home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / blic.recipe < prev    next >
Encoding:
Text File  |  2011-09-09  |  4.4 KB  |  80 lines

  1.  
  2. __license__   = 'GPL v3'
  3. __copyright__ = '2008-2011, Darko Miletic <darko.miletic at gmail.com>'
  4. '''
  5. blic.rs
  6. '''
  7.  
  8. import re
  9. from calibre.web.feeds.news import BasicNewsRecipe
  10.  
  11. class Blic(BasicNewsRecipe):
  12.     title                 = 'Blic'
  13.     __author__            = 'Darko Miletic'
  14.     description           = 'Blic.rs online verzija najtiraznije novine u Srbiji donosi najnovije vesti iz Srbije i sveta, komentare, politicke analize, poslovne i ekonomske vesti, vesti iz regiona, intervjue, informacije iz kulture, reportaze, pokriva sve sportske dogadjaje, detaljan tv program, nagradne igre, zabavu, fenomenalni Blic strip, dnevni horoskop, arhivu svih dogadjaja'
  15.     publisher             = 'RINGIER d.o.o.'
  16.     category              = 'news, politics, Serbia'
  17.     oldest_article        = 2
  18.     max_articles_per_feed = 100
  19.     no_stylesheets        = True
  20.     use_embedded_content  = False
  21.     masthead_url          = 'http://www.blic.rs/resources/images/header/header_back.png'
  22.     language              = 'sr'
  23.     publication_type      = 'newspaper'
  24.     extra_css             = """ 
  25.                                @font-face {font-family: "serif1";src:url(res:///opt/sony/ebook/FONT/tt0011m_.ttf)} 
  26.                                @font-face {font-family: "sans1";src:url(res:///opt/sony/ebook/FONT/tt0003m_.ttf)} 
  27.                                body{font-family: Georgia, serif1, serif} 
  28.                                .articledescription,#nadnaslov,.article_info{font-family: Arial, sans1, sans-serif} 
  29.                                .img_full{float: none}
  30.                                #nadnaslov{font-size: small}
  31.                                #article_lead{font-size: 1.5em}
  32.                                h1{color: red}
  33.                                .potpis{font-size: x-small; color: gray}
  34.                                .article_info{font-size: small}
  35.                                img{margin-bottom: 0.8em; margin-top: 0.8em; display: block} 
  36.                             """
  37.  
  38.     conversion_options = {
  39.                           'comment'  : description
  40.                         , 'tags'     : category
  41.                         , 'publisher': publisher
  42.                         , 'language' : language
  43.                         , 'linearize_tables' : True
  44.                         }
  45.  
  46.     preprocess_regexps = [(re.compile(u'\u0110'), lambda match: u'\u00D0')]
  47.     remove_tags_before = dict(name='div', attrs={'id':'article_info'})
  48.     remove_tags        = [dict(name=['object','link','meta','base','object','embed'])]
  49.     remove_attributes  = ['width','height','m_id','m_ext','mlg_id','poll_id','v_id']
  50.  
  51.     feeds              = [
  52.                             (u'Politika'         , u'http://www.blic.rs/rss/Vesti/Politika')
  53.                            ,(u'Tema Dana'        , u'http://www.blic.rs/rss/Vesti/Tema-Dana')
  54.                            ,(u'Svet'             , u'http://www.blic.rs/rss/Vesti/Svet')
  55.                            ,(u'Drustvo'          , u'http://www.blic.rs/rss/Vesti/Drustvo')
  56.                            ,(u'Ekonomija'        , u'http://www.blic.rs/rss/Vesti/Ekonomija')
  57.                            ,(u'Hronika'          , u'http://www.blic.rs/rss/Vesti/Hronika')
  58.                            ,(u'Beograd'          , u'http://www.blic.rs/rss/Vesti/Beograd')
  59.                            ,(u'Srbija'           , u'http://www.blic.rs/rss/Vesti/Srbija')
  60.                            ,(u'Vojvodina'        , u'http://www.blic.rs/rss/Vesti/Vojvodina')
  61.                            ,(u'Republika Srpska' , u'http://www.blic.rs/rss/Vesti/Republika-Srpska')
  62.                            ,(u'Reportaza'        , u'http://www.blic.rs/rss/Vesti/Reportaza')
  63.                            ,(u'Dodatak'          , u'http://www.blic.rs/rss/Vesti/Dodatak')
  64.                            ,(u'Zabava'           , u'http://www.blic.rs/rss/Zabava')
  65.                            ,(u'Kultura'          , u'http://www.blic.rs/rss/Kultura')
  66.                            ,(u'Slobodno Vreme'   , u'http://www.blic.rs/rss/Slobodno-vreme')
  67.                            ,(u'IT'               , u'http://www.blic.rs/rss/IT')
  68.                            ,(u'Komentar'         , u'http://www.blic.rs/rss/Komentar')
  69.                            ,(u'Intervju'         , u'http://www.blic.rs/rss/Intervju')
  70.                          ]
  71.  
  72.  
  73.     def print_version(self, url):
  74.         return url + '/print'
  75.  
  76.     def preprocess_html(self, soup):
  77.         for item in soup.findAll(style=True):
  78.             del item['style']    
  79.         return soup
  80.