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/douban.com.txt | 40 ++++++++++++------------ 1 file changed, 20 insertions(+), 20 deletions(-) mode change 100644 => 100755 inc/3rdparty/site_config/standard/douban.com.txt (limited to 'inc/3rdparty/site_config/standard/douban.com.txt') diff --git a/inc/3rdparty/site_config/standard/douban.com.txt b/inc/3rdparty/site_config/standard/douban.com.txt old mode 100644 new mode 100755 index 99d7e5dc..d72a2223 --- a/inc/3rdparty/site_config/standard/douban.com.txt +++ b/inc/3rdparty/site_config/standard/douban.com.txt @@ -1,21 +1,21 @@ -# This filter is tested on: -# http://www.douban.com/note/215003067/ -# http://www.douban.com/note/213540049/ -# http://www.douban.com/group/topic/31140104/ - -title: //div[@class='note-header']/h1 -title: //div[@id='content']/h1 - -author: //div[@class='info']/ul/li/a -author: //h3/span/a - -date://div[@class='note-header']/div/span -date://h3/span[contains(@class, 'color-green')] - -body://div[contains(@class, 'note')] -body://div[contains(@class, 'topic-content')] - -strip://h3 - -convert_double_br_tags: yes +# This filter is tested on: +# http://www.douban.com/note/215003067/ +# http://www.douban.com/note/213540049/ +# http://www.douban.com/group/topic/31140104/ + +title: //div[@class='note-header']/h1 +title: //div[@id='content']/h1 + +author: //div[@class='info']/ul/li/a +author: //h3/span/a + +date://div[@class='note-header']/div/span +date://h3/span[contains(@class, 'color-green')] + +body://div[contains(@class, 'note')] +body://div[contains(@class, 'topic-content')] + +strip://h3 + +convert_double_br_tags: yes test_url: http://www.douban.com/group/topic/31140104/ \ No newline at end of file -- cgit v1.2.3