home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / globes_co_il.recipe < prev    next >
Text File  |  2011-09-09  |  3KB  |  48 lines

  1. from calibre.web.feeds.news import BasicNewsRecipe
  2. from calibre.ebooks.BeautifulSoup import re
  3.  
  4. class AdvancedUserRecipe1283848012(BasicNewsRecipe):
  5.     description   = 'This is Globes.co.il.'
  6.     cover_url      = 'http://www.the7eye.org.il/SiteCollectionImages/BAKTANA/arye_avnery_010709_377.jpg'
  7.     title          = u'Globes'
  8.     language              = 'he'
  9.     __author__ = 'marbs'
  10.     extra_css='img {max-width:100%;} body{direction: rtl;max-width:100%;}title{direction: rtl; } article_description{direction: rtl; }, a.article{direction: rtl;max-width:100%;} calibre_feed_description{direction: rtl; }'
  11.     simultaneous_downloads = 5
  12.     remove_javascript     = True
  13.     timefmt        = '[%a, %d %b, %Y]'
  14.     oldest_article = 1
  15.     max_articles_per_feed = 100
  16.     remove_attributes = ['width','style']
  17.  
  18.  
  19.     feeds          = [(u'╫⌐╫ò╫º ╫ö╫ö╫ò╫ƒ', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=585'),
  20.                            (u'╫á╫ô╫£"╫ƒ', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=607'),
  21.                            (u'╫ò╫ò╫£ ╫í╫ÿ╫¿╫Ö╫ÿ ╫ò╫⌐╫ò╫ò╫º╫Ö ╫ö╫ó╫ò╫£╫¥', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=1225'),
  22.                            (u'╫á╫Ö╫¬╫ò╫ù ╫ÿ╫¢╫á╫Ö', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=1294'),
  23.                            (u'╫ö╫Ö╫Ö ╫ÿ╫º', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=594'),
  24.                            (u'╫á╫¬╫ù ╫⌐╫ò╫º ╫ò╫ª╫¿╫¢╫á╫ò╫¬', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=821'),
  25.                            (u'╫ô╫Ö╫ƒ ╫ò╫ù╫⌐╫æ╫ò╫ƒ', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=829'),
  26.                            (u'╫¿╫¢╫æ', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=3220'),
  27.                            (u'╫ô╫ó╫ò╫¬', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=845'),
  28.                            (u'╫º╫á╫Ö╫ò╫ƒ ╫ö╫₧╫á╫Ö╫ò╫¬ - ╫ÿ╫ò╫¿ ╫⌐╫æ╫ò╫ó╫Ö', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=3175'),
  29.                            (u'╫í╫æ╫Ö╫æ╫ö', u'http://www.globes.co.il/webservice/rss/rssfeeder.asmx/FeederNode?iID=3221')]
  30.  
  31.     def print_version(self, url):
  32.         split1 = url.split("=")
  33.         print_url = 'http://www.globes.co.il/serve/globes/printwindow.asp?did=' + split1[1]
  34.         return print_url
  35.  
  36.  
  37.     def preprocess_html(self, soup):
  38.         soup.find('tr',attrs={'bgcolor':'black'}).findPrevious('tr').extract()
  39.         soup.find('tr',attrs={'bgcolor':'black'}).extract()
  40.         return soup
  41.  
  42.     def fixChars(self,string):
  43.         # Replace lsquo (\x91)
  44.         fixed = re.sub("Γûá","Γûá",string)
  45.         return fixed
  46.  
  47.  
  48.