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 --- inc/3rdparty/site_config/standard/ftchinese.com.txt | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) create mode 100755 inc/3rdparty/site_config/standard/ftchinese.com.txt (limited to 'inc/3rdparty/site_config/standard/ftchinese.com.txt') diff --git a/inc/3rdparty/site_config/standard/ftchinese.com.txt b/inc/3rdparty/site_config/standard/ftchinese.com.txt new file mode 100755 index 00000000..5c94d9b0 --- /dev/null +++ b/inc/3rdparty/site_config/standard/ftchinese.com.txt @@ -0,0 +1,18 @@ +# Modified to define the single_page_link +# This filter is tested on: +# http://www.ftchinese.com/story/001047373 +# http://www.ftchinese.com/story/001047631 +# http://www.ftchinese.com/story/001047622/?print=y +# http://www.ftchinese.com/story/001049052 +# http://www.ftchinese.com/story/001049088 + +title:substring-before(//title, '-') +author: //div[@class='byline']/a +date: //a[@class='storytime'] +#Set date in print view +#date: //div[@class='byline']/a/following-sibling::a +body: //div[@id="bodytext"] +strip://div[@class='pagination'] +single_page_link://div[@class='pagination']/a[.='全文'] +#next_page_link: //div[@class='pagination']//a[.='下一页'] +test_url: http://www.ftchinese.com/story/001049088 \ No newline at end of file -- cgit v1.2.3