def mates(conf, dlcs, *args, **opts):
"""The following was adapted from delicious_mates.
http://www.aiplayground.org/artikel/delicious-mates/
% dlcs mates [max_mates[, min_bookmarks[, min_common]]]
"""
max_mates, min_bookmarks, min_common = map(int, opts['mates'].split(','))
delicious_users = {}
posts = cached_posts(conf, dlcs, opts['keep_cache'])
print "Getting mates for collection of %i bookmarks" % len(posts['posts'])
print "\nUsers for each bookmark:"
for i, post in enumerate(posts['posts']):
hash = md5.md5(post['href']).hexdigest()
#urlfeed = pydelicious.dlcs_feed('urlinfo', urlmd5=hash)
posts = dlcs_feed('url', count='all', format='rss', urlmd5=hash)
usernames = [e['author'] for e in posts['entries']]
print " %i. %s (%i)" % (i+1, post['href'], len(usernames))
for username in usernames:
if username != dlcs.user:
delicious_users.setdefault(username, (0.0, 0))
(weight, num_common) = delicious_users[username]
new_weight = weight + 1.0/math.log(len(usernames)+1.0)
delicious_users[username] = (new_weight, num_common + 1)
print "\n%i candidates from list of %i users" % (max_mates, len(delicious_users))
friends = {}
for (username, (weight, num_common)) in value_sorted(delicious_users):
if num_common >= min_common:
num_bookmarks = float([e['summary'] for e in
dlcs_feed('user_info', format='rss', username='******')['entries']
if e['id'] == 'items'][0])
print " %s (%i/%i)" % (username, num_common, num_bookmarks),
if num_bookmarks >= min_bookmarks:
friends[username] = (weight*(num_common/num_bookmarks), num_common, num_bookmarks)
if len(friends) >= max_mates:
break
else:
print
time.sleep(1)
print "\nTop %i del.icio.us mates:" % max_mates
print "username".ljust(20), "weight".ljust(20), "# common bookmarks".ljust(20), "# total bookmarks".ljust(20), "% common"
print "--------------------------------------------------------------------------------------------"
for (username, (weight, num_common, num_total)) in value_sorted(friends)[:max_mates]:
print username.ljust(20),
print ("%.5f" % (weight*100)).ljust(20),
print str(num_common).ljust(20),
print str(int(num_total)).ljust(20),
print "%.5f" % ((num_common/num_total)*100.0)
def main(argv):
optparser, opts, args = parse_argv(__options__, argv, __usage__)
kwds = {}
for k in ('format', 'username', 'tag', 'urlmd5', 'key'):
v = getattr(opts, k)
if type(v) != type(None):
kwds[k] = v
if opts.clear_screen:
if os.name == 'posix':
os.system('clear')
else:
os.system('cls')
matches, candidates = feeds_for_params(**kwds)
if opts.list_feeds:
print >> sys.stderr, "Feeds for current parameters (%s)" % kwds
if matches:
print "Exact matches:"
for m in matches:
print '\t' + m + ':', pydelicious.delicious_v2_feeds[m] % kwds
if candidates:
print "Candidates:"
for m in candidates:
print '\t' + pydelicious.delicious_v2_feeds[m]
sys.exit()
path = args and args.pop(0)
if not path:
if len(matches) == 1:
path = matches[0]
print >> sys.stderr, "Setting path to %s" % path
elif matches:
print >> sys.stderr, "Multiple paths for given parameters, see -L"
sys.exit()
assert path in pydelicious.delicious_v2_feeds
return pydelicious.dlcs_feed(path, **kwds)
def __init__(self, username, password, key, logfile, verbose=False):
self.inbox = pydelicious.dlcs_feed('user_inbox', format='rss',
username=username, key=key)
self.api = pydelicious.DeliciousAPI(username, password)
self.posters = json_load(pydelicious.json_network(username))
self.logfile = logfile
self.verbose = verbose
def main(argv):
optparser, opts, args = parse_argv(__options__, argv, __usage__)
kwds = {}
for k in ('format','username','tag','urlmd5','key'):
v = getattr(opts, k)
if type(v) != type(None):
kwds[k] = v
if opts.clear_screen:
if os.name == 'posix':
os.system('clear')
else:
os.system('cls')
matches, candidates = feeds_for_params(**kwds)
if opts.list_feeds:
print >>sys.stderr,"Feeds for current parameters (%s)" % kwds
if matches:
print "Exact matches:"
for m in matches:
print '\t'+m+':', pydelicious.delicious_v2_feeds[m] % kwds
if candidates:
print "Candidates:"
for m in candidates:
print '\t'+pydelicious.delicious_v2_feeds[m]
sys.exit()
path = args and args.pop(0)
if not path:
if len(matches) == 1:
path = matches[0]
print >>sys.stderr, "Setting path to %s" % path
elif matches:
print >>sys.stderr, "Multiple paths for given parameters, see -L"
sys.exit()
assert path in pydelicious.delicious_v2_feeds
return pydelicious.dlcs_feed(path, **kwds)
def mates(conf, dlcs, *args, **opts):
"""The following was adapted from delicious_mates.
http://www.aiplayground.org/artikel/delicious-mates/
% dlcs mates [max_mates[, min_bookmarks[, min_common]]]
"""
max_mates, min_bookmarks, min_common = map(int, opts['mates'].split(','))
delicious_users = {}
posts = cached_posts(conf, dlcs, opts['keep_cache'])
print "Getting mates for collection of %i bookmarks" % len(posts['posts'])
print "\nUsers for each bookmark:"
for i, post in enumerate(posts['posts']):
hash = md5.md5(post['href']).hexdigest()
#urlfeed = pydelicious.dlcs_feed('urlinfo', urlmd5=hash)
posts = dlcs_feed('url', count='all', format='rss', urlmd5=hash)
usernames = [e['author'] for e in posts['entries']]
print " %i. %s (%i)" % (i + 1, post['href'], len(usernames))
for username in usernames:
if username != dlcs.user:
delicious_users.setdefault(username, (0.0, 0))
(weight, num_common) = delicious_users[username]
new_weight = weight + 1.0 / math.log(len(usernames) + 1.0)
delicious_users[username] = (new_weight, num_common + 1)
print "\n%i candidates from list of %i users" % (max_mates,
len(delicious_users))
friends = {}
for (username, (weight, num_common)) in value_sorted(delicious_users):
if num_common >= min_common:
num_bookmarks = float([
e['summary'] for e in dlcs_feed(
'user_info', format='rss', username='******')['entries']
if e['id'] == 'items'
][0])
print " %s (%i/%i)" % (username, num_common, num_bookmarks),
if num_bookmarks >= min_bookmarks:
friends[username] = (weight * (num_common / num_bookmarks),
num_common, num_bookmarks)
if len(friends) >= max_mates:
break
else:
print
time.sleep(1)
print "\nTop %i del.icio.us mates:" % max_mates
print "username".ljust(20), "weight".ljust(20), "# common bookmarks".ljust(
20), "# total bookmarks".ljust(20), "% common"
print "--------------------------------------------------------------------------------------------"
for (username, (weight, num_common,
num_total)) in value_sorted(friends)[:max_mates]:
print username.ljust(20),
print("%.5f" % (weight * 100)).ljust(20),
print str(num_common).ljust(20),
print str(int(num_total)).ljust(20),
print "%.5f" % ((num_common / num_total) * 100.0)
def check(self):
"""Check inbox for new items."""
logfh = None
if os.access(self.logfile, os.F_OK):
if self.verbose:
print "[LOG] Log file found. Trying to resume...",
try:
# read in previous log data for resuming
logfh = open(self.logfile, 'r')
logfh.readline()
logfh.close()
if self.verbose:
print "OK"
except IOError:
# most probably, the log file does not exist (yet)
if self.verbose:
print "failed"
else:
# log file does not exist, so there isn't any resume data
# to read in
pass
try:
if self.verbose:
print "[LOG] Open log file for appending...",
logfh = open(self.logfile, 'a')
if self.verbose:
print "done"
except IOError:
if self.verbose:
print "failed"
print "[LOG] ERROR: could not open log file for appending"
self._cleanup()
return
# query metadata about each entry from delicious.com
for index, entry in enumerate(self.inbox.entries):
url = entry.link
urlmd5 = md5.new(url).hexdigest()
if entry.author in self.posters:
if self.verbose:
logfh.write("[LOG] %s Processing entry #%s: '%s'\n" % \
(strftime("%Y-%m-%d %H:%M:%S"), index + 1, url))
try:
sleep(1) # be nice and wait 1 sec between connects
urlinfo = json_load(pydelicious.dlcs_feed('urlinfo',
urlmd5=urlmd5))
if urlinfo:
urlinfo = urlinfo[0]
else:
urlinfo = {}
title = urlinfo['title']
top_tags = urlinfo['top_tags'] or []
tagstr = 'via:%s ' % entry.author + \
' '.join([tag.replace(' ','_') for tag in top_tags])
self.api.posts_add(url, title, tags=tagstr.strip())
if self.verbose:
logfh.write("[LOG] %s Saved %s\n" % \
(strftime("%Y-%m-%d %H:%M:%S"), url))
except KeyError:
pass
except pydelicious.DeliciousItemExistsError:
if self.verbose:
logfh.write("[LOG] %s %s already added\n" % \
(strftime("%Y-%m-%d %H:%M:%S"), url))
except:
logfh.write("[LOG] %s ERROR: %s\n" % \
(strftime("%Y-%m-%d %H:%M:%S"), sys.exc_info()[0]))
# clean up
logfh.close()
raise
else:
logfh.write("[LOG] %s ERROR: %s not in network-aborting %s\n" % :\
(strftime("%Y-%m-%d %H:%M:%S"), entry.author, url))
# clean up
logfh.close()
