muutils.misc.hashing

 1from __future__ import annotations
 2
 3import base64
 4import hashlib
 5import json
 6from typing import Any
 7
 8
 9def stable_hash(s: str | bytes) -> int:
10    """Returns a stable hash of the given string. not cryptographically secure, but stable between runs"""
11    # init hash object and update with string
12    s_bytes: bytes
13    if isinstance(s, str):
14        s_bytes = s.encode("utf-8")
15    else:
16        s_bytes = s
17    hash_obj: hashlib._Hash = hashlib.md5(s_bytes)  # pyright: ignore[reportPrivateUsage]
18    # get digest and convert to int
19    return int.from_bytes(hash_obj.digest(), "big")
20
21
22def stable_json_dumps(d: Any) -> str:  # pyright: ignore[reportAny]
23    return json.dumps(
24        d,
25        sort_keys=True,
26        indent=None,
27    )
28
29
30def base64_hash(s: str | bytes) -> str:
31    """Returns a base64 representation of the hash of the given string. not cryptographically secure"""
32    s_bytes: bytes
33    if isinstance(s, str):
34        s_bytes = bytes(s, "UTF-8")
35    else:
36        s_bytes = s
37    hash_bytes: bytes = hashlib.md5(s_bytes).digest()
38    hash_b64: str = base64.b64encode(hash_bytes, altchars=b"-_").decode()
39    return hash_b64