Jetpack Sync: fix extracting UTF-8 characters from image alt text (#4…

…1451)
Automattic · Feb 4, 2025 · 6c87a8b · 6c87a8b
1 parent b83c45c
commit 6c87a8b
Show file tree

Hide file tree

Showing 3 changed files with 26 additions and 1 deletion.
diff --git a/projects/plugins/jetpack/changelog/fix-media-extractor-image-alt-text-encoding b/projects/plugins/jetpack/changelog/fix-media-extractor-image-alt-text-encoding
@@ -0,0 +1,4 @@
+Significance: patch
+Type: other
+
+Jetpack Sync: Fixed extracting UTF-8 characters from image alt-text
diff --git a/projects/plugins/jetpack/class.jetpack-post-images.php b/projects/plugins/jetpack/class.jetpack-post-images.php
@@ -495,10 +495,18 @@ public static function from_html( $html_or_id, $width = 200, $height = 200 ) {
 		// Let's grab all image tags from the HTML.
 		$dom_doc = new DOMDocument();
 
+		// DOMDocument defaults to ISO-8859 because we're loading only the post content, without head tag.
+		// Fix: Enforce encoding with meta tag.
+		$charset = get_option( 'blog_charset' );
+		if ( empty( $charset ) || ! preg_match( '/^[a-zA-Z0-9_-]+$/', $charset ) ) {
+			$charset = 'UTF-8';
+		}
+		$html_prefix = sprintf( '<meta http-equiv="Content-Type" content="text/html; charset=%s">', esc_attr( $charset ) );
+
 		// The @ is not enough to suppress errors when dealing with libxml,
 		// we have to tell it directly how we want to handle errors.
 		libxml_use_internal_errors( true );
-		@$dom_doc->loadHTML( $html_info['html'] ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
+		@$dom_doc->loadHTML( $html_prefix . $html_info['html'] ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
 		libxml_use_internal_errors( false );
 
 		$image_tags = $dom_doc->getElementsByTagName( 'img' );

diff --git a/projects/plugins/jetpack/tests/php/media/test-class.jetpack-post-images.php b/projects/plugins/jetpack/tests/php/media/test-class.jetpack-post-images.php
@@ -83,6 +83,19 @@ public function test_from_html_no_size() {
 		$this->assertEquals( array(), $result );
 	}
 
+	/**
+	 * @covers Jetpack_PostImages::from_html
+	 */
+	public function test_from_html_alt_utf8() {
+		$s = '<img src="bob.jpg" width="200" height="200" alt="Ḽơᶉëᶆ ȋṕšᶙṁ ḍỡḽǭᵳ ʂǐť ӓṁệẗ" />';
+
+		$result = Jetpack_PostImages::from_html( $s );
+
+		$this->assertIsArray( $result );
+		$this->assertNotEmpty( $result );
+		$this->assertEquals( 'Ḽơᶉëᶆ ȋṕšᶙṁ ḍỡḽǭᵳ ʂǐť ӓṁệẗ', $result[0]['alt_text'] );
+	}
+
 	/**
 	 * @author scotchfield
 	 * @covers Jetpack_PostImages::from_slideshow