From bc27596f70cb170203655a55c0f53ee55d8d6316 Mon Sep 17 00:00:00 2001
From: Robin Appelman <robin@icewind.nl>
Date: Wed, 4 Jan 2017 11:10:19 +0100
Subject: Add postprocessor for getting lwn subscriber articles (#72)

---
 lib/Fetcher/FeedFetcher.php            |   8 ++-
 lib/PostProcessor/LWNProcessor.php     | 102 +++++++++++++++++++++++++++++++++
 tests/Unit/Fetcher/FeedFetcherTest.php |  12 +++-
 3 files changed, 120 insertions(+), 2 deletions(-)
 create mode 100644 lib/PostProcessor/LWNProcessor.php

diff --git a/lib/Fetcher/FeedFetcher.php b/lib/Fetcher/FeedFetcher.php
index 194a927f9..587e64af2 100644
--- a/lib/Fetcher/FeedFetcher.php
+++ b/lib/Fetcher/FeedFetcher.php
@@ -15,6 +15,8 @@ namespace OCA\News\Fetcher;
 
 use Exception;
 
+use OCA\News\PostProcessor\LWNProcessor;
+use OCP\Http\Client\IClientService;
 use PicoFeed\Parser\MalFormedXmlException;
 use PicoFeed\Reader\Reader;
 use PicoFeed\Parser\Parser;
@@ -42,15 +44,18 @@ class FeedFetcher implements IFeedFetcher {
     private $reader;
     private $l10n;
     private $time;
+    private $clientService;
 
     public function __construct(Reader $reader,
                                 PicoFeedFaviconFactory $faviconFactory,
                                 IL10N $l10n,
-                                Time $time) {
+                                Time $time,
+                                IClientService $clientService) {
         $this->faviconFactory = $faviconFactory;
         $this->reader = $reader;
         $this->time = $time;
         $this->l10n = $l10n;
+        $this->clientService = $clientService;
     }
 
 
@@ -108,6 +113,7 @@ class FeedFetcher implements IFeedFetcher {
 
             if ($fullTextEnabled) {
                 $parser->enableContentGrabber();
+                $parser->getItemPostProcessor()->register(new LWNProcessor($basicAuthUser, $basicAuthPassword, $this->clientService));
             }
 
             $parsedFeed = $parser->execute();
diff --git a/lib/PostProcessor/LWNProcessor.php b/lib/PostProcessor/LWNProcessor.php
new file mode 100644
index 000000000..f931bb41c
--- /dev/null
+++ b/lib/PostProcessor/LWNProcessor.php
@@ -0,0 +1,102 @@
+<?php
+/**
+ * Nextcloud - News
+ *
+ * This file is licensed under the Affero General Public License version 3 or
+ * later. See the COPYING file.
+ *
+ * @author Robin Appelman <robin@icewind.nl>
+ */
+
+namespace OCA\News\PostProcessor;
+
+use GuzzleHttp\Cookie\CookieJar;
+use OCP\Http\Client\IClientService;
+use PicoFeed\Parser\Feed;
+use PicoFeed\Parser\Item;
+use PicoFeed\Processor\ItemProcessorInterface;
+use PicoFeed\Scraper\RuleParser;
+
+class LWNProcessor implements ItemProcessorInterface {
+    private $user;
+
+    private $password;
+
+    private $clientService;
+
+    private $cookieJar;
+
+    /**
+     * @param $user
+     * @param $password
+     */
+    public function __construct($user, $password, IClientService $clientService) {
+        $this->user = $user;
+        $this->password = $password;
+        $this->clientService = $clientService;
+        $this->cookieJar = new CookieJar();
+    }
+
+    private function login() {
+        if ($this->cookieJar->count() > 0) {
+            return true;
+        }
+        if (!$this->user || !$this->password) {
+            return false;
+        }
+
+        $client = $this->clientService->newClient();
+        $response = $client->post('https://lwn.net/login', [
+            'cookies' => $this->cookieJar,
+            'body' => [
+                'Username' => $this->user,
+                'Password' => $this->password,
+                'target' => '/'
+            ]
+        ]);
+        return ($response->getStatusCode() === 200 && $this->cookieJar->count() > 0);
+    }
+
+    private function getBody($url) {
+        $client = $this->clientService->newClient();
+        $response = $client->get($url, [
+            'cookies' => $this->cookieJar
+        ]);
+        $parser = new RuleParser($response->getBody(), [
+            'body' => array(
+                '//div[@class="ArticleText"]',
+            ),
+            'strip' => array(
+                '//div[@class="FeatureByline"]'
+            )
+        ]);
+        $articleBody = $parser->execute();
+        // make all links absolute
+        return str_replace('href="/', 'href="https://lwn.net/', $articleBody);
+    }
+
+    private function canHandle($url) {
+        $regex = '%(?:https?://|//)?(?:www.)?lwn.net%';
+
+        return (bool)preg_match($regex, $url);
+    }
+
+    /**
+     * Execute Item Processor
+     *
+     * @access public
+     * @param  Feed $feed
+     * @param  Item $item
+     * @return bool
+     */
+    public function execute(Feed $feed, Item $item) {
+        if ($this->canHandle($item->getUrl())) {
+            $loggedIn = $this->login();
+
+            $item->setUrl(str_replace('/rss', '', $item->getUrl()));
+            if ($loggedIn) {
+                $item->setContent($this->getBody($item->getUrl()));
+            }
+        }
+    }
+}
diff --git a/tests/Unit/Fetcher/FeedFetcherTest.php b/tests/Unit/Fetcher/FeedFetcherTest.php
index 930cf4c99..ce09bb0e3 100644
--- a/tests/Unit/Fetcher/FeedFetcherTest.php
+++ b/tests/Unit/Fetcher/FeedFetcherTest.php
@@ -15,6 +15,8 @@ namespace OCA\News\Fetcher;
 
 use \OCA\News\Db\Item;
 use \OCA\News\Db\Feed;
+use OCP\Http\Client\IClientService;
+use PicoFeed\Processor\ItemPostProcessor;
 
 
 class FeedFetcherTest extends \PHPUnit_Framework_TestCase {
@@ -97,11 +99,19 @@ class FeedFetcherTest extends \PHPUnit_Framework_TestCase {
         $timeFactory->expects($this->any())
             ->method('getTime')
             ->will($this->returnValue($this->time));
+        $postProcessor = $this->getMockBuilder(ItemPostProcessor::class)
+            ->getMock();
+        $this->parser->expects($this->any())
+            ->method('getItemPostProcessor')
+            ->will($this->returnValue($postProcessor));
+        $clientService = $this->getMockBuilder(IClientService::class)
+            ->getMock();
         $this->fetcher = new FeedFetcher(
                         $this->reader,
                         $this->faviconFactory,
                         $this->l10n,
-                        $timeFactory);
+                        $timeFactory,
+                        $clientService);
         $this->url = 'http://tests';
 
         $this->permalink = 'http://permalink';
-- 
cgit v1.2.3