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 --- inc/3rdparty/site_config/standard/gigaom.com.txt | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) create mode 100644 inc/3rdparty/site_config/standard/gigaom.com.txt (limited to 'inc/3rdparty/site_config/standard/gigaom.com.txt') diff --git a/inc/3rdparty/site_config/standard/gigaom.com.txt b/inc/3rdparty/site_config/standard/gigaom.com.txt new file mode 100644 index 00000000..348bdf23 --- /dev/null +++ b/inc/3rdparty/site_config/standard/gigaom.com.txt @@ -0,0 +1,17 @@ +date: //meta[@name='DC.date.issued']/@content +date: //span[@class='post-meta the-date'] + +title: //meta[@property='og:title']/@content + +author: //meta[@name='DC.creator']/@content + +body: //div[contains(@class, 'post-sub-head') or starts-with(@id, 'post-content-')] + +find_string: id="content" +replace_string: id="content-ignore" + +strip_id_or_class: sharedaddy + +prune: no + +test_url: http://gigaom.com/2011/10/24/groupon-google-lawsuit/ \ No newline at end of file -- cgit v1.2.3