Properly support template elements in DOM

NEWS

@@ -5,6 +5,9 @@ PHP                                                                        NEWS
 - Core:
   . Fixed bug GH-14801 (Fix build for armv7). (andypost)
 
+- DOM:
+  . Improve support for template elements. (nielsdos)
+
 - GD:
   . Check overflow/underflow for imagescale/imagefilter. (David Carlier)
 

ext/dom/config.m4

@@ -27,7 +27,7 @@ if test "$PHP_DOM" != "no"; then
       $LEXBOR_DIR/ns/ns.c \
       $LEXBOR_DIR/tag/tag.c"
     PHP_NEW_EXTENSION(dom, [php_dom.c attr.c document.c infra.c \
-                            xml_document.c html_document.c xml_serializer.c html5_serializer.c html5_parser.c namespace_compat.c \
+                            xml_document.c html_document.c xml_serializer.c html5_serializer.c html5_parser.c namespace_compat.c private_data.c \
                             domexception.c \
                             parentnode/tree.c parentnode/css_selectors.c \
                             processinginstruction.c cdatasection.c \

ext/dom/config.w32

@@ -8,7 +8,7 @@ if (PHP_DOM == "yes") {
 		CHECK_HEADER_ADD_INCLUDE("libxml/parser.h", "CFLAGS_DOM", PHP_PHP_BUILD + "\\include\\libxml2")
 	) {
 		EXTENSION("dom", "php_dom.c attr.c document.c infra.c \
-			xml_document.c html_document.c xml_serializer.c html5_serializer.c html5_parser.c namespace_compat.c \
+			xml_document.c html_document.c xml_serializer.c html5_serializer.c html5_parser.c namespace_compat.c private_data.c \
 			domexception.c processinginstruction.c \
 			cdatasection.c documentfragment.c domimplementation.c element.c  inner_html_mixin.c \
 			node.c characterdata.c documenttype.c \

ext/dom/document.c

@@ -23,6 +23,7 @@
 #if defined(HAVE_LIBXML) && defined(HAVE_DOM)
 #include "php_dom.h"
 #include "namespace_compat.h"
+#include "private_data.h"
 #include "xml_serializer.h"
 #include "internal_helpers.h"
 #include "dom_properties.h"

ext/dom/domimplementation.c

@@ -23,6 +23,7 @@
 #if defined(HAVE_LIBXML) && defined(HAVE_DOM)
 #include "php_dom.h"
 #include "namespace_compat.h"
+#include "private_data.h"
 
 /*
 * class DOMImplementation
@@ -266,7 +267,8 @@ PHP_METHOD(Dom_Implementation, createDocument)
 
 	xmlDocPtr document = NULL;
 	xmlChar *localname = NULL, *prefix = NULL;
-	php_dom_libxml_ns_mapper *ns_mapper = php_dom_libxml_ns_mapper_create();
+	php_dom_private_data *private_data = php_dom_private_data_create();
+	php_dom_libxml_ns_mapper *ns_mapper = php_dom_ns_mapper_from_private(private_data);
 
 	/* 1. Let document be a new XMLDocument. */
 	document = xmlNewDoc(BAD_CAST "1.0");
@@ -307,7 +309,7 @@ PHP_METHOD(Dom_Implementation, createDocument)
 		NULL
 	);
 	dom_set_xml_class(intern->document);
-	intern->document->private_data = php_dom_libxml_ns_mapper_header(ns_mapper);
+	intern->document->private_data = php_dom_libxml_private_data_header(private_data);
 
 	/* 4. If doctype is non-null, append doctype to document. */
 	if (doctype != NULL) {
@@ -337,7 +339,7 @@ PHP_METHOD(Dom_Implementation, createDocument)
 	xmlFree(localname);
 	xmlFree(prefix);
 	xmlFreeDoc(document);
-	php_dom_libxml_ns_mapper_destroy(ns_mapper);
+	php_dom_private_data_destroy(private_data);
 	RETURN_THROWS();
 }
 /* }}} end dom_domimplementation_create_document */
@@ -366,7 +368,8 @@ PHP_METHOD(Dom_Implementation, createHTMLDocument)
 	/* 3. Append a new doctype, with "html" as its name and with its node document set to doc, to doc. */
 	xmlDtdPtr dtd = xmlCreateIntSubset(doc, BAD_CAST "html", NULL, NULL);
 
-	php_dom_libxml_ns_mapper *ns_mapper = php_dom_libxml_ns_mapper_create();
+	php_dom_private_data *private_data = php_dom_private_data_create();
+	php_dom_libxml_ns_mapper *ns_mapper = php_dom_ns_mapper_from_private(private_data);
 	xmlNsPtr html_ns = php_dom_libxml_ns_mapper_ensure_html_ns(ns_mapper);
 
 	/* 4. Append the result of creating an element given doc, html, and the HTML namespace, to doc. */
@@ -396,7 +399,7 @@ PHP_METHOD(Dom_Implementation, createHTMLDocument)
 	if (UNEXPECTED(dtd == NULL || html_element == NULL || head_element == NULL || (title != NULL && title_element == NULL) || body_element == NULL)) {
 		php_dom_throw_error(INVALID_STATE_ERR, true);
 		xmlFreeDoc(doc);
-		php_dom_libxml_ns_mapper_destroy(ns_mapper);
+		php_dom_private_data_destroy(private_data);
 		RETURN_THROWS();
 	}
 
@@ -408,7 +411,7 @@ PHP_METHOD(Dom_Implementation, createHTMLDocument)
 		NULL
 	);
 	dom_set_xml_class(intern->document);
-	intern->document->private_data = php_dom_libxml_ns_mapper_header(ns_mapper);
+	intern->document->private_data = php_dom_libxml_private_data_header(private_data);
 }
 /* }}} */
 

ext/dom/element.c

@@ -24,6 +24,7 @@
 #include "zend_enum.h"
 #include "php_dom.h"
 #include "namespace_compat.h"
+#include "private_data.h"
 #include "internal_helpers.h"
 #include "dom_properties.h"
 #include "token_list.h"
@@ -2030,6 +2031,16 @@ PHP_METHOD(Dom_Element, rename)
 			}
 			goto cleanup;
 		}
+
+		/* If we currently have a template but the new element type won't be a template, then throw away the templated content. */
+		if (is_currently_html_ns && xmlStrEqual(nodep->name, BAD_CAST "template") && !xmlStrEqual(localname, BAD_CAST "template")) {
+			php_dom_throw_error_with_message(
+				INVALID_MODIFICATION_ERR,
+				"It is not possible to rename the template element because it hosts a document fragment",
+				/* strict */ true
+			);
+			goto cleanup;
+		}
 	}
 
 	php_libxml_invalidate_node_list_cache(intern->document);

ext/dom/html5_parser.c

@@ -22,12 +22,13 @@
 #if defined(HAVE_LIBXML) && defined(HAVE_DOM)
 #include "php_dom.h"
 #include "html5_parser.h"
+#include "private_data.h"
 #include <lexbor/html/parser.h>
 #include <lexbor/html/interfaces/element.h>
+#include <lexbor/html/interfaces/template_element.h>
 #include <lexbor/dom/dom.h>
 #include <libxml/parserInternals.h>
 #include <libxml/HTMLtree.h>
-#include <Zend/zend.h>
 
 #define WORK_LIST_INIT_SIZE 128
 /* libxml2 reserves 2 pointer-sized words for interned strings */
@@ -63,14 +64,20 @@ static unsigned short sanitize_line_nr(size_t line)
     return (unsigned short) line;
 }
 
-static const php_dom_ns_magic_token *get_libxml_namespace_href(uintptr_t lexbor_namespace)
+struct lxml_ns {
+	const php_dom_ns_magic_token *token;
+	const char *href;
+	size_t href_len;
+};
+
+static struct lxml_ns get_libxml_namespace_href(uintptr_t lexbor_namespace)
 {
     if (lexbor_namespace == LXB_NS_SVG) {
-        return php_dom_ns_is_svg_magic_token;
+        return (struct lxml_ns) { php_dom_ns_is_svg_magic_token, ZEND_STRL(DOM_SVG_NS_URI) };
     } else if (lexbor_namespace == LXB_NS_MATH) {
-        return php_dom_ns_is_mathml_magic_token;
+        return (struct lxml_ns) { php_dom_ns_is_mathml_magic_token, ZEND_STRL(DOM_MATHML_NS_URI) };
     } else {
-        return php_dom_ns_is_html_magic_token;
+        return (struct lxml_ns) { php_dom_ns_is_html_magic_token, ZEND_STRL(DOM_XHTML_NS_URI) };
     }
 }
 
@@ -102,11 +109,12 @@ static lexbor_libxml2_bridge_status lexbor_libxml2_bridge_convert(
     xmlNodePtr root,
     bool compact_text_nodes,
     bool create_default_ns,
-    php_dom_libxml_ns_mapper *ns_mapper
+    php_dom_private_data *private_data
 )
 {
     lexbor_libxml2_bridge_status retval = LEXBOR_LIBXML2_BRIDGE_STATUS_OK;
 
+	php_dom_libxml_ns_mapper *ns_mapper = php_dom_ns_mapper_from_private(private_data);
     xmlNsPtr html_ns = php_dom_libxml_ns_mapper_ensure_html_ns(ns_mapper);
     xmlNsPtr xlink_ns = NULL;
     xmlNsPtr prefixed_xmlns_ns = NULL;
@@ -146,24 +154,47 @@ static lexbor_libxml2_bridge_status lexbor_libxml2_bridge_convert(
                 if (entering_namespace == LXB_NS_HTML) {
                     current_lxml_ns = html_ns;
                 } else {
-                    const php_dom_ns_magic_token *magic_token = get_libxml_namespace_href(entering_namespace);
-                    zend_string *uri = zend_string_init((char *) magic_token, strlen((char *) magic_token), false);
+					struct lxml_ns ns = get_libxml_namespace_href(entering_namespace);
+                    zend_string *uri = zend_string_init(ns.href, ns.href_len, false);
                     current_lxml_ns = php_dom_libxml_ns_mapper_get_ns(ns_mapper, NULL, uri);
                     zend_string_release_ex(uri, false);
                     if (EXPECTED(current_lxml_ns != NULL)) {
-                        current_lxml_ns->_private = (void *) magic_token;
+                        current_lxml_ns->_private = (void *) ns.token;
                     }
                 }
             }
             /* Instead of xmlSetNs() because we know the arguments are valid. Prevents overhead. */
             lxml_element->ns = current_lxml_ns;
 
-            for (lxb_dom_node_t *child_node = element->node.last_child; child_node != NULL; child_node = child_node->prev) {
+			/* Handle template element by creating a fragment node to contain its children.
+			 * Other types of nodes contain their children directly. */
+			xmlNodePtr lxml_child_parent = lxml_element;
+			lxb_dom_node_t *child_node = element->node.last_child;
+			if (lxb_html_tree_node_is(&element->node, LXB_TAG_TEMPLATE)) {
+				if (create_default_ns) {
+					lxml_child_parent = xmlNewDocFragment(lxml_doc);
+					if (UNEXPECTED(lxml_child_parent == NULL)) {
+						retval = LEXBOR_LIBXML2_BRIDGE_STATUS_OOM;
+						break;
+					}
+
+					lxml_child_parent->parent = lxml_element;
+					dom_add_element_ns_hook(private_data, lxml_element);
+					php_dom_add_templated_content(private_data, lxml_element, lxml_child_parent);
+				}
+
+				lxb_html_template_element_t *template = lxb_html_interface_template(&element->node);
+				if (template->content != NULL) {
+					child_node = template->content->node.last_child;
+				}
+			}
+
+            for (; child_node != NULL; child_node = child_node->prev) {
                 lexbor_libxml2_bridge_work_list_item_push(
                     &work_list,
                     child_node,
                     entering_namespace,
-                    lxml_element,
+                    lxml_child_parent,
                     current_lxml_ns
                 );
             }
@@ -307,7 +338,7 @@ lexbor_libxml2_bridge_status lexbor_libxml2_bridge_convert_document(
     xmlDocPtr *doc_out,
     bool compact_text_nodes,
     bool create_default_ns,
-    php_dom_libxml_ns_mapper *ns_mapper
+	php_dom_private_data *private_data
 )
 {
     xmlDocPtr lxml_doc = php_dom_create_html_doc();
@@ -320,7 +351,7 @@ lexbor_libxml2_bridge_status lexbor_libxml2_bridge_convert_document(
         (xmlNodePtr) lxml_doc,
         compact_text_nodes,
         create_default_ns,
-        ns_mapper
+        private_data
     );
     if (status != LEXBOR_LIBXML2_BRIDGE_STATUS_OK) {
         xmlFreeDoc(lxml_doc);
@@ -336,7 +367,7 @@ lexbor_libxml2_bridge_status lexbor_libxml2_bridge_convert_fragment(
     xmlNodePtr *fragment_out,
     bool compact_text_nodes,
     bool create_default_ns,
-    php_dom_libxml_ns_mapper *ns_mapper
+	php_dom_private_data *private_data
 )
 {
     xmlNodePtr fragment = xmlNewDocFragment(lxml_doc);
@@ -349,7 +380,7 @@ lexbor_libxml2_bridge_status lexbor_libxml2_bridge_convert_fragment(
         fragment,
         compact_text_nodes,
         create_default_ns,
-        ns_mapper
+        private_data
     );
     if (status != LEXBOR_LIBXML2_BRIDGE_STATUS_OK) {
         xmlFreeNode(fragment);

ext/dom/html5_parser.h

@@ -71,15 +71,15 @@ lexbor_libxml2_bridge_status lexbor_libxml2_bridge_convert_document(
     xmlDocPtr *doc_out,
     bool compact_text_nodes,
     bool create_default_ns,
-    php_dom_libxml_ns_mapper *ns_mapper
+    php_dom_private_data *private_data
 );
 lexbor_libxml2_bridge_status lexbor_libxml2_bridge_convert_fragment(
     lxb_dom_node_t *start_node,
     xmlDocPtr lxml_doc,
     xmlNodePtr *fragment_out,
     bool compact_text_nodes,
     bool create_default_ns,
-    php_dom_libxml_ns_mapper *ns_mapper
+    php_dom_private_data *private_data
 );
 void lexbor_libxml2_bridge_report_errors(
     const lexbor_libxml2_bridge_parse_context *ctx,

ext/dom/html5_serializer.c

@@ -289,9 +289,13 @@ static zend_result dom_html5_serialize_node(dom_html5_serialize_context *ctx, co
 
 			case XML_ELEMENT_NODE: {
 				TRY(dom_html5_serialize_element_start(ctx, node));
-				if (node->children) {
+				const xmlNode *children = node->children;
+				if (php_dom_ns_is_fast(node, php_dom_ns_is_html_magic_token) && xmlStrEqual(node->name, BAD_CAST "template")) {
+					children = php_dom_retrieve_templated_content(ctx->private_data, node);
+				}
+				if (children) {
 					if (!dom_html5_serializes_as_void(node)) {
-						node = node->children;
+						node = children;
 						continue;
 					}
 				} else {
@@ -301,6 +305,14 @@ static zend_result dom_html5_serialize_node(dom_html5_serialize_context *ctx, co
 				break;
 			}
 
+			case XML_DOCUMENT_FRAG_NODE: {
+				if (node->children) {
+					node = node->children;
+					continue;
+				}
+				break;
+			}
+
 			/* Only exists for compatibility with XML and old DOM. */
 			case XML_ENTITY_REF_NODE: {
 				TRY(dom_html5_serialize_entity_ref(ctx, node));
@@ -346,10 +358,15 @@ zend_result dom_html5_serialize(dom_html5_serialize_context *ctx, const xmlNode
 	}
 
 	/* Step 2 not needed because we're not using a string to store the serialized data */
-	/* Step 3 not needed because we don't support template contents yet */
+
+	/* Step 3. If the node is a template element, then let the node instead be the template element's template contents (a DocumentFragment node). */
+	xmlNodePtr children = php_dom_retrieve_templated_content(ctx->private_data, node);
+	if (!children) {
+		children = node->children;
+	}
 
 	/* Step 4 */
-	return dom_html5_serialize_node(ctx, node->children, node);
+	return dom_html5_serialize_node(ctx, children, node);
 }
 
 /* Variant on the above that is equivalent to the "outer HTML". */

ext/dom/html5_serializer.h

@@ -19,11 +19,13 @@
 
 #include <Zend/zend_types.h>
 #include <libxml/tree.h>
+#include "private_data.h"
 
 typedef struct {
 	zend_result (*write_string)(void *application_data, const char *buf);
 	zend_result (*write_string_len)(void *application_data, const char *buf, size_t len);
 	void *application_data;
+	php_dom_private_data *private_data;
 } dom_html5_serialize_context;
 
 zend_result dom_html5_serialize(dom_html5_serialize_context *ctx, const xmlNode *node);