From 5893b1ea08156e3edc58692b45bc5946a13709a4 Mon Sep 17 00:00:00 2001 From: Jeremy Benoist Date: Tue, 16 Feb 2016 14:57:02 +0100 Subject: Update website - remove redirected website where html structure isn't the same - remove redirected website which already have a siteconfig for the redirected url - rename website where html structure is the same as the previous one --- online.wsj.com.txt | 25 ------------------------- 1 file changed, 25 deletions(-) delete mode 100755 online.wsj.com.txt (limited to 'online.wsj.com.txt') diff --git a/online.wsj.com.txt b/online.wsj.com.txt deleted file mode 100755 index 448bb7e..0000000 --- a/online.wsj.com.txt +++ /dev/null @@ -1,25 +0,0 @@ -title: //meta[@property="og:title"]/@content -body: //div[@id='article_story_body'] - -author: //h3[@class='byline']/a -# for slide show content -body: //ul[@id='imageSlide']//li[@class='firstSlide']//img | (//div[@class='txt_body']//p)[1] -date: //li[@class='dateStamp']/small - -strip_id_or_class: insetFullBracket -strip_id_or_class: insettipBox -#strip_id_or_class: legacyInset -strip_id_or_class: recipeACShopAndBuyText - -strip: //div[contains(@class, 'insetContent')]//cite -strip: //*[contains(@style, 'visibility: hidden;')] -strip: //div[contains(@class, 'insetContent') and not(contains(@class, 'image'))] -strip: //div[contains(@class, 'carousel')] - -prune: no -tidy: no - -test_url: http://online.wsj.com/news/articles/SB10001424052702304626304579509100018004342 -test_url: http://online.wsj.com/article/SB10001424052970203363504577185322849515102.html -# slide show -test_url: http://online.wsj.com/article/SB10001424052970204791104577110550376458164.html -- cgit v1.3.1