def in_past_week(result, now, ordered_before=None):
sent_collection_time_str = result[_SENT_COLLECTION_TIME_INDEX]
received_time_str = result[_RECEIVED_TIME_INDEX]
max_time = None
if sent_collection_time_str:
max_time = parse_datetime(sent_collection_time_str)
if ordered_before and max_time > ordered_before:
return False
if received_time_str:
received_time = parse_datetime(received_time_str)
if received_time and max_time:
max_time = max(received_time, max_time)
else:
max_time = received_time
if max_time:
return (now - max_time).days <= 7
return False
def reduce_participant_data_to_hpo_metric_date_deltas(reducer_key, reducer_values, now=None):
"""Input:
reducer_key - participant ID
reducer_values - strings of the form date|metric, or DOB|date_of_birth.
Sorts everything by date, and emits hpoId|participant_type|metric|date|delta strings representing
increments or decrements of metrics based on this participant.
"""
#pylint: disable=unused-argument
metrics_conf = get_config()
metric_fields = get_fieldnames()
summary_fields = metrics_conf['summary_fields']
last_state = {}
last_hpo_id = None
dates_and_metrics = []
date_of_birth = None
for reducer_value in reducer_values:
t = parse_tuple(reducer_value)
if t[0] == DATE_OF_BIRTH_PREFIX:
date_of_birth = datetime.strptime(t[1], DATE_FORMAT).date()
else:
dates_and_metrics.append((parse_datetime(t[0]), t[1]))
if not dates_and_metrics:
return
# Sort the dates and metrics, date first then metric.
dates_and_metrics = sorted(dates_and_metrics)
initial_state = {f.name: UNSET for f in metrics_conf['fields']}
initial_state[TOTAL_SENTINEL] = 1
last_hpo_id = UNSET
# Look for the starting HPO, update the initial state with it, and remove it from
# the list of date-and-metrics pairs.
for i in range(0, len(dates_and_metrics)):
metric = dates_and_metrics[i][1]
metric_name, value = parse_metric(metric)
if metric_name == HPO_ID_METRIC:
last_hpo_id = value
initial_state[HPO_ID_METRIC] = last_hpo_id
break
# If we know the participant's date of birth, and a starting age range
# and entries for when it changes over time.
if date_of_birth:
initial_state[AGE_RANGE_METRIC] = _add_age_range_metrics(dates_and_metrics, date_of_birth, now)
# Re-sort with the new entries for age range changes.
dates_and_metrics = sorted(dates_and_metrics)
# Run summary functions on the initial state.
_update_summary_fields(summary_fields, initial_state)
# Emit 1 values for the initial state before any metrics change.
initial_date = dates_and_metrics[0][0]
for k, v in initial_state.iteritems():
yield reduce_result_value(map_result_key(last_hpo_id, _REGISTERED_PARTICIPANT, k, v),
initial_date.date().isoformat(), '1')
last_state = initial_state
full_participant = False
# Loop through all the metric changes for the participant.
for dt, metric in dates_and_metrics:
date = dt.date()
new_state = copy.deepcopy(last_state)
if not _process_metric(metric_fields, summary_fields, metric, new_state):
continue # No changes so there's nothing to do.
hpo_id = new_state.get(HPO_ID_METRIC)
hpo_change = last_hpo_id != hpo_id
last_full_participant = full_participant
for k, v in new_state.iteritems():
# Output a delta for this field if it is either the first value we have,
# or if it has changed. In the case that one of the facets has changed,
# we need deltas for all fields.
old_val = last_state and last_state.get(k, None)
if hpo_change or v != old_val:
formatted_date = date.isoformat()
if (k == ENROLLMENT_STATUS_METRIC and v == EnrollmentStatus.FULL_PARTICIPANT and
not full_participant):
full_participant = True
# Emit 1 values for the current state for all fields for the full participant type.
for k2, v2 in new_state.iteritems():
yield reduce_result_value(map_result_key(hpo_id, _FULL_PARTICIPANT, k2, v2),
formatted_date, '1')
yield reduce_result_value(map_result_key(hpo_id, _REGISTERED_PARTICIPANT, k, v),
formatted_date, '1')
if last_full_participant:
yield reduce_result_value(map_result_key(hpo_id, _FULL_PARTICIPANT, k, v), formatted_date,
'1')
if last_state:
# If the value changed, output -1 delta for the old value.
yield reduce_result_value(map_result_key(last_hpo_id, _REGISTERED_PARTICIPANT,
k, old_val),
formatted_date, '-1')
if last_full_participant:
yield reduce_result_value(map_result_key(last_hpo_id, _FULL_PARTICIPANT, k, old_val),
formatted_date, '-1')
last_state = new_state
last_hpo_id = hpo_id