Improve handling of feeds when discover encounters feel links without…

… a protocol (#405) Signed-off-by: Accalia <Accalia@Elementia.me> Signed-off-by: Accalia <Accalia@Elementia.me> Co-authored-by: alexdebril <alexdebril@users.noreply.github.com>
alexdebril · Oct 26, 2022 · 5808bc2 · 5808bc2
1 parent 63dd6ea
commit 5808bc2
Show file tree

Hide file tree

Showing 4 changed files with 72 additions and 4 deletions.
diff --git a/src/FeedIo/Explorer.php b/src/FeedIo/Explorer.php
@@ -45,7 +45,15 @@ protected function extractFeeds(string $html): array
         $feeds = [];
         foreach ($links as $link) {
             if ($this->isFeedLink($link)) {
-                $feeds[] = $link->getAttribute('href');
+                $href = $link->getAttribute('href');
+                if (strpos($href, '//') === 0) {
+                    // Link href is protocol-less, Implies feed supports http
+                    // and https. Consumers will often assume that feed url
+                    // includes protocol, so we will assign a protocol before
+                    // returning
+                    $href = 'https:' . $href;
+                }
+                $feeds[] = $href;
             }
         }
 

diff --git a/tests/FeedIo/ExplorerTest.php b/tests/FeedIo/ExplorerTest.php
@@ -26,19 +26,24 @@ class ExplorerTest extends TestCase
     protected $object;
 
     public function setUp(): void
+    {
+
+    }
+
+    protected function createTestObject(string $htmlPath)
     {
         $this->object = new Explorer(
-            $this->getClientMock(),
+            $this->getClientMock($htmlPath),
             new NullLogger()
         );
     }
 
     /**
      * @return \FeedIo\Adapter\ClientInterface
      */
-    protected function getClientMock()
+    protected function getClientMock(string $htmlPath)
     {
-        $html = file_get_contents(dirname(__FILE__)."/../samples/discovery.html");
+        $html = file_get_contents(dirname(__FILE__) . $htmlPath);
         $client = $this->createMock('FeedIo\Adapter\ClientInterface');
         $response = $this->createMock('FeedIo\Adapter\ResponseInterface');
         $response->expects($this->any())->method('getBody')->will($this->returnValue($html));
@@ -52,8 +57,31 @@ protected function getClientMock()
      */
     public function testDiscover()
     {
+        $this->createTestObject("/../samples/discovery.html");
         $feeds = $this->object->discover('https://exmple.org/feed.atom');
 
         $this->assertEquals(['http://example.org/feed.xml', 'http://example.org/comments.xml'], $feeds);
     }
+
+    /**
+     * @covers \FeedIo\Reader::addParser
+     */
+    public function testDiscoverHttps()
+    {
+        $this->createTestObject("/../samples/discovery-https.html");
+        $feeds = $this->object->discover('https://exmple.org/feed.atom');
+
+        $this->assertEquals(['https://example.org/feed.xml', 'https://example.org/comments.xml'], $feeds);
+    }
+
+    /**
+     * @covers \FeedIo\Reader::addParser
+     */
+    public function testDiscoverProtocolless()
+    {
+        $this->createTestObject("/../samples/discovery-mixed.html");
+        $feeds = $this->object->discover('https://exmple.org/feed.atom');
+
+        $this->assertEquals(['https://example.org/feed.xml', 'https://example.org/comments.xml'], $feeds);
+    }
 }
diff --git a/tests/samples/discovery-dualprotocol.html b/tests/samples/discovery-dualprotocol.html
@@ -0,0 +1,16 @@
+<html>
+    <head>
+
+         <link rel="shortcut icon" href="/favicon.ico">
+         <link rel="search" type="application/opensearchdescription+xml" href="/improvedsearch.src" title="Add search">
+         <link rel="alternate" type="application/atom+xml" href="//example.org/feed.xml" title="Main feed">
+         <link rel="alternate" type="application/atom+xml" href="//example.org/comments.xml" title="Comments feed">
+
+         <link rel="canonical" href="//example.org/index.php">
+         <link rel="shorturl" href="//example.org/index">
+         <link rel="alternate" href="//example.org/index" hreflang="x-default">
+
+        <link rel="stylesheet" type="text/css" href="//example.org/styles/home.css" media="screen">
+    </head>
+    <body></body>
+</html>
diff --git a/tests/samples/discovery-https.html b/tests/samples/discovery-https.html
@@ -0,0 +1,16 @@
+<html>
+    <head>
+
+         <link rel="shortcut icon" href="/favicon.ico">
+         <link rel="search" type="application/opensearchdescription+xml" href="/improvedsearch.src" title="Add search">
+         <link rel="alternate" type="application/atom+xml" href="https://example.org/feed.xml" title="Main feed">
+         <link rel="alternate" type="application/atom+xml" href="https://example.org/comments.xml" title="Comments feed">
+
+         <link rel="canonical" href="https://example.org/index.php">
+         <link rel="shorturl" href="https://example.org/index">
+         <link rel="alternate" href="https://example.org/index" hreflang="x-default">
+
+        <link rel="stylesheet" type="text/css" href="https://example.org/styles/home.css" media="screen">
+    </head>
+    <body></body>
+</html>