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 --- inc/3rdparty/site_config/standard/newyorker.com.txt | 21 +++++++++++---------- 1 file changed, 11 insertions(+), 10 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/newyorker.com.txt (limited to 'inc/3rdparty/site_config/standard/newyorker.com.txt') diff --git a/inc/3rdparty/site_config/standard/newyorker.com.txt b/inc/3rdparty/site_config/standard/newyorker.com.txt old mode 100644 new mode 100755 index 5624aa8c..950324a3 --- a/inc/3rdparty/site_config/standard/newyorker.com.txt +++ b/inc/3rdparty/site_config/standard/newyorker.com.txt @@ -1,10 +1,11 @@ -title: //h1[@id='articlehed'] | //h2[@id="articleintro"] -body: //div[@id='articletext'] - -strip: //ul[@id="bc"] | //div[@id="yrail"] | //div[@class="entry-keywords"] | //div[@class="entry-categories"] | //div[@class="socialUtils"] | //div[@id="footer"] - -date: //h4[@id='articleauthor']/span[@class='dd dds'] -date: //div[@id="pagebody"]/div[@class='hentry entry']/div[@class='published'] - -single_page_link: //div[@class='paginationViewSinglePage']/a -test_url: http://www.newyorker.com/online/blogs/culture/2012/06/mug-shot-web-sites.html \ No newline at end of file +title: //h1[@id='articlehed'] | //h2[@id="articleintro"] +body: //div[@id='articletext'] + +strip: //ul[@id="bc"] | //div[@id="yrail"] | //div[@class="entry-keywords"] | //div[@class="entry-categories"] | //div[@class="socialUtils"] | //div[@id="footer"] | //div[@class="cartoon"] + +date: //h4[@id='articleauthor']/span[@class='dd dds'] +date: //div[@id="pagebody"]/div[@class='hentry entry']/div[@class='published'] + +single_page_link: //div[@class='paginationViewSinglePage']/a +test_url: http://www.newyorker.com/online/blogs/culture/2012/06/mug-shot-web-sites.html +test_url: http://www.newyorker.com/reporting/2013/04/22/130422fa_fact_bilger?currentPage=all&mobify=0 \ No newline at end of file -- cgit v1.2.3