# CBS News siteconfig
# Handles both regular articles and live-update stories

title: //h1[@class="content__title"]

# Regular articles
body: //section[@class="content__body"]

# Live-update stories
body: //section[contains(@class,"content-updating-story__content-wrapper")]

date: //p[contains(@class,"content__meta--timestamp")]/time/@datetime

author: //div[contains(@class,"byline__author__link-wrapper")]/a
author: //a[contains(@class,"byline__author__link")]

# Remove global navigation and header
strip: //header[contains(@class,"site-header__container")]

# Remove ads
strip: //div[contains(@class,"top-ad-container")]
strip: //div[contains(@class,"ad-intromercial")]

# Remove sidebar widget area
strip: //section[contains(@class,"content__widget-wrapper")]

# Remove Google Follow button
strip: //div[@class="content__google"]

# Remove newsletter promo blocks
strip: //section[contains(@class,"post-newsletter-promo")]

# Remove per-post share bars
strip: //div[contains(@class,"post-update__share")]
strip: //ul[contains(@class,"sharebar")]

# Remove footer elements
strip: //footer[contains(@class,"content__footer")]
strip: //aside[contains(@class,"recirculation")]

# more clean-up
strip: //button
strip_id_or_class: top-ad-container
strip_id_or_class: post-update__share
strip_id_or_class: sharebar
strip_id_or_class: content__google
strip_id_or_class: content__widget-wrapper
strip_id_or_class: post-newsletter-promo

prune: no

test_url: https://www.cbsnews.com/news/drunken-raccoon-virginia-liquor-store-bathroom-floor/
test_contains: He fell through one of the ceiling tiles and went on a full-blown rampage, drinking everything.

test_url: https://www.cbsnews.com/live-updates/iran-us-war-day-3-american-deaths-israel-gulf-allies-hit-missile-strikes/
test_contains: has been dragged into the Iran war with and attacks on its territory and missile interceptions
