cartography 0.113.0__py3-none-any.whl → 0.115.0__py3-none-any.whl

cartography/intel/entra/app_role_assignments.py

@@ -0,0 +1,277 @@
+import gc
+from typing import Any
+from typing import AsyncGenerator
+
+import neo4j
+from azure.identity import ClientSecretCredential
+from msgraph import GraphServiceClient
+from msgraph.generated.models.app_role_assignment_collection_response import (
+    AppRoleAssignmentCollectionResponse,
+)
+
+from cartography.client.core.tx import load
+from cartography.client.core.tx import read_list_of_values_tx
+from cartography.client.core.tx import read_single_value_tx
+from cartography.graph.job import GraphJob
+from cartography.intel.entra.applications import APP_ROLE_ASSIGNMENTS_PAGE_SIZE
+from cartography.intel.entra.applications import logger
+from cartography.models.entra.app_role_assignment import EntraAppRoleAssignmentSchema
+from cartography.util import timeit
+
+
+@timeit
+async def get_app_role_assignments_for_app(
+    client: GraphServiceClient, neo4j_session: neo4j.Session, app_id: str
+) -> AsyncGenerator[dict[str, Any], None]:
+    """
+    Gets app role assignments for a single application by querying the graph for service principal ID.
+
+    :param client: GraphServiceClient
+    :param neo4j_session: Neo4j session for querying service principal
+    :param app_id: Application ID
+    :return: Generator of app role assignment data as dicts
+    """
+    logger.info(f"Fetching role assignments for application: {app_id}")
+
+    # Query the graph to get the service principal ID for this application
+    query = """
+    MATCH (sp:EntraServicePrincipal {app_id: $app_id})
+    RETURN sp.id as service_principal_id
+    """
+    service_principal_id = neo4j_session.execute_read(
+        read_single_value_tx, query, app_id=app_id
+    )
+
+    if not service_principal_id:
+        logger.warning(
+            f"No service principal found in graph for application {app_id}. Continuing."
+        )
+        return
+
+    # Get assignments for this service principal with pagination and limits
+    # Use maximum page size (999) to get more data per request
+    # Memory is managed through streaming and batching, not page size
+    request_config = client.service_principals.by_service_principal_id(
+        service_principal_id
+    ).app_role_assigned_to.AppRoleAssignedToRequestBuilderGetRequestConfiguration(
+        query_parameters=client.service_principals.by_service_principal_id(
+            service_principal_id
+        ).app_role_assigned_to.AppRoleAssignedToRequestBuilderGetQueryParameters(
+            top=APP_ROLE_ASSIGNMENTS_PAGE_SIZE  # Maximum allowed by Microsoft Graph API
+        )
+    )
+
+    assignments_page: AppRoleAssignmentCollectionResponse | None = (
+        await client.service_principals.by_service_principal_id(
+            service_principal_id
+        ).app_role_assigned_to.get(request_configuration=request_config)
+    )
+
+    assignment_count = 0
+    page_count = 0
+
+    while assignments_page:
+        page_count += 1
+
+        if assignments_page.value:
+            page_valid_count = 0
+            page_skipped_count = 0
+
+            # Process assignments and immediately yield to avoid accumulation
+            for assignment in assignments_page.value:
+                # Only yield if we have valid data since it's possible (but unlikely) for assignment.id to be None
+                if assignment.principal_id:
+                    assignment_count += 1
+                    page_valid_count += 1
+                    yield {
+                        "id": assignment.id,
+                        "app_role_id": assignment.app_role_id,
+                        "created_date_time": assignment.created_date_time,
+                        "principal_id": assignment.principal_id,
+                        "principal_display_name": assignment.principal_display_name,
+                        "principal_type": assignment.principal_type,
+                        "resource_display_name": assignment.resource_display_name,
+                        "resource_id": assignment.resource_id,
+                        "application_app_id": app_id,
+                    }
+                else:
+                    page_skipped_count += 1
+
+            # Log page results with details about skipped objects
+            if page_skipped_count > 0:
+                logger.warning(
+                    f"Page {page_count} for {app_id}: {page_valid_count} valid assignments, "
+                    f"{page_skipped_count} skipped objects. Total valid: {assignment_count}"
+                )
+            else:
+                logger.debug(
+                    f"Page {page_count} for {app_id}: {page_valid_count} assignments. "
+                    f"Total: {assignment_count}"
+                )
+
+            # Force garbage collection after each page
+            gc.collect()
+
+        # Check if we have more pages to fetch
+        if not assignments_page.odata_next_link:
+            break
+
+        # Clear previous page before fetching next
+        assignments_page.value = None
+
+        # Fetch next page
+        logger.debug(f"Fetching page {page_count + 1} of assignments for {app_id}")
+        next_page_url = assignments_page.odata_next_link
+        assignments_page = await client.service_principals.with_url(next_page_url).get()
+
+    logger.info(
+        f"Successfully retrieved {assignment_count} assignments for application {app_id} (pages: {page_count})"
+    )
+
+
+def transform_app_role_assignments(
+    assignments: list[dict[str, Any]],
+) -> list[dict[str, Any]]:
+    """
+    Transform app role assignment data for graph loading.
+
+    :param assignments: Raw app role assignment data as dicts
+    :return: Transformed assignment data for graph loading
+    """
+    transformed = []
+    for assign in assignments:
+        transformed.append(
+            {
+                "id": assign["id"],
+                "app_role_id": (
+                    str(assign["app_role_id"]) if assign["app_role_id"] else None
+                ),
+                "created_date_time": assign["created_date_time"],
+                "principal_id": (
+                    str(assign["principal_id"]) if assign["principal_id"] else None
+                ),
+                "principal_display_name": assign["principal_display_name"],
+                "principal_type": assign["principal_type"],
+                "resource_display_name": assign["resource_display_name"],
+                "resource_id": (
+                    str(assign["resource_id"]) if assign["resource_id"] else None
+                ),
+                "application_app_id": assign["application_app_id"],
+            }
+        )
+    return transformed
+
+
+@timeit
+def load_app_role_assignments(
+    neo4j_session: neo4j.Session,
+    assignments_data: list[dict[str, Any]],
+    update_tag: int,
+    tenant_id: str,
+) -> None:
+    """
+    Load Entra app role assignments to the graph.
+
+    :param neo4j_session: Neo4j session
+    :param assignments_data: Assignment data to load
+    :param update_tag: Update tag for tracking data freshness
+    :param tenant_id: Entra tenant ID
+    """
+    load(
+        neo4j_session,
+        EntraAppRoleAssignmentSchema(),
+        assignments_data,
+        lastupdated=update_tag,
+        TENANT_ID=tenant_id,
+    )
+
+
+@timeit
+def cleanup_app_role_assignments(
+    neo4j_session: neo4j.Session, common_job_parameters: dict[str, Any]
+) -> None:
+    """
+    Delete Entra app role assignments and their relationships from the graph if they were not updated in the last sync.
+
+    :param neo4j_session: Neo4j session
+    :param common_job_parameters: Common job parameters containing UPDATE_TAG and TENANT_ID
+    """
+    GraphJob.from_node_schema(
+        EntraAppRoleAssignmentSchema(), common_job_parameters
+    ).run(neo4j_session)
+
+
+@timeit
+async def sync_app_role_assignments(
+    neo4j_session: neo4j.Session,
+    tenant_id: str,
+    client_id: str,
+    client_secret: str,
+    update_tag: int,
+    common_job_parameters: dict[str, Any],
+) -> None:
+    """
+    Sync Entra app role assignments to the graph.
+
+    :param neo4j_session: Neo4j session
+    :param tenant_id: Entra tenant ID
+    :param client_id: Azure application client ID
+    :param client_secret: Azure application client secret
+    :param update_tag: Update tag for tracking data freshness
+    :param common_job_parameters: Common job parameters for cleanup
+    """
+    # Create credentials and client
+    credential = ClientSecretCredential(
+        tenant_id=tenant_id,
+        client_id=client_id,
+        client_secret=client_secret,
+    )
+
+    client = GraphServiceClient(
+        credential,
+        scopes=["https://graph.microsoft.com/.default"],
+    )
+    assignment_batch_size = 200  # Batch size for assignments
+    assignments_batch = []
+    total_assignment_count = 0
+
+    # Get app_ids from graph instead of streaming from API again
+    query = "MATCH (app:EntraApplication) RETURN app.app_id"
+    app_ids = neo4j_session.execute_read(read_list_of_values_tx, query)
+
+    for app_id in app_ids:
+        # Stream app role assignments (now using graph query for service principal ID)
+        async for assignment in get_app_role_assignments_for_app(
+            client, neo4j_session, app_id
+        ):
+            assignments_batch.append(assignment)
+            total_assignment_count += 1
+
+            # Transform and load assignments in batches
+            if len(assignments_batch) >= assignment_batch_size:
+                transformed_assignments = transform_app_role_assignments(
+                    assignments_batch
+                )
+                load_app_role_assignments(
+                    neo4j_session, transformed_assignments, update_tag, tenant_id
+                )
+                logger.debug(f"Loaded batch of {len(assignments_batch)} assignments")
+                assignments_batch.clear()
+                transformed_assignments.clear()
+
+                # Force garbage collection after batch load
+                gc.collect()
+
+    # Process remaining assignments
+    if assignments_batch:
+        transformed_assignments = transform_app_role_assignments(assignments_batch)
+        load_app_role_assignments(
+            neo4j_session, transformed_assignments, update_tag, tenant_id
+        )
+        assignments_batch.clear()
+        transformed_assignments.clear()
+
+    cleanup_app_role_assignments(neo4j_session, common_job_parameters)
+    logger.info(f"Completed syncing {total_assignment_count} app role assignments")
+    # Final garbage collection
+    gc.collect()

cartography/intel/entra/applications.py

@@ -6,16 +6,11 @@ from typing import Generator
 
 import neo4j
 from azure.identity import ClientSecretCredential
-from msgraph.generated.models.app_role_assignment_collection_response import (
-    AppRoleAssignmentCollectionResponse,
-)
 from msgraph.generated.models.application import Application
-from msgraph.generated.models.service_principal import ServicePrincipal
 from msgraph.graph_service_client import GraphServiceClient
 
 from cartography.client.core.tx import load
 from cartography.graph.job import GraphJob
-from cartography.models.entra.app_role_assignment import EntraAppRoleAssignmentSchema
 from cartography.models.entra.application import EntraApplicationSchema
 from cartography.util import timeit
 
@@ -65,125 +60,6 @@ async def get_entra_applications(
     logger.info(f"Retrieved {count} Entra applications total")
 
 
-@timeit
-async def get_app_role_assignments_for_app(
-    client: GraphServiceClient, app: Application
-) -> AsyncGenerator[dict[str, Any], None]:
-    """
-    Gets app role assignments for a single application with safety limits.
-
-    :param client: GraphServiceClient
-    :param app: Application object
-    :return: Generator of app role assignment data as dicts
-    """
-    if not app.app_id:
-        logger.warning(f"Application {app.id} has no app_id, skipping")
-        return
-
-    logger.info(
-        f"Fetching role assignments for application: {app.display_name} ({app.app_id})"
-    )
-
-    # First, get the service principal for this application
-    service_principals_page = await client.service_principals.get(
-        request_configuration=client.service_principals.ServicePrincipalsRequestBuilderGetRequestConfiguration(
-            query_parameters=client.service_principals.ServicePrincipalsRequestBuilderGetQueryParameters(
-                filter=f"appId eq '{app.app_id}'"
-            )
-        )
-    )
-
-    if not service_principals_page or not service_principals_page.value:
-        logger.warning(
-            f"No service principal found for application {app.app_id} ({app.display_name}). Continuing."
-        )
-        return
-
-    service_principal: ServicePrincipal = service_principals_page.value[0]
-
-    # Get assignments for this service principal with pagination and limits
-    # Use maximum page size (999) to get more data per request
-    # Memory is managed through streaming and batching, not page size
-    request_config = client.service_principals.by_service_principal_id(
-        service_principal.id
-    ).app_role_assigned_to.AppRoleAssignedToRequestBuilderGetRequestConfiguration(
-        query_parameters=client.service_principals.by_service_principal_id(
-            service_principal.id
-        ).app_role_assigned_to.AppRoleAssignedToRequestBuilderGetQueryParameters(
-            top=APP_ROLE_ASSIGNMENTS_PAGE_SIZE  # Maximum allowed by Microsoft Graph API
-        )
-    )
-
-    assignments_page: AppRoleAssignmentCollectionResponse | None = (
-        await client.service_principals.by_service_principal_id(
-            service_principal.id
-        ).app_role_assigned_to.get(request_configuration=request_config)
-    )
-
-    assignment_count = 0
-    page_count = 0
-
-    while assignments_page:
-        page_count += 1
-
-        if assignments_page.value:
-            page_valid_count = 0
-            page_skipped_count = 0
-
-            # Process assignments and immediately yield to avoid accumulation
-            for assignment in assignments_page.value:
-                # Only yield if we have valid data since it's possible (but unlikely) for assignment.id to be None
-                if assignment.principal_id:
-                    assignment_count += 1
-                    page_valid_count += 1
-                    yield {
-                        "id": assignment.id,
-                        "app_role_id": assignment.app_role_id,
-                        "created_date_time": assignment.created_date_time,
-                        "principal_id": assignment.principal_id,
-                        "principal_display_name": assignment.principal_display_name,
-                        "principal_type": assignment.principal_type,
-                        "resource_display_name": assignment.resource_display_name,
-                        "resource_id": assignment.resource_id,
-                        "application_app_id": app.app_id,
-                    }
-                else:
-                    page_skipped_count += 1
-
-            # Log page results with details about skipped objects
-            if page_skipped_count > 0:
-                logger.warning(
-                    f"Page {page_count} for {app.display_name}: {page_valid_count} valid assignments, "
-                    f"{page_skipped_count} skipped objects. Total valid: {assignment_count}"
-                )
-            else:
-                logger.debug(
-                    f"Page {page_count} for {app.display_name}: {page_valid_count} assignments. "
-                    f"Total: {assignment_count}"
-                )
-
-            # Force garbage collection after each page
-            gc.collect()
-
-        # Check if we have more pages to fetch
-        if not assignments_page.odata_next_link:
-            break
-
-        # Clear previous page before fetching next
-        assignments_page.value = None
-
-        # Fetch next page
-        logger.debug(
-            f"Fetching page {page_count + 1} of assignments for {app.display_name}"
-        )
-        next_page_url = assignments_page.odata_next_link
-        assignments_page = await client.service_principals.with_url(next_page_url).get()
-
-    logger.info(
-        f"Successfully retrieved {assignment_count} assignments for application {app.display_name} (pages: {page_count})"
-    )
-
-
 def transform_applications(
     applications: list[Application],
 ) -> Generator[dict[str, Any], None, None]:
@@ -203,39 +79,6 @@ def transform_applications(
         }
 
 
-def transform_app_role_assignments(
-    assignments: list[dict[str, Any]],
-) -> list[dict[str, Any]]:
-    """
-    Transform app role assignment data for graph loading.
-
-    :param assignments: Raw app role assignment data as dicts
-    :return: Transformed assignment data for graph loading
-    """
-    transformed = []
-    for assign in assignments:
-        transformed.append(
-            {
-                "id": assign["id"],
-                "app_role_id": (
-                    str(assign["app_role_id"]) if assign["app_role_id"] else None
-                ),
-                "created_date_time": assign["created_date_time"],
-                "principal_id": (
-                    str(assign["principal_id"]) if assign["principal_id"] else None
-                ),
-                "principal_display_name": assign["principal_display_name"],
-                "principal_type": assign["principal_type"],
-                "resource_display_name": assign["resource_display_name"],
-                "resource_id": (
-                    str(assign["resource_id"]) if assign["resource_id"] else None
-                ),
-                "application_app_id": assign["application_app_id"],
-            }
-        )
-    return transformed
-
-
 @timeit
 def load_applications(
     neo4j_session: neo4j.Session,
@@ -260,30 +103,6 @@ def load_applications(
     )
 
 
-@timeit
-def load_app_role_assignments(
-    neo4j_session: neo4j.Session,
-    assignments_data: list[dict[str, Any]],
-    update_tag: int,
-    tenant_id: str,
-) -> None:
-    """
-    Load Entra app role assignments to the graph.
-
-    :param neo4j_session: Neo4j session
-    :param assignments_data: Assignment data to load
-    :param update_tag: Update tag for tracking data freshness
-    :param tenant_id: Entra tenant ID
-    """
-    load(
-        neo4j_session,
-        EntraAppRoleAssignmentSchema(),
-        assignments_data,
-        lastupdated=update_tag,
-        TENANT_ID=tenant_id,
-    )
-
-
 @timeit
 def cleanup_applications(
     neo4j_session: neo4j.Session, common_job_parameters: dict[str, Any]
@@ -299,21 +118,6 @@ def cleanup_applications(
     )
 
 
-@timeit
-def cleanup_app_role_assignments(
-    neo4j_session: neo4j.Session, common_job_parameters: dict[str, Any]
-) -> None:
-    """
-    Delete Entra app role assignments and their relationships from the graph if they were not updated in the last sync.
-
-    :param neo4j_session: Neo4j session
-    :param common_job_parameters: Common job parameters containing UPDATE_TAG and TENANT_ID
-    """
-    GraphJob.from_node_schema(
-        EntraAppRoleAssignmentSchema(), common_job_parameters
-    ).run(neo4j_session)
-
-
 @timeit
 async def sync_entra_applications(
     neo4j_session: neo4j.Session,
@@ -345,18 +149,12 @@ async def sync_entra_applications(
         scopes=["https://graph.microsoft.com/.default"],
     )
 
-    # Process applications and their assignments in batches
+    # Step 1: Sync applications
     app_batch_size = 10  # Batch size for applications
-    assignment_batch_size = (
-        200  # Batch size for assignments (increased since we handle memory better now)
-    )
-
     apps_batch = []
-    assignments_batch = []
-    total_assignment_count = 0
     total_app_count = 0
 
-    # Stream apps
+    # Stream and load applications
     async for app in get_entra_applications(client):
         total_app_count += 1
         apps_batch.append(app)
@@ -372,45 +170,13 @@ async def sync_entra_applications(
             transformed_apps.clear()
             gc.collect()  # Force garbage collection
 
-        # Stream app role assignments
-        async for assignment in get_app_role_assignments_for_app(client, app):
-            assignments_batch.append(assignment)
-            total_assignment_count += 1
-
-            # Transform and load assignments in batches
-            if len(assignments_batch) >= assignment_batch_size:
-                transformed_assignments = transform_app_role_assignments(
-                    assignments_batch
-                )
-                load_app_role_assignments(
-                    neo4j_session, transformed_assignments, update_tag, tenant_id
-                )
-                logger.debug(f"Loaded batch of {len(assignments_batch)} assignments")
-                assignments_batch.clear()
-                transformed_assignments.clear()
-
-                # Force garbage collection after batch load
-                gc.collect()
-
     # Process remaining applications
     if apps_batch:
         transformed_apps = list(transform_applications(apps_batch))
         load_applications(neo4j_session, transformed_apps, update_tag, tenant_id)
         apps_batch.clear()
         transformed_apps.clear()
-
-    # Process remaining assignments
-    if assignments_batch:
-        transformed_assignments = transform_app_role_assignments(assignments_batch)
-        load_app_role_assignments(
-            neo4j_session, transformed_assignments, update_tag, tenant_id
-        )
-        assignments_batch.clear()
-        transformed_assignments.clear()
-
+    cleanup_applications(neo4j_session, common_job_parameters)
+    logger.info(f"Completed syncing {total_app_count} applications")
     # Final garbage collection
     gc.collect()
-
-    # Cleanup stale data
-    cleanup_applications(neo4j_session, common_job_parameters)
-    cleanup_app_role_assignments(neo4j_session, common_job_parameters)

cartography/intel/entra/federation/aws_identity_center.py

@@ -0,0 +1,77 @@
+from typing import Any
+
+import neo4j
+
+from cartography.client.core.tx import load_matchlinks
+from cartography.client.core.tx import read_list_of_dicts_tx
+from cartography.graph.job import GraphJob
+from cartography.models.entra.entra_user_to_aws_sso import (
+    EntraUserToAWSSSOUserMatchLink,
+)
+from cartography.util import timeit
+
+
+@timeit
+def sync_entra_to_aws_identity_center(
+    neo4j_session: neo4j.Session,
+    update_tag: int,
+    tenant_id: str,
+    common_job_parameters: dict[str, Any],
+) -> None:
+    query = """
+    MATCH (:EntraTenant{id: $TENANT_ID})-[:RESOURCE]->(e:EntraUser)
+          -[:HAS_APP_ROLE]->(ar:EntraAppRoleAssignment)
+          -[:ASSIGNED_TO]->(n:EntraApplication)
+          -[:SERVICE_PRINCIPAL]->(spn:EntraServicePrincipal)
+          -[:FEDERATES_TO]->(ic:AWSIdentityCenter)
+    MATCH (sso:AWSSSOUser{identity_store_id:ic.identity_store_id})
+    WHERE e.user_principal_name = sso.user_name
+    RETURN e.user_principal_name as entra_user_principal_name, sso.user_name as aws_user_name, sso.identity_store_id as identity_store_id
+    """
+    entrauser_to_awssso_users = neo4j_session.execute_read(
+        read_list_of_dicts_tx, query, TENANT_ID=tenant_id
+    )
+
+    # Load MatchLink relationships from Entra users to AWS SSO users
+    load_matchlinks(
+        neo4j_session,
+        EntraUserToAWSSSOUserMatchLink(),
+        entrauser_to_awssso_users,
+        lastupdated=update_tag,
+        _sub_resource_label="EntraTenant",
+        _sub_resource_id=tenant_id,
+    )
+
+    cleanup_entra_user_to_aws_sso_user_matchlinks(neo4j_session, common_job_parameters)
+
+
+@timeit
+async def sync_entra_federation(
+    neo4j_session: neo4j.Session,
+    update_tag: int,
+    tenant_id: str,
+    common_job_parameters: dict[str, Any],
+) -> None:
+    """
+    Sync Entra federation relationships to the graph.
+
+    :param neo4j_session: Neo4j session
+    :param update_tag: Update tag for tracking data freshness
+    :param tenant_id: Entra tenant ID
+    :param common_job_parameters: Common job parameters for cleanup
+    """
+    sync_entra_to_aws_identity_center(
+        neo4j_session, update_tag, tenant_id, common_job_parameters
+    )
+
+
+@timeit
+def cleanup_entra_user_to_aws_sso_user_matchlinks(
+    neo4j_session: neo4j.Session, common_job_parameters: dict[str, Any]
+) -> None:
+    GraphJob.from_matchlink(
+        EntraUserToAWSSSOUserMatchLink(),
+        "EntraTenant",
+        common_job_parameters["TENANT_ID"],
+        common_job_parameters["UPDATE_TAG"],
+    ).run(neo4j_session)