merge refactor and dev

inc/3rdparty/libraries/readability/Readability.php

@@ -679,6 +679,7 @@ class Readability
 				} else {
 					$topCandidate->innerHTML = $page->documentElement->innerHTML;
 					$page->documentElement->innerHTML = '';
+					$this->reinitBody();
 					$page->documentElement->appendChild($topCandidate);
 				}
 			} else {
@@ -794,8 +795,7 @@ class Readability
 		{
 			// TODO: find out why element disappears sometimes, e.g. for this URL http://www.businessinsider.com/6-hedge-fund-etfs-for-average-investors-2011-7
 			// in the meantime, we check and create an empty element if it's not there.
-			if (!isset($this->body->childNodes)) $this->body = $this->dom->createElement('body');
-			$this->body->innerHTML = $this->bodyCache;
+			$this->reinitBody();
 			
 			if ($this->flagIsActive(self::FLAG_STRIP_UNLIKELYS)) {
 				$this->removeFlag(self::FLAG_STRIP_UNLIKELYS);
@@ -1134,5 +1134,18 @@ class Readability
 	public function removeFlag($flag) {
 		$this->flags = $this->flags & ~$flag;
 	}
+	
+	/**
+	 * Will recreate previously deleted body property
+	 *
+	 * @return void
+	 */
+	protected function reinitBody() {
+		if (!isset($this->body->childNodes)) {
+			$this->body = $this->dom->createElement('body');
+			$this->body->innerHTML = $this->bodyCache;
+		}
+	}
+		
 }
 ?>

inc/3rdparty/site_config/custom/blogs.faz.net.txt

@@ -0,0 +1,45 @@
+# Author: zinnober
+
+tidy: no
+prune: no
+
+# Set author
+author: //a[@rel='author']
+
+# Set date
+date: //span[@class='Datum']
+
+# Content is here
+body: //div[@class='Artikel']
+
+# Tidy up before article
+strip: //div[@id='FAZHeaderNeu']
+strip: //h2[@itemprop='headline']
+strip: //span[@class='Datum']
+strip: //span[@class='Autor']
+strip_id_or_class: ArticlePagerTop
+strip: //div[@class='FAZArtikelEinleitung']/h2
+
+# General cleanup
+strip: //div[@class='clear']
+strip: //span[@class='Bildnachweis']
+strip: //iframe
+strip_id_or_class: Community
+strip: ' ·  '
+
+# Remove tracking and ads
+strip_image_src: /l.gif?
+strip: //img[@width='1']
+strip_id_or_class: invisible
+strip_id_or_class: Anzeige
+strip_id_or_class: billboard
+
+# Remove clutter after article
+strip_id_or_class: Tagline
+strip_id_or_class: ArtikelAbbinder
+strip_id_or_class: FAZArtikelKommentare
+strip_id_or_class: ArtikelKommentieren
+strip_id_or_class: FAZContentRight
+
+# Try it yourself
+test_url: http://blogs.faz.net/wost/2014/08/17/viel-fuck-und-wenig-guter-sex-1239/

inc/3rdparty/site_config/standard/.about.com.txt

@@ -0,0 +1,14 @@
+body: //div[@id='articlebody']
+title: //h1
+author: //p[@id='by']//a
+
+next_page_link: //span[@class='next']/a
+# Not the same as below!
+
+prune: yes
+tidy: no
+
+# Annoying 'next' links plainly inside the article body
+strip: //*[text()[contains(.,'Next: ')]]
+
+test_url: http://psychology.about.com/od/theoriesofpersonality/ss/defensemech.htm

inc/3rdparty/site_config/standard/dn.pt.txt

@@ -0,0 +1,9 @@
+single_page_link: concat('http://www.dn.pt/Common/print.aspx?content_id=', //input[@type='hidden' and @name='link-comments']/@value)
+#<input type="hidden" name="link-comments" class="link-comments" value="3972244">
+
+title: //h1
+author: //div[@class="Author"]
+
+strip: //div[@class="Patrocinio"]
+
+test_url: http://www.dn.pt/inicio/opiniao/interior.aspx?content_id=3972244&seccao=Alberto%20Gon%E7alves&tag=Opini%E3o%20-%20Em%20Foco&page=1

inc/3rdparty/site_config/standard/faz.net.txt

@@ -1,36 +1,101 @@
+# Author: zinnober
+# Complete rewrite of the faz.net template as the standard one is broken
+# I tried to consider as many page variants as possible, which was some serious work
+
+tidy: no
+prune: no
+
 # Title
 title: //p[@class='Content HeadlineShort']
 
-# Authors
-# some are known and have a link, others don't
-author: substring-after(//span[@class='Autor'], 'Von')
+# Set author
+author: substring-after(//span[@class='Autor'], 'von ')
+author: //span[@class='caps last']/span[@class='caps last']
+author: //a[@rel='author']
 
-# Date
+# Set date
 date: //span[@class='Datum']
+date: //span[@class='Datum'],/span
 
-# Body
+# Fetch full multipage articles
+next_page_link: //a[@title='Nächste Seite']
+
+# Content is here
 body: //div[@class='Artikel']
 
-# Removements before body text
-strip: //div[@class='Breadcrumbs']
-strip: //div[@class='QuickSearchBox']
-strip: //div[@class='FAZArtikelEinleitung']
-strip: //div[@class='FAZArtikelReiter']
+# Tidy up before article
+strip: //div[@id='FAZHeaderNeu']
+strip: //h2[@itemprop='headline']
+strip: //span[@class='Datum']
+strip: //span[@class='Autor']
+strip_id_or_class: ArticlePagerTop
+
+# General cleanup
 strip: //div[@class='clear']
+strip: //a[@title='Zur Homepage FAZ.NET']
+strip: //iframe
+replace_string( ·  ):
 
-# General removements
-strip: //span[@class='Bildnachweis']
-strip: //img[@class='MediaIcon']
-strip: //div[@class='ArtikelMediaLink']
-dissolve: //a[img]
+# Remove tracking and ads
+strip_image_src: /l.gif?
+strip: //div[contains(@style, 'background-image')]
+strip: //img[@width='1']
+strip_id_or_class: invisible
+strip_id_or_class: Anzeige
+strip_id_or_class: billboard
 
-# Removements after body text
-strip: //div[@class='ArtikelAbbinder']
-strip: //div[@class='ArtikelKommentieren Artikelfuss GETS;tk;boxen.top-lesermeinungen;tp;content']
-strip: //div[@class='FAZArtikelKommentare FAZArtikelContent']
-strip: //div[@class='FAZArtikelFunktionen']
-strip: //div[@id='FAZContentRight']
+# Remove various text boxes and social media foo
+strip_id_or_class: WeitereBeitraege
+strip_id_or_class: WBListe
+strip_id_or_class: AutorenModul
+strip_id_or_class: Community
+strip_id_or_class: SocialMediaStatus
+strip_id_or_class: RelatedLinkBox
+strip_id_or_class: MultimediaNavigation
+strip_id_or_class: IndexTitel
+
+# Fix picture caps and pictures (use better resolution and remove clutter)
+strip_id_or_class: LightBoxOverlay
+strip_id_or_class: exitLarge
+strip_id_or_class: PagerBox
+strip_id_or_class: Bildnachweis
+strip_id_or_class: Bildueberschrift
+strip_id_or_class: Bildbeschreibung
+strip_id_or_class: ArtikelBild610
+strip_id_or_class: MediaLink
+strip_id_or_class: FotoBoxInnerLeft
+strip_id_or_class: BilderRelatedLinks
+
+# Remove clutter after article
+strip_id_or_class: ArticlePagerBottom
+strip_id_or_class: backToHome
+strip_id_or_class: ArtikelAbbinder
+strip_id_or_class: lesermeinungscontainer
+strip_id_or_class: ThemenLinks
+strip_id_or_class: rechtehinweis
+strip_id_or_class: FAZArtikelMap
+strip_id_or_class: FAZArtikelKommentare
+strip_id_or_class: ArtikelKommentieren
+strip_id_or_class: FAZArtikelFunktionen
+strip_id_or_class: mailLB
+strip_id_or_class: FAZContentRight
+strip_id_or_class: stageModule
+strip_id_or_class: ContentFooter
+strip_id_or_class: ServicesFooter
+strip_id_or_class: FAZFooter
+
+# Clean up stuff present just in some articles
+strip_id_or_class: Teaser620
+strip_id_or_class: TeaserMultimedia
+strip_id_or_class: VideoBox
+
+# Remove as soon as Wallabag maight be able to embed flash video
+strip_id_or_class: mmoObjectAsTeaserInArticle
+strip_id_or_class: additionalStylesAudioVideo
+strip_id_or_class: hideMMElements
+
+# Try it yourself
+test_url: http://www.faz.net/aktuell/feuilleton/zum-tod-von-margaret-thatcher-die-reizfigur-12141919.html#Drucken
+test_url: http://www.faz.net/aktuell/politik/inland/allensbach-analyse-im-namen-des-volkes-13106492.html
+test_url: http://www.faz.net/aktuell/feuilleton/kino/video-filmkritiken/video-filmkritik-when-animals-dream-zerrissene-jugend-13105772.html
 
-# Fix picture captions
-wrap_in(small): //span[@class='Bildunterschrift']/text()
-test_url: http://www.faz.net/aktuell/feuilleton/zum-tod-von-margaret-thatcher-die-reizfigur-12141919.html#Drucken

inc/3rdparty/site_config/standard/habrahabr.ru.txt

@@ -0,0 +1,21 @@
+title: //span[@class="post_title"]
+author: //div[@class="author"]
+date: //div[@class="published
+
+body: //div[@class='content html_format'] | //div[@id='comments']
+
+strip: //a[@class="link_to_comment"]
+strip: //div[@class="show_tree"]
+strip: //a[@class="to_parent"]
+
+
+replace_string(class="reply_comments"): style="padding-left: 20px"
+replace_string(class="voting   "): style="float: right"
+replace_string(src="//habrastorage.org/getpro/habr/avatars/): style="width:24px; height:24px;" class="123" src="//habrastorage.org/getpro/habr/avatars/
+replace_string(class="info  "): style="padding-top:5px;font-size:0.85em;line-height:24px;"
+
+
+prune: no
+tidy: no
+
+test_url: http://habrahabr.ru/post/229883/

inc/poche/Database.class.php

@@ -23,6 +23,10 @@ class Database {
     {
         switch (STORAGE) {
             case 'sqlite':
+                // Check if /db is writeable
+                if ( !is_writable(STORAGE_SQLITE) || !is_writable(dirname(STORAGE_SQLITE))) {
+                	die('An error occured: "db" directory must be writeable for your web server user!');
+                }
                 $db_path = 'sqlite:' . STORAGE_SQLITE;
                 $this->handle = new PDO($db_path);
                 break;

inc/poche/Poche.class.php

@@ -524,6 +524,14 @@ class Poche
                 $longlastingsession = isset($_POST['longlastingsession']);
                 $passwordTest = ($isauthenticated) ? $user['password'] : Tools::encodeString($password . $login);
                 Session::login($user['username'], $user['password'], $login, $passwordTest, $longlastingsession, array('poche_user' => new User($user)));
+
+                # reload l10n
+                $language = $user['config']['language'];
+                @putenv('LC_ALL=' . $language);
+                setlocale(LC_ALL, $language);
+                bindtextdomain($language, LOCALE);
+                textdomain($language);
+
                 $this->messages->add('s', _('welcome to your wallabag'));
                 Tools::logm('login successful');
                 Tools::redirect($referer);
@@ -551,42 +559,39 @@ class Poche
      * import datas into your wallabag
      * @return boolean
      */
-    public function import()
-    {
-        if (isset($_FILES['file'])) {
-            Tools::logm('Import stated: parsing file');
 
-            // assume, that file is in json format
+    public function import() {
 
-            $str_data = file_get_contents($_FILES['file']['tmp_name']);
-            $data = json_decode($str_data, true);
-            if ($data === null) {
+      if ( isset($_FILES['file']) && $_FILES['file']['tmp_name'] ) {
+        Tools::logm('Import stated: parsing file');
 
-                // not json - assume html
+        // assume, that file is in json format
+        $str_data = file_get_contents($_FILES['file']['tmp_name']);
+        $data = json_decode($str_data, true);
 
-                $html = new simple_html_dom();
-                $html->load_file($_FILES['file']['tmp_name']);
-                $data = array();
-                $read = 0;
-                foreach(array('ol','ul') as $list) {
-                    foreach($html->find($list) as $ul) {
-                        foreach($ul->find('li') as $li) {
-                            $tmpEntry = array();
-                            $a = $li->find('a');
-                            $tmpEntry['url'] = $a[0]->href;
-                            $tmpEntry['tags'] = $a[0]->tags;
-                            $tmpEntry['is_read'] = $read;
-                            if ($tmpEntry['url']) {
-                                $data[] = $tmpEntry;
-                            }
-                        }
-
-                        // the second <ol/ul> is for read links
-
-                        $read = ((sizeof($data) && $read) ? 0 : 1);
-                    }
-                }
+        if ( $data === null ) {
+          //not json - assume html
+          $html = new simple_html_dom();
+          $html->load_file($_FILES['file']['tmp_name']);
+          $data = array();
+          $read = 0;
+          foreach (array('ol','ul') as $list) {
+            foreach ($html->find($list) as $ul) {
+              foreach ($ul->find('li') as $li) {
+                $tmpEntry = array();
+                  $a = $li->find('a');
+                  $tmpEntry['url'] = $a[0]->href;
+                  $tmpEntry['tags'] = $a[0]->tags;
+                  $tmpEntry['is_read'] = $read;
+                  if ($tmpEntry['url']) {
+                    $data[] = $tmpEntry;
+                  }
+              }
+              # the second <ol/ul> is for read links
+              $read = ((sizeof($data) && $read)?0:1);
             }
+          }
+    	}
 
             // for readability structure
 
@@ -629,9 +634,11 @@ class Poche
                 $this->messages->add('s', _('Articles inserted: ') . $i . _('. Please note, that some may be marked as "read".'));
             }
 
-            Tools::logm('Import of articles finished: ' . $i . ' articles added (w/o content if not provided).');
-        }
-
+        Tools::logm('Import of articles finished: '.$i.' articles added (w/o content if not provided).');
+      }
+      else {
+        $this->messages->add('s', _('Did you forget to select a file?'));
+      }
         // file parsing finished here
         // now download article contents if any
         // check if we need to download any content
@@ -750,8 +757,8 @@ class Poche
             die(sprintf(_('User with this id (%d) does not exist.'), $user_id));
         }
 
-        if (!in_array($type, $allowed_types) || $token != $config['token']) {
-            die(_('Uh, there is a problem while generating feeds.'));
+        if (!in_array($type, $allowed_types) || !isset($config['token']) || $token != $config['token']) {
+            die(_('Uh, there is a problem while generating feed. Wrong token used?'));
         }
 
         $feed = new FeedWriter(RSS2);

inc/poche/Routing.class.php

@@ -98,49 +98,50 @@ class Routing
     private function _launchAction()
     {
         if (isset($_GET['login'])) {
-            // hello you
-            $this->wallabag->login($this->referer);
-        } elseif (isset($_GET['logout'])) {
-            // see you soon !
-            $this->wallabag->logout();
-        } elseif (isset($_GET['config'])) {
-            // update password
-            $this->wallabag->updatePassword($_POST['password'], $_POST['password_repeat']);
-        } elseif (isset($_GET['newuser'])) {
-            $this->wallabag->createNewUser($_POST['newusername'], $_POST['password4newuser']);
-        } elseif (isset($_GET['deluser'])) {
-            $this->wallabag->deleteUser($_POST['password4deletinguser']);
-        } elseif (isset($_GET['epub'])) {
-            $epub = new WallabagEpub($this->wallabag, $_GET['method'], $_GET['value']);
-            $epub->run();
-        } elseif (isset($_GET['import'])) {
-            $import = $this->wallabag->import();
-            $tplVars = array_merge($this->vars, $import);
-        } elseif (isset($_GET['download'])) {
-            Tools::downloadDb();
-        } elseif (isset($_GET['empty-cache'])) {
-            Tools::emptyCache();
-        } elseif (isset($_GET['export'])) {
-            $this->wallabag->export();
-        } elseif (isset($_GET['updatetheme'])) {
-            $this->wallabag->tpl->updateTheme($_POST['theme']);
-        } elseif (isset($_GET['updatelanguage'])) {
-            $this->wallabag->language->updateLanguage($_POST['language']);
-        } elseif (isset($_GET['uploadfile'])) {
-            $this->wallabag->uploadFile();
-        } elseif (isset($_GET['feed'])) {
-            if (isset($_GET['action']) && $_GET['action'] == 'generate') {
+        	// hello to you
+        	$this->wallabag->login($this->referer);
+        } elseif (isset($_GET['feed']) && isset($_GET['user_id'])) {
+            $tag_id = (isset($_GET['tag_id']) ? intval($_GET['tag_id']) : 0);
+            $this->wallabag->generateFeeds($_GET['token'], filter_var($_GET['user_id'],FILTER_SANITIZE_NUMBER_INT), $tag_id, $_GET['type']);
+        }
+        
+        //allowed ONLY to logged in user
+        if (\Session::isLogged() === true) 
+        {
+            if (isset($_GET['logout'])) {
+                // see you soon !
+                $this->wallabag->logout();
+            } elseif (isset($_GET['config'])) {
+                // update password
+                $this->wallabag->updatePassword($_POST['password'], $_POST['password_repeat']);
+            } elseif (isset($_GET['newuser'])) {
+                $this->wallabag->createNewUser($_POST['newusername'], $_POST['password4newuser']);
+            } elseif (isset($_GET['deluser'])) {
+                $this->wallabag->deleteUser($_POST['password4deletinguser']);
+            } elseif (isset($_GET['epub'])) {
+                $epub = new WallabagEpub($this->wallabag, $_GET['method'], $_GET['id'], $_GET['value']);
+                $epub->run();
+            } elseif (isset($_GET['import'])) {
+                $import = $this->wallabag->import();
+                $tplVars = array_merge($this->vars, $import);
+            } elseif (isset($_GET['empty-cache'])) {
+                Tools::emptyCache();
+            } elseif (isset($_GET['export'])) {
+                $this->wallabag->export();
+            } elseif (isset($_GET['updatetheme'])) {
+                $this->wallabag->tpl->updateTheme($_POST['theme']);
+            } elseif (isset($_GET['updatelanguage'])) {
+                $this->wallabag->language->updateLanguage($_POST['language']);
+            } elseif (isset($_GET['uploadfile'])) {
+                $this->wallabag->uploadFile();
+            } elseif (isset($_GET['feed']) && isset($_GET['action']) && $_GET['action'] == 'generate') {
                 $this->wallabag->updateToken();
             }
-            else {
-                $tag_id = (isset($_GET['tag_id']) ? intval($_GET['tag_id']) : 0);
-                $this->wallabag->generateFeeds($_GET['token'], filter_var($_GET['user_id'],FILTER_SANITIZE_NUMBER_INT), $tag_id, $_GET['type']);
+            elseif (isset($_GET['plainurl']) && !empty($_GET['plainurl'])) {
+                $plainUrl = new Url(base64_encode($_GET['plainurl']));
+                $this->wallabag->action('add', $plainUrl);
             }
         }
-        elseif (isset($_GET['plainurl']) && !empty($_GET['plainurl'])) {
-            $plainUrl = new Url(base64_encode($_GET['plainurl']));
-            $this->wallabag->action('add', $plainUrl);
-        }
     }
 
     public function _render($file, $vars)

inc/poche/Tools.class.php

@@ -54,6 +54,10 @@ final class Tools
             || ($https && $_SERVER["SERVER_PORT"] == '443')
             || ($https && $_SERVER["SERVER_PORT"]==SSL_PORT) //Custom HTTPS port detection
             ? '' : ':' . $_SERVER["SERVER_PORT"]);
+        
+        if (isset($_SERVER["HTTP_X_FORWARDED_PORT"])) {
+            $serverport = ':' . $_SERVER["HTTP_X_FORWARDED_PORT"];
+        }
 
         $scriptname = str_replace('/index.php', '/', $_SERVER["SCRIPT_NAME"]);
 
@@ -294,21 +298,6 @@ final class Tools
         }
     }
 
-    /**
-     * Download the sqlite database
-     */
-    public static function downloadDb()
-    {
-        header('Content-Disposition: attachment; filename="poche.sqlite.gz"');
-        self::_status(200);
-
-        header('Content-Transfer-Encoding: binary');
-        header('Content-Type: application/octet-stream');
-        echo gzencode(file_get_contents(STORAGE_SQLITE));
-
-        exit;
-    }
-
     /**
      * Get the content for a given URL (by a call to FullTextFeed)
      *

inc/poche/config.inc.default.php

@@ -59,7 +59,7 @@
 @define ('LOCALE', ROOT . '/locale');
 @define ('CACHE', ROOT . '/cache');
 
-@define ('PAGINATION', '10');
+@define ('PAGINATION', '12');
 
 //limit for download of articles during import
 @define ('IMPORT_LIMIT', 5);

inc/poche/pochePictures.php

@@ -33,7 +33,7 @@ final class Picture
                 }
 
                 if (self::_downloadPictures($absolute_path, $fullpath) === true) {
-                    $content = str_replace($matches[$i][2], $fullpath, $content);
+                    $content = str_replace($matches[$i][2], Tools::getPocheUrl() . $fullpath, $content);
                 }
 
                 $processing_pictures[] = $absolute_path;