home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / hna.recipe < prev    next >
Encoding:
Text File  |  2011-09-09  |  2.2 KB  |  64 lines
  1. __license__   = 'GPL v3'
  2. __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
  3.  
  4. '''
  5. Fetch Hessisch Niedersachsische Allgemeine.
  6. '''
  7.  
  8. from calibre.web.feeds.news import BasicNewsRecipe
  9.  
  10.  
  11. class hnaDe(BasicNewsRecipe):
  12.  
  13.     title = 'HNA'
  14.     description = 'local news from Hessen/Germany'
  15.     __author__ = 'Oliver Niesner'
  16.     use_embedded_content   = False
  17.     language = 'de'
  18.  
  19.     use_embedded_content   = False
  20.     timefmt = ' [%d %b %Y]'
  21.     max_articles_per_feed = 40
  22.     no_stylesheets = True
  23.     remove_javascript = True
  24.     encoding = 'utf-8'
  25.  
  26.     remove_tags = [dict(id='topnav'),
  27.            dict(id='nav_main'),
  28.            dict(id='teaser'),
  29.            dict(id='suchen'),
  30.            dict(id='superbanner'),
  31.            dict(id='navigation'),
  32.            dict(id='skyscraper'),
  33.            dict(id='idNavigationWrap'),
  34.            dict(id='idHeaderSearchForm'),
  35.            dict(id='idLoginBarWrap'),
  36.            dict(id='idAccountButtons'),
  37.            dict(id='idHeadButtons'),
  38.            dict(id='idBoxesWrap'),
  39.            dict(id=''),
  40.                    dict(name='span'),
  41.            dict(name='ul', attrs={'class':'linklist'}),
  42.            dict(name='a', attrs={'href':'#'}),
  43.            dict(name='div', attrs={'class':'hlist'}),
  44.            dict(name='li', attrs={'class':'idButton idIsLoginGroup idHeaderRegister '}),
  45.            dict(name='li', attrs={'class':'idVideoBar idFirst'}),
  46.            dict(name='li', attrs={'class':'idSetStartPageLink idLast'}),
  47.            dict(name='li', attrs={'class':'idKinderNetzBar idLast'}),
  48.            dict(name='li', attrs={'class':'idFotoBar '}),
  49.            dict(name='div', attrs={'class':'subc noprint'}),
  50.            dict(name='div', attrs={'class':'idBreadcrumb'}),
  51.            dict(name='div', attrs={'class':'idLay idAdvertising idClStandard '}),
  52.            dict(name='span', attrs={'class':'idHeadLineIntro'}),
  53.            dict(name='p', attrs={'class':'breadcrumb'}),
  54.            dict(name='a', attrs={'style':'cursor:hand'}),
  55.            dict(name='p', attrs={'class':'h5'}),
  56.            dict(name='p', attrs={'class':'idMoreEnd'})]
  57.     #remove_tags_after = [dict(name='div', attrs={'class':'rahmenbreaking'})]
  58.     remove_tags_after = [dict(name='p', attrs={'class':'idMoreEnd'})]
  59.  
  60.     feeds =  [ ('hna_soehre', 'http://feeds2.feedburner.com/hna/soehre'),
  61.            ('hna_kassel', 'http://feeds2.feedburner.com/hna/kassel') ]
  62.  
  63.  
  64.