9 files changed, 198 insertions, 91 deletions

diff --git a/inc/3rdparty/site_config/custom/blogs.faz.net.txt b/inc/3rdparty/site_config/custom/blogs.faz.net.txt
new file mode 100644
index 00000000..4f2626f1
--- /dev/null
+++ b/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/
diff --git a/inc/3rdparty/site_config/standard/faz.net.txt b/inc/3rdparty/site_config/standard/faz.net.txt
index d087d2aa..47048a1b 100755..100644
--- a/inc/3rdparty/site_config/standard/faz.net.txt
+++ b/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
+
+# Fetch full multipage articles
+next_page_link: //a[@title='Nächste Seite']
 
-# Body
+# 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( ·  ):
+
+# 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
+
+# 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
 
-# General removements
-strip: //span[@class='Bildnachweis']
-strip: //img[@class='MediaIcon']
-strip: //div[@class='ArtikelMediaLink']
-dissolve: //a[img]
-
-# 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']
-
-# 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
\ No newline at end of file
diff --git a/inc/poche/Database.class.php b/inc/poche/Database.class.php
index 2c80b64b..b5dd2120 100755
--- a/inc/poche/Database.class.php
+++ b/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;
@@ -113,10 +117,10 @@ class Database {
         $query = $this->executeQuery($sql, array());
     }
 
-    public function install($login, $password)
+    public function install($login, $password, $email = '')
     {
         $sql = 'INSERT INTO users ( username, password, name, email) VALUES (?, ?, ?, ?)';
-        $params = array($login, $password, $login, ' ');
+        $params = array($login, $password, $login, $email);
         $query = $this->executeQuery($sql, $params);
 
         $sequence = '';
diff --git a/inc/poche/Poche.class.php b/inc/poche/Poche.class.php
index 098dd7c1..8cebafa3 100755
--- a/inc/poche/Poche.class.php
+++ b/inc/poche/Poche.class.php
@@ -74,12 +74,13 @@ class Poche
     /**
      * Creates a new user
      */
-    public function createNewUser($username, $password)
+    public function createNewUser($username, $password, $email = "")
     {
         if (!empty($username) && !empty($password)){
             $newUsername = filter_var($username, FILTER_SANITIZE_STRING);
+            $email = filter_var($email, FILTER_SANITIZE_STRING);
             if (!$this->store->userExists($newUsername)){
-                if ($this->store->install($newUsername, Tools::encodeString($password . $newUsername))) {
+                if ($this->store->install($newUsername, Tools::encodeString($password . $newUsername), $email)) {
                     Tools::logm('The new user ' . $newUsername . ' has been installed');
                     $this->messages->add('s', sprintf(_('The new user %s has been installed. Do you want to <a href="?logout">logout ?</a>'), $newUsername));
                     Tools::redirect();
@@ -524,6 +525,14 @@ 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 +560,39 @@ 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
-
-            $str_data = file_get_contents($_FILES['file']['tmp_name']);
-            $data = json_decode($str_data, true);
-            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) && $read) ? 0 : 1);
-                    }
-                }
+    public function import() {
+
+      if ( isset($_FILES['file']) && $_FILES['file']['tmp_name'] ) {
+        Tools::logm('Import stated: parsing file');
+
+        // assume, that file is in json format
+        $str_data = file_get_contents($_FILES['file']['tmp_name']);
+        $data = json_decode($str_data, true);
+
+        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 +635,11 @@ 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 {
+        $this->messages->add('s', _('Did you forget to select a file?'));
+      }
         // file parsing finished here
         // now download article contents if any
         // check if we need to download any content
diff --git a/inc/poche/Routing.class.php b/inc/poche/Routing.class.php
index 004bd45a..0b373058 100755
--- a/inc/poche/Routing.class.php
+++ b/inc/poche/Routing.class.php
@@ -124,8 +124,6 @@ class Routing
             } elseif (isset($_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'])) {
diff --git a/inc/poche/Tools.class.php b/inc/poche/Tools.class.php
index 63137d76..93ec3fc6 100755
--- a/inc/poche/Tools.class.php
+++ b/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"]);
 
@@ -295,21 +299,6 @@ 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)
      *
      * @param Url $url
diff --git a/inc/poche/WallabagEpub.class.php b/inc/poche/WallabagEpub.class.php
index b81d9bfd..9c4d3566 100644
--- a/inc/poche/WallabagEpub.class.php
+++ b/inc/poche/WallabagEpub.class.php
@@ -12,14 +12,12 @@ class WallabagEpub
 {
     protected $wallabag;
     protected $method;
-    protected $id;
     protected $value;
 
-    public function __construct(Poche $wallabag, $method, $id, $value)
+    public function __construct(Poche $wallabag, $method, $value)
     {
         $this->wallabag = $wallabag;
         $this->method   = $method;
-        $this->id       = $id;
         $this->value    = $value;
     }
 
@@ -30,7 +28,7 @@ class WallabagEpub
     {
         switch ($this->method) {
             case 'id':
-                $entryID = filter_var($this->id, FILTER_SANITIZE_NUMBER_INT);
+                $entryID = filter_var($this->value, FILTER_SANITIZE_NUMBER_INT);
                 $entry = $this->wallabag->store->retrieveOneById($entryID, $this->wallabag->user->getId());
                 $entries = array($entry);
                 $bookTitle = $entry['title'];
@@ -87,7 +85,7 @@ class WallabagEpub
         $log->logLine("getCurrentServerURL: " . $book->getCurrentServerURL());
         $log->logLine("getCurrentPageURL..: " . $book->getCurrentPageURL());
 
-        $book->setTitle(_('wallabag\'s articles'));
+        $book->setTitle($bookTitle);
         $book->setIdentifier("http://$_SERVER[HTTP_HOST]", EPub::IDENTIFIER_URI); // Could also be the ISBN number, prefered for published books, or a UUID.
         //$book->setLanguage("en"); // Not needed, but included for the example, Language is mandatory, but EPub defaults to "en". Use RFC3066 Language codes, such as "en", "da", "fr" etc.
         $book->setDescription(_("Some articles saved on my wallabag"));
diff --git a/inc/poche/config.inc.default.php b/inc/poche/config.inc.default.php
index 6f03af18..2a458544 100755
--- a/inc/poche/config.inc.default.php
+++ b/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);
diff --git a/inc/poche/pochePictures.php b/inc/poche/pochePictures.php
index 7a914f90..52394c70 100644
--- a/inc/poche/pochePictures.php
+++ b/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;