import hashlib
import json
import runpy
import sys
import tarfile
from pathlib import Path


def test_model_bundle_embeds_manifest_without_mutating_source(tmp_path: Path, monkeypatch):
    source = tmp_path / "source"
    image_dir = source / "visual" / "image"
    text_dir = source / "visual" / "text"
    image_dir.mkdir(parents=True)
    text_dir.mkdir(parents=True)
    model = image_dir / "model.bin"
    model.write_bytes(b"image model")
    (text_dir / "config.json").write_text("{}", encoding="utf-8")
    source_manifest = source / "manifest.json"
    source_manifest.write_text("source sentinel", encoding="utf-8")
    output = tmp_path / "models.tar.gz"

    script = Path(__file__).parents[1] / "scripts" / "build-model-bundle.py"
    namespace = runpy.run_path(str(script))
    monkeypatch.setattr(sys, "argv", [str(script), str(source), str(output), "--version", "test-v1"])
    namespace["main"]()

    assert source_manifest.read_text(encoding="utf-8") == "source sentinel"
    with tarfile.open(output, "r:gz") as archive:
        assert archive.getnames().count("manifest.json") == 1
        manifest_file = archive.extractfile("manifest.json")
        assert manifest_file is not None
        manifest = json.load(manifest_file)
    assert manifest["version"] == "test-v1"
    assert "manifest.json" not in manifest["files"]
    assert manifest["files"]["visual/image/model.bin"] == hashlib.sha256(model.read_bytes()).hexdigest()