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/foreignpolicy.com.txt | 26 +++++++++++++--------- 1 file changed, 15 insertions(+), 11 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/foreignpolicy.com.txt (limited to 'inc/3rdparty/site_config/standard/foreignpolicy.com.txt') diff --git a/inc/3rdparty/site_config/standard/foreignpolicy.com.txt b/inc/3rdparty/site_config/standard/foreignpolicy.com.txt old mode 100644 new mode 100755 index 6ab7a091..4e84b989 --- a/inc/3rdparty/site_config/standard/foreignpolicy.com.txt +++ b/inc/3rdparty/site_config/standard/foreignpolicy.com.txt @@ -1,11 +1,15 @@ -title: //div[@id='art-mast']//h1 -author: substring-after(//span[@id='by-line'], 'BY ') -date: //span[@id='pub-date'] -body: //div[@id='art-mast']//h2 | //div[@id='art-mast']/h3 | //div[@id='art-body']//div[@class='translateBody'] -strip: //div[@id='share-box'] -prune: no - -single_page_link: //span[@id='controls']/a[contains(@href, 'print=yes')] - -test_url: http://www.foreignpolicy.com/articles/2011/08/01/a_murderers_manifesto_and_me -test_url: test_url: http://www.foreignpolicy.com/articles/2012/02/29/five_years_in_damascus \ No newline at end of file +title: //div[@class='translateHead']//h1 | //div[@id='art-mast']//h1 +author: substring-after(//span[@id='by-line'], 'BY ') +date: //span[@id='pub-date'] +body: //div[@id='art-mast']/h2 | //div[@class='translateBody'] | //div[@id='art-body'] +#Strip inside article content +strip: //div[@id='share-box'] +strip: //div[@id='special-box'] + +prune: no + +single_page_link: //span[@id='controls']/a[contains(@href, 'print=yes')] +single_page_link: //a[text()='SINGLE PAGE'] + +test_url: http://www.foreignpolicy.com/articles/2011/08/01/a_murderers_manifesto_and_me +test_url: http://www.foreignpolicy.com/articles/2012/02/29/five_years_in_damascus \ No newline at end of file -- cgit v1.2.3