Add unit tests for oembed autodiscover and thumbnail detection

oembed/Oembed.php

@@ -101,7 +101,8 @@ class Oembed {
 	 * @param $url Human readable URL.
 	 * @returns string/bool Oembed URL, or false.
 	 */
-	protected static function autodiscover_from_url($url) {
+	protected static function autodiscover_from_url($url)
+	{
 		// Fetch the URL (cache for a week by default)
 		$service = new RestfulService($url, 60 * 60 * 24 * 7);
 		$body = $service->request();
@@ -109,19 +110,28 @@ class Oembed {
 			return false;
 		}
 		$body = $body->getBody();
+		return static::autodiscover_from_body($body);
+	}
 
+	/**
+	 * Given a response body, determine if there is an autodiscover url
+	 *
+	 * @param string $body
+	 * @return bool|string
+	 */
+	public static function autodiscover_from_body($body) {
 		// Look within the body for an oembed link.
-		$pcreOmbed = '#<link[^>]+?(?:href=[\'"](.+?)[\'"][^>]+?)'
+		$pcreOmbed = '#<link[^>]+?(?:href=[\'"](?<first>[^\'"]+?)[\'"][^>]+?)'
 			. '?type=["\']application/json\+oembed["\']'
-			. '(?:[^>]+?href=[\'"](.+?)[\'"])?#';
+			. '(?:[^>]+?href=[\'"](?<second>[^\'"]+?)[\'"])?#';
 
 		if(preg_match_all($pcreOmbed, $body, $matches, PREG_SET_ORDER)) {
 			$match = $matches[0];
-			if(!empty($match[2])) {
+			if(!empty($match['second'])) {
-				return html_entity_decode($match[2]);
+				return html_entity_decode($match['second']);
 			}
-			if(!empty($match[1])) {
+			if(!empty($match['first'])) {
-				return html_entity_decode($match[1]);
+				return html_entity_decode($match['first']);
 			}
 		}
 		return false;
@@ -297,9 +307,9 @@ class Oembed_Result extends ViewableData {
 		// Convert all keys to lowercase
 		$data = array_change_key_case($data, CASE_LOWER);
 
-		if( isset($data['provider_name']) && $data['provider_name'] == "Facebook" ) {
+		// Check if we can guess thumbnail
-			$id = preg_replace("/.*\/(\d+?)\/?($|\?.*)/", "$1", $data["url"]);
+		if(empty($data['thumbnail_url']) && $thumbnail = $this->findThumbnail($data)) {
-			$data['thumbnail_url'] = "https://graph.facebook.com/{$id}/picture";
+			$data['thumbnail_url'] = $thumbnail;
 		}
 
 		// Purge everything if the type does not match.
@@ -310,6 +320,27 @@ class Oembed_Result extends ViewableData {
 		$this->data = $data;
 	}
 
+	/**
+	 * Find thumbnail if omitted from data
+	 *
+	 * @param array $data
+	 * @return string
+	 */
+	public function findThumbnail($data) {
+		if(!empty($data['thumbnail_url'])) {
+			return $data['thumbnail_url'];
+		}
+
+		// Hack in facebook graph thumbnail
+		if(!empty($data['provider_name']) && $data['provider_name'] === 'Facebook') {
+			$id = preg_replace("/.*\\/(\\d+?)\\/?($|\\?.*)/", "$1", $data["url"]);
+			return "https://graph.facebook.com/{$id}/picture";
+		}
+
+		// no thumbnail found
+		return null;
+	}
+
 	/**
 	 * Wrap the check for looking into Oembed JSON within $this->data.
 	 */

tests/oembed/OembedTest.php

@@ -38,6 +38,70 @@ class OembedTest extends SapphireTest {
 		$this->assertEquals($query['maxwidth'], 'bar', 'Magically creates maxwidth option');
 	}
 
+	public function testAutodiscover() {
+		// Test href after type tag
+		$body = <<<EOS
+<title id="pageTitle">Some content</title>
+<link rel="search" type="application/opensearchdescription+xml" href="/osd.xml" title="Facebook" />
+<link
+	rel="alternate" type="application/json+oembed"
+	href="https://www.facebook.com/plugins/post/oembed.json/?url=https%3A%2F%2Fwww.facebook.com%2Fsomeusername%2Fposts%2F10209305859558135"
+	title="My post"
+/>
+EOS;
+		$this->assertEquals(
+			'https://www.facebook.com/plugins/post/oembed.json/?url=https%3A%2F%2Fwww.facebook.com%2Fsomeusername%2Fposts%2F10209305859558135',
+			Oembed::autodiscover_from_body($body)
+		);
+
+		// Test href before the type tag
+		$body2 = <<<EOS
+<title id="pageTitle">Some content</title>
+<link rel="search" type="application/opensearchdescription+xml" href="/osd.xml" title="Facebook" />
+<link
+	href="https://www.facebook.com/plugins/post/oembed.json/?url=https%3A%2F%2Fwww.facebook.com%2Fsomeusername%2Fposts%2F10209305859558135"
+	rel="alternate" type="application/json+oembed"
+
+	title="My post"
+/>
+EOS;
+		$this->assertEquals(
+			'https://www.facebook.com/plugins/post/oembed.json/?url=https%3A%2F%2Fwww.facebook.com%2Fsomeusername%2Fposts%2F10209305859558135',
+			Oembed::autodiscover_from_body($body2)
+		);
+	}
+
+	public function testFindThumbnail()
+	{
+		$data = array(
+			"author_name"=> "Some User",
+			"author_url"=> null,
+			"provider_url" => "https://www.facebook.com",
+			"provider_name" => "Facebook",
+			"success" => true,
+			"height" => null,
+			"html" => "<div />",
+			"type" => "rich",
+			"version" => "1.0",
+			"url" => "https://www.facebook.com/someuser/posts/6465132161654421654",
+			"width" => 552
+		);
+
+		// Test facebook url
+		$result = new Oembed_Result('https://www.facebook.com/someuser/posts/6465132161654421654');
+		$this->assertEquals(
+			"https://graph.facebook.com/6465132161654421654/picture",
+			$result->findThumbnail($data)
+		);
+
+		// Test respect existing url
+		$data['thumbnail_url'] = 'http://www.silverstripe.com/picture.jpg';
+		$this->assertEquals(
+			"http://www.silverstripe.com/picture.jpg",
+			$result->findThumbnail($data)
+		);
+	}
+
 	public function testRequestProtocolReflectedInGetOembedFromUrl() {
 		Config::inst()->update('Oembed', 'providers', array(
 			'http://*.silverstripe.com/watch*'=> array(