HTML API: Implement "reconstruct the active formatting elements" algorithm.

As part of work to add more spec support to the HTML API, this patch fills
out the active format reconstruction algorithm so that more HTML can be
supported in situations requiring that reconstruction, for example, when
a formatting element such as an A tag or a CODE tag is implicitly closed.

See Core-61576

Author: dmsnell

src/wp-includes/html-api/class-wp-html-active-formatting-elements.php

@@ -43,6 +43,22 @@ class WP_HTML_Active_Formatting_Elements {
 	 */
 	private $stack = array();
 
+	/**
+	 * Returns the node at the given index in the list of active formatting elements.
+	 *
+	 * Do not use this method; it is meant to be used only by the HTML Processor.
+	 *
+	 * @since 6.7.0
+	 *
+	 * @access private
+	 *
+	 * @param int $index Number of nodes from the top node to return.
+	 * @return WP_HTML_Token|null Node at the given index in the stack, if one exists, otherwise null.
+	 */
+	public function at( $index ) {
+		return $this->stack[ $index ];
+	}
+
 	/**
 	 * Reports if a specific node is in the stack of active formatting elements.
 	 *
@@ -86,6 +102,22 @@ public function current_node() {
 		return $current_node ? $current_node : null;
 	}
 
+	/**
+	 * Inserts a "marker" at the end of the list of active formatting elements.
+	 *
+	 * > The markers are inserted when entering applet, object, marquee,
+	 * > template, td, th, and caption elements, and are used to prevent
+	 * > formatting from "leaking" into applet, object, marquee, template,
+	 * > td, th, and caption elements.
+	 *
+	 * @see https://html.spec.whatwg.org/#concept-parser-marker
+	 *
+	 * @since 6.7.0
+	 */
+	public function insert_marker() {
+		$this->push( new WP_HTML_Token( null, 'marker', false ) );
+	}
+
 	/**
 	 * Pushes a node onto the stack of active formatting elements.
 	 *

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

@@ -2864,7 +2864,7 @@ private function generate_implied_end_tags_thoroughly() {
 	 * > in the current body, cell, or caption (whichever is youngest) that haven't
 	 * > been explicitly closed.
 	 *
-	 * @since 6.4.0
+	 * @since 6.7.0
 	 *
 	 * @throws WP_HTML_Unsupported_Exception When encountering unsupported HTML input.
 	 *
@@ -2873,15 +2873,19 @@ private function generate_implied_end_tags_thoroughly() {
 	 * @return bool Whether any formatting elements needed to be reconstructed.
 	 */
 	private function reconstruct_active_formatting_elements() {
+		$count = $this->state->active_formatting_elements->count();
+
 		/*
 		 * > If there are no entries in the list of active formatting elements, then there is nothing
 		 * > to reconstruct; stop this algorithm.
 		 */
-		if ( 0 === $this->state->active_formatting_elements->count() ) {
+		if ( 0 === $count ) {
 			return false;
 		}
 
-		$last_entry = $this->state->active_formatting_elements->current_node();
+		// Start at the last node in the list of active formatting elements.
+		$currently_at = $count - 1;
+		$last_entry   = $this->state->active_formatting_elements->at( $currently_at );
 		if (
 
 			/*
@@ -2900,8 +2904,39 @@ private function reconstruct_active_formatting_elements() {
 			return false;
 		}
 
-		$this->last_error = self::ERROR_UNSUPPORTED;
-		throw new WP_HTML_Unsupported_Exception( 'Cannot reconstruct active formatting elements when advancing and rewinding is required.' );
+		$entry = $last_entry;
+
+		while ( $currently_at >= 0 ) {
+			if ( 0 === $currently_at ) {
+				goto create;
+			}
+			$entry = $this->state->active_formatting_elements->at( --$currently_at );
+
+			/*
+			 * > If entry is neither a marker nor an element that is also in the stack of open elements,
+			 * > go to the step labeled rewind.
+			 */
+			if ( 'marker' === $entry->node_name || $this->state->stack_of_open_elements->contains_node( $entry ) ) {
+				break;
+			}
+		}
+
+		advance:
+		$entry = $this->state->active_formatting_elements->at( ++$currently_at );
+
+		create:
+		$this->insert_html_element( $entry );
+
+		/*
+		 * > Replace the entry for entry in the list with an entry for new element.
+		 * This doesn't need to happen here since no DOM is being created.
+		 */
+
+		if ( $count - 1 !== $currently_at ) {
+			goto advance;
+		}
+
+		return true;
 	}
 
 	/**

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

@@ -112,18 +112,23 @@ public function test_clear_to_navigate_after_seeking() {
 	}
 
 	/**
-	 * Ensures that support is added for reconstructing active formatting elements
-	 * before the HTML Processor handles situations with unclosed formats requiring it.
+	 * Ensures that support is added for reconstructing active formatting elements.
 	 *
 	 * @ticket 58517
 	 *
 	 * @covers WP_HTML_Processor::reconstruct_active_formatting_elements
 	 */
-	public function test_fails_to_reconstruct_formatting_elements() {
-		$processor = WP_HTML_Processor::create_fragment( '<p><em>One<p><em>Two<p><em>Three<p><em>Four' );
+	public function test_reconstructs_formatting_elements() {
+		$processor = WP_HTML_Processor::create_fragment( '<p><em>One<p><em><span>Two<p><em>Three<p><em>Four' );
 
 		$this->assertTrue( $processor->next_tag( 'EM' ), 'Could not find first EM.' );
-		$this->assertFalse( $processor->next_tag( 'EM' ), 'Should have aborted before finding second EM as it required reconstructing the first EM.' );
+		$this->assertSame( array( 'HTML', 'BODY', 'P', 'EM' ), $processor->get_breadcrumbs(), 'Found incorrect breadcrumbs for first EM.' );
+		$this->assertTrue( $processor->next_tag( 'SPAN' ), 'Could not find test span.' );
+		$this->assertSame(
+			array( 'HTML', 'BODY', 'P', 'EM', 'EM', 'SPAN' ),
+			$processor->get_breadcrumbs(),
+			'Found incorrect breadcrumbs for test SPAN; should have created two EMs.'
+		);
 	}
 
 	/**

tests/phpunit/tests/html-api/wpHtmlProcessorBreadcrumbs.php

@@ -219,45 +219,55 @@ public static function data_unsupported_elements() {
 	}
 
 	/**
-	 * @ticket 58517
-	 *
-	 * @dataProvider data_unsupported_markup
+	 * Ensures that formats inside unclosed A elements are reconstructed.
 	 *
-	 * @param string $html HTML containing unsupported markup.
+	 * @ticket 61576
 	 */
-	public function test_fails_when_encountering_unsupported_markup( $html, $description ) {
-		$processor = WP_HTML_Processor::create_fragment( $html );
-
-		while ( $processor->next_token() && null === $processor->get_attribute( 'supported' ) ) {
-			continue;
-		}
+	public function test_reconstructs_formatting_from_unclosed_a_elements() {
+		$processor = WP_HTML_Processor::create_fragment( '<a><strong>Click <a><big>Here</big></a></strong></a>' );
 
-		$this->assertNull(
-			$processor->get_last_error(),
-			'Bailed on unsupported input before finding supported checkpoint: check test code.'
+		$processor->next_tag( 'STRONG' );
+		$this->assertSame(
+			array( 'HTML', 'BODY', 'A', 'STRONG' ),
+			$processor->get_breadcrumbs(),
+			'Failed to construct starting breadcrumbs properly.'
 		);
 
-		$this->assertTrue( $processor->get_attribute( 'supported' ), 'Did not find required supported element.' );
-		$processor->next_token();
-		$this->assertNotNull( $processor->get_last_error(), "Didn't properly reject unsupported markup: {$description}" );
+		$processor->next_tag( 'BIG' );
+		$this->assertSame(
+			array( 'HTML', 'BODY', 'STRONG', 'A', 'BIG' ),
+			$processor->get_breadcrumbs(),
+			'Failed to reconstruct the active formatting elements after an unclosed A element.'
+		);
 	}
 
 	/**
-	 * Data provider.
+	 * Ensures that unclosed A elements are reconstructed.
 	 *
-	 * @return array[]
+	 * @ticket 61576
 	 */
-	public static function data_unsupported_markup() {
-		return array(
-			'A with formatting following unclosed A' => array(
-				'<a><strong>Click <span supported><a unsupported><big>Here</big></a></strong></a>',
-				'Unclosed formatting requires complicated reconstruction.',
-			),
+	public function test_reconstructs_unclosed_a_elements() {
+		$processor = WP_HTML_Processor::create_fragment( '<a><div><a></div></a>' );
 
-			'A after unclosed A inside DIV'          => array(
-				'<a><div supported><a unsupported></div></a>',
-				'A is a formatting element, which requires more complicated reconstruction.',
-			),
+		$processor->next_tag( 'DIV' );
+		$this->assertSame(
+			array( 'HTML', 'BODY', 'DIV' ),
+			$processor->get_breadcrumbs(),
+			'Failed to construct breadcrumbs properly - the DIV should have closed the A element.'
+		);
+
+		// When the DIV re-opens, it reconstructs an unclosed A, then the A in the text is a second A.
+		$processor->next_tag( 'A' );
+		$this->assertSame(
+			array( 'HTML', 'BODY', 'DIV', 'A' ),
+			'Failed to create proper breadcrumbs for recreated A element.'
+		);
+
+		// This is the one that's second in the raw text.
+		$processor->next_tag( 'A' );
+		$this->assertSame(
+			array( 'HTML', 'BODY', 'DIV', 'A' ),
+			'Failed to create proper breadcrumbs for explicit A element - this A should have closed the reconstructed A.'
 		);
 	}