def test_validate_seed_lines_too_many_fields(test_data):
batches = DataFrameBatch(df=test_data, config=config_template, batch_size=3)
with pytest.raises(RuntimeError) as err:
_validate_batch_seed_values(
batches.batches[0],
{"ID_Code": "foo", "target": 0, "var_0": 33, "var_1": 33},
)
assert "number of seed fields" in str(err.value)
def test_validate_seed_lines_field_not_present(test_data):
batches = DataFrameBatch(df=test_data, config=config_template, batch_size=3)
with pytest.raises(RuntimeError) as err:
_validate_batch_seed_values(
batches.batches[0],
{
"ID_code": "foo",
"target": 0,
"var_1": 33,
},
)
assert "The header: var_0 is not in the seed" in str(err.value)
def test_validate_seed_lines_ok_two_field(test_data):
batches = DataFrameBatch(df=test_data, config=config_template, batch_size=3)
check = _validate_batch_seed_values(
batches.batches[0], {"ID_code": "foo", "target": 1}
)
assert check == "foo|1|"
def test_validate_seed_lines_ok_full_size(test_data):
batches = DataFrameBatch(df=test_data, config=config_template, batch_size=3)
check = _validate_batch_seed_values(
batches.batches[0],
{
"ID_code": "foo",
"target": 0,
"var_0": 33,
},
)
assert check == "foo|0|33|"
