From ac4d114214d820b20e18518a2dbc809337e39043 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= Date: Fri, 6 Dec 2013 10:13:03 +0100 Subject: [add] new specific configuration files --- .../site_config/standard/sportschau.de.txt | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) create mode 100644 inc/3rdparty/site_config/standard/sportschau.de.txt (limited to 'inc/3rdparty/site_config/standard/sportschau.de.txt') diff --git a/inc/3rdparty/site_config/standard/sportschau.de.txt b/inc/3rdparty/site_config/standard/sportschau.de.txt new file mode 100644 index 00000000..6500e75c --- /dev/null +++ b/inc/3rdparty/site_config/standard/sportschau.de.txt @@ -0,0 +1,22 @@ +title://div[@id='ardContent']/h1 + +author://p[@id='ardAutor'] +author://span[@id='ardQuelle'] +author:string('sportschau.de') + +date:substring-after(//span[@id='ardStand'], 'Stand: ') + +body://div[@id='ardContent'] + +strip://div[@id='ardContent']/h1 +strip://p[@id='ardAutor'] +strip: //div[@class='embeddedPlayer_clipinfo'] +strip: //div[@class='ardMehrZumThemaRechts'] +strip: //*[contains(@class, 'inv')] + +strip: //p[@id='ardAbbinder'] +strip: //div[@class='socialBookmarks'] +strip: //div[@id='ardContentEnd'] +strip: //div[@id='ardDisclaimer'] +strip: //div[@id='ardRechteSpalte'] +test_url: http://www.sportschau.de/sp/fussball/news201203/17/analyse_leverkusen_gladbach.jsp \ No newline at end of file -- cgit v1.2.3