def test_daily_dataframe_verification_raises_error_for_dates_with_repeating_timeslots(self):
test_df = pd.DataFrame(
data=[
{'v': 1},
{'v': 1},
{'v': 1},
{'v': 1},
{'v': 1},
{'v': 1},
{'v': 1},
],
index=[
dt.datetime(2020,1,1,0,0),
dt.datetime(2020,1,1,1,0),
dt.datetime(2020,1,2,0,0),
dt.datetime(2020,1,2,1,0),
dt.datetime(2020,1,2,1,0),
dt.datetime(2020,1,3,0,0),
dt.datetime(2020,1,3,1,0),
]
)
with pytest.raises(InvalidTimeSlots) as e:
daily_dataframe.verify_data_integrity(test_df, [dt.time(0,0), dt.time(1,0)])
error_message = str(e.value)
assert 'datetime.date(2020, 1, 2)' in error_message
def test_daily_dataframe_verification_raises_error_for_real_dataset_with_local_DST_included_time(self):
test_df = pd.read_hdf('tests/test_data/manually_processed_dataframes', key='raw_df')
test_df.set_index(test_df.index.tz_localize(None), inplace=True)
time_slots = [dt.time(i,0) for i in range(24)]
with pytest.raises(InvalidTimeSlots) as e:
daily_dataframe.verify_data_integrity(test_df, time_slots)
error_message = str(e.value)
assert 'datetime.date(2015, 3, 29)' in error_message
assert 'datetime.date(2015, 10, 25)' in error_message
assert 'datetime.date(2016, 3, 27)' in error_message
assert 'datetime.date(2016, 10, 30)' in error_message
assert 'datetime.date(2017, 3, 26)' in error_message
assert 'datetime.date(2017, 10, 29)' in error_message
assert 'datetime.date(2018, 3, 25)' in error_message
assert 'datetime.date(2018, 10, 28)' in error_message
assert 'datetime.date(2019, 3, 31)' in error_message
assert 'datetime.date(2019, 10, 27)' in error_message
def test_daily_dataframe_raises_error_for_non_contiguous_dates(self):
test_df = pd.DataFrame(
data=[
{'v': 1},
{'v': 1},
{'v': 1},
{'v': 1},
],
index=[
dt.datetime(2020,1,1,0,0),
dt.datetime(2020,1,1,2,0),
dt.datetime(2020,1,3,0,0),
dt.datetime(2020,1,3,2,0),
]
)
with pytest.raises(MoreThanOneContiguousDatapoint) as e:
daily_dataframe.verify_data_integrity(test_df, [dt.time(0,0), dt.time(2,0)])
error_message = str(e.value)
assert '(datetime.date(2020, 1, 1), datetime.date(2020, 1, 1))' in error_message
assert '(datetime.date(2020, 1, 3), datetime.date(2020, 1, 3))' in error_message
def test_daily_dataframe_verification_works_for_correct_dataframe(self):
test_df = pd.DataFrame(
data=[
{'v': 1},
{'v': 1},
{'v': 1},
{'v': 1},
{'v': 1},
{'v': 1},
],
index=[
dt.datetime(2020,1,1,0,0),
dt.datetime(2020,1,1,1,0),
dt.datetime(2020,1,2,0,0),
dt.datetime(2020,1,2,1,0),
dt.datetime(2020,1,3,0,0),
dt.datetime(2020,1,3,1,0),
]
)
assert daily_dataframe.verify_data_integrity(test_df, [dt.time(0,0), dt.time(1,0)]) == (
dt.date(2020,1,1),
dt.date(2020,1,3),
)