Changeset 58828

Timestamp:

07/29/2024 05:37:48 PM (8 months ago)

Author:

dmsnell

Message:

HTML API: Close all elements at the end of a document.

When the model of breadcrumb generation in the HTML Processor and node
traversal was simplified, the change introduced a bug whereby unclosed
nodes at the end of a document would remain unvisited and unclosed.

In this patch, a fix is applied to ensure that all open elements close
while traversing a document. A couple of minor documentation typos are
fixed in the patch as well.

Developed in ​https://github.com/wordpress/wordpress-develop/pull/7085
Discussed in https://core.trac.wordpress.org/ticket/61576

Follow-up to [58713].

Props: dmsnell, gziolo, jonsurrell.
See #61576.

Location:

trunk

Files:

• src/wp-includes/html-api/class-wp-html-processor.php (modified) (4 diffs)
• tests/phpunit/tests/html-api/wpHtmlProcessor.php (modified) (1 diff)

trunk/src/wp-includes/html-api/class-wp-html-processor.php

@@ -610 +610 @@
          *       tokens works in the meantime and isn't obviously wrong.
          */
-        while ( empty( $this->element_queue ) && $this->step() ) {
-            continue;
+        if ( empty( $this->element_queue ) && $this->step() ) {
+            return $this->next_token();
         }
 
@@ -617 +617 @@
         $this->current_element = array_shift( $this->element_queue );
         if ( ! isset( $this->current_element ) ) {
-            return false;
+            // There are no tokens left, so close all remaining open elements.
+            while ( $this->state->stack_of_open_elements->pop() ) {
+                continue;
+            }
+
+            return empty( $this->element_queue ) ? false : $this->next_token();
         }
 
@@ -624 +629 @@
         /*
          * The root node only exists in the fragment parser, and closing it
-         * indicates that the parse is complete. Stop before popping if from
+         * indicates that the parse is complete. Stop before popping it from
          * the breadcrumbs.
          */
         if ( 'root-node' === $this->current_element->token->bookmark_name ) {
-            return ! $is_pop && $this->next_token();
+            return $this->next_token();
         }
 
@@ -639 +644 @@
 
         // Avoid sending close events for elements which don't expect a closing.
-        if ( $is_pop && ! static::expects_closer( $this->current_element->token ) ) {
+        if ( $is_pop && ! $this->expects_closer( $this->current_element->token ) ) {
             return $this->next_token();
         }

trunk/tests/phpunit/tests/html-api/wpHtmlProcessor.php

@@ -477 +477 @@
 
     /**
+     * Ensures that elements which are unopened at the end of a document are implicitly closed.
+     *
+     * @ticket 61576
+     */
+    public function test_closes_unclosed_elements() {
+        $processor = WP_HTML_Processor::create_fragment( '<div><p><span>' );
+
+        $this->assertTrue(
+            $processor->next_tag( 'SPAN' ),
+            'Could not find SPAN element: check test setup.'
+        );
+
+        // This is the end of the document, but there should be three closing events.
+        $processor->next_token();
+        $this->assertSame(
+            'SPAN',
+            $processor->get_tag(),
+            'Should have found implicit SPAN closing tag.'
+        );
+
+        $processor->next_token();
+        $this->assertSame(
+            'P',
+            $processor->get_tag(),
+            'Should have found implicit P closing tag.'
+        );
+
+        $processor->next_token();
+        $this->assertSame(
+            'DIV',
+            $processor->get_tag(),
+            'Should have found implicit DIV closing tag.'
+        );
+
+        $this->assertFalse(
+            $processor->next_token(),
+            "Should have failed to find any more tokens but found a '{$processor->get_token_name()}'"
+        );
+    }
+
+    /**
      * Ensures that subclasses can be created from ::create_fragment method.
      *