aboutsummaryrefslogtreecommitdiffhomepage
path: root/inc/3rdparty/site_config/standard/tagesspiegel.de.txt
diff options
context:
space:
mode:
Diffstat (limited to 'inc/3rdparty/site_config/standard/tagesspiegel.de.txt')
-rwxr-xr-xinc/3rdparty/site_config/standard/tagesspiegel.de.txt60
1 files changed, 60 insertions, 0 deletions
diff --git a/inc/3rdparty/site_config/standard/tagesspiegel.de.txt b/inc/3rdparty/site_config/standard/tagesspiegel.de.txt
new file mode 100755
index 00000000..57e7d3df
--- /dev/null
+++ b/inc/3rdparty/site_config/standard/tagesspiegel.de.txt
@@ -0,0 +1,60 @@
1# Author: zinnober
2# Should work with "normal" articles as well as with image galleries
3
4prune: no
5
6# Title
7title: //h1/span[@class='hcf-headline']
8
9# Set author
10author: //a[@rel='author']
11
12# Set date
13date: //span[@class='date hcf-atlas']
14
15# Fetch full multipage articles
16next_page_link: //a[contains(@class, 'hcf-forward')]
17
18# Content is here
19body: //article
20body: //div[contains(@class, 'hcf-screen')]
21
22# Remove tracking and ads
23strip_id_or_class: hcf-ad
24strip_id_or_class: hcf-autoload-ad
25strip_id_or_class: hcf-content-ad
26
27# Tidy up before article
28strip: //article/h1
29strip_id_or_class: hcf-atlas
30strip_id_or_class: hcf-author
31strip_id_or_class: date hcf-atlas
32strip_id_or_class: date hcf-atlas
33
34# General cleanup
35strip: //div[contains(@class, 'hcf-screen')]//h1
36strip: //div[@class='hcf-subpage-titles']//ul
37strip_id_or_class: hcf-doctype-media
38strip_id_or_class: hcf-inline-gallery
39strip_id_or_class: hcf-doctype-video
40strip_id_or_class: hcf-links
41strip_id_or_class: hcf-mini-navi
42strip_id_or_class: hcf-media-control
43strip_id_or_class: hcf-hidden
44replace_string(<span class="hcf-update">Update</span>): <strong>Update: </strong>
45
46# Fix pictures and captions
47replace_string(<a class="hcf-doctype-gallery): <p class="hcf-doctype-gallery
48replace_string(<a class="hcf-doctype-enlarge): <p class="hcf-doctype-enlarge
49replace_string(<figcaption class="hcf-caption">): <br><small><em>
50replace_string(</figcaption>): </em></small>
51
52# Fix image galleries
53replace_string(<a class=" ajaxify): <p class="ajaxify
54replace_string(<div class="hcf-caption"><div><p>): <small><em>
55
56# Try it yourself
57test_url: http://www.tagesspiegel.de/berlin/bezirke/wedding/wedding-jetzt/auf-der-suche-nach-einem-stadtteil-wilder-weiter-wedding/8757156.html
58test_url: http://www.tagesspiegel.de/berlin/olympia-in-berlin-der-flughafen-tegel-soll-das-olympische-dorf-werden/10645036.html
59test_url: http://www.tagesspiegel.de/mediacenter/fotostrecken/berlin/bildergalerie-kreuzberger-der-woche/9305534.html
60