From 4e067ceabd705201a16b4c92cf4b23f3b990326c Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= Date: Sun, 13 Jul 2014 10:15:40 +0200 Subject: updated specific configuration for parsing --- .../site_config/standard/thestreet.com.txt | 48 +++++++++++----------- 1 file changed, 24 insertions(+), 24 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/thestreet.com.txt (limited to 'inc/3rdparty/site_config/standard/thestreet.com.txt') diff --git a/inc/3rdparty/site_config/standard/thestreet.com.txt b/inc/3rdparty/site_config/standard/thestreet.com.txt old mode 100644 new mode 100755 index 5de75637..58eabf00 --- a/inc/3rdparty/site_config/standard/thestreet.com.txt +++ b/inc/3rdparty/site_config/standard/thestreet.com.txt @@ -1,25 +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(

):

- -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 +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(

):

+ +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 -- cgit v1.2.3