planet.py

   1 import shelve
   2
   3 from feedcache.cache import Cache
   4
   5 from django.conf import settings
   6
   7 from django.template.defaultfilters import truncatewords_html
   8
   9
  10 def TryEncoding(content):
  11     for body_charset in 'UTF-8', 'US-ASCII', 'ISO-8859-1', :
  12         try:
  13             return content.encode(body_charset)
  14         except UnicodeError:
  15             pass
  16         except Exception, e:
  17             if not type(content) is str:
  18                 pass
  19
  20 def GetContent(feed):
  21     if hasattr(feed, "content"):
  22         return feed.content[0]["value"]
  23     else:
  24         return feed.summary
  25
  26 class Parser:
  27     """
  28     "http://planet.gentoo.org/atom.xml"
  29     "http://overlays.gentoo.org/rss20.xml"
  30     >>> f = Parser("http://www.gentoo.org/rdf/en/gentoo-news.rdf")
  31     >>> f.GetTitle()
  32     u'Planet Gentoo'
  33     >>> f.GetLink()
  34     u'http://planet.gentoo.org/'
  35     >>> for e in f: print e["title"], e["content"]
  36     """
  37
  38     def __init__(self, url, summary=False):
  39         storage = shelve.open(settings.ROOT_PATH + "feedcache")
  40         try :
  41             fc = Cache(storage)
  42             self.feed = fc.fetch(url)
  43             self.iterator = 0
  44             self.summary = summary
  45         finally:
  46             storage.close()
  47     def GetTitle(self):
  48         return self.feed.feed.title
  49
  50     def GetLink(self):
  51         return self.feed.feed.link
  52
  53     def next(self):
  54         if self.iterator >= len(self.feed.entries): raise StopIteration
  55
  56         entry = self.feed.entries[self.iterator]
  57
  58         title = TryEncoding(entry["title"])
  59         content = TryEncoding(GetContent(entry))
  60         link = entry.link
  61
  62         if self.summary:
  63             content = TryEncoding(truncatewords_html(content, 30))
  64             content = "".join((content , "..."))
  65
  66         entry = dict((("title", title), ("content", content), ("link", link)))
  67
  68         self.iterator += 1
  69         return entry
  70
  71     def __iter__(self):
  72         return self