Merge pull request ampproject#756 from Automattic/fix/normalized-dime…

…nsion-urls

Dimension Extractor should return original URLs

includes/utils/class-amp-image-dimension-extractor.php

@@ -10,18 +10,32 @@ static public function extract( $urls ) {
 			self::register_callbacks();
 		}
 
-		$valid_urls = array();
-		foreach ( $urls as $url ) {
-			$url = self::normalize_url( $url );
-			if ( false !== $url ) {
-				$valid_urls[] = $url;
+		$return_dimensions = array();
+
+		// Normalize URLs and also track a map of normalized-to-original as we'll need it to reformat things when returning the data.
+		$url_map = array();
+		$normalized_urls = array();
+		foreach ( $urls as $original_url ) {
+			$normalized_url = self::normalize_url( $original_url );
+			if ( false !== $normalized_url ) {
+				$url_map[ $normalized_url ] = $original_url;
+				$normalized_urls[] = $normalized_url;
+			} else {
+				// This is not a URL we can extract dimensions from, so default to false.
+				$return_dimensions[ $original_url ] = false;
 			}
 		}
 
-		$dimensions = array_fill_keys( $valid_urls, false );
-		$dimensions = apply_filters( 'amp_extract_image_dimensions_batch', $dimensions );
+		$extracted_dimensions = array_fill_keys( $normalized_urls, false );
+		$extracted_dimensions = apply_filters( 'amp_extract_image_dimensions_batch', $extracted_dimensions );
 
-		return $dimensions;
+		// We need to return a map with the original (un-normalized URL) as we that to match nodes that need dimensions.
+		foreach ( $extracted_dimensions as $normalized_url => $dimension ) {
+			$original_url = $url_map[ $normalized_url ];
+			$return_dimensions[ $original_url ] = $dimension;
+		}
+
+		return $return_dimensions;
 	}
 
 	public static function normalize_url( $url ) {

tests/test-amp-image-dimension-extractor.php

@@ -7,6 +7,39 @@
 define( 'IMG_350', 'http://i0.wp.com/amptest.files.wordpress.com/2017/03/350x150.png' );
 define( 'IMG_1024', 'http://i0.wp.com/amptest.files.wordpress.com/2017/03/1024x768.png' );
 
+class AMP_Image_Dimension_Extractor__Extract__Test extends
+WP_UnitTestCase {
+	public function setUp() {
+		parent::setUp();
+
+		// We don't want to actually download the images; just testing the extract method
+		add_action( 'amp_extract_image_dimensions_batch_callbacks_registered', array( $this, 'disable_downloads' ) );
+	}
+
+	public function disable_downloads() {
+		remove_all_filters( 'amp_extract_image_dimensions_batch' );
+	}
+
+	public function test__should_return_original_urls() {
+		$source_urls = array(
+			'https://example.com',
+			'//example.com/no-protocol',
+			'/absolute-url/no-host',
+			'data:image/gif;base64,R0lGODl...', // can't normalize
+		);
+		$expected = array(
+			'https://example.com' => false,
+			'//example.com/no-protocol' => false,
+			'/absolute-url/no-host' => false,
+			'data:image/gif;base64,R0lGODl...' => false,
+		);
+
+		$actual = AMP_Image_Dimension_Extractor::extract( $source_urls );
+
+		$this->assertEquals( $expected, $actual );
+	}
+}
+
 class AMP_Image_Dimension_Extractor__Normalize_URL__Test extends WP_UnitTestCase {
 	function get_data() {
 		$site_url = site_url();