from mltrail_verifiable_provenance_ledger_for.reprobundle import ReproBundle
from mltrail_verifiable_provenance_ledger_for.contracts import Environment


def make_sample_environment():
    return Environment(
        id="env1",
        language="python",
        version="3.9",
        dependencies={"numpy": "1.24.0", "pandas": "1.5.3"},
        container_hash=None,
    )


def test_bundle_hash_deterministic():
    env = make_sample_environment()
    bundle1 = ReproBundle(
        code_commit="deadbeef",
        environment=env,
        dataset_refs=[{"id": "ds1", "fingerprint": "fp1"}],
        model_ref={"id": "m1", "fingerprint": "mf1"},
        run_manifest={"seed": 42, "cmd": "python train.py"},
    )
    bundle2 = ReproBundle(
        code_commit="deadbeef",
        environment=make_sample_environment(),
        dataset_refs=[{"id": "ds1", "fingerprint": "fp1"}],
        model_ref={"id": "m1", "fingerprint": "mf1"},
        run_manifest={"seed": 42, "cmd": "python train.py"},
    )

    h1 = bundle1.compute_bundle_hash()
    h2 = bundle2.compute_bundle_hash()
    assert h1 == h2


def test_bundle_hash_changes_on_difference():
    env = make_sample_environment()
    base = ReproBundle(
        code_commit="deadbeef",
        environment=env,
        dataset_refs=[{"id": "ds1", "fingerprint": "fp1"}],
        model_ref={"id": "m1", "fingerprint": "mf1"},
        run_manifest={"seed": 42, "cmd": "python train.py"},
    )
    modified = ReproBundle(
        code_commit="cafebabe",  # different commit
        environment=env,
        dataset_refs=[{"id": "ds1", "fingerprint": "fp1"}],
        model_ref={"id": "m1", "fingerprint": "mf1"},
        run_manifest={"seed": 42, "cmd": "python train.py"},
    )

    assert base.compute_bundle_hash() != modified.compute_bundle_hash()