# Emphasize lead paragraph (replace needs to be done before parsing body)
find_string: <header
replace_string: <em

body: //em[@data-testid='SplitScreenContentHeaderWrapper'] | //div[@class='body__inner-container']

author: //meta[@property="article:author"]/@content

date: //meta[@property="article:published_time"]/@content
date: //article//time[contains(@class, 'content-header__publish-date')]

strip_id_or_class: full-bleed-ad

strip: //button[@id="bookmark"]

# Clean up header
strip: //div[@data-testid="ResponsiveClipVideoContainer"]
strip: //div[@data-testid="ContentHeaderRubric"]
strip: //h1
strip: //div[@data-testid="BylinesWrapper"]
strip: //time
strip: //ul[@data-testid="socialIconslist"]
strip: //div[contains(concat(' ', normalize-space(@class), ' '), ' caption ')]

# Clean up body
strip: //div[@data-testid="GenericCallout"]
strip: //figure[@data-testid="IframeEmbed"]

# Strip unrelated cartoons
strip: //div[contains(concat(' ', normalize-space(@class), ' '), ' responsive-cartoon ')]

prune: no
tidy: no

# Bypass "1 free article" paywall with random cookie value
http_header(Cookie): pay_ent_msmp=1

test_url: https://www.newyorker.com/magazine/2016/02/08/mothers-day-fiction-george-saunders
test_url: https://www.newyorker.com/online/blogs/culture/2012/06/mug-shot-web-sites.html
test_url: https://www.newyorker.com/reporting/2013/04/22/130422fa_fact_bilger?currentPage=all&mobify=0
test_url: https://www.newyorker.com/magazine/2023/11/20/a-coder-considers-the-waning-days-of-the-craft
# Cartoons
test_url: https://www.newyorker.com/cartoons/blitts-kvetchbook/a-man-of-substance
