HTML API: Prevent HTML newline normalization on foreign elements.

HTML and foreign element normalization differ in some cases. Ensure the HTML-specific newline injection is not applied to foreign elements like `svg:textarea`.

Developed in WordPress#12322.

Follow-up to [61747].

See #65372.


git-svn-id: https://develop.svn.wordpress.org/trunk@62574 602fd350-edb4-49c9-b593-d223f7449a82

Author: sirreal

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

@@ -1468,8 +1468,8 @@ public function serialize_token(): string {
 
 		/*
 		 * The HTML parser strips a leading newline immediately after the start
-		 * tag of TEXTAREA, PRE, and LISTING elements. When serializing, prepend
-		 * a leading newline to ensure the semantic HTML content is preserved.
+		 * tag of TEXTAREA, PRE, and LISTING elements in HTML content. When serializing,
+		 * prepend a leading newline to ensure the semantic HTML content is preserved.
 		 *
 		 * For example, `<pre>\n\nX</pre>` must not become `<pre>\nX</pre>` because its content
 		 * has changed. However, `<pre>X</pre>` and `<pre>\nX</pre>` are _equivalent_.
@@ -1488,7 +1488,7 @@ public function serialize_token(): string {
 		 *
 		 * @see https://html.spec.whatwg.org/multipage/parsing.html
 		 */
-		if ( 'TEXTAREA' === $tag_name || 'PRE' === $tag_name || 'LISTING' === $tag_name ) {
+		if ( $in_html && ( 'TEXTAREA' === $tag_name || 'PRE' === $tag_name || 'LISTING' === $tag_name ) ) {
 			$html .= "\n";
 		}
 

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

@@ -463,21 +463,30 @@ public static function data_provider_serialize_doctype() {
 	}
 
 	/**
-	 * Ensures that leading newlines in PRE, LISTING, and TEXTAREA elements are preserved upon normalization,
-	 * and that normalization is idempotent in these cases.
+	 * Ensures that leading newlines in PRE, LISTING, and TEXTAREA elements are normalized
+	 * according to their parsing namespace, and that normalization is idempotent in these cases.
 	 *
 	 * @ticket 64607
 	 *
 	 * @dataProvider data_provider_normalize_special_leading_newline_cases
 	 *
 	 * @param string $input    HTML input containing leading newlines in PRE, LISTING, or TEXTAREA elements.
-	 * @param string $expected Expected output after normalization, which should preserve leading newlines.
+	 * @param string $expected Expected exact output after normalization.
 	 */
 	public function test_normalize_special_leading_newline_handling( string $input, string $expected ) {
 		$normalized = WP_HTML_Processor::normalize( $input );
-		$this->assertEqualHTML( $expected, $normalized );
+
+		/*
+		 * Byte equality pins normalize()'s serialized form; HTML equality verifies
+		 * semantic equivalence. This distinction matters because HTML parsing ignores
+		 * one leading LF after PRE, LISTING, and TEXTAREA start tags.
+		 */
+		$this->assertSame( $expected, $normalized );
+		$this->assertEqualHTML( $input, $normalized );
+
 		$normalized_twice = WP_HTML_Processor::normalize( $normalized );
-		$this->assertEqualHTML( $expected, $normalized_twice );
+		$this->assertSame( $expected, $normalized_twice );
+		$this->assertEqualHTML( $normalized, $normalized_twice );
 	}
 
 	/**
@@ -653,50 +662,66 @@ public static function data_provider_normalized_fuzzer_cases_that_should_be_idem
 	/**
 	 * Data provider.
 	 *
-	 * @return array[]
+	 * @return array<string, array{string, string}>
 	 */
-	public static function data_provider_normalize_special_leading_newline_cases() {
+	public static function data_provider_normalize_special_leading_newline_cases(): array {
 		return array(
 			'Leading newline in PRE'             => array(
 				"<pre>\nline 1\nline 2</pre>",
-				"<pre>line 1\nline 2</pre>",
+				"<pre>\nline 1\nline 2</pre>",
 			),
 			'Double leading newline in PRE'      => array(
 				"<pre>\n\nline 2\nline 3</pre>",
 				"<pre>\n\nline 2\nline 3</pre>",
 			),
 			'Multiple text nodes inside PRE'     => array(
 				"<pre>\nline 1<!--comment--> still line 1</pre>",
-				'<pre>line 1<!--comment--> still line 1</pre>',
+				"<pre>\nline 1<!--comment--> still line 1</pre>",
 			),
 			'Multiple text nodes inside PRE with leading newlines' => array(
 				"<pre>\n\nline 2<!--comment--> still line 2</pre>",
 				"<pre>\n\nline 2<!--comment--> still line 2</pre>",
 			),
 			'Leading newline in LISTING'         => array(
 				"<listing>\nline 1\nline 2</listing>",
-				"<listing>line 1\nline 2</listing>",
+				"<listing>\nline 1\nline 2</listing>",
 			),
 			'Double leading newline in LISTING'  => array(
 				"<listing>\n\nline 2\nline 3</listing>",
 				"<listing>\n\nline 2\nline 3</listing>",
 			),
 			'Multiple text nodes inside LISTING' => array(
 				"<listing>\nline 1<!--comment--> still line 1</listing>",
-				'<listing>line 1<!--comment--> still line 1</listing>',
+				"<listing>\nline 1<!--comment--> still line 1</listing>",
 			),
 			'Multiple text nodes inside LISTING with leading newlines' => array(
 				"<listing>\n\nline 2<!--comment--> still line 2</listing>",
 				"<listing>\n\nline 2<!--comment--> still line 2</listing>",
 			),
 			'Leading newline in TEXTAREA'        => array(
 				"<textarea>\nline 1\nline 2</textarea>",
-				"<textarea>line 1\nline 2</textarea>",
+				"<textarea>\nline 1\nline 2</textarea>",
 			),
 			'Double leading newline in TEXTAREA' => array(
 				"<textarea>\n\nline 2\nline 3</textarea>",
 				"<textarea>\n\nline 2\nline 3</textarea>",
 			),
+			'Foreign MathML TEXTAREA does not ignore leading newlines' => array(
+				'<math><textarea>X</textarea></math>',
+				'<math><textarea>X</textarea></math>',
+			),
+			'Foreign MathML TEXTAREA preserves leading newline' => array(
+				"<math><textarea>\nX</textarea></math>",
+				"<math><textarea>\nX</textarea></math>",
+			),
+			'Foreign SVG TEXTAREA does not ignore leading newlines' => array(
+				'<svg><textarea>X</textarea></svg>',
+				'<svg><textarea>X</textarea></svg>',
+			),
+			'Foreign SVG TEXTAREA preserves leading newline' => array(
+				"<svg><textarea>\nX</textarea></svg>",
+				"<svg><textarea>\nX</textarea></svg>",
+			),
 		);
 	}
 }