def test_LocationsGeocoder_test_real_geocoder(self, tmp_path):
"""Tests the .run() and associated methods with a real geocoder
"""
p = tmp_path/"test.csv"
gl = gc.LocationsGeocoder(self.address_df, valid_email, p,
batch_size = 1)
gl.run(num_batches=3)
assert all(gl.locations.all())
def test_LocationsGeocoder_make_file_header(self, tmp_path):
"""Tests the ._make_file() and ._make_header() methods
for LocationsGeocoder
"""
p = tmp_path/"test.csv"
gl = gc.LocationsGeocoder(self.address_df, valid_email, p)
test = pd.read_csv(p, sep = "\t")
error_msg = "Saved File has improper header."
assert "latitude" in test.columns, error_msg
assert "longitude" in test.columns, error_msg
assert "address" in test.columns, error_msg
assert len(test.columns) <= 4, error_msg
def test_LocationsGeocoder_reset(self, tmp_path):
"""
"""
#Mocks the input asking for DELETE
gc.input = lambda *args : "DELETE"
p = tmp_path/"test.csv"
gl = gc.LocationsGeocoder(self.address_df, valid_email, p)
gl._set_dummy_geocoder()
gl.run()
gl.reset()
test = pd.read_csv(p, sep = "\t")
assert test.empty, "File was not deleted"
assert gl.curr_batch == 1, "Batches were not reset"
assert len(gl._queue) == 234, "Queue was not reset"
def test_LocationsGeocoder_run_one_batch(self, tmp_path):
"""Tests the .run() and associated methods for
LocationsGeocoder.
"""
p = tmp_path/"test.csv"
gl = gc.LocationsGeocoder(self.address_df, valid_email, p)
gl._set_dummy_geocoder()
gl.run()
test = pd.read_csv(p, sep = "\t")
assert not test.empty, "No results saved to disk."
assert not gl.locations.empty, "No results stored in object."
assert len(test) == 234, "Not all lines were saved to disk"
assert len(gl.locations) == 234, "Not all lines were saved in object"
assert len(test.columns) <= 4, "Saved too many columns in tsv format"
def test_LocationsGeocoder_address_dict(self, tmp_path):
"""Tests the .run() and associated methods when constructing df from
list of dicts.
"""
p = tmp_path/"test.csv"
gl = gc.LocationsGeocoder(self.address_dict, valid_email, p,
batch_size = 100)
gl._set_dummy_geocoder()
gl.run()
test = pd.read_csv(p, sep = "\t")
assert gl.tot_batches == 3, "Batches are not constructed properly."
assert not test.empty, "No results saved to disk."
assert not gl.locations.empty, "No results stored in object."
assert len(test) == 234, "Not all lines were saved to disk."
assert len(gl.locations) == 234, "Not all lines were saved in object."
def test_LocationsGeocoder_run_less_than_all_batches(self, tmp_path):
"""Tests the .run() and associated methods for LocationsGeocoder when
running less than all batches.
"""
p = tmp_path/"test.csv"
gl = gc.LocationsGeocoder(self.address_df, valid_email, p,
batch_size = 100)
gl._set_dummy_geocoder()
gl.run(num_batches = 1)
test = pd.read_csv(p)
assert gl.tot_batches == 3, "Batches are not constructed properly."
assert gl.curr_batch < 3, "Run running too many batches."
assert not test.empty, "No results saved to disk."
assert not gl.locations.empty, "No results stored in object."
assert len(test) < 234, "Not all lines were saved to disk."
assert len(gl.locations) < 234, "Not all lines were saved in object."