diff --git a/lib/html2rss/configs/adb.org/news.yml b/lib/html2rss/configs/adb.org/news.yml
index 28c15220..59b50b04 100644
--- a/lib/html2rss/configs/adb.org/news.yml
+++ b/lib/html2rss/configs/adb.org/news.yml
@@ -3,23 +3,17 @@ directory:
topics:
- finance
- civic
- title: "Asian Development Bank — Newsroom"
- summary: "Development financing updates and press releases from the Asian Development Bank."
-
+ title: Asian Development Bank — News Releases
+ summary:
+ Development financing updates and press releases from the Asian Development
+ Bank.
channel:
- title: "Asian Development Bank — Newsroom"
- url: https://www.adb.org/news
+ title: Asian Development Bank — News Releases
+ url: https://www.adb.org/news/releases
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: "div.card"
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: h3 a[href^="/news/"]
+ enhance: true
diff --git a/lib/html2rss/configs/agora-energiewende.org/news-events.yml b/lib/html2rss/configs/agora-energiewende.org/news-events.yml
index be19890d..e916aa78 100644
--- a/lib/html2rss/configs/agora-energiewende.org/news-events.yml
+++ b/lib/html2rss/configs/agora-energiewende.org/news-events.yml
@@ -3,23 +3,17 @@ directory:
topics:
- energy
- environment
- title: "Agora Energiewende — News & Events"
- summary: "Decarbonisation roadmaps, power sector transformation, and climate neutrality strategies."
-
+ title: Agora Energiewende — News & Events
+ summary:
+ Decarbonisation roadmaps, power sector transformation, and climate neutrality
+ strategies.
channel:
- title: "Agora Energiewende — News & Events"
+ title: Agora Energiewende — News & Events
url: https://www.agora-energiewende.org/news-events
language: en
- time_zone: UTC
+ time_zone: Europe/Berlin
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: .teaser a
+ enhance: true
diff --git a/lib/html2rss/configs/bundesfinanzhof.de/de_presse_pressemitteilungen.yml b/lib/html2rss/configs/bundesfinanzhof.de/de_presse_pressemitteilungen.yml
index e8790bde..4d6d64f1 100644
--- a/lib/html2rss/configs/bundesfinanzhof.de/de_presse_pressemitteilungen.yml
+++ b/lib/html2rss/configs/bundesfinanzhof.de/de_presse_pressemitteilungen.yml
@@ -9,17 +9,15 @@ directory:
channel:
title: "Bundesfinanzhof — Pressemitteilungen"
url: https://www.bundesfinanzhof.de/de/presse/pressemitteilungen/
- language: en
- time_zone: UTC
+ language: de
+ time_zone: Europe/Berlin
ttl: 360
selectors:
items:
- selector: "article"
+ selector: "ul.o-article-list > li > article"
enhance: false
title:
- selector: "h1, h2, h3, h4, h5, h6"
+ selector: "h2.m-article__headline"
url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
+ selector: "a.a-link--arrowed"
extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/bundeskanzleramt.gv.at/bundeskanzleramt_nachrichten-der-bundesregierung.yml b/lib/html2rss/configs/bundeskanzleramt.gv.at/bundeskanzleramt_nachrichten-der-bundesregierung.yml
index e1e262c9..1c0604d0 100644
--- a/lib/html2rss/configs/bundeskanzleramt.gv.at/bundeskanzleramt_nachrichten-der-bundesregierung.yml
+++ b/lib/html2rss/configs/bundeskanzleramt.gv.at/bundeskanzleramt_nachrichten-der-bundesregierung.yml
@@ -3,23 +3,15 @@ directory:
topics:
- civic
- news
- title: "Bundeskanzleramt Österreich — Nachrichten"
- summary: "Ministerratsbeschlüsse und Regierungserklärungen aus dem Bundeskanzleramt Österreich."
-
+ title: Bundeskanzleramt Österreich — Nachrichten
+ summary: Austrian federal government news releases and announcements.
channel:
- title: "Bundeskanzleramt Österreich — Nachrichten"
- url: https://www.bundeskanzleramt.gv.at/bundeskanzleramt/nachrichten-der-bundesregierung.html
- language: en
- time_zone: UTC
+ title: Bundeskanzleramt Österreich — Nachrichten
+ url: https://www.bundeskanzleramt.gv.at/bundeskanzleramt/nachrichten-der-bundesregierung/2026.html
+ language: de
+ time_zone: Europe/Vienna
ttl: 360
selectors:
items:
- selector: "div.card"
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: a[href*="/nachrichten-der-bundesregierung/2026/"]
+ enhance: true
diff --git a/lib/html2rss/configs/canada.ca/en_stories.yml b/lib/html2rss/configs/canada.ca/en_stories.yml
index f938eac3..988a6f1c 100644
--- a/lib/html2rss/configs/canada.ca/en_stories.yml
+++ b/lib/html2rss/configs/canada.ca/en_stories.yml
@@ -3,23 +3,18 @@ directory:
topics:
- science
- research
- title: "NRC Canada — Research Stories"
- summary: "Industrial research, quantum computing, and metrology innovations from NRC Canada."
-
+ title: NRC Canada — Research Stories
+ summary:
+ Industrial research, quantum computing, and metrology innovations from
+ NRC Canada.
channel:
- title: "NRC Canada — Research Stories"
+ title: NRC Canada — Research Stories
url: https://www.nrc.canada.ca/en/stories
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: main > section > div > article > div > div > div
+ enhance: true
+strategy: faraday
diff --git a/lib/html2rss/configs/cer-rec.gc.ca/en_about_news-room_news-releases.yml b/lib/html2rss/configs/cer-rec.gc.ca/en_about_news-room_news-releases.yml
index 3d626573..d52932d4 100644
--- a/lib/html2rss/configs/cer-rec.gc.ca/en_about_news-room_news-releases.yml
+++ b/lib/html2rss/configs/cer-rec.gc.ca/en_about_news-room_news-releases.yml
@@ -3,23 +3,17 @@ directory:
topics:
- energy
- environment
- title: "Canada Energy Regulator — News Releases"
- summary: "Pipeline safety, energy infrastructure approvals, and energy export data from Canada."
-
+ title: Canada Energy Regulator — News Releases
+ summary:
+ Pipeline safety, energy infrastructure approvals, and energy export data
+ from Canada.
channel:
- title: "Canada Energy Regulator — News Releases"
+ title: Canada Energy Regulator — News Releases
url: https://www.cer-rec.gc.ca/en/about/news-room/news-releases/
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: 'ul[class*="list"] > li'
- enhance: false
- title:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: ul[class*="list"] > li a
+ enhance: true
diff --git a/lib/html2rss/configs/conseil-constitutionnel.fr/actualites.yml b/lib/html2rss/configs/conseil-constitutionnel.fr/actualites.yml
index 440bb643..621b4c2b 100644
--- a/lib/html2rss/configs/conseil-constitutionnel.fr/actualites.yml
+++ b/lib/html2rss/configs/conseil-constitutionnel.fr/actualites.yml
@@ -3,23 +3,17 @@ directory:
topics:
- civic
- news
- title: "Conseil Constitutionnel France — Actualités"
- summary: "Décisions de conformité, déclarations et actualités du Conseil constitutionnel de France."
-
+ title: Conseil Constitutionnel France — Actualités
+ summary:
+ Décisions de conformité, déclarations et actualités du Conseil constitutionnel
+ de France.
channel:
- title: "Conseil Constitutionnel France — Actualités"
+ title: Conseil Constitutionnel France — Actualités
url: https://www.conseil-constitutionnel.fr/actualites
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: div[class*="item"] a
+ enhance: true
diff --git a/lib/html2rss/configs/csem.ch/en_news.yml b/lib/html2rss/configs/csem.ch/en_news.yml
index 3a3fd001..159beca8 100644
--- a/lib/html2rss/configs/csem.ch/en_news.yml
+++ b/lib/html2rss/configs/csem.ch/en_news.yml
@@ -3,23 +3,23 @@ directory:
topics:
- tech
- research
- title: "CSEM Switzerland — News & Innovation"
- summary: "Microtechnology, photovoltaics, and precision manufacturing innovations from CSEM Switzerland."
-
+ title: CSEM Switzerland — News & Innovation
+ summary:
+ Microtechnology, photovoltaics, and precision manufacturing innovations
+ from CSEM Switzerland.
channel:
- title: "CSEM Switzerland — News & Innovation"
+ title: CSEM Switzerland — News & Innovation
url: https://www.csem.ch/en/news/
language: en
- time_zone: UTC
+ time_zone: Europe/Zurich
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
+ selector: main article.item
title:
- selector: "h1, h2, h3, h4, h5, h6"
+ selector: h3
url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
+ selector: a
extractor: href
- published_at:
- selector: "time"
+ description:
+ selector: p
diff --git a/lib/html2rss/configs/csiro.au/en_news.yml b/lib/html2rss/configs/csiro.au/en_news.yml
index 2740b9ba..7be40b23 100644
--- a/lib/html2rss/configs/csiro.au/en_news.yml
+++ b/lib/html2rss/configs/csiro.au/en_news.yml
@@ -1,25 +1,17 @@
# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
directory:
topics:
- - science
- research
- title: "CSIRO — News Highlights"
- summary: "Australian national science agency research across agriculture, energy, space, and environment."
-
+ - science
+ title: CSIRO — News
+ summary: Australian scientific research news and expert commentary from CSIRO.
channel:
- title: "CSIRO — News Highlights"
+ title: CSIRO — News
url: https://www.csiro.au/en/news
language: en
- time_zone: UTC
+ time_zone: Australia/Sydney
ttl: 360
selectors:
items:
- selector: "div.card"
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: a[href*="/Articles/"]
+ enhance: true
diff --git a/lib/html2rss/configs/diw.de/en_diw_01.c.603338.en_news.yml b/lib/html2rss/configs/diw.de/en_diw_01.c.603338.en_news.yml
deleted file mode 100644
index 50a9fb12..00000000
--- a/lib/html2rss/configs/diw.de/en_diw_01.c.603338.en_news.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - finance
- - research
- title: "DIW Berlin — Economic Research News"
- summary: "German Institute for Economic Research macroeconomic forecasts and policy briefs."
-
-channel:
- title: "DIW Berlin — Economic Research News"
- url: https://www.diw.de/en/diw_01.c.603338.en/news.html
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: 'div[class*="item"]'
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/dlr.de/en_latest_news.yml b/lib/html2rss/configs/dlr.de/en_latest_news.yml
deleted file mode 100644
index f88c7aa1..00000000
--- a/lib/html2rss/configs/dlr.de/en_latest_news.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - science
- - research
- title: "DLR — News & Highlights"
- summary: "Aeronautics, space, energy, and transport research highlights from the German Aerospace Center."
-
-channel:
- title: "DLR — News & Highlights"
- url: https://www.dlr.de/en/latest/news
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: 'div[class*="item"]'
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/ecologie.gouv.fr/presse.yml b/lib/html2rss/configs/ecologie.gouv.fr/presse.yml
index 6057b2da..0e2d614c 100644
--- a/lib/html2rss/configs/ecologie.gouv.fr/presse.yml
+++ b/lib/html2rss/configs/ecologie.gouv.fr/presse.yml
@@ -3,23 +3,18 @@ directory:
topics:
- environment
- energy
- title: "Transition Écologique — Espace Presse"
- summary: "Décarbonation, énergies renouvelables, transports propres et biodiversité en France."
-
+ title: Transition Écologique — Espace Presse
+ summary:
+ Décarbonation, énergies renouvelables, transports propres et biodiversité
+ en France.
channel:
- title: "Transition Écologique — Espace Presse"
+ title: Transition Écologique — Espace Presse
url: https://www.ecologie.gouv.fr/presse
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: 'ul[class*="list"] > li'
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: div.fr-col
+ enhance: true
+strategy: faraday
diff --git a/lib/html2rss/configs/ed.ac.uk/news.yml b/lib/html2rss/configs/ed.ac.uk/news.yml
index d2287be6..4c47ae34 100644
--- a/lib/html2rss/configs/ed.ac.uk/news.yml
+++ b/lib/html2rss/configs/ed.ac.uk/news.yml
@@ -3,23 +3,15 @@ directory:
topics:
- research
- science
- title: "University of Edinburgh — Newsroom"
- summary: "Scottish research innovations, medical science, and artificial intelligence discoveries from Edinburgh."
-
+ title: University of Edinburgh — Latest News
+ summary: Research and university news from the University of Edinburgh.
channel:
- title: "University of Edinburgh — Newsroom"
- url: https://www.ed.ac.uk/news
+ title: University of Edinburgh — Latest News
+ url: https://www.ed.ac.uk/news/latest
language: en
- time_zone: UTC
+ time_zone: Europe/London
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: div.card
+ enhance: true
diff --git a/lib/html2rss/configs/finma.ch/en_news.yml b/lib/html2rss/configs/finma.ch/en_news.yml
deleted file mode 100644
index f58b1619..00000000
--- a/lib/html2rss/configs/finma.ch/en_news.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - finance
- - civic
- title: "FINMA — News & Media"
- summary: "Financial market regulation, resolution, and supervision news from FINMA Switzerland."
-
-channel:
- title: "FINMA — News & Media"
- url: https://www.finma.ch/en/news/
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: 'div[class*="teaser"]'
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/forskningsradet.no/en_news.yml b/lib/html2rss/configs/forskningsradet.no/en_news.yml
index 865814a8..df9ce2d2 100644
--- a/lib/html2rss/configs/forskningsradet.no/en_news.yml
+++ b/lib/html2rss/configs/forskningsradet.no/en_news.yml
@@ -3,23 +3,18 @@ directory:
topics:
- science
- research
- title: "Research Council of Norway — News"
- summary: "Norwegian national scientific funding, green transition grants, and research policy."
-
+ title: Research Council of Norway — News
+ summary:
+ Norwegian national scientific funding, green transition grants, and research
+ policy.
channel:
- title: "Research Council of Norway — News"
+ title: Research Council of Norway — News
url: https://www.forskningsradet.no/en/news/
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: 'ul[class*="list"] > li'
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: main > div > div > div
+ enhance: true
+strategy: faraday
diff --git a/lib/html2rss/configs/fz-juelich.de/en_news_press-releases.yml b/lib/html2rss/configs/fz-juelich.de/en_news_press-releases.yml
deleted file mode 100644
index 51ce40fa..00000000
--- a/lib/html2rss/configs/fz-juelich.de/en_news_press-releases.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - science
- - research
- title: "Forschungszentrum Jülich — Press Releases"
- summary: "Supercomputing, quantum computing, brain research, and bioeconomy from Jülich Germany."
-
-channel:
- title: "Forschungszentrum Jülich — Press Releases"
- url: https://www.fz-juelich.de/en/news/press-releases
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: "li.item"
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/hkma.gov.hk/eng_news-and-media_press-releases.yml b/lib/html2rss/configs/hkma.gov.hk/eng_news-and-media_press-releases.yml
index 9fb13175..71505fd4 100644
--- a/lib/html2rss/configs/hkma.gov.hk/eng_news-and-media_press-releases.yml
+++ b/lib/html2rss/configs/hkma.gov.hk/eng_news-and-media_press-releases.yml
@@ -3,23 +3,16 @@ directory:
topics:
- finance
- civic
- title: "Hong Kong Monetary Authority — Press Releases"
- summary: "Monetary base, banking stability, and exchange fund releases from the HKMA."
-
+ title: Hong Kong Monetary Authority — Press Releases
+ summary: Monetary base, banking stability, and exchange fund releases from the HKMA.
channel:
- title: "Hong Kong Monetary Authority — Press Releases"
+ title: Hong Kong Monetary Authority — Press Releases
url: https://www.hkma.gov.hk/eng/news-and-media/press-releases/
language: en
- time_zone: UTC
+ time_zone: Asia/Hong_Kong
ttl: 360
selectors:
items:
- selector: 'div[class*="item"]'
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: a[href*="/press-releases/20"]
+ enhance: true
+strategy: botasaurus
diff --git a/lib/html2rss/configs/indec.gob.ar/indec_web_noticias-prensa.yml b/lib/html2rss/configs/indec.gob.ar/indec_web_noticias-prensa.yml
deleted file mode 100644
index 4b4c6611..00000000
--- a/lib/html2rss/configs/indec.gob.ar/indec_web_noticias-prensa.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - civic
- - research
- title: "INDEC Argentina — Noticias de Prensa"
- summary: "Instituto Nacional de Estadística y Censos estadísticas públicas, precios y actividad económica en Argentina."
-
-channel:
- title: "INDEC Argentina — Noticias de Prensa"
- url: https://www.indec.gob.ar/indec/web/Noticias-Prensa
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: 'div[class*="card"]'
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/inegi.org.mx/saladeprensa.yml b/lib/html2rss/configs/inegi.org.mx/saladeprensa.yml
deleted file mode 100644
index 9f9f6f36..00000000
--- a/lib/html2rss/configs/inegi.org.mx/saladeprensa.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - civic
- - research
- title: "INEGI Mexico — Sala de Prensa"
- summary: "Instituto Nacional de Estadística y Geografía indicadores económicos, inflación y demografía de México."
-
-channel:
- title: "INEGI Mexico — Sala de Prensa"
- url: https://www.inegi.org.mx/saladeprensa/
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: "div.card"
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/mak.at/en_press.yml b/lib/html2rss/configs/mak.at/en_press.yml
deleted file mode 100644
index 9d06af0b..00000000
--- a/lib/html2rss/configs/mak.at/en_press.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - entertainment
- - civic
- title: "MAK Vienna — Applied Arts Museum Press"
- summary: "Applied arts, contemporary design, and architectural exhibitions from MAK Vienna."
-
-channel:
- title: "MAK Vienna — Applied Arts Museum Press"
- url: https://www.mak.at/en/press
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: 'div[class*="item"]'
- enhance: false
- title:
- selector: 'a.title, a.heading, a[class*="title"], a[class*="heading"]'
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/ndl.go.jp/en_news.yml b/lib/html2rss/configs/ndl.go.jp/en_news.yml
index e0ba442a..3d6760cc 100644
--- a/lib/html2rss/configs/ndl.go.jp/en_news.yml
+++ b/lib/html2rss/configs/ndl.go.jp/en_news.yml
@@ -3,23 +3,18 @@ directory:
topics:
- entertainment
- research
- title: "National Diet Library Japan — News & Events"
- summary: "Parliamentary library collections, bibliographic data, and preservation news from NDL Japan."
-
+ title: National Diet Library Japan — News & Events
+ summary:
+ Parliamentary library collections, bibliographic data, and preservation
+ news from NDL Japan.
channel:
- title: "National Diet Library Japan — News & Events"
+ title: National Diet Library Japan — News & Events
url: https://www.ndl.go.jp/en/news/
language: en
- time_zone: UTC
+ time_zone: Asia/Tokyo
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: a[href*="/en/news/fy"]
+ enhance: true
+strategy: botasaurus
diff --git a/lib/html2rss/configs/nea.gov.sg/media_news_news.yml b/lib/html2rss/configs/nea.gov.sg/media_news_news.yml
deleted file mode 100644
index ad7b3e45..00000000
--- a/lib/html2rss/configs/nea.gov.sg/media_news_news.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - environment
- - civic
- title: "National Environment Agency Singapore — News"
- summary: "Air quality monitoring, waste reduction, carbon tax administration, and dengue prevention."
-
-channel:
- title: "National Environment Agency Singapore — News"
- url: https://www.nea.gov.sg/media/news/news
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: 'ul[class*="list"] > li'
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/parks.vic.gov.au/news.yml b/lib/html2rss/configs/parks.vic.gov.au/news.yml
index 9f722a14..ff0ac0b1 100644
--- a/lib/html2rss/configs/parks.vic.gov.au/news.yml
+++ b/lib/html2rss/configs/parks.vic.gov.au/news.yml
@@ -3,23 +3,18 @@ directory:
topics:
- environment
- travel
- title: "Parks Victoria — News & Media"
- summary: "Bushland conservation, park access updates, and wildlife projects across Victoria Australia."
-
+ title: Parks Victoria — News & Media
+ summary:
+ Bushland conservation, park access updates, and wildlife projects across
+ Victoria Australia.
channel:
- title: "Parks Victoria — News & Media"
+ title: Parks Victoria — News & Media
url: https://www.parks.vic.gov.au/news
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: 'li[class*="item"]'
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: div.col-sm-10
+ enhance: true
+strategy: faraday
diff --git a/lib/html2rss/configs/piie.com/research.yml b/lib/html2rss/configs/piie.com/research.yml
index c17665b4..762b467a 100644
--- a/lib/html2rss/configs/piie.com/research.yml
+++ b/lib/html2rss/configs/piie.com/research.yml
@@ -3,23 +3,17 @@ directory:
topics:
- finance
- research
- title: "PIIE — Peterson Institute for International Economics"
- summary: "Global trade policy, tariffs, macroeconomic models, and central banking from PIIE."
-
+ title: PIIE — Peterson Institute for International Economics
+ summary:
+ Global trade policy, tariffs, macroeconomic models, and central banking
+ from PIIE.
channel:
- title: "PIIE — Peterson Institute for International Economics"
+ title: PIIE — Peterson Institute for International Economics
url: https://www.piie.com/research
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: a[href*="/research/"][href*="-"]
+ enhance: true
diff --git a/lib/html2rss/configs/resbank.co.za/en_home_publications_media-releases.yml b/lib/html2rss/configs/resbank.co.za/en_home_publications_media-releases.yml
deleted file mode 100644
index 938a2c14..00000000
--- a/lib/html2rss/configs/resbank.co.za/en_home_publications_media-releases.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - finance
- - civic
- title: "South African Reserve Bank — Media Releases"
- summary: "Monetary Policy Committee statements and banking regulation updates from SARB South Africa."
-
-channel:
- title: "South African Reserve Bank — Media Releases"
- url: https://www.resbank.co.za/en/home/publications/media-releases
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: 'div[class*="item"]'
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/sahpra.org.za/press-releases.yml b/lib/html2rss/configs/sahpra.org.za/press-releases.yml
deleted file mode 100644
index 98ae718a..00000000
--- a/lib/html2rss/configs/sahpra.org.za/press-releases.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - science
- - consumer
- title: "SAHPRA — South African Health Products Authority"
- summary: "Health product regulation, clinical trial authorizations, and safety alerts in South Africa."
-
-channel:
- title: "SAHPRA — South African Health Products Authority"
- url: https://www.sahpra.org.za/press-releases/
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: 'div[class*="post"]'
- enhance: false
- title:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/standards.org.au/news.yml b/lib/html2rss/configs/standards.org.au/news.yml
index d2ce82ff..d82f47c9 100644
--- a/lib/html2rss/configs/standards.org.au/news.yml
+++ b/lib/html2rss/configs/standards.org.au/news.yml
@@ -3,23 +3,18 @@ directory:
topics:
- product
- consumer
- title: "Standards Australia — News & Insights"
- summary: "National industry standards, safety codes, and technological benchmarks from Standards Australia."
-
+ title: Standards Australia — News & Insights
+ summary:
+ National industry standards, safety codes, and technological benchmarks
+ from Standards Australia.
channel:
- title: "Standards Australia — News & Insights"
+ title: Standards Australia — News & Insights
url: https://www.standards.org.au/news
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: 'div[class*="article"]'
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: div.articles-text-div
+ enhance: true
+strategy: faraday
diff --git a/lib/html2rss/configs/swissmedic.ch/swissmedic_en_home_news_mitteilungen.yml b/lib/html2rss/configs/swissmedic.ch/swissmedic_en_home_news_mitteilungen.yml
index 73f2ac93..bae6905d 100644
--- a/lib/html2rss/configs/swissmedic.ch/swissmedic_en_home_news_mitteilungen.yml
+++ b/lib/html2rss/configs/swissmedic.ch/swissmedic_en_home_news_mitteilungen.yml
@@ -3,23 +3,17 @@ directory:
topics:
- science
- consumer
- title: "Swissmedic — Media Releases"
- summary: "Swiss Agency for Therapeutic Products medicine authorizations and medical device surveillance."
-
+ title: Swissmedic — Media Releases
+ summary:
+ Swiss Agency for Therapeutic Products medicine authorizations and medical
+ device surveillance.
channel:
- title: "Swissmedic — Media Releases"
+ title: Swissmedic — Media Releases
url: https://www.swissmedic.ch/swissmedic/en/home/news/mitteilungen.html
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: 'div[class*="teaser"]'
- enhance: false
- title:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: div[class*="teaser"]
+ enhance: true
diff --git a/lib/html2rss/configs/tum.de/en_news-and-events_all-news_press-releases.yml b/lib/html2rss/configs/tum.de/en_news-and-events_all-news_press-releases.yml
index f6bff34e..97898ff0 100644
--- a/lib/html2rss/configs/tum.de/en_news-and-events_all-news_press-releases.yml
+++ b/lib/html2rss/configs/tum.de/en_news-and-events_all-news_press-releases.yml
@@ -3,23 +3,25 @@ directory:
topics:
- tech
- research
- title: "Technical University of Munich — Press Releases"
- summary: "Robotics, aerospace, green tech, and engineering breakthroughs from TU Munich Germany."
-
+ title: Technical University of Munich — Press Releases
+ summary:
+ Robotics, aerospace, green tech, and engineering breakthroughs from TU
+ Munich Germany.
channel:
- title: "Technical University of Munich — Press Releases"
+ title: Technical University of Munich — Press Releases
url: https://www.tum.de/en/news-and-events/all-news/press-releases
language: en
- time_zone: UTC
+ time_zone: Europe/Berlin
ttl: 360
selectors:
items:
- selector: 'ul[class*="news"] > li'
- enhance: false
+ selector: div.listitemnews
title:
- selector: "a"
+ selector: h4
url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
+ selector: a[href*="/press-releases/details/"]
extractor: href
+ description:
+ selector: .listitemnews__text
published_at:
- selector: "time"
+ selector: .listitemnews__date
diff --git a/lib/html2rss/configs/ucl.ac.uk/news.yml b/lib/html2rss/configs/ucl.ac.uk/news.yml
index b97c4049..ee47a42c 100644
--- a/lib/html2rss/configs/ucl.ac.uk/news.yml
+++ b/lib/html2rss/configs/ucl.ac.uk/news.yml
@@ -3,23 +3,17 @@ directory:
topics:
- research
- science
- title: "UCL — News Highlights"
- summary: "Interdisciplinary science, public health findings, and academic publications from University College London."
-
+ title: UCL — News Highlights
+ summary:
+ Interdisciplinary science, public health findings, and academic publications
+ from University College London.
channel:
- title: "UCL — News Highlights"
+ title: UCL — News Highlights
url: https://www.ucl.ac.uk/news
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: main article a
+ enhance: true
diff --git a/lib/html2rss/configs/unep-wcmc.org/en_news.yml b/lib/html2rss/configs/unep-wcmc.org/en_news.yml
index ba774186..7f54275f 100644
--- a/lib/html2rss/configs/unep-wcmc.org/en_news.yml
+++ b/lib/html2rss/configs/unep-wcmc.org/en_news.yml
@@ -3,23 +3,18 @@ directory:
topics:
- environment
- science
- title: "UNEP-WCMC — Biodiversity Conservation News"
- summary: "World Conservation Monitoring Centre biodiversity data, species assessments, and marine protected areas."
-
+ title: UNEP-WCMC — Biodiversity Conservation News
+ summary:
+ World Conservation Monitoring Centre biodiversity data, species assessments,
+ and marine protected areas.
channel:
- title: "UNEP-WCMC — Biodiversity Conservation News"
+ title: UNEP-WCMC — Biodiversity Conservation News
url: https://www.unep-wcmc.org/en/news
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: li.listing-grid__item
+ enhance: true
+strategy: faraday
diff --git a/lib/html2rss/configs/uni-heidelberg.de/en_newsroom.yml b/lib/html2rss/configs/uni-heidelberg.de/en_newsroom.yml
index 82ec3e8b..1c8d5b2c 100644
--- a/lib/html2rss/configs/uni-heidelberg.de/en_newsroom.yml
+++ b/lib/html2rss/configs/uni-heidelberg.de/en_newsroom.yml
@@ -3,23 +3,17 @@ directory:
topics:
- research
- science
- title: "Heidelberg University — Newsroom"
- summary: "Fundamental physics, medical discoveries, and humanities research from Heidelberg Germany."
-
+ title: Heidelberg University — Research News
+ summary:
+ Fundamental physics, medical discoveries, and humanities research from
+ Heidelberg Germany.
channel:
- title: "Heidelberg University — Newsroom"
- url: https://www.uni-heidelberg.de/en/newsroom
+ title: Heidelberg University — Research News
+ url: https://www.uni-heidelberg.de/en/newsroom/news-research
language: en
- time_zone: UTC
+ time_zone: Europe/Berlin
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: a.link-local[href^="/en/newsroom/"][href*="-"]
+ enhance: true
diff --git a/lib/html2rss/configs/unsw.edu.au/newsroom_news.yml b/lib/html2rss/configs/unsw.edu.au/newsroom_news.yml
deleted file mode 100644
index 6c88fdbb..00000000
--- a/lib/html2rss/configs/unsw.edu.au/newsroom_news.yml
+++ /dev/null
@@ -1,25 +0,0 @@
-# yaml-language-server: $schema=https://raw.githubusercontent.com/html2rss/html2rss/refs/heads/master/schema/html2rss-config.schema.json
-directory:
- topics:
- - research
- - science
- title: "UNSW Sydney — Newsroom"
- summary: "Photovoltaic engineering, clean energy, and medical research from UNSW Sydney."
-
-channel:
- title: "UNSW Sydney — Newsroom"
- url: https://www.unsw.edu.au/newsroom/news
- language: en
- time_zone: UTC
- ttl: 360
-selectors:
- items:
- selector: 'ul[class*="news"] > li'
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
diff --git a/lib/html2rss/configs/uoc.edu/en_news.yml b/lib/html2rss/configs/uoc.edu/en_news.yml
index 9a6b7228..ead8d81c 100644
--- a/lib/html2rss/configs/uoc.edu/en_news.yml
+++ b/lib/html2rss/configs/uoc.edu/en_news.yml
@@ -3,23 +3,15 @@ directory:
topics:
- research
- tech
- title: "Universitat Oberta de Catalunya — Newsroom"
- summary: "Digital education, e-learning research, and knowledge transfer from UOC."
-
+ title: Universitat Oberta de Catalunya — Newsroom
+ summary: Digital education, e-learning research, and knowledge transfer from UOC.
channel:
- title: "Universitat Oberta de Catalunya — Newsroom"
+ title: Universitat Oberta de Catalunya — Newsroom
url: https://www.uoc.edu/en/news
language: en
- time_zone: UTC
+ time_zone: Europe/Madrid
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "h1, h2, h3, h4, h5, h6"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: article.uoc-bcard
+ enhance: true
diff --git a/lib/html2rss/configs/uu.se/en_news.yml b/lib/html2rss/configs/uu.se/en_news.yml
index 9fa5c123..4658770e 100644
--- a/lib/html2rss/configs/uu.se/en_news.yml
+++ b/lib/html2rss/configs/uu.se/en_news.yml
@@ -3,23 +3,17 @@ directory:
topics:
- research
- science
- title: "Uppsala University — News Highlights"
- summary: "Pharmaceutical science, energy storage, and historical research from Uppsala Sweden."
-
+ title: Uppsala University — News Highlights
+ summary:
+ Pharmaceutical science, energy storage, and historical research from Uppsala
+ Sweden.
channel:
- title: "Uppsala University — News Highlights"
+ title: Uppsala University — News Highlights
url: https://www.uu.se/en/news
language: en
time_zone: UTC
ttl: 360
selectors:
items:
- selector: "article"
- enhance: false
- title:
- selector: "a"
- url:
- selector: "h1 a, h2 a, h3 a, h4 a, h5 a, h6 a"
- extractor: href
- published_at:
- selector: "time"
+ selector: div.card
+ enhance: true