Add $outerHTML property to DOM

ext/dom/config.m4

@@ -206,7 +206,7 @@ if test "$PHP_DOM" != "no"; then
         html5_parser.c
         html5_serializer.c
         infra.c
-        inner_html_mixin.c
+        inner_outer_html_mixin.c
         namednodemap.c
         namespace_compat.c
         node.c

ext/dom/config.w32

@@ -10,7 +10,7 @@ if (PHP_DOM == "yes") {
 		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 private_data.c \
 			domexception.c processinginstruction.c \
-			cdatasection.c documentfragment.c domimplementation.c element.c  inner_html_mixin.c \
+			cdatasection.c documentfragment.c domimplementation.c element.c inner_outer_html_mixin.c \
 			node.c characterdata.c documenttype.c \
 			entity.c nodelist.c html_collection.c text.c comment.c \
 			entityreference.c \

ext/dom/dom_properties.h

@@ -86,6 +86,8 @@ zend_result dom_element_id_write(dom_object *obj, zval *newval);
 zend_result dom_element_schema_type_info_read(dom_object *obj, zval *retval);
 zend_result dom_element_inner_html_read(dom_object *obj, zval *retval);
 zend_result dom_element_inner_html_write(dom_object *obj, zval *newval);
+zend_result dom_element_outer_html_read(dom_object *obj, zval *retval);
+zend_result dom_element_outer_html_write(dom_object *obj, zval *newval);
 zend_result dom_element_class_list_read(dom_object *obj, zval *retval);
 zend_result dom_modern_element_substituted_node_value_read(dom_object *obj, zval *retval);
 zend_result dom_modern_element_substituted_node_value_write(dom_object *obj, zval *newval);

ext/dom/inner_html_mixin.c → ext/dom/inner_outer_html_mixin.c

@@ -55,12 +55,9 @@ static int dom_write_smart_str(void *context, const char *buffer, int len)
 	return len;
 }
 
-/* https://w3c.github.io/DOM-Parsing/#the-innerhtml-mixin
- * and https://w3c.github.io/DOM-Parsing/#dfn-fragment-serializing-algorithm */
-zend_result dom_element_inner_html_read(dom_object *obj, zval *retval)
+/* https://html.spec.whatwg.org/multipage/dynamic-markup-insertion.html#fragment-serializing-algorithm-steps */
+static zend_string *dom_element_html_fragment_serialize(dom_object *obj, xmlNodePtr node)
 {
-	DOM_PROP_NODE(xmlNodePtr, node, obj);
-
 	/* 1. Let context document be the value of node's node document. */
 	const xmlDoc *context_document = node->doc;
 
@@ -73,7 +70,7 @@ zend_result dom_element_inner_html_read(dom_object *obj, zval *retval)
 		ctx.write_string = dom_inner_html_write_string;
 		ctx.write_string_len = dom_inner_html_write_string_len;
 		dom_html5_serialize(&ctx, node);
-		ZVAL_STR(retval, smart_str_extract(&output));
+		return smart_str_extract(&output);
 	}
 	/* 3. Otherwise, context document is an XML document; return an XML serialization of node passing the flag require well-formed. */
 	else {
@@ -104,11 +101,21 @@ zend_result dom_element_inner_html_read(dom_object *obj, zval *retval)
 		if (UNEXPECTED(status < 0)) {
 			smart_str_free_ex(&str, false);
 			php_dom_throw_error_with_message(SYNTAX_ERR, "The resulting XML serialization is not well-formed", true);
-			return FAILURE;
+			return NULL;
 		}
-		ZVAL_STR(retval, smart_str_extract(&str));
+		return smart_str_extract(&str);
 	}
+}
 
+/* https://w3c.github.io/DOM-Parsing/#the-innerhtml-mixin */
+zend_result dom_element_inner_html_read(dom_object *obj, zval *retval)
+{
+	DOM_PROP_NODE(xmlNodePtr, node, obj);
+	zend_string *serialization = dom_element_html_fragment_serialize(obj, node);
+	if (serialization == NULL) {
+		return FAILURE;
+	}
+	ZVAL_STR(retval, serialization);
 	return SUCCESS;
 }
 
@@ -334,23 +341,31 @@ static xmlNodePtr dom_xml_fragment_parsing_algorithm(dom_object *obj, const xmlN
 	return NULL;
 }
 
-/* https://w3c.github.io/DOM-Parsing/#the-innerhtml-mixin
- * and https://w3c.github.io/DOM-Parsing/#dfn-fragment-parsing-algorithm */
-zend_result dom_element_inner_html_write(dom_object *obj, zval *newval)
+/* https://w3c.github.io/DOM-Parsing/#dfn-fragment-parsing-algorithm */
+static xmlNodePtr dom_parse_fragment(dom_object *obj, xmlNodePtr context_node, const zend_string *input)
 {
-	DOM_PROP_NODE(xmlNodePtr, context_node, obj);
-
-	xmlNodePtr fragment;
 	if (context_node->doc->type == XML_DOCUMENT_NODE) {
-		fragment = dom_xml_fragment_parsing_algorithm(obj, context_node, Z_STR_P(newval));
+		return dom_xml_fragment_parsing_algorithm(obj, context_node, input);
 	} else {
-		fragment = dom_html_fragment_parsing_algorithm(obj, context_node, Z_STR_P(newval), obj->document->quirks_mode);
+		return dom_html_fragment_parsing_algorithm(obj, context_node, input, obj->document->quirks_mode);
 	}
+}
 
+/* https://w3c.github.io/DOM-Parsing/#the-innerhtml-mixin */
+zend_result dom_element_inner_html_write(dom_object *obj, zval *newval)
+{
+	/* 1. We don't do injection sinks, skip. */
+
+	/* 2. Let context be this. */
+	DOM_PROP_NODE(xmlNodePtr, context_node, obj);
+
+	/* 3. Let fragment be the result of invoking the fragment parsing algorithm steps with context and compliantString. */
+	xmlNodePtr fragment = dom_parse_fragment(obj, context_node, Z_STR_P(newval));
 	if (fragment == NULL) {
 		return FAILURE;
 	}
 
+	/* 4. If context is a template element, then set context to the template element's template contents (a DocumentFragment). */
 	if (php_dom_ns_is_fast(context_node, php_dom_ns_is_html_magic_token) && xmlStrEqual(context_node->name, BAD_CAST "template")) {
 		context_node = php_dom_ensure_templated_content(php_dom_get_private_data(obj), context_node);
 		if (context_node == NULL) {
@@ -359,8 +374,94 @@ zend_result dom_element_inner_html_write(dom_object *obj, zval *newval)
 		}
 	}
 
+	/* 5. Replace all with fragment within context. */
 	dom_remove_all_children(context_node);
 	return php_dom_pre_insert(obj->document, fragment, context_node, NULL) ? SUCCESS : FAILURE;
 }
 
+/* https://html.spec.whatwg.org/multipage/dynamic-markup-insertion.html#the-outerhtml-property */
+zend_result dom_element_outer_html_read(dom_object *obj, zval *retval)
+{
+	DOM_PROP_NODE(xmlNodePtr, this, obj);
+
+	/* 1. Let element be a fictional node whose only child is this. */
+	xmlNode element;
+	memset(&element, 0, sizeof(element));
+	element.type = XML_DOCUMENT_FRAG_NODE;
+	element.children = element.last = this;
+	element.doc = this->doc;
+
+	xmlNodePtr old_parent = this->parent;
+	this->parent = &element;
+
+	/* 2. Return the result of running fragment serializing algorithm steps with element and true. */
+	zend_string *serialization = dom_element_html_fragment_serialize(obj, &element);
+
+	this->parent = old_parent;
+
+	if (serialization == NULL) {
+		return FAILURE;
+	}
+	ZVAL_STR(retval, serialization);
+	return SUCCESS;
+}
+
+/* https://html.spec.whatwg.org/multipage/dynamic-markup-insertion.html#the-outerhtml-property */
+zend_result dom_element_outer_html_write(dom_object *obj, zval *newval)
+{
+	/* 1. We don't do injection sinks, skip. */
+
+	/* 2. Let parent be this's parent. */
+	DOM_PROP_NODE(xmlNodePtr, this, obj);
+	xmlNodePtr parent = this->parent;
+	bool created_parent = false;
+
+	/* 3. If parent is null, return. */
+	if (parent == NULL) {
+		return SUCCESS;
+	}
+
+	/* 4. If parent is a Document, throw. */
+	if (parent->type == XML_DOCUMENT_NODE || parent->type == XML_HTML_DOCUMENT_NODE) {
+		php_dom_throw_error(INVALID_MODIFICATION_ERR, true);
+		return FAILURE;
+	}
+
+	/* 5. If parent is a DocumentFragment, set parent to the result of creating an element given this's node document, body, and the HTML namespace. */
+	if (parent->type == XML_DOCUMENT_FRAG_NODE) {
+		xmlNsPtr html_ns = php_dom_libxml_ns_mapper_ensure_html_ns(php_dom_get_ns_mapper(obj));
+
+		parent = xmlNewDocNode(parent->doc, html_ns, BAD_CAST "body", NULL);
+		created_parent = true;
+		if (UNEXPECTED(parent == NULL)) {
+			php_dom_throw_error(INVALID_STATE_ERR, true);
+			return FAILURE;
+		}
+	}
+
+	/* 6. Let fragment be the result of invoking the fragment parsing algorithm steps given parent and compliantString. */
+	xmlNodePtr fragment = dom_parse_fragment(obj, parent, Z_STR_P(newval));
+	if (fragment == NULL) {
+		if (created_parent) {
+			xmlFreeNode(parent);
+		}
+		return FAILURE;
+	}
+
+	/* 7. Replace this with fragment within this's parent. */
+	if (!php_dom_pre_insert(obj->document, fragment, this->parent, this)) {
+		xmlFreeNode(fragment);
+		if (created_parent) {
+			xmlFreeNode(parent);
+		}
+		return FAILURE;
+	}
+	xmlUnlinkNode(this);
+	if (created_parent) {
+		ZEND_ASSERT(parent->children == NULL);
+		xmlFreeNode(parent);
+	}
+	return SUCCESS;
+}
+
 #endif

ext/dom/php_dom.c

@@ -1111,6 +1111,7 @@ PHP_MINIT_FUNCTION(dom)
 	DOM_REGISTER_PROP_HANDLER(&dom_modern_element_prop_handlers, "previousElementSibling", dom_node_previous_element_sibling_read, NULL);
 	DOM_REGISTER_PROP_HANDLER(&dom_modern_element_prop_handlers, "nextElementSibling", dom_node_next_element_sibling_read, NULL);
 	DOM_REGISTER_PROP_HANDLER(&dom_modern_element_prop_handlers, "innerHTML", dom_element_inner_html_read, dom_element_inner_html_write);
+	DOM_REGISTER_PROP_HANDLER(&dom_modern_element_prop_handlers, "outerHTML", dom_element_outer_html_read, dom_element_outer_html_write);
 	DOM_REGISTER_PROP_HANDLER(&dom_modern_element_prop_handlers, "substitutedNodeValue", dom_modern_element_substituted_node_value_read, dom_modern_element_substituted_node_value_write);
 	zend_hash_merge(&dom_modern_element_prop_handlers, &dom_modern_node_prop_handlers, NULL, false);
 	DOM_OVERWRITE_PROP_HANDLER(&dom_modern_element_prop_handlers, "textContent", dom_node_text_content_read, dom_node_text_content_write);

ext/dom/php_dom.stub.php

@@ -1688,6 +1688,9 @@ public function matches(string $selectors): bool {}
         /** @virtual */
         public string $innerHTML;
 
+        /** @virtual */
+        public string $outerHTML;
+
         /** @virtual */
         public string $substitutedNodeValue;
 

ext/dom/tests/gh15192.phpt

@@ -10,8 +10,8 @@ $element = $dom2->firstChild;
 $dom = new DomDocument();
 var_dump($element);
 ?>
---EXPECTF--
-object(Dom\HTMLElement)#3 (29) {
+--EXPECT--
+object(Dom\HTMLElement)#3 (30) {
   ["namespaceURI"]=>
   string(28) "http://www.w3.org/1999/xhtml"
   ["prefix"]=>
@@ -40,6 +40,8 @@ object(Dom\HTMLElement)#3 (29) {
   NULL
   ["innerHTML"]=>
   string(36) "<head></head><body><p>foo</p></body>"
+  ["outerHTML"]=>
+  string(49) "<html><head></head><body><p>foo</p></body></html>"
   ["substitutedNodeValue"]=>
   string(3) "foo"
   ["nodeType"]=>

ext/dom/tests/modern/html/parser/Element_outerHTML.phpt

@@ -0,0 +1,34 @@
+--TEST--
+Test writing Element::$outerHTML on HTML documents
+--EXTENSIONS--
+dom
+--FILE--
+<?php
+
+$dom = Dom\HTMLDocument::createFromString('<p>foo</p>', LIBXML_NOERROR);
+$p = $dom->body->firstChild;
+$p->outerHTML = '<div></div>&nbsp;<p>'; // intentionally unclosed
+echo $dom->saveXML(), "\n";
+echo $dom->saveHtml(), "\n";
+$div = $dom->body->firstChild;
+$div->outerHTML = "invalid\xffutf-8𐍈𐍈𐍈";
+echo $dom->saveXML(), "\n";
+echo $dom->saveHtml(), "\n";
+
+$dom->body->outerHTML = '<template><p>foo</p></template>';
+var_dump($dom->body->querySelector('p')); // Should be NULL because the template contents do not participate in the DOM tree
+echo $dom->saveXML(), "\n";
+echo $dom->saveHtml(), "\n";
+
+?>
+--EXPECT--
+<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+<html xmlns="http://www.w3.org/1999/xhtml"><head></head><body><div></div> <p></p></body></html>
+<html><head></head><body><div></div>&nbsp;<p></p></body></html>
+<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+<html xmlns="http://www.w3.org/1999/xhtml"><head></head><body>invalid�utf-8𐍈𐍈𐍈 <p></p></body></html>
+<html><head></head><body>invalid�utf-8𐍈𐍈𐍈&nbsp;<p></p></body></html>
+NULL
+<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+<html xmlns="http://www.w3.org/1999/xhtml"><head></head><head><template><p>foo</p></template></head><body></body></html>
+<html><head></head><head><template><p>foo</p></template></head><body></body></html>

ext/dom/tests/modern/html/serializer/Element_outerHTML.phpt

@@ -0,0 +1,28 @@
+--TEST--
+Test reading Element::$outerHTML on HTML documents
+--EXTENSIONS--
+dom
+--FILE--
+<?php
+
+$dom = Dom\HTMLDocument::createFromString('<p>foo</p>', LIBXML_NOERROR);
+
+$p = $dom->body->firstChild;
+var_dump($p->outerHTML);
+
+$root = $dom->documentElement;
+var_dump($root->outerHTML);
+
+$unattached_element = $dom->createElement('unattached');
+var_dump($unattached_element->outerHTML);
+
+$template = $dom->createElement('template');
+$template->innerHTML = '<p>foo</p>';
+var_dump($template->outerHTML);
+
+?>
+--EXPECT--
+string(10) "<p>foo</p>"
+string(49) "<html><head></head><body><p>foo</p></body></html>"
+string(25) "<unattached></unattached>"
+string(31) "<template><p>foo</p></template>"

ext/dom/tests/modern/html/serializer/Element_outerHTML_invalid_tree.phpt

@@ -0,0 +1,17 @@
+--TEST--
+Test reading Element::$outerHTML on HTML documents - invalid tree variation
+--EXTENSIONS--
+dom
+--CREDITS--
+Dennis Snell
+--FILE--
+<?php
+
+$dom = Dom\HTMLDocument::createFromString('<a href="#one"><p>Link</p></a>', LIBXML_NOERROR);
+$p = $dom->body->querySelector('p');
+$p->outerHTML = '<a href="#two">Another Link</a>';
+echo $dom->saveHTML();
+
+?>
+--EXPECT--
+<html><head></head><body><a href="#one"><a href="#two">Another Link</a></a></body></html>