pod.py

   1 #!/usr/bin/env python
   2
   3 from __future__ import with_statement
   4 import urllib2
   5 from email.utils import parsedate
   6 import time
   7 import feedparser
   8
   9 class Feed:
  10     # consider timestamps earlier then this non-existant
  11     reasonable_timestamp = time.mktime(time.strptime("2008", "%Y"))
  12
  13     def __init__(self, url):
  14         self.url = url
  15         # last url timestamp if any
  16         self.timestamp = None
  17         # ignore episodes older then this
  18         self.watermark = time.mktime(time.strptime("03 2008", "%m %Y"))
  19
  20     def get_remote_timestamp(self):
  21         conn = urllib2.urlopen(self.url)
  22         try:
  23             if 'Last-Modified' in conn.info():
  24                 t = time.mktime( parsedate( conn.info()['Last-Modified'] ) )
  25                 if t > self.reasonable_timestamp:
  26                     return t
  27             return None
  28         finally:
  29             conn.close()
  30
  31     def is_updated(self):
  32         if self.timestamp is not None:
  33             remote_timestamp = self.get_remote_timestamp()
  34             return remote_timestamp is None or \
  35                 self.timestamp < remote_timestamp
  36         else:
  37             return True
  38
  39     def get_new_episodes(self):
  40         tree = feedparser.parse( self.url )
  41         watermark = self.watermark
  42         for e in tree.entries:
  43             timestamp = time.mktime(e.modified_parsed)
  44             if timestamp > self.watermark and 'enclosures' in e:
  45                 for encl in e.enclosures:
  46                     yield encl.href
  47                 if timestamp > watermark:
  48                     watermark = timestamp
  49         self.watermark = watermark
  50
  51 def get_subscribed_feeds():
  52     with open("feeds.lst","r") as f:
  53         for url in map(str.strip, f.readlines()):
  54             yield Feed(url)
  55
  56 def get_updated_feeds():
  57     for feed in get_subscribed_feeds():
  58         if feed.is_updated():
  59             yield feed
  60
  61 def download_episode(url):
  62     pass
  63
  64 dl_queue = []
  65
  66 for feed in get_updated_feeds():
  67     print feed.url
  68     for url in feed.get_new_episodes():
  69         print url
  70         dl_queue.append(url)
  71     # update feed watermark
  72
  73 file_manager = []
  74
  75 for url in dl_queue:
  76     if download_episode(url):
  77         dl_queue.remove(url)
  78         file_manager.append(url)
  79
  80 # file_manager.expire_and_prune()
  81
  82