Fix xkcd commics

SimonSchubert · Mar 26, 2019 · b37e237 · b37e237
1 parent 08e44b8
commit b37e237
Show file tree

Hide file tree

Showing 3 changed files with 87 additions and 34 deletions.
diff --git a/composer.lock b/composer.lock
diff --git a/lib/Fetcher/FeedFetcher.php b/lib/Fetcher/FeedFetcher.php
@@ -185,13 +185,9 @@ protected function buildItem(ItemInterface $parsedItem, FeedInterface $parsedFee
             $pubDT = $lastmodified;
         }
 
-        $item->setPubDate(
-            $pubDT->getTimestamp()
-        );
+        $item->setPubDate($pubDT->getTimestamp());
 
-        $item->setLastModified(
-            $lastmodified->getTimestamp()
-        );
+        $item->setLastModified($lastmodified->getTimestamp());
         $item->setRtl($this->determineRtl($parsedFeed));
 
         // unescape content because angularjs helps against XSS
@@ -208,14 +204,19 @@ protected function buildItem(ItemInterface $parsedItem, FeedInterface $parsedFee
             'HTML-ENTITIES',
             mb_detect_encoding($body)
         );
-        libxml_use_internal_errors(true);
-        $data = simplexml_load_string(
-            "<?xml version=\"1.0\"?><item>$body</item>",
-            SimpleXMLElement::class,
-            LIBXML_NOCDATA
-        );
-        libxml_clear_errors();
-        $body = ($data === false) ? $body : (string) $data;
+        if (strpos($body, 'CDATA') !== false) {
+            libxml_use_internal_errors(true);
+            $data = simplexml_load_string(
+                "<?xml version=\"1.0\"?><item>$body</item>",
+                SimpleXMLElement::class,
+                LIBXML_NOCDATA
+            );
+            if ($data !== false && libxml_get_last_error() === false) {
+                $body = (string) $data;
+            }
+            libxml_clear_errors();
+        }
+
         $item->setBody($body);
 
         if ($parsedItem->hasMedia()) {

diff --git a/tests/Unit/Fetcher/FeedFetcherTest.php b/tests/Unit/Fetcher/FeedFetcherTest.php
@@ -18,15 +18,11 @@
 use FeedIo\Feed\Item\MediaInterface;
 use FeedIo\Feed\ItemInterface;
 use FeedIo\FeedInterface;
-use Favicon\Favicon;
 use OC\L10N\L10N;
-use OCA\AdminAudit\Actions\Auth;
 use \OCA\News\Db\Feed;
 use \OCA\News\Db\Item;
 use OCA\News\Fetcher\FeedFetcher;
 use OCA\News\Utility\PsrLogger;
-use OCA\News\Utility\Time;
-use OCP\IL10N;
 
 use PHPUnit\Framework\TestCase;
 
@@ -243,6 +239,60 @@ public function testFetchIfNoModifiedExists()
         $this->assertEquals([$feed, [$item]], $result);
     }
 
+    /**
+     * Return body options
+     * @return array
+     */
+    public function feedBodyProvider()
+    {
+        return [
+            [
+                '<![CDATA[let the bodies hit the floor <a href="test">test</a>]]>',
+                'let the bodies hit the floor <a href="test">test</a>'
+            ],
+            [
+                'let the bodies hit the floor <a href="test">test</a>',
+                'let the bodies hit the floor <a href="test">test</a>'
+            ],
+            [
+                'let the bodies hit the floor "test" test',
+                'let the bodies hit the floor "test" test'
+            ],
+            [
+                '<img src="https://imgs.xkcd.com/google_trends_maps.png" title="It\'s early 2020. The entire country is gripped with Marco Rubio" />',
+                '<img src="https://imgs.xkcd.com/google_trends_maps.png" title="It\'s early 2020. The entire country is gripped with Marco Rubio" />'
+            ],
+        ];
+    }
+
+    /**
+     * Test if body is set correctly.
+     *
+     * @dataProvider feedBodyProvider
+     *
+     * @param string $body        The body before parsing.
+     * @param string $parsed_body The body after parsing.
+     */
+    public function testFetchWithFeedContent($body, $parsed_body)
+    {
+        $bodyBackup = $this->body;
+        $parsedBackup = $this->parsed_body;
+
+        $this->body = $body;
+        $this->parsed_body = $parsed_body;
+
+        $this->setUpReader($this->url, null, true);
+        $item = $this->createItem();
+        $feed = $this->createFeed();
+        $this->mockIterator($this->feed_mock, [$this->item_mock]);
+        $result = $this->fetcher->fetch($this->url, false, '0', null, null);
+
+        $this->assertEquals([$feed, [$item]], $result);
+
+        $this->body = $bodyBackup;
+        $this->parsed_body = $parsedBackup;
+    }
+
     /**
      * Test if the fetch function fetches a feed.
      */