calibre/recipes/hna.recipe
2014-03-05 22:15:05 +05:30

88 lines
3.8 KiB
Plaintext

__license__ = 'GPL v3'
__copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
'''
Fetch Hessisch Niedersachsische Allgemeine.
'''
from calibre.web.feeds.news import BasicNewsRecipe
class hnaDe(BasicNewsRecipe):
title = 'HNA'
description = 'local news from Hessen/Germany'
__author__ = 'Oliver Niesner'
use_embedded_content = False
language = 'de'
use_embedded_content = False
timefmt = ' [%d %b %Y]'
max_articles_per_feed = 40
no_stylesheets = True
remove_javascript = True
auto_cleanup = True
encoding = 'utf-8'
remove_tags = [dict(id='topnav'),
dict(id='nav_main'),
dict(id='teaser'),
dict(id='suchen'),
dict(id='superbanner'),
dict(id='navigation'),
dict(id='skyscraper'),
dict(id='idHeaderSearchForm'),
dict(id='idHeaderSearchBar'),
dict(id='idLoginBarWrap'),
dict(id='idAccountButtons'),
dict(id='idHeadButtons'),
dict(id='idBoxesWrap'),
dict(id='idJSMainNavigation'),
dict(id=''),
dict(name='span'),
dict(name='ul', attrs={'class':'linklist'}),
dict(name='ul', attrs={'class':'idMainNavi idJSActive idHeadHomeBtn'}),
dict(name='ul', attrs={'class':'idHiddenNavi idNaviSubcategories'}),
dict(name='a', attrs={'href':'#'}),
dict(name='a', attrs={'class':'idImgLink'}),
dict(name='a', attrs={'class':'idListLink'}),
dict(name='div', attrs={'class':'hlist'}),
dict(name='div', attrs={'class':'idTabWrap'}),
dict(name='li', attrs={'class':'idButton idIsLoginGroup idHeaderRegister '}),
dict(name='li', attrs={'class':'idVideoBar idFirst'}),
dict(name='li', attrs={'class':'idSetStartPageLink idLast'}),
dict(name='li', attrs={'class':'idKinderNetzBar idLast'}),
dict(name='li', attrs={'class':'idFotoBar '}),
dict(name='div', attrs={'class':'subc noprint'}),
dict(name='div', attrs={'class':'idTxtLay'}),
dict(name='div', attrs={'class':'idLay idClStandard idStaticHtml'}),
dict(name='div', attrs={'class':'idHeaderWrap'}),
dict(name='div', attrs={'class':'idLay idRss idClStandard'}),
#dict(name='div', attrs={'class':' idHide idChannelChooser idJSComponent idParam-component-IDChannelChooser'}),
dict(name='div', attrs={'class':'idLay idClStandard idLeadStoriesFocus idLeadStoriesFocusOverlay '}),
dict(name='div', attrs={'class':'idTeaserLay idTeaserFloat idMediaLeft idLast'}),
dict(name='div', attrs={'class':'idHeaderButtons idAccountButtons'}),
dict(name='div', attrs={'class':'idTeaserLay idTeaserWithImg idSize4 idMediaLeft'}),
dict(name='div', attrs={'class':'idHeaderButtons idHeadButtons'}),
dict(name='div', attrs={'class':'idHeaderButtons idSetStartPage'}),
dict(name='div', attrs={'class':'idLay idClHl idTeaserList '}),
dict(name='div', attrs={'class':'idNavigationWrap'}),
dict(name='div', attrs={'class':'idBreadcrumbWrap'}),
dict(name='div', attrs={'class':'idBoxesWrap'}),
dict(name='div', attrs={'class':'idBreadcrumb'}),
dict(name='div', attrs={'class':'idLay idAdvertising idClStandard '}),
dict(name='span', attrs={'class':'idHeadLineIntro'}),
dict(name='p', attrs={'class':'breadcrumb'}),
dict(name='a', attrs={'style':'cursor:hand'}),
dict(name='p', attrs={'class':'h5'}),
dict(name='p', attrs={'class':'idMoreEnd'})]
#remove_tags_after = [dict(name='div', attrs={'class':'rahmenbreaking'})]
#remove_tags_after = [dict(name='p', attrs={'class':'idMoreEnd'})]
remove_tags_after = [dict(name='div', attrs={'class':'idTxtLay idStaticHtmlIEHelper'})]
feeds = [ ('hna_soehre', 'http://feeds2.feedburner.com/hna/soehre'),
('hna_kassel', 'http://feeds2.feedburner.com/hna/kassel'),
('hna_KSV', 'http://feeds2.feedburner.com/hna/ksv'),
('hna_kultur', 'http://feeds2.feedburner.com/hna/kultur') ]