[github/wallabag/wallabag.git] / inc / 3rdparty / site_config / standard / thestreet.com.txt

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
Commit	Line	Data
4e067cea NL	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
ac4d1142	25	test_url: http://www.thestreet.com/story/11387090/1/7-ubs-stock-picks-for-2012.html