From 3a0e60c8f74f4ad29f915bb1ffb5de82606d9ceb Mon Sep 17 00:00:00 2001 From: Jeremy Benoist Date: Wed, 17 Feb 2016 16:29:15 +0100 Subject: Remove already removed websites --- carpeaqua.com.txt | 6 ------ cedarrepublican.com.txt | 2 -- crazybutable.com.txt | 3 --- dukebasketballreport.com.txt | 11 ----------- dvice.com.txt | 9 --------- interest.co.nz.txt | 2 -- next-gen.biz.txt | 16 ---------------- on.net.mk.txt | 5 ----- pittnews.com.txt | 8 -------- quepasa.cl.txt | 6 ------ smokingapples.com.txt | 5 ----- sportschau.de.txt | 22 ---------------------- sportsillustrated.cnn.com.txt | 26 -------------------------- theappleblog.com.txt | 3 --- thecarton.net.txt | 10 ---------- theeuropean-magazine.com.txt | 17 ----------------- uzivatelsketestovani.cz.txt | 2 -- varingen.no.txt | 5 ----- vemedio.com.txt | 6 ------ webpaper.nzz.ch.txt | 8 -------- wmnf.org.txt | 13 ------------- 21 files changed, 185 deletions(-) delete mode 100755 carpeaqua.com.txt delete mode 100755 cedarrepublican.com.txt delete mode 100755 crazybutable.com.txt delete mode 100755 dukebasketballreport.com.txt delete mode 100755 dvice.com.txt delete mode 100755 interest.co.nz.txt delete mode 100755 next-gen.biz.txt delete mode 100755 on.net.mk.txt delete mode 100755 pittnews.com.txt delete mode 100755 quepasa.cl.txt delete mode 100755 smokingapples.com.txt delete mode 100755 sportschau.de.txt delete mode 100755 sportsillustrated.cnn.com.txt delete mode 100755 theappleblog.com.txt delete mode 100755 thecarton.net.txt delete mode 100755 theeuropean-magazine.com.txt delete mode 100755 uzivatelsketestovani.cz.txt delete mode 100755 varingen.no.txt delete mode 100755 vemedio.com.txt delete mode 100755 webpaper.nzz.ch.txt delete mode 100755 wmnf.org.txt diff --git a/carpeaqua.com.txt b/carpeaqua.com.txt deleted file mode 100755 index 5ea302e..0000000 --- a/carpeaqua.com.txt +++ /dev/null @@ -1,6 +0,0 @@ -title: //h2 -body: //div[@class='entry'] - -prune: no -# otherwise the footnotes are removed -test_url: http://carpeaqua.com/2011/03/27/the-intersection-of-power-and-portability/ \ No newline at end of file diff --git a/cedarrepublican.com.txt b/cedarrepublican.com.txt deleted file mode 100755 index 42faa52..0000000 --- a/cedarrepublican.com.txt +++ /dev/null @@ -1,2 +0,0 @@ -body: //div[@class='frame']//img[@class='horizontal'] | //div[@class='content'] -test_url: http://cedarrepublican.com/online_features/gift_ideas/sending-mother-s-day-flowers-how-to-be-sure-they/article_b69af9b8-1f05-5352-8621-16ce007e5623.html diff --git a/crazybutable.com.txt b/crazybutable.com.txt deleted file mode 100755 index 037cd17..0000000 --- a/crazybutable.com.txt +++ /dev/null @@ -1,3 +0,0 @@ -title://h2 -body://div[contains(@class, 'entrytext')] -test_url: http://www.crazybutable.com/weblog/archives/2010/07/01/house-ideas-that-worked/ \ No newline at end of file diff --git a/dukebasketballreport.com.txt b/dukebasketballreport.com.txt deleted file mode 100755 index 2978797..0000000 --- a/dukebasketballreport.com.txt +++ /dev/null @@ -1,11 +0,0 @@ -title: //h2/a -author: substring-before(substring-after(//span[@class='byline'], 'by'), ',') -date: substring-before(substring-after(//span[@class='byline'], ','), '|') -body: //div[@class='entry'] - - -# strip out auction stuff at the end of posts -# tidy kills the center tag, so disable it -tidy: no -strip: //center//table -test_url: http://www.dukebasketballreport.com/articles/?p=42660 \ No newline at end of file diff --git a/dvice.com.txt b/dvice.com.txt deleted file mode 100755 index 1a1990e..0000000 --- a/dvice.com.txt +++ /dev/null @@ -1,9 +0,0 @@ -strip://*[@id = 'blog_top_stories'] -strip://*[@id = 'takeover_off'] -strip://*[@id = 'right_gray_box'] -strip://*[@class = 'blog_topics'] -strip://*[@class = 'section_titles'] - -author://div[@class = 'post_author_info']/a -date://div[@class = 'post_date_info'] -test_url: http://dvice.com/archives/2012/05/is-nfc-and-smar.php \ No newline at end of file diff --git a/interest.co.nz.txt b/interest.co.nz.txt deleted file mode 100755 index 28c3310..0000000 --- a/interest.co.nz.txt +++ /dev/null @@ -1,2 +0,0 @@ -body: //div[@id='content'] -test_url: http://www.interest.co.nz/opinion/opinion-when-our-fear-corporate-way-and-our-love-small-business-man-dangerous-thing \ No newline at end of file diff --git a/next-gen.biz.txt b/next-gen.biz.txt deleted file mode 100755 index b8d235d..0000000 --- a/next-gen.biz.txt +++ /dev/null @@ -1,16 +0,0 @@ -# 2011-08-22 [carlo@...] initial version -# 2011-08-22 [carlo@...] removed comments & social links - -tidy: no - -single_page_link: //a[@class="single active"] - -body: //div[@id="main"]//div[@class="content-region"]/article -author: //span[@class="author-name"] -date: //time/text() - -strip_id_or_class: //aside[@id="related"] -strip: //footer - -title: //h1 -test_url: http://www.next-gen.biz/reviews/deus-ex-human-revolution-review \ No newline at end of file diff --git a/on.net.mk.txt b/on.net.mk.txt deleted file mode 100755 index a95c2b0..0000000 --- a/on.net.mk.txt +++ /dev/null @@ -1,5 +0,0 @@ -body: //div[(@class = "statija")] -strip: //div[(@class = "relatedBlock")] -strip: //div[(@class = "swftools")] -strip: //table[(@class = "links")] -test_url: http://on.net.mk/video/na-trkala/lamborghini-aventador-avionot-shto-ne-leta \ No newline at end of file diff --git a/pittnews.com.txt b/pittnews.com.txt deleted file mode 100755 index c302526..0000000 --- a/pittnews.com.txt +++ /dev/null @@ -1,8 +0,0 @@ -title: //h2[@class='post-title'] -author: substring-before(substring-after(//h3[@class='post-byline'],'By:'),'/') -date: substring-before(substring-after(//p[@class='post-details'],'Posted on '),'in') -strip: //h2[@class='post-title'] -strip: //p[@class='post-details'] -strip: //h3[@class='post-byline'] -body: //div[@id='content'] -test_url: http://pittnews.com/newsstory/mens-basketball-pitt-recruit-robinson-to-bring-leadership/ \ No newline at end of file diff --git a/quepasa.cl.txt b/quepasa.cl.txt deleted file mode 100755 index fb09a8f..0000000 --- a/quepasa.cl.txt +++ /dev/null @@ -1,6 +0,0 @@ -title: //h1 - -body: //div[@class="cuerpoArticulo"] - - -test_url: http://www.quepasa.cl/magazine/articulo/print.html?id=5299 \ No newline at end of file diff --git a/smokingapples.com.txt b/smokingapples.com.txt deleted file mode 100755 index c68c132..0000000 --- a/smokingapples.com.txt +++ /dev/null @@ -1,5 +0,0 @@ -title: //h2[@class='custom-entry-title'] -author: substring-after(//span[@class='author vcard'],'by ') -date: substring-after(//span[@class='publ'],'Published on ') -body: //div[@class='postentry-content'] -test_url: http://smokingapples.com/software/popclip-for-mac/ \ No newline at end of file diff --git a/sportschau.de.txt b/sportschau.de.txt deleted file mode 100755 index 1e58b52..0000000 --- a/sportschau.de.txt +++ /dev/null @@ -1,22 +0,0 @@ -title://div[@id='ardContent']/h1 - -author://p[@id='ardAutor'] -author://span[@id='ardQuelle'] -author:string('sportschau.de') - -date:substring-after(//span[@id='ardStand'], 'Stand: ') - -body://div[@id='ardContent'] - -strip://div[@id='ardContent']/h1 -strip://p[@id='ardAutor'] -strip: //div[@class='embeddedPlayer_clipinfo'] -strip: //div[@class='ardMehrZumThemaRechts'] -strip: //*[contains(@class, 'inv')] - -strip: //p[@id='ardAbbinder'] -strip: //div[@class='socialBookmarks'] -strip: //div[@id='ardContentEnd'] -strip: //div[@id='ardDisclaimer'] -strip: //div[@id='ardRechteSpalte'] -test_url: http://www.sportschau.de/sp/fussball/news201203/17/analyse_leverkusen_gladbach.jsp \ No newline at end of file diff --git a/sportsillustrated.cnn.com.txt b/sportsillustrated.cnn.com.txt deleted file mode 100755 index b3da813..0000000 --- a/sportsillustrated.cnn.com.txt +++ /dev/null @@ -1,26 +0,0 @@ -# main sportsillustrated.com articles -# -body: //div[@id="cnnStoryContent"] -title: //div[@id="cnnStoryHeadline"]//h1 -author: //div[@id="cnnSubBanner"]//strong -date: substring-after(//div[@id="cnnTimeStamp"], "Updated: ") -date: substring-after(//div[@id="cnnTimeStamp"], "Posted: ") - -# kill ugly font buttons -strip: //div[@id="cnnSCFontButtons"] - -# kill misc filler videos & etc -strip: //div[@class="cnnDivideContent"] -strip: //*[@class="cnnTMbox"] - -# si vault articles -# ------------- -body: //div[@class="siv_artPara"] -title: //div[@class="siv_artHeader"]//h1 -author: //div[@class="byline"] -date: //div[@class="date"] - -next_page_link: //div[@id='cnnStoryContinue']/a -strip_id_or_class: cnnstorypagination - -test_url: http://sportsillustrated.cnn.com/2012/writers/peter_king/02/27/combine/index.html \ No newline at end of file diff --git a/theappleblog.com.txt b/theappleblog.com.txt deleted file mode 100755 index caa5ae0..0000000 --- a/theappleblog.com.txt +++ /dev/null @@ -1,3 +0,0 @@ -# Remove home link -strip: //div[@id='blog-title']/a -test_url: http://theappleblog.com/2010/10/21/the-new-macbook-air-is-underwhelming/ \ No newline at end of file diff --git a/thecarton.net.txt b/thecarton.net.txt deleted file mode 100755 index 13fa35a..0000000 --- a/thecarton.net.txt +++ /dev/null @@ -1,10 +0,0 @@ -title: substring-before(//title, ' – ') -author:string('Shawn') -date: //*/time/@pubdate - - -strip: //header -strip: //div[@id='prev_next'] -strip: //div[@id='masthead'] - -test_url: http://thecarton.net/2012/12/20/imdb \ No newline at end of file diff --git a/theeuropean-magazine.com.txt b/theeuropean-magazine.com.txt deleted file mode 100755 index a19bae1..0000000 --- a/theeuropean-magazine.com.txt +++ /dev/null @@ -1,17 +0,0 @@ -# Tested on: -# http://theeuropean-magazine.com/352-dyson-george/353-evolution-and-innovation -# http://theeuropean-magazine.com/522-casertano-stefano/919-morsi-and-the-future-of-egypt - -title://h2[@class='article-title'] -author:substring-before(substring-after(//p[@class='article-meta'], 'by'), '—') -date:substring-after(//p[@class='article-meta'], '—') -body://div[@class='article'] - -wrap_in(strong)://p[@class='article-teaser'] -move_into(//div[@class='article-head'])://li/img - -strip://h2[@class='article-title'] -strip://p[@class='article-meta'] -strip://div[@class='copyright'] -strip://div[@class='opinions-of-readers'] -test_url: http://theeuropean-magazine.com/522-casertano-stefano/919-morsi-and-the-future-of-egypt \ No newline at end of file diff --git a/uzivatelsketestovani.cz.txt b/uzivatelsketestovani.cz.txt deleted file mode 100755 index 3661b06..0000000 --- a/uzivatelsketestovani.cz.txt +++ /dev/null @@ -1,2 +0,0 @@ -title:h1 -test_url: http://www.uzivatelsketestovani.cz/wiki/doku.php/skoleni-axure-rp \ No newline at end of file diff --git a/varingen.no.txt b/varingen.no.txt deleted file mode 100755 index c0133c9..0000000 --- a/varingen.no.txt +++ /dev/null @@ -1,5 +0,0 @@ -title: //div[@class='ArticleHeadlineDetailedView'] -date: //span[@class='ArticlePublicationDateTimeDetailedView'] -author://span[@class='ArticleBylineDetailedView'] -body: //div[@class='ArticleTextDetailedView'] -test_url: http://www.varingen.no/Nyheter/tabid/392/Default.aspx?ModuleId=56651&articleView=true \ No newline at end of file diff --git a/vemedio.com.txt b/vemedio.com.txt deleted file mode 100755 index d22fc5c..0000000 --- a/vemedio.com.txt +++ /dev/null @@ -1,6 +0,0 @@ -title: //h2 -date: substring-before(//small," • Permalink") -author:string('Martin Hering') - -Strip: //p/small -test_url: http://vemedio.com/blog/posts/state-of-support-and-icloud \ No newline at end of file diff --git a/webpaper.nzz.ch.txt b/webpaper.nzz.ch.txt deleted file mode 100755 index cea1014..0000000 --- a/webpaper.nzz.ch.txt +++ /dev/null @@ -1,8 +0,0 @@ -tidy: no -dissolve: //div[@id="content"]/div/article/header -body: //div[@id="content"]/div/article -title: //div[@id="content"]/div/article/h1 -date: //div[@id="content"]/div/article/header/div[@id="issueSelectTrigger"] -strip: //div[@id="content"]/div/article/h1 - -test_url: http://webpaper.nzz.ch/2012/06/23/front/JJKMS/aphrodite-und-die-kommunisten?guest_pass=24a3ca5b6d%3AJJKMS%3Ad30e1be8628c099669671d4da56cdce4187790ba \ No newline at end of file diff --git a/wmnf.org.txt b/wmnf.org.txt deleted file mode 100755 index 1d403a9..0000000 --- a/wmnf.org.txt +++ /dev/null @@ -1,13 +0,0 @@ -title: //div[@class="bodyText"]/h1/text() -body: //div[@class="bodyText"] - -# author and date are separated by only a newline -# can't figure out how to tokenize that yet -author: //div[@class="bodyText"]/span[@class="info"]/text() -date: //div[@class="bodyText"]/span[@class="info"]/text() - -# strip metdata from body text -strip: //div[@class="bodyText"]/h1/text() -strip: //div[@class="bodyText"]/span[@class="info"] -strip: //div[@class="bodyText"]/span[@class="info"] -test_url: http://www.wmnf.org/news_stories/light-rail-advocates-join-forces-to-combat-opposition-in-pinellas \ No newline at end of file -- cgit v1.3.1