]>
Commit | Line | Data |
---|---|---|
ac4d1142 NL |
1 | title: //div[@class='story_headline']\r |
2 | author: substring-before(substring-after(//div[@class='story_byline'],'By'),'/')\r | |
3 | date: //div[@class='story_lastupdate'] \r | |
4 | body: //div[@id='story']\r | |
5 | strip: //div[@class='story_byline']\r | |
6 | strip: //div[@class='story_lastupdate']\r | |
7 | strip: //div[@class='story_headline']\r | |
8 | strip: //div[@id='abuse']\r | |
9 | strip: //h2\r | |
10 | strip: //div[@class='pagenumbers_wrap']\r | |
11 | strip: //ul[@class='pagenumbers']\r | |
12 | strip: //div[starts-with(., 'To report inappropriate comments')]\r | |
13 | \r | |
14 | strip_id_or_class: story_share\r | |
15 | strip_id_or_class: OUTBRAIN\r | |
16 | strip_id_or_class: story_box_right\r | |
17 | strip: //div[a[@href='http://www.post-gazette.com/pg/12062/1213990-42.stm']]\r | |
18 | strip: //ul[@id='pikame']/li[position()>1]\r | |
19 | \r | |
20 | prune: no\r | |
21 | tidy: no\r | |
22 | \r | |
23 | single_page_link: //a[contains(@href, '?p=0')]\r | |
24 | \r | |
25 | test_url: http://www.post-gazette.com/stories/sports/penguins/pens-crosby-expects-to-return-thursday-226648/\r | |
26 | test_url: http://www.post-gazette.com/stories/sports/pirates/pirates-fork-over-changes-for-fans-at-pnc-park-629789 |