From 8f74210637bd04debff38dcb12da62c150ae1c2a Mon Sep 17 00:00:00 2001 From: "FiveFilters.org" Date: Fri, 17 Apr 2015 16:30:31 +0200 Subject: tagesschau.de --- tagesschau.de.txt | 29 +++++++++-------------------- 1 file changed, 9 insertions(+), 20 deletions(-) diff --git a/tagesschau.de.txt b/tagesschau.de.txt index be76cd0..a01624c 100644 --- a/tagesschau.de.txt +++ b/tagesschau.de.txt @@ -1,23 +1,12 @@ -title://h1[1] +body: //div[contains(@class, 'sectionArticle')]//div[contains(concat(' ',normalize-space(@class),' '),' box ')] -author: substring-after(//em, 'Von ') -author:string('tagesschau.de') +strip_id_or_class: infokasten +strip_id_or_class: teaserImTeaser +strip_id_or_class: Comments +strip_id_or_class: mediaInfo +strip: //div[contains(@class, 'mediaCon')]//iframe -date:substring-after(//div[@class='standDatum'], 'Stand: ') +prune: no -body://div[contains(@class, 'article')] | //div[contains(@class, 'centerCol')] - -strip://h1[1] -strip: //div[contains(@class, 'directLinks')] -strip: //div[contains(@class, 'zitatBox')] -strip: //div[contains(@class, 'teaserBox metaBlock')] -strip: //*[contains(@class, 'inv')] -strip: //span[@class='imgSubline'] -strip: //*[contains(@class, 'topline')][1] -strip: //div[@id='rightCol'][1] -strip: //div[@id="footer"][1] -strip: //div[@class="fPlayer"] -strip: //div[@id='seitenanfang'] -strip: //div[@class='standDatum'] -strip: //em -test_url: http://www.tagesschau.de/ausland/wahlkampffrankreich102.html \ No newline at end of file +test_url: http://www.tagesschau.de/inland/germanwings-trauerfeier-111.html +test_url: http://www.tagesschau.de/xml/rss2 \ No newline at end of file -- cgit v1.3.1