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_HTML_ENABLED
1109 if ((ctxt->html) &&
1110 (value == NULL) && (htmlIsBooleanAttr(fullname))) {
1111 nval = xmlStrdup(fullname);
1112 value = (const xmlChar *) nval;
1113 } else
1114 #endif
1115 {
1116 #ifdef LIBXML_VALID_ENABLED
1117 /*
1118 * Do the last stage of the attribute normalization
1119 * Needed for HTML too:
1120 * http://www.w3.org/TR/html4/types.html#h-6.2
1121 */
1122 ctxt->vctxt.valid = 1;
1123 nval = xmlValidCtxtNormalizeAttributeValue(&ctxt->vctxt,
1124 ctxt->myDoc, ctxt->node,
1125 fullname, value);
1126 if (ctxt->vctxt.valid != 1) {
1127 ctxt->valid = 0;
1128 }
1129 if (nval != NULL)
1130 value = nval;
1131 #else
1132 nval = NULL;
1133 #endif /* LIBXML_VALID_ENABLED */
1134 }
1135
1136 /*
1137 * Check whether it's a namespace definition
1138 */
1139 if ((!ctxt->html) && (ns == NULL) &&
1140 (name[0] == 'x') && (name[1] == 'm') && (name[2] == 'l') &&
1141 (name[3] == 'n') && (name[4] == 's') && (name[5] == 0)) {
1142 xmlNsPtr nsret;
1143 xmlChar *val;
1144
1145 if (!ctxt->replaceEntities) {
1146 ctxt->depth++;
1147 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF,
1148 0,0,0);
1149 ctxt->depth--;
1150 } else {
1151 val = (xmlChar *) value;
1152 }
1153
1154 if (val[0] != 0) {
1155 xmlURIPtr uri;
1156
1157 uri = xmlParseURI((const char *)val);
1158 if (uri == NULL) {
1159 if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL))
1160 ctxt->sax->warning(ctxt->userData,
1161 "xmlns: %s not a valid URI\n", val);
1162 } else {
1163 if (uri->scheme == NULL) {
1164 if ((ctxt->sax != NULL) && (ctxt->sax->warning != NULL))
1165 ctxt->sax->warning(ctxt->userData,
1166 "xmlns: URI %s is not absolute\n", val);
1167 }
1168 xmlFreeURI(uri);
1169 }
1170 }
1171
1172 /* a default namespace definition */
1173 nsret = xmlNewNs(ctxt->node, val, NULL);
1174
1175 #ifdef LIBXML_VALID_ENABLED
1176 /*
1177 * Validate also for namespace decls, they are attributes from
1178 * an XML-1.0 perspective
1179 */
1180 if (nsret != NULL && ctxt->validate && ctxt->wellFormed &&
1181 ctxt->myDoc && ctxt->myDoc->intSubset)
1182 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc,
1183 ctxt->node, prefix, nsret, val);
1184 #endif /* LIBXML_VALID_ENABLED */
1185 if (name != NULL)
1186 xmlFree(name);
1187 if (nval != NULL)
1188 xmlFree(nval);
1189 if (val != value)
1190 xmlFree(val);
1191 return;
1192 }
1193 if ((!ctxt->html) &&
1194 (ns != NULL) && (ns[0] == 'x') && (ns[1] == 'm') && (ns[2] == 'l') &&
1195 (ns[3] == 'n') && (ns[4] == 's') && (ns[5] == 0)) {
1196 xmlNsPtr nsret;
1197 xmlChar *val;
1198
1199 if (!ctxt->replaceEntities) {
1200 ctxt->depth++;
1201 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF,
1202 0,0,0);
1203 ctxt->depth--;
1204 if (val == NULL) {
1205 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement");
1206 xmlFree(ns);
1207 if (name != NULL)
1208 xmlFree(name);
1209 return;
1210 }
1211 } else {
1212 val = (xmlChar *) value;
1213 }
1214
1215 if (val[0] == 0) {
1216 xmlNsErrMsg(ctxt, XML_NS_ERR_EMPTY,
1217 "Empty namespace name for prefix %s\n", name, NULL);
1218 }
1219 if ((ctxt->pedantic != 0) && (val[0] != 0)) {
1220 xmlURIPtr uri;
1221
1222 uri = xmlParseURI((const char *)val);
1223 if (uri == NULL) {
1224 xmlNsWarnMsg(ctxt, XML_WAR_NS_URI,
1225 "xmlns:%s: %s not a valid URI\n", name, value);
1226 } else {
1227 if (uri->scheme == NULL) {
1228 xmlNsWarnMsg(ctxt, XML_WAR_NS_URI_RELATIVE,
1229 "xmlns:%s: URI %s is not absolute\n", name, value);
1230 }
1231 xmlFreeURI(uri);
1232 }
1233 }
1234
1235 /* a standard namespace definition */
1236 nsret = xmlNewNs(ctxt->node, val, name);
1237 xmlFree(ns);
1238 #ifdef LIBXML_VALID_ENABLED
1239 /*
1240 * Validate also for namespace decls, they are attributes from
1241 * an XML-1.0 perspective
1242 */
1243 if (nsret != NULL && ctxt->validate && ctxt->wellFormed &&
1244 ctxt->myDoc && ctxt->myDoc->intSubset)
1245 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc,
1246 ctxt->node, prefix, nsret, value);
1247 #endif /* LIBXML_VALID_ENABLED */
1248 if (name != NULL)
1249 xmlFree(name);
1250 if (nval != NULL)
1251 xmlFree(nval);
1252 if (val != value)
1253 xmlFree(val);
1254 return;
1255 }
1256
1257 if (ns != NULL) {
1258 namespace = xmlSearchNs(ctxt->myDoc, ctxt->node, ns);
1259
1260 if (namespace == NULL) {
1261 xmlNsErrMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
1262 "Namespace prefix %s of attribute %s is not defined\n",
1263 ns, name);
1264 } else {
1265 xmlAttrPtr prop;
1266
1267 prop = ctxt->node->properties;
1268 while (prop != NULL) {
1269 if (prop->ns != NULL) {
1270 if ((xmlStrEqual(name, prop->name)) &&
1271 ((namespace == prop->ns) ||
1272 (xmlStrEqual(namespace->href, prop->ns->href)))) {
1273 xmlNsErrMsg(ctxt, XML_ERR_ATTRIBUTE_REDEFINED,
1274 "Attribute %s in %s redefined\n",
1275 name, namespace->href);
1276 ctxt->wellFormed = 0;
1277 if (ctxt->recovery == 0) ctxt->disableSAX = 1;
1278 goto error;
1279 }
1280 }
1281 prop = prop->next;
1282 }
1283 }
1284 } else {
1285 namespace = NULL;
1286 }
1287
1288 /* !!!!!! <a toto:arg="" xmlns:toto="http://toto.com"> */
1289 ret = xmlNewNsPropEatName(ctxt->node, namespace, name, NULL);
1290
1291 if (ret != NULL) {
1292 if ((ctxt->replaceEntities == 0) && (!ctxt->html)) {
1293 xmlNodePtr tmp;
1294
1295 ret->children = xmlStringGetNodeList(ctxt->myDoc, value);
1296 tmp = ret->children;
1297 while (tmp != NULL) {
1298 tmp->parent = (xmlNodePtr) ret;
1299 if (tmp->next == NULL)
1300 ret->last = tmp;
1301 tmp = tmp->next;
1302 }
1303 } else if (value != NULL) {
1304 ret->children = xmlNewDocText(ctxt->myDoc, value);
1305 ret->last = ret->children;
1306 if (ret->children != NULL)
1307 ret->children->parent = (xmlNodePtr) ret;
1308 }
1309 }
1310
1311 #ifdef LIBXML_VALID_ENABLED
1312 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed &&
1313 ctxt->myDoc && ctxt->myDoc->intSubset) {
1314
1315 /*
1316 * If we don't substitute entities, the validation should be
1317 * done on a value with replaced entities anyway.
1318 */
1319 if (!ctxt->replaceEntities) {
1320 xmlChar *val;
1321
1322 ctxt->depth++;
1323 val = xmlStringDecodeEntities(ctxt, value, XML_SUBSTITUTE_REF,
1324 0,0,0);
1325 ctxt->depth--;
1326
1327 if (val == NULL)
1328 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
1329 ctxt->myDoc, ctxt->node, ret, value);
1330 else {
1331 xmlChar *nvalnorm;
1332
1333 /*
1334 * Do the last stage of the attribute normalization
1335 * It need to be done twice ... it's an extra burden related
1336 * to the ability to keep xmlSAX2References in attributes
1337 */
1338 nvalnorm = xmlValidNormalizeAttributeValue(ctxt->myDoc,
1339 ctxt->node, fullname, val);
1340 if (nvalnorm != NULL) {
1341 xmlFree(val);
1342 val = nvalnorm;
1343 }
1344
1345 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
1346 ctxt->myDoc, ctxt->node, ret, val);
1347 xmlFree(val);
1348 }
1349 } else {
1350 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt, ctxt->myDoc,
1351 ctxt->node, ret, value);
1352 }
1353 } else
1354 #endif /* LIBXML_VALID_ENABLED */
1355 if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) &&
1356 (((ctxt->replaceEntities == 0) && (ctxt->external != 2)) ||
1357 ((ctxt->replaceEntities != 0) && (ctxt->inSubset == 0)))) {
1358 /*
1359 * when validating, the ID registration is done at the attribute
1360 * validation level. Otherwise we have to do specific handling here.
1361 */
1362 if (xmlStrEqual(fullname, BAD_CAST "xml:id")) {
1363 /*
1364 * Add the xml:id value
1365 *
1366 * Open issue: normalization of the value.
1367 */
1368 if (xmlValidateNCName(value, 1) != 0) {
1369 xmlErrValid(ctxt, XML_DTD_XMLID_VALUE,
1370 "xml:id : attribute value %s is not an NCName\n",
1371 (const char *) value, NULL);
1372 }
1373 xmlAddID(&ctxt->vctxt, ctxt->myDoc, value, ret);
1374 } else if (xmlIsID(ctxt->myDoc, ctxt->node, ret))
1375 xmlAddID(&ctxt->vctxt, ctxt->myDoc, value, ret);
1376 else if (xmlIsRef(ctxt->myDoc, ctxt->node, ret))
1377 xmlAddRef(&ctxt->vctxt, ctxt->myDoc, value, ret);
1378 }
1379
1380 error:
1381 if (nval != NULL)
1382 xmlFree(nval);
1383 if (ns != NULL)
1384 xmlFree(ns);
1385 }
1386
1387 /*
1388 * xmlCheckDefaultedAttributes:
1389 *
1390 * Check defaulted attributes from the DTD
1391 */
1392 static void
xmlCheckDefaultedAttributes(xmlParserCtxtPtr ctxt,const xmlChar * name,const xmlChar * prefix,const xmlChar ** atts)1393 xmlCheckDefaultedAttributes(xmlParserCtxtPtr ctxt, const xmlChar *name,
1394 const xmlChar *prefix, const xmlChar **atts) {
1395 xmlElementPtr elemDecl;
1396 const xmlChar *att;
1397 int internal = 1;
1398 int i;
1399
1400 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->intSubset, name, prefix);
1401 if (elemDecl == NULL) {
1402 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->extSubset, name, prefix);
1403 internal = 0;
1404 }
1405
1406 process_external_subset:
1407
1408 if (elemDecl != NULL) {
1409 xmlAttributePtr attr = elemDecl->attributes;
1410 /*
1411 * Check against defaulted attributes from the external subset
1412 * if the document is stamped as standalone
1413 */
1414 if ((ctxt->myDoc->standalone == 1) &&
1415 (ctxt->myDoc->extSubset != NULL) &&
1416 (ctxt->validate)) {
1417 while (attr != NULL) {
1418 if ((attr->defaultValue != NULL) &&
1419 (xmlGetDtdQAttrDesc(ctxt->myDoc->extSubset,
1420 attr->elem, attr->name,
1421 attr->prefix) == attr) &&
1422 (xmlGetDtdQAttrDesc(ctxt->myDoc->intSubset,
1423 attr->elem, attr->name,
1424 attr->prefix) == NULL)) {
1425 xmlChar *fulln;
1426
1427 if (attr->prefix != NULL) {
1428 fulln = xmlStrdup(attr->prefix);
1429 fulln = xmlStrcat(fulln, BAD_CAST ":");
1430 fulln = xmlStrcat(fulln, attr->name);
1431 } else {
1432 fulln = xmlStrdup(attr->name);
1433 }
1434 if (fulln == NULL) {
1435 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement");
1436 break;
1437 }
1438
1439 /*
1440 * Check that the attribute is not declared in the
1441 * serialization
1442 */
1443 att = NULL;
1444 if (atts != NULL) {
1445 i = 0;
1446 att = atts[i];
1447 while (att != NULL) {
1448 if (xmlStrEqual(att, fulln))
1449 break;
1450 i += 2;
1451 att = atts[i];
1452 }
1453 }
1454 if (att == NULL) {
1455 xmlErrValid(ctxt, XML_DTD_STANDALONE_DEFAULTED,
1456 "standalone: attribute %s on %s defaulted from external subset\n",
1457 (const char *)fulln,
1458 (const char *)attr->elem);
1459 }
1460 xmlFree(fulln);
1461 }
1462 attr = attr->nexth;
1463 }
1464 }
1465
1466 /*
1467 * Actually insert defaulted values when needed
1468 */
1469 attr = elemDecl->attributes;
1470 while (attr != NULL) {
1471 /*
1472 * Make sure that attributes redefinition occuring in the
1473 * internal subset are not overriden by definitions in the
1474 * external subset.
1475 */
1476 if (attr->defaultValue != NULL) {
1477 /*
1478 * the element should be instantiated in the tree if:
1479 * - this is a namespace prefix
1480 * - the user required for completion in the tree
1481 * like XSLT
1482 * - there isn't already an attribute definition
1483 * in the internal subset overriding it.
1484 */
1485 if (((attr->prefix != NULL) &&
1486 (xmlStrEqual(attr->prefix, BAD_CAST "xmlns"))) ||
1487 ((attr->prefix == NULL) &&
1488 (xmlStrEqual(attr->name, BAD_CAST "xmlns"))) ||
1489 (ctxt->loadsubset & XML_COMPLETE_ATTRS)) {
1490 xmlAttributePtr tst;
1491
1492 tst = xmlGetDtdQAttrDesc(ctxt->myDoc->intSubset,
1493 attr->elem, attr->name,
1494 attr->prefix);
1495 if ((tst == attr) || (tst == NULL)) {
1496 xmlChar fn[50];
1497 xmlChar *fulln;
1498
1499 fulln = xmlBuildQName(attr->name, attr->prefix, fn, 50);
1500 if (fulln == NULL) {
1501 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement");
1502 return;
1503 }
1504
1505 /*
1506 * Check that the attribute is not declared in the
1507 * serialization
1508 */
1509 att = NULL;
1510 if (atts != NULL) {
1511 i = 0;
1512 att = atts[i];
1513 while (att != NULL) {
1514 if (xmlStrEqual(att, fulln))
1515 break;
1516 i += 2;
1517 att = atts[i];
1518 }
1519 }
1520 if (att == NULL) {
1521 xmlSAX2AttributeInternal(ctxt, fulln,
1522 attr->defaultValue, prefix);
1523 }
1524 if ((fulln != fn) && (fulln != attr->name))
1525 xmlFree(fulln);
1526 }
1527 }
1528 }
1529 attr = attr->nexth;
1530 }
1531 if (internal == 1) {
1532 elemDecl = xmlGetDtdQElementDesc(ctxt->myDoc->extSubset,
1533 name, prefix);
1534 internal = 0;
1535 goto process_external_subset;
1536 }
1537 }
1538 }
1539
1540 /**
1541 * xmlSAX2StartElement:
1542 * @ctx: the user data (XML parser context)
1543 * @fullname: The element name, including namespace prefix
1544 * @atts: An array of name/value attributes pairs, NULL terminated
1545 *
1546 * called when an opening tag has been processed.
1547 */
1548 void
xmlSAX2StartElement(void * ctx,const xmlChar * fullname,const xmlChar ** atts)1549 xmlSAX2StartElement(void *ctx, const xmlChar *fullname, const xmlChar **atts)
1550 {
1551 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
1552 xmlNodePtr ret;
1553 xmlNodePtr parent;
1554 xmlNsPtr ns;
1555 xmlChar *name;
1556 xmlChar *prefix;
1557 const xmlChar *att;
1558 const xmlChar *value;
1559 int i;
1560
1561 if ((ctx == NULL) || (fullname == NULL) || (ctxt->myDoc == NULL)) return;
1562 parent = ctxt->node;
1563 #ifdef DEBUG_SAX
1564 xmlGenericError(xmlGenericErrorContext,
1565 "SAX.xmlSAX2StartElement(%s)\n", fullname);
1566 #endif
1567
1568 /*
1569 * First check on validity:
1570 */
1571 if (ctxt->validate && (ctxt->myDoc->extSubset == NULL) &&
1572 ((ctxt->myDoc->intSubset == NULL) ||
1573 ((ctxt->myDoc->intSubset->notations == NULL) &&
1574 (ctxt->myDoc->intSubset->elements == NULL) &&
1575 (ctxt->myDoc->intSubset->attributes == NULL) &&
1576 (ctxt->myDoc->intSubset->entities == NULL)))) {
1577 xmlErrValid(ctxt, XML_ERR_NO_DTD,
1578 "Validation failed: no DTD found !", NULL, NULL);
1579 ctxt->validate = 0;
1580 }
1581
1582
1583 /*
1584 * Split the full name into a namespace prefix and the tag name
1585 */
1586 name = xmlSplitQName(ctxt, fullname, &prefix);
1587
1588
1589 /*
1590 * Note : the namespace resolution is deferred until the end of the
1591 * attributes parsing, since local namespace can be defined as
1592 * an attribute at this level.
1593 */
1594 ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL, name, NULL);
1595 if (ret == NULL) {
1596 if (prefix != NULL)
1597 xmlFree(prefix);
1598 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElement");
1599 return;
1600 }
1601 if (ctxt->myDoc->children == NULL) {
1602 #ifdef DEBUG_SAX_TREE
1603 xmlGenericError(xmlGenericErrorContext, "Setting %s as root\n", name);
1604 #endif
1605 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret);
1606 } else if (parent == NULL) {
1607 parent = ctxt->myDoc->children;
1608 }
1609 ctxt->nodemem = -1;
1610 if (ctxt->linenumbers) {
1611 if (ctxt->input != NULL) {
1612 if (ctxt->input->line < 65535)
1613 ret->line = (short) ctxt->input->line;
1614 else
1615 ret->line = 65535;
1616 }
1617 }
1618
1619 /*
1620 * We are parsing a new node.
1621 */
1622 #ifdef DEBUG_SAX_TREE
1623 xmlGenericError(xmlGenericErrorContext, "pushing(%s)\n", name);
1624 #endif
1625 nodePush(ctxt, ret);
1626
1627 /*
1628 * Link the child element
1629 */
1630 if (parent != NULL) {
1631 if (parent->type == XML_ELEMENT_NODE) {
1632 #ifdef DEBUG_SAX_TREE
1633 xmlGenericError(xmlGenericErrorContext,
1634 "adding child %s to %s\n", name, parent->name);
1635 #endif
1636 xmlAddChild(parent, ret);
1637 } else {
1638 #ifdef DEBUG_SAX_TREE
1639 xmlGenericError(xmlGenericErrorContext,
1640 "adding sibling %s to ", name);
1641 xmlDebugDumpOneNode(stderr, parent, 0);
1642 #endif
1643 xmlAddSibling(parent, ret);
1644 }
1645 }
1646
1647 /*
1648 * Insert all the defaulted attributes from the DTD especially namespaces
1649 */
1650 if ((!ctxt->html) &&
1651 ((ctxt->myDoc->intSubset != NULL) ||
1652 (ctxt->myDoc->extSubset != NULL))) {
1653 xmlCheckDefaultedAttributes(ctxt, name, prefix, atts);
1654 }
1655
1656 /*
1657 * process all the attributes whose name start with "xmlns"
1658 */
1659 if (atts != NULL) {
1660 i = 0;
1661 att = atts[i++];
1662 value = atts[i++];
1663 if (!ctxt->html) {
1664 while ((att != NULL) && (value != NULL)) {
1665 if ((att[0] == 'x') && (att[1] == 'm') && (att[2] == 'l') &&
1666 (att[3] == 'n') && (att[4] == 's'))
1667 xmlSAX2AttributeInternal(ctxt, att, value, prefix);
1668
1669 att = atts[i++];
1670 value = atts[i++];
1671 }
1672 }
1673 }
1674
1675 /*
1676 * Search the namespace, note that since the attributes have been
1677 * processed, the local namespaces are available.
1678 */
1679 ns = xmlSearchNs(ctxt->myDoc, ret, prefix);
1680 if ((ns == NULL) && (parent != NULL))
1681 ns = xmlSearchNs(ctxt->myDoc, parent, prefix);
1682 if ((prefix != NULL) && (ns == NULL)) {
1683 ns = xmlNewNs(ret, NULL, prefix);
1684 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
1685 "Namespace prefix %s is not defined\n",
1686 prefix, NULL);
1687 }
1688
1689 /*
1690 * set the namespace node, making sure that if the default namspace
1691 * is unbound on a parent we simply kee it NULL
1692 */
1693 if ((ns != NULL) && (ns->href != NULL) &&
1694 ((ns->href[0] != 0) || (ns->prefix != NULL)))
1695 xmlSetNs(ret, ns);
1696
1697 /*
1698 * process all the other attributes
1699 */
1700 if (atts != NULL) {
1701 i = 0;
1702 att = atts[i++];
1703 value = atts[i++];
1704 if (ctxt->html) {
1705 while (att != NULL) {
1706 xmlSAX2AttributeInternal(ctxt, att, value, NULL);
1707 att = atts[i++];
1708 value = atts[i++];
1709 }
1710 } else {
1711 while ((att != NULL) && (value != NULL)) {
1712 if ((att[0] != 'x') || (att[1] != 'm') || (att[2] != 'l') ||
1713 (att[3] != 'n') || (att[4] != 's'))
1714 xmlSAX2AttributeInternal(ctxt, att, value, NULL);
1715
1716 /*
1717 * Next ones
1718 */
1719 att = atts[i++];
1720 value = atts[i++];
1721 }
1722 }
1723 }
1724
1725 #ifdef LIBXML_VALID_ENABLED
1726 /*
1727 * If it's the Document root, finish the DTD validation and
1728 * check the document root element for validity
1729 */
1730 if ((ctxt->validate) && (ctxt->vctxt.finishDtd == XML_CTXT_FINISH_DTD_0)) {
1731 int chk;
1732
1733 chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc);
1734 if (chk <= 0)
1735 ctxt->valid = 0;
1736 if (chk < 0)
1737 ctxt->wellFormed = 0;
1738 ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
1739 ctxt->vctxt.finishDtd = XML_CTXT_FINISH_DTD_1;
1740 }
1741 #endif /* LIBXML_VALID_ENABLED */
1742
1743 if (prefix != NULL)
1744 xmlFree(prefix);
1745
1746 }
1747
1748 /**
1749 * xmlSAX2EndElement:
1750 * @ctx: the user data (XML parser context)
1751 * @name: The element name
1752 *
1753 * called when the end of an element has been detected.
1754 */
1755 void
xmlSAX2EndElement(void * ctx,const xmlChar * name ATTRIBUTE_UNUSED)1756 xmlSAX2EndElement(void *ctx, const xmlChar *name ATTRIBUTE_UNUSED)
1757 {
1758 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
1759 xmlParserNodeInfo node_info;
1760 xmlNodePtr cur;
1761
1762 if (ctx == NULL) return;
1763 cur = ctxt->node;
1764 #ifdef DEBUG_SAX
1765 if (name == NULL)
1766 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2EndElement(NULL)\n");
1767 else
1768 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2EndElement(%s)\n", name);
1769 #endif
1770
1771 /* Capture end position and add node */
1772 if (cur != NULL && ctxt->record_info) {
1773 node_info.end_pos = ctxt->input->cur - ctxt->input->base;
1774 node_info.end_line = ctxt->input->line;
1775 node_info.node = cur;
1776 xmlParserAddNodeInfo(ctxt, &node_info);
1777 }
1778 ctxt->nodemem = -1;
1779
1780 #ifdef LIBXML_VALID_ENABLED
1781 if (ctxt->validate && ctxt->wellFormed &&
1782 ctxt->myDoc && ctxt->myDoc->intSubset)
1783 ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc,
1784 cur);
1785 #endif /* LIBXML_VALID_ENABLED */
1786
1787
1788 /*
1789 * end of parsing of this node.
1790 */
1791 #ifdef DEBUG_SAX_TREE
1792 xmlGenericError(xmlGenericErrorContext, "popping(%s)\n", cur->name);
1793 #endif
1794 nodePop(ctxt);
1795 }
1796 #endif /* LIBXML_SAX1_ENABLED || LIBXML_HTML_ENABLE */
1797
1798 /*
1799 * xmlSAX2TextNode:
1800 * @ctxt: the parser context
1801 * @str: the input string
1802 * @len: the string length
1803 *
1804 * Remove the entities from an attribute value
1805 *
1806 * Returns the newly allocated string or NULL if not needed or error
1807 */
1808 static xmlNodePtr
xmlSAX2TextNode(xmlParserCtxtPtr ctxt,const xmlChar * str,int len)1809 xmlSAX2TextNode(xmlParserCtxtPtr ctxt, const xmlChar *str, int len) {
1810 xmlNodePtr ret;
1811 const xmlChar *intern = NULL;
1812
1813 /*
1814 * Allocate
1815 */
1816 if (ctxt->freeElems != NULL) {
1817 ret = ctxt->freeElems;
1818 ctxt->freeElems = ret->next;
1819 ctxt->freeElemsNr--;
1820 } else {
1821 ret = (xmlNodePtr) xmlMalloc(sizeof(xmlNode));
1822 }
1823 if (ret == NULL) {
1824 xmlErrMemory(ctxt, "xmlSAX2Characters");
1825 return(NULL);
1826 }
1827 memset(ret, 0, sizeof(xmlNode));
1828 /*
1829 * intern the formatting blanks found between tags, or the
1830 * very short strings
1831 */
1832 if (ctxt->dictNames) {
1833 xmlChar cur = str[len];
1834
1835 if ((len < (int) (2 * sizeof(void *))) &&
1836 (ctxt->options & XML_PARSE_COMPACT)) {
1837 /* store the string in the node overrithing properties and nsDef */
1838 xmlChar *tmp = (xmlChar *) &(ret->properties);
1839 memcpy(tmp, str, len);
1840 tmp[len] = 0;
1841 intern = tmp;
1842 } else if ((len <= 3) && ((cur == '"') || (cur == '\'') ||
1843 ((cur == '<') && (str[len + 1] != '!')))) {
1844 intern = xmlDictLookup(ctxt->dict, str, len);
1845 } else if (IS_BLANK_CH(*str) && (len < 60) && (cur == '<') &&
1846 (str[len + 1] != '!')) {
1847 int i;
1848
1849 for (i = 1;i < len;i++) {
1850 if (!IS_BLANK_CH(str[i])) goto skip;
1851 }
1852 intern = xmlDictLookup(ctxt->dict, str, len);
1853 }
1854 }
1855 skip:
1856 ret->type = XML_TEXT_NODE;
1857
1858 ret->name = xmlStringText;
1859 if (intern == NULL) {
1860 ret->content = xmlStrndup(str, len);
1861 if (ret->content == NULL) {
1862 xmlSAX2ErrMemory(ctxt, "xmlSAX2TextNode");
1863 xmlFree(ret);
1864 return(NULL);
1865 }
1866 } else
1867 ret->content = (xmlChar *) intern;
1868
1869 if (ctxt->input != NULL)
1870 ret->line = ctxt->input->line;
1871
1872 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue))
1873 xmlRegisterNodeDefaultValue(ret);
1874 return(ret);
1875 }
1876
1877 #ifdef LIBXML_VALID_ENABLED
1878 /*
1879 * xmlSAX2DecodeAttrEntities:
1880 * @ctxt: the parser context
1881 * @str: the input string
1882 * @len: the string length
1883 *
1884 * Remove the entities from an attribute value
1885 *
1886 * Returns the newly allocated string or NULL if not needed or error
1887 */
1888 static xmlChar *
xmlSAX2DecodeAttrEntities(xmlParserCtxtPtr ctxt,const xmlChar * str,const xmlChar * end)1889 xmlSAX2DecodeAttrEntities(xmlParserCtxtPtr ctxt, const xmlChar *str,
1890 const xmlChar *end) {
1891 const xmlChar *in;
1892 xmlChar *ret;
1893
1894 in = str;
1895 while (in < end)
1896 if (*in++ == '&')
1897 goto decode;
1898 return(NULL);
1899 decode:
1900 ctxt->depth++;
1901 ret = xmlStringLenDecodeEntities(ctxt, str, end - str,
1902 XML_SUBSTITUTE_REF, 0,0,0);
1903 ctxt->depth--;
1904 return(ret);
1905 }
1906 #endif /* LIBXML_VALID_ENABLED */
1907
1908 /**
1909 * xmlSAX2AttributeNs:
1910 * @ctx: the user data (XML parser context)
1911 * @localname: the local name of the attribute
1912 * @prefix: the attribute namespace prefix if available
1913 * @URI: the attribute namespace name if available
1914 * @value: Start of the attribute value
1915 * @valueend: end of the attribute value
1916 *
1917 * Handle an attribute that has been read by the parser.
1918 * The default handling is to convert the attribute into an
1919 * DOM subtree and past it in a new xmlAttr element added to
1920 * the element.
1921 */
1922 static void
xmlSAX2AttributeNs(xmlParserCtxtPtr ctxt,const xmlChar * localname,const xmlChar * prefix,const xmlChar * value,const xmlChar * valueend)1923 xmlSAX2AttributeNs(xmlParserCtxtPtr ctxt,
1924 const xmlChar * localname,
1925 const xmlChar * prefix,
1926 const xmlChar * value,
1927 const xmlChar * valueend)
1928 {
1929 xmlAttrPtr ret;
1930 xmlNsPtr namespace = NULL;
1931 xmlChar *dup = NULL;
1932
1933 /*
1934 * Note: if prefix == NULL, the attribute is not in the default namespace
1935 */
1936 if (prefix != NULL)
1937 namespace = xmlSearchNs(ctxt->myDoc, ctxt->node, prefix);
1938
1939 /*
1940 * allocate the node
1941 */
1942 if (ctxt->freeAttrs != NULL) {
1943 ret = ctxt->freeAttrs;
1944 ctxt->freeAttrs = ret->next;
1945 ctxt->freeAttrsNr--;
1946 memset(ret, 0, sizeof(xmlAttr));
1947 ret->type = XML_ATTRIBUTE_NODE;
1948
1949 ret->parent = ctxt->node;
1950 ret->doc = ctxt->myDoc;
1951 ret->ns = namespace;
1952
1953 if (ctxt->dictNames)
1954 ret->name = localname;
1955 else
1956 ret->name = xmlStrdup(localname);
1957
1958 /* link at the end to preserv order, TODO speed up with a last */
1959 if (ctxt->node->properties == NULL) {
1960 ctxt->node->properties = ret;
1961 } else {
1962 xmlAttrPtr prev = ctxt->node->properties;
1963
1964 while (prev->next != NULL) prev = prev->next;
1965 prev->next = ret;
1966 ret->prev = prev;
1967 }
1968
1969 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue))
1970 xmlRegisterNodeDefaultValue((xmlNodePtr)ret);
1971 } else {
1972 if (ctxt->dictNames)
1973 ret = xmlNewNsPropEatName(ctxt->node, namespace,
1974 (xmlChar *) localname, NULL);
1975 else
1976 ret = xmlNewNsProp(ctxt->node, namespace, localname, NULL);
1977 if (ret == NULL) {
1978 xmlErrMemory(ctxt, "xmlSAX2AttributeNs");
1979 return;
1980 }
1981 }
1982
1983 if ((ctxt->replaceEntities == 0) && (!ctxt->html)) {
1984 xmlNodePtr tmp;
1985
1986 /*
1987 * We know that if there is an entity reference, then
1988 * the string has been dup'ed and terminates with 0
1989 * otherwise with ' or "
1990 */
1991 if (*valueend != 0) {
1992 tmp = xmlSAX2TextNode(ctxt, value, valueend - value);
1993 ret->children = tmp;
1994 ret->last = tmp;
1995 if (tmp != NULL) {
1996 tmp->doc = ret->doc;
1997 tmp->parent = (xmlNodePtr) ret;
1998 }
1999 } else {
2000 ret->children = xmlStringLenGetNodeList(ctxt->myDoc, value,
2001 valueend - value);
2002 tmp = ret->children;
2003 while (tmp != NULL) {
2004 tmp->doc = ret->doc;
2005 tmp->parent = (xmlNodePtr) ret;
2006 if (tmp->next == NULL)
2007 ret->last = tmp;
2008 tmp = tmp->next;
2009 }
2010 }
2011 } else if (value != NULL) {
2012 xmlNodePtr tmp;
2013
2014 tmp = xmlSAX2TextNode(ctxt, value, valueend - value);
2015 ret->children = tmp;
2016 ret->last = tmp;
2017 if (tmp != NULL) {
2018 tmp->doc = ret->doc;
2019 tmp->parent = (xmlNodePtr) ret;
2020 }
2021 }
2022
2023 #ifdef LIBXML_VALID_ENABLED
2024 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed &&
2025 ctxt->myDoc && ctxt->myDoc->intSubset) {
2026 /*
2027 * If we don't substitute entities, the validation should be
2028 * done on a value with replaced entities anyway.
2029 */
2030 if (!ctxt->replaceEntities) {
2031 dup = xmlSAX2DecodeAttrEntities(ctxt, value, valueend);
2032 if (dup == NULL) {
2033 if (*valueend == 0) {
2034 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
2035 ctxt->myDoc, ctxt->node, ret, value);
2036 } else {
2037 /*
2038 * That should already be normalized.
2039 * cheaper to finally allocate here than duplicate
2040 * entry points in the full validation code
2041 */
2042 dup = xmlStrndup(value, valueend - value);
2043
2044 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
2045 ctxt->myDoc, ctxt->node, ret, dup);
2046 }
2047 } else {
2048 /*
2049 * dup now contains a string of the flattened attribute
2050 * content with entities substitued. Check if we need to
2051 * apply an extra layer of normalization.
2052 * It need to be done twice ... it's an extra burden related
2053 * to the ability to keep references in attributes
2054 */
2055 if (ctxt->attsSpecial != NULL) {
2056 xmlChar *nvalnorm;
2057 xmlChar fn[50];
2058 xmlChar *fullname;
2059
2060 fullname = xmlBuildQName(localname, prefix, fn, 50);
2061 if (fullname != NULL) {
2062 ctxt->vctxt.valid = 1;
2063 nvalnorm = xmlValidCtxtNormalizeAttributeValue(
2064 &ctxt->vctxt, ctxt->myDoc,
2065 ctxt->node, fullname, dup);
2066 if (ctxt->vctxt.valid != 1)
2067 ctxt->valid = 0;
2068
2069 if ((fullname != fn) && (fullname != localname))
2070 xmlFree(fullname);
2071 if (nvalnorm != NULL) {
2072 xmlFree(dup);
2073 dup = nvalnorm;
2074 }
2075 }
2076 }
2077
2078 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
2079 ctxt->myDoc, ctxt->node, ret, dup);
2080 }
2081 } else {
2082 /*
2083 * if entities already have been substitued, then
2084 * the attribute as passed is already normalized
2085 */
2086 dup = xmlStrndup(value, valueend - value);
2087
2088 ctxt->valid &= xmlValidateOneAttribute(&ctxt->vctxt,
2089 ctxt->myDoc, ctxt->node, ret, dup);
2090 }
2091 } else
2092 #endif /* LIBXML_VALID_ENABLED */
2093 if (((ctxt->loadsubset & XML_SKIP_IDS) == 0) &&
2094 (((ctxt->replaceEntities == 0) && (ctxt->external != 2)) ||
2095 ((ctxt->replaceEntities != 0) && (ctxt->inSubset == 0)))) {
2096 /*
2097 * when validating, the ID registration is done at the attribute
2098 * validation level. Otherwise we have to do specific handling here.
2099 */
2100 if ((prefix == ctxt->str_xml) &&
2101 (localname[0] == 'i') && (localname[1] == 'd') &&
2102 (localname[2] == 0)) {
2103 /*
2104 * Add the xml:id value
2105 *
2106 * Open issue: normalization of the value.
2107 */
2108 if (dup == NULL)
2109 dup = xmlStrndup(value, valueend - value);
2110 #ifdef LIBXML_VALID_ENABLED
2111 if (xmlValidateNCName(dup, 1) != 0) {
2112 xmlErrValid(ctxt, XML_DTD_XMLID_VALUE,
2113 "xml:id : attribute value %s is not an NCName\n",
2114 (const char *) dup, NULL);
2115 }
2116 #endif
2117 xmlAddID(&ctxt->vctxt, ctxt->myDoc, dup, ret);
2118 } else if (xmlIsID(ctxt->myDoc, ctxt->node, ret)) {
2119 /* might be worth duplicate entry points and not copy */
2120 if (dup == NULL)
2121 dup = xmlStrndup(value, valueend - value);
2122 xmlAddID(&ctxt->vctxt, ctxt->myDoc, dup, ret);
2123 } else if (xmlIsRef(ctxt->myDoc, ctxt->node, ret)) {
2124 if (dup == NULL)
2125 dup = xmlStrndup(value, valueend - value);
2126 xmlAddRef(&ctxt->vctxt, ctxt->myDoc, dup, ret);
2127 }
2128 }
2129 if (dup != NULL)
2130 xmlFree(dup);
2131 }
2132
2133 /**
2134 * xmlSAX2StartElementNs:
2135 * @ctx: the user data (XML parser context)
2136 * @localname: the local name of the element
2137 * @prefix: the element namespace prefix if available
2138 * @URI: the element namespace name if available
2139 * @nb_namespaces: number of namespace definitions on that node
2140 * @namespaces: pointer to the array of prefix/URI pairs namespace definitions
2141 * @nb_attributes: the number of attributes on that node
2142 * @nb_defaulted: the number of defaulted attributes.
2143 * @attributes: pointer to the array of (localname/prefix/URI/value/end)
2144 * attribute values.
2145 *
2146 * SAX2 callback when an element start has been detected by the parser.
2147 * It provides the namespace informations for the element, as well as
2148 * the new namespace declarations on the element.
2149 */
2150 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)2151 xmlSAX2StartElementNs(void *ctx,
2152 const xmlChar *localname,
2153 const xmlChar *prefix,
2154 const xmlChar *URI,
2155 int nb_namespaces,
2156 const xmlChar **namespaces,
2157 int nb_attributes,
2158 int nb_defaulted,
2159 const xmlChar **attributes)
2160 {
2161 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2162 xmlNodePtr ret;
2163 xmlNodePtr parent;
2164 xmlNsPtr last = NULL, ns;
2165 const xmlChar *uri, *pref;
2166 int i, j;
2167
2168 if (ctx == NULL) return;
2169 parent = ctxt->node;
2170 /*
2171 * First check on validity:
2172 */
2173 if (ctxt->validate && (ctxt->myDoc->extSubset == NULL) &&
2174 ((ctxt->myDoc->intSubset == NULL) ||
2175 ((ctxt->myDoc->intSubset->notations == NULL) &&
2176 (ctxt->myDoc->intSubset->elements == NULL) &&
2177 (ctxt->myDoc->intSubset->attributes == NULL) &&
2178 (ctxt->myDoc->intSubset->entities == NULL)))) {
2179 xmlErrValid(ctxt, XML_ERR_NO_DTD,
2180 "Validation failed: no DTD found !", NULL, NULL);
2181 ctxt->validate = 0;
2182 }
2183
2184 /*
2185 * allocate the node
2186 */
2187 if (ctxt->freeElems != NULL) {
2188 ret = ctxt->freeElems;
2189 ctxt->freeElems = ret->next;
2190 ctxt->freeElemsNr--;
2191 memset(ret, 0, sizeof(xmlNode));
2192 ret->type = XML_ELEMENT_NODE;
2193
2194 if (ctxt->dictNames)
2195 ret->name = localname;
2196 else {
2197 ret->name = xmlStrdup(localname);
2198 if (ret->name == NULL) {
2199 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs");
2200 return;
2201 }
2202 }
2203 if ((__xmlRegisterCallbacks) && (xmlRegisterNodeDefaultValue))
2204 xmlRegisterNodeDefaultValue(ret);
2205 } else {
2206 if (ctxt->dictNames)
2207 ret = xmlNewDocNodeEatName(ctxt->myDoc, NULL,
2208 (xmlChar *) localname, NULL);
2209 else
2210 ret = xmlNewDocNode(ctxt->myDoc, NULL, localname, NULL);
2211 if (ret == NULL) {
2212 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs");
2213 return;
2214 }
2215 }
2216 if (ctxt->linenumbers) {
2217 if (ctxt->input != NULL) {
2218 if (ctxt->input->line < 65535)
2219 ret->line = (short) ctxt->input->line;
2220 else
2221 ret->line = 65535;
2222 }
2223 }
2224
2225 if ((ctxt->myDoc->children == NULL) || (parent == NULL)) {
2226 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret);
2227 }
2228 /*
2229 * Build the namespace list
2230 */
2231 for (i = 0,j = 0;j < nb_namespaces;j++) {
2232 pref = namespaces[i++];
2233 uri = namespaces[i++];
2234 ns = xmlNewNs(NULL, uri, pref);
2235 if (ns != NULL) {
2236 if (last == NULL) {
2237 ret->nsDef = last = ns;
2238 } else {
2239 last->next = ns;
2240 last = ns;
2241 }
2242 if ((URI != NULL) && (prefix == pref))
2243 ret->ns = ns;
2244 } else {
2245 /*
2246 * any out of memory error would already have been raised
2247 * but we can't be garanteed it's the actual error due to the
2248 * API, best is to skip in this case
2249 */
2250 continue;
2251 }
2252 #ifdef LIBXML_VALID_ENABLED
2253 if ((!ctxt->html) && ctxt->validate && ctxt->wellFormed &&
2254 ctxt->myDoc && ctxt->myDoc->intSubset) {
2255 ctxt->valid &= xmlValidateOneNamespace(&ctxt->vctxt, ctxt->myDoc,
2256 ret, prefix, ns, uri);
2257 }
2258 #endif /* LIBXML_VALID_ENABLED */
2259 }
2260 ctxt->nodemem = -1;
2261
2262 /*
2263 * We are parsing a new node.
2264 */
2265 nodePush(ctxt, ret);
2266
2267 /*
2268 * Link the child element
2269 */
2270 if (parent != NULL) {
2271 if (parent->type == XML_ELEMENT_NODE) {
2272 xmlAddChild(parent, ret);
2273 } else {
2274 xmlAddSibling(parent, ret);
2275 }
2276 }
2277
2278 /*
2279 * Insert the defaulted attributes from the DTD only if requested:
2280 */
2281 if ((nb_defaulted != 0) &&
2282 ((ctxt->loadsubset & XML_COMPLETE_ATTRS) == 0))
2283 nb_attributes -= nb_defaulted;
2284
2285 /*
2286 * Search the namespace if it wasn't already found
2287 * Note that, if prefix is NULL, this searches for the default Ns
2288 */
2289 if ((URI != NULL) && (ret->ns == NULL)) {
2290 ret->ns = xmlSearchNs(ctxt->myDoc, parent, prefix);
2291 if ((ret->ns == NULL) && (xmlStrEqual(prefix, BAD_CAST "xml"))) {
2292 ret->ns = xmlSearchNs(ctxt->myDoc, ret, prefix);
2293 }
2294 if (ret->ns == NULL) {
2295 ns = xmlNewNs(ret, NULL, prefix);
2296 if (ns == NULL) {
2297
2298 xmlSAX2ErrMemory(ctxt, "xmlSAX2StartElementNs");
2299 return;
2300 }
2301 if (prefix != NULL)
2302 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
2303 "Namespace prefix %s was not found\n",
2304 prefix, NULL);
2305 else
2306 xmlNsWarnMsg(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE,
2307 "Namespace default prefix was not found\n",
2308 NULL, NULL);
2309 }
2310 }
2311
2312 /*
2313 * process all the other attributes
2314 */
2315 if (nb_attributes > 0) {
2316 for (j = 0,i = 0;i < nb_attributes;i++,j+=5) {
2317 xmlSAX2AttributeNs(ctxt, attributes[j], attributes[j+1],
2318 attributes[j+3], attributes[j+4]);
2319 }
2320 }
2321
2322 #ifdef LIBXML_VALID_ENABLED
2323 /*
2324 * If it's the Document root, finish the DTD validation and
2325 * check the document root element for validity
2326 */
2327 if ((ctxt->validate) && (ctxt->vctxt.finishDtd == XML_CTXT_FINISH_DTD_0)) {
2328 int chk;
2329
2330 chk = xmlValidateDtdFinal(&ctxt->vctxt, ctxt->myDoc);
2331 if (chk <= 0)
2332 ctxt->valid = 0;
2333 if (chk < 0)
2334 ctxt->wellFormed = 0;
2335 ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc);
2336 ctxt->vctxt.finishDtd = XML_CTXT_FINISH_DTD_1;
2337 }
2338 #endif /* LIBXML_VALID_ENABLED */
2339 }
2340
2341 /**
2342 * xmlSAX2EndElementNs:
2343 * @ctx: the user data (XML parser context)
2344 * @localname: the local name of the element
2345 * @prefix: the element namespace prefix if available
2346 * @URI: the element namespace name if available
2347 *
2348 * SAX2 callback when an element end has been detected by the parser.
2349 * It provides the namespace informations for the element.
2350 */
2351 void
xmlSAX2EndElementNs(void * ctx,const xmlChar * localname ATTRIBUTE_UNUSED,const xmlChar * prefix ATTRIBUTE_UNUSED,const xmlChar * URI ATTRIBUTE_UNUSED)2352 xmlSAX2EndElementNs(void *ctx,
2353 const xmlChar * localname ATTRIBUTE_UNUSED,
2354 const xmlChar * prefix ATTRIBUTE_UNUSED,
2355 const xmlChar * URI ATTRIBUTE_UNUSED)
2356 {
2357 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2358 xmlParserNodeInfo node_info;
2359 xmlNodePtr cur;
2360
2361 if (ctx == NULL) return;
2362 cur = ctxt->node;
2363 /* Capture end position and add node */
2364 if ((ctxt->record_info) && (cur != NULL)) {
2365 node_info.end_pos = ctxt->input->cur - ctxt->input->base;
2366 node_info.end_line = ctxt->input->line;
2367 node_info.node = cur;
2368 xmlParserAddNodeInfo(ctxt, &node_info);
2369 }
2370 ctxt->nodemem = -1;
2371
2372 #ifdef LIBXML_VALID_ENABLED
2373 if (ctxt->validate && ctxt->wellFormed &&
2374 ctxt->myDoc && ctxt->myDoc->intSubset)
2375 ctxt->valid &= xmlValidateOneElement(&ctxt->vctxt, ctxt->myDoc, cur);
2376 #endif /* LIBXML_VALID_ENABLED */
2377
2378 /*
2379 * end of parsing of this node.
2380 */
2381 nodePop(ctxt);
2382 }
2383
2384 /**
2385 * xmlSAX2Reference:
2386 * @ctx: the user data (XML parser context)
2387 * @name: The entity name
2388 *
2389 * called when an entity xmlSAX2Reference is detected.
2390 */
2391 void
xmlSAX2Reference(void * ctx,const xmlChar * name)2392 xmlSAX2Reference(void *ctx, const xmlChar *name)
2393 {
2394 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2395 xmlNodePtr ret;
2396
2397 if (ctx == NULL) return;
2398 #ifdef DEBUG_SAX
2399 xmlGenericError(xmlGenericErrorContext,
2400 "SAX.xmlSAX2Reference(%s)\n", name);
2401 #endif
2402 if (name[0] == '#')
2403 ret = xmlNewCharRef(ctxt->myDoc, name);
2404 else
2405 ret = xmlNewReference(ctxt->myDoc, name);
2406 #ifdef DEBUG_SAX_TREE
2407 xmlGenericError(xmlGenericErrorContext,
2408 "add xmlSAX2Reference %s to %s \n", name, ctxt->node->name);
2409 #endif
2410 if (xmlAddChild(ctxt->node, ret) == NULL) {
2411 xmlFreeNode(ret);
2412 }
2413 }
2414
2415 /**
2416 * xmlSAX2Characters:
2417 * @ctx: the user data (XML parser context)
2418 * @ch: a xmlChar string
2419 * @len: the number of xmlChar
2420 *
2421 * receiving some chars from the parser.
2422 */
2423 void
xmlSAX2Characters(void * ctx,const xmlChar * ch,int len)2424 xmlSAX2Characters(void *ctx, const xmlChar *ch, int len)
2425 {
2426 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2427 xmlNodePtr lastChild;
2428
2429 if (ctx == NULL) return;
2430 #ifdef DEBUG_SAX
2431 xmlGenericError(xmlGenericErrorContext,
2432 "SAX.xmlSAX2Characters(%.30s, %d)\n", ch, len);
2433 #endif
2434 /*
2435 * Handle the data if any. If there is no child
2436 * add it as content, otherwise if the last child is text,
2437 * concatenate it, else create a new node of type text.
2438 */
2439
2440 if (ctxt->node == NULL) {
2441 #ifdef DEBUG_SAX_TREE
2442 xmlGenericError(xmlGenericErrorContext,
2443 "add chars: ctxt->node == NULL !\n");
2444 #endif
2445 return;
2446 }
2447 lastChild = ctxt->node->last;
2448 #ifdef DEBUG_SAX_TREE
2449 xmlGenericError(xmlGenericErrorContext,
2450 "add chars to %s \n", ctxt->node->name);
2451 #endif
2452
2453 /*
2454 * Here we needed an accelerator mechanism in case of very large
2455 * elements. Use an attribute in the structure !!!
2456 */
2457 if (lastChild == NULL) {
2458 lastChild = xmlSAX2TextNode(ctxt, ch, len);
2459 if (lastChild != NULL) {
2460 ctxt->node->children = lastChild;
2461 ctxt->node->last = lastChild;
2462 lastChild->parent = ctxt->node;
2463 lastChild->doc = ctxt->node->doc;
2464 ctxt->nodelen = len;
2465 ctxt->nodemem = len + 1;
2466 } else {
2467 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters");
2468 return;
2469 }
2470 } else {
2471 int coalesceText = (lastChild != NULL) &&
2472 (lastChild->type == XML_TEXT_NODE) &&
2473 (lastChild->name == xmlStringText);
2474 if ((coalesceText) && (ctxt->nodemem != 0)) {
2475 /*
2476 * The whole point of maintaining nodelen and nodemem,
2477 * xmlTextConcat is too costly, i.e. compute length,
2478 * reallocate a new buffer, move data, append ch. Here
2479 * We try to minimaze realloc() uses and avoid copying
2480 * and recomputing length over and over.
2481 */
2482 if (lastChild->content == (xmlChar *)&(lastChild->properties)) {
2483 lastChild->content = xmlStrdup(lastChild->content);
2484 lastChild->properties = NULL;
2485 } else if ((ctxt->nodemem == ctxt->nodelen + 1) &&
2486 (xmlDictOwns(ctxt->dict, lastChild->content))) {
2487 lastChild->content = xmlStrdup(lastChild->content);
2488 }
2489 if (((size_t)ctxt->nodelen + (size_t)len > XML_MAX_TEXT_LENGTH) &&
2490 ((ctxt->options & XML_PARSE_HUGE) == 0)) {
2491 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters: huge text node");
2492 return;
2493 }
2494 if ((size_t)ctxt->nodelen > SIZE_T_MAX - (size_t)len ||
2495 (size_t)ctxt->nodemem + (size_t)len > SIZE_T_MAX / 2) {
2496 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters overflow prevented");
2497 return;
2498 }
2499 if (ctxt->nodelen + len >= ctxt->nodemem) {
2500 xmlChar *newbuf;
2501 size_t size;
2502
2503 size = ctxt->nodemem + len;
2504 size *= 2;
2505 newbuf = (xmlChar *) xmlRealloc(lastChild->content,size);
2506 if (newbuf == NULL) {
2507 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters");
2508 return;
2509 }
2510 ctxt->nodemem = size;
2511 lastChild->content = newbuf;
2512 }
2513 memcpy(&lastChild->content[ctxt->nodelen], ch, len);
2514 ctxt->nodelen += len;
2515 lastChild->content[ctxt->nodelen] = 0;
2516 } else if (coalesceText) {
2517 if (xmlTextConcat(lastChild, ch, len)) {
2518 xmlSAX2ErrMemory(ctxt, "xmlSAX2Characters");
2519 }
2520 if (ctxt->node->children != NULL) {
2521 ctxt->nodelen = xmlStrlen(lastChild->content);
2522 ctxt->nodemem = ctxt->nodelen + 1;
2523 }
2524 } else {
2525 /* Mixed content, first time */
2526 lastChild = xmlSAX2TextNode(ctxt, ch, len);
2527 if (lastChild != NULL) {
2528 xmlAddChild(ctxt->node, lastChild);
2529 if (ctxt->node->children != NULL) {
2530 ctxt->nodelen = len;
2531 ctxt->nodemem = len + 1;
2532 }
2533 }
2534 }
2535 }
2536 }
2537
2538 /**
2539 * xmlSAX2IgnorableWhitespace:
2540 * @ctx: the user data (XML parser context)
2541 * @ch: a xmlChar string
2542 * @len: the number of xmlChar
2543 *
2544 * receiving some ignorable whitespaces from the parser.
2545 * UNUSED: by default the DOM building will use xmlSAX2Characters
2546 */
2547 void
xmlSAX2IgnorableWhitespace(void * ctx ATTRIBUTE_UNUSED,const xmlChar * ch ATTRIBUTE_UNUSED,int len ATTRIBUTE_UNUSED)2548 xmlSAX2IgnorableWhitespace(void *ctx ATTRIBUTE_UNUSED, const xmlChar *ch ATTRIBUTE_UNUSED, int len ATTRIBUTE_UNUSED)
2549 {
2550 /* xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx; */
2551 #ifdef DEBUG_SAX
2552 xmlGenericError(xmlGenericErrorContext,
2553 "SAX.xmlSAX2IgnorableWhitespace(%.30s, %d)\n", ch, len);
2554 #endif
2555 }
2556
2557 /**
2558 * xmlSAX2ProcessingInstruction:
2559 * @ctx: the user data (XML parser context)
2560 * @target: the target name
2561 * @data: the PI data's
2562 *
2563 * A processing instruction has been parsed.
2564 */
2565 void
xmlSAX2ProcessingInstruction(void * ctx,const xmlChar * target,const xmlChar * data)2566 xmlSAX2ProcessingInstruction(void *ctx, const xmlChar *target,
2567 const xmlChar *data)
2568 {
2569 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2570 xmlNodePtr ret;
2571 xmlNodePtr parent;
2572
2573 if (ctx == NULL) return;
2574 parent = ctxt->node;
2575 #ifdef DEBUG_SAX
2576 xmlGenericError(xmlGenericErrorContext,
2577 "SAX.xmlSAX2ProcessingInstruction(%s, %s)\n", target, data);
2578 #endif
2579
2580 ret = xmlNewDocPI(ctxt->myDoc, target, data);
2581 if (ret == NULL) return;
2582
2583 if (ctxt->linenumbers) {
2584 if (ctxt->input != NULL) {
2585 if (ctxt->input->line < 65535)
2586 ret->line = (short) ctxt->input->line;
2587 else
2588 ret->line = 65535;
2589 }
2590 }
2591 if (ctxt->inSubset == 1) {
2592 xmlAddChild((xmlNodePtr) ctxt->myDoc->intSubset, ret);
2593 return;
2594 } else if (ctxt->inSubset == 2) {
2595 xmlAddChild((xmlNodePtr) ctxt->myDoc->extSubset, ret);
2596 return;
2597 }
2598 if ((ctxt->myDoc->children == NULL) || (parent == NULL)) {
2599 #ifdef DEBUG_SAX_TREE
2600 xmlGenericError(xmlGenericErrorContext,
2601 "Setting PI %s as root\n", target);
2602 #endif
2603 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret);
2604 return;
2605 }
2606 if (parent->type == XML_ELEMENT_NODE) {
2607 #ifdef DEBUG_SAX_TREE
2608 xmlGenericError(xmlGenericErrorContext,
2609 "adding PI %s child to %s\n", target, parent->name);
2610 #endif
2611 xmlAddChild(parent, ret);
2612 } else {
2613 #ifdef DEBUG_SAX_TREE
2614 xmlGenericError(xmlGenericErrorContext,
2615 "adding PI %s sibling to ", target);
2616 xmlDebugDumpOneNode(stderr, parent, 0);
2617 #endif
2618 xmlAddSibling(parent, ret);
2619 }
2620 }
2621
2622 /**
2623 * xmlSAX2Comment:
2624 * @ctx: the user data (XML parser context)
2625 * @value: the xmlSAX2Comment content
2626 *
2627 * A xmlSAX2Comment has been parsed.
2628 */
2629 void
xmlSAX2Comment(void * ctx,const xmlChar * value)2630 xmlSAX2Comment(void *ctx, const xmlChar *value)
2631 {
2632 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2633 xmlNodePtr ret;
2634 xmlNodePtr parent;
2635
2636 if (ctx == NULL) return;
2637 parent = ctxt->node;
2638 #ifdef DEBUG_SAX
2639 xmlGenericError(xmlGenericErrorContext, "SAX.xmlSAX2Comment(%s)\n", value);
2640 #endif
2641 ret = xmlNewDocComment(ctxt->myDoc, value);
2642 if (ret == NULL) return;
2643 if (ctxt->linenumbers) {
2644 if (ctxt->input != NULL) {
2645 if (ctxt->input->line < 65535)
2646 ret->line = (short) ctxt->input->line;
2647 else
2648 ret->line = 65535;
2649 }
2650 }
2651
2652 if (ctxt->inSubset == 1) {
2653 xmlAddChild((xmlNodePtr) ctxt->myDoc->intSubset, ret);
2654 return;
2655 } else if (ctxt->inSubset == 2) {
2656 xmlAddChild((xmlNodePtr) ctxt->myDoc->extSubset, ret);
2657 return;
2658 }
2659 if ((ctxt->myDoc->children == NULL) || (parent == NULL)) {
2660 #ifdef DEBUG_SAX_TREE
2661 xmlGenericError(xmlGenericErrorContext,
2662 "Setting xmlSAX2Comment as root\n");
2663 #endif
2664 xmlAddChild((xmlNodePtr) ctxt->myDoc, (xmlNodePtr) ret);
2665 return;
2666 }
2667 if (parent->type == XML_ELEMENT_NODE) {
2668 #ifdef DEBUG_SAX_TREE
2669 xmlGenericError(xmlGenericErrorContext,
2670 "adding xmlSAX2Comment child to %s\n", parent->name);
2671 #endif
2672 xmlAddChild(parent, ret);
2673 } else {
2674 #ifdef DEBUG_SAX_TREE
2675 xmlGenericError(xmlGenericErrorContext,
2676 "adding xmlSAX2Comment sibling to ");
2677 xmlDebugDumpOneNode(stderr, parent, 0);
2678 #endif
2679 xmlAddSibling(parent, ret);
2680 }
2681 }
2682
2683 /**
2684 * xmlSAX2CDataBlock:
2685 * @ctx: the user data (XML parser context)
2686 * @value: The pcdata content
2687 * @len: the block length
2688 *
2689 * called when a pcdata block has been parsed
2690 */
2691 void
xmlSAX2CDataBlock(void * ctx,const xmlChar * value,int len)2692 xmlSAX2CDataBlock(void *ctx, const xmlChar *value, int len)
2693 {
2694 xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) ctx;
2695 xmlNodePtr ret, lastChild;
2696
2697 if (ctx == NULL) return;
2698 #ifdef DEBUG_SAX
2699 xmlGenericError(xmlGenericErrorContext,
2700 "SAX.pcdata(%.10s, %d)\n", value, len);
2701 #endif
2702 lastChild = xmlGetLastChild(ctxt->node);
2703 #ifdef DEBUG_SAX_TREE
2704 xmlGenericError(xmlGenericErrorContext,
2705 "add chars to %s \n", ctxt->node->name);
2706 #endif
2707 if ((lastChild != NULL) &&
2708 (lastChild->type == XML_CDATA_SECTION_NODE)) {
2709 xmlTextConcat(lastChild, value, len);
2710 } else {
2711 ret = xmlNewCDataBlock(ctxt->myDoc, value, len);
2712 xmlAddChild(ctxt->node, ret);
2713 }
2714 }
2715
2716 static int xmlSAX2DefaultVersionValue = 2;
2717
2718 #ifdef LIBXML_SAX1_ENABLED
2719 /**
2720 * xmlSAXDefaultVersion:
2721 * @version: the version, 1 or 2
2722 *
2723 * Set the default version of SAX used globally by the library.
2724 * By default, during initialization the default is set to 2.
2725 * Note that it is generally a better coding style to use
2726 * xmlSAXVersion() to set up the version explicitly for a given
2727 * parsing context.
2728 *
2729 * Returns the previous value in case of success and -1 in case of error.
2730 */
2731 int
xmlSAXDefaultVersion(int version)2732 xmlSAXDefaultVersion(int version)
2733 {
2734 int ret = xmlSAX2DefaultVersionValue;
2735
2736 if ((version != 1) && (version != 2))
2737 return(-1);
2738 xmlSAX2DefaultVersionValue = version;
2739 return(ret);
2740 }
2741 #endif /* LIBXML_SAX1_ENABLED */
2742
2743 /**
2744 * xmlSAXVersion:
2745 * @hdlr: the SAX handler
2746 * @version: the version, 1 or 2
2747 *
2748 * Initialize the default XML SAX handler according to the version
2749 *
2750 * Returns 0 in case of success and -1 in case of error.
2751 */
2752 int
xmlSAXVersion(xmlSAXHandler * hdlr,int version)2753 xmlSAXVersion(xmlSAXHandler *hdlr, int version)
2754 {
2755 if (hdlr == NULL) return(-1);
2756 if (version == 2) {
2757 hdlr->startElement = NULL;
2758 hdlr->endElement = NULL;
2759 hdlr->startElementNs = xmlSAX2StartElementNs;
2760 hdlr->endElementNs = xmlSAX2EndElementNs;
2761 hdlr->serror = NULL;
2762 hdlr->initialized = XML_SAX2_MAGIC;
2763 #ifdef LIBXML_SAX1_ENABLED
2764 } else if (version == 1) {
2765 hdlr->startElement = xmlSAX2StartElement;
2766 hdlr->endElement = xmlSAX2EndElement;
2767 hdlr->initialized = 1;
2768 #endif /* LIBXML_SAX1_ENABLED */
2769 } else
2770 return(-1);
2771 hdlr->internalSubset = xmlSAX2InternalSubset;
2772 hdlr->externalSubset = xmlSAX2ExternalSubset;
2773 hdlr->isStandalone = xmlSAX2IsStandalone;
2774 hdlr->hasInternalSubset = xmlSAX2HasInternalSubset;
2775 hdlr->hasExternalSubset = xmlSAX2HasExternalSubset;
2776 hdlr->resolveEntity = xmlSAX2ResolveEntity;
2777 hdlr->getEntity = xmlSAX2GetEntity;
2778 hdlr->getParameterEntity = xmlSAX2GetParameterEntity;
2779 hdlr->entityDecl = xmlSAX2EntityDecl;
2780 hdlr->attributeDecl = xmlSAX2AttributeDecl;
2781 hdlr->elementDecl = xmlSAX2ElementDecl;
2782 hdlr->notationDecl = xmlSAX2NotationDecl;
2783 hdlr->unparsedEntityDecl = xmlSAX2UnparsedEntityDecl;
2784 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator;
2785 hdlr->startDocument = xmlSAX2StartDocument;
2786 hdlr->endDocument = xmlSAX2EndDocument;
2787 hdlr->reference = xmlSAX2Reference;
2788 hdlr->characters = xmlSAX2Characters;
2789 hdlr->cdataBlock = xmlSAX2CDataBlock;
2790 hdlr->ignorableWhitespace = xmlSAX2Characters;
2791 hdlr->processingInstruction = xmlSAX2ProcessingInstruction;
2792 hdlr->comment = xmlSAX2Comment;
2793 hdlr->warning = xmlParserWarning;
2794 hdlr->error = xmlParserError;
2795 hdlr->fatalError = xmlParserError;
2796
2797 return(0);
2798 }
2799
2800 /**
2801 * xmlSAX2InitDefaultSAXHandler:
2802 * @hdlr: the SAX handler
2803 * @warning: flag if non-zero sets the handler warning procedure
2804 *
2805 * Initialize the default XML SAX2 handler
2806 */
2807 void
xmlSAX2InitDefaultSAXHandler(xmlSAXHandler * hdlr,int warning)2808 xmlSAX2InitDefaultSAXHandler(xmlSAXHandler *hdlr, int warning)
2809 {
2810 if ((hdlr == NULL) || (hdlr->initialized != 0))
2811 return;
2812
2813 xmlSAXVersion(hdlr, xmlSAX2DefaultVersionValue);
2814 if (warning == 0)
2815 hdlr->warning = NULL;
2816 else
2817 hdlr->warning = xmlParserWarning;
2818 }
2819
2820 /**
2821 * xmlDefaultSAXHandlerInit:
2822 *
2823 * Initialize the default SAX2 handler
2824 */
2825 void
xmlDefaultSAXHandlerInit(void)2826 xmlDefaultSAXHandlerInit(void)
2827 {
2828 #ifdef LIBXML_SAX1_ENABLED
2829 xmlSAXVersion((xmlSAXHandlerPtr) &xmlDefaultSAXHandler, 1);
2830 #endif /* LIBXML_SAX1_ENABLED */
2831 }
2832
2833 #ifdef LIBXML_HTML_ENABLED
2834
2835 /**
2836 * xmlSAX2InitHtmlDefaultSAXHandler:
2837 * @hdlr: the SAX handler
2838 *
2839 * Initialize the default HTML SAX2 handler
2840 */
2841 void
xmlSAX2InitHtmlDefaultSAXHandler(xmlSAXHandler * hdlr)2842 xmlSAX2InitHtmlDefaultSAXHandler(xmlSAXHandler *hdlr)
2843 {
2844 if ((hdlr == NULL) || (hdlr->initialized != 0))
2845 return;
2846
2847 hdlr->internalSubset = xmlSAX2InternalSubset;
2848 hdlr->externalSubset = NULL;
2849 hdlr->isStandalone = NULL;
2850 hdlr->hasInternalSubset = NULL;
2851 hdlr->hasExternalSubset = NULL;
2852 hdlr->resolveEntity = NULL;
2853 hdlr->getEntity = xmlSAX2GetEntity;
2854 hdlr->getParameterEntity = NULL;
2855 hdlr->entityDecl = NULL;
2856 hdlr->attributeDecl = NULL;
2857 hdlr->elementDecl = NULL;
2858 hdlr->notationDecl = NULL;
2859 hdlr->unparsedEntityDecl = NULL;
2860 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator;
2861 hdlr->startDocument = xmlSAX2StartDocument;
2862 hdlr->endDocument = xmlSAX2EndDocument;
2863 hdlr->startElement = xmlSAX2StartElement;
2864 hdlr->endElement = xmlSAX2EndElement;
2865 hdlr->reference = NULL;
2866 hdlr->characters = xmlSAX2Characters;
2867 hdlr->cdataBlock = xmlSAX2CDataBlock;
2868 hdlr->ignorableWhitespace = xmlSAX2IgnorableWhitespace;
2869 hdlr->processingInstruction = xmlSAX2ProcessingInstruction;
2870 hdlr->comment = xmlSAX2Comment;
2871 hdlr->warning = xmlParserWarning;
2872 hdlr->error = xmlParserError;
2873 hdlr->fatalError = xmlParserError;
2874
2875 hdlr->initialized = 1;
2876 }
2877
2878 /**
2879 * htmlDefaultSAXHandlerInit:
2880 *
2881 * Initialize the default SAX handler
2882 */
2883 void
htmlDefaultSAXHandlerInit(void)2884 htmlDefaultSAXHandlerInit(void)
2885 {
2886 xmlSAX2InitHtmlDefaultSAXHandler((xmlSAXHandlerPtr) &htmlDefaultSAXHandler);
2887 }
2888
2889 #endif /* LIBXML_HTML_ENABLED */
2890
2891 #ifdef LIBXML_DOCB_ENABLED
2892
2893 /**
2894 * xmlSAX2InitDocbDefaultSAXHandler:
2895 * @hdlr: the SAX handler
2896 *
2897 * Initialize the default DocBook SAX2 handler
2898 */
2899 void
xmlSAX2InitDocbDefaultSAXHandler(xmlSAXHandler * hdlr)2900 xmlSAX2InitDocbDefaultSAXHandler(xmlSAXHandler *hdlr)
2901 {
2902 if ((hdlr == NULL) || (hdlr->initialized != 0))
2903 return;
2904
2905 hdlr->internalSubset = xmlSAX2InternalSubset;
2906 hdlr->externalSubset = NULL;
2907 hdlr->isStandalone = xmlSAX2IsStandalone;
2908 hdlr->hasInternalSubset = xmlSAX2HasInternalSubset;
2909 hdlr->hasExternalSubset = xmlSAX2HasExternalSubset;
2910 hdlr->resolveEntity = xmlSAX2ResolveEntity;
2911 hdlr->getEntity = xmlSAX2GetEntity;
2912 hdlr->getParameterEntity = NULL;
2913 hdlr->entityDecl = xmlSAX2EntityDecl;
2914 hdlr->attributeDecl = NULL;
2915 hdlr->elementDecl = NULL;
2916 hdlr->notationDecl = NULL;
2917 hdlr->unparsedEntityDecl = NULL;
2918 hdlr->setDocumentLocator = xmlSAX2SetDocumentLocator;
2919 hdlr->startDocument = xmlSAX2StartDocument;
2920 hdlr->endDocument = xmlSAX2EndDocument;
2921 hdlr->startElement = xmlSAX2StartElement;
2922 hdlr->endElement = xmlSAX2EndElement;
2923 hdlr->reference = xmlSAX2Reference;
2924 hdlr->characters = xmlSAX2Characters;
2925 hdlr->cdataBlock = NULL;
2926 hdlr->ignorableWhitespace = xmlSAX2IgnorableWhitespace;
2927 hdlr->processingInstruction = NULL;
2928 hdlr->comment = xmlSAX2Comment;
2929 hdlr->warning = xmlParserWarning;
2930 hdlr->error = xmlParserError;
2931 hdlr->fatalError = xmlParserError;
2932
2933 hdlr->initialized = 1;
2934 }
2935
2936 /**
2937 * docbDefaultSAXHandlerInit:
2938 *
2939 * Initialize the default SAX handler
2940 */
2941 void
docbDefaultSAXHandlerInit(void)2942 docbDefaultSAXHandlerInit(void)
2943 {
2944 xmlSAX2InitDocbDefaultSAXHandler((xmlSAXHandlerPtr) &docbDefaultSAXHandler);
2945 }
2946
2947 #endif /* LIBXML_DOCB_ENABLED */
2948 #define bottom_SAX2
2949 #include "elfgcchack.h"
2950