diff options
| -rw-r--r-- | buzzfeed.com.txt | 2 | ||||
| -rw-r--r-- | finance.yahoo.com.txt | 4 |
2 files changed, 4 insertions, 2 deletions
diff --git a/buzzfeed.com.txt b/buzzfeed.com.txt index 97dddae..ea8ad4f 100644 --- a/buzzfeed.com.txt +++ b/buzzfeed.com.txt @@ -10,6 +10,8 @@ date: //time[@data-print='date'] body: //div[@data-print='body'] body: //section[@data-print='body'] +native_ad_clue: //meta[@property="article:section" and @content="Advertiser"] + # For various things... strip: *[@data-print="ignore"] test_url: http://www.buzzfeed.com/hgrant/35-reasons-why-dogs-hate-the-holidays
\ No newline at end of file diff --git a/finance.yahoo.com.txt b/finance.yahoo.com.txt index 248522c..0c967db 100644 --- a/finance.yahoo.com.txt +++ b/finance.yahoo.com.txt @@ -5,8 +5,8 @@ strip: //div[contains(@class, 'related-companies')] strip: //div[@id='y-article-related'] strip: //div[@id='ypf-article-related'] prune: no +tidy: no single_page_link: //div[@class='ft']//a[contains(@href, 'page=all')] -test_url: http://sg.finance.yahoo.com/news/Motorola-takes-wraps-249-rsg-3508842732.html?x=0&.v=1 -test_url: http://finance.yahoo.com/news/super-young-retirement-savers.html
\ No newline at end of file +test_url: http://finance.yahoo.com/news/canadian-orebodies-gives-notice-exercise-130000032.html
\ No newline at end of file |