tidy: no
prune: no

# Title
title: //p[@class='Content HeadlineShort']

# Set author
author: substring-after(//span[@class='Autor'], 'von ')
author: //span[@class='caps last']/span[@class='caps last']
author: //a[@rel='author']

# Set date
date: //span[@class='Datum']
date: //span[@class='Datum'],/span

# Fetch full multipage articles
#single_page_link: //a[contains(@href, 'printPagedArticle')]

# Content is here

body: //article[@class='storytelling']
body: //article[contains(@class='article')]//div[contains(@class,'body-elements')] | (//div[contains(@class,'header-teaser__image')])[1] | (//div[contains(concat(' ',normalize-space(@class),' '),' header-teaser ')])[last()]
body: //article[1]
body: //div[@class='Artikel']


# Tidy up before article
strip: //div[@id='FAZHeaderNeu']
strip: //h2[@itemprop='headline']
strip: //span[@class='Datum']
strip: //span[@class='Autor']
strip_id_or_class: ArticlePagerTop
strip_id_or_class: header-detail
strip_id_or_class: intro-text
strip: //button[contains(@class,'image-toggle')]

# General cleanup
strip: //div[@class='clear']
strip: //a[@title='Zur Homepage FAZ.NET']
#strip: //iframe
replace_string( ·  ):
strip_id_or_class: TeaserMore
strip_id_or_class: plista_alternativ
strip_id_or_class: paywall
strip: //button
strip_id_or_class: header-teaser__image-details
strip_id_or_class: tik4-sharing
strip: //aside[contains(concat(' ',normalize-space(@class),' '),' btn ')]
strip: //div[contains(concat(' ',normalize-space(@class),' '),' meta2 ')]
strip: //footer

# Remove tracking and ads
strip_image_src: /l.gif?
strip: //div[contains(@style, 'background-image')]
strip: //img[@width='1']
strip_id_or_class: invisible
strip_id_or_class: Anzeige
strip_id_or_class: billboard

# Remove various text boxes and social media foo
strip_id_or_class: WeitereBeitraege
strip_id_or_class: WBListe
strip_id_or_class: AutorenModul
strip_id_or_class: Community
strip_id_or_class: SocialMediaStatus
strip_id_or_class: RelatedLinkBox
strip_id_or_class: MultimediaNavigation
strip_id_or_class: IndexTitel
strip_id_or_class: cbx-Author-is-in-article-container-info
strip_id_or_class: BigBox
strip_id_or_class: upper-toolbar
strip_id_or_class: print:hidden

# Fix picture caps and pictures (use better resolution and remove clutter)
strip_id_or_class: LightBoxOverlay
strip_id_or_class: exitLarge
strip_id_or_class: PagerBox
strip_id_or_class: Bildnachweis
strip_id_or_class: Bildueberschrift
strip_id_or_class: Bildbeschreibung
strip_id_or_class: ArtikelBild610
strip_id_or_class: MediaLink
strip_id_or_class: FotoBoxInnerLeft
strip_id_or_class: BilderRelatedLinks
# Handle <img src='https://www.faz.net/img/leer.gif' width='62' data-retina-src='[real img URL]'
strip_attr: //img/@width
strip_attr: //img/@height
find_string: src='https://www.faz.net/img/leer.gif'
replace_string: removed-src=''
replace_string(data-retina-src=): src= 

# Remove clutter after article
strip_id_or_class: article-footer
strip_id_or_class: ArticlePagerBottom
strip_id_or_class: backToHome
strip_id_or_class: ArtikelAbbinder
strip_id_or_class: lesermeinungscontainer
strip_id_or_class: ThemenLinks
strip_id_or_class: rechtehinweis
strip_id_or_class: FAZArtikelMap
strip_id_or_class: FAZArtikelKommentare
strip_id_or_class: ArtikelKommentieren
strip_id_or_class: FAZArtikelFunktionen
strip_id_or_class: mailLB
strip_id_or_class: FAZContentRight
strip_id_or_class: stageModule
strip_id_or_class: ContentFooter
strip_id_or_class: ServicesFooter
strip_id_or_class: FAZFooter

# Clean up stuff present just in some articles
strip_id_or_class: Teaser620
strip_id_or_class: TeaserMultimedia
strip_id_or_class: VideoBox

# Remove as soon as wallabag might be able to embed flash video
strip_id_or_class: mmoObjectAsTeaserInArticle
strip_id_or_class: additionalStylesAudioVideo
strip_id_or_class: hideMMElements

# Remove F+ paywall adverts and more tidy-up ### by HolgerAusB
strip_id_or_class: js-aco-ContentPayment_Content
strip_id_or_class: js-aco-ContentPayment_Panel
strip_id_or_class: aco-ContentPayment_Link
strip_id_or_class: mm-adbox
strip: //footer[@class='tsr-Base_ContentMeta']
strip_id_or_class: aut-Teaser
strip_id_or_class: related-articles

# Try it yourself
test_url: https://www.faz.net/aktuell/feuilleton/bilder-und-zeiten/lord-byron-in-venedig-als-pilger-popstar-und-poet-19648440.html
test_url: https://www.faz.net/aktuell/politik/europawahl/europawahl-2024-wer-in-die-eu-einzahlt-und-wer-mittel-erhaelt-19707069.html
test_url: http://www.faz.net/aktuell/feuilleton/zum-tod-von-margaret-thatcher-die-reizfigur-12141919.html#Drucken
test_url: http://www.faz.net/aktuell/politik/inland/allensbach-analyse-im-namen-des-volkes-13106492.html
test_url: http://www.faz.net/aktuell/feuilleton/kino/video-filmkritiken/video-filmkritik-when-animals-dream-zerrissene-jugend-13105772.html
test_url: https://www.faz.net/aktuell/feuilleton/debatten/keine-smart-city-in-toronto-google-stadt-ist-abgesagt-16763217.html?GEPC=s5
test_url: https://www.faz.net/aktuell/wirtschaft/wohnen/christian-voelkers-der-immobilienmakler-der-superreichen-17778869.html
