Python-Projekte/sandbox/hash_list.py

import hashlib
import io
from pathlib import Path


def get_hash(filename):
    with open(filename, "r", encoding="latin-1", buffering=0) as frh:
        with open(filename + ".sha", "w", encoding="latin-1") as fwh:
            chunk = io.StringIO()
            for i, line in enumerate(frh.readlines()):
                chunk.write(line)
                if i % 1000 == 0:
                    m = hashlib.sha256(chunk.getvalue().encode())
                    chunk.truncate(0)
                    fwh.write(m.hexdigest() + "\n")


def main():
    for f in Path("datev/data").glob("*.csv"):
        get_hash(str(f))


if __name__ == "__main__":
    main()