1 /*
2 * SAX2.c : Default SAX2 handler to build a tree.
3 *
4 * See Copyright for the status of this software.
5 *
6 * Daniel Veillard <daniel@veillard.com>
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 <libxml/xmlmemory.h>
16 #include <libxml/tree.h>
17 #include <libxml/parser.h>
18 #include <libxml/parserInternals.h>
19 #include <libxml/valid.h>
20 #include <libxml/entities.h>
21 #include <libxml/xmlerror.h>
22 #include <libxml/debugXML.h>
23 #include <libxml/xmlIO.h>
24 #include <libxml/SAX.h>
25 #include <libxml/uri.h>
26 #include <libxml/valid.h>
27 #include <libxml/HTMLtree.h>
28 #include <libxml/globals.h>
29
30 /* Define SIZE_T_MAX unless defined through <limits.h>. */
31 #ifndef SIZE_T_MAX
32 # define SIZE_T_MAX ((size_t)-1)
33 #endif /* !SIZE_T_MAX */
34
35 /* #define DEBUG_SAX2 */
36 /* #define DEBUG_SAX2_TREE */
37
38 /**
39 * TODO:
40 *
41 * macro to flag unimplemented blocks
42 * XML_CATALOG_PREFER user env to select between system/public prefered
43 * option. C.f. Richard Tobin <richard@cogsci.ed.ac.uk>
44 *> Just FYI, I am using an environment variable XML_CATALOG_PREFER with
45 *> values "system" and "public". I have made the default be "system" to
46 *> match yours.
47 */
48 #define TODO \
49 xmlGenericError(xmlGenericErrorContext, \
50 "Unimplemented block at %s:%d\n", \
51 __FILE__, __LINE__);
52
53 /*
54 * xmlSAX2ErrMemory:
55 * @ctxt: an XML validation parser context
56 * @msg: a string to accompany the error message
57 */
58 static void
xmlSAX2ErrMemory(xmlParserCtxtPtr ctxt,const char * msg)59 xmlSAX2ErrMemory(xmlParserCtxtPtr ctxt, const char *msg) {
60 if (ctxt != NULL) {
61 if ((ctxt->sax != NULL) && (ctxt->sax->error != NULL))
62 ctxt->sax->error(ctxt->userData, "%s: out of memory\n", msg);
63 ctxt->errNo = XML_ERR_NO_MEMORY;
64 ctxt->instate = XML_PARSER_EOF;
65 ctxt->disableSAX = 1;
66 }
67 }
68
69 /**
70 * xmlValidError:
71 * @ctxt: an XML validation parser context
72 * @error: the error number
73 * @msg: the error message
74 * @str1: extra data
75 * @str2: extra data
76 *
77 * Handle a validation error
78 */
79 static void
xmlErrValid(xmlParserCtxtPtr ctxt,xmlParserErrors error,const char * msg,const char * str1,const char * str2)80 xmlErrValid(xmlParserCtxtPtr ctxt, xmlParserErrors error,
81 const char *msg, const char *str1, const char *str2)
82 {
83 xmlStructuredErrorFunc schannel = NULL;
84
85 if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
86 (ctxt->instate == XML_PARSER_EOF))
87 return;
88 if (ctxt != NULL) {
89 ctxt->errNo = error;
90 if ((ctxt->sax != NULL) && (ctxt->sax->initialized == XML_SAX2_MAGIC))
91 schannel = ctxt->sax->serror;
92 __xmlRaiseError(schannel,
93 ctxt->vctxt.error, ctxt->vctxt.userData,
94 ctxt, NULL, XML_FROM_DTD, error,
95 XML_ERR_ERROR, NULL, 0, (const char *) str1,
96 (const char *) str2, NULL, 0, 0,
97 msg, (const char *) str1, (const char *) str2);
98 ctxt->valid = 0;
99 } else {
100 __xmlRaiseError(schannel,
101 NULL, NULL,
102 ctxt, NULL, XML_FROM_DTD, error,
103 XML_ERR_ERROR, NULL, 0, (const char *) str1,
104 (const char *) str2, NULL, 0, 0,
105 msg, (const char *) str1, (const char *) str2);
106 }
107 }
108
109 /**
110 * xmlFatalErrMsg:
111 * @ctxt: an XML parser context
112 * @error: the error number
113 * @msg: the error message
114 * @str1: an error string
115 * @str2: an error string
116 *
117 * Handle a fatal parser error, i.e. violating Well-Formedness constraints
118 */
119 static void
xmlFatalErrMsg(xmlParserCtxtPtr ctxt,xmlParserErrors error,const char * msg,const xmlChar * str1,const xmlChar * str2)120 xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
121 const char *msg, const xmlChar *str1, const xmlChar *str2)
122 {
123 if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
124 (ctxt->instate == XML_PARSER_EOF))
125 return;
126 if (ctxt != NULL)
127 ctxt->errNo = error;
128 __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error,
129 XML_ERR_FATAL, NULL, 0,
130 (const char *) str1, (const char *) str2,
131 NULL, 0, 0, msg, str1, str2);
132 if (ctxt != NULL) {
133 ctxt->wellFormed = 0;
134 ctxt->valid = 0;
135 if (ctxt->recovery == 0)
136 ctxt->disableSAX = 1;
137 }
138 }
139
140 /**
141 * xmlWarnMsg:
142 * @ctxt: an XML parser context
143 * @error: the error number
144 * @msg: the error message
145 * @str1: an error string
146 * @str2: an error string
147 *
148 * Handle a parser warning
149 */
150 static void
xmlWarnMsg(xmlParserCtxtPtr ctxt,xmlParserErrors error,const char * msg,const xmlChar * str1)151 xmlWarnMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
152 const char *msg, const xmlChar *str1)
153 {
154 if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
155 (ctxt->instate == XML_PARSER_EOF))
156 return;
157 if (ctxt != NULL)
158 ctxt->errNo = error;
159 __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error,
160 XML_ERR_WARNING, NULL, 0,
161 (const char *) str1, NULL,
162 NULL, 0, 0, msg, str1);
163 }
164
165 /**
166 * xmlNsErrMsg:
167 * @ctxt: an XML parser context
168 * @error: the error number
169 * @msg: the error message
170 * @str1: an error string
171 * @str2: an error string
172 *
173 * Handle a namespace error
174 */
175 static void
xmlNsErrMsg(xmlParserCtxtPtr ctxt,xmlParserErrors error,const char * msg,const xmlChar * str1,const xmlChar * str2)176 xmlNsErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
177 const char *msg, const xmlChar *str1, const xmlChar *str2)
178 {
179 if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
180 (ctxt->instate == XML_PARSER_EOF))
181 return;
182 if (ctxt != NULL)
183 ctxt->errNo = error;
184 __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error,
185 XML_ERR_ERROR, NULL, 0,
186 (const char *) str1, (const char *) str2,
187 NULL, 0, 0, msg, str1, str2);
188 }
189
190 /**
191 * xmlNsWarnMsg:
192 * @ctxt: an XML parser context
193 * @error: the error number
194 * @msg: the error message
195 * @str1: an error string
196 *
197 * Handle a namespace warning
198 */
199 static void
xmlNsWarnMsg(xmlParserCtxtPtr ctxt,xmlParserErrors error,const char * msg,const xmlChar * str1,const xmlChar * str2)200 xmlNsWarnMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error,
201 const char *msg, const xmlChar *str1, const xmlChar *str2)
202 {
203 if ((ctxt != NULL) && (ctxt->disableSAX != 0) &&
204 (ctxt->instate == XML_PARSER_EOF))
205 return;
206 if (ctxt != NULL)
207 ctxt->errNo = error;
208 __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error,
209 XML_ERR_WARNING, NULL, 0,
210 (const char *) str1, (const char *) str2,
211 NULL, 0, 0, msg, str1, str2);
212 }
213
214 /**
215 * xmlSAX2GetPublicId:
216 * @ctx: the user data (XML parser context)
217 *
218 * Provides the public ID e.g. "-//SGMLSOURCE//DTD DEMO//EN"
219 *
220 * Returns a xmlChar *
221 */
222 const xmlChar *
xmlSAX2GetPublicId(void * ctx ATTRIBUTE_UNUSED)223 xmlSAX2GetPublicId(void *ctx ATTRIBUTE_UNUSED)
224 {
225 /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */
226 return(NULL);
227 }
228
229 /**
230 * xmlSAX2GetSystemId:
231 * @ctx: the user data (XML parser context)
232 *
233 * Provides the system ID, basically URL or filename e.g.
234 * http://www.sgmlsource.com/dtds/memo.dtd
235 *
236 * Returns a xmlChar *
237 */
238 const xmlChar *
xmlSAX2GetSystemId(void * ctx)239 xmlSAX2GetSystemId(void *ctx)
240 {
241 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
242 if ((ctx == NULL) || (ctxt->input == NULL)) return(NULL);
243 return((const xmlChar *) ctxt->input->filename);
244 }
245
246 /**
247 * xmlSAX2GetLineNumber:
248 * @ctx: the user data (XML parser context)
249 *
250 * Provide the line number of the current parsing point.
251 *
252 * Returns an int
253 */
254 int
xmlSAX2GetLineNumber(void * ctx)255 xmlSAX2GetLineNumber(void *ctx)
256 {
257 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
258 if ((ctx == NULL) || (ctxt->input == NULL)) return(0);
259 return(ctxt->input->line);
260 }
261
262 /**
263 * xmlSAX2GetColumnNumber:
264 * @ctx: the user data (XML parser context)
265 *
266 * Provide the column number of the current parsing point.
267 *
268 * Returns an int
269 */
270 int
xmlSAX2GetColumnNumber(void * ctx)271 xmlSAX2GetColumnNumber(void *ctx)
272 {
273 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
274 if ((ctx == NULL) || (ctxt->input == NULL)) return(0);
275 return(ctxt->input->col);
276 }
277
278 /**
279 * xmlSAX2IsStandalone:
280 * @ctx: the user data (XML parser context)
281 *
282 * Is this document tagged standalone ?
283 *
284 * Returns 1 if true
285 */
286 int
xmlSAX2IsStandalone(void * ctx)287 xmlSAX2IsStandalone(void *ctx)
288 {
289 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
290 if ((ctx == NULL) || (ctxt->myDoc == NULL)) return(0);
291 return(ctxt->myDoc->standalone == 1);
292 }
293
294 /**
295 * xmlSAX2HasInternalSubset:
296 * @ctx: the user data (XML parser context)
297 *
298 * Does this document has an internal subset
299 *
300 * Returns 1 if true
301 */
302 int
xmlSAX2HasInternalSubset(void * ctx)303 xmlSAX2HasInternalSubset(void *ctx)
304 {
305 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
306 if ((ctxt == NULL) || (ctxt->myDoc == NULL)) return(0);
307 return(ctxt->myDoc->intSubset != NULL);
308 }
309
310 /**
311 * xmlSAX2HasExternalSubset:
312 * @ctx: the user data (XML parser context)
313 *
314 * Does this document has an external subset
315 *
316 * Returns 1 if true
317 */
318 int
xmlSAX2HasExternalSubset(void * ctx)319 xmlSAX2HasExternalSubset(void *ctx)
320 {
321 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
322 if ((ctxt == NULL) || (ctxt->myDoc == NULL)) return(0);
323 return(ctxt->myDoc->extSubset != NULL);
324 }
325
326 /**
327 * xmlSAX2InternalSubset:
328 * @ctx: the user data (XML parser context)
329 * @name: the root element name
330 * @ExternalID: the external ID
331 * @SystemID: the SYSTEM ID (e.g. filename or URL)
332 *
333 * Callback on internal subset declaration.
334 */
335 void
xmlSAX2InternalSubset(void * ctx,const xmlChar * name,const xmlChar * ExternalID,const xmlChar * SystemID)336 xmlSAX2InternalSubset(void *ctx, const xmlChar *name,
337 const xmlChar *ExternalID, const xmlChar *SystemID)
338 {
339 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
340 xmlDtdPtr dtd;
341 if (ctx == NULL) return;
342 #ifdef DEBUG_SAX
343 xmlGenericError(xmlGenericErrorContext,
344 "SAX.xmlSAX2InternalSubset(%s, %s, %s)\n",
345 name, ExternalID, SystemID);
346 #endif
347
348 if (ctxt->myDoc == NULL)
349 return;
350 dtd = xmlGetIntSubset(ctxt->myDoc);
351 if (dtd != NULL) {
352 if (ctxt->html)
353 return;
354 xmlUnlinkNode((xmlNodePtr) dtd);
355 xmlFreeDtd(dtd);
356 ctxt->myDoc->intSubset = NULL;
357 }
358 ctxt->myDoc->intSubset =
359 xmlCreateIntSubset(ctxt->myDoc, name, ExternalID, SystemID);
360 if (ctxt->myDoc->intSubset == NULL)
361 xmlSAX2ErrMemory(ctxt, "xmlSAX2InternalSubset");
362 }
363
364 /**
365 * xmlSAX2ExternalSubset:
366 * @ctx: the user data (XML parser context)
367 * @name: the root element name
368 * @ExternalID: the external ID
369 * @SystemID: the SYSTEM ID (e.g. filename or URL)
370 *
371 * Callback on external subset declaration.
372 */
373 void
xmlSAX2ExternalSubset(void * ctx,const xmlChar * name,const xmlChar * ExternalID,const xmlChar * SystemID)374 xmlSAX2ExternalSubset(void *ctx, const xmlChar *name,
375 const xmlChar *ExternalID, const xmlChar *SystemID)
376 {
377 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
378 if (ctx == NULL) return;
379 #ifdef DEBUG_SAX
380 xmlGenericError(xmlGenericErrorContext,
381 "SAX.xmlSAX2ExternalSubset(%s, %s, %s)\n",
382 name, ExternalID, SystemID);
383 #endif
384 if (((ExternalID != NULL) || (SystemID != NULL)) &&
385 (((ctxt->validate) || (ctxt->loadsubset != 0)) &&
386 (ctxt->wellFormed && ctxt->myDoc))) {
387 /*
388 * Try to fetch and parse the external subset.
389 */
390 xmlParserInputPtr oldinput;
391 int oldinputNr;
392 int oldinputMax;
393 xmlParserInputPtr *oldinputTab;
394 xmlParserInputPtr input = NULL;
395 xmlCharEncoding enc;
396 int oldcharset;
397
398 /*
399 * Ask the Entity resolver to load the damn thing
400 */
401 if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL))
402 input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID,
403 SystemID);
404 if (input == NULL) {
405 return;
406 }
407
408 xmlNewDtd(ctxt->myDoc, name, ExternalID, SystemID);
409
410 /*
411 * make sure we won't destroy the main document context
412 */
413 oldinput = ctxt->input;
414 oldinputNr = ctxt->inputNr;
415 oldinputMax = ctxt->inputMax;
416 oldinputTab = ctxt->inputTab;
417 oldcharset = ctxt->charset;
418
419 ctxt->inputTab = (xmlParserInputPtr *)
420 xmlMalloc(5 * sizeof(xmlParserInputPtr));
421 if (ctxt->inputTab == NULL) {
422 xmlSAX2ErrMemory(ctxt, "xmlSAX2ExternalSubset");
423 ctxt->input = oldinput;
424 ctxt->inputNr = oldinputNr;
425 ctxt->inputMax = oldinputMax;
426 ctxt->inputTab = oldinputTab;
427 ctxt->charset = oldcharset;
428 return;
429 }
430 ctxt->inputNr = 0;
431 ctxt->inputMax = 5;
432 ctxt->input = NULL;
433 xmlPushInput(ctxt, input);
434
435 /*
436 * On the fly encoding conversion if needed
437 */
438 if (ctxt->input->length >= 4) {
439 enc = xmlDetectCharEncoding(ctxt->input->cur, 4);
440 xmlSwitchEncoding(ctxt, enc);
441 }
442
443 if (input->filename == NULL)
444 input->filename = (char *) xmlCanonicPath(SystemID);
445 input->line = 1;
446 input->col = 1;
447 input->base = ctxt->input->cur;
448 input->cur = ctxt->input->cur;
449 input->free = NULL;
450
451 /*
452 * let's parse that entity knowing it's an external subset.
453 */
454 xmlParseExternalSubset(ctxt, ExternalID, SystemID);
455
456 /*
457 * Free up the external entities
458 */
459
460 while (ctxt->inputNr > 1)
461 xmlPopInput(ctxt);
462 xmlFreeInputStream(ctxt->input);
463 xmlFree(ctxt->inputTab);
464
465 /*
466 * Restore the parsing context of the main entity
467 */
468 ctxt->input = oldinput;
469 ctxt->inputNr = oldinputNr;
470 ctxt->inputMax = oldinputMax;
471 ctxt->inputTab = oldinputTab;
472 ctxt->charset = oldcharset;
473 /* ctxt->wellFormed = oldwellFormed; */
474 }
475 }
476
477 /**
478 * xmlSAX2ResolveEntity:
479 * @ctx: the user data (XML parser context)
480 * @publicId: The public ID of the entity
481 * @systemId: The system ID of the entity
482 *
483 * The entity loader, to control the loading of external entities,
484 * the application can either:
485 * - override this xmlSAX2ResolveEntity() callback in the SAX block
486 * - or better use the xmlSetExternalEntityLoader() function to
487 * set up it's own entity resolution routine
488 *
489 * Returns the xmlParserInputPtr if inlined or NULL for DOM behaviour.
490 */
491 xmlParserInputPtr
xmlSAX2ResolveEntity(void * ctx,const xmlChar * publicId,const xmlChar * systemId)492 xmlSAX2ResolveEntity(void *ctx, const xmlChar *publicId, const xmlChar *systemId)
493 {
494 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
495 xmlParserInputPtr ret;
496 xmlChar *URI;
497 const char *base = NULL;
498
499 if (ctx == NULL) return(NULL);
500 if (ctxt->input != NULL)
501 base = ctxt->input->filename;
502 if (base == NULL)
503 base = ctxt->directory;
504
505 URI = xmlBuildURI(systemId, (const xmlChar *) base);
506
507 #ifdef DEBUG_SAX
508 xmlGenericError(xmlGenericErrorContext,
509 "SAX.xmlSAX2ResolveEntity(%s, %s)\n", publicId, systemId);
510 #endif
511
512 ret = xmlLoadExternalEntity((const char *) URI,
513 (const char *) publicId, ctxt);
514 if (URI != NULL)
515 xmlFree(URI);
516 return(ret);
517 }
518
519 /**
520 * xmlSAX2GetEntity:
521 * @ctx: the user data (XML parser context)
522 * @name: The entity name
523 *
524 * Get an entity by name
525 *
526 * Returns the xmlEntityPtr if found.
527 */
528 xmlEntityPtr
xmlSAX2GetEntity(void * ctx,const xmlChar * name)529 xmlSAX2GetEntity(void *ctx, const xmlChar *name)
530 {
531 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
532 xmlEntityPtr ret = NULL;
533
534 if (ctx == NULL) return(NULL);
535 #ifdef DEBUG_SAX
536 xmlGenericError(xmlGenericErrorContext,
537 "SAX.xmlSAX2GetEntity(%s)\n", name);
538 #endif
539
540 if (ctxt->inSubset == 0) {
541 ret = xmlGetPredefinedEntity(name);
542 if (ret != NULL)
543 return(ret);
544 }
545 if ((ctxt->myDoc != NULL) && (ctxt->myDoc->standalone == 1)) {
546 if (ctxt->inSubset == 2) {
547 ctxt->myDoc->standalone = 0;
548 ret = xmlGetDocEntity(ctxt->myDoc, name);
549 ctxt->myDoc->standalone = 1;
550 } else {
551 ret = xmlGetDocEntity(ctxt->myDoc, name);
552 if (ret == NULL) {
553 ctxt->myDoc->standalone = 0;
554 ret = xmlGetDocEntity(ctxt->myDoc, name);
555 if (ret != NULL) {
556 xmlFatalErrMsg(ctxt, XML_ERR_NOT_STANDALONE,
557 "Entity(%s) document marked standalone but requires external subset\n",
558 name, NULL);
559 }
560 ctxt->myDoc->standalone = 1;
561 }
562 }
563 } else {
564 ret = xmlGetDocEntity(ctxt->myDoc, name);
565 }
566 if ((ret != NULL) &&
567 ((ctxt->validate) || (ctxt->replaceEntities)) &&
568 (ret->children == NULL) &&
569 (ret->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) {
570 int val;
571
572 /*
573 * for validation purposes we really need to fetch and
574 * parse the external entity
575 */
576 xmlNodePtr children;
577
578 val = xmlParseCtxtExternalEntity(ctxt, ret->URI,
579 ret->ExternalID, &children);
580 if (val == 0) {
581 xmlAddChildList((xmlNodePtr) ret, children);
582 } else {
583 xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
584 "Failure to process entity %s\n", name, NULL);
585 ctxt->validate = 0;
586 return(NULL);
587 }
588 ret->owner = 1;
589 if (ret->checked == 0)
590 ret->checked = 1;
591 }
592 return(ret);
593 }
594
595 /**
596 * xmlSAX2GetParameterEntity:
597 * @ctx: the user data (XML parser context)
598 * @name: The entity name
599 *
600 * Get a parameter entity by name
601 *
602 * Returns the xmlEntityPtr if found.
603 */
604 xmlEntityPtr
xmlSAX2GetParameterEntity(void * ctx,const xmlChar * name)605 xmlSAX2GetParameterEntity(void *ctx, const xmlChar *name)
606 {
607 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
608 xmlEntityPtr ret;
609
610 if (ctx == NULL) return(NULL);
611 #ifdef DEBUG_SAX
612 xmlGenericError(xmlGenericErrorContext,
613 "SAX.xmlSAX2GetParameterEntity(%s)\n", name);
614 #endif
615
616 ret = xmlGetParameterEntity(ctxt->myDoc, name);
617 return(ret);
618 }
619
620
621 /**
622 * xmlSAX2EntityDecl:
623 * @ctx: the user data (XML parser context)
624 * @name: the entity name
625 * @type: the entity type
626 * @publicId: The public ID of the entity
627 * @systemId: The system ID of the entity
628 * @content: the entity value (without processing).
629 *
630 * An entity definition has been parsed
631 */
632 void
xmlSAX2EntityDecl(void * ctx,const xmlChar * name,int type,const xmlChar * publicId,const xmlChar * systemId,xmlChar * content)633 xmlSAX2EntityDecl(void *ctx, const xmlChar *name, int type,
634 const xmlChar *publicId, const xmlChar *systemId, xmlChar *content)
635 {
636 xmlEntityPtr ent;
637 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
638
639 if (ctx == NULL) return;
640 #ifdef DEBUG_SAX
641 xmlGenericError(xmlGenericErrorContext,
642 "SAX.xmlSAX2EntityDecl(%s, %d, %s, %s, %s)\n",
643 name, type, publicId, systemId, content);
644 #endif
645 if (ctxt->inSubset == 1) {
646 ent = xmlAddDocEntity(ctxt->myDoc, name, type, publicId,
647 systemId, content);
648 if ((ent == NULL) && (ctxt->pedantic))
649 xmlWarnMsg(ctxt, XML_WAR_ENTITY_REDEFINED,
650 "Entity(%s) already defined in the internal subset\n",
651 name);
652 if ((ent != NULL) && (ent->URI == NULL) && (systemId != NULL)) {
653 xmlChar *URI;
654 const char *base = NULL;
655
656 if (ctxt->input != NULL)
657 base = ctxt->input->filename;
658 if (base == NULL)
659 base = ctxt->directory;
660
661 URI = xmlBuildURI(systemId, (const xmlChar *) base);
662 ent->URI = URI;
663 }
664 } else if (ctxt->inSubset == 2) {
665 ent = xmlAddDtdEntity(ctxt->myDoc, name, type, publicId,
666 systemId, content);
667 if ((ent == NULL) && (ctxt->pedantic) &&
668 (ctxt->sax != NULL) && (ctxt->sax->warning != NULL))
669 ctxt->sax->warning(ctxt->userData,
670 "Entity(%s) already defined in the external subset\n", name);
671 if ((ent != NULL) && (ent->URI == NULL) && (systemId != NULL)) {
672 xmlChar *URI;
673 const char *base = NULL;
674
675 if (ctxt->input != NULL)
676 base = ctxt->input->filename;
677 if (base == NULL)
678 base = ctxt->directory;
679
680 URI = xmlBuildURI(systemId, (const xmlChar *) base);
681 ent->URI = URI;
682 }
683 } else {
684 xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_PROCESSING,
685 "SAX.xmlSAX2EntityDecl(%s) called while not in subset\n",
686 name, NULL);
687 }
688 }
689
690 /**
691 * xmlSAX2AttributeDecl:
692 * @ctx: the user data (XML parser context)
693 * @elem: the name of the element
694 * @fullname: the attribute name
695 * @type: the attribute type
696 * @def: the type of default value
697 * @defaultValue: the attribute default value
698 * @tree: the tree of enumerated value set
699 *
700 * An attribute definition has been parsed
701 */
702 void
xmlSAX2AttributeDecl(void * ctx,const xmlChar * elem,const xmlChar * fullname,int type,int def,const xmlChar * defaultValue,xmlEnumerationPtr tree)703 xmlSAX2AttributeDecl(void *ctx, const xmlChar *elem, const xmlChar *fullname,
704 int type, int def, const xmlChar *defaultValue,
705 xmlEnumerationPtr tree)
706 {
707 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
708 xmlAttributePtr attr;
709 xmlChar *name = NULL, *prefix = NULL;
710
711 if ((ctxt == NULL) || (ctxt->myDoc == NULL))
712 return;
713
714 #ifdef DEBUG_SAX
715 xmlGenericError(xmlGenericErrorContext,
716 "SAX.xmlSAX2AttributeDecl(%s, %s, %d, %d, %s, ...)\n",
717 elem, fullname, type, def, defaultValue);
718 #endif
719 if ((xmlStrEqual(fullname, BAD_CAST "xml:id")) &&
720 (type != XML_ATTRIBUTE_ID)) {
721 /*
722 * Raise the error but keep the validity flag
723 */
724 int tmp = ctxt->valid;
725 xmlErrValid(ctxt, XML_DTD_XMLID_TYPE,
726 "xml:id : attribute type should be ID\n", NULL, NULL);
727 ctxt->valid = tmp;
728 }
729 /* TODO: optimize name/prefix allocation */
730 name = xmlSplitQName(ctxt, fullname, &prefix);
731 ctxt->vctxt.valid = 1;
732 if (ctxt->inSubset == 1)
733 attr = xmlAddAttributeDecl(&ctxt->vctxt, ctxt->myDoc->intSubset, elem,
734 name, prefix, (xmlAttributeType) type,
735 (xmlAttributeDefault) def, defaultValue, tree);
736 else if (ctxt->inSubset == 2)
737 attr = xmlAddAttributeDecl(&ctxt->vctxt, ctxt->myDoc->extSubset, elem,
738 name, prefix, (xmlAttributeType) type,
739 (xmlAttributeDefault) def, defaultValue, tree);
740 else {
741 xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
742 "SAX.xmlSAX2AttributeDecl(%s) called while not in subset\n",
743 name, NULL);
744 xmlFreeEnumeration(tree);
745 return;
746 }
747 #ifdef LIBXML_VALID_ENABLED
748 if (ctxt->vctxt.valid == 0)
749 ctxt->valid = 0;
750 if ((attr != NULL) && (ctxt->validate) && (ctxt->wellFormed) &&
751 (ctxt->myDoc->intSubset != NULL))
752 ctxt->valid &= xmlValidateAttributeDecl(&ctxt->vctxt, ctxt->myDoc,
753 attr);
754 #endif /* LIBXML_VALID_ENABLED */
755 if (prefix != NULL)
756 xmlFree(prefix);
757 if (name != NULL)
758 xmlFree(name);
759 }
760
761 /**
762 * xmlSAX2ElementDecl:
763 * @ctx: the user data (XML parser context)
764 * @name: the element name
765 * @type: the element type
766 * @content: the element value tree
767 *
768 * An element definition has been parsed
769 */
770 void
xmlSAX2ElementDecl(void * ctx,const xmlChar * name,int type,xmlElementContentPtr content)771 xmlSAX2ElementDecl(void *ctx, const xmlChar * name, int type,
772 xmlElementContentPtr content)
773 {
774 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
775 xmlElementPtr elem = NULL;
776
777 if ((ctxt == NULL) || (ctxt->myDoc == NULL))
778 return;
779
780 #ifdef DEBUG_SAX
781 xmlGenericError(xmlGenericErrorContext,
782 "SAX.xmlSAX2ElementDecl(%s, %d, ...)\n", name, type);
783 #endif
784
785 if (ctxt->inSubset == 1)
786 elem = xmlAddElementDecl(&ctxt->vctxt, ctxt->myDoc->intSubset,
787 name, (xmlElementTypeVal) type, content);
788 else if (ctxt->inSubset == 2)
789 elem = xmlAddElementDecl(&ctxt->vctxt, ctxt->myDoc->extSubset,
790 name, (xmlElementTypeVal) type, content);
791 else {
792 xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
793 "SAX.xmlSAX2ElementDecl(%s) called while not in subset\n",
794 name, NULL);
795 return;
796 }
797 #ifdef LIBXML_VALID_ENABLED
798 if (elem == NULL)
799 ctxt->valid = 0;
800 if (ctxt->validate && ctxt->wellFormed &&
801 ctxt->myDoc && ctxt->myDoc->intSubset)
802 ctxt->valid &=
803 xmlValidateElementDecl(&ctxt->vctxt, ctxt->myDoc, elem);
804 #endif /* LIBXML_VALID_ENABLED */
805 }
806
807 /**
808 * xmlSAX2NotationDecl:
809 * @ctx: the user data (XML parser context)
810 * @name: The name of the notation
811 * @publicId: The public ID of the entity
812 * @systemId: The system ID of the entity
813 *
814 * What to do when a notation declaration has been parsed.
815 */
816 void
xmlSAX2NotationDecl(void * ctx,const xmlChar * name,const xmlChar * publicId,const xmlChar * systemId)817 xmlSAX2NotationDecl(void *ctx, const xmlChar *name,
818 const xmlChar *publicId, const xmlChar *systemId)
819 {
820 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
821 xmlNotationPtr nota = NULL;
822
823 if ((ctxt == NULL) || (ctxt->myDoc == NULL))
824 return;
825
826 #ifdef DEBUG_SAX
827 xmlGenericError(xmlGenericErrorContext,
828 "SAX.xmlSAX2NotationDecl(%s, %s, %s)\n", name, publicId, systemId);
829 #endif
830
831 if ((publicId == NULL) && (systemId == NULL)) {
832 xmlFatalErrMsg(ctxt, XML_ERR_NOTATION_PROCESSING,
833 "SAX.xmlSAX2NotationDecl(%s) externalID or PublicID missing\n",
834 name, NULL);
835 return;
836 } else if (ctxt->inSubset == 1)
837 nota = xmlAddNotationDecl(&ctxt->vctxt, ctxt->myDoc->intSubset, name,
838 publicId, systemId);
839 else if (ctxt->inSubset == 2)
840 nota = xmlAddNotationDecl(&ctxt->vctxt, ctxt->myDoc->extSubset, name,
841 publicId, systemId);
842 else {
843 xmlFatalErrMsg(ctxt, XML_ERR_NOTATION_PROCESSING,
844 "SAX.xmlSAX2NotationDecl(%s) called while not in subset\n",
845 name, NULL);
846 return;
847 }
848 #ifdef LIBXML_VALID_ENABLED
849 if (nota == NULL) ctxt->valid = 0;
850 if ((ctxt->validate) && (ctxt->wellFormed) &&
851 (ctxt->myDoc->intSubset != NULL))
852 ctxt->valid &= xmlValidateNotationDecl(&ctxt->vctxt, ctxt->myDoc,
853 nota);
854 #endif /* LIBXML_VALID_ENABLED */
855 }
856
857 /**
858 * xmlSAX2UnparsedEntityDecl:
859 * @ctx: the user data (XML parser context)
860 * @name: The name of the entity
861 * @publicId: The public ID of the entity
862 * @systemId: The system ID of the entity
863 * @notationName: the name of the notation
864 *
865 * What to do when an unparsed entity declaration is parsed
866 */
867 void
xmlSAX2UnparsedEntityDecl(void * ctx,const xmlChar * name,const xmlChar * publicId,const xmlChar * systemId,const xmlChar * notationName)868 xmlSAX2UnparsedEntityDecl(void *ctx, const xmlChar *name,
869 const xmlChar *publicId, const xmlChar *systemId,
870 const xmlChar *notationName)
871 {
872 xmlEntityPtr ent;
873 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
874 if (ctx == NULL) return;
875 #ifdef DEBUG_SAX
876 xmlGenericError(xmlGenericErrorContext,
877 "SAX.xmlSAX2UnparsedEntityDecl(%s, %s, %s, %s)\n",
878 name, publicId, systemId, notationName);
879 #endif
880 if (ctxt->inSubset == 1) {
881 ent = xmlAddDocEntity(ctxt->myDoc, name,
882 XML_EXTERNAL_GENERAL_UNPARSED_ENTITY,
883 publicId, systemId, notationName);
884 if ((ent == NULL) && (ctxt->pedantic) &&
885 (ctxt->sax != NULL) && (ctxt->sax->warning != NULL))
886 ctxt->sax->warning(ctxt->userData,
887 "Entity(%s) already defined in the internal subset\n", name);
888 if ((ent != NULL) && (ent->URI == NULL) && (systemId != NULL)) {
889 xmlChar *URI;
890 const char *base = NULL;
891
892 if (ctxt->input != NULL)
893 base = ctxt->input->filename;
894 if (base == NULL)
895 base = ctxt->directory;
896
897 URI = xmlBuildURI(systemId, (const xmlChar *) base);
898 ent->URI = URI;
899 }
900 } else if (ctxt->inSubset == 2) {
901 ent = xmlAddDtdEntity(ctxt->myDoc, name,
902 XML_EXTERNAL_GENERAL_UNPARSED_ENTITY,
903 publicId, systemId, notationName);
904 if ((ent == NULL) && (ctxt->pedantic) &&
905 (ctxt->sax != NULL) && (ctxt->sax->warning != NULL))
906 ctxt->sax->warning(ctxt->userData,
907 "Entity(%s) already defined in the external subset\n", name);
908 if ((ent != NULL) && (ent->URI == NULL) && (systemId != NULL)) {
909 xmlChar *URI;
910 const char *base = NULL;
911
912 if (ctxt->input != NULL)
913 base = ctxt->input->filename;
914 if (base == NULL)
915 base = ctxt->directory;
916
917 URI = xmlBuildURI(systemId, (const xmlChar *) base);
918 ent->URI = URI;
919 }
920 } else {
921 xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR,
922 "SAX.xmlSAX2UnparsedEntityDecl(%s) called while not in subset\n",
923 name, NULL);
924 }
925 }
926
927 /**
928 * xmlSAX2SetDocumentLocator:
929 * @ctx: the user data (XML parser context)
930 * @loc: A SAX Locator
931 *
932 * Receive the document locator at startup, actually xmlDefaultSAXLocator
933 * Everything is available on the context, so this is useless in our case.
934 */
935 void
xmlSAX2SetDocumentLocator(void * ctx ATTRIBUTE_UNUSED,xmlSAXLocatorPtr loc ATTRIBUTE_UNUSED)936 xmlSAX2SetDocumentLocator(void *ctx ATTRIBUTE_UNUSED, xmlSAXLocatorPtr loc ATTRIBUTE_UNUSED)
937 {
938 /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */
939 #ifdef DEBUG_SAX
940 xmlGenericError(xmlGenericErrorContext,
941 "SAX.xmlSAX2SetDocumentLocator()\n");
942 #endif
943 }
944
945 /**
946 * xmlSAX2StartDocument:
947 * @ctx: the user data (XML parser context)
948 *
949 * called when the document start being processed.
950 */
951 void
xmlSAX2StartDocument(void * ctx)952 xmlSAX2StartDocument(void *ctx)
953 {
954 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
955 xmlDocPtr doc;
956
957 if (ctx == NULL) return;
958
959 #ifdef DEBUG_SAX
960 xmlGenericError(xmlGenericErrorContext,
961 "SAX.xmlSAX2StartDocument()\n");
962 #endif
963 if (ctxt->html) {
964 #ifdef LIBXML_HTML_ENABLED
965 if (ctxt->myDoc == NULL)
966 ctxt->myDoc = htmlNewDocNoDtD(NULL, NULL);
967 ctxt->myDoc->properties = XML_DOC_HTML;
968 ctxt->myDoc->parseFlags = ctxt->options;
969 if (ctxt->myDoc == NULL) {
970 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartDocument");
971 return;
972 }
973 #else
974 xmlGenericError(xmlGenericErrorContext,
975 "libxml2 built without HTML support\n");
976 ctxt->errNo = XML_ERR_INTERNAL_ERROR;
977 ctxt->instate = XML_PARSER_EOF;
978 ctxt->disableSAX = 1;
979 return;
980 #endif
981 } else {
982 doc = ctxt->myDoc = xmlNewDoc(ctxt->version);
983 if (doc != NULL) {
984 doc->properties = 0;
985 if (ctxt->options & XML_PARSE_OLD10)
986 doc->properties |= XML_DOC_OLD10;
987 doc->parseFlags = ctxt->options;
988 if (ctxt->encoding != NULL)
989 doc->encoding = xmlStrdup(ctxt->encoding);
990 else
991 doc->encoding = NULL;
992 doc->standalone = ctxt->standalone;
993 } else {
994 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartDocument");
995 return;
996 }
997 if ((ctxt->dictNames) && (doc != NULL)) {
998 doc->dict = ctxt->dict;
999 xmlDictReference(doc->dict);
1000 }
1001 }
1002 if ((ctxt->myDoc != NULL) && (ctxt->myDoc->URL == NULL) &&
1003 (ctxt->input != NULL) && (ctxt->input->filename != NULL)) {
1004 ctxt->myDoc->URL = xmlPathToURI((const xmlChar *)ctxt->input->filename);
1005 if (ctxt->myDoc->URL == NULL)
1006 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartDocument");
1007 }
1008 }
1009
1010 /**
1011 * xmlSAX2EndDocument:
1012 * @ctx: the user data (XML parser context)
1013 *
1014 * called when the document end has been detected.
1015 */
1016 void
xmlSAX2EndDocument(void * ctx)1017 xmlSAX2EndDocument(void *ctx)
1018 {
1019 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
1020 #ifdef DEBUG_SAX
1021 xmlGenericError(xmlGenericErrorContext,
1022 "SAX.xmlSAX2EndDocument()\n");
1023 #endif
1024 if (ctx == NULL) return;
1025 #ifdef LIBXML_VALID_ENABLED
1026 if (ctxt->validate && ctxt->wellFormed &&
1027 ctxt->myDoc && ctxt->myDoc->intSubset)
1028 ctxt->valid &= xmlValidateDocumentFinal(&ctxt->vctxt, ctxt->myDoc);
1029 #endif /* LIBXML_VALID_ENABLED */
1030
1031 /*
1032 * Grab the encoding if it was added on-the-fly
1033 */
1034 if ((ctxt->encoding != NULL) && (ctxt->myDoc != NULL) &&
1035 (ctxt->myDoc->encoding == NULL)) {
1036 ctxt->myDoc->encoding = ctxt->encoding;
1037 ctxt->encoding = NULL;
1038 }
1039 if ((ctxt->inputTab != NULL) &&
1040 (ctxt->inputNr > 0) && (ctxt->inputTab[0] != NULL) &&
1041 (ctxt->inputTab[0]->encoding != NULL) && (ctxt->myDoc != NULL) &&
1042 (ctxt->myDoc->encoding == NULL)) {
1043 ctxt->myDoc->encoding = xmlStrdup(ctxt->inputTab[0]->encoding);
1044 }
1045 if ((ctxt->charset != XML_CHAR_ENCODING_NONE) && (ctxt->myDoc != NULL) &&
1046 (ctxt->myDoc->charset == XML_CHAR_ENCODING_NONE)) {
1047 ctxt->myDoc->charset = ctxt->charset;
1048 }
1049 }
1050
1051 #if defined(LIBXML_SAX1_ENABLED) || defined(LIBXML_HTML_ENABLED) || defined(LIBXML_WRITER_ENABLED) || defined(LIBXML_DOCB_ENABLED)
1052 /**
1053 * xmlSAX2AttributeInternal:
1054 * @ctx: the user data (XML parser context)
1055 * @fullname: The attribute name, including namespace prefix
1056 * @value: The attribute value
1057 * @prefix: the prefix on the element node
1058 *
1059 * Handle an attribute that has been read by the parser.
1060 * The default handling is to convert the attribute into an
1061 * DOM subtree and past it in a new xmlAttr element added to
1062 * the element.
1063 */
1064 static void
xmlSAX2AttributeInternal(void * ctx,const xmlChar * fullname,const xmlChar * value,const xmlChar * prefix ATTRIBUTE_UNUSED)1065 xmlSAX2AttributeInternal(void *ctx, const xmlChar *fullname,
1066 const xmlChar *value, const xmlChar *prefix ATTRIBUTE_UNUSED)
1067 {
1068 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
1069 xmlAttrPtr ret;
1070 xmlChar *name;
1071 xmlChar *ns;
1072 xmlChar *nval;
1073 xmlNsPtr namespace;
1074
1075 if (ctxt->html) {
1076 name = xmlStrdup(fullname);
1077 ns = NULL;
1078 namespace = NULL;
1079 } else {
1080 /*
1081 * Split the full name into a namespace prefix and the tag name
1082 */
1083 name = xmlSplitQName(ctxt, fullname, &ns);
1084 if ((name != NULL) && (name[0] == 0)) {
1085 if (xmlStrEqual(ns, BAD_CAST "xmlns")) {
1086 xmlNsErrMsg(ctxt, XML_ERR_NS_DECL_ERROR,
1087 "invalid namespace declaration '%s'\n",
1088 fullname, NULL);
1089 } else {
1090 xmlNsWarnMsg(ctxt, XML_WAR_NS_COLUMN,
1091 "Avoid attribute ending with ':' like '%s'\n",
1092 fullname, NULL);
1093 }
1094 if (ns != NULL)
1095 xmlFree(ns);
1096 ns = NULL;
1097 xmlFree(name);
1098 name = xmlStrdup(fullname);
1099 }
1100 }
1101 if (name == NULL) {
1102 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement");
1103 if (ns != NULL)
1104 xmlFree(ns);
1105 return;
1106 }
1107
1108 #ifdef LIBXML_VALID_ENABLED
1109 /*
1110 * Do the last stage of the attribute normalization
1111 * Needed for HTML too:
1112 * http://www.w3.org/TR/html4/types.html#h-6.2
1113 */
1114 ctxt->vctxt.valid = 1;
1115 nval = xmlValidCtxtNormalizeAttributeValue(&ctxt->vctxt,
1116 ctxt->myDoc, ctxt->node,
1117 fullname, value);
1118 if (ctxt->vctxt.valid != 1) {
1119 ctxt->valid = 0;
1120 }
1121 if (nval != NULL)
1122 value = nval;
1123 #else
1124 nval = NULL;
1125 #endif /* LIBXML_VALID_ENABLED */
1126
1127 /*
1128 * Check whether it's a namespace definition
1129 */
1130 if ((!ctxt->html) && (ns == NULL) &&
1131 (name[0] == 'x') && (name[1] == 'm') && (name[2] == 'l') &&
1132 (name[3] == 'n') && (name[4] == 's') && (name[5] == 0)) {
1133 xmlNsPtr nsret;
1134 xmlChar *val;
1135
1136 if (!ctxt->replaceEntities) {
1137 ctxt->depth++;
1138 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF,
1139 0,0,0);
1140 ctxt->depth--;
1141 } else {
1142 val = (xmlChar *) value;
1143 }
1144
1145 if (val[0] != 0) {
1146 xmlURIPtr uri;
1147
1148 uri = xmlParseURI((const char *)val);
1149 if (uri == NULL) {
1150 if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL))
1151 ctxt->sax->warning(ctxt->userData,
1152 "xmlns: %s not a valid URI\n", val);
1153 } else {
1154 if (uri->scheme == NULL) {
1155 if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL))
1156 ctxt->sax->warning(ctxt->userData,
1157 "xmlns: URI %s is not absolute\n", val);
1158 }
1159 xmlFreeURI(uri);
1160 }
1161 }
1162
1163 /* a default namespace definition */
1164 nsret = xmlNewNs(ctxt->node, val, NULL);
1165
1166 #ifdef LIBXML_VALID_ENABLED
1167 /*
1168 * Validate also for namespace decls, they are attributes from
1169 * an XML-1.0 perspective
1170 */
1171 if (nsret != NULL && ctxt->validate && ctxt->wellFormed &&
1172 ctxt->myDoc && ctxt->myDoc->intSubset)
1173 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc,
1174 ctxt->node, prefix, nsret, val);
1175 #endif /* LIBXML_VALID_ENABLED */
1176 if (name != NULL)
1177 xmlFree(name);
1178 if (nval != NULL)
1179 xmlFree(nval);
1180 if (val != value)
1181 xmlFree(val);
1182 return;
1183 }
1184 if ((!ctxt->html) &&
1185 (ns != NULL) && (ns[0] == 'x') && (ns[1] == 'm') && (ns[2] == 'l') &&
1186 (ns[3] == 'n') && (ns[4] == 's') && (ns[5] == 0)) {
1187 xmlNsPtr nsret;
1188 xmlChar *val;
1189
1190 if (!ctxt->replaceEntities) {
1191 ctxt->depth++;
1192 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF,
1193 0,0,0);
1194 ctxt->depth--;
1195 if (val == NULL) {
1196 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement");
1197 xmlFree(ns);
1198 if (name != NULL)
1199 xmlFree(name);
1200 return;
1201 }
1202 } else {
1203 val = (xmlChar *) value;
1204 }
1205
1206 if (val[0] == 0) {
1207 xmlNsErrMsg(ctxt, XML_NS_ERR_EMPTY,
1208 "Empty namespace name for prefix %s\n", name, NULL);
1209 }
1210 if ((ctxt->pedantic != 0) && (val[0] != 0)) {
1211 xmlURIPtr uri;
1212
1213 uri = xmlParseURI((const char *)val);
1214 if (uri == NULL) {
1215 xmlNsWarnMsg(ctxt, XML_WAR_NS_URI,
1216 "xmlns:%s: %s not a valid URI\n", name, value);
1217 } else {
1218 if (uri->scheme == NULL) {
1219 xmlNsWarnMsg(ctxt, XML_WAR_NS_URI_RELATIVE,
1220 "xmlns:%s: URI %s is not absolute\n", name, value);
1221 }
1222 xmlFreeURI(uri);
1223 }
1224 }
1225
1226 /* a standard namespace definition */
1227 nsret = xmlNewNs(ctxt->node, val, name);
1228 xmlFree(ns);
1229 #ifdef LIBXML_VALID_ENABLED
1230 /*
1231 * Validate also for namespace decls, they are attributes from
1232 * an XML-1.0 perspective
1233 */
1234 if (nsret != NULL && ctxt->validate && ctxt->wellFormed &&
1235 ctxt->myDoc && ctxt->myDoc->intSubset)
1236 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc,
1237 ctxt->node, prefix, nsret, value);
1238 #endif /* LIBXML_VALID_ENABLED */
1239 if (name != NULL)
1240 xmlFree(name);
1241 if (nval != NULL)
1242 xmlFree(nval);
1243 if (val != value)
1244 xmlFree(val);
1245 return;
1246 }
1247
1248 if (ns != NULL) {
1249 xmlAttrPtr prop;
1250 namespace = xmlSearchNs(ctxt->myDoc, ctxt->node, ns);
1251 if (namespace == NULL) {
1252 xmlNsErrMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
1253 "Namespace prefix %s of attribute %s is not defined\n",
1254 ns, name);
1255 }
1256
1257 prop = ctxt->node->properties;
1258 while (prop != NULL) {
1259 if (prop->ns != NULL) {
1260 if ((xmlStrEqual(name, prop->name)) &&
1261 ((namespace == prop->ns) ||
1262 (xmlStrEqual(namespace->href, prop->ns->href)))) {
1263 xmlNsErrMsg(ctxt, XML_ERR_ATTRIBUTE_REDEFINED,
1264 "Attribute %s in %s redefined\n",
1265 name, namespace->href);
1266 ctxt->wellFormed = 0;
1267 if (ctxt->recovery == 0) ctxt->disableSAX = 1;
1268 goto error;
1269 }
1270 }
1271 prop = prop->next;
1272 }
1273 } else {
1274 namespace = NULL;
1275 }
1276
1277 /* !!!!!! <a toto:arg="" xmlns:toto="http://toto.com"> */
1278 ret = xmlNewNsPropEatName(ctxt->node, namespace, name, NULL);
1279
1280 if (ret != NULL) {
1281 if ((ctxt->replaceEntities == 0) && (!ctxt->html)) {
1282 xmlNodePtr tmp;
1283
1284 ret->children = xmlStringGetNodeList(ctxt->myDoc, value);
1285 tmp = ret->children;
1286 while (tmp != NULL) {
1287 tmp->parent = (xmlNodePtr) ret;
1288 if (tmp->next == NULL)
1289 ret->last = tmp;
1290 tmp = tmp->next;
1291 }
1292 } else if (value != NULL) {
1293 ret->children = xmlNewDocText(ctxt->myDoc, value);
1294 ret->last = ret->children;
1295 if (ret->children != NULL)
1296 ret->children->parent = (xmlNodePtr) ret;
1297 }
1298 }
1299
1300 #ifdef LIBXML_VALID_ENABLED
1301 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed &&
1302 ctxt->myDoc && ctxt->myDoc->intSubset) {
1303
1304 /*
1305 * If we don't substitute entities, the validation should be
1306 * done on a value with replaced entities anyway.
1307 */
1308 if (!ctxt->replaceEntities) {
1309 xmlChar *val;
1310
1311 ctxt->depth++;
1312 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF,
1313 0,0,0);
1314 ctxt->depth--;
1315
1316 if (val == NULL)
1317 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
1318 ctxt->myDoc, ctxt->node, ret, value);
1319 else {
1320 xmlChar *nvalnorm;
1321
1322 /*
1323 * Do the last stage of the attribute normalization
1324 * It need to be done twice ... it's an extra burden related
1325 * to the ability to keep xmlSAX2References in attributes
1326 */
1327 nvalnorm = xmlValidNormalizeAttributeValue(ctxt->myDoc,
1328 ctxt->node, fullname, val);
1329 if (nvalnorm != NULL) {
1330 xmlFree(val);
1331 val = nvalnorm;
1332 }
1333
1334 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
1335 ctxt->myDoc, ctxt->node, ret, val);
1336 xmlFree(val);
1337 }
1338 } else {
1339 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, ctxt->myDoc,
1340 ctxt->node, ret, value);
1341 }
1342 } else
1343 #endif /* LIBXML_VALID_ENABLED */
1344 if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) &&
1345 (((ctxt->replaceEntities == 0) && (ctxt->external != 2)) ||
1346 ((ctxt->replaceEntities != 0) && (ctxt->inSubset == 0)))) {
1347 /*
1348 * when validating, the ID registration is done at the attribute
1349 * validation level. Otherwise we have to do specific handling here.
1350 */
1351 if (xmlStrEqual(fullname, BAD_CAST "xml:id")) {
1352 /*
1353 * Add the xml:id value
1354 *
1355 * Open issue: normalization of the value.
1356 */
1357 if (xmlValidateNCName(value, 1) != 0) {
1358 xmlErrValid(ctxt, XML_DTD_XMLID_VALUE,
1359 "xml:id : attribute value %s is not an NCName\n",
1360 (const char *) value, NULL);
1361 }
1362 xmlAddID(&ctxt->vctxt, ctxt->myDoc, value, ret);
1363 } else if (xmlIsID(ctxt->myDoc, ctxt->node, ret))
1364 xmlAddID(&ctxt->vctxt, ctxt->myDoc, value, ret);
1365 else if (xmlIsRef(ctxt->myDoc, ctxt->node, ret))
1366 xmlAddRef(&ctxt->vctxt, ctxt->myDoc, value, ret);
1367 }
1368
1369 error:
1370 if (nval != NULL)
1371 xmlFree(nval);
1372 if (ns != NULL)
1373 xmlFree(ns);
1374 }
1375
1376 /*
1377 * xmlCheckDefaultedAttributes:
1378 *
1379 * Check defaulted attributes from the DTD
1380 */
1381 static void
xmlCheckDefaultedAttributes(xmlParserCtxtPtr ctxt,const xmlChar * name,const xmlChar * prefix,const xmlChar ** atts)1382 xmlCheckDefaultedAttributes(xmlParserCtxtPtr ctxt, const xmlChar *name,
1383 const xmlChar *prefix, const xmlChar **atts) {
1384 xmlElementPtr elemDecl;
1385 const xmlChar *att;
1386 int internal = 1;
1387 int i;
1388
1389 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->intSubset, name, prefix);
1390 if (elemDecl == NULL) {
1391 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->extSubset, name, prefix);
1392 internal = 0;
1393 }
1394
1395 process_external_subset:
1396
1397 if (elemDecl != NULL) {
1398 xmlAttributePtr attr = elemDecl->attributes;
1399 /*
1400 * Check against defaulted attributes from the external subset
1401 * if the document is stamped as standalone
1402 */
1403 if ((ctxt->myDoc->standalone == 1) &&
1404 (ctxt->myDoc->extSubset != NULL) &&
1405 (ctxt->validate)) {
1406 while (attr != NULL) {
1407 if ((attr->defaultValue != NULL) &&
1408 (xmlGetDtdQAttrDesc(ctxt->myDoc->extSubset,
1409 attr->elem, attr->name,
1410 attr->prefix) == attr) &&
1411 (xmlGetDtdQAttrDesc(ctxt->myDoc->intSubset,
1412 attr->elem, attr->name,
1413 attr->prefix) == NULL)) {
1414 xmlChar *fulln;
1415
1416 if (attr->prefix != NULL) {
1417 fulln = xmlStrdup(attr->prefix);
1418 fulln = xmlStrcat(fulln, BAD_CAST ":");
1419 fulln = xmlStrcat(fulln, attr->name);
1420 } else {
1421 fulln = xmlStrdup(attr->name);
1422 }
1423
1424 /*
1425 * Check that the attribute is not declared in the
1426 * serialization
1427 */
1428 att = NULL;
1429 if (atts != NULL) {
1430 i = 0;
1431 att = atts[i];
1432 while (att != NULL) {
1433 if (xmlStrEqual(att, fulln))
1434 break;
1435 i += 2;
1436 att = atts[i];
1437 }
1438 }
1439 if (att == NULL) {
1440 xmlErrValid(ctxt, XML_DTD_STANDALONE_DEFAULTED,
1441 "standalone: attribute %s on %s defaulted from external subset\n",
1442 (const char *)fulln,
1443 (const char *)attr->elem);
1444 }
1445 }
1446 attr = attr->nexth;
1447 }
1448 }
1449
1450 /*
1451 * Actually insert defaulted values when needed
1452 */
1453 attr = elemDecl->attributes;
1454 while (attr != NULL) {
1455 /*
1456 * Make sure that attributes redefinition occuring in the
1457 * internal subset are not overriden by definitions in the
1458 * external subset.
1459 */
1460 if (attr->defaultValue != NULL) {
1461 /*
1462 * the element should be instantiated in the tree if:
1463 * - this is a namespace prefix
1464 * - the user required for completion in the tree
1465 * like XSLT
1466 * - there isn't already an attribute definition
1467 * in the internal subset overriding it.
1468 */
1469 if (((attr->prefix != NULL) &&
1470 (xmlStrEqual(attr->prefix, BAD_CAST "xmlns"))) ||
1471 ((attr->prefix == NULL) &&
1472 (xmlStrEqual(attr->name, BAD_CAST "xmlns"))) ||
1473 (ctxt->loadsubset & XML_COMPLETE_ATTRS)) {
1474 xmlAttributePtr tst;
1475
1476 tst = xmlGetDtdQAttrDesc(ctxt->myDoc->intSubset,
1477 attr->elem, attr->name,
1478 attr->prefix);
1479 if ((tst == attr) || (tst == NULL)) {
1480 xmlChar fn[50];
1481 xmlChar *fulln;
1482
1483 fulln = xmlBuildQName(attr->name, attr->prefix, fn, 50);
1484 if (fulln == NULL) {
1485 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement");
1486 return;
1487 }
1488
1489 /*
1490 * Check that the attribute is not declared in the
1491 * serialization
1492 */
1493 att = NULL;
1494 if (atts != NULL) {
1495 i = 0;
1496 att = atts[i];
1497 while (att != NULL) {
1498 if (xmlStrEqual(att, fulln))
1499 break;
1500 i += 2;
1501 att = atts[i];
1502 }
1503 }
1504 if (att == NULL) {
1505 xmlSAX2AttributeInternal(ctxt, fulln,
1506 attr->defaultValue, prefix);
1507 }
1508 if ((fulln != fn) && (fulln != attr->name))
1509 xmlFree(fulln);
1510 }
1511 }
1512 }
1513 attr = attr->nexth;
1514 }
1515 if (internal == 1) {
1516 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->extSubset,
1517 name, prefix);
1518 internal = 0;
1519 goto process_external_subset;
1520 }
1521 }
1522 }
1523
1524 /**
1525 * xmlSAX2StartElement:
1526 * @ctx: the user data (XML parser context)
1527 * @fullname: The element name, including namespace prefix
1528 * @atts: An array of name/value attributes pairs, NULL terminated
1529 *
1530 * called when an opening tag has been processed.
1531 */
1532 void
xmlSAX2StartElement(void * ctx,const xmlChar * fullname,const xmlChar ** atts)1533 xmlSAX2StartElement(void *ctx, const xmlChar *fullname, const xmlChar **atts)
1534 {
1535 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
1536 xmlNodePtr ret;
1537 xmlNodePtr parent;
1538 xmlNsPtr ns;
1539 xmlChar *name;
1540 xmlChar *prefix;
1541 const xmlChar *att;
1542 const xmlChar *value;
1543 int i;
1544
1545 if ((ctx == NULL) || (fullname == NULL) || (ctxt->myDoc == NULL)) return;
1546 parent = ctxt->node;
1547 #ifdef DEBUG_SAX
1548 xmlGenericError(xmlGenericErrorContext,
1549 "SAX.xmlSAX2StartElement(%s)\n", fullname);
1550 #endif
1551
1552 /*
1553 * First check on validity:
1554 */
1555 if (ctxt->validate && (ctxt->myDoc->extSubset == NULL) &&
1556 ((ctxt->myDoc->intSubset == NULL) ||
1557 ((ctxt->myDoc->intSubset->notations == NULL) &&
1558 (ctxt->myDoc->intSubset->elements == NULL) &&
1559 (ctxt->myDoc->intSubset->attributes == NULL) &&
1560 (ctxt->myDoc->intSubset->entities == NULL)))) {
1561 xmlErrValid(ctxt, XML_ERR_NO_DTD,
1562 "Validation failed: no DTD found !", NULL, NULL);
1563 ctxt->validate = 0;
1564 }
1565
1566
1567 /*
1568 * Split the full name into a namespace prefix and the tag name
1569 */
1570 name = xmlSplitQName(ctxt, fullname, &prefix);
1571
1572
1573 /*
1574 * Note : the namespace resolution is deferred until the end of the
1575 * attributes parsing, since local namespace can be defined as
1576 * an attribute at this level.
1577 */
1578 ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, name, NULL);
1579 if (ret == NULL) {
1580 if (prefix != NULL)
1581 xmlFree(prefix);
1582 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement");
1583 return;
1584 }
1585 if (ctxt->myDoc->children == NULL) {
1586 #ifdef DEBUG_SAX_TREE
1587 xmlGenericError(xmlGenericErrorContext, "Setting %s as root\n", name);
1588 #endif
1589 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret);
1590 } else if (parent == NULL) {
1591 parent = ctxt->myDoc->children;
1592 }
1593 ctxt->nodemem = -1;
1594 if (ctxt->linenumbers) {
1595 if (ctxt->input != NULL) {
1596 if (ctxt->input->line < 65535)
1597 ret->line = (short) ctxt->input->line;
1598 else
1599 ret->line = 65535;
1600 }
1601 }
1602
1603 /*
1604 * We are parsing a new node.
1605 */
1606 #ifdef DEBUG_SAX_TREE
1607 xmlGenericError(xmlGenericErrorContext, "pushing(%s)\n", name);
1608 #endif
1609 nodePush(ctxt, ret);
1610
1611 /*
1612 * Link the child element
1613 */
1614 if (parent != NULL) {
1615 if (parent->type == XML_ELEMENT_NODE) {
1616 #ifdef DEBUG_SAX_TREE
1617 xmlGenericError(xmlGenericErrorContext,
1618 "adding child %s to %s\n", name, parent->name);
1619 #endif
1620 xmlAddChild(parent, ret);
1621 } else {
1622 #ifdef DEBUG_SAX_TREE
1623 xmlGenericError(xmlGenericErrorContext,
1624 "adding sibling %s to ", name);
1625 xmlDebugDumpOneNode(stderr, parent, 0);
1626 #endif
1627 xmlAddSibling(parent, ret);
1628 }
1629 }
1630
1631 /*
1632 * Insert all the defaulted attributes from the DTD especially namespaces
1633 */
1634 if ((!ctxt->html) &&
1635 ((ctxt->myDoc->intSubset != NULL) ||
1636 (ctxt->myDoc->extSubset != NULL))) {
1637 xmlCheckDefaultedAttributes(ctxt, name, prefix, atts);
1638 }
1639
1640 /*
1641 * process all the attributes whose name start with "xmlns"
1642 */
1643 if (atts != NULL) {
1644 i = 0;
1645 att = atts[i++];
1646 value = atts[i++];
1647 if (!ctxt->html) {
1648 while ((att != NULL) && (value != NULL)) {
1649 if ((att[0] == 'x') && (att[1] == 'm') && (att[2] == 'l') &&
1650 (att[3] == 'n') && (att[4] == 's'))
1651 xmlSAX2AttributeInternal(ctxt, att, value, prefix);
1652
1653 att = atts[i++];
1654 value = atts[i++];
1655 }
1656 }
1657 }
1658
1659 /*
1660 * Search the namespace, note that since the attributes have been
1661 * processed, the local namespaces are available.
1662 */
1663 ns = xmlSearchNs(ctxt->myDoc, ret, prefix);
1664 if ((ns == NULL) && (parent != NULL))
1665 ns = xmlSearchNs(ctxt->myDoc, parent, prefix);
1666 if ((prefix != NULL) && (ns == NULL)) {
1667 ns = xmlNewNs(ret, NULL, prefix);
1668 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
1669 "Namespace prefix %s is not defined\n",
1670 prefix, NULL);
1671 }
1672
1673 /*
1674 * set the namespace node, making sure that if the default namspace
1675 * is unbound on a parent we simply kee it NULL
1676 */
1677 if ((ns != NULL) && (ns->href != NULL) &&
1678 ((ns->href[0] != 0) || (ns->prefix != NULL)))
1679 xmlSetNs(ret, ns);
1680
1681 /*
1682 * process all the other attributes
1683 */
1684 if (atts != NULL) {
1685 i = 0;
1686 att = atts[i++];
1687 value = atts[i++];
1688 if (ctxt->html) {
1689 while (att != NULL) {
1690 xmlSAX2AttributeInternal(ctxt, att, value, NULL);
1691 att = atts[i++];
1692 value = atts[i++];
1693 }
1694 } else {
1695 while ((att != NULL) && (value != NULL)) {
1696 if ((att[0] != 'x') || (att[1] != 'm') || (att[2] != 'l') ||
1697 (att[3] != 'n') || (att[4] != 's'))
1698 xmlSAX2AttributeInternal(ctxt, att, value, NULL);
1699
1700 /*
1701 * Next ones
1702 */
1703 att = atts[i++];
1704 value = atts[i++];
1705 }
1706 }
1707 }
1708
1709 #ifdef LIBXML_VALID_ENABLED
1710 /*
1711 * If it's the Document root, finish the DTD validation and
1712 * check the document root element for validity
1713 */
1714 if ((ctxt->validate) && (ctxt->vctxt.finishDtd == XML_CTXT_FINISH_DTD_0)) {
1715 int chk;
1716
1717 chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc);
1718 if (chk <= 0)
1719 ctxt->valid = 0;
1720 if (chk < 0)
1721 ctxt->wellFormed = 0;
1722 ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
1723 ctxt->vctxt.finishDtd = XML_CTXT_FINISH_DTD_1;
1724 }
1725 #endif /* LIBXML_VALID_ENABLED */
1726
1727 if (prefix != NULL)
1728 xmlFree(prefix);
1729
1730 }
1731
1732 /**
1733 * xmlSAX2EndElement:
1734 * @ctx: the user data (XML parser context)
1735 * @name: The element name
1736 *
1737 * called when the end of an element has been detected.
1738 */
1739 void
xmlSAX2EndElement(void * ctx,const xmlChar * name ATTRIBUTE_UNUSED)1740 xmlSAX2EndElement(void *ctx, const xmlChar *name ATTRIBUTE_UNUSED)
1741 {
1742 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
1743 xmlParserNodeInfo node_info;
1744 xmlNodePtr cur;
1745
1746 if (ctx == NULL) return;
1747 cur = ctxt->node;
1748 #ifdef DEBUG_SAX
1749 if (name == NULL)
1750 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2EndElement(NULL)\n");
1751 else
1752 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2EndElement(%s)\n", name);
1753 #endif
1754
1755 /* Capture end position and add node */
1756 if (cur != NULL && ctxt->record_info) {
1757 node_info.end_pos = ctxt->input->cur - ctxt->input->base;
1758 node_info.end_line = ctxt->input->line;
1759 node_info.node = cur;
1760 xmlParserAddNodeInfo(ctxt, &node_info);
1761 }
1762 ctxt->nodemem = -1;
1763
1764 #ifdef LIBXML_VALID_ENABLED
1765 if (ctxt->validate && ctxt->wellFormed &&
1766 ctxt->myDoc && ctxt->myDoc->intSubset)
1767 ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc,
1768 cur);
1769 #endif /* LIBXML_VALID_ENABLED */
1770
1771
1772 /*
1773 * end of parsing of this node.
1774 */
1775 #ifdef DEBUG_SAX_TREE
1776 xmlGenericError(xmlGenericErrorContext, "popping(%s)\n", cur->name);
1777 #endif
1778 nodePop(ctxt);
1779 }
1780 #endif /* LIBXML_SAX1_ENABLED || LIBXML_HTML_ENABLE */
1781
1782 /*
1783 * xmlSAX2TextNode:
1784 * @ctxt: the parser context
1785 * @str: the input string
1786 * @len: the string length
1787 *
1788 * Remove the entities from an attribute value
1789 *
1790 * Returns the newly allocated string or NULL if not needed or error
1791 */
1792 static xmlNodePtr
xmlSAX2TextNode(xmlParserCtxtPtr ctxt,const xmlChar * str,int len)1793 xmlSAX2TextNode(xmlParserCtxtPtr ctxt, const xmlChar *str, int len) {
1794 xmlNodePtr ret;
1795 const xmlChar *intern = NULL;
1796
1797 /*
1798 * Allocate
1799 */
1800 if (ctxt->freeElems != NULL) {
1801 ret = ctxt->freeElems;
1802 ctxt->freeElems = ret->next;
1803 ctxt->freeElemsNr--;
1804 } else {
1805 ret = (xmlNodePtr) xmlMalloc(sizeof(xmlNode));
1806 }
1807 if (ret == NULL) {
1808 xmlErrMemory(ctxt, "xmlSAX2Characters");
1809 return(NULL);
1810 }
1811 memset(ret, 0, sizeof(xmlNode));
1812 /*
1813 * intern the formatting blanks found between tags, or the
1814 * very short strings
1815 */
1816 if (ctxt->dictNames) {
1817 xmlChar cur = str[len];
1818
1819 if ((len < (int) (2 * sizeof(void *))) &&
1820 (ctxt->options & XML_PARSE_COMPACT)) {
1821 /* store the string in the node overrithing properties and nsDef */
1822 xmlChar *tmp = (xmlChar *) &(ret->properties);
1823 memcpy(tmp, str, len);
1824 tmp[len] = 0;
1825 intern = tmp;
1826 } else if ((len <= 3) && ((cur == '"') || (cur == '\'') ||
1827 ((cur == '<') && (str[len + 1] != '!')))) {
1828 intern = xmlDictLookup(ctxt->dict, str, len);
1829 } else if (IS_BLANK_CH(*str) && (len < 60) && (cur == '<') &&
1830 (str[len + 1] != '!')) {
1831 int i;
1832
1833 for (i = 1;i < len;i++) {
1834 if (!IS_BLANK_CH(str[i])) goto skip;
1835 }
1836 intern = xmlDictLookup(ctxt->dict, str, len);
1837 }
1838 }
1839 skip:
1840 ret->type = XML_TEXT_NODE;
1841
1842 ret->name = xmlStringText;
1843 if (intern == NULL) {
1844 ret->content = xmlStrndup(str, len);
1845 if (ret->content == NULL) {
1846 xmlSAX2ErrMemory(ctxt, "xmlSAX2TextNode");
1847 xmlFree(ret);
1848 return(NULL);
1849 }
1850 } else
1851 ret->content = (xmlChar *) intern;
1852
1853 if (ctxt->input != NULL)
1854 ret->line = ctxt->input->line;
1855
1856 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue))
1857 xmlRegisterNodeDefaultValue(ret);
1858 return(ret);
1859 }
1860
1861 #ifdef LIBXML_VALID_ENABLED
1862 /*
1863 * xmlSAX2DecodeAttrEntities:
1864 * @ctxt: the parser context
1865 * @str: the input string
1866 * @len: the string length
1867 *
1868 * Remove the entities from an attribute value
1869 *
1870 * Returns the newly allocated string or NULL if not needed or error
1871 */
1872 static xmlChar *
xmlSAX2DecodeAttrEntities(xmlParserCtxtPtr ctxt,const xmlChar * str,const xmlChar * end)1873 xmlSAX2DecodeAttrEntities(xmlParserCtxtPtr ctxt, const xmlChar *str,
1874 const xmlChar *end) {
1875 const xmlChar *in;
1876 xmlChar *ret;
1877
1878 in = str;
1879 while (in < end)
1880 if (*in++ == '&')
1881 goto decode;
1882 return(NULL);
1883 decode:
1884 ctxt->depth++;
1885 ret = xmlStringLenDecodeEntities(ctxt, str, end - str,
1886 XML_SUBSTITUTE_REF, 0,0,0);
1887 ctxt->depth--;
1888 return(ret);
1889 }
1890 #endif /* LIBXML_VALID_ENABLED */
1891
1892 /**
1893 * xmlSAX2AttributeNs:
1894 * @ctx: the user data (XML parser context)
1895 * @localname: the local name of the attribute
1896 * @prefix: the attribute namespace prefix if available
1897 * @URI: the attribute namespace name if available
1898 * @value: Start of the attribute value
1899 * @valueend: end of the attribute value
1900 *
1901 * Handle an attribute that has been read by the parser.
1902 * The default handling is to convert the attribute into an
1903 * DOM subtree and past it in a new xmlAttr element added to
1904 * the element.
1905 */
1906 static void
xmlSAX2AttributeNs(xmlParserCtxtPtr ctxt,const xmlChar * localname,const xmlChar * prefix,const xmlChar * value,const xmlChar * valueend)1907 xmlSAX2AttributeNs(xmlParserCtxtPtr ctxt,
1908 const xmlChar * localname,
1909 const xmlChar * prefix,
1910 const xmlChar * value,
1911 const xmlChar * valueend)
1912 {
1913 xmlAttrPtr ret;
1914 xmlNsPtr namespace = NULL;
1915 xmlChar *dup = NULL;
1916
1917 /*
1918 * Note: if prefix == NULL, the attribute is not in the default namespace
1919 */
1920 if (prefix != NULL)
1921 namespace = xmlSearchNs(ctxt->myDoc, ctxt->node, prefix);
1922
1923 /*
1924 * allocate the node
1925 */
1926 if (ctxt->freeAttrs != NULL) {
1927 ret = ctxt->freeAttrs;
1928 ctxt->freeAttrs = ret->next;
1929 ctxt->freeAttrsNr--;
1930 memset(ret, 0, sizeof(xmlAttr));
1931 ret->type = XML_ATTRIBUTE_NODE;
1932
1933 ret->parent = ctxt->node;
1934 ret->doc = ctxt->myDoc;
1935 ret->ns = namespace;
1936
1937 if (ctxt->dictNames)
1938 ret->name = localname;
1939 else
1940 ret->name = xmlStrdup(localname);
1941
1942 /* link at the end to preserv order, TODO speed up with a last */
1943 if (ctxt->node->properties == NULL) {
1944 ctxt->node->properties = ret;
1945 } else {
1946 xmlAttrPtr prev = ctxt->node->properties;
1947
1948 while (prev->next != NULL) prev = prev->next;
1949 prev->next = ret;
1950 ret->prev = prev;
1951 }
1952
1953 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue))
1954 xmlRegisterNodeDefaultValue((xmlNodePtr)ret);
1955 } else {
1956 if (ctxt->dictNames)
1957 ret = xmlNewNsPropEatName(ctxt->node, namespace,
1958 (xmlChar *) localname, NULL);
1959 else
1960 ret = xmlNewNsProp(ctxt->node, namespace, localname, NULL);
1961 if (ret == NULL) {
1962 xmlErrMemory(ctxt, "xmlSAX2AttributeNs");
1963 return;
1964 }
1965 }
1966
1967 if ((ctxt->replaceEntities == 0) && (!ctxt->html)) {
1968 xmlNodePtr tmp;
1969
1970 /*
1971 * We know that if there is an entity reference, then
1972 * the string has been dup'ed and terminates with 0
1973 * otherwise with ' or "
1974 */
1975 if (*valueend != 0) {
1976 tmp = xmlSAX2TextNode(ctxt, value, valueend - value);
1977 ret->children = tmp;
1978 ret->last = tmp;
1979 if (tmp != NULL) {
1980 tmp->doc = ret->doc;
1981 tmp->parent = (xmlNodePtr) ret;
1982 }
1983 } else {
1984 ret->children = xmlStringLenGetNodeList(ctxt->myDoc, value,
1985 valueend - value);
1986 tmp = ret->children;
1987 while (tmp != NULL) {
1988 tmp->doc = ret->doc;
1989 tmp->parent = (xmlNodePtr) ret;
1990 if (tmp->next == NULL)
1991 ret->last = tmp;
1992 tmp = tmp->next;
1993 }
1994 }
1995 } else if (value != NULL) {
1996 xmlNodePtr tmp;
1997
1998 tmp = xmlSAX2TextNode(ctxt, value, valueend - value);
1999 ret->children = tmp;
2000 ret->last = tmp;
2001 if (tmp != NULL) {
2002 tmp->doc = ret->doc;
2003 tmp->parent = (xmlNodePtr) ret;
2004 }
2005 }
2006
2007 #ifdef LIBXML_VALID_ENABLED
2008 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed &&
2009 ctxt->myDoc && ctxt->myDoc->intSubset) {
2010 /*
2011 * If we don't substitute entities, the validation should be
2012 * done on a value with replaced entities anyway.
2013 */
2014 if (!ctxt->replaceEntities) {
2015 dup = xmlSAX2DecodeAttrEntities(ctxt, value, valueend);
2016 if (dup == NULL) {
2017 if (*valueend == 0) {
2018 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
2019 ctxt->myDoc, ctxt->node, ret, value);
2020 } else {
2021 /*
2022 * That should already be normalized.
2023 * cheaper to finally allocate here than duplicate
2024 * entry points in the full validation code
2025 */
2026 dup = xmlStrndup(value, valueend - value);
2027
2028 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
2029 ctxt->myDoc, ctxt->node, ret, dup);
2030 }
2031 } else {
2032 /*
2033 * dup now contains a string of the flattened attribute
2034 * content with entities substitued. Check if we need to
2035 * apply an extra layer of normalization.
2036 * It need to be done twice ... it's an extra burden related
2037 * to the ability to keep references in attributes
2038 */
2039 if (ctxt->attsSpecial != NULL) {
2040 xmlChar *nvalnorm;
2041 xmlChar fn[50];
2042 xmlChar *fullname;
2043
2044 fullname = xmlBuildQName(localname, prefix, fn, 50);
2045 if (fullname != NULL) {
2046 ctxt->vctxt.valid = 1;
2047 nvalnorm = xmlValidCtxtNormalizeAttributeValue(
2048 &ctxt->vctxt, ctxt->myDoc,
2049 ctxt->node, fullname, dup);
2050 if (ctxt->vctxt.valid != 1)
2051 ctxt->valid = 0;
2052
2053 if ((fullname != fn) && (fullname != localname))
2054 xmlFree(fullname);
2055 if (nvalnorm != NULL) {
2056 xmlFree(dup);
2057 dup = nvalnorm;
2058 }
2059 }
2060 }
2061
2062 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
2063 ctxt->myDoc, ctxt->node, ret, dup);
2064 }
2065 } else {
2066 /*
2067 * if entities already have been substitued, then
2068 * the attribute as passed is already normalized
2069 */
2070 dup = xmlStrndup(value, valueend - value);
2071
2072 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
2073 ctxt->myDoc, ctxt->node, ret, dup);
2074 }
2075 } else
2076 #endif /* LIBXML_VALID_ENABLED */
2077 if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) &&
2078 (((ctxt->replaceEntities == 0) && (ctxt->external != 2)) ||
2079 ((ctxt->replaceEntities != 0) && (ctxt->inSubset == 0)))) {
2080 /*
2081 * when validating, the ID registration is done at the attribute
2082 * validation level. Otherwise we have to do specific handling here.
2083 */
2084 if ((prefix == ctxt->str_xml) &&
2085 (localname[0] == 'i') && (localname[1] == 'd') &&
2086 (localname[2] == 0)) {
2087 /*
2088 * Add the xml:id value
2089 *
2090 * Open issue: normalization of the value.
2091 */
2092 if (dup == NULL)
2093 dup = xmlStrndup(value, valueend - value);
2094 #ifdef LIBXML_VALID_ENABLED
2095 if (xmlValidateNCName(dup, 1) != 0) {
2096 xmlErrValid(ctxt, XML_DTD_XMLID_VALUE,
2097 "xml:id : attribute value %s is not an NCName\n",
2098 (const char *) dup, NULL);
2099 }
2100 #endif
2101 xmlAddID(&ctxt->vctxt, ctxt->myDoc, dup, ret);
2102 } else if (xmlIsID(ctxt->myDoc, ctxt->node, ret)) {
2103 /* might be worth duplicate entry points and not copy */
2104 if (dup == NULL)
2105 dup = xmlStrndup(value, valueend - value);
2106 xmlAddID(&ctxt->vctxt, ctxt->myDoc, dup, ret);
2107 } else if (xmlIsRef(ctxt->myDoc, ctxt->node, ret)) {
2108 if (dup == NULL)
2109 dup = xmlStrndup(value, valueend - value);
2110 xmlAddRef(&ctxt->vctxt, ctxt->myDoc, dup, ret);
2111 }
2112 }
2113 if (dup != NULL)
2114 xmlFree(dup);
2115 }
2116
2117 /**
2118 * xmlSAX2StartElementNs:
2119 * @ctx: the user data (XML parser context)
2120 * @localname: the local name of the element
2121 * @prefix: the element namespace prefix if available
2122 * @URI: the element namespace name if available
2123 * @nb_namespaces: number of namespace definitions on that node
2124 * @namespaces: pointer to the array of prefix/URI pairs namespace definitions
2125 * @nb_attributes: the number of attributes on that node
2126 * @nb_defaulted: the number of defaulted attributes.
2127 * @attributes: pointer to the array of (localname/prefix/URI/value/end)
2128 * attribute values.
2129 *
2130 * SAX2 callback when an element start has been detected by the parser.
2131 * It provides the namespace informations for the element, as well as
2132 * the new namespace declarations on the element.
2133 */
2134 void
xmlSAX2StartElementNs(void * ctx,const xmlChar * localname,const xmlChar * prefix,const xmlChar * URI,int nb_namespaces,const xmlChar ** namespaces,int nb_attributes,int nb_defaulted,const xmlChar ** attributes)2135 xmlSAX2StartElementNs(void *ctx,
2136 const xmlChar *localname,
2137 const xmlChar *prefix,
2138 const xmlChar *URI,
2139 int nb_namespaces,
2140 const xmlChar **namespaces,
2141 int nb_attributes,
2142 int nb_defaulted,
2143 const xmlChar **attributes)
2144 {
2145 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2146 xmlNodePtr ret;
2147 xmlNodePtr parent;
2148 xmlNsPtr last = NULL, ns;
2149 const xmlChar *uri, *pref;
2150 int i, j;
2151
2152 if (ctx == NULL) return;
2153 parent = ctxt->node;
2154 /*
2155 * First check on validity:
2156 */
2157 if (ctxt->validate && (ctxt->myDoc->extSubset == NULL) &&
2158 ((ctxt->myDoc->intSubset == NULL) ||
2159 ((ctxt->myDoc->intSubset->notations == NULL) &&
2160 (ctxt->myDoc->intSubset->elements == NULL) &&
2161 (ctxt->myDoc->intSubset->attributes == NULL) &&
2162 (ctxt->myDoc->intSubset->entities == NULL)))) {
2163 xmlErrValid(ctxt, XML_ERR_NO_DTD,
2164 "Validation failed: no DTD found !", NULL, NULL);
2165 ctxt->validate = 0;
2166 }
2167
2168 /*
2169 * allocate the node
2170 */
2171 if (ctxt->freeElems != NULL) {
2172 ret = ctxt->freeElems;
2173 ctxt->freeElems = ret->next;
2174 ctxt->freeElemsNr--;
2175 memset(ret, 0, sizeof(xmlNode));
2176 ret->type = XML_ELEMENT_NODE;
2177
2178 if (ctxt->dictNames)
2179 ret->name = localname;
2180 else {
2181 ret->name = xmlStrdup(localname);
2182 if (ret->name == NULL) {
2183 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs");
2184 return;
2185 }
2186 }
2187 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue))
2188 xmlRegisterNodeDefaultValue(ret);
2189 } else {
2190 if (ctxt->dictNames)
2191 ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL,
2192 (xmlChar *) localname, NULL);
2193 else
2194 ret = xmlNewDocNode(ctxt->myDoc, NULL, localname, NULL);
2195 if (ret == NULL) {
2196 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs");
2197 return;
2198 }
2199 }
2200 if (ctxt->linenumbers) {
2201 if (ctxt->input != NULL) {
2202 if (ctxt->input->line < 65535)
2203 ret->line = (short) ctxt->input->line;
2204 else
2205 ret->line = 65535;
2206 }
2207 }
2208
2209 if ((ctxt->myDoc->children == NULL) || (parent == NULL)) {
2210 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret);
2211 }
2212 /*
2213 * Build the namespace list
2214 */
2215 for (i = 0,j = 0;j < nb_namespaces;j++) {
2216 pref = namespaces[i++];
2217 uri = namespaces[i++];
2218 ns = xmlNewNs(NULL, uri, pref);
2219 if (ns != NULL) {
2220 if (last == NULL) {
2221 ret->nsDef = last = ns;
2222 } else {
2223 last->next = ns;
2224 last = ns;
2225 }
2226 if ((URI != NULL) && (prefix == pref))
2227 ret->ns = ns;
2228 } else {
2229 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs");
2230 return;
2231 }
2232 #ifdef LIBXML_VALID_ENABLED
2233 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed &&
2234 ctxt->myDoc && ctxt->myDoc->intSubset) {
2235 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc,
2236 ret, prefix, ns, uri);
2237 }
2238 #endif /* LIBXML_VALID_ENABLED */
2239 }
2240 ctxt->nodemem = -1;
2241
2242 /*
2243 * We are parsing a new node.
2244 */
2245 nodePush(ctxt, ret);
2246
2247 /*
2248 * Link the child element
2249 */
2250 if (parent != NULL) {
2251 if (parent->type == XML_ELEMENT_NODE) {
2252 xmlAddChild(parent, ret);
2253 } else {
2254 xmlAddSibling(parent, ret);
2255 }
2256 }
2257
2258 /*
2259 * Insert the defaulted attributes from the DTD only if requested:
2260 */
2261 if ((nb_defaulted != 0) &&
2262 ((ctxt->loadsubset & XML_COMPLETE_ATTRS) == 0))
2263 nb_attributes -= nb_defaulted;
2264
2265 /*
2266 * Search the namespace if it wasn't already found
2267 * Note that, if prefix is NULL, this searches for the default Ns
2268 */
2269 if ((URI != NULL) && (ret->ns == NULL)) {
2270 ret->ns = xmlSearchNs(ctxt->myDoc, parent, prefix);
2271 if ((ret->ns == NULL) && (xmlStrEqual(prefix, BAD_CAST "xml"))) {
2272 ret->ns = xmlSearchNs(ctxt->myDoc, ret, prefix);
2273 }
2274 if (ret->ns == NULL) {
2275 ns = xmlNewNs(ret, NULL, prefix);
2276 if (ns == NULL) {
2277
2278 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs");
2279 return;
2280 }
2281 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
2282 "Namespace prefix %s was not found\n",
2283 prefix, NULL);
2284 }
2285 }
2286
2287 /*
2288 * process all the other attributes
2289 */
2290 if (nb_attributes > 0) {
2291 for (j = 0,i = 0;i < nb_attributes;i++,j+=5) {
2292 xmlSAX2AttributeNs(ctxt, attributes[j], attributes[j+1],
2293 attributes[j+3], attributes[j+4]);
2294 }
2295 }
2296
2297 #ifdef LIBXML_VALID_ENABLED
2298 /*
2299 * If it's the Document root, finish the DTD validation and
2300 * check the document root element for validity
2301 */
2302 if ((ctxt->validate) && (ctxt->vctxt.finishDtd == XML_CTXT_FINISH_DTD_0)) {
2303 int chk;
2304
2305 chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc);
2306 if (chk <= 0)
2307 ctxt->valid = 0;
2308 if (chk < 0)
2309 ctxt->wellFormed = 0;
2310 ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
2311 ctxt->vctxt.finishDtd = XML_CTXT_FINISH_DTD_1;
2312 }
2313 #endif /* LIBXML_VALID_ENABLED */
2314 }
2315
2316 /**
2317 * xmlSAX2EndElementNs:
2318 * @ctx: the user data (XML parser context)
2319 * @localname: the local name of the element
2320 * @prefix: the element namespace prefix if available
2321 * @URI: the element namespace name if available
2322 *
2323 * SAX2 callback when an element end has been detected by the parser.
2324 * It provides the namespace informations for the element.
2325 */
2326 void
xmlSAX2EndElementNs(void * ctx,const xmlChar * localname ATTRIBUTE_UNUSED,const xmlChar * prefix ATTRIBUTE_UNUSED,const xmlChar * URI ATTRIBUTE_UNUSED)2327 xmlSAX2EndElementNs(void *ctx,
2328 const xmlChar * localname ATTRIBUTE_UNUSED,
2329 const xmlChar * prefix ATTRIBUTE_UNUSED,
2330 const xmlChar * URI ATTRIBUTE_UNUSED)
2331 {
2332 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2333 xmlParserNodeInfo node_info;
2334 xmlNodePtr cur;
2335
2336 if (ctx == NULL) return;
2337 cur = ctxt->node;
2338 /* Capture end position and add node */
2339 if ((ctxt->record_info) && (cur != NULL)) {
2340 node_info.end_pos = ctxt->input->cur - ctxt->input->base;
2341 node_info.end_line = ctxt->input->line;
2342 node_info.node = cur;
2343 xmlParserAddNodeInfo(ctxt, &node_info);
2344 }
2345 ctxt->nodemem = -1;
2346
2347 #ifdef LIBXML_VALID_ENABLED
2348 if (ctxt->validate && ctxt->wellFormed &&
2349 ctxt->myDoc && ctxt->myDoc->intSubset)
2350 ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc, cur);
2351 #endif /* LIBXML_VALID_ENABLED */
2352
2353 /*
2354 * end of parsing of this node.
2355 */
2356 nodePop(ctxt);
2357 }
2358
2359 /**
2360 * xmlSAX2Reference:
2361 * @ctx: the user data (XML parser context)
2362 * @name: The entity name
2363 *
2364 * called when an entity xmlSAX2Reference is detected.
2365 */
2366 void
xmlSAX2Reference(void * ctx,const xmlChar * name)2367 xmlSAX2Reference(void *ctx, const xmlChar *name)
2368 {
2369 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2370 xmlNodePtr ret;
2371
2372 if (ctx == NULL) return;
2373 #ifdef DEBUG_SAX
2374 xmlGenericError(xmlGenericErrorContext,
2375 "SAX.xmlSAX2Reference(%s)\n", name);
2376 #endif
2377 if (name[0] == '#')
2378 ret = xmlNewCharRef(ctxt->myDoc, name);
2379 else
2380 ret = xmlNewReference(ctxt->myDoc, name);
2381 #ifdef DEBUG_SAX_TREE
2382 xmlGenericError(xmlGenericErrorContext,
2383 "add xmlSAX2Reference %s to %s \n", name, ctxt->node->name);
2384 #endif
2385 if (xmlAddChild(ctxt->node, ret) == NULL) {
2386 xmlFreeNode(ret);
2387 }
2388 }
2389
2390 /**
2391 * xmlSAX2Characters:
2392 * @ctx: the user data (XML parser context)
2393 * @ch: a xmlChar string
2394 * @len: the number of xmlChar
2395 *
2396 * receiving some chars from the parser.
2397 */
2398 void
xmlSAX2Characters(void * ctx,const xmlChar * ch,int len)2399 xmlSAX2Characters(void *ctx, const xmlChar *ch, int len)
2400 {
2401 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2402 xmlNodePtr lastChild;
2403
2404 if (ctx == NULL) return;
2405 #ifdef DEBUG_SAX
2406 xmlGenericError(xmlGenericErrorContext,
2407 "SAX.xmlSAX2Characters(%.30s, %d)\n", ch, len);
2408 #endif
2409 /*
2410 * Handle the data if any. If there is no child
2411 * add it as content, otherwise if the last child is text,
2412 * concatenate it, else create a new node of type text.
2413 */
2414
2415 if (ctxt->node == NULL) {
2416 #ifdef DEBUG_SAX_TREE
2417 xmlGenericError(xmlGenericErrorContext,
2418 "add chars: ctxt->node == NULL !\n");
2419 #endif
2420 return;
2421 }
2422 lastChild = ctxt->node->last;
2423 #ifdef DEBUG_SAX_TREE
2424 xmlGenericError(xmlGenericErrorContext,
2425 "add chars to %s \n", ctxt->node->name);
2426 #endif
2427
2428 /*
2429 * Here we needed an accelerator mechanism in case of very large
2430 * elements. Use an attribute in the structure !!!
2431 */
2432 if (lastChild == NULL) {
2433 lastChild = xmlSAX2TextNode(ctxt, ch, len);
2434 if (lastChild != NULL) {
2435 ctxt->node->children = lastChild;
2436 ctxt->node->last = lastChild;
2437 lastChild->parent = ctxt->node;
2438 lastChild->doc = ctxt->node->doc;
2439 ctxt->nodelen = len;
2440 ctxt->nodemem = len + 1;
2441 } else {
2442 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters");
2443 return;
2444 }
2445 } else {
2446 int coalesceText = (lastChild != NULL) &&
2447 (lastChild->type == XML_TEXT_NODE) &&
2448 (lastChild->name == xmlStringText);
2449 if ((coalesceText) && (ctxt->nodemem != 0)) {
2450 /*
2451 * The whole point of maintaining nodelen and nodemem,
2452 * xmlTextConcat is too costly, i.e. compute length,
2453 * reallocate a new buffer, move data, append ch. Here
2454 * We try to minimaze realloc() uses and avoid copying
2455 * and recomputing length over and over.
2456 */
2457 if (lastChild->content == (xmlChar *)&(lastChild->properties)) {
2458 lastChild->content = xmlStrdup(lastChild->content);
2459 lastChild->properties = NULL;
2460 } else if ((ctxt->nodemem == ctxt->nodelen + 1) &&
2461 (xmlDictOwns(ctxt->dict, lastChild->content))) {
2462 lastChild->content = xmlStrdup(lastChild->content);
2463 }
2464 if (((size_t)ctxt->nodelen + (size_t)len > XML_MAX_TEXT_LENGTH) &&
2465 ((ctxt->options & XML_PARSE_HUGE) == 0)) {
2466 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters: huge text node");
2467 return;
2468 }
2469 if ((size_t)ctxt->nodelen > SIZE_T_MAX - (size_t)len ||
2470 (size_t)ctxt->nodemem + (size_t)len > SIZE_T_MAX / 2) {
2471 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters overflow prevented");
2472 return;
2473 }
2474 if (ctxt->nodelen + len >= ctxt->nodemem) {
2475 xmlChar *newbuf;
2476 size_t size;
2477
2478 size = ctxt->nodemem + len;
2479 size *= 2;
2480 newbuf = (xmlChar *) xmlRealloc(lastChild->content,size);
2481 if (newbuf == NULL) {
2482 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters");
2483 return;
2484 }
2485 ctxt->nodemem = size;
2486 lastChild->content = newbuf;
2487 }
2488 memcpy(&lastChild->content[ctxt->nodelen], ch, len);
2489 ctxt->nodelen += len;
2490 lastChild->content[ctxt->nodelen] = 0;
2491 } else if (coalesceText) {
2492 if (xmlTextConcat(lastChild, ch, len)) {
2493 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters");
2494 }
2495 if (ctxt->node->children != NULL) {
2496 ctxt->nodelen = xmlStrlen(lastChild->content);
2497 ctxt->nodemem = ctxt->nodelen + 1;
2498 }
2499 } else {
2500 /* Mixed content, first time */
2501 lastChild = xmlSAX2TextNode(ctxt, ch, len);
2502 if (lastChild != NULL) {
2503 xmlAddChild(ctxt->node, lastChild);
2504 if (ctxt->node->children != NULL) {
2505 ctxt->nodelen = len;
2506 ctxt->nodemem = len + 1;
2507 }
2508 }
2509 }
2510 }
2511 }
2512
2513 /**
2514 * xmlSAX2IgnorableWhitespace:
2515 * @ctx: the user data (XML parser context)
2516 * @ch: a xmlChar string
2517 * @len: the number of xmlChar
2518 *
2519 * receiving some ignorable whitespaces from the parser.
2520 * UNUSED: by default the DOM building will use xmlSAX2Characters
2521 */
2522 void
xmlSAX2IgnorableWhitespace(void * ctx ATTRIBUTE_UNUSED,const xmlChar * ch ATTRIBUTE_UNUSED,int len ATTRIBUTE_UNUSED)2523 xmlSAX2IgnorableWhitespace(void *ctx ATTRIBUTE_UNUSED, const xmlChar *ch ATTRIBUTE_UNUSED, int len ATTRIBUTE_UNUSED)
2524 {
2525 /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */
2526 #ifdef DEBUG_SAX
2527 xmlGenericError(xmlGenericErrorContext,
2528 "SAX.xmlSAX2IgnorableWhitespace(%.30s, %d)\n", ch, len);
2529 #endif
2530 }
2531
2532 /**
2533 * xmlSAX2ProcessingInstruction:
2534 * @ctx: the user data (XML parser context)
2535 * @target: the target name
2536 * @data: the PI data's
2537 *
2538 * A processing instruction has been parsed.
2539 */
2540 void
xmlSAX2ProcessingInstruction(void * ctx,const xmlChar * target,const xmlChar * data)2541 xmlSAX2ProcessingInstruction(void *ctx, const xmlChar *target,
2542 const xmlChar *data)
2543 {
2544 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2545 xmlNodePtr ret;
2546 xmlNodePtr parent;
2547
2548 if (ctx == NULL) return;
2549 parent = ctxt->node;
2550 #ifdef DEBUG_SAX
2551 xmlGenericError(xmlGenericErrorContext,
2552 "SAX.xmlSAX2ProcessingInstruction(%s, %s)\n", target, data);
2553 #endif
2554
2555 ret = xmlNewDocPI(ctxt->myDoc, target, data);
2556 if (ret == NULL) return;
2557 parent = ctxt->node;
2558
2559 if (ctxt->linenumbers) {
2560 if (ctxt->input != NULL) {
2561 if (ctxt->input->line < 65535)
2562 ret->line = (short) ctxt->input->line;
2563 else
2564 ret->line = 65535;
2565 }
2566 }
2567 if (ctxt->inSubset == 1) {
2568 xmlAddChild((xmlNodePtr) ctxt->myDoc->intSubset, ret);
2569 return;
2570 } else if (ctxt->inSubset == 2) {
2571 xmlAddChild((xmlNodePtr) ctxt->myDoc->extSubset, ret);
2572 return;
2573 }
2574 if ((ctxt->myDoc->children == NULL) || (parent == NULL)) {
2575 #ifdef DEBUG_SAX_TREE
2576 xmlGenericError(xmlGenericErrorContext,
2577 "Setting PI %s as root\n", target);
2578 #endif
2579 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret);
2580 return;
2581 }
2582 if (parent->type == XML_ELEMENT_NODE) {
2583 #ifdef DEBUG_SAX_TREE
2584 xmlGenericError(xmlGenericErrorContext,
2585 "adding PI %s child to %s\n", target, parent->name);
2586 #endif
2587 xmlAddChild(parent, ret);
2588 } else {
2589 #ifdef DEBUG_SAX_TREE
2590 xmlGenericError(xmlGenericErrorContext,
2591 "adding PI %s sibling to ", target);
2592 xmlDebugDumpOneNode(stderr, parent, 0);
2593 #endif
2594 xmlAddSibling(parent, ret);
2595 }
2596 }
2597
2598 /**
2599 * xmlSAX2Comment:
2600 * @ctx: the user data (XML parser context)
2601 * @value: the xmlSAX2Comment content
2602 *
2603 * A xmlSAX2Comment has been parsed.
2604 */
2605 void
xmlSAX2Comment(void * ctx,const xmlChar * value)2606 xmlSAX2Comment(void *ctx, const xmlChar *value)
2607 {
2608 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2609 xmlNodePtr ret;
2610 xmlNodePtr parent;
2611
2612 if (ctx == NULL) return;
2613 parent = ctxt->node;
2614 #ifdef DEBUG_SAX
2615 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2Comment(%s)\n", value);
2616 #endif
2617 ret = xmlNewDocComment(ctxt->myDoc, value);
2618 if (ret == NULL) return;
2619 if (ctxt->linenumbers) {
2620 if (ctxt->input != NULL) {
2621 if (ctxt->input->line < 65535)
2622 ret->line = (short) ctxt->input->line;
2623 else
2624 ret->line = 65535;
2625 }
2626 }
2627
2628 if (ctxt->inSubset == 1) {
2629 xmlAddChild((xmlNodePtr) ctxt->myDoc->intSubset, ret);
2630 return;
2631 } else if (ctxt->inSubset == 2) {
2632 xmlAddChild((xmlNodePtr) ctxt->myDoc->extSubset, ret);
2633 return;
2634 }
2635 if ((ctxt->myDoc->children == NULL) || (parent == NULL)) {
2636 #ifdef DEBUG_SAX_TREE
2637 xmlGenericError(xmlGenericErrorContext,
2638 "Setting xmlSAX2Comment as root\n");
2639 #endif
2640 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret);
2641 return;
2642 }
2643 if (parent->type == XML_ELEMENT_NODE) {
2644 #ifdef DEBUG_SAX_TREE
2645 xmlGenericError(xmlGenericErrorContext,
2646 "adding xmlSAX2Comment child to %s\n", parent->name);
2647 #endif
2648 xmlAddChild(parent, ret);
2649 } else {
2650 #ifdef DEBUG_SAX_TREE
2651 xmlGenericError(xmlGenericErrorContext,
2652 "adding xmlSAX2Comment sibling to ");
2653 xmlDebugDumpOneNode(stderr, parent, 0);
2654 #endif
2655 xmlAddSibling(parent, ret);
2656 }
2657 }
2658
2659 /**
2660 * xmlSAX2CDataBlock:
2661 * @ctx: the user data (XML parser context)
2662 * @value: The pcdata content
2663 * @len: the block length
2664 *
2665 * called when a pcdata block has been parsed
2666 */
2667 void
xmlSAX2CDataBlock(void * ctx,const xmlChar * value,int len)2668 xmlSAX2CDataBlock(void *ctx, const xmlChar *value, int len)
2669 {
2670 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2671 xmlNodePtr ret, lastChild;
2672
2673 if (ctx == NULL) return;
2674 #ifdef DEBUG_SAX
2675 xmlGenericError(xmlGenericErrorContext,
2676 "SAX.pcdata(%.10s, %d)\n", value, len);
2677 #endif
2678 lastChild = xmlGetLastChild(ctxt->node);
2679 #ifdef DEBUG_SAX_TREE
2680 xmlGenericError(xmlGenericErrorContext,
2681 "add chars to %s \n", ctxt->node->name);
2682 #endif
2683 if ((lastChild != NULL) &&
2684 (lastChild->type == XML_CDATA_SECTION_NODE)) {
2685 xmlTextConcat(lastChild, value, len);
2686 } else {
2687 ret = xmlNewCDataBlock(ctxt->myDoc, value, len);
2688 xmlAddChild(ctxt->node, ret);
2689 }
2690 }
2691
2692 static int xmlSAX2DefaultVersionValue = 2;
2693
2694 #ifdef LIBXML_SAX1_ENABLED
2695 /**
2696 * xmlSAXDefaultVersion:
2697 * @version: the version, 1 or 2
2698 *
2699 * Set the default version of SAX used globally by the library.
2700 * By default, during initialization the default is set to 2.
2701 * Note that it is generally a better coding style to use
2702 * xmlSAXVersion() to set up the version explicitly for a given
2703 * parsing context.
2704 *
2705 * Returns the previous value in case of success and -1 in case of error.
2706 */
2707 int
xmlSAXDefaultVersion(int version)2708 xmlSAXDefaultVersion(int version)
2709 {
2710 int ret = xmlSAX2DefaultVersionValue;
2711
2712 if ((version != 1) && (version != 2))
2713 return(-1);
2714 xmlSAX2DefaultVersionValue = version;
2715 return(ret);
2716 }
2717 #endif /* LIBXML_SAX1_ENABLED */
2718
2719 /**
2720 * xmlSAXVersion:
2721 * @hdlr: the SAX handler
2722 * @version: the version, 1 or 2
2723 *
2724 * Initialize the default XML SAX handler according to the version
2725 *
2726 * Returns 0 in case of success and -1 in case of error.
2727 */
2728 int
xmlSAXVersion(xmlSAXHandler * hdlr,int version)2729 xmlSAXVersion(xmlSAXHandler *hdlr, int version)
2730 {
2731 if (hdlr == NULL) return(-1);
2732 if (version == 2) {
2733 hdlr->startElement = NULL;
2734 hdlr->endElement = NULL;
2735 hdlr->startElementNs = xmlSAX2StartElementNs;
2736 hdlr->endElementNs = xmlSAX2EndElementNs;
2737 hdlr->serror = NULL;
2738 hdlr->initialized = XML_SAX2_MAGIC;
2739 #ifdef LIBXML_SAX1_ENABLED
2740 } else if (version == 1) {
2741 hdlr->startElement = xmlSAX2StartElement;
2742 hdlr->endElement = xmlSAX2EndElement;
2743 hdlr->initialized = 1;
2744 #endif /* LIBXML_SAX1_ENABLED */
2745 } else
2746 return(-1);
2747 hdlr->internalSubset = xmlSAX2InternalSubset;
2748 hdlr->externalSubset = xmlSAX2ExternalSubset;
2749 hdlr->isStandalone = xmlSAX2IsStandalone;
2750 hdlr->hasInternalSubset = xmlSAX2HasInternalSubset;
2751 hdlr->hasExternalSubset = xmlSAX2HasExternalSubset;
2752 hdlr->resolveEntity = xmlSAX2ResolveEntity;
2753 hdlr->getEntity = xmlSAX2GetEntity;
2754 hdlr->getParameterEntity = xmlSAX2GetParameterEntity;
2755 hdlr->entityDecl = xmlSAX2EntityDecl;
2756 hdlr->attributeDecl = xmlSAX2AttributeDecl;
2757 hdlr->elementDecl = xmlSAX2ElementDecl;
2758 hdlr->notationDecl = xmlSAX2NotationDecl;
2759 hdlr->unparsedEntityDecl = xmlSAX2UnparsedEntityDecl;
2760 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator;
2761 hdlr->startDocument = xmlSAX2StartDocument;
2762 hdlr->endDocument = xmlSAX2EndDocument;
2763 hdlr->reference = xmlSAX2Reference;
2764 hdlr->characters = xmlSAX2Characters;
2765 hdlr->cdataBlock = xmlSAX2CDataBlock;
2766 hdlr->ignorableWhitespace = xmlSAX2Characters;
2767 hdlr->processingInstruction = xmlSAX2ProcessingInstruction;
2768 hdlr->comment = xmlSAX2Comment;
2769 hdlr->warning = xmlParserWarning;
2770 hdlr->error = xmlParserError;
2771 hdlr->fatalError = xmlParserError;
2772
2773 return(0);
2774 }
2775
2776 /**
2777 * xmlSAX2InitDefaultSAXHandler:
2778 * @hdlr: the SAX handler
2779 * @warning: flag if non-zero sets the handler warning procedure
2780 *
2781 * Initialize the default XML SAX2 handler
2782 */
2783 void
xmlSAX2InitDefaultSAXHandler(xmlSAXHandler * hdlr,int warning)2784 xmlSAX2InitDefaultSAXHandler(xmlSAXHandler *hdlr, int warning)
2785 {
2786 if ((hdlr == NULL) || (hdlr->initialized != 0))
2787 return;
2788
2789 xmlSAXVersion(hdlr, xmlSAX2DefaultVersionValue);
2790 if (warning == 0)
2791 hdlr->warning = NULL;
2792 else
2793 hdlr->warning = xmlParserWarning;
2794 }
2795
2796 /**
2797 * xmlDefaultSAXHandlerInit:
2798 *
2799 * Initialize the default SAX2 handler
2800 */
2801 void
xmlDefaultSAXHandlerInit(void)2802 xmlDefaultSAXHandlerInit(void)
2803 {
2804 #ifdef LIBXML_SAX1_ENABLED
2805 xmlSAXVersion((xmlSAXHandlerPtr) &xmlDefaultSAXHandler, 1);
2806 #endif /* LIBXML_SAX1_ENABLED */
2807 }
2808
2809 #ifdef LIBXML_HTML_ENABLED
2810
2811 /**
2812 * xmlSAX2InitHtmlDefaultSAXHandler:
2813 * @hdlr: the SAX handler
2814 *
2815 * Initialize the default HTML SAX2 handler
2816 */
2817 void
xmlSAX2InitHtmlDefaultSAXHandler(xmlSAXHandler * hdlr)2818 xmlSAX2InitHtmlDefaultSAXHandler(xmlSAXHandler *hdlr)
2819 {
2820 if ((hdlr == NULL) || (hdlr->initialized != 0))
2821 return;
2822
2823 hdlr->internalSubset = xmlSAX2InternalSubset;
2824 hdlr->externalSubset = NULL;
2825 hdlr->isStandalone = NULL;
2826 hdlr->hasInternalSubset = NULL;
2827 hdlr->hasExternalSubset = NULL;
2828 hdlr->resolveEntity = NULL;
2829 hdlr->getEntity = xmlSAX2GetEntity;
2830 hdlr->getParameterEntity = NULL;
2831 hdlr->entityDecl = NULL;
2832 hdlr->attributeDecl = NULL;
2833 hdlr->elementDecl = NULL;
2834 hdlr->notationDecl = NULL;
2835 hdlr->unparsedEntityDecl = NULL;
2836 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator;
2837 hdlr->startDocument = xmlSAX2StartDocument;
2838 hdlr->endDocument = xmlSAX2EndDocument;
2839 hdlr->startElement = xmlSAX2StartElement;
2840 hdlr->endElement = xmlSAX2EndElement;
2841 hdlr->reference = NULL;
2842 hdlr->characters = xmlSAX2Characters;
2843 hdlr->cdataBlock = xmlSAX2CDataBlock;
2844 hdlr->ignorableWhitespace = xmlSAX2IgnorableWhitespace;
2845 hdlr->processingInstruction = xmlSAX2ProcessingInstruction;
2846 hdlr->comment = xmlSAX2Comment;
2847 hdlr->warning = xmlParserWarning;
2848 hdlr->error = xmlParserError;
2849 hdlr->fatalError = xmlParserError;
2850
2851 hdlr->initialized = 1;
2852 }
2853
2854 /**
2855 * htmlDefaultSAXHandlerInit:
2856 *
2857 * Initialize the default SAX handler
2858 */
2859 void
htmlDefaultSAXHandlerInit(void)2860 htmlDefaultSAXHandlerInit(void)
2861 {
2862 xmlSAX2InitHtmlDefaultSAXHandler((xmlSAXHandlerPtr) &htmlDefaultSAXHandler);
2863 }
2864
2865 #endif /* LIBXML_HTML_ENABLED */
2866
2867 #ifdef LIBXML_DOCB_ENABLED
2868
2869 /**
2870 * xmlSAX2InitDocbDefaultSAXHandler:
2871 * @hdlr: the SAX handler
2872 *
2873 * Initialize the default DocBook SAX2 handler
2874 */
2875 void
xmlSAX2InitDocbDefaultSAXHandler(xmlSAXHandler * hdlr)2876 xmlSAX2InitDocbDefaultSAXHandler(xmlSAXHandler *hdlr)
2877 {
2878 if ((hdlr == NULL) || (hdlr->initialized != 0))
2879 return;
2880
2881 hdlr->internalSubset = xmlSAX2InternalSubset;
2882 hdlr->externalSubset = NULL;
2883 hdlr->isStandalone = xmlSAX2IsStandalone;
2884 hdlr->hasInternalSubset = xmlSAX2HasInternalSubset;
2885 hdlr->hasExternalSubset = xmlSAX2HasExternalSubset;
2886 hdlr->resolveEntity = xmlSAX2ResolveEntity;
2887 hdlr->getEntity = xmlSAX2GetEntity;
2888 hdlr->getParameterEntity = NULL;
2889 hdlr->entityDecl = xmlSAX2EntityDecl;
2890 hdlr->attributeDecl = NULL;
2891 hdlr->elementDecl = NULL;
2892 hdlr->notationDecl = NULL;
2893 hdlr->unparsedEntityDecl = NULL;
2894 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator;
2895 hdlr->startDocument = xmlSAX2StartDocument;
2896 hdlr->endDocument = xmlSAX2EndDocument;
2897 hdlr->startElement = xmlSAX2StartElement;
2898 hdlr->endElement = xmlSAX2EndElement;
2899 hdlr->reference = xmlSAX2Reference;
2900 hdlr->characters = xmlSAX2Characters;
2901 hdlr->cdataBlock = NULL;
2902 hdlr->ignorableWhitespace = xmlSAX2IgnorableWhitespace;
2903 hdlr->processingInstruction = NULL;
2904 hdlr->comment = xmlSAX2Comment;
2905 hdlr->warning = xmlParserWarning;
2906 hdlr->error = xmlParserError;
2907 hdlr->fatalError = xmlParserError;
2908
2909 hdlr->initialized = 1;
2910 }
2911
2912 /**
2913 * docbDefaultSAXHandlerInit:
2914 *
2915 * Initialize the default SAX handler
2916 */
2917 void
docbDefaultSAXHandlerInit(void)2918 docbDefaultSAXHandlerInit(void)
2919 {
2920 xmlSAX2InitDocbDefaultSAXHandler((xmlSAXHandlerPtr) &docbDefaultSAXHandler);
2921 }
2922
2923 #endif /* LIBXML_DOCB_ENABLED */
2924 #define bottom_SAX2
2925 #include "elfgcchack.h"
2926