4 # Parse arbitrary XML news streams into an object type
5 # understandable by Planet UCC.
6 # Now uses feedparser to parse 9 different types of RSS _and_ Atom
13 sys.path.insert(0, 'extra')
16 feedparser.USER_AGENT = "PlanetUCC/1.0b +http://planet.ucc.asn.au/ %s" % feedparser.USER_AGENT
36 def __init__(self, URL, blogObject):
38 self.blogObject = blogObject
41 "Return a single Blog object"
43 if self.blogObject and self.blogObject.cache:
44 sys.stdout.write('Downloading feed %s...' % self.feedURL)
46 data = feedparser.parse(self.feedURL, self.blogObject.cache.etag, self.blogObject.cache.date)
47 sys.stdout.write('done.\n')
49 sys.stdout.write('failed.\n')
51 # check to see what we got returned
52 if data['items'] == [] and data['channel'] == {}:
53 sys.stdout.write('Feed %s is upto date.\n' % self.feedURL)
54 return self.blogObject
56 sys.stdout.write('Downloading feed from %s (no cache)...' % self.feedURL)
58 data = feedparser.parse(self.feedURL)
59 sys.stdout.write('done.\n')
61 sys.stdout.write('failed.\n')
65 cache = CacheHandler.CacheObject()
67 cache.etag = data['etag']
71 cache.date = data['modified']
76 sys.stderr.write('DEBUG: XMLParse2: cache item generation failed\n')
78 # parse the return of data into a blog
79 if data['channel'].has_key('title'):
80 item.blogTitle = data['channel']['title']
82 item.blogTitle = '(Unknown)'
83 if data['channel'].has_key('link'):
84 item.blogURL = data['channel']['link']
86 item.blogURL = self.feedURL
87 for entry in data['items']:
89 if entry.has_key('title'):
90 blogItem.itemTitle = entry['title']
92 blogItem.itemTitle = '(Untitled)'
93 if entry.has_key('link'):
94 blogItem.itemURL = entry['link']
96 blogItem.itemURL = item.blogURL
97 if entry.has_key('date_parsed'):
98 blogItem.itemDate = time.mktime(entry['date_parsed']) + 28800
100 blogItem.itemDate = 0
101 if entry.has_key('description'):
102 blogItem.contents = entry['description']
104 blogItem.contents = '(entry could not be retrieved)'
105 item.items.append(blogItem)