A script I use to cleanup Python "scripts" provided in form of copy-pasted REPL sessions

repl2script.py

# Python 3.8 is required, but this was only tested manually with 3.12
from __future__ import annotations

import argparse
import re
import sys
from dataclasses import dataclass
from functools import cached_property
from pathlib import Path


@dataclass(frozen=True)
class Repl:
    prompt: str
    continuation: str

    @cached_property
    def regexp(self):
        return re.compile(rf"^({self.prompt}|{self.continuation})")


PythonRepl = Repl(prompt=r">>> ", continuation=r"\.\.\. ")
IPythonRepl = Repl(prompt=r"In \[\d+\]: ", continuation=r"   \.\.\.: ")
NullRepl = Repl(prompt=r"", continuation=r"")  # does nothing

ALL_REPLS = [PythonRepl, IPythonRepl]


def get_repl(lines: list[str]) -> Repl:
    for repl in ALL_REPLS:
        for L in lines:
            if repl.regexp.match(L):
                return repl
    return NullRepl


def cleanup_outputs(lines: list[str], repl: Repl) -> list[str]:
    return [L for L in lines if repl.regexp.match(L)]


def cleanup_prompts(lines: list[str], repl: Repl) -> list[str]:
    return [repl.regexp.sub("", L) for L in lines]


def cleanup_whitespace(lines: list[str]) -> list[str]:
    return [L.rstrip() for L in lines]


def main(argv: list[str] | None = None) -> int:
    parser = argparse.ArgumentParser()

    parser.add_argument("input_file", type=Path, help="the file to be read")
    parser.add_argument(
        "--no-output-cleaning",
        dest="output_cleaning",
        action="store_false",
        help="disable output cleaning (intended for debugging)",
    )
    parser.add_argument(
        "--overwrite",
        action="store_true",
        help="redirect output to the input file instead of stdout",
    )

    repl_selector = parser.add_mutually_exclusive_group(required=False)
    repl_selector.add_argument(
        "--python",
        action="store_true",
        help="assume input comes from the basic Python REPL",
    )
    repl_selector.add_argument(
        "--ipython",
        action="store_true",
        help="assume input comes from the IPython REPL",
    )

    args = parser.parse_args(argv)

    if not args.input_file.is_file():
        print(f"Error: no such file '{args.input_file!s}'", file=sys.stderr)
        return 1

    body = args.input_file.read_text()
    lines = body.splitlines()

    if args.python:
        repl = PythonRepl
    elif args.ipython:
        repl = IPythonRepl
    else:
        repl = get_repl(lines)

    if args.output_cleaning:
        lines = cleanup_outputs(lines, repl)

    lines = cleanup_prompts(lines, repl)
    lines = cleanup_whitespace(lines)
    result = "\n".join(lines)
    if args.overwrite:
        if result != body:
            args.input_file.write_text(result)
    else:
        print(result)
    return 0


if __name__ == "__main__":
    sys.exit(main())