diff options
Diffstat (limited to 'thestreet.com.txt')
| -rw-r--r-- | thestreet.com.txt | 25 |
1 files changed, 25 insertions, 0 deletions
diff --git a/thestreet.com.txt b/thestreet.com.txt new file mode 100644 index 0000000..58eabf0 --- /dev/null +++ b/thestreet.com.txt @@ -0,0 +1,25 @@ +title: //div[@id='storyHdr']/h1 +title: //div[@id='print']//h2 +body: //div[@class="virtualpage"] +body: //div[@id='print']//div[@id='bd'] +author: //meta[@name="AUTHOR"]/@content +author: (//div[@id='print']//div[@id='bd']/h4)[1] +date: //meta[@name="DATE"]/@content +date: //div[@id='print']//div[@id='dte'] + +strip_id_or_class: articleFooter +strip_id_or_class: sidebar +strip_id_or_class: ie6PrintSubhead +strip_id_or_class: subHdr + + +replace_string(<P/>): </p><p> + +prune: no + +#TODO: redirects back - perhaps needs referer to work +single_page_link: //div[@id='storyDetail']//a[contains(@href, '/print/')] + +test_url: http://www.thestreet.com/story/11386556/1/which-of-these-10-dividend-stocks-is-worth-the-risk.html +# multi page +test_url: http://www.thestreet.com/story/11387090/1/7-ubs-stock-picks-for-2012.html
\ No newline at end of file |