Beispiel #1
0
def source_file2_partial_dtypes(source_data_test_dir: Path) -> SourceFile:
    """Get SourceFile instance of source_file2.tsv with dtypes set for
    only 2/4 columns."""
    file_path = source_data_test_dir / 'test_dir1' / 'source_file2.tsv'
    dtypes = {
        'column_b': 'Int64',
        'column_c': 'datetime64[ns]',
    }
    params = get_file_params(dtypes=dtypes)
    return SourceFile(file_path, params)
Beispiel #2
0
def source_file2(source_data_test_dir: Path) -> SourceFile:
    """Get SourceFile instance of source_file2.tsv with dtypes."""
    file_path = source_data_test_dir / 'test_dir1' / 'source_file2.tsv'
    dtypes = {
        'column_a': 'object',
        'column_b': 'Int64',
        'column_c': 'datetime64[ns]',
        'column_d': 'float64',
    }
    params = get_file_params(dtypes=dtypes)
    return SourceFile(file_path, params)
Beispiel #3
0
def sas_source_file(source_data_test_dir: Path) -> SourceFile:
    """Get DataFrame of beer.sas7bdat."""
    file_path = source_data_test_dir / 'test_dir1' / 'beer.sas7bdat'
    params = get_file_params(binary=True)
    return SourceFile(file_path, params)
Beispiel #4
0
def source_file1_df(source_data_test_dir: Path) -> pd.DataFrame:
    """Get DataFrame of source_file1.csv without setting dtypes."""
    file_path = source_data_test_dir / 'test_dir1' / 'source_file1.csv'
    params = get_file_params(delimiter=',')
    source_file = SourceFile(file_path, params)
    return source_file.get_csv_as_df(apply_dtypes=False)