# Author: zinnober
# Should work with "normal" articles as well as with image galleries

prune: no

# Title
title: //h1/span[@class='hcf-headline']

# Set author
author: //a[@rel='author']

# Set date
date: //span[@class='date hcf-atlas']

# Fetch full multipage articles
next_page_link: //a[contains(@class, 'hcf-forward')]

# Content is here
body: //article
body: //div[contains(@class, 'hcf-screen')]

# Remove tracking and ads
strip_id_or_class: hcf-ad
strip_id_or_class: hcf-autoload-ad
strip_id_or_class: hcf-content-ad

# Tidy up before article
strip: //article/h1
strip_id_or_class: hcf-atlas
strip_id_or_class: hcf-author
strip_id_or_class: date hcf-atlas
strip_id_or_class: date hcf-atlas

# General cleanup
strip: //div[contains(@class, 'hcf-screen')]//h1
strip: //div[@class='hcf-subpage-titles']//ul
strip_id_or_class: hcf-doctype-media
strip_id_or_class: hcf-inline-gallery
strip_id_or_class: hcf-doctype-video
strip_id_or_class: hcf-links
strip_id_or_class: hcf-mini-navi
strip_id_or_class: hcf-media-control
strip_id_or_class: hcf-hidden
replace_string(<span class="hcf-update">Update</span>): <strong>Update: </strong>

# Fix pictures and captions
replace_string(<a class="hcf-doctype-gallery): <p class="hcf-doctype-gallery
replace_string(<a class="hcf-doctype-enlarge): <p class="hcf-doctype-enlarge
replace_string(<figcaption class="hcf-caption">): <br><small><em>
replace_string(</figcaption>): </em></small>

# Fix image galleries
replace_string(<a class=" ajaxify): <p class="ajaxify
replace_string(<div class="hcf-caption"><div><p>): <small><em>

# Try it yourself
test_url: http://www.tagesspiegel.de/berlin/bezirke/wedding/wedding-jetzt/auf-der-suche-nach-einem-stadtteil-wilder-weiter-wedding/8757156.html
test_url: http://www.tagesspiegel.de/berlin/olympia-in-berlin-der-flughafen-tegel-soll-das-olympische-dorf-werden/10645036.html
test_url: http://www.tagesspiegel.de/mediacenter/fotostrecken/berlin/bildergalerie-kreuzberger-der-woche/9305534.html