fix/allow CDATA encoding (#428)

6 files changed, 120 insertions, 109 deletions

diff --git a/lib/AppInfo/Application.php b/lib/AppInfo/Application.php
index 755f3ea70..06d6135dc 100644
--- a/lib/AppInfo/Application.php
+++ b/lib/AppInfo/Application.php
@@ -159,13 +159,9 @@ class Application extends App
          * Fetchers
          */
         $container->registerService(FetcherConfig::class, function (IContainer $c): FetcherConfig {
-            // FIXME: move this into a separate class for testing?
-            $config = $c->query(Config::class);
-            $proxy  = $c->query(ProxyConfigParser::class);
-
             $fConfig = new FetcherConfig();
-            $fConfig->setClientTimeout($config->getFeedFetcherTimeout());
-            $fConfig->setProxy($proxy);
+            $fConfig->setConfig($c->query(Config::class))
+                    ->setProxy($c->query(IConfig::class));
 
             return $fConfig;
         });
diff --git a/lib/Command/ShowFeed.php b/lib/Command/ShowFeed.php
new file mode 100644
index 000000000..bbe1913fa
--- /dev/null
+++ b/lib/Command/ShowFeed.php
@@ -0,0 +1,70 @@
+<?php
+/**
+ * Nextcloud - News
+ *
+ * This file is licensed under the Affero General Public License version 3 or
+ * later. See the COPYING file.
+ *
+ * @author Sean Molenaar <sean@seanmolenaar.eu>
+ * @copyright Sean Molenaar 2019
+ */
+namespace OCA\News\Command;
+
+use FeedIo\FeedIo;
+use Favicon\Favicon;
+
+use OCA\News\Fetcher\Fetcher;
+use Symfony\Component\Console\Command\Command;
+use Symfony\Component\Console\Input\InputInterface;
+use Symfony\Component\Console\Input\InputArgument;
+use Symfony\Component\Console\Input\InputOption;
+use Symfony\Component\Console\Output\OutputInterface;
+
+/**
+ * This is used for debugging feed data:
+ * ./occ news:show-feed www.feed.com
+ */
+class ShowFeed extends Command
+{
+    /**
+     * Feed and favicon fetcher.
+     */
+    protected $feedFetcher;
+
+    /**
+     * Set up class.
+     *
+     * @param Fetcher  $feedFetcher  Feed reader
+     */
+    public function __construct(Fetcher $feedFetcher)
+    {
+        $this->feedFetcher  = $feedFetcher;
+        parent::__construct();
+    }
+
+    protected function configure()
+    {
+        $this->setName('news:show-feed')
+            ->setDescription('Prints a JSON string which represents the given feed as it would be in the DB.')
+            ->addArgument('feed', InputArgument::REQUIRED, 'Feed to parse')
+            ->addOption('user', 'u', InputOption::VALUE_OPTIONAL, 'Username for the feed')
+            ->addOption('password', 'p', InputOption::VALUE_OPTIONAL, 'Password for the feed');
+    }
+
+    protected function execute(InputInterface $input, OutputInterface $output)
+    {
+        $url   = $input->getArgument('feed');
+        $user = $input->getOption('user');
+        $password = $input->getOption('password');
+
+        try {
+            list($feed, $items) = $this->feedFetcher->fetch($url, true, null, $user, $password);
+            $output->writeln("Feed: " . json_encode($feed, JSON_PRETTY_PRINT));
+            $output->writeln("Items: " . json_encode($items, JSON_PRETTY_PRINT));
+        } catch (\Throwable $ex) {
+            $output->writeln('<error>Failed to fetch feed info:</error>');
+            $output->writeln($ex->getMessage());
+            return 1;
+        }
+    }
+}
diff --git a/lib/Config/FetcherConfig.php b/lib/Config/FetcherConfig.php
index 55603c47c..1459ff032 100644
--- a/lib/Config/FetcherConfig.php
+++ b/lib/Config/FetcherConfig.php
@@ -16,6 +16,7 @@ namespace OCA\News\Config;
 use FeedIo\Adapter\ClientInterface;
 use \GuzzleHttp\Client;
 use \FeedIo\Adapter\Guzzle\Client as FeedIoClient;
+use OCP\IConfig;
 
 /**
  * Class FetcherConfig
@@ -26,6 +27,8 @@ class FetcherConfig
 {
     protected $client_timeout;
     protected $proxy;
+    protected $redirects;
+    protected $max_size;
 
     /**
      * Configure a guzzle client
@@ -36,14 +39,18 @@ class FetcherConfig
     {
         if (!class_exists('GuzzleHttp\Collection')) {
             $config = [
-                'timeout' => $this->getClientTimeout(),
+                'timeout' => $this->client_timeout,
             ];
 
             if (!empty($this->proxy)) {
                 $config['proxy'] = $this->proxy;
             }
 
-            $guzzle = new Client();
+            if (!empty($this->redirects)) {
+                $config['redirect.max'] = $this->redirects;
+            }
+
+            $guzzle = new Client($config);
             $client = new FeedIoClient($guzzle);
 
             return $client;
@@ -51,7 +58,7 @@ class FetcherConfig
 
         $config = [
             'request.options' => [
-                'timeout' => $this->getClientTimeout(),
+                'timeout' => $this->client_timeout,
             ],
         ];
 
@@ -59,59 +66,54 @@ class FetcherConfig
             $config['request.options']['proxy'] = $this->proxy;
         }
 
+        if (!empty($this->redirects)) {
+            $config['request.options']['redirect.max'] = $this->redirects;
+        }
+
         $guzzle = new Client($config);
         return new LegacyGuzzleClient($guzzle);
     }
 
     /**
-     * Set a timeout for the client
+     * Set settings for config.
      *
-     * @param int $timeout The timeout
+     * @param Config $config The shared configuration
      *
      * @return self
      */
-    public function setClientTimeout($timeout)
+    public function setConfig(Config $config)
     {
-        $this->client_timeout = $timeout;
+        $this->client_timeout = $config->getFeedFetcherTimeout();
+        $this->redirects = $config->getMaxRedirects();
+        $this->max_size = $config->getMaxSize();
 
         return $this;
     }
 
     /**
-     * Get the client timeout.
-     *
-     * @return mixed
-     */
-    public function getClientTimeout()
-    {
-        return $this->client_timeout;
-    }
-
-    /**
      * Set the proxy
      *
-     * @param \OCA\News\Utility\ProxyConfigParser $proxy The proxy to set.
+     * @param IConfig $config Nextcloud config.
      *
      * @return self
      */
-    public function setProxy($proxy)
+    public function setProxy(IConfig $config)
     {
-        // proxy settings
-        $proxySettings = $proxy->parse();
-        $host = $proxySettings['host'];
-        $port = $proxySettings['port'];
-        $user = $proxySettings['user'];
-        $password = $proxySettings['password'];
-
-        $proxy_string = 'https://';
-        if (!empty($user)) {
-            $proxy_string .= $user . ':' . $password . '@';
+        $proxy = $config->getSystemValue('proxy', null);
+        $creds = $config->getSystemValue('proxyuserpwd', null);
+
+        if (is_null($proxy)) {
+            return $this;
         }
-        $proxy_string .= $host;
-        if (!empty($port)) {
-            $proxy_string .= ':' . $port;
+
+        $url = new \Net_URL2($proxy);
+
+        if ($creds) {
+            $auth = explode(':', $creds, 2);
+            $url->setUserinfo($auth[0], $auth[1]);
         }
-        $this->proxy = $proxy_string;
+
+        $this->proxy = $url->getNormalizedURL();
 
         return $this;
     }
diff --git a/lib/Db/Item.php b/lib/Db/Item.php
index 3a17dd2cb..710f239d4 100644
--- a/lib/Db/Item.php
+++ b/lib/Db/Item.php
@@ -146,7 +146,7 @@ class Item extends Entity implements IAPI, \JsonSerializable
         return $this->enclosureMime;
     }
 
-    public function getFeedId(): int
+    public function getFeedId()
     {
         return $this->feedId;
     }
@@ -169,7 +169,7 @@ class Item extends Entity implements IAPI, \JsonSerializable
         return $this->guidHash;
     }
 
-    public function getId(): int
+    public function getId()
     {
         return $this->id;
     }
diff --git a/lib/Fetcher/FeedFetcher.php b/lib/Fetcher/FeedFetcher.php
index 9b4a2f996..fb9da7919 100755
--- a/lib/Fetcher/FeedFetcher.php
+++ b/lib/Fetcher/FeedFetcher.php
@@ -19,12 +19,14 @@ use FeedIo\Feed\ItemInterface;
 use FeedIo\FeedInterface;
 use FeedIo\FeedIo;
 
+use Net_URL2;
 use OCA\News\Utility\PsrLogger;
 use OCP\IL10N;
 
 use OCA\News\Db\Item;
 use OCA\News\Db\Feed;
 use OCA\News\Utility\Time;
+use SimpleXMLElement;
 
 class FeedFetcher implements IFeedFetcher
 {
@@ -65,10 +67,11 @@ class FeedFetcher implements IFeedFetcher
      */
     public function fetch(string $url, bool $favicon, $lastModified, $user, $password): array
     {
+        $url2 = new Net_URL2($url);
         if (!empty($user) && !empty(trim($user))) {
-            $url = explode('://', $url);
-            $url = $url[0] . '://' . urlencode($user) . ':' . urlencode($password) . '@' . $url[1];
+            $url2->setUserinfo(urlencode($user), urlencode($password));
         }
+        $url = $url2->getNormalizedURL();
         if (is_null($lastModified) || !is_string($lastModified)) {
             $resource = $this->reader->read($url);
         } else {
@@ -204,6 +207,12 @@ class FeedFetcher implements IFeedFetcher
             'HTML-ENTITIES',
             mb_detect_encoding($body)
         );
+        $data = simplexml_load_string(
+            '<?xml version="1.0" encoding="utf-8"?><item>' . $body . '</item>',
+            SimpleXMLElement::class,
+            LIBXML_NOCDATA
+        );
+        $body = ($data === false) ? $body : (string) $data;
         $item->setBody($body);
 
         if ($parsedItem->hasMedia()) {
diff --git a/lib/Utility/ProxyConfigParser.php b/lib/Utility/ProxyConfigParser.php
deleted file mode 100644
index dc9bd3d8b..000000000
--- a/lib/Utility/ProxyConfigParser.php
+++ /dev/null
@@ -1,66 +0,0 @@
-<?php
-/**
- * Nextcloud - News
- *
- * This file is licensed under the Affero General Public License version 3 or
- * later. See the COPYING file.
- *
- * @author    Alessandro Cosentino <cosenal@gmail.com>
- * @author    Bernhard Posselt <dev@bernhard-posselt.com>
- * @copyright 2012 Alessandro Cosentino
- * @copyright 2012-2014 Bernhard Posselt
- */
-
-
-namespace OCA\News\Utility;
-
-use \OCP\IConfig;
-
-class ProxyConfigParser
-{
-
-    private $config;
-
-    public function __construct(IConfig $config)
-    {
-        $this->config = $config;
-    }
-
-
-    /**
-     * Parses the config and splits up the port + url
-     *
-     * @return array
-     */
-    public function parse()
-    {
-        $proxy = $this->config->getSystemValue('proxy');
-        $userpasswd = $this->config->getSystemValue('proxyuserpwd');
-
-        $result = [
-            'host' => null,
-            'port' => null,
-            'user' => null,
-            'password' => null
-        ];
-
-        // we need to filter out the port -.-
-        $url = new \Net_URL2($proxy);
-        $port = $url->getPort();
-
-        $url->setPort(false);
-        $host = $url->getUrl();
-
-
-        $result['host'] = $host;
-        $result['port'] = $port;
-
-        if ($userpasswd) {
-            $auth = explode(':', $userpasswd, 2);
-            $result['user'] = $auth[0];
-            $result['password'] = $auth[1];
-        }
-
-        return $result;
-    }
-}