test(03-01): add Wave 0 xfail stubs and shared fixtures for Phase 3

- Add auth_user, admin_user, mock_minio_presigned, mock_minio_stat fixtures to conftest.py
- Create test_quota.py with 4 xfail stubs (STORE-03, STORE-05, STORE-06, SC2 race)
- Append test_migration_0003 to test_alembic.py (full pre-seed + post-migration assertions)
- Append 3 classifier xfail stubs (DOC-03, DOC-05, D-15)
- Append 6 document xfail stubs (D-05, STORE-04, SEC-04, D-16)
- Append 4 topic xfail stubs (DOC-04, D-09, D-17)
- Append test_settings_endpoint_removed stub (D-12)
- All 19 new test IDs collect cleanly with xfail(strict=False)

backend/tests/test_alembic.py

@@ -114,3 +114,133 @@ def test_documents_user_id_nullable(tmp_path, monkeypatch):
     assert columns["user_id"]["notnull"] == 0, (
         "documents.user_id is NOT NULL but D-03 requires it to be nullable in Phase 1"
     )
+
+
+# ---------------------------------------------------------------------------
+# Test 3: migration 0003 — null-user cleanup + NOT NULL + quota reconciliation
+# ---------------------------------------------------------------------------
+
+@pytest.mark.xfail(strict=False, reason="implemented in plan 03-01 migration step")
+def test_migration_0003(tmp_path, monkeypatch):
+    """After alembic upgrade head applying 0003:
+
+    - The documents row with user_id=None is deleted (D-01, D-02).
+    - The documents row with a real user_id is preserved.
+    - PRAGMA table_info shows documents.user_id notnull=1.
+    - All topics rows are deleted (D-10).
+    - ix_topics_user_id exists in sqlite_master.
+    - quotas.used_bytes for the populated user equals SUM(size_bytes).
+    """
+    try:
+        import alembic.command
+        from alembic.config import Config
+    except ImportError as exc:
+        pytest.skip(f"alembic not installed: {exc}")
+
+    import sqlite3
+    import uuid
+
+    db_path = tmp_path / "test_0003.db"
+    db_url = f"sqlite+aiosqlite:///{db_path}"
+
+    monkeypatch.setenv("DATABASE_MIGRATE_URL", db_url)
+
+    alembic_cfg = Config("alembic.ini")
+    alembic_cfg.set_main_option("sqlalchemy.url", db_url)
+
+    # Apply migrations up to 0002 so the schema is in the pre-0003 state
+    try:
+        alembic.command.upgrade(alembic_cfg, "0002")
+    except Exception as exc:
+        pytest.skip(f"alembic upgrade to 0002 failed: {exc}")
+
+    # Pre-seed test data using raw sqlite3 (synchronous)
+    user_id = str(uuid.uuid4())
+    doc_id_null = str(uuid.uuid4())
+    doc_id_user = str(uuid.uuid4())
+    topic_id = str(uuid.uuid4())
+
+    conn = sqlite3.connect(str(db_path))
+    try:
+        # Insert a user so we can seed a quota and a user-owned document
+        conn.execute(
+            "INSERT INTO users (id, handle, email, password_hash, role, is_active, "
+            "password_must_change, default_storage_backend) "
+            "VALUES (?, ?, ?, ?, ?, 1, 0, 'minio')",
+            (user_id, "testuser", "test@example.com", "hash", "user"),
+        )
+        conn.execute(
+            "INSERT INTO quotas (user_id, limit_bytes, used_bytes) VALUES (?, ?, ?)",
+            (user_id, 104857600, 0),
+        )
+        # Null-user document (to be deleted)
+        conn.execute(
+            "INSERT INTO documents (id, user_id, filename, object_key, content_type, "
+            "size_bytes, storage_backend, status) VALUES (?, NULL, ?, ?, ?, ?, 'minio', 'uploaded')",
+            (doc_id_null, "null_doc.txt", "null/key.txt", "text/plain", 1000),
+        )
+        # User-owned document (to be preserved)
+        conn.execute(
+            "INSERT INTO documents (id, user_id, filename, object_key, content_type, "
+            "size_bytes, storage_backend, status) VALUES (?, ?, ?, ?, ?, ?, 'minio', 'uploaded')",
+            (doc_id_user, user_id, "user_doc.txt", "user/key.txt", "text/plain", 2048),
+        )
+        # A topic row (all topics deleted in 0003)
+        conn.execute(
+            "INSERT INTO topics (id, user_id, name, description, color) "
+            "VALUES (?, NULL, ?, '', ?)",
+            (topic_id, "Finance", "#6366f1"),
+        )
+        conn.commit()
+    finally:
+        conn.close()
+
+    # Apply migration 0003 (no MinIO env set — MinIO step is skipped safely)
+    try:
+        alembic.command.upgrade(alembic_cfg, "0003")
+    except Exception as exc:
+        pytest.fail(f"alembic upgrade to 0003 failed: {exc}")
+
+    conn = sqlite3.connect(str(db_path))
+    try:
+        # 1. Null-user document must be gone
+        cursor = conn.execute(
+            "SELECT id FROM documents WHERE id = ?", (doc_id_null,)
+        )
+        assert cursor.fetchone() is None, "Null-user document was not deleted by migration 0003"
+
+        # 2. User-owned document must be preserved
+        cursor = conn.execute(
+            "SELECT id FROM documents WHERE id = ?", (doc_id_user,)
+        )
+        assert cursor.fetchone() is not None, "User-owned document was incorrectly deleted"
+
+        # 3. documents.user_id must now be NOT NULL
+        cursor = conn.execute("PRAGMA table_info(documents)")
+        columns = {row[1]: {"notnull": row[3]} for row in cursor.fetchall()}
+        assert columns["user_id"]["notnull"] == 1, (
+            "documents.user_id is still nullable after migration 0003"
+        )
+
+        # 4. All topics rows must be deleted
+        cursor = conn.execute("SELECT COUNT(*) FROM topics")
+        count = cursor.fetchone()[0]
+        assert count == 0, f"Expected 0 topics after migration 0003, found {count}"
+
+        # 5. ix_topics_user_id must exist
+        cursor = conn.execute(
+            "SELECT name FROM sqlite_master WHERE type='index' AND name='ix_topics_user_id'"
+        )
+        assert cursor.fetchone() is not None, "ix_topics_user_id index not created by migration 0003"
+
+        # 6. quotas.used_bytes must equal SUM(size_bytes) for the user
+        cursor = conn.execute(
+            "SELECT used_bytes FROM quotas WHERE user_id = ?", (user_id,)
+        )
+        row = cursor.fetchone()
+        assert row is not None, "Quota row not found for test user"
+        assert row[0] == 2048, (
+            f"quotas.used_bytes should be 2048 (SUM of user docs) but is {row[0]}"
+        )
+    finally:
+        conn.close()