def test_invalid_arguments():
with pytest.raises(ValueError,
match="Argument `freq` must be one of: 'day', 'week'."):
HourInterval(start_hour="00:00", stop_hour="08:00", freq="foobar")
with pytest.raises(
ValueError,
match="If freq='week' then the `weekday` argument must be provided."
):
HourInterval(start_hour="00:00",
stop_hour="08:00",
freq="week",
weekday=None)
with pytest.raises(ValueError, match="Invalid value for `weekday`."):
HourInterval(start_hour="00:00",
stop_hour="08:00",
freq="week",
weekday="foobar")
with pytest.raises(
ValueError,
match=
"If freq='day' then the `weekday` argument must not be provided.",
):
HourInterval(start_hour="00:00",
stop_hour="08:00",
freq="day",
weekday="Monday")
def test_daily_hour_slice_without_stop_hour():
hs = HourInterval(start_hour="07:20", stop_hour=None, freq="day")
assert hs.start_hour == "07:20"
assert hs.stop_hour.is_missing
assert hs.period.freq == "day"
expr = hs.filter_timestamp_column(EventsCallsTable.datetime)
expected = "to_char(events.calls.datetime, 'HH24:MI') >= '07:20'"
assert expected == get_string_representation(expr)
def test_daily_hour_slice():
hs = HourInterval(start_hour="00:00", stop_hour="06:30", freq="day")
assert hs.start_hour == "00:00"
assert hs.stop_hour == "06:30"
assert hs.period.freq == "day"
expr = hs.filter_timestamp_column(EventsCallsTable.datetime)
expected = "to_char(events.calls.datetime, 'HH24:MI') >= '00:00' AND to_char(events.calls.datetime, 'HH24:MI') < '06:30'"
assert expected == get_string_representation(expr)
def test_weekly_hour_slice_without_stop_value():
hs = HourInterval(start_hour="10:00",
stop_hour=None,
freq="week",
weekday="Saturday")
assert hs.start_hour == "10:00"
assert hs.stop_hour.is_missing
assert hs.period.freq == "week"
assert hs.period.weekday == "Saturday"
ts_col = EventsCallsTable.datetime
expr = hs.filter_timestamp_column(ts_col)
expected = ("to_char(events.calls.datetime, 'HH24:MI') >= '10:00' AND "
"EXTRACT(isodow FROM events.calls.datetime) = 6")
assert expected == get_string_representation(expr)
def test_weekly_hour_slice():
hs = HourInterval(start_hour="04:00",
stop_hour="07:45",
freq="week",
weekday="tuesday")
assert hs.start_hour == "04:00"
assert hs.stop_hour == "07:45"
assert hs.period.freq == "week"
assert hs.period.weekday == "Tuesday"
ts_col = EventsCallsTable.datetime
expr = hs.filter_timestamp_column(ts_col)
expected = ("to_char(events.calls.datetime, 'HH24:MI') >= '04:00' AND "
"to_char(events.calls.datetime, 'HH24:MI') < '07:45' AND "
"EXTRACT(isodow FROM events.calls.datetime) = 2")
assert expected == get_string_representation(expr)
def test_multiple_our_slices():
hs1 = HourInterval(start_hour="08:00", stop_hour="16:30", freq="day")
hs2 = HourInterval(start_hour="10:00",
stop_hour="18:45",
freq="week",
weekday="Thursday")
mhs = HourSlice(hour_intervals=[hs1, hs2])
ts_col = EventsCallsTable.datetime
expr = mhs.get_subsetting_condition(ts_col)
expected = ("to_char(events.calls.datetime, 'HH24:MI') >= '08:00' AND "
"to_char(events.calls.datetime, 'HH24:MI') < '16:30' OR "
"to_char(events.calls.datetime, 'HH24:MI') >= '10:00' AND "
"to_char(events.calls.datetime, 'HH24:MI') < '18:45' AND "
"EXTRACT(isodow FROM events.calls.datetime) = 4")
assert expected == get_string_representation(expr)
def __init__(
self,
*,
start=None,
stop=None,
hours="all",
hour_slices=None,
table="events.calls",
subscriber_subset=None,
columns=["*"],
subscriber_identifier="msisdn",
):
if hours != "all" and hour_slices is not None:
raise ValueError(
"The arguments `hours` and `hour_slice` are mutually exclusive."
)
if hours != "all":
assert (isinstance(hours, tuple) and len(hours) == 2
and isinstance(hours[0], int)
and isinstance(hours[1], int)) # sanity check
start_hour = hours[0]
stop_hour = hours[1]
start_hour_str = f"{start_hour:02d}:00"
stop_hour_str = f"{stop_hour:02d}:00"
if start_hour <= stop_hour:
hs = HourInterval(start_hour=start_hour_str,
stop_hour=stop_hour_str,
freq="day")
self.hour_slices = HourSlice(hour_intervals=[hs])
else:
# If hours are backwards, then this is interpreted as spanning midnight,
# so we split it into two time slices for the beginning/end of the day.
hs1 = HourInterval(start_hour=None,
stop_hour=stop_hour_str,
freq="day")
hs2 = HourInterval(start_hour=start_hour_str,
stop_hour=None,
freq="day")
self.hour_slices = HourSlice(hour_intervals=[hs1, hs2])
else:
self.hour_slices = HourSlice(hour_intervals=[])
self.start = standardise_date(start)
self.stop = standardise_date(stop)
self.hours = hours
self.subscriber_subsetter = make_subscriber_subsetter(
subscriber_subset)
self.subscriber_identifier = subscriber_identifier.lower()
if columns == ["*"]:
self.table_ORIG = Table(table)
columns = self.table_ORIG.column_names
else:
self.table_ORIG = Table(table, columns=columns)
self.columns = set(columns)
try:
self.columns.remove(subscriber_identifier)
self.columns.add(f"{subscriber_identifier} AS subscriber")
except KeyError:
if self.subscriber_subsetter.is_proper_subset:
warnings.warn(
f"No subscriber column requested, did you mean to include {subscriber_identifier} in columns? "
"Since you passed a subscriber_subset the data will still be subset by your subscriber subset, "
"but the subscriber column will not be present in the output.",
stacklevel=2,
)
self.columns = sorted(self.columns)
self.sqlalchemy_table = get_sqlalchemy_table_definition(
self.table_ORIG.fully_qualified_table_name,
engine=get_db().engine,
)
if self.start == self.stop:
raise ValueError("Start and stop are the same.")
super().__init__()
# This needs to happen after the parent classes init method has been
# called as it relies upon the connection object existing
self._check_dates()