# website-engine: medium.com
http_header(user-agent): Mozilla/5.0 (compatible; Yahoo! Slurp; http://help.yahoo.com/help/us/ysearch/slurp)

title: //meta[@property="og:title"]/@content
title: //h1
body: //article
author: //meta[@name="author"]/@content
date: //meta[@property="article:published_time"]/@content

strip: //button
strip: //svg

# Remove low quality images
strip: //article//img[contains(@src, '?q=20') or contains(@src, 'max/34/')]
# Remove empty images (width attribute but no src attribute)
strip: //article//img[not(@src)]
# Use the high-quality copies in <noscript> elements
replace_string(<noscript>): <div class="ftr-noscript">
replace_string(</noscript>): </div>

# Remove parts of header
strip: //header[contains(@class, 'pw-post-byline-header')]
strip: //h1[contains(@class, 'pw-post-title')]
strip_id_or_class: speechify-ignore
strip: //a[@rel='noopener  ugc nofollow']/parent::div

# Remove footer and everything below
strip: //footer/self::footer | //footer/following-sibling::*

# remove duplicate images in this form... <img><noscript><img></noscript>
# handled above, but if browser HTML is submitted, the above rules won't apply
strip: //article//img[@src]/following-sibling::*[1][self::div and @class='ftr-noscript']
# less accurate...
# strip: //article//img[@src]/following-sibling::div[@class="ftr-noscript"]

tidy: no
prune: no

###  selfhosters may uncomment this to have images in their articles.
###  rewrites image-links to plain html
find_string: source srcSet="http
replace_string: img src="http
find_string: source srcset="http
replace_string: img src="http
find_string: .png 640w
replace_string: .png"><foo bar="
find_string: .jpg 1400w
replace_string: .jpg"><foo bar="
find_string: .jpg 640w
replace_string: .jpg"><foo bar="
find_string: .jpeg 1400w
replace_string: .jpeg"><foo bar="
find_string: .jpeg 640w
replace_string: .jpeg"><foo bar="
find_string: .gif 640w
replace_string: .gif"><foo bar="

### VERY BRAVE self-hosters may try to acticate this block INSTEAD
### helps to show also images with no filename extension.
#find_string: source srcSet="http
#replace_string: img src="http
#find_string: source srcset="http
#replace_string: img src="http
#replace_string( 1400w,): "><foo bar="
#replace_string( 720w,): "><foo bar="
#replace_string( 640w,): "><foo bar="

test_url: https://entrepreneurshandbook.co/the-bang-with-friends-story-under-the-covers-77f757bdb822
