fix(api): scope metadata endpoints to role provider groups

api/changelog.d/metadata-endpoints-provider-scope.security.md

@@ -0,0 +1 @@
+`GET /resources/metadata`, `GET /resources/metadata/latest`, `GET /findings/metadata` and `GET /findings/metadata/latest` now honour the provider groups of the caller role instead of aggregating every provider in the tenant

api/src/backend/api/tests/test_views.py

@@ -97,6 +97,11 @@ from rest_framework_simplejwt.token_blacklist.models import (
     OutstandingToken,
 )
 from rest_framework_simplejwt.tokens import RefreshToken
+from tasks.jobs.backfill import (
+    aggregate_scan_category_summaries,
+    aggregate_scan_resource_group_summaries,
+    backfill_resource_scan_summaries,
+)
 
 
 class TestViewSet:
@@ -6949,6 +6954,99 @@ class TestAttackPathsScanViewSet:
         assert response.status_code == status.HTTP_400_BAD_REQUEST
 
 
+def _client_restricted_to_provider(tenant, provider, factory):
+    """Authenticated client for a fresh member whose role only sees `provider`."""
+    user = User.objects.create_user(
+        name=f"restricted-{uuid4()}",
+        email=f"restricted-{uuid4()}@prowler.com",
+        password=TEST_PASSWORD,
+    )
+    Membership.objects.create(
+        user=user, tenant=tenant, role=Membership.RoleChoices.MEMBER
+    )
+    role = Role.objects.create(
+        name=f"restricted-{uuid4()}",
+        tenant_id=tenant.id,
+        manage_users=False,
+        manage_account=False,
+        manage_billing=False,
+        manage_providers=False,
+        manage_integrations=False,
+        manage_scans=False,
+        unlimited_visibility=False,
+    )
+    UserRoleRelationship.objects.create(user=user, role=role, tenant_id=tenant.id)
+    provider_group = ProviderGroup.objects.create(
+        name=f"restricted-{uuid4()}", tenant_id=tenant.id
+    )
+    ProviderGroupMembership.objects.create(
+        tenant_id=tenant.id, provider_group=provider_group, provider=provider
+    )
+    RoleProviderGroupRelationship.objects.create(
+        tenant_id=tenant.id, role=role, provider_group=provider_group
+    )
+    return factory(user, tenant)
+
+
+def _completed_scan_with_summaries(
+    tenant, provider, resources, *, categories, resource_group
+):
+    """Completed scan for `provider` with one finding over `resources`, summary tables filled."""
+    tenant_id = str(tenant.id)
+    scan = Scan.objects.create(
+        name=f"scan-{uuid4()}",
+        provider=provider,
+        trigger=Scan.TriggerChoices.MANUAL,
+        state=StateChoices.COMPLETED,
+        tenant_id=tenant_id,
+    )
+    finding = Finding.objects.create(
+        tenant_id=tenant_id,
+        uid=f"finding-{uuid4()}",
+        scan=scan,
+        delta="new",
+        status=Status.FAIL,
+        status_extended="test status",
+        impact=Severity.critical,
+        impact_extended="test impact",
+        severity=Severity.critical,
+        raw_result={"status": Status.FAIL},
+        check_id="scoped_check",
+        check_metadata={"CheckId": "scoped_check"},
+        categories=categories,
+        resource_groups=resource_group,
+        first_seen_at="2024-01-02T00:00:00Z",
+    )
+    finding.add_resources(resources)
+    backfill_resource_scan_summaries(tenant_id, str(scan.id))
+    aggregate_scan_category_summaries(tenant_id, str(scan.id))
+    aggregate_scan_resource_group_summaries(tenant_id, str(scan.id))
+    return scan
+
+
+@pytest.fixture
+def scans_per_provider_fixture(tenants_fixture, aws_provider_pair, resources_fixture):
+    """One completed scan per provider; provider 2 only owns the `test` typed resource."""
+    tenant = tenants_fixture[0]
+    provider1, provider2 = aws_provider_pair
+    resource1, resource2, resource3 = resources_fixture
+    _completed_scan_with_summaries(
+        tenant,
+        provider1,
+        [resource1, resource2],
+        categories=["iam"],
+        resource_group="identity",
+    )
+    _completed_scan_with_summaries(
+        tenant,
+        provider2,
+        [resource3],
+        categories=["data-protection"],
+        resource_group="data",
+    )
+    return tenant, provider2
+
+
 @pytest.mark.django_db
 class TestResourceViewSet:
     def test_resources_list_none(self, authenticated_client):
@@ -7320,6 +7418,42 @@ class TestResourceViewSet:
         assert attributes["types"] == [latest_scan_resource.type]
         assert "groups" in attributes
 
+    def test_resources_metadata_restricted_to_provider_group(
+        self, authenticated_client_for_tenant_factory, scans_per_provider_fixture
+    ):
+        tenant, provider2 = scans_per_provider_fixture
+        client = _client_restricted_to_provider(
+            tenant, provider2, authenticated_client_for_tenant_factory
+        )
+
+        response = client.get(
+            reverse("resource-metadata"), {"filter[updated_at]": TODAY}
+        )
+
+        assert response.status_code == status.HTTP_200_OK
+        attributes = response.json()["data"]["attributes"]
+        assert set(attributes["services"]) == {"ec2"}
+        assert set(attributes["regions"]) == {"us-east-1"}
+        assert set(attributes["types"]) == {"test"}
+        assert set(attributes["groups"]) == {"compute"}
+
+    def test_resources_metadata_latest_restricted_to_provider_group(
+        self, authenticated_client_for_tenant_factory, scans_per_provider_fixture
+    ):
+        tenant, provider2 = scans_per_provider_fixture
+        client = _client_restricted_to_provider(
+            tenant, provider2, authenticated_client_for_tenant_factory
+        )
+
+        response = client.get(reverse("resource-metadata_latest"))
+
+        assert response.status_code == status.HTTP_200_OK
+        attributes = response.json()["data"]["attributes"]
+        assert set(attributes["services"]) == {"ec2"}
+        assert set(attributes["regions"]) == {"us-east-1"}
+        assert set(attributes["types"]) == {"test"}
+        assert set(attributes["groups"]) == {"compute"}
+
     def test_resources_latest_filter_by_provider_id(
         self, authenticated_client, latest_scan_resource
     ):
@@ -8952,6 +9086,53 @@ class TestFindingViewSet:
         assert "groups" in attributes
         assert "ai_ml" in attributes["groups"]
 
+    @patch("api.v1.views.backfill_scan_resource_summaries_task.apply_async")
+    def test_findings_metadata_restricted_to_provider_group(
+        self,
+        mock_backfill,
+        authenticated_client_for_tenant_factory,
+        scans_per_provider_fixture,
+    ):
+        tenant, provider2 = scans_per_provider_fixture
+        client = _client_restricted_to_provider(
+            tenant, provider2, authenticated_client_for_tenant_factory
+        )
+
+        response = client.get(
+            reverse("finding-metadata"), {"filter[inserted_at]": TODAY}
+        )
+
+        assert response.status_code == status.HTTP_200_OK
+        mock_backfill.assert_not_called()
+        attributes = response.json()["data"]["attributes"]
+        assert set(attributes["services"]) == {"ec2"}
+        assert set(attributes["regions"]) == {"us-east-1"}
+        assert set(attributes["resource_types"]) == {"test"}
+        assert set(attributes["categories"]) == {"data-protection"}
+
+    @patch("api.v1.views.backfill_scan_resource_summaries_task.apply_async")
+    def test_findings_metadata_latest_restricted_to_provider_group(
+        self,
+        mock_backfill,
+        authenticated_client_for_tenant_factory,
+        scans_per_provider_fixture,
+    ):
+        tenant, provider2 = scans_per_provider_fixture
+        client = _client_restricted_to_provider(
+            tenant, provider2, authenticated_client_for_tenant_factory
+        )
+
+        response = client.get(reverse("finding-metadata_latest"))
+
+        assert response.status_code == status.HTTP_200_OK
+        mock_backfill.assert_not_called()
+        attributes = response.json()["data"]["attributes"]
+        assert set(attributes["services"]) == {"ec2"}
+        assert set(attributes["regions"]) == {"us-east-1"}
+        assert set(attributes["resource_types"]) == {"test"}
+        assert set(attributes["categories"]) == {"data-protection"}
+        assert set(attributes["groups"]) == {"data"}
+
     def test_findings_filter_by_category(
         self, authenticated_client, findings_with_categories
     ):

api/src/backend/api/v1/views.py

@@ -351,6 +351,15 @@ from tasks.tasks import (
 
 logger = logging.getLogger(BackendLogger.API)
 
+
+def _scans_visible_to(role, tenant_id):
+    """Tenant scans, narrowed to the role's provider groups unless it has unlimited visibility."""
+    scans = Scan.all_objects.filter(tenant_id=tenant_id)
+    if role.unlimited_visibility:
+        return scans
+    return scans.filter(provider__in=get_providers(role))
+
+
 CACHE_DECORATOR = cache_control(
     max_age=django_settings.CACHE_MAX_AGE,
     stale_while_revalidate=django_settings.CACHE_STALE_WHILE_REVALIDATE,
@@ -3544,8 +3553,15 @@ class ResourceViewSet(PaginateByPkMixin, BaseRLSViewSet):
 
         tenant_id = request.tenant_id
         query_params = request.query_params
+        user_roles = get_role(request.user, tenant_id)
 
         queryset = ResourceScanSummary.objects.filter(tenant_id=tenant_id)
+        if not user_roles.unlimited_visibility:
+            queryset = queryset.filter(
+                scan_id__in=_scans_visible_to(user_roles, tenant_id).values_list(
+                    "id", flat=True
+                )
+            )
 
         if scans := query_params.get("filter[scan__in]") or query_params.get(
             "filter[scan]"
@@ -3623,7 +3639,10 @@ class ResourceViewSet(PaginateByPkMixin, BaseRLSViewSet):
         all_groups = Resource.objects.filter(
             tenant_id=tenant_id,
             groups__isnull=False,
-        ).values_list("groups", flat=True)
+        )
+        if not user_roles.unlimited_visibility:
+            all_groups = all_groups.filter(provider__in=get_providers(user_roles))
+        all_groups = all_groups.values_list("groups", flat=True)
         groups = sorted(
             {g for groups_list in all_groups if groups_list for g in groups_list}
         )
@@ -3648,9 +3667,10 @@ class ResourceViewSet(PaginateByPkMixin, BaseRLSViewSet):
     def metadata_latest(self, request):
         tenant_id = request.tenant_id
         query_params = request.query_params
+        user_roles = get_role(request.user, tenant_id)
 
-        latest_scans_queryset = Scan.all_objects.filter(
-            tenant_id=tenant_id
+        latest_scans_queryset = _scans_visible_to(
+            user_roles, tenant_id
         ).latest_per_provider()
 
         queryset = ResourceScanSummary.objects.filter(
@@ -3691,7 +3711,10 @@ class ResourceViewSet(PaginateByPkMixin, BaseRLSViewSet):
         all_groups = Resource.objects.filter(
             tenant_id=tenant_id,
             groups__isnull=False,
-        ).values_list("groups", flat=True)
+        )
+        if not user_roles.unlimited_visibility:
+            all_groups = all_groups.filter(provider__in=get_providers(user_roles))
+        all_groups = all_groups.values_list("groups", flat=True)
         groups = sorted(
             {g for groups_list in all_groups if groups_list for g in groups_list}
         )
@@ -4102,8 +4125,14 @@ class FindingViewSet(PaginateByPkMixin, BaseRLSViewSet):
 
         tenant_id = request.tenant_id
         query_params = request.query_params
+        user_roles = get_role(request.user, tenant_id)
+        visible_scan_ids = _scans_visible_to(user_roles, tenant_id).values_list(
+            "id", flat=True
+        )
 
         queryset = ResourceScanSummary.objects.filter(tenant_id=tenant_id)
+        if not user_roles.unlimited_visibility:
+            queryset = queryset.filter(scan_id__in=visible_scan_ids)
         scan_based_filters = {}
         category_scan_filters = {}  # Filters for ScanCategorySummary
 
@@ -4161,7 +4190,10 @@ class FindingViewSet(PaginateByPkMixin, BaseRLSViewSet):
         if not queryset.exists():
             raw_scans_ids = Scan.objects.filter(
                 tenant_id=tenant_id, **scan_based_filters
-            ).values_list("id", "unique_resource_count")
+            )
+            if not user_roles.unlimited_visibility:
+                raw_scans_ids = raw_scans_ids.filter(id__in=visible_scan_ids)
+            raw_scans_ids = raw_scans_ids.values_list("id", "unique_resource_count")
             scan_ids = [
                 scan_id for scan_id, count in raw_scans_ids if count and count > 0
             ]
@@ -4203,11 +4235,13 @@ class FindingViewSet(PaginateByPkMixin, BaseRLSViewSet):
         )
 
         # Get categories from ScanCategorySummary using same scan filters
+        category_summaries = ScanCategorySummary.objects.filter(
+            tenant_id=tenant_id, **category_scan_filters
+        )
+        if not user_roles.unlimited_visibility:
+            category_summaries = category_summaries.filter(scan_id__in=visible_scan_ids)
         categories = list(
-            ScanCategorySummary.objects.filter(
-                tenant_id=tenant_id, **category_scan_filters
-            )
-            .values_list("category", flat=True)
+            category_summaries.values_list("category", flat=True)
             .distinct()
             .order_by("category")
         )
@@ -4262,9 +4296,10 @@ class FindingViewSet(PaginateByPkMixin, BaseRLSViewSet):
     def metadata_latest(self, request):
         tenant_id = request.tenant_id
         query_params = request.query_params
+        user_roles = get_role(request.user, tenant_id)
 
-        latest_scans_queryset = Scan.all_objects.filter(
-            tenant_id=tenant_id
+        latest_scans_queryset = _scans_visible_to(
+            user_roles, tenant_id
         ).latest_per_provider()
         raw_latest_scans_ids = list(
             latest_scans_queryset.values_list("id", "unique_resource_count")