Merge pull request #46 from zinnober/contributions

Three new templates

3 files changed, 112 insertions, 0 deletions

diff --git a/apotheke-adhoc.de.txt b/apotheke-adhoc.de.txt
new file mode 100644
index 0000000..3a702e7
--- /dev/null
+++ b/apotheke-adhoc.de.txt
@@ -0,0 +1,23 @@
+# Author: zinnober
+
+prune: no
+
+title: substring-before(//div[@id='content']/h1, ',')
+
+single_page_link: //a[@title='Seite drucken']
+
+body: //div[@id='detail-body']
+
+replace_string(<span class="description">): <em>
+replace_string(<p class="leadtext"><small>): <p class="leadtext">
+
+# Fix headlines
+replace_string(Patrick Hollstein): &nbsp;
+replace_string(APOTHEKE ADHOC): &nbsp;
+replace_string(dpa): &nbsp;
+replace_string(Katharina Lübke): &nbsp;
+replace_string(Julia Pradel): &nbsp;
+replace_string(Franziska Gerhardt): &nbsp;
+
+test_url: http://www.apotheke-adhoc.de/nachrichten/politik/nachricht-detail-politik/deutscher-apothekertag-antraege-gegen-lieferengpaesse-2/
+
diff --git a/deutsche-apotheker-zeitung.de.txt b/deutsche-apotheker-zeitung.de.txt
new file mode 100644
index 0000000..36709ca
--- /dev/null
+++ b/deutsche-apotheker-zeitung.de.txt
@@ -0,0 +1,29 @@
+# Author: zinnober
+
+prune: yes
+tidy: yes
+
+title: //h1
+date: //p[@class='news_datum']
+author: //span[@class='author']
+
+body: //div[@class='tagesnews-content']
+
+# General clenaup
+strip_id_or_class: dachzeile
+strip: //h3
+strip: //p[@class='bodytext']//a
+strip_id_or_class: autor_datum
+strip_id_or_class: comments
+strip_id_or_class: banner-
+
+strip: //p[contains(., 'Lesen Sie')]
+strip: //p[contains(., '– in DAZ')]
+
+# Fix image captions
+replace_string(<p class="image_caption">): <p><small><em>
+replace_string(</dd>): </em></small></dd>
+
+test_url: http://www.deutsche-apotheker-zeitung.de/pharmazie/news/2014/09/03/weniger-nebenwirkungen-aber-kein-zusatznutzen/13715.html
+test_url: http://www.deutsche-apotheker-zeitung.de/recht/news/2014/09/02/urteile-zum-cannabis-eigenanbau-bfarm-geht-in-berufung/13716.html
+
diff --git a/tagesspiegel.de.txt b/tagesspiegel.de.txt
new file mode 100644
index 0000000..57e7d3d
--- /dev/null
+++ b/tagesspiegel.de.txt
@@ -0,0 +1,60 @@
+# Author: zinnober
+# Should work with "normal" articles as well as with image galleries
+
+prune: no
+
+# Title
+title: //h1/span[@class='hcf-headline']
+
+# Set author
+author: //a[@rel='author']
+
+# Set date
+date: //span[@class='date hcf-atlas']
+
+# Fetch full multipage articles
+next_page_link: //a[contains(@class, 'hcf-forward')]
+
+# Content is here
+body: //article
+body: //div[contains(@class, 'hcf-screen')]
+
+# Remove tracking and ads
+strip_id_or_class: hcf-ad
+strip_id_or_class: hcf-autoload-ad
+strip_id_or_class: hcf-content-ad
+
+# Tidy up before article
+strip: //article/h1
+strip_id_or_class: hcf-atlas
+strip_id_or_class: hcf-author
+strip_id_or_class: date hcf-atlas
+strip_id_or_class: date hcf-atlas
+
+# General cleanup
+strip: //div[contains(@class, 'hcf-screen')]//h1
+strip: //div[@class='hcf-subpage-titles']//ul
+strip_id_or_class: hcf-doctype-media
+strip_id_or_class: hcf-inline-gallery
+strip_id_or_class: hcf-doctype-video
+strip_id_or_class: hcf-links
+strip_id_or_class: hcf-mini-navi
+strip_id_or_class: hcf-media-control
+strip_id_or_class: hcf-hidden
+replace_string(<span class="hcf-update">Update</span>): <strong>Update: </strong>
+
+# Fix pictures and captions
+replace_string(<a class="hcf-doctype-gallery): <p class="hcf-doctype-gallery
+replace_string(<a class="hcf-doctype-enlarge): <p class="hcf-doctype-enlarge
+replace_string(<figcaption class="hcf-caption">): <br><small><em>
+replace_string(</figcaption>): </em></small>
+
+# Fix image galleries
+replace_string(<a class=" ajaxify): <p class="ajaxify
+replace_string(<div class="hcf-caption"><div><p>): <small><em>
+
+# Try it yourself
+test_url: http://www.tagesspiegel.de/berlin/bezirke/wedding/wedding-jetzt/auf-der-suche-nach-einem-stadtteil-wilder-weiter-wedding/8757156.html
+test_url: http://www.tagesspiegel.de/berlin/olympia-in-berlin-der-flughafen-tegel-soll-das-olympische-dorf-werden/10645036.html
+test_url: http://www.tagesspiegel.de/mediacenter/fotostrecken/berlin/bildergalerie-kreuzberger-der-woche/9305534.html
+