# NOTE:  If testing this configuration yields bad results, including junk text like "Try BostonGlobe.com today" and "THIS STORY APPEARED IN", please replace the Test URL with a current-day headline link from bostonglobe.com.

title: //article//h1[contains(@class, 'title')]
author: //div[contains(@class, "byline")]//span[contains(@class, 'author')]
date: //meta[@name="datePublished"]/@content
#date: //div[@class="byline"]/p[last()]
body: //article//h2[contains(@class, 'subheader')] | (//article//div[contains(@class, 'image')])[1] | //article[@id="article-body"]

strip_id_or_class: bgmp-comments
strip_id_or_class: aside
strip_id_or_class: promo
strip_id_or_class: skip-nav
strip_id_or_class: article-more
strip_id_or_class: article-bar
strip_id_or_class: post-comment
strip_id_or_class: link-related_links_link
strip_id_or_class: newsletter
strip_id_or_class: arc_ad
strip_id_or_class: teads-inread
strip_id_or_class: continue_button

# remove tiny image (full size not present in element)
strip: //img[contains(@src, '/20x0/')]

# This removes image captions.
strip_id_or_class: figure

test_url: https://www.bostonglobe.com/2020/04/07/metro/coronavirus-cases-surge-citys-safety-net-hospital-four-out-10-cases-are-now-related-pandemic/
test_url: https://www.bostonglobe.com/2021/03/01/metro/charlie-youre-making-big-mistake-experts-criticize-states-monday-reopening/