xref: /PHP-7.2/ext/xml/compat.c (revision 7a7ec01a)
1 /*
2    +----------------------------------------------------------------------+
3    | PHP Version 7                                                        |
4    +----------------------------------------------------------------------+
5    | Copyright (c) 1997-2018 The PHP Group                                |
6    +----------------------------------------------------------------------+
7    | This source file is subject to version 3.01 of the PHP license,      |
8    | that is bundled with this package in the file LICENSE, and is        |
9    | available through the world-wide-web at the following url:           |
10    | http://www.php.net/license/3_01.txt                                  |
11    | If you did not receive a copy of the PHP license and are unable to   |
12    | obtain it through the world-wide-web, please send a note to          |
13    | license@php.net so we can mail you a copy immediately.               |
14    +----------------------------------------------------------------------+
15    | Authors: Sterling Hughes <sterling@php.net>                          |
16    +----------------------------------------------------------------------+
17  */
18 
19 #include "php.h"
20 #if defined(HAVE_LIBXML) && (defined(HAVE_XML) || defined(HAVE_XMLRPC)) && !defined(HAVE_LIBEXPAT)
21 #include "expat_compat.h"
22 
23 typedef struct _php_xml_ns {
24 	xmlNsPtr nsptr;
25 	int ref_count;
26 	void *next;
27 	void *prev;
28 } php_xml_ns;
29 
30 #ifdef LIBXML_EXPAT_COMPAT
31 
32 #define IS_NS_DECL(__ns) \
33 	((__ns) != NULL && strlen(__ns) == 5 && *(__ns) == 'x' && *((__ns)+1) == 'm' && \
34 	 *((__ns)+2) == 'l' && *((__ns)+3) == 'n' && *((__ns)+4) == 's')
35 
36 static void
_qualify_namespace(XML_Parser parser,const xmlChar * name,const xmlChar * URI,xmlChar ** qualified)37 _qualify_namespace(XML_Parser parser, const xmlChar *name, const xmlChar *URI, xmlChar **qualified)
38 {
39 	if (URI) {
40 			/* Use libxml functions otherwise its memory deallocation is screwed up */
41 			*qualified = xmlStrdup(URI);
42 			*qualified = xmlStrncat(*qualified, parser->_ns_separator, 1);
43 			*qualified = xmlStrncat(*qualified, name, xmlStrlen(name));
44 	} else {
45 		*qualified = xmlStrdup(name);
46 	}
47 }
48 
49 static void
_start_element_handler(void * user,const xmlChar * name,const xmlChar ** attributes)50 _start_element_handler(void *user, const xmlChar *name, const xmlChar **attributes)
51 {
52 	XML_Parser  parser = (XML_Parser) user;
53 	xmlChar    *qualified_name = NULL;
54 
55 	if (parser->h_start_element == NULL) {
56 		if (parser->h_default) {
57 			int attno = 0;
58 
59 			qualified_name = xmlStrncatNew((xmlChar *)"<", name, xmlStrlen(name));
60 			if (attributes) {
61 				while (attributes[attno] != NULL) {
62 					int att_len;
63 					char *att_string, *att_name, *att_value;
64 
65 					att_name = (char *)attributes[attno++];
66 					att_value = (char *)attributes[attno++];
67 
68 					att_len = spprintf(&att_string, 0, " %s=\"%s\"", att_name, att_value);
69 
70 					qualified_name = xmlStrncat(qualified_name, (xmlChar *)att_string, att_len);
71 					efree(att_string);
72 				}
73 
74 			}
75 			qualified_name = xmlStrncat(qualified_name, (xmlChar *)">", 1);
76 			parser->h_default(parser->user, (const XML_Char *) qualified_name, xmlStrlen(qualified_name));
77 			xmlFree(qualified_name);
78 		}
79 		return;
80 	}
81 
82 	qualified_name = xmlStrdup(name);
83 
84 	parser->h_start_element(parser->user, (const XML_Char *) qualified_name, (const XML_Char **) attributes);
85 
86 	xmlFree(qualified_name);
87 }
88 
89 static void
_start_element_handler_ns(void * user,const xmlChar * name,const xmlChar * prefix,const xmlChar * URI,int nb_namespaces,const xmlChar ** namespaces,int nb_attributes,int nb_defaulted,const xmlChar ** attributes)90 _start_element_handler_ns(void *user, const xmlChar *name, const xmlChar *prefix, const xmlChar *URI, int nb_namespaces, const xmlChar ** namespaces, int nb_attributes, int nb_defaulted, const xmlChar ** attributes)
91 {
92 	XML_Parser  parser = (XML_Parser) user;
93 	xmlChar    *qualified_name = NULL;
94 	xmlChar **attrs = NULL;
95 	int i;
96 	int z = 0;
97 	int y = 0;
98 
99 	if (nb_namespaces > 0 && parser->h_start_ns != NULL) {
100 		for (i = 0; i < nb_namespaces; i += 1) {
101 			parser->h_start_ns(parser->user, (const XML_Char *) namespaces[y], (const XML_Char *) namespaces[y+1]);
102 			y += 2;
103 		}
104 		y = 0;
105 	}
106 
107 	if (parser->h_start_element == NULL) {
108 	 	if (parser->h_default) {
109 
110 			if (prefix) {
111 				qualified_name = xmlStrncatNew((xmlChar *)"<", prefix, xmlStrlen(prefix));
112 				qualified_name = xmlStrncat(qualified_name, (xmlChar *)":", 1);
113 				qualified_name = xmlStrncat(qualified_name, name, xmlStrlen(name));
114 			} else {
115 				qualified_name = xmlStrncatNew((xmlChar *)"<", name, xmlStrlen(name));
116 			}
117 
118 			if (namespaces) {
119 				int i, j;
120 				for (i = 0,j = 0;j < nb_namespaces;j++) {
121 					int ns_len;
122 					char *ns_string, *ns_prefix, *ns_url;
123 
124 					ns_prefix = (char *) namespaces[i++];
125 					ns_url = (char *) namespaces[i++];
126 
127 					if (ns_prefix) {
128 						ns_len = spprintf(&ns_string, 0, " xmlns:%s=\"%s\"", ns_prefix, ns_url);
129 					} else {
130 						ns_len = spprintf(&ns_string, 0, " xmlns=\"%s\"", ns_url);
131 					}
132 					qualified_name = xmlStrncat(qualified_name, (xmlChar *)ns_string, ns_len);
133 
134 					efree(ns_string);
135 				}
136 			}
137 
138 			if (attributes) {
139 				for (i = 0; i < nb_attributes; i += 1) {
140 					int att_len;
141 					char *att_string, *att_name, *att_value, *att_prefix, *att_valueend;
142 
143 					att_name = (char *) attributes[y++];
144 					att_prefix = (char *)attributes[y++];
145 					y++;
146 					att_value = (char *)attributes[y++];
147 					att_valueend = (char *)attributes[y++];
148 
149 					if (att_prefix) {
150 						att_len = spprintf(&att_string, 0, " %s:%s=\"", att_prefix, att_name);
151 					} else {
152 						att_len = spprintf(&att_string, 0, " %s=\"", att_name);
153 					}
154 
155 					qualified_name = xmlStrncat(qualified_name, (xmlChar *)att_string, att_len);
156 					qualified_name = xmlStrncat(qualified_name, (xmlChar *)att_value, att_valueend - att_value);
157 					qualified_name = xmlStrncat(qualified_name, (xmlChar *)"\"", 1);
158 
159 					efree(att_string);
160 				}
161 
162 			}
163 			qualified_name = xmlStrncat(qualified_name, (xmlChar *)">", 1);
164 			parser->h_default(parser->user, (const XML_Char *) qualified_name, xmlStrlen(qualified_name));
165 			xmlFree(qualified_name);
166 		}
167 		return;
168 	}
169 	_qualify_namespace(parser, name, URI, &qualified_name);
170 
171 	if (attributes != NULL) {
172 		xmlChar    *qualified_name_attr = NULL;
173 		attrs = safe_emalloc((nb_attributes  * 2) + 1, sizeof(int *), 0);
174 
175 		for (i = 0; i < nb_attributes; i += 1) {
176 
177 			if (attributes[y+1] != NULL) {
178 				_qualify_namespace(parser, attributes[y] , attributes[y + 2], &qualified_name_attr);
179 			} else {
180 				qualified_name_attr = xmlStrdup(attributes[y]);
181 			}
182 			attrs[z] = qualified_name_attr;
183 			attrs[z + 1] = xmlStrndup(attributes[y + 3] , (int) (attributes[y + 4] - attributes[y + 3]));
184 			z += 2;
185 			y += 5;
186 		}
187 
188 		attrs[z] = NULL;
189 	}
190 	parser->h_start_element(parser->user, (const XML_Char *) qualified_name, (const XML_Char **) attrs);
191 	if (attrs) {
192 		for (i = 0; i < z; i++) {
193 			xmlFree(attrs[i]);
194 		}
195 		efree(attrs);
196 	}
197 	xmlFree(qualified_name);
198 }
199 
200 static void
_end_element_handler(void * user,const xmlChar * name)201 _end_element_handler(void *user, const xmlChar *name)
202 {
203 	xmlChar    *qualified_name;
204 	XML_Parser  parser = (XML_Parser) user;
205 
206 	if (parser->h_end_element == NULL) {
207 		if (parser->h_default) {
208 			char *end_element;
209 
210 			spprintf(&end_element, 0, "</%s>", (char *)name);
211 			parser->h_default(parser->user, (const XML_Char *) end_element, strlen(end_element));
212 			efree(end_element);
213 		}
214 		return;
215 	}
216 
217 	qualified_name = xmlStrdup(name);
218 
219 	parser->h_end_element(parser->user, (const XML_Char *) qualified_name);
220 
221 	xmlFree(qualified_name);
222 }
223 
224 static void
_end_element_handler_ns(void * user,const xmlChar * name,const xmlChar * prefix,const xmlChar * URI)225 _end_element_handler_ns(void *user, const xmlChar *name, const xmlChar * prefix, const xmlChar *URI)
226 {
227 	xmlChar    *qualified_name;
228 	XML_Parser  parser = (XML_Parser) user;
229 
230 	if (parser->h_end_element == NULL) {
231 		if (parser->h_default) {
232 			char *end_element;
233 			int end_element_len;
234 
235 			if (prefix) {
236 				end_element_len = spprintf(&end_element, 0, "</%s:%s>", (char *) prefix, (char *)name);
237 			} else {
238 				end_element_len = spprintf(&end_element, 0, "</%s>", (char *)name);
239 			}
240 			parser->h_default(parser->user, (const XML_Char *) end_element, end_element_len);
241 			efree(end_element);
242 		}
243 		return;
244 	}
245 
246 	_qualify_namespace(parser, name, URI,  &qualified_name);
247 
248 	parser->h_end_element(parser->user, (const XML_Char *) qualified_name);
249 
250 	xmlFree(qualified_name);
251 }
252 
253 static void
_cdata_handler(void * user,const xmlChar * cdata,int cdata_len)254 _cdata_handler(void *user, const xmlChar *cdata, int cdata_len)
255 {
256 	XML_Parser parser = (XML_Parser) user;
257 
258 	if (parser->h_cdata == NULL) {
259 		if (parser->h_default) {
260 			parser->h_default(parser->user, (const XML_Char *) cdata, cdata_len);
261 		}
262 		return;
263 	}
264 
265 	parser->h_cdata(parser->user, (const XML_Char *) cdata, cdata_len);
266 }
267 
268 static void
_pi_handler(void * user,const xmlChar * target,const xmlChar * data)269 _pi_handler(void *user, const xmlChar *target, const xmlChar *data)
270 {
271 	XML_Parser parser = (XML_Parser) user;
272 
273 	if (parser->h_pi == NULL) {
274 		if (parser->h_default) {
275 			char    *full_pi;
276 			spprintf(&full_pi, 0, "<?%s %s?>", (char *)target, (char *)data);
277 			parser->h_default(parser->user, (const XML_Char *) full_pi, strlen(full_pi));
278 			efree(full_pi);
279 		}
280 		return;
281 	}
282 
283 	parser->h_pi(parser->user, (const XML_Char *) target, (const XML_Char *) data);
284 }
285 
286 static void
_unparsed_entity_decl_handler(void * user,const xmlChar * name,const xmlChar * pub_id,const xmlChar * sys_id,const xmlChar * notation)287 _unparsed_entity_decl_handler(void *user,
288                               const xmlChar *name,
289 							  const xmlChar *pub_id,
290 							  const xmlChar *sys_id,
291 							  const xmlChar *notation)
292 {
293 	XML_Parser parser = (XML_Parser) user;
294 
295 	if (parser->h_unparsed_entity_decl == NULL) {
296 		return;
297 	}
298 
299 	parser->h_unparsed_entity_decl(parser->user, name, NULL, sys_id, pub_id, notation);
300 }
301 
302 static void
_notation_decl_handler(void * user,const xmlChar * notation,const xmlChar * pub_id,const xmlChar * sys_id)303 _notation_decl_handler(void *user, const xmlChar *notation, const xmlChar *pub_id, const xmlChar *sys_id)
304 {
305 	XML_Parser parser = (XML_Parser) user;
306 
307 	if (parser->h_notation_decl == NULL) {
308 		return;
309 	}
310 
311 	parser->h_notation_decl(parser->user, notation, NULL, sys_id, pub_id);
312 }
313 
314 static void
_build_comment(const xmlChar * data,int data_len,xmlChar ** comment,int * comment_len)315 _build_comment(const xmlChar *data, int data_len, xmlChar **comment, int *comment_len)
316 {
317 	*comment_len = data_len + 7;
318 
319 	*comment = xmlMalloc(*comment_len + 1);
320 	memcpy(*comment, "<!--", 4);
321 	memcpy(*comment + 4, data, data_len);
322 	memcpy(*comment + 4 + data_len, "-->", 3);
323 
324 	(*comment)[*comment_len] = '\0';
325 }
326 
327 static void
_comment_handler(void * user,const xmlChar * comment)328 _comment_handler(void *user, const xmlChar *comment)
329 {
330 	XML_Parser parser = (XML_Parser) user;
331 
332 	if (parser->h_default) {
333 		xmlChar *d_comment;
334 		int      d_comment_len;
335 
336 		_build_comment(comment, xmlStrlen(comment), &d_comment, &d_comment_len);
337 		parser->h_default(parser->user, d_comment, d_comment_len);
338 		xmlFree(d_comment);
339 	}
340 }
341 
342 static void
_build_entity(const xmlChar * name,int len,xmlChar ** entity,int * entity_len)343 _build_entity(const xmlChar *name, int len, xmlChar **entity, int *entity_len)
344 {
345 	*entity_len = len + 2;
346 	*entity = xmlMalloc(*entity_len + 1);
347 	(*entity)[0] = '&';
348 	memcpy(*entity+1, name, len);
349 	(*entity)[len+1] = ';';
350 	(*entity)[*entity_len] = '\0';
351 }
352 
353 static void
_external_entity_ref_handler(void * user,const xmlChar * names,int type,const xmlChar * sys_id,const xmlChar * pub_id,xmlChar * content)354 _external_entity_ref_handler(void *user, const xmlChar *names, int type, const xmlChar *sys_id, const xmlChar *pub_id, xmlChar *content)
355 {
356 	XML_Parser parser = (XML_Parser) user;
357 
358 	if (parser->h_external_entity_ref == NULL) {
359 		return;
360 	}
361 
362 	parser->h_external_entity_ref(parser, names, (XML_Char *) "", sys_id, pub_id);
363 }
364 
365 static xmlEntityPtr
_get_entity(void * user,const xmlChar * name)366 _get_entity(void *user, const xmlChar *name)
367 {
368 	XML_Parser parser = (XML_Parser) user;
369 	xmlEntityPtr ret = NULL;
370 
371 	if (parser->parser->inSubset == 0) {
372 		ret = xmlGetPredefinedEntity(name);
373 		if (ret == NULL)
374 			ret = xmlGetDocEntity(parser->parser->myDoc, name);
375 
376 		if (ret == NULL || (parser->parser->instate != XML_PARSER_ENTITY_VALUE && parser->parser->instate != XML_PARSER_ATTRIBUTE_VALUE)) {
377 			if (ret == NULL || ret->etype == XML_INTERNAL_GENERAL_ENTITY || ret->etype == XML_INTERNAL_PARAMETER_ENTITY || ret->etype == XML_INTERNAL_PREDEFINED_ENTITY) {
378 				/* Predefined entities will expand unless no cdata handler is present */
379 				if (parser->h_default && ! (ret && ret->etype == XML_INTERNAL_PREDEFINED_ENTITY && parser->h_cdata)) {
380 					xmlChar *entity;
381 					int      len;
382 
383 					_build_entity(name, xmlStrlen(name), &entity, &len);
384 					parser->h_default(parser->user, (const xmlChar *) entity, len);
385 					xmlFree(entity);
386 				} else {
387 					/* expat will not expand internal entities if default handler is present otherwise
388 					it will expand and pass them to cdata handler */
389 					if (parser->h_cdata && ret) {
390 						parser->h_cdata(parser->user, ret->content, xmlStrlen(ret->content));
391 					}
392 				}
393 			} else {
394 				if (ret->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) {
395 					_external_entity_ref_handler(user, ret->name, ret->etype, ret->SystemID, ret->ExternalID, NULL);
396 				}
397 			}
398 		}
399 	}
400 
401 	return ret;
402 }
403 
404 static xmlSAXHandler
405 php_xml_compat_handlers = {
406 	NULL, /* internalSubset */
407 	NULL, /* isStandalone */
408 	NULL, /* hasInternalSubset */
409 	NULL, /* hasExternalSubset */
410 	NULL, /* resolveEntity */
411 	_get_entity, /* getEntity */
412 	NULL, /* entityDecl */
413 	_notation_decl_handler,
414 	NULL, /* attributeDecl */
415 	NULL, /* elementDecl */
416 	_unparsed_entity_decl_handler, /* unparsedEntity */
417 	NULL, /* setDocumentLocator */
418 	NULL, /* startDocument */
419 	NULL, /* endDocument */
420 	_start_element_handler, /* startElement */
421 	_end_element_handler, /* endElement */
422 	NULL, /* reference */
423 	_cdata_handler,
424 	NULL, /* ignorableWhitespace */
425 	_pi_handler,
426 	_comment_handler, /* comment */
427 	NULL, /* warning */
428 	NULL, /* error */
429 	NULL,  /* fatalError */
430 	NULL,  /* getParameterEntity */
431 	_cdata_handler, /* cdataBlock */
432 	NULL, /* externalSubset */
433 	XML_SAX2_MAGIC,
434 	NULL,
435 	_start_element_handler_ns,
436 	_end_element_handler_ns,
437 	NULL
438 };
439 
440 PHP_XML_API XML_Parser
XML_ParserCreate(const XML_Char * encoding)441 XML_ParserCreate(const XML_Char *encoding)
442 {
443 	return XML_ParserCreate_MM(encoding, NULL, NULL);
444 }
445 
446 PHP_XML_API XML_Parser
XML_ParserCreateNS(const XML_Char * encoding,const XML_Char sep)447 XML_ParserCreateNS(const XML_Char *encoding, const XML_Char sep)
448 {
449 	XML_Char tmp[2];
450 	tmp[0] = sep;
451 	tmp[1] = '\0';
452 	return XML_ParserCreate_MM(encoding, NULL, tmp);
453 }
454 
455 PHP_XML_API XML_Parser
XML_ParserCreate_MM(const XML_Char * encoding,const XML_Memory_Handling_Suite * memsuite,const XML_Char * sep)456 XML_ParserCreate_MM(const XML_Char *encoding, const XML_Memory_Handling_Suite *memsuite, const XML_Char *sep)
457 {
458 	XML_Parser parser;
459 
460 	parser = (XML_Parser) emalloc(sizeof(struct _XML_Parser));
461 	memset(parser, 0, sizeof(struct _XML_Parser));
462 	parser->use_namespace = 0;
463 	parser->_ns_separator = NULL;
464 
465 	parser->parser = xmlCreatePushParserCtxt((xmlSAXHandlerPtr) &php_xml_compat_handlers, (void *) parser, NULL, 0, NULL);
466 	if (parser->parser == NULL) {
467 		efree(parser);
468 		return NULL;
469 	}
470 #if LIBXML_VERSION <= 20617
471 	/* for older versions of libxml2, allow correct detection of
472 	 * charset in documents with a BOM: */
473 	parser->parser->charset = XML_CHAR_ENCODING_NONE;
474 #endif
475 
476 #if LIBXML_VERSION >= 20703
477 	xmlCtxtUseOptions(parser->parser, XML_PARSE_OLDSAX);
478 #endif
479 
480 	parser->parser->replaceEntities = 1;
481 	parser->parser->wellFormed = 0;
482 	if (sep != NULL) {
483 		parser->use_namespace = 1;
484 		parser->parser->sax2 = 1;
485 		parser->_ns_separator = xmlStrdup(sep);
486 	} else {
487 		/* Reset flag as XML_SAX2_MAGIC is needed for xmlCreatePushParserCtxt
488 		so must be set in the handlers */
489 		parser->parser->sax->initialized = 1;
490 	}
491 	return parser;
492 }
493 
494 PHP_XML_API void
XML_SetUserData(XML_Parser parser,void * user)495 XML_SetUserData(XML_Parser parser, void *user)
496 {
497 	parser->user = user;
498 }
499 
500 PHP_XML_API void *
XML_GetUserData(XML_Parser parser)501 XML_GetUserData(XML_Parser parser)
502 {
503 	return parser->user;
504 }
505 
506 PHP_XML_API void
XML_SetElementHandler(XML_Parser parser,XML_StartElementHandler start,XML_EndElementHandler end)507 XML_SetElementHandler(XML_Parser parser, XML_StartElementHandler start, XML_EndElementHandler end)
508 {
509 	parser->h_start_element = start;
510 	parser->h_end_element = end;
511 }
512 
513 PHP_XML_API void
XML_SetCharacterDataHandler(XML_Parser parser,XML_CharacterDataHandler cdata)514 XML_SetCharacterDataHandler(XML_Parser parser, XML_CharacterDataHandler cdata)
515 {
516 	parser->h_cdata = cdata;
517 }
518 
519 PHP_XML_API void
XML_SetProcessingInstructionHandler(XML_Parser parser,XML_ProcessingInstructionHandler pi)520 XML_SetProcessingInstructionHandler(XML_Parser parser, XML_ProcessingInstructionHandler pi)
521 {
522 	parser->h_pi = pi;
523 }
524 
525 PHP_XML_API void
XML_SetCommentHandler(XML_Parser parser,XML_CommentHandler comment)526 XML_SetCommentHandler(XML_Parser parser, XML_CommentHandler comment)
527 {
528 	parser->h_comment = comment;
529 }
530 
531 PHP_XML_API void
XML_SetDefaultHandler(XML_Parser parser,XML_DefaultHandler d)532 XML_SetDefaultHandler(XML_Parser parser, XML_DefaultHandler d)
533 {
534 	parser->h_default = d;
535 }
536 
537 PHP_XML_API void
XML_SetUnparsedEntityDeclHandler(XML_Parser parser,XML_UnparsedEntityDeclHandler unparsed_decl)538 XML_SetUnparsedEntityDeclHandler(XML_Parser parser, XML_UnparsedEntityDeclHandler unparsed_decl)
539 {
540 	parser->h_unparsed_entity_decl = unparsed_decl;
541 }
542 
543 PHP_XML_API void
XML_SetNotationDeclHandler(XML_Parser parser,XML_NotationDeclHandler notation_decl)544 XML_SetNotationDeclHandler(XML_Parser parser, XML_NotationDeclHandler notation_decl)
545 {
546 	parser->h_notation_decl = notation_decl;
547 }
548 
549 PHP_XML_API void
XML_SetExternalEntityRefHandler(XML_Parser parser,XML_ExternalEntityRefHandler ext_entity)550 XML_SetExternalEntityRefHandler(XML_Parser parser, XML_ExternalEntityRefHandler ext_entity)
551 {
552 	parser->h_external_entity_ref = ext_entity;
553 }
554 
555 PHP_XML_API void
XML_SetStartNamespaceDeclHandler(XML_Parser parser,XML_StartNamespaceDeclHandler start_ns)556 XML_SetStartNamespaceDeclHandler(XML_Parser parser, XML_StartNamespaceDeclHandler start_ns)
557 {
558 	parser->h_start_ns = start_ns;
559 }
560 
561 PHP_XML_API void
XML_SetEndNamespaceDeclHandler(XML_Parser parser,XML_EndNamespaceDeclHandler end_ns)562 XML_SetEndNamespaceDeclHandler(XML_Parser parser, XML_EndNamespaceDeclHandler end_ns)
563 {
564 	parser->h_end_ns = end_ns;
565 }
566 
567 PHP_XML_API int
XML_Parse(XML_Parser parser,const XML_Char * data,int data_len,int is_final)568 XML_Parse(XML_Parser parser, const XML_Char *data, int data_len, int is_final)
569 {
570 	int error;
571 
572 /* The following is a hack to keep BC with PHP 4 while avoiding
573 the inifite loop in libxml <= 2.6.17 which occurs when no encoding
574 has been defined and none can be detected */
575 #if LIBXML_VERSION <= 20617
576 	if (parser->parser->charset == XML_CHAR_ENCODING_NONE) {
577 		if (data_len >= 4 || (parser->parser->input->buf->buffer->use + data_len >= 4)) {
578 			xmlChar start[4];
579 			int char_count;
580 
581 			char_count = parser->parser->input->buf->buffer->use;
582 			if (char_count > 4) {
583 				char_count = 4;
584 			}
585 
586 			memcpy(start, parser->parser->input->buf->buffer->content, (size_t)char_count);
587 			memcpy(start + char_count, data, (size_t)(4 - char_count));
588 
589 			if (xmlDetectCharEncoding(&start[0], 4) == XML_CHAR_ENCODING_NONE) {
590 				parser->parser->charset = XML_CHAR_ENCODING_UTF8;
591 			}
592 		}
593 	}
594 #endif
595 
596 	if (parser->parser->lastError.level >= XML_ERR_WARNING) {
597 		return 0;
598 	}
599 
600 	error = xmlParseChunk(parser->parser, (char *) data, data_len, is_final);
601 	if (error) {
602 		return 0;
603 	} else {
604 		return 1;
605 	}
606 }
607 
608 PHP_XML_API int
XML_GetErrorCode(XML_Parser parser)609 XML_GetErrorCode(XML_Parser parser)
610 {
611 	return parser->parser->errNo;
612 }
613 
614 static const XML_Char *const error_mapping[] = {
615 	(const XML_Char *)"No error",
616 	(const XML_Char *)"No memory",
617 	(const XML_Char *)"Invalid document start",
618 	(const XML_Char *)"Empty document",
619 	(const XML_Char *)"Not well-formed (invalid token)",
620 	(const XML_Char *)"Invalid document end",
621 	(const XML_Char *)"Invalid hexadecimal character reference",
622 	(const XML_Char *)"Invalid decimal character reference",
623 	(const XML_Char *)"Invalid character reference",
624 	(const XML_Char *)"Invalid character",
625 	(const XML_Char *)"XML_ERR_CHARREF_AT_EOF",
626 	(const XML_Char *)"XML_ERR_CHARREF_IN_PROLOG",
627 	(const XML_Char *)"XML_ERR_CHARREF_IN_EPILOG",
628 	(const XML_Char *)"XML_ERR_CHARREF_IN_DTD",
629 	(const XML_Char *)"XML_ERR_ENTITYREF_AT_EOF",
630 	(const XML_Char *)"XML_ERR_ENTITYREF_IN_PROLOG",
631 	(const XML_Char *)"XML_ERR_ENTITYREF_IN_EPILOG",
632 	(const XML_Char *)"XML_ERR_ENTITYREF_IN_DTD",
633 	(const XML_Char *)"PEReference at end of document",
634 	(const XML_Char *)"PEReference in prolog",
635 	(const XML_Char *)"PEReference in epilog",
636 	(const XML_Char *)"PEReference: forbidden within markup decl in internal subset",
637 	(const XML_Char *)"XML_ERR_ENTITYREF_NO_NAME",
638 	(const XML_Char *)"EntityRef: expecting ';'",
639 	(const XML_Char *)"PEReference: no name",
640 	(const XML_Char *)"PEReference: expecting ';'",
641 	(const XML_Char *)"Undeclared entity error",
642 	(const XML_Char *)"Undeclared entity warning",
643 	(const XML_Char *)"Unparsed Entity",
644 	(const XML_Char *)"XML_ERR_ENTITY_IS_EXTERNAL",
645 	(const XML_Char *)"XML_ERR_ENTITY_IS_PARAMETER",
646 	(const XML_Char *)"Unknown encoding",
647 	(const XML_Char *)"Unsupported encoding",
648 	(const XML_Char *)"String not started expecting ' or \"",
649 	(const XML_Char *)"String not closed expecting \" or '",
650 	(const XML_Char *)"Namespace declaration error",
651 	(const XML_Char *)"EntityValue: \" or ' expected",
652 	(const XML_Char *)"EntityValue: \" or ' expected",
653 	(const XML_Char *)"< in attribute",
654 	(const XML_Char *)"Attribute not started",
655 	(const XML_Char *)"Attribute not finished",
656 	(const XML_Char *)"Attribute without value",
657 	(const XML_Char *)"Attribute redefined",
658 	(const XML_Char *)"SystemLiteral \" or ' expected",
659 	(const XML_Char *)"SystemLiteral \" or ' expected",
660 	/* (const XML_Char *)"XML_ERR_COMMENT_NOT_STARTED", <= eliminated on purpose */
661 	(const XML_Char *)"Comment not finished",
662 	(const XML_Char *)"Processing Instruction not started",
663 	(const XML_Char *)"Processing Instruction not finished",
664 	(const XML_Char *)"NOTATION: Name expected here",
665 	(const XML_Char *)"'>' required to close NOTATION declaration",
666 	(const XML_Char *)"'(' required to start ATTLIST enumeration",
667 	(const XML_Char *)"'(' required to start ATTLIST enumeration",
668 	(const XML_Char *)"MixedContentDecl : '|' or ')*' expected",
669 	(const XML_Char *)"XML_ERR_MIXED_NOT_FINISHED",
670 	(const XML_Char *)"ELEMENT in DTD not started",
671 	(const XML_Char *)"ELEMENT in DTD not finished",
672 	(const XML_Char *)"XML declaration not started",
673 	(const XML_Char *)"XML declaration not finished",
674 	(const XML_Char *)"XML_ERR_CONDSEC_NOT_STARTED",
675 	(const XML_Char *)"XML conditional section not closed",
676 	(const XML_Char *)"Content error in the external subset",
677 	(const XML_Char *)"DOCTYPE not finished",
678 	(const XML_Char *)"Sequence ']]>' not allowed in content",
679 	(const XML_Char *)"CDATA not finished",
680 	(const XML_Char *)"Reserved XML Name",
681 	(const XML_Char *)"Space required",
682 	(const XML_Char *)"XML_ERR_SEPARATOR_REQUIRED",
683 	(const XML_Char *)"NmToken expected in ATTLIST enumeration",
684 	(const XML_Char *)"XML_ERR_NAME_REQUIRED",
685 	(const XML_Char *)"MixedContentDecl : '#PCDATA' expected",
686 	(const XML_Char *)"SYSTEM or PUBLIC, the URI is missing",
687 	(const XML_Char *)"PUBLIC, the Public Identifier is missing",
688 	(const XML_Char *)"< required",
689 	(const XML_Char *)"> required",
690 	(const XML_Char *)"</ required",
691 	(const XML_Char *)"= required",
692 	(const XML_Char *)"Mismatched tag",
693 	(const XML_Char *)"Tag not finished",
694 	(const XML_Char *)"standalone accepts only 'yes' or 'no'",
695 	(const XML_Char *)"Invalid XML encoding name",
696 	(const XML_Char *)"Comment must not contain '--' (double-hyphen)",
697 	(const XML_Char *)"Invalid encoding",
698 	(const XML_Char *)"external parsed entities cannot be standalone",
699 	(const XML_Char *)"XML conditional section '[' expected",
700 	(const XML_Char *)"Entity value required",
701 	(const XML_Char *)"chunk is not well balanced",
702 	(const XML_Char *)"extra content at the end of well balanced chunk",
703 	(const XML_Char *)"XML_ERR_ENTITY_CHAR_ERROR",
704 	(const XML_Char *)"PEReferences forbidden in internal subset",
705 	(const XML_Char *)"Detected an entity reference loop",
706 	(const XML_Char *)"XML_ERR_ENTITY_BOUNDARY",
707 	(const XML_Char *)"Invalid URI",
708 	(const XML_Char *)"Fragment not allowed",
709 	(const XML_Char *)"XML_WAR_CATALOG_PI",
710 	(const XML_Char *)"XML_ERR_NO_DTD",
711 	(const XML_Char *)"conditional section INCLUDE or IGNORE keyword expected", /* 95 */
712 	(const XML_Char *)"Version in XML Declaration missing", /* 96 */
713 	(const XML_Char *)"XML_WAR_UNKNOWN_VERSION", /* 97 */
714 	(const XML_Char *)"XML_WAR_LANG_VALUE", /* 98 */
715 	(const XML_Char *)"XML_WAR_NS_URI", /* 99 */
716 	(const XML_Char *)"XML_WAR_NS_URI_RELATIVE", /* 100 */
717 	(const XML_Char *)"Missing encoding in text declaration" /* 101 */
718 };
719 
720 PHP_XML_API const XML_Char *
XML_ErrorString(int code)721 XML_ErrorString(int code)
722 {
723 	if (code < 0 || code >= (int)(sizeof(error_mapping) / sizeof(error_mapping[0]))) {
724 		return (const XML_Char *) "Unknown";
725 	}
726 	return error_mapping[code];
727 }
728 
729 PHP_XML_API int
XML_GetCurrentLineNumber(XML_Parser parser)730 XML_GetCurrentLineNumber(XML_Parser parser)
731 {
732 	return parser->parser->input->line;
733 }
734 
735 PHP_XML_API int
XML_GetCurrentColumnNumber(XML_Parser parser)736 XML_GetCurrentColumnNumber(XML_Parser parser)
737 {
738 	return parser->parser->input->col;
739 }
740 
741 PHP_XML_API int
XML_GetCurrentByteIndex(XML_Parser parser)742 XML_GetCurrentByteIndex(XML_Parser parser)
743 {
744 	return parser->parser->input->consumed +
745 			(parser->parser->input->cur - parser->parser->input->base);
746 }
747 
748 PHP_XML_API int
XML_GetCurrentByteCount(XML_Parser parser)749 XML_GetCurrentByteCount(XML_Parser parser)
750 {
751 	/* WARNING: this is identical to ByteIndex; it should probably
752 	 * be different */
753 	return parser->parser->input->consumed +
754 			(parser->parser->input->cur - parser->parser->input->base);
755 }
756 
XML_ExpatVersion(void)757 PHP_XML_API const XML_Char *XML_ExpatVersion(void)
758 {
759 	return (const XML_Char *) "1.0";
760 }
761 
762 PHP_XML_API void
XML_ParserFree(XML_Parser parser)763 XML_ParserFree(XML_Parser parser)
764 {
765 	if (parser->use_namespace) {
766 		if (parser->_ns_separator) {
767 			xmlFree(parser->_ns_separator);
768 		}
769 	}
770 	if (parser->parser->myDoc) {
771 		xmlFreeDoc(parser->parser->myDoc);
772 		parser->parser->myDoc = NULL;
773 	}
774 	xmlFreeParserCtxt(parser->parser);
775 	efree(parser);
776 }
777 
778 #endif /* LIBXML_EXPAT_COMPAT */
779 #endif
780 
781 /**
782  * Local Variables:
783  * tab-width: 4
784  * c-basic-offset: 4
785  * indent-tabs-mode: t
786  * End:
787  * vim600: fdm=marker
788  * vim: ts=4 noet sw=4
789  */
790