Implement GH-18550: Implement getElementsByClassName() (#19108)

Spec: https://dom.spec.whatwg.org/#ref-for-dom-element-getelementsbyclassname

Author: nielsdos

NEWS

@@ -8,6 +8,7 @@ PHP NEWS
 
 - DOM:
  . Make cloning DOM node lists, maps, and collections fail. (nielsdos)
+ . Added Dom\Element::getElementsByClassName(). (nielsdos)
 
 - PDO_ODBC
  . Fetch larger block sizes and better handle SQL_NO_TOTAL when calling

UPGRADING

@@ -408,6 +408,7 @@ PHP 8.5 UPGRADE NOTES
  RFC: https://wiki.php.net/rfc/curl_share_persistence_improvement
 
 - DOM:
+ . Added Dom\Element::getElementsByClassName().
  . Added Dom\Element::insertAdjacentHTML().
 
 - Enchant:

ext/dom/element.c

@@ -842,6 +842,44 @@ PHP_METHOD(Dom_Element, getElementsByTagName)
 }
 /* }}} end dom_element_get_elements_by_tag_name */
 
+PHP_METHOD(Dom_Element, getElementsByClassName)
+{
+dom_object *intern, *namednode;
+zend_string *class_names;
+
+if (zend_parse_parameters(ZEND_NUM_ARGS(), "P", &class_names) == FAILURE) {
+RETURN_THROWS();
+}
+
+if (ZSTR_LEN(class_names) > INT_MAX) {
+zend_argument_value_error(1, "is too long");
+RETURN_THROWS();
+}
+
+DOM_GET_THIS_INTERN(intern);
+
+object_init_ex(return_value, dom_html_collection_class_entry);
+namednode = Z_DOMOBJ_P(return_value);
+
+HashTable *token_set;
+ALLOC_HASHTABLE(token_set);
+zend_hash_init(token_set, 0, NULL, NULL, false);
+dom_ordered_set_parser(token_set, ZSTR_VAL(class_names), intern->document->quirks_mode == PHP_LIBXML_QUIRKS);
+
+if (zend_hash_num_elements(token_set) == 0) {
+php_dom_create_obj_map(intern, namednode, NULL, NULL, NULL, &php_dom_obj_map_noop);
+
+zend_hash_destroy(token_set);
+FREE_HASHTABLE(token_set);
+} else {
+php_dom_create_obj_map(intern, namednode, NULL, NULL, NULL, &php_dom_obj_map_by_class_name);
+
+dom_nnodemap_object *map = namednode->ptr;
+map->array = token_set;
+map->release_array = true;
+}
+}
+
 /* should_free_result must be initialized to false */
 static const xmlChar *dom_get_attribute_ns(dom_object *intern, xmlNodePtr elemp, const char *uri, size_t uri_len, const char *name, bool *should_free_result)
 {

ext/dom/obj_map.c

@@ -24,6 +24,7 @@
 #if defined(HAVE_LIBXML) && defined(HAVE_DOM)
 #include "php_dom.h"
 #include "obj_map.h"
+#include "token_list.h"
 
 static zend_always_inline void objmap_cache_release_cached_obj(dom_nnodemap_object *objmap)
 {
@@ -40,6 +41,30 @@ static zend_always_inline void reset_objmap_cache(dom_nnodemap_object *objmap)
 objmap->cached_length = -1;
 }
 
+static bool dom_matches_class_name(const dom_nnodemap_object *map, const xmlNode *nodep)
+{
+bool ret = false;
+
+if (nodep->type == XML_ELEMENT_NODE) {
+xmlAttrPtr classes = xmlHasNsProp(nodep, BAD_CAST "class", NULL);
+if (classes != NULL) {
+bool should_free;
+xmlChar *value = php_libxml_attr_value(classes, &should_free);
+
+bool quirks = map->baseobj->document->quirks_mode == PHP_LIBXML_QUIRKS;
+if (dom_ordered_set_all_contained(map->array, (const char *) value, quirks)) {
+ret = true;
+}
+
+if (should_free) {
+xmlFree(value);
+}
+}
+}
+
+return ret;
+}
+
 /**************************
  * === Length methods === *
  **************************/
@@ -106,6 +131,24 @@ static zend_long dom_map_get_by_tag_name_length(dom_nnodemap_object *map)
 return count;
 }
 
+static zend_long dom_map_get_by_class_name_length(dom_nnodemap_object *map)
+{
+xmlNodePtr nodep = dom_object_get_node(map->baseobj);
+zend_long count = 0;
+if (nodep) {
+xmlNodePtr basep = nodep;
+nodep = php_dom_first_child_of_container_node(basep);
+
+while (nodep != NULL) {
+if (dom_matches_class_name(map, nodep)) {
+count++;
+}
+nodep = php_dom_next_in_tree_order(nodep, basep);
+}
+}
+return count;
+}
+
 static zend_long dom_map_get_zero_length(dom_nnodemap_object *map)
 {
 return 0;
@@ -276,6 +319,10 @@ static void dom_map_collection_named_item_elements_iter(dom_nnodemap_object *map
 }
 }
 
+static void dom_map_collection_named_item_null(dom_nnodemap_object *map, php_dom_obj_map_collection_iter *iter)
+{
+}
+
 static void dom_map_get_by_tag_name_item(dom_nnodemap_object *map, zend_long index, zval *return_value)
 {
 xmlNodePtr nodep = dom_object_get_node(map->baseobj);
@@ -292,12 +339,54 @@ static void dom_map_get_by_tag_name_item(dom_nnodemap_object *map, zend_long ind
 }
 }
 
+static void dom_map_get_by_class_name_item(dom_nnodemap_object *map, zend_long index, zval *return_value)
+{
+xmlNodePtr nodep = dom_object_get_node(map->baseobj);
+xmlNodePtr itemnode = NULL;
+if (nodep && index >= 0) {
+dom_node_idx_pair start_point = dom_obj_map_get_start_point(map, nodep, index);
+if (start_point.node) {
+if (start_point.index > 0) {
+/* Only start iteration at next point if we actually have an index to seek to. */
+itemnode = php_dom_next_in_tree_order(start_point.node, nodep);
+} else {
+itemnode = start_point.node;
+}
+} else {
+itemnode = php_dom_first_child_of_container_node(nodep);
+}
+
+do {
+--start_point.index;
+while (itemnode != NULL && !dom_matches_class_name(map, itemnode)) {
+itemnode = php_dom_next_in_tree_order(itemnode, nodep);
+}
+} while (start_point.index > 0 && itemnode);
+}
+dom_ret_node_to_zobj(map, itemnode, return_value);
+if (itemnode) {
+dom_map_cache_obj(map, itemnode, index, return_value);
+}
+}
+
 static void dom_map_collection_named_item_by_tag_name_iter(dom_nnodemap_object *map, php_dom_obj_map_collection_iter *iter)
 {
 iter->candidate = dom_get_elements_by_tag_name_ns_raw(iter->basep, iter->candidate, map->ns, map->local, map->local_lower, &iter->cur, iter->next);
 iter->next = iter->cur + 1;
 }
 
+static void dom_map_collection_named_item_by_class_name_iter(dom_nnodemap_object *map, php_dom_obj_map_collection_iter *iter)
+{
+xmlNodePtr basep = iter->basep;
+xmlNodePtr nodep = iter->candidate ? php_dom_next_in_tree_order(iter->candidate, basep) : php_dom_first_child_of_container_node(basep);
+
+while (nodep != NULL && !dom_matches_class_name(map, nodep)) {
+nodep = php_dom_next_in_tree_order(nodep, basep);
+}
+
+iter->candidate = nodep;
+}
+
 static void dom_map_get_null_item(dom_nnodemap_object *map, zend_long index, zval *return_value)
 {
 RETURN_NULL();
@@ -478,6 +567,16 @@ const php_dom_obj_map_handler php_dom_obj_map_by_tag_name = {
 .nameless = true,
 };
 
+const php_dom_obj_map_handler php_dom_obj_map_by_class_name = {
+.length = dom_map_get_by_class_name_length,
+.get_item = dom_map_get_by_class_name_item,
+.get_ns_named_item = dom_map_get_ns_named_item_null,
+.has_ns_named_item = dom_map_has_ns_named_item_null,
+.collection_named_item_iter = dom_map_collection_named_item_by_class_name_iter,
+.use_cache = true,
+.nameless = true,
+};
+
 const php_dom_obj_map_handler php_dom_obj_map_child_nodes = {
 .length = dom_map_get_nodes_length,
 .get_item = dom_map_get_nodes_item,
@@ -533,7 +632,7 @@ const php_dom_obj_map_handler php_dom_obj_map_noop = {
 .get_item = dom_map_get_null_item,
 .get_ns_named_item = dom_map_get_ns_named_item_null,
 .has_ns_named_item = dom_map_has_ns_named_item_null,
-.collection_named_item_iter = NULL,
+.collection_named_item_iter = dom_map_collection_named_item_null,
 .use_cache = false,
 .nameless = true,
 };

ext/dom/obj_map.h

@@ -63,6 +63,7 @@ zend_long php_dom_get_nodelist_length(dom_object *obj);
 
 extern const php_dom_obj_map_handler php_dom_obj_map_attributes;
 extern const php_dom_obj_map_handler php_dom_obj_map_by_tag_name;
+extern const php_dom_obj_map_handler php_dom_obj_map_by_class_name;
 extern const php_dom_obj_map_handler php_dom_obj_map_child_elements;
 extern const php_dom_obj_map_handler php_dom_obj_map_child_nodes;
 extern const php_dom_obj_map_handler php_dom_obj_map_nodeset;

ext/dom/php_dom.stub.php

@@ -1659,6 +1659,7 @@ public function removeAttributeNode(Attr $attr) : Attr {}
 
  public function getElementsByTagName(string $qualifiedName): HTMLCollection {}
  public function getElementsByTagNameNS(?string $namespace, string $localName): HTMLCollection {}
+ public function getElementsByClassName(string $classNames): HTMLCollection {}
 
  public function insertAdjacentElement(AdjacentPosition $where, Element $element): ?Element {}
  public function insertAdjacentText(AdjacentPosition $where, string $data): void {}
@@ -1986,6 +1987,8 @@ abstract class Document extends Node implements ParentNode
  public function getElementsByTagName(string $qualifiedName): HTMLCollection {}
  /** @implementation-alias Dom\Element::getElementsByTagNameNS */
  public function getElementsByTagNameNS(?string $namespace, string $localName): HTMLCollection {}
+ /** @implementation-alias Dom\Element::getElementsByClassName */
+ public function getElementsByClassName(string $classNames): HTMLCollection {}
 
  public function createElement(string $localName): Element {}
  public function createElementNS(?string $namespace, string $qualifiedName): Element {}

ext/dom/php_dom_arginfo.h

@@ -0,0 +1,32 @@
+--TEST--
+Dom\Element::getElementsByClassName() empty class names
+--EXTENSIONS--
+dom
+--FILE--
+<?php
+
+$dom = Dom\HTMLDocument::createFromString(<<<HTML
+<div class=" foo bar ">
+ <p id="child"></p>
+</div>
+HTML, LIBXML_NOERROR);
+
+$collection = $dom->documentElement->getElementsByClassName("");
+var_dump($collection->count());
+
+foreach ($collection as $node) {
+ throw new Error("unreachable");
+}
+
+var_dump($dom->getElementsByClassName(" ")->count());
+var_dump($dom->getElementsByClassName("\t")->count());
+var_dump($dom->getElementsByClassName("\t\n\f\v")->count());
+var_dump($dom->getElementsByClassName("\t\n\f\v")->namedItem("child"));
+
+?>
+--EXPECT--
+int(0)
+int(0)
+int(0)
+int(0)
+NULL

ext/dom/tests/modern/common/Element_getElementsByClassName_empty.phpt

@@ -0,0 +1,55 @@
+--TEST--
+Dom\Element::getElementsByClassName() non quirks mode
+--EXTENSIONS--
+dom
+--FILE--
+<?php
+
+$dom = Dom\HTMLDocument::createFromString(<<<HTML
+<!DOCTYPE html>
+<div id="container">
+ <p class="Bar">1</p>
+ <p class="bar">2</p>
+ <p class="Bar Foo">3</p>
+ <p class="Bar foo">4</p>
+ <p class="foo bar">5</p>
+ <p class="foo bar" name="here">6</p>
+</div>
+<div>
+ <p class="Bar">7</p>
+ <p class="bar">8</p>
+ <p class="Bar Foo">9</p>
+ <p class="Bar foo">10</p>
+ <p class="foo bar">11</p>
+</div>
+HTML);
+
+$collection = $dom->getElementsByClassName("foo \n bar");
+
+echo "There are {$collection->length} items in the document in total that have both \"foo\" and \"bar\"\n";
+
+$collection = $dom->getElementById('container')->getElementsByClassName("foo \n bar");
+
+echo "There are {$collection->length} items in #container in total that have both \"foo\" and \"bar\"\n";
+
+foreach ($collection as $key => $node) {
+ echo "--- Key $key ---\n";
+ var_dump($node->tagName, $node->textContent);
+ var_dump($node === $collection->item($key));
+}
+
+var_dump($collection->namedItem("here")->textContent);
+
+?>
+--EXPECT--
+There are 3 items in the document in total that have both "foo" and "bar"
+There are 2 items in #container in total that have both "foo" and "bar"
+--- Key 0 ---
+string(1) "P"
+string(1) "5"
+bool(true)
+--- Key 1 ---
+string(1) "P"
+string(1) "6"
+bool(true)
+string(1) "6"