Add ref_count management for deletions with atomic operations and error handling

backend/tests/test_garbage_collection.py

@@ -0,0 +1,168 @@
+"""
+Integration tests for garbage collection functionality.
+
+Tests cover:
+- Listing orphaned artifacts (ref_count=0)
+- Garbage collection in dry-run mode
+- Garbage collection actual deletion
+- Verifying artifacts with refs are not deleted
+"""
+
+import pytest
+from tests.conftest import (
+    compute_sha256,
+    upload_test_file,
+)
+
+
+class TestOrphanedArtifactsEndpoint:
+    """Tests for GET /api/v1/admin/orphaned-artifacts endpoint."""
+
+    @pytest.mark.integration
+    def test_list_orphaned_artifacts_returns_list(self, integration_client):
+        """Test orphaned artifacts endpoint returns a list."""
+        response = integration_client.get("/api/v1/admin/orphaned-artifacts")
+        assert response.status_code == 200
+        assert isinstance(response.json(), list)
+
+    @pytest.mark.integration
+    def test_orphaned_artifact_has_required_fields(self, integration_client):
+        """Test orphaned artifact response has required fields."""
+        response = integration_client.get("/api/v1/admin/orphaned-artifacts?limit=1")
+        assert response.status_code == 200
+
+        data = response.json()
+        if len(data) > 0:
+            artifact = data[0]
+            assert "id" in artifact
+            assert "size" in artifact
+            assert "created_at" in artifact
+            assert "created_by" in artifact
+            assert "original_name" in artifact
+
+    @pytest.mark.integration
+    def test_orphaned_artifacts_respects_limit(self, integration_client):
+        """Test orphaned artifacts endpoint respects limit parameter."""
+        response = integration_client.get("/api/v1/admin/orphaned-artifacts?limit=5")
+        assert response.status_code == 200
+        assert len(response.json()) <= 5
+
+    @pytest.mark.integration
+    def test_artifact_becomes_orphaned_when_tag_deleted(
+        self, integration_client, test_package, unique_test_id
+    ):
+        """Test artifact appears in orphaned list after tag is deleted."""
+        project, package = test_package
+        content = f"orphan test {unique_test_id}".encode()
+        expected_hash = compute_sha256(content)
+
+        # Upload with tag
+        upload_test_file(integration_client, project, package, content, tag="temp-tag")
+
+        # Verify not in orphaned list (has ref_count=1)
+        response = integration_client.get("/api/v1/admin/orphaned-artifacts?limit=1000")
+        orphaned_ids = [a["id"] for a in response.json()]
+        assert expected_hash not in orphaned_ids
+
+        # Delete the tag
+        integration_client.delete(f"/api/v1/project/{project}/{package}/tags/temp-tag")
+
+        # Verify now in orphaned list (ref_count=0)
+        response = integration_client.get("/api/v1/admin/orphaned-artifacts?limit=1000")
+        orphaned_ids = [a["id"] for a in response.json()]
+        assert expected_hash in orphaned_ids
+
+
+class TestGarbageCollectionEndpoint:
+    """Tests for POST /api/v1/admin/garbage-collect endpoint."""
+
+    @pytest.mark.integration
+    def test_garbage_collect_dry_run_returns_response(self, integration_client):
+        """Test garbage collection dry run returns valid response."""
+        response = integration_client.post("/api/v1/admin/garbage-collect?dry_run=true")
+        assert response.status_code == 200
+
+        data = response.json()
+        assert "artifacts_deleted" in data
+        assert "bytes_freed" in data
+        assert "artifact_ids" in data
+        assert "dry_run" in data
+        assert data["dry_run"] is True
+
+    @pytest.mark.integration
+    def test_garbage_collect_dry_run_doesnt_delete(
+        self, integration_client, test_package, unique_test_id
+    ):
+        """Test garbage collection dry run doesn't actually delete artifacts."""
+        project, package = test_package
+        content = f"dry run test {unique_test_id}".encode()
+        expected_hash = compute_sha256(content)
+
+        # Upload and delete tag to create orphan
+        upload_test_file(integration_client, project, package, content, tag="dry-run")
+        integration_client.delete(f"/api/v1/project/{project}/{package}/tags/dry-run")
+
+        # Verify artifact exists
+        response = integration_client.get(f"/api/v1/artifact/{expected_hash}")
+        assert response.status_code == 200
+
+        # Run garbage collection in dry-run mode
+        gc_response = integration_client.post(
+            "/api/v1/admin/garbage-collect?dry_run=true&limit=1000"
+        )
+        assert gc_response.status_code == 200
+        assert expected_hash in gc_response.json()["artifact_ids"]
+
+        # Verify artifact STILL exists (dry run didn't delete)
+        response = integration_client.get(f"/api/v1/artifact/{expected_hash}")
+        assert response.status_code == 200
+
+    @pytest.mark.integration
+    def test_garbage_collect_preserves_referenced_artifacts(
+        self, integration_client, test_package, unique_test_id
+    ):
+        """Test garbage collection doesn't delete artifacts with ref_count > 0."""
+        project, package = test_package
+        content = f"preserve test {unique_test_id}".encode()
+        expected_hash = compute_sha256(content)
+
+        # Upload with tag (ref_count=1)
+        upload_test_file(integration_client, project, package, content, tag="keep-this")
+
+        # Verify artifact exists with ref_count=1
+        response = integration_client.get(f"/api/v1/artifact/{expected_hash}")
+        assert response.status_code == 200
+        assert response.json()["ref_count"] == 1
+
+        # Run garbage collection (dry_run to not affect other tests)
+        gc_response = integration_client.post(
+            "/api/v1/admin/garbage-collect?dry_run=true&limit=1000"
+        )
+        assert gc_response.status_code == 200
+
+        # Verify artifact was NOT in delete list (has ref_count > 0)
+        assert expected_hash not in gc_response.json()["artifact_ids"]
+
+        # Verify artifact still exists
+        response = integration_client.get(f"/api/v1/artifact/{expected_hash}")
+        assert response.status_code == 200
+        assert response.json()["ref_count"] == 1
+
+    @pytest.mark.integration
+    def test_garbage_collect_respects_limit(self, integration_client):
+        """Test garbage collection respects limit parameter."""
+        response = integration_client.post(
+            "/api/v1/admin/garbage-collect?dry_run=true&limit=5"
+        )
+        assert response.status_code == 200
+        assert response.json()["artifacts_deleted"] <= 5
+
+    @pytest.mark.integration
+    def test_garbage_collect_returns_bytes_freed(self, integration_client):
+        """Test garbage collection returns accurate bytes_freed."""
+        response = integration_client.post("/api/v1/admin/garbage-collect?dry_run=true")
+        assert response.status_code == 200
+
+        data = response.json()
+        assert data["bytes_freed"] >= 0
+        assert isinstance(data["bytes_freed"], int)