home *** CD-ROM | disk | FTP | other *** search
/ Chip 2011 November / CHIP_2011_11.iso / Programy / Narzedzia / Calibre / calibre-0.8.18.msi / file_280 / cynewslive.recipe < prev    next >
Encoding:
Text File  |  2011-09-09  |  4.0 KB  |  102 lines
  1. from calibre.web.feeds.news import BasicNewsRecipe
  2. from datetime import datetime, timedelta
  3.  
  4. class CyNewsLiveRecipe(BasicNewsRecipe):
  5.     __license__  = 'GPL v3'
  6.     __author__ = 'kwetal'
  7.     language = 'en_CY'
  8.     version = 1
  9.  
  10.     title = u'Cyprus News Live'
  11.     publisher = u'The Cyprus Weekly'
  12.     category = u'News, Newspaper'
  13.     description = u'News from Cyprus'
  14.  
  15.     use_embedded_content = False
  16.     remove_empty_feeds = True
  17.     oldest_article = 7
  18.     max_articles_per_feed = 100
  19.  
  20.     no_stylesheets = True
  21.     remove_javascript = True
  22.  
  23.     pubTime = None
  24.     minTime = None
  25.     articleCount = 0
  26.  
  27.     INDEX = 'http://www.cynewslive.com'
  28.  
  29.     feeds = []
  30.     feeds.append(('News: Cyprus', 'http://www.cynewslive.com/main/92,0,0,0-CYPRUS.aspx'))
  31.     feeds.append(('News: World', 'http://www.cynewslive.com/main/78,0,0,0-UKWORLD.aspx'))
  32.     feeds.append(('Sport: Football', 'http://www.cynewslive.com/main/82,0,0,0-FOOTBALL.aspx'))
  33.     feeds.append(('Sport: Rugby', 'http://www.cynewslive.com/main/83,0,0,0-RUGBY.aspx'))
  34.     feeds.append(('Sport: Cricket', 'http://www.cynewslive.com/main/85,0,0,0-CRICKET.aspx'))
  35.     feeds.append(('Sport: Tennis', 'http://www.cynewslive.com/main/84,0,0,0-TENNIS.aspx'))
  36.     feeds.append(('Sport: Other', 'http://www.cynewslive.com/main/86,0,0,0-OTHER.aspx'))
  37.     feeds.append(('Business: Local', 'http://www.cynewslive.com/main/100,0,0,0-LOCAL.aspx'))
  38.     feeds.append(('Business: Foreign', 'http://www.cynewslive.com/main/101,0,0,0-FOREIGN.aspx'))
  39.     feeds.append(('Environment', 'http://www.cynewslive.com/main/93,0,0,0-ENVIRONMENT.aspx'))
  40.     feeds.append(('Culture', 'http://www.cynewslive.com/main/208,0,0,0-CULTURE.aspx'))
  41.  
  42.     keep_only_tags = []
  43.     keep_only_tags.append(dict(name = 'div', attrs = {'class': 'ArticleCategories'}))
  44.  
  45.     extra_css = '''
  46.                 body{font-family:verdana,arial,helvetica,geneva,sans-serif ;}
  47.                 '''
  48.  
  49.     def parse_index(self):
  50.         answer = []
  51.         for feed in self.feeds:
  52.             self.articleCount = 0
  53.             articles = []
  54.             soup = self.index_to_soup(feed[1])
  55.  
  56.             table = soup.find('table', attrs = {'id': 'ctl00_cp_ctl01_listp'})
  57.             if table:
  58.                 self.pubTime = datetime.now()
  59.                 self.minTime = self.pubTime - timedelta(days = self.oldest_article)
  60.  
  61.                 self.find_articles(table, articles)
  62.  
  63.             answer.append((feed[0], articles))
  64.  
  65.         return answer
  66.  
  67.     def postprocess_html(self, soup, first):
  68.         for el in soup.findAll(attrs = {'style': True}):
  69.             del el['style']
  70.  
  71.         for el in soup.findAll('font'):
  72.             el.name = 'div'
  73.             for attr, value in el:
  74.                 del el[attr]
  75.  
  76.         return soup
  77.  
  78.     def find_articles(self, table, articles):
  79.         for div in table.findAll('div', attrs = {'class': 'ListArticle'}):
  80.             el = div.find('div', attrs = {'class': 'ListArticle_T'})
  81.             title = self.tag_to_string(el.a)
  82.             url = self.INDEX + el.a['href']
  83.  
  84.             description = self.tag_to_string(div.find('div', attrs = {'class': 'ListArticle_BODY300'}))
  85.  
  86.             el = div.find('div', attrs = {'class': 'ListArticle_D'})
  87.             if el:
  88.                 dateParts = self.tag_to_string(el).split(' ')
  89.                 monthNames = {'January': 1, 'February': 2, 'March': 3, 'April': 4, 'May': 5, 'June': 6,
  90.                               'July': 7, 'August': 8, 'September': 9, 'October': 10, 'November': 11,
  91.                               'December': 12}
  92.                 timeParts = dateParts[3].split(':')
  93.                 self.pubTime = datetime(year = int(dateParts[2]), month = int(monthNames[dateParts[1]]),
  94.                                         day = int(dateParts[0]), hour = int(timeParts[0]),
  95.                                         minute = int(timeParts[1]))
  96.  
  97.             if self.pubTime >= self.minTime and self.articleCount <= self.max_articles_per_feed:
  98.                 articles.append({'title': title, 'date': self.pubTime, 'url': url, 'description': description})
  99.                 self.articleCount += 1
  100.             else:
  101.                 return
  102.