WIP checkpoint: Creator Studio: Asset Import Pipeline — glTF model import backend

backend/tests/test_gltf_import.py

@@ -0,0 +1,493 @@
+"""Tests for the glTF import pipeline."""
+
+import json
+from pathlib import Path
+
+import pytest
+from fastapi.testclient import TestClient
+from sqlalchemy import select
+
+from app.config import settings
+from app.database import get_db, SessionLocal
+from app.models.asset import Asset, AssetType
+from app.models.user import User
+from app.main import app
+
+
+# --- Fixtures ---
+
+
+@pytest.fixture
+def db_session():
+    """Provide a test database session with rollback."""
+    async_session = SessionLocal()
+
+    async def override_get_db():
+        return async_session
+
+    app.dependency_overrides[get_db] = override_get_db
+
+    yield async_session
+
+    async def cleanup():
+        await async_session.rollback()
+        await async_session.close()
+
+    import asyncio
+
+    asyncio.get_event_loop().run_until_complete(cleanup())
+    app.dependency_overrides.clear()
+
+
+@pytest.fixture
+def test_user(db_session):
+    """Create a test user."""
+    user = User(
+        username="test_gltf_user",
+        email="gltf_test@example.com",
+        hashed_password="$2b$12$EixZaYVK1fsbw1ZfbX3OXePaWxn96p36PQm3iV8jEGPym2fA1Wile",  # pass
+        is_active=True,
+    )
+    db_session.add(user)
+    import asyncio
+
+    asyncio.get_event_loop().run_until_complete(db_session.commit())
+    asyncio.get_event_loop().run_until_complete(db_session.refresh(user))
+    return user
+
+
+@pytest.fixture
+def client(db_session, test_user):
+    """Create a test client with auth."""
+    from jose import jwt
+    from datetime import datetime, timedelta, timezone
+
+    token = jwt.encode(
+        {
+            "sub": str(test_user.id),
+            "exp": datetime.now(timezone.utc)
+            + timedelta(minutes=30),
+        },
+        settings.SECRET_KEY,
+        algorithm=settings.ALGORITHM,
+    )
+
+    from fastapi.testclient import TestClient
+
+    client = TestClient(app)
+    client.headers["Authorization"] = f"Bearer {token}"
+    return client
+
+
+# --- Helpers for building test glTF files ---
+
+
+def make_gltf_json(
+    scenes=True,
+    meshes=True,
+    nodes=False,
+    materials=0,
+    animations=0,
+    vertex_count=0,
+    has_bounds=True,
+):
+    """Build a minimal valid .gltf JSON structure."""
+    data = {
+        "asset": {"version": "2.0", "generator": "test"},
+    }
+    if scenes:
+        data["scenes"] = [{"nodes": [0]}]
+    if nodes:
+        data["nodes"] = [{"name": "root"}]
+    if meshes:
+        accessors = []
+        if vertex_count > 0 and has_bounds:
+            accessors.append(
+                {
+                    "count": vertex_count,
+                    "componentType": 5126,
+                    "type": "VEC3",
+                    "min": [0.0, 0.0, 0.0],
+                    "max": [1.0, 1.0, 1.0],
+                }
+            )
+        elif vertex_count > 0:
+            accessors.append(
+                {
+                    "count": vertex_count,
+                    "componentType": 5126,
+                    "type": "VEC3",
+                }
+            )
+        meshes_list = []
+        for _ in range(1):
+            prim = {"attributes": {"POSITION": 0}} if accessors else {}
+            meshes_list.append({"primitives": [prim]})
+        data["meshes"] = meshes_list
+        if accessors:
+            data["accessors"] = accessors
+    if materials > 0:
+        data["materials"] = [{"name": f"mat_{i}"} for i in range(materials)]
+    if animations > 0:
+        data["animations"] = [{"name": f"anim_{i}"} for i in range(animations)]
+    return data
+
+
+def make_glb(data_dict):
+    """Build a minimal valid .glb binary file from a dict."""
+    json_bytes = json.dumps(data_dict).encode("utf-8")
+    # Pad JSON to 4-byte boundary
+    json_padding = (4 - len(json_bytes) % 4) % 4
+    json_bytes += b" " * json_padding
+
+    json_chunk_length = len(json_bytes)
+    json_chunk_type = b"JSON"
+
+    # No binary chunks for simplicity
+    total_length = 12 + 8 + json_chunk_length + 8  # header + json chunk header + data + BIN padding
+
+    header = b"glTF"
+    header += int.to_bytes(2, 4, "little")  # version
+    header += int.to_bytes(total_length, 4, "little")  # total length
+
+    chunk_header = int.to_bytes(json_chunk_length, 4, "little")
+    chunk_header += json_chunk_type
+
+    # Add a BIN chunk with zero length to pad
+    bin_chunk_length = 0
+    bin_chunk_type = b"BIN"
+    bin_padding = (4 - bin_chunk_length % 4) % 4
+
+    return (
+        header
+        + chunk_header
+        + json_bytes
+        + bin_chunk_type
+        + int.to_bytes(bin_chunk_length, 4, "little")
+        + b" " * bin_padding
+    )
+
+
+# --- Tests: Extension validation ---
+
+
+class TestImportGltfExtensionValidation:
+    """Test that only .gltf and .glb files are accepted."""
+
+    def test_rejects_non_gltf_extension(self, client):
+        """Files without .gltf or .glb extension return 400."""
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("test.png", b"fake png data", "image/png")},
+        )
+        assert response.status_code == 400
+        assert ".gltf" in response.json()["detail"].lower() or ".glb" in response.json()["detail"].lower()
+
+    def test_rejects_txt_extension(self, client):
+        """A .txt file is rejected."""
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.txt", b"not a gltf", "text/plain")},
+        )
+        assert response.status_code == 400
+
+
+# --- Tests: .gltf validation ---
+
+
+class TestImportGltfJsonValidation:
+    """Test validation of .gltf (JSON) files."""
+
+    def test_valid_minimal_gltf(self, client):
+        """A minimal valid .gltf is accepted."""
+        data = make_gltf_json(vertex_count=24, materials=2, animations=1)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+        body = response.json()
+        assert body["asset_type"] == "model_gltf"
+        assert body["file_size"] == len(json_bytes)
+        assert body["mime_type"] == "model/gltf+json"
+        metadata = json.loads(body["metadata_json"])
+        assert metadata["vertex_count"] == 24
+        assert metadata["material_count"] == 2
+        assert metadata["has_animation"] is True
+
+    def test_rejects_invalid_json(self, client):
+        """Malformed JSON returns 400."""
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("bad.gltf", b"{not valid json", "model/gltf+json")},
+        )
+        assert response.status_code == 400
+        assert "Invalid JSON" in response.json()["detail"]
+
+    def test_rejects_gltf_missing_asset(self, client):
+        """glTF without 'asset' property returns 400."""
+        data = {"scenes": [{}]}
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("no_asset.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 400
+        assert "Missing required 'asset'" in response.json()["detail"]
+
+    def test_rejects_gltf_no_meshes_scenes_nodes(self, client):
+        """glTF with neither scenes, meshes, nor nodes returns 400."""
+        data = {"asset": {"version": "2.0"}, "materials": []}
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("empty.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 400
+        assert "scenes, meshes, or nodes" in response.json()["detail"]
+
+
+# --- Tests: .glb validation ---
+
+
+class TestImportGlbValidation:
+    """Test validation of .glb (binary) files."""
+
+    def test_valid_minimal_glb(self, client):
+        """A minimal valid .glb is accepted."""
+        data = make_gltf_json(vertex_count=12, materials=1)
+        glb_bytes = make_glb(data)
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.glb", glb_bytes, "model/gltf+binary")},
+        )
+        assert response.status_code == 201
+        body = response.json()
+        assert body["asset_type"] == "model_gltf"
+        assert body["file_size"] == len(glb_bytes)
+        assert body["mime_type"] == "model/gltf+binary"
+        metadata = json.loads(body["metadata_json"])
+        assert metadata["vertex_count"] == 12
+        assert metadata["material_count"] == 1
+
+    def test_rejects_invalid_magic_bytes(self, client):
+        """File not starting with 'glTF' magic returns 400."""
+        bad_glb = b"xxxx" + int.to_bytes(2, 4, "little") + int.to_bytes(16, 4, "little")
+        bad_glb += int.to_bytes(4, 4, "little") + b"JSON" + b"{}" + b"BIN" + int.to_bytes(0, 4, "little")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("bad.glb", bad_glb, "model/gltf+binary")},
+        )
+        assert response.status_code == 400
+        assert "magic" in response.json()["detail"].lower()
+
+    def test_rejects_wrong_version(self, client):
+        """glTF version 1 returns 400."""
+        data = {"asset": {"version": "1.0"}}
+        json_bytes = json.dumps(data).encode("utf-8")
+        json_padding = (4 - len(json_bytes) % 4) % 4
+        json_bytes += b" " * json_padding
+        json_len = len(json_bytes)
+
+        header = b"glTF" + int.to_bytes(1, 4, "little") + int.to_bytes(12 + 8 + json_len + 8, 4, "little")
+        chunk = int.to_bytes(json_len, 4, "little") + b"JSON"
+        bin_pad = b"BIN" + int.to_bytes(0, 4, "little")
+
+        glb_bytes = header + chunk + json_bytes + bin_pad
+
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("v1.glb", glb_bytes, "model/gltf+binary")},
+        )
+        assert response.status_code == 400
+        assert "Unsupported glTF version" in response.json()["detail"]
+
+    def test_rejects_too_short(self, client):
+        """A file shorter than 12 bytes returns 400."""
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("tiny.glb", b"glTF123", "model/gltf+binary")},
+        )
+        assert response.status_code == 400
+        assert "too short" in response.json()["detail"].lower()
+
+
+# --- Tests: File size validation ---
+
+
+class TestImportGltfFileSize:
+    """Test file size limits."""
+
+    def test_rejects_oversized_file(self, client):
+        """Files exceeding MAX_UPLOAD_SIZE_MB return 413."""
+        # Create a file larger than the default 100 MB limit
+        large_content = b"X" * (settings.MAX_UPLOAD_SIZE_MB * 1024 * 1024 + 1)
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("big.gltf", large_content, "model/gltf+json")},
+        )
+        assert response.status_code == 413
+        assert "exceeds maximum" in response.json()["detail"].lower()
+
+
+# --- Tests: Metadata extraction ---
+
+
+class TestMetadataExtraction:
+    """Test that metadata is correctly extracted from glTF data."""
+
+    def test_vertex_count_from_meshes(self, client):
+        """Vertex count is extracted from mesh primitives."""
+        data = make_gltf_json(vertex_count=100, materials=0, animations=0)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+        metadata = json.loads(response.json()["metadata_json"])
+        assert metadata["vertex_count"] == 100
+
+    def test_material_count(self, client):
+        """Material count is extracted from materials array."""
+        data = make_gltf_json(materials=5, vertex_count=0)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+        metadata = json.loads(response.json()["metadata_json"])
+        assert metadata["material_count"] == 5
+
+    def test_has_animation_flag(self, client):
+        """has_animation is True when animations exist."""
+        data = make_gltf_json(animations=3, vertex_count=0)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+        metadata = json.loads(response.json()["metadata_json"])
+        assert metadata["has_animation"] is True
+
+    def test_no_animation_flag(self, client):
+        """has_animation is False when no animations exist."""
+        data = make_gltf_json(animations=0, vertex_count=0)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+        metadata = json.loads(response.json()["metadata_json"])
+        assert metadata["has_animation"] is False
+
+    def test_bounding_box_extraction(self, client):
+        """Bounding box is extracted from accessor min/max."""
+        data = make_gltf_json(vertex_count=8, has_bounds=True)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+        metadata = json.loads(response.json()["metadata_json"])
+        assert metadata["bounding_box"] is not None
+        assert "min" in metadata["bounding_box"]
+        assert "max" in metadata["bounding_box"]
+
+    def test_no_bounding_box_without_accessor_bounds(self, client):
+        """Bounding box is None when accessor has no min/max."""
+        data = make_gltf_json(vertex_count=8, has_bounds=False)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+        metadata = json.loads(response.json()["metadata_json"])
+        assert metadata["bounding_box"] is None
+
+
+# --- Tests: Database persistence ---
+
+
+class TestDatabasePersistence:
+    """Test that assets are correctly stored in the database."""
+
+    def test_asset_created_in_db(self, client, db_session, test_user):
+        """An asset record is created in the database."""
+        data = make_gltf_json(vertex_count=10)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+
+        # Verify in database
+        result = db_session.execute(select(Asset).where(Asset.id == response.json()["id"]))
+        asset = result.scalar_one_or_none()
+        assert asset is not None
+        assert asset.asset_type == AssetType.MODEL_GLTF
+        assert asset.owner_id == test_user.id
+        assert asset.file_size == len(json_bytes)
+        assert asset.metadata_json is not None
+
+    def test_file_stored_on_disk(self, client, db_session, test_user):
+        """The file is stored in the upload directory."""
+        data = make_gltf_json(vertex_count=10)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+
+        asset_id = response.json()["id"]
+        upload_dir = Path(settings.UPLOAD_DIR)
+        # The stored file should exist
+        found = False
+        for f in upload_dir.iterdir():
+            if f.name.startswith(f"{asset_id}_"):
+                found = True
+                assert f.read_bytes() == json_bytes
+                break
+        assert found, f"File not found in {upload_dir}"
+
+
+# --- Tests: Filename sanitization ---
+
+
+class TestFilenameSanitization:
+    """Test that filenames are properly sanitized."""
+
+    def test_sanitizes_special_chars(self, client):
+        """Special characters in filename are stripped."""
+        data = make_gltf_json(vertex_count=10)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("my model (1).gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+        # The name should be the stem without special chars
+        assert response.json()["name"] == "my model 1"
+
+    def test_preserves_extension(self, client):
+        """The .gltf extension is preserved in the stored file."""
+        data = make_gltf_json(vertex_count=10)
+        json_bytes = json.dumps(data).encode("utf-8")
+        response = client.post(
+            "/api/assets/import-gltf",
+            files={"file": ("model.gltf", json_bytes, "model/gltf+json")},
+        )
+        assert response.status_code == 201
+        upload_dir = Path(settings.UPLOAD_DIR)
+        for f in upload_dir.iterdir():
+            if f.name.endswith(".gltf"):
+                assert f.read_bytes() == json_bytes
+                break