aboutsummaryrefslogtreecommitdiffhomepage
path: root/inc/3rdparty/site_config/standard/smithsonianmag.com.txt
diff options
context:
space:
mode:
Diffstat (limited to 'inc/3rdparty/site_config/standard/smithsonianmag.com.txt')
-rwxr-xr-x[-rw-r--r--]inc/3rdparty/site_config/standard/smithsonianmag.com.txt36
1 files changed, 18 insertions, 18 deletions
diff --git a/inc/3rdparty/site_config/standard/smithsonianmag.com.txt b/inc/3rdparty/site_config/standard/smithsonianmag.com.txt
index 10a3f717..3e8fee95 100644..100755
--- a/inc/3rdparty/site_config/standard/smithsonianmag.com.txt
+++ b/inc/3rdparty/site_config/standard/smithsonianmag.com.txt
@@ -1,20 +1,20 @@
1# meta data 1# meta data
2title://h1[@id = 'articleTitle'] 2title://h1[@id = 'articleTitle']
3author:substring-after(//ul[@id = 'byLine']/li[1],'By ') 3author:substring-after(//ul[@id = 'byLine']/li[1],'By ')
4date:substring-before(substring-after(//ul[@id = 'byLine']/li[last()],','),',') 4date:substring-before(substring-after(//ul[@id = 'byLine']/li[last()],','),',')
5body://div[@id = 'article-body'] 5body://div[@id = 'article-body']
6 6
7# full content 7# full content
8single_page_link://td/li[@class = 'article-singlepage']/a 8single_page_link://td/li[@class = 'article-singlepage']/a
9 9
10# caption clean up 10# caption clean up
11wrap_in(i)://span[@class='articleImageCaptionwide'] 11wrap_in(i)://span[@class='articleImageCaptionwide']
12move_into (//span[@class='articleImageCaptionwide'])://div[@id = 'articleImage']/p 12move_into (//span[@class='articleImageCaptionwide'])://div[@id = 'articleImage']/p
13 13
14 14
15# clean up 15# clean up
16strip://p[@id = 'articlePaginationWrapper'] 16strip://p[@id = 'articlePaginationWrapper']
17strip://ul[contains(@class, 'cat-breadcrumb')] 17strip://ul[contains(@class, 'cat-breadcrumb')]
18strip://div [@class= 'viewMorePhotos'] 18strip://div [@class= 'viewMorePhotos']
19 19
20test_url: http://www.smithsonianmag.com/history-archaeology/The-Goddess-Goes-Home.html \ No newline at end of file 20test_url: http://www.smithsonianmag.com/history-archaeology/The-Goddess-Goes-Home.html \ No newline at end of file