fix: use updated user agent when fetching feeds and favicons

Signed-off-by: Ben Vidulich <ben@vidulich.nz>
nextcloud · Oct 3, 2024 · 32d02c0 · 32d02c0
1 parent ad9f614
commit 32d02c0
Show file tree

Hide file tree

Showing 6 changed files with 127 additions and 2 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -10,6 +10,7 @@ You can also check [on GitHub](https://github.com/nextcloud/news/releases), the
 
 
 ### Fixed
+- Use updated user agent when fetching feeds and favicons (#2788)
 - Allow feed title to be null in DB. (#2745)
 - Store HTTP last modified date from response header (#2724)
 - Admin settings could not be saved (#2533)

diff --git a/lib/AppInfo/Application.php b/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;
         });
     }

diff --git a/lib/Config/FetcherConfig.php b/lib/Config/FetcherConfig.php
@@ -44,6 +44,12 @@ class FetcherConfig
      */
     protected $redirects;
 
+    /**
+     * Version number for the news application.
+     * @var string
+     */
+    private $version;
+
     /**
      * User agent for the client.
      * @var string
@@ -75,6 +81,11 @@ public function __construct(IConfig $config)
             'maxRedirects',
             Application::DEFAULT_SETTINGS['maxRedirects']
         );
+        $this->version = $config->getAppValue(
+            Application::NAME,
+            'installed_version',
+            '1.0'
+        );
 
         $proxy = $config->getSystemValue('proxy', null);
         if (is_null($proxy)) {
@@ -127,7 +138,7 @@ public function getClient(): ClientInterface
         $config = [
             'timeout' => $this->client_timeout,
             'headers' =>  [
-                'User-Agent' => static::DEFAULT_USER_AGENT,
+                'User-Agent' => $this->getUserAgent(),
                 'Accept' => static::DEFAULT_ACCEPT,
                 'Accept-Encoding' => $this->checkEncoding()
             ],
@@ -143,4 +154,18 @@ public function getClient(): ClientInterface
         $client = new Client($config);
         return new FeedIoClient($client);
     }
+
+    /**
+     * Gets a user agent name for the client
+     *
+     * @return string
+     */
+    public function getUserAgent(): string
+    {
+        if (is_null($this->version)) {
+            return self::DEFAULT_USER_AGENT;
+        }
+
+        return 'NextCloud-News/' . $this->version;
+    }
 }
diff --git a/lib/Fetcher/FaviconDataAccess.php b/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 <ben@vidulich.nz>
+ * @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(),
+                ]
+            ]
+        );
+    }
+}
diff --git a/lib/Fetcher/FeedFetcher.php b/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("");

diff --git a/tests/Unit/Config/FetcherConfigTest.php b/tests/Unit/Config/FetcherConfigTest.php
@@ -62,4 +62,35 @@ public function testGetClient()
 
         $this->assertInstanceOf(FeedIoClient::class, $this->class->getClient());
     }
+
+    public function testGetUserAgent()
+    {
+        $this->config->expects($this->exactly(3))
+            ->method('getAppValue')
+            ->willReturnMap([
+                ['news', 'feedFetcherTimeout', 60, '60'],
+                ['news', 'maxRedirects', 10, '10'],
+                ['news', 'installed_version', '1.0', '123.45']
+            ]);
+        $this->class = new FetcherConfig($this->config);
+
+        $expected = 'NextCloud-News/123.45';
+        $response = $this->class->getUserAgent();
+        $this->assertEquals($expected, $response);
+    }
+
+    public function testGetUserAgentUnknownVersion()
+    {
+        $this->config->expects($this->exactly(3))
+            ->method('getAppValue')
+            ->willReturnMap([
+                ['news', 'feedFetcherTimeout', 60, '60'],
+                ['news', 'maxRedirects', 10, '10']
+            ]);
+        $this->class = new FetcherConfig($this->config);
+
+        $expected = 'NextCloud-News/1.0';
+        $response = $this->class->getUserAgent();
+        $this->assertEquals($expected, $response);
+    }
 }