├── .gitignore ├── _config.yml ├── vendor ├── composer │ ├── installed.json │ ├── autoload_psr4.php │ ├── autoload_classmap.php │ └── autoload_namespaces.php └── autoload.php ├── Procfile ├── site_config ├── standard │ ├── version.txt │ ├── getpocket.com.txt │ ├── feeds.feedblitz.com.txt │ ├── ganglia.info.txt │ ├── autocrypt.org.txt │ ├── index.php │ ├── what-if.xkcd.com.txt │ ├── lukew.com.txt │ ├── soundcity.tv.txt │ ├── crimemagazine.com.txt │ ├── ht.ly.txt │ ├── the-magazine.org.txt │ ├── pxlnv.com.txt │ ├── alex.mullr.net.txt │ ├── chareidi.org.txt │ ├── opensource.org.txt │ ├── snip.ly.txt │ ├── techmeme.com.txt │ ├── blog.fefe.de.txt │ ├── axesslab.com.txt │ ├── bitelia.com.txt │ ├── blog.spu.edu.txt │ ├── ecetia.com.txt │ ├── extracine.com.txt │ ├── help.fivefilters.org.txt │ ├── jjahnke.net.txt │ ├── kumailplus.com.txt │ ├── luxuo.com.txt │ ├── t3n.de.txt │ ├── unherd.com.txt │ ├── zhuanlan.zhihu.com.txt │ ├── facta.co.jp.txt │ ├── mattcutts.com.txt │ ├── meowni.ca.txt │ ├── motorfull.com.txt │ ├── roy.gbiv.com.txt │ ├── 512pixels.net.txt │ ├── apple.news.txt │ ├── elance.com.txt │ ├── getnews.jp.txt │ ├── jamesclear.com.txt │ ├── lostgarden.com.txt │ ├── mbl.is.txt │ ├── n.survol.fr.txt │ ├── acidcow.com.txt │ ├── alseraj.net.txt │ ├── blogs.gnome.org.txt │ ├── kachestvo.ru.txt │ ├── monkeyzen.com.txt │ ├── news.ycombinator.com.txt │ ├── paquier.xyz.txt │ ├── allafrica.com.txt │ ├── carlchenet.com.txt │ ├── doc.wallabag.org.txt │ ├── eternabuenosaires.com.txt │ ├── fakirpresse.info.txt │ ├── fivefilters.org.txt │ ├── fok.nl.txt │ ├── gist.github.com.txt │ ├── gsmarena.com.txt │ ├── indehekken.net.txt │ ├── macdrifter.com.txt │ ├── marksdailyapple.com.txt │ ├── osmand.net.txt │ ├── summitroute.com.txt │ ├── zerokspot.com.txt │ ├── blog.imirhil.fr.txt │ ├── dagogtid.no.txt │ ├── dailydot.com.txt │ ├── drdobbs.com.txt │ ├── flyingmachinestudios.com.txt │ ├── gizmovil.com.txt │ ├── goodfil.ms.txt │ ├── je-suis-papa.com.txt │ ├── staltz.com.txt │ ├── stjv.fr.txt │ ├── appleweblog.com.txt │ ├── dropbox.com.txt │ ├── gihyo.jp.txt │ ├── hiperpop.com.txt │ ├── jungle-world.com.txt │ ├── krone.at.txt │ ├── panic.com.txt │ ├── tofugu.com.txt │ ├── urbandictionary.com.txt │ ├── vg.no.txt │ ├── warnerbros.fr.txt │ ├── zataz.com.txt │ ├── LICENSE.txt │ ├── altfoto.com.txt │ ├── blogs.forbes.com.txt │ ├── brucelawson.co.uk.txt │ ├── cucharasonica.com.txt │ ├── dansdata.com.txt │ ├── hipertextual.com.txt │ ├── interviewmagazine.com.txt │ ├── jvt.me.txt │ ├── phototrend.fr.txt │ ├── portertech.ca.txt │ ├── renenekuda.cz.txt │ ├── usccb.org.txt │ ├── crn.de.txt │ ├── eff.org.txt │ ├── finexpert.e15.cz.txt │ ├── marigold.cz.txt │ ├── news.rub.de.txt │ ├── news.techmeme.com.txt │ ├── pentaxforums.com.txt │ ├── pieria.co.uk.txt │ ├── tldp.org.txt │ ├── gizmologia.com.txt │ ├── happyassassin.net.txt │ ├── iansommerville.com.txt │ ├── iplaysoft.com.txt │ ├── lefilrouge.media.txt │ ├── lupa.cz.txt │ ├── matt.might.net.txt │ ├── monkeyuser.com.txt │ ├── radionz.co.nz.txt │ ├── sivers.org.txt │ ├── thenews.coop.txt │ ├── turnoff.us.txt │ ├── utdailybeacon.com.txt │ ├── worldwidewords.org.txt │ ├── betabeat.com.txt │ ├── cashless.pl.txt │ ├── davidwalsh.name.txt │ ├── democracynow.org.txt │ ├── explosm.net.txt │ ├── freelancer.com.txt │ ├── gauchiste.fr.txt │ ├── hs.fi.txt │ ├── izismile.com.txt │ ├── jobbank.gc.ca.txt │ ├── mein-mmo.de.txt │ ├── mytotalretail.com.txt │ ├── nplusonemag.com.txt │ ├── signal.org.txt │ ├── stumbleupon.com.txt │ ├── www3.imperial.ac.uk.txt │ ├── cooper.com.txt │ ├── hackertarget.com.txt │ ├── juliareda.eu.txt │ ├── kresus.org.txt │ ├── letraslibres.com.txt │ ├── longform.org.txt │ ├── oschina.net.txt │ ├── php.net.txt │ ├── alexduner.com.txt │ ├── archdaily.com.txt │ ├── berlingske.dk.txt │ ├── codeproject.com.txt │ ├── doughellmann.com.txt │ ├── esglobal.org.txt │ ├── fokus.se.txt │ ├── geenstijl.nl.txt │ ├── gold.ac.uk.txt │ ├── ianlewis.org.txt │ ├── nextcloud.com.txt │ ├── sme.sk.txt │ ├── sourcebooks.com.txt │ ├── sprengsatz.de.txt │ ├── thefilmexperience.net.txt │ ├── triplebyte.com.txt │ ├── vivirmexico.com.txt │ ├── w3.org.txt │ ├── www.seriouseats.com.txt │ ├── a11ywithlindsey.com.txt │ ├── blog.eng.xogrp.com.txt │ ├── blog.nightly.mozilla.org.txt │ ├── caffereggio.net.txt │ ├── economie.gouv.fr.txt │ ├── elblogsalmon.com.txt │ ├── lvsl.fr.txt │ ├── marriedtothesea.com.txt │ ├── pastepad.fivefilters.org.txt │ ├── pixellibre.net.txt │ ├── pmf.silvrback.com.txt │ ├── pogue.blogs.nytimes.com.txt │ ├── queerty.com.txt │ ├── theses.enc.sorbonne.fr.txt │ ├── tidbits.com.txt │ ├── viget.com.txt │ ├── yourerie.com.txt │ ├── 9to5mac.com.txt │ ├── cn.engadget.com.txt │ ├── dictionary.reference.com.txt │ ├── ericsuh.com.txt │ ├── f-droid.org.txt │ ├── github.blog.txt │ ├── grumpygamer.com.txt │ ├── hmercer.com.txt │ ├── laughingsquid.com.txt │ ├── lawfareblog.com.txt │ ├── osmc.tv.txt │ ├── radar.oreilly.com.txt │ ├── shahrsakhtafzar.com.txt │ ├── theoaklandpress.com.txt │ ├── thisamericanlife.org.txt │ ├── zaknrw.de.txt │ ├── blogs.reuters.com.txt │ ├── crimethinc.com.txt │ ├── good.is.txt │ ├── highscalability.com.txt │ ├── marketresearchdirect.com.txt │ ├── prolost.com.txt │ ├── redalemeden.com.txt │ ├── techno-science.net.txt │ ├── tthfanfic.org.txt │ ├── web-libre.org.txt │ ├── americanthinker.com.txt │ ├── autoactu.com.txt │ ├── bobbyromeo.com.txt │ ├── code.google.com.txt │ ├── historic-uk.com.txt │ ├── mebedo.de.txt │ ├── mises.org.txt │ ├── pymotw.com.txt │ ├── saltyworld.net.txt │ ├── sfweekly.com.txt │ ├── share.ez.no.txt │ ├── sqlite.org.txt │ ├── varsity.co.uk.txt │ ├── brandingstrategyinsider.com.txt │ ├── brettterpstra.com.txt │ ├── fair.org.txt │ ├── ft.com.txt │ ├── iphonetweak.fr.txt │ ├── m.theregister.co.uk.txt │ ├── motorcyclistonline.com.txt │ ├── publications.parliament.uk.txt │ ├── robertsspaceindustries.com.txt │ ├── scottohara.me.txt │ ├── spectrum.ieee.org.txt │ ├── splitsider.com.txt │ ├── theintercept.com.txt │ ├── therumpus.net.txt │ ├── will-self.com.txt │ ├── 5by5.tv.txt │ ├── bez.es.txt │ ├── catb.org.txt │ ├── hanselman.com.txt │ ├── houstonchronicle.com.txt │ ├── indiehackers.com.txt │ ├── instagr.am.txt │ ├── kathimerini.gr.txt │ ├── labs.mwrinfosecurity.com.txt │ ├── luminous-landscape.com.txt │ ├── mobilenet.cz.txt │ ├── muycomputerpro.com.txt │ ├── okfn.de.txt │ ├── scnsrc.me.txt │ ├── singularityhub.com.txt │ ├── smarthomebeginner.com.txt │ ├── tabletmag.com.txt │ ├── vedomosti.ru.txt │ ├── zoomit.ir.txt │ ├── 43folders.com.txt │ ├── bernama.com.txt │ ├── brooksreview.net.txt │ ├── chomsky.info.txt │ ├── ciperchile.cl.txt │ ├── dcurt.is.txt │ ├── developers.facebook.com.txt │ ├── elmalpensante.com.txt │ ├── indiatimes.com.txt │ ├── openthemagazine.com.txt │ ├── plzkthxbai.com.txt │ ├── scinfolex.com.txt │ ├── toolsandtoys.net.txt │ ├── book.douban.com.txt │ ├── delong.typepad.com.txt │ ├── gameswirtschaft.de.txt │ ├── groups.drupal.org.txt │ ├── informationclearinghouse.info.txt │ ├── itavisen.no.txt │ ├── keyboardmag.com.txt │ ├── kingarthurflour.com.txt │ ├── lifehack.org.txt │ ├── m.xkcd.com.txt │ ├── nj.com.txt │ ├── onlinewelten.com.txt │ ├── parliament.uk.txt │ ├── pinterest.com.txt │ ├── pittsburghpanthers.com.txt │ ├── spin.com.txt │ ├── thethaovanhoa.vn.txt │ ├── trailer.web-view.net.txt │ ├── xlsemanal.com.txt │ ├── amptoons.com.txt │ ├── dadall.info.txt │ ├── linuxnix.com.txt │ ├── mactechnews.de.txt │ ├── marcvidal.net.txt │ ├── moo.nac.uci.edu.txt │ ├── pjmedia.com.txt │ ├── real.gr.txt │ ├── steamcommunity.com.txt │ ├── thenetworkgarden.blogs.com.txt │ ├── thepointmag.com.txt │ ├── ux.artu.tv.txt │ ├── arduino-tutorial.de.txt │ ├── ascarter.net.txt │ ├── buquad.com.txt │ ├── hazlitt.net.txt │ ├── health.com.txt │ ├── histoire-filante.fr.txt │ ├── mesec.cz.txt │ ├── pandodaily.com.txt │ ├── philstar.com.txt │ ├── racjonalista.pl.txt │ ├── rom-game.fr.txt │ ├── tbray.org.txt │ ├── 24a11y.com.txt │ ├── deia.com.txt │ ├── halo.bungie.org.txt │ ├── jandan.net.txt │ ├── singaporeanstocksinvestor.blogspot.com.txt │ ├── slog.thestranger.com.txt │ ├── utiliser-lightroom.com.txt │ ├── www2.cnrs.fr.txt │ ├── cars.com.txt │ ├── gocomics.com.txt │ ├── gurusblog.com.txt │ ├── isource.com.txt │ ├── maitre-eolas.fr.txt │ ├── protonmail.com.txt │ ├── roughtype.com.txt │ ├── ruhlman.com.txt │ ├── wordpress.org.txt │ ├── wz-newsline.de.txt │ ├── 37signals.com.txt │ ├── 7newsbelize.com.txt │ ├── bjango.com.txt │ ├── blog.chriszacharias.com.txt │ ├── blog.twitter.com.txt │ ├── blog.wells.ee.txt │ ├── borderhouseblog.com.txt │ ├── business2community.com.txt │ ├── communities-dominate.blogs.com.txt │ ├── hespress.com.txt │ ├── hometheaterreview.com.txt │ ├── jacobinmag.com.txt │ ├── linux-magazin.de.txt │ ├── manga-news.com.txt │ ├── maxim.com.txt │ ├── nosalty.hu.txt │ ├── palmbeachpost.com.txt │ ├── propakistani.pk.txt │ ├── retractionwatch.com.txt │ ├── shifteleven.com.txt │ ├── techpinions.com.txt │ ├── thewirecutter.com.txt │ ├── tuaw.com.txt │ ├── wpmayor.com.txt │ ├── alsacreations.com.txt │ ├── ciaosamin.com.txt │ ├── cjr.org.txt │ ├── filamentgroup.com.txt │ ├── geeksofdoom.com.txt │ ├── higcapital.com.txt │ ├── mforum.cari.com.my.txt │ ├── mikeash.com.txt │ ├── neunetz.com.txt │ ├── omicrono.elespanol.com.txt │ ├── pharmazeutische-zeitung.de.txt │ ├── readability.com.txt │ ├── robots.thoughtbot.com.txt │ ├── seattletransitblog.com.txt │ ├── thinkprogress.org.txt │ ├── voltairenet.org.txt │ ├── basicthinking.de.txt │ ├── blog.eleven-labs.com.txt │ ├── diagonalperiodico.net.txt │ ├── ecranlarge.com.txt │ ├── heise-online.mobi.txt │ ├── humantransit.org.txt │ ├── inhabitat.com.txt │ ├── inquirer.com.txt │ ├── menshealth.com.sg.txt │ ├── rasgolatente.es.txt │ ├── sayidaty.net.txt │ ├── squashed.tumblr.com.txt │ ├── theage.com.au.txt │ ├── tijd.be.txt │ ├── toolinux.com.txt │ ├── adslzone.net.txt │ ├── blog.pinboard.in.txt │ ├── boundlessline.org.txt │ ├── engineering.tumblr.com.txt │ ├── gurumed.org.txt │ ├── hvg.hu.txt │ ├── papodehomem.com.br.txt │ ├── quantamagazine.org.txt │ ├── rachelandrew.co.uk.txt │ ├── rezeptwelt.de.txt │ ├── techcommunity.microsoft.com.txt │ ├── devlinsangle.blogspot.co.at.txt │ ├── glazman.org.txt │ ├── gnppn.fr.txt │ ├── hiphopleeft.nl.txt │ ├── icannabis.tumblr.com.txt │ ├── ilyabirman.ru.txt │ ├── itmedia.co.jp.txt │ ├── lehollandaisvolant.net.txt │ ├── nojesguiden.se.txt │ ├── rpgsite.net.txt │ ├── useit.com.txt │ ├── vitispr.com.txt │ ├── yostivanich.com.txt │ ├── 24.ae.txt │ ├── acroswing.fr.txt │ ├── gawker.com.txt │ ├── huffingtonpost.fr.txt │ ├── inessential.com.txt │ ├── omiliya.org.txt │ ├── popehat.com.txt │ ├── timeshighereducation.com.txt │ ├── accesstoinsight.org.txt │ ├── addendum.org.txt │ ├── americandrink.net.txt │ ├── blog.dropbox.com.txt │ ├── da.feedsportal.com.txt │ ├── ebay.com.txt │ ├── elmundo.es.txt │ ├── firstthings.com.txt │ ├── grafikart.fr.txt │ ├── itwire.com.txt │ ├── lezephyrmag.com.txt │ ├── libcom.org.txt │ ├── mintpressnews.com.txt │ ├── msdn.microsoft.com.txt │ ├── n0where.net.txt │ ├── pcmag.com.txt │ ├── politifact.com.txt │ ├── searchenginejournal.com.txt │ ├── telerama.fr.txt │ ├── thedrive.com.txt │ ├── timeshighereducation.co.uk.txt │ ├── wiki.guildwars.com.txt │ ├── 36kr.com.txt │ ├── annouchka.fr.txt │ ├── folklore.org.txt │ ├── gizmodo.uol.com.br.txt │ ├── haberler.com.txt │ ├── linuxjournal.com.txt │ ├── marco.org.txt │ ├── medialens.org.txt │ ├── nakedsecurity.sophos.com.txt │ ├── pastebin.com.txt │ ├── resume.se.txt │ ├── splinternews.com.txt │ ├── thesimpledollar.com.txt │ ├── wmpoweruser.com.txt │ ├── albayan.ae.txt │ ├── amandala.com.bz.txt │ ├── annatravelling.wordpress.com.txt │ ├── blog.kaelig.fr.txt │ ├── econlog.econlib.org.txt │ ├── ekultura.hu.txt │ ├── franceculture.fr.txt │ ├── gizmodo.co.uk.txt │ ├── gofugyourself.com.txt │ ├── healthletter.mayoclinic.com.txt │ ├── kottke.org.txt │ ├── netzpolitik.org.txt │ ├── phoronix.com.txt │ ├── usfirst.org.txt │ ├── warriordudimanche.net.txt │ ├── wiki.guildwars2.com.txt │ ├── aclu.org.txt │ ├── actualitte.com.txt │ ├── blog.mozilla.org.txt │ ├── coffeecircle.com.txt │ ├── csswizardry.com.txt │ ├── digital-photography-school.com.txt │ ├── enikos.gr.txt │ ├── fictionpress.com.txt │ ├── framablog.org.txt │ ├── generation-nt.com.txt │ ├── information.dk.txt │ ├── itsfoss.com.txt │ ├── kickstarter.com.txt │ ├── linux.com.txt │ ├── marmiton.org.txt │ ├── mentalfloss.com.txt │ ├── mirrorfootball.co.uk.txt │ ├── n-tv.de.txt │ ├── outsideonline.com.txt │ ├── parislemon.com.txt │ ├── venturebeat.com.txt │ ├── watoday.com.au.txt │ ├── alternet.org.txt │ ├── android-developers.googleblog.com.txt │ ├── autoblog.com.txt │ ├── gizmodo.com.txt │ ├── globalgrind.com.txt │ ├── hacks.mozilla.org.txt │ ├── jalopnik.com.txt │ ├── kotaku.com.txt │ ├── mlssoccer.com.txt │ ├── news.rambler.ru.txt │ ├── theinventory.com.txt │ ├── tvtropes.org.txt │ ├── whatever.scalzi.com.txt │ ├── aftenposten.no.txt │ ├── andy-bell.design.txt │ ├── asymco.com.txt │ ├── chrisltd.com.txt │ ├── christies.com.txt │ ├── css-tricks.com.txt │ ├── deadspin.com.txt │ ├── dev.to.txt │ ├── fivethirtyeight.com.txt │ ├── globalresearch.ca.txt │ ├── lifestyle.inquirer.net.txt │ ├── thetakeout.com.txt │ ├── blog.naver.com.txt │ ├── blogs.hbr.org.txt │ ├── computerworld.dk.txt │ ├── core77.com.txt │ ├── jezebel.com.txt │ ├── juppy.org.txt │ ├── mobile.lemondeinformatique.fr.txt │ ├── netzoekonom.de.txt │ ├── ontologicalgeek.com.txt │ ├── preparedfoods.com.txt │ ├── sydsvenskan.se.txt │ ├── techstage.de.txt │ ├── theonion.com.txt │ ├── wow.joystiq.com.txt │ ├── boagworld.com.txt │ ├── eleconomista.es.txt │ ├── idlewords.com.txt │ ├── mactalk.com.au.txt │ ├── newrepublic.com.txt │ ├── pathawks.com.txt │ ├── recode.net.txt │ ├── straightdope.com.txt │ ├── thrillist.com.txt │ ├── tomshardware.com.txt │ ├── trailerzone.de.txt │ ├── uefa.com.txt │ ├── washingtoninstitute.org.txt │ ├── worldpoultry.net.txt │ ├── xatakaciencia.com.txt │ ├── americastestkitchenfeed.com.txt │ ├── antirez.com.txt │ ├── blog.instagram.com.txt │ ├── blog.pchome.net.txt │ ├── daringfireball.net.txt │ ├── estadao.com.br.txt │ ├── gatopardo.com.txt │ ├── hackmake.org.txt │ ├── hurriyet.com.tr.txt │ ├── martinfowler.com.txt │ ├── nbnnews.com.au.txt │ ├── net-security.org.txt │ ├── rodrigo.sharpcube.com.txt │ ├── thegreatdiscontent.com.txt │ ├── theroot.com.txt │ ├── tomdispatch.com.txt │ ├── brainpickings.org.txt │ ├── champeau.info.txt │ ├── fanfiction.net.txt │ ├── iphonehacks.com.txt │ ├── news.com.au.txt │ ├── nrc.nl.txt │ ├── ntoskrnl.org.txt │ ├── planetvita.de.txt │ ├── puri.sm.txt │ ├── winfuture.de.txt │ ├── 24ways.org.txt │ ├── abc.es.txt │ ├── aws.amazon.com.txt │ ├── pittsburghmagazine.com.txt │ ├── privacyinternational.org.txt │ ├── prog21.dadgum.com.txt │ ├── scotusblog.com.txt │ ├── smbc-comics.com.txt │ ├── songshuhui.net.txt │ ├── trailers.apple.com.txt │ ├── cmswire.com.txt │ ├── habrahabr.ru.txt │ ├── hosted.ap.org.txt │ ├── ipadclub.nl.txt │ └── kenrockwell.com.txt ├── index.php ├── custom │ └── index.php └── README.txt ├── cache ├── index.php ├── rss │ └── index.php ├── rss-with-key │ └── index.php └── urls │ └── index.php ├── robots.txt ├── images └── agplv3.png ├── libraries ├── DisableSimplePieSanitize.php └── html5php │ └── HTML5 │ ├── Exception.php │ └── Parser │ └── ParseError.php └── app.json /.gitignore: -------------------------------------------------------------------------------- 1 | custom_config.php -------------------------------------------------------------------------------- /_config.yml: -------------------------------------------------------------------------------- 1 | theme: jekyll-theme-cayman -------------------------------------------------------------------------------- /vendor/composer/installed.json: -------------------------------------------------------------------------------- 1 | [] 2 | -------------------------------------------------------------------------------- /Procfile: -------------------------------------------------------------------------------- 1 | web: vendor/bin/heroku-php-nginx 2 | -------------------------------------------------------------------------------- /site_config/standard/version.txt: -------------------------------------------------------------------------------- 1 | 2014-05-05T08:36:15Z -------------------------------------------------------------------------------- /site_config/standard/getpocket.com.txt: -------------------------------------------------------------------------------- 1 | http_header(user-agent): PHP/5.3 -------------------------------------------------------------------------------- /cache/index.php: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /site_config/custom/index.php: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /site_config/standard/what-if.xkcd.com.txt: -------------------------------------------------------------------------------- 1 | autodetect_next_page: no 2 | test_url: http://what-if.xkcd.com/1/ -------------------------------------------------------------------------------- /site_config/standard/lukew.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | 3 | test_url: https://www.lukew.com/ff/entry.asp?1995 4 | -------------------------------------------------------------------------------- /site_config/standard/soundcity.tv.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: sharing 2 | 3 | test_url: http://soundcity.tv/feed/ 4 | -------------------------------------------------------------------------------- /site_config/standard/crimemagazine.com.txt: -------------------------------------------------------------------------------- 1 | autodetect_next_page: no 2 | test_url: http://www.crimemagazine.com/son-sam -------------------------------------------------------------------------------- /site_config/standard/ht.ly.txt: -------------------------------------------------------------------------------- 1 | single_page_link: //iframe[@id='hootFrame']/@src 2 | 3 | test_url: http://ht.ly/bOiZV -------------------------------------------------------------------------------- /site_config/standard/the-magazine.org.txt: -------------------------------------------------------------------------------- 1 | tidy: no 2 | 3 | test_url: http://the-magazine.org/1/alone-together-again -------------------------------------------------------------------------------- /site_config/standard/pxlnv.com.txt: -------------------------------------------------------------------------------- 1 | date: //main//time/@datetime 2 | 3 | test_url: https://pxlnv.com/blog/bullshit-web/ 4 | -------------------------------------------------------------------------------- /site_config/standard/alex.mullr.net.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class="entry"] 2 | test_url: http://alex.mullr.net/blog/2011/05/on-spotify/ -------------------------------------------------------------------------------- /site_config/standard/chareidi.org.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | test_url: http://www.chareidi.org/archives5772/tetzaveh/TZV72adraft.htm -------------------------------------------------------------------------------- /site_config/standard/opensource.org.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='content clear-block'] 2 | test_url: http://opensource.org/node/537 -------------------------------------------------------------------------------- /site_config/standard/snip.ly.txt: -------------------------------------------------------------------------------- 1 | single_page_link: //meta[@property="og:url"]/@content 2 | 3 | test_url: http://snip.ly/qa1R -------------------------------------------------------------------------------- /site_config/standard/techmeme.com.txt: -------------------------------------------------------------------------------- 1 | single_page_link_in_feed: //b/a 2 | 3 | test_url: http://www.techmeme.com/feed.xml 4 | -------------------------------------------------------------------------------- /site_config/standard/blog.fefe.de.txt: -------------------------------------------------------------------------------- 1 | title: //h2 2 | date: //h3 3 | body: //ul 4 | 5 | test_url: http://blog.fefe.de/?ts=b063bf55 -------------------------------------------------------------------------------- /site_config/standard/axesslab.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@id="main-content"] 2 | 3 | test_url: https://axesslab.com/disabled-buttons-suck/ 4 | -------------------------------------------------------------------------------- /site_config/standard/bitelia.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://bitelia.com/2011/09/klout-midiendo-influencia -------------------------------------------------------------------------------- /site_config/standard/blog.spu.edu.txt: -------------------------------------------------------------------------------- 1 | body://div[@class='post'] 2 | test_url: http://blog.spu.edu/lectio/from-the-frying-pan-into-the-fire/ -------------------------------------------------------------------------------- /site_config/standard/ecetia.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://ecetia.com/2011/09/vida-de-jugon-vii-las-tres-es -------------------------------------------------------------------------------- /site_config/standard/extracine.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://extracine.com/2011/09/straw-dogs-la-original -------------------------------------------------------------------------------- /site_config/standard/help.fivefilters.org.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class="title"]/h3 2 | date: substring-after(//div[@class="meta"], ": ") 3 | -------------------------------------------------------------------------------- /site_config/standard/jjahnke.net.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='entry'] 2 | prune: no 3 | 4 | test_url: http://www.jjahnke.net/rundbr87.html#2514 -------------------------------------------------------------------------------- /site_config/standard/kumailplus.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class = "entry-full"] 2 | 3 | test_url: http://www.kumailplus.com/2011/12/02/24308 -------------------------------------------------------------------------------- /site_config/standard/luxuo.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='post-content'] 2 | prune: no 3 | 4 | test_url: http://www.luxuo.com/watches/feed -------------------------------------------------------------------------------- /site_config/standard/t3n.de.txt: -------------------------------------------------------------------------------- 1 | next_page_link: //link[@rel='next'] 2 | 3 | test_url: http://t3n.de/news/zukunftstag-2017-t3n-818436/ 4 | -------------------------------------------------------------------------------- /site_config/standard/unherd.com.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: ref-ar 2 | 3 | test_url: https://unherd.com/2019/09/labours-brexit-shambles/ 4 | -------------------------------------------------------------------------------- /site_config/standard/zhuanlan.zhihu.com.txt: -------------------------------------------------------------------------------- 1 | http_header(user-agent): PHP/7.2 2 | 3 | test_url: https://zhuanlan.zhihu.com/p/67184988 4 | -------------------------------------------------------------------------------- /site_config/standard/facta.co.jp.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='content'] 2 | 3 | test_url: http://facta.co.jp/blog/archives/20111026001026.html 4 | -------------------------------------------------------------------------------- /site_config/standard/mattcutts.com.txt: -------------------------------------------------------------------------------- 1 | date: //*[@class = 'published'] 2 | test_url: http://www.mattcutts.com/blog/internet-censorship-sopa/ -------------------------------------------------------------------------------- /site_config/standard/meowni.ca.txt: -------------------------------------------------------------------------------- 1 | author: //meta[@name="author"]/@content 2 | 3 | test_url: https://meowni.ca/posts/2017-puppeteer-tests/ 4 | -------------------------------------------------------------------------------- /site_config/standard/motorfull.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://motorfull.com/2011/09/aparca-valeo-park4u-remote -------------------------------------------------------------------------------- /site_config/standard/roy.gbiv.com.txt: -------------------------------------------------------------------------------- 1 | strip_comments: no 2 | test_url: http://roy.gbiv.com/untangled/2008/rest-apis-must-be-hypertext-driven -------------------------------------------------------------------------------- /site_config/standard/512pixels.net.txt: -------------------------------------------------------------------------------- 1 | title: //meta[@property='og:title']/@content 2 | test_url: http://www.512pixels.net/blog/2014/10/the-move 3 | -------------------------------------------------------------------------------- /site_config/standard/apple.news.txt: -------------------------------------------------------------------------------- 1 | single_page_link: //p//a[contains(., 'Click here')] 2 | test_url: https://apple.news/AHQREjzH0Ts6iikKhNe6o8w 3 | -------------------------------------------------------------------------------- /site_config/standard/elance.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='jobDesc-bd']/p 2 | 3 | test_url: http://www.elance.com/j/xml-technical-intergration/23687172/ -------------------------------------------------------------------------------- /site_config/standard/getnews.jp.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='post'] 2 | strip: //ul[@id='bookmark_single'] 3 | test_url: http://getnews.jp/archives/117312 -------------------------------------------------------------------------------- /site_config/standard/jamesclear.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, 'entry-content')] 2 | 3 | test_url: https://jamesclear.com/procrastination 4 | -------------------------------------------------------------------------------- /site_config/standard/lostgarden.com.txt: -------------------------------------------------------------------------------- 1 | prune: no 2 | convert_double_br_tags: yes 3 | test_url: http://www.lostgarden.com/2012/04/loops-and-arcs.html -------------------------------------------------------------------------------- /site_config/standard/mbl.is.txt: -------------------------------------------------------------------------------- 1 | body: //div[class="frett-main"] 2 | test_url: http://mbl.is/frettir/innlent/2012/02/21/litill_munur_a_fargjaldaverdi/ -------------------------------------------------------------------------------- /site_config/standard/n.survol.fr.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | date: //header//time/@datetime 3 | 4 | test_url: https://n.survol.fr/n/gerer-son-potager 5 | -------------------------------------------------------------------------------- /site_config/standard/acidcow.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[starts-with(@id, 'news-id-')] 2 | 3 | test_url: http://acidcow.com/fun/20933-acid-picdump-83-pics.html -------------------------------------------------------------------------------- /site_config/standard/alseraj.net.txt: -------------------------------------------------------------------------------- 1 | title: //*[@id='normalfontyellow'] 2 | test_url: http://www.alseraj.net/cgi-bin/pros/av/LeqaTextDisplay.cgi?display&2 -------------------------------------------------------------------------------- /site_config/standard/blogs.gnome.org.txt: -------------------------------------------------------------------------------- 1 | http_header(user-agent): PHP/7.2 2 | 3 | test_url: https://blogs.gnome.org/aday/2017/08/08/the-gnome-way/ 4 | -------------------------------------------------------------------------------- /site_config/standard/kachestvo.ru.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, 'inner_content')] 2 | 3 | test_url: http://kachestvo.ru/promtovar/odezhda/denim.html -------------------------------------------------------------------------------- /site_config/standard/monkeyzen.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://monkeyzen.com/2011/09/siluetas-de-clasicos-a-modo-de-vinilos -------------------------------------------------------------------------------- /site_config/standard/news.ycombinator.com.txt: -------------------------------------------------------------------------------- 1 | strip_comments: no 2 | strip: //a[. = 'reply'] 3 | test_url: http://news.ycombinator.com/item?id=1516461 -------------------------------------------------------------------------------- /site_config/standard/paquier.xyz.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='post'] 2 | 3 | test_url: http://paquier.xyz/postgresql-2/postgres-10-incompatible-changes/ 4 | -------------------------------------------------------------------------------- /site_config/standard/allafrica.com.txt: -------------------------------------------------------------------------------- 1 | http_header(user-agent): PHP/7.0 2 | 3 | test_url: http://allafrica.com/tools/headlines/rdf/latest/headlines.rdf 4 | -------------------------------------------------------------------------------- /site_config/standard/carlchenet.com.txt: -------------------------------------------------------------------------------- 1 | date: //time/@datetime 2 | 3 | test_url: https://carlchenet.com/foss-passive-consumerism-kills-our-community/ 4 | -------------------------------------------------------------------------------- /site_config/standard/doc.wallabag.org.txt: -------------------------------------------------------------------------------- 1 | body: //section[contains(@class, 'normal')] 2 | 3 | test_url: https://doc.wallabag.org/en/user/filters.html 4 | -------------------------------------------------------------------------------- /site_config/standard/eternabuenosaires.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://eternabuenosaires.com/2011/09/calle-adolfo-bioy-casares -------------------------------------------------------------------------------- /site_config/standard/fakirpresse.info.txt: -------------------------------------------------------------------------------- 1 | author: //a[@class="url fn spip_in"] 2 | 3 | test_url: https://www.fakirpresse.info/les-patrons-ca-osent-tout 4 | -------------------------------------------------------------------------------- /site_config/standard/fivefilters.org.txt: -------------------------------------------------------------------------------- 1 | body: //section[contains(@class, 'container')] 2 | prune: no 3 | 4 | test_url: http://fivefilters.org/kindle-it/ 5 | -------------------------------------------------------------------------------- /site_config/standard/fok.nl.txt: -------------------------------------------------------------------------------- 1 | # skip cookie warning 2 | single_page_link: concat(//form/@action, '?allowcookies=yes') 3 | 4 | test_url: http://fok.nl/687116 -------------------------------------------------------------------------------- /site_config/standard/gist.github.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class="highlight"]/pre 2 | 3 | prune: no 4 | tidy: no 5 | 6 | test_url: https://gist.github.com/1258908 -------------------------------------------------------------------------------- /site_config/standard/gsmarena.com.txt: -------------------------------------------------------------------------------- 1 | next_page_link: //a[@class='pages-next'] 2 | 3 | test_url: http://www.gsmarena.com/samsung_galaxy_j2-review-1348.php -------------------------------------------------------------------------------- /site_config/standard/indehekken.net.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='post-entry']/p 2 | 3 | test_url: http://www.indehekken.net/you-only-sing-when-youre-rowing/ 4 | -------------------------------------------------------------------------------- /site_config/standard/macdrifter.com.txt: -------------------------------------------------------------------------------- 1 | title: substring-before(//title,' « Macdrifter') 2 | test_url: http://www.macdrifter.com/2012/03/instacast-on-my-mac/ -------------------------------------------------------------------------------- /site_config/standard/marksdailyapple.com.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: wwsgd 2 | test_url: http://www.marksdailyapple.com/are-detoxes-and-cleanses-safe-and-effective/ -------------------------------------------------------------------------------- /site_config/standard/osmand.net.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class="article"]//h2 2 | date: //meta/@pubdate 3 | 4 | test_url: https://osmand.net/blog/guideline-pt 5 | -------------------------------------------------------------------------------- /site_config/standard/summitroute.com.txt: -------------------------------------------------------------------------------- 1 | prune: no 2 | 3 | test_url: https://summitroute.com/blog/2015/12/24/instagram_bounty_case_study_for_defense/ 4 | -------------------------------------------------------------------------------- /site_config/standard/zerokspot.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | body: //div[@id="primarycontent"] 3 | test_url: http://zerokspot.com/weblog/2011/06/26/europython2011/ -------------------------------------------------------------------------------- /site_config/standard/blog.imirhil.fr.txt: -------------------------------------------------------------------------------- 1 | date: //section[@id="post"]//h1//time 2 | 3 | test_url: https://blog.imirhil.fr/2019/11/13/first-party-tracker.html 4 | -------------------------------------------------------------------------------- /site_config/standard/dagogtid.no.txt: -------------------------------------------------------------------------------- 1 | title: //span[@class = 'overskriftEkstrastor'] 2 | author: //em/a 3 | 4 | test_url: http://dagogtid.no/nyhet.cfm?nyhetid=2414 -------------------------------------------------------------------------------- /site_config/standard/dailydot.com.txt: -------------------------------------------------------------------------------- 1 | tidy: no 2 | body: //article 3 | 4 | test_url: http://www.dailydot.com/entertainment/tumblr-christopher-price-topherchris/ -------------------------------------------------------------------------------- /site_config/standard/drdobbs.com.txt: -------------------------------------------------------------------------------- 1 | single_page_link: //a[contains(@href, '/article/print')] 2 | test_url: http://www.drdobbs.com/architecture-and-design/240001128 -------------------------------------------------------------------------------- /site_config/standard/flyingmachinestudios.com.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: linenos 2 | test_url: http://www.flyingmachinestudios.com/programming/whoops-dci-refactoring/ -------------------------------------------------------------------------------- /site_config/standard/gizmovil.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://gizmovil.com/2011/09/hipertextual-labs-receptor-bluetooth-nokia-bh-214 -------------------------------------------------------------------------------- /site_config/standard/goodfil.ms.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: gutter 2 | test_url: http://goodfil.ms/blog/posts/2012/08/13/angularjs-and-the-goodfilms-mobile-site-part-1/ -------------------------------------------------------------------------------- /site_config/standard/je-suis-papa.com.txt: -------------------------------------------------------------------------------- 1 | strip: //noscript 2 | 3 | test_url: http://www.je-suis-papa.com/pandacraft-kit-educatif-creatif-abonnement-12-ans/ 4 | -------------------------------------------------------------------------------- /site_config/standard/staltz.com.txt: -------------------------------------------------------------------------------- 1 | author: //h2[@class="name-title"] 2 | 3 | test_url: https://staltz.com/a-plan-to-rescue-the-web-from-the-internet.html 4 | -------------------------------------------------------------------------------- /site_config/standard/stjv.fr.txt: -------------------------------------------------------------------------------- 1 | date: //time[contains(@class, 'published')]/@datetime 2 | 3 | test_url: https://www.stjv.fr/2018/01/au-sujet-quantic-dream/ 4 | -------------------------------------------------------------------------------- /site_config/standard/appleweblog.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://appleweblog.com/2011/09/encontrada-vulnerabilidad-grave-en-skype-para-ios -------------------------------------------------------------------------------- /site_config/standard/dropbox.com.txt: -------------------------------------------------------------------------------- 1 | single_page_link: //a[@id='download_button_link'] 2 | 3 | test_url: https://www.dropbox.com/s/qmocfrco2t0d28o/Fluffbeast.docx 4 | -------------------------------------------------------------------------------- /site_config/standard/gihyo.jp.txt: -------------------------------------------------------------------------------- 1 | single_page_link: //p[@id='skip']//a[contains(@href, 'skip')] 2 | 3 | test_url: http://gihyo.jp/dev/serial/01/machine-learning/0010 -------------------------------------------------------------------------------- /site_config/standard/hiperpop.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://hiperpop.com/2011/09/marc-anthony-celebra-su-cumpleanos-con-jennifer-lopez -------------------------------------------------------------------------------- /site_config/standard/jungle-world.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | body: //div[contains(@class,'story')] 3 | test_url: http://jungle-world.com/artikel/2015/02/51207.html 4 | -------------------------------------------------------------------------------- /site_config/standard/krone.at.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' kmm-article-box ')] 2 | 3 | test_url: https://www.krone.at/1941895 4 | -------------------------------------------------------------------------------- /site_config/standard/panic.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='entry'] 2 | date: //h3[@class='postDate'] 3 | test_url: http://www.panic.com/blog/2011/07/panic-is-ready-for-lion/ -------------------------------------------------------------------------------- /site_config/standard/tofugu.com.txt: -------------------------------------------------------------------------------- 1 | body://div[@class='entry-content'] 2 | 3 | test_url: http://www.tofugu.com/2015/07/20/interview-with-toriena-japanese-chiptune/ 4 | -------------------------------------------------------------------------------- /site_config/standard/urbandictionary.com.txt: -------------------------------------------------------------------------------- 1 | title: //title 2 | body: //table[@id='entries'] 3 | test_url: http://www.urbandictionary.com/define.php?term=Grown-Ass 4 | -------------------------------------------------------------------------------- /site_config/standard/vg.no.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='artikkelspalte'] 2 | strip_id_or_class: 'breadcrumb' 3 | test_url: http://www.vg.no/spill/artikkel.php?artid=10003628 -------------------------------------------------------------------------------- /site_config/standard/warnerbros.fr.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class="article-body"] 2 | test_url: https://www.warnerbros.fr/articles/magic-mike-xxl-adam-rodriguez-portrait 3 | -------------------------------------------------------------------------------- /site_config/standard/zataz.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class="blog-title"] 2 | 3 | test_url: https://www.zataz.com/bientot-noel-securisons-le-prochain-pc-de-la-famille/ 4 | -------------------------------------------------------------------------------- /site_config/standard/LICENSE.txt: -------------------------------------------------------------------------------- 1 | These files are released to the public domain. 2 | 3 | See https://creativecommons.org/publicdomain/zero/1.0/ for more information. 4 | -------------------------------------------------------------------------------- /site_config/standard/altfoto.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://altfoto.com/2011/09/nikon-presenta-su-nuevo-sistema-nikon-1-y-dos-nuevas-camaras -------------------------------------------------------------------------------- /site_config/standard/blogs.forbes.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='entry'] 2 | test_url: http://blogs.forbes.com/adamhartung/2011/04/08/apple-is-better-managed-than-microsoft/ -------------------------------------------------------------------------------- /site_config/standard/brucelawson.co.uk.txt: -------------------------------------------------------------------------------- 1 | title: //h2[@itemprop="title"] 2 | 3 | test_url: https://www.brucelawson.co.uk/2018/the-practical-value-of-semantic-html/ 4 | -------------------------------------------------------------------------------- /site_config/standard/cucharasonica.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://cucharasonica.com/2011/09/queen-busca-candidatos-para-su-propia-banda-tributo -------------------------------------------------------------------------------- /site_config/standard/dansdata.com.txt: -------------------------------------------------------------------------------- 1 | autodetect_next_page: no 2 | tidy: no 3 | prune: no 4 | body: //div[@class='NoOverflow'] 5 | test_url: http://www.dansdata.com/gz129.htm -------------------------------------------------------------------------------- /site_config/standard/hipertextual.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://hipertextual.com/2011/09/banda-ancha-en-america-latina-insignificante 3 | -------------------------------------------------------------------------------- /site_config/standard/interviewmagazine.com.txt: -------------------------------------------------------------------------------- 1 | title: //title 2 | body: //div[contains(@class, 'block')] 3 | 4 | test_url: http://www.interviewmagazine.com/film/spike-jonze -------------------------------------------------------------------------------- /site_config/standard/jvt.me.txt: -------------------------------------------------------------------------------- 1 | author: //div[@class="post-details"]//a[contains(@class, "p-name")] 2 | 3 | test_url: https://www.jvt.me/posts/2019/10/20/indieweb-talk/ 4 | -------------------------------------------------------------------------------- /site_config/standard/phototrend.fr.txt: -------------------------------------------------------------------------------- 1 | author: //div[contains(@class, "vcard")] 2 | 3 | test_url: https://phototrend.fr/2019/04/mise-a-jour-luminar-3-1-0-accent-ai-2-0/ 4 | -------------------------------------------------------------------------------- /site_config/standard/portertech.ca.txt: -------------------------------------------------------------------------------- 1 | author: //*[(@class = "author")] 2 | date: //*[(@class = "date")] 3 | test_url: http://portertech.ca/2012/12/10/iac-morning-market/ -------------------------------------------------------------------------------- /site_config/standard/renenekuda.cz.txt: -------------------------------------------------------------------------------- 1 | title: //*[@class='entry-title'] 2 | body: //div[@class='entry-content'] 3 | test_url: http://www.renenekuda.cz/recept-na-produktivitu/ -------------------------------------------------------------------------------- /site_config/standard/usccb.org.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='CS_Element_maincontent'] 2 | 3 | tidy: no 4 | prune: no 5 | 6 | test_url: http://www.usccb.org/bible/readings/072412.cfm -------------------------------------------------------------------------------- /site_config/standard/crn.de.txt: -------------------------------------------------------------------------------- 1 | author: //p[contains(@class,'author')]/a 2 | date: //div[contains(@class,'date')] 3 | test_url: http://www.crn.de/netzwerke-tk/artikel-93103.html -------------------------------------------------------------------------------- /site_config/standard/eff.org.txt: -------------------------------------------------------------------------------- 1 | author: //meta[@name="author"]/@content 2 | 3 | test_url: https://www.eff.org/deeplinks/2018/02/john-perry-barlow-internet-pioneer-1947-2018 4 | -------------------------------------------------------------------------------- /site_config/standard/finexpert.e15.cz.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: article-linktoanother 2 | 3 | test_url: http://finexpert.e15.cz/budiz-teplo-eu-stedre-zadotuje-nejen-plynovy-kotel -------------------------------------------------------------------------------- /site_config/standard/marigold.cz.txt: -------------------------------------------------------------------------------- 1 | http_header(accept): */* 2 | 3 | test_url: https://www.marigold.cz/item/projektovy-manazer-je-v-cesku-sproste-slovo-ke-skode-projektu 4 | -------------------------------------------------------------------------------- /site_config/standard/news.rub.de.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' content-inner-inner-wrapper ')] 2 | 3 | http_header(Cookie): has_js=1 4 | -------------------------------------------------------------------------------- /site_config/standard/news.techmeme.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='main']/div[@class='item'] 2 | strip: //div[@class='right'] 3 | 4 | test_url: http://news.techmeme.com/110516/fh-rip -------------------------------------------------------------------------------- /site_config/standard/pentaxforums.com.txt: -------------------------------------------------------------------------------- 1 | next_page_link: //a[contains(., 'Next:')] 2 | test_url: http://www.pentaxforums.com/reviews/long-exposure-handhelds/introduction.html -------------------------------------------------------------------------------- /site_config/standard/pieria.co.uk.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='article'] 2 | 3 | test_url: http://www.pieria.co.uk/articles/need_a_fiscal_rule_george__try_get_the_economy_growing 4 | -------------------------------------------------------------------------------- /site_config/standard/tldp.org.txt: -------------------------------------------------------------------------------- 1 | title: //title 2 | body: //h2 | //p | //ul 3 | prune: no 4 | tidy: no 5 | 6 | test_url: http://www.tldp.org/HOWTO/Plug-and-Play-HOWTO-7.html -------------------------------------------------------------------------------- /site_config/standard/gizmologia.com.txt: -------------------------------------------------------------------------------- 1 | body: //*[(@class = "historia")] 2 | test_url: http://gizmologia.com/2011/09/amd-trinity-el-sucesor-de-llano-en-una-demostracion-muy-interesante -------------------------------------------------------------------------------- /site_config/standard/happyassassin.net.txt: -------------------------------------------------------------------------------- 1 | http_header(user-agent): PHP/7.2 2 | 3 | test_url: https://www.happyassassin.net/2014/01/25/uefi-boot-how-does-that-actually-work-then/ 4 | -------------------------------------------------------------------------------- /site_config/standard/iansommerville.com.txt: -------------------------------------------------------------------------------- 1 | http_header(user-agent): PHP/5.3 2 | 3 | test_url: http://iansommerville.com/blog/a-guide-to-scottish-delicacies-for-tgo-challengers/ 4 | -------------------------------------------------------------------------------- /site_config/standard/iplaysoft.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='content']//div[@class='entry-banner' or @class='entry-content'] 2 | test_url: http://www.iplaysoft.com/webbrowserpassview.html -------------------------------------------------------------------------------- /site_config/standard/lefilrouge.media.txt: -------------------------------------------------------------------------------- 1 | 2 | strip_id_or_class: the_champ_sharing_container 3 | 4 | test_url: https://lefilrouge.media/etats-unis-nevada-burning-man-festival/ 5 | -------------------------------------------------------------------------------- /site_config/standard/lupa.cz.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: promo-in-article 2 | 3 | test_url: http://www.lupa.cz/clanky/michal-zamec-parfums-za-dva-roky-presahneme-obrat-6-miliard-korun/ -------------------------------------------------------------------------------- /site_config/standard/matt.might.net.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | author: string("Matt Might") 3 | strip: //h1/following-sibling::div 4 | 5 | test_url: http://matt.might.net/articles/oo-cesk/ -------------------------------------------------------------------------------- /site_config/standard/monkeyuser.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ', normalize-space(@class), ' '), ' content ')] 2 | 3 | test_url: https://www.monkeyuser.com/2019/v-201/ 4 | -------------------------------------------------------------------------------- /site_config/standard/radionz.co.nz.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='body'] 2 | title: //div[@class='newsstory']/h2 3 | test_url: http://www.radionz.co.nz/news/stories/2010/07/18/12481029a86d -------------------------------------------------------------------------------- /site_config/standard/sivers.org.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='content']//article 2 | 3 | strip: //header 4 | strip: //footer 5 | 6 | prune: no 7 | 8 | test_url: https://sivers.org/exex 9 | -------------------------------------------------------------------------------- /site_config/standard/thenews.coop.txt: -------------------------------------------------------------------------------- 1 | body: //div[@itemprop='articleBody'] 2 | 3 | test_url: http://www.thenews.coop/98221/news/co-operatives/jeremy-corbyn-needs-co-op-movement/ 4 | -------------------------------------------------------------------------------- /site_config/standard/turnoff.us.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class='post-title'] 2 | body: //article[@class='post-content'] 3 | test_url: https://turnoff.us/geek/the-depressed-developer-13/ 4 | -------------------------------------------------------------------------------- /site_config/standard/utdailybeacon.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='blox-story-text'] 2 | test_url: http://www.utdailybeacon.com/news/article_ccf6d024-0f15-11e5-ae29-9f63598deb81.html 3 | -------------------------------------------------------------------------------- /site_config/standard/worldwidewords.org.txt: -------------------------------------------------------------------------------- 1 | title: //p[@id='content'] 2 | 3 | body: //div[@class='contentblock'] 4 | test_url: http://www.worldwidewords.org/weirdwords/ww-gro1.htm -------------------------------------------------------------------------------- /site_config/standard/betabeat.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class="entry-content"] 2 | test_url: http://www.betabeat.com/2011/07/04/sheryl-sandberg-breaks-through-silicon-valleys-boys-club-sort-of/ -------------------------------------------------------------------------------- /site_config/standard/cashless.pl.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' post-page-content ')] 2 | 3 | test_url: https://www.cashless.pl/5465-revolut-kryzys 4 | -------------------------------------------------------------------------------- /site_config/standard/davidwalsh.name.txt: -------------------------------------------------------------------------------- 1 | author: //article//span[@itemprop="name"] 2 | strip: //div[@class="article-block"] 3 | 4 | test_url: https://davidwalsh.name/optional-chaining 5 | -------------------------------------------------------------------------------- /site_config/standard/democracynow.org.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, 'blog_body')] 2 | 3 | prune: no 4 | 5 | test_url: http://www.democracynow.org/blog/2014/1/9/the_fbi_the_nsa_and_a -------------------------------------------------------------------------------- /site_config/standard/explosm.net.txt: -------------------------------------------------------------------------------- 1 | body: //img[@id='main-comic'] 2 | author: substring(//small[@class="author-credit-name"], 4) 3 | 4 | test_url: http://explosm.net/comics/3954/ 5 | -------------------------------------------------------------------------------- /site_config/standard/freelancer.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id="projectDetailsContent"]//td 2 | 3 | test_url: http://www.freelancer.com/projects/PHP-Website-Design/debug-Forum-website-code.html -------------------------------------------------------------------------------- /site_config/standard/gauchiste.fr.txt: -------------------------------------------------------------------------------- 1 | date: //meta[@name="date"]/@content 2 | 3 | test_url: https://gauchiste.fr/post/2019/11/08/Un-catadioptre-vous-sauvera-(peut-%C3%AAtre)-la-vie 4 | -------------------------------------------------------------------------------- /site_config/standard/hs.fi.txt: -------------------------------------------------------------------------------- 1 | prune: yes 2 | tidy: yes 3 | test_url: http://www.hs.fi/kotimaa/Teollisuushallin%20palo%20levitt%C3%A4%C3%A4%20vaarallista%20savua%20Tuusulassa/a1305571582405 -------------------------------------------------------------------------------- /site_config/standard/izismile.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[starts-with(@id, 'news-id-')] 2 | prune: no 3 | 4 | test_url: http://izismile.com/2011/06/13/uncanny_factoid_fashion_or_creepy_2_pics.html -------------------------------------------------------------------------------- /site_config/standard/jobbank.gc.ca.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='formatCont_en'] 2 | 3 | prune: no 4 | 5 | test_url: http://www.jobbank.gc.ca/detail-eng.aspx?Source=JobPosting&OrderNum=6397922 -------------------------------------------------------------------------------- /site_config/standard/mein-mmo.de.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='gp-entry-text'] 2 | strip: //aside 3 | 4 | test_url: http://mein-mmo.de/pokemon-go-15-staerksten-pokemon-der-2-generation/ 5 | -------------------------------------------------------------------------------- /site_config/standard/mytotalretail.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | body: //div[@itemprop="content"] 3 | 4 | test_url: https://www.mytotalretail.com/article/how-site-search-could-kill-amazon/ 5 | -------------------------------------------------------------------------------- /site_config/standard/nplusonemag.com.txt: -------------------------------------------------------------------------------- 1 | title: /html/body/div[3]/div/div/h1 2 | 3 | body: //*[@id="article-body"] 4 | 5 | 6 | test_url: http://nplusonemag.com/the-outskirts-of-progress -------------------------------------------------------------------------------- /site_config/standard/signal.org.txt: -------------------------------------------------------------------------------- 1 | title: //h2[@class="post-title"] 2 | author: //h3[@class="post-subtitle"]//a 3 | 4 | test_url: https://signal.org/blog/signal-private-group-system/ 5 | -------------------------------------------------------------------------------- /site_config/standard/stumbleupon.com.txt: -------------------------------------------------------------------------------- 1 | single_page_link: //iframe[@id='tb-stumble-frame']/@src 2 | 3 | test_url: http://www.stumbleupon.com/su/35V0wB/zouchmagazine.com/poetry-violet/ -------------------------------------------------------------------------------- /site_config/standard/www3.imperial.ac.uk.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: hidelabel 2 | test_url: http://www3.imperial.ac.uk/newsandeventspggrp/imperialcollege/newssummary/news_14-7-2010-15-53-18 -------------------------------------------------------------------------------- /libraries/DisableSimplePieSanitize.php: -------------------------------------------------------------------------------- 1 | ):

4 | 5 | prune: no 6 | 7 | test_url: http://www.sqlite.org/fileformat2.html -------------------------------------------------------------------------------- /site_config/standard/varsity.co.uk.txt: -------------------------------------------------------------------------------- 1 | # FB comments are inside an h2. Weird. Without this, the line 'Comments' is preserved by the text parser 2 | 3 | strip: //h2 4 | test_url: http://www.varsity.co.uk/reviews/2662 -------------------------------------------------------------------------------- /libraries/html5php/HTML5/Parser/ParseError.php: -------------------------------------------------------------------------------- 1 | '), '>') 2 | body: //div[@class='NewsArticleContent'] 3 | test_url: http://www.mactechnews.de/news/index/Apple-Pressekonferenz-zum-iPhone-4-147316.html -------------------------------------------------------------------------------- /site_config/standard/marcvidal.net.txt: -------------------------------------------------------------------------------- 1 | title: //div[contains(@class, 'post-title')] 2 | body: //article/div 3 | 4 | skip_json_ld: yes 5 | 6 | test_url: https://www.marcvidal.net/blog/2019/1/8/google-ya-es-un-banco-europeo-y-ahora-que 7 | -------------------------------------------------------------------------------- /site_config/standard/moo.nac.uci.edu.txt: -------------------------------------------------------------------------------- 1 | title: //div[@id='header']//h1[1] 2 | 3 | body: //div[@id='content'] 4 | 5 | strip_id_or_class: toc 6 | 7 | prune: no 8 | 9 | test_url: http://moo.nac.uci.edu/~hjm/HOWTO_move_data.html 10 | -------------------------------------------------------------------------------- /site_config/standard/pjmedia.com.txt: -------------------------------------------------------------------------------- 1 | find_string:display:none 2 | replace_string: .. 3 | 4 | single_page_link: //div[@class='single-page-button']//a 5 | 6 | test_url: https://pjmedia.com/eddriscoll/2016/08/31/tom-wolfe-kingdom-of-speech/ 7 | -------------------------------------------------------------------------------- /site_config/standard/real.gr.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, 'article-photo-wrapper')] 2 | prune: no 3 | 4 | test_url: http://www.real.gr/DefaultArthro.aspx?page=arthro&id=360962&catID=1 5 | test_contains: Επισήμως το αποψινό υπουργικό 6 | -------------------------------------------------------------------------------- /site_config/standard/steamcommunity.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, "announcement_body")] 2 | author: //a[@class="whiteLink"] 3 | 4 | test_url: https://steamcommunity.com/games/elitedangerous/announcements/detail/1711828601535410336 5 | -------------------------------------------------------------------------------- /site_config/standard/thenetworkgarden.blogs.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id="beta-inner"] 2 | title: //h3[@class="entry-header"] 3 | 4 | test_url: http://thenetworkgarden.blogs.com/weblog/2011/09/microsoft-metro-and-the-next-wave-in-computing.html -------------------------------------------------------------------------------- /site_config/standard/thepointmag.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='article'] 2 | strip_id_or_class: z-max 3 | strip_id_or_class: readLaterMenu 4 | 5 | test_url: https://thepointmag.com/2016/examined-life/can-liberal-education-save-the-sciences 6 | -------------------------------------------------------------------------------- /site_config/standard/ux.artu.tv.txt: -------------------------------------------------------------------------------- 1 | author: ("Arturo Toledo") 2 | title: //div[@class="post"]/h2 3 | body: //div[@class="entry"] 4 | 5 | # Remove Twitter button 6 | strip: //div[@class="entry"]/p[2]/a/img 7 | test_url: http://ux.artu.tv/?p=192 -------------------------------------------------------------------------------- /site_config/standard/arduino-tutorial.de.txt: -------------------------------------------------------------------------------- 1 | // Stripping of Crayon Syntax Highlighter in duplicate 2 | strip: //textarea[contains(@class, 'crayon-plain')] 3 | 4 | test_url: https://www.arduino-tutorial.de/ein-altes-handy-als-arduino-input/ 5 | -------------------------------------------------------------------------------- /site_config/standard/ascarter.net.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class='article_title'] 2 | author: //span[@class='author'] 3 | date: //h2[@class='dateline'] 4 | body: //div[@class='article_body'] 5 | test_url: http://ascarter.net/2012/02/20/enough-is-enough.html -------------------------------------------------------------------------------- /site_config/standard/buquad.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | author: //h2/a 3 | date: substring-after(//h2, '|') 4 | strip_id_or_class: 'attachment' 5 | strip: //h3 6 | 7 | body: //div[@class='entry'] 8 | test_url: http://buquad.com/2012/04/09/paul-ryan/ -------------------------------------------------------------------------------- /site_config/standard/hazlitt.net.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='article-wrap'] 2 | title: //h1[@class='article-title'] 3 | author: //div[@class='article-footer']//div[@class='author-name']/a 4 | 5 | test_url: https://hazlitt.net/longreads/real-lolita 6 | -------------------------------------------------------------------------------- /site_config/standard/health.com.txt: -------------------------------------------------------------------------------- 1 | http_header(cookie): euConsent=true; euConsentId=61c78ceb-c244-4016-b707-3c640ba09311 2 | 3 | test_url: https://www.health.com/news/cell-phone-elbow-new-ill-wired-age 4 | test_contains: As symptoms progress 5 | -------------------------------------------------------------------------------- /site_config/standard/histoire-filante.fr.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class="entry-title"] 2 | date: //header[@class="entry-header"]//time[@class="entry-date published"]/@datetime 3 | 4 | test_url: https://histoire-filante.fr/2019/09/05/notre-itineraire/ 5 | -------------------------------------------------------------------------------- /site_config/standard/mesec.cz.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='urs'] | //div[@itemprop='articleBody'] 2 | strip_id_or_class: promo-in-article 3 | 4 | test_url: http://www.mesec.cz/aktuality/ceske-drahy-pridaji-zamestnancum-jejich-mzdy-vzrostou-o-1-7/ 5 | -------------------------------------------------------------------------------- /site_config/standard/pandodaily.com.txt: -------------------------------------------------------------------------------- 1 | tidy: no 2 | body: //article 3 | date: //time/@datetime 4 | strip_id_or_class: sharedaddy 5 | test_url: http://pandodaily.com/2012/01/19/ibooks-author-is-not-going-to-hurt-publishers-it-might-even-help-them/ -------------------------------------------------------------------------------- /site_config/standard/philstar.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@property="content:encoded"] 2 | 3 | test_url: http://www.philstar.com/headlines/2017/03/06/1678561/samuel-martires-named-new-sc-justice 4 | test_url: http://www.philstar.com/rss/breakingnews 5 | -------------------------------------------------------------------------------- /site_config/standard/racjonalista.pl.txt: -------------------------------------------------------------------------------- 1 | author: /html/body/center/b 2 | date: /html/body/table/tr[2]/td/i 3 | single_page_link: //*[@id='oTxt']/table[3]/tr[2]/td/a[1] 4 | 5 | test_url: http://www.racjonalista.pl/kk.php/s,7214/q,Geneza.szubrawstwa -------------------------------------------------------------------------------- /site_config/standard/rom-game.fr.txt: -------------------------------------------------------------------------------- 1 | author: //div[@id="main"]//a[contains(@href, "/auteurs/")] 2 | date: //meta[@itemprop="datePublished"]/@content 3 | 4 | test_url: https://www.rom-game.fr/news/3371-Goldeneye+007+devient+Goldeneye+25.html 5 | -------------------------------------------------------------------------------- /site_config/standard/tbray.org.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='centercontent'] 2 | strip: //div[@id='rightcontent'] 3 | date: substring-before( //div[@id='cats'], '·') 4 | title: //h1 5 | test_url: http://www.tbray.org/ongoing/When/201x/2012/03/04/Mobile-Money -------------------------------------------------------------------------------- /site_config/standard/24a11y.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | author: substring-after(//div[contains(@class, "author-description")]//h2[@class="author-title"], 'About') 3 | 4 | test_url: https://www.24a11y.com/2018/web-components-still-need-to-be-accessible/ 5 | -------------------------------------------------------------------------------- /site_config/standard/deia.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='widget full_article'] 2 | strip: //div[@class='Herramientas'] 3 | 4 | test_url: http://www.deia.com/2015/10/03/politica/euskadi/el-envilecimiento-de-la-politica-y-de-los-medios-de-comunicacion 5 | -------------------------------------------------------------------------------- /site_config/standard/halo.bungie.org.txt: -------------------------------------------------------------------------------- 1 | title:substring-before(id("maincontent")/table, 'Posted') 2 | body:id("maincontent")/p 3 | # eventually convert linebreaks better 4 | 5 | test_url: http://halo.bungie.org/fanfic/?story=Delahunt0312112316071.html -------------------------------------------------------------------------------- /site_config/standard/jandan.net.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='content']//div[@class = 'post f'] 2 | strip_id_or_class: comment-big 3 | strip_id_or_class: avatar 4 | strip: //div[@class='time_s'] 5 | 6 | test_url: http://jandan.net/2011/04/03/iphone-5-sony.html -------------------------------------------------------------------------------- /site_config/standard/singaporeanstocksinvestor.blogspot.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='post-body'] 2 | strip: //div[@id='lws_0'] 3 | prune: no 4 | 5 | test_url: http://singaporeanstocksinvestor.blogspot.com/2011/04/aims-amp-capital-industrial-reit.html -------------------------------------------------------------------------------- /site_config/standard/slog.thestranger.com.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: postCategory 2 | title: //h3[@class='postTitle'] 3 | body: //div[@class='postBody'] 4 | test_url: http://slog.thestranger.com/slog/archives/2010/10/12/sl-letter-of-the-day-leave-it-alone -------------------------------------------------------------------------------- /site_config/standard/utiliser-lightroom.com.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class="titrearticle"] 2 | body: //div[@class="article"] 3 | 4 | test_url: http://www.utiliser-lightroom.com/blog/2019/04/23/skylum-flex-et-airmagic-pour-les-utilisateurs-de-lightroom/ 5 | -------------------------------------------------------------------------------- /site_config/standard/www2.cnrs.fr.txt: -------------------------------------------------------------------------------- 1 | # Site configuration for CNRS press releases 2 | 3 | body: //div[@id="contenu"]//h2[@id="chapeau"] | //div[@id="contenu"]/div[@id="textContenu"] 4 | 5 | test_url: http://www2.cnrs.fr/presse/communique/5327.htm 6 | -------------------------------------------------------------------------------- /app.json: -------------------------------------------------------------------------------- 1 | { 2 | "name": "full-text-rss", 3 | "description": "[Updated and moded] Full-Text RSS can transform partial feeds to deliver the full content stripped of clutter and ads", 4 | "repository": "https://github.com/reycn/full-text-rss" 5 | } -------------------------------------------------------------------------------- /site_config/standard/cars.com.txt: -------------------------------------------------------------------------------- 1 | title: //div[contains(@class, 'basicInfo')]//h1 2 | 3 | body: //img[@id='chosenPhotoIMG'] | //div[@id='aboutThisVehicleBox'] 4 | 5 | prune: no 6 | 7 | test_url: http://www.cars.com/go/search/detail.jsp?listingId=115364779 -------------------------------------------------------------------------------- /site_config/standard/gocomics.com.txt: -------------------------------------------------------------------------------- 1 | body: //a[@class="photo"]/img[@class="strip"] 2 | author: //meta[@name="author"]/@content 3 | date: //meta[@property="gocomics:publish_date"]/@content 4 | 5 | test_url: http://www.gocomics.com/garfield/2015/06/13 6 | -------------------------------------------------------------------------------- /site_config/standard/gurusblog.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class='entry-title'] 2 | body: //div[@class='content'] 3 | 4 | test_url: https://www.gurusblog.com/archives/las-subastas-de-joyas-de-christies-y-sothebys-rompen-todos-los-records-historicos/15/11/2018/ 5 | -------------------------------------------------------------------------------- /site_config/standard/isource.com.txt: -------------------------------------------------------------------------------- 1 | # Remove social buttons 2 | strip: //div[@id='temp_Content_Right'] 3 | 4 | # Remove duplicate article title 5 | strip: //*[(@class='storytitle')] 6 | test_url: http://isource.com/2010/10/24/swearch-a-cool-iphone-web-app/ -------------------------------------------------------------------------------- /site_config/standard/maitre-eolas.fr.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class="post-content"] 2 | author: //meta[@name="author"]/@content 3 | date: //meta[@name="date"]/@content 4 | 5 | test_url: http://www.maitre-eolas.fr/post/2018/05/13/Pour-en-finir-avec-les-fiches-S 6 | -------------------------------------------------------------------------------- /site_config/standard/protonmail.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class="entry-title"] 2 | author: //header[@class="entry-header"]//a[@class="url fn n"] 3 | 4 | strip: //div[@class="csbtns"] 5 | 6 | test_url: https://protonmail.com/blog/biometric-authentication/ 7 | -------------------------------------------------------------------------------- /site_config/standard/roughtype.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='content'] 2 | strip: //p[@class='postmeta']/following::* 3 | strip: //p[@class='postmeta'] 4 | strip: //p[@align='left'] 5 | test_url: http://www.roughtype.com/archives/2012/01/power_to_the_da.php -------------------------------------------------------------------------------- /site_config/standard/ruhlman.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class='entry-title'] 2 | author: ///span[@class='author vcard'] 3 | date: //abbr[@class='published'] 4 | body: //div[@class='entry-content'] 5 | 6 | test_url: http://ruhlman.com/2009/05/cookbooks-that-teach/ -------------------------------------------------------------------------------- /site_config/standard/wordpress.org.txt: -------------------------------------------------------------------------------- 1 | author: //div[@class="meta"]/a[1] 2 | 3 | strip: //div[@class="meta"] 4 | strip: //div[@id="likes-other-gravatars"] 5 | strip: //noscript 6 | 7 | test_url: https://wordpress.org/news/2018/07/update-on-gutenberg/ 8 | -------------------------------------------------------------------------------- /site_config/standard/wz-newsline.de.txt: -------------------------------------------------------------------------------- 1 | title://h1 2 | 3 | date://p[@class='articleDate'] 4 | body://div[@class='articleBody wzStandardArticle'] 5 | test_url: http://www.wz-newsline.de/home/sport/tennis/federer-zum-vierten-mal-sieger-in-indian-wells-1.938050 -------------------------------------------------------------------------------- /site_config/standard/37signals.com.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class='post_header']//h2/a 2 | author: //span[@class='author'] 3 | date: //span[@class='date'] 4 | body: //div[@id='Content'] 5 | 6 | test_url: http://37signals.com/svn/posts/2785-the-end-of-the-it-department -------------------------------------------------------------------------------- /site_config/standard/7newsbelize.com.txt: -------------------------------------------------------------------------------- 1 | title: //*[@id='sstitle'] 2 | body: //div[@id='sstory'] 3 | strip_id_or_class: newsoptions 4 | prune: no 5 | 6 | test_url: http://www.7newsbelize.com/sstory.php?nid=25654 7 | test_url: http://www.7newsbelize.com/7news.xml -------------------------------------------------------------------------------- /site_config/standard/bjango.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class='articlehead'] 2 | body: //div[@class='column'] 3 | strip: //h1 4 | strip: //div[@class='help'] 5 | 6 | #no author or date/time provided in current layout 7 | test_url: http://bjango.com/articles/actions/ -------------------------------------------------------------------------------- /site_config/standard/blog.chriszacharias.com.txt: -------------------------------------------------------------------------------- 1 | author: //h2[@id="nav_title"]//a 2 | date: //time[@class="article_time"]/@datetime 3 | 4 | strip: //figure[contains(@class, "kudo")] 5 | 6 | test_url: http://blog.chriszacharias.com/a-conspiracy-to-kill-ie6 7 | -------------------------------------------------------------------------------- /site_config/standard/blog.twitter.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id="component-wrapper"] 2 | 3 | strip_id_or_class: tweet-error-text 4 | 5 | test_url: https://blog.twitter.com/developer/en_us/topics/tools/2018/new-developer-requirements-to-protect-our-platform.html 6 | -------------------------------------------------------------------------------- /site_config/standard/blog.wells.ee.txt: -------------------------------------------------------------------------------- 1 | title: //h2/a[@class="no-link title"] 2 | author: //h2[@id="blog_owner"] 3 | date: //time 4 | strip: //h2/a[@class="no-link title"] 5 | test_url: http://blog.wells.ee/retina 6 | test_url: http://blog.wells.ee/skeuomorphism -------------------------------------------------------------------------------- /site_config/standard/borderhouseblog.com.txt: -------------------------------------------------------------------------------- 1 | title://h1 2 | author://div[@class="meta"]/span/a 3 | date://div[@class="date"] 4 | body://div[@class="content article"] 5 | strip://div[@class="content article"]/h1 6 | 7 | test_url: http://borderhouseblog.com/?p=7832 -------------------------------------------------------------------------------- /site_config/standard/business2community.com.txt: -------------------------------------------------------------------------------- 1 | date: substring-after(//p[@class='byline'],'Published') 2 | 3 | strip: //div[@class='article-meta'] 4 | 5 | test_url: http://www.business2community.com/social-media/funky-ways-to-print-instagram-photos-0485340 6 | -------------------------------------------------------------------------------- /site_config/standard/communities-dominate.blogs.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class="entry-body"] 2 | test_url: http://communities-dominate.blogs.com/brands/2012/03/brutal-truth-about-lumia-cannot-sustain-even-1-to-1-replacement-of-symbian-windows-phone-strategy-do.html -------------------------------------------------------------------------------- /site_config/standard/hespress.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='article_holder']//div[@class='image'] | //div[@id='article_body'] 2 | 3 | prune: no 4 | tidy: no 5 | 6 | test_url: http://hespress.com/videos/73684.html 7 | test_url: http://hespress.com/permalink/73678.html -------------------------------------------------------------------------------- /site_config/standard/hometheaterreview.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='entry-body'] 2 | strip_id_or_class: paginate 3 | strip: //p[contains(., 'Additional Resources')] 4 | test_url: http://hometheaterreview.com/dreamvision-starlight-3-three-chip-d-ila-projector-reviewed/ -------------------------------------------------------------------------------- /site_config/standard/jacobinmag.com.txt: -------------------------------------------------------------------------------- 1 | body: //article[contains(@class, 'po__article')] 2 | 3 | strip: //aside 4 | strip: //footer 5 | 6 | prune: no 7 | 8 | test_url: https://www.jacobinmag.com/2019/07/noam-chomsky-interview-climate-change-imperialism 9 | -------------------------------------------------------------------------------- /site_config/standard/linux-magazin.de.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='td-post-content'] 2 | next_page_link: //link[@rel="next"] 3 | 4 | test_url: https://www.linux-magazin.de/ausgaben/2019/10/interview-2/ 5 | test_contains: eine solche Durchschlagskraft gewonnen 6 | -------------------------------------------------------------------------------- /site_config/standard/manga-news.com.txt: -------------------------------------------------------------------------------- 1 | title: //h2[@class='actu-title'] 2 | 3 | body: //div[@class='actu-content'] 4 | 5 | prune: no 6 | 7 | test_url: http://www.manga-news.com/index.php/actus/2016/02/02/Deux-retours-dans-le-planning-des-editions-Pika 8 | -------------------------------------------------------------------------------- /site_config/standard/maxim.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, 'field-name-body') or contains(@class, 'featured-image')] 2 | 3 | test_url: http://www.maxim.com/rss-feeds.xml 4 | test_url: http://www.maxim.com/entertainment/article/second-city-chicago-goes-flames -------------------------------------------------------------------------------- /site_config/standard/nosalty.hu.txt: -------------------------------------------------------------------------------- 1 | title: //div[@id='tab-recept']//h1 2 | body: //div[@id='tab-recept']//div[contains(@class, 'column-container')] 3 | strip_id_or_class: ajanlo-box 4 | prune: no 5 | 6 | test_url: http://www.nosalty.hu/recept/szupergyors-fank -------------------------------------------------------------------------------- /site_config/standard/palmbeachpost.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' story-text ')] 2 | 3 | test_url: http://www.palmbeachpost.com/news/crime--law/new-pbso-releases-sketch-person-interest-shooting/IcToH2Mij0hAw3EIFnw5tL/ 4 | -------------------------------------------------------------------------------- /site_config/standard/propakistani.pk.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' content-post ')] 2 | test_url: https://propakistani.pk/2017/06/23/eid-mubarak-happy-holidays-everyone/ 3 | test_contains: away from work for Eid Holidays 4 | -------------------------------------------------------------------------------- /site_config/standard/retractionwatch.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | body: //div[contains(concat(' ',normalize-space(@class),' '),' entry-content ')] 3 | test_url: https://retractionwatch.com/2018/07/30/how-institutions-gaslight-whistleblowers-and-what-can-be-done/ 4 | -------------------------------------------------------------------------------- /site_config/standard/shifteleven.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[ @class='entry-content' ] 2 | 3 | strip: //div[ contains(@class, 'sharing') ] 4 | 5 | date: //div[ @class='entry-meta' ]/a 6 | test_url: http://shifteleven.com/articles/2008/05/10/issue-tracking-git-ticgit -------------------------------------------------------------------------------- /site_config/standard/techpinions.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class="post"] 2 | 3 | strip: //div[@class="post-meta"] 4 | strip: //div[@id="socialicons"] 5 | strip: //div[@id="authorbox"] 6 | 7 | test_url: http://techpinions.com/why-google-and-microsoft-hate-siri/3572 -------------------------------------------------------------------------------- /site_config/standard/thewirecutter.com.txt: -------------------------------------------------------------------------------- 1 | body: //main/div/article 2 | 3 | strip: //header 4 | strip: //aside 5 | 6 | author: //p[@data-scp="author_name"] 7 | date: //time/@datetime 8 | 9 | test_url: https://thewirecutter.com/money/credit-cards/ 10 | -------------------------------------------------------------------------------- /site_config/standard/tuaw.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class='posttitle'] 2 | author: //span[@class='author']/a 3 | date: //span[@class='timestamp'] 4 | body: //div[@class='body'] 5 | 6 | test_url: http://www.tuaw.com/2011/10/19/apple-posts-fans-memories-of-steve-jobs/ -------------------------------------------------------------------------------- /site_config/standard/wpmayor.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='nrelate_flyout_placeholder'] 2 | 3 | strip_id_or_class: share 4 | 5 | prune: no 6 | 7 | test_url: http://www.wpmayor.com/themes/wordpress-portfolio-resume-themes/ 8 | test_url: http://www.wpmayor.com/feed/ -------------------------------------------------------------------------------- /site_config/standard/alsacreations.com.txt: -------------------------------------------------------------------------------- 1 | author: //div[@class="auteur-meta"]//a[contains(@href, '/profil/')] 2 | date: //div[@class="auteur-meta"]//time/@datetime 3 | 4 | test_url: https://www.alsacreations.com/tuto/lire/1771-css-grid-layout-en-production.html 5 | -------------------------------------------------------------------------------- /site_config/standard/ciaosamin.com.txt: -------------------------------------------------------------------------------- 1 | body://div[contains(@class, 'entry-content')] 2 | date://h2[contains(@class, 'date-header')] 3 | title://h3[contains(@class, 'post-title')] 4 | test_url: http://ciaosamin.com/ciao/2015/12/28/recipe-million-dollar-caramels 5 | -------------------------------------------------------------------------------- /site_config/standard/cjr.org.txt: -------------------------------------------------------------------------------- 1 | body: //p[@class='subhead' or @class='attribution'] | //div[@class='article-body'] 2 | prune: no 3 | 4 | single_page_link: //li[@class='print']/a 5 | 6 | test_url: http://www.cjr.org/behind_the_news/from_breaking_news_to_baseless.php -------------------------------------------------------------------------------- /site_config/standard/filamentgroup.com.txt: -------------------------------------------------------------------------------- 1 | title: //div[@id="content"]//h1 2 | author: //p[@class="article_author"]//a[contains(@href, "/about")] 3 | 4 | strip: //p[@class="articles_viewall"] 5 | 6 | test_url: https://www.filamentgroup.com/lab/select-css.html 7 | -------------------------------------------------------------------------------- /site_config/standard/geeksofdoom.com.txt: -------------------------------------------------------------------------------- 1 | author: substring-after(//span[@class='storyauthor'],'Posted by') 2 | date: //span[@class='storydate'] 3 | test_url: http://www.geeksofdoom.com/2012/03/14/robert-rodriguez-says-machete-kills-and-sin-city-2-will-film-this-year/ -------------------------------------------------------------------------------- /site_config/standard/higcapital.com.txt: -------------------------------------------------------------------------------- 1 | title: //h2[contains(@class, 'news-release-title')] 2 | body: //article 3 | strip: //div[contains(@class, 'intro')] 4 | strip: //div[contains(@class, 'contact')] 5 | 6 | test_url: https://higcapital.com/news/release/1251 7 | -------------------------------------------------------------------------------- /site_config/standard/mforum.cari.com.my.txt: -------------------------------------------------------------------------------- 1 | body: (//td[starts-with(@id, 'postmessage_')])[1] 2 | 3 | prune: no 4 | 5 | test_url: http://mforum.cari.com.my/forum.php?mod=viewthread&tid=788033 6 | test_url: http://mforum.cari.com.my/forum.php?mod=rss&fid=265&auth=0 -------------------------------------------------------------------------------- /site_config/standard/mikeash.com.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class="blogtitle"] 2 | strip: //div[@class="blogtitle"] 3 | 4 | author: substring-after(//span[@class="blogheader"], 'Author: ') 5 | test_url: http://www.mikeash.com/pyblog/friday-qa-2012-01-13-the-mac-toolbox.html -------------------------------------------------------------------------------- /site_config/standard/neunetz.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, 'entry-content')] 2 | prune: no 3 | 4 | strip_id_or_class: sharedaddy 5 | 6 | test_url: https://neunetz.com/2019/09/13/apple-tv-quasi-geschenkt/ 7 | test_contains: Apple muss hier noch aufholen 8 | -------------------------------------------------------------------------------- /site_config/standard/omicrono.elespanol.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1/a 2 | body: //div[@id='singlePostContent'] 3 | 4 | strip: //blockquote 5 | 6 | test_url: https://omicrono.elespanol.com/2018/08/samsung-comienza-producir-discos-ssd-gran-capacidad-mucho-mas-baratos/ 7 | -------------------------------------------------------------------------------- /site_config/standard/pharmazeutische-zeitung.de.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='row pageheaderAnlauf'] | //div[@class='row tt-content-layout-0'] 2 | strip: //div[@class='content_7'] 3 | 4 | test_url: https://www.pharmazeutische-zeitung.de/54-apotheken-weniger-in-2019/ 5 | -------------------------------------------------------------------------------- /site_config/standard/readability.com.txt: -------------------------------------------------------------------------------- 1 | single_page_link: //link[@rel='canonical']/@href 2 | 3 | test_url: http://www.readability.com/read?url=http://feeds.gawker.com/~r/lifehacker/full/~3/jaxAjSay_Rw/add-a-rain-gutter-to-a-picnic-table-for-a-built+in-drink-cooler -------------------------------------------------------------------------------- /site_config/standard/robots.thoughtbot.com.txt: -------------------------------------------------------------------------------- 1 | body: //section[@class='post text'] 2 | title: //h1[@class='title'] 3 | date: //p[@class='post-date'] 4 | strip: //section[@class='meta-info'] 5 | test_url: http://robots.thoughtbot.com/post/32455387133/four-phase-test -------------------------------------------------------------------------------- /site_config/standard/seattletransitblog.com.txt: -------------------------------------------------------------------------------- 1 | title: //h3[@class="storytitle"] 2 | date: //div[@class='meta'] 3 | body: //div[@class='storycontent'] 4 | 5 | test_url: http://seattletransitblog.com/2012/06/19/times-st-louis-interested-in-buying-waterfront-streetcars/ -------------------------------------------------------------------------------- /site_config/standard/thinkprogress.org.txt: -------------------------------------------------------------------------------- 1 | author: //p[@class="byline"]/a 2 | body: //div[@class="post"] 3 | 4 | test_url: http://thinkprogress.org/special/2011/11/12/367040/harvard-law-professor-criticizes-homeland-security-feel-of-overreaction-to-occupy-harvard/ -------------------------------------------------------------------------------- /site_config/standard/voltairenet.org.txt: -------------------------------------------------------------------------------- 1 | title: //h1[contains(concat(' ',normalize-space(@class),' '),' titre_serif_1 ')] 2 | body: //div[contains(concat(' ',normalize-space(@class),' '),' texte_sans ')] 3 | 4 | test_url: http://www.voltairenet.org/article195149.html -------------------------------------------------------------------------------- /site_config/standard/basicthinking.de.txt: -------------------------------------------------------------------------------- 1 | title: //h2 2 | date: //span[@class='date'] 3 | body: //div[@class='entry'] 4 | 5 | strip: //div[@class='zusatz'] 6 | 7 | test_url: http://www.basicthinking.de/blog/2011/12/13/sagt-social-networks-adieu-begrust-private-networks/ -------------------------------------------------------------------------------- /site_config/standard/blog.eleven-labs.com.txt: -------------------------------------------------------------------------------- 1 | body://div[@class='post_inner_wrapper'] 2 | date://div[@class='sub_page_caption'] 3 | strip://div[contains(concat(' ',normalize-space(@class),' '),' post_img ')] 4 | test_url: http://blog.eleven-labs.com/en/cqrs-pattern-2/ 5 | -------------------------------------------------------------------------------- /site_config/standard/diagonalperiodico.net.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='field__items'] 2 | title: //div[@class='art_titulo'] 3 | 4 | test_url: https://www.diagonalperiodico.net/global/27947-cuanto-mas-tiempo-nos-aferremos-este-sistema-peor-y-menores-seran-nuestras-opciones 5 | -------------------------------------------------------------------------------- /site_config/standard/ecranlarge.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' content-description ')] 2 | 3 | test_url: https://www.ecranlarge.com/films/news/997080-star-wars-les-derniers-jedi-de-nouvelles-photos-de-rey-et-luke-en-mode-badass 4 | -------------------------------------------------------------------------------- /site_config/standard/heise-online.mobi.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='content']/div 2 | date: //p[@class='author_date']/span[@class='date'] 3 | test_url: http://heise-online.mobi/newsticker/meldung/Amazons-Appstore-in-der-Kritik-Ein-Desaster-fuer-Kunden-und-Entwickler-1273936.html -------------------------------------------------------------------------------- /site_config/standard/humantransit.org.txt: -------------------------------------------------------------------------------- 1 | title: //h3[@class="entry-header"] 2 | date: //h2[@class="date-header"] 3 | body: //div[contains(@class, 'entry')] 4 | 5 | test_url: http://www.humantransit.org/2012/06/can-network-primers-reduce-grief-about-network-design.html -------------------------------------------------------------------------------- /site_config/standard/inhabitat.com.txt: -------------------------------------------------------------------------------- 1 | # set body 2 | body: //div[@class='post-listing'] 3 | 4 | # remove clutter 5 | strip: //a/big 6 | strip: //a/em 7 | strip: //p/em 8 | test_url: http://inhabitat.com/2010/11/18/sliding-walls-transform-this-tokyo-house-into-an-office/ -------------------------------------------------------------------------------- /site_config/standard/inquirer.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='article-body'] 2 | 3 | strip_id_or_class: related-stories 4 | 5 | test_url: https://www.inquirer.com/news/berks-detention-center-child-migrants-20191029.html 6 | test_contains: candidates running for Berks County 7 | -------------------------------------------------------------------------------- /site_config/standard/menshealth.com.sg.txt: -------------------------------------------------------------------------------- 1 | strip: //div[contains(@style, 'float:right') and contains(., 'advertisement')] 2 | body: //div[@style="float:left;width:740px;"] 3 | 4 | tidy: no 5 | 6 | # broken feed? 7 | test_url: http://www.menshealth.com.sg/fitness/feed 8 | -------------------------------------------------------------------------------- /site_config/standard/rasgolatente.es.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class='entry-title'] 2 | author: //div[@class='entry-author'] 3 | date: //div[@class='entry-time'] 4 | body: //div[@class='entry-content'] 5 | 6 | test_url: http://rasgolatente.es/estupidez-psicologia-estupidos/ 7 | -------------------------------------------------------------------------------- /site_config/standard/sayidaty.net.txt: -------------------------------------------------------------------------------- 1 | date: //meta[@property='article:published_time']/@content 2 | body: (//div[contains(@class, 'article-slider')]//img)[1] | //div[contains(@class, 'bottom-article-con')] 3 | 4 | test_url: http://www.sayidaty.net/taxonomy/term/10/all/feed -------------------------------------------------------------------------------- /site_config/standard/squashed.tumblr.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='content'] 2 | date: substring-before( //div[@class='unit dateAndNotes'], 'with') 3 | title: //h3 4 | test_url: http://squashed.tumblr.com/post/17613522228/lets-stop-blaming-the-victims-of-predatory-lending -------------------------------------------------------------------------------- /site_config/standard/theage.com.au.txt: -------------------------------------------------------------------------------- 1 | author: //h3[@class='authorName'] 2 | date: //time 3 | body: //div[@class='articleBody'] 4 | strip_id_or_class: adspot 5 | test_url: http://www.theage.com.au/victoria/top-cops-warns-outlaw-bikies-we-have-a-gang-too-20130331-2h1l8.html -------------------------------------------------------------------------------- /site_config/standard/tijd.be.txt: -------------------------------------------------------------------------------- 1 | strip: //div[@class="ac_paragraph-hidden"] 2 | 3 | test_url: https://www.tijd.be/politiek-economie/internationaal/algemeen/ik-ben-vooral-bezorgd-dat-corona-uitbreekt-in-afrika/10204834.html 4 | test_contains: U klinkt gealarmeerder 5 | -------------------------------------------------------------------------------- /site_config/standard/toolinux.com.txt: -------------------------------------------------------------------------------- 1 | title: //h2[contains(@class,'news')] 2 | body: //div[contains(@class,'articleContent')] 3 | date: substring-after(//div[@class = 'SupaDate']/text(), 'le') 4 | 5 | test_url: http://www.toolinux.com/Wi-Fi-Linksys-WRT-la-legende-de 6 | -------------------------------------------------------------------------------- /site_config/standard/adslzone.net.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class='entry-title'] 2 | body: //article 3 | 4 | strip: //div[@class='io-word-count'] 5 | strip: //ol[@class='breadcrumb'] 6 | strip: //aside 7 | 8 | test_url: https://www.adslzone.net/2017/12/19/xiaomi-sneaky-santa-17/ 9 | -------------------------------------------------------------------------------- /site_config/standard/blog.pinboard.in.txt: -------------------------------------------------------------------------------- 1 | title: //a[@class="blog_title"] 2 | date: //p[@class="when"]/a 3 | body: //div[@class="blog_entry"] 4 | strip_id_or_class:blog_title 5 | strip_id_or_class:when 6 | test_url: http://blog.pinboard.in/2011/11/the_social_graph_is_neither/ -------------------------------------------------------------------------------- /site_config/standard/boundlessline.org.txt: -------------------------------------------------------------------------------- 1 | title: substring-before(//title, '|') 2 | body: //div[@class="entry"] 3 | # Remove the author's picture 4 | strip: //div[@class="entry"]/a[1] 5 | test_url: http://www.boundlessline.org/2011/06/the-nyts-on-gender-over-the-weekend.html -------------------------------------------------------------------------------- /site_config/standard/engineering.tumblr.com.txt: -------------------------------------------------------------------------------- 1 | title: //h2 2 | body: //div[@class="post_content"] 3 | author: //p[@class="author"]/a 4 | date: //p[@class="date"] 5 | strip: //h2 6 | strip: //header 7 | test_url: http://engineering.tumblr.com/post/21276808338/tumblr-firehose -------------------------------------------------------------------------------- /site_config/standard/gurumed.org.txt: -------------------------------------------------------------------------------- 1 | prune: no 2 | body: //div[@class='entry'] 3 | strip: //div[@class='addthis_toolbox'] 4 | strip: //div[@class='yarpp-related'] 5 | 6 | test_url: http://www.gurumed.org/2015/06/22/nous-entrons-dsormais-dans-la-sixime-extinction-massive/ 7 | -------------------------------------------------------------------------------- /site_config/standard/hvg.hu.txt: -------------------------------------------------------------------------------- 1 | title: //div[@id='pg-content']//h1 2 | body: //div[@id='articleBody0'] 3 | replace_string():

4 | 5 | single_page_link: //div[@class="up-header"]/a 6 | 7 | prune: no 8 | 9 | test_url: http://hvg.hu/w/20111125_sparta -------------------------------------------------------------------------------- /site_config/standard/papodehomem.com.br.txt: -------------------------------------------------------------------------------- 1 | title: //h2[@class="page_title"] 2 | body: //div[@class="entry arquivo"] 3 | author: //span[@class="author"] 4 | footnotes: yes 5 | prune: yes 6 | test_url: http://papodehomem.com.br/um-relato-confessional-sobre-a-maioridade-penal/ -------------------------------------------------------------------------------- /site_config/standard/quantamagazine.org.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, 'post__content__section')] 2 | 3 | test_url: https://www.quantamagazine.org/a-path-less-taken-to-the-peak-of-the-math-world-20170627/ 4 | test_contains: Mathematicians are interested in the following 5 | -------------------------------------------------------------------------------- /site_config/standard/rachelandrew.co.uk.txt: -------------------------------------------------------------------------------- 1 | date: //time/@datetime 2 | 3 | strip: //aside[@class="article-meta"] 4 | strip: //div[@class="keep-in-touch"] 5 | 6 | test_url: https://www.rachelandrew.co.uk/archives/2019/01/30/html-css-and-our-vanishing-industry-entry-points/ 7 | -------------------------------------------------------------------------------- /site_config/standard/rezeptwelt.de.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='step-content'] | //div[@class='global-active ingredients-box'] 2 | title: //div[@class='step-1-container'] 3 | 4 | tidy: no 5 | test_url: http://www.rezeptwelt.de/backen-herzhaft-rezepte/w%C3%BCrstchen-schlangen/530372 -------------------------------------------------------------------------------- /site_config/standard/techcommunity.microsoft.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' lia-message-body-content ')] 2 | 3 | test_url: https://techcommunity.microsoft.com/t5/Azure-Active-Directory-Identity/Your-Pa-word-doesn-t-matter/ba-p/731984 4 | -------------------------------------------------------------------------------- /site_config/standard/devlinsangle.blogspot.co.at.txt: -------------------------------------------------------------------------------- 1 | date: //h2[@class='date-header'] 2 | body: //div[@class='post hentry'] 3 | title: //h3 4 | strip: //div[@class='post-footer'] 5 | 6 | test_url: http://devlinsangle.blogspot.co.at/2012/03/difference-between-teaching-and_01.html -------------------------------------------------------------------------------- /site_config/standard/glazman.org.txt: -------------------------------------------------------------------------------- 1 | title: //h2[@class="post-title"] 2 | date: //meta[@name="date"]/@content 3 | author: //meta[@name="dc.publisher"]/@content 4 | 5 | test_url: http://www.glazman.org/weblog/dotclear/index.php?post/2019/01/23/WebExtensions-v3-considered-harmful 6 | -------------------------------------------------------------------------------- /site_config/standard/gnppn.fr.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class="entry-title"] 2 | author: //p[@class="site-title"] 3 | date: //time[@class="entry-date published"]/@datetime 4 | 5 | test_url: https://gnppn.fr/2018/10/21/quotidiens-nationaux-et-abonnement-en-ligne-prime-a-la-penibilite/ 6 | -------------------------------------------------------------------------------- /site_config/standard/hiphopleeft.nl.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class = 'pd'] 2 | strip: //div[@id = 'overzicht-albumrecensies'] 3 | strip: //div[@id = 'jc'] 4 | test_url: http://hiphopleeft.nl/index.php?option=com_content&view=article&id=2767:mark-ronson-record-collection&catid=66:m&Itemid=142 -------------------------------------------------------------------------------- /site_config/standard/icannabis.tumblr.com.txt: -------------------------------------------------------------------------------- 1 | tidy:no 2 | prune:no 3 | 4 | body://div[contains(@id,'content')] 5 | 6 | strip_id_or_class:meta 7 | strip_id_or_class:notes 8 | strip_id_or_class:pagination 9 | test_url: http://icannabis.tumblr.com/post/28660592471/reviewmswireless3000 -------------------------------------------------------------------------------- /site_config/standard/ilyabirman.ru.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class='published visible e2-smart-title']//span 2 | author: //span[@id='e2-blog-title'] 3 | date: //p[@class='super-h'] 4 | body: //div[@class='text published visible'] 5 | test_url: http://ilyabirman.ru/meanwhile/2011/11/15/2/ -------------------------------------------------------------------------------- /site_config/standard/itmedia.co.jp.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='cmsBody'] 2 | 3 | next_page_link: //span[@id='next']/a 4 | 5 | strip_id_or_class: cmsCopyright 6 | strip_id_or_class: masterSocialbuttonBtm 7 | 8 | test_url: http://www.itmedia.co.jp/enterprise/articles/0912/05/news002.html -------------------------------------------------------------------------------- /site_config/standard/lehollandaisvolant.net.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class="blog-post-title"] 2 | date: //time[contains(@itemprop, "datePublished")]/@datetime 3 | 4 | test_url: https://lehollandaisvolant.net/?d=2019/12/07/11/41/38-non-respirer-nemet-pas-plus-de-co2-dans-latmosphere 5 | -------------------------------------------------------------------------------- /site_config/standard/nojesguiden.se.txt: -------------------------------------------------------------------------------- 1 | author: //span[@class='meta']/span[@class='username'] 2 | body: //div[@class='article-content'] 3 | 4 | strip_id_or_class: 'article-actions' 5 | test_url: http://nojesguiden.se/blogg/maja-bredberg/maja-laser-tidningen-en-helt-vanlig-lordag-i -------------------------------------------------------------------------------- /site_config/standard/rpgsite.net.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='news-text'] 2 | prune: no 3 | test_url: http://www.rpgsite.net/news/1964-tetsuya-nomura-says-hell-soon-show-the-future-of-final-fantasy 4 | test_url: http://www.rpgsite.net/news/1965-new-atelier-totori-plus-screens-and-artwork -------------------------------------------------------------------------------- /site_config/standard/useit.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | 3 | date: substring-after(//p[@class='overline']/strong, ',') 4 | body: //div[@class="maintext"] 5 | strip: //p[@class='overline'] 6 | strip: //h1 7 | tidy: no 8 | test_url: http://www.useit.com/alertbox/mobile-startup-screen.html -------------------------------------------------------------------------------- /site_config/standard/vitispr.com.txt: -------------------------------------------------------------------------------- 1 | strip: //*[(@id = "ja-search")] 2 | body: //*[(@id = "ja-mainbody")] 3 | body: //*[(@id = "content-mass-bottom")] 4 | strip://h3[contains(span,'Related Posts')] 5 | strip://img 6 | test_url: http://vitispr.com/blog/coventry-is-a-technology-hotspot -------------------------------------------------------------------------------- /site_config/standard/yostivanich.com.txt: -------------------------------------------------------------------------------- 1 | title://div[@class='entry-title'] 2 | body://div[@class='entry-content'] 3 | strip_comments:yes 4 | convert_double_br_tags:yes 5 | test_url: http://www.yostivanich.com/2010/07/11/wired-com-with-world-watching-wikileaks-falls-into-disrepair/ -------------------------------------------------------------------------------- /site_config/README.txt: -------------------------------------------------------------------------------- 1 | Full-Text RSS Site Patterns 2 | --------------------------- 3 | 4 | Site patterns allow you to specify what should be extracted from specific sites. 5 | 6 | Please see http://help.fivefilters.org/customer/portal/articles/223153-site-patterns for more information. -------------------------------------------------------------------------------- /site_config/standard/24.ae.txt: -------------------------------------------------------------------------------- 1 | title: //div[@id='DivTitle'] 2 | body: //div[@id='divImages' or @id='Divkhabarcontent'] 3 | author: //div[@id='DivAuthor'] 4 | 5 | prune: no 6 | 7 | test_url: http://24.ae/article.aspx?ArticleId=123304 8 | test_url: http://24.ae/rss.aspx?pageId=30 9 | -------------------------------------------------------------------------------- /site_config/standard/acroswing.fr.txt: -------------------------------------------------------------------------------- 1 | tidy:no 2 | date: //time[@class='updated'] 3 | dissolve: //ul[@class='video-gallery']/li 4 | dissolve: //ul[@class='video-gallery'] 5 | test_url: http://www.acroswing.fr/actualites/competition_rock/selectif_bellegarde_sur_valserine__2012-02-26.php -------------------------------------------------------------------------------- /site_config/standard/gawker.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class="post-body"] 2 | 3 | # Remove 'content is restricted' 4 | strip: //div[@id='agegate_IDHERE'] 5 | 6 | http_header(user-agent): PHP/5.3 7 | 8 | test_url: http://gawker.com/#!5782070/russian-bomb-squad-successfully-defuses-sex-toy -------------------------------------------------------------------------------- /site_config/standard/huffingtonpost.fr.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: tag-cloud 2 | strip_id_or_class: follow-us__correction 3 | strip_id_or_class: corrections_container 4 | 5 | test_url: http://www.huffingtonpost.fr/michael-dias/pourquoi-la-generation-y-est-elle-en-train-de-demissionner/ 6 | -------------------------------------------------------------------------------- /site_config/standard/inessential.com.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class='weblogPost']/h3[1] 2 | author: ("Brent Simmons") 3 | date: //span[@class="weblogPostDisplayDate"] 4 | body: //div[@class='weblogPostBody'] 5 | test_url: http://inessential.com/2011/10/25/why_just_store_the_app_data_on_dropbo -------------------------------------------------------------------------------- /site_config/standard/omiliya.org.txt: -------------------------------------------------------------------------------- 1 | title: //div[@id='squeeze']/h1 2 | strip: //div[@id='squeeze']/h1 3 | author: //div[@class='submitted']/a 4 | strip: //div[@class='submitted']/a 5 | convert_double_br_tags: yes 6 | 7 | 8 | 9 | test_url: http://omiliya.org/content/predchuvstvie.html -------------------------------------------------------------------------------- /site_config/standard/popehat.com.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class='entry-title'] 2 | author: //div[@class='entry-author'] 3 | date: //div[@class='entry-time'] 4 | body: //div[@class='entry-content'] 5 | 6 | test_url: https://popehat.com/2015/12/16/eric-posner-the-first-amendments-nemesis/ 7 | -------------------------------------------------------------------------------- /site_config/standard/timeshighereducation.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class="col-md-12 radix-layouts-contentheader panel-panel"] 2 | strip: //div[@class="htmlContent subscribe_box"] 3 | test_url: https://www.timeshighereducation.com/blog/jeremy-corbyn-serious-about-free-higher-education 4 | -------------------------------------------------------------------------------- /site_config/standard/accesstoinsight.org.txt: -------------------------------------------------------------------------------- 1 | title: //div[@id='H_docTitle'] 2 | 3 | body: //div[@id='H_meta' or @id='H_content' or @id='F_footer'] 4 | 5 | strip_id_or_class: F_toenail 6 | 7 | prune: no 8 | 9 | test_url: http://www.accesstoinsight.org/lib/authors/nyanaponika/wheel026.html -------------------------------------------------------------------------------- /site_config/standard/addendum.org.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | body: //div[@class='copy__content'] | //div[@class='heroStage__introText'] 3 | 4 | test_url: https://www.addendum.org/pendler/problem-pendlerpauschale 5 | test_url: https://www.addendum.org/repraesentation/zusammensetzung-nationalrat/ 6 | -------------------------------------------------------------------------------- /site_config/standard/americandrink.net.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class='head']/h2/a 2 | author: //div[@class='head']/a 3 | date: //div[@class='head']/p[@class='date']/a 4 | body: //div[@class='copy'] 5 | strip: //p[@class='meta'] 6 | test_url: http://americandrink.net/post/10567188712/free-the-hooch -------------------------------------------------------------------------------- /site_config/standard/blog.dropbox.com.txt: -------------------------------------------------------------------------------- 1 | body: (//article)[1] 2 | 3 | strip_id_or_class: quote-plank 4 | strip_id_or_class: article-next-prev-plank 5 | strip_id_or_class: author-bios__description-mobile 6 | 7 | test_url: https://blog.dropbox.com/topics/work-culture/meet-the-lasermonks 8 | -------------------------------------------------------------------------------- /site_config/standard/da.feedsportal.com.txt: -------------------------------------------------------------------------------- 1 | single_page_link: //a 2 | tidy: no 3 | prune: no 4 | 5 | test_url: http://da.feedsportal.com/c/585/f/413794/s/17037b5a/l/0L0Stelegraaf0Bnl0Cbinnenland0C10A2757860C0I0IKlacht0Itegen0Idr0B0IFrank0Iniet0I0Eontvankelijk0I0I0Bhtml0Dcid0Frss/ia1.htm 6 | -------------------------------------------------------------------------------- /site_config/standard/ebay.com.txt: -------------------------------------------------------------------------------- 1 | body: //h1[@class='it-ttl'] | //div[@id='mainImgHldr'] | //span[@id='prcIsum'] 2 | 3 | strip_image_src: imgLoading_30x30.gif 4 | 5 | test_url: http://www.ebay.com/itm/BRAND-NEW-FM-Transmitter-Ca-r-Charger-iPhone-4S-4-4G-3GS-3G-2G-iPod-Touch-/190657497204 -------------------------------------------------------------------------------- /site_config/standard/elmundo.es.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class='js-headline'] 2 | body: //div[@itemprop='articleBody'] 3 | author: //li[@class='author-name'] 4 | 5 | strip: //figure[0] 6 | strip: //aside 7 | 8 | test_url: http://www.elmundo.es/cataluna/2018/01/14/5a5b3be9e5fdea3f118b45c6.html 9 | -------------------------------------------------------------------------------- /site_config/standard/firstthings.com.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class='articleTitle'] 2 | author: //div[@class='articleAuthor'] 3 | body: //div[@class='articleContent'] 4 | prune: no 5 | convert_double_br_tags: yes 6 | 7 | test_url: http://www.firstthings.com/article/2011/05/the-trouble-with-ayn-rand -------------------------------------------------------------------------------- /site_config/standard/grafikart.fr.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' formatted ')] 2 | date: //div[contains(concat(' ',normalize-space(@class),' '),' tutoriel_date ')] 3 | 4 | test_url: https://www.grafikart.fr/tutoriels/json-web-token-presentation-958 5 | -------------------------------------------------------------------------------- /site_config/standard/itwire.com.txt: -------------------------------------------------------------------------------- 1 | author: //a[@rel="author"] 2 | date: //li[@class="itemDateCreated"] 3 | strip: //div[contains(@class, 'legend-rounded')] 4 | 5 | test_url: http://www.itwire.com/it-industry-news/market/59661-ibm-looks-to-high-value-solutions-to-meet-changing-demands 6 | -------------------------------------------------------------------------------- /site_config/standard/lezephyrmag.com.txt: -------------------------------------------------------------------------------- 1 | 2 | body: //div[contains(concat(' ',normalize-space(@class),' '),' entry-content ')] 3 | 4 | strip_id_or_class: swp_social_panel 5 | 6 | test_url: https://lezephyrmag.com/libre/paroles/jean-roch-waro-voyager-avec-la-musique-est-un-acte-poetique/ 7 | -------------------------------------------------------------------------------- /site_config/standard/libcom.org.txt: -------------------------------------------------------------------------------- 1 | date: //span[contains(@class, 'page-date')] 2 | body: //div[@id='node-page'] 3 | strip_id_or_class: book-navigation 4 | prune: no 5 | 6 | test_url: http://libcom.org/library/what-was-the-ussr-aufheben-1 7 | test_url: http://libcom.org/library-latest/feed -------------------------------------------------------------------------------- /site_config/standard/mintpressnews.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' entry-content ')] 2 | 3 | test_url: https://www.mintpressnews.com/mega-group-maxwells-mossad-spy-story-jeffrey-epstein-scandal/261172/ 4 | test_contains: One of those businessmen 5 | -------------------------------------------------------------------------------- /site_config/standard/msdn.microsoft.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id="mainBody"] 2 | prune: no 3 | test_url: http://msdn.microsoft.com/en-us/library/hh542796(VS.103).aspx 4 | test_url: https://msdn.microsoft.com/library/hh191443(vs.110).aspx 5 | test_contains: An async method typically contains 6 | -------------------------------------------------------------------------------- /site_config/standard/n0where.net.txt: -------------------------------------------------------------------------------- 1 | title://div[@class='main-title single-title entry-title'] 2 | body://div[@id="content-anchor-inner"] 3 | date://meta[@content] 4 | 5 | strip: //noscript 6 | 7 | test_url: https://n0where.net/dump-and-analyze-net-applications-memory-memoscope-net/ 8 | -------------------------------------------------------------------------------- /site_config/standard/pcmag.com.txt: -------------------------------------------------------------------------------- 1 | prune:yes 2 | 3 | date://*[contains(@class,'date')] 4 | 5 | body://div[contains(@id,'content')] 6 | 7 | next_page_link://a[contains(.,'Next >')] 8 | 9 | strip_id_or_class:sponsors 10 | test_url: http://www.pcmag.com/article2/0,2817,2401676,00.asp -------------------------------------------------------------------------------- /site_config/standard/politifact.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id="content"] 2 | 3 | strip: //div[@class="pfcontentmid"]/div[position()>4]|//div[@class="pfad"] 4 | test_url: http://www.politifact.com/truth-o-meter/statements/2011/may/30/barbara-boxer/barbara-boxer-says-medicare-overhead-far-lower-pri/ -------------------------------------------------------------------------------- /site_config/standard/searchenginejournal.com.txt: -------------------------------------------------------------------------------- 1 | strip: //ul[contains(@id, "social")] 2 | strip: //div[contains(@class, "ts-fab-wrapper")] 3 | strip: //div[contains(@id, 'gpt-ad')] 4 | 5 | test_url: http://www.searchenginejournal.com/web-design-vs-seo-it-doesnt-make-much-sense/62294/ 6 | -------------------------------------------------------------------------------- /site_config/standard/telerama.fr.txt: -------------------------------------------------------------------------------- 1 | author: //div[@itemprop="author"]//img/@alt 2 | strip: //div[@id="article--sidebar-right"] 3 | strip: //div[@class="article--tags"] 4 | 5 | test_url: https://www.telerama.fr/medias/peertube,-la-plateforme-qui-defie-youtube-avec-lethique,n5715829.php 6 | -------------------------------------------------------------------------------- /site_config/standard/thedrive.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='has-ad-column'] 2 | 3 | strip: //div[contains(@class, 'articleFragment') and not(contains(@class, 'paragraph'))] 4 | 5 | test_url: http://www.thedrive.com/new-cars/12579/why-this-4000-renault-is-as-disruptive-as-the-tesla-model-3 6 | -------------------------------------------------------------------------------- /site_config/standard/timeshighereducation.co.uk.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | body: //div[@class="storytext"] 3 | strip: //div[@id="thelogin"] 4 | strip: //*[@class="hide"] 5 | strip: //div[@id="anchored"] 6 | test_url: http://www.timeshighereducation.co.uk/story.asp?sectioncode=26&storycode=416124&c=1 -------------------------------------------------------------------------------- /site_config/standard/wiki.guildwars.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | body: //div[@id='content'] 3 | strip_id_or_class: editsection 4 | strip_id_or_class: toc 5 | strip: //div[@id='siteNotice'] 6 | strip: //div[@id='content']//table[last()] 7 | prune: no 8 | test_url: http://wiki.guildwars.com/wiki/Monk -------------------------------------------------------------------------------- /site_config/standard/36kr.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[contains(@class, 'entry-title')] 2 | date: //meta[@name='weibo: article:create_at']/@content 3 | body: //div[contains(@class, 'mainContent')] 4 | strip_id_or_class: related_topics 5 | 6 | prune: no 7 | 8 | test_url: http://www.36kr.com/p/207879.html -------------------------------------------------------------------------------- /site_config/standard/annouchka.fr.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: code-block 2 | strip_id_or_class: hellocoton_vote 3 | 4 | strip: //a[@class='c3'] 5 | strip: //a[@class='c4'] 6 | strip: //a[@class='c5'] 7 | 8 | test_url: http://www.annouchka.fr/5-conseils-pour-profiter-un-peu-plus-de-ses-enfants/ 9 | -------------------------------------------------------------------------------- /site_config/standard/folklore.org.txt: -------------------------------------------------------------------------------- 1 | author: /html/body/table[3]/tbody/tr/td[1]/table[2]/tbody/tr[1]/td[2] 2 | date: /html/body/table[3]/tbody/tr/td[1]/table[2]/tbody/tr[2]/td[2] 3 | body: //div[@class='main'] 4 | test_url: http://www.folklore.org/StoryView.py?story=Calculator_Construction_Set.txt -------------------------------------------------------------------------------- /site_config/standard/gizmodo.uol.com.br.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | 3 | body: //div[@id='destaques']//div[contains(@class, 'img')] | //div[@id='maincontent']//p 4 | 5 | test_url: http://gizmodo.uol.com.br/nvidia-gtx-titan-z/ 6 | test_url: http://gizmodo.uol.com.br/perfil-mark-zuckerberg-hackeado/ 7 | -------------------------------------------------------------------------------- /site_config/standard/haberler.com.txt: -------------------------------------------------------------------------------- 1 | title: //div[@id="habermetni"]/h1[@id="haber_baslik"] 2 | body: //div[@id="habermetni"]/p 3 | strip: //img[@class='newsDetailLeft'] 4 | strip_image_src: /haber-resimleri/ 5 | test_url: http://www.haberler.com/emniyete-atacakti-elinde-patladi-3198733-haberi/ -------------------------------------------------------------------------------- /site_config/standard/linuxjournal.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='content-area'] 2 | next_page_link: //a[@title='Go to next page'] 3 | author: //a[@title='View user profile.'] 4 | strip_id_or_class: comments 5 | 6 | test_url: http://www.linuxjournal.com/content/be-mechanicwith-android-and-linux 7 | -------------------------------------------------------------------------------- /site_config/standard/marco.org.txt: -------------------------------------------------------------------------------- 1 | tidy: no 2 | prune: no 3 | date: //article//time[@pubdate] 4 | title: //article/header/h2 5 | body: //article 6 | strip: //header 7 | test_url: http://www.marco.org/2012/09/08/businessweek-gruber 8 | test_url: http://www.marco.org/2012/04/24/might-upgrade-someday -------------------------------------------------------------------------------- /site_config/standard/medialens.org.txt: -------------------------------------------------------------------------------- 1 | strip_id_or_class: article-tools 2 | strip_id_or_class: pagenav 3 | prune: no 4 | test_url: http://www.medialens.org/index.php/alerts/alert-archive/2012/713-the-illusion-of-democracy.html 5 | test_contains: In an era of permanent war, economic meltdown 6 | -------------------------------------------------------------------------------- /site_config/standard/nakedsecurity.sophos.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='entry-content'] 2 | title: //div[@class='entry-title'] 3 | strip: //div[@class='entry-sharing'] 4 | 5 | test_url: https://nakedsecurity.sophos.com/2016/04/04/new-ransomware-with-an-old-trick-petya-parties-like-its-1989/ 6 | -------------------------------------------------------------------------------- /site_config/standard/pastebin.com.txt: -------------------------------------------------------------------------------- 1 | title://div[@class="paste_box_line1"]/h1 2 | author://div[@class="paste_box_line2"]/a 3 | body://div[@class="text"] 4 | date:substring-before(substring-after(//div[@class="paste_box_line2"],'|'),'|') 5 | dissolve://li 6 | test_url: http://pastebin.com/LAykd1es -------------------------------------------------------------------------------- /site_config/standard/resume.se.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, 'article-body')] 2 | 3 | prune: no 4 | 5 | test_url: https://www.resume.se/nyheter/artiklar/2017/09/19/blossa-lanserar-arets-glogg-via-livestream--hamtar-inspiration-fran-indien 6 | test_url: http://www.resume.se/rss-nyheter 7 | -------------------------------------------------------------------------------- /site_config/standard/splinternews.com.txt: -------------------------------------------------------------------------------- 1 | title: //head/title 2 | author: //meta[@name="author"]/@content 3 | body: //div[contains(@class, 'post-content')] 4 | strip: //div[contains(@class, 'content-summary')] 5 | 6 | test_url: https://splinternews.com/jonathan-chait-has-no-excuse-1831076209 7 | -------------------------------------------------------------------------------- /site_config/standard/thesimpledollar.com.txt: -------------------------------------------------------------------------------- 1 | title: //h3[@class='post-title']/a[@class='post-title-link'] 2 | body: //div[@class='post-content'] 3 | author: //div[@class='post-meta-under-title']/a 4 | test_url: http://www.thesimpledollar.com/2011/09/13/determining-the-size-of-your-emergency-fund/ -------------------------------------------------------------------------------- /site_config/standard/wmpoweruser.com.txt: -------------------------------------------------------------------------------- 1 | date://*[@class="entry-date"] 2 | author://*[@class="author vcard"] 3 | strip://*[@style="position:relative;left:72px;top:2px;"]|//*[@id="authorbox"] 4 | test_url: http://wmpoweruser.com/breaking-nokia-announces-nfc-support-in-lumia-610-windows-phone-device/ -------------------------------------------------------------------------------- /site_config/standard/albayan.ae.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='main-column']//div[@class='content'] 2 | 3 | strip_id_or_class: social-buttons 4 | 5 | prune: no 6 | 7 | test_url: http://www.albayan.ae/across-the-uae/education/2013-08-29-1.1949645 8 | test_url: http://www.albayan.ae/1.448?ot=ot.AjaxPageLayout -------------------------------------------------------------------------------- /site_config/standard/amandala.com.bz.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='content']//div[contains(@class, 'content')] 2 | strip_id_or_class: widget 3 | strip: //a[contains(@href, 'upm_export=')] 4 | 5 | test_url: http://amandala.com.bz/news/feed/ 6 | test_url: http://amandala.com.bz/news/poor-pse-results-30-raise/ -------------------------------------------------------------------------------- /site_config/standard/annatravelling.wordpress.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class="title"] 2 | 3 | author: ("Anna Manasova") 4 | # is ignored, unfortunately 5 | 6 | date: //p[@class="date"] 7 | 8 | body: //div[@class="entry"] 9 | test_url: http://annatravelling.wordpress.com/2011/11/07/a-day-of-cooking-thai/ -------------------------------------------------------------------------------- /site_config/standard/blog.kaelig.fr.txt: -------------------------------------------------------------------------------- 1 | body: //*[contains(@class, 'post_content')] 2 | author: string('Kaelig Deloumeau-Prigent') 3 | title: //h1[@class='title'] 4 | date: //span[@class='date'] 5 | test_url: http://blog.kaelig.fr/post/24877648508/pr%C3%A9processeurs-css-renoncer-par-choix-ou-par 6 | -------------------------------------------------------------------------------- /site_config/standard/econlog.econlib.org.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class="title"] 2 | author: //div[@class="hosted"]/a 3 | date: substring-after(//div[@class="dateline"]/text(), '|') 4 | 5 | strip: //a[@class="top" and @href="#"] 6 | test_url: http://econlog.econlib.org/archives/2012/04/blinder_on_heal.html -------------------------------------------------------------------------------- /site_config/standard/ekultura.hu.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class='style6 nevek'] 2 | 3 | body: //div[@class='bal3'] 4 | 5 | 6 | prune: yes 7 | 8 | tidy: yes 9 | convert_double_br_tags: yes 10 | 11 | test_url: http://ekultura.hu/olvasnivalo/egyeb/cikk/2010-12-15/interju-galvolgyi-judit-2010-december -------------------------------------------------------------------------------- /site_config/standard/franceculture.fr.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' text-zone ')] 2 | src_lazy_load_attr: data-dejavu-src 3 | 4 | test_url: https://www.franceculture.fr/emissions/entendez-vous-leco/paroles-deconomistes-24-passer-leconomie-le-role-de-luniversite 5 | -------------------------------------------------------------------------------- /site_config/standard/gizmodo.co.uk.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id="leadimage" or @class="postcontent"] 2 | author: //div[@class="contentauthor"] 3 | date: //div[@class="timestamp"] 4 | 5 | prune: no 6 | 7 | test_url: http://www.gizmodo.co.uk/2013/02/bbc-forcing-poor-old-sir-david-attenborough-to-go-on-twitter/ -------------------------------------------------------------------------------- /site_config/standard/gofugyourself.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, 'featured-embed-holder')] | //div[contains(@class, 'article-content') and @itemprop="articleBody"] 2 | 3 | strip_id_or_class: wp-polls 4 | 5 | test_url: https://www.gofugyourself.com/can-maren-morris-be-a-cinderella-03-2019 6 | -------------------------------------------------------------------------------- /site_config/standard/healthletter.mayoclinic.com.txt: -------------------------------------------------------------------------------- 1 | body: //main//div[contains(@class, 'article')] 2 | 3 | prune: no 4 | 5 | test_url: https://healthletter.mayoclinic.com/issues/june-2018/heat-illness 6 | test_url: https://healthletter.mayoclinic.com/issues/january-2019/freshening-bad-breath 7 | -------------------------------------------------------------------------------- /site_config/standard/kottke.org.txt: -------------------------------------------------------------------------------- 1 | title: //h2 2 | author: //*[@id='main']/div/a[1] 3 | date: substring-before(substring-after(//div[@class='meta'],'•'),'•') 4 | body: //div[@id='main'] 5 | strip: //div[@class='meta'] 6 | test_url: http://kottke.org/08/02/king-of-kong-a-fistful-of-quarters -------------------------------------------------------------------------------- /site_config/standard/netzpolitik.org.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class='entry-title'] 2 | author: //a[@ref='author'] 3 | date: //span[@class='entry-date'] 4 | body: //div[@class='entry-content'] 5 | strip_id_or_class: netzpolitik-cta 6 | 7 | test_url: http://netzpolitik.org/2011/buch-generation-facebook/ 8 | -------------------------------------------------------------------------------- /site_config/standard/phoronix.com.txt: -------------------------------------------------------------------------------- 1 | # based on the grabber rules of picofeed 2 | 3 | body: //div[@class="content"] 4 | test_url: http://www.phoronix.com/scan.php?page=article&item=amazon_ec2_bare&num=1 5 | 6 | # replace_string(

):

7 | 8 | next_page_link: //a[@title='Go To Next Page'] 9 | -------------------------------------------------------------------------------- /site_config/standard/usfirst.org.txt: -------------------------------------------------------------------------------- 1 | title: //meta[@property='dc:title']/@content 2 | date: //div[@class='content']//span[@property='dc:date']/@content 3 | body: //div[@property='content:encoded'] 4 | prune: no 5 | 6 | test_url: http://www.usfirst.org/roboticsprograms/frc/Photo-From-Kickoff-Filming -------------------------------------------------------------------------------- /site_config/standard/warriordudimanche.net.txt: -------------------------------------------------------------------------------- 1 | title: //article[contains(concat(' ',normalize-space(@class),' '),' article ')]//header//h1 2 | body: //article[contains(concat(' ',normalize-space(@class),' '),' article ')]//section 3 | 4 | test_url: http://warriordudimanche.net/article458/589065212a599 -------------------------------------------------------------------------------- /site_config/standard/wiki.guildwars2.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | body: //div[@id='content'] 3 | strip_id_or_class: editsection 4 | strip_id_or_class: toc 5 | strip: //div[@id='siteNotice'] 6 | strip: //div[@id='content']//table[last()] 7 | prune: no 8 | test_url: http://wiki.guildwars2.com/wiki/Guardian -------------------------------------------------------------------------------- /site_config/standard/aclu.org.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class='panel-panel panel-main-3 content-column'] 2 | title: //div[@class='panel-pane pane-node-title'] 3 | date: //div[@class='updated-date'] 4 | 5 | test_url: https://www.aclu.org/blog/free-future/chinas-nightmarish-citizen-scores-are-warning-americans 6 | -------------------------------------------------------------------------------- /site_config/standard/actualitte.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' article ')] 2 | author: //p[@class="auteur"]/a 3 | 4 | test_url: https://www.actualitte.com/article/monde-edition/jean-miguel-pire-n-est-plus-le-conseiller-culture-livre-et-lecture-de-nyssen/84556 5 | -------------------------------------------------------------------------------- /site_config/standard/blog.mozilla.org.txt: -------------------------------------------------------------------------------- 1 | title: //main[@id="content"]//h1 2 | author: //address 3 | date: //div[@class="entry-info"]//time/@datetime 4 | body: //div[@class="entry-content"] 5 | 6 | test_url: https://blog.mozilla.org/blog/2018/06/07/parlez-vous-deutsch-rhagor-o-leisiau-i-common-voice/ 7 | -------------------------------------------------------------------------------- /site_config/standard/coffeecircle.com.txt: -------------------------------------------------------------------------------- 1 | tidy: no 2 | prune: no 3 | body: //div[@class='post--title'] | //div[@class='blog-content'] 4 | strip_id_or_class: tag 5 | strip_id_or_class: photoset-grid 6 | strip_id_or_class: newsletter 7 | 8 | test_url: https://www.coffeecircle.com/de/b/cold-brew-eiskaffee 9 | -------------------------------------------------------------------------------- /site_config/standard/csswizardry.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | author: //b[@itemprop="author"] 3 | date: //time[@itemprop="datePublished"]/@datetime 4 | 5 | strip: //nav[@class="c-pagination"] 6 | strip: //a[@href="/services/"] 7 | 8 | test_url: https://csswizardry.com/2019/05/self-host-your-static-assets/ 9 | -------------------------------------------------------------------------------- /site_config/standard/digital-photography-school.com.txt: -------------------------------------------------------------------------------- 1 | title: //div[@class='post-title']/h1 2 | author: //a[@href='#author'] 3 | body: //div[@class='post-content'] 4 | strip: //div[@class='post-meta'] 5 | 6 | test_url: http://www.digital-photography-school.com/10-ways-to-develop-yourself-photographically -------------------------------------------------------------------------------- /site_config/standard/enikos.gr.txt: -------------------------------------------------------------------------------- 1 | body: //div[@id='article']//div[contains(@class, 'inside')] 2 | 3 | strip_id_or_class: tags 4 | strip_id_or_class: actions 5 | strip_id_or_class: google-ads 6 | 7 | prune: no 8 | 9 | test_url: http://www.enikos.gr/politics/98606,To_oxi_toy_Agorastoy_stoys_Germanoys.html -------------------------------------------------------------------------------- /site_config/standard/fictionpress.com.txt: -------------------------------------------------------------------------------- 1 | body: id('storytext') 2 | author: //a[starts-with(@href, '/u/')] 3 | #next_page_link: substring-after(//input[contains(@value, 'Next')]/@onclick, "self.location='") 4 | strip_id_or_class: 'a2a_kit' 5 | test_url: http://www.fictionpress.com/s/2897964/1/All_We_Knew -------------------------------------------------------------------------------- /site_config/standard/framablog.org.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | 3 | strip: //div[contains(@class, 'pdfprnt-buttons')] 4 | strip: //div[contains(@class, 'social-sharing')] 5 | strip: //div[@id="diaspora-button-container"] 6 | 7 | test_url: https://framablog.org/2018/05/31/peertube-vers-la-version-1-et-au-dela/ 8 | -------------------------------------------------------------------------------- /site_config/standard/generation-nt.com.txt: -------------------------------------------------------------------------------- 1 | author: //div[@class="titleinfo"]//a[@itemprop="author"] 2 | date: //div[@class="titleinfo"]//time[@itemprop="datePublished"]/@datetime 3 | 4 | test_url: https://www.generation-nt.com/connexion-sans-mot-passe-webauthn-w3c-standard-web-actualite-1962765.html 5 | -------------------------------------------------------------------------------- /site_config/standard/information.dk.txt: -------------------------------------------------------------------------------- 1 | title: //meta[@property='og:title']/@content 2 | author: //*[@property='dc:creator'] 3 | date: //*[@property='dc:date']/@content 4 | body: //div[@id='page-content']//div[contains(@class, 'article-body')] 5 | 6 | tidy: no 7 | test_url: http://www.information.dk/282307 -------------------------------------------------------------------------------- /site_config/standard/itsfoss.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1 2 | body: //article[contains(@class,'type-post')]/div[@itemprop='text'] 3 | strip: //a[@rel='dofollow'] 4 | strip: //div[contains(@class,'zem_rp_wrap')] 5 | strip: //div[@id='shr_canvas2'] 6 | test_url: https://itsfoss.com/fix-gvfsd-smb-high-cpu-ubuntu/ 7 | -------------------------------------------------------------------------------- /site_config/standard/kickstarter.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@id='name'] 2 | body: //*[@id='leftcol'] 3 | 4 | strip_id_or_class: 'share-box' 5 | strip_id_or_class: 'project-faqs' 6 | strip_id_or_class: 'report-issue-wrap' 7 | test_url: http://www.kickstarter.com/projects/hop/elevation-dock-the-best-dock-for-iphone -------------------------------------------------------------------------------- /site_config/standard/linux.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(concat(' ',normalize-space(@class),' '),' article-content ')] 2 | 3 | author: //div[contains(concat(' ',normalize-space(@class),' '),' td-post-author-name ')]/a 4 | 5 | test_url: https://www.linux.com/tutorials/lf-networking-releases-onap-dublin/ 6 | -------------------------------------------------------------------------------- /site_config/standard/marmiton.org.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class="main-title"] 2 | author: //span[@class="recipe-author__name"] 3 | body: //div[@id="sticky-desktop-only"] 4 | strip: //div[@id="bloc-video"] 5 | 6 | test_url: http://www.marmiton.org/recettes/recette_gateau-au-chocolat-fondant-rapide_166352.aspx 7 | -------------------------------------------------------------------------------- /site_config/standard/mentalfloss.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[contains(@class, 'article-content')][1] 2 | author: //div[contains(@class, 'author')][1] 3 | 4 | strip_id_or_class: date-text 5 | strip_id_or_class: after-article-subscribe-nl 6 | strip_id_or_class: eyebrow-text 7 | strip_id_or_class: eyebrow-tail 8 | -------------------------------------------------------------------------------- /site_config/standard/mirrorfootball.co.uk.txt: -------------------------------------------------------------------------------- 1 | # Remove extra links 2 | strip: //*[@class='appended_html'] 3 | test_url: http://www.mirrorfootball.co.uk/news/West-Ham-crisis-Carlton-Cole-slams-diabolical-performance-and-rips-into-Avram-Grant-lack-of-tactical-nous-following-Liverpool-mauling-article636151.html -------------------------------------------------------------------------------- /site_config/standard/n-tv.de.txt: -------------------------------------------------------------------------------- 1 | date: //span[@class='article__date'] 2 | title: //title 3 | body: //div[@class='article__text'] 4 | 5 | test_url: https://www.n-tv.de/politik/Die-alten-Maenner-werden-es-nie-blicken-article21447318.html 6 | 7 | # Last update: 2019-12-12 8 | # Tested with: Wallabag 2.3.8 -------------------------------------------------------------------------------- /site_config/standard/outsideonline.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@class="article__column--right"] 2 | strip_id_or_class: block-recirc 3 | strip_id_or_class: social 4 | strip: //a[@data-pin-custom="true"] 5 | 6 | test_url: http://www.outsideonline.com/2108066/emerald-citys-velo-thieves-have-problem-bike-batman 7 | -------------------------------------------------------------------------------- /site_config/standard/parislemon.com.txt: -------------------------------------------------------------------------------- 1 | title: //h2[@class="post-title"] 2 | author: substring-after(//div[@class="description"],'Words by ') 3 | date: //li[@class="date"] 4 | strip: //h2[@class="post-title"] 5 | body: //div[@class="copy"] 6 | test_url: http://parislemon.com/post/13462682469/the-15-inch-air -------------------------------------------------------------------------------- /site_config/standard/venturebeat.com.txt: -------------------------------------------------------------------------------- 1 | title: //h1[@class="entry-title"] 2 | author: //div[@class="author-name"] 3 | date: //span[@class="the-time"] 4 | body: //div[@class="entry-content"] 5 | strip: //div[@class="vb-gallery"] 6 | test_url: http://venturebeat.com/2012/07/17/marissa-mayer-yahoo/#s:mayer-1 -------------------------------------------------------------------------------- /site_config/standard/watoday.com.au.txt: -------------------------------------------------------------------------------- 1 | # strip the breadcrumb 2 | strip: //div//ul 3 | strip: //aside 4 | strip_id_or_class: adWrapper 5 | strip_id_or_class: noPrint 6 | 7 | test_url: https://www.watoday.com.au/business/the-economy/no-one-died-behind-the-typo-on-the-new-50-note-20190618-p51yqg.html 8 | -------------------------------------------------------------------------------- /site_config/standard/alternet.org.txt: -------------------------------------------------------------------------------- 1 | single_page_link: //div[contains(@class, 'story_tools')]//a[contains(@href, '/print/')] 2 | 3 | test_url: http://www.alternet.org/civil-liberties/noam-chomsky-surveillance-state-beyond-imagination-being-created-one-freest 4 | test_url: http://feeds.feedblitz.com/alternet -------------------------------------------------------------------------------- /site_config/standard/android-developers.googleblog.com.txt: -------------------------------------------------------------------------------- 1 | body: //div[@itemprop="articleBody"] 2 | 3 | find_string: