hms-mediaengine/tests/unit/test_content_manifest.py

"""Unit-Tests Content-Manifest (PLAN.md §6.4, §10.1, §29.1).

Manifest-Chunks/-Hashes, Diff für Sync, Verifikation gegen Fälschung
und Rundreise – Grundlage für Preflight und resumierbare Übertragung.
"""

from __future__ import annotations

import json
from pathlib import Path

import pytest
from hms_content_sync import CHUNK_SIZE, ContentManifest, hash_file


@pytest.fixture()
def content_tree(tmp_path: Path) -> Path:
    root = tmp_path / "content"
    (root / "clips").mkdir(parents=True)
    (root / "clips" / "intro.mp4").write_bytes(b"A" * 1000)
    (root / "clips" / "loop.mp4").write_bytes(b"B" * 500)
    (root / "presets").mkdir()
    (root / "presets" / "look.json").write_text("{}", encoding="utf-8")
    return root


def test_hash_file_chunks_match_content(tmp_path: Path) -> None:
    f = tmp_path / "f.bin"
    f.write_bytes(b"x" * (CHUNK_SIZE + 100))  # > 1 Chunk
    file_hash, chunks = hash_file(f)
    assert len(chunks) == 2  # voller + Rest-Chunk
    assert len(file_hash) == 64  # SHA-256 Hex


def test_hash_file_empty_file_yields_empty_chunk(tmp_path: Path) -> None:
    f = tmp_path / "empty.bin"
    f.write_bytes(b"")
    file_hash, chunks = hash_file(f)
    assert len(chunks) == 1
    assert file_hash  # auch leere Dateien haben einen definierten Hash


def test_manifest_build_covers_all_files(content_tree: Path) -> None:
    manifest = ContentManifest.build(content_tree, revision=7)
    assert len(manifest) == 3
    assert manifest.revision == 7
    assert "clips/intro.mp4" in manifest.entries
    assert "presets/look.json" in manifest.entries
    entry = manifest.entries["clips/intro.mp4"]
    assert entry.size_bytes == 1000
    assert len(entry.sha256) == 64
    assert entry.kind == "media"  # Default-Art


def test_manifest_kind_classification(content_tree: Path) -> None:
    manifest = ContentManifest.build(
        content_tree,
        kind_of=lambda rel: "plugin" if rel.endswith(".json") else "media",
    )
    assert manifest.entries["presets/look.json"].kind == "plugin"
    assert manifest.entries["clips/intro.mp4"].kind == "media"


def test_manifest_roundtrip_preserves_entries(content_tree: Path) -> None:
    manifest = ContentManifest.build(content_tree, revision=3)
    restored = ContentManifest.from_dict(json.loads(json.dumps(manifest.to_dict())))
    assert restored.revision == 3
    assert set(restored.entries) == set(manifest.entries)
    orig = manifest.entries["clips/intro.mp4"]
    copy = restored.entries["clips/intro.mp4"]
    assert copy.sha256 == orig.sha256
    assert copy.chunk_hashes == orig.chunk_hashes


def test_manifest_is_deterministic_across_builds(content_tree: Path) -> None:
    """Gleicher Inhalt → gleiche Einträge (nur manifest_id ist neu)."""
    a = ContentManifest.build(content_tree)
    b = ContentManifest.build(content_tree)
    assert a.to_dict()["entries"] == b.to_dict()["entries"]
    assert a.manifest_id != b.manifest_id  # Instanz-ID ist zeitlich unique


def test_verify_detects_tampered_file(content_tree: Path) -> None:
    manifest = ContentManifest.build(content_tree)
    assert manifest.verify_file(content_tree, "clips/intro.mp4")
    (content_tree / "clips" / "intro.mp4").write_bytes(b"TAMPERED" * 50)
    assert not manifest.verify_file(content_tree, "clips/intro.mp4")  # §6.4 Hashprüfung


def test_verify_detects_missing_and_unknown(content_tree: Path) -> None:
    manifest = ContentManifest.build(content_tree)
    (content_tree / "clips" / "loop.mp4").unlink()
    assert not manifest.verify_file(content_tree, "clips/loop.mp4")
    assert not manifest.verify_file(content_tree, "gibts/nicht.mp4")


def test_diff_drives_sync_plan(content_tree: Path) -> None:
    """Diff liefert exakt die Übertragungsliste (§6.4: added/removed/changed)."""
    old = ContentManifest.build(content_tree, revision=1)
    # Ändern + Hinzufügen + Entfernen
    (content_tree / "clips" / "intro.mp4").write_bytes(b"GEAENDERT")
    (content_tree / "clips" / "neu.mp4").write_bytes(b"N")
    (content_tree / "clips" / "loop.mp4").unlink()
    new = ContentManifest.build(content_tree, revision=2)

    diff = old.diff(new)
    assert diff.added == frozenset({"clips/neu.mp4"})
    assert diff.removed == frozenset({"clips/loop.mp4"})
    assert diff.changed == frozenset({"clips/intro.mp4"})
    assert not diff.empty

    # identische Manifeste: leerer Diff
    same = old.diff(ContentManifest.from_dict(old.to_dict()))
    assert same.empty


def test_build_rejects_missing_root(tmp_path: Path) -> None:
    with pytest.raises(FileNotFoundError):
        ContentManifest.build(tmp_path / "gibts-nicht")