Sometimes we need to create a special test function to help us solve a specific issue over and over. In this case, we’re going to make a special test helper that checks if the first row of a pandas dataframe is equal to one we specify with the caveat that the row we are comparing to can have more columns or the columns in a different order than our truth values. This is helpful when the returned data may have more columns (i.e. a certain station may have a radiation sensor and another may not).
testing.py in the meteogram directory.from numpy.testing import assert_almost_equal
import pandas as pd
def assert_dataseries_equal(actual, desired):
"""
Intelligently compare dataseries (i.e. dictionary) for testing.
Only compares the keys that are in desired, i.e. there can be extra keys
in actual and the test will pass. Handles comparison of floating points
in a more intelligent way and does not require fields to be lined up in
the same order.
"""
for key in desired.keys():
# If this is a float, then check accordingly
if 'float' in str(type(desired.get(key))):
assert_almost_equal(actual.get(key), desired.get(key))
else:
assert actual.get(key) == desired.get(key)
Now, let’s write a test using that helper.
from meteogram.testing import assert_dataseries_equal
@patch('meteogram.meteogram.current_utc_time', new=mocked_current_utc_time)
def test_download_asos_data():
"""Test downloading ASOS data."""
# Setup
url = meteogram.build_asos_request_url('AMW')
# Exercise
df = meteogram.download_asos_data(url)
# Verify
first_row_truth = pd.Series(
{'station_id': 'AMW',
'station_name': 'Ames',
'latitude_deg': 41.990439,
'longitude_deg': -93.618515,
'UTC': pd.Timestamp('2018-03-25 12:00:00'),
'temperature_degF': 29,
'dewpoint_degF': 24,
'wind_speed_knots': 8,
'wind_direction_degrees': 113})
assert_dataseries_equal(df.iloc[0], first_row_truth)
# Cleanup - none necessary