xref: /php-src/ext/dom/dom_iterators.c (revision 78ccea4e)
1 /*
2    +----------------------------------------------------------------------+
3    | Copyright (c) The PHP Group                                          |
4    +----------------------------------------------------------------------+
5    | This source file is subject to version 3.01 of the PHP license,      |
6    | that is bundled with this package in the file LICENSE, and is        |
7    | available through the world-wide-web at the following url:           |
8    | https://www.php.net/license/3_01.txt                                 |
9    | If you did not receive a copy of the PHP license and are unable to   |
10    | obtain it through the world-wide-web, please send a note to          |
11    | license@php.net so we can mail you a copy immediately.               |
12    +----------------------------------------------------------------------+
13    | Authors: Christian Stocker <chregu@php.net>                          |
14    |          Rob Richards <rrichards@php.net>                            |
15    +----------------------------------------------------------------------+
16 */
17 
18 #ifdef HAVE_CONFIG_H
19 #include "config.h"
20 #endif
21 
22 #include "php.h"
23 #if defined(HAVE_LIBXML) && defined(HAVE_DOM)
24 #include "php_dom.h"
25 #include "dom_ce.h"
26 
27 typedef struct _nodeIterator nodeIterator;
28 struct _nodeIterator {
29 	int cur;
30 	int index;
31 	xmlNode *node;
32 };
33 
34 /* Function pointer typedef changed in 2.9.8, see https://github.com/GNOME/libxml2/commit/e03f0a199a67017b2f8052354cf732b2b4cae787 */
35 #if LIBXML_VERSION >= 20908
itemHashScanner(void * payload,void * data,const xmlChar * name)36 static void itemHashScanner (void *payload, void *data, const xmlChar *name) /* {{{ */
37 #else
38 static void itemHashScanner (void *payload, void *data, xmlChar *name)
39 #endif
40 {
41 	nodeIterator *priv = data;
42 
43 	if (priv->cur < priv->index) {
44 		priv->cur++;
45 	} else {
46 		if (priv->node == NULL) {
47 			priv->node = payload;
48 		}
49 	}
50 }
51 /* }}} */
52 
create_notation(const xmlChar * name,const xmlChar * ExternalID,const xmlChar * SystemID)53 xmlNodePtr create_notation(const xmlChar *name, const xmlChar *ExternalID, const xmlChar *SystemID) /* {{{ */
54 {
55 	xmlEntityPtr ret = xmlMalloc(sizeof(xmlEntity));
56 	memset(ret, 0, sizeof(xmlEntity));
57 	ret->type = XML_NOTATION_NODE;
58 	ret->name = xmlStrdup(name);
59 	ret->ExternalID = xmlStrdup(ExternalID);
60 	ret->SystemID = xmlStrdup(SystemID);
61 	return (xmlNodePtr) ret;
62 }
63 /* }}} */
64 
php_dom_libxml_hash_iter_ex(xmlHashTable * ht,int index)65 static xmlNode *php_dom_libxml_hash_iter_ex(xmlHashTable *ht, int index)
66 {
67 	int htsize;
68 
69 	if ((htsize = xmlHashSize(ht)) > 0 && index < htsize) {
70 		nodeIterator iter;
71 		iter.cur = 0;
72 		iter.index = index;
73 		iter.node = NULL;
74 		xmlHashScan(ht, itemHashScanner, &iter);
75 		return iter.node;
76 	} else {
77 		return NULL;
78 	}
79 }
80 
php_dom_libxml_hash_iter(dom_nnodemap_object * objmap,int index)81 xmlNode *php_dom_libxml_hash_iter(dom_nnodemap_object *objmap, int index)
82 {
83 	xmlNode *curnode = php_dom_libxml_hash_iter_ex(objmap->ht, index);
84 
85 	if (curnode != NULL && objmap->nodetype != XML_ENTITY_NODE) {
86 		xmlNotation *notation = (xmlNotation *) curnode;
87 		curnode = create_notation(notation->name, notation->PublicID, notation->SystemID);
88 	}
89 
90 	return curnode;
91 }
92 
php_dom_iterator_dtor(zend_object_iterator * iter)93 static void php_dom_iterator_dtor(zend_object_iterator *iter) /* {{{ */
94 {
95 	php_dom_iterator *iterator = (php_dom_iterator *)iter;
96 
97 	zval_ptr_dtor(&iterator->intern.data);
98 	zval_ptr_dtor(&iterator->curobj);
99 }
100 /* }}} */
101 
php_dom_iterator_valid(zend_object_iterator * iter)102 static zend_result php_dom_iterator_valid(zend_object_iterator *iter) /* {{{ */
103 {
104 	php_dom_iterator *iterator = (php_dom_iterator *)iter;
105 
106 	if (Z_TYPE(iterator->curobj) != IS_UNDEF) {
107 		return SUCCESS;
108 	} else {
109 		return FAILURE;
110 	}
111 }
112 /* }}} */
113 
php_dom_iterator_current_data(zend_object_iterator * iter)114 zval *php_dom_iterator_current_data(zend_object_iterator *iter) /* {{{ */
115 {
116 	php_dom_iterator *iterator = (php_dom_iterator *)iter;
117 
118 	return &iterator->curobj;
119 }
120 /* }}} */
121 
php_dom_iterator_current_key(zend_object_iterator * iter,zval * key)122 static void php_dom_iterator_current_key(zend_object_iterator *iter, zval *key) /* {{{ */
123 {
124 	php_dom_iterator *iterator = (php_dom_iterator *)iter;
125 	zval *object = &iterator->intern.data;
126 	zend_class_entry *ce = Z_OBJCE_P(object);
127 
128 	/* Nodelists have the index as a key while named node maps have the name as a key. */
129 	if (instanceof_function(ce, dom_nodelist_class_entry) || instanceof_function(ce, dom_modern_nodelist_class_entry)) {
130 		ZVAL_LONG(key, iter->index);
131 	} else {
132 		dom_object *intern = Z_DOMOBJ_P(&iterator->curobj);
133 
134 		if (intern != NULL && intern->ptr != NULL) {
135 			xmlNodePtr curnode = (xmlNodePtr)((php_libxml_node_ptr *)intern->ptr)->node;
136 			ZVAL_STRINGL(key, (char *) curnode->name, xmlStrlen(curnode->name));
137 		} else {
138 			ZVAL_NULL(key);
139 		}
140 	}
141 }
142 /* }}} */
143 
dom_fetch_first_iteration_item(dom_nnodemap_object * objmap)144 static xmlNodePtr dom_fetch_first_iteration_item(dom_nnodemap_object *objmap)
145 {
146 	xmlNodePtr basep = dom_object_get_node(objmap->baseobj);
147 	if (!basep) {
148 		return NULL;
149 	}
150 	if (objmap->nodetype == XML_ATTRIBUTE_NODE || objmap->nodetype == XML_ELEMENT_NODE) {
151 		if (objmap->nodetype == XML_ATTRIBUTE_NODE) {
152 			return (xmlNodePtr) basep->properties;
153 		} else {
154 			return basep->children;
155 		}
156 	} else {
157 		int curindex = 0;
158 		xmlNodePtr nodep = php_dom_first_child_of_container_node(basep);
159 		return dom_get_elements_by_tag_name_ns_raw(
160 			basep, nodep, objmap->ns, objmap->local, objmap->local_lower, &curindex, 0);
161 	}
162 }
163 
php_dom_iterator_move_forward(zend_object_iterator * iter)164 static void php_dom_iterator_move_forward(zend_object_iterator *iter) /* {{{ */
165 {
166 	xmlNodePtr curnode = NULL;
167 
168 	php_dom_iterator *iterator = (php_dom_iterator *)iter;
169 
170 	zval *object = &iterator->intern.data;
171 	dom_object *nnmap = Z_DOMOBJ_P(object);
172 	dom_nnodemap_object *objmap = nnmap->ptr;
173 
174 	dom_object *intern = Z_DOMOBJ_P(&iterator->curobj);
175 
176 	if (intern != NULL && intern->ptr != NULL) {
177 		if (objmap->nodetype != XML_ENTITY_NODE &&
178 			objmap->nodetype != XML_NOTATION_NODE) {
179 			if (objmap->nodetype == DOM_NODESET) {
180 				HashTable *nodeht = HASH_OF(&objmap->baseobj_zv);
181 				zval *entry;
182 				zend_hash_move_forward_ex(nodeht, &iterator->pos);
183 				if ((entry = zend_hash_get_current_data_ex(nodeht, &iterator->pos))) {
184 					zval_ptr_dtor(&iterator->curobj);
185 					ZVAL_COPY(&iterator->curobj, entry);
186 					return;
187 				}
188 			} else {
189 				if (objmap->nodetype == XML_ATTRIBUTE_NODE ||
190 					objmap->nodetype == XML_ELEMENT_NODE) {
191 
192 					/* Note: keep legacy behaviour for non-spec mode. */
193 					if (php_dom_follow_spec_intern(intern) && php_dom_is_cache_tag_stale_from_doc_ptr(&iterator->cache_tag, intern->document)) {
194 						php_dom_mark_cache_tag_up_to_date_from_doc_ref(&iterator->cache_tag, intern->document);
195 						curnode = dom_fetch_first_iteration_item(objmap);
196 						zend_ulong index = 0;
197 						while (curnode != NULL && index++ < iter->index) {
198 							curnode = curnode->next;
199 						}
200 					} else {
201 						curnode = (xmlNodePtr)((php_libxml_node_ptr *)intern->ptr)->node;
202 						curnode = curnode->next;
203 					}
204 				} else {
205 					/* The collection is live, we nav the tree from the base object if we cannot
206 					 * use the cache to restart from the last point. */
207 					xmlNodePtr basenode = dom_object_get_node(objmap->baseobj);
208 
209 					/* We have a strong reference to the base node via baseobj_zv, this cannot become NULL */
210 					ZEND_ASSERT(basenode != NULL);
211 
212 					int previndex;
213 					if (php_dom_is_cache_tag_stale_from_node(&iterator->cache_tag, basenode)) {
214 						php_dom_mark_cache_tag_up_to_date_from_node(&iterator->cache_tag, basenode);
215 						previndex = 0;
216 						curnode = php_dom_first_child_of_container_node(basenode);
217 					} else {
218 						previndex = iter->index - 1;
219 						curnode = (xmlNodePtr)((php_libxml_node_ptr *)intern->ptr)->node;
220 					}
221 					curnode = dom_get_elements_by_tag_name_ns_raw(
222 						basenode, curnode, objmap->ns, objmap->local, objmap->local_lower, &previndex, iter->index);
223 				}
224 			}
225 		} else {
226 			curnode = php_dom_libxml_hash_iter(objmap, iter->index);
227 		}
228 	}
229 
230 	zval_ptr_dtor(&iterator->curobj);
231 	ZVAL_UNDEF(&iterator->curobj);
232 
233 	if (curnode) {
234 		php_dom_create_object(curnode, &iterator->curobj, objmap->baseobj);
235 	}
236 }
237 /* }}} */
238 
239 static const zend_object_iterator_funcs php_dom_iterator_funcs = {
240 	php_dom_iterator_dtor,
241 	php_dom_iterator_valid,
242 	php_dom_iterator_current_data,
243 	php_dom_iterator_current_key,
244 	php_dom_iterator_move_forward,
245 	NULL,
246 	NULL,
247 	NULL, /* get_gc */
248 };
249 
php_dom_get_iterator(zend_class_entry * ce,zval * object,int by_ref)250 zend_object_iterator *php_dom_get_iterator(zend_class_entry *ce, zval *object, int by_ref) /* {{{ */
251 {
252 	dom_object *intern;
253 	dom_nnodemap_object *objmap;
254 	xmlNodePtr curnode=NULL;
255 	HashTable *nodeht;
256 	zval *entry;
257 	php_dom_iterator *iterator;
258 
259 	if (by_ref) {
260 		zend_throw_error(NULL, "An iterator cannot be used with foreach by reference");
261 		return NULL;
262 	}
263 	iterator = emalloc(sizeof(php_dom_iterator));
264 	zend_iterator_init(&iterator->intern);
265 	iterator->cache_tag.modification_nr = 0;
266 
267 	ZVAL_OBJ_COPY(&iterator->intern.data, Z_OBJ_P(object));
268 	iterator->intern.funcs = &php_dom_iterator_funcs;
269 
270 	ZVAL_UNDEF(&iterator->curobj);
271 
272 	intern = Z_DOMOBJ_P(object);
273 	objmap = (dom_nnodemap_object *)intern->ptr;
274 	if (objmap != NULL) {
275 		if (objmap->nodetype != XML_ENTITY_NODE &&
276 			objmap->nodetype != XML_NOTATION_NODE) {
277 			if (objmap->nodetype == DOM_NODESET) {
278 				nodeht = HASH_OF(&objmap->baseobj_zv);
279 				zend_hash_internal_pointer_reset_ex(nodeht, &iterator->pos);
280 				if ((entry = zend_hash_get_current_data_ex(nodeht, &iterator->pos))) {
281 					ZVAL_COPY(&iterator->curobj, entry);
282 				}
283 			} else {
284 				curnode = dom_fetch_first_iteration_item(objmap);
285 			}
286 		} else {
287 			curnode = php_dom_libxml_hash_iter(objmap, 0);
288 		}
289 	}
290 
291 	if (curnode) {
292 		php_dom_create_object(curnode, &iterator->curobj, objmap->baseobj);
293 	}
294 
295 	return &iterator->intern;
296 }
297 /* }}} */
298 
299 #endif
300