updated specific configuration for parsing

1 files changed, 19 insertions, 17 deletions

diff --git a/inc/3rdparty/site_config/standard/theatlantic.com.txt b/inc/3rdparty/site_config/standard/theatlantic.com.txt
index 267fd39c..aa41b153 100644..100755
--- a/inc/3rdparty/site_config/standard/theatlantic.com.txt
+++ b/inc/3rdparty/site_config/standard/theatlantic.com.txt
@@ -1,18 +1,20 @@
-title: //div[@id='article']/h1

-title: //h1

-

-body: //div[@class='articleText']

-body: //div[@class='articleContent']

-body: //div[@id='article']

-date: //*[contains(@class, 'date')]

-author: //div[@id='profile']//*[@class='authors']//a[1]

-author: //*[@class='author']/span

-prune: no

-

-strip: //div[@class='moreOnBoxWithImages']

-

-single_page_link: //a[@class='print']

-

-test_url: http://www.theatlantic.com/technology/archive/2011/04/want-to-see-how-crazy-a-bot-run-market-can-be/237773/

-test_url: http://www.theatlantic.com/magazine/archive/2007/11/the-autumn-of-the-multitaskers/6342/

+title: //div[contains(@class, 'articleHead')]//h1
+
+body: //div[@class='articleText']
+body: //div[@class='articleContent']
+body: //div[@id='article']
+date: //*[contains(@class, 'date')]
+author: //div[@id='profile']//*[@class='authors']//a[1]
+author: //*[@class='author']/span
+prune: no
+
+strip: //div[@class='moreOnBoxWithImages']
+strip: //p[contains(., 'This article available online at:')]
+strip: //p[contains(., 'This article available online at:')]/following::*
+strip: //div[@class='earthbox']
+
+single_page_link: //article//a[contains(@class, 'print')]
+
+test_url: http://www.theatlantic.com/technology/archive/2011/04/want-to-see-how-crazy-a-bot-run-market-can-be/237773/
+test_url: http://www.theatlantic.com/magazine/archive/2007/11/the-autumn-of-the-multitaskers/6342/
 test_url: http://www.theatlantic.com/entertainment/archive/2012/04/30-rock-live-a-funny-reminder-of-why-sitcoms-arent-shot-live-anymore/256447/
\ No newline at end of file