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/pittnews.com.txt | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/pittnews.com.txt (limited to 'inc/3rdparty/site_config/standard/pittnews.com.txt') diff --git a/inc/3rdparty/site_config/standard/pittnews.com.txt b/inc/3rdparty/site_config/standard/pittnews.com.txt old mode 100644 new mode 100755 index 92777073..c302526d --- a/inc/3rdparty/site_config/standard/pittnews.com.txt +++ b/inc/3rdparty/site_config/standard/pittnews.com.txt @@ -1,8 +1,8 @@ -title: //h2[@class='post-title'] -author: substring-before(substring-after(//h3[@class='post-byline'],'By:'),'/') -date: substring-before(substring-after(//p[@class='post-details'],'Posted on '),'in') -strip: //h2[@class='post-title'] -strip: //p[@class='post-details'] -strip: //h3[@class='post-byline'] +title: //h2[@class='post-title'] +author: substring-before(substring-after(//h3[@class='post-byline'],'By:'),'/') +date: substring-before(substring-after(//p[@class='post-details'],'Posted on '),'in') +strip: //h2[@class='post-title'] +strip: //p[@class='post-details'] +strip: //h3[@class='post-byline'] body: //div[@id='content'] test_url: http://pittnews.com/newsstory/mens-basketball-pitt-recruit-robinson-to-bring-leadership/ \ No newline at end of file -- cgit v1.2.3