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/walrusmagazine.com.txt | 24 +++++++++++----------- 1 file changed, 12 insertions(+), 12 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/walrusmagazine.com.txt (limited to 'inc/3rdparty/site_config/standard/walrusmagazine.com.txt') diff --git a/inc/3rdparty/site_config/standard/walrusmagazine.com.txt b/inc/3rdparty/site_config/standard/walrusmagazine.com.txt old mode 100644 new mode 100755 index 3ab22172..c53eb0dd --- a/inc/3rdparty/site_config/standard/walrusmagazine.com.txt +++ b/inc/3rdparty/site_config/standard/walrusmagazine.com.txt @@ -1,14 +1,14 @@ -title: //div[@id='pr']/h3 -author: //div[@class='dateline']//a[contains(@href, '/author/')] - -# print page -body: //div[@id='prbody'] -# standard page -body: //div[@id='pgbody'] - -# for multi-page articles -single_page_link: //div[@class='tipjar']//a[contains(@href, '/printerFriendly.php?')] - -prune: no +title: //div[@id='pr']/h3 +author: //div[@class='dateline']//a[contains(@href, '/author/')] + +# print page +body: //div[@id='prbody'] +# standard page +body: //div[@id='pgbody'] + +# for multi-page articles +single_page_link: //div[@class='tipjar']//a[contains(@href, '/printerFriendly.php?')] + +prune: no test_url: http://www.walrusmagazine.com/articles/2011.12-memoir-kidnapped \ No newline at end of file -- cgit v1.2.3