3 files changed, 98 insertions, 81 deletions

diff --git a/inc/poche/Database.class.php b/inc/poche/Database.class.php
index 6244df88..036c9d1b 100755
--- a/inc/poche/Database.class.php
+++ b/inc/poche/Database.class.php
@@ -77,7 +77,7 @@ class Database {
         }
         else {
             $sql = '
-                CREATE TABLE tags (
+                CREATE TABLE IF NOT EXISTS tags (
                   id bigserial primary key,
                   value varchar(255) NOT NULL
                 );
@@ -110,7 +110,7 @@ class Database {
         }
         else {
             $sql = '
-                CREATE TABLE tags_entries (
+                CREATE TABLE IF NOT EXISTS tags_entries (
                   id bigserial primary key,
                   entry_id integer NOT NULL,
                   tag_id integer NOT NULL
@@ -245,7 +245,7 @@ class Database {
             $sql_limit = "LIMIT ".$limit." OFFSET 0";
         }
 
-        $sql        = "SELECT * FROM entries WHERE (content = '' OR content IS NULL) AND user_id=? ORDER BY id " . $sql_limit;
+        $sql        = "SELECT * FROM entries WHERE (content = '' OR content IS NULL) AND title LIKE 'Untitled - Import%' AND user_id=? ORDER BY id " . $sql_limit;
         $query      = $this->executeQuery($sql, array($user_id));
         $entries    = $query->fetchAll();
 
@@ -253,7 +253,7 @@ class Database {
     }
 
     public function retrieveUnfetchedEntriesCount($user_id) {
-      $sql        = "SELECT count(*) FROM entries WHERE (content = '' OR content IS NULL) AND user_id=?";
+      $sql        = "SELECT count(*) FROM entries WHERE (content = '' OR content IS NULL) AND title LIKE 'Untitled - Import%' AND user_id=?";
       $query      = $this->executeQuery($sql, array($user_id));
       list($count) = $query->fetch();
 
@@ -374,7 +374,7 @@ class Database {
           $id = null;
         }
         else {
-          $id = intval($this->getLastId( (STORAGE == 'postgres') ? 'users_id_seq' : '' ));
+          $id = intval($this->getLastId( (STORAGE == 'postgres') ? 'entries_id_seq' : '') );
         }
         return $id;
     }
@@ -407,7 +407,7 @@ class Database {
     public function getLastId($column = '') {
         return $this->getHandle()->lastInsertId($column);
     }
-        
+
     public function search($term, $user_id, $limit = '') {
         $search = '%'.$term.'%';
         $sql_action = "SELECT * FROM entries WHERE user_id=? AND (content LIKE ? OR title LIKE ? OR url LIKE ?) "; //searches in content, title and URL
diff --git a/inc/poche/Poche.class.php b/inc/poche/Poche.class.php
index a662f695..811895dc 100755
--- a/inc/poche/Poche.class.php
+++ b/inc/poche/Poche.class.php
@@ -373,9 +373,7 @@ class Poche
                 $body = $content['rss']['channel']['item']['description'];
 
                 // clean content from prevent xss attack
-                $config = HTMLPurifier_Config::createDefault();
-                $config->set('Cache.SerializerPath', CACHE);
-                $purifier = new HTMLPurifier($config);
+                $purifier = $this->getPurifier();
                 $title = $purifier->purify($title);
                 $body = $purifier->purify($body);
 
@@ -828,10 +826,12 @@ class Poche
         define('IMPORT_LIMIT', 5);
       }
       if (!defined('IMPORT_DELAY')) {
-        define('IMPORT_DELAY', 5);
+        define('IMPORT_DELAY', 5);
       }
 
       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);
@@ -844,18 +844,18 @@ class Poche
           $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) && $read)?0:1);
+              foreach ($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);
             }
           }
         }
@@ -866,16 +866,16 @@ class Poche
             $data[] = $record;
             foreach ($record as $record2) {
               if (is_array($record2)) {
-                $data[] = $record2;
+                $data[] = $record2;
               }
             }
           }
         }
 
-        $i = 0; //counter for articles inserted
+        $urlsInserted = array(); //urls of articles inserted
         foreach ($data as $record) {
           $url = trim( isset($record['article__url']) ? $record['article__url'] : (isset($record['url']) ? $record['url'] : '') );
-          if ( $url ) {
+          if ( $url and !in_array($url, $urlsInserted) ) {
             $title = (isset($record['title']) ? $record['title'] :  _('Untitled - Import - ').'</a> <a href="./?import">'._('click to finish import').'</a><a>');
             $body = (isset($record['content']) ? $record['content'] : '');
             $isRead = (isset($record['is_read']) ? intval($record['is_read']) : (isset($record['archive'])?intval($record['archive']):0));
@@ -883,19 +883,21 @@ class Poche
             //insert new record
             $id = $this->store->add($url, $title, $body, $this->user->getId(), $isFavorite, $isRead);
             if ( $id ) {
-              //increment no of records inserted
-              $i++;
+              $urlsInserted[] = $url; //add
+
               if ( isset($record['tags']) && trim($record['tags']) ) {
-                //@TODO: set tags
+                //@TODO: set tags
 
               }
             }
           }
         }
 
+        $i = sizeof($urlsInserted);
         if ( $i > 0 ) {
           $this->messages->add('s', _('Articles inserted: ').$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).');
       }
       //file parsing finished here
 
@@ -906,30 +908,32 @@ class Poche
       if ( $recordsDownloadRequired == 0 ) {
         //nothing to download
         $this->messages->add('s', _('Import finished.'));
+        Tools::logm('Import finished completely');
         Tools::redirect();
       }
       else {
         //if just inserted - don't download anything, download will start in next reload
         if ( !isset($_FILES['file']) ) {
           //download next batch
+          Tools::logm('Fetching next batch of articles...');
           $items = $this->store->retrieveUnfetchedEntries($this->user->getId(), IMPORT_LIMIT);
 
-          $config = HTMLPurifier_Config::createDefault();
-          $config->set('Cache.SerializerPath', CACHE);
-          $purifier = new HTMLPurifier($config);
+          $purifier = $this->getPurifier();
 
           foreach ($items as $item) {
-                $url = new Url(base64_encode($item['url']));
-                $content = Tools::getPageContent($url);
+            $url = new Url(base64_encode($item['url']));
+            Tools::logm('Fetching article '.$item['id']);
+            $content = Tools::getPageContent($url);
 
-                $title = (($content['rss']['channel']['item']['title'] != '') ? $content['rss']['channel']['item']['title'] : _('Untitled'));
-                $body = (($content['rss']['channel']['item']['description'] != '') ? $content['rss']['channel']['item']['description'] : _('Undefined'));
+            $title = (($content['rss']['channel']['item']['title'] != '') ? $content['rss']['channel']['item']['title'] : _('Untitled'));
+            $body = (($content['rss']['channel']['item']['description'] != '') ? $content['rss']['channel']['item']['description'] : _('Undefined'));
 
-                //clean content to prevent xss attack
-                $title = $purifier->purify($title);
-                $body = $purifier->purify($body);
+            //clean content to prevent xss attack
+            $title = $purifier->purify($title);
+            $body = $purifier->purify($body);
 
-                $this->store->updateContentAndTitle($item['id'], $title, $body, $this->user->getId());
+            $this->store->updateContentAndTitle($item['id'], $title, $body, $this->user->getId());
+            Tools::logm('Article '.$item['id'].' updated.');
           }
 
         }
@@ -942,16 +946,15 @@ class Poche
      * export poche entries in json
      * @return json all poche entries
      */
-    public function export()
-    {
-                $filename = "wallabag-export-".$this->user->getId()."-".date("Y-m-d").".json";
-                header('Content-Disposition: attachment; filename='.$filename);
-
-        $entries = $this->store->retrieveAll($this->user->getId());
-        echo $this->tpl->render('export.twig', array(
-            'export' => Tools::renderJson($entries),
-        ));
-        Tools::logm('export view');
+    public function export() {
+      $filename = "wallabag-export-".$this->user->getId()."-".date("Y-m-d").".json";
+      header('Content-Disposition: attachment; filename='.$filename);
+
+      $entries = $this->store->retrieveAll($this->user->getId());
+      echo $this->tpl->render('export.twig', array(
+          'export' => Tools::renderJson($entries),
+      ));
+      Tools::logm('export view');
     }
 
     /**
@@ -959,43 +962,42 @@ class Poche
      * @param  string $which 'prod' or 'dev'
      * @return string        latest $which version
      */
-    private function getPocheVersion($which = 'prod')
-    {
-        $cache_file = CACHE . '/' . $which;
-        $check_time = time();
-
-        # checks if the cached version file exists
-        if (file_exists($cache_file) && (filemtime($cache_file) > (time() - 86400 ))) {
-           $version = file_get_contents($cache_file);
-           $check_time = filemtime($cache_file);
-        } else {
-           $version = file_get_contents('http://static.wallabag.org/versions/' . $which);
-           file_put_contents($cache_file, $version, LOCK_EX);
-        }
-        return array($version, $check_time);
+    private function getPocheVersion($which = 'prod') {
+      $cache_file = CACHE . '/' . $which;
+      $check_time = time();
+
+      # checks if the cached version file exists
+      if (file_exists($cache_file) && (filemtime($cache_file) > (time() - 86400 ))) {
+         $version = file_get_contents($cache_file);
+         $check_time = filemtime($cache_file);
+      } else {
+         $version = file_get_contents('http://static.wallabag.org/versions/' . $which);
+         file_put_contents($cache_file, $version, LOCK_EX);
+      }
+      return array($version, $check_time);
     }
 
     public function generateToken()
     {
-        if (ini_get('open_basedir') === '') {
-                        if (strtoupper(substr(PHP_OS, 0, 3)) === 'WIN') {
-                        echo 'This is a server using Windows!';
-                        // alternative to /dev/urandom for Windows
-                        $token = substr(base64_encode(uniqid(mt_rand(), true)), 0, 20);
-                        } else {
-                        $token = substr(base64_encode(file_get_contents('/dev/urandom', false, null, 0, 20)), 0, 15);
-                        }
-        }
-        else {
-            $token = substr(base64_encode(uniqid(mt_rand(), true)), 0, 20);
+      if (ini_get('open_basedir') === '') {
+        if (strtoupper(substr(PHP_OS, 0, 3)) === 'WIN') {
+          echo 'This is a server using Windows!';
+          // alternative to /dev/urandom for Windows
+          $token = substr(base64_encode(uniqid(mt_rand(), true)), 0, 20);
+        } else {
+          $token = substr(base64_encode(file_get_contents('/dev/urandom', false, null, 0, 20)), 0, 15);
         }
+      }
+      else {
+        $token = substr(base64_encode(uniqid(mt_rand(), true)), 0, 20);
+      }
 
-        $token = str_replace('+', '', $token);
-        $this->store->updateUserConfig($this->user->getId(), 'token', $token);
-        $currentConfig = $_SESSION['poche_user']->config;
-        $currentConfig['token'] = $token;
-        $_SESSION['poche_user']->setConfig($currentConfig);
-        Tools::redirect();
+      $token = str_replace('+', '', $token);
+      $this->store->updateUserConfig($this->user->getId(), 'token', $token);
+      $currentConfig = $_SESSION['poche_user']->config;
+      $currentConfig['token'] = $token;
+      $_SESSION['poche_user']->setConfig($currentConfig);
+      Tools::redirect();
     }
 
     public function generateFeeds($token, $user_id, $tag_id, $type = 'home')
@@ -1031,6 +1033,7 @@ class Poche
             foreach ($entries as $entry) {
                 $newItem = $feed->createNewItem();
                 $newItem->setTitle($entry['title']);
+                $newItem->setSource(Tools::getPocheUrl() . '?view=view&amp;id=' . $entry['id']);
                 $newItem->setLink($entry['url']);
                 $newItem->setDate(time());
                 $newItem->setDescription($entry['content']);
@@ -1057,4 +1060,16 @@ class Poche
         $this->messages->add('s', _('Cache deleted.'));
         Tools::redirect();
     }
+
+    /**
+     * return new purifier object with actual config
+     */
+    protected function getPurifier() {
+      $config = HTMLPurifier_Config::createDefault();

+      $config->set('Cache.SerializerPath', CACHE);

+      $config->set('HTML.SafeIframe', true);

+      $config->set('URI.SafeIframeRegexp', '%^(https?:)?//(www\.youtube(?:-nocookie)?\.com/embed/|player\.vimeo\.com/video/)%'); //allow YouTube and Vimeo$purifier = new HTMLPurifier($config);
+

+      return new HTMLPurifier($config);
+    }
 }
diff --git a/inc/poche/Tools.class.php b/inc/poche/Tools.class.php
index a130e94b..7f064020 100755
--- a/inc/poche/Tools.class.php
+++ b/inc/poche/Tools.class.php
@@ -59,8 +59,10 @@ class Tools
             return $scriptname;
         }
 
+        $host = (isset($_SERVER['HTTP_X_FORWARDED_HOST']) ? $_SERVER['HTTP_X_FORWARDED_HOST'] : (isset($_SERVER['HTTP_HOST']) ? $_SERVER['HTTP_HOST'] : $_SERVER['SERVER_NAME']));
+        
         return 'http' . ($https ? 's' : '') . '://'
-            . $_SERVER["HTTP_HOST"] . $serverport . $scriptname;
+            . $host . $serverport . $scriptname;
     }
 
     public static function redirect($url = '')