From 90a1a78b1e2f4d40e1d9b8e6f46aca129a9d7bcf Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= Date: Mon, 27 Oct 2014 06:46:13 +0100 Subject: updated site_config --- inc/3rdparty/site_config/standard/golem.de.txt | 53 +++++++++++++++----------- 1 file changed, 31 insertions(+), 22 deletions(-) (limited to 'inc/3rdparty/site_config/standard/golem.de.txt') diff --git a/inc/3rdparty/site_config/standard/golem.de.txt b/inc/3rdparty/site_config/standard/golem.de.txt index 6afdebe8..c64860c0 100755 --- a/inc/3rdparty/site_config/standard/golem.de.txt +++ b/inc/3rdparty/site_config/standard/golem.de.txt @@ -1,25 +1,34 @@ -# Jens Kohl, jens.kohl@... -# - Added publication date -# - Striped pagination block -# - Added single page link -# - Added xpath-querys for the printer friendly version +# Author: zinnober +# Rewrite of original template which fetched the printer-version without pictures -title: //h1 -body: //div[@class='formatted'] +tidy: no 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] -test_url: http://www.golem.de/1112/88696.html \ No newline at end of file +# Set full title +title: //h1 + +date: //time + +# Content is here +body: //article + +# Fetch full multipage articles +next_page_link: //a[@id='atoc_next'] + +# Remove tracking and ads +strip_id_or_class: iqadtile4 + +# General Cleanup +strip_id_or_class: list-jtoc +strip_id_or_class: table-jtoc +strip_id_or_class: implied +strip_id_or_class: social- +strip_id_or_class: comments +strip_id_or_class: footer + +# Tidy up galleries (could still be improved, though) +strip: //img[@src=''] + +# Try yourself +test_url: http://www.golem.de/news/intel-core-i7-5960x-im-test-die-pc-revolution-beginnt-mit-octacore-und-ddr4-1408-108893.html +test_url: http://www.golem.de/news/test-infamous-first-light-neonbunter-actionspass-1408-108914.html -- cgit v1.2.3