Remove locally kept entity definitions if the tree was not built |
10400 | | */ |
10401 | 21.9k | doc = ctxt->myDoc; |
10402 | 21.9k | if ((doc != NULL) && |
10403 | 21.9k | (xmlStrEqual(doc->version, SAX_COMPAT_MODE))) { |
10404 | 135 | xmlFreeDoc(doc); |
10405 | 135 | ctxt->myDoc = NULL; |
10406 | 135 | } |
10407 | 21.9k | } |
10408 | | |
10409 | | /** |
10410 | | * Parse an XML document and invoke the SAX handlers. This is useful |
10411 | | * if you're only interested in custom SAX callbacks. If you want a |
10412 | | * document tree, use xmlCtxtParseDocument(). |
10413 | | * |
10414 | | * @param ctxt an XML parser context |
10415 | | * @returns 0, -1 in case of error. |
10416 | | */ |
10417 | | |
10418 | | int |
10419 | 21.9k | xmlParseDocument(xmlParserCtxtPtr ctxt) { |
10420 | 21.9k | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10421 | 0 | return(-1); |
10422 | | |
10423 | 21.9k | GROW; |
10424 | | |
10425 | | /* |
10426 | | * SAX: detecting the level. |
10427 | | */ |
10428 | 21.9k | xmlCtxtInitializeLate(ctxt); |
10429 | | |
10430 | 21.9k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
10431 | 21.9k | ctxt->sax->setDocumentLocator(ctxt->userData, |
10432 | 21.9k | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
10433 | 21.9k | } |
10434 | | |
10435 | 21.9k | xmlDetectEncoding(ctxt); |
10436 | | |
10437 | 21.9k | if (CUR == 0) { |
10438 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10439 | 0 | return(-1); |
10440 | 0 | } |
10441 | | |
10442 | 21.9k | GROW; |
10443 | 21.9k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10444 | | |
10445 | | /* |
10446 | | * Note that we will switch encoding on the fly. |
10447 | | */ |
10448 | 11.5k | xmlParseXMLDecl(ctxt); |
10449 | 11.5k | SKIP_BLANKS; |
10450 | 11.5k | } else { |
10451 | 10.4k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10452 | 10.4k | if (ctxt->version == NULL) { |
10453 | 0 | xmlErrMemory(ctxt); |
10454 | 0 | return(-1); |
10455 | 0 | } |
10456 | 10.4k | } |
10457 | 21.9k | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10458 | 14.5k | ctxt->sax->startDocument(ctxt->userData); |
10459 | 21.9k | if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) && |
10460 | 21.9k | (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) { |
10461 | 0 | ctxt->myDoc->compression = ctxt->input->buf->compressed; |
10462 | 0 | } |
10463 | | |
10464 | | /* |
10465 | | * The Misc part of the Prolog |
10466 | | */ |
10467 | 21.9k | xmlParseMisc(ctxt); |
10468 | | |
10469 | | /* |
10470 | | * Then possibly doc type declaration(s) and more Misc |
10471 | | * (doctypedecl Misc*)? |
10472 | | */ |
10473 | 21.9k | GROW; |
10474 | 21.9k | if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) { |
10475 | | |
10476 | 10.7k | ctxt->inSubset = 1; |
10477 | 10.7k | xmlParseDocTypeDecl(ctxt); |
10478 | 10.7k | if (RAW == '[') { |
10479 | 9.02k | xmlParseInternalSubset(ctxt); |
10480 | 9.02k | } else if (RAW == '>') { |
10481 | 439 | NEXT; |
10482 | 439 | } |
10483 | | |
10484 | | /* |
10485 | | * Create and update the external subset. |
10486 | | */ |
10487 | 10.7k | ctxt->inSubset = 2; |
10488 | 10.7k | if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) && |
10489 | 10.7k | (!ctxt->disableSAX)) |
10490 | 6.51k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
10491 | 6.51k | ctxt->extSubSystem, ctxt->extSubURI); |
10492 | 10.7k | ctxt->inSubset = 0; |
10493 | | |
10494 | 10.7k | xmlCleanSpecialAttr(ctxt); |
10495 | | |
10496 | 10.7k | xmlParseMisc(ctxt); |
10497 | 10.7k | } |
10498 | | |
10499 | | /* |
10500 | | * Time to start parsing the tree itself |
10501 | | */ |
10502 | 21.9k | GROW; |
10503 | 21.9k | if (RAW != '<') { |
10504 | 4.49k | if (ctxt->wellFormed) |
10505 | 2 | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
10506 | 2 | "Start tag expected, '<' not found\n"); |
10507 | 17.4k | } else { |
10508 | 17.4k | xmlParseElement(ctxt); |
10509 | | |
10510 | | /* |
10511 | | * The Misc part at the end |
10512 | | */ |
10513 | 17.4k | xmlParseMisc(ctxt); |
10514 | | |
10515 | 17.4k | xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END); |
10516 | 17.4k | } |
10517 | | |
10518 | 21.9k | ctxt->instate = XML_PARSER_EOF; |
10519 | 21.9k | xmlFinishDocument(ctxt); |
10520 | | |
10521 | 21.9k | if (! ctxt->wellFormed) { |
10522 | 21.9k | ctxt->valid = 0; |
10523 | 21.9k | return(-1); |
10524 | 21.9k | } |
10525 | | |
10526 | 50 | return(0); |
10527 | 21.9k | } |
10528 | | |
10529 | | /** |
10530 | | * parse a general parsed entity |
10531 | | * An external general parsed entity is well-formed if it matches the |
10532 | | * production labeled extParsedEnt. |
10533 | | * |
10534 | | * @deprecated Internal function, don't use. |
10535 | | * |
10536 | | * [78] extParsedEnt ::= TextDecl? content |
10537 | | * |
10538 | | * @param ctxt an XML parser context |
10539 | | * @returns 0, -1 in case of error. the parser context is augmented |
10540 | | * as a result of the parsing. |
10541 | | */ |
10542 | | |
10543 | | int |
10544 | 0 | xmlParseExtParsedEnt(xmlParserCtxtPtr ctxt) { |
10545 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10546 | 0 | return(-1); |
10547 | | |
10548 | 0 | xmlCtxtInitializeLate(ctxt); |
10549 | |
|
10550 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
10551 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
10552 | 0 | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
10553 | 0 | } |
10554 | |
|
10555 | 0 | xmlDetectEncoding(ctxt); |
10556 | |
|
10557 | 0 | if (CUR == 0) { |
10558 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10559 | 0 | } |
10560 | | |
10561 | | /* |
10562 | | * Check for the XMLDecl in the Prolog. |
10563 | | */ |
10564 | 0 | GROW; |
10565 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10566 | | |
10567 | | /* |
10568 | | * Note that we will switch encoding on the fly. |
10569 | | */ |
10570 | 0 | xmlParseXMLDecl(ctxt); |
10571 | 0 | SKIP_BLANKS; |
10572 | 0 | } else { |
10573 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10574 | 0 | } |
10575 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10576 | 0 | ctxt->sax->startDocument(ctxt->userData); |
10577 | | |
10578 | | /* |
10579 | | * Doing validity checking on chunk doesn't make sense |
10580 | | */ |
10581 | 0 | ctxt->options &= ~XML_PARSE_DTDVALID; |
10582 | 0 | ctxt->validate = 0; |
10583 | 0 | ctxt->depth = 0; |
10584 | |
|
10585 | 0 | xmlParseContentInternal(ctxt); |
10586 | |
|
10587 | 0 | if (ctxt->input->cur < ctxt->input->end) |
10588 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
10589 | | |
10590 | | /* |
10591 | | * SAX: end of the document processing. |
10592 | | */ |
10593 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10594 | 0 | ctxt->sax->endDocument(ctxt->userData); |
10595 | |
|
10596 | 0 | if (! ctxt->wellFormed) return(-1); |
10597 | 0 | return(0); |
10598 | 0 | } |
10599 | | |
10600 | | #ifdef LIBXML_PUSH_ENABLED |
10601 | | /************************************************************************ |
10602 | | * * |
10603 | | * Progressive parsing interfaces * |
10604 | | * * |
10605 | | ************************************************************************/ |
10606 | | |
10607 | | /** |
10608 | | * Check whether the input buffer contains a character. |
10609 | | * |
10610 | | * @param ctxt an XML parser context |
10611 | | * @param c character |
10612 | | */ |
10613 | | static int |
10614 | 0 | xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) { |
10615 | 0 | const xmlChar *cur; |
10616 | |
|
10617 | 0 | if (ctxt->checkIndex == 0) { |
10618 | 0 | cur = ctxt->input->cur + 1; |
10619 | 0 | } else { |
10620 | 0 | cur = ctxt->input->cur + ctxt->checkIndex; |
10621 | 0 | } |
10622 | |
|
10623 | 0 | if (memchr(cur, c, ctxt->input->end - cur) == NULL) { |
10624 | 0 | size_t index = ctxt->input->end - ctxt->input->cur; |
10625 | |
|
10626 | 0 | if (index > LONG_MAX) { |
10627 | 0 | ctxt->checkIndex = 0; |
10628 | 0 | return(1); |
10629 | 0 | } |
10630 | 0 | ctxt->checkIndex = index; |
10631 | 0 | return(0); |
10632 | 0 | } else { |
10633 | 0 | ctxt->checkIndex = 0; |
10634 | 0 | return(1); |
10635 | 0 | } |
10636 | 0 | } |
10637 | | |
10638 | | /** |
10639 | | * Check whether the input buffer contains a string. |
10640 | | * |
10641 | | * @param ctxt an XML parser context |
10642 | | * @param startDelta delta to apply at the start |
10643 | | * @param str string |
10644 | | * @param strLen length of string |
10645 | | */ |
10646 | | static const xmlChar * |
10647 | | xmlParseLookupString(xmlParserCtxtPtr ctxt, size_t startDelta, |
10648 | 0 | const char *str, size_t strLen) { |
10649 | 0 | const xmlChar *cur, *term; |
10650 | |
|
10651 | 0 | if (ctxt->checkIndex == 0) { |
10652 | 0 | cur = ctxt->input->cur + startDelta; |
10653 | 0 | } else { |
10654 | 0 | cur = ctxt->input->cur + ctxt->checkIndex; |
10655 | 0 | } |
10656 | |
|
10657 | 0 | term = BAD_CAST strstr((const char *) cur, str); |
10658 | 0 | if (term == NULL) { |
10659 | 0 | const xmlChar *end = ctxt->input->end; |
10660 | 0 | size_t index; |
10661 | | |
10662 | | /* Rescan (strLen - 1) characters. */ |
10663 | 0 | if ((size_t) (end - cur) < strLen) |
10664 | 0 | end = cur; |
10665 | 0 | else |
10666 | 0 | end -= strLen - 1; |
10667 | 0 | index = end - ctxt->input->cur; |
10668 | 0 | if (index > LONG_MAX) { |
10669 | 0 | ctxt->checkIndex = 0; |
10670 | 0 | return(ctxt->input->end - strLen); |
10671 | 0 | } |
10672 | 0 | ctxt->checkIndex = index; |
10673 | 0 | } else { |
10674 | 0 | ctxt->checkIndex = 0; |
10675 | 0 | } |
10676 | | |
10677 | 0 | return(term); |
10678 | 0 | } |
10679 | | |
10680 | | /** |
10681 | | * Check whether the input buffer contains terminated char data. |
10682 | | * |
10683 | | * @param ctxt an XML parser context |
10684 | | */ |
10685 | | static int |
10686 | 0 | xmlParseLookupCharData(xmlParserCtxtPtr ctxt) { |
10687 | 0 | const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex; |
10688 | 0 | const xmlChar *end = ctxt->input->end; |
10689 | 0 | size_t index; |
10690 | |
|
10691 | 0 | while (cur < end) { |
10692 | 0 | if ((*cur == '<') || (*cur == '&')) { |
10693 | 0 | ctxt->checkIndex = 0; |
10694 | 0 | return(1); |
10695 | 0 | } |
10696 | 0 | cur++; |
10697 | 0 | } |
10698 | | |
10699 | 0 | index = cur - ctxt->input->cur; |
10700 | 0 | if (index > LONG_MAX) { |
10701 | 0 | ctxt->checkIndex = 0; |
10702 | 0 | return(1); |
10703 | 0 | } |
10704 | 0 | ctxt->checkIndex = index; |
10705 | 0 | return(0); |
10706 | 0 | } |
10707 | | |
10708 | | /** |
10709 | | * Check whether there's enough data in the input buffer to finish parsing |
10710 | | * a start tag. This has to take quotes into account. |
10711 | | * |
10712 | | * @param ctxt an XML parser context |
10713 | | */ |
10714 | | static int |
10715 | 0 | xmlParseLookupGt(xmlParserCtxtPtr ctxt) { |
10716 | 0 | const xmlChar *cur; |
10717 | 0 | const xmlChar *end = ctxt->input->end; |
10718 | 0 | int state = ctxt->endCheckState; |
10719 | 0 | size_t index; |
10720 | |
|
10721 | 0 | if (ctxt->checkIndex == 0) |
10722 | 0 | cur = ctxt->input->cur + 1; |
10723 | 0 | else |
10724 | 0 | cur = ctxt->input->cur + ctxt->checkIndex; |
10725 | |
|
10726 | 0 | while (cur < end) { |
10727 | 0 | if (state) { |
10728 | 0 | if (*cur == state) |
10729 | 0 | state = 0; |
10730 | 0 | } else if (*cur == '\'' || *cur == '"') { |
10731 | 0 | state = *cur; |
10732 | 0 | } else if (*cur == '>') { |
10733 | 0 | ctxt->checkIndex = 0; |
10734 | 0 | ctxt->endCheckState = 0; |
10735 | 0 | return(1); |
10736 | 0 | } |
10737 | 0 | cur++; |
10738 | 0 | } |
10739 | | |
10740 | 0 | index = cur - ctxt->input->cur; |
10741 | 0 | if (index > LONG_MAX) { |
10742 | 0 | ctxt->checkIndex = 0; |
10743 | 0 | ctxt->endCheckState = 0; |
10744 | 0 | return(1); |
10745 | 0 | } |
10746 | 0 | ctxt->checkIndex = index; |
10747 | 0 | ctxt->endCheckState = state; |
10748 | 0 | return(0); |
10749 | 0 | } |
10750 | | |
10751 | | /** |
10752 | | * Check whether there's enough data in the input buffer to finish parsing |
10753 | | * the internal subset. |
10754 | | * |
10755 | | * @param ctxt an XML parser context |
10756 | | */ |
10757 | | static int |
10758 | 0 | xmlParseLookupInternalSubset(xmlParserCtxtPtr ctxt) { |
10759 | | /* |
10760 | | * Sorry, but progressive parsing of the internal subset is not |
10761 | | * supported. We first check that the full content of the internal |
10762 | | * subset is available and parsing is launched only at that point. |
10763 | | * Internal subset ends with "']' S? '>'" in an unescaped section and |
10764 | | * not in a ']]>' sequence which are conditional sections. |
10765 | | */ |
10766 | 0 | const xmlChar *cur, *start; |
10767 | 0 | const xmlChar *end = ctxt->input->end; |
10768 | 0 | int state = ctxt->endCheckState; |
10769 | 0 | size_t index; |
10770 | |
|
10771 | 0 | if (ctxt->checkIndex == 0) { |
10772 | 0 | cur = ctxt->input->cur + 1; |
10773 | 0 | } else { |
10774 | 0 | cur = ctxt->input->cur + ctxt->checkIndex; |
10775 | 0 | } |
10776 | 0 | start = cur; |
10777 | |
|
10778 | 0 | while (cur < end) { |
10779 | 0 | if (state == '-') { |
10780 | 0 | if ((*cur == '-') && |
10781 | 0 | (cur[1] == '-') && |
10782 | 0 | (cur[2] == '>')) { |
10783 | 0 | state = 0; |
10784 | 0 | cur += 3; |
10785 | 0 | start = cur; |
10786 | 0 | continue; |
10787 | 0 | } |
10788 | 0 | } |
10789 | 0 | else if (state == ']') { |
10790 | 0 | if (*cur == '>') { |
10791 | 0 | ctxt->checkIndex = 0; |
10792 | 0 | ctxt->endCheckState = 0; |
10793 | 0 | return(1); |
10794 | 0 | } |
10795 | 0 | if (IS_BLANK_CH(*cur)) { |
10796 | 0 | state = ' '; |
10797 | 0 | } else if (*cur != ']') { |
10798 | 0 | state = 0; |
10799 | 0 | start = cur; |
10800 | 0 | continue; |
10801 | 0 | } |
10802 | 0 | } |
10803 | 0 | else if (state == ' ') { |
10804 | 0 | if (*cur == '>') { |
10805 | 0 | ctxt->checkIndex = 0; |
10806 | 0 | ctxt->endCheckState = 0; |
10807 | 0 | return(1); |
10808 | 0 | } |
10809 | 0 | if (!IS_BLANK_CH(*cur)) { |
10810 | 0 | state = 0; |
10811 | 0 | start = cur; |
10812 | 0 | continue; |
10813 | 0 | } |
10814 | 0 | } |
10815 | 0 | else if (state != 0) { |
10816 | 0 | if (*cur == state) { |
10817 | 0 | state = 0; |
10818 | 0 | start = cur + 1; |
10819 | 0 | } |
10820 | 0 | } |
10821 | 0 | else if (*cur == '<') { |
10822 | 0 | if ((cur[1] == '!') && |
10823 | 0 | (cur[2] == '-') && |
10824 | 0 | (cur[3] == '-')) { |
10825 | 0 | state = '-'; |
10826 | 0 | cur += 4; |
10827 | | /* Don't treat <!--> as comment */ |
10828 | 0 | start = cur; |
10829 | 0 | continue; |
10830 | 0 | } |
10831 | 0 | } |
10832 | 0 | else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) { |
10833 | 0 | state = *cur; |
10834 | 0 | } |
10835 | | |
10836 | 0 | cur++; |
10837 | 0 | } |
10838 | | |
10839 | | /* |
10840 | | * Rescan the three last characters to detect "<!--" and "-->" |
10841 | | * split across chunks. |
10842 | | */ |
10843 | 0 | if ((state == 0) || (state == '-')) { |
10844 | 0 | if (cur - start < 3) |
10845 | 0 | cur = start; |
10846 | 0 | else |
10847 | 0 | cur -= 3; |
10848 | 0 | } |
10849 | 0 | index = cur - ctxt->input->cur; |
10850 | 0 | if (index > LONG_MAX) { |
10851 | 0 | ctxt->checkIndex = 0; |
10852 | 0 | ctxt->endCheckState = 0; |
10853 | 0 | return(1); |
10854 | 0 | } |
10855 | 0 | ctxt->checkIndex = index; |
10856 | 0 | ctxt->endCheckState = state; |
10857 | 0 | return(0); |
10858 | 0 | } |
10859 | | |
10860 | | /** |
10861 | | * Try to progress on parsing |
10862 | | * |
10863 | | * @param ctxt an XML parser context |
10864 | | * @param terminate last chunk indicator |
10865 | | * @returns zero if no parsing was possible |
10866 | | */ |
10867 | | static int |
10868 | 0 | xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) { |
10869 | 0 | int ret = 0; |
10870 | 0 | size_t avail; |
10871 | 0 | xmlChar cur, next; |
10872 | |
|
10873 | 0 | if (ctxt->input == NULL) |
10874 | 0 | return(0); |
10875 | | |
10876 | 0 | if ((ctxt->input != NULL) && |
10877 | 0 | (ctxt->input->cur - ctxt->input->base > 4096)) { |
10878 | 0 | xmlParserShrink(ctxt); |
10879 | 0 | } |
10880 | |
|
10881 | 0 | while (ctxt->disableSAX == 0) { |
10882 | 0 | avail = ctxt->input->end - ctxt->input->cur; |
10883 | 0 | if (avail < 1) |
10884 | 0 | goto done; |
10885 | 0 | switch (ctxt->instate) { |
10886 | 0 | case XML_PARSER_EOF: |
10887 | | /* |
10888 | | * Document parsing is done ! |
10889 | | */ |
10890 | 0 | goto done; |
10891 | 0 | case XML_PARSER_START: |
10892 | | /* |
10893 | | * Very first chars read from the document flow. |
10894 | | */ |
10895 | 0 | if ((!terminate) && (avail < 4)) |
10896 | 0 | goto done; |
10897 | | |
10898 | | /* |
10899 | | * We need more bytes to detect EBCDIC code pages. |
10900 | | * See xmlDetectEBCDIC. |
10901 | | */ |
10902 | 0 | if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) && |
10903 | 0 | (!terminate) && (avail < 200)) |
10904 | 0 | goto done; |
10905 | | |
10906 | 0 | xmlDetectEncoding(ctxt); |
10907 | 0 | ctxt->instate = XML_PARSER_XML_DECL; |
10908 | 0 | break; |
10909 | | |
10910 | 0 | case XML_PARSER_XML_DECL: |
10911 | 0 | if ((!terminate) && (avail < 2)) |
10912 | 0 | goto done; |
10913 | 0 | cur = ctxt->input->cur[0]; |
10914 | 0 | next = ctxt->input->cur[1]; |
10915 | 0 | if ((cur == '<') && (next == '?')) { |
10916 | | /* PI or XML decl */ |
10917 | 0 | if ((!terminate) && |
10918 | 0 | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
10919 | 0 | goto done; |
10920 | 0 | if ((ctxt->input->cur[2] == 'x') && |
10921 | 0 | (ctxt->input->cur[3] == 'm') && |
10922 | 0 | (ctxt->input->cur[4] == 'l') && |
10923 | 0 | (IS_BLANK_CH(ctxt->input->cur[5]))) { |
10924 | 0 | ret += 5; |
10925 | 0 | xmlParseXMLDecl(ctxt); |
10926 | 0 | } else { |
10927 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10928 | 0 | if (ctxt->version == NULL) { |
10929 | 0 | xmlErrMemory(ctxt); |
10930 | 0 | break; |
10931 | 0 | } |
10932 | 0 | } |
10933 | 0 | } else { |
10934 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10935 | 0 | if (ctxt->version == NULL) { |
10936 | 0 | xmlErrMemory(ctxt); |
10937 | 0 | break; |
10938 | 0 | } |
10939 | 0 | } |
10940 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
10941 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
10942 | 0 | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
10943 | 0 | } |
10944 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
10945 | 0 | (!ctxt->disableSAX)) |
10946 | 0 | ctxt->sax->startDocument(ctxt->userData); |
10947 | 0 | ctxt->instate = XML_PARSER_MISC; |
10948 | 0 | break; |
10949 | 0 | case XML_PARSER_START_TAG: { |
10950 | 0 | const xmlChar *name; |
10951 | 0 | const xmlChar *prefix = NULL; |
10952 | 0 | const xmlChar *URI = NULL; |
10953 | 0 | int line = ctxt->input->line; |
10954 | 0 | int nbNs = 0; |
10955 | |
|
10956 | 0 | if ((!terminate) && (avail < 2)) |
10957 | 0 | goto done; |
10958 | 0 | cur = ctxt->input->cur[0]; |
10959 | 0 | if (cur != '<') { |
10960 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
10961 | 0 | "Start tag expected, '<' not found"); |
10962 | 0 | ctxt->instate = XML_PARSER_EOF; |
10963 | 0 | xmlFinishDocument(ctxt); |
10964 | 0 | goto done; |
10965 | 0 | } |
10966 | 0 | if ((!terminate) && (!xmlParseLookupGt(ctxt))) |
10967 | 0 | goto done; |
10968 | 0 | if (ctxt->spaceNr == 0) |
10969 | 0 | spacePush(ctxt, -1); |
10970 | 0 | else if (*ctxt->space == -2) |
10971 | 0 | spacePush(ctxt, -1); |
10972 | 0 | else |
10973 | 0 | spacePush(ctxt, *ctxt->space); |
10974 | 0 | #ifdef LIBXML_SAX1_ENABLED |
10975 | 0 | if (ctxt->sax2) |
10976 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
10977 | 0 | name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs); |
10978 | 0 | #ifdef LIBXML_SAX1_ENABLED |
10979 | 0 | else |
10980 | 0 | name = xmlParseStartTag(ctxt); |
10981 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
10982 | 0 | if (name == NULL) { |
10983 | 0 | spacePop(ctxt); |
10984 | 0 | ctxt->instate = XML_PARSER_EOF; |
10985 | 0 | xmlFinishDocument(ctxt); |
10986 | 0 | goto done; |
10987 | 0 | } |
10988 | 0 | #ifdef LIBXML_VALID_ENABLED |
10989 | | /* |
10990 | | * [ VC: Root Element Type ] |
10991 | | * The Name in the document type declaration must match |
10992 | | * the element type of the root element. |
10993 | | */ |
10994 | 0 | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
10995 | 0 | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
10996 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
10997 | 0 | #endif /* LIBXML_VALID_ENABLED */ |
10998 | | |
10999 | | /* |
11000 | | * Check for an Empty Element. |
11001 | | */ |
11002 | 0 | if ((RAW == '/') && (NXT(1) == '>')) { |
11003 | 0 | SKIP(2); |
11004 | |
|
11005 | 0 | if (ctxt->sax2) { |
11006 | 0 | if ((ctxt->sax != NULL) && |
11007 | 0 | (ctxt->sax->endElementNs != NULL) && |
11008 | 0 | (!ctxt->disableSAX)) |
11009 | 0 | ctxt->sax->endElementNs(ctxt->userData, name, |
11010 | 0 | prefix, URI); |
11011 | 0 | if (nbNs > 0) |
11012 | 0 | xmlParserNsPop(ctxt, nbNs); |
11013 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11014 | 0 | } else { |
11015 | 0 | if ((ctxt->sax != NULL) && |
11016 | 0 | (ctxt->sax->endElement != NULL) && |
11017 | 0 | (!ctxt->disableSAX)) |
11018 | 0 | ctxt->sax->endElement(ctxt->userData, name); |
11019 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11020 | 0 | } |
11021 | 0 | spacePop(ctxt); |
11022 | 0 | } else if (RAW == '>') { |
11023 | 0 | NEXT; |
11024 | 0 | nameNsPush(ctxt, name, prefix, URI, line, nbNs); |
11025 | 0 | } else { |
11026 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED, |
11027 | 0 | "Couldn't find end of Start Tag %s\n", |
11028 | 0 | name); |
11029 | 0 | nodePop(ctxt); |
11030 | 0 | spacePop(ctxt); |
11031 | 0 | if (nbNs > 0) |
11032 | 0 | xmlParserNsPop(ctxt, nbNs); |
11033 | 0 | } |
11034 | |
|
11035 | 0 | if (ctxt->nameNr == 0) |
11036 | 0 | ctxt->instate = XML_PARSER_EPILOG; |
11037 | 0 | else |
11038 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11039 | 0 | break; |
11040 | 0 | } |
11041 | 0 | case XML_PARSER_CONTENT: { |
11042 | 0 | cur = ctxt->input->cur[0]; |
11043 | |
|
11044 | 0 | if (cur == '<') { |
11045 | 0 | if ((!terminate) && (avail < 2)) |
11046 | 0 | goto done; |
11047 | 0 | next = ctxt->input->cur[1]; |
11048 | |
|
11049 | 0 | if (next == '/') { |
11050 | 0 | ctxt->instate = XML_PARSER_END_TAG; |
11051 | 0 | break; |
11052 | 0 | } else if (next == '?') { |
11053 | 0 | if ((!terminate) && |
11054 | 0 | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11055 | 0 | goto done; |
11056 | 0 | xmlParsePI(ctxt); |
11057 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11058 | 0 | break; |
11059 | 0 | } else if (next == '!') { |
11060 | 0 | if ((!terminate) && (avail < 3)) |
11061 | 0 | goto done; |
11062 | 0 | next = ctxt->input->cur[2]; |
11063 | |
|
11064 | 0 | if (next == '-') { |
11065 | 0 | if ((!terminate) && (avail < 4)) |
11066 | 0 | goto done; |
11067 | 0 | if (ctxt->input->cur[3] == '-') { |
11068 | 0 | if ((!terminate) && |
11069 | 0 | (!xmlParseLookupString(ctxt, 4, "-->", 3))) |
11070 | 0 | goto done; |
11071 | 0 | xmlParseComment(ctxt); |
11072 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11073 | 0 | break; |
11074 | 0 | } |
11075 | 0 | } else if (next == '[') { |
11076 | 0 | if ((!terminate) && (avail < 9)) |
11077 | 0 | goto done; |
11078 | 0 | if ((ctxt->input->cur[2] == '[') && |
11079 | 0 | (ctxt->input->cur[3] == 'C') && |
11080 | 0 | (ctxt->input->cur[4] == 'D') && |
11081 | 0 | (ctxt->input->cur[5] == 'A') && |
11082 | 0 | (ctxt->input->cur[6] == 'T') && |
11083 | 0 | (ctxt->input->cur[7] == 'A') && |
11084 | 0 | (ctxt->input->cur[8] == '[')) { |
11085 | 0 | if ((!terminate) && |
11086 | 0 | (!xmlParseLookupString(ctxt, 9, "]]>", 3))) |
11087 | 0 | goto done; |
11088 | 0 | ctxt->instate = XML_PARSER_CDATA_SECTION; |
11089 | 0 | xmlParseCDSect(ctxt); |
11090 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11091 | 0 | break; |
11092 | 0 | } |
11093 | 0 | } |
11094 | 0 | } |
11095 | 0 | } else if (cur == '&') { |
11096 | 0 | if ((!terminate) && (!xmlParseLookupChar(ctxt, ';'))) |
11097 | 0 | goto done; |
11098 | 0 | xmlParseReference(ctxt); |
11099 | 0 | break; |
11100 | 0 | } else { |
11101 | | /* TODO Avoid the extra copy, handle directly !!! */ |
11102 | | /* |
11103 | | * Goal of the following test is: |
11104 | | * - minimize calls to the SAX 'character' callback |
11105 | | * when they are mergeable |
11106 | | * - handle an problem for isBlank when we only parse |
11107 | | * a sequence of blank chars and the next one is |
11108 | | * not available to check against '<' presence. |
11109 | | * - tries to homogenize the differences in SAX |
11110 | | * callbacks between the push and pull versions |
11111 | | * of the parser. |
11112 | | */ |
11113 | 0 | if (avail < XML_PARSER_BIG_BUFFER_SIZE) { |
11114 | 0 | if ((!terminate) && (!xmlParseLookupCharData(ctxt))) |
11115 | 0 | goto done; |
11116 | 0 | } |
11117 | 0 | ctxt->checkIndex = 0; |
11118 | 0 | xmlParseCharDataInternal(ctxt, !terminate); |
11119 | 0 | break; |
11120 | 0 | } |
11121 | | |
11122 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
11123 | 0 | break; |
11124 | 0 | } |
11125 | 0 | case XML_PARSER_END_TAG: |
11126 | 0 | if ((!terminate) && (!xmlParseLookupChar(ctxt, '>'))) |
11127 | 0 | goto done; |
11128 | 0 | if (ctxt->sax2) { |
11129 | 0 | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
11130 | 0 | nameNsPop(ctxt); |
11131 | 0 | } |
11132 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11133 | 0 | else |
11134 | 0 | xmlParseEndTag1(ctxt, 0); |
11135 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11136 | 0 | if (ctxt->nameNr == 0) { |
11137 | 0 | ctxt->instate = XML_PARSER_EPILOG; |
11138 | 0 | } else { |
11139 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11140 | 0 | } |
11141 | 0 | break; |
11142 | 0 | case XML_PARSER_MISC: |
11143 | 0 | case XML_PARSER_PROLOG: |
11144 | 0 | case XML_PARSER_EPILOG: |
11145 | 0 | SKIP_BLANKS; |
11146 | 0 | avail = ctxt->input->end - ctxt->input->cur; |
11147 | 0 | if (avail < 1) |
11148 | 0 | goto done; |
11149 | 0 | if (ctxt->input->cur[0] == '<') { |
11150 | 0 | if ((!terminate) && (avail < 2)) |
11151 | 0 | goto done; |
11152 | 0 | next = ctxt->input->cur[1]; |
11153 | 0 | if (next == '?') { |
11154 | 0 | if ((!terminate) && |
11155 | 0 | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11156 | 0 | goto done; |
11157 | 0 | xmlParsePI(ctxt); |
11158 | 0 | break; |
11159 | 0 | } else if (next == '!') { |
11160 | 0 | if ((!terminate) && (avail < 3)) |
11161 | 0 | goto done; |
11162 | | |
11163 | 0 | if (ctxt->input->cur[2] == '-') { |
11164 | 0 | if ((!terminate) && (avail < 4)) |
11165 | 0 | goto done; |
11166 | 0 | if (ctxt->input->cur[3] == '-') { |
11167 | 0 | if ((!terminate) && |
11168 | 0 | (!xmlParseLookupString(ctxt, 4, "-->", 3))) |
11169 | 0 | goto done; |
11170 | 0 | xmlParseComment(ctxt); |
11171 | 0 | break; |
11172 | 0 | } |
11173 | 0 | } else if (ctxt->instate == XML_PARSER_MISC) { |
11174 | 0 | if ((!terminate) && (avail < 9)) |
11175 | 0 | goto done; |
11176 | 0 | if ((ctxt->input->cur[2] == 'D') && |
11177 | 0 | (ctxt->input->cur[3] == 'O') && |
11178 | 0 | (ctxt->input->cur[4] == 'C') && |
11179 | 0 | (ctxt->input->cur[5] == 'T') && |
11180 | 0 | (ctxt->input->cur[6] == 'Y') && |
11181 | 0 | (ctxt->input->cur[7] == 'P') && |
11182 | 0 | (ctxt->input->cur[8] == 'E')) { |
11183 | 0 | if ((!terminate) && (!xmlParseLookupGt(ctxt))) |
11184 | 0 | goto done; |
11185 | 0 | ctxt->inSubset = 1; |
11186 | 0 | xmlParseDocTypeDecl(ctxt); |
11187 | 0 | if (RAW == '[') { |
11188 | 0 | ctxt->instate = XML_PARSER_DTD; |
11189 | 0 | } else { |
11190 | 0 | if (RAW == '>') |
11191 | 0 | NEXT; |
11192 | | /* |
11193 | | * Create and update the external subset. |
11194 | | */ |
11195 | 0 | ctxt->inSubset = 2; |
11196 | 0 | if ((ctxt->sax != NULL) && |
11197 | 0 | (!ctxt->disableSAX) && |
11198 | 0 | (ctxt->sax->externalSubset != NULL)) |
11199 | 0 | ctxt->sax->externalSubset( |
11200 | 0 | ctxt->userData, |
11201 | 0 | ctxt->intSubName, |
11202 | 0 | ctxt->extSubSystem, |
11203 | 0 | ctxt->extSubURI); |
11204 | 0 | ctxt->inSubset = 0; |
11205 | 0 | xmlCleanSpecialAttr(ctxt); |
11206 | 0 | ctxt->instate = XML_PARSER_PROLOG; |
11207 | 0 | } |
11208 | 0 | break; |
11209 | 0 | } |
11210 | 0 | } |
11211 | 0 | } |
11212 | 0 | } |
11213 | | |
11214 | 0 | if (ctxt->instate == XML_PARSER_EPILOG) { |
11215 | 0 | if (ctxt->errNo == XML_ERR_OK) |
11216 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
11217 | 0 | ctxt->instate = XML_PARSER_EOF; |
11218 | 0 | xmlFinishDocument(ctxt); |
11219 | 0 | } else { |
11220 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
11221 | 0 | } |
11222 | 0 | break; |
11223 | 0 | case XML_PARSER_DTD: { |
11224 | 0 | if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt))) |
11225 | 0 | goto done; |
11226 | 0 | xmlParseInternalSubset(ctxt); |
11227 | 0 | ctxt->inSubset = 2; |
11228 | 0 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
11229 | 0 | (ctxt->sax->externalSubset != NULL)) |
11230 | 0 | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
11231 | 0 | ctxt->extSubSystem, ctxt->extSubURI); |
11232 | 0 | ctxt->inSubset = 0; |
11233 | 0 | xmlCleanSpecialAttr(ctxt); |
11234 | 0 | ctxt->instate = XML_PARSER_PROLOG; |
11235 | 0 | break; |
11236 | 0 | } |
11237 | 0 | default: |
11238 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
11239 | 0 | "PP: internal error\n"); |
11240 | 0 | ctxt->instate = XML_PARSER_EOF; |
11241 | 0 | break; |
11242 | 0 | } |
11243 | 0 | } |
11244 | 0 | done: |
11245 | 0 | return(ret); |
11246 | 0 | } |
11247 | | |
11248 | | /** |
11249 | | * Parse a chunk of memory in push parser mode. |
11250 | | * |
11251 | | * Assumes that the parser context was initialized with |
11252 | | * xmlCreatePushParserCtxt(). |
11253 | | * |
11254 | | * The last chunk, which will often be empty, must be marked with |
11255 | | * the `terminate` flag. With the default SAX callbacks, the resulting |
11256 | | * document will be available in ctxt->myDoc. This pointer will not |
11257 | | * be freed when calling xmlFreeParserCtxt() and must be freed by the |
11258 | | * caller. If the document isn't well-formed, it will still be returned |
11259 | | * in ctxt->myDoc. |
11260 | | * |
11261 | | * As an exception, xmlCtxtResetPush() will free the document in |
11262 | | * ctxt->myDoc. So ctxt->myDoc should be set to NULL after extracting |
11263 | | * the document. |
11264 | | * |
11265 | | * @param ctxt an XML parser context |
11266 | | * @param chunk chunk of memory |
11267 | | * @param size size of chunk in bytes |
11268 | | * @param terminate last chunk indicator |
11269 | | * @returns an xmlParserErrors code (0 on success). |
11270 | | */ |
11271 | | int |
11272 | | xmlParseChunk(xmlParserCtxtPtr ctxt, const char *chunk, int size, |
11273 | 0 | int terminate) { |
11274 | 0 | size_t curBase; |
11275 | 0 | size_t maxLength; |
11276 | 0 | size_t pos; |
11277 | 0 | int end_in_lf = 0; |
11278 | 0 | int res; |
11279 | |
|
11280 | 0 | if ((ctxt == NULL) || (size < 0)) |
11281 | 0 | return(XML_ERR_ARGUMENT); |
11282 | 0 | if ((chunk == NULL) && (size > 0)) |
11283 | 0 | return(XML_ERR_ARGUMENT); |
11284 | 0 | if ((ctxt->input == NULL) || (ctxt->input->buf == NULL)) |
11285 | 0 | return(XML_ERR_ARGUMENT); |
11286 | 0 | if (ctxt->disableSAX != 0) |
11287 | 0 | return(ctxt->errNo); |
11288 | | |
11289 | 0 | ctxt->input->flags |= XML_INPUT_PROGRESSIVE; |
11290 | 0 | if (ctxt->instate == XML_PARSER_START) |
11291 | 0 | xmlCtxtInitializeLate(ctxt); |
11292 | 0 | if ((size > 0) && (chunk != NULL) && (!terminate) && |
11293 | 0 | (chunk[size - 1] == '\r')) { |
11294 | 0 | end_in_lf = 1; |
11295 | 0 | size--; |
11296 | 0 | } |
11297 | | |
11298 | | /* |
11299 | | * Also push an empty chunk to make sure that the raw buffer |
11300 | | * will be flushed if there is an encoder. |
11301 | | */ |
11302 | 0 | pos = ctxt->input->cur - ctxt->input->base; |
11303 | 0 | res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
11304 | 0 | xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos); |
11305 | 0 | if (res < 0) { |
11306 | 0 | xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL); |
11307 | 0 | xmlHaltParser(ctxt); |
11308 | 0 | return(ctxt->errNo); |
11309 | 0 | } |
11310 | | |
11311 | 0 | xmlParseTryOrFinish(ctxt, terminate); |
11312 | |
|
11313 | 0 | curBase = ctxt->input->cur - ctxt->input->base; |
11314 | 0 | maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
11315 | 0 | XML_MAX_HUGE_LENGTH : |
11316 | 0 | XML_MAX_LOOKUP_LIMIT; |
11317 | 0 | if (curBase > maxLength) { |
11318 | 0 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, |
11319 | 0 | "Buffer size limit exceeded, try XML_PARSE_HUGE\n"); |
11320 | 0 | xmlHaltParser(ctxt); |
11321 | 0 | } |
11322 | |
|
11323 | 0 | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0)) |
11324 | 0 | return(ctxt->errNo); |
11325 | | |
11326 | 0 | if (end_in_lf == 1) { |
11327 | 0 | pos = ctxt->input->cur - ctxt->input->base; |
11328 | 0 | res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r"); |
11329 | 0 | xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos); |
11330 | 0 | if (res < 0) { |
11331 | 0 | xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL); |
11332 | 0 | xmlHaltParser(ctxt); |
11333 | 0 | return(ctxt->errNo); |
11334 | 0 | } |
11335 | 0 | } |
11336 | 0 | if (terminate) { |
11337 | | /* |
11338 | | * Check for termination |
11339 | | */ |
11340 | 0 | if ((ctxt->instate != XML_PARSER_EOF) && |
11341 | 0 | (ctxt->instate != XML_PARSER_EPILOG)) { |
11342 | 0 | if (ctxt->nameNr > 0) { |
11343 | 0 | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
11344 | 0 | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
11345 | 0 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
11346 | 0 | "Premature end of data in tag %s line %d\n", |
11347 | 0 | name, line, NULL); |
11348 | 0 | } else if (ctxt->instate == XML_PARSER_START) { |
11349 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11350 | 0 | } else { |
11351 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
11352 | 0 | "Start tag expected, '<' not found\n"); |
11353 | 0 | } |
11354 | 0 | } else { |
11355 | 0 | xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END); |
11356 | 0 | } |
11357 | 0 | if (ctxt->instate != XML_PARSER_EOF) { |
11358 | 0 | ctxt->instate = XML_PARSER_EOF; |
11359 | 0 | xmlFinishDocument(ctxt); |
11360 | 0 | } |
11361 | 0 | } |
11362 | 0 | if (ctxt->wellFormed == 0) |
11363 | 0 | return((xmlParserErrors) ctxt->errNo); |
11364 | 0 | else |
11365 | 0 | return(0); |
11366 | 0 | } |
11367 | | |
11368 | | /************************************************************************ |
11369 | | * * |
11370 | | * I/O front end functions to the parser * |
11371 | | * * |
11372 | | ************************************************************************/ |
11373 | | |
11374 | | /** |
11375 | | * Create a parser context for using the XML parser in push mode. |
11376 | | * See xmlParseChunk(). |
11377 | | * |
11378 | | * Passing an initial chunk is useless and deprecated. |
11379 | | * |
11380 | | * The push parser doesn't support recovery mode or the |
11381 | | * XML_PARSE_NOBLANKS option. |
11382 | | * |
11383 | | * `filename` is used as base URI to fetch external entities and for |
11384 | | * error reports. |
11385 | | * |
11386 | | * @param sax a SAX handler (optional) |
11387 | | * @param user_data user data for SAX callbacks (optional) |
11388 | | * @param chunk initial chunk (optional, deprecated) |
11389 | | * @param size size of initial chunk in bytes |
11390 | | * @param filename file name or URI (optional) |
11391 | | * @returns the new parser context or NULL if a memory allocation |
11392 | | * failed. |
11393 | | */ |
11394 | | |
11395 | | xmlParserCtxtPtr |
11396 | | xmlCreatePushParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
11397 | 0 | const char *chunk, int size, const char *filename) { |
11398 | 0 | xmlParserCtxtPtr ctxt; |
11399 | 0 | xmlParserInputPtr input; |
11400 | |
|
11401 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
11402 | 0 | if (ctxt == NULL) |
11403 | 0 | return(NULL); |
11404 | | |
11405 | 0 | ctxt->options &= ~XML_PARSE_NODICT; |
11406 | 0 | ctxt->dictNames = 1; |
11407 | |
|
11408 | 0 | input = xmlNewPushInput(filename, chunk, size); |
11409 | 0 | if (input == NULL) { |
11410 | 0 | xmlFreeParserCtxt(ctxt); |
11411 | 0 | return(NULL); |
11412 | 0 | } |
11413 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
11414 | 0 | xmlFreeInputStream(input); |
11415 | 0 | xmlFreeParserCtxt(ctxt); |
11416 | 0 | return(NULL); |
11417 | 0 | } |
11418 | | |
11419 | 0 | return(ctxt); |
11420 | 0 | } |
11421 | | #endif /* LIBXML_PUSH_ENABLED */ |
11422 | | |
11423 | | /** |
11424 | | * Blocks further parser processing |
11425 | | * |
11426 | | * @param ctxt an XML parser context |
11427 | | */ |
11428 | | void |
11429 | 0 | xmlStopParser(xmlParserCtxtPtr ctxt) { |
11430 | 0 | if (ctxt == NULL) |
11431 | 0 | return; |
11432 | 0 | xmlHaltParser(ctxt); |
11433 | 0 | if (ctxt->errNo != XML_ERR_NO_MEMORY) |
11434 | 0 | ctxt->errNo = XML_ERR_USER_STOP; |
11435 | 0 | } |
11436 | | |
11437 | | /** |
11438 | | * Create a parser context for using the XML parser with an existing |
11439 | | * I/O stream |
11440 | | * |
11441 | | * @param sax a SAX handler (optional) |
11442 | | * @param user_data user data for SAX callbacks (optional) |
11443 | | * @param ioread an I/O read function |
11444 | | * @param ioclose an I/O close function (optional) |
11445 | | * @param ioctx an I/O handler |
11446 | | * @param enc the charset encoding if known (deprecated) |
11447 | | * @returns the new parser context or NULL |
11448 | | */ |
11449 | | xmlParserCtxtPtr |
11450 | | xmlCreateIOParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
11451 | | xmlInputReadCallback ioread, |
11452 | | xmlInputCloseCallback ioclose, |
11453 | 0 | void *ioctx, xmlCharEncoding enc) { |
11454 | 0 | xmlParserCtxtPtr ctxt; |
11455 | 0 | xmlParserInputPtr input; |
11456 | 0 | const char *encoding; |
11457 | |
|
11458 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
11459 | 0 | if (ctxt == NULL) |
11460 | 0 | return(NULL); |
11461 | | |
11462 | 0 | encoding = xmlGetCharEncodingName(enc); |
11463 | 0 | input = xmlCtxtNewInputFromIO(ctxt, NULL, ioread, ioclose, ioctx, |
11464 | 0 | encoding, 0); |
11465 | 0 | if (input == NULL) { |
11466 | 0 | xmlFreeParserCtxt(ctxt); |
11467 | 0 | return (NULL); |
11468 | 0 | } |
11469 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
11470 | 0 | xmlFreeInputStream(input); |
11471 | 0 | xmlFreeParserCtxt(ctxt); |
11472 | 0 | return(NULL); |
11473 | 0 | } |
11474 | | |
11475 | 0 | return(ctxt); |
11476 | 0 | } |
11477 | | |
11478 | | #ifdef LIBXML_VALID_ENABLED |
11479 | | /************************************************************************ |
11480 | | * * |
11481 | | * Front ends when parsing a DTD * |
11482 | | * * |
11483 | | ************************************************************************/ |
11484 | | |
11485 | | /** |
11486 | | * Parse a DTD. |
11487 | | * |
11488 | | * Option XML_PARSE_DTDLOAD should be enabled in the parser context |
11489 | | * to make external entities work. |
11490 | | * |
11491 | | * @since 2.14.0 |
11492 | | * |
11493 | | * @param ctxt a parser context |
11494 | | * @param input a parser input |
11495 | | * @param publicId public ID of the DTD (optional) |
11496 | | * @param systemId system ID of the DTD (optional) |
11497 | | * @returns the resulting xmlDtdPtr or NULL in case of error. |
11498 | | * `input` will be freed by the function in any case. |
11499 | | */ |
11500 | | xmlDtdPtr |
11501 | | xmlCtxtParseDtd(xmlParserCtxtPtr ctxt, xmlParserInputPtr input, |
11502 | 0 | const xmlChar *publicId, const xmlChar *systemId) { |
11503 | 0 | xmlDtdPtr ret = NULL; |
11504 | |
|
11505 | 0 | if ((ctxt == NULL) || (input == NULL)) { |
11506 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
11507 | 0 | xmlFreeInputStream(input); |
11508 | 0 | return(NULL); |
11509 | 0 | } |
11510 | | |
11511 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
11512 | 0 | xmlFreeInputStream(input); |
11513 | 0 | return(NULL); |
11514 | 0 | } |
11515 | | |
11516 | 0 | if (publicId == NULL) |
11517 | 0 | publicId = BAD_CAST "none"; |
11518 | 0 | if (systemId == NULL) |
11519 | 0 | systemId = BAD_CAST "none"; |
11520 | |
|
11521 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
11522 | 0 | if (ctxt->myDoc == NULL) { |
11523 | 0 | xmlErrMemory(ctxt); |
11524 | 0 | goto error; |
11525 | 0 | } |
11526 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
11527 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
11528 | 0 | publicId, systemId); |
11529 | 0 | if (ctxt->myDoc->extSubset == NULL) { |
11530 | 0 | xmlErrMemory(ctxt); |
11531 | 0 | xmlFreeDoc(ctxt->myDoc); |
11532 | 0 | goto error; |
11533 | 0 | } |
11534 | | |
11535 | 0 | xmlParseExternalSubset(ctxt, publicId, systemId); |
11536 | |
|
11537 | 0 | if (ctxt->wellFormed) { |
11538 | 0 | ret = ctxt->myDoc->extSubset; |
11539 | 0 | ctxt->myDoc->extSubset = NULL; |
11540 | 0 | if (ret != NULL) { |
11541 | 0 | xmlNodePtr tmp; |
11542 | |
|
11543 | 0 | ret->doc = NULL; |
11544 | 0 | tmp = ret->children; |
11545 | 0 | while (tmp != NULL) { |
11546 | 0 | tmp->doc = NULL; |
11547 | 0 | tmp = tmp->next; |
11548 | 0 | } |
11549 | 0 | } |
11550 | 0 | } else { |
11551 | 0 | ret = NULL; |
11552 | 0 | } |
11553 | 0 | xmlFreeDoc(ctxt->myDoc); |
11554 | 0 | ctxt->myDoc = NULL; |
11555 | |
|
11556 | 0 | error: |
11557 | 0 | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
11558 | |
|
11559 | 0 | return(ret); |
11560 | 0 | } |
11561 | | |
11562 | | /** |
11563 | | * Load and parse a DTD |
11564 | | * |
11565 | | * @deprecated Use xmlCtxtParseDtd(). |
11566 | | * |
11567 | | * @param sax the SAX handler block or NULL |
11568 | | * @param input an Input Buffer |
11569 | | * @param enc the charset encoding if known |
11570 | | * @returns the resulting xmlDtdPtr or NULL in case of error. |
11571 | | * `input` will be freed by the function in any case. |
11572 | | */ |
11573 | | |
11574 | | xmlDtdPtr |
11575 | | xmlIOParseDTD(xmlSAXHandlerPtr sax, xmlParserInputBufferPtr input, |
11576 | 0 | xmlCharEncoding enc) { |
11577 | 0 | xmlDtdPtr ret = NULL; |
11578 | 0 | xmlParserCtxtPtr ctxt; |
11579 | 0 | xmlParserInputPtr pinput = NULL; |
11580 | |
|
11581 | 0 | if (input == NULL) |
11582 | 0 | return(NULL); |
11583 | | |
11584 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
11585 | 0 | if (ctxt == NULL) { |
11586 | 0 | xmlFreeParserInputBuffer(input); |
11587 | 0 | return(NULL); |
11588 | 0 | } |
11589 | 0 | xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD); |
11590 | | |
11591 | | /* |
11592 | | * generate a parser input from the I/O handler |
11593 | | */ |
11594 | |
|
11595 | 0 | pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
11596 | 0 | if (pinput == NULL) { |
11597 | 0 | xmlFreeParserInputBuffer(input); |
11598 | 0 | xmlFreeParserCtxt(ctxt); |
11599 | 0 | return(NULL); |
11600 | 0 | } |
11601 | | |
11602 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
11603 | 0 | xmlSwitchEncoding(ctxt, enc); |
11604 | 0 | } |
11605 | |
|
11606 | 0 | ret = xmlCtxtParseDtd(ctxt, pinput, NULL, NULL); |
11607 | |
|
11608 | 0 | xmlFreeParserCtxt(ctxt); |
11609 | 0 | return(ret); |
11610 | 0 | } |
11611 | | |
11612 | | /** |
11613 | | * Load and parse an external subset. |
11614 | | * |
11615 | | * @deprecated Use xmlCtxtParseDtd(). |
11616 | | * |
11617 | | * @param sax the SAX handler block |
11618 | | * @param ExternalID a NAME* containing the External ID of the DTD |
11619 | | * @param SystemID a NAME* containing the URL to the DTD |
11620 | | * @returns the resulting xmlDtdPtr or NULL in case of error. |
11621 | | */ |
11622 | | |
11623 | | xmlDtdPtr |
11624 | | xmlSAXParseDTD(xmlSAXHandlerPtr sax, const xmlChar *ExternalID, |
11625 | 0 | const xmlChar *SystemID) { |
11626 | 0 | xmlDtdPtr ret = NULL; |
11627 | 0 | xmlParserCtxtPtr ctxt; |
11628 | 0 | xmlParserInputPtr input = NULL; |
11629 | 0 | xmlChar* systemIdCanonic; |
11630 | |
|
11631 | 0 | if ((ExternalID == NULL) && (SystemID == NULL)) return(NULL); |
11632 | | |
11633 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
11634 | 0 | if (ctxt == NULL) { |
11635 | 0 | return(NULL); |
11636 | 0 | } |
11637 | 0 | xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD); |
11638 | | |
11639 | | /* |
11640 | | * Canonicalise the system ID |
11641 | | */ |
11642 | 0 | systemIdCanonic = xmlCanonicPath(SystemID); |
11643 | 0 | if ((SystemID != NULL) && (systemIdCanonic == NULL)) { |
11644 | 0 | xmlFreeParserCtxt(ctxt); |
11645 | 0 | return(NULL); |
11646 | 0 | } |
11647 | | |
11648 | | /* |
11649 | | * Ask the Entity resolver to load the damn thing |
11650 | | */ |
11651 | | |
11652 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) |
11653 | 0 | input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID, |
11654 | 0 | systemIdCanonic); |
11655 | 0 | if (input == NULL) { |
11656 | 0 | xmlFreeParserCtxt(ctxt); |
11657 | 0 | if (systemIdCanonic != NULL) |
11658 | 0 | xmlFree(systemIdCanonic); |
11659 | 0 | return(NULL); |
11660 | 0 | } |
11661 | | |
11662 | 0 | if (input->filename == NULL) |
11663 | 0 | input->filename = (char *) systemIdCanonic; |
11664 | 0 | else |
11665 | 0 | xmlFree(systemIdCanonic); |
11666 | |
|
11667 | 0 | ret = xmlCtxtParseDtd(ctxt, input, ExternalID, SystemID); |
11668 | |
|
11669 | 0 | xmlFreeParserCtxt(ctxt); |
11670 | 0 | return(ret); |
11671 | 0 | } |
11672 | | |
11673 | | |
11674 | | /** |
11675 | | * Load and parse an external subset. |
11676 | | * |
11677 | | * @param ExternalID a NAME* containing the External ID of the DTD |
11678 | | * @param SystemID a NAME* containing the URL to the DTD |
11679 | | * @returns the resulting xmlDtdPtr or NULL in case of error. |
11680 | | */ |
11681 | | |
11682 | | xmlDtdPtr |
11683 | 0 | xmlParseDTD(const xmlChar *ExternalID, const xmlChar *SystemID) { |
11684 | 0 | return(xmlSAXParseDTD(NULL, ExternalID, SystemID)); |
11685 | 0 | } |
11686 | | #endif /* LIBXML_VALID_ENABLED */ |
11687 | | |
11688 | | /************************************************************************ |
11689 | | * * |
11690 | | * Front ends when parsing an Entity * |
11691 | | * * |
11692 | | ************************************************************************/ |
11693 | | |
11694 | | static xmlNodePtr |
11695 | | xmlCtxtParseContentInternal(xmlParserCtxtPtr ctxt, xmlParserInputPtr input, |
11696 | 1.03k | int hasTextDecl, int buildTree) { |
11697 | 1.03k | xmlNodePtr root = NULL; |
11698 | 1.03k | xmlNodePtr list = NULL; |
11699 | 1.03k | xmlChar *rootName = BAD_CAST "#root"; |
11700 | 1.03k | int result; |
11701 | | |
11702 | 1.03k | if (buildTree) { |
11703 | 1.03k | root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL); |
11704 | 1.03k | if (root == NULL) { |
11705 | 0 | xmlErrMemory(ctxt); |
11706 | 0 | goto error; |
11707 | 0 | } |
11708 | 1.03k | } |
11709 | | |
11710 | 1.03k | if (xmlCtxtPushInput(ctxt, input) < 0) |
11711 | 0 | goto error; |
11712 | | |
11713 | 1.03k | nameNsPush(ctxt, rootName, NULL, NULL, 0, 0); |
11714 | 1.03k | spacePush(ctxt, -1); |
11715 | | |
11716 | 1.03k | if (buildTree) |
11717 | 1.03k | nodePush(ctxt, root); |
11718 | | |
11719 | 1.03k | if (hasTextDecl) { |
11720 | 0 | xmlDetectEncoding(ctxt); |
11721 | | |
11722 | | /* |
11723 | | * Parse a possible text declaration first |
11724 | | */ |
11725 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
11726 | 0 | (IS_BLANK_CH(NXT(5)))) { |
11727 | 0 | xmlParseTextDecl(ctxt); |
11728 | | /* |
11729 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
11730 | | */ |
11731 | 0 | if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) && |
11732 | 0 | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
11733 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
11734 | 0 | "Version mismatch between document and " |
11735 | 0 | "entity\n"); |
11736 | 0 | } |
11737 | 0 | } |
11738 | 0 | } |
11739 | | |
11740 | 1.03k | xmlParseContentInternal(ctxt); |
11741 | | |
11742 | 1.03k | if (ctxt->input->cur < ctxt->input->end) |
11743 | 48 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
11744 | | |
11745 | 1.03k | if ((ctxt->wellFormed) || |
11746 | 1.03k | ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) { |
11747 | 858 | if (root != NULL) { |
11748 | 858 | xmlNodePtr cur; |
11749 | | |
11750 | | /* |
11751 | | * Unlink newly created node list. |
11752 | | */ |
11753 | 858 | list = root->children; |
11754 | 858 | root->children = NULL; |
11755 | 858 | root->last = NULL; |
11756 | 3.29k | for (cur = list; cur != NULL; cur = cur->next) |
11757 | 2.43k | cur->parent = NULL; |
11758 | 858 | } |
11759 | 858 | } |
11760 | | |
11761 | | /* |
11762 | | * Read the rest of the stream in case of errors. We want |
11763 | | * to account for the whole entity size. |
11764 | | */ |
11765 | 1.03k | do { |
11766 | 1.03k | ctxt->input->cur = ctxt->input->end; |
11767 | 1.03k | xmlParserShrink(ctxt); |
11768 | 1.03k | result = xmlParserGrow(ctxt); |
11769 | 1.03k | } while (result > 0); |
11770 | | |
11771 | 1.03k | if (buildTree) |
11772 | 1.03k | nodePop(ctxt); |
11773 | | |
11774 | 1.03k | namePop(ctxt); |
11775 | 1.03k | spacePop(ctxt); |
11776 | | |
11777 | 1.03k | xmlCtxtPopInput(ctxt); |
11778 | | |
11779 | 1.03k | error: |
11780 | 1.03k | xmlFreeNode(root); |
11781 | | |
11782 | 1.03k | return(list); |
11783 | 1.03k | } |
11784 | | |
11785 | | static void |
11786 | 1.04k | xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) { |
11787 | 1.04k | xmlParserInputPtr input; |
11788 | 1.04k | xmlNodePtr list; |
11789 | 1.04k | unsigned long consumed; |
11790 | 1.04k | int isExternal; |
11791 | 1.04k | int buildTree; |
11792 | 1.04k | int oldMinNsIndex; |
11793 | 1.04k | int oldNodelen, oldNodemem; |
11794 | | |
11795 | 1.04k | isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY); |
11796 | 1.04k | buildTree = (ctxt->node != NULL); |
11797 | | |
11798 | | /* |
11799 | | * Recursion check |
11800 | | */ |
11801 | 1.04k | if (ent->flags & XML_ENT_EXPANDING) { |
11802 | 9 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
11803 | 9 | xmlHaltParser(ctxt); |
11804 | 9 | goto error; |
11805 | 9 | } |
11806 | | |
11807 | | /* |
11808 | | * Load entity |
11809 | | */ |
11810 | 1.03k | input = xmlNewEntityInputStream(ctxt, ent); |
11811 | 1.03k | if (input == NULL) |
11812 | 0 | goto error; |
11813 | | |
11814 | | /* |
11815 | | * When building a tree, we need to limit the scope of namespace |
11816 | | * declarations, so that entities don't reference xmlNs structs |
11817 | | * from the parent of a reference. |
11818 | | */ |
11819 | 1.03k | oldMinNsIndex = ctxt->nsdb->minNsIndex; |
11820 | 1.03k | if (buildTree) |
11821 | 1.03k | ctxt->nsdb->minNsIndex = ctxt->nsNr; |
11822 | | |
11823 | 1.03k | oldNodelen = ctxt->nodelen; |
11824 | 1.03k | oldNodemem = ctxt->nodemem; |
11825 | 1.03k | ctxt->nodelen = 0; |
11826 | 1.03k | ctxt->nodemem = 0; |
11827 | | |
11828 | | /* |
11829 | | * Parse content |
11830 | | * |
11831 | | * This initiates a recursive call chain: |
11832 | | * |
11833 | | * - xmlCtxtParseContentInternal |
11834 | | * - xmlParseContentInternal |
11835 | | * - xmlParseReference |
11836 | | * - xmlCtxtParseEntity |
11837 | | * |
11838 | | * The nesting depth is limited by the maximum number of inputs, |
11839 | | * see xmlCtxtPushInput. |
11840 | | * |
11841 | | * It's possible to make this non-recursive (minNsIndex must be |
11842 | | * stored in the input struct) at the expense of code readability. |
11843 | | */ |
11844 | | |
11845 | 1.03k | ent->flags |= XML_ENT_EXPANDING; |
11846 | | |
11847 | 1.03k | list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree); |
11848 | | |
11849 | 1.03k | ent->flags &= ~XML_ENT_EXPANDING; |
11850 | | |
11851 | 1.03k | ctxt->nsdb->minNsIndex = oldMinNsIndex; |
11852 | 1.03k | ctxt->nodelen = oldNodelen; |
11853 | 1.03k | ctxt->nodemem = oldNodemem; |
11854 | | |
11855 | | /* |
11856 | | * Entity size accounting |
11857 | | */ |
11858 | 1.03k | consumed = input->consumed; |
11859 | 1.03k | xmlSaturatedAddSizeT(&consumed, input->end - input->base); |
11860 | | |
11861 | 1.03k | if ((ent->flags & XML_ENT_CHECKED) == 0) |
11862 | 552 | xmlSaturatedAdd(&ent->expandedSize, consumed); |
11863 | | |
11864 | 1.03k | if ((ent->flags & XML_ENT_PARSED) == 0) { |
11865 | 552 | if (isExternal) |
11866 | 0 | xmlSaturatedAdd(&ctxt->sizeentities, consumed); |
11867 | | |
11868 | 552 | ent->children = list; |
11869 | | |
11870 | 2.98k | while (list != NULL) { |
11871 | 2.43k | list->parent = (xmlNodePtr) ent; |
11872 | | |
11873 | | /* |
11874 | | * Downstream code like the nginx xslt module can set |
11875 | | * ctxt->myDoc->extSubset to a separate DTD, so the entity |
11876 | | * might have a different or a NULL document. |
11877 | | */ |
11878 | 2.43k | if (list->doc != ent->doc) |
11879 | 0 | xmlSetTreeDoc(list, ent->doc); |
11880 | | |
11881 | 2.43k | if (list->next == NULL) |
11882 | 356 | ent->last = list; |
11883 | 2.43k | list = list->next; |
11884 | 2.43k | } |
11885 | 552 | } else { |
11886 | 481 | xmlFreeNodeList(list); |
11887 | 481 | } |
11888 | | |
11889 | 1.03k | xmlFreeInputStream(input); |
11890 | | |
11891 | 1.04k | error: |
11892 | 1.04k | ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED; |
11893 | 1.04k | } |
11894 | | |
11895 | | /** |
11896 | | * Parse an external general entity within an existing parsing context |
11897 | | * An external general parsed entity is well-formed if it matches the |
11898 | | * production labeled extParsedEnt. |
11899 | | * |
11900 | | * [78] extParsedEnt ::= TextDecl? content |
11901 | | * |
11902 | | * @param ctxt the existing parsing context |
11903 | | * @param URL the URL for the entity to load |
11904 | | * @param ID the System ID for the entity to load |
11905 | | * @param listOut the return value for the set of parsed nodes |
11906 | | * @returns 0 if the entity is well formed, -1 in case of args problem and |
11907 | | * the parser error code otherwise |
11908 | | */ |
11909 | | |
11910 | | int |
11911 | | xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctxt, const xmlChar *URL, |
11912 | 0 | const xmlChar *ID, xmlNodePtr *listOut) { |
11913 | 0 | xmlParserInputPtr input; |
11914 | 0 | xmlNodePtr list; |
11915 | |
|
11916 | 0 | if (listOut != NULL) |
11917 | 0 | *listOut = NULL; |
11918 | |
|
11919 | 0 | if (ctxt == NULL) |
11920 | 0 | return(XML_ERR_ARGUMENT); |
11921 | | |
11922 | 0 | input = xmlLoadResource(ctxt, (char *) URL, (char *) ID, |
11923 | 0 | XML_RESOURCE_GENERAL_ENTITY); |
11924 | 0 | if (input == NULL) |
11925 | 0 | return(ctxt->errNo); |
11926 | | |
11927 | 0 | xmlCtxtInitializeLate(ctxt); |
11928 | |
|
11929 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1); |
11930 | 0 | if (listOut != NULL) |
11931 | 0 | *listOut = list; |
11932 | 0 | else |
11933 | 0 | xmlFreeNodeList(list); |
11934 | |
|
11935 | 0 | xmlFreeInputStream(input); |
11936 | 0 | return(ctxt->errNo); |
11937 | 0 | } |
11938 | | |
11939 | | #ifdef LIBXML_SAX1_ENABLED |
11940 | | /** |
11941 | | * Parse an external general entity |
11942 | | * An external general parsed entity is well-formed if it matches the |
11943 | | * production labeled extParsedEnt. |
11944 | | * |
11945 | | * @deprecated Use xmlParseCtxtExternalEntity(). |
11946 | | * |
11947 | | * [78] extParsedEnt ::= TextDecl? content |
11948 | | * |
11949 | | * @param doc the document the chunk pertains to |
11950 | | * @param sax the SAX handler block (possibly NULL) |
11951 | | * @param user_data The user data returned on SAX callbacks (possibly NULL) |
11952 | | * @param depth Used for loop detection, use 0 |
11953 | | * @param URL the URL for the entity to load |
11954 | | * @param ID the System ID for the entity to load |
11955 | | * @param list the return value for the set of parsed nodes |
11956 | | * @returns 0 if the entity is well formed, -1 in case of args problem and |
11957 | | * the parser error code otherwise |
11958 | | */ |
11959 | | |
11960 | | int |
11961 | | xmlParseExternalEntity(xmlDocPtr doc, xmlSAXHandlerPtr sax, void *user_data, |
11962 | 0 | int depth, const xmlChar *URL, const xmlChar *ID, xmlNodePtr *list) { |
11963 | 0 | xmlParserCtxtPtr ctxt; |
11964 | 0 | int ret; |
11965 | |
|
11966 | 0 | if (list != NULL) |
11967 | 0 | *list = NULL; |
11968 | |
|
11969 | 0 | if (doc == NULL) |
11970 | 0 | return(XML_ERR_ARGUMENT); |
11971 | | |
11972 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
11973 | 0 | if (ctxt == NULL) |
11974 | 0 | return(XML_ERR_NO_MEMORY); |
11975 | | |
11976 | 0 | ctxt->depth = depth; |
11977 | 0 | ctxt->myDoc = doc; |
11978 | 0 | ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list); |
11979 | |
|
11980 | 0 | xmlFreeParserCtxt(ctxt); |
11981 | 0 | return(ret); |
11982 | 0 | } |
11983 | | |
11984 | | /** |
11985 | | * Parse a well-balanced chunk of an XML document |
11986 | | * called by the parser |
11987 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
11988 | | * the content production in the XML grammar: |
11989 | | * |
11990 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | |
11991 | | * Comment)* |
11992 | | * |
11993 | | * @param doc the document the chunk pertains to (must not be NULL) |
11994 | | * @param sax the SAX handler block (possibly NULL) |
11995 | | * @param user_data The user data returned on SAX callbacks (possibly NULL) |
11996 | | * @param depth Used for loop detection, use 0 |
11997 | | * @param string the input string in UTF8 or ISO-Latin (zero terminated) |
11998 | | * @param lst the return value for the set of parsed nodes |
11999 | | * @returns 0 if the chunk is well balanced, -1 in case of args problem and |
12000 | | * the parser error code otherwise |
12001 | | */ |
12002 | | |
12003 | | int |
12004 | | xmlParseBalancedChunkMemory(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
12005 | 0 | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst) { |
12006 | 0 | return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data, |
12007 | 0 | depth, string, lst, 0 ); |
12008 | 0 | } |
12009 | | #endif /* LIBXML_SAX1_ENABLED */ |
12010 | | |
12011 | | /** |
12012 | | * Parse a well-balanced chunk of XML matching the 'content' production. |
12013 | | * |
12014 | | * Namespaces in scope of `node` and entities of `node`'s document are |
12015 | | * recognized. When validating, the DTD of `node`'s document is used. |
12016 | | * |
12017 | | * Always consumes `input` even in error case. |
12018 | | * |
12019 | | * @since 2.14.0 |
12020 | | * |
12021 | | * @param ctxt parser context |
12022 | | * @param input parser input |
12023 | | * @param node target node or document |
12024 | | * @param hasTextDecl whether to parse text declaration |
12025 | | * @returns a node list or NULL in case of error. |
12026 | | */ |
12027 | | xmlNodePtr |
12028 | | xmlCtxtParseContent(xmlParserCtxtPtr ctxt, xmlParserInputPtr input, |
12029 | 0 | xmlNodePtr node, int hasTextDecl) { |
12030 | 0 | xmlDocPtr doc; |
12031 | 0 | xmlNodePtr cur, list = NULL; |
12032 | 0 | int nsnr = 0; |
12033 | 0 | xmlDictPtr oldDict; |
12034 | 0 | int oldOptions, oldDictNames, oldLoadSubset; |
12035 | |
|
12036 | 0 | if ((ctxt == NULL) || (input == NULL) || (node == NULL)) { |
12037 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12038 | 0 | goto exit; |
12039 | 0 | } |
12040 | | |
12041 | 0 | doc = node->doc; |
12042 | 0 | if (doc == NULL) { |
12043 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12044 | 0 | goto exit; |
12045 | 0 | } |
12046 | | |
12047 | 0 | switch (node->type) { |
12048 | 0 | case XML_ELEMENT_NODE: |
12049 | 0 | case XML_DOCUMENT_NODE: |
12050 | 0 | case XML_HTML_DOCUMENT_NODE: |
12051 | 0 | break; |
12052 | | |
12053 | 0 | case XML_ATTRIBUTE_NODE: |
12054 | 0 | case XML_TEXT_NODE: |
12055 | 0 | case XML_CDATA_SECTION_NODE: |
12056 | 0 | case XML_ENTITY_REF_NODE: |
12057 | 0 | case XML_PI_NODE: |
12058 | 0 | case XML_COMMENT_NODE: |
12059 | 0 | for (cur = node->parent; cur != NULL; cur = node->parent) { |
12060 | 0 | if ((cur->type == XML_ELEMENT_NODE) || |
12061 | 0 | (cur->type == XML_DOCUMENT_NODE) || |
12062 | 0 | (cur->type == XML_HTML_DOCUMENT_NODE)) { |
12063 | 0 | node = cur; |
12064 | 0 | break; |
12065 | 0 | } |
12066 | 0 | } |
12067 | 0 | break; |
12068 | | |
12069 | 0 | default: |
12070 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12071 | 0 | goto exit; |
12072 | 0 | } |
12073 | | |
12074 | 0 | #ifdef LIBXML_HTML_ENABLED |
12075 | 0 | if (ctxt->html) |
12076 | 0 | htmlCtxtReset(ctxt); |
12077 | 0 | else |
12078 | 0 | #endif |
12079 | 0 | xmlCtxtReset(ctxt); |
12080 | |
|
12081 | 0 | oldDict = ctxt->dict; |
12082 | 0 | oldOptions = ctxt->options; |
12083 | 0 | oldDictNames = ctxt->dictNames; |
12084 | 0 | oldLoadSubset = ctxt->loadsubset; |
12085 | | |
12086 | | /* |
12087 | | * Use input doc's dict if present, else assure XML_PARSE_NODICT is set. |
12088 | | */ |
12089 | 0 | if (doc->dict != NULL) { |
12090 | 0 | ctxt->dict = doc->dict; |
12091 | 0 | } else { |
12092 | 0 | ctxt->options |= XML_PARSE_NODICT; |
12093 | 0 | ctxt->dictNames = 0; |
12094 | 0 | } |
12095 | | |
12096 | | /* |
12097 | | * Disable IDs |
12098 | | */ |
12099 | 0 | ctxt->loadsubset |= XML_SKIP_IDS; |
12100 | |
|
12101 | 0 | ctxt->myDoc = doc; |
12102 | |
|
12103 | 0 | #ifdef LIBXML_HTML_ENABLED |
12104 | 0 | if (ctxt->html) { |
12105 | | /* |
12106 | | * When parsing in context, it makes no sense to add implied |
12107 | | * elements like html/body/etc... |
12108 | | */ |
12109 | 0 | ctxt->options |= HTML_PARSE_NOIMPLIED; |
12110 | |
|
12111 | 0 | list = htmlCtxtParseContentInternal(ctxt, input); |
12112 | 0 | } else |
12113 | 0 | #endif |
12114 | 0 | { |
12115 | 0 | xmlCtxtInitializeLate(ctxt); |
12116 | | |
12117 | | /* |
12118 | | * initialize the SAX2 namespaces stack |
12119 | | */ |
12120 | 0 | cur = node; |
12121 | 0 | while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) { |
12122 | 0 | xmlNsPtr ns = cur->nsDef; |
12123 | 0 | xmlHashedString hprefix, huri; |
12124 | |
|
12125 | 0 | while (ns != NULL) { |
12126 | 0 | hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1); |
12127 | 0 | huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1); |
12128 | 0 | if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0) |
12129 | 0 | nsnr++; |
12130 | 0 | ns = ns->next; |
12131 | 0 | } |
12132 | 0 | cur = cur->parent; |
12133 | 0 | } |
12134 | |
|
12135 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1); |
12136 | |
|
12137 | 0 | if (nsnr > 0) |
12138 | 0 | xmlParserNsPop(ctxt, nsnr); |
12139 | 0 | } |
12140 | |
|
12141 | 0 | ctxt->dict = oldDict; |
12142 | 0 | ctxt->options = oldOptions; |
12143 | 0 | ctxt->dictNames = oldDictNames; |
12144 | 0 | ctxt->loadsubset = oldLoadSubset; |
12145 | 0 | ctxt->myDoc = NULL; |
12146 | 0 | ctxt->node = NULL; |
12147 | |
|
12148 | 0 | exit: |
12149 | 0 | xmlFreeInputStream(input); |
12150 | 0 | return(list); |
12151 | 0 | } |
12152 | | |
12153 | | /** |
12154 | | * Parse a well-balanced chunk of an XML document |
12155 | | * within the context (DTD, namespaces, etc ...) of the given node. |
12156 | | * |
12157 | | * The allowed sequence for the data is a Well Balanced Chunk defined by |
12158 | | * the content production in the XML grammar: |
12159 | | * |
12160 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | |
12161 | | * Comment)* |
12162 | | * |
12163 | | * This function assumes the encoding of `node`'s document which is |
12164 | | * typically not what you want. A better alternative is |
12165 | | * xmlCtxtParseContent(). |
12166 | | * |
12167 | | * @param node the context node |
12168 | | * @param data the input string |
12169 | | * @param datalen the input string length in bytes |
12170 | | * @param options a combination of xmlParserOption |
12171 | | * @param listOut the return value for the set of parsed nodes |
12172 | | * @returns XML_ERR_OK if the chunk is well balanced, and the parser |
12173 | | * error code otherwise |
12174 | | */ |
12175 | | xmlParserErrors |
12176 | | xmlParseInNodeContext(xmlNodePtr node, const char *data, int datalen, |
12177 | 0 | int options, xmlNodePtr *listOut) { |
12178 | 0 | xmlParserCtxtPtr ctxt; |
12179 | 0 | xmlParserInputPtr input; |
12180 | 0 | xmlDocPtr doc; |
12181 | 0 | xmlNodePtr list; |
12182 | 0 | xmlParserErrors ret; |
12183 | |
|
12184 | 0 | if (listOut == NULL) |
12185 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12186 | 0 | *listOut = NULL; |
12187 | |
|
12188 | 0 | if ((node == NULL) || (data == NULL) || (datalen < 0)) |
12189 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12190 | | |
12191 | 0 | doc = node->doc; |
12192 | 0 | if (doc == NULL) |
12193 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12194 | | |
12195 | 0 | #ifdef LIBXML_HTML_ENABLED |
12196 | 0 | if (doc->type == XML_HTML_DOCUMENT_NODE) { |
12197 | 0 | ctxt = htmlNewParserCtxt(); |
12198 | 0 | } |
12199 | 0 | else |
12200 | 0 | #endif |
12201 | 0 | ctxt = xmlNewParserCtxt(); |
12202 | |
|
12203 | 0 | if (ctxt == NULL) |
12204 | 0 | return(XML_ERR_NO_MEMORY); |
12205 | | |
12206 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen, |
12207 | 0 | (const char *) doc->encoding, |
12208 | 0 | XML_INPUT_BUF_STATIC); |
12209 | 0 | if (input == NULL) { |
12210 | 0 | xmlFreeParserCtxt(ctxt); |
12211 | 0 | return(XML_ERR_NO_MEMORY); |
12212 | 0 | } |
12213 | | |
12214 | 0 | xmlCtxtUseOptions(ctxt, options); |
12215 | |
|
12216 | 0 | list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0); |
12217 | |
|
12218 | 0 | if (list == NULL) { |
12219 | 0 | ret = ctxt->errNo; |
12220 | 0 | if (ret == XML_ERR_ARGUMENT) |
12221 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
12222 | 0 | } else { |
12223 | 0 | ret = XML_ERR_OK; |
12224 | 0 | *listOut = list; |
12225 | 0 | } |
12226 | |
|
12227 | 0 | xmlFreeParserCtxt(ctxt); |
12228 | |
|
12229 | 0 | return(ret); |
12230 | 0 | } |
12231 | | |
12232 | | #ifdef LIBXML_SAX1_ENABLED |
12233 | | /** |
12234 | | * Parse a well-balanced chunk of an XML document |
12235 | | * |
12236 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
12237 | | * the content production in the XML grammar: |
12238 | | * |
12239 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | |
12240 | | * Comment)* |
12241 | | * |
12242 | | * In case recover is set to 1, the nodelist will not be empty even if |
12243 | | * the parsed chunk is not well balanced, assuming the parsing succeeded to |
12244 | | * some extent. |
12245 | | * |
12246 | | * @param doc the document the chunk pertains to (must not be NULL) |
12247 | | * @param sax the SAX handler block (possibly NULL) |
12248 | | * @param user_data The user data returned on SAX callbacks (possibly NULL) |
12249 | | * @param depth Used for loop detection, use 0 |
12250 | | * @param string the input string in UTF8 or ISO-Latin (zero terminated) |
12251 | | * @param listOut the return value for the set of parsed nodes |
12252 | | * @param recover return nodes even if the data is broken (use 0) |
12253 | | * @returns 0 if the chunk is well balanced, or thehe parser error code |
12254 | | * otherwise. |
12255 | | */ |
12256 | | int |
12257 | | xmlParseBalancedChunkMemoryRecover(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
12258 | | void *user_data, int depth, const xmlChar *string, xmlNodePtr *listOut, |
12259 | 0 | int recover) { |
12260 | 0 | xmlParserCtxtPtr ctxt; |
12261 | 0 | xmlParserInputPtr input; |
12262 | 0 | xmlNodePtr list; |
12263 | 0 | int ret; |
12264 | |
|
12265 | 0 | if (listOut != NULL) |
12266 | 0 | *listOut = NULL; |
12267 | |
|
12268 | 0 | if (string == NULL) |
12269 | 0 | return(XML_ERR_ARGUMENT); |
12270 | | |
12271 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
12272 | 0 | if (ctxt == NULL) |
12273 | 0 | return(XML_ERR_NO_MEMORY); |
12274 | | |
12275 | 0 | xmlCtxtInitializeLate(ctxt); |
12276 | |
|
12277 | 0 | ctxt->depth = depth; |
12278 | 0 | ctxt->myDoc = doc; |
12279 | 0 | if (recover) { |
12280 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12281 | 0 | ctxt->recovery = 1; |
12282 | 0 | } |
12283 | |
|
12284 | 0 | input = xmlNewStringInputStream(ctxt, string); |
12285 | 0 | if (input == NULL) { |
12286 | 0 | ret = ctxt->errNo; |
12287 | 0 | goto error; |
12288 | 0 | } |
12289 | | |
12290 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1); |
12291 | 0 | if (listOut != NULL) |
12292 | 0 | *listOut = list; |
12293 | 0 | else |
12294 | 0 | xmlFreeNodeList(list); |
12295 | |
|
12296 | 0 | if (!ctxt->wellFormed) |
12297 | 0 | ret = ctxt->errNo; |
12298 | 0 | else |
12299 | 0 | ret = XML_ERR_OK; |
12300 | |
|
12301 | 0 | error: |
12302 | 0 | xmlFreeInputStream(input); |
12303 | 0 | xmlFreeParserCtxt(ctxt); |
12304 | 0 | return(ret); |
12305 | 0 | } |
12306 | | |
12307 | | /** |
12308 | | * parse an XML external entity out of context and build a tree. |
12309 | | * It use the given SAX function block to handle the parsing callback. |
12310 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12311 | | * |
12312 | | * @deprecated Don't use. |
12313 | | * |
12314 | | * [78] extParsedEnt ::= TextDecl? content |
12315 | | * |
12316 | | * This correspond to a "Well Balanced" chunk |
12317 | | * |
12318 | | * @param sax the SAX handler block |
12319 | | * @param filename the filename |
12320 | | * @returns the resulting document tree |
12321 | | */ |
12322 | | |
12323 | | xmlDocPtr |
12324 | 0 | xmlSAXParseEntity(xmlSAXHandlerPtr sax, const char *filename) { |
12325 | 0 | xmlDocPtr ret; |
12326 | 0 | xmlParserCtxtPtr ctxt; |
12327 | |
|
12328 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
12329 | 0 | if (ctxt == NULL) { |
12330 | 0 | return(NULL); |
12331 | 0 | } |
12332 | 0 | if (sax != NULL) { |
12333 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
12334 | 0 | *ctxt->sax = *sax; |
12335 | 0 | } else { |
12336 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
12337 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12338 | 0 | } |
12339 | 0 | ctxt->userData = NULL; |
12340 | 0 | } |
12341 | |
|
12342 | 0 | xmlParseExtParsedEnt(ctxt); |
12343 | |
|
12344 | 0 | if (ctxt->wellFormed) { |
12345 | 0 | ret = ctxt->myDoc; |
12346 | 0 | } else { |
12347 | 0 | ret = NULL; |
12348 | 0 | xmlFreeDoc(ctxt->myDoc); |
12349 | 0 | } |
12350 | |
|
12351 | 0 | xmlFreeParserCtxt(ctxt); |
12352 | |
|
12353 | 0 | return(ret); |
12354 | 0 | } |
12355 | | |
12356 | | /** |
12357 | | * parse an XML external entity out of context and build a tree. |
12358 | | * |
12359 | | * [78] extParsedEnt ::= TextDecl? content |
12360 | | * |
12361 | | * This correspond to a "Well Balanced" chunk |
12362 | | * |
12363 | | * @param filename the filename |
12364 | | * @returns the resulting document tree |
12365 | | */ |
12366 | | |
12367 | | xmlDocPtr |
12368 | 0 | xmlParseEntity(const char *filename) { |
12369 | 0 | return(xmlSAXParseEntity(NULL, filename)); |
12370 | 0 | } |
12371 | | #endif /* LIBXML_SAX1_ENABLED */ |
12372 | | |
12373 | | /** |
12374 | | * Create a parser context for an external entity |
12375 | | * Automatic support for ZLIB/Compress compressed document is provided |
12376 | | * by default if found at compile-time. |
12377 | | * |
12378 | | * @deprecated Don't use. |
12379 | | * |
12380 | | * @param URL the entity URL |
12381 | | * @param ID the entity PUBLIC ID |
12382 | | * @param base a possible base for the target URI |
12383 | | * @returns the new parser context or NULL |
12384 | | */ |
12385 | | xmlParserCtxtPtr |
12386 | | xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID, |
12387 | 0 | const xmlChar *base) { |
12388 | 0 | xmlParserCtxtPtr ctxt; |
12389 | 0 | xmlParserInputPtr input; |
12390 | 0 | xmlChar *uri = NULL; |
12391 | |
|
12392 | 0 | ctxt = xmlNewParserCtxt(); |
12393 | 0 | if (ctxt == NULL) |
12394 | 0 | return(NULL); |
12395 | | |
12396 | 0 | if (base != NULL) { |
12397 | 0 | if (xmlBuildURISafe(URL, base, &uri) < 0) |
12398 | 0 | goto error; |
12399 | 0 | if (uri != NULL) |
12400 | 0 | URL = uri; |
12401 | 0 | } |
12402 | | |
12403 | 0 | input = xmlLoadResource(ctxt, (char *) URL, (char *) ID, |
12404 | 0 | XML_RESOURCE_UNKNOWN); |
12405 | 0 | if (input == NULL) |
12406 | 0 | goto error; |
12407 | | |
12408 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
12409 | 0 | xmlFreeInputStream(input); |
12410 | 0 | goto error; |
12411 | 0 | } |
12412 | | |
12413 | 0 | xmlFree(uri); |
12414 | 0 | return(ctxt); |
12415 | | |
12416 | 0 | error: |
12417 | 0 | xmlFree(uri); |
12418 | 0 | xmlFreeParserCtxt(ctxt); |
12419 | 0 | return(NULL); |
12420 | 0 | } |
12421 | | |
12422 | | /************************************************************************ |
12423 | | * * |
12424 | | * Front ends when parsing from a file * |
12425 | | * * |
12426 | | ************************************************************************/ |
12427 | | |
12428 | | /** |
12429 | | * Create a parser context for a file or URL content. |
12430 | | * Automatic support for ZLIB/Compress compressed document is provided |
12431 | | * by default if found at compile-time and for file accesses |
12432 | | * |
12433 | | * @deprecated Use xmlNewParserCtxt() and xmlCtxtReadFile(). |
12434 | | * |
12435 | | * @param filename the filename or URL |
12436 | | * @param options a combination of xmlParserOption |
12437 | | * @returns the new parser context or NULL |
12438 | | */ |
12439 | | xmlParserCtxtPtr |
12440 | | xmlCreateURLParserCtxt(const char *filename, int options) |
12441 | 0 | { |
12442 | 0 | xmlParserCtxtPtr ctxt; |
12443 | 0 | xmlParserInputPtr input; |
12444 | |
|
12445 | 0 | ctxt = xmlNewParserCtxt(); |
12446 | 0 | if (ctxt == NULL) |
12447 | 0 | return(NULL); |
12448 | | |
12449 | 0 | options |= XML_PARSE_UNZIP; |
12450 | |
|
12451 | 0 | xmlCtxtUseOptions(ctxt, options); |
12452 | 0 | ctxt->linenumbers = 1; |
12453 | |
|
12454 | 0 | input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT); |
12455 | 0 | if (input == NULL) { |
12456 | 0 | xmlFreeParserCtxt(ctxt); |
12457 | 0 | return(NULL); |
12458 | 0 | } |
12459 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
12460 | 0 | xmlFreeInputStream(input); |
12461 | 0 | xmlFreeParserCtxt(ctxt); |
12462 | 0 | return(NULL); |
12463 | 0 | } |
12464 | | |
12465 | 0 | return(ctxt); |
12466 | 0 | } |
12467 | | |
12468 | | /** |
12469 | | * Create a parser context for a file content. |
12470 | | * Automatic support for ZLIB/Compress compressed document is provided |
12471 | | * by default if found at compile-time. |
12472 | | * |
12473 | | * @deprecated Use xmlNewParserCtxt() and xmlCtxtReadFile(). |
12474 | | * |
12475 | | * @param filename the filename |
12476 | | * @returns the new parser context or NULL |
12477 | | */ |
12478 | | xmlParserCtxtPtr |
12479 | | xmlCreateFileParserCtxt(const char *filename) |
12480 | 0 | { |
12481 | 0 | return(xmlCreateURLParserCtxt(filename, 0)); |
12482 | 0 | } |
12483 | | |
12484 | | #ifdef LIBXML_SAX1_ENABLED |
12485 | | /** |
12486 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12487 | | * compressed document is provided by default if found at compile-time. |
12488 | | * It use the given SAX function block to handle the parsing callback. |
12489 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12490 | | * |
12491 | | * @deprecated Use xmlNewSAXParserCtxt() and xmlCtxtReadFile(). |
12492 | | * |
12493 | | * User data (void *) is stored within the parser context in the |
12494 | | * context's _private member, so it is available nearly everywhere in libxml |
12495 | | * |
12496 | | * @param sax the SAX handler block |
12497 | | * @param filename the filename |
12498 | | * @param recovery work in recovery mode, i.e. tries to read no Well Formed |
12499 | | * documents |
12500 | | * @param data the userdata |
12501 | | * @returns the resulting document tree |
12502 | | */ |
12503 | | |
12504 | | xmlDocPtr |
12505 | | xmlSAXParseFileWithData(xmlSAXHandlerPtr sax, const char *filename, |
12506 | 0 | int recovery, void *data) { |
12507 | 0 | xmlDocPtr ret = NULL; |
12508 | 0 | xmlParserCtxtPtr ctxt; |
12509 | 0 | xmlParserInputPtr input; |
12510 | |
|
12511 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12512 | 0 | if (ctxt == NULL) |
12513 | 0 | return(NULL); |
12514 | | |
12515 | 0 | if (data != NULL) |
12516 | 0 | ctxt->_private = data; |
12517 | |
|
12518 | 0 | if (recovery) { |
12519 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12520 | 0 | ctxt->recovery = 1; |
12521 | 0 | } |
12522 | |
|
12523 | 0 | if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0)) |
12524 | 0 | input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0); |
12525 | 0 | else |
12526 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0); |
12527 | |
|
12528 | 0 | if (input != NULL) |
12529 | 0 | ret = xmlCtxtParseDocument(ctxt, input); |
12530 | |
|
12531 | 0 | xmlFreeParserCtxt(ctxt); |
12532 | 0 | return(ret); |
12533 | 0 | } |
12534 | | |
12535 | | /** |
12536 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12537 | | * compressed document is provided by default if found at compile-time. |
12538 | | * It use the given SAX function block to handle the parsing callback. |
12539 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12540 | | * |
12541 | | * @deprecated Use xmlNewSAXParserCtxt() and xmlCtxtReadFile(). |
12542 | | * |
12543 | | * @param sax the SAX handler block |
12544 | | * @param filename the filename |
12545 | | * @param recovery work in recovery mode, i.e. tries to read no Well Formed |
12546 | | * documents |
12547 | | * @returns the resulting document tree |
12548 | | */ |
12549 | | |
12550 | | xmlDocPtr |
12551 | | xmlSAXParseFile(xmlSAXHandlerPtr sax, const char *filename, |
12552 | 0 | int recovery) { |
12553 | 0 | return(xmlSAXParseFileWithData(sax,filename,recovery,NULL)); |
12554 | 0 | } |
12555 | | |
12556 | | /** |
12557 | | * parse an XML in-memory document and build a tree. |
12558 | | * In the case the document is not Well Formed, a attempt to build a |
12559 | | * tree is tried anyway |
12560 | | * |
12561 | | * @deprecated Use xmlReadDoc() with XML_PARSE_RECOVER. |
12562 | | * |
12563 | | * @param cur a pointer to an array of xmlChar |
12564 | | * @returns the resulting document tree or NULL in case of failure |
12565 | | */ |
12566 | | |
12567 | | xmlDocPtr |
12568 | 0 | xmlRecoverDoc(const xmlChar *cur) { |
12569 | 0 | return(xmlSAXParseDoc(NULL, cur, 1)); |
12570 | 0 | } |
12571 | | |
12572 | | /** |
12573 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12574 | | * compressed document is provided by default if found at compile-time. |
12575 | | * |
12576 | | * @deprecated Use xmlReadFile(). |
12577 | | * |
12578 | | * @param filename the filename |
12579 | | * @returns the resulting document tree if the file was wellformed, |
12580 | | * NULL otherwise. |
12581 | | */ |
12582 | | |
12583 | | xmlDocPtr |
12584 | 0 | xmlParseFile(const char *filename) { |
12585 | 0 | return(xmlSAXParseFile(NULL, filename, 0)); |
12586 | 0 | } |
12587 | | |
12588 | | /** |
12589 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12590 | | * compressed document is provided by default if found at compile-time. |
12591 | | * In the case the document is not Well Formed, it attempts to build |
12592 | | * a tree anyway |
12593 | | * |
12594 | | * @deprecated Use xmlReadFile() with XML_PARSE_RECOVER. |
12595 | | * |
12596 | | * @param filename the filename |
12597 | | * @returns the resulting document tree or NULL in case of failure |
12598 | | */ |
12599 | | |
12600 | | xmlDocPtr |
12601 | 0 | xmlRecoverFile(const char *filename) { |
12602 | 0 | return(xmlSAXParseFile(NULL, filename, 1)); |
12603 | 0 | } |
12604 | | |
12605 | | |
12606 | | /** |
12607 | | * Setup the parser context to parse a new buffer; Clears any prior |
12608 | | * contents from the parser context. The buffer parameter must not be |
12609 | | * NULL, but the filename parameter can be |
12610 | | * |
12611 | | * @deprecated Don't use. |
12612 | | * |
12613 | | * @param ctxt an XML parser context |
12614 | | * @param buffer a xmlChar * buffer |
12615 | | * @param filename a file name |
12616 | | */ |
12617 | | void |
12618 | | xmlSetupParserForBuffer(xmlParserCtxtPtr ctxt, const xmlChar* buffer, |
12619 | | const char* filename) |
12620 | 0 | { |
12621 | 0 | xmlParserInputPtr input; |
12622 | |
|
12623 | 0 | if ((ctxt == NULL) || (buffer == NULL)) |
12624 | 0 | return; |
12625 | | |
12626 | 0 | xmlCtxtReset(ctxt); |
12627 | |
|
12628 | 0 | input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer, |
12629 | 0 | NULL, 0); |
12630 | 0 | if (input == NULL) |
12631 | 0 | return; |
12632 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) |
12633 | 0 | xmlFreeInputStream(input); |
12634 | 0 | } |
12635 | | |
12636 | | /** |
12637 | | * parse an XML file and call the given SAX handler routines. |
12638 | | * Automatic support for ZLIB/Compress compressed document is provided |
12639 | | * |
12640 | | * @deprecated Use xmlNewSAXParserCtxt() and xmlCtxtReadFile(). |
12641 | | * |
12642 | | * @param sax a SAX handler |
12643 | | * @param user_data The user data returned on SAX callbacks |
12644 | | * @param filename a file name |
12645 | | * @returns 0 in case of success or a error number otherwise |
12646 | | */ |
12647 | | int |
12648 | | xmlSAXUserParseFile(xmlSAXHandlerPtr sax, void *user_data, |
12649 | 0 | const char *filename) { |
12650 | 0 | int ret = 0; |
12651 | 0 | xmlParserCtxtPtr ctxt; |
12652 | |
|
12653 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
12654 | 0 | if (ctxt == NULL) return -1; |
12655 | 0 | if (sax != NULL) { |
12656 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
12657 | 0 | *ctxt->sax = *sax; |
12658 | 0 | } else { |
12659 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
12660 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12661 | 0 | } |
12662 | 0 | ctxt->userData = user_data; |
12663 | 0 | } |
12664 | |
|
12665 | 0 | xmlParseDocument(ctxt); |
12666 | |
|
12667 | 0 | if (ctxt->wellFormed) |
12668 | 0 | ret = 0; |
12669 | 0 | else { |
12670 | 0 | if (ctxt->errNo != 0) |
12671 | 0 | ret = ctxt->errNo; |
12672 | 0 | else |
12673 | 0 | ret = -1; |
12674 | 0 | } |
12675 | 0 | if (ctxt->myDoc != NULL) { |
12676 | 0 | xmlFreeDoc(ctxt->myDoc); |
12677 | 0 | ctxt->myDoc = NULL; |
12678 | 0 | } |
12679 | 0 | xmlFreeParserCtxt(ctxt); |
12680 | |
|
12681 | 0 | return ret; |
12682 | 0 | } |
12683 | | #endif /* LIBXML_SAX1_ENABLED */ |
12684 | | |
12685 | | /************************************************************************ |
12686 | | * * |
12687 | | * Front ends when parsing from memory * |
12688 | | * * |
12689 | | ************************************************************************/ |
12690 | | |
12691 | | /** |
12692 | | * Create a parser context for an XML in-memory document. The input buffer |
12693 | | * must not contain a terminating null byte. |
12694 | | * |
12695 | | * @param buffer a pointer to a char array |
12696 | | * @param size the size of the array |
12697 | | * @returns the new parser context or NULL |
12698 | | */ |
12699 | | xmlParserCtxtPtr |
12700 | 0 | xmlCreateMemoryParserCtxt(const char *buffer, int size) { |
12701 | 0 | xmlParserCtxtPtr ctxt; |
12702 | 0 | xmlParserInputPtr input; |
12703 | |
|
12704 | 0 | if (size < 0) |
12705 | 0 | return(NULL); |
12706 | | |
12707 | 0 | ctxt = xmlNewParserCtxt(); |
12708 | 0 | if (ctxt == NULL) |
12709 | 0 | return(NULL); |
12710 | | |
12711 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0); |
12712 | 0 | if (input == NULL) { |
12713 | 0 | xmlFreeParserCtxt(ctxt); |
12714 | 0 | return(NULL); |
12715 | 0 | } |
12716 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
12717 | 0 | xmlFreeInputStream(input); |
12718 | 0 | xmlFreeParserCtxt(ctxt); |
12719 | 0 | return(NULL); |
12720 | 0 | } |
12721 | | |
12722 | 0 | return(ctxt); |
12723 | 0 | } |
12724 | | |
12725 | | #ifdef LIBXML_SAX1_ENABLED |
12726 | | /** |
12727 | | * parse an XML in-memory block and use the given SAX function block |
12728 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
12729 | | * DOM tree building routines. |
12730 | | * |
12731 | | * @deprecated Use xmlNewSAXParserCtxt() and xmlCtxtReadMemory(). |
12732 | | * |
12733 | | * User data (void *) is stored within the parser context in the |
12734 | | * context's _private member, so it is available nearly everywhere in libxml |
12735 | | * |
12736 | | * @param sax the SAX handler block |
12737 | | * @param buffer an pointer to a char array |
12738 | | * @param size the size of the array |
12739 | | * @param recovery work in recovery mode, i.e. tries to read no Well Formed |
12740 | | * documents |
12741 | | * @param data the userdata |
12742 | | * @returns the resulting document tree |
12743 | | */ |
12744 | | |
12745 | | xmlDocPtr |
12746 | | xmlSAXParseMemoryWithData(xmlSAXHandlerPtr sax, const char *buffer, |
12747 | 0 | int size, int recovery, void *data) { |
12748 | 0 | xmlDocPtr ret = NULL; |
12749 | 0 | xmlParserCtxtPtr ctxt; |
12750 | 0 | xmlParserInputPtr input; |
12751 | |
|
12752 | 0 | if (size < 0) |
12753 | 0 | return(NULL); |
12754 | | |
12755 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12756 | 0 | if (ctxt == NULL) |
12757 | 0 | return(NULL); |
12758 | | |
12759 | 0 | if (data != NULL) |
12760 | 0 | ctxt->_private=data; |
12761 | |
|
12762 | 0 | if (recovery) { |
12763 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12764 | 0 | ctxt->recovery = 1; |
12765 | 0 | } |
12766 | |
|
12767 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, |
12768 | 0 | XML_INPUT_BUF_STATIC); |
12769 | |
|
12770 | 0 | if (input != NULL) |
12771 | 0 | ret = xmlCtxtParseDocument(ctxt, input); |
12772 | |
|
12773 | 0 | xmlFreeParserCtxt(ctxt); |
12774 | 0 | return(ret); |
12775 | 0 | } |
12776 | | |
12777 | | /** |
12778 | | * parse an XML in-memory block and use the given SAX function block |
12779 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
12780 | | * DOM tree building routines. |
12781 | | * |
12782 | | * @deprecated Use xmlNewSAXParserCtxt() and xmlCtxtReadMemory(). |
12783 | | * |
12784 | | * @param sax the SAX handler block |
12785 | | * @param buffer an pointer to a char array |
12786 | | * @param size the size of the array |
12787 | | * @param recovery work in recovery mode, i.e. tries to read not Well Formed |
12788 | | * documents |
12789 | | * @returns the resulting document tree |
12790 | | */ |
12791 | | xmlDocPtr |
12792 | | xmlSAXParseMemory(xmlSAXHandlerPtr sax, const char *buffer, |
12793 | 0 | int size, int recovery) { |
12794 | 0 | return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL); |
12795 | 0 | } |
12796 | | |
12797 | | /** |
12798 | | * parse an XML in-memory block and build a tree. |
12799 | | * |
12800 | | * @deprecated Use xmlReadMemory(). |
12801 | | * |
12802 | | * @param buffer an pointer to a char array |
12803 | | * @param size the size of the array |
12804 | | * @returns the resulting document tree |
12805 | | */ |
12806 | | |
12807 | 0 | xmlDocPtr xmlParseMemory(const char *buffer, int size) { |
12808 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 0)); |
12809 | 0 | } |
12810 | | |
12811 | | /** |
12812 | | * parse an XML in-memory block and build a tree. |
12813 | | * In the case the document is not Well Formed, an attempt to |
12814 | | * build a tree is tried anyway |
12815 | | * |
12816 | | * @deprecated Use xmlReadMemory() with XML_PARSE_RECOVER. |
12817 | | * |
12818 | | * @param buffer an pointer to a char array |
12819 | | * @param size the size of the array |
12820 | | * @returns the resulting document tree or NULL in case of error |
12821 | | */ |
12822 | | |
12823 | 0 | xmlDocPtr xmlRecoverMemory(const char *buffer, int size) { |
12824 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 1)); |
12825 | 0 | } |
12826 | | |
12827 | | /** |
12828 | | * parse an XML in-memory buffer and call the given SAX handler routines. |
12829 | | * |
12830 | | * @deprecated Use xmlNewSAXParserCtxt() and xmlCtxtReadMemory(). |
12831 | | * |
12832 | | * @param sax a SAX handler |
12833 | | * @param user_data The user data returned on SAX callbacks |
12834 | | * @param buffer an in-memory XML document input |
12835 | | * @param size the length of the XML document in bytes |
12836 | | * @returns 0 in case of success or a error number otherwise |
12837 | | */ |
12838 | | int xmlSAXUserParseMemory(xmlSAXHandlerPtr sax, void *user_data, |
12839 | 0 | const char *buffer, int size) { |
12840 | 0 | int ret = 0; |
12841 | 0 | xmlParserCtxtPtr ctxt; |
12842 | |
|
12843 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
12844 | 0 | if (ctxt == NULL) return -1; |
12845 | 0 | if (sax != NULL) { |
12846 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
12847 | 0 | *ctxt->sax = *sax; |
12848 | 0 | } else { |
12849 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
12850 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12851 | 0 | } |
12852 | 0 | ctxt->userData = user_data; |
12853 | 0 | } |
12854 | |
|
12855 | 0 | xmlParseDocument(ctxt); |
12856 | |
|
12857 | 0 | if (ctxt->wellFormed) |
12858 | 0 | ret = 0; |
12859 | 0 | else { |
12860 | 0 | if (ctxt->errNo != 0) |
12861 | 0 | ret = ctxt->errNo; |
12862 | 0 | else |
12863 | 0 | ret = -1; |
12864 | 0 | } |
12865 | 0 | if (ctxt->myDoc != NULL) { |
12866 | 0 | xmlFreeDoc(ctxt->myDoc); |
12867 | 0 | ctxt->myDoc = NULL; |
12868 | 0 | } |
12869 | 0 | xmlFreeParserCtxt(ctxt); |
12870 | |
|
12871 | 0 | return ret; |
12872 | 0 | } |
12873 | | #endif /* LIBXML_SAX1_ENABLED */ |
12874 | | |
12875 | | /** |
12876 | | * Creates a parser context for an XML in-memory document. |
12877 | | * |
12878 | | * @param str a pointer to an array of xmlChar |
12879 | | * @returns the new parser context or NULL |
12880 | | */ |
12881 | | xmlParserCtxtPtr |
12882 | 0 | xmlCreateDocParserCtxt(const xmlChar *str) { |
12883 | 0 | xmlParserCtxtPtr ctxt; |
12884 | 0 | xmlParserInputPtr input; |
12885 | |
|
12886 | 0 | ctxt = xmlNewParserCtxt(); |
12887 | 0 | if (ctxt == NULL) |
12888 | 0 | return(NULL); |
12889 | | |
12890 | 0 | input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0); |
12891 | 0 | if (input == NULL) { |
12892 | 0 | xmlFreeParserCtxt(ctxt); |
12893 | 0 | return(NULL); |
12894 | 0 | } |
12895 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
12896 | 0 | xmlFreeInputStream(input); |
12897 | 0 | xmlFreeParserCtxt(ctxt); |
12898 | 0 | return(NULL); |
12899 | 0 | } |
12900 | | |
12901 | 0 | return(ctxt); |
12902 | 0 | } |
12903 | | |
12904 | | #ifdef LIBXML_SAX1_ENABLED |
12905 | | /** |
12906 | | * parse an XML in-memory document and build a tree. |
12907 | | * It use the given SAX function block to handle the parsing callback. |
12908 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12909 | | * |
12910 | | * @deprecated Use xmlNewSAXParserCtxt() and xmlCtxtReadDoc(). |
12911 | | * |
12912 | | * @param sax the SAX handler block |
12913 | | * @param cur a pointer to an array of xmlChar |
12914 | | * @param recovery work in recovery mode, i.e. tries to read no Well Formed |
12915 | | * documents |
12916 | | * @returns the resulting document tree |
12917 | | */ |
12918 | | |
12919 | | xmlDocPtr |
12920 | 0 | xmlSAXParseDoc(xmlSAXHandlerPtr sax, const xmlChar *cur, int recovery) { |
12921 | 0 | xmlDocPtr ret; |
12922 | 0 | xmlParserCtxtPtr ctxt; |
12923 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
12924 | |
|
12925 | 0 | if (cur == NULL) return(NULL); |
12926 | | |
12927 | | |
12928 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
12929 | 0 | if (ctxt == NULL) return(NULL); |
12930 | 0 | if (sax != NULL) { |
12931 | 0 | oldsax = ctxt->sax; |
12932 | 0 | ctxt->sax = sax; |
12933 | 0 | ctxt->userData = NULL; |
12934 | 0 | } |
12935 | |
|
12936 | 0 | xmlParseDocument(ctxt); |
12937 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
12938 | 0 | else { |
12939 | 0 | ret = NULL; |
12940 | 0 | xmlFreeDoc(ctxt->myDoc); |
12941 | 0 | ctxt->myDoc = NULL; |
12942 | 0 | } |
12943 | 0 | if (sax != NULL) |
12944 | 0 | ctxt->sax = oldsax; |
12945 | 0 | xmlFreeParserCtxt(ctxt); |
12946 | |
|
12947 | 0 | return(ret); |
12948 | 0 | } |
12949 | | |
12950 | | /** |
12951 | | * parse an XML in-memory document and build a tree. |
12952 | | * |
12953 | | * @deprecated Use xmlReadDoc(). |
12954 | | * |
12955 | | * @param cur a pointer to an array of xmlChar |
12956 | | * @returns the resulting document tree |
12957 | | */ |
12958 | | |
12959 | | xmlDocPtr |
12960 | 0 | xmlParseDoc(const xmlChar *cur) { |
12961 | 0 | return(xmlSAXParseDoc(NULL, cur, 0)); |
12962 | 0 | } |
12963 | | #endif /* LIBXML_SAX1_ENABLED */ |
12964 | | |
12965 | | /************************************************************************ |
12966 | | * * |
12967 | | * New set (2.6.0) of simpler and more flexible APIs * |
12968 | | * * |
12969 | | ************************************************************************/ |
12970 | | |
12971 | | /** |
12972 | | * Free a string if it is not owned by the "dict" dictionary in the |
12973 | | * current scope |
12974 | | * |
12975 | | * @param str a string |
12976 | | */ |
12977 | | #define DICT_FREE(str) \ |
12978 | 0 | if ((str) && ((!dict) || \ |
12979 | 0 | (xmlDictOwns(dict, (const xmlChar *)(str)) == 0))) \ |
12980 | 0 | xmlFree((char *)(str)); |
12981 | | |
12982 | | /** |
12983 | | * Reset a parser context |
12984 | | * |
12985 | | * @param ctxt an XML parser context |
12986 | | */ |
12987 | | void |
12988 | | xmlCtxtReset(xmlParserCtxtPtr ctxt) |
12989 | 0 | { |
12990 | 0 | xmlParserInputPtr input; |
12991 | 0 | xmlDictPtr dict; |
12992 | |
|
12993 | 0 | if (ctxt == NULL) |
12994 | 0 | return; |
12995 | | |
12996 | 0 | dict = ctxt->dict; |
12997 | |
|
12998 | 0 | while ((input = xmlCtxtPopInput(ctxt)) != NULL) { /* Non consuming */ |
12999 | 0 | xmlFreeInputStream(input); |
13000 | 0 | } |
13001 | 0 | ctxt->inputNr = 0; |
13002 | 0 | ctxt->input = NULL; |
13003 | |
|
13004 | 0 | ctxt->spaceNr = 0; |
13005 | 0 | if (ctxt->spaceTab != NULL) { |
13006 | 0 | ctxt->spaceTab[0] = -1; |
13007 | 0 | ctxt->space = &ctxt->spaceTab[0]; |
13008 | 0 | } else { |
13009 | 0 | ctxt->space = NULL; |
13010 | 0 | } |
13011 | | |
13012 | |
|
13013 | 0 | ctxt->nodeNr = 0; |
13014 | 0 | ctxt->node = NULL; |
13015 | |
|
13016 | 0 | ctxt->nameNr = 0; |
13017 | 0 | ctxt->name = NULL; |
13018 | |
|
13019 | 0 | ctxt->nsNr = 0; |
13020 | 0 | xmlParserNsReset(ctxt->nsdb); |
13021 | |
|
13022 | 0 | DICT_FREE(ctxt->version); |
13023 | 0 | ctxt->version = NULL; |
13024 | 0 | DICT_FREE(ctxt->encoding); |
13025 | 0 | ctxt->encoding = NULL; |
13026 | 0 | DICT_FREE(ctxt->extSubURI); |
13027 | 0 | ctxt->extSubURI = NULL; |
13028 | 0 | DICT_FREE(ctxt->extSubSystem); |
13029 | 0 | ctxt->extSubSystem = NULL; |
13030 | |
|
13031 | 0 | if (ctxt->directory != NULL) { |
13032 | 0 | xmlFree(ctxt->directory); |
13033 | 0 | ctxt->directory = NULL; |
13034 | 0 | } |
13035 | |
|
13036 | 0 | if (ctxt->myDoc != NULL) |
13037 | 0 | xmlFreeDoc(ctxt->myDoc); |
13038 | 0 | ctxt->myDoc = NULL; |
13039 | |
|
13040 | 0 | ctxt->standalone = -1; |
13041 | 0 | ctxt->hasExternalSubset = 0; |
13042 | 0 | ctxt->hasPErefs = 0; |
13043 | 0 | ctxt->html = 0; |
13044 | 0 | ctxt->instate = XML_PARSER_START; |
13045 | |
|
13046 | 0 | ctxt->wellFormed = 1; |
13047 | 0 | ctxt->nsWellFormed = 1; |
13048 | 0 | ctxt->disableSAX = 0; |
13049 | 0 | ctxt->valid = 1; |
13050 | 0 | ctxt->record_info = 0; |
13051 | 0 | ctxt->checkIndex = 0; |
13052 | 0 | ctxt->endCheckState = 0; |
13053 | 0 | ctxt->inSubset = 0; |
13054 | 0 | ctxt->errNo = XML_ERR_OK; |
13055 | 0 | ctxt->depth = 0; |
13056 | 0 | ctxt->catalogs = NULL; |
13057 | 0 | ctxt->sizeentities = 0; |
13058 | 0 | ctxt->sizeentcopy = 0; |
13059 | 0 | xmlInitNodeInfoSeq(&ctxt->node_seq); |
13060 | |
|
13061 | 0 | if (ctxt->attsDefault != NULL) { |
13062 | 0 | xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator); |
13063 | 0 | ctxt->attsDefault = NULL; |
13064 | 0 | } |
13065 | 0 | if (ctxt->attsSpecial != NULL) { |
13066 | 0 | xmlHashFree(ctxt->attsSpecial, NULL); |
13067 | 0 | ctxt->attsSpecial = NULL; |
13068 | 0 | } |
13069 | |
|
13070 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
13071 | 0 | if (ctxt->catalogs != NULL) |
13072 | 0 | xmlCatalogFreeLocal(ctxt->catalogs); |
13073 | 0 | #endif |
13074 | 0 | ctxt->nbErrors = 0; |
13075 | 0 | ctxt->nbWarnings = 0; |
13076 | 0 | if (ctxt->lastError.code != XML_ERR_OK) |
13077 | 0 | xmlResetError(&ctxt->lastError); |
13078 | 0 | } |
13079 | | |
13080 | | /** |
13081 | | * Reset a push parser context |
13082 | | * |
13083 | | * @param ctxt an XML parser context |
13084 | | * @param chunk a pointer to an array of chars |
13085 | | * @param size number of chars in the array |
13086 | | * @param filename an optional file name or URI |
13087 | | * @param encoding the document encoding, or NULL |
13088 | | * @returns 0 in case of success and 1 in case of error |
13089 | | */ |
13090 | | int |
13091 | | xmlCtxtResetPush(xmlParserCtxtPtr ctxt, const char *chunk, |
13092 | | int size, const char *filename, const char *encoding) |
13093 | 0 | { |
13094 | 0 | xmlParserInputPtr input; |
13095 | |
|
13096 | 0 | if (ctxt == NULL) |
13097 | 0 | return(1); |
13098 | | |
13099 | 0 | xmlCtxtReset(ctxt); |
13100 | |
|
13101 | 0 | input = xmlNewPushInput(filename, chunk, size); |
13102 | 0 | if (input == NULL) |
13103 | 0 | return(1); |
13104 | | |
13105 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
13106 | 0 | xmlFreeInputStream(input); |
13107 | 0 | return(1); |
13108 | 0 | } |
13109 | | |
13110 | 0 | if (encoding != NULL) |
13111 | 0 | xmlSwitchEncodingName(ctxt, encoding); |
13112 | |
|
13113 | 0 | return(0); |
13114 | 0 | } |
13115 | | |
13116 | | static int |
13117 | | xmlCtxtSetOptionsInternal(xmlParserCtxtPtr ctxt, int options, int keepMask) |
13118 | 21.9k | { |
13119 | 21.9k | int allMask; |
13120 | | |
13121 | 21.9k | if (ctxt == NULL) |
13122 | 0 | return(-1); |
13123 | | |
13124 | | /* |
13125 | | * XInclude options aren't handled by the parser. |
13126 | | * |
13127 | | * XML_PARSE_XINCLUDE |
13128 | | * XML_PARSE_NOXINCNODE |
13129 | | * XML_PARSE_NOBASEFIX |
13130 | | */ |
13131 | 21.9k | allMask = XML_PARSE_RECOVER | |
13132 | 21.9k | XML_PARSE_NOENT | |
13133 | 21.9k | XML_PARSE_DTDLOAD | |
13134 | 21.9k | XML_PARSE_DTDATTR | |
13135 | 21.9k | XML_PARSE_DTDVALID | |
13136 | 21.9k | XML_PARSE_NOERROR | |
13137 | 21.9k | XML_PARSE_NOWARNING | |
13138 | 21.9k | XML_PARSE_PEDANTIC | |
13139 | 21.9k | XML_PARSE_NOBLANKS | |
13140 | 21.9k | #ifdef LIBXML_SAX1_ENABLED |
13141 | 21.9k | XML_PARSE_SAX1 | |
13142 | 21.9k | #endif |
13143 | 21.9k | XML_PARSE_NONET | |
13144 | 21.9k | XML_PARSE_NODICT | |
13145 | 21.9k | XML_PARSE_NSCLEAN | |
13146 | 21.9k | XML_PARSE_NOCDATA | |
13147 | 21.9k | XML_PARSE_COMPACT | |
13148 | 21.9k | XML_PARSE_OLD10 | |
13149 | 21.9k | XML_PARSE_HUGE | |
13150 | 21.9k | XML_PARSE_OLDSAX | |
13151 | 21.9k | XML_PARSE_IGNORE_ENC | |
13152 | 21.9k | XML_PARSE_BIG_LINES | |
13153 | 21.9k | XML_PARSE_NO_XXE | |
13154 | 21.9k | XML_PARSE_UNZIP | |
13155 | 21.9k | XML_PARSE_NO_SYS_CATALOG | |
13156 | 21.9k | XML_PARSE_CATALOG_PI; |
13157 | | |
13158 | 21.9k | ctxt->options = (ctxt->options & keepMask) | (options & allMask); |
13159 | | |
13160 | | /* |
13161 | | * For some options, struct members are historically the source |
13162 | | * of truth. The values are initalized from global variables and |
13163 | | * old code could also modify them directly. Several older API |
13164 | | * functions that don't take an options argument rely on these |
13165 | | * deprecated mechanisms. |
13166 | | * |
13167 | | * Once public access to struct members and the globals are |
13168 | | * disabled, we can use the options bitmask as source of |
13169 | | * truth, making all these struct members obsolete. |
13170 | | * |
13171 | | * The XML_DETECT_IDS flags is misnamed. It simply enables |
13172 | | * loading of the external subset. |
13173 | | */ |
13174 | 21.9k | ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0; |
13175 | 21.9k | ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0; |
13176 | 21.9k | ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0; |
13177 | 21.9k | ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0; |
13178 | 21.9k | ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0; |
13179 | 21.9k | ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0; |
13180 | 21.9k | ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1; |
13181 | 21.9k | ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1; |
13182 | | |
13183 | 21.9k | if (options & XML_PARSE_HUGE) { |
13184 | 0 | if (ctxt->dict != NULL) |
13185 | 0 | xmlDictSetLimit(ctxt->dict, 0); |
13186 | 0 | } |
13187 | | |
13188 | 21.9k | ctxt->linenumbers = 1; |
13189 | | |
13190 | 21.9k | return(options & ~allMask); |
13191 | 21.9k | } |
13192 | | |
13193 | | /** |
13194 | | * Applies the options to the parser context. Unset options are |
13195 | | * cleared. |
13196 | | * |
13197 | | * @since 2.13.0 |
13198 | | * |
13199 | | * With older versions, you can use xmlCtxtUseOptions(). |
13200 | | * |
13201 | | * @param ctxt an XML parser context |
13202 | | * @param options a bitmask of xmlParserOption values |
13203 | | * @returns 0 in case of success, the set of unknown or unimplemented options |
13204 | | * in case of error. |
13205 | | */ |
13206 | | int |
13207 | | xmlCtxtSetOptions(xmlParserCtxtPtr ctxt, int options) |
13208 | 0 | { |
13209 | 0 | #ifdef LIBXML_HTML_ENABLED |
13210 | 0 | if ((ctxt != NULL) && (ctxt->html)) |
13211 | 0 | return(htmlCtxtSetOptions(ctxt, options)); |
13212 | 0 | #endif |
13213 | | |
13214 | 0 | return(xmlCtxtSetOptionsInternal(ctxt, options, 0)); |
13215 | 0 | } |
13216 | | |
13217 | | /** |
13218 | | * Get the current options of the parser context. |
13219 | | * |
13220 | | * @since 2.14.0 |
13221 | | * |
13222 | | * @param ctxt an XML parser context |
13223 | | * @returns the current options set in the parser context, or -1 if ctxt is NULL. |
13224 | | */ |
13225 | | int |
13226 | | xmlCtxtGetOptions(xmlParserCtxtPtr ctxt) |
13227 | 0 | { |
13228 | 0 | if (ctxt == NULL) |
13229 | 0 | return(-1); |
13230 | | |
13231 | 0 | return(ctxt->options); |
13232 | 0 | } |
13233 | | |
13234 | | /** |
13235 | | * Applies the options to the parser context. The following options |
13236 | | * are never cleared and can only be enabled: |
13237 | | * |
13238 | | * @deprecated Use xmlCtxtSetOptions(). |
13239 | | * |
13240 | | * - XML_PARSE_NOERROR |
13241 | | * - XML_PARSE_NOWARNING |
13242 | | * - XML_PARSE_NONET |
13243 | | * - XML_PARSE_NSCLEAN |
13244 | | * - XML_PARSE_NOCDATA |
13245 | | * - XML_PARSE_COMPACT |
13246 | | * - XML_PARSE_OLD10 |
13247 | | * - XML_PARSE_HUGE |
13248 | | * - XML_PARSE_OLDSA- X |
13249 | | * - XML_PARSE_IGNORE_ENC |
13250 | | * - XML_PARSE_BIG_LINES |
13251 | | * |
13252 | | * @param ctxt an XML parser context |
13253 | | * @param options a combination of xmlParserOption |
13254 | | * @returns 0 in case of success, the set of unknown or unimplemented options |
13255 | | * in case of error. |
13256 | | */ |
13257 | | int |
13258 | | xmlCtxtUseOptions(xmlParserCtxtPtr ctxt, int options) |
13259 | 21.9k | { |
13260 | 21.9k | int keepMask; |
13261 | | |
13262 | 21.9k | #ifdef LIBXML_HTML_ENABLED |
13263 | 21.9k | if ((ctxt != NULL) && (ctxt->html)) |
13264 | 0 | return(htmlCtxtUseOptions(ctxt, options)); |
13265 | 21.9k | #endif |
13266 | | |
13267 | | /* |
13268 | | * For historic reasons, some options can only be enabled. |
13269 | | */ |
13270 | 21.9k | keepMask = XML_PARSE_NOERROR | |
13271 | 21.9k | XML_PARSE_NOWARNING | |
13272 | 21.9k | XML_PARSE_NONET | |
13273 | 21.9k | XML_PARSE_NSCLEAN | |
13274 | 21.9k | XML_PARSE_NOCDATA | |
13275 | 21.9k | XML_PARSE_COMPACT | |
13276 | 21.9k | XML_PARSE_OLD10 | |
13277 | 21.9k | XML_PARSE_HUGE | |
13278 | 21.9k | XML_PARSE_OLDSAX | |
13279 | 21.9k | XML_PARSE_IGNORE_ENC | |
13280 | 21.9k | XML_PARSE_BIG_LINES; |
13281 | | |
13282 | 21.9k | return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask)); |
13283 | 21.9k | } |
13284 | | |
13285 | | /** |
13286 | | * To protect against exponential entity expansion ("billion laughs"), the |
13287 | | * size of serialized output is (roughly) limited to the input size |
13288 | | * multiplied by this factor. The default value is 5. |
13289 | | * |
13290 | | * When working with documents making heavy use of entity expansion, it can |
13291 | | * be necessary to increase the value. For security reasons, this should only |
13292 | | * be considered when processing trusted input. |
13293 | | * |
13294 | | * @param ctxt an XML parser context |
13295 | | * @param maxAmpl maximum amplification factor |
13296 | | */ |
13297 | | void |
13298 | | xmlCtxtSetMaxAmplification(xmlParserCtxtPtr ctxt, unsigned maxAmpl) |
13299 | 0 | { |
13300 | 0 | if (ctxt == NULL) |
13301 | 0 | return; |
13302 | 0 | ctxt->maxAmpl = maxAmpl; |
13303 | 0 | } |
13304 | | |
13305 | | /** |
13306 | | * Parse an XML document and return the resulting document tree. |
13307 | | * Takes ownership of the input object. |
13308 | | * |
13309 | | * @since 2.13.0 |
13310 | | * |
13311 | | * @param ctxt an XML parser context |
13312 | | * @param input parser input |
13313 | | * @returns the resulting document tree or NULL |
13314 | | */ |
13315 | | xmlDocPtr |
13316 | | xmlCtxtParseDocument(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) |
13317 | 21.9k | { |
13318 | 21.9k | xmlDocPtr ret = NULL; |
13319 | | |
13320 | 21.9k | if ((ctxt == NULL) || (input == NULL)) { |
13321 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
13322 | 0 | xmlFreeInputStream(input); |
13323 | 0 | return(NULL); |
13324 | 0 | } |
13325 | | |
13326 | | /* assert(ctxt->inputNr == 0); */ |
13327 | 21.9k | while (ctxt->inputNr > 0) |
13328 | 0 | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
13329 | | |
13330 | 21.9k | if (xmlCtxtPushInput(ctxt, input) < 0) { |
13331 | 0 | xmlFreeInputStream(input); |
13332 | 0 | return(NULL); |
13333 | 0 | } |
13334 | | |
13335 | 21.9k | xmlParseDocument(ctxt); |
13336 | | |
13337 | 21.9k | ret = xmlCtxtGetDocument(ctxt); |
13338 | | |
13339 | | /* assert(ctxt->inputNr == 1); */ |
13340 | 43.9k | while (ctxt->inputNr > 0) |
13341 | 21.9k | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
13342 | | |
13343 | 21.9k | return(ret); |
13344 | 21.9k | } |
13345 | | |
13346 | | /** |
13347 | | * Convenience function to parse an XML document from a |
13348 | | * zero-terminated string. |
13349 | | * |
13350 | | * See xmlCtxtReadDoc() for details. |
13351 | | * |
13352 | | * @param cur a pointer to a zero terminated string |
13353 | | * @param URL base URL (optional) |
13354 | | * @param encoding the document encoding (optional) |
13355 | | * @param options a combination of xmlParserOption |
13356 | | * @returns the resulting document tree |
13357 | | */ |
13358 | | xmlDocPtr |
13359 | | xmlReadDoc(const xmlChar *cur, const char *URL, const char *encoding, |
13360 | | int options) |
13361 | 0 | { |
13362 | 0 | xmlParserCtxtPtr ctxt; |
13363 | 0 | xmlParserInputPtr input; |
13364 | 0 | xmlDocPtr doc = NULL; |
13365 | |
|
13366 | 0 | ctxt = xmlNewParserCtxt(); |
13367 | 0 | if (ctxt == NULL) |
13368 | 0 | return(NULL); |
13369 | | |
13370 | 0 | xmlCtxtUseOptions(ctxt, options); |
13371 | |
|
13372 | 0 | input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding, |
13373 | 0 | XML_INPUT_BUF_STATIC); |
13374 | |
|
13375 | 0 | if (input != NULL) |
13376 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13377 | |
|
13378 | 0 | xmlFreeParserCtxt(ctxt); |
13379 | 0 | return(doc); |
13380 | 0 | } |
13381 | | |
13382 | | /** |
13383 | | * Convenience function to parse an XML file from the filesystem, |
13384 | | * the network or a global user-define resource loader. |
13385 | | * |
13386 | | * This function always enables the XML_PARSE_UNZIP option for |
13387 | | * backward compatibility. If a "-" filename is passed, it will |
13388 | | * read from stdin. Both of these features are potentially |
13389 | | * insecure and might be removed from later versions. |
13390 | | * |
13391 | | * See xmlCtxtReadFile() for details. |
13392 | | * |
13393 | | * @param filename a file or URL |
13394 | | * @param encoding the document encoding (optional) |
13395 | | * @param options a combination of xmlParserOption |
13396 | | * @returns the resulting document tree |
13397 | | */ |
13398 | | xmlDocPtr |
13399 | | xmlReadFile(const char *filename, const char *encoding, int options) |
13400 | 0 | { |
13401 | 0 | xmlParserCtxtPtr ctxt; |
13402 | 0 | xmlParserInputPtr input; |
13403 | 0 | xmlDocPtr doc = NULL; |
13404 | |
|
13405 | 0 | ctxt = xmlNewParserCtxt(); |
13406 | 0 | if (ctxt == NULL) |
13407 | 0 | return(NULL); |
13408 | | |
13409 | 0 | options |= XML_PARSE_UNZIP; |
13410 | |
|
13411 | 0 | xmlCtxtUseOptions(ctxt, options); |
13412 | | |
13413 | | /* |
13414 | | * Backward compatibility for users of command line utilities like |
13415 | | * xmlstarlet expecting "-" to mean stdin. This is dangerous and |
13416 | | * should be removed at some point. |
13417 | | */ |
13418 | 0 | if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0)) |
13419 | 0 | input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, |
13420 | 0 | encoding, 0); |
13421 | 0 | else |
13422 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0); |
13423 | |
|
13424 | 0 | if (input != NULL) |
13425 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13426 | |
|
13427 | 0 | xmlFreeParserCtxt(ctxt); |
13428 | 0 | return(doc); |
13429 | 0 | } |
13430 | | |
13431 | | /** |
13432 | | * Parse an XML in-memory document and build a tree. The input buffer must |
13433 | | * not contain a terminating null byte. |
13434 | | * |
13435 | | * See xmlCtxtReadMemory() for details. |
13436 | | * |
13437 | | * @param buffer a pointer to a char array |
13438 | | * @param size the size of the array |
13439 | | * @param url base URL (optional) |
13440 | | * @param encoding the document encoding (optional) |
13441 | | * @param options a combination of xmlParserOption |
13442 | | * @returns the resulting document tree |
13443 | | */ |
13444 | | xmlDocPtr |
13445 | | xmlReadMemory(const char *buffer, int size, const char *url, |
13446 | | const char *encoding, int options) |
13447 | 21.9k | { |
13448 | 21.9k | xmlParserCtxtPtr ctxt; |
13449 | 21.9k | xmlParserInputPtr input; |
13450 | 21.9k | xmlDocPtr doc = NULL; |
13451 | | |
13452 | 21.9k | if (size < 0) |
13453 | 0 | return(NULL); |
13454 | | |
13455 | 21.9k | ctxt = xmlNewParserCtxt(); |
13456 | 21.9k | if (ctxt == NULL) |
13457 | 0 | return(NULL); |
13458 | | |
13459 | 21.9k | xmlCtxtUseOptions(ctxt, options); |
13460 | | |
13461 | 21.9k | input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding, |
13462 | 21.9k | XML_INPUT_BUF_STATIC); |
13463 | | |
13464 | 21.9k | if (input != NULL) |
13465 | 21.9k | doc = xmlCtxtParseDocument(ctxt, input); |
13466 | | |
13467 | 21.9k | xmlFreeParserCtxt(ctxt); |
13468 | 21.9k | return(doc); |
13469 | 21.9k | } |
13470 | | |
13471 | | /** |
13472 | | * Parse an XML from a file descriptor and build a tree. |
13473 | | * |
13474 | | * See xmlCtxtReadFd() for details. |
13475 | | * |
13476 | | * NOTE that the file descriptor will not be closed when the |
13477 | | * context is freed or reset. |
13478 | | * |
13479 | | * @param fd an open file descriptor |
13480 | | * @param URL base URL (optional) |
13481 | | * @param encoding the document encoding (optional) |
13482 | | * @param options a combination of xmlParserOption |
13483 | | * @returns the resulting document tree |
13484 | | */ |
13485 | | xmlDocPtr |
13486 | | xmlReadFd(int fd, const char *URL, const char *encoding, int options) |
13487 | 0 | { |
13488 | 0 | xmlParserCtxtPtr ctxt; |
13489 | 0 | xmlParserInputPtr input; |
13490 | 0 | xmlDocPtr doc = NULL; |
13491 | |
|
13492 | 0 | ctxt = xmlNewParserCtxt(); |
13493 | 0 | if (ctxt == NULL) |
13494 | 0 | return(NULL); |
13495 | | |
13496 | 0 | xmlCtxtUseOptions(ctxt, options); |
13497 | |
|
13498 | 0 | input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0); |
13499 | |
|
13500 | 0 | if (input != NULL) |
13501 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13502 | |
|
13503 | 0 | xmlFreeParserCtxt(ctxt); |
13504 | 0 | return(doc); |
13505 | 0 | } |
13506 | | |
13507 | | /** |
13508 | | * Parse an XML document from I/O functions and context and build a tree. |
13509 | | * |
13510 | | * See xmlCtxtReadIO() for details. |
13511 | | * |
13512 | | * @param ioread an I/O read function |
13513 | | * @param ioclose an I/O close function (optional) |
13514 | | * @param ioctx an I/O handler |
13515 | | * @param URL base URL (optional) |
13516 | | * @param encoding the document encoding (optional) |
13517 | | * @param options a combination of xmlParserOption |
13518 | | * @returns the resulting document tree |
13519 | | */ |
13520 | | xmlDocPtr |
13521 | | xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
13522 | | void *ioctx, const char *URL, const char *encoding, int options) |
13523 | 0 | { |
13524 | 0 | xmlParserCtxtPtr ctxt; |
13525 | 0 | xmlParserInputPtr input; |
13526 | 0 | xmlDocPtr doc = NULL; |
13527 | |
|
13528 | 0 | ctxt = xmlNewParserCtxt(); |
13529 | 0 | if (ctxt == NULL) |
13530 | 0 | return(NULL); |
13531 | | |
13532 | 0 | xmlCtxtUseOptions(ctxt, options); |
13533 | |
|
13534 | 0 | input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx, |
13535 | 0 | encoding, 0); |
13536 | |
|
13537 | 0 | if (input != NULL) |
13538 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13539 | |
|
13540 | 0 | xmlFreeParserCtxt(ctxt); |
13541 | 0 | return(doc); |
13542 | 0 | } |
13543 | | |
13544 | | /** |
13545 | | * Parse an XML in-memory document and build a tree. |
13546 | | * |
13547 | | * `URL` is used as base to resolve external entities and for error |
13548 | | * reporting. |
13549 | | * |
13550 | | * See xmlCtxtUseOptions() for details. |
13551 | | * |
13552 | | * @param ctxt an XML parser context |
13553 | | * @param str a pointer to a zero terminated string |
13554 | | * @param URL base URL (optional) |
13555 | | * @param encoding the document encoding (optional) |
13556 | | * @param options a combination of xmlParserOption |
13557 | | * @returns the resulting document tree |
13558 | | */ |
13559 | | xmlDocPtr |
13560 | | xmlCtxtReadDoc(xmlParserCtxtPtr ctxt, const xmlChar *str, |
13561 | | const char *URL, const char *encoding, int options) |
13562 | 0 | { |
13563 | 0 | xmlParserInputPtr input; |
13564 | |
|
13565 | 0 | if (ctxt == NULL) |
13566 | 0 | return(NULL); |
13567 | | |
13568 | 0 | xmlCtxtReset(ctxt); |
13569 | 0 | xmlCtxtUseOptions(ctxt, options); |
13570 | |
|
13571 | 0 | input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) str, encoding, |
13572 | 0 | XML_INPUT_BUF_STATIC); |
13573 | 0 | if (input == NULL) |
13574 | 0 | return(NULL); |
13575 | | |
13576 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13577 | 0 | } |
13578 | | |
13579 | | /** |
13580 | | * Parse an XML file from the filesystem, the network or a user-defined |
13581 | | * resource loader. |
13582 | | * |
13583 | | * This function always enables the XML_PARSE_UNZIP option for |
13584 | | * backward compatibility. This feature is potentially insecure |
13585 | | * and might be removed from later versions. |
13586 | | * |
13587 | | * @param ctxt an XML parser context |
13588 | | * @param filename a file or URL |
13589 | | * @param encoding the document encoding (optional) |
13590 | | * @param options a combination of xmlParserOption |
13591 | | * @returns the resulting document tree |
13592 | | */ |
13593 | | xmlDocPtr |
13594 | | xmlCtxtReadFile(xmlParserCtxtPtr ctxt, const char *filename, |
13595 | | const char *encoding, int options) |
13596 | 0 | { |
13597 | 0 | xmlParserInputPtr input; |
13598 | |
|
13599 | 0 | if (ctxt == NULL) |
13600 | 0 | return(NULL); |
13601 | | |
13602 | 0 | options |= XML_PARSE_UNZIP; |
13603 | |
|
13604 | 0 | xmlCtxtReset(ctxt); |
13605 | 0 | xmlCtxtUseOptions(ctxt, options); |
13606 | |
|
13607 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0); |
13608 | 0 | if (input == NULL) |
13609 | 0 | return(NULL); |
13610 | | |
13611 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13612 | 0 | } |
13613 | | |
13614 | | /** |
13615 | | * Parse an XML in-memory document and build a tree. The input buffer must |
13616 | | * not contain a terminating null byte. |
13617 | | * |
13618 | | * `URL` is used as base to resolve external entities and for error |
13619 | | * reporting. |
13620 | | * |
13621 | | * See xmlCtxtUseOptions() for details. |
13622 | | * |
13623 | | * @param ctxt an XML parser context |
13624 | | * @param buffer a pointer to a char array |
13625 | | * @param size the size of the array |
13626 | | * @param URL base URL (optional) |
13627 | | * @param encoding the document encoding (optional) |
13628 | | * @param options a combination of xmlParserOption |
13629 | | * @returns the resulting document tree |
13630 | | */ |
13631 | | xmlDocPtr |
13632 | | xmlCtxtReadMemory(xmlParserCtxtPtr ctxt, const char *buffer, int size, |
13633 | | const char *URL, const char *encoding, int options) |
13634 | 0 | { |
13635 | 0 | xmlParserInputPtr input; |
13636 | |
|
13637 | 0 | if ((ctxt == NULL) || (size < 0)) |
13638 | 0 | return(NULL); |
13639 | | |
13640 | 0 | xmlCtxtReset(ctxt); |
13641 | 0 | xmlCtxtUseOptions(ctxt, options); |
13642 | |
|
13643 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding, |
13644 | 0 | XML_INPUT_BUF_STATIC); |
13645 | 0 | if (input == NULL) |
13646 | 0 | return(NULL); |
13647 | | |
13648 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13649 | 0 | } |
13650 | | |
13651 | | /** |
13652 | | * Parse an XML document from a file descriptor and build a tree. |
13653 | | * |
13654 | | * NOTE that the file descriptor will not be closed when the |
13655 | | * context is freed or reset. |
13656 | | * |
13657 | | * `URL` is used as base to resolve external entities and for error |
13658 | | * reporting. |
13659 | | * |
13660 | | * See xmlCtxtUseOptions() for details. |
13661 | | * |
13662 | | * @param ctxt an XML parser context |
13663 | | * @param fd an open file descriptor |
13664 | | * @param URL base URL (optional) |
13665 | | * @param encoding the document encoding (optional) |
13666 | | * @param options a combination of xmlParserOption |
13667 | | * @returns the resulting document tree |
13668 | | */ |
13669 | | xmlDocPtr |
13670 | | xmlCtxtReadFd(xmlParserCtxtPtr ctxt, int fd, |
13671 | | const char *URL, const char *encoding, int options) |
13672 | 0 | { |
13673 | 0 | xmlParserInputPtr input; |
13674 | |
|
13675 | 0 | if (ctxt == NULL) |
13676 | 0 | return(NULL); |
13677 | | |
13678 | 0 | xmlCtxtReset(ctxt); |
13679 | 0 | xmlCtxtUseOptions(ctxt, options); |
13680 | |
|
13681 | 0 | input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0); |
13682 | 0 | if (input == NULL) |
13683 | 0 | return(NULL); |
13684 | | |
13685 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13686 | 0 | } |
13687 | | |
13688 | | /** |
13689 | | * parse an XML document from I/O functions and source and build a tree. |
13690 | | * This reuses the existing `ctxt` parser context |
13691 | | * |
13692 | | * `URL` is used as base to resolve external entities and for error |
13693 | | * reporting. |
13694 | | * |
13695 | | * See xmlCtxtUseOptions() for details. |
13696 | | * |
13697 | | * @param ctxt an XML parser context |
13698 | | * @param ioread an I/O read function |
13699 | | * @param ioclose an I/O close function |
13700 | | * @param ioctx an I/O handler |
13701 | | * @param URL the base URL to use for the document |
13702 | | * @param encoding the document encoding, or NULL |
13703 | | * @param options a combination of xmlParserOption |
13704 | | * @returns the resulting document tree |
13705 | | */ |
13706 | | xmlDocPtr |
13707 | | xmlCtxtReadIO(xmlParserCtxtPtr ctxt, xmlInputReadCallback ioread, |
13708 | | xmlInputCloseCallback ioclose, void *ioctx, |
13709 | | const char *URL, |
13710 | | const char *encoding, int options) |
13711 | 0 | { |
13712 | 0 | xmlParserInputPtr input; |
13713 | |
|
13714 | 0 | if (ctxt == NULL) |
13715 | 0 | return(NULL); |
13716 | | |
13717 | 0 | xmlCtxtReset(ctxt); |
13718 | 0 | xmlCtxtUseOptions(ctxt, options); |
13719 | |
|
13720 | 0 | input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx, |
13721 | 0 | encoding, 0); |
13722 | 0 | if (input == NULL) |
13723 | 0 | return(NULL); |
13724 | | |
13725 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13726 | 0 | } |
13727 | | |