def main():
# read the uid list
uid_list = read_uid_list()
# get user weibo 7day {user:[weibos]}
user_weibo_dict = read_user_weibo(uid_list)
uid_list = user_weibo_dict.keys()
#print 'uid_list:', len(uid_list)
#print 'user weibo dict:', len(user_weibo_dict)
flow_result = get_flow_information(uid_list)
register_result = get_profile_information(uid_list)
# compute text attribute
bulk_action = []
for user in user_weibo_dict:
weibo_list = user_weibo_dict[user]
uname = weibo_list[0]['uname']
results = compute_text_attribute(user, weibo_list)
results['uid'] = str(user)
flow_dict = flow_result[str(user)]
results = dict(results, **flow_dict)
# deal to the bulk action
user_info = {'uid':str(user), 'domain':results['domain'], 'topic':results['topic'], 'activity_geo':results['activity_geo']}
evaluation_index = get_evaluate_index(user_info, status='insert')
results = dict(results, **evaluation_index)
#print 'register_result:', register_result
register_dict = register_result[str(user)]
results = dict(results, **register_dict)
action = {'index':{'_id': str(user)}}
bulk_action.extend([action, results])
status = save_user_results(bulk_action)
return True # save by bulk
def update_atttribute_week():
# scan the user_portrait and bulk action to update
status = False
results = {}
count = 0
index_name = 'user_portriat'
index_type = 'user'
s_re = scan(es, query={'query':{'match_all':{}}, 'size':1000}, index=index_name, doc_type=index_type)
while True:
bulk_action = []
uid_list = []
while True:
try:
scan_re = s_re.next()['_source']
count += 1
except StopIteration:
print 'all done'
sys.exit(0)
except Exception, r:
print Exception, r
sys.exit(0)
uid = scan_re['uid']
uid_list.append(uid)
if count%1000==0:
break
if uid_list:
# get user list weibo dict from weibo api
user_weibo_dict = read_user_weibo(uid_list)
status = compute2in(uid_list, user_weibo_dict, status='update')
print 'status:', status
def compute_attribute(uid_list=[]):
# test
user_weibo_dict = read_user_weibo(uid_list)
uid_list = user_weibo_dict.keys()
flow_result = get_flow_information(uid_list)
register_result = get_profile_information(uid_list)
bulk_action = []
count = 0
count_list = set()
for user in uid_list:
weibo_list = user_weibo_dict[user]
uname = weibo_list[0]['uname']
results = compute_text_attribute(user, weibo_list)
results['uname'] = uname
results['uid'] = str(user)
flow_dict = flow_result[str(user)]
results.update(flow_dict)
user_info = {'uid':str(user), 'domain':results['domain'], 'topic':results['topic'], 'activity_geo':results['geo_string']}
evaluation_index = get_evaluate_index(user_info, status='insert')
results.update(evaluation_index)
register_dict = register_result[user]
results.update(register_dict)
action = {'index':{'_id':str(user)}}
bulk_action.extend([action, results])
count_list.add(user)
count += 1
if count % 200 == 0:
es.bulk(bulk_action, index=index_name, doc_type="user", timeout=60)
bulk_action = []
print count
if bulk_action:
status = save_user_results(bulk_action)
return "1"
def update_portrait():
user_weibo_dict = read_user_weibo()
uid_list = user_weibo_dict.keys()
flow_result = get_flow_information(uid_list)
bulk_action = []
count = 0
for user in uid_list:
action = {'update':{'_id': str(user)}}
result = {'doc':flow_result[user]}
bulk_action.extend([action, result])
count += 1
if count % 500 == 0:
es.bulk(bulk_action, index='sensitive_user_portrait', doc_type='user', timeout=60)
bulk_action = []
print count
if bulk_action:
es.bulk(bulk_action, index='sensitive_user_portrait', doc_type='user', timeout=60)
return '1'