Add integration tests for deduplication and ref_count

- Add test_integration_uploads.py with 12 tests for duplicate upload scenarios
- Add test_ref_count.py with 7 tests for ref_count management
- Fix ArtifactDetailResponse to include sha256 and checksum fields
- Fix health check SQL warning by wrapping in text()
- Update tests to use unique content per test run for idempotency

backend/tests/test_ref_count.py

@@ -0,0 +1,176 @@
+"""
+Unit and integration tests for reference counting behavior.
+
+Tests cover:
+- ref_count is set correctly for new artifacts
+- ref_count increments on duplicate uploads
+- ref_count query correctly identifies existing artifacts
+- Artifact lookup by SHA256 hash works correctly
+"""
+
+import pytest
+import io
+from tests.conftest import (
+    compute_sha256,
+    upload_test_file,
+    TEST_CONTENT_HELLO,
+    TEST_HASH_HELLO,
+)
+
+
+class TestRefCountQuery:
+    """Tests for ref_count querying and artifact lookup."""
+
+    @pytest.mark.integration
+    def test_artifact_lookup_by_sha256(self, integration_client, test_package):
+        """Test artifact lookup by SHA256 hash (primary key) works correctly."""
+        project, package = test_package
+        content = b"unique content for lookup test"
+        expected_hash = compute_sha256(content)
+
+        # Upload a file
+        upload_result = upload_test_file(
+            integration_client, project, package, content, tag="v1"
+        )
+        assert upload_result["artifact_id"] == expected_hash
+
+        # Look up artifact by ID (SHA256)
+        response = integration_client.get(f"/api/v1/artifact/{expected_hash}")
+        assert response.status_code == 200
+
+        artifact = response.json()
+        assert artifact["id"] == expected_hash
+        assert artifact["sha256"] == expected_hash
+        assert artifact["size"] == len(content)
+
+    @pytest.mark.integration
+    def test_ref_count_query_identifies_existing_artifact(
+        self, integration_client, test_package
+    ):
+        """Test ref_count query correctly identifies existing artifacts by hash."""
+        project, package = test_package
+        content = b"content for ref count query test"
+        expected_hash = compute_sha256(content)
+
+        # Upload a file with a tag
+        upload_result = upload_test_file(
+            integration_client, project, package, content, tag="v1"
+        )
+
+        # Query artifact and check ref_count
+        response = integration_client.get(f"/api/v1/artifact/{expected_hash}")
+        assert response.status_code == 200
+
+        artifact = response.json()
+        assert artifact["ref_count"] >= 1  # At least 1 from the tag
+
+    @pytest.mark.integration
+    def test_ref_count_set_to_1_for_new_artifact_with_tag(
+        self, integration_client, test_package, unique_test_id
+    ):
+        """Test ref_count is set to 1 for new artifacts when created with a tag."""
+        project, package = test_package
+        content = f"brand new content for ref count test {unique_test_id}".encode()
+        expected_hash = compute_sha256(content)
+
+        # Upload a new file with a tag
+        upload_result = upload_test_file(
+            integration_client, project, package, content, tag="initial"
+        )
+
+        assert upload_result["artifact_id"] == expected_hash
+        assert upload_result["ref_count"] == 1
+        assert upload_result["deduplicated"] is False
+
+    @pytest.mark.integration
+    def test_ref_count_increments_on_duplicate_upload_with_tag(
+        self, integration_client, test_package, unique_test_id
+    ):
+        """Test ref_count is incremented when duplicate content is uploaded with a new tag."""
+        project, package = test_package
+        content = f"content that will be uploaded twice {unique_test_id}".encode()
+        expected_hash = compute_sha256(content)
+
+        # First upload with tag
+        result1 = upload_test_file(
+            integration_client, project, package, content, tag="v1"
+        )
+        assert result1["ref_count"] == 1
+        assert result1["deduplicated"] is False
+
+        # Second upload with different tag (same content)
+        result2 = upload_test_file(
+            integration_client, project, package, content, tag="v2"
+        )
+        assert result2["artifact_id"] == expected_hash
+        assert result2["ref_count"] == 2
+        assert result2["deduplicated"] is True
+
+    @pytest.mark.integration
+    def test_ref_count_after_multiple_tags(self, integration_client, test_package):
+        """Test ref_count correctly reflects number of tags pointing to artifact."""
+        project, package = test_package
+        content = b"content for multiple tag test"
+        expected_hash = compute_sha256(content)
+
+        # Upload with multiple tags
+        tags = ["v1", "v2", "v3", "latest"]
+        for i, tag in enumerate(tags):
+            result = upload_test_file(
+                integration_client, project, package, content, tag=tag
+            )
+            assert result["artifact_id"] == expected_hash
+            assert result["ref_count"] == i + 1
+
+        # Verify final ref_count via artifact endpoint
+        response = integration_client.get(f"/api/v1/artifact/{expected_hash}")
+        assert response.status_code == 200
+        assert response.json()["ref_count"] == len(tags)
+
+
+class TestRefCountWithDeletion:
+    """Tests for ref_count behavior when tags are deleted."""
+
+    @pytest.mark.integration
+    def test_ref_count_decrements_on_tag_delete(self, integration_client, test_package):
+        """Test ref_count decrements when a tag is deleted."""
+        project, package = test_package
+        content = b"content for delete test"
+        expected_hash = compute_sha256(content)
+
+        # Upload with two tags
+        upload_test_file(integration_client, project, package, content, tag="v1")
+        upload_test_file(integration_client, project, package, content, tag="v2")
+
+        # Verify ref_count is 2
+        response = integration_client.get(f"/api/v1/artifact/{expected_hash}")
+        assert response.json()["ref_count"] == 2
+
+        # Delete one tag
+        delete_response = integration_client.delete(
+            f"/api/v1/project/{project}/{package}/tags/v1"
+        )
+        assert delete_response.status_code == 204
+
+        # Verify ref_count is now 1
+        response = integration_client.get(f"/api/v1/artifact/{expected_hash}")
+        assert response.json()["ref_count"] == 1
+
+    @pytest.mark.integration
+    def test_ref_count_zero_after_all_tags_deleted(
+        self, integration_client, test_package
+    ):
+        """Test ref_count goes to 0 when all tags are deleted."""
+        project, package = test_package
+        content = b"content that will be orphaned"
+        expected_hash = compute_sha256(content)
+
+        # Upload with one tag
+        upload_test_file(integration_client, project, package, content, tag="only-tag")
+
+        # Delete the tag
+        integration_client.delete(f"/api/v1/project/{project}/{package}/tags/only-tag")
+
+        # Verify ref_count is 0
+        response = integration_client.get(f"/api/v1/artifact/{expected_hash}")
+        assert response.json()["ref_count"] == 0