diff options
Diffstat (limited to 'inc/3rdparty/site_config/standard/thestreet.com.txt')
-rw-r--r-- | inc/3rdparty/site_config/standard/thestreet.com.txt | 25 |
1 files changed, 25 insertions, 0 deletions
diff --git a/inc/3rdparty/site_config/standard/thestreet.com.txt b/inc/3rdparty/site_config/standard/thestreet.com.txt new file mode 100644 index 00000000..5de75637 --- /dev/null +++ b/inc/3rdparty/site_config/standard/thestreet.com.txt | |||
@@ -0,0 +1,25 @@ | |||
1 | title: //div[@id='storyHdr']/h1 | ||
2 | title: //div[@id='print']//h2 | ||
3 | body: //div[@class="virtualpage"] | ||
4 | body: //div[@id='print']//div[@id='bd'] | ||
5 | author: //meta[@name="AUTHOR"]/@content | ||
6 | author: (//div[@id='print']//div[@id='bd']/h4)[1] | ||
7 | date: //meta[@name="DATE"]/@content | ||
8 | date: //div[@id='print']//div[@id='dte'] | ||
9 | |||
10 | strip_id_or_class: articleFooter | ||
11 | strip_id_or_class: sidebar | ||
12 | strip_id_or_class: ie6PrintSubhead | ||
13 | strip_id_or_class: subHdr | ||
14 | |||
15 | |||
16 | replace_string(<P/>): </p><p> | ||
17 | |||
18 | prune: no | ||
19 | |||
20 | #TODO: redirects back - perhaps needs referer to work | ||
21 | single_page_link: //div[@id='storyDetail']//a[contains(@href, '/print/')] | ||
22 | |||
23 | test_url: http://www.thestreet.com/story/11386556/1/which-of-these-10-dividend-stocks-is-worth-the-risk.html | ||
24 | # multi page | ||
25 | test_url: http://www.thestreet.com/story/11387090/1/7-ubs-stock-picks-for-2012.html \ No newline at end of file | ||