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
37 def __init__(self, URL, blogObject):
39 self.blogObject = blogObject
42 "Return a single Blog object"
44 if self.blogObject and self.blogObject.cache:
45 sys.stdout.write('Downloading feed %s...' % self.feedURL)
47 data = feedparser.parse(self.feedURL, self.blogObject.cache.etag, self.blogObject.cache.date)
48 sys.stdout.write('done.\n')
50 sys.stdout.write('failed.\n')
52 # check to see what we got returned
53 if data['items'] == [] and data['channel'] == {}:
54 sys.stdout.write('Feed %s is upto date.\n' % self.feedURL)
55 return self.blogObject
57 sys.stdout.write('Downloading feed from %s (no cache)...' % self.feedURL)
59 data = feedparser.parse(self.feedURL)
60 sys.stdout.write('done.\n')
62 sys.stdout.write('failed.\n')
66 cache = CacheHandler.CacheObject()
68 cache.etag = data['etag']
72 cache.date = data['modified']
77 sys.stderr.write('DEBUG: XMLParse2: cache item generation failed\n')
79 # parse the return of data into a blog
80 if data['channel'].has_key('title'):
81 item.blogTitle = data['channel']['title']
83 item.blogTitle = '(Unknown)'
84 if data['channel'].has_key('link'):
85 item.blogURL = data['channel']['link']
87 item.blogURL = self.feedURL
88 for entry in data['items']:
90 if entry.has_key('title'):
91 blogItem.itemTitle = entry['title']
93 blogItem.itemTitle = '(Untitled)'
94 if entry.has_key('link'):
95 blogItem.itemURL = entry['link']
97 blogItem.itemURL = item.blogURL
98 if entry.has_key('date_parsed'):
99 blogItem.itemDate = time.mktime(entry['date_parsed']) + 28800
101 blogItem.itemDate = 0
102 if entry.has_key('description'):
103 blogItem.contents = entry['description']
105 blogItem.contents = '(entry could not be retrieved)'
106 item.items.append(blogItem)