Introduce NamespaceInfo and add getDescendantNamespaces

Author: ndossche

ext/dom/dom_ce.h

@@ -66,5 +66,6 @@ extern PHP_DOM_EXPORT zend_class_entry *dom_xpath_class_entry;
 extern PHP_DOM_EXPORT zend_class_entry *dom_modern_xpath_class_entry;
 #endif
 extern PHP_DOM_EXPORT zend_class_entry *dom_namespace_node_class_entry;
+extern PHP_DOM_EXPORT zend_class_entry *dom_namespace_info_class_entry;
 
 #endif /* DOM_CE_H */

ext/dom/element.c

@@ -1742,6 +1742,59 @@ zend_result dom_modern_element_substituted_node_value_write(dom_object *obj, zva
 	return SUCCESS;
 }
 
+static void dom_element_get_in_scope_namespace_info(php_dom_libxml_ns_mapper *ns_mapper, HashTable *result, xmlNodePtr nodep, dom_object *intern)
+{
+	HashTable prefix_to_ns_table;
+	zend_hash_init(&prefix_to_ns_table, 0, NULL, NULL, false);
+	zend_hash_real_init_mixed(&prefix_to_ns_table);
+
+	/* https://www.w3.org/TR/1999/REC-xpath-19991116/#namespace-nodes */
+	for (const xmlNode *cur = nodep; cur != NULL; cur = cur->parent) {
+		if (cur->type == XML_ELEMENT_NODE) {
+			for (const xmlAttr *attr = cur->properties; attr != NULL; attr = attr->next) {
+				if (attr->ns != NULL && php_dom_ns_is_fast_ex(attr->ns, php_dom_ns_is_xmlns_magic_token)
+					&& attr->children != NULL && attr->children->content != NULL) {
+					const char *prefix = attr->ns->prefix == NULL ? NULL : (const char *) attr->name;
+					const char *key = prefix == NULL ? "" : prefix;
+					xmlNsPtr ns = php_dom_libxml_ns_mapper_get_ns_raw_strings_nullsafe(ns_mapper, prefix, (const char *) attr->children->content);
+					zend_hash_str_add_ptr(&prefix_to_ns_table, key, strlen(key), ns);
+				}
+			}
+		}
+	}
+
+	xmlNsPtr ns;
+	zend_string *prefix;
+	ZEND_HASH_MAP_FOREACH_STR_KEY_PTR(&prefix_to_ns_table, prefix, ns) {
+		if (ZSTR_LEN(prefix) == 0 && (ns == NULL || ns->href == NULL || *ns->href == '\0')) {
+			/* Exception: "the value of the xmlns attribute for the nearest such element is non-empty" */
+			continue;
+		}
+
+		zval zv;
+		object_init_ex(&zv, dom_namespace_info_class_entry);
+		zend_object *obj = Z_OBJ(zv);
+
+		if (ZSTR_LEN(prefix) != 0) {
+			ZVAL_STR_COPY(OBJ_PROP_NUM(obj, 0), prefix);
+		} else {
+			ZVAL_NULL(OBJ_PROP_NUM(obj, 0));
+		}
+
+		if (ns != NULL && ns->href != NULL && *ns->href != '\0') {
+			ZVAL_STRING(OBJ_PROP_NUM(obj, 1), (const char *) ns->href);
+		} else {
+			ZVAL_NULL(OBJ_PROP_NUM(obj, 1));
+		}
+
+		php_dom_create_object(nodep, OBJ_PROP_NUM(obj, 2), intern);
+
+		zend_hash_next_index_insert_new(result, &zv);
+	} ZEND_HASH_FOREACH_END();
+
+	zend_hash_destroy(&prefix_to_ns_table);
+}
+
 PHP_METHOD(DOM_Element, getInScopeNamespaces)
 {
 	zval *id;
@@ -1755,22 +1808,47 @@ PHP_METHOD(DOM_Element, getInScopeNamespaces)
 	DOM_GET_THIS_OBJ(nodep, id, xmlNodePtr, intern);
 
 	php_dom_libxml_ns_mapper *ns_mapper = php_dom_get_ns_mapper(intern);
-	php_dom_in_scope_ns namespaces = php_dom_get_in_scope_ns(ns_mapper, nodep, true);
 
-	array_init_size(return_value, namespaces.count);
+	array_init(return_value);
+	HashTable *result = Z_ARRVAL_P(return_value);
 
-	for (size_t i = 0; i < namespaces.count; i++) {
-		xmlNsPtr ns = namespaces.list[i];
-		if (ns == NULL) {
-			/* This case corresponds to xmlns="" */
-			add_assoc_null(return_value, "");
-		} else {
-			const char *prefix = (const char *) ns->prefix;
-			add_assoc_stringl(return_value, prefix == NULL ? "" : prefix, (const char *) ns->href, xmlStrlen(ns->href));
-		}
+	dom_element_get_in_scope_namespace_info(ns_mapper, result, nodep, intern);
+}
+
+PHP_METHOD(DOM_Element, getDescendantNamespaces)
+{
+	zval *id;
+	xmlNode *nodep;
+	dom_object *intern;
+
+	if (zend_parse_parameters_none() != SUCCESS) {
+		RETURN_THROWS();
 	}
 
-	php_dom_in_scope_ns_destroy(&namespaces);
+	DOM_GET_THIS_OBJ(nodep, id, xmlNodePtr, intern);
+
+	php_dom_libxml_ns_mapper *ns_mapper = php_dom_get_ns_mapper(intern);
+
+	array_init(return_value);
+	HashTable *result = Z_ARRVAL_P(return_value);
+
+	dom_element_get_in_scope_namespace_info(ns_mapper, result, nodep, intern);
+
+	xmlNodePtr cur = nodep->children;
+	while (cur != NULL) {
+		if (cur->type == XML_ELEMENT_NODE) {
+			/* TODO: this could be more optimized by updating the same HashTable repeatedly
+			 * instead of recreating it on every node. */
+			dom_element_get_in_scope_namespace_info(ns_mapper, result, cur, intern);
+
+			if (cur->children) {
+				cur = cur->children;
+				continue;
+			}
+		}
+
+		cur = php_dom_next_in_tree_order(cur, nodep);
+	}
 }
 
 PHP_METHOD(DOM_Element, rename)

ext/dom/namespace_compat.c

@@ -455,17 +455,18 @@ PHP_DOM_EXPORT void php_dom_libxml_reconcile_modern(php_dom_libxml_ns_mapper *ns
 	zend_hash_destroy(&ctx.old_ns_to_new_ns_ptr);
 }
 
-PHP_DOM_EXPORT php_dom_in_scope_ns php_dom_get_in_scope_ns(php_dom_libxml_ns_mapper *ns_mapper, const xmlNode *node, bool register_element_default_ns)
+PHP_DOM_EXPORT php_dom_in_scope_ns php_dom_get_in_scope_ns(php_dom_libxml_ns_mapper *ns_mapper, const xmlNode *node)
 {
 	ZEND_ASSERT(node != NULL);
 
-	php_dom_in_scope_ns in_scope_ns;
-	in_scope_ns.origin_is_ns_compat = true;
-
 	/* libxml fetches all nsDef items from bottom to top - left to right, ignoring prefixes already in the list.
 	 * We don't have nsDef, but we can use the ns pointer (as that is necessarily in scope),
 	 * and check the xmlns attributes. */
 	HashTable tmp_prefix_to_ns_table;
+
+	/* libxml fetches all nsDef items from bottom to top - left to right, ignoring prefixes already in the list.
+	 * We don't have nsDef, but we can use the ns pointer (as that is necessarily in scope),
+	 * and check the xmlns attributes. */
 	zend_hash_init(&tmp_prefix_to_ns_table, 0, NULL, NULL, false);
 	zend_hash_real_init_mixed(&tmp_prefix_to_ns_table);
 
@@ -474,15 +475,9 @@ PHP_DOM_EXPORT php_dom_in_scope_ns php_dom_get_in_scope_ns(php_dom_libxml_ns_map
 			/* Register namespace of element */
 			if (cur->ns != NULL) {
 				const char *prefix = (const char *) cur->ns->prefix;
-				if (register_element_default_ns || prefix != NULL) {
-					if (prefix == NULL) {
-						zend_hash_str_add_ptr(&tmp_prefix_to_ns_table, "", 0, cur->ns);
-					} else {
-						zend_hash_str_add_ptr(&tmp_prefix_to_ns_table, prefix, strlen(prefix), cur->ns);
-					}
+				if (prefix != NULL) {
+					zend_hash_str_add_ptr(&tmp_prefix_to_ns_table, prefix, strlen(prefix), cur->ns);
 				}
-			} else if (register_element_default_ns) {
-				zend_hash_str_add_ptr(&tmp_prefix_to_ns_table, "", 0, NULL);
 			}
 
 			/* Register xmlns attributes */
@@ -499,6 +494,8 @@ PHP_DOM_EXPORT php_dom_in_scope_ns php_dom_get_in_scope_ns(php_dom_libxml_ns_map
 		}
 	}
 
+	php_dom_in_scope_ns in_scope_ns;
+	in_scope_ns.origin_is_ns_compat = true;
 	in_scope_ns.count = zend_hash_num_elements(&tmp_prefix_to_ns_table);
 	in_scope_ns.list = safe_emalloc(in_scope_ns.count, sizeof(xmlNsPtr), 0);
 

ext/dom/namespace_compat.h

@@ -70,7 +70,7 @@ typedef struct _php_dom_in_scope_ns {
    bool origin_is_ns_compat;
 } php_dom_in_scope_ns;
 
-PHP_DOM_EXPORT php_dom_in_scope_ns php_dom_get_in_scope_ns(php_dom_libxml_ns_mapper *ns_mapper, const xmlNode *node, bool register_element_default_ns);
+PHP_DOM_EXPORT php_dom_in_scope_ns php_dom_get_in_scope_ns(php_dom_libxml_ns_mapper *ns_mapper, const xmlNode *node);
 PHP_DOM_EXPORT php_dom_in_scope_ns php_dom_get_in_scope_ns_legacy(const xmlNode *node);
 PHP_DOM_EXPORT void php_dom_in_scope_ns_destroy(php_dom_in_scope_ns *in_scope_ns);
 

ext/dom/php_dom.c

@@ -82,6 +82,7 @@ PHP_DOM_EXPORT zend_class_entry *dom_xpath_class_entry;
 PHP_DOM_EXPORT zend_class_entry *dom_modern_xpath_class_entry;
 #endif
 PHP_DOM_EXPORT zend_class_entry *dom_namespace_node_class_entry;
+PHP_DOM_EXPORT zend_class_entry *dom_namespace_info_class_entry;
 /* }}} */
 
 static zend_object_handlers dom_object_handlers;
@@ -802,6 +803,8 @@ PHP_MINIT_FUNCTION(dom)
 	DOM_REGISTER_PROP_HANDLER(&dom_namespace_node_prop_handlers, "parentElement", dom_node_parent_element_read, NULL);
 	zend_hash_add_new_ptr(&classes, dom_namespace_node_class_entry->name, &dom_namespace_node_prop_handlers);
 
+	dom_namespace_info_class_entry = register_class_DOM_NamespaceInfo();
+
 	dom_documentfragment_class_entry = register_class_DOMDocumentFragment(dom_node_class_entry, dom_parentnode_class_entry);
 	dom_documentfragment_class_entry->create_object = dom_objects_new;
 	dom_documentfragment_class_entry->default_object_handlers = &dom_object_handlers;

ext/dom/php_dom.stub.php

@@ -1376,8 +1376,12 @@ public function replaceChildren(Node|string ...$nodes): void {}
 
         public string $substitutedNodeValue;
 
+        /** @return list<NamespaceInfo> */
         public function getInScopeNamespaces(): array {}
 
+        /** @return list<NamespaceInfo> */
+        public function getDescendantNamespaces(): array {}
+
         public function rename(?string $namespaceURI, string $qualifiedName): void {}
     }
 
@@ -1645,6 +1649,20 @@ public function saveXML(?Node $node = null, int $options = 0): string|false {}
         public function saveXMLFile(string $filename, int $options = 0): int|false {}
     }
 
+    /**
+     * @not-serializable
+     * @strict-properties
+     */
+    final class NamespaceInfo
+    {
+        public readonly ?string $prefix;
+        public readonly ?string $namespaceURI;
+        public readonly Element $element;
+
+        /** @implementation-alias DOM\Node::__construct */
+        private function __construct() {}
+    }
+
 #ifdef LIBXML_XPATH_ENABLED
     /** @not-serializable */
     final class XPath