2014-07-13 04:15:40 -04:00
|
|
|
# Jens Kohl, jens.kohl@...
|
|
|
|
|
# - Added publication date
|
|
|
|
|
# - Striped pagination block
|
|
|
|
|
# - Added single page link
|
|
|
|
|
# - Added xpath-querys for the printer friendly version
|
|
|
|
|
|
|
|
|
|
title: //h1
|
|
|
|
|
body: //div[@class='formatted']
|
|
|
|
|
prune: no
|
|
|
|
|
|
|
|
|
|
date: substring-after(//li[2][@class="text1"], 'Datum:')
|
|
|
|
|
strip: //ol[@class="list-chapters"]
|
|
|
|
|
strip_comments: yes
|
|
|
|
|
|
|
|
|
|
# next: commands for printer friendly pages
|
|
|
|
|
single_page_link: //a[contains(@href, 'print.php?a=')]/@href
|
|
|
|
|
title: //body/h3
|
|
|
|
|
strip_image_src: staticrl/images/logo.jpg
|
|
|
|
|
strip_image_src: http://cpx.golem.de/cpx.php?class=7
|
|
|
|
|
strip: //body/h3
|
|
|
|
|
strip: //body/b[1]
|
|
|
|
|
strip: //body/b[2]
|
|
|
|
|
strip: //body/b[3]
|
|
|
|
|
strip: //div[1]
|
2013-12-06 04:13:03 -05:00
|
|
|
test_url: http://www.golem.de/1112/88696.html
|
