diff options
author | Nicolas LÅ“uillet <nicolas@loeuillet.org> | 2014-07-13 10:15:40 +0200 |
---|---|---|
committer | Nicolas LÅ“uillet <nicolas@loeuillet.org> | 2014-07-13 10:15:40 +0200 |
commit | 4e067ceabd705201a16b4c92cf4b23f3b990326c (patch) | |
tree | 939f3a8e5ff3ab9ee414a57a895d3e78e1d46ce3 /inc/3rdparty/site_config/standard/golem.de.txt | |
parent | 58dbe103889148def78b0fc8744d3f94c56a1561 (diff) | |
download | wallabag-4e067ceabd705201a16b4c92cf4b23f3b990326c.tar.gz wallabag-4e067ceabd705201a16b4c92cf4b23f3b990326c.tar.zst wallabag-4e067ceabd705201a16b4c92cf4b23f3b990326c.zip |
updated specific configuration for parsing
Diffstat (limited to 'inc/3rdparty/site_config/standard/golem.de.txt')
-rwxr-xr-x[-rw-r--r--] | inc/3rdparty/site_config/standard/golem.de.txt | 48 |
1 files changed, 24 insertions, 24 deletions
diff --git a/inc/3rdparty/site_config/standard/golem.de.txt b/inc/3rdparty/site_config/standard/golem.de.txt index 6c5d1c4f..6afdebe8 100644..100755 --- a/inc/3rdparty/site_config/standard/golem.de.txt +++ b/inc/3rdparty/site_config/standard/golem.de.txt | |||
@@ -1,25 +1,25 @@ | |||
1 | # Jens Kohl, jens.kohl@... | 1 | # Jens Kohl, jens.kohl@... |
2 | # - Added publication date | 2 | # - Added publication date |
3 | # - Striped pagination block | 3 | # - Striped pagination block |
4 | # - Added single page link | 4 | # - Added single page link |
5 | # - Added xpath-querys for the printer friendly version | 5 | # - Added xpath-querys for the printer friendly version |
6 | 6 | ||
7 | title: //h1 | 7 | title: //h1 |
8 | body: //div[@class='formatted'] | 8 | body: //div[@class='formatted'] |
9 | prune: no | 9 | prune: no |
10 | 10 | ||
11 | date: substring-after(//li[2][@class="text1"], 'Datum:') | 11 | date: substring-after(//li[2][@class="text1"], 'Datum:') |
12 | strip: //ol[@class="list-chapters"] | 12 | strip: //ol[@class="list-chapters"] |
13 | strip_comments: yes | 13 | strip_comments: yes |
14 | 14 | ||
15 | # next: commands for printer friendly pages | 15 | # next: commands for printer friendly pages |
16 | single_page_link: //a[contains(@href, 'print.php?a=')]/@href | 16 | single_page_link: //a[contains(@href, 'print.php?a=')]/@href |
17 | title: //body/h3 | 17 | title: //body/h3 |
18 | strip_image_src: staticrl/images/logo.jpg | 18 | strip_image_src: staticrl/images/logo.jpg |
19 | strip_image_src: http://cpx.golem.de/cpx.php?class=7 | 19 | strip_image_src: http://cpx.golem.de/cpx.php?class=7 |
20 | strip: //body/h3 | 20 | strip: //body/h3 |
21 | strip: //body/b[1] | 21 | strip: //body/b[1] |
22 | strip: //body/b[2] | 22 | strip: //body/b[2] |
23 | strip: //body/b[3] | 23 | strip: //body/b[3] |
24 | strip: //div[1] | 24 | strip: //div[1] |
25 | test_url: http://www.golem.de/1112/88696.html \ No newline at end of file | 25 | test_url: http://www.golem.de/1112/88696.html \ No newline at end of file |