def _initial_populate(self):
'''Gather a group of TwitterUsers. Tries to choose a highly
interconnected group.'''
root_user = TwitterUser(self._root_user_id)
root_node = root_user.get_all_data()
self._community_members.append(root_node)
root_score = self._filled_user_score(root_node)
#user scores determine how interconnected a user is
self._node_pool[self._root_user_id] = {
'user': root_node,
'score': root_score
}
logging.debug('Adding root to node_pool')
self.save()
self._resume_populate()
def _resume_populate(self):
'''Gather a group of TwitterUsers. Tries to choose a highly
interconnected group. Picks up where a previous _populate()
call stopped'''
while self._node_pool and len(self._community_members) < \
self._max_population:
'''
choose person on list with highest interconnection
first 0 at end of line is to take first user, second
0 get's user's id from tuple result
'''
self._rescore_node_pool()
highest_scoring_id = sorted(self._node_pool.items(),
key=lambda item: item[1]['score'],
reverse=True)[0][0]
curr_node = self._node_pool[highest_scoring_id]['user']
#choose friends by rank to add to community, seems to work
#better then followers
friend_ids = self._sort_by_empty_score(curr_node['friend_ids'])
added_count = 0
community_ids = []
for member in self._community_members:
community_ids.append(member['uid'])
#flag to delete the id we are using from the node pool when finished with it
delete_highest_scoring_id = True
for friend_id in friend_ids:
if added_count <= self._max_friends_per_user:
try:
if friend_id not in community_ids:
tu = TwitterUser(friend_id)
sleep(1)
new_user = tu.get_all_data()
#TODO add some sort of conditions for addition to the community
self._community_members.append(new_user)
logging.debug('TwitterUser accepted to community')
new_user_score = self._filled_user_score(new_user)
self._node_pool[friend_id] = {
'user': new_user,
'score': new_user_score
}
if self._safe:
self.save()
print len(self._community_members), "members"
added_count += 1
logging.debug('TwitterUser accepted to node_pool')
delete_highest_scoring_id = True
except TwitterHTTPError as error:
logging.debug('Twitter error: %s' % error)
self.save()
print "Number of members: ", len(
self._community_members)
#rate limiting error
if '400' in str(error) or '420' in str(error):
delete_highest_scoring_id = False
logging.debug('Hit rate limit, quitting')
return
#unauthorized for user error
elif '401' in str(error) or '404' in str(error):
delete_highest_scoring_id = True
if self._safe:
self.save()
#otherwise it's probably just a twitter server issue
else:
delete_highest_scoring_id = False
logging.debug('Server error, sleeping for 5 secs')
sleep(5)
except BadUser as error:
delete_highest_scoring_id = True
logging.debug('TwitterUser rejected: %s' % error)
except URLError:
delete_highest_scoring_id = False
logging.debug('URLError, sleeping for 5 secs')
sleep(5)
#once a user is chosen for evaluation pop him off the node list
del self._node_pool[highest_scoring_id]
if self._safe:
self.save()
logging.debug('Deleting node from node pool')
self.save()
print "Maximum community size reached."
print "Number of members: ", len(self._community_members)