Merge pull request #45 from zinnober/contributions

Updates for heise.de and golem.de

2 files changed, 69 insertions, 25 deletions

diff --git a/golem.de.txt b/golem.de.txt
index 6afdebe..c64860c 100644
--- a/golem.de.txt
+++ b/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
+# 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='']
 
-# 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
+# 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
diff --git a/heise.de.txt b/heise.de.txt
index c51af56..9433104 100644..100755
--- a/heise.de.txt
+++ b/heise.de.txt
@@ -1,7 +1,42 @@
-single_page_link: //p[@class='news_option']/a
+# Author: zinnober
+# Template should work well with either desktop or mobile version (m.heise.de)
 
+prune: no
+
+title: //article/h1 | //h1
 date: //p[@class='news_datum']
-title: //h1
-body: //div[@class='meldung_wrapper']
+author: //h4[@class='author']
+
+body: //article | //div[@class='meldung_wrapper']
+
+# General cleanup
+strip: //time
+strip: //h4[@class='author']
+strip: //p[@class='news_datum']
+strip: //p[@class='artikel_datum']
+strip: //a[contains(@href, 'mailto')]
+strip_id_or_class: comments
+strip_id_or_class: ISI_IGNORE
+strip_id_or_class: clear
+
+strip_id_or_class: linkurl_grossbild
+strip_id_or_class: image-num
+strip_id_or_class: heisebox_right
+strip_id_or_class: dossier
+
+# Strip Ads
+strip_id_or_class: ad_
+
+# Some optimizations
+replace_string(<h5>): <h2>
+replace_string(</h5>): </h2>
+replace_string(<span class="bild_rechts"): <p
+replace_string(<div class="heisebox">): <blockquote>
+
+
+next_page_link: //a[@class='next']
+next_page_link: //a[@title='vor']
 
-test_url: http://www.heise.de/newsticker/meldung/Europa-soll-Grundrechteschutz-im-Netz-staerken-1392664.html
\ No newline at end of file
+test_url: http://www.heise.de/open/artikel/Die-Neuerungen-von-Linux-3-15-2196231.html
+test_url: http://m.heise.de/open/artikel/Die-Neuerungen-von-Linux-3-15-2196231.html
+test_url: http://www.heise.de/newsticker/meldung/Ueberwachungstechnik-Die-globale-Handy-Standortueberwachung-2301494.html