def get(self, request):
user = request.user
sleep_activities = SleepLog.objects.filter(user=user)
serializer = SleepActivityDataframeBuilder(sleep_activities)
sleep_aggregate = serializer.get_sleep_history_series()
# because pandas uses a timeindex, when we go to json - it doesn't
# play nicely with a typical json dump, so we do an additional load so drf can transmit nicely
result = sleep_aggregate.to_json(date_format='iso')
result = json.loads(result)
return Response(data=result)
def get(self, request):
user = request.user
sleep_activities = SleepLog.objects.filter(user=user)
serializer = SleepActivityDataframeBuilder(sleep_activities)
sleep_aggregate = serializer.get_sleep_history_series()
# because pandas uses a timeindex, when we go to json - it doesn't
# play nicely with a typical json dump, so we do an additional load so drf can transmit nicely
result = sleep_aggregate.to_json(date_format='iso')
result = json.loads(result)
return Response(data=result)
def _get_sleep_series_last_year(user):
"""
:param user:
:return: Series data of how much sleep that person has gotten minutes
"""
start_date = get_current_date_years_ago(1)
sleep_events = SleepLog.objects.filter(user=user, start_time__date__gte=start_date)
builder = SleepActivityDataframeBuilder(sleep_events)
series = builder.get_sleep_history_series()
# anytime sleep is actually set at zero, the value should be NaN
series[series == 0] = np.NaN
return series
def get(self, request):
try:
window = int(request.query_params[LOOKBACK_PARAM_NAME])
except MultiValueDictKeyError:
# MultiValueDictKeyError happens when a key doesn't exist
window = 1
except ValueError:
# ValueError if something entered for a window that couldn't be interpreted
return Response(status=400)
user = request.user
sleep_activities = SleepLog.objects.filter(user=user)
builder = SleepActivityDataframeBuilder(sleep_activities)
sleep_aggregate = builder.get_sleep_history_series()
sleep_average = sleep_aggregate.rolling(window=window, min_periods=1).mean()
result = sleep_average.to_json(date_format='iso')
result = json.loads(result)
return Response(data=result)
def get(self, request):
try:
window = int(request.query_params[LOOKBACK_PARAM_NAME])
except MultiValueDictKeyError:
# MultiValueDictKeyError happens when a key doesn't exist
window = 1
except ValueError:
# ValueError if something entered for a window that couldn't be interpreted
return Response(status=400)
user = request.user
sleep_activities = SleepLog.objects.filter(user=user)
builder = SleepActivityDataframeBuilder(sleep_activities)
sleep_aggregate = builder.get_sleep_history_series()
sleep_average = sleep_aggregate.rolling(window=window,
min_periods=1).mean()
result = sleep_average.to_json(date_format='iso')
result = json.loads(result)
return Response(data=result)
def get(self, request, supplement_uuid):
# TODO - Refactor this garbage, you can add some smart redis caching level to this
supplement = get_object_or_404(Supplement,
uuid=supplement_uuid,
user=request.user)
user = request.user
serializer = SupplementLogRequestParametersSerializer(
data=request.query_params)
serializer.is_valid(raise_exception=True)
params = serializer.validated_data
start_date = params['start_date']
end_date = get_current_userdate(user)
supplement_events = SupplementLog.objects.filter(
user=user,
supplement=supplement,
time__date__gte=start_date,
time__date__lte=end_date)
# no point if nothing exists
if not supplement_events.exists():
return Response([])
# lots of crappy templating here, sorry.
supplement_builder = SupplementEventsDataframeBuilder(
supplement_events)
# TODO - Really feels like you should build a helper on the builder to do this since you do it so often
supplement_series = supplement_builder.build_dataframe(
)['Quantity'].sort_index()
# because the dataframe will also get things like "source" etc, and we only care about
# quantity, take that series and then recast it as a numeric
supplement_series = pd.to_numeric(supplement_series)
productivity_logs = DailyProductivityLog.objects.filter(
user=user, date__gte=start_date, date__lte=end_date)
productivity_builder = ProductivityLogEventsDataframeBuilder(
productivity_logs)
productivity_series = productivity_builder.get_productive_timeseries()
sleep_logs = SleepLog.objects.filter(user=user,
start_time__date__gte=start_date)
sleep_builder = SleepActivityDataframeBuilder(sleep_logs, user)
sleep_series = sleep_builder.get_sleep_history_series()
dataframe_details = {
'sleep_time': sleep_series,
'productivity_time': productivity_series,
'quantity': supplement_series,
}
dataframe = pd.DataFrame(dataframe_details)
# don't really need to convert it to local, just makes debugging make easier
dataframe_localized = dataframe.tz_convert(user.pytz_timezone)
"""
because events are datetime based, but productivity and sleep are date-based
this parts get a little hairy, but we want the nans for 8/30 and 9/01 to be filled
however, we cant just pad fill because if a log for productivity and sleep was missing
the wrong result would be filled. so ... the code below is slightly magical
productivity_time sleep_time quantity
2017-08-30 00:00:00-04:00 1336.0 647.013778 0.0
2017-08-30 19:51:36.483443-04:00 NaN NaN 1.0
2017-08-31 00:00:00-04:00 1476.0 726.132314 0.0
2017-09-01 00:00:00-04:00 730.0 513.894938 0.0
2017-09-01 14:51:36.483443-04:00 NaN NaN 1.0
"""
if not params['frequency']:
dataframe_localized_date_index = dataframe_localized.index.date
dataframe_localized_date_index = pd.DatetimeIndex(
dataframe_localized_date_index, tz=request.user.pytz_timezone)
productivity_series = dataframe_localized[
'productivity_time'].dropna()
productivity_series_filled = productivity_series[
dataframe_localized_date_index]
sleep_series = dataframe_localized['sleep_time'].dropna()
sleep_series_filled = sleep_series[dataframe_localized_date_index]
dataframe_localized[
'productivity_time'] = productivity_series_filled.values
dataframe_localized['sleep_time'] = sleep_series_filled.values
valid_supplement_index = dataframe_localized['quantity'].dropna(
).index
dataframe_localized = dataframe_localized.ix[
valid_supplement_index]
elif params['frequency'] == DAILY_FREQUENCY:
dataframe_localized = dataframe_localized.resample('D').sum()
elif params['frequency'] == MONTHLY_FREQUENCY:
dataframe_localized = dataframe_localized.resample('M').sum()
dataframe_localized = update_dataframe_to_be_none_instead_of_nan_for_api_responses(
dataframe_localized)
results = []
for index, values in dataframe_localized.iterrows():
time = index.isoformat()
result = values.to_dict()
result['time'] = time
result['uniqueKey'] = '{}-{}'.format(time, result['quantity'])
results.append(result)
return Response(results)
def get_sleep_activity_series(queryset):
builder = SleepActivityDataframeBuilder(queryset)
series = builder.get_sleep_history_series()
return series
def get(self, request, supplement_uuid):
# TODO - Refactor this garbage, you can add some smart redis caching level to this
supplement = get_object_or_404(Supplement, uuid=supplement_uuid, user=request.user)
user = request.user
serializer = SupplementLogRequestParametersSerializer(data=request.query_params)
serializer.is_valid(raise_exception=True)
params = serializer.validated_data
start_date = params['start_date']
end_date = get_current_userdate(user)
supplement_events = SupplementLog.objects.filter(
user=user, supplement=supplement, time__date__gte=start_date, time__date__lte=end_date)
# no point if nothing exists
if not supplement_events.exists():
return Response([])
# lots of crappy templating here, sorry.
supplement_builder = SupplementEventsDataframeBuilder(supplement_events)
# TODO - Really feels like you should build a helper on the builder to do this since you do it so often
supplement_series = supplement_builder.build_dataframe()['Quantity'].sort_index()
# because the dataframe will also get things like "source" etc, and we only care about
# quantity, take that series and then recast it as a numeric
supplement_series = pd.to_numeric(supplement_series)
productivity_logs = DailyProductivityLog.objects.filter(
user=user, date__gte=start_date, date__lte=end_date)
productivity_builder = ProductivityLogEventsDataframeBuilder(productivity_logs)
productivity_series = productivity_builder.get_productive_timeseries()
sleep_logs = SleepLog.objects.filter(user=user, start_time__date__gte=start_date)
sleep_builder = SleepActivityDataframeBuilder(sleep_logs, user)
sleep_series = sleep_builder.get_sleep_history_series()
dataframe_details = {
'sleep_time': sleep_series,
'productivity_time': productivity_series,
'quantity': supplement_series,
}
dataframe = pd.DataFrame(dataframe_details)
# don't really need to convert it to local, just makes debugging make easier
dataframe_localized = dataframe.tz_convert(user.pytz_timezone)
"""
because events are datetime based, but productivity and sleep are date-based
this parts get a little hairy, but we want the nans for 8/30 and 9/01 to be filled
however, we cant just pad fill because if a log for productivity and sleep was missing
the wrong result would be filled. so ... the code below is slightly magical
productivity_time sleep_time quantity
2017-08-30 00:00:00-04:00 1336.0 647.013778 0.0
2017-08-30 19:51:36.483443-04:00 NaN NaN 1.0
2017-08-31 00:00:00-04:00 1476.0 726.132314 0.0
2017-09-01 00:00:00-04:00 730.0 513.894938 0.0
2017-09-01 14:51:36.483443-04:00 NaN NaN 1.0
"""
if not params['frequency']:
dataframe_localized_date_index = dataframe_localized.index.date
dataframe_localized_date_index = pd.DatetimeIndex(dataframe_localized_date_index,
tz=request.user.pytz_timezone)
productivity_series = dataframe_localized['productivity_time'].dropna()
productivity_series_filled = productivity_series[dataframe_localized_date_index]
sleep_series = dataframe_localized['sleep_time'].dropna()
sleep_series_filled = sleep_series[dataframe_localized_date_index]
dataframe_localized['productivity_time'] = productivity_series_filled.values
dataframe_localized['sleep_time'] = sleep_series_filled.values
valid_supplement_index = dataframe_localized['quantity'].dropna().index
dataframe_localized = dataframe_localized.ix[valid_supplement_index]
elif params['frequency'] == DAILY_FREQUENCY:
dataframe_localized = dataframe_localized.resample('D').sum()
elif params['frequency'] == MONTHLY_FREQUENCY:
dataframe_localized = dataframe_localized.resample('M').sum()
dataframe_localized = update_dataframe_to_be_none_instead_of_nan_for_api_responses(dataframe_localized)
results = []
for index, values in dataframe_localized.iterrows():
time = index.isoformat()
result = values.to_dict()
result['time'] = time
result['uniqueKey'] = '{}-{}'.format(time, result['quantity'])
results.append(result)
return Response(results)