mirror of
https://gitlab.com/news-flash/article_scraper.git
synced 2025-07-13 10:55:31 +02:00
42 lines
1,022 B
Text
42 lines
1,022 B
Text
# Author: zinnober
|
|
# Rewrite of original template which fetched the printer-version without pictures
|
|
|
|
tidy: no
|
|
prune: no
|
|
|
|
# Set full title
|
|
title: //h1/span
|
|
|
|
date: //time
|
|
author: //a[@rel='author']
|
|
|
|
# Content is here
|
|
body: //article
|
|
|
|
# Fetch full multipage articles
|
|
next_page_link: //a[@id='atoc_next']
|
|
|
|
# Remove tracking and ads
|
|
strip_id_or_class: iqadtile4
|
|
|
|
# General Cleanup
|
|
strip_id_or_class: list-jtoc
|
|
strip_id_or_class: table-jtoc
|
|
strip_id_or_class: implied
|
|
strip_id_or_class: social-
|
|
strip_id_or_class: comments
|
|
strip_id_or_class: footer
|
|
strip_id_or_class: job-market
|
|
strip_id_or_class: tags
|
|
|
|
# Tidy up galleries (could still be improved, though)
|
|
strip: //img[@src='']
|
|
strip: //li[not(*)]
|
|
strip: //div[contains(@style,'margin')]
|
|
strip: //figure[contains(@id,'gvideo')]
|
|
|
|
|
|
# Try yourself
|
|
test_url: http://www.golem.de/news/intel-core-i7-5960x-im-test-die-pc-revolution-beginnt-mit-octacore-und-ddr4-1408-108893.html
|
|
test_url: http://www.golem.de/news/test-infamous-first-light-neonbunter-actionspass-1408-108914.html
|
|
|