# complete feed:
# https://www.diepresse.com/rss

# or add a category, e.g.:
# https://www.diepresse.com/rss/kultur/kunst


title: //div[@class='article']/h1
date: substring-before(//p[@class='articletime'],'|')
author: //div[@class='author']/descendant::a
body: //article | //div[@id='articletext'] | //div[@id='article-body']

strip: //div[@class='meta__date']
strip: //div[@class='meta__authors']
strip: //h1[@class='article__title']
strip: //div[contains(@class, 'fm-relation')]

strip_id_or_class: skyline
strip_id_or_class: article__actions
strip_id_or_class: article__byline


# strip gerneric text images when there is no article photo/image

find_string: https://www.diepresse.com/assets_v2/images/
replace_string:

# 'data-src' sometimes prevents images to load, replaced with 'src'

find_string: class="figure__image lazyload" data-src=
replace_string: class="figure__image lazyload" src=

find_string: <gallery-inline
replace_string: <div

#################
# change path for images to get rid of the text overlays

find_string: /alternates/DPR_SOCIAL_NACHRICHTEN_PAID/
replace_string: /alternates/ORIGINAL/

find_string: /alternates/DPR_SOCIAL_NACHRICHTEN/
replace_string: /alternates/ORIGINAL/

find_string: /alternates/DPR_SOCIAL_MAGAZIN_PAID/
replace_string: /alternates/ORIGINAL/

find_string: /alternates/DPR_SOCIAL_MAGAZIN/
replace_string: /alternates/ORIGINAL/

find_string: /alternates/DPR_SOCIAL_MEINUNG_PAID/
replace_string: /alternates/ORIGINAL/

find_string: /alternates/DPR_SOCIAL_MEINUNG/
replace_string: /alternates/ORIGINAL/

#################

prune: no

test_url: https://www.diepresse.com/701905/tibeter-proteste-nonne-verbrennt-sich-selbst
test_url: https://www.diepresse.com/6115784/omikron-ba2-wann-infizierte-hoch-ansteckend-sind
test_url: https://www.diepresse.com/rss/panorama/wien
