clean_roll.py

#!/usr/bin/env python
"""Text file cleaning utility (primarily for secured roll preparation)."""
from pathlib import Path
import typing


def clean_file(
    in_path: Path,
    out_path: Path,
    in_encoding: str,
    out_encoding: str,
):
    with open(in_path, "r", encoding=in_encoding) as f:
        with open(out_path, "w", encoding=out_encoding) as g:
            g.write(f.read().replace("\\", ""))


def clean_file_name(p: Path) -> Path:
    s = str(p.name).lower().replace(" ", "_")
    i = s.find(".")
    j = i + 1
    return Path(f"{s[:i]}_cleaned.{s[j:]}")

DEFAULT_IN_ENCODING: str = "latin-1"
DEFAULT_OUT_ENCODING: str = "utf8"

Encoding = typing.Literal["latin-1", "utf8"]

DESCRIPTION = """\
Take a (hopefully) ASCII text file and re-write it as a cleaned UTF-8 text file.
Removes any all back-slashes.
"""

if __name__ == "__main__":
    import argparse

    parser = argparse.ArgumentParser("clean_ascii", description=DESCRIPTION)

    _encodings = typing.get_args(Encoding)

    parser.add_argument("in_path", type=Path)
    parser.add_argument("-o", "--out_path", type=Path)
    parser.add_argument(
        "-e",
        "--in_encoding",
        type=str,
        choices=_encodings,
        default=DEFAULT_IN_ENCODING,
    )
    parser.add_argument(
        "-E",
        "--out_encoding",
        type=str,
        choices=_encodings,
        default=DEFAULT_OUT_ENCODING,
    )

    args = parser.parse_args()

    out_path = args.out_path or args.in_path.parent / clean_file_name(args.in_path)
    clean_file(
        args.in_path,
        out_path,
        args.in_encoding,
        args.out_encoding,
    )