Line | Count | Source (jump to first uncovered line) |
1 | | /* |
2 | | * SAX2.c : Default SAX2 handler to build a tree. |
3 | | * |
4 | | * See Copyright for the status of this software. |
5 | | * |
6 | | * Author: Daniel Veillard |
7 | | */ |
8 | | |
9 | | |
10 | | #define IN_LIBXML |
11 | | #include "libxml.h" |
12 | | #include <stdlib.h> |
13 | | #include <string.h> |
14 | | #include <limits.h> |
15 | | #include <stddef.h> |
16 | | #include <libxml/SAX2.h> |
17 | | #include <libxml/xmlmemory.h> |
18 | | #include <libxml/tree.h> |
19 | | #include <libxml/parser.h> |
20 | | #include <libxml/parserInternals.h> |
21 | | #include <libxml/valid.h> |
22 | | #include <libxml/entities.h> |
23 | | #include <libxml/xmlerror.h> |
24 | | #include <libxml/xmlIO.h> |
25 | | #include <libxml/uri.h> |
26 | | #include <libxml/valid.h> |
27 | | #include <libxml/HTMLtree.h> |
28 | | |
29 | | #include "private/error.h" |
30 | | #include "private/parser.h" |
31 | | #include "private/tree.h" |
32 | | |
33 | | #ifndef SIZE_MAX |
34 | | #define SIZE_MAX ((size_t) -1) |
35 | | #endif |
36 | | |
37 | | /* |
38 | | * @param ctxt an XML validation parser context |
39 | | * @param msg a string to accompany the error message |
40 | | */ |
41 | | static void |
42 | 427 | xmlSAX2ErrMemory(xmlParserCtxtPtr ctxt) { |
43 | 427 | xmlCtxtErrMemory(ctxt); |
44 | 427 | } |
45 | | |
46 | | #ifdef LIBXML_VALID_ENABLED |
47 | | /** |
48 | | * Handle a validation error |
49 | | * |
50 | | * @param ctxt an XML validation parser context |
51 | | * @param error the error number |
52 | | * @param msg the error message |
53 | | * @param str1 extra data |
54 | | * @param str2 extra data |
55 | | */ |
56 | | static void LIBXML_ATTR_FORMAT(3,0) |
57 | | xmlErrValid(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
58 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
59 | 0 | { |
60 | 0 | xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR, |
61 | 0 | str1, str2, NULL, 0, msg, str1, str2); |
62 | 0 | if (ctxt != NULL) |
63 | 0 | ctxt->valid = 0; |
64 | 0 | } |
65 | | #endif /* LIBXML_VALID_ENABLED */ |
66 | | |
67 | | /** |
68 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
69 | | * |
70 | | * @param ctxt an XML parser context |
71 | | * @param error the error number |
72 | | * @param msg the error message |
73 | | * @param str1 an error string |
74 | | * @param str2 an error string |
75 | | */ |
76 | | static void LIBXML_ATTR_FORMAT(3,0) |
77 | | xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
78 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
79 | 281 | { |
80 | 281 | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
81 | 281 | str1, str2, NULL, 0, msg, str1, str2); |
82 | 281 | } |
83 | | |
84 | | /** |
85 | | * Handle an xml:id error |
86 | | * |
87 | | * @param ctxt an XML validation parser context |
88 | | * @param error the error number |
89 | | * @param msg the error message |
90 | | * @param str1 extra data |
91 | | */ |
92 | | static void LIBXML_ATTR_FORMAT(3,0) |
93 | | xmlErrId(xmlParserCtxtPtr ctxt, xmlParserErrors error, const char *msg, |
94 | | const xmlChar *str1) |
95 | 2.48k | { |
96 | 2.48k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR, |
97 | 2.48k | str1, NULL, NULL, 0, msg, str1); |
98 | 2.48k | } |
99 | | |
100 | | /** |
101 | | * Handle a parser warning |
102 | | * |
103 | | * @param ctxt an XML parser context |
104 | | * @param error the error number |
105 | | * @param msg the error message |
106 | | * @param str1 an error string |
107 | | */ |
108 | | static void LIBXML_ATTR_FORMAT(3,0) |
109 | | xmlWarnMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
110 | | const char *msg, const xmlChar *str1) |
111 | 41.9k | { |
112 | 41.9k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING, |
113 | 41.9k | str1, NULL, NULL, 0, msg, str1); |
114 | 41.9k | } |
115 | | |
116 | | /** |
117 | | * Handle a namespace warning |
118 | | * |
119 | | * @param ctxt an XML parser context |
120 | | * @param error the error number |
121 | | * @param msg the error message |
122 | | * @param str1 an error string |
123 | | * @param str2 an error string |
124 | | */ |
125 | | static void LIBXML_ATTR_FORMAT(3,0) |
126 | | xmlNsWarnMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
127 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
128 | 24 | { |
129 | 24 | xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING, |
130 | 24 | str1, str2, NULL, 0, msg, str1, str2); |
131 | 24 | } |
132 | | |
133 | | /** |
134 | | * Provides the public ID e.g. "-//SGMLSOURCE//DTD DEMO//EN" |
135 | | * |
136 | | * @param ctx the user data (XML parser context) |
137 | | * @returns a xmlChar * |
138 | | */ |
139 | | const xmlChar * |
140 | | xmlSAX2GetPublicId(void *ctx ATTRIBUTE_UNUSED) |
141 | 0 | { |
142 | | /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */ |
143 | 0 | return(NULL); |
144 | 0 | } |
145 | | |
146 | | /** |
147 | | * Provides the system ID, basically URL or filename e.g. |
148 | | * http://www.sgmlsource.com/dtds/memo.dtd |
149 | | * |
150 | | * @param ctx the user data (XML parser context) |
151 | | * @returns a xmlChar * |
152 | | */ |
153 | | const xmlChar * |
154 | | xmlSAX2GetSystemId(void *ctx) |
155 | 0 | { |
156 | 0 | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
157 | 0 | if ((ctx == NULL) || (ctxt->input == NULL)) return(NULL); |
158 | 0 | return((const xmlChar *) ctxt->input->filename); |
159 | 0 | } |
160 | | |
161 | | /** |
162 | | * Provide the line number of the current parsing point. |
163 | | * |
164 | | * @param ctx the user data (XML parser context) |
165 | | * @returns an int |
166 | | */ |
167 | | int |
168 | | xmlSAX2GetLineNumber(void *ctx) |
169 | 0 | { |
170 | 0 | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
171 | 0 | if ((ctx == NULL) || (ctxt->input == NULL)) return(0); |
172 | 0 | return(ctxt->input->line); |
173 | 0 | } |
174 | | |
175 | | /** |
176 | | * Provide the column number of the current parsing point. |
177 | | * |
178 | | * @param ctx the user data (XML parser context) |
179 | | * @returns an int |
180 | | */ |
181 | | int |
182 | | xmlSAX2GetColumnNumber(void *ctx) |
183 | 0 | { |
184 | 0 | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
185 | 0 | if ((ctx == NULL) || (ctxt->input == NULL)) return(0); |
186 | 0 | return(ctxt->input->col); |
187 | 0 | } |
188 | | |
189 | | /** |
190 | | * Is this document tagged standalone ? |
191 | | * |
192 | | * @param ctx the user data (XML parser context) |
193 | | * @returns 1 if true |
194 | | */ |
195 | | int |
196 | | xmlSAX2IsStandalone(void *ctx) |
197 | 0 | { |
198 | 0 | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
199 | 0 | if ((ctx == NULL) || (ctxt->myDoc == NULL)) return(0); |
200 | 0 | return(ctxt->myDoc->standalone == 1); |
201 | 0 | } |
202 | | |
203 | | /** |
204 | | * Does this document has an internal subset |
205 | | * |
206 | | * @param ctx the user data (XML parser context) |
207 | | * @returns 1 if true |
208 | | */ |
209 | | int |
210 | | xmlSAX2HasInternalSubset(void *ctx) |
211 | 0 | { |
212 | 0 | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
213 | 0 | if ((ctxt == NULL) || (ctxt->myDoc == NULL)) return(0); |
214 | 0 | return(ctxt->myDoc->intSubset != NULL); |
215 | 0 | } |
216 | | |
217 | | /** |
218 | | * Does this document has an external subset |
219 | | * |
220 | | * @param ctx the user data (XML parser context) |
221 | | * @returns 1 if true |
222 | | */ |
223 | | int |
224 | | xmlSAX2HasExternalSubset(void *ctx) |
225 | 0 | { |
226 | 0 | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
227 | 0 | if ((ctxt == NULL) || (ctxt->myDoc == NULL)) return(0); |
228 | 0 | return(ctxt->myDoc->extSubset != NULL); |
229 | 0 | } |
230 | | |
231 | | /** |
232 | | * Callback on internal subset declaration. |
233 | | * |
234 | | * @param ctx the user data (XML parser context) |
235 | | * @param name the root element name |
236 | | * @param publicId public identifier of the DTD (optional) |
237 | | * @param systemId system identifier (URL) of the DTD |
238 | | */ |
239 | | void |
240 | | xmlSAX2InternalSubset(void *ctx, const xmlChar *name, |
241 | | const xmlChar *publicId, const xmlChar *systemId) |
242 | 12.7k | { |
243 | 12.7k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
244 | 12.7k | xmlDtdPtr dtd; |
245 | 12.7k | if (ctx == NULL) return; |
246 | | |
247 | 12.7k | if (ctxt->myDoc == NULL) |
248 | 0 | return; |
249 | 12.7k | if ((ctxt->html) && (ctxt->instate != XML_PARSER_MISC)) |
250 | 0 | return; |
251 | 12.7k | dtd = xmlGetIntSubset(ctxt->myDoc); |
252 | 12.7k | if (dtd != NULL) { |
253 | 0 | xmlUnlinkNode((xmlNodePtr) dtd); |
254 | 0 | xmlFreeDtd(dtd); |
255 | 0 | ctxt->myDoc->intSubset = NULL; |
256 | 0 | } |
257 | 12.7k | ctxt->myDoc->intSubset = |
258 | 12.7k | xmlCreateIntSubset(ctxt->myDoc, name, publicId, systemId); |
259 | 12.7k | if (ctxt->myDoc->intSubset == NULL) |
260 | 6 | xmlSAX2ErrMemory(ctxt); |
261 | 12.7k | } |
262 | | |
263 | | /** |
264 | | * Callback on external subset declaration. |
265 | | * |
266 | | * @param ctx the user data (XML parser context) |
267 | | * @param name the root element name |
268 | | * @param publicId public identifier of the DTD (optional) |
269 | | * @param systemId system identifier (URL) of the DTD |
270 | | */ |
271 | | void |
272 | | xmlSAX2ExternalSubset(void *ctx, const xmlChar *name, |
273 | | const xmlChar *publicId, const xmlChar *systemId) |
274 | 11.5k | { |
275 | 11.5k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
276 | 11.5k | if (ctx == NULL) return; |
277 | 11.5k | if ((systemId != NULL) && |
278 | 11.5k | ((ctxt->options & XML_PARSE_NO_XXE) == 0) && |
279 | 11.5k | (((ctxt->validate) || (ctxt->loadsubset & ~XML_SKIP_IDS)) && |
280 | 2.87k | (ctxt->wellFormed && ctxt->myDoc))) { |
281 | | /* |
282 | | * Try to fetch and parse the external subset. |
283 | | */ |
284 | 2.76k | xmlParserInputPtr oldinput; |
285 | 2.76k | int oldinputNr; |
286 | 2.76k | int oldinputMax; |
287 | 2.76k | xmlParserInputPtr *oldinputTab; |
288 | 2.76k | xmlParserInputPtr input = NULL; |
289 | 2.76k | xmlChar *oldencoding; |
290 | 2.76k | unsigned long consumed; |
291 | 2.76k | size_t buffered; |
292 | 2.76k | #ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION |
293 | 2.76k | int inputMax = 1; |
294 | | #else |
295 | | int inputMax = 5; |
296 | | #endif |
297 | | |
298 | | /* |
299 | | * Ask the Entity resolver to load the damn thing |
300 | | */ |
301 | 2.76k | if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) |
302 | 2.76k | input = ctxt->sax->resolveEntity(ctxt->userData, publicId, |
303 | 2.76k | systemId); |
304 | 2.76k | if (input == NULL) { |
305 | 727 | return; |
306 | 727 | } |
307 | | |
308 | 2.03k | if (xmlNewDtd(ctxt->myDoc, name, publicId, systemId) == NULL) { |
309 | 2 | xmlSAX2ErrMemory(ctxt); |
310 | 2 | xmlFreeInputStream(input); |
311 | 2 | return; |
312 | 2 | } |
313 | | |
314 | | /* |
315 | | * make sure we won't destroy the main document context |
316 | | */ |
317 | 2.03k | oldinput = ctxt->input; |
318 | 2.03k | oldinputNr = ctxt->inputNr; |
319 | 2.03k | oldinputMax = ctxt->inputMax; |
320 | 2.03k | oldinputTab = ctxt->inputTab; |
321 | 2.03k | oldencoding = ctxt->encoding; |
322 | 2.03k | ctxt->encoding = NULL; |
323 | | |
324 | 2.03k | ctxt->inputTab = xmlMalloc(inputMax * sizeof(xmlParserInputPtr)); |
325 | 2.03k | if (ctxt->inputTab == NULL) { |
326 | 2 | xmlSAX2ErrMemory(ctxt); |
327 | 2 | goto error; |
328 | 2 | } |
329 | 2.03k | ctxt->inputNr = 0; |
330 | 2.03k | ctxt->inputMax = inputMax; |
331 | 2.03k | ctxt->input = NULL; |
332 | 2.03k | if (xmlCtxtPushInput(ctxt, input) < 0) |
333 | 1 | goto error; |
334 | | |
335 | 2.02k | if (input->filename == NULL) |
336 | 0 | input->filename = (char *) xmlCanonicPath(systemId); |
337 | 2.02k | input->line = 1; |
338 | 2.02k | input->col = 1; |
339 | 2.02k | input->base = ctxt->input->cur; |
340 | 2.02k | input->cur = ctxt->input->cur; |
341 | 2.02k | input->free = NULL; |
342 | | |
343 | | /* |
344 | | * let's parse that entity knowing it's an external subset. |
345 | | */ |
346 | 2.02k | xmlParseExternalSubset(ctxt, publicId, systemId); |
347 | | |
348 | | /* |
349 | | * Free up the external entities |
350 | | */ |
351 | | |
352 | 2.10k | while (ctxt->inputNr > 1) |
353 | 76 | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
354 | | |
355 | 2.02k | consumed = ctxt->input->consumed; |
356 | 2.02k | buffered = ctxt->input->cur - ctxt->input->base; |
357 | 2.02k | if (buffered > ULONG_MAX - consumed) |
358 | 0 | consumed = ULONG_MAX; |
359 | 2.02k | else |
360 | 2.02k | consumed += buffered; |
361 | 2.02k | if (consumed > ULONG_MAX - ctxt->sizeentities) |
362 | 0 | ctxt->sizeentities = ULONG_MAX; |
363 | 2.02k | else |
364 | 2.02k | ctxt->sizeentities += consumed; |
365 | | |
366 | 2.03k | error: |
367 | 2.03k | xmlFreeInputStream(input); |
368 | 2.03k | xmlFree(ctxt->inputTab); |
369 | | |
370 | | /* |
371 | | * Restore the parsing context of the main entity |
372 | | */ |
373 | 2.03k | ctxt->input = oldinput; |
374 | 2.03k | ctxt->inputNr = oldinputNr; |
375 | 2.03k | ctxt->inputMax = oldinputMax; |
376 | 2.03k | ctxt->inputTab = oldinputTab; |
377 | 2.03k | if (ctxt->encoding != NULL) |
378 | 21 | xmlFree(ctxt->encoding); |
379 | 2.03k | ctxt->encoding = oldencoding; |
380 | | /* ctxt->wellFormed = oldwellFormed; */ |
381 | 2.03k | } |
382 | 11.5k | } |
383 | | |
384 | | /** |
385 | | * This is only used to load DTDs. The preferred way to install |
386 | | * custom resolvers is #xmlCtxtSetResourceLoader. |
387 | | * |
388 | | * @param ctx the user data (XML parser context) |
389 | | * @param publicId The public ID of the entity |
390 | | * @param systemId The system ID (URL) of the entity |
391 | | * @returns a parser input. |
392 | | */ |
393 | | xmlParserInput * |
394 | | xmlSAX2ResolveEntity(void *ctx, const xmlChar *publicId, |
395 | | const xmlChar *systemId) |
396 | 2.76k | { |
397 | 2.76k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
398 | 2.76k | xmlParserInputPtr ret = NULL; |
399 | 2.76k | xmlChar *URI = NULL; |
400 | | |
401 | 2.76k | if (ctx == NULL) return(NULL); |
402 | | |
403 | 2.76k | if (systemId != NULL) { |
404 | 2.76k | const xmlChar *base = NULL; |
405 | 2.76k | int res; |
406 | | |
407 | 2.76k | if (ctxt->input != NULL) |
408 | 2.76k | base = BAD_CAST ctxt->input->filename; |
409 | | |
410 | | /* |
411 | | * We don't really need the 'directory' struct member, but some |
412 | | * users set it manually to a base URI for memory streams. |
413 | | */ |
414 | 2.76k | if (base == NULL) |
415 | 0 | base = BAD_CAST ctxt->directory; |
416 | | |
417 | 2.76k | if ((xmlStrlen(systemId) > XML_MAX_URI_LENGTH) || |
418 | 2.76k | (xmlStrlen(base) > XML_MAX_URI_LENGTH)) { |
419 | 1 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, "URI too long"); |
420 | 1 | return(NULL); |
421 | 1 | } |
422 | 2.76k | res = xmlBuildURISafe(systemId, base, &URI); |
423 | 2.76k | if (URI == NULL) { |
424 | 62 | if (res < 0) |
425 | 36 | xmlSAX2ErrMemory(ctxt); |
426 | 26 | else |
427 | 26 | xmlWarnMsg(ctxt, XML_ERR_INVALID_URI, |
428 | 26 | "Can't resolve URI: %s\n", systemId); |
429 | 62 | return(NULL); |
430 | 62 | } |
431 | 2.69k | if (xmlStrlen(URI) > XML_MAX_URI_LENGTH) { |
432 | 3 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, "URI too long"); |
433 | 3 | xmlFree(URI); |
434 | 3 | return(NULL); |
435 | 3 | } |
436 | 2.69k | } |
437 | | |
438 | 2.69k | ret = xmlLoadResource(ctxt, (const char *) URI, |
439 | 2.69k | (const char *) publicId, XML_RESOURCE_DTD); |
440 | | |
441 | 2.69k | xmlFree(URI); |
442 | 2.69k | return(ret); |
443 | 2.76k | } |
444 | | |
445 | | /** |
446 | | * Get an entity by name |
447 | | * |
448 | | * @param ctx the user data (XML parser context) |
449 | | * @param name The entity name |
450 | | * @returns the xmlEntity if found. |
451 | | */ |
452 | | xmlEntity * |
453 | | xmlSAX2GetEntity(void *ctx, const xmlChar *name) |
454 | 1.09M | { |
455 | 1.09M | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
456 | 1.09M | xmlEntityPtr ret = NULL; |
457 | | |
458 | 1.09M | if (ctx == NULL) return(NULL); |
459 | | |
460 | 1.09M | if (ctxt->inSubset == 0) { |
461 | 983k | ret = xmlGetPredefinedEntity(name); |
462 | 983k | if (ret != NULL) |
463 | 100k | return(ret); |
464 | 983k | } |
465 | 991k | if ((ctxt->myDoc != NULL) && (ctxt->myDoc->standalone == 1)) { |
466 | 863 | if (ctxt->inSubset == 2) { |
467 | 76 | ctxt->myDoc->standalone = 0; |
468 | 76 | ret = xmlGetDocEntity(ctxt->myDoc, name); |
469 | 76 | ctxt->myDoc->standalone = 1; |
470 | 787 | } else { |
471 | 787 | ret = xmlGetDocEntity(ctxt->myDoc, name); |
472 | 787 | if (ret == NULL) { |
473 | 589 | ctxt->myDoc->standalone = 0; |
474 | 589 | ret = xmlGetDocEntity(ctxt->myDoc, name); |
475 | 589 | if (ret != NULL) { |
476 | 198 | xmlFatalErrMsg(ctxt, XML_ERR_NOT_STANDALONE, |
477 | 198 | "Entity(%s) document marked standalone but requires external subset\n", |
478 | 198 | name, NULL); |
479 | 198 | } |
480 | 589 | ctxt->myDoc->standalone = 1; |
481 | 589 | } |
482 | 787 | } |
483 | 990k | } else { |
484 | 990k | ret = xmlGetDocEntity(ctxt->myDoc, name); |
485 | 990k | } |
486 | 991k | return(ret); |
487 | 1.09M | } |
488 | | |
489 | | /** |
490 | | * Get a parameter entity by name |
491 | | * |
492 | | * @param ctx the user data (XML parser context) |
493 | | * @param name The entity name |
494 | | * @returns the xmlEntity if found. |
495 | | */ |
496 | | xmlEntity * |
497 | | xmlSAX2GetParameterEntity(void *ctx, const xmlChar *name) |
498 | 233k | { |
499 | 233k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
500 | 233k | xmlEntityPtr ret; |
501 | | |
502 | 233k | if (ctx == NULL) return(NULL); |
503 | | |
504 | 233k | ret = xmlGetParameterEntity(ctxt->myDoc, name); |
505 | 233k | return(ret); |
506 | 233k | } |
507 | | |
508 | | |
509 | | /** |
510 | | * An entity definition has been parsed |
511 | | * |
512 | | * @param ctx the user data (XML parser context) |
513 | | * @param name the entity name |
514 | | * @param type the entity type |
515 | | * @param publicId The public ID of the entity |
516 | | * @param systemId The system ID of the entity |
517 | | * @param content the entity value (without processing). |
518 | | */ |
519 | | void |
520 | | xmlSAX2EntityDecl(void *ctx, const xmlChar *name, int type, |
521 | | const xmlChar *publicId, const xmlChar *systemId, xmlChar *content) |
522 | 84.4k | { |
523 | 84.4k | xmlEntityPtr ent; |
524 | 84.4k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
525 | 84.4k | int extSubset; |
526 | 84.4k | int res; |
527 | | |
528 | 84.4k | if ((ctxt == NULL) || (ctxt->myDoc == NULL)) |
529 | 0 | return; |
530 | | |
531 | 84.4k | extSubset = ctxt->inSubset == 2; |
532 | 84.4k | res = xmlAddEntity(ctxt->myDoc, extSubset, name, type, publicId, systemId, |
533 | 84.4k | content, &ent); |
534 | 84.4k | switch (res) { |
535 | 19.9k | case XML_ERR_OK: |
536 | 19.9k | break; |
537 | 25 | case XML_ERR_NO_MEMORY: |
538 | 25 | xmlSAX2ErrMemory(ctxt); |
539 | 25 | return; |
540 | 41.0k | case XML_WAR_ENTITY_REDEFINED: |
541 | 41.0k | if (ctxt->pedantic) { |
542 | 18.3k | if (extSubset) |
543 | 536 | xmlWarnMsg(ctxt, res, "Entity(%s) already defined in the" |
544 | 536 | " external subset\n", name); |
545 | 17.7k | else |
546 | 17.7k | xmlWarnMsg(ctxt, res, "Entity(%s) already defined in the" |
547 | 17.7k | " internal subset\n", name); |
548 | 18.3k | } |
549 | 41.0k | return; |
550 | 23.4k | case XML_ERR_REDECL_PREDEF_ENTITY: |
551 | | /* |
552 | | * Technically an error but it's a common mistake to get double |
553 | | * escaping according to "4.6 Predefined Entities" wrong. |
554 | | */ |
555 | 23.4k | xmlWarnMsg(ctxt, res, "Invalid redeclaration of predefined" |
556 | 23.4k | " entity '%s'", name); |
557 | 23.4k | return; |
558 | 0 | default: |
559 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
560 | 0 | "Unexpected error code from xmlAddEntity\n", |
561 | 0 | NULL, NULL); |
562 | 0 | return; |
563 | 84.4k | } |
564 | | |
565 | 19.9k | if ((ent->URI == NULL) && (systemId != NULL)) { |
566 | 7.90k | xmlChar *URI; |
567 | 7.90k | const char *base = NULL; |
568 | 7.90k | int i; |
569 | | |
570 | 7.95k | for (i = ctxt->inputNr - 1; i >= 0; i--) { |
571 | 7.95k | if (ctxt->inputTab[i]->filename != NULL) { |
572 | 7.90k | base = ctxt->inputTab[i]->filename; |
573 | 7.90k | break; |
574 | 7.90k | } |
575 | 7.95k | } |
576 | | |
577 | | /* |
578 | | * We don't really need the 'directory' struct member, but some |
579 | | * users set it manually to a base URI for memory streams. |
580 | | */ |
581 | 7.90k | if (base == NULL) |
582 | 0 | base = ctxt->directory; |
583 | | |
584 | 7.90k | res = xmlBuildURISafe(systemId, (const xmlChar *) base, &URI); |
585 | | |
586 | 7.90k | if (URI == NULL) { |
587 | 187 | if (res < 0) { |
588 | 10 | xmlSAX2ErrMemory(ctxt); |
589 | 177 | } else { |
590 | 177 | xmlWarnMsg(ctxt, XML_ERR_INVALID_URI, |
591 | 177 | "Can't resolve URI: %s\n", systemId); |
592 | 177 | } |
593 | 7.71k | } else if (xmlStrlen(URI) > XML_MAX_URI_LENGTH) { |
594 | 6 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, "URI too long"); |
595 | 6 | xmlFree(URI); |
596 | 7.71k | } else { |
597 | 7.71k | ent->URI = URI; |
598 | 7.71k | } |
599 | 7.90k | } |
600 | 19.9k | } |
601 | | |
602 | | /** |
603 | | * An attribute definition has been parsed |
604 | | * |
605 | | * @param ctx the user data (XML parser context) |
606 | | * @param elem the name of the element |
607 | | * @param fullname the attribute name |
608 | | * @param type the attribute type |
609 | | * @param def the type of default value |
610 | | * @param defaultValue the attribute default value |
611 | | * @param tree the tree of enumerated value set |
612 | | */ |
613 | | void |
614 | | xmlSAX2AttributeDecl(void *ctx, const xmlChar *elem, const xmlChar *fullname, |
615 | | int type, int def, const xmlChar *defaultValue, |
616 | | xmlEnumeration *tree) |
617 | 42.1k | { |
618 | 42.1k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
619 | 42.1k | xmlAttributePtr attr; |
620 | 42.1k | const xmlChar *name = NULL; |
621 | 42.1k | xmlChar *prefix = NULL; |
622 | | |
623 | | /* Avoid unused variable warning if features are disabled. */ |
624 | 42.1k | (void) attr; |
625 | | |
626 | 42.1k | if ((ctxt == NULL) || (ctxt->myDoc == NULL)) |
627 | 0 | return; |
628 | | |
629 | 42.1k | if ((xmlStrEqual(fullname, BAD_CAST "xml:id")) && |
630 | 42.1k | (type != XML_ATTRIBUTE_ID)) { |
631 | 115 | xmlErrId(ctxt, XML_DTD_XMLID_TYPE, |
632 | 115 | "xml:id : attribute type should be ID\n", NULL); |
633 | 115 | } |
634 | 42.1k | name = xmlSplitQName4(fullname, &prefix); |
635 | 42.1k | if (name == NULL) |
636 | 9 | xmlSAX2ErrMemory(ctxt); |
637 | 42.1k | ctxt->vctxt.valid = 1; |
638 | 42.1k | if (ctxt->inSubset == 1) |
639 | 24.2k | attr = xmlAddAttributeDecl(&ctxt->vctxt, ctxt->myDoc->intSubset, elem, |
640 | 24.2k | name, prefix, (xmlAttributeType) type, |
641 | 24.2k | (xmlAttributeDefault) def, defaultValue, tree); |
642 | 17.9k | else if (ctxt->inSubset == 2) |
643 | 17.9k | attr = xmlAddAttributeDecl(&ctxt->vctxt, ctxt->myDoc->extSubset, elem, |
644 | 17.9k | name, prefix, (xmlAttributeType) type, |
645 | 17.9k | (xmlAttributeDefault) def, defaultValue, tree); |
646 | 0 | else { |
647 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
648 | 0 | "SAX.xmlSAX2AttributeDecl(%s) called while not in subset\n", |
649 | 0 | name, NULL); |
650 | 0 | xmlFree(prefix); |
651 | 0 | xmlFreeEnumeration(tree); |
652 | 0 | return; |
653 | 0 | } |
654 | 42.1k | #ifdef LIBXML_VALID_ENABLED |
655 | 42.1k | if (ctxt->vctxt.valid == 0) |
656 | 0 | ctxt->valid = 0; |
657 | 42.1k | if ((attr != NULL) && (ctxt->validate) && (ctxt->wellFormed) && |
658 | 42.1k | (ctxt->myDoc->intSubset != NULL)) |
659 | 0 | ctxt->valid &= xmlValidateAttributeDecl(&ctxt->vctxt, ctxt->myDoc, |
660 | 0 | attr); |
661 | 42.1k | #endif /* LIBXML_VALID_ENABLED */ |
662 | 42.1k | if (prefix != NULL) |
663 | 2.55k | xmlFree(prefix); |
664 | 42.1k | } |
665 | | |
666 | | /** |
667 | | * An element definition has been parsed |
668 | | * |
669 | | * @param ctx the user data (XML parser context) |
670 | | * @param name the element name |
671 | | * @param type the element type |
672 | | * @param content the element value tree |
673 | | */ |
674 | | void |
675 | | xmlSAX2ElementDecl(void *ctx, const xmlChar * name, int type, |
676 | | xmlElementContent *content) |
677 | 43.0k | { |
678 | 43.0k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
679 | 43.0k | xmlElementPtr elem = NULL; |
680 | | |
681 | | /* Avoid unused variable warning if features are disabled. */ |
682 | 43.0k | (void) elem; |
683 | | |
684 | 43.0k | if ((ctxt == NULL) || (ctxt->myDoc == NULL)) |
685 | 0 | return; |
686 | | |
687 | 43.0k | if (ctxt->inSubset == 1) |
688 | 9.79k | elem = xmlAddElementDecl(&ctxt->vctxt, ctxt->myDoc->intSubset, |
689 | 9.79k | name, (xmlElementTypeVal) type, content); |
690 | 33.2k | else if (ctxt->inSubset == 2) |
691 | 33.2k | elem = xmlAddElementDecl(&ctxt->vctxt, ctxt->myDoc->extSubset, |
692 | 33.2k | name, (xmlElementTypeVal) type, content); |
693 | 0 | else { |
694 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
695 | 0 | "SAX.xmlSAX2ElementDecl(%s) called while not in subset\n", |
696 | 0 | name, NULL); |
697 | 0 | return; |
698 | 0 | } |
699 | 43.0k | #ifdef LIBXML_VALID_ENABLED |
700 | 43.0k | if (elem == NULL) |
701 | 37.5k | ctxt->valid = 0; |
702 | 43.0k | if (ctxt->validate && ctxt->wellFormed && |
703 | 43.0k | ctxt->myDoc && ctxt->myDoc->intSubset) |
704 | 0 | ctxt->valid &= |
705 | 0 | xmlValidateElementDecl(&ctxt->vctxt, ctxt->myDoc, elem); |
706 | 43.0k | #endif /* LIBXML_VALID_ENABLED */ |
707 | 43.0k | } |
708 | | |
709 | | /** |
710 | | * What to do when a notation declaration has been parsed. |
711 | | * |
712 | | * @param ctx the user data (XML parser context) |
713 | | * @param name The name of the notation |
714 | | * @param publicId The public ID of the entity |
715 | | * @param systemId The system ID of the entity |
716 | | */ |
717 | | void |
718 | | xmlSAX2NotationDecl(void *ctx, const xmlChar *name, |
719 | | const xmlChar *publicId, const xmlChar *systemId) |
720 | 2.09k | { |
721 | 2.09k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
722 | 2.09k | xmlNotationPtr nota = NULL; |
723 | | |
724 | | /* Avoid unused variable warning if features are disabled. */ |
725 | 2.09k | (void) nota; |
726 | | |
727 | 2.09k | if ((ctxt == NULL) || (ctxt->myDoc == NULL)) |
728 | 0 | return; |
729 | | |
730 | 2.09k | if ((publicId == NULL) && (systemId == NULL)) { |
731 | 83 | xmlFatalErrMsg(ctxt, XML_ERR_NOTATION_PROCESSING, |
732 | 83 | "SAX.xmlSAX2NotationDecl(%s) externalID or PublicID missing\n", |
733 | 83 | name, NULL); |
734 | 83 | return; |
735 | 2.01k | } else if (ctxt->inSubset == 1) |
736 | 1.47k | nota = xmlAddNotationDecl(&ctxt->vctxt, ctxt->myDoc->intSubset, name, |
737 | 1.47k | publicId, systemId); |
738 | 540 | else if (ctxt->inSubset == 2) |
739 | 540 | nota = xmlAddNotationDecl(&ctxt->vctxt, ctxt->myDoc->extSubset, name, |
740 | 540 | publicId, systemId); |
741 | 0 | else { |
742 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NOTATION_PROCESSING, |
743 | 0 | "SAX.xmlSAX2NotationDecl(%s) called while not in subset\n", |
744 | 0 | name, NULL); |
745 | 0 | return; |
746 | 0 | } |
747 | 2.01k | #ifdef LIBXML_VALID_ENABLED |
748 | 2.01k | if (nota == NULL) ctxt->valid = 0; |
749 | 2.01k | if ((ctxt->validate) && (ctxt->wellFormed) && |
750 | 2.01k | (ctxt->myDoc->intSubset != NULL)) |
751 | 0 | ctxt->valid &= xmlValidateNotationDecl(&ctxt->vctxt, ctxt->myDoc, |
752 | 0 | nota); |
753 | 2.01k | #endif /* LIBXML_VALID_ENABLED */ |
754 | 2.01k | } |
755 | | |
756 | | /** |
757 | | * What to do when an unparsed entity declaration is parsed |
758 | | * |
759 | | * @param ctx the user data (XML parser context) |
760 | | * @param name The name of the entity |
761 | | * @param publicId The public ID of the entity |
762 | | * @param systemId The system ID of the entity |
763 | | * @param notationName the name of the notation |
764 | | */ |
765 | | void |
766 | | xmlSAX2UnparsedEntityDecl(void *ctx, const xmlChar *name, |
767 | | const xmlChar *publicId, const xmlChar *systemId, |
768 | | const xmlChar *notationName) |
769 | 2.24k | { |
770 | 2.24k | xmlSAX2EntityDecl(ctx, name, XML_EXTERNAL_GENERAL_UNPARSED_ENTITY, |
771 | 2.24k | publicId, systemId, (xmlChar *) notationName); |
772 | 2.24k | } |
773 | | |
774 | | /** |
775 | | * Receive the document locator at startup, actually xmlDefaultSAXLocator |
776 | | * Everything is available on the context, so this is useless in our case. |
777 | | * |
778 | | * @param ctx the user data (XML parser context) |
779 | | * @param loc A SAX Locator |
780 | | */ |
781 | | void |
782 | | xmlSAX2SetDocumentLocator(void *ctx ATTRIBUTE_UNUSED, xmlSAXLocator *loc ATTRIBUTE_UNUSED) |
783 | 31.5k | { |
784 | 31.5k | } |
785 | | |
786 | | /** |
787 | | * called when the document start being processed. |
788 | | * |
789 | | * @param ctx the user data (XML parser context) |
790 | | */ |
791 | | void |
792 | | xmlSAX2StartDocument(void *ctx) |
793 | 30.7k | { |
794 | 30.7k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
795 | 30.7k | xmlDocPtr doc; |
796 | | |
797 | 30.7k | if (ctx == NULL) return; |
798 | | |
799 | 30.7k | #ifdef LIBXML_HTML_ENABLED |
800 | 30.7k | if (ctxt->html) { |
801 | 0 | if (ctxt->myDoc == NULL) |
802 | 0 | ctxt->myDoc = htmlNewDocNoDtD(NULL, NULL); |
803 | 0 | if (ctxt->myDoc == NULL) { |
804 | 0 | xmlSAX2ErrMemory(ctxt); |
805 | 0 | return; |
806 | 0 | } |
807 | 0 | ctxt->myDoc->properties = XML_DOC_HTML; |
808 | 0 | ctxt->myDoc->parseFlags = ctxt->options; |
809 | 0 | } else |
810 | 30.7k | #endif |
811 | 30.7k | { |
812 | 30.7k | doc = ctxt->myDoc = xmlNewDoc(ctxt->version); |
813 | 30.7k | if (doc != NULL) { |
814 | 30.6k | doc->properties = 0; |
815 | 30.6k | if (ctxt->options & XML_PARSE_OLD10) |
816 | 10.2k | doc->properties |= XML_DOC_OLD10; |
817 | 30.6k | doc->parseFlags = ctxt->options; |
818 | 30.6k | doc->standalone = ctxt->standalone; |
819 | 30.6k | } else { |
820 | 35 | xmlSAX2ErrMemory(ctxt); |
821 | 35 | return; |
822 | 35 | } |
823 | 30.6k | if ((ctxt->dictNames) && (doc != NULL)) { |
824 | 20.7k | doc->dict = ctxt->dict; |
825 | 20.7k | xmlDictReference(doc->dict); |
826 | 20.7k | } |
827 | 30.6k | } |
828 | 30.6k | if ((ctxt->myDoc != NULL) && (ctxt->myDoc->URL == NULL) && |
829 | 30.6k | (ctxt->input != NULL) && (ctxt->input->filename != NULL)) { |
830 | 30.6k | ctxt->myDoc->URL = xmlPathToURI((const xmlChar *)ctxt->input->filename); |
831 | 30.6k | if (ctxt->myDoc->URL == NULL) |
832 | 18 | xmlSAX2ErrMemory(ctxt); |
833 | 30.6k | } |
834 | 30.6k | } |
835 | | |
836 | | /** |
837 | | * called when the document end has been detected. |
838 | | * |
839 | | * @param ctx the user data (XML parser context) |
840 | | */ |
841 | | void |
842 | | xmlSAX2EndDocument(void *ctx) |
843 | 31.2k | { |
844 | 31.2k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
845 | 31.2k | xmlDocPtr doc; |
846 | | |
847 | 31.2k | if (ctx == NULL) return; |
848 | 31.2k | #ifdef LIBXML_VALID_ENABLED |
849 | 31.2k | if (ctxt->validate && ctxt->wellFormed && |
850 | 31.2k | ctxt->myDoc && ctxt->myDoc->intSubset) |
851 | 0 | ctxt->valid &= xmlValidateDocumentFinal(&ctxt->vctxt, ctxt->myDoc); |
852 | 31.2k | #endif /* LIBXML_VALID_ENABLED */ |
853 | | |
854 | 31.2k | doc = ctxt->myDoc; |
855 | 31.2k | if (doc == NULL) |
856 | 513 | return; |
857 | | |
858 | 30.7k | if (doc->encoding == NULL) { |
859 | 30.7k | const xmlChar *encoding = xmlGetActualEncoding(ctxt); |
860 | | |
861 | 30.7k | if (encoding != NULL) { |
862 | 544 | doc->encoding = xmlStrdup(encoding); |
863 | 544 | if (doc->encoding == NULL) |
864 | 2 | xmlSAX2ErrMemory(ctxt); |
865 | 544 | } |
866 | 30.7k | } |
867 | | |
868 | 30.7k | #ifdef LIBXML_HTML_ENABLED |
869 | 30.7k | if (ctxt->html) { |
870 | 0 | if (((ctxt->options & HTML_PARSE_NODEFDTD) == 0) && |
871 | 0 | (doc->intSubset == NULL)) { |
872 | 0 | doc->intSubset = xmlCreateIntSubset(doc, BAD_CAST "html", |
873 | 0 | BAD_CAST "-//W3C//DTD HTML 4.0 Transitional//EN", |
874 | 0 | BAD_CAST "http://www.w3.org/TR/REC-html40/loose.dtd"); |
875 | 0 | if (doc->intSubset == NULL) |
876 | 0 | xmlSAX2ErrMemory(ctxt); |
877 | 0 | } |
878 | 0 | } else |
879 | 30.7k | #endif /* LIBXML_HTML_ENABLED */ |
880 | 30.7k | { |
881 | 30.7k | if (ctxt->wellFormed) { |
882 | 274 | doc->properties |= XML_DOC_WELLFORMED; |
883 | 274 | if (ctxt->valid) |
884 | 264 | doc->properties |= XML_DOC_DTDVALID; |
885 | 274 | if (ctxt->nsWellFormed) |
886 | 197 | doc->properties |= XML_DOC_NSVALID; |
887 | 274 | } |
888 | | |
889 | 30.7k | if (ctxt->options & XML_PARSE_OLD10) |
890 | 10.3k | doc->properties |= XML_DOC_OLD10; |
891 | 30.7k | } |
892 | 30.7k | } |
893 | | |
894 | | static void |
895 | 1.85M | xmlSAX2AppendChild(xmlParserCtxtPtr ctxt, xmlNodePtr node) { |
896 | 1.85M | xmlNodePtr parent; |
897 | 1.85M | xmlNodePtr last; |
898 | | |
899 | 1.85M | if (ctxt->inSubset == 1) { |
900 | 49.4k | parent = (xmlNodePtr) ctxt->myDoc->intSubset; |
901 | 1.80M | } else if (ctxt->inSubset == 2) { |
902 | 357k | parent = (xmlNodePtr) ctxt->myDoc->extSubset; |
903 | 1.44M | } else { |
904 | 1.44M | parent = ctxt->node; |
905 | 1.44M | if (parent == NULL) |
906 | 21.8k | parent = (xmlNodePtr) ctxt->myDoc; |
907 | 1.44M | } |
908 | | |
909 | 1.85M | last = parent->last; |
910 | 1.85M | if (last == NULL) { |
911 | 156k | parent->children = node; |
912 | 1.70M | } else { |
913 | 1.70M | last->next = node; |
914 | 1.70M | node->prev = last; |
915 | 1.70M | } |
916 | | |
917 | 1.85M | parent->last = node; |
918 | 1.85M | node->parent = parent; |
919 | | |
920 | 1.85M | if ((node->type != XML_TEXT_NODE) && |
921 | 1.85M | (ctxt->input != NULL)) { |
922 | 1.59M | if ((unsigned) ctxt->input->line < (unsigned) USHRT_MAX) |
923 | 1.48M | node->line = ctxt->input->line; |
924 | 105k | else |
925 | 105k | node->line = USHRT_MAX; |
926 | 1.59M | } |
927 | 1.85M | } |
928 | | |
929 | | #if defined(LIBXML_SAX1_ENABLED) |
930 | | /** |
931 | | * Handle a namespace error |
932 | | * |
933 | | * @param ctxt an XML parser context |
934 | | * @param error the error number |
935 | | * @param msg the error message |
936 | | * @param str1 an error string |
937 | | * @param str2 an error string |
938 | | */ |
939 | | static void LIBXML_ATTR_FORMAT(3,0) |
940 | | xmlNsErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
941 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
942 | 0 | { |
943 | 0 | xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR, |
944 | 0 | str1, str2, NULL, 0, msg, str1, str2); |
945 | 0 | } |
946 | | |
947 | | /** |
948 | | * Handle an attribute that has been read by the parser. |
949 | | * |
950 | | * Deprecated SAX1 interface. |
951 | | * |
952 | | * @param ctxt the parser context |
953 | | * @param fullname the attribute name, including namespace prefix |
954 | | * @param value the attribute value |
955 | | * @param prefix the namespace prefix |
956 | | */ |
957 | | static void |
958 | | xmlSAX1Attribute(xmlParserCtxtPtr ctxt, const xmlChar *fullname, |
959 | | const xmlChar *value, const xmlChar *prefix) |
960 | 0 | { |
961 | 0 | xmlAttrPtr ret; |
962 | 0 | const xmlChar *name; |
963 | 0 | xmlChar *ns; |
964 | 0 | xmlNsPtr namespace; |
965 | | |
966 | | /* |
967 | | * Split the full name into a namespace prefix and the tag name |
968 | | */ |
969 | 0 | name = xmlSplitQName4(fullname, &ns); |
970 | 0 | if (name == NULL) { |
971 | 0 | xmlSAX2ErrMemory(ctxt); |
972 | 0 | return; |
973 | 0 | } |
974 | | |
975 | | /* |
976 | | * Check whether it's a namespace definition |
977 | | */ |
978 | 0 | if ((ns == NULL) && |
979 | 0 | (name[0] == 'x') && (name[1] == 'm') && (name[2] == 'l') && |
980 | 0 | (name[3] == 'n') && (name[4] == 's') && (name[5] == 0)) { |
981 | 0 | xmlNsPtr nsret; |
982 | 0 | xmlChar *val; |
983 | | |
984 | | /* Avoid unused variable warning if features are disabled. */ |
985 | 0 | (void) nsret; |
986 | |
|
987 | 0 | if (!ctxt->replaceEntities) { |
988 | | /* TODO: normalize if needed */ |
989 | 0 | val = xmlExpandEntitiesInAttValue(ctxt, value, /* normalize */ 0); |
990 | 0 | if (val == NULL) { |
991 | 0 | xmlSAX2ErrMemory(ctxt); |
992 | 0 | return; |
993 | 0 | } |
994 | 0 | } else { |
995 | 0 | val = (xmlChar *) value; |
996 | 0 | } |
997 | | |
998 | 0 | if (val[0] != 0) { |
999 | 0 | xmlURIPtr uri; |
1000 | |
|
1001 | 0 | if (xmlParseURISafe((const char *)val, &uri) < 0) |
1002 | 0 | xmlSAX2ErrMemory(ctxt); |
1003 | 0 | if (uri == NULL) { |
1004 | 0 | xmlNsWarnMsg(ctxt, XML_WAR_NS_URI, |
1005 | 0 | "xmlns:%s: %s not a valid URI\n", name, value); |
1006 | 0 | } else { |
1007 | 0 | if (uri->scheme == NULL) { |
1008 | 0 | xmlNsWarnMsg(ctxt, XML_WAR_NS_URI_RELATIVE, |
1009 | 0 | "xmlns:%s: URI %s is not absolute\n", |
1010 | 0 | name, value); |
1011 | 0 | } |
1012 | 0 | xmlFreeURI(uri); |
1013 | 0 | } |
1014 | 0 | } |
1015 | | |
1016 | | /* a default namespace definition */ |
1017 | 0 | nsret = xmlNewNs(ctxt->node, val, NULL); |
1018 | 0 | if (nsret == NULL) { |
1019 | 0 | xmlSAX2ErrMemory(ctxt); |
1020 | 0 | } |
1021 | 0 | #ifdef LIBXML_VALID_ENABLED |
1022 | | /* |
1023 | | * Validate also for namespace decls, they are attributes from |
1024 | | * an XML-1.0 perspective |
1025 | | */ |
1026 | 0 | else if (ctxt->validate && ctxt->wellFormed && |
1027 | 0 | ctxt->myDoc && ctxt->myDoc->intSubset) { |
1028 | 0 | ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc, |
1029 | 0 | ctxt->node, prefix, nsret, val); |
1030 | 0 | } |
1031 | 0 | #endif /* LIBXML_VALID_ENABLED */ |
1032 | 0 | if (val != value) |
1033 | 0 | xmlFree(val); |
1034 | 0 | return; |
1035 | 0 | } |
1036 | 0 | if ((ns != NULL) && (ns[0] == 'x') && (ns[1] == 'm') && (ns[2] == 'l') && |
1037 | 0 | (ns[3] == 'n') && (ns[4] == 's') && (ns[5] == 0)) { |
1038 | 0 | xmlNsPtr nsret; |
1039 | 0 | xmlChar *val; |
1040 | | |
1041 | | /* Avoid unused variable warning if features are disabled. */ |
1042 | 0 | (void) nsret; |
1043 | |
|
1044 | 0 | if (!ctxt->replaceEntities) { |
1045 | | /* TODO: normalize if needed */ |
1046 | 0 | val = xmlExpandEntitiesInAttValue(ctxt, value, /* normalize */ 0); |
1047 | 0 | if (val == NULL) { |
1048 | 0 | xmlSAX2ErrMemory(ctxt); |
1049 | 0 | xmlFree(ns); |
1050 | 0 | return; |
1051 | 0 | } |
1052 | 0 | } else { |
1053 | 0 | val = (xmlChar *) value; |
1054 | 0 | } |
1055 | | |
1056 | 0 | if (val[0] == 0) { |
1057 | 0 | xmlNsErrMsg(ctxt, XML_NS_ERR_EMPTY, |
1058 | 0 | "Empty namespace name for prefix %s\n", name, NULL); |
1059 | 0 | } |
1060 | 0 | if ((ctxt->pedantic != 0) && (val[0] != 0)) { |
1061 | 0 | xmlURIPtr uri; |
1062 | |
|
1063 | 0 | if (xmlParseURISafe((const char *)val, &uri) < 0) |
1064 | 0 | xmlSAX2ErrMemory(ctxt); |
1065 | 0 | if (uri == NULL) { |
1066 | 0 | xmlNsWarnMsg(ctxt, XML_WAR_NS_URI, |
1067 | 0 | "xmlns:%s: %s not a valid URI\n", name, value); |
1068 | 0 | } else { |
1069 | 0 | if (uri->scheme == NULL) { |
1070 | 0 | xmlNsWarnMsg(ctxt, XML_WAR_NS_URI_RELATIVE, |
1071 | 0 | "xmlns:%s: URI %s is not absolute\n", name, value); |
1072 | 0 | } |
1073 | 0 | xmlFreeURI(uri); |
1074 | 0 | } |
1075 | 0 | } |
1076 | | |
1077 | | /* a standard namespace definition */ |
1078 | 0 | nsret = xmlNewNs(ctxt->node, val, name); |
1079 | 0 | xmlFree(ns); |
1080 | |
|
1081 | 0 | if (nsret == NULL) { |
1082 | 0 | xmlSAX2ErrMemory(ctxt); |
1083 | 0 | } |
1084 | 0 | #ifdef LIBXML_VALID_ENABLED |
1085 | | /* |
1086 | | * Validate also for namespace decls, they are attributes from |
1087 | | * an XML-1.0 perspective |
1088 | | */ |
1089 | 0 | else if (ctxt->validate && ctxt->wellFormed && |
1090 | 0 | ctxt->myDoc && ctxt->myDoc->intSubset) { |
1091 | 0 | ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc, |
1092 | 0 | ctxt->node, prefix, nsret, value); |
1093 | 0 | } |
1094 | 0 | #endif /* LIBXML_VALID_ENABLED */ |
1095 | 0 | if (val != value) |
1096 | 0 | xmlFree(val); |
1097 | 0 | return; |
1098 | 0 | } |
1099 | | |
1100 | 0 | if (ns != NULL) { |
1101 | 0 | int res; |
1102 | |
|
1103 | 0 | res = xmlSearchNsSafe(ctxt->node, ns, &namespace); |
1104 | 0 | if (res < 0) |
1105 | 0 | xmlSAX2ErrMemory(ctxt); |
1106 | |
|
1107 | 0 | if (namespace == NULL) { |
1108 | 0 | xmlNsErrMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
1109 | 0 | "Namespace prefix %s of attribute %s is not defined\n", |
1110 | 0 | ns, name); |
1111 | 0 | } else { |
1112 | 0 | xmlAttrPtr prop; |
1113 | |
|
1114 | 0 | prop = ctxt->node->properties; |
1115 | 0 | while (prop != NULL) { |
1116 | 0 | if (prop->ns != NULL) { |
1117 | 0 | if ((xmlStrEqual(name, prop->name)) && |
1118 | 0 | ((namespace == prop->ns) || |
1119 | 0 | (xmlStrEqual(namespace->href, prop->ns->href)))) { |
1120 | 0 | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, |
1121 | 0 | XML_ERR_ATTRIBUTE_REDEFINED, XML_ERR_FATAL, |
1122 | 0 | name, NULL, NULL, 0, |
1123 | 0 | "Attribute %s in %s redefined\n", |
1124 | 0 | name, namespace->href); |
1125 | 0 | goto error; |
1126 | 0 | } |
1127 | 0 | } |
1128 | 0 | prop = prop->next; |
1129 | 0 | } |
1130 | 0 | } |
1131 | 0 | } else { |
1132 | 0 | namespace = NULL; |
1133 | 0 | } |
1134 | | |
1135 | | /* !!!!!! <a toto:arg="" xmlns:toto="http://toto.com"> */ |
1136 | 0 | ret = xmlNewNsProp(ctxt->node, namespace, name, NULL); |
1137 | 0 | if (ret == NULL) { |
1138 | 0 | xmlSAX2ErrMemory(ctxt); |
1139 | 0 | goto error; |
1140 | 0 | } |
1141 | | |
1142 | 0 | if (ctxt->replaceEntities == 0) { |
1143 | 0 | if (xmlNodeParseAttValue(ret->doc, ret, value, SIZE_MAX, NULL) < 0) |
1144 | 0 | xmlSAX2ErrMemory(ctxt); |
1145 | 0 | } else if (value != NULL) { |
1146 | 0 | ret->children = xmlNewDocText(ctxt->myDoc, value); |
1147 | 0 | if (ret->children == NULL) { |
1148 | 0 | xmlSAX2ErrMemory(ctxt); |
1149 | 0 | } else { |
1150 | 0 | ret->last = ret->children; |
1151 | 0 | ret->children->parent = (xmlNodePtr) ret; |
1152 | 0 | } |
1153 | 0 | } |
1154 | |
|
1155 | 0 | #ifdef LIBXML_VALID_ENABLED |
1156 | 0 | if (ctxt->validate && ctxt->wellFormed && |
1157 | 0 | ctxt->myDoc && ctxt->myDoc->intSubset) { |
1158 | | |
1159 | | /* |
1160 | | * If we don't substitute entities, the validation should be |
1161 | | * done on a value with replaced entities anyway. |
1162 | | */ |
1163 | 0 | if (!ctxt->replaceEntities) { |
1164 | 0 | xmlChar *val; |
1165 | | |
1166 | | /* TODO: normalize if needed */ |
1167 | 0 | val = xmlExpandEntitiesInAttValue(ctxt, value, /* normalize */ 0); |
1168 | |
|
1169 | 0 | if (val == NULL) |
1170 | 0 | ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, |
1171 | 0 | ctxt->myDoc, ctxt->node, ret, value); |
1172 | 0 | else { |
1173 | 0 | xmlChar *nvalnorm; |
1174 | | |
1175 | | /* |
1176 | | * Do the last stage of the attribute normalization |
1177 | | * It need to be done twice ... it's an extra burden related |
1178 | | * to the ability to keep xmlSAX2References in attributes |
1179 | | */ |
1180 | 0 | nvalnorm = xmlValidCtxtNormalizeAttributeValue( |
1181 | 0 | &ctxt->vctxt, ctxt->myDoc, |
1182 | 0 | ctxt->node, fullname, val); |
1183 | 0 | if (nvalnorm != NULL) { |
1184 | 0 | xmlFree(val); |
1185 | 0 | val = nvalnorm; |
1186 | 0 | } |
1187 | |
|
1188 | 0 | ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, |
1189 | 0 | ctxt->myDoc, ctxt->node, ret, val); |
1190 | 0 | xmlFree(val); |
1191 | 0 | } |
1192 | 0 | } else { |
1193 | | /* |
1194 | | * When replacing entities, make sure that IDs in |
1195 | | * entities aren't registered. This also shouldn't be |
1196 | | * done when entities aren't replaced, but this would |
1197 | | * require to rework IDREF checks. |
1198 | | */ |
1199 | 0 | if (ctxt->input->entity != NULL) |
1200 | 0 | ctxt->vctxt.flags |= XML_VCTXT_IN_ENTITY; |
1201 | |
|
1202 | 0 | ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, ctxt->myDoc, |
1203 | 0 | ctxt->node, ret, value); |
1204 | |
|
1205 | 0 | ctxt->vctxt.flags &= ~XML_VCTXT_IN_ENTITY; |
1206 | 0 | } |
1207 | 0 | } else |
1208 | 0 | #endif /* LIBXML_VALID_ENABLED */ |
1209 | 0 | if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) && |
1210 | 0 | (ctxt->input->entity == NULL) && |
1211 | | /* Don't create IDs containing entity references */ |
1212 | 0 | (ret->children != NULL) && |
1213 | 0 | (ret->children->type == XML_TEXT_NODE) && |
1214 | 0 | (ret->children->next == NULL)) { |
1215 | 0 | xmlChar *content = ret->children->content; |
1216 | | /* |
1217 | | * when validating, the ID registration is done at the attribute |
1218 | | * validation level. Otherwise we have to do specific handling here. |
1219 | | */ |
1220 | 0 | if (xmlStrEqual(fullname, BAD_CAST "xml:id")) { |
1221 | | /* |
1222 | | * Add the xml:id value |
1223 | | * |
1224 | | * Open issue: normalization of the value. |
1225 | | */ |
1226 | 0 | if (xmlValidateNCName(content, 1) != 0) { |
1227 | 0 | xmlErrId(ctxt, XML_DTD_XMLID_VALUE, |
1228 | 0 | "xml:id : attribute value %s is not an NCName\n", |
1229 | 0 | content); |
1230 | 0 | } |
1231 | 0 | xmlAddID(&ctxt->vctxt, ctxt->myDoc, content, ret); |
1232 | 0 | } else { |
1233 | 0 | int res = xmlIsID(ctxt->myDoc, ctxt->node, ret); |
1234 | |
|
1235 | 0 | if (res < 0) |
1236 | 0 | xmlCtxtErrMemory(ctxt); |
1237 | 0 | else if (res > 0) |
1238 | 0 | xmlAddID(&ctxt->vctxt, ctxt->myDoc, content, ret); |
1239 | 0 | else if (xmlIsRef(ctxt->myDoc, ctxt->node, ret)) |
1240 | 0 | xmlAddRef(&ctxt->vctxt, ctxt->myDoc, content, ret); |
1241 | 0 | } |
1242 | 0 | } |
1243 | |
|
1244 | 0 | error: |
1245 | 0 | if (ns != NULL) |
1246 | 0 | xmlFree(ns); |
1247 | 0 | } |
1248 | | |
1249 | | /* |
1250 | | * |
1251 | | * Check defaulted attributes from the DTD |
1252 | | * |
1253 | | * Deprecated SAX1 interface. |
1254 | | */ |
1255 | | static void |
1256 | | xmlCheckDefaultedAttributes(xmlParserCtxtPtr ctxt, const xmlChar *name, |
1257 | 0 | const xmlChar *prefix, const xmlChar **atts) { |
1258 | 0 | xmlElementPtr elemDecl; |
1259 | 0 | const xmlChar *att; |
1260 | 0 | int internal = 1; |
1261 | 0 | int i; |
1262 | |
|
1263 | 0 | elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->intSubset, name, prefix); |
1264 | 0 | if (elemDecl == NULL) { |
1265 | 0 | elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->extSubset, name, prefix); |
1266 | 0 | internal = 0; |
1267 | 0 | } |
1268 | |
|
1269 | 0 | process_external_subset: |
1270 | |
|
1271 | 0 | if (elemDecl != NULL) { |
1272 | 0 | xmlAttributePtr attr = elemDecl->attributes; |
1273 | |
|
1274 | 0 | #ifdef LIBXML_VALID_ENABLED |
1275 | | /* |
1276 | | * Check against defaulted attributes from the external subset |
1277 | | * if the document is stamped as standalone. |
1278 | | * |
1279 | | * This should be moved to valid.c, but we don't keep track |
1280 | | * whether an attribute was defaulted. |
1281 | | */ |
1282 | 0 | if ((ctxt->myDoc->standalone == 1) && |
1283 | 0 | (ctxt->myDoc->extSubset != NULL) && |
1284 | 0 | (ctxt->validate)) { |
1285 | 0 | while (attr != NULL) { |
1286 | 0 | if ((attr->defaultValue != NULL) && |
1287 | 0 | (xmlGetDtdQAttrDesc(ctxt->myDoc->extSubset, |
1288 | 0 | attr->elem, attr->name, |
1289 | 0 | attr->prefix) == attr) && |
1290 | 0 | (xmlGetDtdQAttrDesc(ctxt->myDoc->intSubset, |
1291 | 0 | attr->elem, attr->name, |
1292 | 0 | attr->prefix) == NULL)) { |
1293 | 0 | xmlChar *fulln; |
1294 | |
|
1295 | 0 | if (attr->prefix != NULL) { |
1296 | 0 | fulln = xmlStrdup(attr->prefix); |
1297 | 0 | if (fulln != NULL) |
1298 | 0 | fulln = xmlStrcat(fulln, BAD_CAST ":"); |
1299 | 0 | if (fulln != NULL) |
1300 | 0 | fulln = xmlStrcat(fulln, attr->name); |
1301 | 0 | } else { |
1302 | 0 | fulln = xmlStrdup(attr->name); |
1303 | 0 | } |
1304 | 0 | if (fulln == NULL) { |
1305 | 0 | xmlSAX2ErrMemory(ctxt); |
1306 | 0 | break; |
1307 | 0 | } |
1308 | | |
1309 | | /* |
1310 | | * Check that the attribute is not declared in the |
1311 | | * serialization |
1312 | | */ |
1313 | 0 | att = NULL; |
1314 | 0 | if (atts != NULL) { |
1315 | 0 | i = 0; |
1316 | 0 | att = atts[i]; |
1317 | 0 | while (att != NULL) { |
1318 | 0 | if (xmlStrEqual(att, fulln)) |
1319 | 0 | break; |
1320 | 0 | i += 2; |
1321 | 0 | att = atts[i]; |
1322 | 0 | } |
1323 | 0 | } |
1324 | 0 | if (att == NULL) { |
1325 | 0 | xmlErrValid(ctxt, XML_DTD_STANDALONE_DEFAULTED, |
1326 | 0 | "standalone: attribute %s on %s defaulted from external subset\n", |
1327 | 0 | fulln, |
1328 | 0 | attr->elem); |
1329 | 0 | } |
1330 | 0 | xmlFree(fulln); |
1331 | 0 | } |
1332 | 0 | attr = attr->nexth; |
1333 | 0 | } |
1334 | 0 | } |
1335 | 0 | #endif |
1336 | | |
1337 | | /* |
1338 | | * Actually insert defaulted values when needed |
1339 | | */ |
1340 | 0 | attr = elemDecl->attributes; |
1341 | 0 | while (attr != NULL) { |
1342 | | /* |
1343 | | * Make sure that attributes redefinition occurring in the |
1344 | | * internal subset are not overridden by definitions in the |
1345 | | * external subset. |
1346 | | */ |
1347 | 0 | if (attr->defaultValue != NULL) { |
1348 | | /* |
1349 | | * the element should be instantiated in the tree if: |
1350 | | * - this is a namespace prefix |
1351 | | * - the user required for completion in the tree |
1352 | | * like XSLT |
1353 | | * - there isn't already an attribute definition |
1354 | | * in the internal subset overriding it. |
1355 | | */ |
1356 | 0 | if (((attr->prefix != NULL) && |
1357 | 0 | (xmlStrEqual(attr->prefix, BAD_CAST "xmlns"))) || |
1358 | 0 | ((attr->prefix == NULL) && |
1359 | 0 | (xmlStrEqual(attr->name, BAD_CAST "xmlns"))) || |
1360 | 0 | (ctxt->loadsubset & XML_COMPLETE_ATTRS)) { |
1361 | 0 | xmlAttributePtr tst; |
1362 | |
|
1363 | 0 | tst = xmlGetDtdQAttrDesc(ctxt->myDoc->intSubset, |
1364 | 0 | attr->elem, attr->name, |
1365 | 0 | attr->prefix); |
1366 | 0 | if ((tst == attr) || (tst == NULL)) { |
1367 | 0 | xmlChar fn[50]; |
1368 | 0 | xmlChar *fulln; |
1369 | |
|
1370 | 0 | fulln = xmlBuildQName(attr->name, attr->prefix, fn, 50); |
1371 | 0 | if (fulln == NULL) { |
1372 | 0 | xmlSAX2ErrMemory(ctxt); |
1373 | 0 | return; |
1374 | 0 | } |
1375 | | |
1376 | | /* |
1377 | | * Check that the attribute is not declared in the |
1378 | | * serialization |
1379 | | */ |
1380 | 0 | att = NULL; |
1381 | 0 | if (atts != NULL) { |
1382 | 0 | i = 0; |
1383 | 0 | att = atts[i]; |
1384 | 0 | while (att != NULL) { |
1385 | 0 | if (xmlStrEqual(att, fulln)) |
1386 | 0 | break; |
1387 | 0 | i += 2; |
1388 | 0 | att = atts[i]; |
1389 | 0 | } |
1390 | 0 | } |
1391 | 0 | if (att == NULL) { |
1392 | 0 | xmlSAX1Attribute(ctxt, fulln, |
1393 | 0 | attr->defaultValue, prefix); |
1394 | 0 | } |
1395 | 0 | if ((fulln != fn) && (fulln != attr->name)) |
1396 | 0 | xmlFree(fulln); |
1397 | 0 | } |
1398 | 0 | } |
1399 | 0 | } |
1400 | 0 | attr = attr->nexth; |
1401 | 0 | } |
1402 | 0 | if (internal == 1) { |
1403 | 0 | elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->extSubset, |
1404 | 0 | name, prefix); |
1405 | 0 | internal = 0; |
1406 | 0 | goto process_external_subset; |
1407 | 0 | } |
1408 | 0 | } |
1409 | 0 | } |
1410 | | |
1411 | | /** |
1412 | | * called when an opening tag has been processed. |
1413 | | * |
1414 | | * Deprecated SAX1 interface. |
1415 | | * |
1416 | | * @param ctx the user data (XML parser context) |
1417 | | * @param fullname The element name, including namespace prefix |
1418 | | * @param atts An array of name/value attributes pairs, NULL terminated |
1419 | | */ |
1420 | | static void |
1421 | | xmlSAX1StartElement(void *ctx, const xmlChar *fullname, const xmlChar **atts) |
1422 | 0 | { |
1423 | 0 | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
1424 | 0 | xmlNodePtr ret; |
1425 | 0 | xmlNodePtr parent; |
1426 | 0 | xmlNsPtr ns; |
1427 | 0 | const xmlChar *name; |
1428 | 0 | xmlChar *prefix; |
1429 | 0 | const xmlChar *att; |
1430 | 0 | const xmlChar *value; |
1431 | 0 | int i, res; |
1432 | |
|
1433 | 0 | if ((ctx == NULL) || (fullname == NULL) || (ctxt->myDoc == NULL)) return; |
1434 | | |
1435 | 0 | #ifdef LIBXML_VALID_ENABLED |
1436 | | /* |
1437 | | * First check on validity: |
1438 | | */ |
1439 | 0 | if (ctxt->validate && (ctxt->myDoc->extSubset == NULL) && |
1440 | 0 | ((ctxt->myDoc->intSubset == NULL) || |
1441 | 0 | ((ctxt->myDoc->intSubset->notations == NULL) && |
1442 | 0 | (ctxt->myDoc->intSubset->elements == NULL) && |
1443 | 0 | (ctxt->myDoc->intSubset->attributes == NULL) && |
1444 | 0 | (ctxt->myDoc->intSubset->entities == NULL)))) { |
1445 | 0 | xmlErrValid(ctxt, XML_ERR_NO_DTD, |
1446 | 0 | "Validation failed: no DTD found !", NULL, NULL); |
1447 | 0 | ctxt->validate = 0; |
1448 | 0 | } |
1449 | 0 | #endif |
1450 | | |
1451 | | /* |
1452 | | * Split the full name into a namespace prefix and the tag name |
1453 | | */ |
1454 | 0 | name = xmlSplitQName4(fullname, &prefix); |
1455 | 0 | if (name == NULL) { |
1456 | 0 | xmlSAX2ErrMemory(ctxt); |
1457 | 0 | return; |
1458 | 0 | } |
1459 | | |
1460 | | /* |
1461 | | * Note : the namespace resolution is deferred until the end of the |
1462 | | * attributes parsing, since local namespace can be defined as |
1463 | | * an attribute at this level. |
1464 | | */ |
1465 | 0 | ret = xmlNewDocNode(ctxt->myDoc, NULL, name, NULL); |
1466 | 0 | if (ret == NULL) { |
1467 | 0 | xmlFree(prefix); |
1468 | 0 | xmlSAX2ErrMemory(ctxt); |
1469 | 0 | return; |
1470 | 0 | } |
1471 | 0 | ctxt->nodemem = -1; |
1472 | | |
1473 | | /* Initialize parent before pushing node */ |
1474 | 0 | parent = ctxt->node; |
1475 | 0 | if (parent == NULL) |
1476 | 0 | parent = (xmlNodePtr) ctxt->myDoc; |
1477 | | |
1478 | | /* |
1479 | | * Link the child element |
1480 | | */ |
1481 | 0 | xmlSAX2AppendChild(ctxt, ret); |
1482 | | |
1483 | | /* |
1484 | | * We are parsing a new node. |
1485 | | */ |
1486 | 0 | if (nodePush(ctxt, ret) < 0) { |
1487 | 0 | xmlUnlinkNode(ret); |
1488 | 0 | xmlFreeNode(ret); |
1489 | 0 | if (prefix != NULL) |
1490 | 0 | xmlFree(prefix); |
1491 | 0 | return; |
1492 | 0 | } |
1493 | | |
1494 | | /* |
1495 | | * Insert all the defaulted attributes from the DTD especially |
1496 | | * namespaces |
1497 | | */ |
1498 | 0 | if ((ctxt->myDoc->intSubset != NULL) || |
1499 | 0 | (ctxt->myDoc->extSubset != NULL)) { |
1500 | 0 | xmlCheckDefaultedAttributes(ctxt, name, prefix, atts); |
1501 | 0 | } |
1502 | | |
1503 | | /* |
1504 | | * process all the attributes whose name start with "xmlns" |
1505 | | */ |
1506 | 0 | if (atts != NULL) { |
1507 | 0 | i = 0; |
1508 | 0 | att = atts[i++]; |
1509 | 0 | value = atts[i++]; |
1510 | 0 | while ((att != NULL) && (value != NULL)) { |
1511 | 0 | if ((att[0] == 'x') && (att[1] == 'm') && (att[2] == 'l') && |
1512 | 0 | (att[3] == 'n') && (att[4] == 's')) |
1513 | 0 | xmlSAX1Attribute(ctxt, att, value, prefix); |
1514 | |
|
1515 | 0 | att = atts[i++]; |
1516 | 0 | value = atts[i++]; |
1517 | 0 | } |
1518 | 0 | } |
1519 | | |
1520 | | /* |
1521 | | * Search the namespace, note that since the attributes have been |
1522 | | * processed, the local namespaces are available. |
1523 | | */ |
1524 | 0 | res = xmlSearchNsSafe(ret, prefix, &ns); |
1525 | 0 | if (res < 0) |
1526 | 0 | xmlSAX2ErrMemory(ctxt); |
1527 | 0 | if ((ns == NULL) && (parent != NULL)) { |
1528 | 0 | res = xmlSearchNsSafe(parent, prefix, &ns); |
1529 | 0 | if (res < 0) |
1530 | 0 | xmlSAX2ErrMemory(ctxt); |
1531 | 0 | } |
1532 | 0 | if ((prefix != NULL) && (ns == NULL)) { |
1533 | 0 | xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
1534 | 0 | "Namespace prefix %s is not defined\n", |
1535 | 0 | prefix, NULL); |
1536 | 0 | ns = xmlNewNs(ret, NULL, prefix); |
1537 | 0 | if (ns == NULL) |
1538 | 0 | xmlSAX2ErrMemory(ctxt); |
1539 | 0 | } |
1540 | | |
1541 | | /* |
1542 | | * set the namespace node, making sure that if the default namespace |
1543 | | * is unbound on a parent we simply keep it NULL |
1544 | | */ |
1545 | 0 | if ((ns != NULL) && (ns->href != NULL) && |
1546 | 0 | ((ns->href[0] != 0) || (ns->prefix != NULL))) |
1547 | 0 | xmlSetNs(ret, ns); |
1548 | | |
1549 | | /* |
1550 | | * process all the other attributes |
1551 | | */ |
1552 | 0 | if (atts != NULL) { |
1553 | 0 | i = 0; |
1554 | 0 | att = atts[i++]; |
1555 | 0 | value = atts[i++]; |
1556 | 0 | while ((att != NULL) && (value != NULL)) { |
1557 | 0 | if ((att[0] != 'x') || (att[1] != 'm') || (att[2] != 'l') || |
1558 | 0 | (att[3] != 'n') || (att[4] != 's')) |
1559 | 0 | xmlSAX1Attribute(ctxt, att, value, NULL); |
1560 | | |
1561 | | /* |
1562 | | * Next ones |
1563 | | */ |
1564 | 0 | att = atts[i++]; |
1565 | 0 | value = atts[i++]; |
1566 | 0 | } |
1567 | 0 | } |
1568 | |
|
1569 | 0 | #ifdef LIBXML_VALID_ENABLED |
1570 | | /* |
1571 | | * If it's the Document root, finish the DTD validation and |
1572 | | * check the document root element for validity |
1573 | | */ |
1574 | 0 | if ((ctxt->validate) && |
1575 | 0 | ((ctxt->vctxt.flags & XML_VCTXT_DTD_VALIDATED) == 0)) { |
1576 | 0 | int chk; |
1577 | |
|
1578 | 0 | chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc); |
1579 | 0 | if (chk <= 0) |
1580 | 0 | ctxt->valid = 0; |
1581 | 0 | if (chk < 0) |
1582 | 0 | ctxt->wellFormed = 0; |
1583 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
1584 | 0 | ctxt->vctxt.flags |= XML_VCTXT_DTD_VALIDATED; |
1585 | 0 | } |
1586 | 0 | #endif /* LIBXML_VALID_ENABLED */ |
1587 | |
|
1588 | 0 | if (prefix != NULL) |
1589 | 0 | xmlFree(prefix); |
1590 | |
|
1591 | 0 | } |
1592 | | #endif /* LIBXML_SAX1_ENABLED */ |
1593 | | |
1594 | | #ifdef LIBXML_HTML_ENABLED |
1595 | | static void |
1596 | | xmlSAX2HtmlAttribute(xmlParserCtxtPtr ctxt, const xmlChar *fullname, |
1597 | 0 | const xmlChar *value) { |
1598 | 0 | xmlAttrPtr ret; |
1599 | 0 | xmlChar *nval = NULL; |
1600 | |
|
1601 | 0 | ret = xmlNewNsProp(ctxt->node, NULL, fullname, NULL); |
1602 | 0 | if (ret == NULL) { |
1603 | 0 | xmlSAX2ErrMemory(ctxt); |
1604 | 0 | return; |
1605 | 0 | } |
1606 | | |
1607 | 0 | if ((value == NULL) && (htmlIsBooleanAttr(fullname))) { |
1608 | 0 | nval = xmlStrdup(fullname); |
1609 | 0 | if (nval == NULL) { |
1610 | 0 | xmlSAX2ErrMemory(ctxt); |
1611 | 0 | return; |
1612 | 0 | } |
1613 | 0 | value = nval; |
1614 | 0 | } |
1615 | | |
1616 | 0 | if (value != NULL) { |
1617 | 0 | ret->children = xmlNewDocText(ctxt->myDoc, value); |
1618 | 0 | if (ret->children == NULL) { |
1619 | 0 | xmlSAX2ErrMemory(ctxt); |
1620 | 0 | } else { |
1621 | 0 | ret->last = ret->children; |
1622 | 0 | ret->children->parent = (xmlNodePtr) ret; |
1623 | 0 | } |
1624 | 0 | } |
1625 | |
|
1626 | 0 | if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) && |
1627 | | /* |
1628 | | * Don't create IDs containing entity references (should |
1629 | | * be always the case with HTML) |
1630 | | */ |
1631 | 0 | (ret->children != NULL) && |
1632 | 0 | (ret->children->type == XML_TEXT_NODE) && |
1633 | 0 | (ret->children->next == NULL)) { |
1634 | 0 | int res = xmlIsID(ctxt->myDoc, ctxt->node, ret); |
1635 | |
|
1636 | 0 | if (res < 0) |
1637 | 0 | xmlCtxtErrMemory(ctxt); |
1638 | 0 | else if (res > 0) |
1639 | 0 | xmlAddID(&ctxt->vctxt, ctxt->myDoc, ret->children->content, ret); |
1640 | 0 | } |
1641 | |
|
1642 | 0 | if (nval != NULL) |
1643 | 0 | xmlFree(nval); |
1644 | 0 | } |
1645 | | |
1646 | | /** |
1647 | | * Called when an opening tag has been processed. |
1648 | | * |
1649 | | * @param ctxt parser context |
1650 | | * @param fullname The element name, including namespace prefix |
1651 | | * @param atts An array of name/value attributes pairs, NULL terminated |
1652 | | */ |
1653 | | static void |
1654 | | xmlSAX2StartHtmlElement(xmlParserCtxtPtr ctxt, const xmlChar *fullname, |
1655 | 0 | const xmlChar **atts) { |
1656 | 0 | xmlNodePtr ret; |
1657 | 0 | xmlNodePtr parent; |
1658 | 0 | const xmlChar *att; |
1659 | 0 | const xmlChar *value; |
1660 | 0 | int i; |
1661 | |
|
1662 | 0 | ret = xmlNewDocNode(ctxt->myDoc, NULL, fullname, NULL); |
1663 | 0 | if (ret == NULL) { |
1664 | 0 | xmlSAX2ErrMemory(ctxt); |
1665 | 0 | return; |
1666 | 0 | } |
1667 | 0 | ctxt->nodemem = -1; |
1668 | | |
1669 | | /* Initialize parent before pushing node */ |
1670 | 0 | parent = ctxt->node; |
1671 | 0 | if (parent == NULL) |
1672 | 0 | parent = (xmlNodePtr) ctxt->myDoc; |
1673 | | |
1674 | | /* |
1675 | | * Link the child element |
1676 | | */ |
1677 | 0 | xmlSAX2AppendChild(ctxt, ret); |
1678 | | |
1679 | | /* |
1680 | | * We are parsing a new node. |
1681 | | */ |
1682 | 0 | if (nodePush(ctxt, ret) < 0) { |
1683 | 0 | xmlUnlinkNode(ret); |
1684 | 0 | xmlFreeNode(ret); |
1685 | 0 | return; |
1686 | 0 | } |
1687 | | |
1688 | 0 | if (atts != NULL) { |
1689 | 0 | i = 0; |
1690 | 0 | att = atts[i++]; |
1691 | 0 | value = atts[i++]; |
1692 | 0 | while (att != NULL) { |
1693 | 0 | xmlSAX2HtmlAttribute(ctxt, att, value); |
1694 | 0 | att = atts[i++]; |
1695 | 0 | value = atts[i++]; |
1696 | 0 | } |
1697 | 0 | } |
1698 | 0 | } |
1699 | | #endif /* LIBXML_HTML_ENABLED */ |
1700 | | |
1701 | | /** |
1702 | | * Called when an opening tag has been processed. |
1703 | | * |
1704 | | * @deprecated Don't call this function directly. |
1705 | | * |
1706 | | * Used for HTML and SAX1. |
1707 | | * |
1708 | | * @param ctx the user data (XML parser context) |
1709 | | * @param fullname The element name, including namespace prefix |
1710 | | * @param atts An array of name/value attributes pairs, NULL terminated |
1711 | | */ |
1712 | | void |
1713 | 0 | xmlSAX2StartElement(void *ctx, const xmlChar *fullname, const xmlChar **atts) { |
1714 | 0 | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
1715 | |
|
1716 | 0 | (void) atts; |
1717 | |
|
1718 | 0 | if ((ctxt == NULL) || (fullname == NULL) || (ctxt->myDoc == NULL)) |
1719 | 0 | return; |
1720 | | |
1721 | 0 | #ifdef LIBXML_SAX1_ENABLED |
1722 | 0 | if (!ctxt->html) { |
1723 | 0 | xmlSAX1StartElement(ctxt, fullname, atts); |
1724 | 0 | return; |
1725 | 0 | } |
1726 | 0 | #endif |
1727 | | |
1728 | 0 | #ifdef LIBXML_HTML_ENABLED |
1729 | 0 | if (ctxt->html) { |
1730 | 0 | xmlSAX2StartHtmlElement(ctxt, fullname, atts); |
1731 | 0 | return; |
1732 | 0 | } |
1733 | 0 | #endif |
1734 | 0 | } |
1735 | | |
1736 | | /** |
1737 | | * called when the end of an element has been detected. |
1738 | | * |
1739 | | * @deprecated Don't call this function directly. |
1740 | | * |
1741 | | * Used for HTML and SAX1. |
1742 | | * |
1743 | | * @param ctx the user data (XML parser context) |
1744 | | * @param name The element name |
1745 | | */ |
1746 | | void |
1747 | | xmlSAX2EndElement(void *ctx, const xmlChar *name ATTRIBUTE_UNUSED) |
1748 | 0 | { |
1749 | 0 | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
1750 | |
|
1751 | 0 | if (ctxt == NULL) |
1752 | 0 | return; |
1753 | | |
1754 | 0 | #if defined(LIBXML_SAX1_ENABLED) && defined(LIBXML_VALID_ENABLED) |
1755 | 0 | if (!ctxt->html && ctxt->validate && ctxt->wellFormed && |
1756 | 0 | ctxt->myDoc && ctxt->myDoc->intSubset) |
1757 | 0 | ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc, |
1758 | 0 | ctxt->node); |
1759 | 0 | #endif /* LIBXML_VALID_ENABLED */ |
1760 | |
|
1761 | 0 | #if defined(LIBXML_SAX1_ENABLED) || defined(LIBXML_HTML_ENABLED) |
1762 | 0 | ctxt->nodemem = -1; |
1763 | | |
1764 | | /* |
1765 | | * end of parsing of this node. |
1766 | | */ |
1767 | 0 | nodePop(ctxt); |
1768 | 0 | #endif |
1769 | 0 | } |
1770 | | |
1771 | | /* |
1772 | | * @param ctxt the parser context |
1773 | | * @param doc the document |
1774 | | * @param str the input string |
1775 | | * @param len the string length |
1776 | | * |
1777 | | * Callback for a text node |
1778 | | * |
1779 | | * @returns the newly allocated string or NULL if not needed or error |
1780 | | */ |
1781 | | static xmlNodePtr |
1782 | | xmlSAX2TextNode(xmlParserCtxtPtr ctxt, xmlDocPtr doc, const xmlChar *str, |
1783 | 485k | int len) { |
1784 | 485k | xmlNodePtr ret; |
1785 | 485k | const xmlChar *intern = NULL; |
1786 | | |
1787 | | /* |
1788 | | * Allocate |
1789 | | */ |
1790 | 485k | if (ctxt->freeElems != NULL) { |
1791 | 0 | ret = ctxt->freeElems; |
1792 | 0 | ctxt->freeElems = ret->next; |
1793 | 0 | ctxt->freeElemsNr--; |
1794 | 485k | } else { |
1795 | 485k | ret = (xmlNodePtr) xmlMalloc(sizeof(xmlNode)); |
1796 | 485k | } |
1797 | 485k | if (ret == NULL) { |
1798 | 37 | xmlCtxtErrMemory(ctxt); |
1799 | 37 | return(NULL); |
1800 | 37 | } |
1801 | 485k | memset(ret, 0, sizeof(xmlNode)); |
1802 | | /* |
1803 | | * intern the formatting blanks found between tags, or the |
1804 | | * very short strings |
1805 | | */ |
1806 | 485k | if ((!ctxt->html) && (ctxt->dictNames)) { |
1807 | 364k | xmlChar cur = str[len]; |
1808 | | |
1809 | 364k | if ((len < (int) (2 * sizeof(void *))) && |
1810 | 364k | (ctxt->options & XML_PARSE_COMPACT)) { |
1811 | | /* store the string in the node overriding properties and nsDef */ |
1812 | 91.5k | xmlChar *tmp = (xmlChar *) &(ret->properties); |
1813 | 91.5k | memcpy(tmp, str, len); |
1814 | 91.5k | tmp[len] = 0; |
1815 | 91.5k | intern = tmp; |
1816 | 273k | } else if ((len <= 3) && ((cur == '"') || (cur == '\'') || |
1817 | 118k | ((cur == '<') && (str[len + 1] != '!')))) { |
1818 | 79.8k | intern = xmlDictLookup(ctxt->dict, str, len); |
1819 | 79.8k | if (intern == NULL) { |
1820 | 2 | xmlSAX2ErrMemory(ctxt); |
1821 | 2 | xmlFree(ret); |
1822 | 2 | return(NULL); |
1823 | 2 | } |
1824 | 193k | } else if (IS_BLANK_CH(*str) && (len < 60) && (cur == '<') && |
1825 | 193k | (str[len + 1] != '!')) { |
1826 | 7.07k | int i; |
1827 | | |
1828 | 30.4k | for (i = 1;i < len;i++) { |
1829 | 27.3k | if (!IS_BLANK_CH(str[i])) goto skip; |
1830 | 27.3k | } |
1831 | 3.11k | intern = xmlDictLookup(ctxt->dict, str, len); |
1832 | 3.11k | if (intern == NULL) { |
1833 | 1 | xmlSAX2ErrMemory(ctxt); |
1834 | 1 | xmlFree(ret); |
1835 | 1 | return(NULL); |
1836 | 1 | } |
1837 | 3.11k | } |
1838 | 364k | } |
1839 | 485k | skip: |
1840 | 485k | ret->type = XML_TEXT_NODE; |
1841 | 485k | ret->doc = doc; |
1842 | | |
1843 | 485k | ret->name = xmlStringText; |
1844 | 485k | if (intern == NULL) { |
1845 | 311k | ret->content = xmlStrndup(str, len); |
1846 | 311k | if (ret->content == NULL) { |
1847 | 24 | xmlSAX2ErrMemory(ctxt); |
1848 | 24 | xmlFree(ret); |
1849 | 24 | return(NULL); |
1850 | 24 | } |
1851 | 311k | } else |
1852 | 174k | ret->content = (xmlChar *) intern; |
1853 | | |
1854 | 485k | if ((xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue)) |
1855 | 0 | xmlRegisterNodeDefaultValue(ret); |
1856 | 485k | return(ret); |
1857 | 485k | } |
1858 | | |
1859 | | #ifdef LIBXML_VALID_ENABLED |
1860 | | /* |
1861 | | * @param ctxt the parser context |
1862 | | * @param str the input string |
1863 | | * @param len the string length |
1864 | | * |
1865 | | * Remove the entities from an attribute value |
1866 | | * |
1867 | | * @returns the newly allocated string or NULL if not needed or error |
1868 | | */ |
1869 | | static xmlChar * |
1870 | | xmlSAX2DecodeAttrEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, |
1871 | 0 | const xmlChar *end) { |
1872 | 0 | const xmlChar *in; |
1873 | |
|
1874 | 0 | in = str; |
1875 | 0 | while (in < end) |
1876 | 0 | if (*in++ == '&') |
1877 | 0 | goto decode; |
1878 | 0 | return(NULL); |
1879 | 0 | decode: |
1880 | | /* |
1881 | | * If the value contains '&', we can be sure it was allocated and is |
1882 | | * zero-terminated. |
1883 | | */ |
1884 | | /* TODO: normalize if needed */ |
1885 | 0 | return(xmlExpandEntitiesInAttValue(ctxt, str, /* normalize */ 0)); |
1886 | 0 | } |
1887 | | #endif /* LIBXML_VALID_ENABLED */ |
1888 | | |
1889 | | /** |
1890 | | * Handle an attribute that has been read by the parser. |
1891 | | * The default handling is to convert the attribute into an |
1892 | | * DOM subtree and past it in a new xmlAttr element added to |
1893 | | * the element. |
1894 | | * |
1895 | | * @param ctxt the parser context |
1896 | | * @param localname the local name of the attribute |
1897 | | * @param prefix the attribute namespace prefix if available |
1898 | | * @param value start of the attribute value |
1899 | | * @param valueend end of the attribute value |
1900 | | * @returns the new attribute or NULL in case of error. |
1901 | | */ |
1902 | | static xmlAttrPtr |
1903 | | xmlSAX2AttributeNs(xmlParserCtxtPtr ctxt, |
1904 | | const xmlChar * localname, |
1905 | | const xmlChar * prefix, |
1906 | | const xmlChar * value, |
1907 | | const xmlChar * valueend) |
1908 | 267k | { |
1909 | 267k | xmlAttrPtr ret; |
1910 | 267k | xmlNsPtr namespace = NULL; |
1911 | 267k | xmlChar *dup = NULL; |
1912 | | |
1913 | | /* |
1914 | | * Note: if prefix == NULL, the attribute is not in the default namespace |
1915 | | */ |
1916 | 267k | if (prefix != NULL) { |
1917 | 55.3k | namespace = xmlParserNsLookupSax(ctxt, prefix); |
1918 | 55.3k | if ((namespace == NULL) && (xmlStrEqual(prefix, BAD_CAST "xml"))) { |
1919 | 35.8k | int res; |
1920 | | |
1921 | 35.8k | res = xmlSearchNsSafe(ctxt->node, prefix, &namespace); |
1922 | 35.8k | if (res < 0) |
1923 | 14 | xmlSAX2ErrMemory(ctxt); |
1924 | 35.8k | } |
1925 | 55.3k | } |
1926 | | |
1927 | | /* |
1928 | | * allocate the node |
1929 | | */ |
1930 | 267k | if (ctxt->freeAttrs != NULL) { |
1931 | 0 | ret = ctxt->freeAttrs; |
1932 | 0 | ctxt->freeAttrs = ret->next; |
1933 | 0 | ctxt->freeAttrsNr--; |
1934 | 267k | } else { |
1935 | 267k | ret = xmlMalloc(sizeof(*ret)); |
1936 | 267k | if (ret == NULL) { |
1937 | 15 | xmlSAX2ErrMemory(ctxt); |
1938 | 15 | return(NULL); |
1939 | 15 | } |
1940 | 267k | } |
1941 | | |
1942 | 267k | memset(ret, 0, sizeof(xmlAttr)); |
1943 | 267k | ret->type = XML_ATTRIBUTE_NODE; |
1944 | | |
1945 | | /* |
1946 | | * xmlParseBalancedChunkMemoryRecover had a bug that could result in |
1947 | | * a mismatch between ctxt->node->doc and ctxt->myDoc. We use |
1948 | | * ctxt->node->doc here, but we should somehow make sure that the |
1949 | | * document pointers match. |
1950 | | */ |
1951 | | |
1952 | | /* assert(ctxt->node->doc == ctxt->myDoc); */ |
1953 | | |
1954 | 267k | ret->parent = ctxt->node; |
1955 | 267k | ret->doc = ctxt->node->doc; |
1956 | 267k | ret->ns = namespace; |
1957 | | |
1958 | 267k | if (ctxt->dictNames) { |
1959 | 202k | ret->name = localname; |
1960 | 202k | } else { |
1961 | 64.3k | ret->name = xmlStrdup(localname); |
1962 | 64.3k | if (ret->name == NULL) |
1963 | 9 | xmlSAX2ErrMemory(ctxt); |
1964 | 64.3k | } |
1965 | | |
1966 | 267k | if ((xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue)) |
1967 | 0 | xmlRegisterNodeDefaultValue((xmlNodePtr)ret); |
1968 | | |
1969 | 267k | if ((ctxt->replaceEntities == 0) && (!ctxt->html)) { |
1970 | 135k | xmlNodePtr tmp; |
1971 | | |
1972 | | /* |
1973 | | * We know that if there is an entity reference, then |
1974 | | * the string has been dup'ed and terminates with 0 |
1975 | | * otherwise with ' or " |
1976 | | */ |
1977 | 135k | if (*valueend != 0) { |
1978 | 14.3k | tmp = xmlSAX2TextNode(ctxt, ret->doc, value, valueend - value); |
1979 | 14.3k | ret->children = tmp; |
1980 | 14.3k | ret->last = tmp; |
1981 | 14.3k | if (tmp != NULL) { |
1982 | 14.3k | tmp->parent = (xmlNodePtr) ret; |
1983 | 14.3k | } |
1984 | 120k | } else if (valueend > value) { |
1985 | 120k | if (xmlNodeParseAttValue(ret->doc, ret, value, valueend - value, |
1986 | 120k | NULL) < 0) |
1987 | 41 | xmlSAX2ErrMemory(ctxt); |
1988 | 120k | } |
1989 | 135k | } else if (value != NULL) { |
1990 | 131k | xmlNodePtr tmp; |
1991 | | |
1992 | 131k | tmp = xmlSAX2TextNode(ctxt, ret->doc, value, valueend - value); |
1993 | 131k | ret->children = tmp; |
1994 | 131k | ret->last = tmp; |
1995 | 131k | if (tmp != NULL) { |
1996 | 131k | tmp->parent = (xmlNodePtr) ret; |
1997 | 131k | } |
1998 | 131k | } |
1999 | | |
2000 | 267k | #ifdef LIBXML_VALID_ENABLED |
2001 | 267k | if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed && |
2002 | 267k | ctxt->myDoc && ctxt->myDoc->intSubset) { |
2003 | | /* |
2004 | | * If we don't substitute entities, the validation should be |
2005 | | * done on a value with replaced entities anyway. |
2006 | | */ |
2007 | 0 | if (!ctxt->replaceEntities) { |
2008 | 0 | dup = xmlSAX2DecodeAttrEntities(ctxt, value, valueend); |
2009 | 0 | if (dup == NULL) { |
2010 | 0 | if (*valueend == 0) { |
2011 | 0 | ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, |
2012 | 0 | ctxt->myDoc, ctxt->node, ret, value); |
2013 | 0 | } else { |
2014 | | /* |
2015 | | * That should already be normalized. |
2016 | | * cheaper to finally allocate here than duplicate |
2017 | | * entry points in the full validation code |
2018 | | */ |
2019 | 0 | dup = xmlStrndup(value, valueend - value); |
2020 | 0 | if (dup == NULL) |
2021 | 0 | xmlSAX2ErrMemory(ctxt); |
2022 | |
|
2023 | 0 | ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, |
2024 | 0 | ctxt->myDoc, ctxt->node, ret, dup); |
2025 | 0 | } |
2026 | 0 | } else { |
2027 | | /* |
2028 | | * dup now contains a string of the flattened attribute |
2029 | | * content with entities substituted. Check if we need to |
2030 | | * apply an extra layer of normalization. |
2031 | | * It need to be done twice ... it's an extra burden related |
2032 | | * to the ability to keep references in attributes |
2033 | | */ |
2034 | 0 | if (ctxt->attsSpecial != NULL) { |
2035 | 0 | xmlChar *nvalnorm; |
2036 | 0 | xmlChar fn[50]; |
2037 | 0 | xmlChar *fullname; |
2038 | |
|
2039 | 0 | fullname = xmlBuildQName(localname, prefix, fn, 50); |
2040 | 0 | if (fullname == NULL) { |
2041 | 0 | xmlSAX2ErrMemory(ctxt); |
2042 | 0 | } else { |
2043 | 0 | ctxt->vctxt.valid = 1; |
2044 | 0 | nvalnorm = xmlValidCtxtNormalizeAttributeValue( |
2045 | 0 | &ctxt->vctxt, ctxt->myDoc, |
2046 | 0 | ctxt->node, fullname, dup); |
2047 | 0 | if (ctxt->vctxt.valid != 1) |
2048 | 0 | ctxt->valid = 0; |
2049 | |
|
2050 | 0 | if ((fullname != fn) && (fullname != localname)) |
2051 | 0 | xmlFree(fullname); |
2052 | 0 | if (nvalnorm != NULL) { |
2053 | 0 | xmlFree(dup); |
2054 | 0 | dup = nvalnorm; |
2055 | 0 | } |
2056 | 0 | } |
2057 | 0 | } |
2058 | |
|
2059 | 0 | ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, |
2060 | 0 | ctxt->myDoc, ctxt->node, ret, dup); |
2061 | 0 | } |
2062 | 0 | } else { |
2063 | | /* |
2064 | | * if entities already have been substituted, then |
2065 | | * the attribute as passed is already normalized |
2066 | | */ |
2067 | 0 | dup = xmlStrndup(value, valueend - value); |
2068 | 0 | if (dup == NULL) |
2069 | 0 | xmlSAX2ErrMemory(ctxt); |
2070 | | |
2071 | | /* |
2072 | | * When replacing entities, make sure that IDs in |
2073 | | * entities aren't registered. This also shouldn't be |
2074 | | * done when entities aren't replaced, but this would |
2075 | | * require to rework IDREF checks. |
2076 | | */ |
2077 | 0 | if (ctxt->input->entity != NULL) |
2078 | 0 | ctxt->vctxt.flags |= XML_VCTXT_IN_ENTITY; |
2079 | |
|
2080 | 0 | ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, |
2081 | 0 | ctxt->myDoc, ctxt->node, ret, dup); |
2082 | |
|
2083 | 0 | ctxt->vctxt.flags &= ~XML_VCTXT_IN_ENTITY; |
2084 | 0 | } |
2085 | 0 | } else |
2086 | 267k | #endif /* LIBXML_VALID_ENABLED */ |
2087 | 267k | if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) && |
2088 | 267k | (ctxt->input->entity == NULL) && |
2089 | | /* Don't create IDs containing entity references */ |
2090 | 267k | (ret->children != NULL) && |
2091 | 267k | (ret->children->type == XML_TEXT_NODE) && |
2092 | 267k | (ret->children->next == NULL)) { |
2093 | 108k | xmlChar *content = ret->children->content; |
2094 | | /* |
2095 | | * when validating, the ID registration is done at the attribute |
2096 | | * validation level. Otherwise we have to do specific handling here. |
2097 | | */ |
2098 | 108k | if ((prefix == ctxt->str_xml) && |
2099 | 108k | (localname[0] == 'i') && (localname[1] == 'd') && |
2100 | 108k | (localname[2] == 0)) { |
2101 | | /* |
2102 | | * Add the xml:id value |
2103 | | * |
2104 | | * Open issue: normalization of the value. |
2105 | | */ |
2106 | 2.84k | if (xmlValidateNCName(content, 1) != 0) { |
2107 | 2.36k | xmlErrId(ctxt, XML_DTD_XMLID_VALUE, |
2108 | 2.36k | "xml:id : attribute value %s is not an NCName\n", |
2109 | 2.36k | content); |
2110 | 2.36k | } |
2111 | 2.84k | xmlAddID(&ctxt->vctxt, ctxt->myDoc, content, ret); |
2112 | 105k | } else { |
2113 | 105k | int res = xmlIsID(ctxt->myDoc, ctxt->node, ret); |
2114 | | |
2115 | 105k | if (res < 0) |
2116 | 2 | xmlCtxtErrMemory(ctxt); |
2117 | 105k | else if (res > 0) |
2118 | 2.02k | xmlAddID(&ctxt->vctxt, ctxt->myDoc, content, ret); |
2119 | 103k | else if (xmlIsRef(ctxt->myDoc, ctxt->node, ret)) |
2120 | 36.8k | xmlAddRef(&ctxt->vctxt, ctxt->myDoc, content, ret); |
2121 | 105k | } |
2122 | 108k | } |
2123 | 267k | if (dup != NULL) |
2124 | 0 | xmlFree(dup); |
2125 | | |
2126 | 267k | return(ret); |
2127 | 267k | } |
2128 | | |
2129 | | /** |
2130 | | * SAX2 callback when an element start has been detected by the parser. |
2131 | | * It provides the namespace information for the element, as well as |
2132 | | * the new namespace declarations on the element. |
2133 | | * |
2134 | | * @param ctx the user data (XML parser context) |
2135 | | * @param localname the local name of the element |
2136 | | * @param prefix the element namespace prefix if available |
2137 | | * @param URI the element namespace name if available |
2138 | | * @param nb_namespaces number of namespace definitions on that node |
2139 | | * @param namespaces pointer to the array of prefix/URI pairs namespace definitions |
2140 | | * @param nb_attributes the number of attributes on that node |
2141 | | * @param nb_defaulted the number of defaulted attributes. |
2142 | | * @param attributes pointer to the array of (localname/prefix/URI/value/end) |
2143 | | * attribute values. |
2144 | | */ |
2145 | | void |
2146 | | xmlSAX2StartElementNs(void *ctx, |
2147 | | const xmlChar *localname, |
2148 | | const xmlChar *prefix, |
2149 | | const xmlChar *URI, |
2150 | | int nb_namespaces, |
2151 | | const xmlChar **namespaces, |
2152 | | int nb_attributes, |
2153 | | int nb_defaulted, |
2154 | | const xmlChar **attributes) |
2155 | 1.13M | { |
2156 | 1.13M | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
2157 | 1.13M | xmlNodePtr ret; |
2158 | 1.13M | xmlNsPtr last = NULL, ns; |
2159 | 1.13M | const xmlChar *uri, *pref; |
2160 | 1.13M | xmlChar *lname = NULL; |
2161 | 1.13M | int i, j; |
2162 | | |
2163 | 1.13M | if (ctx == NULL) return; |
2164 | | |
2165 | 1.13M | #ifdef LIBXML_VALID_ENABLED |
2166 | | /* |
2167 | | * First check on validity: |
2168 | | */ |
2169 | 1.13M | if (ctxt->validate && |
2170 | 1.13M | ((ctxt->myDoc == NULL) || |
2171 | 0 | ((ctxt->myDoc->extSubset == NULL) && |
2172 | 0 | ((ctxt->myDoc->intSubset == NULL) || |
2173 | 0 | ((ctxt->myDoc->intSubset->notations == NULL) && |
2174 | 0 | (ctxt->myDoc->intSubset->elements == NULL) && |
2175 | 0 | (ctxt->myDoc->intSubset->attributes == NULL) && |
2176 | 0 | (ctxt->myDoc->intSubset->entities == NULL)))))) { |
2177 | 0 | xmlErrValid(ctxt, XML_DTD_NO_DTD, |
2178 | 0 | "Validation failed: no DTD found !", NULL, NULL); |
2179 | 0 | ctxt->validate = 0; |
2180 | 0 | } |
2181 | 1.13M | #endif /* LIBXML_VALID_ENABLED */ |
2182 | | |
2183 | | /* |
2184 | | * Take care of the rare case of an undefined namespace prefix |
2185 | | */ |
2186 | 1.13M | if ((prefix != NULL) && (URI == NULL)) { |
2187 | 18.0k | if (ctxt->dictNames) { |
2188 | 11.9k | const xmlChar *fullname; |
2189 | | |
2190 | 11.9k | fullname = xmlDictQLookup(ctxt->dict, prefix, localname); |
2191 | 11.9k | if (fullname == NULL) { |
2192 | 2 | xmlSAX2ErrMemory(ctxt); |
2193 | 2 | return; |
2194 | 2 | } |
2195 | 11.9k | localname = fullname; |
2196 | 11.9k | } else { |
2197 | 6.10k | lname = xmlBuildQName(localname, prefix, NULL, 0); |
2198 | 6.10k | if (lname == NULL) { |
2199 | 1 | xmlSAX2ErrMemory(ctxt); |
2200 | 1 | return; |
2201 | 1 | } |
2202 | 6.10k | } |
2203 | 18.0k | } |
2204 | | /* |
2205 | | * allocate the node |
2206 | | */ |
2207 | 1.13M | if (ctxt->freeElems != NULL) { |
2208 | 0 | ret = ctxt->freeElems; |
2209 | 0 | ctxt->freeElems = ret->next; |
2210 | 0 | ctxt->freeElemsNr--; |
2211 | 0 | memset(ret, 0, sizeof(xmlNode)); |
2212 | 0 | ret->doc = ctxt->myDoc; |
2213 | 0 | ret->type = XML_ELEMENT_NODE; |
2214 | |
|
2215 | 0 | if (ctxt->dictNames) |
2216 | 0 | ret->name = localname; |
2217 | 0 | else { |
2218 | 0 | if (lname == NULL) |
2219 | 0 | ret->name = xmlStrdup(localname); |
2220 | 0 | else |
2221 | 0 | ret->name = lname; |
2222 | 0 | if (ret->name == NULL) { |
2223 | 0 | xmlSAX2ErrMemory(ctxt); |
2224 | 0 | xmlFree(ret); |
2225 | 0 | return; |
2226 | 0 | } |
2227 | 0 | } |
2228 | 0 | if ((xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue)) |
2229 | 0 | xmlRegisterNodeDefaultValue(ret); |
2230 | 1.13M | } else { |
2231 | 1.13M | if (ctxt->dictNames) |
2232 | 812k | ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, |
2233 | 812k | (xmlChar *) localname, NULL); |
2234 | 323k | else if (lname == NULL) |
2235 | 317k | ret = xmlNewDocNode(ctxt->myDoc, NULL, localname, NULL); |
2236 | 6.10k | else |
2237 | 6.10k | ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, lname, NULL); |
2238 | 1.13M | if (ret == NULL) { |
2239 | 46 | xmlSAX2ErrMemory(ctxt); |
2240 | 46 | return; |
2241 | 46 | } |
2242 | 1.13M | } |
2243 | | |
2244 | | /* |
2245 | | * Build the namespace list |
2246 | | */ |
2247 | 1.23M | for (i = 0,j = 0;j < nb_namespaces;j++) { |
2248 | 99.3k | pref = namespaces[i++]; |
2249 | 99.3k | uri = namespaces[i++]; |
2250 | 99.3k | ns = xmlNewNs(NULL, uri, pref); |
2251 | 99.3k | if (ns != NULL) { |
2252 | 99.3k | if (last == NULL) { |
2253 | 55.7k | ret->nsDef = last = ns; |
2254 | 55.7k | } else { |
2255 | 43.5k | last->next = ns; |
2256 | 43.5k | last = ns; |
2257 | 43.5k | } |
2258 | 99.3k | if ((URI != NULL) && (prefix == pref)) |
2259 | 12.3k | ret->ns = ns; |
2260 | 99.3k | } else { |
2261 | 45 | xmlSAX2ErrMemory(ctxt); |
2262 | 45 | continue; |
2263 | 45 | } |
2264 | | |
2265 | 99.3k | xmlParserNsUpdateSax(ctxt, pref, ns); |
2266 | | |
2267 | 99.3k | #ifdef LIBXML_VALID_ENABLED |
2268 | 99.3k | if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed && |
2269 | 99.3k | ctxt->myDoc && ctxt->myDoc->intSubset) { |
2270 | 0 | ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc, |
2271 | 0 | ret, prefix, ns, uri); |
2272 | 0 | } |
2273 | 99.3k | #endif /* LIBXML_VALID_ENABLED */ |
2274 | 99.3k | } |
2275 | 1.13M | ctxt->nodemem = -1; |
2276 | | |
2277 | | /* |
2278 | | * Link the child element |
2279 | | */ |
2280 | 1.13M | xmlSAX2AppendChild(ctxt, ret); |
2281 | | |
2282 | | /* |
2283 | | * We are parsing a new node. |
2284 | | */ |
2285 | 1.13M | if (nodePush(ctxt, ret) < 0) { |
2286 | 20 | xmlUnlinkNode(ret); |
2287 | 20 | xmlFreeNode(ret); |
2288 | 20 | return; |
2289 | 20 | } |
2290 | | |
2291 | | /* |
2292 | | * Insert the defaulted attributes from the DTD only if requested: |
2293 | | */ |
2294 | 1.13M | if ((nb_defaulted != 0) && |
2295 | 1.13M | ((ctxt->loadsubset & XML_COMPLETE_ATTRS) == 0)) |
2296 | 3.57k | nb_attributes -= nb_defaulted; |
2297 | | |
2298 | | /* |
2299 | | * Search the namespace if it wasn't already found |
2300 | | * Note that, if prefix is NULL, this searches for the default Ns |
2301 | | */ |
2302 | 1.13M | if ((URI != NULL) && (ret->ns == NULL)) { |
2303 | 581k | ret->ns = xmlParserNsLookupSax(ctxt, prefix); |
2304 | 581k | if ((ret->ns == NULL) && (xmlStrEqual(prefix, BAD_CAST "xml"))) { |
2305 | 570 | int res; |
2306 | | |
2307 | 570 | res = xmlSearchNsSafe(ret, prefix, &ret->ns); |
2308 | 570 | if (res < 0) |
2309 | 3 | xmlSAX2ErrMemory(ctxt); |
2310 | 570 | } |
2311 | 581k | if (ret->ns == NULL) { |
2312 | 24 | ns = xmlNewNs(ret, NULL, prefix); |
2313 | 24 | if (ns == NULL) { |
2314 | |
|
2315 | 0 | xmlSAX2ErrMemory(ctxt); |
2316 | 0 | return; |
2317 | 0 | } |
2318 | 24 | if (prefix != NULL) |
2319 | 5 | xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
2320 | 5 | "Namespace prefix %s was not found\n", |
2321 | 5 | prefix, NULL); |
2322 | 19 | else |
2323 | 19 | xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
2324 | 19 | "Namespace default prefix was not found\n", |
2325 | 19 | NULL, NULL); |
2326 | 24 | } |
2327 | 581k | } |
2328 | | |
2329 | | /* |
2330 | | * process all the other attributes |
2331 | | */ |
2332 | 1.13M | if (nb_attributes > 0) { |
2333 | 136k | xmlAttrPtr prev = NULL; |
2334 | | |
2335 | 403k | for (j = 0,i = 0;i < nb_attributes;i++,j+=5) { |
2336 | 267k | xmlAttrPtr attr = NULL; |
2337 | | |
2338 | | /* |
2339 | | * Handle the rare case of an undefined attribute prefix |
2340 | | */ |
2341 | 267k | if ((attributes[j+1] != NULL) && (attributes[j+2] == NULL)) { |
2342 | 29.0k | if (ctxt->dictNames) { |
2343 | 25.3k | const xmlChar *fullname; |
2344 | | |
2345 | 25.3k | fullname = xmlDictQLookup(ctxt->dict, attributes[j+1], |
2346 | 25.3k | attributes[j]); |
2347 | 25.3k | if (fullname == NULL) { |
2348 | 1 | xmlSAX2ErrMemory(ctxt); |
2349 | 1 | return; |
2350 | 1 | } |
2351 | 25.3k | attr = xmlSAX2AttributeNs(ctxt, fullname, NULL, |
2352 | 25.3k | attributes[j+3], |
2353 | 25.3k | attributes[j+4]); |
2354 | 25.3k | goto have_attr; |
2355 | 25.3k | } else { |
2356 | 3.63k | lname = xmlBuildQName(attributes[j], attributes[j+1], |
2357 | 3.63k | NULL, 0); |
2358 | 3.63k | if (lname == NULL) { |
2359 | 1 | xmlSAX2ErrMemory(ctxt); |
2360 | 1 | return; |
2361 | 1 | } |
2362 | 3.63k | attr = xmlSAX2AttributeNs(ctxt, lname, NULL, |
2363 | 3.63k | attributes[j+3], |
2364 | 3.63k | attributes[j+4]); |
2365 | 3.63k | xmlFree(lname); |
2366 | 3.63k | goto have_attr; |
2367 | 3.63k | } |
2368 | 29.0k | } |
2369 | 238k | attr = xmlSAX2AttributeNs(ctxt, attributes[j], attributes[j+1], |
2370 | 238k | attributes[j+3], attributes[j+4]); |
2371 | 267k | have_attr: |
2372 | 267k | if (attr == NULL) |
2373 | 15 | continue; |
2374 | | |
2375 | | /* link at the end to preserve order */ |
2376 | 267k | if (prev == NULL) { |
2377 | 136k | ctxt->node->properties = attr; |
2378 | 136k | } else { |
2379 | 130k | prev->next = attr; |
2380 | 130k | attr->prev = prev; |
2381 | 130k | } |
2382 | | |
2383 | 267k | prev = attr; |
2384 | 267k | } |
2385 | 136k | } |
2386 | | |
2387 | 1.13M | #ifdef LIBXML_VALID_ENABLED |
2388 | | /* |
2389 | | * If it's the Document root, finish the DTD validation and |
2390 | | * check the document root element for validity |
2391 | | */ |
2392 | 1.13M | if ((ctxt->validate) && |
2393 | 1.13M | ((ctxt->vctxt.flags & XML_VCTXT_DTD_VALIDATED) == 0)) { |
2394 | 0 | int chk; |
2395 | |
|
2396 | 0 | chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc); |
2397 | 0 | if (chk <= 0) |
2398 | 0 | ctxt->valid = 0; |
2399 | 0 | if (chk < 0) |
2400 | 0 | ctxt->wellFormed = 0; |
2401 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
2402 | 0 | ctxt->vctxt.flags |= XML_VCTXT_DTD_VALIDATED; |
2403 | 0 | } |
2404 | 1.13M | #endif /* LIBXML_VALID_ENABLED */ |
2405 | 1.13M | } |
2406 | | |
2407 | | /** |
2408 | | * SAX2 callback when an element end has been detected by the parser. |
2409 | | * It provides the namespace information for the element. |
2410 | | * |
2411 | | * @param ctx the user data (XML parser context) |
2412 | | * @param localname the local name of the element |
2413 | | * @param prefix the element namespace prefix if available |
2414 | | * @param URI the element namespace name if available |
2415 | | */ |
2416 | | void |
2417 | | xmlSAX2EndElementNs(void *ctx, |
2418 | | const xmlChar * localname ATTRIBUTE_UNUSED, |
2419 | | const xmlChar * prefix ATTRIBUTE_UNUSED, |
2420 | | const xmlChar * URI ATTRIBUTE_UNUSED) |
2421 | 97.7k | { |
2422 | 97.7k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
2423 | | |
2424 | 97.7k | if (ctx == NULL) return; |
2425 | 97.7k | ctxt->nodemem = -1; |
2426 | | |
2427 | 97.7k | #ifdef LIBXML_VALID_ENABLED |
2428 | 97.7k | if (ctxt->validate && ctxt->wellFormed && |
2429 | 97.7k | ctxt->myDoc && ctxt->myDoc->intSubset) |
2430 | 0 | ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc, |
2431 | 0 | ctxt->node); |
2432 | 97.7k | #endif /* LIBXML_VALID_ENABLED */ |
2433 | | |
2434 | | /* |
2435 | | * end of parsing of this node. |
2436 | | */ |
2437 | 97.7k | nodePop(ctxt); |
2438 | 97.7k | } |
2439 | | |
2440 | | /** |
2441 | | * called when an entity #xmlSAX2Reference is detected. |
2442 | | * |
2443 | | * @param ctx the user data (XML parser context) |
2444 | | * @param name The entity name |
2445 | | */ |
2446 | | void |
2447 | | xmlSAX2Reference(void *ctx, const xmlChar *name) |
2448 | 24.5k | { |
2449 | 24.5k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
2450 | 24.5k | xmlNodePtr ret; |
2451 | | |
2452 | 24.5k | if (ctx == NULL) return; |
2453 | 24.5k | ret = xmlNewReference(ctxt->myDoc, name); |
2454 | 24.5k | if (ret == NULL) { |
2455 | 4 | xmlSAX2ErrMemory(ctxt); |
2456 | 4 | return; |
2457 | 4 | } |
2458 | | |
2459 | 24.5k | xmlSAX2AppendChild(ctxt, ret); |
2460 | 24.5k | } |
2461 | | |
2462 | | /** |
2463 | | * Append characters. |
2464 | | * |
2465 | | * @param ctxt the parser context |
2466 | | * @param ch a xmlChar string |
2467 | | * @param len the number of xmlChar |
2468 | | * @param type text or cdata |
2469 | | */ |
2470 | | static void |
2471 | | xmlSAX2Text(xmlParserCtxtPtr ctxt, const xmlChar *ch, int len, |
2472 | | xmlElementType type) |
2473 | 930k | { |
2474 | 930k | xmlNodePtr lastChild; |
2475 | 930k | xmlNodePtr parent; |
2476 | | |
2477 | 930k | if (ctxt == NULL) |
2478 | 0 | return; |
2479 | | |
2480 | 930k | parent = ctxt->node; |
2481 | 930k | if (parent == NULL) |
2482 | 0 | return; |
2483 | 930k | lastChild = parent->last; |
2484 | | |
2485 | | /* |
2486 | | * Try to merge with previous text node using size and capacity |
2487 | | * stored in the parser context to avoid naive concatenation. |
2488 | | * |
2489 | | * Don't merge CDATA sections. In HTML mode, CDATA is used for |
2490 | | * raw text which should be merged. |
2491 | | */ |
2492 | 930k | if ((lastChild == NULL) || |
2493 | 930k | (lastChild->type != type) || |
2494 | 930k | ((!ctxt->html) && (type != XML_TEXT_NODE))) { |
2495 | 352k | xmlNode *node; |
2496 | | |
2497 | 352k | if (type == XML_TEXT_NODE) |
2498 | 339k | node = xmlSAX2TextNode(ctxt, parent->doc, ch, len); |
2499 | 13.0k | else |
2500 | 13.0k | node = xmlNewCDataBlock(parent->doc, ch, len); |
2501 | 352k | if (node == NULL) { |
2502 | 41 | xmlSAX2ErrMemory(ctxt); |
2503 | 41 | return; |
2504 | 41 | } |
2505 | | |
2506 | 352k | if (lastChild == NULL) { |
2507 | 76.2k | parent->children = node; |
2508 | 76.2k | parent->last = node; |
2509 | 76.2k | node->parent = parent; |
2510 | 276k | } else { |
2511 | 276k | xmlSAX2AppendChild(ctxt, node); |
2512 | 276k | } |
2513 | | |
2514 | 352k | ctxt->nodelen = len; |
2515 | 352k | ctxt->nodemem = len + 1; |
2516 | 352k | lastChild = node; |
2517 | 577k | } else { |
2518 | 577k | xmlChar *content; |
2519 | 577k | int oldSize, newSize, capacity; |
2520 | 577k | int maxSize = (ctxt->options & XML_PARSE_HUGE) ? |
2521 | 224k | XML_MAX_HUGE_LENGTH : |
2522 | 577k | XML_MAX_TEXT_LENGTH; |
2523 | | |
2524 | 577k | content = lastChild->content; |
2525 | 577k | oldSize = ctxt->nodelen; |
2526 | 577k | capacity = ctxt->nodemem; |
2527 | | |
2528 | | /* Shouldn't happen */ |
2529 | 577k | if ((content == NULL) || (capacity <= 0)) { |
2530 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
2531 | 0 | "xmlSAX2Text: no content"); |
2532 | 0 | return; |
2533 | 0 | } |
2534 | | |
2535 | 577k | if ((len > maxSize) || (oldSize > maxSize - len)) { |
2536 | 0 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, |
2537 | 0 | "Text node too long, try XML_PARSE_HUGE"); |
2538 | 0 | return; |
2539 | 0 | } |
2540 | | |
2541 | 577k | newSize = oldSize + len; |
2542 | | |
2543 | 577k | if (newSize >= capacity) { |
2544 | 169k | if (newSize <= 20) |
2545 | 75.0k | capacity = 40; |
2546 | 94.6k | else |
2547 | 94.6k | capacity = newSize > INT_MAX / 2 ? INT_MAX : newSize * 2; |
2548 | | |
2549 | | /* |
2550 | | * If the content was stored in properties or in |
2551 | | * the dictionary, don't realloc. |
2552 | | */ |
2553 | 169k | if ((content == (xmlChar *) &lastChild->properties) || |
2554 | 169k | ((ctxt->nodemem == oldSize + 1) && |
2555 | 133k | (xmlDictOwns(ctxt->dict, content)))) { |
2556 | 40.7k | xmlChar *newContent; |
2557 | | |
2558 | 40.7k | newContent = xmlMalloc(capacity); |
2559 | 40.7k | if (newContent == NULL) { |
2560 | 3 | xmlSAX2ErrMemory(ctxt); |
2561 | 3 | return; |
2562 | 3 | } |
2563 | | |
2564 | 40.7k | memcpy(newContent, content, oldSize); |
2565 | 40.7k | lastChild->properties = NULL; |
2566 | 40.7k | content = newContent; |
2567 | 128k | } else { |
2568 | 128k | content = xmlRealloc(content, capacity); |
2569 | 128k | if (content == NULL) { |
2570 | 9 | xmlSAX2ErrMemory(ctxt); |
2571 | 9 | return; |
2572 | 9 | } |
2573 | 128k | } |
2574 | | |
2575 | 169k | ctxt->nodemem = capacity; |
2576 | 169k | lastChild->content = content; |
2577 | 169k | } |
2578 | | |
2579 | 577k | memcpy(&content[oldSize], ch, len); |
2580 | 577k | content[newSize] = 0; |
2581 | 577k | ctxt->nodelen = newSize; |
2582 | 577k | } |
2583 | | |
2584 | 930k | if ((lastChild != NULL) && |
2585 | 930k | (type == XML_TEXT_NODE) && |
2586 | 930k | (ctxt->input != NULL)) { |
2587 | 917k | if ((unsigned) ctxt->input->line < (unsigned) USHRT_MAX) |
2588 | 889k | lastChild->line = ctxt->input->line; |
2589 | 27.9k | else { |
2590 | 27.9k | lastChild->line = USHRT_MAX; |
2591 | 27.9k | if (ctxt->options & XML_PARSE_BIG_LINES) |
2592 | 5.47k | lastChild->psvi = XML_INT_TO_PTR(ctxt->input->line); |
2593 | 27.9k | } |
2594 | 917k | } |
2595 | 930k | } |
2596 | | |
2597 | | /** |
2598 | | * receiving some chars from the parser. |
2599 | | * |
2600 | | * @param ctx the user data (XML parser context) |
2601 | | * @param ch a xmlChar string |
2602 | | * @param len the number of xmlChar |
2603 | | */ |
2604 | | void |
2605 | | xmlSAX2Characters(void *ctx, const xmlChar *ch, int len) |
2606 | 917k | { |
2607 | 917k | xmlSAX2Text((xmlParserCtxtPtr) ctx, ch, len, XML_TEXT_NODE); |
2608 | 917k | } |
2609 | | |
2610 | | /** |
2611 | | * receiving some ignorable whitespaces from the parser. |
2612 | | * UNUSED: by default the DOM building will use #xmlSAX2Characters |
2613 | | * |
2614 | | * @param ctx the user data (XML parser context) |
2615 | | * @param ch a xmlChar string |
2616 | | * @param len the number of xmlChar |
2617 | | */ |
2618 | | void |
2619 | | xmlSAX2IgnorableWhitespace(void *ctx ATTRIBUTE_UNUSED, const xmlChar *ch ATTRIBUTE_UNUSED, int len ATTRIBUTE_UNUSED) |
2620 | 0 | { |
2621 | 0 | } |
2622 | | |
2623 | | /** |
2624 | | * A processing instruction has been parsed. |
2625 | | * |
2626 | | * @param ctx the user data (XML parser context) |
2627 | | * @param target the target name |
2628 | | * @param data the PI data's |
2629 | | */ |
2630 | | void |
2631 | | xmlSAX2ProcessingInstruction(void *ctx, const xmlChar *target, |
2632 | | const xmlChar *data) |
2633 | 18.5k | { |
2634 | 18.5k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
2635 | 18.5k | xmlNodePtr ret; |
2636 | | |
2637 | 18.5k | if (ctx == NULL) return; |
2638 | | |
2639 | 18.5k | ret = xmlNewDocPI(ctxt->myDoc, target, data); |
2640 | 18.5k | if (ret == NULL) { |
2641 | 7 | xmlSAX2ErrMemory(ctxt); |
2642 | 7 | return; |
2643 | 7 | } |
2644 | | |
2645 | 18.5k | xmlSAX2AppendChild(ctxt, ret); |
2646 | 18.5k | } |
2647 | | |
2648 | | /** |
2649 | | * A #xmlSAX2Comment has been parsed. |
2650 | | * |
2651 | | * @param ctx the user data (XML parser context) |
2652 | | * @param value the #xmlSAX2Comment content |
2653 | | */ |
2654 | | void |
2655 | | xmlSAX2Comment(void *ctx, const xmlChar *value) |
2656 | 401k | { |
2657 | 401k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; |
2658 | 401k | xmlNodePtr ret; |
2659 | | |
2660 | 401k | if (ctx == NULL) return; |
2661 | | |
2662 | 401k | ret = xmlNewDocComment(ctxt->myDoc, value); |
2663 | 401k | if (ret == NULL) { |
2664 | 13 | xmlSAX2ErrMemory(ctxt); |
2665 | 13 | return; |
2666 | 13 | } |
2667 | | |
2668 | 401k | xmlSAX2AppendChild(ctxt, ret); |
2669 | 401k | } |
2670 | | |
2671 | | /** |
2672 | | * called when a pcdata block has been parsed |
2673 | | * |
2674 | | * @param ctx the user data (XML parser context) |
2675 | | * @param value The pcdata content |
2676 | | * @param len the block length |
2677 | | */ |
2678 | | void |
2679 | | xmlSAX2CDataBlock(void *ctx, const xmlChar *value, int len) |
2680 | 13.0k | { |
2681 | 13.0k | xmlSAX2Text((xmlParserCtxtPtr) ctx, value, len, XML_CDATA_SECTION_NODE); |
2682 | 13.0k | } |
2683 | | |
2684 | | #ifdef LIBXML_SAX1_ENABLED |
2685 | | /** |
2686 | | * Has no effect. |
2687 | | * |
2688 | | * @deprecated Use parser option XML_PARSE_SAX1. |
2689 | | * |
2690 | | * @param version the version, must be 2 |
2691 | | * @returns 2 in case of success and -1 in case of error. |
2692 | | */ |
2693 | | int |
2694 | | xmlSAXDefaultVersion(int version) |
2695 | 0 | { |
2696 | 0 | if (version != 2) |
2697 | 0 | return(-1); |
2698 | 0 | return(2); |
2699 | 0 | } |
2700 | | #endif /* LIBXML_SAX1_ENABLED */ |
2701 | | |
2702 | | /** |
2703 | | * Initialize the default XML SAX handler according to the version |
2704 | | * |
2705 | | * @param hdlr the SAX handler |
2706 | | * @param version the version, 1 or 2 |
2707 | | * @returns 0 in case of success and -1 in case of error. |
2708 | | */ |
2709 | | int |
2710 | | xmlSAXVersion(xmlSAXHandler *hdlr, int version) |
2711 | 34.5k | { |
2712 | 34.5k | if (hdlr == NULL) return(-1); |
2713 | 34.5k | if (version == 2) { |
2714 | 34.5k | hdlr->startElementNs = xmlSAX2StartElementNs; |
2715 | 34.5k | hdlr->endElementNs = xmlSAX2EndElementNs; |
2716 | 34.5k | hdlr->serror = NULL; |
2717 | 34.5k | hdlr->initialized = XML_SAX2_MAGIC; |
2718 | 34.5k | #ifdef LIBXML_SAX1_ENABLED |
2719 | 34.5k | } else if (version == 1) { |
2720 | 0 | hdlr->initialized = 1; |
2721 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
2722 | 0 | } else |
2723 | 0 | return(-1); |
2724 | 34.5k | #ifdef LIBXML_SAX1_ENABLED |
2725 | 34.5k | hdlr->startElement = xmlSAX2StartElement; |
2726 | 34.5k | hdlr->endElement = xmlSAX2EndElement; |
2727 | | #else |
2728 | | hdlr->startElement = NULL; |
2729 | | hdlr->endElement = NULL; |
2730 | | #endif /* LIBXML_SAX1_ENABLED */ |
2731 | 34.5k | hdlr->internalSubset = xmlSAX2InternalSubset; |
2732 | 34.5k | hdlr->externalSubset = xmlSAX2ExternalSubset; |
2733 | 34.5k | hdlr->isStandalone = xmlSAX2IsStandalone; |
2734 | 34.5k | hdlr->hasInternalSubset = xmlSAX2HasInternalSubset; |
2735 | 34.5k | hdlr->hasExternalSubset = xmlSAX2HasExternalSubset; |
2736 | 34.5k | hdlr->resolveEntity = xmlSAX2ResolveEntity; |
2737 | 34.5k | hdlr->getEntity = xmlSAX2GetEntity; |
2738 | 34.5k | hdlr->getParameterEntity = xmlSAX2GetParameterEntity; |
2739 | 34.5k | hdlr->entityDecl = xmlSAX2EntityDecl; |
2740 | 34.5k | hdlr->attributeDecl = xmlSAX2AttributeDecl; |
2741 | 34.5k | hdlr->elementDecl = xmlSAX2ElementDecl; |
2742 | 34.5k | hdlr->notationDecl = xmlSAX2NotationDecl; |
2743 | 34.5k | hdlr->unparsedEntityDecl = xmlSAX2UnparsedEntityDecl; |
2744 | 34.5k | hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator; |
2745 | 34.5k | hdlr->startDocument = xmlSAX2StartDocument; |
2746 | 34.5k | hdlr->endDocument = xmlSAX2EndDocument; |
2747 | 34.5k | hdlr->reference = xmlSAX2Reference; |
2748 | 34.5k | hdlr->characters = xmlSAX2Characters; |
2749 | 34.5k | hdlr->cdataBlock = xmlSAX2CDataBlock; |
2750 | 34.5k | hdlr->ignorableWhitespace = xmlSAX2Characters; |
2751 | 34.5k | hdlr->processingInstruction = xmlSAX2ProcessingInstruction; |
2752 | 34.5k | hdlr->comment = xmlSAX2Comment; |
2753 | 34.5k | hdlr->warning = xmlParserWarning; |
2754 | 34.5k | hdlr->error = xmlParserError; |
2755 | 34.5k | hdlr->fatalError = xmlParserError; |
2756 | | |
2757 | 34.5k | return(0); |
2758 | 34.5k | } |
2759 | | |
2760 | | /** |
2761 | | * Initialize the default XML SAX2 handler |
2762 | | * |
2763 | | * @param hdlr the SAX handler |
2764 | | * @param warning flag if non-zero sets the handler warning procedure |
2765 | | */ |
2766 | | void |
2767 | | xmlSAX2InitDefaultSAXHandler(xmlSAXHandler *hdlr, int warning) |
2768 | 0 | { |
2769 | 0 | if ((hdlr == NULL) || (hdlr->initialized != 0)) |
2770 | 0 | return; |
2771 | | |
2772 | 0 | xmlSAXVersion(hdlr, 2); |
2773 | 0 | if (warning == 0) |
2774 | 0 | hdlr->warning = NULL; |
2775 | 0 | } |
2776 | | |
2777 | | /** |
2778 | | * Initialize the default SAX2 handler |
2779 | | * |
2780 | | * @deprecated This function is a no-op. Call #xmlInitParser to |
2781 | | * initialize the library. |
2782 | | * |
2783 | | */ |
2784 | | void |
2785 | | xmlDefaultSAXHandlerInit(void) |
2786 | 0 | { |
2787 | 0 | } |
2788 | | |
2789 | | #ifdef LIBXML_HTML_ENABLED |
2790 | | |
2791 | | /** |
2792 | | * Initialize the default HTML SAX2 handler |
2793 | | * |
2794 | | * @param hdlr the SAX handler |
2795 | | */ |
2796 | | void |
2797 | | xmlSAX2InitHtmlDefaultSAXHandler(xmlSAXHandler *hdlr) |
2798 | 0 | { |
2799 | 0 | if ((hdlr == NULL) || (hdlr->initialized != 0)) |
2800 | 0 | return; |
2801 | | |
2802 | 0 | hdlr->internalSubset = xmlSAX2InternalSubset; |
2803 | 0 | hdlr->externalSubset = NULL; |
2804 | 0 | hdlr->isStandalone = NULL; |
2805 | 0 | hdlr->hasInternalSubset = NULL; |
2806 | 0 | hdlr->hasExternalSubset = NULL; |
2807 | 0 | hdlr->resolveEntity = NULL; |
2808 | 0 | hdlr->getEntity = xmlSAX2GetEntity; |
2809 | 0 | hdlr->getParameterEntity = NULL; |
2810 | 0 | hdlr->entityDecl = NULL; |
2811 | 0 | hdlr->attributeDecl = NULL; |
2812 | 0 | hdlr->elementDecl = NULL; |
2813 | 0 | hdlr->notationDecl = NULL; |
2814 | 0 | hdlr->unparsedEntityDecl = NULL; |
2815 | 0 | hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator; |
2816 | 0 | hdlr->startDocument = xmlSAX2StartDocument; |
2817 | 0 | hdlr->endDocument = xmlSAX2EndDocument; |
2818 | 0 | hdlr->startElement = xmlSAX2StartElement; |
2819 | 0 | hdlr->endElement = xmlSAX2EndElement; |
2820 | 0 | hdlr->reference = NULL; |
2821 | 0 | hdlr->characters = xmlSAX2Characters; |
2822 | 0 | hdlr->cdataBlock = xmlSAX2CDataBlock; |
2823 | 0 | hdlr->ignorableWhitespace = xmlSAX2IgnorableWhitespace; |
2824 | 0 | hdlr->processingInstruction = xmlSAX2ProcessingInstruction; |
2825 | 0 | hdlr->comment = xmlSAX2Comment; |
2826 | 0 | hdlr->warning = xmlParserWarning; |
2827 | 0 | hdlr->error = xmlParserError; |
2828 | 0 | hdlr->fatalError = xmlParserError; |
2829 | |
|
2830 | 0 | hdlr->initialized = 1; |
2831 | 0 | } |
2832 | | |
2833 | | /** |
2834 | | * @deprecated This function is a no-op. Call #xmlInitParser to |
2835 | | * initialize the library. |
2836 | | */ |
2837 | | void |
2838 | | htmlDefaultSAXHandlerInit(void) |
2839 | 0 | { |
2840 | 0 | } |
2841 | | |
2842 | | #endif /* LIBXML_HTML_ENABLED */ |