"""Dataset downloader — URL registry + SHA256 verification."""

import hashlib
from pathlib import Path

import pytest

from gps_denied.testing.download import (
    DATASET_REGISTRY,
    DatasetSpec,
    verify_sha256,
    download_dataset,
)


def test_registry_has_euroc():
    assert "euroc_mh01" in DATASET_REGISTRY
    spec = DATASET_REGISTRY["euroc_mh01"]
    assert isinstance(spec, DatasetSpec)
    assert spec.url.startswith("http")
    assert len(spec.sha256) == 64


def test_verify_sha256_matches(tmp_path: Path):
    data = b"hello world"
    f = tmp_path / "x.bin"
    f.write_bytes(data)
    expected = hashlib.sha256(data).hexdigest()
    assert verify_sha256(f, expected) is True


def test_verify_sha256_mismatch(tmp_path: Path):
    f = tmp_path / "x.bin"
    f.write_bytes(b"hello world")
    assert verify_sha256(f, "0" * 64) is False


def test_download_skip_if_present(tmp_path: Path, monkeypatch):
    f = tmp_path / "cached.zip"
    f.write_bytes(b"cached")
    spec = DatasetSpec(
        url="http://example.invalid/cached.zip",
        sha256=hashlib.sha256(b"cached").hexdigest(),
        target_subdir="cached",
    )
    # Should return the path without hitting the network
    called = {"n": 0}

    def fake_get(*args, **kwargs):
        called["n"] += 1
        raise AssertionError("download should have been skipped")

    monkeypatch.setattr("urllib.request.urlretrieve", fake_get)
    result = download_dataset(spec, f)
    assert result == f
    assert called["n"] == 0