"""Canonicalisation: same data must always hash to the same bytes.

Article 1 (source of truth) and Article 8 (audit ledger) both depend on
a stable canonical form: votes are signed over canonical bytes and the
ledger chains canonical hashes. These tests pin the properties that the
rest of the pipeline relies on.
"""

from __future__ import annotations

import re

from govtool import canonical

from tests.helpers import resolve_fn

HEX64 = re.compile(r"^[0-9a-f]{64}$")


def _dumps(obj) -> bytes:
    fn = resolve_fn(canonical, "canonical_json", "canonical_dumps", "dumps", "to_canonical")
    out = fn(obj)
    return out if isinstance(out, bytes) else str(out).encode("utf-8")


def _hash(obj) -> str:
    fn = resolve_fn(canonical, "canonical_hash", "hash_obj", "sha256_hex", "digest", "content_hash")
    return str(fn(obj))


def test_key_order_does_not_matter():
    a = {"beta": 1, "alpha": [1, 2, {"z": 0, "y": None}], "gamma": "x"}
    b = {"gamma": "x", "alpha": [1, 2, {"y": None, "z": 0}], "beta": 1}
    assert _dumps(a) == _dumps(b)
    assert _hash(a) == _hash(b)


def test_list_order_does_matter():
    assert _hash({"votes": ["yes", "no"]}) != _hash({"votes": ["no", "yes"]})


def test_deterministic_across_calls():
    obj = {"id": "prop-001", "tally": {"yes": 4, "no": 1}, "nested": [{"a": 1}, {"b": 2}]}
    assert _dumps(obj) == _dumps(obj)
    assert _hash(obj) == _hash(obj)


def test_hash_is_sha256_hex():
    digest = _hash({"hello": "world"})
    assert HEX64.match(digest), f"expected 64 lowercase hex chars, got {digest!r}"


def test_distinct_objects_distinct_hashes():
    assert _hash({"yes": 1}) != _hash({"yes": 2})
    assert _hash({"yes": 1}) != _hash({"no": 1})


def test_unicode_is_stable():
    obj = {"name": "Citizen \u00e9\u00e8\u00ea \u4e16\u754c", "emoji": "\U0001f5f3\ufe0f"}
    first = _dumps(obj)
    second = _dumps(obj)
    assert first == second
    assert _hash(obj) == _hash(obj)


def test_scalar_types_are_distinguished():
    # The string "1" and the integer 1 must not collide in canonical form.
    assert _hash({"v": 1}) != _hash({"v": "1"})