Merge remote-tracking branch 'upstream/master'

j0k3r · Aug 21, 2023 · e543537 · e543537
2 parents e9f1439 + 19c1290
commit e543537
Show file tree

Hide file tree

Showing 67 changed files with 1,266 additions and 432 deletions.
diff --git a/24auto.de.txt b/24auto.de.txt
@@ -1,12 +1,27 @@
-# Author: HolgerAusB  |  Version: 2022-03-22
+# Author: HolgerAusB  |  Version: 2023-08-17
 #
 # to get a source-feed, try to add 'rssfeed.rdf' to the category-URL e.g.
 #    https://www.example.com/hessen/rssfeed.rdf
 #
+# @ippen.media site
 #==========================
 
-# strip author box and social media box
-# these boxes sometimes prevented main picure to show up
+body: //article
+author: substring-after(//p[contains(@class, 'id-Story-authors')] , 'on:')
+author: //div[contains(@class, 'id-AuthorList')]/descendant::*[contains(@class, 'id-Link')]
+date: //p[contains(@class, 'id-Story-timestamp')]/descendant::time/@datetime
+
+strip_id_or_class: id-Article-dateActionboxCombo
+strip_id_or_class: id-Article-kicker
+strip_id_or_class: id-Article-headline
+strip_id_or_class: id-AuthorList
+strip_id_or_class: id-StoryElement-inArticleReco
+strip_id_or_class: id-Comments
+strip_id_or_class: id-Story-timestamp
+strip_id_or_class: id-Story-authors
+strip_id_or_class: id-Story-interactionBar
+strip: //a[@title='Bilderzoom']
+
 strip_id_or_class: idjs-simpletab-nav-item
 strip_id_or_class: idjs-simpletab-content-close
 strip_id_or_class: id-AuthorList
@@ -27,7 +42,6 @@ strip_id_or_class: id-Article-content-item.id-Article-advert
 strip_id_or_class: id-Article-advert--ad3
 strip_id_or_class: id-Article-advert
 
-tidy: yes
-prune: yes
-
+tidy: no
+prune: no
 test_url: https://www.24auto.de/news/adac-panne-elektro-auto-starterbatterie-akku-wachstum-verbrenner-tuev-wartung-muenchen-starthilfe-91426845.html
diff --git a/24garten.de.txt b/24garten.de.txt
@@ -1,12 +1,27 @@
-# Author: HolgerAusB  |  Version: 2022-03-22
+# Author: HolgerAusB  |  Version: 2023-08-17
 #
 # to get a source-feed, try to add 'rssfeed.rdf' to the category-URL e.g.
 #    https://www.example.com/hessen/rssfeed.rdf
 #
+# @ippen.media site
 #==========================
 
-# strip author box and social media box
-# these boxes sometimes prevented main picure to show up
+body: //article
+author: substring-after(//p[contains(@class, 'id-Story-authors')] , 'on:')
+author: //div[contains(@class, 'id-AuthorList')]/descendant::*[contains(@class, 'id-Link')]
+date: //p[contains(@class, 'id-Story-timestamp')]/descendant::time/@datetime
+
+strip_id_or_class: id-Article-dateActionboxCombo
+strip_id_or_class: id-Article-kicker
+strip_id_or_class: id-Article-headline
+strip_id_or_class: id-AuthorList
+strip_id_or_class: id-StoryElement-inArticleReco
+strip_id_or_class: id-Comments
+strip_id_or_class: id-Story-timestamp
+strip_id_or_class: id-Story-authors
+strip_id_or_class: id-Story-interactionBar
+strip: //a[@title='Bilderzoom']
+
 strip_id_or_class: idjs-simpletab-nav-item
 strip_id_or_class: idjs-simpletab-content-close
 strip_id_or_class: id-AuthorList
@@ -27,7 +42,6 @@ strip_id_or_class: id-Article-content-item.id-Article-advert
 strip_id_or_class: id-Article-advert--ad3
 strip_id_or_class: id-Article-advert
 
-tidy: yes
-prune: yes
-
+tidy: no
+prune: no
 test_url: https://www.24garten.de/mein-garten/basilikum-trocknen-einlegen-haltbarkeit-ernte-kraeuter-pesto-einfrieren-rosmarin-muenchen-zr-91424291.html
diff --git a/24hamburg.de.txt b/24hamburg.de.txt
@@ -1,12 +1,27 @@
-# Author: HolgerAusB  |  Version: 2022-03-22
+# Author: HolgerAusB  |  Version: 2023-08-17
 #
 # to get a source-feed, try to add 'rssfeed.rdf' to the category-URL e.g.
 #    https://www.example.com/hessen/rssfeed.rdf
 #
+# @ippen.media site
 #==========================
 
-# strip author box and social media box
-# these boxes sometimes prevented main picure to show up
+body: //article
+author: substring-after(//p[contains(@class, 'id-Story-authors')] , 'on:')
+author: //div[contains(@class, 'id-AuthorList')]/descendant::*[contains(@class, 'id-Link')]
+date: //p[contains(@class, 'id-Story-timestamp')]/descendant::time/@datetime
+
+strip_id_or_class: id-Article-dateActionboxCombo
+strip_id_or_class: id-Article-kicker
+strip_id_or_class: id-Article-headline
+strip_id_or_class: id-AuthorList
+strip_id_or_class: id-StoryElement-inArticleReco
+strip_id_or_class: id-Comments
+strip_id_or_class: id-Story-timestamp
+strip_id_or_class: id-Story-authors
+strip_id_or_class: id-Story-interactionBar
+strip: //a[@title='Bilderzoom']
+
 strip_id_or_class: idjs-simpletab-nav-item
 strip_id_or_class: idjs-simpletab-content-close
 strip_id_or_class: id-AuthorList
@@ -27,7 +42,6 @@ strip_id_or_class: id-Article-content-item.id-Article-advert
 strip_id_or_class: id-Article-advert--ad3
 strip_id_or_class: id-Article-advert
 
-tidy: yes
-prune: yes
-
+tidy: no
+prune: no
 test_url: https://www.24hamburg.de/hamburg/neue-corona-regeln-im-hvv-ab-sofort-entfaellt-3g-91422638.html
diff --git a/24rhein.de.txt b/24rhein.de.txt
@@ -1,12 +1,27 @@
-# Author: HolgerAusB  |  Version: 2022-03-22
+# Author: HolgerAusB  |  Version: 2023-08-17
 #
 # to get a source-feed, try to add 'rssfeed.rdf' to the category-URL e.g.
 #    https://www.example.com/hessen/rssfeed.rdf
 #
+# @ippen.media site
 #==========================
 
-# strip author box and social media box
-# these boxes sometimes prevented main picure to show up
+body: //article
+author: substring-after(//p[contains(@class, 'id-Story-authors')] , 'on:')
+author: //div[contains(@class, 'id-AuthorList')]/descendant::*[contains(@class, 'id-Link')]
+date: //p[contains(@class, 'id-Story-timestamp')]/descendant::time/@datetime
+
+strip_id_or_class: id-Article-dateActionboxCombo
+strip_id_or_class: id-Article-kicker
+strip_id_or_class: id-Article-headline
+strip_id_or_class: id-AuthorList
+strip_id_or_class: id-StoryElement-inArticleReco
+strip_id_or_class: id-Comments
+strip_id_or_class: id-Story-timestamp
+strip_id_or_class: id-Story-authors
+strip_id_or_class: id-Story-interactionBar
+strip: //a[@title='Bilderzoom']
+
 strip_id_or_class: idjs-simpletab-nav-item
 strip_id_or_class: idjs-simpletab-content-close
 strip_id_or_class: id-AuthorList
@@ -27,7 +42,6 @@ strip_id_or_class: id-Article-content-item.id-Article-advert
 strip_id_or_class: id-Article-advert--ad3
 strip_id_or_class: id-Article-advert
 
-tidy: yes
-prune: yes
-
+tidy: no
+prune: no
 test_url: https://www.24rhein.de/koeln/kalk/koeln-videoueberwachung-kameras-kalk-orte-ueberblick-polizei-ausbau-91388746.html
diff --git a/24vita.de.txt b/24vita.de.txt
@@ -1,12 +1,27 @@
-# Author: HolgerAusB  |  Version: 2022-03-22
+# Author: HolgerAusB  |  Version: 2023-08-17
 #
 # to get a source-feed, try to add 'rssfeed.rdf' to the category-URL e.g.
 #    https://www.example.com/hessen/rssfeed.rdf
 #
+# @ippen.media site
 #==========================
 
-# strip author box and social media box
-# these boxes sometimes prevented main picure to show up
+body: //article
+author: substring-after(//p[contains(@class, 'id-Story-authors')] , 'on:')
+author: //div[contains(@class, 'id-AuthorList')]/descendant::*[contains(@class, 'id-Link')]
+date: //p[contains(@class, 'id-Story-timestamp')]/descendant::time/@datetime
+
+strip_id_or_class: id-Article-dateActionboxCombo
+strip_id_or_class: id-Article-kicker
+strip_id_or_class: id-Article-headline
+strip_id_or_class: id-AuthorList
+strip_id_or_class: id-StoryElement-inArticleReco
+strip_id_or_class: id-Comments
+strip_id_or_class: id-Story-timestamp
+strip_id_or_class: id-Story-authors
+strip_id_or_class: id-Story-interactionBar
+strip: //a[@title='Bilderzoom']
+
 strip_id_or_class: idjs-simpletab-nav-item
 strip_id_or_class: idjs-simpletab-content-close
 strip_id_or_class: id-AuthorList
@@ -27,7 +42,6 @@ strip_id_or_class: id-Article-content-item.id-Article-advert
 strip_id_or_class: id-Article-advert--ad3
 strip_id_or_class: id-Article-advert
 
-tidy: yes
-prune: yes
-
+tidy: no
+prune: no
 test_url: https://www.24vita.de/verbraucher/allesesser-vegetarier-veganer-deutschland-daab-fleisch-herstellung-speiseplan-dge-bonn-91417014.html
diff --git a/abc.net.au.txt b/abc.net.au.txt
@@ -1,18 +1,16 @@
-title: //div[@class='article section']//h1
-author: //div[@class="byline"]/a
-date: //span[@class="timestamp"]
-body: //div[@class="page section"]
+body: //article/div[@data-component="ArticleWeb"]
 
-strip: //a[@class="inline-caption"]
-strip: //p[@class="ticker section noprint"]
-strip: //p[@class="topics"]
-strip: //h1
-strip: //div[@class="byline"]
-strip: //p[@class="published"]
-strip: //div[contains(@class,"featured-scroller")]
-strip_id_or_class: footer
+strip: //aside
+strip: //button
+strip: //header/div[contains(@class, 'Headline_meta')]
+strip: //div[contains(@class, 'ArticleWeb_publishedDate')]
+strip: //div[contains(@class, 'ArticleWeb_shareBottom')]/self::div | //div[contains(@class, 'ArticleWeb_shareBottom')]/following-sibling::*
+strip: //div[contains(@aria-label, 'media') and contains(@aria-label, 'embed')]
 
-tidy: no
+prune: no
 
-test_url: http://www.abc.net.au/news/2013-03-27/open-speed-highways-change-clp-giles/4597892
-test_url: http://www.abc.net.au/news/2013-04-30/credit-growth-remains-subdued/4660054?section=business
+test_url: https://www.abc.net.au/news/2013-04-30/credit-growth-remains-subdued/4660054
+test_url: https://www.abc.net.au/news/2013-03-27/open-speed-highways-change-clp-giles/4597892
+
+test_url: https://www.abc.net.au/news/2023-06-16/fact-check-jacinta-nampijinpa-price-secret-documents-niaa/102485040
+test_contains: widespread but bogus claim about a list of secret "demands" linked to the Voice referendum
diff --git a/abendzeitung-muenchen.de.txt b/abendzeitung-muenchen.de.txt
@@ -0,0 +1,22 @@
+body: //article
+
+strip_id_or_class: teaserbox
+strip_id_or_class: aufmacherbox
+strip_id_or_class: mehrzumthema
+strip_id_or_class: artdetail_time
+strip_id_or_class: artdetail_social
+strip_id_or_class: artdetail_desc
+
+strip: //footer
+
+# needed for wallabag:
+strip: //picture/source
+
+# strip additional text within figure after caption
+# wallabag/f43.me shows here double captions
+# see second image at
+# https://www.abendzeitung-muenchen.de/muenchen/wahlkampf-in-bayern-soeder-und-aiwanger-ringen-um-die-bierzelt-herrschaft-art-921576
+strip: //figure/figcaption/following-sibling::text() | //figure/figcaption/following-sibling::p
+
+test_url: https://www.abendzeitung-muenchen.de/muenchen/mvg-fast-alle-fahrschein-automaten-in-muenchen-kommen-weg-art-921634
+test_url: https://www.abendzeitung-muenchen.de/muenchen/wahlkampf-in-bayern-soeder-und-aiwanger-ringen-um-die-bierzelt-herrschaft-art-921576
diff --git a/az-online.de.txt b/az-online.de.txt
@@ -1,12 +1,27 @@
-# Author: HolgerAusB  |  Version: 2022-03-22
+# Author: HolgerAusB  |  Version: 2023-08-17
 #
 # to get a source-feed, try to add 'rssfeed.rdf' to the category-URL e.g.
 #    https://www.example.com/hessen/rssfeed.rdf
 #
+# @ippen.media site
 #==========================
 
-# strip author box and social media box
-# these boxes sometimes prevented main picure to show up
+body: //article
+author: substring-after(//p[contains(@class, 'id-Story-authors')] , 'on:')
+author: //div[contains(@class, 'id-AuthorList')]/descendant::*[contains(@class, 'id-Link')]
+date: //p[contains(@class, 'id-Story-timestamp')]/descendant::time/@datetime
+
+strip_id_or_class: id-Article-dateActionboxCombo
+strip_id_or_class: id-Article-kicker
+strip_id_or_class: id-Article-headline
+strip_id_or_class: id-AuthorList
+strip_id_or_class: id-StoryElement-inArticleReco
+strip_id_or_class: id-Comments
+strip_id_or_class: id-Story-timestamp
+strip_id_or_class: id-Story-authors
+strip_id_or_class: id-Story-interactionBar
+strip: //a[@title='Bilderzoom']
+
 strip_id_or_class: idjs-simpletab-nav-item
 strip_id_or_class: idjs-simpletab-content-close
 strip_id_or_class: id-AuthorList
@@ -27,7 +42,6 @@ strip_id_or_class: id-Article-content-item.id-Article-advert
 strip_id_or_class: id-Article-advert--ad3
 strip_id_or_class: id-Article-advert
 
-tidy: yes
-prune: yes
-
+tidy: no
+prune: no
 test_url: https://www.az-online.de/uelzen/stadt-uelzen/mehr-gaeste-in-der-uelzener-stadthalle-als-erlaubt-91282522.html
diff --git a/bgland24.de.txt b/bgland24.de.txt
@@ -1,12 +1,27 @@
-# Author: HolgerAusB  |  Version: 2022-03-22
+# Author: HolgerAusB  |  Version: 2023-08-17
 #
 # to get a source-feed, try to add 'rssfeed.rdf' to the category-URL e.g.
 #    https://www.example.com/hessen/rssfeed.rdf
 #
+# @ippen.media site
 #==========================
 
-# strip author box and social media box
-# these boxes sometimes prevented main picure to show up
+body: //article
+author: substring-after(//p[contains(@class, 'id-Story-authors')] , 'on:')
+author: //div[contains(@class, 'id-AuthorList')]/descendant::*[contains(@class, 'id-Link')]
+date: //p[contains(@class, 'id-Story-timestamp')]/descendant::time/@datetime
+
+strip_id_or_class: id-Article-dateActionboxCombo
+strip_id_or_class: id-Article-kicker
+strip_id_or_class: id-Article-headline
+strip_id_or_class: id-AuthorList
+strip_id_or_class: id-StoryElement-inArticleReco
+strip_id_or_class: id-Comments
+strip_id_or_class: id-Story-timestamp
+strip_id_or_class: id-Story-authors
+strip_id_or_class: id-Story-interactionBar
+strip: //a[@title='Bilderzoom']
+
 strip_id_or_class: idjs-simpletab-nav-item
 strip_id_or_class: idjs-simpletab-content-close
 strip_id_or_class: id-AuthorList
@@ -27,7 +42,6 @@ strip_id_or_class: id-Article-content-item.id-Article-advert
 strip_id_or_class: id-Article-advert--ad3
 strip_id_or_class: id-Article-advert
 
-tidy: yes
-prune: yes
-
+tidy: no
+prune: no
 test_url: https://www.bgland24.de/bgland/region-berchtesgaden/bischofswiesen-ort28409/schoenau-am-koenigssee-wieder-grosses-interesse-bei-fit-durch-unser-gmoa-2022-91285772.html
diff --git a/buzzfeed.de.txt b/buzzfeed.de.txt
@@ -1,12 +1,27 @@
-# Author: HolgerAusB  |  Version: 2022-03-22
+# Author: HolgerAusB  |  Version: 2023-08-17
 #
 # to get a source-feed, try to add 'rssfeed.rdf' to the category-URL e.g.
 #    https://www.example.com/hessen/rssfeed.rdf
 #
+# @ippen.media site
 #==========================
 
-# strip author box and social media box
-# these boxes sometimes prevented main picure to show up
+body: //article
+author: substring-after(//p[contains(@class, 'id-Story-authors')] , 'on:')
+author: //div[contains(@class, 'id-AuthorList')]/descendant::*[contains(@class, 'id-Link')]
+date: //p[contains(@class, 'id-Story-timestamp')]/descendant::time/@datetime
+
+strip_id_or_class: id-Article-dateActionboxCombo
+strip_id_or_class: id-Article-kicker
+strip_id_or_class: id-Article-headline
+strip_id_or_class: id-AuthorList
+strip_id_or_class: id-StoryElement-inArticleReco
+strip_id_or_class: id-Comments
+strip_id_or_class: id-Story-timestamp
+strip_id_or_class: id-Story-authors
+strip_id_or_class: id-Story-interactionBar
+strip: //a[@title='Bilderzoom']
+
 strip_id_or_class: idjs-simpletab-nav-item
 strip_id_or_class: idjs-simpletab-content-close
 strip_id_or_class: id-AuthorList
@@ -27,7 +42,6 @@ strip_id_or_class: id-Article-content-item.id-Article-advert
 strip_id_or_class: id-Article-advert--ad3
 strip_id_or_class: id-Article-advert
 
-tidy: yes
-prune: yes
-
+tidy: no
+prune: no
 test_url: https://www.buzzfeed.de/buzz/19-buecher-die-leute-einfach-nicht-zu-ende-lesen-konnten-91400849.html