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/theatlantic.com.txt | 36 ++++++++++++---------- 1 file changed, 19 insertions(+), 17 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/theatlantic.com.txt (limited to 'inc/3rdparty/site_config/standard/theatlantic.com.txt') diff --git a/inc/3rdparty/site_config/standard/theatlantic.com.txt b/inc/3rdparty/site_config/standard/theatlantic.com.txt old mode 100644 new mode 100755 index 267fd39c..aa41b153 --- 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 -- cgit v1.2.3