Script to transform UKMap and LSOA data into a form readable by CiMo.

ukmapconvert.py

from pathlib import Path

import geopandas as gpd
import click
click.disable_unicode_literals_warning = True

ACTUAL_LSOA_COLUMN_NAME = 'LSOA_CODE'
NEEDED_LSOA_COLUMN_NAME = 'LSOAcode'


def cast_gtn_to_int(path_to_file, path_to_modified_file):
    data = gpd.read_file(path_to_file.as_posix())
    data.GTN = [int(gtn) for gtn in data.GTN]
    data.to_file(path_to_modified_file.as_posix())
    print('Casted GTN to integer and wrote results to {}.'.format(path_to_modified_file))


def rename_lsoa_column(path_to_file, path_to_modified_file):
    data = gpd.read_file(path_to_file.as_posix())
    data[NEEDED_LSOA_COLUMN_NAME] = data[ACTUAL_LSOA_COLUMN_NAME]
    del data[ACTUAL_LSOA_COLUMN_NAME]
    data.to_file(path_to_modified_file.as_posix())
    print('Renamed column "{}" to "{}" and wrote results to {}.'.format(ACTUAL_LSOA_COLUMN_NAME,
                                                                        NEEDED_LSOA_COLUMN_NAME,
                                                                        path_to_modified_file))


class ShapefileParamType(click.ParamType):
    name = 'Shapefile'

    def convert(self, value, param, ctx):
        path = Path(value)
        if not path.suffix == '.shp':
            self.fail('File "{}" is not a shape file.'.format(value))
        if not path.exists():
            self.fail('Path "{}" does not exist.'.format(value))
        return path


@click.command(name='ukmap-converter')
@click.argument('shape_file', type=ShapefileParamType())
def command_line_interface(shape_file):
    """Convertes UKMap files to be read by CiMo.

    The script will convert GTN from float to integer, and rename the LSOAcode column.
    Result files will be written to the same folder with a 'modified' extension.

    \b
    Parameters:
        shape_file:  The UKMap based shape file to be modified.
    """
    path_to_modified_file = shape_file.with_name(shape_file.stem + '_modified')
    cast_gtn_to_int(shape_file, path_to_modified_file)
    rename_lsoa_column(path_to_modified_file, path_to_modified_file)


if __name__ == '__main__':
    command_line_interface()

Could this script be incorporated into MasterBeforePreProcess.py?