From: Nicolas Lœuillet Date: Fri, 28 Feb 2014 20:49:38 +0000 (+0100) Subject: [add] cron to fetch content on imported entries X-Git-Tag: 1.6.0^2~41^2~1 X-Git-Url: https://git.immae.eu/?a=commitdiff_plain;h=53e3158dfe697ea59da1fa0e401e8da75ae13030;p=github%2Fwallabag%2Fwallabag.git [add] cron to fetch content on imported entries --- diff --git a/cron.php b/cron.php new file mode 100644 index 00000000..cc137ca2 --- /dev/null +++ b/cron.php @@ -0,0 +1,53 @@ +getConfigUser($user_id); + +if ($token != $config['token']) { + die(_('Uh, there is a problem with the cron.')); +} + +$items = $store->retrieveUnfetchedEntries($user_id, $limit); + +foreach ($items as $item) { + $url = new Url(base64_encode($item['url'])); + $content = Tools::getPageContent($url); + + $title = ($content['rss']['channel']['item']['title'] != '') ? $content['rss']['channel']['item']['title'] : _('Untitled'); + $body = $content['rss']['channel']['item']['description']; + + // // clean content from prevent xss attack + // $config = HTMLPurifier_Config::createDefault(); + // $purifier = new HTMLPurifier($config); + // $title = $purifier->purify($title); + // $body = $purifier->purify($body); + + + $store->updateContentAndTitle($item['id'], $title, $body, $user_id); +} \ No newline at end of file diff --git a/inc/poche/Database.class.php b/inc/poche/Database.class.php index c998fe14..edc775f5 100755 --- a/inc/poche/Database.class.php +++ b/inc/poche/Database.class.php @@ -230,8 +230,30 @@ class Database { } } + public function updateContentAndTitle($id, $title, $body, $user_id) { + $sql_action = 'UPDATE entries SET content = ?, title = ? WHERE id=? AND user_id=?'; + $params_action = array($body, $title, $id, $user_id); + $query = $this->executeQuery($sql_action, $params_action); + + return $query; + } + + public function retrieveUnfetchedEntries($user_id, $limit) { + + $sql_limit = "LIMIT 0,".$limit; + if (STORAGE == 'postgres') { + $sql_limit = "LIMIT ".$limit." OFFSET 0"; + } + + $sql = "SELECT * FROM entries WHERE (content = '' OR content IS NULL) AND user_id=? ORDER BY id " . $sql_limit; + $query = $this->executeQuery($sql, array($user_id)); + $entries = $query->fetchAll(); + + return $entries; + } + public function retrieveAll($user_id) { - $sql = "SELECT * FROM entries WHERE user_id=? ORDER BY id"; + $sql = "SELECT * FROM entries WHERE content <> '' AND user_id=? ORDER BY id"; $query = $this->executeQuery($sql, array($user_id)); $entries = $query->fetchAll(); @@ -250,7 +272,7 @@ class Database { public function retrieveOneByURL($url, $user_id) { $entry = NULL; - $sql = "SELECT * FROM entries WHERE url=? AND user_id=?"; + $sql = "SELECT * FROM entries WHERE content <> '' AND url=? AND user_id=?"; $params = array($url, $user_id); $query = $this->executeQuery($sql, $params); $entry = $query->fetchAll(); @@ -267,21 +289,22 @@ class Database { public function getEntriesByView($view, $user_id, $limit = '', $tag_id = 0) { switch ($view) { case 'archive': - $sql = "SELECT * FROM entries WHERE user_id=? AND is_read=? "; + $sql = "SELECT * FROM entries WHERE content <> '' AND user_id=? AND is_read=? "; $params = array($user_id, 1); break; case 'fav' : - $sql = "SELECT * FROM entries WHERE user_id=? AND is_fav=? "; + $sql = "SELECT * FROM entries WHERE content <> '' AND user_id=? AND is_fav=? "; $params = array($user_id, 1); break; case 'tag' : $sql = "SELECT entries.* FROM entries LEFT JOIN tags_entries ON tags_entries.entry_id=entries.id - WHERE entries.user_id=? AND tags_entries.tag_id = ? "; + WHERE entries.content <> '' AND + entries.user_id=? AND tags_entries.tag_id = ? "; $params = array($user_id, $tag_id); break; default: - $sql = "SELECT * FROM entries WHERE user_id=? AND is_read=? "; + $sql = "SELECT * FROM entries WHERE content <> '' AND user_id=? AND is_read=? "; $params = array($user_id, 0); break; } @@ -294,24 +317,25 @@ class Database { return $entries; } - public function getEntriesByViewCount($view, $user_id, $tag_id = 0) { - switch ($view) { + public function getEntriesByViewCount($view, $user_id, $tag_id = 0) { + switch ($view) { case 'archive': - $sql = "SELECT count(*) FROM entries WHERE user_id=? AND is_read=? "; + $sql = "SELECT count(*) FROM entries WHERE content <> '' AND user_id=? AND is_read=? "; $params = array($user_id, 1); break; case 'fav' : - $sql = "SELECT count(*) FROM entries WHERE user_id=? AND is_fav=? "; + $sql = "SELECT count(*) FROM entries WHERE content <> '' AND user_id=? AND is_fav=? "; $params = array($user_id, 1); break; - case 'tag' : - $sql = "SELECT count(*) FROM entries - LEFT JOIN tags_entries ON tags_entries.entry_id=entries.id - WHERE entries.user_id=? AND tags_entries.tag_id = ? "; - $params = array($user_id, $tag_id); - break; + case 'tag' : + $sql = "SELECT count(*) FROM entries + LEFT JOIN tags_entries ON tags_entries.entry_id=entries.id + WHERE entries.content <> '' AND + entries.user_id=? AND tags_entries.tag_id = ? "; + $params = array($user_id, $tag_id); + break; default: - $sql = "SELECT count(*) FROM entries WHERE user_id=? AND is_read=? "; + $sql = "SELECT count(*) FROM entries WHERE content <> '' AND user_id=? AND is_read=? "; $params = array($user_id, 0); break; } @@ -319,7 +343,7 @@ class Database { $query = $this->executeQuery($sql, $params); list($count) = $query->fetch(); - return $count; + return $count; } public function updateContent($id, $content, $user_id) { @@ -369,7 +393,7 @@ class Database { $sql = "SELECT DISTINCT tags.* FROM tags LEFT JOIN tags_entries ON tags_entries.tag_id=tags.id LEFT JOIN entries ON tags_entries.entry_id=entries.id - WHERE entries.user_id=?"; + WHERE entries.content <> '' AND entries.user_id=?"; $query = $this->executeQuery($sql, array($user_id)); $tags = $query->fetchAll(); @@ -381,7 +405,7 @@ class Database { $sql = "SELECT DISTINCT tags.* FROM tags LEFT JOIN tags_entries ON tags_entries.tag_id=tags.id LEFT JOIN entries ON tags_entries.entry_id=entries.id - WHERE tags.id=? AND entries.user_id=?"; + WHERE entries.content <> '' AND tags.id=? AND entries.user_id=?"; $params = array(intval($id), $user_id); $query = $this->executeQuery($sql, $params); $tag = $query->fetchAll(); @@ -393,7 +417,8 @@ class Database { $sql = "SELECT entries.* FROM entries LEFT JOIN tags_entries ON tags_entries.entry_id=entries.id - WHERE tags_entries.tag_id = ? AND entries.user_id=?"; + WHERE entries.content <> '' AND + tags_entries.tag_id = ? AND entries.user_id=?"; $query = $this->executeQuery($sql, array($tag_id, $user_id)); $entries = $query->fetchAll(); diff --git a/inc/poche/Poche.class.php b/inc/poche/Poche.class.php index 026b0b4e..5a89a8d2 100755 --- a/inc/poche/Poche.class.php +++ b/inc/poche/Poche.class.php @@ -362,60 +362,6 @@ class Poche ); } - protected function getPageContent(Url $url) - { - // Saving and clearing context - $REAL = array(); - foreach( $GLOBALS as $key => $value ) { - if( $key != 'GLOBALS' && $key != '_SESSION' && $key != 'HTTP_SESSION_VARS' ) { - $GLOBALS[$key] = array(); - $REAL[$key] = $value; - } - } - // Saving and clearing session - $REAL_SESSION = array(); - foreach( $_SESSION as $key => $value ) { - $REAL_SESSION[$key] = $value; - unset($_SESSION[$key]); - } - - // Running code in different context - $scope = function() { - extract( func_get_arg(1) ); - $_GET = $_REQUEST = array( - "url" => $url->getUrl(), - "max" => 5, - "links" => "preserve", - "exc" => "", - "format" => "json", - "submit" => "Create Feed" - ); - ob_start(); - require func_get_arg(0); - $json = ob_get_flush(); - return $json; - }; - $json = $scope( "inc/3rdparty/makefulltextfeed.php", array("url" => $url) ); - - // Clearing and restoring context - foreach( $GLOBALS as $key => $value ) { - if( $key != "GLOBALS" && $key != "_SESSION" ) { - unset($GLOBALS[$key]); - } - } - foreach( $REAL as $key => $value ) { - $GLOBALS[$key] = $value; - } - // Clearing and restoring session - foreach( $_SESSION as $key => $value ) { - unset($_SESSION[$key]); - } - foreach( $REAL_SESSION as $key => $value ) { - $_SESSION[$key] = $value; - } - return json_decode($json, true); - } - /** * Call action (mark as fav, archive, delete, etc.) */ @@ -424,15 +370,21 @@ class Poche switch ($action) { case 'add': - $content = $this->getPageContent($url); - $title = ($content['rss']['channel']['item']['title'] != '') ? $content['rss']['channel']['item']['title'] : _('Untitled'); - $body = $content['rss']['channel']['item']['description']; - - // clean content from prevent xss attack - $config = HTMLPurifier_Config::createDefault(); - $purifier = new HTMLPurifier($config); - $title = $purifier->purify($title); - $body = $purifier->purify($body); + if (!$import) { + $content = Tools::getPageContent($url); + $title = ($content['rss']['channel']['item']['title'] != '') ? $content['rss']['channel']['item']['title'] : _('Untitled'); + $body = $content['rss']['channel']['item']['description']; + + // clean content from prevent xss attack + $config = HTMLPurifier_Config::createDefault(); + $purifier = new HTMLPurifier($config); + $title = $purifier->purify($title); + $body = $purifier->purify($body); + } + else { + $title = ''; + $body = ''; + } //search for possible duplicate if not in import mode if (!$import) { @@ -897,7 +849,7 @@ class Poche # the second
    is for read links $read = 1; } - $this->messages->add('s', _('import from instapaper completed')); + $this->messages->add('s', _('import from instapaper completed. You have to execute the cron to fetch content.')); Tools::logm('import from instapaper completed'); Tools::redirect(); } @@ -941,7 +893,7 @@ class Poche # the second + {% if token == '' %} +

    {% trans "3. Your feed token is currently empty and must first be generated to fetch content. Click here to generate it." %}

    + {% else %} +

    3. {% trans "You can fetch content for imported items." %} Click here to fetch content for 10 articles.

    +

    {% trans "You can also create a cron task:" %}

    + 
    0 */4 * * *  cd /path/to/wallabag && php cron.php --limit=10 --user-id={{user_id}} --token={{token}} >/dev/null 2>&1
    + {% endif %} +

    {% trans "Export your wallabag data" %}

    {% if constant('STORAGE') == 'sqlite' %}

    {% trans "Click here" %} {% trans "to download your database." %}

    {% endif %}