chrome-to-bitwarden.py

#!/usr/bin/python3
#
# Copyright (C) 2020 Julian Andres Klode <[email protected]>
#
# SPDX-License-Identifier: AGPL-3.0-or-later

"""Takes a Chrome CSV on stdin, and produces a Bitwarden JSON export on stdout.

The output only contains relevant fields, and logins are merged together based
on user name and password.
"""

import csv, sys, uuid, json


def remove_prefix(s, prefix):
    """Remove a prefix from a string."""
    return s[len(prefix) :] if s.startswith(prefix) else s


def clean_name(name, x):
    """Produce a clean name without useless noise like www in it"""

    if not name and x["url"].startswith("android://"):
        name = ".".join(reversed(x["url"].split("/")[2].split("@")[-1].split(".")))
        print("Found android app", name, file=sys.stderr)

    if not any(c.isalpha() for c in name):
        print("Ignoring IP", name, file=sys.stderr)
        return name

    inp = name

    name = remove_prefix(name, "www.")
    name = remove_prefix(name, "login.")
    name = remove_prefix(name, "signin.")
    name = remove_prefix(name, "passport.")
    name = remove_prefix(name, "auth.")
    name = remove_prefix(name, "console.")
    name = remove_prefix(name, "config.")
    name = remove_prefix(name, "app.")
    name = remove_prefix(name, "m.")
    name = remove_prefix(name, "mobile.")
    name = remove_prefix(name, "my.")
    name = remove_prefix(name, "id.")
    name = remove_prefix(name, "account.")
    name = remove_prefix(name, "shop.")
    name = remove_prefix(name, "accounts.login.idm.")
    name = remove_prefix(name, "identity.")
    name = remove_prefix(name, "co2offset.")
    name = remove_prefix(name, "sso.")

    if name.endswith(".ikea.com"):
        name = "ikea.com"
    if name.endswith(".auth0.com"):
        name = name.replace(".auth0.com", "")
    if name.endswith(".shopgate.com"):
        name = name.replace(".shopgate.com", "")

    if name.endswith("wikipedia.org"):
        name = "wikipedia.org"

    if name.endswith("tutanota.com"):
        name = "tutanota.com"

    out = ".".join(name.split(".")[-2:])
    pre = ".".join(name.split(".")[:-2])

    if pre:
        out = f"{out} - {pre}"
        print("Replacing", pre, "in", name, "=>", out, file=sys.stderr)

    return out


def convert(fobj=sys.stdin):
    """Convert Chrome CSV file in fobj to BitWarden JSON file."""
    passwds = list(csv.DictReader(fobj))
    by_login = [((k["username"], k["password"]), k) for k in passwds]
    by_login1 = {k: [v1 for k1, v1 in by_login if k1 == k] for k, v in by_login}
    return {
        "folders": [],
        "items": [
            {
                "type": 1,
                "name": (", ".join(set(clean_name(x["name"], x) for x in v))),
                "login": {
                    "uris": [{"match": None, "uri": x["url"]} for x in v],
                    "username": v[0]["username"],
                    "password": v[0]["password"],
                },
            }
            for v in by_login1.values()
        ],
    }


print(json.dumps(convert(), indent=4,))