2014-07-13 04:15:40 -04:00
|
|
|
#keep all body text
|
|
|
|
|
prune: no
|
|
|
|
|
|
|
|
|
|
#title, body, metadata
|
|
|
|
|
title: //div[@class='story_header']/h1
|
|
|
|
|
body: //div[@id='content']
|
|
|
|
|
author: substring-after(//span[@class='byline'], "by ")
|
|
|
|
|
author: substring-after(//span[@class='byline'], "By ")
|
|
|
|
|
author: //span[@class='byline']
|
|
|
|
|
date: //span[@class='date']
|
|
|
|
|
|
|
|
|
|
#formatting
|
|
|
|
|
convert_double_br_tags: yes
|
|
|
|
|
dissolve: //div[@class='slides_full']/ul/li
|
|
|
|
|
|
|
|
|
|
# cleanup
|
|
|
|
|
strip: //a[@id='story_note']
|
|
|
|
|
strip: //br
|
|
|
|
|
strip: //div[@class='intro']
|
|
|
|
|
strip: //div[@class='share-block']
|
|
|
|
|
strip: //div[@class='sidebar-social']
|
|
|
|
|
strip: //div[@class='top-stories']
|
|
|
|
|
strip: //div[@class='prevnext']
|
2013-12-06 04:13:03 -05:00
|
|
|
test_url: http://www.thedaily.com/page/2012/01/09/010912-news-college-costs-1-5/
|
