diff options
Diffstat (limited to 'inc/3rdparty/site_config/standard/deutsche-apotheker-zeitung.de.txt')
-rwxr-xr-x | inc/3rdparty/site_config/standard/deutsche-apotheker-zeitung.de.txt | 29 |
1 files changed, 29 insertions, 0 deletions
diff --git a/inc/3rdparty/site_config/standard/deutsche-apotheker-zeitung.de.txt b/inc/3rdparty/site_config/standard/deutsche-apotheker-zeitung.de.txt new file mode 100755 index 00000000..36709cab --- /dev/null +++ b/inc/3rdparty/site_config/standard/deutsche-apotheker-zeitung.de.txt | |||
@@ -0,0 +1,29 @@ | |||
1 | # Author: zinnober | ||
2 | |||
3 | prune: yes | ||
4 | tidy: yes | ||
5 | |||
6 | title: //h1 | ||
7 | date: //p[@class='news_datum'] | ||
8 | author: //span[@class='author'] | ||
9 | |||
10 | body: //div[@class='tagesnews-content'] | ||
11 | |||
12 | # General clenaup | ||
13 | strip_id_or_class: dachzeile | ||
14 | strip: //h3 | ||
15 | strip: //p[@class='bodytext']//a | ||
16 | strip_id_or_class: autor_datum | ||
17 | strip_id_or_class: comments | ||
18 | strip_id_or_class: banner- | ||
19 | |||
20 | strip: //p[contains(., 'Lesen Sie')] | ||
21 | strip: //p[contains(., '– in DAZ')] | ||
22 | |||
23 | # Fix image captions | ||
24 | replace_string(<p class="image_caption">): <p><small><em> | ||
25 | replace_string(</dd>): </em></small></dd> | ||
26 | |||
27 | test_url: http://www.deutsche-apotheker-zeitung.de/pharmazie/news/2014/09/03/weniger-nebenwirkungen-aber-kein-zusatznutzen/13715.html | ||
28 | test_url: http://www.deutsche-apotheker-zeitung.de/recht/news/2014/09/02/urteile-zum-cannabis-eigenanbau-bfarm-geht-in-berufung/13716.html | ||
29 | |||