From a67542b691044470a8b6d2295efd19dec169b2b5 Mon Sep 17 00:00:00 2001
From: gunjanjaswal <gunjanjaswal@gmail.com>
Date: Thu, 13 Aug 2026 19:03:28 +0530
Subject: [PATCH 1/3] Formatting: make get_tag_regex() match tags non-greedily.

get_tag_regex() built a pattern with a greedy body match, so content with more
than one tag of the same type matched from the first opening tag through the
last closing tag, merging every tag in between (and the text between them) into
a single match.

Make the attribute and body matches lazy, aligning get_tag_regex() with the
pattern get_media_embedded_in_content() already uses. Adds unit tests for
get_tag_regex(), which previously had none, and refreshes the stale patch.

Props kopepasah, afercia.
Fixes #26674.
---
 src/wp-includes/functions.php                 |  2 +-
 tests/phpunit/tests/functions/getTagRegex.php | 87 +++++++++++++++++++
 2 files changed, 88 insertions(+), 1 deletion(-)
 create mode 100644 tests/phpunit/tests/functions/getTagRegex.php

diff --git a/src/wp-includes/functions.php b/src/wp-includes/functions.php
index f5002a45de1e8..80880c3121a0b 100644
--- a/src/wp-includes/functions.php
+++ b/src/wp-includes/functions.php
@@ -7679,7 +7679,7 @@ function get_tag_regex( $tag ) {
 	if ( empty( $tag ) ) {
 		return '';
 	}
-	return sprintf( '<%1$s[^<]*(?:>[\s\S]*<\/%1$s>|\s*\/>)', tag_escape( $tag ) );
+	return sprintf( '<%1$s[^<]*?(?:>[\s\S]*?<\/%1$s>|\s*\/>)', tag_escape( $tag ) );
 }
 
 /**
diff --git a/tests/phpunit/tests/functions/getTagRegex.php b/tests/phpunit/tests/functions/getTagRegex.php
new file mode 100644
index 0000000000000..45c3c883dd3ae
--- /dev/null
+++ b/tests/phpunit/tests/functions/getTagRegex.php
@@ -0,0 +1,87 @@
+<?php
+
+/**
+ * Tests for the get_tag_regex() function.
+ *
+ * @group functions
+ *
+ * @covers ::get_tag_regex
+ */
+class Tests_Functions_GetTagRegex extends WP_UnitTestCase {
+
+	/**
+	 * @ticket 26674
+	 *
+	 * @dataProvider data_get_tag_regex_matches
+	 *
+	 * @param string   $tag      Tag name to build the regex for.
+	 * @param string   $content  Content to match against.
+	 * @param string[] $expected Expected full matches, in order.
+	 */
+	public function test_get_tag_regex_matches( $tag, $content, $expected ) {
+		preg_match_all( '#' . get_tag_regex( $tag ) . '#', $content, $matches );
+		$this->assertSame( $expected, $matches[0] );
+	}
+
+	/**
+	 * Data provider.
+	 *
+	 * @return array[]
+	 */
+	public function data_get_tag_regex_matches() {
+		return array(
+			'a single tag with a body'                  => array(
+				'iframe',
+				'<iframe src="https://example.com/a"></iframe>',
+				array( '<iframe src="https://example.com/a"></iframe>' ),
+			),
+
+			// The regression: a greedy match ran from the first opening tag to
+			// the last closing tag, merging both embeds and the text between
+			// them into a single match. See #26674.
+			'two adjacent tags are matched separately'  => array(
+				'iframe',
+				'<iframe src="https://example.com/a"></iframe> text <iframe src="https://example.com/b"></iframe>',
+				array(
+					'<iframe src="https://example.com/a"></iframe>',
+					'<iframe src="https://example.com/b"></iframe>',
+				),
+			),
+
+			'a self-closing tag'                        => array(
+				'iframe',
+				'<iframe src="https://example.com/a" />',
+				array( '<iframe src="https://example.com/a" />' ),
+			),
+
+			'a tag with a multiline body'               => array(
+				'video',
+				"<video>\n<source src=\"a.mp4\">\n</video>",
+				array( "<video>\n<source src=\"a.mp4\">\n</video>" ),
+			),
+
+			'no match when the tag is absent'           => array(
+				'iframe',
+				'<p>No embeds here.</p>',
+				array(),
+			),
+		);
+	}
+
+	/**
+	 * @ticket 26674
+	 */
+	public function test_get_tag_regex_returns_empty_string_for_empty_tag() {
+		$this->assertSame( '', get_tag_regex( '' ) );
+	}
+
+	/**
+	 * The tag name is passed through tag_escape(), so casing and invalid
+	 * characters do not change the generated pattern.
+	 *
+	 * @ticket 26674
+	 */
+	public function test_get_tag_regex_escapes_the_tag_name() {
+		$this->assertSame( get_tag_regex( 'iframe' ), get_tag_regex( 'IFRAME' ) );
+	}
+}

From 7fa9a4114ab389d37f1434d3a1f10367c6c13e89 Mon Sep 17 00:00:00 2001
From: gunjanjaswal <gunjanjaswal@gmail.com>
Date: Thu, 13 Aug 2026 22:55:13 +0530
Subject: [PATCH 2/3] Align data provider array arrows to satisfy WPCS.

---
 tests/phpunit/tests/functions/getTagRegex.php | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/tests/phpunit/tests/functions/getTagRegex.php b/tests/phpunit/tests/functions/getTagRegex.php
index 45c3c883dd3ae..0d3d7a8a85ed3 100644
--- a/tests/phpunit/tests/functions/getTagRegex.php
+++ b/tests/phpunit/tests/functions/getTagRegex.php
@@ -30,7 +30,7 @@ public function test_get_tag_regex_matches( $tag, $content, $expected ) {
 	 */
 	public function data_get_tag_regex_matches() {
 		return array(
-			'a single tag with a body'                  => array(
+			'a single tag with a body'                 => array(
 				'iframe',
 				'<iframe src="https://example.com/a"></iframe>',
 				array( '<iframe src="https://example.com/a"></iframe>' ),
@@ -39,7 +39,7 @@ public function data_get_tag_regex_matches() {
 			// The regression: a greedy match ran from the first opening tag to
 			// the last closing tag, merging both embeds and the text between
 			// them into a single match. See #26674.
-			'two adjacent tags are matched separately'  => array(
+			'two adjacent tags are matched separately' => array(
 				'iframe',
 				'<iframe src="https://example.com/a"></iframe> text <iframe src="https://example.com/b"></iframe>',
 				array(
@@ -48,19 +48,19 @@ public function data_get_tag_regex_matches() {
 				),
 			),
 
-			'a self-closing tag'                        => array(
+			'a self-closing tag'                       => array(
 				'iframe',
 				'<iframe src="https://example.com/a" />',
 				array( '<iframe src="https://example.com/a" />' ),
 			),
 
-			'a tag with a multiline body'               => array(
+			'a tag with a multiline body'              => array(
 				'video',
 				"<video>\n<source src=\"a.mp4\">\n</video>",
 				array( "<video>\n<source src=\"a.mp4\">\n</video>" ),
 			),
 
-			'no match when the tag is absent'           => array(
+			'no match when the tag is absent'          => array(
 				'iframe',
 				'<p>No embeds here.</p>',
 				array(),

From d805e4e3e998f1899f9b81320476f1b814e72ac4 Mon Sep 17 00:00:00 2001
From: gunjanjaswal <gunjanjaswal@gmail.com>
Date: Thu, 13 Aug 2026 23:48:50 +0530
Subject: [PATCH 3/3] Tests: cover a self-closing void element with and without
 a trailing space.

Per review on #26674: an iframe cannot self-close, so exercise the
self-closing branch with an input, both '<input />' and '<input/>'.
---
 tests/phpunit/tests/functions/getTagRegex.php | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

diff --git a/tests/phpunit/tests/functions/getTagRegex.php b/tests/phpunit/tests/functions/getTagRegex.php
index 0d3d7a8a85ed3..8988de987548b 100644
--- a/tests/phpunit/tests/functions/getTagRegex.php
+++ b/tests/phpunit/tests/functions/getTagRegex.php
@@ -48,10 +48,18 @@ public function data_get_tag_regex_matches() {
 				),
 			),
 
-			'a self-closing tag'                       => array(
-				'iframe',
-				'<iframe src="https://example.com/a" />',
-				array( '<iframe src="https://example.com/a" />' ),
+			// A self-closing void element (an iframe cannot self-close), matched
+			// both with and without the space before the slash (comment:16).
+			'a self-closing tag with a space'          => array(
+				'input',
+				'<input type="text" />',
+				array( '<input type="text" />' ),
+			),
+
+			'a self-closing tag without a space'       => array(
+				'input',
+				'<input type="text"/>',
+				array( '<input type="text"/>' ),
 			),
 
 			'a tag with a multiline body'              => array(