"""Canonical JSON serialization + sha256 hashing for content-addressed identity."""

from __future__ import annotations

import hashlib
import json
from typing import Any


def canonicalize(value: Any) -> str:
    """Return canonical JSON: keys sorted, no insignificant whitespace, UTF-16 codepoint order.

    json.dumps with sort_keys=True uses Python's default dict key sort which is by Unicode
    codepoint. For ASCII keys this is equivalent to UTF-16 codepoint order which is what
    we want. For non-ASCII keys outside the BMP, this is a documented approximation.
    """
    return json.dumps(
        value,
        sort_keys=True,
        ensure_ascii=False,
        separators=(",", ":"),
        allow_nan=False,
    )


def sha256(value: Any) -> str:
    """Return sha256 hex digest of canonical JSON of value."""
    return hashlib.sha256(canonicalize(value).encode("utf-8")).hexdigest()