From ac4d114214d820b20e18518a2dbc809337e39043 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Nicolas=20L=C5=93uillet?= Date: Fri, 6 Dec 2013 10:13:03 +0100 Subject: [add] new specific configuration files --- .../site_config/standard/technologyreview.com.txt | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) create mode 100644 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 new file mode 100644 index 00000000..41f21d46 --- /dev/null +++ b/inc/3rdparty/site_config/standard/technologyreview.com.txt @@ -0,0 +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')] +test_url: http://www.technologyreview.com/news/427567/facebooks-telescope-on-human-behavior/ \ No newline at end of file -- cgit v1.2.3