"""Tests for canonical JSON encoding (``mnema.core.canonical``).

The canonical encoding is the foundation of every content id and signature in
the protocol: two nodes must derive byte-identical encodings for semantically
identical values, or ids and signatures will not interoperate.
"""

import json

import pytest

from mnema.core.canonical import canonical_json


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


def test_key_order_independence():
    a = {"b": 2, "a": 1, "c": [1, 2]}
    b = {"c": [1, 2], "a": 1, "b": 2}
    assert canonical_json(a) == canonical_json(b)


def test_nested_key_order_independence():
    a = {"outer": {"y": [{"k": 1, "j": 2}], "x": 0}}
    b = {"outer": {"x": 0, "y": [{"j": 2, "k": 1}]}}
    assert canonical_json(a) == canonical_json(b)


def test_array_order_is_significant():
    # Canonicalisation must never reorder arrays - only object keys.
    assert canonical_json([1, 2]) != canonical_json([2, 1])


def test_round_trip_preserves_value():
    value = {
        "s": "héllo",
        "n": 3,
        "f": 0.5,
        "b": True,
        "z": None,
        "l": [1, "two", {"three": 3}],
    }
    assert json.loads(canonical_json(value).decode("utf-8")) == value


def test_deterministic_across_calls():
    value = {"k": list(range(50)), "m": {"x": "y"}}
    assert canonical_json(value) == canonical_json(value)


def test_distinct_values_produce_distinct_bytes():
    assert canonical_json({"a": 1}) != canonical_json({"a": 2})
    # Type distinctions must survive: the number 1 is not the string "1".
    assert canonical_json({"a": 1}) != canonical_json({"a": "1"})


def test_empty_containers_are_distinct():
    assert canonical_json({}) != canonical_json([])


@pytest.mark.parametrize("bad", [float("nan"), float("inf"), float("-inf")])
def test_rejects_non_finite_numbers(bad):
    # NaN/Infinity have no JSON representation; allowing them would produce
    # encodings other implementations cannot parse.
    with pytest.raises((ValueError, TypeError)):
        canonical_json({"x": bad})


def test_unicode_round_trip():
    value = {"text": "café ☕ — 東京"}
    decoded = json.loads(canonical_json(value).decode("utf-8"))
    assert decoded == value