set user agent when retrieving favicons

Signed-off-by: Ben Vidulich <[email protected]>

lib/AppInfo/Application.php

@@ -32,6 +32,7 @@
 use OCP\AppFramework\Bootstrap\IRegistrationContext;
 use OCP\AppFramework\App;
 
+use OCA\News\Fetcher\FaviconDataAccess;
 use OCA\News\Fetcher\FeedFetcher;
 use OCA\News\Fetcher\Fetcher;
 use OCP\User\Events\BeforeUserDeletedEvent;
@@ -135,9 +136,15 @@ public function register(IRegistrationContext $context): void
             return new Explorer($config->getClient(), $c->get(LoggerInterface::class));
         });
 
+        $context->registerService(FaviconDataAccess::class, function (ContainerInterface $c): FaviconDataAccess {
+            $config = $c->get(FetcherConfig::class);
+            return new FaviconDataAccess($config);
+        });
+
         $context->registerService(Favicon::class, function (ContainerInterface $c): Favicon {
             $favicon = new Favicon();
             $favicon->cache(['dir' => $c->get(Cache::class)->getCache("feedFavicon")]);
+            $favicon->setDataAccess($c->get(FaviconDataAccess::class));
             return $favicon;
         });
     }

lib/Fetcher/FaviconDataAccess.php

@@ -0,0 +1,61 @@
+<?php
+/**
+ * Nextcloud - News
+ *
+ * This file is licensed under the Affero General Public License version 3 or
+ * later. See the COPYING file.
+ *
+ * @author    Ben Vidulich <[email protected]>
+ * @copyright 2024 Ben Vidulich
+ */
+
+namespace OCA\News\Fetcher;
+
+use Favicon\DataAccess;
+
+use OCA\News\Config\FetcherConfig;
+
+/**
+ * Modified version of DataAccess with a configurable user agent header.
+ */
+class FaviconDataAccess extends DataAccess
+{
+    /**
+     * @var FetcherConfig
+     */
+    private $fetcherConfig;
+
+    public function __construct(
+        FetcherConfig $fetcherConfig,
+    ) {
+        $this->fetcherConfig  = $fetcherConfig;
+    }
+
+    public function retrieveUrl($url)
+    {
+        $this->setContext();
+        return @file_get_contents($url);
+    }
+
+    public function retrieveHeader($url)
+    {
+        $this->setContext();
+        $headers = @get_headers($url, 1);
+        return is_array($headers) ? array_change_key_case($headers) : [];
+    }
+
+    private function setContext()
+    {
+        stream_context_set_default(
+            [
+                'http' => [
+                    'method' => 'GET',
+                    'follow_location' => 0,
+                    'max_redirects' => 1,
+                    'timeout' => 10,
+                    'header' => 'User-Agent: ' . $this->fetcherConfig->getUserAgent(),
+                ]
+            ]
+        );
+    }
+}

lib/Fetcher/FeedFetcher.php

@@ -384,7 +384,7 @@ protected function getFavicon(FeedInterface $feed, string $url): ?string
             $favicon = trim($feed_logo);
         }
 
-        ini_set('user_agent', FetcherConfig::DEFAULT_USER_AGENT);
+        ini_set('user_agent', $this->fetcherConfig->getUserAgent());
 
         $base_url = new Net_URL2($url);
         $base_url->setPath("");