From d50ffaee66a180b0f9f0fe096cbe93a01b8b3e0d Mon Sep 17 00:00:00 2001 From: Jeremy Benoist Date: Mon, 22 Feb 2016 09:06:53 +0100 Subject: Last cleanup --- aachener-zeitung.de.txt | 10 ---------- alexduner.squarespace.com.txt | 4 ---- americascup.com.txt | 10 ---------- bostonreview.net.txt | 15 --------------- community.servicenow.com.txt | 8 -------- csnbayarea.com.txt | 7 ------- danleech.com.txt | 6 ------ datanami.com.txt | 4 ---- edge.org.txt | 5 ----- etc.se.txt | 6 ------ gotomanager.com.txt | 21 --------------------- laphamsquarterly.org.txt | 13 ------------- lovefm.com.txt | 6 ------ lovetv.com.bz.txt | 9 --------- myfoxboston.com.txt | 4 ---- narenji.ir.txt | 2 -- newsle.com.txt | 2 -- observer.com.txt | 7 ------- sebbo.net.txt | 4 ---- sivers.org.txt | 6 ------ smartinvestor.de.txt | 5 ----- spectator.co.uk.txt | 7 ------- technicallyjordan.tumblr.com.txt | 8 -------- thebrowser.com.txt | 10 ---------- thesiasat.com.txt | 11 ----------- westhamtillidie.com.txt | 6 ------ winsupersite.com.txt | 3 --- xfgjls.com.txt | 11 ----------- zingtrain.com.txt | 3 --- 29 files changed, 213 deletions(-) delete mode 100755 aachener-zeitung.de.txt delete mode 100755 alexduner.squarespace.com.txt delete mode 100755 americascup.com.txt delete mode 100755 bostonreview.net.txt delete mode 100755 community.servicenow.com.txt delete mode 100755 csnbayarea.com.txt delete mode 100755 danleech.com.txt delete mode 100755 datanami.com.txt delete mode 100755 edge.org.txt delete mode 100755 etc.se.txt delete mode 100755 gotomanager.com.txt delete mode 100755 laphamsquarterly.org.txt delete mode 100755 lovefm.com.txt delete mode 100755 lovetv.com.bz.txt delete mode 100755 myfoxboston.com.txt delete mode 100755 narenji.ir.txt delete mode 100755 newsle.com.txt delete mode 100755 observer.com.txt delete mode 100755 sebbo.net.txt delete mode 100755 sivers.org.txt delete mode 100755 smartinvestor.de.txt delete mode 100755 spectator.co.uk.txt delete mode 100755 technicallyjordan.tumblr.com.txt delete mode 100755 thebrowser.com.txt delete mode 100755 thesiasat.com.txt delete mode 100755 westhamtillidie.com.txt delete mode 100755 winsupersite.com.txt delete mode 100755 xfgjls.com.txt delete mode 100755 zingtrain.com.txt diff --git a/aachener-zeitung.de.txt b/aachener-zeitung.de.txt deleted file mode 100755 index 013afa4..0000000 --- a/aachener-zeitung.de.txt +++ /dev/null @@ -1,10 +0,0 @@ -title: //meta[@property='og:title']/@content -body: //*[@class='fliesstext_detail' or @class='detail_fliesstext'] | //img[@itemprop="image" and starts-with(@src, "/sixcms/media.php/")] - -strip_id_or_class: socialshareprivacy1 -strip_id_or_class: zvaFacebookButton - -tidy: no -prune: no - -test_url: http://www.aachener-zeitung.de/sixcms/detail.php?template=az_detail&id=2552718 \ No newline at end of file diff --git a/alexduner.squarespace.com.txt b/alexduner.squarespace.com.txt deleted file mode 100755 index 875405e..0000000 --- a/alexduner.squarespace.com.txt +++ /dev/null @@ -1,4 +0,0 @@ -body: //section[@class='content'] -date: //span[1] -author: //h1[@id='sitetitle'] -test_url: https://alexduner.squarespace.com/blog/2013/1/tech-culture-from-the-outside-looking-in \ No newline at end of file diff --git a/americascup.com.txt b/americascup.com.txt deleted file mode 100755 index 31723f8..0000000 --- a/americascup.com.txt +++ /dev/null @@ -1,10 +0,0 @@ -title: //div[@class="editorial-content"]/h3 -body: //div[@class="hero-image" or @class="editorial-content"] - -strip: //ul[@class="hero-caption"] -strip_id_or_class: footer - -prune: no -tidy: no - -test_url: http://www.americascup.com/en/Latest/News/2012/3/Coutts-and-Peyron-tell-transformative-tale-at-Global-Sports-Forum/ \ No newline at end of file diff --git a/bostonreview.net.txt b/bostonreview.net.txt deleted file mode 100755 index 64e04a1..0000000 --- a/bostonreview.net.txt +++ /dev/null @@ -1,15 +0,0 @@ -#basics -title://h3[@class = 'article_title'] -date://span[@class = 'article_date'] -body://div[@id = 'center_column_article'] -#correct, but author not being picked up in preview -author://span[@class = 'article_author'] - -#strips basics from article -strip_id_or_class:article_title -strip_id_or_class:article_date -strip_id_or_class:article_author - -#strips pull quotes -strip_id_or_class:pull_quote -test_url: http://www.bostonreview.net/BR36.4/megan_pugh_agnes_de_mille_dance.php \ No newline at end of file diff --git a/community.servicenow.com.txt b/community.servicenow.com.txt deleted file mode 100755 index 98fb688..0000000 --- a/community.servicenow.com.txt +++ /dev/null @@ -1,8 +0,0 @@ -body: //div[@id="center"]//div[@class="node"] -title: //div[@id="center"]//h2 -author: substring-after(//div[@id="center"]//div[@class="node"]//span[@class="submitted"], "—") -date: substring-before(//div[@id="center"]//div[@class="node"]//span[@class="submitted"], "—") -strip: //div[@id="center"]//h2[1] -strip: //span[@class="submitted"][1] -move_into(//div[@class="node"])://div[@class="breadcrumb"] -test_url: http://community.servicenow.com/blog/lawrenceeng/seasons-greetings-servicenow-team diff --git a/csnbayarea.com.txt b/csnbayarea.com.txt deleted file mode 100755 index 1da60b4..0000000 --- a/csnbayarea.com.txt +++ /dev/null @@ -1,7 +0,0 @@ -title: //div[@id='csn_blogST_headline']/h1 - -body: //div[@id='csn_blogST_main'] -strip_id_or_class: ipfootnotes -strip: //div[@id='csn_blogST_main']/p[1]/img -strip: //div[@id='csn_blogST_sidebar'] -test_url: http://www.csnbayarea.com/blog/giants-talk/post/-?blog%2Fgiants-talk%2Fpost%2F-=&blockID=578902&feedID=5987 \ No newline at end of file diff --git a/danleech.com.txt b/danleech.com.txt deleted file mode 100755 index 1d4cec7..0000000 --- a/danleech.com.txt +++ /dev/null @@ -1,6 +0,0 @@ -tidy: no -prune: no -date: //article//time[@pubdate] -title: //article/h1//span[contains(@class, 'entry-title')] -body: //article/div[contains(@class, 'entry-content')] -test_url: http://danleech.com/post/36822126876/simple-icons \ No newline at end of file diff --git a/datanami.com.txt b/datanami.com.txt deleted file mode 100755 index e9111a4..0000000 --- a/datanami.com.txt +++ /dev/null @@ -1,4 +0,0 @@ -body: //div[@id="article"] -date: //p[@class="date"] -author: //p[@class="byline"] -test_url: http://www.datanami.com/datanami/2011-12-07/new_path_for_sap:_in_memory_computing,_predictive_analysis_converge.html?featured=top \ No newline at end of file diff --git a/edge.org.txt b/edge.org.txt deleted file mode 100755 index 95805f6..0000000 --- a/edge.org.txt +++ /dev/null @@ -1,5 +0,0 @@ -title: //div[@class='HomeLeftPannel IMGCTRL']/h2 -body: //div[@class='HomeLeftPannel IMGCTRL']//div[@class='Brownalink' or @id='shortdesc'] -tidy: no - -test_url: http://edge.org/print/conversation.php?cid=the-argumentative-theory \ No newline at end of file diff --git a/etc.se.txt b/etc.se.txt deleted file mode 100755 index 95f8cf7..0000000 --- a/etc.se.txt +++ /dev/null @@ -1,6 +0,0 @@ -strip_id_or_class: 'left' -strip_id_or_class: 'right' -strip_id_or_class: 'block-belowcontent' -author: //span[@class = 'name']/a -date: //div[@class= 'datum'] -test_url: http://www.etc.se/intervju/lonsamt-att-radda-jorden \ No newline at end of file diff --git a/gotomanager.com.txt b/gotomanager.com.txt deleted file mode 100755 index f8af732..0000000 --- a/gotomanager.com.txt +++ /dev/null @@ -1,21 +0,0 @@ -title: //span[@id="showTitle"] -author: //span[@id="showAuthor"] -date: //span[@id="showRefDate"] - -strip: //span[@class="black_bold"] -strip: //div[@id="sectionName"] -strip: //div[@id="storyHeader"] - -body: //div[@id="newsBodyText"] - -strip_image_src: "http://www.gotomanager.com/img/mgrm/space.gif" -strip_image_src: "http://www.gotomanager.com/images/separator.gif" -strip_image_src: "http://www.gotomanager.com/images/spaces.gif" - -convert_double_br_tags: yes -tidy: yes - -strip: //div[@id="smallLeadImage"] -strip: //div[@id="truehitsSurvey"] -strip: //table[@id="relatedInfoTable"] -test_url: http://www.gotomanager.com/news/details.aspx?id=86759 \ No newline at end of file diff --git a/laphamsquarterly.org.txt b/laphamsquarterly.org.txt deleted file mode 100755 index d25999d..0000000 --- a/laphamsquarterly.org.txt +++ /dev/null @@ -1,13 +0,0 @@ -title: //h1[@class='headline'] -body: //div[@class='article'] -strip: //div[@class='article']//h3[contains(@class, 'section')] -strip: //div[@class='article']//ul[contains(@class, 'article-actions')] -strip: //div[@id='syndication-upper'] -strip: //a[@id='syndication'] -strip: //dl[@id='article-tags'] -strip: //div[@id='article-like'] -prune: no - -single_page_link: //li[@class='single-page']/a - -test_url: http://www.laphamsquarterly.org/essays/balanced-diets.php \ No newline at end of file diff --git a/lovefm.com.txt b/lovefm.com.txt deleted file mode 100755 index 20d26c5..0000000 --- a/lovefm.com.txt +++ /dev/null @@ -1,6 +0,0 @@ -title: //*[@id='title'] -date: //*[@id='date'] -body: //*[@id='desc'] -tidy: no - -test_url: http://www.lovefm.com/local_news.php?item=2176 \ No newline at end of file diff --git a/lovetv.com.bz.txt b/lovetv.com.bz.txt deleted file mode 100755 index a71fccd..0000000 --- a/lovetv.com.bz.txt +++ /dev/null @@ -1,9 +0,0 @@ -title: //div[contains(@class, 'post')]//h1 -body: //div[contains(@class, 'post')] -strip: //hr -strip_id_or_class: post-meta - -prune: no - -test_url: http://www.lovetv.com.bz/2013/06/28/recently-discovered-ancient-maya-wooden-canoe-paddle-to-be-handed-over-to-archaeology/ -test_url: http://www.lovetv.com.bz/feed/ \ No newline at end of file diff --git a/myfoxboston.com.txt b/myfoxboston.com.txt deleted file mode 100755 index 9ad8ce0..0000000 --- a/myfoxboston.com.txt +++ /dev/null @@ -1,4 +0,0 @@ -body: //div[@class="col1"]//div[@class="photo"] | //div[@class="detail"]/p[@class="fontStyle21"] | //div[@class="story last"] -tidy: no - -test_url: http://www.myfoxboston.com/dpp/news/local/transit-police-say-woman-spat-on-mbta-bus-driver-2010611 \ No newline at end of file diff --git a/narenji.ir.txt b/narenji.ir.txt deleted file mode 100755 index 6c3d0c2..0000000 --- a/narenji.ir.txt +++ /dev/null @@ -1,2 +0,0 @@ -body: //div[@class='node'] -test_url: http://www.narenji.ir/2806 \ No newline at end of file diff --git a/newsle.com.txt b/newsle.com.txt deleted file mode 100755 index e500ddc..0000000 --- a/newsle.com.txt +++ /dev/null @@ -1,2 +0,0 @@ -single_page_link: //iframe/@src -test_url: http://newsle.com/article/0/15831103/ \ No newline at end of file diff --git a/observer.com.txt b/observer.com.txt deleted file mode 100755 index 0b10753..0000000 --- a/observer.com.txt +++ /dev/null @@ -1,7 +0,0 @@ -body: //article[contains(@class, 'instapaper_body')] - -prune: no - -single_page_link: //a[@id='print-button'] - -test_url: http://www.observer.com/2008/would-you-take-tumblr-man \ No newline at end of file diff --git a/sebbo.net.txt b/sebbo.net.txt deleted file mode 100755 index b6d9c92..0000000 --- a/sebbo.net.txt +++ /dev/null @@ -1,4 +0,0 @@ -title: substring-before(//title, '«') -body: //div[@class = 'entry'] -strip_id_or_class: 'postmetabox' -test_url: http://sebbo.net/2010/12/akkus/ \ No newline at end of file diff --git a/sivers.org.txt b/sivers.org.txt deleted file mode 100755 index a88f30d..0000000 --- a/sivers.org.txt +++ /dev/null @@ -1,6 +0,0 @@ -title: //article[@class='post']/header[@class='wrapper']//h1/a -author: //header[@id='masthead']//h1/a -date: //article[@class='post']/header[@class='wrapper']//p[@class='postdate'] -body: //div[@id='body-content'] - -test_url: http://sivers.org/delegate/ \ No newline at end of file diff --git a/smartinvestor.de.txt b/smartinvestor.de.txt deleted file mode 100755 index 85ca46d..0000000 --- a/smartinvestor.de.txt +++ /dev/null @@ -1,5 +0,0 @@ -title: //td[@class='hweissblau2'] -body: //p[@class='copy'] | //div[@class='Section1'] -prune: no - -test_url: http://www.smartinvestor.de/news/smartinvestor/detail.hbs?itemid=item949496655&recnr=14593 \ No newline at end of file diff --git a/spectator.co.uk.txt b/spectator.co.uk.txt deleted file mode 100755 index d0605ed..0000000 --- a/spectator.co.uk.txt +++ /dev/null @@ -1,7 +0,0 @@ -author: /html/body/div[@id='wrapper']/div[@id='main-content']/div[@class='article_body']/a[@class='author-link']/child::text() - -body: /html/body/div[@id='wrapper']/div[@id='main-content']/div[@class='article_body'] - -# Not very helpfull, the title and author are container by the same element that contains the body -strip: /html/body/div[@id='wrapper']/div[@id='main-content']/div[@class='article_body']/h2 | /html/body/div[@id='wrapper']/div[@id='main-content']/div[@class='article_body']/a[@class='author-link'] -test_url: http://www.spectator.co.uk/arts-and-culture/night-and-day/7449683/spotify-sunday-my-personal-soundtrack.thtml \ No newline at end of file diff --git a/technicallyjordan.tumblr.com.txt b/technicallyjordan.tumblr.com.txt deleted file mode 100755 index d871b60..0000000 --- a/technicallyjordan.tumblr.com.txt +++ /dev/null @@ -1,8 +0,0 @@ -title: //h2 -author: //meta[@name="author"]/@content -date: //h3 -body: //div[@class="postBody"] -strip: //h1 -strip: //h2 -strip: //h3 -test_url: http://technicallyjordan.tumblr.com/post/22914659822/facebook-to-launch-app-store-knock-off \ No newline at end of file diff --git a/thebrowser.com.txt b/thebrowser.com.txt deleted file mode 100755 index 807e7da..0000000 --- a/thebrowser.com.txt +++ /dev/null @@ -1,10 +0,0 @@ -title: //h2[contains(@class, 'page-title')] -body: //div[@id='content']/div[contains(@id, 'node-')]/div[@class='content'] - -prune: no - -strip: //div[contains(@class, 'node-book')]//a[@class='button'] - -single_page_link: //a[@class='tool-print'] - -test_url: http://thebrowser.com/interviews/yotam-ottolenghi-on-his-favourite-cookery-books \ No newline at end of file diff --git a/thesiasat.com.txt b/thesiasat.com.txt deleted file mode 100755 index 68a8bc8..0000000 --- a/thesiasat.com.txt +++ /dev/null @@ -1,11 +0,0 @@ -#body: (//div[@class='ftr-yt-vid'])[1] -body: (//blockquote[contains(@class, 'postcontent')])[1] -body: (//div[starts-with(@id, 'post_message')])[1] - -prune: no -tidy: no - -#replace_string():   - -test_url: http://www.thesiasat.com/showthread.php?19220-Dunya-News-HASB-E-HAAL-16-06-2012-Part-1-5 \ No newline at end of file diff --git a/westhamtillidie.com.txt b/westhamtillidie.com.txt deleted file mode 100755 index 3132e98..0000000 --- a/westhamtillidie.com.txt +++ /dev/null @@ -1,6 +0,0 @@ -title: substring-before(//title, '«') - -body: //div[@class='entry'] -strip: //div[@class='sharing_label'] -strip: //div[@class='snap_nopreview sharing robots-nocontent'] -test_url: http://www.westhamtillidie.com/2012/03/11/twelve-things-we-learned-from-the-doncaster-game/ \ No newline at end of file diff --git a/winsupersite.com.txt b/winsupersite.com.txt deleted file mode 100755 index f725b67..0000000 --- a/winsupersite.com.txt +++ /dev/null @@ -1,3 +0,0 @@ -date: //*[@class='kicker'] -body: //*[@class='KonaBody'] -test_url: http://www.winsupersite.com/article/paul-thurrotts-wininfo/android-malware-surges-separate-studies-141364 \ No newline at end of file diff --git a/xfgjls.com.txt b/xfgjls.com.txt deleted file mode 100755 index 2dc247a..0000000 --- a/xfgjls.com.txt +++ /dev/null @@ -1,11 +0,0 @@ -# This filter is tested on: -# http://www.xfgjls.com/magazine/html/?131.html -# http://www.xfgjls.com/magazine/html/?170.html - -body://h3/following-sibling::div -title: //h3 -date: substring-before(//h3/following-sibling::div/p, ' ') -author: substring-before(substring-after(//h3/following-sibling::div/p, '作者:'), '来源') -wrap_in(strong)://span[contains(@style, "FONT-WEIGHT: bold")] -dissolve://span[@style="FONT-FAMILY: '宋体'; FONT-SIZE: 10.5pt; FONT-WEIGHT: bold; mso-spacerun: 'yes'"] -test_url: http://www.xfgjls.com/magazine/html/?170.html \ No newline at end of file diff --git a/zingtrain.com.txt b/zingtrain.com.txt deleted file mode 100755 index 188d4dd..0000000 --- a/zingtrain.com.txt +++ /dev/null @@ -1,3 +0,0 @@ -title: substring-after(id, 'post')/h2 -body://div[@class = 'entry'] -test_url: http://www.zingtrain.com/category/ontrack/january-2007/ \ No newline at end of file -- cgit v1.3.1