1/*
2 +----------------------------------------------------------------------+
3 | PHP Version 5 |
4 +----------------------------------------------------------------------+
5 | Copyright (c) 1997-2015 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: Stig Sæther Bakken <ssb@php.net> |
16 | Thies C. Arntzen <thies@thieso.net> |
17 | Sterling Hughes <sterling@php.net> |
18 +----------------------------------------------------------------------+
19 */
20
21/* $Id$ */
22
23#define IS_EXT_MODULE
24
25#ifdef HAVE_CONFIG_H
26#include "config.h"
27#endif
28
29#include "php.h"
30
31#define PHP_XML_INTERNAL
32#include "zend_variables.h"
33#include "ext/standard/php_string.h"
34#include "ext/standard/info.h"
35#include "ext/standard/html.h"
36
37#if HAVE_XML
38
39#include "php_xml.h"
40# include "ext/standard/head.h"
41#ifdef LIBXML_EXPAT_COMPAT
42#include "ext/libxml/php_libxml.h"
43#endif
44
45/* Short-term TODO list:
46 * - Implement XML_ExternalEntityParserCreate()
47 * - XML_SetCommentHandler
48 * - XML_SetCdataSectionHandler
49 * - XML_SetParamEntityParsing
50 */
51
52/* Long-term TODO list:
53 * - Fix the expat library so you can install your own memory manager
54 * functions
55 */
56
57/* Known bugs:
58 * - Weird things happen with <![CDATA[]]> sections.
59 */
60
61ZEND_DECLARE_MODULE_GLOBALS(xml)
62
63/* {{{ dynamically loadable module stuff */
64#ifdef COMPILE_DL_XML
65ZEND_GET_MODULE(xml)
66#endif /* COMPILE_DL_XML */
67/* }}} */
68
69/* {{{ function prototypes */
70PHP_MINIT_FUNCTION(xml);
71PHP_MINFO_FUNCTION(xml);
72static PHP_GINIT_FUNCTION(xml);
73
74static void xml_parser_dtor(zend_rsrc_list_entry *rsrc TSRMLS_DC);
75static void xml_set_handler(zval **, zval **);
76inline static unsigned short xml_encode_iso_8859_1(unsigned char);
77inline static char xml_decode_iso_8859_1(unsigned short);
78inline static unsigned short xml_encode_us_ascii(unsigned char);
79inline static char xml_decode_us_ascii(unsigned short);
80static zval *xml_call_handler(xml_parser *, zval *, zend_function *, int, zval **);
81static zval *_xml_xmlchar_zval(const XML_Char *, int, const XML_Char *);
82static int _xml_xmlcharlen(const XML_Char *);
83static void _xml_add_to_info(xml_parser *parser,char *name);
84inline static char *_xml_decode_tag(xml_parser *parser, const char *tag);
85
86void _xml_startElementHandler(void *, const XML_Char *, const XML_Char **);
87void _xml_endElementHandler(void *, const XML_Char *);
88void _xml_characterDataHandler(void *, const XML_Char *, int);
89void _xml_processingInstructionHandler(void *, const XML_Char *, const XML_Char *);
90void _xml_defaultHandler(void *, const XML_Char *, int);
91void _xml_unparsedEntityDeclHandler(void *, const XML_Char *, const XML_Char *, const XML_Char *, const XML_Char *, const XML_Char *);
92void _xml_notationDeclHandler(void *, const XML_Char *, const XML_Char *, const XML_Char *, const XML_Char *);
93int _xml_externalEntityRefHandler(XML_Parser, const XML_Char *, const XML_Char *, const XML_Char *, const XML_Char *);
94
95void _xml_startNamespaceDeclHandler(void *, const XML_Char *, const XML_Char *);
96void _xml_endNamespaceDeclHandler(void *, const XML_Char *);
97/* }}} */
98
99/* {{{ extension definition structures */
100ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_parser_create, 0, 0, 0)
101 ZEND_ARG_INFO(0, encoding)
102ZEND_END_ARG_INFO()
103
104ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_parser_create_ns, 0, 0, 0)
105 ZEND_ARG_INFO(0, encoding)
106 ZEND_ARG_INFO(0, sep)
107ZEND_END_ARG_INFO()
108
109ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_set_object, 0, 0, 2)
110 ZEND_ARG_INFO(0, parser)
111 ZEND_ARG_INFO(1, obj)
112ZEND_END_ARG_INFO()
113
114ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_set_element_handler, 0, 0, 3)
115 ZEND_ARG_INFO(0, parser)
116 ZEND_ARG_INFO(0, shdl)
117 ZEND_ARG_INFO(0, ehdl)
118ZEND_END_ARG_INFO()
119
120ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_set_character_data_handler, 0, 0, 2)
121 ZEND_ARG_INFO(0, parser)
122 ZEND_ARG_INFO(0, hdl)
123ZEND_END_ARG_INFO()
124
125ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_set_processing_instruction_handler, 0, 0, 2)
126 ZEND_ARG_INFO(0, parser)
127 ZEND_ARG_INFO(0, hdl)
128ZEND_END_ARG_INFO()
129
130ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_set_default_handler, 0, 0, 2)
131 ZEND_ARG_INFO(0, parser)
132 ZEND_ARG_INFO(0, hdl)
133ZEND_END_ARG_INFO()
134
135ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_set_unparsed_entity_decl_handler, 0, 0, 2)
136 ZEND_ARG_INFO(0, parser)
137 ZEND_ARG_INFO(0, hdl)
138ZEND_END_ARG_INFO()
139
140ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_set_notation_decl_handler, 0, 0, 2)
141 ZEND_ARG_INFO(0, parser)
142 ZEND_ARG_INFO(0, hdl)
143ZEND_END_ARG_INFO()
144
145ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_set_external_entity_ref_handler, 0, 0, 2)
146 ZEND_ARG_INFO(0, parser)
147 ZEND_ARG_INFO(0, hdl)
148ZEND_END_ARG_INFO()
149
150ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_set_start_namespace_decl_handler, 0, 0, 2)
151 ZEND_ARG_INFO(0, parser)
152 ZEND_ARG_INFO(0, hdl)
153ZEND_END_ARG_INFO()
154
155ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_set_end_namespace_decl_handler, 0, 0, 2)
156 ZEND_ARG_INFO(0, parser)
157 ZEND_ARG_INFO(0, hdl)
158ZEND_END_ARG_INFO()
159
160ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_parse, 0, 0, 2)
161 ZEND_ARG_INFO(0, parser)
162 ZEND_ARG_INFO(0, data)
163 ZEND_ARG_INFO(0, isfinal)
164ZEND_END_ARG_INFO()
165
166ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_parse_into_struct, 0, 0, 3)
167 ZEND_ARG_INFO(0, parser)
168 ZEND_ARG_INFO(0, data)
169 ZEND_ARG_INFO(1, values)
170 ZEND_ARG_INFO(1, index)
171ZEND_END_ARG_INFO()
172
173ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_get_error_code, 0, 0, 1)
174 ZEND_ARG_INFO(0, parser)
175ZEND_END_ARG_INFO()
176
177ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_error_string, 0, 0, 1)
178 ZEND_ARG_INFO(0, code)
179ZEND_END_ARG_INFO()
180
181ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_get_current_line_number, 0, 0, 1)
182 ZEND_ARG_INFO(0, parser)
183ZEND_END_ARG_INFO()
184
185ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_get_current_column_number, 0, 0, 1)
186 ZEND_ARG_INFO(0, parser)
187ZEND_END_ARG_INFO()
188
189ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_get_current_byte_index, 0, 0, 1)
190 ZEND_ARG_INFO(0, parser)
191ZEND_END_ARG_INFO()
192
193ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_parser_free, 0, 0, 1)
194 ZEND_ARG_INFO(0, parser)
195ZEND_END_ARG_INFO()
196
197ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_parser_set_option, 0, 0, 3)
198 ZEND_ARG_INFO(0, parser)
199 ZEND_ARG_INFO(0, option)
200 ZEND_ARG_INFO(0, value)
201ZEND_END_ARG_INFO()
202
203ZEND_BEGIN_ARG_INFO_EX(arginfo_xml_parser_get_option, 0, 0, 2)
204 ZEND_ARG_INFO(0, parser)
205 ZEND_ARG_INFO(0, option)
206ZEND_END_ARG_INFO()
207
208ZEND_BEGIN_ARG_INFO_EX(arginfo_utf8_encode, 0, 0, 1)
209 ZEND_ARG_INFO(0, data)
210ZEND_END_ARG_INFO()
211
212ZEND_BEGIN_ARG_INFO_EX(arginfo_utf8_decode, 0, 0, 1)
213 ZEND_ARG_INFO(0, data)
214ZEND_END_ARG_INFO()
215
216const zend_function_entry xml_functions[] = {
217 PHP_FE(xml_parser_create, arginfo_xml_parser_create)
218 PHP_FE(xml_parser_create_ns, arginfo_xml_parser_create_ns)
219 PHP_FE(xml_set_object, arginfo_xml_set_object)
220 PHP_FE(xml_set_element_handler, arginfo_xml_set_element_handler)
221 PHP_FE(xml_set_character_data_handler, arginfo_xml_set_character_data_handler)
222 PHP_FE(xml_set_processing_instruction_handler, arginfo_xml_set_processing_instruction_handler)
223 PHP_FE(xml_set_default_handler, arginfo_xml_set_default_handler)
224 PHP_FE(xml_set_unparsed_entity_decl_handler,arginfo_xml_set_unparsed_entity_decl_handler)
225 PHP_FE(xml_set_notation_decl_handler, arginfo_xml_set_notation_decl_handler)
226 PHP_FE(xml_set_external_entity_ref_handler, arginfo_xml_set_external_entity_ref_handler)
227 PHP_FE(xml_set_start_namespace_decl_handler,arginfo_xml_set_start_namespace_decl_handler)
228 PHP_FE(xml_set_end_namespace_decl_handler, arginfo_xml_set_end_namespace_decl_handler)
229 PHP_FE(xml_parse, arginfo_xml_parse)
230 PHP_FE(xml_parse_into_struct, arginfo_xml_parse_into_struct)
231 PHP_FE(xml_get_error_code, arginfo_xml_get_error_code)
232 PHP_FE(xml_error_string, arginfo_xml_error_string)
233 PHP_FE(xml_get_current_line_number, arginfo_xml_get_current_line_number)
234 PHP_FE(xml_get_current_column_number, arginfo_xml_get_current_column_number)
235 PHP_FE(xml_get_current_byte_index, arginfo_xml_get_current_byte_index)
236 PHP_FE(xml_parser_free, arginfo_xml_parser_free)
237 PHP_FE(xml_parser_set_option, arginfo_xml_parser_set_option)
238 PHP_FE(xml_parser_get_option, arginfo_xml_parser_get_option)
239 PHP_FE(utf8_encode, arginfo_utf8_encode)
240 PHP_FE(utf8_decode, arginfo_utf8_decode)
241 PHP_FE_END
242};
243
244#ifdef LIBXML_EXPAT_COMPAT
245static const zend_module_dep xml_deps[] = {
246 ZEND_MOD_REQUIRED("libxml")
247 ZEND_MOD_END
248};
249#endif
250
251zend_module_entry xml_module_entry = {
252#ifdef LIBXML_EXPAT_COMPAT
253 STANDARD_MODULE_HEADER_EX, NULL,
254 xml_deps,
255#else
256 STANDARD_MODULE_HEADER,
257#endif
258 "xml", /* extension name */
259 xml_functions, /* extension function list */
260 PHP_MINIT(xml), /* extension-wide startup function */
261 NULL, /* extension-wide shutdown function */
262 NULL, /* per-request startup function */
263 NULL, /* per-request shutdown function */
264 PHP_MINFO(xml), /* information function */
265 NO_VERSION_YET,
266 PHP_MODULE_GLOBALS(xml), /* globals descriptor */
267 PHP_GINIT(xml), /* globals ctor */
268 NULL, /* globals dtor */
269 NULL, /* post deactivate */
270 STANDARD_MODULE_PROPERTIES_EX
271};
272
273/* All the encoding functions are set to NULL right now, since all
274 * the encoding is currently done internally by expat/xmltok.
275 */
276xml_encoding xml_encodings[] = {
277 { (XML_Char *)"ISO-8859-1", xml_decode_iso_8859_1, xml_encode_iso_8859_1 },
278 { (XML_Char *)"US-ASCII", xml_decode_us_ascii, xml_encode_us_ascii },
279 { (XML_Char *)"UTF-8", NULL, NULL },
280 { (XML_Char *)NULL, NULL, NULL }
281};
282
283static XML_Memory_Handling_Suite php_xml_mem_hdlrs;
284
285/* True globals, no need for thread safety */
286static int le_xml_parser;
287
288/* }}} */
289
290/* {{{ startup, shutdown and info functions */
291static PHP_GINIT_FUNCTION(xml)
292{
293 xml_globals->default_encoding = "UTF-8";
294}
295
296static void *php_xml_malloc_wrapper(size_t sz)
297{
298 return emalloc(sz);
299}
300
301static void *php_xml_realloc_wrapper(void *ptr, size_t sz)
302{
303 return erealloc(ptr, sz);
304}
305
306static void php_xml_free_wrapper(void *ptr)
307{
308 if (ptr != NULL) {
309 efree(ptr);
310 }
311}
312
313PHP_MINIT_FUNCTION(xml)
314{
315 le_xml_parser = zend_register_list_destructors_ex(xml_parser_dtor, NULL, "xml", module_number);
316
317 REGISTER_LONG_CONSTANT("XML_ERROR_NONE", XML_ERROR_NONE, CONST_CS|CONST_PERSISTENT);
318 REGISTER_LONG_CONSTANT("XML_ERROR_NO_MEMORY", XML_ERROR_NO_MEMORY, CONST_CS|CONST_PERSISTENT);
319 REGISTER_LONG_CONSTANT("XML_ERROR_SYNTAX", XML_ERROR_SYNTAX, CONST_CS|CONST_PERSISTENT);
320 REGISTER_LONG_CONSTANT("XML_ERROR_NO_ELEMENTS", XML_ERROR_NO_ELEMENTS, CONST_CS|CONST_PERSISTENT);
321 REGISTER_LONG_CONSTANT("XML_ERROR_INVALID_TOKEN", XML_ERROR_INVALID_TOKEN, CONST_CS|CONST_PERSISTENT);
322 REGISTER_LONG_CONSTANT("XML_ERROR_UNCLOSED_TOKEN", XML_ERROR_UNCLOSED_TOKEN, CONST_CS|CONST_PERSISTENT);
323 REGISTER_LONG_CONSTANT("XML_ERROR_PARTIAL_CHAR", XML_ERROR_PARTIAL_CHAR, CONST_CS|CONST_PERSISTENT);
324 REGISTER_LONG_CONSTANT("XML_ERROR_TAG_MISMATCH", XML_ERROR_TAG_MISMATCH, CONST_CS|CONST_PERSISTENT);
325 REGISTER_LONG_CONSTANT("XML_ERROR_DUPLICATE_ATTRIBUTE", XML_ERROR_DUPLICATE_ATTRIBUTE, CONST_CS|CONST_PERSISTENT);
326 REGISTER_LONG_CONSTANT("XML_ERROR_JUNK_AFTER_DOC_ELEMENT", XML_ERROR_JUNK_AFTER_DOC_ELEMENT, CONST_CS|CONST_PERSISTENT);
327 REGISTER_LONG_CONSTANT("XML_ERROR_PARAM_ENTITY_REF", XML_ERROR_PARAM_ENTITY_REF, CONST_CS|CONST_PERSISTENT);
328 REGISTER_LONG_CONSTANT("XML_ERROR_UNDEFINED_ENTITY", XML_ERROR_UNDEFINED_ENTITY, CONST_CS|CONST_PERSISTENT);
329 REGISTER_LONG_CONSTANT("XML_ERROR_RECURSIVE_ENTITY_REF", XML_ERROR_RECURSIVE_ENTITY_REF, CONST_CS|CONST_PERSISTENT);
330 REGISTER_LONG_CONSTANT("XML_ERROR_ASYNC_ENTITY", XML_ERROR_ASYNC_ENTITY, CONST_CS|CONST_PERSISTENT);
331 REGISTER_LONG_CONSTANT("XML_ERROR_BAD_CHAR_REF", XML_ERROR_BAD_CHAR_REF, CONST_CS|CONST_PERSISTENT);
332 REGISTER_LONG_CONSTANT("XML_ERROR_BINARY_ENTITY_REF", XML_ERROR_BINARY_ENTITY_REF, CONST_CS|CONST_PERSISTENT);
333 REGISTER_LONG_CONSTANT("XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF", XML_ERROR_ATTRIBUTE_EXTERNAL_ENTITY_REF, CONST_CS|CONST_PERSISTENT);
334 REGISTER_LONG_CONSTANT("XML_ERROR_MISPLACED_XML_PI", XML_ERROR_MISPLACED_XML_PI, CONST_CS|CONST_PERSISTENT);
335 REGISTER_LONG_CONSTANT("XML_ERROR_UNKNOWN_ENCODING", XML_ERROR_UNKNOWN_ENCODING, CONST_CS|CONST_PERSISTENT);
336 REGISTER_LONG_CONSTANT("XML_ERROR_INCORRECT_ENCODING", XML_ERROR_INCORRECT_ENCODING, CONST_CS|CONST_PERSISTENT);
337 REGISTER_LONG_CONSTANT("XML_ERROR_UNCLOSED_CDATA_SECTION", XML_ERROR_UNCLOSED_CDATA_SECTION, CONST_CS|CONST_PERSISTENT);
338 REGISTER_LONG_CONSTANT("XML_ERROR_EXTERNAL_ENTITY_HANDLING", XML_ERROR_EXTERNAL_ENTITY_HANDLING, CONST_CS|CONST_PERSISTENT);
339
340 REGISTER_LONG_CONSTANT("XML_OPTION_CASE_FOLDING", PHP_XML_OPTION_CASE_FOLDING, CONST_CS|CONST_PERSISTENT);
341 REGISTER_LONG_CONSTANT("XML_OPTION_TARGET_ENCODING", PHP_XML_OPTION_TARGET_ENCODING, CONST_CS|CONST_PERSISTENT);
342 REGISTER_LONG_CONSTANT("XML_OPTION_SKIP_TAGSTART", PHP_XML_OPTION_SKIP_TAGSTART, CONST_CS|CONST_PERSISTENT);
343 REGISTER_LONG_CONSTANT("XML_OPTION_SKIP_WHITE", PHP_XML_OPTION_SKIP_WHITE, CONST_CS|CONST_PERSISTENT);
344
345 /* this object should not be pre-initialised at compile time,
346 as the order of members may vary */
347
348 php_xml_mem_hdlrs.malloc_fcn = php_xml_malloc_wrapper;
349 php_xml_mem_hdlrs.realloc_fcn = php_xml_realloc_wrapper;
350 php_xml_mem_hdlrs.free_fcn = php_xml_free_wrapper;
351
352#ifdef LIBXML_EXPAT_COMPAT
353 REGISTER_STRING_CONSTANT("XML_SAX_IMPL", "libxml", CONST_CS|CONST_PERSISTENT);
354#else
355 REGISTER_STRING_CONSTANT("XML_SAX_IMPL", "expat", CONST_CS|CONST_PERSISTENT);
356#endif
357
358 return SUCCESS;
359}
360
361PHP_MINFO_FUNCTION(xml)
362{
363 php_info_print_table_start();
364 php_info_print_table_row(2, "XML Support", "active");
365 php_info_print_table_row(2, "XML Namespace Support", "active");
366#if defined(LIBXML_DOTTED_VERSION) && defined(LIBXML_EXPAT_COMPAT)
367 php_info_print_table_row(2, "libxml2 Version", LIBXML_DOTTED_VERSION);
368#else
369 php_info_print_table_row(2, "EXPAT Version", XML_ExpatVersion());
370#endif
371 php_info_print_table_end();
372}
373/* }}} */
374
375/* {{{ extension-internal functions */
376static zval *_xml_resource_zval(long value)
377{
378 zval *ret;
379 TSRMLS_FETCH();
380
381 MAKE_STD_ZVAL(ret);
382
383 Z_TYPE_P(ret) = IS_RESOURCE;
384 Z_LVAL_P(ret) = value;
385
386 zend_list_addref(value);
387
388 return ret;
389}
390
391static zval *_xml_string_zval(const char *str)
392{
393 zval *ret;
394 int len = strlen(str);
395 MAKE_STD_ZVAL(ret);
396
397 Z_TYPE_P(ret) = IS_STRING;
398 Z_STRLEN_P(ret) = len;
399 Z_STRVAL_P(ret) = estrndup(str, len);
400 return ret;
401}
402
403static zval *_xml_xmlchar_zval(const XML_Char *s, int len, const XML_Char *encoding)
404{
405 zval *ret;
406 MAKE_STD_ZVAL(ret);
407
408 if (s == NULL) {
409 ZVAL_FALSE(ret);
410 return ret;
411 }
412 if (len == 0) {
413 len = _xml_xmlcharlen(s);
414 }
415 Z_TYPE_P(ret) = IS_STRING;
416 Z_STRVAL_P(ret) = xml_utf8_decode(s, len, &Z_STRLEN_P(ret), encoding);
417 return ret;
418}
419/* }}} */
420
421/* {{{ xml_parser_dtor() */
422static void xml_parser_dtor(zend_rsrc_list_entry *rsrc TSRMLS_DC)
423{
424 xml_parser *parser = (xml_parser *)rsrc->ptr;
425
426 if (parser->parser) {
427 XML_ParserFree(parser->parser);
428 }
429 if (parser->ltags) {
430 int inx;
431 for (inx = 0; ((inx < parser->level) && (inx < XML_MAXLEVEL)); inx++)
432 efree(parser->ltags[ inx ]);
433 efree(parser->ltags);
434 }
435 if (parser->startElementHandler) {
436 zval_ptr_dtor(&parser->startElementHandler);
437 }
438 if (parser->endElementHandler) {
439 zval_ptr_dtor(&parser->endElementHandler);
440 }
441 if (parser->characterDataHandler) {
442 zval_ptr_dtor(&parser->characterDataHandler);
443 }
444 if (parser->processingInstructionHandler) {
445 zval_ptr_dtor(&parser->processingInstructionHandler);
446 }
447 if (parser->defaultHandler) {
448 zval_ptr_dtor(&parser->defaultHandler);
449 }
450 if (parser->unparsedEntityDeclHandler) {
451 zval_ptr_dtor(&parser->unparsedEntityDeclHandler);
452 }
453 if (parser->notationDeclHandler) {
454 zval_ptr_dtor(&parser->notationDeclHandler);
455 }
456 if (parser->externalEntityRefHandler) {
457 zval_ptr_dtor(&parser->externalEntityRefHandler);
458 }
459 if (parser->unknownEncodingHandler) {
460 zval_ptr_dtor(&parser->unknownEncodingHandler);
461 }
462 if (parser->startNamespaceDeclHandler) {
463 zval_ptr_dtor(&parser->startNamespaceDeclHandler);
464 }
465 if (parser->endNamespaceDeclHandler) {
466 zval_ptr_dtor(&parser->endNamespaceDeclHandler);
467 }
468 if (parser->baseURI) {
469 efree(parser->baseURI);
470 }
471 if (parser->object) {
472 zval_ptr_dtor(&parser->object);
473 }
474
475 efree(parser);
476}
477/* }}} */
478
479/* {{{ xml_set_handler() */
480static void xml_set_handler(zval **handler, zval **data)
481{
482 /* If we have already a handler, release it */
483 if (*handler) {
484 zval_ptr_dtor(handler);
485 }
486
487 /* IS_ARRAY might indicate that we're using array($obj, 'method') syntax */
488 if (Z_TYPE_PP(data) != IS_ARRAY && Z_TYPE_PP(data) != IS_OBJECT) {
489
490 convert_to_string_ex(data);
491 if (Z_STRLEN_PP(data) == 0) {
492 *handler = NULL;
493 return;
494 }
495 }
496
497 zval_add_ref(data);
498
499 *handler = *data;
500}
501/* }}} */
502
503/* {{{ xml_call_handler() */
504static zval *xml_call_handler(xml_parser *parser, zval *handler, zend_function *function_ptr, int argc, zval **argv)
505{
506 int i;
507 TSRMLS_FETCH();
508
509 if (parser && handler && !EG(exception)) {
510 zval ***args;
511 zval *retval;
512 int result;
513 zend_fcall_info fci;
514
515 args = safe_emalloc(sizeof(zval **), argc, 0);
516 for (i = 0; i < argc; i++) {
517 args[i] = &argv[i];
518 }
519
520 fci.size = sizeof(fci);
521 fci.function_table = EG(function_table);
522 fci.function_name = handler;
523 fci.symbol_table = NULL;
524 fci.object_ptr = parser->object;
525 fci.retval_ptr_ptr = &retval;
526 fci.param_count = argc;
527 fci.params = args;
528 fci.no_separation = 0;
529 /*fci.function_handler_cache = &function_ptr;*/
530
531 result = zend_call_function(&fci, NULL TSRMLS_CC);
532 if (result == FAILURE) {
533 zval **method;
534 zval **obj;
535
536 if (Z_TYPE_P(handler) == IS_STRING) {
537 php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unable to call handler %s()", Z_STRVAL_P(handler));
538 } else if (zend_hash_index_find(Z_ARRVAL_P(handler), 0, (void **) &obj) == SUCCESS &&
539 zend_hash_index_find(Z_ARRVAL_P(handler), 1, (void **) &method) == SUCCESS &&
540 Z_TYPE_PP(obj) == IS_OBJECT &&
541 Z_TYPE_PP(method) == IS_STRING) {
542 php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unable to call handler %s::%s()", Z_OBJCE_PP(obj)->name, Z_STRVAL_PP(method));
543 } else
544 php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unable to call handler");
545 }
546
547 for (i = 0; i < argc; i++) {
548 zval_ptr_dtor(args[i]);
549 }
550 efree(args);
551
552 if (result == FAILURE) {
553 return NULL;
554 } else {
555 return EG(exception) ? NULL : retval;
556 }
557 } else {
558 for (i = 0; i < argc; i++) {
559 zval_ptr_dtor(&argv[i]);
560 }
561 return NULL;
562 }
563}
564/* }}} */
565
566/* {{{ xml_encode_iso_8859_1() */
567inline static unsigned short xml_encode_iso_8859_1(unsigned char c)
568{
569 return (unsigned short)c;
570}
571/* }}} */
572
573/* {{{ xml_decode_iso_8859_1() */
574inline static char xml_decode_iso_8859_1(unsigned short c)
575{
576 return (char)(c > 0xff ? '?' : c);
577}
578/* }}} */
579
580/* {{{ xml_encode_us_ascii() */
581inline static unsigned short xml_encode_us_ascii(unsigned char c)
582{
583 return (unsigned short)c;
584}
585/* }}} */
586
587/* {{{ xml_decode_us_ascii() */
588inline static char xml_decode_us_ascii(unsigned short c)
589{
590 return (char)(c > 0x7f ? '?' : c);
591}
592/* }}} */
593
594/* {{{ xml_get_encoding() */
595static xml_encoding *xml_get_encoding(const XML_Char *name)
596{
597 xml_encoding *enc = &xml_encodings[0];
598
599 while (enc && enc->name) {
600 if (strcasecmp(name, enc->name) == 0) {
601 return enc;
602 }
603 enc++;
604 }
605 return NULL;
606}
607/* }}} */
608
609/* {{{ xml_utf8_encode */
610PHPAPI char *xml_utf8_encode(const char *s, int len, int *newlen, const XML_Char *encoding)
611{
612 int pos = len;
613 char *newbuf;
614 unsigned int c;
615 unsigned short (*encoder)(unsigned char) = NULL;
616 xml_encoding *enc = xml_get_encoding(encoding);
617
618 *newlen = 0;
619 if (enc) {
620 encoder = enc->encoding_function;
621 } else {
622 /* If the target encoding was unknown, fail */
623 return NULL;
624 }
625 if (encoder == NULL) {
626 /* If no encoder function was specified, return the data as-is.
627 */
628 newbuf = emalloc(len + 1);
629 memcpy(newbuf, s, len);
630 *newlen = len;
631 newbuf[*newlen] = '\0';
632 return newbuf;
633 }
634 /* This is the theoretical max (will never get beyond len * 2 as long
635 * as we are converting from single-byte characters, though) */
636 newbuf = safe_emalloc(len, 4, 1);
637 while (pos > 0) {
638 c = encoder ? encoder((unsigned char)(*s)) : (unsigned short)(*s);
639 if (c < 0x80) {
640 newbuf[(*newlen)++] = (char) c;
641 } else if (c < 0x800) {
642 newbuf[(*newlen)++] = (0xc0 | (c >> 6));
643 newbuf[(*newlen)++] = (0x80 | (c & 0x3f));
644 } else if (c < 0x10000) {
645 newbuf[(*newlen)++] = (0xe0 | (c >> 12));
646 newbuf[(*newlen)++] = (0xc0 | ((c >> 6) & 0x3f));
647 newbuf[(*newlen)++] = (0x80 | (c & 0x3f));
648 } else if (c < 0x200000) {
649 newbuf[(*newlen)++] = (0xf0 | (c >> 18));
650 newbuf[(*newlen)++] = (0xe0 | ((c >> 12) & 0x3f));
651 newbuf[(*newlen)++] = (0xc0 | ((c >> 6) & 0x3f));
652 newbuf[(*newlen)++] = (0x80 | (c & 0x3f));
653 }
654 pos--;
655 s++;
656 }
657 newbuf[*newlen] = 0;
658 newbuf = erealloc(newbuf, (*newlen)+1);
659 return newbuf;
660}
661/* }}} */
662
663/* {{{ xml_utf8_decode */
664PHPAPI char *xml_utf8_decode(const XML_Char *s, int len, int *newlen, const XML_Char *encoding)
665{
666 size_t pos = 0;
667 char *newbuf = emalloc(len + 1);
668 unsigned int c;
669 char (*decoder)(unsigned short) = NULL;
670 xml_encoding *enc = xml_get_encoding(encoding);
671
672 *newlen = 0;
673 if (enc) {
674 decoder = enc->decoding_function;
675 }
676 if (decoder == NULL) {
677 /* If the target encoding was unknown, or no decoder function
678 * was specified, return the UTF-8-encoded data as-is.
679 */
680 memcpy(newbuf, s, len);
681 *newlen = len;
682 newbuf[*newlen] = '\0';
683 return newbuf;
684 }
685
686 while (pos < (size_t)len) {
687 int status = FAILURE;
688 c = php_next_utf8_char((const unsigned char*)s, (size_t) len, &pos, &status);
689
690 if (status == FAILURE || c > 0xFFU) {
691 c = '?';
692 }
693
694 newbuf[*newlen] = decoder ? decoder(c) : c;
695 ++*newlen;
696 }
697 if (*newlen < len) {
698 newbuf = erealloc(newbuf, *newlen + 1);
699 }
700 newbuf[*newlen] = '\0';
701 return newbuf;
702}
703/* }}} */
704
705/* {{{ _xml_xmlcharlen() */
706static int _xml_xmlcharlen(const XML_Char *s)
707{
708 int len = 0;
709
710 while (*s) {
711 len++;
712 s++;
713 }
714 return len;
715}
716/* }}} */
717
718/* {{{ _xml_zval_strdup() */
719PHPAPI char *_xml_zval_strdup(zval *val)
720{
721 if (Z_TYPE_P(val) == IS_STRING) {
722 char *buf = emalloc(Z_STRLEN_P(val) + 1);
723 memcpy(buf, Z_STRVAL_P(val), Z_STRLEN_P(val));
724 buf[Z_STRLEN_P(val)] = '\0';
725 return buf;
726 }
727 return NULL;
728}
729/* }}} */
730
731/* {{{ _xml_add_to_info */
732static void _xml_add_to_info(xml_parser *parser,char *name)
733{
734 zval **element, *values;
735
736 if (! parser->info) {
737 return;
738 }
739
740 if (zend_hash_find(Z_ARRVAL_P(parser->info),name,strlen(name) + 1,(void **) &element) == FAILURE) {
741 MAKE_STD_ZVAL(values);
742
743 array_init(values);
744
745 zend_hash_update(Z_ARRVAL_P(parser->info), name, strlen(name)+1, (void *) &values, sizeof(zval*), (void **) &element);
746 }
747
748 add_next_index_long(*element,parser->curtag);
749
750 parser->curtag++;
751}
752/* }}} */
753
754/* {{{ _xml_decode_tag() */
755static char *_xml_decode_tag(xml_parser *parser, const char *tag)
756{
757 char *newstr;
758 int out_len;
759
760 newstr = xml_utf8_decode(tag, strlen(tag), &out_len, parser->target_encoding);
761
762 if (parser->case_folding) {
763 php_strtoupper(newstr, out_len);
764 }
765
766 return newstr;
767}
768/* }}} */
769
770/* {{{ _xml_startElementHandler() */
771void _xml_startElementHandler(void *userData, const XML_Char *name, const XML_Char **attributes)
772{
773 xml_parser *parser = (xml_parser *)userData;
774 const char **attrs = (const char **) attributes;
775 char *tag_name;
776 char *att, *val;
777 int val_len;
778 zval *retval, *args[3];
779
780 if (parser) {
781 parser->level++;
782
783 tag_name = _xml_decode_tag(parser, name);
784
785 if (parser->startElementHandler) {
786 args[0] = _xml_resource_zval(parser->index);
787 args[1] = _xml_string_zval(((char *) tag_name) + parser->toffset);
788 MAKE_STD_ZVAL(args[2]);
789 array_init(args[2]);
790
791 while (attributes && *attributes) {
792 att = _xml_decode_tag(parser, attributes[0]);
793 val = xml_utf8_decode(attributes[1], strlen(attributes[1]), &val_len, parser->target_encoding);
794
795 add_assoc_stringl(args[2], att, val, val_len, 0);
796
797 attributes += 2;
798
799 efree(att);
800 }
801
802 if ((retval = xml_call_handler(parser, parser->startElementHandler, parser->startElementPtr, 3, args))) {
803 zval_ptr_dtor(&retval);
804 }
805 }
806
807 if (parser->data) {
808 if (parser->level <= XML_MAXLEVEL) {
809 zval *tag, *atr;
810 int atcnt = 0;
811
812 MAKE_STD_ZVAL(tag);
813 MAKE_STD_ZVAL(atr);
814
815 array_init(tag);
816 array_init(atr);
817
818 _xml_add_to_info(parser,((char *) tag_name) + parser->toffset);
819
820 add_assoc_string(tag,"tag",((char *) tag_name) + parser->toffset,1); /* cast to avoid gcc-warning */
821 add_assoc_string(tag,"type","open",1);
822 add_assoc_long(tag,"level",parser->level);
823
824 parser->ltags[parser->level-1] = estrdup(tag_name);
825 parser->lastwasopen = 1;
826
827 attributes = (const XML_Char **) attrs;
828
829 while (attributes && *attributes) {
830 att = _xml_decode_tag(parser, attributes[0]);
831 val = xml_utf8_decode(attributes[1], strlen(attributes[1]), &val_len, parser->target_encoding);
832
833 add_assoc_stringl(atr,att,val,val_len,0);
834
835 atcnt++;
836 attributes += 2;
837
838 efree(att);
839 }
840
841 if (atcnt) {
842 zend_hash_add(Z_ARRVAL_P(tag),"attributes",sizeof("attributes"),&atr,sizeof(zval*),NULL);
843 } else {
844 zval_ptr_dtor(&atr);
845 }
846
847 zend_hash_next_index_insert(Z_ARRVAL_P(parser->data),&tag,sizeof(zval*),(void *) &parser->ctag);
848 } else if (parser->level == (XML_MAXLEVEL + 1)) {
849 TSRMLS_FETCH();
850 php_error_docref(NULL TSRMLS_CC, E_WARNING, "Maximum depth exceeded - Results truncated");
851 }
852 }
853
854 efree(tag_name);
855 }
856}
857/* }}} */
858
859/* {{{ _xml_endElementHandler() */
860void _xml_endElementHandler(void *userData, const XML_Char *name)
861{
862 xml_parser *parser = (xml_parser *)userData;
863 char *tag_name;
864
865 if (parser) {
866 zval *retval, *args[2];
867
868 tag_name = _xml_decode_tag(parser, name);
869
870 if (parser->endElementHandler) {
871 args[0] = _xml_resource_zval(parser->index);
872 args[1] = _xml_string_zval(((char *) tag_name) + parser->toffset);
873
874 if ((retval = xml_call_handler(parser, parser->endElementHandler, parser->endElementPtr, 2, args))) {
875 zval_ptr_dtor(&retval);
876 }
877 }
878
879 if (parser->data) {
880 zval *tag;
881
882 if (parser->lastwasopen) {
883 add_assoc_string(*(parser->ctag),"type","complete",1);
884 } else {
885 MAKE_STD_ZVAL(tag);
886
887 array_init(tag);
888
889 _xml_add_to_info(parser,((char *) tag_name) + parser->toffset);
890
891 add_assoc_string(tag,"tag",((char *) tag_name) + parser->toffset,1); /* cast to avoid gcc-warning */
892 add_assoc_string(tag,"type","close",1);
893 add_assoc_long(tag,"level",parser->level);
894
895 zend_hash_next_index_insert(Z_ARRVAL_P(parser->data),&tag,sizeof(zval*),NULL);
896 }
897
898 parser->lastwasopen = 0;
899 }
900
901 efree(tag_name);
902
903 if ((parser->ltags) && (parser->level <= XML_MAXLEVEL)) {
904 efree(parser->ltags[parser->level-1]);
905 }
906
907 parser->level--;
908 }
909}
910/* }}} */
911
912/* {{{ _xml_characterDataHandler() */
913void _xml_characterDataHandler(void *userData, const XML_Char *s, int len)
914{
915 xml_parser *parser = (xml_parser *)userData;
916
917 if (parser) {
918 zval *retval, *args[2];
919
920 if (parser->characterDataHandler) {
921 args[0] = _xml_resource_zval(parser->index);
922 args[1] = _xml_xmlchar_zval(s, len, parser->target_encoding);
923 if ((retval = xml_call_handler(parser, parser->characterDataHandler, parser->characterDataPtr, 2, args))) {
924 zval_ptr_dtor(&retval);
925 }
926 }
927
928 if (parser->data) {
929 int i;
930 int doprint = 0;
931
932 char *decoded_value;
933 int decoded_len;
934
935 decoded_value = xml_utf8_decode(s,len,&decoded_len,parser->target_encoding);
936 for (i = 0; i < decoded_len; i++) {
937 switch (decoded_value[i]) {
938 case ' ':
939 case '\t':
940 case '\n':
941 continue;
942 default:
943 doprint = 1;
944 break;
945 }
946 if (doprint) {
947 break;
948 }
949 }
950 if (doprint || (! parser->skipwhite)) {
951 if (parser->lastwasopen) {
952 zval **myval;
953
954 /* check if the current tag already has a value - if yes append to that! */
955 if (zend_hash_find(Z_ARRVAL_PP(parser->ctag),"value",sizeof("value"),(void **) &myval) == SUCCESS) {
956 int newlen = Z_STRLEN_PP(myval) + decoded_len;
957 Z_STRVAL_PP(myval) = erealloc(Z_STRVAL_PP(myval),newlen+1);
958 strncpy(Z_STRVAL_PP(myval) + Z_STRLEN_PP(myval), decoded_value, decoded_len + 1);
959 Z_STRLEN_PP(myval) += decoded_len;
960 efree(decoded_value);
961 } else {
962 add_assoc_string(*(parser->ctag),"value",decoded_value,0);
963 }
964
965 } else {
966 zval *tag;
967 zval **curtag, **mytype, **myval;
968 HashPosition hpos=NULL;
969
970 zend_hash_internal_pointer_end_ex(Z_ARRVAL_P(parser->data), &hpos);
971
972 if (hpos && (zend_hash_get_current_data_ex(Z_ARRVAL_P(parser->data), (void **) &curtag, &hpos) == SUCCESS)) {
973 if (zend_hash_find(Z_ARRVAL_PP(curtag),"type",sizeof("type"),(void **) &mytype) == SUCCESS) {
974 if (!strcmp(Z_STRVAL_PP(mytype), "cdata")) {
975 if (zend_hash_find(Z_ARRVAL_PP(curtag),"value",sizeof("value"),(void **) &myval) == SUCCESS) {
976 int newlen = Z_STRLEN_PP(myval) + decoded_len;
977 Z_STRVAL_PP(myval) = erealloc(Z_STRVAL_PP(myval),newlen+1);
978 strncpy(Z_STRVAL_PP(myval) + Z_STRLEN_PP(myval), decoded_value, decoded_len + 1);
979 Z_STRLEN_PP(myval) += decoded_len;
980 efree(decoded_value);
981 return;
982 }
983 }
984 }
985 }
986
987 if (parser->level <= XML_MAXLEVEL) {
988 MAKE_STD_ZVAL(tag);
989
990 array_init(tag);
991
992 _xml_add_to_info(parser,parser->ltags[parser->level-1] + parser->toffset);
993
994 add_assoc_string(tag,"tag",parser->ltags[parser->level-1] + parser->toffset,1);
995 add_assoc_string(tag,"value",decoded_value,0);
996 add_assoc_string(tag,"type","cdata",1);
997 add_assoc_long(tag,"level",parser->level);
998
999 zend_hash_next_index_insert(Z_ARRVAL_P(parser->data),&tag,sizeof(zval*),NULL);
1000 } else if (parser->level == (XML_MAXLEVEL + 1)) {
1001 TSRMLS_FETCH();
1002 php_error_docref(NULL TSRMLS_CC, E_WARNING, "Maximum depth exceeded - Results truncated");
1003 }
1004 }
1005 } else {
1006 efree(decoded_value);
1007 }
1008 }
1009 }
1010}
1011/* }}} */
1012
1013/* {{{ _xml_processingInstructionHandler() */
1014void _xml_processingInstructionHandler(void *userData, const XML_Char *target, const XML_Char *data)
1015{
1016 xml_parser *parser = (xml_parser *)userData;
1017
1018 if (parser && parser->processingInstructionHandler) {
1019 zval *retval, *args[3];
1020
1021 args[0] = _xml_resource_zval(parser->index);
1022 args[1] = _xml_xmlchar_zval(target, 0, parser->target_encoding);
1023 args[2] = _xml_xmlchar_zval(data, 0, parser->target_encoding);
1024 if ((retval = xml_call_handler(parser, parser->processingInstructionHandler, parser->processingInstructionPtr, 3, args))) {
1025 zval_ptr_dtor(&retval);
1026 }
1027 }
1028}
1029/* }}} */
1030
1031/* {{{ _xml_defaultHandler() */
1032void _xml_defaultHandler(void *userData, const XML_Char *s, int len)
1033{
1034 xml_parser *parser = (xml_parser *)userData;
1035
1036 if (parser && parser->defaultHandler) {
1037 zval *retval, *args[2];
1038
1039 args[0] = _xml_resource_zval(parser->index);
1040 args[1] = _xml_xmlchar_zval(s, len, parser->target_encoding);
1041 if ((retval = xml_call_handler(parser, parser->defaultHandler, parser->defaultPtr, 2, args))) {
1042 zval_ptr_dtor(&retval);
1043 }
1044 }
1045}
1046/* }}} */
1047
1048/* {{{ _xml_unparsedEntityDeclHandler() */
1049void _xml_unparsedEntityDeclHandler(void *userData,
1050 const XML_Char *entityName,
1051 const XML_Char *base,
1052 const XML_Char *systemId,
1053 const XML_Char *publicId,
1054 const XML_Char *notationName)
1055{
1056 xml_parser *parser = (xml_parser *)userData;
1057
1058 if (parser && parser->unparsedEntityDeclHandler) {
1059 zval *retval, *args[6];
1060
1061 args[0] = _xml_resource_zval(parser->index);
1062 args[1] = _xml_xmlchar_zval(entityName, 0, parser->target_encoding);
1063 args[2] = _xml_xmlchar_zval(base, 0, parser->target_encoding);
1064 args[3] = _xml_xmlchar_zval(systemId, 0, parser->target_encoding);
1065 args[4] = _xml_xmlchar_zval(publicId, 0, parser->target_encoding);
1066 args[5] = _xml_xmlchar_zval(notationName, 0, parser->target_encoding);
1067 if ((retval = xml_call_handler(parser, parser->unparsedEntityDeclHandler, parser->unparsedEntityDeclPtr, 6, args))) {
1068 zval_ptr_dtor(&retval);
1069 }
1070 }
1071}
1072/* }}} */
1073
1074/* {{{ _xml_notationDeclHandler() */
1075void _xml_notationDeclHandler(void *userData,
1076 const XML_Char *notationName,
1077 const XML_Char *base,
1078 const XML_Char *systemId,
1079 const XML_Char *publicId)
1080{
1081 xml_parser *parser = (xml_parser *)userData;
1082
1083 if (parser && parser->notationDeclHandler) {
1084 zval *retval, *args[5];
1085
1086 args[0] = _xml_resource_zval(parser->index);
1087 args[1] = _xml_xmlchar_zval(notationName, 0, parser->target_encoding);
1088 args[2] = _xml_xmlchar_zval(base, 0, parser->target_encoding);
1089 args[3] = _xml_xmlchar_zval(systemId, 0, parser->target_encoding);
1090 args[4] = _xml_xmlchar_zval(publicId, 0, parser->target_encoding);
1091 if ((retval = xml_call_handler(parser, parser->notationDeclHandler, parser->notationDeclPtr, 5, args))) {
1092 zval_ptr_dtor(&retval);
1093 }
1094 }
1095}
1096/* }}} */
1097
1098/* {{{ _xml_externalEntityRefHandler() */
1099int _xml_externalEntityRefHandler(XML_Parser parserPtr,
1100 const XML_Char *openEntityNames,
1101 const XML_Char *base,
1102 const XML_Char *systemId,
1103 const XML_Char *publicId)
1104{
1105 xml_parser *parser = XML_GetUserData(parserPtr);
1106 int ret = 0; /* abort if no handler is set (should be configurable?) */
1107
1108 if (parser && parser->externalEntityRefHandler) {
1109 zval *retval, *args[5];
1110
1111 args[0] = _xml_resource_zval(parser->index);
1112 args[1] = _xml_xmlchar_zval(openEntityNames, 0, parser->target_encoding);
1113 args[2] = _xml_xmlchar_zval(base, 0, parser->target_encoding);
1114 args[3] = _xml_xmlchar_zval(systemId, 0, parser->target_encoding);
1115 args[4] = _xml_xmlchar_zval(publicId, 0, parser->target_encoding);
1116 if ((retval = xml_call_handler(parser, parser->externalEntityRefHandler, parser->externalEntityRefPtr, 5, args))) {
1117 convert_to_long(retval);
1118 ret = Z_LVAL_P(retval);
1119 efree(retval);
1120 } else {
1121 ret = 0;
1122 }
1123 }
1124 return ret;
1125}
1126/* }}} */
1127
1128/* {{{ _xml_startNamespaceDeclHandler() */
1129void _xml_startNamespaceDeclHandler(void *userData,const XML_Char *prefix, const XML_Char *uri)
1130{
1131 xml_parser *parser = (xml_parser *)userData;
1132
1133 if (parser && parser->startNamespaceDeclHandler) {
1134 zval *retval, *args[3];
1135
1136 args[0] = _xml_resource_zval(parser->index);
1137 args[1] = _xml_xmlchar_zval(prefix, 0, parser->target_encoding);
1138 args[2] = _xml_xmlchar_zval(uri, 0, parser->target_encoding);
1139 if ((retval = xml_call_handler(parser, parser->startNamespaceDeclHandler, parser->startNamespaceDeclPtr, 3, args))) {
1140 zval_ptr_dtor(&retval);
1141 }
1142 }
1143}
1144/* }}} */
1145
1146/* {{{ _xml_endNamespaceDeclHandler() */
1147void _xml_endNamespaceDeclHandler(void *userData, const XML_Char *prefix)
1148{
1149 xml_parser *parser = (xml_parser *)userData;
1150
1151 if (parser && parser->endNamespaceDeclHandler) {
1152 zval *retval, *args[2];
1153
1154 args[0] = _xml_resource_zval(parser->index);
1155 args[1] = _xml_xmlchar_zval(prefix, 0, parser->target_encoding);
1156 if ((retval = xml_call_handler(parser, parser->endNamespaceDeclHandler, parser->endNamespaceDeclPtr, 2, args))) {
1157 zval_ptr_dtor(&retval);
1158 }
1159 }
1160}
1161/* }}} */
1162
1163/************************* EXTENSION FUNCTIONS *************************/
1164
1165static void php_xml_parser_create_impl(INTERNAL_FUNCTION_PARAMETERS, int ns_support) /* {{{ */
1166{
1167 xml_parser *parser;
1168 int auto_detect = 0;
1169
1170 char *encoding_param = NULL;
1171 int encoding_param_len = 0;
1172
1173 char *ns_param = NULL;
1174 int ns_param_len = 0;
1175
1176 XML_Char *encoding;
1177
1178 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, (ns_support ? "|ss": "|s"), &encoding_param, &encoding_param_len, &ns_param, &ns_param_len) == FAILURE) {
1179 RETURN_FALSE;
1180 }
1181
1182 if (encoding_param != NULL) {
1183 /* The supported encoding types are hardcoded here because
1184 * we are limited to the encodings supported by expat/xmltok.
1185 */
1186 if (encoding_param_len == 0) {
1187 encoding = XML(default_encoding);
1188 auto_detect = 1;
1189 } else if (strcasecmp(encoding_param, "ISO-8859-1") == 0) {
1190 encoding = "ISO-8859-1";
1191 } else if (strcasecmp(encoding_param, "UTF-8") == 0) {
1192 encoding = "UTF-8";
1193 } else if (strcasecmp(encoding_param, "US-ASCII") == 0) {
1194 encoding = "US-ASCII";
1195 } else {
1196 php_error_docref(NULL TSRMLS_CC, E_WARNING, "unsupported source encoding \"%s\"", encoding_param);
1197 RETURN_FALSE;
1198 }
1199 } else {
1200 encoding = XML(default_encoding);
1201 }
1202
1203 if (ns_support && ns_param == NULL){
1204 ns_param = ":";
1205 }
1206
1207 parser = ecalloc(1, sizeof(xml_parser));
1208 parser->parser = XML_ParserCreate_MM((auto_detect ? NULL : encoding),
1209 &php_xml_mem_hdlrs, ns_param);
1210
1211 parser->target_encoding = encoding;
1212 parser->case_folding = 1;
1213 parser->object = NULL;
1214 parser->isparsing = 0;
1215
1216 XML_SetUserData(parser->parser, parser);
1217
1218 ZEND_REGISTER_RESOURCE(return_value, parser,le_xml_parser);
1219 parser->index = Z_LVAL_P(return_value);
1220}
1221/* }}} */
1222
1223/* {{{ proto resource xml_parser_create([string encoding])
1224 Create an XML parser */
1225PHP_FUNCTION(xml_parser_create)
1226{
1227 php_xml_parser_create_impl(INTERNAL_FUNCTION_PARAM_PASSTHRU, 0);
1228}
1229/* }}} */
1230
1231/* {{{ proto resource xml_parser_create_ns([string encoding [, string sep]])
1232 Create an XML parser */
1233PHP_FUNCTION(xml_parser_create_ns)
1234{
1235 php_xml_parser_create_impl(INTERNAL_FUNCTION_PARAM_PASSTHRU, 1);
1236}
1237/* }}} */
1238
1239/* {{{ proto int xml_set_object(resource parser, object &obj)
1240 Set up object which should be used for callbacks */
1241PHP_FUNCTION(xml_set_object)
1242{
1243 xml_parser *parser;
1244 zval *pind, *mythis;
1245
1246 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ro", &pind, &mythis) == FAILURE) {
1247 return;
1248 }
1249
1250 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1251
1252 /* please leave this commented - or ask thies@thieso.net before doing it (again) */
1253 if (parser->object) {
1254 zval_ptr_dtor(&parser->object);
1255 }
1256
1257 /* please leave this commented - or ask thies@thieso.net before doing it (again) */
1258/* #ifdef ZEND_ENGINE_2
1259 zval_add_ref(&parser->object);
1260#endif */
1261
1262 ALLOC_ZVAL(parser->object);
1263 MAKE_COPY_ZVAL(&mythis, parser->object);
1264
1265 RETVAL_TRUE;
1266}
1267/* }}} */
1268
1269/* {{{ proto int xml_set_element_handler(resource parser, string shdl, string ehdl)
1270 Set up start and end element handlers */
1271PHP_FUNCTION(xml_set_element_handler)
1272{
1273 xml_parser *parser;
1274 zval *pind, **shdl, **ehdl;
1275
1276 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rZZ", &pind, &shdl, &ehdl) == FAILURE) {
1277 return;
1278 }
1279
1280 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1281
1282 xml_set_handler(&parser->startElementHandler, shdl);
1283 xml_set_handler(&parser->endElementHandler, ehdl);
1284 XML_SetElementHandler(parser->parser, _xml_startElementHandler, _xml_endElementHandler);
1285 RETVAL_TRUE;
1286}
1287/* }}} */
1288
1289/* {{{ proto int xml_set_character_data_handler(resource parser, string hdl)
1290 Set up character data handler */
1291PHP_FUNCTION(xml_set_character_data_handler)
1292{
1293 xml_parser *parser;
1294 zval *pind, **hdl;
1295
1296 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rZ", &pind, &hdl) == FAILURE) {
1297 return;
1298 }
1299
1300 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1301
1302 xml_set_handler(&parser->characterDataHandler, hdl);
1303 XML_SetCharacterDataHandler(parser->parser, _xml_characterDataHandler);
1304 RETVAL_TRUE;
1305}
1306/* }}} */
1307
1308/* {{{ proto int xml_set_processing_instruction_handler(resource parser, string hdl)
1309 Set up processing instruction (PI) handler */
1310PHP_FUNCTION(xml_set_processing_instruction_handler)
1311{
1312 xml_parser *parser;
1313 zval *pind, **hdl;
1314
1315 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rZ", &pind, &hdl) == FAILURE) {
1316 return;
1317 }
1318
1319 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1320
1321 xml_set_handler(&parser->processingInstructionHandler, hdl);
1322 XML_SetProcessingInstructionHandler(parser->parser, _xml_processingInstructionHandler);
1323 RETVAL_TRUE;
1324}
1325/* }}} */
1326
1327/* {{{ proto int xml_set_default_handler(resource parser, string hdl)
1328 Set up default handler */
1329PHP_FUNCTION(xml_set_default_handler)
1330{
1331 xml_parser *parser;
1332 zval *pind, **hdl;
1333
1334 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rZ", &pind, &hdl) == FAILURE) {
1335 return;
1336 }
1337 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1338
1339 xml_set_handler(&parser->defaultHandler, hdl);
1340 XML_SetDefaultHandler(parser->parser, _xml_defaultHandler);
1341 RETVAL_TRUE;
1342}
1343/* }}} */
1344
1345/* {{{ proto int xml_set_unparsed_entity_decl_handler(resource parser, string hdl)
1346 Set up unparsed entity declaration handler */
1347PHP_FUNCTION(xml_set_unparsed_entity_decl_handler)
1348{
1349 xml_parser *parser;
1350 zval *pind, **hdl;
1351
1352 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rZ", &pind, &hdl) == FAILURE) {
1353 return;
1354 }
1355
1356 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1357
1358 xml_set_handler(&parser->unparsedEntityDeclHandler, hdl);
1359 XML_SetUnparsedEntityDeclHandler(parser->parser, _xml_unparsedEntityDeclHandler);
1360 RETVAL_TRUE;
1361}
1362/* }}} */
1363
1364/* {{{ proto int xml_set_notation_decl_handler(resource parser, string hdl)
1365 Set up notation declaration handler */
1366PHP_FUNCTION(xml_set_notation_decl_handler)
1367{
1368 xml_parser *parser;
1369 zval *pind, **hdl;
1370
1371 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rZ", &pind, &hdl) == FAILURE) {
1372 return;
1373 }
1374 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1375
1376 xml_set_handler(&parser->notationDeclHandler, hdl);
1377 XML_SetNotationDeclHandler(parser->parser, _xml_notationDeclHandler);
1378 RETVAL_TRUE;
1379}
1380/* }}} */
1381
1382/* {{{ proto int xml_set_external_entity_ref_handler(resource parser, string hdl)
1383 Set up external entity reference handler */
1384PHP_FUNCTION(xml_set_external_entity_ref_handler)
1385{
1386 xml_parser *parser;
1387 zval *pind, **hdl;
1388
1389 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rZ", &pind, &hdl) == FAILURE) {
1390 return;
1391 }
1392 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1393
1394 xml_set_handler(&parser->externalEntityRefHandler, hdl);
1395 XML_SetExternalEntityRefHandler(parser->parser, (void *) _xml_externalEntityRefHandler);
1396 RETVAL_TRUE;
1397}
1398/* }}} */
1399
1400/* {{{ proto int xml_set_start_namespace_decl_handler(resource parser, string hdl)
1401 Set up character data handler */
1402PHP_FUNCTION(xml_set_start_namespace_decl_handler)
1403{
1404 xml_parser *parser;
1405 zval *pind, **hdl;
1406
1407 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rZ", &pind, &hdl) == FAILURE) {
1408 return;
1409 }
1410
1411 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1412
1413 xml_set_handler(&parser->startNamespaceDeclHandler, hdl);
1414 XML_SetStartNamespaceDeclHandler(parser->parser, _xml_startNamespaceDeclHandler);
1415 RETVAL_TRUE;
1416}
1417/* }}} */
1418
1419/* {{{ proto int xml_set_end_namespace_decl_handler(resource parser, string hdl)
1420 Set up character data handler */
1421PHP_FUNCTION(xml_set_end_namespace_decl_handler)
1422{
1423 xml_parser *parser;
1424 zval *pind, **hdl;
1425
1426 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rZ", &pind, &hdl) == FAILURE) {
1427 return;
1428 }
1429
1430 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1431
1432 xml_set_handler(&parser->endNamespaceDeclHandler, hdl);
1433 XML_SetEndNamespaceDeclHandler(parser->parser, _xml_endNamespaceDeclHandler);
1434 RETVAL_TRUE;
1435}
1436/* }}} */
1437
1438/* {{{ proto int xml_parse(resource parser, string data [, int isFinal])
1439 Start parsing an XML document */
1440PHP_FUNCTION(xml_parse)
1441{
1442 xml_parser *parser;
1443 zval *pind;
1444 char *data;
1445 int data_len, ret;
1446 long isFinal = 0;
1447
1448 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rs|l", &pind, &data, &data_len, &isFinal) == FAILURE) {
1449 return;
1450 }
1451 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1452
1453 parser->isparsing = 1;
1454 ret = XML_Parse(parser->parser, data, data_len, isFinal);
1455 parser->isparsing = 0;
1456 RETVAL_LONG(ret);
1457}
1458
1459/* }}} */
1460
1461/* {{{ proto int xml_parse_into_struct(resource parser, string data, array &values [, array &index ])
1462 Parsing a XML document */
1463
1464PHP_FUNCTION(xml_parse_into_struct)
1465{
1466 xml_parser *parser;
1467 zval *pind, **xdata, **info = NULL;
1468 char *data;
1469 int data_len, ret;
1470
1471 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rsZ|Z", &pind, &data, &data_len, &xdata, &info) == FAILURE) {
1472 return;
1473 }
1474
1475 if (info) {
1476 zval_dtor(*info);
1477 array_init(*info);
1478 }
1479
1480 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1481
1482 zval_dtor(*xdata);
1483 array_init(*xdata);
1484
1485 parser->data = *xdata;
1486
1487 if (info) {
1488 parser->info = *info;
1489 }
1490
1491 parser->level = 0;
1492 parser->ltags = safe_emalloc(XML_MAXLEVEL, sizeof(char *), 0);
1493
1494 XML_SetDefaultHandler(parser->parser, _xml_defaultHandler);
1495 XML_SetElementHandler(parser->parser, _xml_startElementHandler, _xml_endElementHandler);
1496 XML_SetCharacterDataHandler(parser->parser, _xml_characterDataHandler);
1497
1498 parser->isparsing = 1;
1499 ret = XML_Parse(parser->parser, data, data_len, 1);
1500 parser->isparsing = 0;
1501
1502 RETVAL_LONG(ret);
1503}
1504/* }}} */
1505
1506/* {{{ proto int xml_get_error_code(resource parser)
1507 Get XML parser error code */
1508PHP_FUNCTION(xml_get_error_code)
1509{
1510 xml_parser *parser;
1511 zval *pind;
1512
1513 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "r", &pind) == FAILURE) {
1514 return;
1515 }
1516 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1517
1518 RETVAL_LONG((long)XML_GetErrorCode(parser->parser));
1519}
1520/* }}} */
1521
1522/* {{{ proto string xml_error_string(int code)
1523 Get XML parser error string */
1524PHP_FUNCTION(xml_error_string)
1525{
1526 long code;
1527 char *str;
1528
1529 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "l", &code) == FAILURE) {
1530 return;
1531 }
1532
1533 str = (char *)XML_ErrorString((int)code);
1534 if (str) {
1535 RETVAL_STRING(str, 1);
1536 }
1537}
1538/* }}} */
1539
1540/* {{{ proto int xml_get_current_line_number(resource parser)
1541 Get current line number for an XML parser */
1542PHP_FUNCTION(xml_get_current_line_number)
1543{
1544 xml_parser *parser;
1545 zval *pind;
1546
1547 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "r", &pind) == FAILURE) {
1548 return;
1549 }
1550 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1551
1552 RETVAL_LONG(XML_GetCurrentLineNumber(parser->parser));
1553}
1554/* }}} */
1555
1556/* {{{ proto int xml_get_current_column_number(resource parser)
1557 Get current column number for an XML parser */
1558PHP_FUNCTION(xml_get_current_column_number)
1559{
1560 xml_parser *parser;
1561 zval *pind;
1562
1563 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "r", &pind) == FAILURE) {
1564 return;
1565 }
1566 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1567
1568 RETVAL_LONG(XML_GetCurrentColumnNumber(parser->parser));
1569}
1570/* }}} */
1571
1572/* {{{ proto int xml_get_current_byte_index(resource parser)
1573 Get current byte index for an XML parser */
1574PHP_FUNCTION(xml_get_current_byte_index)
1575{
1576 xml_parser *parser;
1577 zval *pind;
1578
1579 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "r", &pind) == FAILURE) {
1580 return;
1581 }
1582 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1583
1584 RETVAL_LONG(XML_GetCurrentByteIndex(parser->parser));
1585}
1586/* }}} */
1587
1588/* {{{ proto int xml_parser_free(resource parser)
1589 Free an XML parser */
1590PHP_FUNCTION(xml_parser_free)
1591{
1592 zval *pind;
1593 xml_parser *parser;
1594
1595 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "r", &pind) == FAILURE) {
1596 return;
1597 }
1598
1599 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1600
1601 if (parser->isparsing == 1) {
1602 php_error_docref(NULL TSRMLS_CC, E_WARNING, "Parser cannot be freed while it is parsing.");
1603 RETURN_FALSE;
1604 }
1605
1606 if (zend_list_delete(parser->index) == FAILURE) {
1607 RETURN_FALSE;
1608 }
1609
1610 RETVAL_TRUE;
1611}
1612/* }}} */
1613
1614/* {{{ proto int xml_parser_set_option(resource parser, int option, mixed value)
1615 Set options in an XML parser */
1616PHP_FUNCTION(xml_parser_set_option)
1617{
1618 xml_parser *parser;
1619 zval *pind, **val;
1620 long opt;
1621
1622 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rlZ", &pind, &opt, &val) == FAILURE) {
1623 return;
1624 }
1625 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1626
1627 switch (opt) {
1628 case PHP_XML_OPTION_CASE_FOLDING:
1629 convert_to_long_ex(val);
1630 parser->case_folding = Z_LVAL_PP(val);
1631 break;
1632 case PHP_XML_OPTION_SKIP_TAGSTART:
1633 convert_to_long_ex(val);
1634 parser->toffset = Z_LVAL_PP(val);
1635 break;
1636 case PHP_XML_OPTION_SKIP_WHITE:
1637 convert_to_long_ex(val);
1638 parser->skipwhite = Z_LVAL_PP(val);
1639 break;
1640 case PHP_XML_OPTION_TARGET_ENCODING: {
1641 xml_encoding *enc;
1642 convert_to_string_ex(val);
1643 enc = xml_get_encoding(Z_STRVAL_PP(val));
1644 if (enc == NULL) {
1645 php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unsupported target encoding \"%s\"", Z_STRVAL_PP(val));
1646 RETURN_FALSE;
1647 }
1648 parser->target_encoding = enc->name;
1649 break;
1650 }
1651 default:
1652 php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown option");
1653 RETURN_FALSE;
1654 break;
1655 }
1656 RETVAL_TRUE;
1657}
1658/* }}} */
1659
1660/* {{{ proto int xml_parser_get_option(resource parser, int option)
1661 Get options from an XML parser */
1662PHP_FUNCTION(xml_parser_get_option)
1663{
1664 xml_parser *parser;
1665 zval *pind;
1666 long opt;
1667
1668 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "rl", &pind, &opt) == FAILURE) {
1669 return;
1670 }
1671 ZEND_FETCH_RESOURCE(parser,xml_parser *, &pind, -1, "XML Parser", le_xml_parser);
1672
1673 switch (opt) {
1674 case PHP_XML_OPTION_CASE_FOLDING:
1675 RETURN_LONG(parser->case_folding);
1676 break;
1677 case PHP_XML_OPTION_TARGET_ENCODING:
1678 RETURN_STRING(parser->target_encoding, 1);
1679 break;
1680 default:
1681 php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown option");
1682 RETURN_FALSE;
1683 break;
1684 }
1685
1686 RETVAL_FALSE; /* never reached */
1687}
1688/* }}} */
1689
1690/* {{{ proto string utf8_encode(string data)
1691 Encodes an ISO-8859-1 string to UTF-8 */
1692PHP_FUNCTION(utf8_encode)
1693{
1694 char *arg;
1695 XML_Char *encoded;
1696 int arg_len, len;
1697
1698 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s", &arg, &arg_len) == FAILURE) {
1699 return;
1700 }
1701
1702 encoded = xml_utf8_encode(arg, arg_len, &len, "ISO-8859-1");
1703 if (encoded == NULL) {
1704 RETURN_FALSE;
1705 }
1706 RETVAL_STRINGL(encoded, len, 0);
1707}
1708/* }}} */
1709
1710/* {{{ proto string utf8_decode(string data)
1711 Converts a UTF-8 encoded string to ISO-8859-1 */
1712PHP_FUNCTION(utf8_decode)
1713{
1714 char *arg;
1715 XML_Char *decoded;
1716 int arg_len, len;
1717
1718 if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s", &arg, &arg_len) == FAILURE) {
1719 return;
1720 }
1721
1722 decoded = xml_utf8_decode(arg, arg_len, &len, "ISO-8859-1");
1723 if (decoded == NULL) {
1724 RETURN_FALSE;
1725 }
1726 RETVAL_STRINGL(decoded, len, 0);
1727}
1728/* }}} */
1729
1730#endif
1731
1732/*
1733 * Local variables:
1734 * tab-width: 4
1735 * c-basic-offset: 4
1736 * End:
1737 * vim600: sw=4 ts=4 fdm=marker
1738 * vim<600: sw=4 ts=4
1739 */
1740