home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / noaa.recipe < prev    next >
Text File  |  2011-09-09  |  1KB  |  43 lines
  1. #!/usr/bin/env  python
  2.  
  3. __license__   = 'GPL v3'
  4. __copyright__ = '2009, Darko Miletic <darko.miletic at gmail.com>'
  5. '''
  6. noaa.com
  7. '''
  8.  
  9. from calibre.web.feeds.news import BasicNewsRecipe
  10. from calibre.ebooks.BeautifulSoup import Tag
  11.  
  12. class NOAA(BasicNewsRecipe):
  13.     title                  = 'NOAA Online'
  14.     __author__             = 'Darko Miletic'
  15.     description            = 'NOAA'
  16.     publisher              = 'NOAA'
  17.     category               = 'news, science, US, ocean'
  18.     oldest_article         = 15
  19.     max_articles_per_feed  = 100
  20.     no_stylesheets         = True
  21.     use_embedded_content   = False
  22.     simultaneous_downloads = 1
  23.     encoding               = 'utf-8'
  24.     lang                   = 'en-US'
  25.     language = 'en'
  26.  
  27.  
  28.  
  29.     remove_tags        = [dict(name=['embed','object'])]
  30.     keep_only_tags     = [dict(name='div', attrs={'id':'contentArea'})]
  31.  
  32.     feeds          = [(u'NOAA articles', u'http://www.rss.noaa.gov/noaarss.xml')]
  33.  
  34.     def preprocess_html(self, soup):
  35.         soup.html['xml:lang'] = self.lang
  36.         soup.html['lang']     = self.lang
  37.         mlang = Tag(soup,'meta',[("http-equiv","Content-Language"),("content",self.lang)])
  38.         mcharset = Tag(soup,'meta',[("http-equiv","Content-Type"),("content","text/html; charset=UTF-8")])
  39.         soup.head.insert(0,mlang)
  40.         soup.head.insert(1,mcharset)
  41.         return self.adeify_images(soup)
  42.  
  43.