def test_extract_float_from_category(self):
table = pd.DataFrame({"A": ["1", "2.1", "note: 3.2", "-3", "-3.0"]},
dtype="category")
form = Form(["A"], True, InputNumberType.FLOAT, error_means_null=True)
result = form.convert_table(table)
assert_frame_equal(result,
pd.DataFrame({"A": [np.nan, 2.1, 3.2, np.nan, -3]}))
def test_extract_float_eu(self):
table = pd.DataFrame({"A": ["1,234", "2,345.67", "3.456"]})
form = Form(["A"],
True,
InputNumberType.FLOAT,
InputLocale.EU,
error_means_null=True)
result = form.convert_table(table)
assert_frame_equal(result, pd.DataFrame({"A": [1.234, 2.345, np.nan]}))
def test_error_on_no_match(self):
table = pd.DataFrame({"A": ["", ".", np.nan, "1", "2.1"]})
form = Form(["A"], True, InputNumberType.INTEGER, InputLocale.US)
result = form.convert_table(table)
self.assertEqual(
result,
("'' in row 1 of 'A' cannot be converted. Overall, there are 2 "
"errors in 1 column. Select 'Convert non-numbers to null' to set "
"these values to null."),
)
def test_match_eu_thousands_must_be_in_groups_of_3(self):
form = Form(["A"],
False,
InputNumberType.ANY,
InputLocale.EU,
error_means_null=True)
result = form.convert_table(
pd.DataFrame({"A": ["123.4", "2.345.1", "3.23,123"]}))
assert_frame_equal(result,
pd.DataFrame({"A": [np.nan, np.nan, np.nan]}))
def test_ignore_numbers(self):
table = pd.DataFrame({"A": [1, 2]})
result = Form(["A"]).convert_table(table)
assert_frame_equal(result, pd.DataFrame({"A": [1, 2]}))
def test_replace_with_null(self):
table = pd.DataFrame({"A": ["", ".", np.nan, "1", "2.1"]})
form = Form(["A"], True, InputNumberType.INTEGER, InputLocale.US, True)
result = form.convert_table(table)
assert_frame_equal(result,
pd.DataFrame({"A": [np.nan, np.nan, np.nan, 1, 2]}))
def test_extract_integer_eu(self):
table = pd.DataFrame({"A": ["1,234", "2,345.67", "3.456"]})
form = Form(["A"], True, InputNumberType.INTEGER, InputLocale.EU)
result = form.convert_table(table)
assert_frame_equal(result, pd.DataFrame({"A": [1, 2, 3456]}))
def test_extract_integer_no_separator(self):
table = pd.DataFrame({"A": ["10000", "20001"]})
result = Form(["A"], True,
InputNumberType.INTEGER).convert_table(table)
assert_frame_equal(result, pd.DataFrame({"A": [10000, 20001]}))
def test_extract_integer_from_category(self):
table = pd.DataFrame({"A": ["1", "2.1", "note: 3.2"]},
dtype="category")
form = Form(["A"], True, InputNumberType.INTEGER)
result = form.convert_table(table)
assert_frame_equal(result, pd.DataFrame({"A": [1, 2, 3]}))
def test_extract_any_eu_thousands_must_be_in_groups_of_3(self):
form = Form(["A"], True, InputNumberType.ANY, InputLocale.EU, True)
result = form.convert_table(
pd.DataFrame({"A": ["123.4", "2.345.1", "3.23,123"]}))
assert_frame_equal(result, pd.DataFrame({"A": [123, 2345, 3]}))
def test_extract_any_many_commas(self):
result = Form(["A"], True, InputNumberType.ANY,
InputLocale.US).convert_table(
pd.DataFrame({"A": ["1,234,567,890"]}))
assert_frame_equal(result, pd.DataFrame({"A": [1234567890]}))
def test_extract_any_us(self):
table = pd.DataFrame({"A": ["1,234", "2,345.67", "3.456"]})
form = Form(["A"], True, InputNumberType.ANY, InputLocale.US)
result = form.convert_table(table)
assert_frame_equal(result, pd.DataFrame({"A": [1234, 2345.67, 3.456]}))
def test_extract_any_from_category(self):
table = pd.DataFrame({"A": ["1", "2.1", "note: 3.2"]},
dtype="category")
form = Form(["A"], True, InputNumberType.ANY)
result = form.convert_table(table)
assert_frame_equal(result, pd.DataFrame({"A": [1.0, 2.1, 3.2]}))
def test_extract_any_from_str(self):
table = pd.DataFrame({"A": ["1", "2.1", "note: 3.2", "-3.1"]})
form = Form(["A"], True, InputNumberType.ANY, error_means_null=True)
result = form.convert_table(table)
assert_frame_equal(result, pd.DataFrame({"A": [1.0, 2.1, 3.2, -3.1]}))
def test_match_unicode_minus(self):
table = pd.DataFrame({"A": ["-1", "\u22122"]})
result = Form(["A"]).convert_table(table)
assert_frame_equal(result, pd.DataFrame({"A": [-1, -2]}))