Convert the locale of WhatsApp exports from DE to EN

whatsapp_convert_locale.py

import re, sys

REGEX_ATTACHMENT = {"de": r"\u200E(.*?) \(Datei angehängt\)"}
SUB_ATTACHMENT = r"<attached: \g<1>>"


def convert(file: str, locale="de"):
    """
    Convert the locale of attachments to the english one.
    The output is written to a file in the same location.

    :param file: The input file name that ends in .txt
    :param locale: The locale of the input file, defaults to "de"
    """
    output = re.sub(r"\.txt$", f".en.txt", file)
    pattern = re.compile(REGEX_ATTACHMENT[locale])

    with open(file, "r", encoding="utf-8") as fread:
        with open(output, "w", encoding="utf-8") as fwrit:
            for line in fread:
                result = pattern.sub(SUB_ATTACHMENT, line.strip())
                fwrit.write(result + "\n")

    print(f"saved converted version to '{output}'")


if __name__ == "__main__":
    if len(sys.argv) == 1:
        print("please specify the file name as an argument")
        exit(1)

    convert(sys.argv[1])