]>
Commit | Line | Data |
---|---|---|
ac4d1142 NL |
1 | # Jens Kohl, jens.kohl@...\r |
2 | # - Added publication date\r | |
3 | # - Striped pagination block\r | |
4 | # - Added single page link\r | |
5 | # - Added xpath-querys for the printer friendly version\r | |
6 | \r | |
7 | title: //h1\r | |
8 | body: //div[@class='formatted']\r | |
9 | prune: no\r | |
10 | \r | |
11 | date: substring-after(//li[2][@class="text1"], 'Datum:')\r | |
12 | strip: //ol[@class="list-chapters"]\r | |
13 | strip_comments: yes\r | |
14 | \r | |
15 | # next: commands for printer friendly pages\r | |
16 | single_page_link: //a[contains(@href, 'print.php?a=')]/@href\r | |
17 | title: //body/h3\r | |
18 | strip_image_src: staticrl/images/logo.jpg\r | |
19 | strip_image_src: http://cpx.golem.de/cpx.php?class=7\r | |
20 | strip: //body/h3\r | |
21 | strip: //body/b[1]\r | |
22 | strip: //body/b[2]\r | |
23 | strip: //body/b[3]\r | |
24 | strip: //div[1]\r | |
25 | test_url: http://www.golem.de/1112/88696.html |