def is_matched_user(user_id, spec):
metric_list = get_metric_list(spec["checks"])
time_type = spec['time_type']
if 'from_local_date' in spec and 'to_local_date' in spec:
freq_metrics = enam.summarize_by_local_date(user_id,
spec["from_local_date"],
spec["to_local_date"],
spec["freq"],
metric_list,
include_aggregate=False)
elif 'start_time' in spec and 'end_time' in spec:
freq_metrics = enam.summarize_by_timestamp(user_id,
spec["start_time"],
spec["end_time"],
spec["freq"],
metric_list,
include_aggregate=False)
else:
# If no start and end times are specified, we assume that this is a
# timestamp query because we can come up with a reasonable start and end
# time for timestamps but not for local_dates, which are basically a filter.
# so if we run this on the first of a month, for example, we won't find
# anything, which seems bogus and not what people would expect
assert time_type == "timestamp", "time_type = %s, expected timestamp" % time_type
freq_metrics = enam.summarize_by_timestamp(user_id,
0,
time.time(),
spec["freq"],
metric_list,
include_aggregate=False)
assert (freq_metrics is not None)
assert ('user_metrics' in freq_metrics)
curr_user_metrics = freq_metrics['user_metrics']
checks = spec['checks']
check_results = np.zeros(len(checks))
for i, check in enumerate(checks):
curr_metric_result = curr_user_metrics[i]
# curr_freq_result is a list of ModeStatTimeSummary objects, one for each
# grouped time interval in the range
# e.g. for daily, 2017-01-19, 2017-01-20, 2017-01-21, 2017-01-22, 2017-01-23, ....
for msts in curr_metric_result:
# We defined our check as being true if it is true for _any_ grouped time
# period in the range. So as long as we find a match for that check, we are
# good!
if matches_check(check, msts):
check_results[i] = True
logging.info("For user_id %s, check result array = %s, all? %s" %
(user_id, check_results, np.all(check_results)))
return np.all(check_results)
def testCountTimestampMetrics(self):
met_result = metrics.summarize_by_timestamp(self.testUUID,
self.aug_start_ts, self.aug_end_ts,
'd', ['count'], True)
logging.debug(met_result)
self.assertEqual(list(met_result.keys()), ['aggregate_metrics', 'user_metrics'])
user_met_result = met_result['user_metrics'][0]
agg_met_result = met_result['aggregate_metrics'][0]
self.assertEqual(len(user_met_result), 2)
self.assertEqual([m.nUsers for m in user_met_result], [1,1])
self.assertEqual(user_met_result[0].local_dt.day, 27)
self.assertEqual(user_met_result[1].local_dt.day, 28)
self.assertEqual(user_met_result[0].ON_FOOT, 4)
self.assertEqual(user_met_result[0].BICYCLING, 2)
# Changed from 3 to 4 - investigation at
# https://github.com/e-mission/e-mission-server/issues/288#issuecomment-242531798
self.assertEqual(user_met_result[0].IN_VEHICLE, 4)
# We are not going to make absolute value assertions about
# the aggregate values since they are affected by other
# entries in the database. However, because we have at least
# data for two days in the database, the aggregate data
# must be at least that much larger than the original data.
self.assertEqual(len(agg_met_result), 8)
# no overlap between users at the daily level
# bunch of intermediate entries with no users since this binning works
# by range
self.assertEqual([m.nUsers for m in agg_met_result], [1,1,0,0,0,0,1,1])
# If there are no users, there are no values for any of the fields
# since these are never negative, it implies that their sum is zero
self.assertTrue('ON_FOOT' not in agg_met_result[2] and
'BICYCLING' not in agg_met_result[2] and
'IN_VEHICLE' not in agg_met_result[2])
def testCountTimestampMetrics(self):
met_result = metrics.summarize_by_timestamp(self.testUUID,
self.aug_start_ts,
self.aug_end_ts, 'd',
'count')
logging.debug(met_result)
self.assertEqual(met_result.keys(),
['aggregate_metrics', 'user_metrics'])
user_met_result = met_result['user_metrics']
agg_met_result = met_result['aggregate_metrics']
self.assertEqual(len(user_met_result), 2)
self.assertEqual([m.nUsers for m in user_met_result], [1, 1])
self.assertEqual(user_met_result[0].local_dt.day, 27)
self.assertEqual(user_met_result[1].local_dt.day, 28)
self.assertEqual(user_met_result[0].ON_FOOT, 4)
self.assertEqual(user_met_result[0].BICYCLING, 2)
self.assertEqual(user_met_result[0].IN_VEHICLE, 3)
# We are not going to make absolute value assertions about
# the aggregate values since they are affected by other
# entries in the database. However, because we have at least
# data for two days in the database, the aggregate data
# must be at least that much larger than the original data.
self.assertEqual(len(agg_met_result), 8)
# no overlap between users at the daily level
# bunch of intermediate entries with no users since this binning works
# by range
self.assertEqual([m.nUsers for m in agg_met_result],
[1, 1, 0, 0, 0, 0, 1, 1])
# If there are no users, there are no values for any of the fields
# since these are never negative, it implies that their sum is zero
self.assertTrue('ON_FOOT' not in agg_met_result[2]
and 'BICYCLING' not in agg_met_result[2]
and 'IN_VEHICLE' not in agg_met_result[2])
def is_matched_user(user_id, spec):
metric_list = get_metric_list(spec["checks"])
time_type = spec['time_type']
if 'from_local_date' in spec and 'to_local_date' in spec:
freq_metrics = enam.summarize_by_local_date(user_id,
spec["from_local_date"], spec["to_local_date"],
spec["freq"], metric_list, include_aggregate=False)
elif 'start_time' in spec and 'end_time' in spec:
freq_metrics = enam.summarize_by_timestamp(user_id,
spec["start_time"], spec["end_time"],
spec["freq"], metric_list, include_aggregate=False)
else:
# If no start and end times are specified, we assume that this is a
# timestamp query because we can come up with a reasonable start and end
# time for timestamps but not for local_dates, which are basically a filter.
# so if we run this on the first of a month, for example, we won't find
# anything, which seems bogus and not what people would expect
assert time_type == "timestamp", "time_type = %s, expected timestamp" % time_type
freq_metrics = enam.summarize_by_timestamp(user_id,
0, time.time(), spec["freq"], metric_list, include_aggregate=False)
assert(freq_metrics is not None)
assert('user_metrics' in freq_metrics)
curr_user_metrics = freq_metrics['user_metrics']
checks = spec['checks']
check_results = np.zeros(len(checks))
for i, check in enumerate(checks):
curr_metric_result = curr_user_metrics[i]
# curr_freq_result is a list of ModeStatTimeSummary objects, one for each
# grouped time interval in the range
# e.g. for daily, 2017-01-19, 2017-01-20, 2017-01-21, 2017-01-22, 2017-01-23, ....
for msts in curr_metric_result:
# We defined our check as being true if it is true for _any_ grouped time
# period in the range. So as long as we find a match for that check, we are
# good!
if matches_check(check, msts):
check_results[i] = True
logging.info("For user_id %s, check result array = %s, all? %s" % (user_id, check_results, np.all(check_results)))
return np.all(check_results)
def testCountNoEntries(self):
# Ensure that we don't crash if we don't find any entries
# Should return empty array instead
# Unlike in https://amplab.cs.berkeley.edu/jenkins/job/e-mission-server-prb/591/
met_result_ld = metrics.summarize_by_local_date(self.testUUID,
ecwl.LocalDate({'year': 2000}),
ecwl.LocalDate({'year': 2001}),
'MONTHLY', ['count'], True)
self.assertEqual(list(met_result_ld.keys()), ['aggregate_metrics', 'user_metrics'])
self.assertEqual(met_result_ld['aggregate_metrics'][0], [])
self.assertEqual(met_result_ld['user_metrics'][0], [])
met_result_ts = metrics.summarize_by_timestamp(self.testUUID,
arrow.get(2000,1,1).timestamp,
arrow.get(2001,1,1).timestamp,
'm', ['count'], True)
self.assertEqual(list(met_result_ts.keys()), ['aggregate_metrics', 'user_metrics'])
self.assertEqual(met_result_ts['aggregate_metrics'][0], [])
self.assertEqual(met_result_ts['user_metrics'][0], [])
def testCountNoEntries(self):
# Ensure that we don't crash if we don't find any entries
# Should return empty array instead
# Unlike in https://amplab.cs.berkeley.edu/jenkins/job/e-mission-server-prb/591/
met_result_ld = metrics.summarize_by_local_date(self.testUUID,
ecwl.LocalDate({'year': 2000}),
ecwl.LocalDate({'year': 2001}),
'MONTHLY', ['count'], True)
self.assertEqual(met_result_ld.keys(), ['aggregate_metrics', 'user_metrics'])
self.assertEqual(met_result_ld['aggregate_metrics'][0], [])
self.assertEqual(met_result_ld['user_metrics'][0], [])
met_result_ts = metrics.summarize_by_timestamp(self.testUUID,
arrow.get(2000,1,1).timestamp,
arrow.get(2001,1,1).timestamp,
'm', ['count'], True)
self.assertEqual(met_result_ts.keys(), ['aggregate_metrics', 'user_metrics'])
self.assertEqual(met_result_ts['aggregate_metrics'][0], [])
self.assertEqual(met_result_ts['user_metrics'][0], [])
def testCountTimestampMetrics(self):
met_result = metrics.summarize_by_timestamp(self.testUUID,
self.aug_start_ts,
self.aug_end_ts, 'd',
['count'], True)
import json
import bson.json_util as bju
logging.debug(json.dumps(met_result, default=bju.default))
self.assertEqual(list(met_result.keys()),
['aggregate_metrics', 'user_metrics'])
user_met_result = met_result['user_metrics'][0]
agg_met_result = met_result['aggregate_metrics'][0]
self.assertEqual(len(user_met_result), 2)
self.assertEqual([m.nUsers for m in user_met_result], [1, 1])
self.assertEqual(user_met_result[0].local_dt.day, 27)
self.assertEqual(user_met_result[1].local_dt.day, 28)
self.assertEqual(user_met_result[0].WALKING, 7)
self.assertNotIn("BICYCLING", user_met_result[0])
# Changed from 3 to 4 - investigation at
# https://github.com/e-mission/e-mission-server/issues/288#issuecomment-242531798
self.assertEqual(user_met_result[0].BUS, 4)
# We are not going to make absolute value assertions about
# the aggregate values since they are affected by other
# entries in the database. However, because we have at least
# data for two days in the database, the aggregate data
# must be at least that much larger than the original data.
self.assertEqual(len(agg_met_result), 8)
# no overlap between users at the daily level
# bunch of intermediate entries with no users since this binning works
# by range
self.assertEqual([m.nUsers for m in agg_met_result],
[1, 1, 0, 0, 0, 0, 1, 1])
# If there are no users, there are no values for any of the fields
# since these are never negative, it implies that their sum is zero
self.assertTrue('WALKING' not in agg_met_result[2]
and 'BICYCLING' not in agg_met_result[2]
and 'IN_VEHICLE' not in agg_met_result[2])
