mirror of
https://github.com/kovidgoyal/calibre.git
synced 2026-01-08 21:20:21 -05:00
44 lines
1.9 KiB
Plaintext
44 lines
1.9 KiB
Plaintext
from calibre.web.feeds.news import BasicNewsRecipe
|
|
import re
|
|
# History:
|
|
# 1: Base Version
|
|
# 2: Added rules for wdr.de, ndr.de, br-online.de
|
|
# 3: Added rules for rbb-online.de, boerse.ard.de, sportschau.de
|
|
# 4: New design of tagesschau.de implemented. Simplyfied.
|
|
# 5: Taken out the pictures.
|
|
|
|
class Tagesschau(BasicNewsRecipe):
|
|
title = 'Tagesschau'
|
|
description = 'Nachrichten der ARD'
|
|
publisher = 'ARD'
|
|
language = 'de'
|
|
version = 5
|
|
|
|
__author__ = 'Florian Andreas Pfaff, a.peter'
|
|
oldest_article = 7
|
|
max_articles_per_feed = 100
|
|
no_stylesheets = True
|
|
remove_javascript = True
|
|
|
|
feeds = [('Tagesschau', 'http://www.tagesschau.de/xml/rss2')]
|
|
|
|
remove_tags = [
|
|
dict(name='div', attrs={'class':['inline']}),
|
|
dict(name='div', attrs={'class':re.compile(r'.*linklist.*')}),
|
|
dict(name='div', attrs={'class':re.compile(r'.*media(Left|Right|Top|Bottom|Info).*')}),
|
|
dict(name='div', attrs={'class':re.compile(r'.*mod(Socialbar|ConComments).*')}),
|
|
dict(name='div', attrs={'class':re.compile(r'.*infokasten.*')}),
|
|
dict(name='div', attrs={'class':re.compile(r'.*articlePictureA.*')}),
|
|
dict(name='div', attrs={'id': re.compile(r'.*zoomTeaser.*')}),
|
|
dict(name='h2', attrs={'class':re.compile(r'.*conHeadline.*')}),
|
|
dict(name='ul', attrs={'class':['iconList']})
|
|
]
|
|
keep_only_tags = [
|
|
dict(name='div', attrs={'class':['section sectionZ sectionArticle']}),
|
|
dict(name='div', attrs={'class':re.compile(r'.*containerArticle.*')})
|
|
]
|
|
|
|
def get_masthead_url(self):
|
|
return 'http://intern.tagesschau.de/html/img/image.jpg'
|
|
|