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/techcrunch.com.txt | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) create mode 100644 inc/3rdparty/site_config/standard/techcrunch.com.txt (limited to 'inc/3rdparty/site_config/standard/techcrunch.com.txt') diff --git a/inc/3rdparty/site_config/standard/techcrunch.com.txt b/inc/3rdparty/site_config/standard/techcrunch.com.txt new file mode 100644 index 00000000..f436acb5 --- /dev/null +++ b/inc/3rdparty/site_config/standard/techcrunch.com.txt @@ -0,0 +1,18 @@ +body: //div[contains(@class, 'media-container') or contains(@class, 'body-copy')] + +author: //a[@class="name"] + +date: //div[@class="post-time"] + +title: //h1[@class="headline"] +strip_id_or_class: module-crunchbase + +# The following is for the mobile site +body: //div[@id="singlentry"] +author: substring-after(//span[@class="single-post-meta-top"],'rsaquo; ') +date: substring-before(//div[@class="single-post-meta-top"],' @') +title: //a[@class="sh2"] + +prune: no + +test_url: http://techcrunch.com/2011/10/18/apples-insanely-great-q1-2012/ \ No newline at end of file -- cgit v1.2.3