Image CDN: URL encode path parts (#39560)

* rawurlencode each path part in the image URL * Fix a typo * Add a test * changelog * Add comment to clarify space usage --------- Co-authored-by: Peter Petrov <peter.petrov89@gmail.com>
Automattic · Oct 8, 2024 · 061b353 · 061b353
1 parent 155c6a3
commit 061b353
Show file tree

Hide file tree

Showing 5 changed files with 41 additions and 2 deletions.
diff --git a/projects/packages/image-cdn/changelog/fix-photon-url-encoding b/projects/packages/image-cdn/changelog/fix-photon-url-encoding
@@ -0,0 +1,4 @@
+Significance: patch
+Type: fixed
+
+URL encode path parts of an image
diff --git a/projects/packages/image-cdn/src/class-image-cdn-core.php b/projects/packages/image-cdn/src/class-image-cdn-core.php
@@ -191,8 +191,7 @@ public static function cdn_url( $image_url, $args = array(), $scheme = null ) {
 			}
 		}
 
-		$image_host_path = $image_url_parts['host'] . $image_url_parts['path'];
-
+		$image_host_path = $image_url_parts['host'] . static::escape_path( $image_url_parts['path'] );
 		/**
 		 * Filters the domain used by the Photon module.
 		 *
@@ -262,6 +261,22 @@ public static function is_cdn_url( $url ) {
 			|| wp_parse_url( $custom_photon_url, PHP_URL_HOST ) === $parsed_url['host'];
 	}
 
+	/**
+	 * URL-encodes each path component.
+	 *
+	 * Example:
+	 * Input: "foo/bar baz/baz"
+	 * Output: "foo/bar%20baz/baz"
+	 *
+	 * @param string $path The path to escape.
+	 * @return string The escaped path.
+	 */
+	private static function escape_path( $path ) {
+		$parts = explode( '/', $path );
+		$parts = array_map( 'rawurlencode', $parts );
+		return implode( '/', $parts );
+	}
+
 	/**
 	 * Parses WP.com-hosted image args to replicate the crop.
 	 *

diff --git a/projects/packages/image-cdn/tests/php/test_class.image_cdn_core.php b/projects/packages/image-cdn/tests/php/test_class.image_cdn_core.php
@@ -283,6 +283,18 @@ public function test_is_cdn_url_method() {
 		$this->assertFalse( Image_CDN_Core::is_cdn_url( '//example.com/img.jpg' ) );
 	}
 
+	/**
+	 * @covers ::Image_CDN_Core::cdn_url
+	 * @since  $$next-version$$
+	 * @group  jetpack_photon_filter_url_encoding
+	 */
+	public function test_photon_url_filter_url_encodes_path_parts() {
+		// The first two spaces are not standard spaces - https://www.compart.com/en/unicode/U+202F
+		$url = Image_CDN_Core::cdn_url( '//example.com/narrow no-break space/name with spaces.jpg', array(), 'https' );
+
+		$this->assertEquals( 'https://i0.wp.com/example.com/narrow%E2%80%AFno-break%E2%80%AFspace/name%20with%20spaces.jpg', $url );
+	}
+
 	/**
 	 * @author aduth
 	 * @covers ::Image_CDN_Core::cdn_url_scheme

diff --git a/projects/plugins/boost/changelog/fix-photon-url-encoding b/projects/plugins/boost/changelog/fix-photon-url-encoding
@@ -0,0 +1,4 @@
+Significance: patch
+Type: fixed
+
+Image CDN: URL encode image path parts for RSS feed compatibility
diff --git a/projects/plugins/jetpack/changelog/fix-photon-url-encoding b/projects/plugins/jetpack/changelog/fix-photon-url-encoding
@@ -0,0 +1,4 @@
+Significance: patch
+Type: compat
+
+Image CDN: URL encode image path parts for RSS feed compatibility