From cbba2d911da7b4dc444be4e9a8569dcc5690ba85 Mon Sep 17 00:00:00 2001 From: Jeremy Benoist Date: Thu, 31 Dec 2015 18:13:20 +0100 Subject: Backport site_config changes from wallabag v1 Fix https://github.com/wallabag/wallabag/issues/1414 --- 944.com.txt | 9 +++++++++ aachener-zeitung.de.txt | 10 ++++++++++ adme.ru.txt | 6 ++++++ alexduner.squarespace.com.txt | 4 ++++ alt1040.com.txt | 2 ++ americascup.com.txt | 10 ++++++++++ asahi.com.txt | 3 +++ bastamag.net.txt | 10 ++++++++++ bearmetal.eu.txt | 6 ++++++ blog.jaysalvat.com.txt | 4 ++++ blog.niqnutn.com.txt | 9 +++++++++ boldizsar.palotas.eu.txt | 3 +++ bostonreview.net.txt | 15 +++++++++++++++ canonrumors.com.txt | 28 ++++++++++++++++++++++++++++ captaineconomics.fr.txt | 6 ++++++ carpeaqua.com.txt | 6 ++++++ cedarrepublican.com.txt | 2 ++ cloudacademy.com.txt | 6 ++++++ cn.reuters.com.txt | 9 +++++++++ cnnsi.com.txt | 26 ++++++++++++++++++++++++++ codebase64.org.txt | 9 +++++++++ community.service-now.com.txt | 8 ++++++++ crazybutable.com.txt | 3 +++ csnbayarea.com.txt | 7 +++++++ cyrille-borne.com.txt | 6 ++++++ danleech.com.txt | 6 ++++++ datanami.com.txt | 4 ++++ des-livres-pour-changer-de-vie.fr.txt | 6 ++++++ dn.pt.txt | 9 +++++++++ dukebasketballreport.com.txt | 11 +++++++++++ dvice.com.txt | 9 +++++++++ eamesinerudition.com.txt | 8 ++++++++ edge.org.txt | 5 +++++ espacepolitique.revues.org.txt | 10 ++++++++++ essentialpublicradio.org.txt | 6 ++++++ etc.se.txt | 6 ++++++ frankdenneman.nl.txt | 2 ++ frontburner.dmagazine.com.txt | 6 ++++++ ftd.de.txt | 5 +++++ ganglia.info.txt | 2 ++ garythink.com.txt | 3 +++ givemesomethingtoread.com.txt | 3 +++ gokulkrishh.github.io.txt | 6 ++++++ gotomanager.com.txt | 21 +++++++++++++++++++++ guardian.co.uk.txt | 7 +++++++ habrahabr.ru.txt | 21 +++++++++++++++++++++ hackersrepublic.org.txt | 6 ++++++ healthland.time.com.txt | 10 ++++++++++ heroturko.me.txt | 6 ++++++ iceblog.free.fr.txt | 7 +++++++ interest.co.nz.txt | 2 ++ iso.500px.com.txt | 6 ++++++ it-connect.fr.txt | 9 +++++++++ jsforcats.com.txt | 6 ++++++ jungle-world.com.txt | 3 +++ kachiblog.com.txt | 7 +++++++ laphamsquarterly.org.txt | 13 +++++++++++++ lapin-blanc.blogs.docteo.net.txt | 6 ++++++ lececil.org.txt | 6 ++++++ lifeandculture.fr.txt | 3 +++ lovefm.com.txt | 6 ++++++ lovetv.com.bz.txt | 9 +++++++++ m.guardian.co.uk.txt | 12 ++++++++++++ m00natic.github.io.txt | 7 +++++++ macformat.techradar.com.txt | 9 +++++++++ metrocop.net.txt | 6 ++++++ mobile.slate.com.txt | 5 +++++ moo.nac.uci.edu.txt | 9 +++++++++ myfoxboston.com.txt | 4 ++++ n0where.net.txt | 5 +++++ narenji.ir.txt | 2 ++ neomoney.co.txt | 3 +++ news.kanaloco.jp.txt | 9 +++++++++ newsle.com.txt | 2 ++ newsmill.se.txt | 12 ++++++++++++ next-gen.biz.txt | 16 ++++++++++++++++ nhk.or.jp.txt | 2 ++ observer.com.txt | 7 +++++++ on.net.mk.txt | 5 +++++ online.wsj.com.txt | 25 +++++++++++++++++++++++++ pakistantvdekho.com.txt | 11 +++++++++++ pcast.me.txt | 2 ++ physicstoday.org.txt | 7 +++++++ pittnews.com.txt | 8 ++++++++ "p\303\251rotin.com.txt" | 6 ++++++ quechoisir.org.txt | 10 ++++++++++ quepasa.cl.txt | 6 ++++++ red-hot-girls.com.txt | 5 +++++ redtape.msnbc.msn.com.txt | 20 ++++++++++++++++++++ retrieverweekly.com.txt | 6 ++++++ rezeptwelt.de.txt | 5 +++++ ritemail.blogspot.com.txt | 5 +++++ ruttloff.org.txt | 3 +++ sct.temple.edu.txt | 5 +++++ sebbo.net.txt | 4 ++++ simonwillison.net.txt | 5 +++++ sivers.org.txt | 6 ++++++ smartinvestor.de.txt | 5 +++++ smokingapples.com.txt | 5 +++++ spectator.co.uk.txt | 7 +++++++ spiil.org.txt | 6 ++++++ sportschau.de.txt | 22 ++++++++++++++++++++++ sportsillustrated.cnn.com.txt | 26 ++++++++++++++++++++++++++ stockholm.etc.se.txt | 5 +++++ stockholmsfria.nu.txt | 7 +++++++ tcmanila.tk.txt | 7 +++++++ tcng.org.txt | 4 ++++ technicallyjordan.tumblr.com.txt | 8 ++++++++ theappleblog.com.txt | 3 +++ thebrowser.com.txt | 10 ++++++++++ thecarton.net.txt | 10 ++++++++++ thedaily.com.txt | 24 ++++++++++++++++++++++++ theeuropean-magazine.com.txt | 17 +++++++++++++++++ thesiasat.com.txt | 11 +++++++++++ thisismynext.com.txt | 8 ++++++++ tofugu.com.txt | 3 +++ toolinux.com.txt | 5 +++++ tthor.com.txt | 2 ++ tuckreview.com.txt | 6 ++++++ tuhdo.github.io.txt | 7 +++++++ uzivatelsketestovani.cz.txt | 2 ++ varingen.no.txt | 5 +++++ vemedio.com.txt | 6 ++++++ walrusmagazine.com.txt | 14 ++++++++++++++ weblog.bignerdranch.com.txt | 5 +++++ webpaper.nzz.ch.txt | 8 ++++++++ westhamtillidie.com.txt | 6 ++++++ wheelyric.com.txt | 11 +++++++++++ winsupersite.com.txt | 3 +++ wmnf.org.txt | 13 +++++++++++++ wordyard.com.txt | 8 ++++++++ wyctim.com.txt | 3 +++ xfgjls.com.txt | 11 +++++++++++ yosko.net.txt | 6 ++++++ zcommunications.org.txt | 7 +++++++ zerodistraction.com.txt | 4 ++++ zingtrain.com.txt | 3 +++ 137 files changed, 1045 insertions(+) create mode 100755 944.com.txt create mode 100755 aachener-zeitung.de.txt create mode 100644 adme.ru.txt create mode 100755 alexduner.squarespace.com.txt create mode 100755 alt1040.com.txt create mode 100755 americascup.com.txt create mode 100755 asahi.com.txt create mode 100644 bastamag.net.txt create mode 100644 bearmetal.eu.txt create mode 100755 blog.jaysalvat.com.txt create mode 100644 blog.niqnutn.com.txt create mode 100755 boldizsar.palotas.eu.txt create mode 100755 bostonreview.net.txt create mode 100755 canonrumors.com.txt create mode 100644 captaineconomics.fr.txt create mode 100755 carpeaqua.com.txt create mode 100755 cedarrepublican.com.txt create mode 100644 cloudacademy.com.txt create mode 100755 cn.reuters.com.txt create mode 100755 cnnsi.com.txt create mode 100644 codebase64.org.txt create mode 100755 community.service-now.com.txt create mode 100755 crazybutable.com.txt create mode 100755 csnbayarea.com.txt create mode 100644 cyrille-borne.com.txt create mode 100755 danleech.com.txt create mode 100755 datanami.com.txt create mode 100644 des-livres-pour-changer-de-vie.fr.txt create mode 100755 dn.pt.txt create mode 100755 dukebasketballreport.com.txt create mode 100755 dvice.com.txt create mode 100755 eamesinerudition.com.txt create mode 100755 edge.org.txt create mode 100644 espacepolitique.revues.org.txt create mode 100755 essentialpublicradio.org.txt create mode 100755 etc.se.txt create mode 100644 frankdenneman.nl.txt create mode 100644 frontburner.dmagazine.com.txt create mode 100755 ftd.de.txt create mode 100644 ganglia.info.txt create mode 100755 garythink.com.txt create mode 100755 givemesomethingtoread.com.txt create mode 100644 gokulkrishh.github.io.txt create mode 100755 gotomanager.com.txt create mode 100644 guardian.co.uk.txt create mode 100755 habrahabr.ru.txt create mode 100644 hackersrepublic.org.txt create mode 100644 healthland.time.com.txt create mode 100755 heroturko.me.txt create mode 100644 iceblog.free.fr.txt create mode 100755 interest.co.nz.txt create mode 100644 iso.500px.com.txt create mode 100644 it-connect.fr.txt create mode 100644 jsforcats.com.txt create mode 100644 jungle-world.com.txt create mode 100755 kachiblog.com.txt create mode 100755 laphamsquarterly.org.txt create mode 100644 lapin-blanc.blogs.docteo.net.txt create mode 100644 lececil.org.txt create mode 100755 lifeandculture.fr.txt create mode 100755 lovefm.com.txt create mode 100755 lovetv.com.bz.txt create mode 100644 m.guardian.co.uk.txt create mode 100644 m00natic.github.io.txt create mode 100755 macformat.techradar.com.txt create mode 100644 metrocop.net.txt create mode 100755 mobile.slate.com.txt create mode 100644 moo.nac.uci.edu.txt create mode 100755 myfoxboston.com.txt create mode 100644 n0where.net.txt create mode 100755 narenji.ir.txt create mode 100755 neomoney.co.txt create mode 100755 news.kanaloco.jp.txt create mode 100755 newsle.com.txt create mode 100755 newsmill.se.txt create mode 100755 next-gen.biz.txt create mode 100755 nhk.or.jp.txt create mode 100755 observer.com.txt create mode 100755 on.net.mk.txt create mode 100755 online.wsj.com.txt create mode 100755 pakistantvdekho.com.txt create mode 100755 pcast.me.txt create mode 100755 physicstoday.org.txt create mode 100755 pittnews.com.txt create mode 100644 "p\303\251rotin.com.txt" create mode 100644 quechoisir.org.txt create mode 100755 quepasa.cl.txt create mode 100755 red-hot-girls.com.txt create mode 100755 redtape.msnbc.msn.com.txt create mode 100755 retrieverweekly.com.txt create mode 100644 rezeptwelt.de.txt create mode 100755 ritemail.blogspot.com.txt create mode 100755 ruttloff.org.txt create mode 100755 sct.temple.edu.txt create mode 100755 sebbo.net.txt create mode 100755 simonwillison.net.txt create mode 100755 sivers.org.txt create mode 100755 smartinvestor.de.txt create mode 100755 smokingapples.com.txt create mode 100755 spectator.co.uk.txt create mode 100644 spiil.org.txt create mode 100755 sportschau.de.txt create mode 100755 sportsillustrated.cnn.com.txt create mode 100755 stockholm.etc.se.txt create mode 100755 stockholmsfria.nu.txt create mode 100755 tcmanila.tk.txt create mode 100755 tcng.org.txt create mode 100755 technicallyjordan.tumblr.com.txt create mode 100755 theappleblog.com.txt create mode 100755 thebrowser.com.txt create mode 100755 thecarton.net.txt create mode 100755 thedaily.com.txt create mode 100755 theeuropean-magazine.com.txt create mode 100755 thesiasat.com.txt create mode 100755 thisismynext.com.txt create mode 100644 tofugu.com.txt create mode 100755 toolinux.com.txt create mode 100755 tthor.com.txt create mode 100755 tuckreview.com.txt create mode 100644 tuhdo.github.io.txt create mode 100755 uzivatelsketestovani.cz.txt create mode 100755 varingen.no.txt create mode 100755 vemedio.com.txt create mode 100755 walrusmagazine.com.txt create mode 100755 weblog.bignerdranch.com.txt create mode 100755 webpaper.nzz.ch.txt create mode 100755 westhamtillidie.com.txt create mode 100755 wheelyric.com.txt create mode 100755 winsupersite.com.txt create mode 100755 wmnf.org.txt create mode 100644 wordyard.com.txt create mode 100755 wyctim.com.txt create mode 100755 xfgjls.com.txt create mode 100644 yosko.net.txt create mode 100755 zcommunications.org.txt create mode 100644 zerodistraction.com.txt create mode 100755 zingtrain.com.txt diff --git a/944.com.txt b/944.com.txt new file mode 100755 index 0000000..8bf6a4c --- /dev/null +++ b/944.com.txt @@ -0,0 +1,9 @@ +title: //h2[@class='border'] +body: //div[@class='padding'] + +convert_double_br_tags: yes + +strip: //div[@id='social_sharing'] +strip: //div[@class='socialLinks'] + +test_url: http://www.944.com/articles/mild-obsessions-frock-la-get-to-know-victoria-tik-s-haute-sustainable-fashion-line/ \ No newline at end of file diff --git a/aachener-zeitung.de.txt b/aachener-zeitung.de.txt new file mode 100755 index 0000000..013afa4 --- /dev/null +++ b/aachener-zeitung.de.txt @@ -0,0 +1,10 @@ +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/adme.ru.txt b/adme.ru.txt new file mode 100644 index 0000000..b929685 --- /dev/null +++ b/adme.ru.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fwww.adme.ru%2Ftvorchestvo-hudozhniki%2Fprostoj-kak-5-kopeek-hudozhnik-557405%2F + +body: //article[contains(concat(' ',normalize-space(@class),' '),' article ')] +test_url: http://www.adme.ru/tvorchestvo-hudozhniki/prostoj-kak-5-kopeek-hudozhnik-557405/ diff --git a/alexduner.squarespace.com.txt b/alexduner.squarespace.com.txt new file mode 100755 index 0000000..875405e --- /dev/null +++ b/alexduner.squarespace.com.txt @@ -0,0 +1,4 @@ +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/alt1040.com.txt b/alt1040.com.txt new file mode 100755 index 0000000..4fd4571 --- /dev/null +++ b/alt1040.com.txt @@ -0,0 +1,2 @@ +body: //*[(@class = "historia")] +test_url: http://alt1040.com/2011/09/banda-ancha-en-america-latina-insignificante \ No newline at end of file diff --git a/americascup.com.txt b/americascup.com.txt new file mode 100755 index 0000000..31723f8 --- /dev/null +++ b/americascup.com.txt @@ -0,0 +1,10 @@ +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/asahi.com.txt b/asahi.com.txt new file mode 100755 index 0000000..b4eec7b --- /dev/null +++ b/asahi.com.txt @@ -0,0 +1,3 @@ +body: //div[@id='HeadLine'] +strip: //div[@id='utility_right'] +test_url: http://www.asahi.com/culture/update/0520/TKY201105200321.html \ No newline at end of file diff --git a/bastamag.net.txt b/bastamag.net.txt new file mode 100644 index 0000000..142c959 --- /dev/null +++ b/bastamag.net.txt @@ -0,0 +1,10 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fwww.bastamag.net%2FEncadrement-des-loyers-pourquoi-Anne-Hidalgo-ferait-bien-de-visiter-Vienne-l + +body: //div[@id='content']//article + +strip_id_or_class: appel-soutien +strip_id_or_class: liste dossiers-content + +test_url: http://www.bastamag.net/Encadrement-des-loyers-pourquoi-Anne-Hidalgo-ferait-bien-de-visiter-Vienne-l diff --git a/bearmetal.eu.txt b/bearmetal.eu.txt new file mode 100644 index 0000000..44d3465 --- /dev/null +++ b/bearmetal.eu.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=https%3A%2F%2Fbearmetal.eu%2Ftheden%2Fits-not-about-you%2F + +body: //div[contains(concat(' ',normalize-space(@class),' '),' entry-content ')] +test_url: https://bearmetal.eu/theden/its-not-about-you/ diff --git a/blog.jaysalvat.com.txt b/blog.jaysalvat.com.txt new file mode 100755 index 0000000..e89ad3a --- /dev/null +++ b/blog.jaysalvat.com.txt @@ -0,0 +1,4 @@ +date: //span[contains(@class, 'date-links')] +author: //span[contains(@class, 'author-links')] +body: //div[contains(@class, 'entry-content')] +test_url: http://blog.jaysalvat.com/article/celui-qui-avait-refait-son-site-web \ No newline at end of file diff --git a/blog.niqnutn.com.txt b/blog.niqnutn.com.txt new file mode 100644 index 0000000..aa2f186 --- /dev/null +++ b/blog.niqnutn.com.txt @@ -0,0 +1,9 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fblog.niqnutn.com%2Farticle35%2Fsteganographie-l-art-de-dissimuler-un-message + +body: //article[contains(concat(' ',normalize-space(@class),' '),' article-content ')]//section + +strip_id_or_class: sommaire + +test_url: http://blog.niqnutn.com/article35/steganographie-l-art-de-dissimuler-un-message diff --git a/boldizsar.palotas.eu.txt b/boldizsar.palotas.eu.txt new file mode 100755 index 0000000..3616bbf --- /dev/null +++ b/boldizsar.palotas.eu.txt @@ -0,0 +1,3 @@ +title: //h2[@class='entry-title'] +body: //div[@class='entry-content'] +test_url: http://boldizsar.palotas.eu/blog/?p=1394 \ No newline at end of file diff --git a/bostonreview.net.txt b/bostonreview.net.txt new file mode 100755 index 0000000..64e04a1 --- /dev/null +++ b/bostonreview.net.txt @@ -0,0 +1,15 @@ +#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/canonrumors.com.txt b/canonrumors.com.txt new file mode 100755 index 0000000..c22cf4f --- /dev/null +++ b/canonrumors.com.txt @@ -0,0 +1,28 @@ +# Author: zinnober + +tidy: no +prune: no + +# Set title +title: //h2 + +date: //li[@class='time'] + +# Set author +author: //a[contains(@rel, 'author')] + +# Content is here +body: //div[@id='content'] + +# Tidy up before article +strip: //div[@class='meta'] + +# Tidy up after article +strip_id_or_class: nr_related_placeholder +strip_id_or_class: twitter-share-button +strip_id_or_class: afterpost +strip_id_or_class: tags + +# Try it yourself +test_url: http://www.canonrumors.com/2014/09/chuck-westfall-talks-canon-eos-7d-mark-ii/ +test_url: http://www.canonrumors.com/2014/09/canon-cinema-eos-captures-space-in-4k-for-new-imax-3d-film/ diff --git a/captaineconomics.fr.txt b/captaineconomics.fr.txt new file mode 100644 index 0000000..036e028 --- /dev/null +++ b/captaineconomics.fr.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fwww.captaineconomics.fr%2F-les-immigres-recemment-arrives-en-france-qui-sont-ils-insee + +body: //div[contains(concat(' ',normalize-space(@class),' '),' post_content ')] +test_url: http://www.captaineconomics.fr/-les-immigres-recemment-arrives-en-france-qui-sont-ils-insee diff --git a/carpeaqua.com.txt b/carpeaqua.com.txt new file mode 100755 index 0000000..5ea302e --- /dev/null +++ b/carpeaqua.com.txt @@ -0,0 +1,6 @@ +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 new file mode 100755 index 0000000..42faa52 --- /dev/null +++ b/cedarrepublican.com.txt @@ -0,0 +1,2 @@ +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/cloudacademy.com.txt b/cloudacademy.com.txt new file mode 100644 index 0000000..f5b6784 --- /dev/null +++ b/cloudacademy.com.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fcloudacademy.com%2Fblog%2Fberkshelf-manage-chef-cookbooks%2F + +body: //div[contains(concat(' ',normalize-space(@class),' '),' entry-content ')] +test_url: http://cloudacademy.com/blog/berkshelf-manage-chef-cookbooks/ \ No newline at end of file diff --git a/cn.reuters.com.txt b/cn.reuters.com.txt new file mode 100755 index 0000000..28f1047 --- /dev/null +++ b/cn.reuters.com.txt @@ -0,0 +1,9 @@ +title: //div[@id='maincontent']//h1 +body: //div[@id='resizeableText'] + +single_page_link: concat(//link[@rel='canonical']/@href, '?sp=true') + +test_url: http://cn.reuters.com/article/CNAnalysesNews/idCNKBS0FF0NM20140710 +test_url: http://cn.reuters.feedsportal.com/CNAnalysesNews +# multipage link +test_url: http://cn.reuters.com/article/idCNKBS0FF0UL20140710 \ No newline at end of file diff --git a/cnnsi.com.txt b/cnnsi.com.txt new file mode 100755 index 0000000..ac49aef --- /dev/null +++ b/cnnsi.com.txt @@ -0,0 +1,26 @@ +# 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://cnnsi.com/2012/writers/peter_king/01/08/wild.card.round/index.html \ No newline at end of file diff --git a/codebase64.org.txt b/codebase64.org.txt new file mode 100644 index 0000000..d992d2f --- /dev/null +++ b/codebase64.org.txt @@ -0,0 +1,9 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fcodebase64.org%2Fdoku.php%3Fid%3Dbase%3Ademo_coding_introduction + +body: //div[contains(concat(' ',normalize-space(@class),' '),' page ')] + +strip_id_or_class: dw__toc + +test_url: http://codebase64.org/doku.php?id=base:demo_coding_introduction diff --git a/community.service-now.com.txt b/community.service-now.com.txt new file mode 100755 index 0000000..c9854b4 --- /dev/null +++ b/community.service-now.com.txt @@ -0,0 +1,8 @@ +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.service-now.com/blog/lawrenceeng/seasons-greetings-servicenow-team \ No newline at end of file diff --git a/crazybutable.com.txt b/crazybutable.com.txt new file mode 100755 index 0000000..037cd17 --- /dev/null +++ b/crazybutable.com.txt @@ -0,0 +1,3 @@ +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/csnbayarea.com.txt b/csnbayarea.com.txt new file mode 100755 index 0000000..1da60b4 --- /dev/null +++ b/csnbayarea.com.txt @@ -0,0 +1,7 @@ +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/cyrille-borne.com.txt b/cyrille-borne.com.txt new file mode 100644 index 0000000..7b41aa0 --- /dev/null +++ b/cyrille-borne.com.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fwww.cyrille-borne.com%2Farticle511%2Fun-an-de-zte-open-c-plus-tard + +body: //article//section +test_url: http://www.cyrille-borne.com/article511/un-an-de-zte-open-c-plus-tard \ No newline at end of file diff --git a/danleech.com.txt b/danleech.com.txt new file mode 100755 index 0000000..1d4cec7 --- /dev/null +++ b/danleech.com.txt @@ -0,0 +1,6 @@ +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 new file mode 100755 index 0000000..e9111a4 --- /dev/null +++ b/datanami.com.txt @@ -0,0 +1,4 @@ +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/des-livres-pour-changer-de-vie.fr.txt b/des-livres-pour-changer-de-vie.fr.txt new file mode 100644 index 0000000..56b0ea2 --- /dev/null +++ b/des-livres-pour-changer-de-vie.fr.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fwww.des-livres-pour-changer-de-vie.fr%2Fbit-literacy-2%2F + +body: //div[@id='getsocialmain'] +test_url: http://www.des-livres-pour-changer-de-vie.fr/bit-literacy-2/ \ No newline at end of file diff --git a/dn.pt.txt b/dn.pt.txt new file mode 100755 index 0000000..051b8cb --- /dev/null +++ b/dn.pt.txt @@ -0,0 +1,9 @@ +single_page_link: concat('http://www.dn.pt/Common/print.aspx?content_id=', //input[@type='hidden' and @name='link-comments']/@value) +# + +title: //h1 +author: //div[@class="Author"] + +strip: //div[@class="Patrocinio"] + +test_url: http://www.dn.pt/inicio/opiniao/interior.aspx?content_id=3972244&seccao=Alberto%20Gon%E7alves&tag=Opini%E3o%20-%20Em%20Foco&page=1 \ No newline at end of file diff --git a/dukebasketballreport.com.txt b/dukebasketballreport.com.txt new file mode 100755 index 0000000..2978797 --- /dev/null +++ b/dukebasketballreport.com.txt @@ -0,0 +1,11 @@ +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 new file mode 100755 index 0000000..1a1990e --- /dev/null +++ b/dvice.com.txt @@ -0,0 +1,9 @@ +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/eamesinerudition.com.txt b/eamesinerudition.com.txt new file mode 100755 index 0000000..89a68bc --- /dev/null +++ b/eamesinerudition.com.txt @@ -0,0 +1,8 @@ +title: //div [@class="post contain"]/h1 +strip: //div [@class="post contain"]/h1 +body: //div [@class="post contain"] +author: substring-before(//title, ':') +author: substring-before(//title, ' ') + + +test_url: http://eamesinerudition.com/2012/03/hospital-numbers-are-bad-for-you \ No newline at end of file diff --git a/edge.org.txt b/edge.org.txt new file mode 100755 index 0000000..95805f6 --- /dev/null +++ b/edge.org.txt @@ -0,0 +1,5 @@ +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/espacepolitique.revues.org.txt b/espacepolitique.revues.org.txt new file mode 100644 index 0000000..c4acb26 --- /dev/null +++ b/espacepolitique.revues.org.txt @@ -0,0 +1,10 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fespacepolitique.revues.org%2F284 + +body: //div[@id='text'] + +strip_id_or_class: paranumber +strip_id_or_class: go-top + +test_url: http://espacepolitique.revues.org/284 diff --git a/essentialpublicradio.org.txt b/essentialpublicradio.org.txt new file mode 100755 index 0000000..9a92239 --- /dev/null +++ b/essentialpublicradio.org.txt @@ -0,0 +1,6 @@ +title: //*[@itemprop='headline'] +author: //*[@itemprop='author'] +date: //*[@itemprop='datePublished'] +body: //*[@itemprop='articleBody'] +strip: //*[contains(@class, 'instapaper_ignore')] +test_url: http://www.essentialpublicradio.org/story/2011-11-14/volunteers-sought-federal-tax-assistance-program-pennsylvania-9421 \ No newline at end of file diff --git a/etc.se.txt b/etc.se.txt new file mode 100755 index 0000000..95f8cf7 --- /dev/null +++ b/etc.se.txt @@ -0,0 +1,6 @@ +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/frankdenneman.nl.txt b/frankdenneman.nl.txt new file mode 100644 index 0000000..c601100 --- /dev/null +++ b/frankdenneman.nl.txt @@ -0,0 +1,2 @@ +body: //*[(@class = "content")] +test_url: http://frankdenneman.nl/ diff --git a/frontburner.dmagazine.com.txt b/frontburner.dmagazine.com.txt new file mode 100644 index 0000000..73f4432 --- /dev/null +++ b/frontburner.dmagazine.com.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Ffrontburner.dmagazine.com%2F2015%2F04%2F23%2Fthe-barrett-brown-review-of-arts-and-letters-and-jail-a-sign-of-things-to-come%2F + +body: //div[@id='article-body'] +test_url: http://frontburner.dmagazine.com/2015/04/23/the-barrett-brown-review-of-arts-and-letters-and-jail-a-sign-of-things-to-come/ diff --git a/ftd.de.txt b/ftd.de.txt new file mode 100755 index 0000000..7d76af0 --- /dev/null +++ b/ftd.de.txt @@ -0,0 +1,5 @@ +body: //div[@class='boxIntroHead']/span[@class='h3'] | //div[@class='section']/div[@class='paragraph' or @class='embObjLeft'] +single_page_link: //a[@class='icon print'] + +test_url: http://www.ftd.de/it-medien/it-telekommunikation/:mobilfunk-vivendi-und-vodafone-trennen-sich-in-frankreich/60034691.html +test_url: http://www.ftd.de/it-medien/medien-internet/:verkauf-von-warner-music-musikbranche-auf-dem-sprung/60048185.html \ No newline at end of file diff --git a/ganglia.info.txt b/ganglia.info.txt new file mode 100644 index 0000000..2ae27b4 --- /dev/null +++ b/ganglia.info.txt @@ -0,0 +1,2 @@ +body: //*[(@id = "mid")] +test_url: http://ganglia.info/ diff --git a/garythink.com.txt b/garythink.com.txt new file mode 100755 index 0000000..327ac55 --- /dev/null +++ b/garythink.com.txt @@ -0,0 +1,3 @@ +tidy: no + +test_url: http://www.garythink.com/eft/testing.html \ No newline at end of file diff --git a/givemesomethingtoread.com.txt b/givemesomethingtoread.com.txt new file mode 100755 index 0000000..0de0750 --- /dev/null +++ b/givemesomethingtoread.com.txt @@ -0,0 +1,3 @@ +single_page_link: //div[@id="content"]//h2/a + +test_url: http://givemesomethingtoread.com/post/6285838917/the-baddest-lawyer-in-the-history-of-jersey \ No newline at end of file diff --git a/gokulkrishh.github.io.txt b/gokulkrishh.github.io.txt new file mode 100644 index 0000000..b395fa9 --- /dev/null +++ b/gokulkrishh.github.io.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=https%3A%2F%2Fgokulkrishh.github.io%2FJavascript-Guidelines%2F + +body: //section[contains(concat(' ',normalize-space(@class),' '),' main-content ')] +test_url: https://gokulkrishh.github.io/Javascript-Guidelines/ diff --git a/gotomanager.com.txt b/gotomanager.com.txt new file mode 100755 index 0000000..f8af732 --- /dev/null +++ b/gotomanager.com.txt @@ -0,0 +1,21 @@ +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/guardian.co.uk.txt b/guardian.co.uk.txt new file mode 100644 index 0000000..299d71c --- /dev/null +++ b/guardian.co.uk.txt @@ -0,0 +1,7 @@ +title: //div[@id='main-article-info']//h1 +body: //div[@id='article-wrapper'] +date: //li[@class='publication']//time[@pubdate] | //li[@class='publication']//data[@pubdate] +author: //li[@class='byline'] +prune: no +tidy: no +test_url: http://www.guardian.co.uk/business/2011/oct/06/quantitative-easing-75bn-bank-of-england \ No newline at end of file diff --git a/habrahabr.ru.txt b/habrahabr.ru.txt new file mode 100755 index 0000000..6753835 --- /dev/null +++ b/habrahabr.ru.txt @@ -0,0 +1,21 @@ +title: //span[@class="post_title"] +author: //div[@class="author"] +date: //div[@class="published + +body: //div[@class='content html_format'] | //div[@id='comments'] + +strip: //a[@class="link_to_comment"] +strip: //div[@class="show_tree"] +strip: //a[@class="to_parent"] + + +replace_string(class="reply_comments"): style="padding-left: 20px" +replace_string(class="voting "): style="float: right" +replace_string(src="//habrastorage.org/getpro/habr/avatars/): style="width:24px; height:24px;" class="123" src="//habrastorage.org/getpro/habr/avatars/ +replace_string(class="info "): style="padding-top:5px;font-size:0.85em;line-height:24px;" + + +prune: no +tidy: no + +test_url: http://habrahabr.ru/post/229883/ \ No newline at end of file diff --git a/hackersrepublic.org.txt b/hackersrepublic.org.txt new file mode 100644 index 0000000..da01c21 --- /dev/null +++ b/hackersrepublic.org.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fwww.hackersrepublic.org%2Fculture-du-hacking%2Fune-histoire-de-privileges + +body: //div[contains(concat(' ',normalize-space(@class),' '),' field-name-body ')]//div[contains(concat(' ',normalize-space(@class),' '),' field-items ')]//div[contains(concat(' ',normalize-space(@class),' '),' field-item ') and (contains(concat(' ',normalize-space(@class),' '),' even '))] +test_url: http://www.hackersrepublic.org/culture-du-hacking/une-histoire-de-privileges diff --git a/healthland.time.com.txt b/healthland.time.com.txt new file mode 100644 index 0000000..0542097 --- /dev/null +++ b/healthland.time.com.txt @@ -0,0 +1,10 @@ +date: //span[@class = 'date'] +body: //div[@class = 'entry-content'] +strip://div[@class='more-ways'] +strip://div[@id = 'stayConnected'] +strip://p[child::a[@rel = 'bookmark']] +strip://p[starts-with(string(.),'(MORE:')] +strip://p[starts-with(string(.),'(PHOTOS:')] +move_into(//p[../@class = 'entry-content'][position() = last()])://div[@id = 'featbox'] + +test_url: http://healthland.time.com/2011/07/24/amy-winehouse-and-the-pain-of-addiction/?preview=true&preview_id=39210&preview_nonce=0777d4e408 \ No newline at end of file diff --git a/heroturko.me.txt b/heroturko.me.txt new file mode 100755 index 0000000..07b6adf --- /dev/null +++ b/heroturko.me.txt @@ -0,0 +1,6 @@ +title: //div[contains(@class, 'title')]//h1 +body: //div[contains(@class, 'story')] + +prune: no + +test_url: http://www.heroturko.me/5223034-ds-catia-p3-v5-6r2014-gasp0-x86x64-multilanguage-english-docs.html \ No newline at end of file diff --git a/iceblog.free.fr.txt b/iceblog.free.fr.txt new file mode 100644 index 0000000..9d1ce21 --- /dev/null +++ b/iceblog.free.fr.txt @@ -0,0 +1,7 @@ + +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Ficeblog.free.fr%2Findex.php%3Farticle102%2Fweb-histoire-geek-4 + +body: //article//section +test_url: http://iceblog.free.fr/index.php?article102/web-histoire-geek-4 diff --git a/interest.co.nz.txt b/interest.co.nz.txt new file mode 100755 index 0000000..28c3310 --- /dev/null +++ b/interest.co.nz.txt @@ -0,0 +1,2 @@ +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/iso.500px.com.txt b/iso.500px.com.txt new file mode 100644 index 0000000..ffa29a1 --- /dev/null +++ b/iso.500px.com.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=https%3A%2F%2Fiso.500px.com%2Fhow-i-post-processed-the-most-successful-image-of-my-career%2F + +body: //div[contains(concat(' ',normalize-space(@class),' '),' post-content ')] +test_url: https://iso.500px.com/how-i-post-processed-the-most-successful-image-of-my-career/ diff --git a/it-connect.fr.txt b/it-connect.fr.txt new file mode 100644 index 0000000..80c95e6 --- /dev/null +++ b/it-connect.fr.txt @@ -0,0 +1,9 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fwww.it-connect.fr%2Fdebuter-avec-docker-et-les-containers-sous-debian-8%2F + +body: //div[contains(concat(' ',normalize-space(@class),' '),' post_inner_wrapper ')] + +strip_id_or_class: toc_container + +test_url: http://www.it-connect.fr/debuter-avec-docker-et-les-containers-sous-debian-8/ diff --git a/jsforcats.com.txt b/jsforcats.com.txt new file mode 100644 index 0000000..f2699ef --- /dev/null +++ b/jsforcats.com.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fjsforcats.com%2F + +body: //div[contains(concat(' ',normalize-space(@class),' '),' post-content ')] +test_url: http://jsforcats.com/ diff --git a/jungle-world.com.txt b/jungle-world.com.txt new file mode 100644 index 0000000..61e0087 --- /dev/null +++ b/jungle-world.com.txt @@ -0,0 +1,3 @@ +title: //h1 +body: //div[contains(@class,'story')] +test_url: http://jungle-world.com/artikel/2015/02/51207.html diff --git a/kachiblog.com.txt b/kachiblog.com.txt new file mode 100755 index 0000000..57ab0de --- /dev/null +++ b/kachiblog.com.txt @@ -0,0 +1,7 @@ +title: //h3[contains(@class, 'entry-title')] +date: //abbr[@itemprop='datePublished']/@title +body: //div[@itemprop='articleBody'] +tidy: no + +test_url: http://www.kachiblog.com/2013/05/samsung-galaxy-s4-vs-samsung-galaxy.html +test_url: http://www.kachiblog.com/feed diff --git a/laphamsquarterly.org.txt b/laphamsquarterly.org.txt new file mode 100755 index 0000000..d25999d --- /dev/null +++ b/laphamsquarterly.org.txt @@ -0,0 +1,13 @@ +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/lapin-blanc.blogs.docteo.net.txt b/lapin-blanc.blogs.docteo.net.txt new file mode 100644 index 0000000..38835a9 --- /dev/null +++ b/lapin-blanc.blogs.docteo.net.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Flapin-blanc.blogs.docteo.net%2F2015%2F01%2F20%2Flencadrement-doctoral-deviendrait-il-un-sujet-dactualite%2F + +body: //div[contains(concat(' ',normalize-space(@class),' '),' entry-content ')] +test_url: http://lapin-blanc.blogs.docteo.net/2015/01/20/lencadrement-doctoral-deviendrait-il-un-sujet-dactualite/ diff --git a/lececil.org.txt b/lececil.org.txt new file mode 100644 index 0000000..8a38bbf --- /dev/null +++ b/lececil.org.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fwww.lececil.org%2Fnode%2F7688 + +body: //div[contains(concat(' ',normalize-space(@class),' '),' field-body ')] +test_url: http://www.lececil.org/node/7688 diff --git a/lifeandculture.fr.txt b/lifeandculture.fr.txt new file mode 100755 index 0000000..0e1dceb --- /dev/null +++ b/lifeandculture.fr.txt @@ -0,0 +1,3 @@ +title: //h2[@class="entry-title"] +body: //div[@class="entry-content"] +test_url: http://www.lifeandculture.fr/digital/facebook-and-the-epiphanator-an-end-to-endings/ \ No newline at end of file diff --git a/lovefm.com.txt b/lovefm.com.txt new file mode 100755 index 0000000..20d26c5 --- /dev/null +++ b/lovefm.com.txt @@ -0,0 +1,6 @@ +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 new file mode 100755 index 0000000..a71fccd --- /dev/null +++ b/lovetv.com.bz.txt @@ -0,0 +1,9 @@ +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/m.guardian.co.uk.txt b/m.guardian.co.uk.txt new file mode 100644 index 0000000..cce0271 --- /dev/null +++ b/m.guardian.co.uk.txt @@ -0,0 +1,12 @@ +title: //p[@class="txhead"] +author: //div[@class='txb'] +wrap_in(p): //div[@class='para'] +date: //div[@class='txb']/following-sibling::p/text()[substring(., 14)] +strip: //table[@class="tlogo"] +strip: //div[@class="cookieText"] +strip: //*[@class="sltb"] +strip: //*[@class="ijobs-x-link"] +strip: //*[@class="sponscolour"] +strip: //*[@class="sponsouter"] +strip: //div[@id="bottom-nav-block"]/following::* +test_url: http://m.guardian.co.uk/ms/p/gnm/op/s3OOwgO3yIhGuj41C1_S3Xg/view.m?id=15&gid=world/2012/jul/26/arctic-climate-change&cat=top-stories \ No newline at end of file diff --git a/m00natic.github.io.txt b/m00natic.github.io.txt new file mode 100644 index 0000000..911fcbd --- /dev/null +++ b/m00natic.github.io.txt @@ -0,0 +1,7 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=https%3A%2F%2Fm00natic.github.io%2Femacs%2Femacs-wiki.html + +body: //div[@id='content'] +strip_id_or_class: table-of-contents +test_url: https://m00natic.github.io/emacs/emacs-wiki.html \ No newline at end of file diff --git a/macformat.techradar.com.txt b/macformat.techradar.com.txt new file mode 100755 index 0000000..522efb4 --- /dev/null +++ b/macformat.techradar.com.txt @@ -0,0 +1,9 @@ +# Remove news feed +strip: //div[@id='news_feed_front'] + +# Remove pull quote +strip: //div[@class='field field-type-text field-field-pull-quote'] + +# Remove login +strip: //div[@class='right_bar_login'] +test_url: http://macformat.techradar.com/blog/solid-state-storage-bringing-parity-back-mac-29-10-10&article=89189666 \ No newline at end of file diff --git a/metrocop.net.txt b/metrocop.net.txt new file mode 100644 index 0000000..9535f12 --- /dev/null +++ b/metrocop.net.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fmetrocop.net%2Fmythic-comics%2Fissue-1%2F + +body: //div[contains(concat(' ',normalize-space(@class),' '),' entry ')] +test_url: http://metrocop.net/mythic-comics/issue-1/ diff --git a/mobile.slate.com.txt b/mobile.slate.com.txt new file mode 100755 index 0000000..6ffcd18 --- /dev/null +++ b/mobile.slate.com.txt @@ -0,0 +1,5 @@ +title: //h2[@class="article_title"] +strip: //a[@class="houseAdLink"] +strip: //h1 +strip: //div[@class="more_articles"] +test_url: http://mobile.slate.com/rss.jsp?rssid=411&item=http%3a%2f%2fwww.slate.com%2fdefault.aspx%3fdisplaymode%3d201%26id%3d2293749%26device%3drss \ No newline at end of file diff --git a/moo.nac.uci.edu.txt b/moo.nac.uci.edu.txt new file mode 100644 index 0000000..24c949e --- /dev/null +++ b/moo.nac.uci.edu.txt @@ -0,0 +1,9 @@ +title: //div[@id='header']//h1[1] + +body: //div[@id='content'] + +strip_id_or_class: toc + +prune: no + +test_url: http://moo.nac.uci.edu/~hjm/HOWTO_move_data.html diff --git a/myfoxboston.com.txt b/myfoxboston.com.txt new file mode 100755 index 0000000..9ad8ce0 --- /dev/null +++ b/myfoxboston.com.txt @@ -0,0 +1,4 @@ +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/n0where.net.txt b/n0where.net.txt new file mode 100644 index 0000000..e91159e --- /dev/null +++ b/n0where.net.txt @@ -0,0 +1,5 @@ +title://div[@class='main-title single-title entry-title'] +body://div[@id="content-anchor-inner"] +date://meta[@content] + +strip: //noscript diff --git a/narenji.ir.txt b/narenji.ir.txt new file mode 100755 index 0000000..6c3d0c2 --- /dev/null +++ b/narenji.ir.txt @@ -0,0 +1,2 @@ +body: //div[@class='node'] +test_url: http://www.narenji.ir/2806 \ No newline at end of file diff --git a/neomoney.co.txt b/neomoney.co.txt new file mode 100755 index 0000000..2089fc3 --- /dev/null +++ b/neomoney.co.txt @@ -0,0 +1,3 @@ +title: //*[@class="header_title"]/h1 +body: //div[contains(@class, 'content')] +test_url: http://neomoney.co/personal/expatriate-and-migrant-loans/expatriate-loans/ \ No newline at end of file diff --git a/news.kanaloco.jp.txt b/news.kanaloco.jp.txt new file mode 100755 index 0000000..5754d47 --- /dev/null +++ b/news.kanaloco.jp.txt @@ -0,0 +1,9 @@ +body: //div[@id='main'] +strip: //div[@id='sbs'] +strip: //div[@id='fsizeSwitch'] +strip: //div[@id='googleAd'] +strip: //div[@id='detailFoot'] +strip_image_src: counter?key +convert_double_br_tags: yes + +test_url: http://news.kanaloco.jp/localnews/article/1105200018/ \ No newline at end of file diff --git a/newsle.com.txt b/newsle.com.txt new file mode 100755 index 0000000..e500ddc --- /dev/null +++ b/newsle.com.txt @@ -0,0 +1,2 @@ +single_page_link: //iframe/@src +test_url: http://newsle.com/article/0/15831103/ \ No newline at end of file diff --git a/newsmill.se.txt b/newsmill.se.txt new file mode 100755 index 0000000..1a99031 --- /dev/null +++ b/newsmill.se.txt @@ -0,0 +1,12 @@ +title: //h1 +body: (//div[@class='articleImg']//img)[1] | //p[contains(@class, 'commentTextArticle') or contains(@class, 'articlePublished')] | //div[@id='articleLeftContent'] +author: //div[@class='byline']//a[contains(@href, '/user/')] + +strip_id_or_class: facts +strip_id_or_class: articleBlogsHolder +strip_id_or_class: byline + +prune: no +tidy: no + +test_url: http://www.newsmill.se/artikel/2012/05/06/medielogiken-v-ger-tyngre-n-reportrarnas-sikter \ No newline at end of file diff --git a/next-gen.biz.txt b/next-gen.biz.txt new file mode 100755 index 0000000..b8d235d --- /dev/null +++ b/next-gen.biz.txt @@ -0,0 +1,16 @@ +# 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/nhk.or.jp.txt b/nhk.or.jp.txt new file mode 100755 index 0000000..0a3bb91 --- /dev/null +++ b/nhk.or.jp.txt @@ -0,0 +1,2 @@ +body: //div[@id = 'news_right'] +test_url: http://www.nhk.or.jp/news/html/20110309/t10014559982000.html \ No newline at end of file diff --git a/observer.com.txt b/observer.com.txt new file mode 100755 index 0000000..0b10753 --- /dev/null +++ b/observer.com.txt @@ -0,0 +1,7 @@ +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/on.net.mk.txt b/on.net.mk.txt new file mode 100755 index 0000000..a95c2b0 --- /dev/null +++ b/on.net.mk.txt @@ -0,0 +1,5 @@ +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/online.wsj.com.txt b/online.wsj.com.txt new file mode 100755 index 0000000..448bb7e --- /dev/null +++ b/online.wsj.com.txt @@ -0,0 +1,25 @@ +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 diff --git a/pakistantvdekho.com.txt b/pakistantvdekho.com.txt new file mode 100755 index 0000000..7e2985e --- /dev/null +++ b/pakistantvdekho.com.txt @@ -0,0 +1,11 @@ +#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://pakistantvdekho.com/showthread.php?647741-Sitam-Gar-by-HUM-TV-Episode-07&p=659080#post659080 \ No newline at end of file diff --git a/pcast.me.txt b/pcast.me.txt new file mode 100755 index 0000000..ae38e8e --- /dev/null +++ b/pcast.me.txt @@ -0,0 +1,2 @@ +prune: no +test_url: http://pcast.me/shownotes/get/16t \ No newline at end of file diff --git a/physicstoday.org.txt b/physicstoday.org.txt new file mode 100755 index 0000000..624055b --- /dev/null +++ b/physicstoday.org.txt @@ -0,0 +1,7 @@ +title: //div[@class='abstitle']//h1 +author: //div[@class='authorList'] +body: //div[@id='fulltext_body'] + +prune: no + +test_url: http://www.physicstoday.org/resource/1/phtoad/v64/i10/p48_s1?bypassSSO=1 \ No newline at end of file diff --git a/pittnews.com.txt b/pittnews.com.txt new file mode 100755 index 0000000..c302526 --- /dev/null +++ b/pittnews.com.txt @@ -0,0 +1,8 @@ +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/p\303\251rotin.com.txt" "b/p\303\251rotin.com.txt" new file mode 100644 index 0000000..7784fa4 --- /dev/null +++ "b/p\303\251rotin.com.txt" @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=xn--protin-bva.com%2Fpost%2F2009%2F06%2F09%2FSAV-Free-un-sketch-kafkaien + +body: //div[contains(concat(' ',normalize-space(@class),' '),' post ')] +test_url: http://xn--protin-bva.com/post/2009/06/09/SAV-Free-un-sketch-kafkaien diff --git a/quechoisir.org.txt b/quechoisir.org.txt new file mode 100644 index 0000000..a8bacdb --- /dev/null +++ b/quechoisir.org.txt @@ -0,0 +1,10 @@ +title: //h1[@id='titre'] +body://h2[@id="surtitre"]|//div[@id="ctn_introarticle"]|//div[@class="ctn_globalcontent"] + +strip_id_or_class: qc-container-main +strip_id_or_class: article_footer + +tidy: no +prune: no + +test_url: http://www.quechoisir.org/alimentation/securite-hygiene/actualite-acrylamide-un-contaminant-trop-present-dans-nos-assiettes diff --git a/quepasa.cl.txt b/quepasa.cl.txt new file mode 100755 index 0000000..fb09a8f --- /dev/null +++ b/quepasa.cl.txt @@ -0,0 +1,6 @@ +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/red-hot-girls.com.txt b/red-hot-girls.com.txt new file mode 100755 index 0000000..0403ee8 --- /dev/null +++ b/red-hot-girls.com.txt @@ -0,0 +1,5 @@ +body: //div[@class='short-text' or starts-with(@id, 'news-id-')] +prune: no +tidy: no + +test_url: http://red-hot-girls.com/2011/06/10/the_red_hot_natalia_maria_53_pics.html \ No newline at end of file diff --git a/redtape.msnbc.msn.com.txt b/redtape.msnbc.msn.com.txt new file mode 100755 index 0000000..8541a0d --- /dev/null +++ b/redtape.msnbc.msn.com.txt @@ -0,0 +1,20 @@ +# Think there might be something up with your parser that it strips out 'print' from the title :) + +title: //meta[@name='title']/@content +author: //meta[@name='author']/@content +date: //meta[@name='date']/@content + +body: //div[@class='articleText'] + +strip: //div[contains(@class, 'day')] +strip: //div[contains(@class, 'month')] +strip: //div[contains(@class, 'year')] +strip: //div[contains(@class, 'time')] +strip: //h1[@class='gl_headline'] +strip: //div[@class='byline'] +strip: //div[@id='left_ear'] +strip: //div[@id='right_ear'] +strip: //div[contains(@class, 'PopularPosts')] +strip ://div[@class='discuss_page_break'] +strip ://div[contains(@class, 'p-content_TagList')] +test_url: http://redtape.msnbc.msn.com/_news/2011/09/28/8020661-sprint-raises-fee-but-wont-free-users-from-two-year-contracts?preview=true \ No newline at end of file diff --git a/retrieverweekly.com.txt b/retrieverweekly.com.txt new file mode 100755 index 0000000..a0a2394 --- /dev/null +++ b/retrieverweekly.com.txt @@ -0,0 +1,6 @@ +single_page_link://a[contains(@href, 'print')] + +# Grab metadata from the "printer-friendly" page, after specifying single_page_link +title://h2 +date://cite +test_url: http://www.retrieverweekly.com/?cmd=displaystory&story_id=7548&format=html \ No newline at end of file diff --git a/rezeptwelt.de.txt b/rezeptwelt.de.txt new file mode 100644 index 0000000..2093573 --- /dev/null +++ b/rezeptwelt.de.txt @@ -0,0 +1,5 @@ +body: //div[@class='step-content'] | //div[@class='global-active ingredients-box'] +title: //div[@class='step-1-container'] + +tidy: no +test_url: http://www.rezeptwelt.de/backen-herzhaft-rezepte/w%C3%BCrstchen-schlangen/530372 \ No newline at end of file diff --git a/ritemail.blogspot.com.txt b/ritemail.blogspot.com.txt new file mode 100755 index 0000000..ed72915 --- /dev/null +++ b/ritemail.blogspot.com.txt @@ -0,0 +1,5 @@ +body: //div[@class='post-body entry-content'] +strip: //div[@id='lws_0'] +prune: no + +test_url: http://ritemail.blogspot.com/2011/06/hayden-panettiere-candids-in-los.html \ No newline at end of file diff --git a/ruttloff.org.txt b/ruttloff.org.txt new file mode 100755 index 0000000..43e130a --- /dev/null +++ b/ruttloff.org.txt @@ -0,0 +1,3 @@ +author: //a[@class='author'] +tidy: no +test_url: http://ruttloff.org/2012/06/13/intervention \ No newline at end of file diff --git a/sct.temple.edu.txt b/sct.temple.edu.txt new file mode 100755 index 0000000..55f2417 --- /dev/null +++ b/sct.temple.edu.txt @@ -0,0 +1,5 @@ +body: //*[@class="entry-content"] +title: //h1[@class="entry-title"] +date: //*[@class="entry-date"] +author: //*[@class="author vcard"] +test_url: http://sct.temple.edu/blogs/news-events/2011/05/congratulations-sct-class-of-2011/ \ No newline at end of file diff --git a/sebbo.net.txt b/sebbo.net.txt new file mode 100755 index 0000000..b6d9c92 --- /dev/null +++ b/sebbo.net.txt @@ -0,0 +1,4 @@ +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/simonwillison.net.txt b/simonwillison.net.txt new file mode 100755 index 0000000..6999969 --- /dev/null +++ b/simonwillison.net.txt @@ -0,0 +1,5 @@ +body: //div[contains(@class, "entry")] + +date: //div[contains(@class, "entryFooter")]/a + +test_url: http://simonwillison.net/2009/Oct/22/redis/ \ No newline at end of file diff --git a/sivers.org.txt b/sivers.org.txt new file mode 100755 index 0000000..a88f30d --- /dev/null +++ b/sivers.org.txt @@ -0,0 +1,6 @@ +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 new file mode 100755 index 0000000..85ca46d --- /dev/null +++ b/smartinvestor.de.txt @@ -0,0 +1,5 @@ +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/smokingapples.com.txt b/smokingapples.com.txt new file mode 100755 index 0000000..c68c132 --- /dev/null +++ b/smokingapples.com.txt @@ -0,0 +1,5 @@ +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/spectator.co.uk.txt b/spectator.co.uk.txt new file mode 100755 index 0000000..d0605ed --- /dev/null +++ b/spectator.co.uk.txt @@ -0,0 +1,7 @@ +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/spiil.org.txt b/spiil.org.txt new file mode 100644 index 0000000..dd12fb8 --- /dev/null +++ b/spiil.org.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=https%3A%2F%2Fwww.spiil.org%2Fguide-bonnes-pratiques-editeurs-de-presse-ligne + +body: //div[contains(concat(' ',normalize-space(@class),' '),' node-content ')] +test_url: https://www.spiil.org/guide-bonnes-pratiques-editeurs-de-presse-ligne diff --git a/sportschau.de.txt b/sportschau.de.txt new file mode 100755 index 0000000..1e58b52 --- /dev/null +++ b/sportschau.de.txt @@ -0,0 +1,22 @@ +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 new file mode 100755 index 0000000..b3da813 --- /dev/null +++ b/sportsillustrated.cnn.com.txt @@ -0,0 +1,26 @@ +# 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/stockholm.etc.se.txt b/stockholm.etc.se.txt new file mode 100755 index 0000000..2f4f8cb --- /dev/null +++ b/stockholm.etc.se.txt @@ -0,0 +1,5 @@ +strip_id_or_class: 'left' +strip_id_or_class: 'right' +strip_id_or_class: 'block-belowcontent' + +test_url: http://stockholm.etc.se/reportage/bakom-stangda-dorrar-pa-fas-3-massa \ No newline at end of file diff --git a/stockholmsfria.nu.txt b/stockholmsfria.nu.txt new file mode 100755 index 0000000..cc8c28b --- /dev/null +++ b/stockholmsfria.nu.txt @@ -0,0 +1,7 @@ +body: //div[contains(@class, 'layout__inner')]//div[contains(@class, 'file-image') or contains(@class, 'node__content')] +author: //article//div[contains(@class, 'field-byline')] +strip_id_or_class: rekommenderade +strip_id_or_class: disqus +strip_id_or_class: annonser + +test_url: http://www.stockholmsfria.nu/artikel/112068 \ No newline at end of file diff --git a/tcmanila.tk.txt b/tcmanila.tk.txt new file mode 100755 index 0000000..f6032ec --- /dev/null +++ b/tcmanila.tk.txt @@ -0,0 +1,7 @@ +title: //h2 +body: //div[@class="post_content"] +author: //span[@class="fn"] +date: //time[@class="updated"] +strip_comments: //yes +footnotes: //yes +test_url: http://tcmanila.tk/post/29189064358/my-2012-roadmap-is-almost-complete-look-at-the \ No newline at end of file diff --git a/tcng.org.txt b/tcng.org.txt new file mode 100755 index 0000000..4873b50 --- /dev/null +++ b/tcng.org.txt @@ -0,0 +1,4 @@ +title: //div[@id='main-content']/h1 +body: //div[@id='main-content'] +strip: //div[@id='main-content']/h1 +test_url: http://www.tcng.org/index.php/blog/view/teaching-basic-health-cutting-down-costs \ No newline at end of file diff --git a/technicallyjordan.tumblr.com.txt b/technicallyjordan.tumblr.com.txt new file mode 100755 index 0000000..d871b60 --- /dev/null +++ b/technicallyjordan.tumblr.com.txt @@ -0,0 +1,8 @@ +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/theappleblog.com.txt b/theappleblog.com.txt new file mode 100755 index 0000000..caa5ae0 --- /dev/null +++ b/theappleblog.com.txt @@ -0,0 +1,3 @@ +# 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/thebrowser.com.txt b/thebrowser.com.txt new file mode 100755 index 0000000..807e7da --- /dev/null +++ b/thebrowser.com.txt @@ -0,0 +1,10 @@ +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/thecarton.net.txt b/thecarton.net.txt new file mode 100755 index 0000000..13fa35a --- /dev/null +++ b/thecarton.net.txt @@ -0,0 +1,10 @@ +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/thedaily.com.txt b/thedaily.com.txt new file mode 100755 index 0000000..e255e6a --- /dev/null +++ b/thedaily.com.txt @@ -0,0 +1,24 @@ +#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'] +test_url: http://www.thedaily.com/page/2012/01/09/010912-news-college-costs-1-5/ \ No newline at end of file diff --git a/theeuropean-magazine.com.txt b/theeuropean-magazine.com.txt new file mode 100755 index 0000000..a19bae1 --- /dev/null +++ b/theeuropean-magazine.com.txt @@ -0,0 +1,17 @@ +# 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/thesiasat.com.txt b/thesiasat.com.txt new file mode 100755 index 0000000..68a8bc8 --- /dev/null +++ b/thesiasat.com.txt @@ -0,0 +1,11 @@ +#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/thisismynext.com.txt b/thisismynext.com.txt new file mode 100755 index 0000000..70b5399 --- /dev/null +++ b/thisismynext.com.txt @@ -0,0 +1,8 @@ +author: //div[@class='meta clearfix']/a +body: //div[@class='post'] + +strip: //div[@class='metaCat'] +strip: //div[@class='post']/h1 +strip: //div[@class='post']/div[@class='meta clearfix'] +strip: //div[@class='post']/div[@class='social-bar clearfix'] +test_url: http://thisismynext.com/2011/10/18/galaxy-nexus-android-ice-cream-sandwich-pictures-video-hands-on/ \ No newline at end of file diff --git a/tofugu.com.txt b/tofugu.com.txt new file mode 100644 index 0000000..5ac9d6a --- /dev/null +++ b/tofugu.com.txt @@ -0,0 +1,3 @@ +body://div[@class='entry-content'] + +test_url: http://www.tofugu.com/2015/07/20/interview-with-toriena-japanese-chiptune/ diff --git a/toolinux.com.txt b/toolinux.com.txt new file mode 100755 index 0000000..3f1a840 --- /dev/null +++ b/toolinux.com.txt @@ -0,0 +1,5 @@ +title: //h2[contains(@class,'news')] +body: //div[contains(@class,'articleContent')] +date: substring-after(//div[@class = 'SupaDate']/text(), 'le') + +test_url: http://www.toolinux.com/Wi-Fi-Linksys-WRT-la-legende-de diff --git a/tthor.com.txt b/tthor.com.txt new file mode 100755 index 0000000..902fcd1 --- /dev/null +++ b/tthor.com.txt @@ -0,0 +1,2 @@ +prune: no +test_url: http://www.tthor.com/06/ \ No newline at end of file diff --git a/tuckreview.com.txt b/tuckreview.com.txt new file mode 100755 index 0000000..6e18e3d --- /dev/null +++ b/tuckreview.com.txt @@ -0,0 +1,6 @@ +title: //h1[@class='post-title'] +author: //div[@class='display-name'] +date: //div[@class='date'] +body: //div[@class='body'] +footnotes: no +test_url: http://tuckreview.com/2012/8/14/migrating-to-v6 \ No newline at end of file diff --git a/tuhdo.github.io.txt b/tuhdo.github.io.txt new file mode 100644 index 0000000..beb551f --- /dev/null +++ b/tuhdo.github.io.txt @@ -0,0 +1,7 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=https%3A%2F%2Ftuhdo.github.io%2Femacs-tutor.html + +body: //div[@id='content'] +strip_id_or_class: table-of-contents +test_url: https://tuhdo.github.io/emacs-tutor.html diff --git a/uzivatelsketestovani.cz.txt b/uzivatelsketestovani.cz.txt new file mode 100755 index 0000000..3661b06 --- /dev/null +++ b/uzivatelsketestovani.cz.txt @@ -0,0 +1,2 @@ +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 new file mode 100755 index 0000000..c0133c9 --- /dev/null +++ b/varingen.no.txt @@ -0,0 +1,5 @@ +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 new file mode 100755 index 0000000..d22fc5c --- /dev/null +++ b/vemedio.com.txt @@ -0,0 +1,6 @@ +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/walrusmagazine.com.txt b/walrusmagazine.com.txt new file mode 100755 index 0000000..c53eb0d --- /dev/null +++ b/walrusmagazine.com.txt @@ -0,0 +1,14 @@ +title: //div[@id='pr']/h3 +author: //div[@class='dateline']//a[contains(@href, '/author/')] + +# print page +body: //div[@id='prbody'] +# standard page +body: //div[@id='pgbody'] + +# for multi-page articles +single_page_link: //div[@class='tipjar']//a[contains(@href, '/printerFriendly.php?')] + +prune: no + +test_url: http://www.walrusmagazine.com/articles/2011.12-memoir-kidnapped \ No newline at end of file diff --git a/weblog.bignerdranch.com.txt b/weblog.bignerdranch.com.txt new file mode 100755 index 0000000..578ba52 --- /dev/null +++ b/weblog.bignerdranch.com.txt @@ -0,0 +1,5 @@ +title://div[@class="post"]/h2 +author://p[@class="postinfo"]/a +date:substring-before(substring-after(//p[@class="postinfo"],' on '),' under ') +body://div[@class="contenttext"] +test_url: http://weblog.bignerdranch.com/?p=304 \ No newline at end of file diff --git a/webpaper.nzz.ch.txt b/webpaper.nzz.ch.txt new file mode 100755 index 0000000..cea1014 --- /dev/null +++ b/webpaper.nzz.ch.txt @@ -0,0 +1,8 @@ +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/westhamtillidie.com.txt b/westhamtillidie.com.txt new file mode 100755 index 0000000..3132e98 --- /dev/null +++ b/westhamtillidie.com.txt @@ -0,0 +1,6 @@ +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/wheelyric.com.txt b/wheelyric.com.txt new file mode 100755 index 0000000..b9eeaa0 --- /dev/null +++ b/wheelyric.com.txt @@ -0,0 +1,11 @@ +body://div[contains(@class,'oAndtLyrics')] +strip://div[contains(@class,'info')] +strip://div[contains(@id,'romanization')] +strip://div[contains(@id,'youtube')] +strip://div[contains(@id,'romanizationSelector')] +strip://div[contains(@id,'langSelectWrap')] +strip://div[contains(@id,'requestTranslationWrap')] +strip://div[contains(@id,'viewMore')] +strip://div[contains(@class,'lyricsListInMainContent')] +strip://div[contains(@class,'descIpNoti')] +test_url: http://wheelyric.com/lyrics/121#2 \ No newline at end of file diff --git a/winsupersite.com.txt b/winsupersite.com.txt new file mode 100755 index 0000000..f725b67 --- /dev/null +++ b/winsupersite.com.txt @@ -0,0 +1,3 @@ +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/wmnf.org.txt b/wmnf.org.txt new file mode 100755 index 0000000..1d403a9 --- /dev/null +++ b/wmnf.org.txt @@ -0,0 +1,13 @@ +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 diff --git a/wordyard.com.txt b/wordyard.com.txt new file mode 100644 index 0000000..d8c753d --- /dev/null +++ b/wordyard.com.txt @@ -0,0 +1,8 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fwww.wordyard.com%2F2014%2F09%2F26%2Fremove-blindfold-before-embarking-to-utopia%2F + +body: //div[contains(concat(' ',normalize-space(@class),' '),' entry-content ')] +strip_id_or_class: robots-nocontent +strip_id_or_class: post-revisions +test_url: http://www.wordyard.com/2014/09/26/remove-blindfold-before-embarking-to-utopia/ \ No newline at end of file diff --git a/wyctim.com.txt b/wyctim.com.txt new file mode 100755 index 0000000..bd7ecf2 --- /dev/null +++ b/wyctim.com.txt @@ -0,0 +1,3 @@ +body: //div[@class='article-body'] +title: //h1 +test_url: http://wyctim.com/icloud-sync-regebbi-rendszereken/ \ No newline at end of file diff --git a/xfgjls.com.txt b/xfgjls.com.txt new file mode 100755 index 0000000..2dc247a --- /dev/null +++ b/xfgjls.com.txt @@ -0,0 +1,11 @@ +# 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/yosko.net.txt b/yosko.net.txt new file mode 100644 index 0000000..99e711e --- /dev/null +++ b/yosko.net.txt @@ -0,0 +1,6 @@ +# Generated by FiveFilters.org's web-based selection tool +# Place this file inside your site_config/custom/ folder +# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Fwww.yosko.net%2Farticle33%2Fsnippet-06-javascript-capturer-des-raccourcis-clavier-utilises-par-votre-navigateur + +body: //article//section +test_url: http://www.yosko.net/article33/snippet-06-javascript-capturer-des-raccourcis-clavier-utilises-par-votre-navigateur diff --git a/zcommunications.org.txt b/zcommunications.org.txt new file mode 100755 index 0000000..4deb49b --- /dev/null +++ b/zcommunications.org.txt @@ -0,0 +1,7 @@ +title: //h1[@id='view_title'] +author: //div[contains(@class, 'content_authors')]//a +body: //div[@id='view_body'] + +prune: no + +test_url: http://www.zcommunications.org/orwellian-language-update-by-edward-s-herman.html \ No newline at end of file diff --git a/zerodistraction.com.txt b/zerodistraction.com.txt new file mode 100644 index 0000000..80ac6c1 --- /dev/null +++ b/zerodistraction.com.txt @@ -0,0 +1,4 @@ +author: //span[@class='author']//a +date: //span[@class='date'] +test_url: http://zerodistraction.com/blog/2012/3/11/retina-ipad-that-means-i-am-going-digital-only-for-comic-boo.html +test_url: http://zerodistraction.com/notes/unreasonably-grumpy \ No newline at end of file diff --git a/zingtrain.com.txt b/zingtrain.com.txt new file mode 100755 index 0000000..188d4dd --- /dev/null +++ b/zingtrain.com.txt @@ -0,0 +1,3 @@ +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 From ca877ec3e88d1a7c7148b9e1e324c624b21bc4b5 Mon Sep 17 00:00:00 2001 From: Jeremy Benoist Date: Tue, 16 Feb 2016 14:19:07 +0100 Subject: Remove dead website --- 944.com.txt | 9 --------- asahi.com.txt | 3 --- blog.jaysalvat.com.txt | 4 ---- boldizsar.palotas.eu.txt | 3 --- cnnsi.com.txt | 26 -------------------------- eamesinerudition.com.txt | 8 -------- essentialpublicradio.org.txt | 6 ------ ftd.de.txt | 5 ----- garythink.com.txt | 3 --- givemesomethingtoread.com.txt | 3 --- heroturko.me.txt | 6 ------ kachiblog.com.txt | 7 ------- lifeandculture.fr.txt | 3 --- news.kanaloco.jp.txt | 9 --------- newsmill.se.txt | 12 ------------ nhk.or.jp.txt | 2 -- pakistantvdekho.com.txt | 11 ----------- pcast.me.txt | 2 -- sct.temple.edu.txt | 5 ----- tcmanila.tk.txt | 7 ------- tcng.org.txt | 4 ---- thedaily.com.txt | 24 ------------------------ tthor.com.txt | 2 -- tuckreview.com.txt | 6 ------ wheelyric.com.txt | 11 ----------- 25 files changed, 181 deletions(-) delete mode 100755 944.com.txt delete mode 100755 asahi.com.txt delete mode 100755 blog.jaysalvat.com.txt delete mode 100755 boldizsar.palotas.eu.txt delete mode 100755 cnnsi.com.txt delete mode 100755 eamesinerudition.com.txt delete mode 100755 essentialpublicradio.org.txt delete mode 100755 ftd.de.txt delete mode 100755 garythink.com.txt delete mode 100755 givemesomethingtoread.com.txt delete mode 100755 heroturko.me.txt delete mode 100755 kachiblog.com.txt delete mode 100755 lifeandculture.fr.txt delete mode 100755 news.kanaloco.jp.txt delete mode 100755 newsmill.se.txt delete mode 100755 nhk.or.jp.txt delete mode 100755 pakistantvdekho.com.txt delete mode 100755 pcast.me.txt delete mode 100755 sct.temple.edu.txt delete mode 100755 tcmanila.tk.txt delete mode 100755 tcng.org.txt delete mode 100755 thedaily.com.txt delete mode 100755 tthor.com.txt delete mode 100755 tuckreview.com.txt delete mode 100755 wheelyric.com.txt diff --git a/944.com.txt b/944.com.txt deleted file mode 100755 index 8bf6a4c..0000000 --- a/944.com.txt +++ /dev/null @@ -1,9 +0,0 @@ -title: //h2[@class='border'] -body: //div[@class='padding'] - -convert_double_br_tags: yes - -strip: //div[@id='social_sharing'] -strip: //div[@class='socialLinks'] - -test_url: http://www.944.com/articles/mild-obsessions-frock-la-get-to-know-victoria-tik-s-haute-sustainable-fashion-line/ \ No newline at end of file diff --git a/asahi.com.txt b/asahi.com.txt deleted file mode 100755 index b4eec7b..0000000 --- a/asahi.com.txt +++ /dev/null @@ -1,3 +0,0 @@ -body: //div[@id='HeadLine'] -strip: //div[@id='utility_right'] -test_url: http://www.asahi.com/culture/update/0520/TKY201105200321.html \ No newline at end of file diff --git a/blog.jaysalvat.com.txt b/blog.jaysalvat.com.txt deleted file mode 100755 index e89ad3a..0000000 --- a/blog.jaysalvat.com.txt +++ /dev/null @@ -1,4 +0,0 @@ -date: //span[contains(@class, 'date-links')] -author: //span[contains(@class, 'author-links')] -body: //div[contains(@class, 'entry-content')] -test_url: http://blog.jaysalvat.com/article/celui-qui-avait-refait-son-site-web \ No newline at end of file diff --git a/boldizsar.palotas.eu.txt b/boldizsar.palotas.eu.txt deleted file mode 100755 index 3616bbf..0000000 --- a/boldizsar.palotas.eu.txt +++ /dev/null @@ -1,3 +0,0 @@ -title: //h2[@class='entry-title'] -body: //div[@class='entry-content'] -test_url: http://boldizsar.palotas.eu/blog/?p=1394 \ No newline at end of file diff --git a/cnnsi.com.txt b/cnnsi.com.txt deleted file mode 100755 index ac49aef..0000000 --- a/cnnsi.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://cnnsi.com/2012/writers/peter_king/01/08/wild.card.round/index.html \ No newline at end of file diff --git a/eamesinerudition.com.txt b/eamesinerudition.com.txt deleted file mode 100755 index 89a68bc..0000000 --- a/eamesinerudition.com.txt +++ /dev/null @@ -1,8 +0,0 @@ -title: //div [@class="post contain"]/h1 -strip: //div [@class="post contain"]/h1 -body: //div [@class="post contain"] -author: substring-before(//title, ':') -author: substring-before(//title, ' ') - - -test_url: http://eamesinerudition.com/2012/03/hospital-numbers-are-bad-for-you \ No newline at end of file diff --git a/essentialpublicradio.org.txt b/essentialpublicradio.org.txt deleted file mode 100755 index 9a92239..0000000 --- a/essentialpublicradio.org.txt +++ /dev/null @@ -1,6 +0,0 @@ -title: //*[@itemprop='headline'] -author: //*[@itemprop='author'] -date: //*[@itemprop='datePublished'] -body: //*[@itemprop='articleBody'] -strip: //*[contains(@class, 'instapaper_ignore')] -test_url: http://www.essentialpublicradio.org/story/2011-11-14/volunteers-sought-federal-tax-assistance-program-pennsylvania-9421 \ No newline at end of file diff --git a/ftd.de.txt b/ftd.de.txt deleted file mode 100755 index 7d76af0..0000000 --- a/ftd.de.txt +++ /dev/null @@ -1,5 +0,0 @@ -body: //div[@class='boxIntroHead']/span[@class='h3'] | //div[@class='section']/div[@class='paragraph' or @class='embObjLeft'] -single_page_link: //a[@class='icon print'] - -test_url: http://www.ftd.de/it-medien/it-telekommunikation/:mobilfunk-vivendi-und-vodafone-trennen-sich-in-frankreich/60034691.html -test_url: http://www.ftd.de/it-medien/medien-internet/:verkauf-von-warner-music-musikbranche-auf-dem-sprung/60048185.html \ No newline at end of file diff --git a/garythink.com.txt b/garythink.com.txt deleted file mode 100755 index 327ac55..0000000 --- a/garythink.com.txt +++ /dev/null @@ -1,3 +0,0 @@ -tidy: no - -test_url: http://www.garythink.com/eft/testing.html \ No newline at end of file diff --git a/givemesomethingtoread.com.txt b/givemesomethingtoread.com.txt deleted file mode 100755 index 0de0750..0000000 --- a/givemesomethingtoread.com.txt +++ /dev/null @@ -1,3 +0,0 @@ -single_page_link: //div[@id="content"]//h2/a - -test_url: http://givemesomethingtoread.com/post/6285838917/the-baddest-lawyer-in-the-history-of-jersey \ No newline at end of file diff --git a/heroturko.me.txt b/heroturko.me.txt deleted file mode 100755 index 07b6adf..0000000 --- a/heroturko.me.txt +++ /dev/null @@ -1,6 +0,0 @@ -title: //div[contains(@class, 'title')]//h1 -body: //div[contains(@class, 'story')] - -prune: no - -test_url: http://www.heroturko.me/5223034-ds-catia-p3-v5-6r2014-gasp0-x86x64-multilanguage-english-docs.html \ No newline at end of file diff --git a/kachiblog.com.txt b/kachiblog.com.txt deleted file mode 100755 index 57ab0de..0000000 --- a/kachiblog.com.txt +++ /dev/null @@ -1,7 +0,0 @@ -title: //h3[contains(@class, 'entry-title')] -date: //abbr[@itemprop='datePublished']/@title -body: //div[@itemprop='articleBody'] -tidy: no - -test_url: http://www.kachiblog.com/2013/05/samsung-galaxy-s4-vs-samsung-galaxy.html -test_url: http://www.kachiblog.com/feed diff --git a/lifeandculture.fr.txt b/lifeandculture.fr.txt deleted file mode 100755 index 0e1dceb..0000000 --- a/lifeandculture.fr.txt +++ /dev/null @@ -1,3 +0,0 @@ -title: //h2[@class="entry-title"] -body: //div[@class="entry-content"] -test_url: http://www.lifeandculture.fr/digital/facebook-and-the-epiphanator-an-end-to-endings/ \ No newline at end of file diff --git a/news.kanaloco.jp.txt b/news.kanaloco.jp.txt deleted file mode 100755 index 5754d47..0000000 --- a/news.kanaloco.jp.txt +++ /dev/null @@ -1,9 +0,0 @@ -body: //div[@id='main'] -strip: //div[@id='sbs'] -strip: //div[@id='fsizeSwitch'] -strip: //div[@id='googleAd'] -strip: //div[@id='detailFoot'] -strip_image_src: counter?key -convert_double_br_tags: yes - -test_url: http://news.kanaloco.jp/localnews/article/1105200018/ \ No newline at end of file diff --git a/newsmill.se.txt b/newsmill.se.txt deleted file mode 100755 index 1a99031..0000000 --- a/newsmill.se.txt +++ /dev/null @@ -1,12 +0,0 @@ -title: //h1 -body: (//div[@class='articleImg']//img)[1] | //p[contains(@class, 'commentTextArticle') or contains(@class, 'articlePublished')] | //div[@id='articleLeftContent'] -author: //div[@class='byline']//a[contains(@href, '/user/')] - -strip_id_or_class: facts -strip_id_or_class: articleBlogsHolder -strip_id_or_class: byline - -prune: no -tidy: no - -test_url: http://www.newsmill.se/artikel/2012/05/06/medielogiken-v-ger-tyngre-n-reportrarnas-sikter \ No newline at end of file diff --git a/nhk.or.jp.txt b/nhk.or.jp.txt deleted file mode 100755 index 0a3bb91..0000000 --- a/nhk.or.jp.txt +++ /dev/null @@ -1,2 +0,0 @@ -body: //div[@id = 'news_right'] -test_url: http://www.nhk.or.jp/news/html/20110309/t10014559982000.html \ No newline at end of file diff --git a/pakistantvdekho.com.txt b/pakistantvdekho.com.txt deleted file mode 100755 index 7e2985e..0000000 --- a/pakistantvdekho.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://pakistantvdekho.com/showthread.php?647741-Sitam-Gar-by-HUM-TV-Episode-07&p=659080#post659080 \ No newline at end of file diff --git a/pcast.me.txt b/pcast.me.txt deleted file mode 100755 index ae38e8e..0000000 --- a/pcast.me.txt +++ /dev/null @@ -1,2 +0,0 @@ -prune: no -test_url: http://pcast.me/shownotes/get/16t \ No newline at end of file diff --git a/sct.temple.edu.txt b/sct.temple.edu.txt deleted file mode 100755 index 55f2417..0000000 --- a/sct.temple.edu.txt +++ /dev/null @@ -1,5 +0,0 @@ -body: //*[@class="entry-content"] -title: //h1[@class="entry-title"] -date: //*[@class="entry-date"] -author: //*[@class="author vcard"] -test_url: http://sct.temple.edu/blogs/news-events/2011/05/congratulations-sct-class-of-2011/ \ No newline at end of file diff --git a/tcmanila.tk.txt b/tcmanila.tk.txt deleted file mode 100755 index f6032ec..0000000 --- a/tcmanila.tk.txt +++ /dev/null @@ -1,7 +0,0 @@ -title: //h2 -body: //div[@class="post_content"] -author: //span[@class="fn"] -date: //time[@class="updated"] -strip_comments: //yes -footnotes: //yes -test_url: http://tcmanila.tk/post/29189064358/my-2012-roadmap-is-almost-complete-look-at-the \ No newline at end of file diff --git a/tcng.org.txt b/tcng.org.txt deleted file mode 100755 index 4873b50..0000000 --- a/tcng.org.txt +++ /dev/null @@ -1,4 +0,0 @@ -title: //div[@id='main-content']/h1 -body: //div[@id='main-content'] -strip: //div[@id='main-content']/h1 -test_url: http://www.tcng.org/index.php/blog/view/teaching-basic-health-cutting-down-costs \ No newline at end of file diff --git a/thedaily.com.txt b/thedaily.com.txt deleted file mode 100755 index e255e6a..0000000 --- a/thedaily.com.txt +++ /dev/null @@ -1,24 +0,0 @@ -#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'] -test_url: http://www.thedaily.com/page/2012/01/09/010912-news-college-costs-1-5/ \ No newline at end of file diff --git a/tthor.com.txt b/tthor.com.txt deleted file mode 100755 index 902fcd1..0000000 --- a/tthor.com.txt +++ /dev/null @@ -1,2 +0,0 @@ -prune: no -test_url: http://www.tthor.com/06/ \ No newline at end of file diff --git a/tuckreview.com.txt b/tuckreview.com.txt deleted file mode 100755 index 6e18e3d..0000000 --- a/tuckreview.com.txt +++ /dev/null @@ -1,6 +0,0 @@ -title: //h1[@class='post-title'] -author: //div[@class='display-name'] -date: //div[@class='date'] -body: //div[@class='body'] -footnotes: no -test_url: http://tuckreview.com/2012/8/14/migrating-to-v6 \ No newline at end of file diff --git a/wheelyric.com.txt b/wheelyric.com.txt deleted file mode 100755 index b9eeaa0..0000000 --- a/wheelyric.com.txt +++ /dev/null @@ -1,11 +0,0 @@ -body://div[contains(@class,'oAndtLyrics')] -strip://div[contains(@class,'info')] -strip://div[contains(@id,'romanization')] -strip://div[contains(@id,'youtube')] -strip://div[contains(@id,'romanizationSelector')] -strip://div[contains(@id,'langSelectWrap')] -strip://div[contains(@id,'requestTranslationWrap')] -strip://div[contains(@id,'viewMore')] -strip://div[contains(@class,'lyricsListInMainContent')] -strip://div[contains(@class,'descIpNoti')] -test_url: http://wheelyric.com/lyrics/121#2 \ No newline at end of file -- cgit v1.3.1 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 --- alt1040.com.txt | 2 -- community.service-now.com.txt | 8 -------- community.servicenow.com.txt | 8 ++++++++ guardian.co.uk.txt | 7 ------- hipertextual.com.txt | 2 ++ iceblog.free.fr.txt | 7 ------- m.guardian.co.uk.txt | 12 ------------ macformat.techradar.com.txt | 9 --------- mobile.slate.com.txt | 5 ----- neomoney.co.txt | 3 --- online.wsj.com.txt | 25 ------------------------- physicstoday.org.txt | 7 ------- red-hot-girls.com.txt | 5 ----- redtape.msnbc.msn.com.txt | 20 -------------------- retrieverweekly.com.txt | 6 ------ ritemail.blogspot.com.txt | 5 ----- ruttloff.org.txt | 3 --- simonwillison.net.txt | 5 ----- stockholmsfria.nu.txt | 7 ------- stockholmsfria.se.txt | 7 +++++++ thisismynext.com.txt | 8 -------- walrusmagazine.com.txt | 14 -------------- weblog.bignerdranch.com.txt | 5 ----- wyctim.com.txt | 3 --- zcommunications.org.txt | 7 ------- 25 files changed, 17 insertions(+), 173 deletions(-) delete mode 100755 alt1040.com.txt delete mode 100755 community.service-now.com.txt create mode 100755 community.servicenow.com.txt delete mode 100644 guardian.co.uk.txt create mode 100755 hipertextual.com.txt delete mode 100644 iceblog.free.fr.txt delete mode 100644 m.guardian.co.uk.txt delete mode 100755 macformat.techradar.com.txt delete mode 100755 mobile.slate.com.txt delete mode 100755 neomoney.co.txt delete mode 100755 online.wsj.com.txt delete mode 100755 physicstoday.org.txt delete mode 100755 red-hot-girls.com.txt delete mode 100755 redtape.msnbc.msn.com.txt delete mode 100755 retrieverweekly.com.txt delete mode 100755 ritemail.blogspot.com.txt delete mode 100755 ruttloff.org.txt delete mode 100755 simonwillison.net.txt delete mode 100755 stockholmsfria.nu.txt create mode 100755 stockholmsfria.se.txt delete mode 100755 thisismynext.com.txt delete mode 100755 walrusmagazine.com.txt delete mode 100755 weblog.bignerdranch.com.txt delete mode 100755 wyctim.com.txt delete mode 100755 zcommunications.org.txt diff --git a/alt1040.com.txt b/alt1040.com.txt deleted file mode 100755 index 4fd4571..0000000 --- a/alt1040.com.txt +++ /dev/null @@ -1,2 +0,0 @@ -body: //*[(@class = "historia")] -test_url: http://alt1040.com/2011/09/banda-ancha-en-america-latina-insignificante \ No newline at end of file diff --git a/community.service-now.com.txt b/community.service-now.com.txt deleted file mode 100755 index c9854b4..0000000 --- a/community.service-now.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.service-now.com/blog/lawrenceeng/seasons-greetings-servicenow-team \ No newline at end of file diff --git a/community.servicenow.com.txt b/community.servicenow.com.txt new file mode 100755 index 0000000..98fb688 --- /dev/null +++ b/community.servicenow.com.txt @@ -0,0 +1,8 @@ +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/guardian.co.uk.txt b/guardian.co.uk.txt deleted file mode 100644 index 299d71c..0000000 --- a/guardian.co.uk.txt +++ /dev/null @@ -1,7 +0,0 @@ -title: //div[@id='main-article-info']//h1 -body: //div[@id='article-wrapper'] -date: //li[@class='publication']//time[@pubdate] | //li[@class='publication']//data[@pubdate] -author: //li[@class='byline'] -prune: no -tidy: no -test_url: http://www.guardian.co.uk/business/2011/oct/06/quantitative-easing-75bn-bank-of-england \ No newline at end of file diff --git a/hipertextual.com.txt b/hipertextual.com.txt new file mode 100755 index 0000000..2cb7ca5 --- /dev/null +++ b/hipertextual.com.txt @@ -0,0 +1,2 @@ +body: //*[(@class = "historia")] +test_url: http://hipertextual.com/2011/09/banda-ancha-en-america-latina-insignificante diff --git a/iceblog.free.fr.txt b/iceblog.free.fr.txt deleted file mode 100644 index 9d1ce21..0000000 --- a/iceblog.free.fr.txt +++ /dev/null @@ -1,7 +0,0 @@ - -# Generated by FiveFilters.org's web-based selection tool -# Place this file inside your site_config/custom/ folder -# Source: http://siteconfig.fivefilters.org/grab.php?url=http%3A%2F%2Ficeblog.free.fr%2Findex.php%3Farticle102%2Fweb-histoire-geek-4 - -body: //article//section -test_url: http://iceblog.free.fr/index.php?article102/web-histoire-geek-4 diff --git a/m.guardian.co.uk.txt b/m.guardian.co.uk.txt deleted file mode 100644 index cce0271..0000000 --- a/m.guardian.co.uk.txt +++ /dev/null @@ -1,12 +0,0 @@ -title: //p[@class="txhead"] -author: //div[@class='txb'] -wrap_in(p): //div[@class='para'] -date: //div[@class='txb']/following-sibling::p/text()[substring(., 14)] -strip: //table[@class="tlogo"] -strip: //div[@class="cookieText"] -strip: //*[@class="sltb"] -strip: //*[@class="ijobs-x-link"] -strip: //*[@class="sponscolour"] -strip: //*[@class="sponsouter"] -strip: //div[@id="bottom-nav-block"]/following::* -test_url: http://m.guardian.co.uk/ms/p/gnm/op/s3OOwgO3yIhGuj41C1_S3Xg/view.m?id=15&gid=world/2012/jul/26/arctic-climate-change&cat=top-stories \ No newline at end of file diff --git a/macformat.techradar.com.txt b/macformat.techradar.com.txt deleted file mode 100755 index 522efb4..0000000 --- a/macformat.techradar.com.txt +++ /dev/null @@ -1,9 +0,0 @@ -# Remove news feed -strip: //div[@id='news_feed_front'] - -# Remove pull quote -strip: //div[@class='field field-type-text field-field-pull-quote'] - -# Remove login -strip: //div[@class='right_bar_login'] -test_url: http://macformat.techradar.com/blog/solid-state-storage-bringing-parity-back-mac-29-10-10&article=89189666 \ No newline at end of file diff --git a/mobile.slate.com.txt b/mobile.slate.com.txt deleted file mode 100755 index 6ffcd18..0000000 --- a/mobile.slate.com.txt +++ /dev/null @@ -1,5 +0,0 @@ -title: //h2[@class="article_title"] -strip: //a[@class="houseAdLink"] -strip: //h1 -strip: //div[@class="more_articles"] -test_url: http://mobile.slate.com/rss.jsp?rssid=411&item=http%3a%2f%2fwww.slate.com%2fdefault.aspx%3fdisplaymode%3d201%26id%3d2293749%26device%3drss \ No newline at end of file diff --git a/neomoney.co.txt b/neomoney.co.txt deleted file mode 100755 index 2089fc3..0000000 --- a/neomoney.co.txt +++ /dev/null @@ -1,3 +0,0 @@ -title: //*[@class="header_title"]/h1 -body: //div[contains(@class, 'content')] -test_url: http://neomoney.co/personal/expatriate-and-migrant-loans/expatriate-loans/ \ No newline at end of file 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 diff --git a/physicstoday.org.txt b/physicstoday.org.txt deleted file mode 100755 index 624055b..0000000 --- a/physicstoday.org.txt +++ /dev/null @@ -1,7 +0,0 @@ -title: //div[@class='abstitle']//h1 -author: //div[@class='authorList'] -body: //div[@id='fulltext_body'] - -prune: no - -test_url: http://www.physicstoday.org/resource/1/phtoad/v64/i10/p48_s1?bypassSSO=1 \ No newline at end of file diff --git a/red-hot-girls.com.txt b/red-hot-girls.com.txt deleted file mode 100755 index 0403ee8..0000000 --- a/red-hot-girls.com.txt +++ /dev/null @@ -1,5 +0,0 @@ -body: //div[@class='short-text' or starts-with(@id, 'news-id-')] -prune: no -tidy: no - -test_url: http://red-hot-girls.com/2011/06/10/the_red_hot_natalia_maria_53_pics.html \ No newline at end of file diff --git a/redtape.msnbc.msn.com.txt b/redtape.msnbc.msn.com.txt deleted file mode 100755 index 8541a0d..0000000 --- a/redtape.msnbc.msn.com.txt +++ /dev/null @@ -1,20 +0,0 @@ -# Think there might be something up with your parser that it strips out 'print' from the title :) - -title: //meta[@name='title']/@content -author: //meta[@name='author']/@content -date: //meta[@name='date']/@content - -body: //div[@class='articleText'] - -strip: //div[contains(@class, 'day')] -strip: //div[contains(@class, 'month')] -strip: //div[contains(@class, 'year')] -strip: //div[contains(@class, 'time')] -strip: //h1[@class='gl_headline'] -strip: //div[@class='byline'] -strip: //div[@id='left_ear'] -strip: //div[@id='right_ear'] -strip: //div[contains(@class, 'PopularPosts')] -strip ://div[@class='discuss_page_break'] -strip ://div[contains(@class, 'p-content_TagList')] -test_url: http://redtape.msnbc.msn.com/_news/2011/09/28/8020661-sprint-raises-fee-but-wont-free-users-from-two-year-contracts?preview=true \ No newline at end of file diff --git a/retrieverweekly.com.txt b/retrieverweekly.com.txt deleted file mode 100755 index a0a2394..0000000 --- a/retrieverweekly.com.txt +++ /dev/null @@ -1,6 +0,0 @@ -single_page_link://a[contains(@href, 'print')] - -# Grab metadata from the "printer-friendly" page, after specifying single_page_link -title://h2 -date://cite -test_url: http://www.retrieverweekly.com/?cmd=displaystory&story_id=7548&format=html \ No newline at end of file diff --git a/ritemail.blogspot.com.txt b/ritemail.blogspot.com.txt deleted file mode 100755 index ed72915..0000000 --- a/ritemail.blogspot.com.txt +++ /dev/null @@ -1,5 +0,0 @@ -body: //div[@class='post-body entry-content'] -strip: //div[@id='lws_0'] -prune: no - -test_url: http://ritemail.blogspot.com/2011/06/hayden-panettiere-candids-in-los.html \ No newline at end of file diff --git a/ruttloff.org.txt b/ruttloff.org.txt deleted file mode 100755 index 43e130a..0000000 --- a/ruttloff.org.txt +++ /dev/null @@ -1,3 +0,0 @@ -author: //a[@class='author'] -tidy: no -test_url: http://ruttloff.org/2012/06/13/intervention \ No newline at end of file diff --git a/simonwillison.net.txt b/simonwillison.net.txt deleted file mode 100755 index 6999969..0000000 --- a/simonwillison.net.txt +++ /dev/null @@ -1,5 +0,0 @@ -body: //div[contains(@class, "entry")] - -date: //div[contains(@class, "entryFooter")]/a - -test_url: http://simonwillison.net/2009/Oct/22/redis/ \ No newline at end of file diff --git a/stockholmsfria.nu.txt b/stockholmsfria.nu.txt deleted file mode 100755 index cc8c28b..0000000 --- a/stockholmsfria.nu.txt +++ /dev/null @@ -1,7 +0,0 @@ -body: //div[contains(@class, 'layout__inner')]//div[contains(@class, 'file-image') or contains(@class, 'node__content')] -author: //article//div[contains(@class, 'field-byline')] -strip_id_or_class: rekommenderade -strip_id_or_class: disqus -strip_id_or_class: annonser - -test_url: http://www.stockholmsfria.nu/artikel/112068 \ No newline at end of file diff --git a/stockholmsfria.se.txt b/stockholmsfria.se.txt new file mode 100755 index 0000000..ee87483 --- /dev/null +++ b/stockholmsfria.se.txt @@ -0,0 +1,7 @@ +body: //div[contains(@class, 'layout__inner')]//div[contains(@class, 'file-image') or contains(@class, 'node__content')] +author: //article//div[contains(@class, 'field-byline')] +strip_id_or_class: rekommenderade +strip_id_or_class: disqus +strip_id_or_class: annonser + +test_url: http://www.stockholmsfria.se/artikel/112068 diff --git a/thisismynext.com.txt b/thisismynext.com.txt deleted file mode 100755 index 70b5399..0000000 --- a/thisismynext.com.txt +++ /dev/null @@ -1,8 +0,0 @@ -author: //div[@class='meta clearfix']/a -body: //div[@class='post'] - -strip: //div[@class='metaCat'] -strip: //div[@class='post']/h1 -strip: //div[@class='post']/div[@class='meta clearfix'] -strip: //div[@class='post']/div[@class='social-bar clearfix'] -test_url: http://thisismynext.com/2011/10/18/galaxy-nexus-android-ice-cream-sandwich-pictures-video-hands-on/ \ No newline at end of file diff --git a/walrusmagazine.com.txt b/walrusmagazine.com.txt deleted file mode 100755 index c53eb0d..0000000 --- a/walrusmagazine.com.txt +++ /dev/null @@ -1,14 +0,0 @@ -title: //div[@id='pr']/h3 -author: //div[@class='dateline']//a[contains(@href, '/author/')] - -# print page -body: //div[@id='prbody'] -# standard page -body: //div[@id='pgbody'] - -# for multi-page articles -single_page_link: //div[@class='tipjar']//a[contains(@href, '/printerFriendly.php?')] - -prune: no - -test_url: http://www.walrusmagazine.com/articles/2011.12-memoir-kidnapped \ No newline at end of file diff --git a/weblog.bignerdranch.com.txt b/weblog.bignerdranch.com.txt deleted file mode 100755 index 578ba52..0000000 --- a/weblog.bignerdranch.com.txt +++ /dev/null @@ -1,5 +0,0 @@ -title://div[@class="post"]/h2 -author://p[@class="postinfo"]/a -date:substring-before(substring-after(//p[@class="postinfo"],' on '),' under ') -body://div[@class="contenttext"] -test_url: http://weblog.bignerdranch.com/?p=304 \ No newline at end of file diff --git a/wyctim.com.txt b/wyctim.com.txt deleted file mode 100755 index bd7ecf2..0000000 --- a/wyctim.com.txt +++ /dev/null @@ -1,3 +0,0 @@ -body: //div[@class='article-body'] -title: //h1 -test_url: http://wyctim.com/icloud-sync-regebbi-rendszereken/ \ No newline at end of file diff --git a/zcommunications.org.txt b/zcommunications.org.txt deleted file mode 100755 index 4deb49b..0000000 --- a/zcommunications.org.txt +++ /dev/null @@ -1,7 +0,0 @@ -title: //h1[@id='view_title'] -author: //div[contains(@class, 'content_authors')]//a -body: //div[@id='view_body'] - -prune: no - -test_url: http://www.zcommunications.org/orwellian-language-update-by-edward-s-herman.html \ No newline at end of file -- cgit v1.3.1 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 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