refeed/trunk/fetch_rss.py

#!/usr/bin/env python
#-----------------------------------------
# $Id: fetch_rss.py,v 1.4 2006/03/22 05:47:54 migurski Exp $
# 
# vim: ts=4 foldcolumn=4 foldmethod=marker
# 
# This file is part of Reblog: http://reblog.org
# A derivative work of Feed On Feeds: http://feedonfeeds.com
# 
# Distributed under the Gnu Public License, see LICENSE
# 
# Copyright 2004 Michael Frumin, Michal Migurski
# mike@stamen.com, http://stamen.com
# mfrumin@eyebeam.org, http://eyebeam.org
# 
# Feed parser of last resort: if Reblog's usual Magpie parser is unable
# to make sense of an RSS source, Mark Pilgrim's Universal Feed Parser is
# called in for duty. This file acccepts an RSS URL as an argument, and
# returns a representation of the RSS data in JSON format.
#-----------------------------------------
import sys, urllib

sys.path.insert(0, 'library/json-py')
import json

sys.path.insert(0, 'library/feedparser')
import feedparser

# (source, destination) for each attribute of a feed
feed_attrs = [('title', 'title'), ('link', 'link'), ('tagline', 'description')]

# (source, destination) for each attribute of an item
item_attrs = [('id', 'guid'), ('link', 'link'), ('title', 'title'), ('summary', 'description'),
              ('author', 'author'), ('category', 'category'), ('modified', 'modified')]

# read the feed URL from the command line and pass it through the feedparser
try:
    feed_url = sys.argv[-1].strip()
    feed_src = urllib.urlopen(feed_url).read()
    feed_obj = feedparser.parse(feed_src)
except IOError:
    sys.exit("Failed to retrieve feed from URL")

# create the final feed object
feed = {'channel': {}, 'items': []}

# assign channel properties
for (src, dest) in feed_attrs:
    try:
        feed['channel'][dest] = getattr(feed_obj.feed, src).encode('UTF-8')
    except:
        pass

# assign item properties for each item
for entry_src in feed_obj.entries:

    item = {}
    
    for (src, dest) in item_attrs:
        try:
            item[dest] = getattr(entry_src, src).encode('UTF-8')
        except:
            pass
                 
    feed['items'].append(item)

# return the result
sys.stdout.write(json.write(feed))
1	dpavlin	2	#!/usr/bin/env python
2			#-----------------------------------------
3			# $Id: fetch_rss.py,v 1.4 2006/03/22 05:47:54 migurski Exp $
4			#
5			# vim: ts=4 foldcolumn=4 foldmethod=marker
6			#
7			# This file is part of Reblog: http://reblog.org
8			# A derivative work of Feed On Feeds: http://feedonfeeds.com
9			#
10			# Distributed under the Gnu Public License, see LICENSE
11			#
12			# Copyright 2004 Michael Frumin, Michal Migurski
13			# mike@stamen.com, http://stamen.com
14			# mfrumin@eyebeam.org, http://eyebeam.org
15			#
16			# Feed parser of last resort: if Reblog's usual Magpie parser is unable
17			# to make sense of an RSS source, Mark Pilgrim's Universal Feed Parser is
18			# called in for duty. This file acccepts an RSS URL as an argument, and
19			# returns a representation of the RSS data in JSON format.
20			#-----------------------------------------
21			import sys, urllib
22
23			sys.path.insert(0, 'library/json-py')
24			import json
25
26			sys.path.insert(0, 'library/feedparser')
27			import feedparser
28
29			# (source, destination) for each attribute of a feed
30			feed_attrs = [('title', 'title'), ('link', 'link'), ('tagline', 'description')]
31
32			# (source, destination) for each attribute of an item
33			item_attrs = [('id', 'guid'), ('link', 'link'), ('title', 'title'), ('summary', 'description'),
34			('author', 'author'), ('category', 'category'), ('modified', 'modified')]
35
36			# read the feed URL from the command line and pass it through the feedparser
37			try:
38			feed_url = sys.argv[-1].strip()
39			feed_src = urllib.urlopen(feed_url).read()
40			feed_obj = feedparser.parse(feed_src)
41			except IOError:
42			sys.exit("Failed to retrieve feed from URL")
43
44			# create the final feed object
45			feed = {'channel': {}, 'items': []}
46
47			# assign channel properties
48			for (src, dest) in feed_attrs:
49			try:
50			feed['channel'][dest] = getattr(feed_obj.feed, src).encode('UTF-8')
51			except:
52			pass
53
54			# assign item properties for each item
55			for entry_src in feed_obj.entries:
56
57			item = {}
58
59			for (src, dest) in item_attrs:
60			try:
61			item[dest] = getattr(entry_src, src).encode('UTF-8')
62			except:
63			pass
64
65			feed['items'].append(item)
66
67			# return the result
68			sys.stdout.write(json.write(feed))