"""Canonicalization tests.

The wire format (milestone #2) requires that every signed operation is
serialised through a single canonical JSON encoding: UTF-8, lexicographically
sorted object keys, no insignificant whitespace, deterministic for equal
values regardless of construction order. Signature and hash stability across
implementations depends entirely on these properties.
"""

from __future__ import annotations

import json

import pytest

from pmp.canonical import canonical_json, hash_bytes, hash_canonical
from pmp.errors import PMPError


def test_returns_bytes():
    out = canonical_json({"a": 1})
    assert isinstance(out, (bytes, bytearray))


def test_key_order_is_sorted_and_deterministic():
    a = canonical_json({"b": 1, "a": 2, "c": 3})
    b = canonical_json({"c": 3, "a": 2, "b": 1})
    assert a == b
    decoded = a.decode("utf-8")
    assert decoded.index('"a"') < decoded.index('"b"') < decoded.index('"c"')


def test_no_insignificant_whitespace():
    out = canonical_json({"a": [1, 2], "b": {"c": True}}).decode("utf-8")
    # Canonical form must not contain spaces outside string values.
    assert " " not in out
    assert "\n" not in out
    assert "\t" not in out


def test_nested_structures_roundtrip():
    value = {
        "list": [3, 1, 2],          # list order must be preserved, not sorted
        "nested": {"y": None, "x": [True, False]},
        "empty_list": [],
        "empty_obj": {},
        "int": 42,
        "str": "hello",
    }
    out = canonical_json(value)
    assert json.loads(out.decode("utf-8")) == value


def test_list_order_preserved():
    a = canonical_json({"k": [2, 1]})
    b = canonical_json({"k": [1, 2]})
    assert a != b


def test_unicode_values_roundtrip():
    value = {"name": "Çağrı", "city": "Zürich", "emoji": "🥐"}
    out = canonical_json(value)
    assert json.loads(out.decode("utf-8")) == value


def test_equal_values_hash_equal():
    h1 = hash_canonical({"x": 1, "y": [1, 2, 3]})
    h2 = hash_canonical({"y": [1, 2, 3], "x": 1})
    assert h1 == h2


def test_different_values_hash_differently():
    assert hash_canonical({"x": 1}) != hash_canonical({"x": 2})
    assert hash_canonical({"x": 1}) != hash_canonical({"x": "1"})


def test_hash_format_is_sha256_hex():
    h = hash_canonical({"x": 1})
    digest = h.split(":")[-1]
    assert len(digest) == 64
    assert set(digest) <= set("0123456789abcdef")


def test_hash_bytes_matches_known_input_shape():
    h1 = hash_bytes(b"hello")
    h2 = hash_bytes(b"hello")
    h3 = hash_bytes(b"hellp")
    assert h1 == h2
    assert h1 != h3
    digest = h1.split(":")[-1]
    assert len(digest) == 64


def test_non_serialisable_values_rejected():
    with pytest.raises((PMPError, TypeError, ValueError)):
        canonical_json({"x": object()})