def query():
"""
given {start_time, end_time, grid_id [optional]}
return -> [{grid_id, volume}, ...]
"""
start_time = datetime.strptime(request.query.get('start_time'), '%Y-%m-%dT%H:%M:%SZ')
end_time = datetime.strptime(request.query.get('end_time'), '%Y-%m-%dT%H:%M:%SZ')
grid_id = request.query.get('grid_id')
mg = MongoDB()
mg.connect()
print('querying grid volumes...')
# if grid_id:
# results = mg.group_by([{'$match': {'created_at': {'$gt': start_time, '$lt': end_time}, 'grid_id': grid_id}},
# {'$group': {'_id': '$grid_id', 'count': {'$sum': 1}}}
# ])
# else:
# results = mg.group_by([{'$match': {'created_at': {'$gt': start_time, '$lt': end_time}}},
# {'$group': {'_id': '$grid_id', 'count': {'$sum': 1}}}
# ])
# group and count distinct user:
results = mg.group_by([ {'$match': {'created_at': {'$gt': start_time, '$lt': end_time}}},
{'$group': {'_id':{ 'grid_id':'$grid_id', 'user_id':'$user_id' }, 'count': {'$sum': 1} } },
{'$group': {'_id': '$_id.grid_id', 'count': {'$sum': 1}}}
])
ret = []
for result in results:
ret.append({'grid_id': result['_id'], 'volume': result['count']})
response.content_type = 'application/json'
return json_dumps(ret, indent=2)
def load_tweets_to_grids():
# mongodb
mg = MongoDB()
mg.connect()
tweets = mg.find()
grid_db = GridDB()
grid_db.add(tweets)
return grid_db
def all_grids():
mg = MongoDB()
mg.connect()
griddb = GridDB()
print('querying grid volumes...')
results = mg.group_by([{'$match': {'created_at': {'$gt': datetime.strptime('2012-10-15T20:00:02Z', '%Y-%m-%dT%H:%M:%SZ'),
'$lt': datetime.strptime('2012-11-15T20:00:02Z', '%Y-%m-%dT%H:%M:%SZ')}}}]) # print(results)
griddb.add(results)
ret = Grid.get_raw_pandas_ts(results, 'D')
STL.seasonal_decomposition(ret)
def single_grid(grid_id):
mg = MongoDB()
mg.connect()
start_time = datetime.strptime('2012-10-15T20:00:02Z', '%Y-%m-%dT%H:%M:%SZ')
end_time = datetime.strptime('2012-11-15T20:00:02Z', '%Y-%m-%dT%H:%M:%SZ')
print('querying grid volumes...')
# results = mg.group_by([{'$match': {'created_at': {'$gt': start_time, '$lt': end_time}, 'grid_id': grid_id}},
# {'$group': {'_id': '$grid_id', 'count': {'$sum': 1}}}])
results = mg.group_by([{'$match': {'created_at': {'$gt': datetime.strptime('2012-10-15T00:00:00Z', '%Y-%m-%dT%H:%M:%SZ'),
'$lt': datetime.strptime('2012-11-15T00:00:00Z', '%Y-%m-%dT%H:%M:%SZ')},
'grid_id': grid_id}}]) # print(results)
print(results)
ret = Grid.get_raw_pandas_ts(results, 'H')
# print('------------')
# print(ret)
print(STL.seasonal_decomposition(ret))
def write_tweets_to_mongo():
""" load tweets, decide the grid of tweets and insert the tweets into mongo
this function serves as the initial step of the pipeline.
all following db-specific operations are based on the dataset inserted here.
This function is only called once for the entire pipeline.
"""
data_file = "data/sandy_all.txt"
kml_file = ["data/nj_ct.kml", "data/nyc_ct_sea.kml"]
# kml_file = ['data/nyc_cb_sea.kml']
tweets = load_tweets(data_file)
# trajectories = tweets_to_trajectories(tweets)
grid_db = GridDB()
grid_db.load_grid_from_file(kml_file[0])
grid_db.load_grid_from_file(kml_file[1])
# grid_db.write_grids_to_json('shapefile.json')
grid_db.check_and_add(tweets)
#####################################################
############## index Tweets into MongoDB ############
#####################################################
# mongodb
mg = MongoDB()
mg.connect()
print("connected...")
mg.drop()
# tweets to dicts;
rst = []
for t in grid_db.get_tweets():
rst.append(t.to_dict())
print("inserting...")
mg.insert_tweets(rst)
def query():
"""
given {start_time, end_time, aggregation}
return -> [{time, freq}, ...]
"""
start_time = datetime.strptime(request.query.get("start_time"), '%Y-%m-%dT%H:%M:%SZ')
end_time = datetime.strptime(request.query.get("end_time"), '%Y-%m-%dT%H:%M:%SZ')
aggregation = request.query.get("aggregation")
mg = MongoDB()
mg.connect()
print("querying time series...")
results = mg.find( {'created_at': {'$gt': start_time, '$lt': end_time}} )
results = Grid.get_ts(results, aggregation)
ret = []
for result in results:
ret.append({'start_time': result[0].to_datetime().strftime('%Y-%m-%dT%H:%M:%SZ'),
'frequency': result[1].item()})
response.content_type = 'application/json'
return json_dumps(ret, indent=2)
return grid_db
if __name__ == "__main__":
# write_tweets_to_mongo()
grid_db = load_tweets_to_grids()
print(grid_db.get_outlier_grid_ids("H"))
print("tweets loaded")
# for grid_id in grid_db.grid_cache:
# print(grid_id, grid_db.grid_cache[grid_id].get_ts(grid_db.get_tweets(), 'H'))
exit()
mg = MongoDB()
mg.connect()
print("querying grid volumes...")
results = mg.group_by(
[
{
"$match": {
"created_at": {
"$gt": datetime.strptime("2012-10-15T20:00:02Z", "%Y-%m-%dT%H:%M:%SZ"),
"$lt": datetime.strptime("2012-10-22T20:00:02Z", "%Y-%m-%dT%H:%M:%SZ"),
}
}
}
]
) # print(results)
def wrapper(func, *args, **kwargs):
mongo = MongoDB()
mongo.connect(app.config['app.mongohost'])
