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/nzz.ch.txt | 22 +++++++++++----------- 1 file changed, 11 insertions(+), 11 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/nzz.ch.txt (limited to 'inc/3rdparty/site_config/standard/nzz.ch.txt') diff --git a/inc/3rdparty/site_config/standard/nzz.ch.txt b/inc/3rdparty/site_config/standard/nzz.ch.txt old mode 100644 new mode 100755 index 81faabae..749f4f2a --- a/inc/3rdparty/site_config/standard/nzz.ch.txt +++ b/inc/3rdparty/site_config/standard/nzz.ch.txt @@ -1,12 +1,12 @@ -body: //*[@class='article-full'] -title: //h3 -strip: //header[@class='group'] -#body: //p[@class='lead'] -#move_into(//p[@class='lead']): //*[@class='article-full']/figure -#move_into(//p[@class='lead']): //div[@id='articleBodyText'] -strip: //div[@id='social-media-floater'] -strip: //div[@class='advertisement'] -strip: //div[@class='infobox'] -strip: //div[@id='articleComments'] - +body: //*[@class='article-full'] +title: //h3 +strip: //header[@class='group'] +#body: //p[@class='lead'] +#move_into(//p[@class='lead']): //*[@class='article-full']/figure +#move_into(//p[@class='lead']): //div[@id='articleBodyText'] +strip: //div[@id='social-media-floater'] +strip: //div[@class='advertisement'] +strip: //div[@class='infobox'] +strip: //div[@id='articleComments'] + test_url: http://www.nzz.ch/wissen/wissenschaft/sonnenschutz-fuer-die-erde-1.17282213 \ No newline at end of file -- cgit v1.2.3