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/technologyreview.com.txt | 30 +++++++++++----------- 1 file changed, 15 insertions(+), 15 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/technologyreview.com.txt (limited to 'inc/3rdparty/site_config/standard/technologyreview.com.txt') diff --git a/inc/3rdparty/site_config/standard/technologyreview.com.txt b/inc/3rdparty/site_config/standard/technologyreview.com.txt old mode 100644 new mode 100755 index 41f21d46..d405eb18 --- a/inc/3rdparty/site_config/standard/technologyreview.com.txt +++ b/inc/3rdparty/site_config/standard/technologyreview.com.txt @@ -1,16 +1,16 @@ -title: //header[@class='article-meta']/h1 -title: substring-before(//title, '|') - -body: //section[contains(@class, 'body')] - -# Author & Date for News and Featured Stories -author: //ul[@class='byline']/li/a -author: substring-before(substring-after(//ul[@class='byline']/li, 'By '), ' on') -date: substring-after(//ul[@class='byline']/li, 'on ') - -# Author & Date for "Views" -author: //div[@class='view-byline']/div[@class='meta']/h2[1] -date: //div[@class='view-byline']/div[@class='meta']/h2[2] - -next_page_link: //section[@class='pagination']/a[contains(@class, 'continue')] +title: //header[@class='article-meta']/h1 +title: substring-before(//title, '|') + +body: //section[contains(@class, 'body')] + +# Author & Date for News and Featured Stories +author: //ul[@class='byline']/li/a +author: substring-before(substring-after(//ul[@class='byline']/li, 'By '), ' on') +date: substring-after(//ul[@class='byline']/li, 'on ') + +# Author & Date for "Views" +author: //div[@class='view-byline']/div[@class='meta']/h2[1] +date: //div[@class='view-byline']/div[@class='meta']/h2[2] + +next_page_link: //section[@class='pagination']/a[contains(@class, 'continue')] test_url: http://www.technologyreview.com/news/427567/facebooks-telescope-on-human-behavior/ \ No newline at end of file -- cgit v1.2.3