home *** CD-ROM | disk | FTP | other *** search
/ Maximum CD 2010 November / maximum-cd-2010-11.iso / DiscContents / calibre-0.7.13.msi / file_3898 < prev    next >
Encoding:
Text File  |  2010-01-16  |  5.4 KB  |  95 lines

  1. #!/usr/bin/env  python
  2. __license__     = 'GPL v3'
  3. __author__      = 'Lorenzo Vigentini'
  4. __copyright__   = '2009, Lorenzo Vigentini <l.vigentini at gmail.com>'
  5. __version__     = 'v1.01'
  6. __date__        = '14, January 2010'
  7. __description__ = 'Macworld is an independent journal not affiliated with Apple Computer.'
  8.  
  9. '''
  10. http://www.macworld.co.uk/
  11. '''
  12.  
  13. from calibre.web.feeds.news import BasicNewsRecipe
  14. from calibre.ptempfile import PersistentTemporaryFile
  15.  
  16. temp_files = []
  17. articles_are_obfuscated = True
  18.  
  19. class macWorld(BasicNewsRecipe):
  20.     __author__    = 'Lorenzo Vigentini'
  21.     description   = 'Macworld is an independent journal not affiliated with Apple Computer.'
  22.     cover_url     = 'http://images.macworld.com/images/templates/v4/mw-logo.gif'
  23.  
  24.     title          = 'Mac World '
  25.     publisher      = 'IDG Communication'
  26.     category       = 'Apple, Mac, video, computing, product reviews, editing, cameras, production'
  27.  
  28.     language       = 'en'
  29.     timefmt        = '[%a, %d %b, %Y]'
  30.  
  31.     oldest_article        = 7
  32.     max_articles_per_feed = 20
  33.     use_embedded_content  = False
  34.     recursion             = 10
  35.  
  36.     remove_javascript     = True
  37.     no_stylesheets        = True
  38.  
  39.     def get_obfuscated_article(self, url):
  40.         br = self.get_browser()
  41.         br.open(url+'&print')
  42.  
  43.         response = br.follow_link(url, nr = 0)
  44.         html = response.read()
  45.  
  46.         self.temp_files.append(PersistentTemporaryFile('_fa.html'))
  47.         self.temp_files[-1].write(html)
  48.         self.temp_files[-1].close()
  49.         return self.temp_files[-1].name
  50.  
  51.     keep_only_tags     = [
  52.                             dict(name='div', attrs={'id':'article'})
  53.                         ]
  54.  
  55.     remove_tags        = [
  56.                             dict(name='div', attrs={'class':['toolBar','mac_tags','toolBar  btmTools','textAds']}),
  57.                             dict(name='p', attrs={'class':'breadcrumbs'}),
  58.                             dict(name='div', attrs={'id':['breadcrumb','sidebar','comments']})
  59.  
  60.                         ]
  61.  
  62.     feeds          = [
  63.                        (u'MacWorld Headlines', u'http://rss.macworld.com/macworld/news'),
  64.                        (u'How-To', u'http://rss.macworld.com/macworld/howto'),
  65.                        (u'Security', u'http://rss.macworld.com/macworld/topics/security'),
  66.                        (u'MAC IT', u'http://rss.macworld.com/macworld/topics/mac_it'),
  67.                        (u'Business Mac', u'http://rss.macworld.com/macworld/topics/business_mac'),
  68.                        (u'Reviews', u'http://rss.macworld.com/macworld/reviews'),
  69.                        (u'Products: Mac', u'http://rss.macworld.com/macworld/products/mac'),
  70.                        (u'Products: iPod', u'http://rss.macworld.com/macworld/products/ipod'),
  71.                        (u'Products: iPhone', u'http://rss.macworld.com/macworld/products/iphone'),
  72.                        (u'Products: Software', u'http://rss.macworld.com/macworld/products/mac/software'),
  73.                        (u'OSX Hints', u'http://rss.macworld.com/macworld/weblogs/macosxhints'),
  74.                        (u'Mac Gems', u'http://rss.macworld.com/macworld/weblogs/macgems'),
  75.                        (u'Mac 911', u'http://rss.macworld.com/macworld/weblogs/mac911'),
  76.                        (u'Game Room', u'http://rss.macworld.com/macworld/topics/games'),
  77.                        (u'Editos notes', u'http://rss.macworld.com/macworld/weblogs/editors'),
  78.                        (u'Creative notes', u'http://rss.macworld.com/macworld/weblogs/creative'),
  79.                        (u'Playlist', u'http://rss.macworld.com/macworld/weblogs/ipodblog'),
  80.                        (u'Mobile', u'http://rss.macworld.com/macworld/weblogs/mobile'),
  81.                        (u'From the lab', u'http://rss.macworld.com/macworld/weblogs/macworldlab'),
  82.                        (u'MacUser', u'http://rss.macworld.com/macworld/weblogs/macuser')
  83.                      ]
  84.  
  85.     extra_css = '''
  86.                 h1 {color:#008852;font-family:Arial,Helvetica,sans-serif; font-size:20px; font-size-adjust:none; font-stretch:normal; font-style:normal; font-variant:normal; font-weight:bold; line-height:18px;}
  87.                 h2 {color:#4D4D4D;font-family:Arial,Helvetica,sans-serif; font-size:16px; font-size-adjust:none; font-stretch:normal; font-style:normal; font-variant:normal; font-weight:bold; line-height:16px; }
  88.                 h3 {color:#4D4D4D;font-family:Arial,Helvetica,sans-serif; font-size:15px; font-size-adjust:none; font-stretch:normal; font-style:normal; font-variant:normal; font-weight:bold; line-height:14px;}
  89.                 h4 {color:#333333; font-family:Arial,Helvetica,sans-serif;font-size:13px; font-size-adjust:none; font-stretch:normal; font-style:normal; font-variant:normal; font-weight:bold; line-height:14px; }
  90.                 h5 {color:#333333; font-family:Arial,Helvetica,sans-serif; font-size:11px; font-size-adjust:none; font-stretch:normal; font-style:normal; font-variant:normal; font-weight:bold; line-height:14px; text-transform:uppercase;}
  91.                 .newsdate {color:#333333;font-family:Arial,Helvetica,sans-serif;font-size:10px; font-size-adjust:none; font-stretch:normal; font-style:italic; font-variant:normal; font-weight:bold; line-height:10px; text-decoration:none;}
  92.                 .articleInfo {color:#4D4D4D;font-family:Arial,Helvetica,sans-serif;font-size:10px; font-size-adjust:none; font-stretch:normal; font-style:bold; font-variant:normal; font-weight:bold; line-height:10px; text-decoration:none;}
  93.                 img {align:left;}
  94.                 '''
  95.