def cmd(config, args):
if (args.command not in avaliable_cmds):
raise Exception("not a valid command...")
nid = args.node_id
logger.info("node_id: %s"%(nid))
node_queue = NodeQueue(nid, redis_config=config['redis_config'])
node_coordinator = NodeCoordinator(config['redis_config'])
# this can be done locally without sending the command to the servers...
if (args.command == 'GET_UIDS_FROM_SCREEN_NAMES'):
apikeys = config["apikeys"].values()[0]
if (not os.path.exists(args.json)):
raise Exception("doesn't exist... ")
with open(os.path.abspath(args.json), 'rb') as f, open(os.path.abspath(args.output), 'wb') as o_f:
screen_names = json.load(f)
user_api = User(apikeys=apikeys)
user_ids = user_api.get_user_ids_by_screen_names(screen_names)
json.dump(list(user_ids), o_f)
elif (args.command == 'GET_USERS_FROM_IDS'):
apikeys = config["apikeys"].values()[0]
if (not os.path.exists(args.json)):
raise Exception("doesn't exist... ")
with open(os.path.abspath(args.json), 'rb') as f, open(os.path.abspath(args.output), 'wb') as o_f:
user_ids = json.load(f)
user_api = User(apikeys=apikeys)
users = user_api.get_users(user_ids)
json.dump(list(users), o_f)
elif (args.command.startswith('BATCH_')):
command = args.command.replace('BATCH_', '')
args_dict = copy.copy(args.__dict__)
if (not os.path.exists(args.json)):
raise Exception("doesn't exist... ")
with open(os.path.abspath(args.json), 'rb') as f:
user_ids = json.load(f)
for user_id in user_ids:
args_dict['user_id'] = user_id
cmd = new_cmd(command, args_dict)
node_queue.put(cmd)
elif (args.command == 'LIST_NODES'):
pp.pprint(node_coordinator.list_nodes())
elif (args.command == 'NODE_QSIZES'):
raise NotImplemented("NotImplemented yet...")
#pp.pprint(node_coordinator.list_nodes())
elif (args.command == 'SHUTDOWN_NODE'):
#node_coordinator.remove_node(nid)
#pp.pprint(node_coordinator.list_nodes())
raise NotImplemented("NotImplemented yet...")
elif (args.command == 'CLEAR_NODE_QUEUES'):
node_queue.clear_all_queues()
else:
args_dict = copy.copy(args.__dict__)
cmd = new_cmd(args.command, args_dict)
node_queue.put(cmd)
logger.info('sent [%s]'%(cmd))
def test_get_user_id(self): from tweetf0rm.twitterapi.users import User from tweetf0rm.handler.inmemory_handler import InMemoryHandler apikeys = self.config["apikeys"]["i0mf0rmer03"] #inmemoryhandler = InMemoryHandler() user_api = User(apikeys=apikeys) userIds = user_api.get_user_ids_by_screen_names(["AmericanCance"]) logger.info(userIds)
def init_user_api(
self
): # this will throw StopIteration if all proxies have been tried...
if (self.proxies):
try:
self.client_args['proxies'] = next(
self.proxies)['proxy_dict'] # this will throw out
#logger.info("client_args: %s"%json.dumps(self.client_args))
except StopIteration as exc:
raise
except Exception as exc:
self.init_user_api()
if (self.user_api):
del self.user_api
#crawler_id=self.crawler_id,
self.user_api = User(apikeys=self.apikeys,
client_args=self.client_args)
def call_user_api(apikeys, client_args):
user_api = User(apikeys=apikeys, client_args=client_args)
user_api.find_all_friend_ids(53039176, [Handler()])
def cmd(config, args):
if (args.command not in avaliable_cmds):
raise Exception("not a valid command...")
nid = args.node_id
logger.info("node_id: %s" % (nid))
node_queue = NodeQueue(nid, redis_config=config['redis_config'])
node_coordinator = NodeCoordinator(config['redis_config'])
# this can be done locally without sending the command to the servers...
if (args.command == 'GET_UIDS_FROM_SCREEN_NAMES'):
apikeys = config["apikeys"].values()[0]
if (not os.path.exists(args.json)):
raise Exception("doesn't exist... ")
with open(os.path.abspath(args.json),
'rb') as f, open(os.path.abspath(args.output), 'wb') as o_f:
screen_names = json.load(f)
user_api = User(apikeys=apikeys)
user_ids = user_api.get_user_ids_by_screen_names(screen_names)
json.dump(list(user_ids), o_f)
elif (args.command == 'GET_USERS_FROM_IDS'):
apikeys = config["apikeys"].values()[0]
if (not os.path.exists(args.json)):
raise Exception("doesn't exist... ")
with open(os.path.abspath(args.json),
'rb') as f, open(os.path.abspath(args.output), 'wb') as o_f:
user_ids = json.load(f)
user_api = User(apikeys=apikeys)
users = user_api.get_users(user_ids)
json.dump(list(users), o_f)
elif (args.command.startswith('BATCH_')):
new_command = args.command.replace('BATCH_', '')
args_dict = copy.copy(args.__dict__)
if (not os.path.exists(args.json)):
raise Exception("doesn't exist... ")
with open(os.path.abspath(args.json), 'rb') as f:
if (args.command == 'BATCH_CRAWL_TWEET'):
tweet_ids = json.load(f)
for tweet_id in tweet_ids:
print "Loading Tweet ID: ", tweet_id
args_dict['tweet_id'] = tweet_id
cmd = new_cmd(new_command, args_dict)
node_queue.put(cmd)
else:
user_ids = json.load(f)
for user_id in user_ids:
args_dict['user_id'] = user_id
cmd = new_cmd(new_command, args_dict)
node_queue.put(cmd)
elif (args.command == 'LIST_NODES'):
pp.pprint(node_coordinator.list_nodes())
elif (args.command == 'NODE_QSIZES'):
raise NotImplemented("NotImplemented yet...")
#pp.pprint(node_coordinator.list_nodes())
elif (args.command == 'SHUTDOWN_NODE'):
#node_coordinator.remove_node(nid)
#pp.pprint(node_coordinator.list_nodes())
raise NotImplemented("NotImplemented yet...")
elif (args.command == 'CLEAR_NODE_QUEUES'):
node_queue.clear_all_queues()
else:
args_dict = copy.copy(args.__dict__)
cmd = new_cmd(args.command, args_dict)
node_queue.put(cmd)
logger.info('sent [%s]' % (cmd))
def call_user_api(apikeys, client_args): user_api = User(apikeys=apikeys, client_args=client_args) user_api.find_all_friend_ids(53039176, [Handler()])
