updated specific configuration for parsing

[github/wallabag/wallabag.git] / inc / 3rdparty / site_config / standard / lemonde.fr.txt

diff --git a/inc/3rdparty/site_config/standard/lemonde.fr.txt b/inc/3rdparty/site_config/standard/lemonde.fr.txt
old mode 100644 (file)
new mode 100755 (executable)
index eb20527..097999b
--- a/inc/3rdparty/site_config/standard/lemonde.fr.txt
+++ b/inc/3rdparty/site_config/standard/lemonde.fr.txt
@@ -1,13 +1,18 @@
-title: //h1\r
-\r
-# they have a single component containing both author and date\r
-#author: //p[@class='source']\r
-#date: //p[@class='source']\r
-\r
-body: //div[@class='contenu_article']\r
-#Shoot the insane "conjugaison.lemonde.fr" links :\r
-strip: //a[contains(@class, 'listLink')]\r
-\r
-prune: no\r
-\r
-test_url: http://www.lemonde.fr/economie/article/2011/07/05/moody-s-abaisse-la-note-du-portugal-de-quatre-crans_1545237_3234.html
\ No newline at end of file
+title: //h1
+
+# We can have multiple authors
+author: //a[@class='auteur']
+
+# Last edition date (if any)
+date: //time[@itemprop='dateModified']/@datetime
+# Publication date
+date: //time[@itemprop='datePublished']/@datetime
+
+
+body: //div[@id='articleBody']
+#Shoot the insane "conjugaison.lemonde.fr" links :
+#strip: //a[contains(@class, 'conjug')]
+
+prune: no
+
+test_url: http://www.lemonde.fr/economie/article/2011/07/05/moody-s-abaisse-la-note-du-portugal-de-quatre-crans_1545237_3234.html