aboutsummaryrefslogtreecommitdiffhomepage
path: root/inc
diff options
context:
space:
mode:
Diffstat (limited to 'inc')
-rw-r--r--inc/3rdparty/libraries/PHPePub/Zip.php2
-rw-r--r--inc/3rdparty/site_config/standard/aclu.org.txt5
-rw-r--r--inc/3rdparty/site_config/standard/bez.es.txt5
-rwxr-xr-xinc/3rdparty/site_config/standard/blog.cloudflare.com.txt7
-rw-r--r--inc/3rdparty/site_config/standard/caffereggio.net.txt3
-rwxr-xr-xinc/3rdparty/site_config/standard/chronicle.com.txt15
-rw-r--r--inc/3rdparty/site_config/standard/deia.com.txt4
-rw-r--r--inc/3rdparty/site_config/standard/diagonalperiodico.net.txt4
-rw-r--r--inc/3rdparty/site_config/standard/esglobal.org.txt3
-rw-r--r--inc/3rdparty/site_config/standard/lawfareblog.com.txt4
-rw-r--r--inc/3rdparty/site_config/standard/letraslibres.com.txt3
-rw-r--r--inc/3rdparty/site_config/standard/naiz.eus.txt4
-rw-r--r--inc/3rdparty/site_config/standard/pieria.co.uk.txt3
-rw-r--r--inc/3rdparty/site_config/standard/popehat.com.txt6
-rw-r--r--inc/3rdparty/site_config/standard/rasgolatente.es.txt6
-rw-r--r--inc/3rdparty/site_config/standard/thenews.coop.txt3
-rw-r--r--inc/3rdparty/site_config/standard/timeshighereducation.com.txt3
-rw-r--r--inc/3rdparty/site_config/standard/wired.de.txt8
-rwxr-xr-xinc/poche/Poche.class.php4
-rwxr-xr-xinc/poche/Tools.class.php23
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 {
569 if ($inline) { 569 if ($inline) {
570 $cd .= "inline"; 570 $cd .= "inline";
571 } else{ 571 } else{
572 $cd .= "attached"; 572 $cd .= "attachment";
573 } 573 }
574 if ($fileName) { 574 if ($fileName) {
575 $cd .= '; filename="' . $fileName . '"'; 575 $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 @@
1body: //div[@class='panel-panel panel-main-3 content-column']
2title: //div[@class='panel-pane pane-node-title']
3date: //div[@class='updated-date']
4
5test_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 @@
1body: //div[@class='text_art']
2strip: //div[@class='cab_datos_opinion']
3strip: //div[@class='sumario2_left']
4
5test_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 @@
1# Instapaper gets this back to front and only gets the blog title instead of the article title. 1# Instapaper gets this back to front and only gets the blog title instead of the article title.
2title: substring-before(//title, '-') 2title: //div[@class='title']
3 3
4author: //a[ contains(@href, '/people') ] 4author: //a[ contains(@href, '/people') ]
5 5
6body: //div[ @class='post' ] 6body: //div[ @class='post-content' ]
7 7
8# Date is impossible to retrieve since they use those stupid "fuzzy" dates, inserted through javascript, at posterous. 8# Date is impossible to retrieve since they use those stupid "fuzzy" dates, inserted through javascript, at posterous.
9test_url: http://blog.cloudflare.com/understanding-analytics-when-is-a-page-view-n \ No newline at end of file 9test_url: http://blog.cloudflare.com/understanding-analytics-when-is-a-page-view-n
10test_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 @@
1body: //div[@class="pf-content"]
2
3test_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 @@
1title: //h1[contains(@class, "entry-title")] 1title: //h1[contains(@class, "entry-title")]
2author: //p[contains(@class, "byline")] 2author: //p[contains(@class, "byline")]
3 3
4# blog articles (chronicle.com/blogs/*)
5body: //div[contains(@class, "abstract")]
6date: //p[contains(@class, "time")]
7
8# all (?) other articles 4# all (?) other articles
9body: //div[@id="article-body"] 5body: //div[@class="content-item__story"]
10date: //p[contains(@class, "dateline")] 6date: //p[contains(@class, "dateline")]
11 7
12# remove sidebars containing images (I assume this is desired for Instapaper) 8# 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 $
13strip: //div[@id="related"] 9test_url: http://chronicle.com/article/In-a-Land-of-Second-Chances/128375/
14strip: //div[contains(@class, "image")] 10test_url: http://chronicle.com/blogs/wiredcampus/university-run-boot-camps-offer-students-marketable-skills-but-not-course-credit/57494
15
16# 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
17test_url: http://chronicle.com/article/In-a-Land-of-Second-Chances/128375/ \ No newline at end of file
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 @@
1body: //div[@class='widget full_article']
2strip: //div[@class='Herramientas']
3
4test_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 @@
1body: //div[@class='field__items']
2title: //div[@class='art_titulo']
3
4test_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 @@
1body: //div[@class='blog-content']
2
3test_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 @@
1body: //div[@class='field-items']
2title: //div[@class='title']
3
4test_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 @@
1single_page_link: concat(link[@rel="canonical"], "?page=full")
2
3test_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 @@
1body: //div[contains(@class,'widget full_article')]
2
3test_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#
4test_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 @@
1body: //div[@id='article']
2
3test_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 @@
1title: //div[@class='entry-title']
2author: //div[@class='entry-author']
3date: //div[@class='entry-time']
4body: //div[@class='entry-content']
5
6test_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 @@
1title: //div[@class='entry-title']
2author: //div[@class='entry-author']
3date: //div[@class='entry-time']
4body: //div[@class='entry-content']
5
6test_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 @@
1body: //div[@class='entry-content']
2
3test_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 @@
1body: //div[@class="col-md-12 radix-layouts-contentheader panel-panel"]
2strip: //div[@class="htmlContent subscribe_box"]
3test_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 @@
1title: //meta[@name='Title']/@content
2author: //meta[@name='Author']/@content
3date: //meta[@name='DisplayDate']/@content
4body: //div[@class='article-content']
5strip: //div[@class='article-sidebar']
6
7test_url: https://www.wired.de/collection/latest/der-vw-skandal-zeigt-ohne-vertrauen-ist-software-nicht-mehr-als-bose-magie
8
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
190 { 190 {
191 case 'add': 191 case 'add':
192 $content = Tools::getPageContent($url); 192 $content = Tools::getPageContent($url);
193 $title = ($content['rss']['channel']['item']['title'] != '') ? $content['rss']['channel']['item']['title'] : _('Untitled'); 193 $title = ($content['rss']['channel']['item']['title'] != '') ? $content['rss']['channel']['item']['title'] : parse_url($url, PHP_URL_HOST);
194 $body = $content['rss']['channel']['item']['description']; 194 $body = $content['rss']['channel']['item']['description'];
195 195
196 // clean content from prevent xss attack 196 // clean content from prevent xss attack
@@ -908,7 +908,7 @@ class Poche
908 $entries = $this->store->retrieveAllWithTags($this->user->getId()); 908 $entries = $this->store->retrieveAllWithTags($this->user->getId());
909 if ($entries) { 909 if ($entries) {
910 echo $this->tpl->render('export.twig', array( 910 echo $this->tpl->render('export.twig', array(
911 'export' => Tools::renderJson($entries), 911 'export' => Tools::renderJson(Tools::utf8ize($entries)),
912 )); 912 ));
913 Tools::logm('export view'); 913 Tools::logm('export view');
914 } else { 914 } else {
diff --git a/inc/poche/Tools.class.php b/inc/poche/Tools.class.php
index 9c748391..263034f0 100755
--- a/inc/poche/Tools.class.php
+++ b/inc/poche/Tools.class.php
@@ -141,7 +141,7 @@ final class Tools
141 public static function getFile($url) 141 public static function getFile($url)
142 { 142 {
143 $timeout = 15; 143 $timeout = 15;
144 $useragent = "Mozilla/5.0 (Windows NT 5.1; rv:18.0) Gecko/20100101 Firefox/18.0"; 144 $useragent = "Mozilla/5.0 (Windows NT 6.3; rv:36.0) Gecko/20100101 Firefox/36.0";
145 145
146 if (in_array ('curl', get_loaded_extensions())) { 146 if (in_array ('curl', get_loaded_extensions())) {
147 # Fetch feed from URL 147 # Fetch feed from URL
@@ -233,6 +233,27 @@ final class Tools
233 } 233 }
234 234
235 /** 235 /**
236 * UTF-8 encode array of string
237 *
238 * @param $data
239 */
240 public static function utf8ize($data)
241 {
242 if (is_array($data))
243 {
244 foreach ($data as $k => $v)
245 {
246 $data[$k] = self::utf8ize($v);
247 }
248 }
249 else if (is_string ($data) && '' == mb_detect_encoding($data))
250 {
251 return utf8_encode($data);
252 }
253 return $data;
254 }
255
256 /**
236 * Create new line in log file 257 * Create new line in log file
237 * 258 *
238 * @param $message 259 * @param $message