rowsandall/rowers/alerts.py

from rowers.models import Alert, Condition, User, Rower, Workout
from rowers.teams import coach_getcoachees
from rowers.dataprep import getsmallrowdata_db, getrowdata_db
import datetime
import numpy as np
import math

def create_alert(manager, rower, measured, period=7, emailalert=True,
                 reststrokes=False, workouttype='water', boattype='1x',
                 name='', **kwargs):

    # check if manager is coach of rower. If not return 0
    if manager.rower != rower:  # pragma: no cover
        if rower not in coach_getcoachees(manager.rower):
            return 0, 'You are not allowed to create this alert'

    m = Condition(
        metric=measured['metric'],
        value1=measured['value1'],
        value2=measured['value2'],
        condition=measured['condition']
    )

    m.save()

    alert = Alert(name=name,
                  manager=manager,
                  rower=rower,
                  measured=m,
                  reststrokes=reststrokes,
                  period=period,
                  emailalert=emailalert,
                  workouttype=workouttype,
                  boattype=boattype,
                  )

    alert.save()

    if 'filter' in kwargs:
        filters = kwargs['filter']
        for f in filters:
            if f['metric'] and f['condition']:
                m = Condition(
                    metric=f['metric'],
                    value1=f['value1'],
                    value2=f['value2'],
                    condition=f['condition']
                )

                m.save()

                alert.filter.add(m)

    return alert.id, 'Your alert was created'


# update alert
def alert_add_filters(alert, filters):
    for f in alert.filter.all():
        alert.filter.remove(f)
        f.delete()

    for f in filters:
        metric = f['metric']
        value1 = f['value1']
        condition = f['condition']

        if condition and metric and value1:
            m = Condition(
                metric=f['metric'],
                value1=f['value1'],
                value2=f['value2'],
                condition=f['condition']
            )

            m.save()

            alert.filter.add(m)

    return 1

# get alert stats
# nperiod = 0: current period, i.e. next_run - n days to today
# nperiod = 1: 1 period ago , i.e. next_run -2n days to next_run -n days


def alert_get_stats(alert, nperiod=0):  # pragma: no cover
    # get strokes
    workstrokesonly = not alert.reststrokes
    startdate = (alert.next_run -
                 datetime.timedelta(days=(nperiod+1)*alert.period-1))
    enddate = alert.next_run - datetime.timedelta(days=(nperiod)*alert.period)
    columns = [alert.measured.metric]

    for condition in alert.filter.all():
        columns.append(condition.metric)

    workouts = Workout.objects.filter(date__gte=startdate, date__lte=enddate, user=alert.rower,
                                      workouttype=alert.workouttype, duplicate=False,
                                      boattype=alert.boattype)
    ids = [w.id for w in workouts]

    try:
        df = getsmallrowdata_db(columns, ids=ids, doclean=True,
                                workstrokesonly=workstrokesonly)
    except:
        return {
            'workouts': workouts.count(),
            'startdate': startdate,
            'enddate': enddate,
            'nr_strokes': 0,
            'nr_strokes_qualifying': 0,
            'percentage': 0,
            'nperiod': nperiod,
            'median': 0,
            'median_q': 0,
            'standard_dev': 0,
        }

    if df.empty:
        return {
            'workouts': workouts.count(),
            'startdate': startdate,
            'enddate': enddate,
            'nr_strokes': 0,
            'nr_strokes_qualifying': 0,
            'percentage': 0,
            'nperiod': nperiod,
            'median': 0,
            'median_q': 0,
            'standard_dev': 0,
        }

    # check if filters are in columns list
    pdcolumns = set(df.columns)  # pragma: no cover

    # drop strokes through filter
    if set(columns) <= pdcolumns:  # pragma: no cover
        for condition in alert.filter.all():
            if condition.condition == '>':
                mask = df[condition.metric] > condition.value1
                df.loc[mask, alert.measured.metric] = np.nan
            elif condition.condition == '<':
                mask = df[condition.metric] < condition.value1
                df.loc[mask, alert.measured.metric] = np.nan
            elif condition.condition == 'between':
                mask = df[condition.metric] > condition.value1
                mask2 = df[condition.metric] < condition.value2
                df.loc[mask & mask2, alert.measured.metric] = np.nan
            elif condition.condition == '=':
                mask = df[condition.metric] == condition.value1
                df.loc[mask, alert.measured.metric] = np.nan

        df.dropna(inplace=True, axis=0)
    else:  # pragma: no cover
        return {
            'workouts': workouts.count(),
            'startdate': startdate,
            'enddate': enddate,
            'nr_strokes': 0,
            'nr_strokes_qualifying': 0,
            'percentage': 0,
            'nperiod': nperiod,
            'median': 0,
            'median_q': 0,
            'standard_dev': 0,
        }

    # count strokes
    nr_strokes = len(df)

    # count qualifying
    if alert.measured.condition == '>':
        mask = df[alert.measured.metric] > alert.measured.value1
        df2 = df[mask].copy()
    elif alert.measured.condition == '<':
        mask = df[alert.measured.metric] < alert.measured.value1
        df2 = df[mask].copy()
    elif alert.measured.condition == 'between':
        mask = df[alert.measured.metric] > alert.measured.value1
        mask2 = df[alert.measured.metric] < alert.measured.value2
        df2 = df[mask & mask2].copy()
    else:
        mask = df[alert.measured.metric] == alert.measured.value1
        df2 = df[mask].copy()

    nr_strokes_qualifying = len(df2)

    if nr_strokes > 0:
        percentage = int(100.*nr_strokes_qualifying/nr_strokes)
    else:
        percentage = 0

    median_q = df2[alert.measured.metric].median()
    median = df[alert.measured.metric].median()
    std = df[alert.measured.metric].std()

    data = {
        'workouts': workouts.count(),
        'startdate': startdate,
        'enddate': enddate,
        'nr_strokes': nr_strokes,
        'nr_strokes_qualifying': nr_strokes_qualifying,
        'percentage': percentage,
        'nperiod': nperiod,
        'median': median,
        'median_q': median_q,
        'standard_dev': std,
    }

    data_clean = {}

    for k in data:
        data_clean[k] = data[k]
        try:
            if math.isnan(data[k]):
                data_clean[k] = 0
        except TypeError:
            pass

    return data_clean

# run alert report

# check alert permission
from django.utils import timezone

def checkalertowner(alert, user):
    if alert.manager == user:
        return True
    if alert.rower.user == user:  # pragma: no cover
        return True
    coaches = alert.manager.rower.get_coaches()
    for coach in coaches:
        if coach.rowerplan == 'coach':
            return True
        if coach.coachtrialexpires >= timezone.now().date():
            return True
    return False  # pragma: no cover