From 4e067ceabd705201a16b4c92cf4b23f3b990326c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= Date: Sun, 13 Jul 2014 10:15:40 +0200 Subject: updated specific configuration for parsing --- inc/3rdparty/site_config/standard/lemonde.fr.txt | 31 ++++++++++++++---------- 1 file changed, 18 insertions(+), 13 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/lemonde.fr.txt (limited to '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 new mode 100755 index eb205275..097999b6 --- a/inc/3rdparty/site_config/standard/lemonde.fr.txt +++ b/inc/3rdparty/site_config/standard/lemonde.fr.txt @@ -1,13 +1,18 @@ -title: //h1 - -# they have a single component containing both author and date -#author: //p[@class='source'] -#date: //p[@class='source'] - -body: //div[@class='contenu_article'] -#Shoot the insane "conjugaison.lemonde.fr" links : -strip: //a[contains(@class, 'listLink')] - -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 \ 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 -- cgit v1.2.3