Merge remote-tracking branch 'nextcloud/master' into master-merge

Signed-off-by: Devlin Junker <devlin.junker@gmail.com>

10 files changed, 229 insertions, 36 deletions

diff --git a/lib/AppInfo/Application.php b/lib/AppInfo/Application.php
index bc5e1b476..8bfa09c86 100644
--- a/lib/AppInfo/Application.php
+++ b/lib/AppInfo/Application.php
@@ -24,11 +24,11 @@ use OCA\News\Hooks\UserDeleteHook;
 use OCA\News\Search\FeedSearchProvider;
 use OCA\News\Search\FolderSearchProvider;
 use OCA\News\Search\ItemSearchProvider;
+use OCA\News\Utility\Cache;
 
 use OCP\AppFramework\Bootstrap\IBootContext;
 use OCP\AppFramework\Bootstrap\IBootstrap;
 use OCP\AppFramework\Bootstrap\IRegistrationContext;
-use OCP\ITempManager;
 use OCP\AppFramework\App;
 
 use OCA\News\Fetcher\FeedFetcher;
@@ -92,15 +92,9 @@ class Application extends App implements IBootstrap
         $context->registerParameter('exploreDir', __DIR__ . '/../Explore/feeds');
 
         $context->registerService(HTMLPurifier::class, function (ContainerInterface $c): HTMLPurifier {
-            $directory = $c->get(ITempManager::class)->getTempBaseDir() . '/news/cache/purifier';
-
-            if (!is_dir($directory)) {
-                mkdir($directory, 0770, true);
-            }
-
             $config = HTMLPurifier_Config::createDefault();
             $config->set('HTML.ForbiddenAttributes', 'class');
-            $config->set('Cache.SerializerPath', $directory);
+            $config->set('Cache.SerializerPath', $c->get(Cache::class)->getCache("purifier"));
             $config->set('HTML.SafeIframe', true);
             $config->set(
                 'URI.SafeIframeRegexp',
@@ -140,7 +134,7 @@ class Application extends App implements IBootstrap
 
         $context->registerService(Favicon::class, function (ContainerInterface $c): Favicon {
             $favicon = new Favicon();
-            $favicon->cache(['dir' => $c->get(ITempManager::class)->getTempBaseDir()]);
+            $favicon->cache(['dir' => $c->get(Cache::class)->getCache("feedFavicon")]);
             return $favicon;
         });
     }
diff --git a/lib/Command/Updater/Job.php b/lib/Command/Updater/Job.php
new file mode 100644
index 000000000..4404a456e
--- /dev/null
+++ b/lib/Command/Updater/Job.php
@@ -0,0 +1,76 @@
+<?php
+/**
+ * Nextcloud - News
+ *
+ * This file is licensed under the Affero General Public License version 3 or
+ * later. See the COPYING file.
+ */
+
+namespace OCA\News\Command\Updater;
+
+use DateTime;
+use OCP\Util;
+use OCA\News\Service\StatusService;
+use OCA\News\Service\UpdaterService;
+use Symfony\Component\Console\Command\Command;
+use Symfony\Component\Console\Input\InputOption;
+use Symfony\Component\Console\Input\InputInterface;
+use Symfony\Component\Console\Output\OutputInterface;
+
+class Job extends Command
+{
+    /**
+     * @var StatusService Status service
+     */
+    private $statusService;
+
+    /**
+     * @var UpdaterService Update service
+     */
+    private $updaterService;
+
+    public function __construct(StatusService $statusService, UpdaterService $updaterService)
+    {
+        parent::__construct();
+        $this->statusService = $statusService;
+        $this->updaterService = $updaterService;
+    }
+
+    /**
+     * @return void
+     */
+    protected function configure()
+    {
+        $this->setName('news:updater:job')
+            ->addOption(
+                'reset',
+                null,
+                InputOption::VALUE_NONE,
+                'If the job should be reset, warning this might lead to issues.'
+            )
+            ->setDescription('Console API for checking the update job status and to reset it.');
+    }
+
+    protected function execute(InputInterface $input, OutputInterface $output): int
+    {
+        $reset = (bool) $input->getOption('reset');
+
+        [$major, $minor, $micro] = Util::getVersion();
+        
+        if ($major < 26) {
+            $output->writeln("Error: This only works with Nextcloud 26 or newer.");
+            return 1;
+        }
+        $output->writeln("Checking update Status");
+        $date = new DateTime();
+        $date->setTimestamp($this->statusService->getUpdateTime());
+        $output->writeln("Last Execution was ".$date->format('Y-m-d H:i:s e'));
+
+        if ($reset) {
+            $output->writeln("Attempting to reset the job.");
+            $this->updaterService->reset();
+            $output->writeln("Done, job should execute on next schedule.");
+        }
+        return 0;
+    }
+}
diff --git a/lib/Config/FetcherConfig.php b/lib/Config/FetcherConfig.php
index 797dae49e..702fccdf1 100644
--- a/lib/Config/FetcherConfig.php
+++ b/lib/Config/FetcherConfig.php
@@ -95,6 +95,29 @@ class FetcherConfig
     }
 
     /**
+     * Checks for available encoding options
+     *
+     * @return String list of supported encoding types
+     */
+    public function checkEncoding()
+    {
+        $supportedEncoding = [];
+
+        // check curl features
+        $curl_features = curl_version()["features"];
+
+        $bitfields = array('CURL_VERSION_LIBZ' => ['gzip', 'deflate'], 'CURL_VERSION_BROTLI' => ['br']);
+        
+        foreach ($bitfields as $feature => $header) {
+            // checking available features via the 'features' bitmask and adding available types to the list
+            if (defined($feature) && $curl_features & constant($feature)) {
+                $supportedEncoding = array_merge($supportedEncoding, $header);
+            }
+        }
+        return implode(", ", $supportedEncoding);
+    }
+
+    /**
      * Configure a guzzle client
      *
      * @return ClientInterface Client to guzzle.
@@ -106,7 +129,7 @@ class FetcherConfig
             'headers' =>  [
                 'User-Agent' => static::DEFAULT_USER_AGENT,
                 'Accept' => static::DEFAULT_ACCEPT,
-                'Accept-Encoding' => 'gzip, deflate',
+                'Accept-Encoding' => $this->checkEncoding()
             ],
         ];
 
diff --git a/lib/Controller/PageController.php b/lib/Controller/PageController.php
index abdd3f2f7..41d0e7f89 100644
--- a/lib/Controller/PageController.php
+++ b/lib/Controller/PageController.php
@@ -96,6 +96,7 @@ class PageController extends Controller
         $csp->addAllowedImageDomain('*')
             ->addAllowedMediaDomain('*')
             ->addAllowedConnectDomain('*')// chrome breaks on audio elements
+            ->allowEvalScript(true)
             ->addAllowedFrameDomain('https://youtube.com')
             ->addAllowedFrameDomain('https://www.youtube.com')
             ->addAllowedFrameDomain('https://player.vimeo.com')
diff --git a/lib/Fetcher/Client/FeedIoClient.php b/lib/Fetcher/Client/FeedIoClient.php
index ac1f62a1f..fe57c5889 100644
--- a/lib/Fetcher/Client/FeedIoClient.php
+++ b/lib/Fetcher/Client/FeedIoClient.php
@@ -38,14 +38,14 @@ class FeedIoClient implements ClientInterface
     }
 
     /**
-     * @param  string                               $url
-     * @param  DateTime                            $modifiedSince
+     * @param  string                                   $url
+     * @param  DateTime|null                            $modifiedSince
      *
      * @return ResponseInterface
      * @throws ServerErrorException|GuzzleException
      * @throws NotFoundException
      */
-    public function getResponse(string $url, DateTime $modifiedSince) : ResponseInterface
+    public function getResponse(string $url, ?DateTime $modifiedSince = null) : ResponseInterface
     {
         $modifiedSince->setTimezone(new \DateTimeZone('GMT'));
         try {
@@ -65,7 +65,7 @@ class FeedIoClient implements ClientInterface
                 case 404:
                     throw new NotFoundException($e->getMessage());
                 default:
-                    throw new ServerErrorException($e->getMessage());
+                    throw new ServerErrorException($e->getResponse());
             }
         }
     }
diff --git a/lib/Fetcher/FeedFetcher.php b/lib/Fetcher/FeedFetcher.php
index 60d798889..c0477407c 100755
--- a/lib/Fetcher/FeedFetcher.php
+++ b/lib/Fetcher/FeedFetcher.php
@@ -30,6 +30,7 @@ use OCP\ITempManager;
 use OCA\News\Db\Item;
 use OCA\News\Db\Feed;
 use OCA\News\Utility\Time;
+use OCA\News\Utility\Cache;
 use OCA\News\Scraper\Scraper;
 use OCA\News\Config\FetcherConfig;
 use Psr\Log\LoggerInterface;
@@ -59,11 +60,6 @@ class FeedFetcher implements IFeedFetcher
     private $l10n;
 
     /**
-     * @var ITempManager
-     */
-    private $ITempManager;
-
-    /**
      * @var Time
      */
     private $time;
@@ -73,22 +69,34 @@ class FeedFetcher implements IFeedFetcher
      */
     private $logger;
 
+    /**
+     * @var FetcherConfig
+     */
+    private $fetcherConfig;
+     
+    /**
+     * @var Cache
+     */
+    private $cache;
+
     public function __construct(
         FeedIo $fetcher,
         Favicon $favicon,
         Scraper $scraper,
         IL10N $l10n,
-        ITempManager $ITempManager,
         Time $time,
-        LoggerInterface $logger
+        LoggerInterface $logger,
+        FetcherConfig $fetcherConfig,
+        Cache $cache
     ) {
         $this->reader         = $fetcher;
         $this->faviconFactory = $favicon;
         $this->scraper        = $scraper;
         $this->l10n           = $l10n;
-        $this->ITempManager   = $ITempManager;
         $this->time           = $time;
         $this->logger         = $logger;
+        $this->fetcherConfig  = $fetcherConfig;
+        $this->cache          = $cache;
     }
 
 
@@ -127,7 +135,6 @@ class FeedFetcher implements IFeedFetcher
             $lastModified = null;
         }
         $url = $url2->getNormalizedURL();
-        $this->reader->resetFilters();
         $resource = $this->reader->read($url, null, $lastModified);
 
         $location     = $resource->getUrl();
@@ -388,8 +395,10 @@ class FeedFetcher implements IFeedFetcher
             return is_string($return) ? $return : null;
         }
 
-        // logo will be saved in the tmp folder provided by Nextcloud, file is named as md5 of the url
-        $favicon_path = join(DIRECTORY_SEPARATOR, [$this->ITempManager->getTempBaseDir(), md5($favicon)]);
+        $logo_cache = $this->cache->getCache("feedLogo");
+
+        // file name of the logo is md5 of the url
+        $favicon_path = join(DIRECTORY_SEPARATOR, [$logo_cache, md5($favicon)]);
         $downloaded = false;
 
         if (file_exists($favicon_path)) {
@@ -409,17 +418,18 @@ class FeedFetcher implements IFeedFetcher
                     'headers' => [
                         'User-Agent'        => FetcherConfig::DEFAULT_USER_AGENT,
                         'Accept'            => 'image/*',
-                        'If-Modified-Since' => date(DateTime::RFC7231, $last_modified)
+                        'If-Modified-Since' => date(DateTime::RFC7231, $last_modified),
+                        'Accept-Encoding'   => $this->fetcherConfig->checkEncoding()
                     ]
                 ]
             );
             $downloaded = true;
 
             $this->logger->debug(
-                "Feed:{url} Logo:{logo} Status:{status}",
+                "Feed:{feed} Logo:{logo} Status:{status}",
                 [
                 'status' => $response->getStatusCode(),
-                'url'    => $favicon_path,
+                'feed'   => $url,
                 'logo'   => $favicon
                 ]
             );
@@ -437,6 +447,14 @@ class FeedFetcher implements IFeedFetcher
 
         // check if file is actually an image
         if (!$is_image) {
+            $this->logger->debug(
+                "Downloaded file:{file} from {url} is not an image",
+                [
+                'file' => $favicon_path,
+                'url'   => $favicon
+                ]
+            );
+
             $return = $this->faviconFactory->get($base_url);
             return is_string($return) ? $return : null;
         }
@@ -444,6 +462,13 @@ class FeedFetcher implements IFeedFetcher
         list($width, $height, $type, $attr) = getimagesize($favicon_path);
         // check if image is square else fall back to favicon
         if ($width !== $height) {
+            $this->logger->debug(
+                "Downloaded file:{file} from {url} is not square",
+                [
+                'file' => $favicon_path,
+                'url'   => $favicon
+                ]
+            );
             $return = $this->faviconFactory->get($base_url);
             return is_string($return) ? $return : null;
         }
diff --git a/lib/Scraper/Scraper.php b/lib/Scraper/Scraper.php
index 998c4464c..c5ae401f9 100644
--- a/lib/Scraper/Scraper.php
+++ b/lib/Scraper/Scraper.php
@@ -16,6 +16,7 @@ use fivefilters\Readability\Configuration;
 use fivefilters\Readability\ParseException;
 use League\Uri\Exceptions\SyntaxError;
 use Psr\Log\LoggerInterface;
+use OCA\News\Config\FetcherConfig;
 
 class Scraper implements IScraper
 {
@@ -37,7 +38,7 @@ class Scraper implements IScraper
             CURLOPT_RETURNTRANSFER => true,     // return web page
             CURLOPT_HEADER         => false,    // do not return headers
             CURLOPT_FOLLOWLOCATION => true,     // follow redirects
-            //CURLOPT_USERAGENT    => "php-news", // who am i
+            CURLOPT_USERAGENT      => FetcherConfig::DEFAULT_USER_AGENT, // who am i
             CURLOPT_AUTOREFERER    => true,     // set referer on redirect
             CURLOPT_CONNECTTIMEOUT => 120,      // timeout on connect
             CURLOPT_TIMEOUT        => 120,      // timeout on response
diff --git a/lib/Service/StatusService.php b/lib/Service/StatusService.php
index 69a621e3c..09cea9e4a 100644
--- a/lib/Service/StatusService.php
+++ b/lib/Service/StatusService.php
@@ -92,12 +92,8 @@ class StatusService
 
         $time = 0;
 
-        [$major, $minor, $micro] = Util::getVersion();
-        
-        if ($major >= 26) {
-            $myJobList = $this->jobList->getJobsIterator(UpdaterJob::class, 1, 0);
-            $time = $myJobList->current()->getLastRun();
-        }
+        $myJobList = $this->jobList->getJobsIterator(UpdaterJob::class, 1, 0);
+        $time = $myJobList->current()->getLastRun();
         
         return $time;
     }
diff --git a/lib/Service/UpdaterService.php b/lib/Service/UpdaterService.php
index 5ab553d10..a67376de7 100644
--- a/lib/Service/UpdaterService.php
+++ b/lib/Service/UpdaterService.php
@@ -14,6 +14,9 @@
 
 namespace OCA\News\Service;
 
+use OCP\BackgroundJob\IJobList;
+use OCA\News\Cron\UpdaterJob;
+
 class UpdaterService
 {
 
@@ -32,14 +35,19 @@ class UpdaterService
      */
     private $itemService;
 
+    /** @var IJobList */
+    private $jobList;
+
     public function __construct(
         FolderServiceV2 $folderService,
         FeedServiceV2 $feedService,
-        ItemServiceV2 $itemService
+        ItemServiceV2 $itemService,
+        IJobList $jobList
     ) {
         $this->folderService = $folderService;
         $this->feedService = $feedService;
         $this->itemService = $itemService;
+        $this->jobList = $jobList;
     }
 
 
@@ -60,4 +68,14 @@ class UpdaterService
     {
         $this->itemService->purgeOverThreshold();
     }
+
+    public function reset(): int
+    {
+        $myJobList = $this->jobList->getJobsIterator(UpdaterJob::class, 1, 0);
+        $job = $myJobList->current();
+
+        $this->jobList->resetBackgroundJob($job);
+
+        return 0;
+    }
 }
diff --git a/lib/Utility/Cache.php b/lib/Utility/Cache.php
new file mode 100644
index 000000000..6337487cf
--- /dev/null
+++ b/lib/Utility/Cache.php
@@ -0,0 +1,59 @@
+<?php
+/**
+ * Nextcloud - News
+ *
+ * This file is licensed under the Affero General Public License version 3 or
+ * later. See the COPYING file.
+ *
+ * @author    Benjamin Brahmer <info@b-brahmer.de>
+ * @copyright 2023 Benjamin Brahmer
+ */
+namespace OCA\News\Utility;
+
+use OCP\ITempManager;
+use OCP\IConfig;
+
+class Cache
+{
+   
+
+    /**
+     * @var ITempManager
+     */
+    private $ITempManager;
+
+    /**
+     * @var IConfig
+     */
+    private $IConfig;
+
+
+    public function __construct(
+        ITempManager $ITempManager,
+        IConfig $IConfig
+    ) {
+        $this->ITempManager   = $ITempManager;
+        $this->IConfig        = $IConfig;
+    }
+
+    /**
+     * Get a news app cache directory
+     *
+     * @param String $name for the sub-directory, is created if not existing
+     *
+     * @return String $directory The path for the cache
+     */
+    public function getCache(String $name): String
+    {
+        $baseDir = $this->ITempManager->getTempBaseDir();
+        $instanceID = $this->IConfig->getSystemValue('instanceid');
+
+        $directory = join(DIRECTORY_SEPARATOR, [$baseDir, "news-" . $instanceID, 'cache', $name]);
+
+        if (!is_dir($directory)) {
+            mkdir($directory, 0770, true);
+        }
+
+        return $directory;
+    }
+}