20 files changed, 97 insertions, 18 deletions

diff --git a/inc/3rdparty/libraries/PHPePub/Zip.php b/inc/3rdparty/libraries/PHPePub/Zip.php
index 01e03566..92e78e70 100644
--- a/inc/3rdparty/libraries/PHPePub/Zip.php
+++ b/inc/3rdparty/libraries/PHPePub/Zip.php
@@ -569,7 +569,7 @@ class Zip {
                 if ($inline) {
                     $cd .= "inline";
                                 } else{
-                    $cd .= "attached";
+                    $cd .= "attachment";
                                 }
                 if ($fileName) {
                     $cd .= '; filename="' . $fileName . '"';
diff --git a/inc/3rdparty/site_config/standard/aclu.org.txt b/inc/3rdparty/site_config/standard/aclu.org.txt
new file mode 100644
index 00000000..74236e2d
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/aclu.org.txt
@@ -0,0 +1,5 @@
+body: //div[@class='panel-panel panel-main-3 content-column']
+title: //div[@class='panel-pane pane-node-title']
+date: //div[@class='updated-date']
+
+test_url: https://www.aclu.org/blog/free-future/chinas-nightmarish-citizen-scores-are-warning-americans
diff --git a/inc/3rdparty/site_config/standard/bez.es.txt b/inc/3rdparty/site_config/standard/bez.es.txt
new file mode 100644
index 00000000..cab7c6f7
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/bez.es.txt
@@ -0,0 +1,5 @@
+body: //div[@class='text_art']
+strip: //div[@class='cab_datos_opinion']
+strip: //div[@class='sumario2_left']
+
+test_url: http://www.bez.es/382758623/otros-fracasos-empresas.html
diff --git a/inc/3rdparty/site_config/standard/blog.cloudflare.com.txt b/inc/3rdparty/site_config/standard/blog.cloudflare.com.txt
index db80a35f..2f9a5a2f 100755
--- a/inc/3rdparty/site_config/standard/blog.cloudflare.com.txt
+++ b/inc/3rdparty/site_config/standard/blog.cloudflare.com.txt
@@ -1,9 +1,10 @@
 # Instapaper gets this back to front and only gets the blog title instead of the article title.
-title: substring-before(//title, '-')
+title: //div[@class='title']
 
 author: //a[ contains(@href, '/people') ]
 
-body: //div[ @class='post' ]
+body: //div[ @class='post-content' ]
 
 # Date is impossible to retrieve since they use those stupid "fuzzy" dates, inserted through javascript, at posterous.
-test_url: http://blog.cloudflare.com/understanding-analytics-when-is-a-page-view-n
\ No newline at end of file
+test_url: http://blog.cloudflare.com/understanding-analytics-when-is-a-page-view-n
+test_url: https://blog.cloudflare.com/sha-1-deprecation-no-browser-left-behind
diff --git a/inc/3rdparty/site_config/standard/caffereggio.net.txt b/inc/3rdparty/site_config/standard/caffereggio.net.txt
new file mode 100644
index 00000000..08e1f05a
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/caffereggio.net.txt
@@ -0,0 +1,3 @@
+body: //div[@class="pf-content"]
+
+test_url: http://www.caffereggio.net/2016/01/16/115327/
diff --git a/inc/3rdparty/site_config/standard/chronicle.com.txt b/inc/3rdparty/site_config/standard/chronicle.com.txt
index e86d3eca..227a03c7 100755
--- a/inc/3rdparty/site_config/standard/chronicle.com.txt
+++ b/inc/3rdparty/site_config/standard/chronicle.com.txt
@@ -1,17 +1,10 @@
 title: //h1[contains(@class, "entry-title")]
 author: //p[contains(@class, "byline")]
 
-# blog articles (chronicle.com/blogs/*)
-body: //div[contains(@class, "abstract")]
-date: //p[contains(@class, "time")]
-
 # all (?) other articles
-body: //div[@id="article-body"]
+body: //div[@class="content-item__story"]
 date: //p[contains(@class, "dateline")]
 
-# remove sidebars containing images (I assume this is desired for Instapaper)
-strip: //div[@id="related"]
-strip: //div[contains(@class, "image")]
-
-# note that if you're not a Chronicle subscriber (personally or institutionally), you'll only see the first couple of paragraphs of the article, and Instapaper will display that with some crap above and below. thank goodness for that bookmarklet
-test_url: http://chronicle.com/article/In-a-Land-of-Second-Chances/128375/
\ No newline at end of file
+# note that if you're not a Chronicle subscriber (personally or institutionally), you'll only see the first couple of paragraphs of the article, and Instapaper $
+test_url: http://chronicle.com/article/In-a-Land-of-Second-Chances/128375/
+test_url: http://chronicle.com/blogs/wiredcampus/university-run-boot-camps-offer-students-marketable-skills-but-not-course-credit/57494
diff --git a/inc/3rdparty/site_config/standard/deia.com.txt b/inc/3rdparty/site_config/standard/deia.com.txt
new file mode 100644
index 00000000..d2f78239
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/deia.com.txt
@@ -0,0 +1,4 @@
+body: //div[@class='widget full_article']
+strip: //div[@class='Herramientas']
+
+test_url: http://www.deia.com/2015/10/03/politica/euskadi/el-envilecimiento-de-la-politica-y-de-los-medios-de-comunicacion
diff --git a/inc/3rdparty/site_config/standard/diagonalperiodico.net.txt b/inc/3rdparty/site_config/standard/diagonalperiodico.net.txt
new file mode 100644
index 00000000..f0681bac
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/diagonalperiodico.net.txt
@@ -0,0 +1,4 @@
+body: //div[@class='field__items']
+title: //div[@class='art_titulo']
+
+test_url: https://www.diagonalperiodico.net/global/27947-cuanto-mas-tiempo-nos-aferremos-este-sistema-peor-y-menores-seran-nuestras-opciones
diff --git a/inc/3rdparty/site_config/standard/esglobal.org.txt b/inc/3rdparty/site_config/standard/esglobal.org.txt
new file mode 100644
index 00000000..6408f441
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/esglobal.org.txt
@@ -0,0 +1,3 @@
+body: //div[@class='blog-content']
+
+test_url: http://www.esglobal.org/el-caos-en-el-este-los-socios-de-la-ue-necesitan-que-se-les-preste-atencion/
diff --git a/inc/3rdparty/site_config/standard/lawfareblog.com.txt b/inc/3rdparty/site_config/standard/lawfareblog.com.txt
new file mode 100644
index 00000000..49d858a7
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/lawfareblog.com.txt
@@ -0,0 +1,4 @@
+body: //div[@class='field-items']
+title: //div[@class='title']
+
+test_url: https://lawfareblog.com/limits-panopticon
diff --git a/inc/3rdparty/site_config/standard/letraslibres.com.txt b/inc/3rdparty/site_config/standard/letraslibres.com.txt
new file mode 100644
index 00000000..cf271bca
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/letraslibres.com.txt
@@ -0,0 +1,3 @@
+single_page_link: concat(link[@rel="canonical"], "?page=full")
+
+test_url: http://www.letraslibres.com/revista/dossier/quien-manda-en-europa
diff --git a/inc/3rdparty/site_config/standard/naiz.eus.txt b/inc/3rdparty/site_config/standard/naiz.eus.txt
new file mode 100644
index 00000000..9e48333a
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/naiz.eus.txt
@@ -0,0 +1,4 @@
+body: //div[contains(@class,'widget full_article')]
+
+test_url: http://www.naiz.eus/eu/actualidad/noticia/20151002/adegi-afirma-que-los-jovenes-viven-una-vida-muy-comoda-y-no-tienen-hambre-para-emprender#
+test_url: http://www.naiz.eus/eu/actualidad/noticia/20151012/podemos-euskadi-critica-que-otegi-y-sus-companeros-quieran-pasar-por-grandes-pacifistas
diff --git a/inc/3rdparty/site_config/standard/pieria.co.uk.txt b/inc/3rdparty/site_config/standard/pieria.co.uk.txt
new file mode 100644
index 00000000..31f2c84a
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/pieria.co.uk.txt
@@ -0,0 +1,3 @@
+body: //div[@id='article']
+
+test_url: http://www.pieria.co.uk/articles/need_a_fiscal_rule_george__try_get_the_economy_growing
diff --git a/inc/3rdparty/site_config/standard/popehat.com.txt b/inc/3rdparty/site_config/standard/popehat.com.txt
new file mode 100644
index 00000000..45b76ae1
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/popehat.com.txt
@@ -0,0 +1,6 @@
+title: //div[@class='entry-title']
+author: //div[@class='entry-author']
+date: //div[@class='entry-time']
+body: //div[@class='entry-content']
+
+test_url: https://popehat.com/2015/12/16/eric-posner-the-first-amendments-nemesis/
diff --git a/inc/3rdparty/site_config/standard/rasgolatente.es.txt b/inc/3rdparty/site_config/standard/rasgolatente.es.txt
new file mode 100644
index 00000000..fcdbbd97
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/rasgolatente.es.txt
@@ -0,0 +1,6 @@
+title: //div[@class='entry-title']
+author: //div[@class='entry-author']
+date: //div[@class='entry-time']
+body: //div[@class='entry-content']
+
+test_url: http://rasgolatente.es/estupidez-psicologia-estupidos/
diff --git a/inc/3rdparty/site_config/standard/thenews.coop.txt b/inc/3rdparty/site_config/standard/thenews.coop.txt
new file mode 100644
index 00000000..7d593805
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/thenews.coop.txt
@@ -0,0 +1,3 @@
+body: //div[@class='entry-content']
+
+test_url: http://www.thenews.coop/98221/news/co-operatives/jeremy-corbyn-needs-co-op-movement/
diff --git a/inc/3rdparty/site_config/standard/timeshighereducation.com.txt b/inc/3rdparty/site_config/standard/timeshighereducation.com.txt
new file mode 100644
index 00000000..ba52788d
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/timeshighereducation.com.txt
@@ -0,0 +1,3 @@
+body: //div[@class="col-md-12 radix-layouts-contentheader panel-panel"]
+strip: //div[@class="htmlContent subscribe_box"]
+test_url: https://www.timeshighereducation.com/blog/jeremy-corbyn-serious-about-free-higher-education
diff --git a/inc/3rdparty/site_config/standard/wired.de.txt b/inc/3rdparty/site_config/standard/wired.de.txt
new file mode 100644
index 00000000..da88362f
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/wired.de.txt
@@ -0,0 +1,8 @@
+title: //meta[@name='Title']/@content
+author: //meta[@name='Author']/@content
+date: //meta[@name='DisplayDate']/@content
+body: //div[@class='article-content']
+strip: //div[@class='article-sidebar']
+
+test_url: https://www.wired.de/collection/latest/der-vw-skandal-zeigt-ohne-vertrauen-ist-software-nicht-mehr-als-bose-magie
+
diff --git a/inc/poche/Poche.class.php b/inc/poche/Poche.class.php
index 9014f455..1476d158 100755
--- a/inc/poche/Poche.class.php
+++ b/inc/poche/Poche.class.php
@@ -190,7 +190,7 @@ class Poche
         {
             case 'add':
                 $content = Tools::getPageContent($url);
-                $title = ($content['rss']['channel']['item']['title'] != '') ? $content['rss']['channel']['item']['title'] : _('Untitled');
+                $title = ($content['rss']['channel']['item']['title'] != '') ? $content['rss']['channel']['item']['title'] : parse_url($url, PHP_URL_HOST);
                 $body = $content['rss']['channel']['item']['description'];
 
                 // clean content from prevent xss attack
@@ -908,7 +908,7 @@ class Poche
         $entries = $this->store->retrieveAllWithTags($this->user->getId());
         if ($entries) {
             echo $this->tpl->render('export.twig', array(
-            'export' => Tools::renderJson($entries),
+            'export' => Tools::renderJson(Tools::utf8ize($entries)),
             ));
             Tools::logm('export view');
         } else {
diff --git a/inc/poche/Tools.class.php b/inc/poche/Tools.class.php
index 9c748391..263034f0 100755
--- a/inc/poche/Tools.class.php
+++ b/inc/poche/Tools.class.php
@@ -141,7 +141,7 @@ final class Tools
     public static function getFile($url)
     {
         $timeout = 15;
-        $useragent = "Mozilla/5.0 (Windows NT 5.1; rv:18.0) Gecko/20100101 Firefox/18.0";
+        $useragent = "Mozilla/5.0 (Windows NT 6.3; rv:36.0) Gecko/20100101 Firefox/36.0";
 
         if (in_array ('curl', get_loaded_extensions())) {
             # Fetch feed from URL
@@ -233,6 +233,27 @@ final class Tools
     }
 
     /**
+     * UTF-8 encode array of string
+     *
+     * @param $data
+     */
+    public static function utf8ize($data)
+    {
+        if (is_array($data))
+        {
+            foreach ($data as $k => $v)
+            {
+                $data[$k] = self::utf8ize($v);
+            }
+        }
+        else if (is_string ($data) && '' == mb_detect_encoding($data))
+        {
+            return utf8_encode($data);
+        }
+        return $data;
+    }
+
+    /**
      * Create new line in log file
      *
      * @param $message