cartography 0.116.1__py3-none-any.whl → 0.118.0__py3-none-any.whl

cartography/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '0.116.1'
-__version_tuple__ = version_tuple = (0, 116, 1)
+__version__ = version = '0.118.0'
+__version_tuple__ = version_tuple = (0, 118, 0)
 
 __commit_id__ = commit_id = None

cartography/cli.py

@@ -279,6 +279,17 @@ class CLI:
                 "Example: 'HIGH' will sync only HIGH and CRITICAL findings, filtering out LOW and MEDIUM severity findings."
             ),
         )
+        parser.add_argument(
+            "--experimental-aws-inspector-batch",
+            type=int,
+            default=1000,
+            help=(
+                "EXPERIMENTAL: This feature is experimental and may be removed in the future. "
+                "Batch size for AWS Inspector findings sync. Controls how many findings are fetched, processed and cleaned up at a time. "
+                "Default is 1000. Increase this value if you have a large number of findings and want to reduce API calls, "
+                "or decrease it if you're experiencing memory issues."
+            ),
+        )
         parser.add_argument(
             "--analysis-job-directory",
             type=str,

cartography/client/core/tx.py

@@ -6,6 +6,7 @@ from typing import Optional
 from typing import Tuple
 from typing import Union
 
+import backoff
 import neo4j
 
 from cartography.graph.querybuilder import build_create_index_queries
@@ -14,11 +15,31 @@ from cartography.graph.querybuilder import build_ingestion_query
 from cartography.graph.querybuilder import build_matchlink_query
 from cartography.models.core.nodes import CartographyNodeSchema
 from cartography.models.core.relationships import CartographyRelSchema
+from cartography.util import backoff_handler
 from cartography.util import batch
 
 logger = logging.getLogger(__name__)
 
 
+@backoff.on_exception(  # type: ignore
+    backoff.expo,
+    (
+        ConnectionResetError,
+        neo4j.exceptions.ServiceUnavailable,
+        neo4j.exceptions.SessionExpired,
+        neo4j.exceptions.TransientError,
+    ),
+    max_tries=5,
+    on_backoff=backoff_handler,
+)
+def _run_index_query_with_retry(neo4j_session: neo4j.Session, query: str) -> None:
+    """
+    Execute an index creation query with retry logic.
+    Index creation requires autocommit transactions and can experience transient errors.
+    """
+    neo4j_session.run(query)
+
+
 def run_write_query(
     neo4j_session: neo4j.Session, query: str, **parameters: Any
 ) -> None:
@@ -269,7 +290,7 @@ def ensure_indexes(
             raise ValueError(
                 'Query provided to `ensure_indexes()` does not start with "CREATE INDEX IF NOT EXISTS".',
             )
-        neo4j_session.run(query)
+        _run_index_query_with_retry(neo4j_session, query)
 
 
 def ensure_indexes_for_matchlinks(
@@ -288,7 +309,7 @@ def ensure_indexes_for_matchlinks(
             raise ValueError(
                 'Query provided to `ensure_indexes_for_matchlinks()` does not start with "CREATE INDEX IF NOT EXISTS".',
             )
-        neo4j_session.run(query)
+        _run_index_query_with_retry(neo4j_session, query)
 
 
 def load(

cartography/config.py

@@ -58,6 +58,9 @@ class Config:
     :type aws_guardduty_severity_threshold: str
     :param aws_guardduty_severity_threshold: GuardDuty severity threshold filter. Only findings at or above this
         severity level will be synced. Valid values: LOW, MEDIUM, HIGH, CRITICAL. Optional.
+    :type experimental_aws_inspector_batch: int
+    :param experimental_aws_inspector_batch: EXPERIMENTAL: Batch size for AWS Inspector findings sync. Controls how
+        many findings are fetched, processed and cleaned up at a time. Default is 1000. Optional.
     :type analysis_job_directory: str
     :param analysis_job_directory: Path to a directory tree containing analysis jobs to run. Optional.
     :type oci_sync_all_profiles: bool
@@ -195,6 +198,7 @@ class Config:
         aws_regions=None,
         aws_best_effort_mode=False,
         aws_cloudtrail_management_events_lookback_hours=None,
+        experimental_aws_inspector_batch=1000,
         azure_sync_all_subscriptions=False,
         azure_sp_auth=None,
         azure_tenant_id=None,
@@ -287,6 +291,7 @@ class Config:
         self.aws_cloudtrail_management_events_lookback_hours = (
             aws_cloudtrail_management_events_lookback_hours
         )
+        self.experimental_aws_inspector_batch = experimental_aws_inspector_batch
         self.azure_sync_all_subscriptions = azure_sync_all_subscriptions
         self.azure_sp_auth = azure_sp_auth
         self.azure_tenant_id = azure_tenant_id

cartography/graph/job.py

@@ -139,11 +139,13 @@ class GraphJob:
         cls,
         node_schema: CartographyNodeSchema,
         parameters: Dict[str, Any],
+        iterationsize: int = 100,
     ) -> "GraphJob":
         """
         Create a cleanup job from a CartographyNodeSchema object.
         For a given node, the fields used in the node_schema.sub_resource_relationship.target_node_node_matcher.keys()
         must be provided as keys and values in the params dict.
+        :param iterationsize: The number of items to process in each iteration. Defaults to 100.
         """
         queries: List[str] = build_cleanup_queries(node_schema)
 
@@ -165,7 +167,7 @@ class GraphJob:
                 query,
                 parameters=parameters,
                 iterative=True,
-                iterationsize=100,
+                iterationsize=iterationsize,
                 parent_job_name=node_schema.label,
                 parent_job_sequence_num=idx,
             )
@@ -185,6 +187,7 @@ class GraphJob:
         sub_resource_label: str,
         sub_resource_id: str,
         update_tag: int,
+        iterationsize: int = 100,
     ) -> "GraphJob":
         """
         Create a cleanup job from a CartographyRelSchema object (specifically, a MatchLink).
@@ -194,6 +197,7 @@ class GraphJob:
         - For a given rel_schema, the fields used in the rel_schema.properties._sub_resource_label.name and
         rel_schema.properties._sub_resource_id.name must be provided as keys and values in the params dict.
         - The rel_schema must have a source_node_matcher and target_node_matcher.
+        :param iterationsize: The number of items to process in each iteration. Defaults to 100.
         """
         cleanup_link_query = build_cleanup_query_for_matchlink(rel_schema)
         logger.debug(f"Cleanup query: {cleanup_link_query}")
@@ -208,7 +212,7 @@ class GraphJob:
             cleanup_link_query,
             parameters=parameters,
             iterative=True,
-            iterationsize=100,
+            iterationsize=iterationsize,
             parent_job_name=rel_schema.rel_label,
         )
 

cartography/graph/statement.py

@@ -52,6 +52,10 @@ class GraphStatement:
         self.parameters = parameters or {}
         self.iterative = iterative
         self.iterationsize = iterationsize
+        if iterationsize < 0:
+            raise ValueError(
+                f"iterationsize must be a positive integer, got {iterationsize}",
+            )
         self.parameters["LIMIT_SIZE"] = self.iterationsize
 
         self.parent_job_name = parent_job_name if parent_job_name else None

cartography/intel/aws/__init__.py

@@ -312,6 +312,7 @@ def start_aws_ingestion(neo4j_session: neo4j.Session, config: Config) -> None:
         "permission_relationships_file": config.permission_relationships_file,
         "aws_guardduty_severity_threshold": config.aws_guardduty_severity_threshold,
         "aws_cloudtrail_management_events_lookback_hours": config.aws_cloudtrail_management_events_lookback_hours,
+        "experimental_aws_inspector_batch": config.experimental_aws_inspector_batch,
     }
     try:
         boto3_session = boto3.Session()

cartography/intel/aws/apigateway.py

@@ -178,11 +178,24 @@ def get_rest_api_resources_methods_integrations(
                 method["apiId"] = api["id"]
                 method["httpMethod"] = http_method
                 methods.append(method)
-                integration = client.get_integration(
-                    restApiId=api["id"],
-                    resourceId=resource_id,
-                    httpMethod=http_method,
-                )
+                try:
+                    integration = client.get_integration(
+                        restApiId=api["id"],
+                        resourceId=resource_id,
+                        httpMethod=http_method,
+                    )
+                except ClientError as e:
+                    error_code = e.response.get("Error", {}).get("Code")
+                    if error_code == "NotFoundException":
+                        logger.warning(
+                            "No integration found for API %s resource %s method %s: %s",
+                            api["id"],
+                            resource_id,
+                            http_method,
+                            e,
+                        )
+                        continue
+                    raise
                 integration["resourceId"] = resource_id
                 integration["apiId"] = api["id"]
                 integration["integrationHttpMethod"] = integration.get("httpMethod")

cartography/intel/aws/ec2/elastic_ip_addresses.py

@@ -6,6 +6,7 @@ import boto3
 import neo4j
 from botocore.exceptions import ClientError
 
+from cartography.client.core.tx import run_write_query
 from cartography.util import aws_handle_regions
 from cartography.util import run_cleanup_job
 from cartography.util import timeit
@@ -83,7 +84,8 @@ def load_elastic_ip_addresses(
         SET r.lastupdated = $update_tag
     """
 
-    neo4j_session.run(
+    run_write_query(
+        neo4j_session,
         ingest_addresses,
         elastic_ip_addresses=elastic_ip_addresses,
         Region=region,

cartography/intel/aws/ec2/internet_gateways.py

@@ -5,6 +5,7 @@ from typing import List
 import boto3
 import neo4j
 
+from cartography.client.core.tx import run_write_query
 from cartography.util import aws_handle_regions
 from cartography.util import run_cleanup_job
 from cartography.util import timeit
@@ -63,13 +64,14 @@ def load_internet_gateways(
         SET r.lastupdated = $aws_update_tag
     """
 
-    neo4j_session.run(
+    run_write_query(
+        neo4j_session,
         query,
         internet_gateways=internet_gateways,
         region=region,
         aws_account_id=current_aws_account_id,
         aws_update_tag=update_tag,
-    ).consume()
+    )
 
 
 @timeit

cartography/intel/aws/ec2/load_balancer_v2s.py

@@ -6,6 +6,7 @@ import boto3
 import botocore
 import neo4j
 
+from cartography.client.core.tx import run_write_query
 from cartography.util import aws_handle_regions
 from cartography.util import run_cleanup_job
 from cartography.util import timeit
@@ -104,7 +105,8 @@ def load_load_balancer_v2s(
             logger.warning("Skipping load balancer entry with missing DNSName: %r", lb)
             continue
 
-        neo4j_session.run(
+        run_write_query(
+            neo4j_session,
             ingest_load_balancer_v2,
             ID=load_balancer_id,
             CREATED_TIME=str(lb["CreatedTime"]),
@@ -138,7 +140,8 @@ def load_load_balancer_v2s(
             SET r.lastupdated = $update_tag
             """
             for group in lb["SecurityGroups"]:
-                neo4j_session.run(
+                run_write_query(
+                    neo4j_session,
                     ingest_load_balancer_v2_security_group,
                     ID=load_balancer_id,
                     GROUP_ID=str(group),
@@ -182,7 +185,8 @@ def load_load_balancer_v2_subnets(
     SET r.lastupdated = $update_tag
     """
     for az in az_data:
-        neo4j_session.run(
+        run_write_query(
+            neo4j_session,
             ingest_load_balancer_subnet,
             ID=load_balancer_id,
             SubnetId=az["SubnetId"],
@@ -219,7 +223,8 @@ def load_load_balancer_v2_target_groups(
             continue
 
         for instance in target_group["Targets"]:
-            neo4j_session.run(
+            run_write_query(
+                neo4j_session,
                 ingest_instances,
                 ID=load_balancer_id,
                 INSTANCE_ID=instance,
@@ -253,7 +258,8 @@ def load_load_balancer_v2_listeners(
         ON CREATE SET r.firstseen = timestamp()
         SET r.lastupdated = $update_tag
     """
-    neo4j_session.run(
+    run_write_query(
+        neo4j_session,
         ingest_listener,
         LoadBalancerId=load_balancer_id,
         Listeners=listener_data,

cartography/intel/aws/ec2/network_interfaces.py

@@ -98,6 +98,10 @@ def transform_network_interface_data(
                 "SourceDestCheck": network_interface["SourceDestCheck"],
                 "Status": network_interface["Status"],
                 "SubnetId": network_interface["SubnetId"],
+                "AttachTime": network_interface.get("Attachment", {}).get("AttachTime"),
+                "DeviceIndex": network_interface.get("Attachment", {}).get(
+                    "DeviceIndex"
+                ),
                 "ElbV1Id": elb_v1_id,
                 "ElbV2Id": elb_v2_id,
             },

cartography/intel/aws/ec2/reserved_instances.py

@@ -6,6 +6,7 @@ import boto3
 import neo4j
 from botocore.exceptions import ClientError
 
+from cartography.client.core.tx import run_write_query
 from cartography.util import aws_handle_regions
 from cartography.util import run_cleanup_job
 from cartography.util import timeit
@@ -64,7 +65,8 @@ def load_reserved_instances(
         r_instance["Start"] = str(r_instance["Start"])
         r_instance["End"] = str(r_instance["End"])
 
-    neo4j_session.run(
+    run_write_query(
+        neo4j_session,
         ingest_reserved_instances,
         reserved_instances_list=data,
         AWS_ACCOUNT_ID=current_aws_account_id,

cartography/intel/aws/ec2/tgw.py

@@ -6,6 +6,7 @@ import boto3
 import botocore.exceptions
 import neo4j
 
+from cartography.client.core.tx import run_write_query
 from cartography.util import aws_handle_regions
 from cartography.util import run_cleanup_job
 from cartography.util import timeit
@@ -120,7 +121,8 @@ def load_transit_gateways(
     for tgw in data:
         tgw_id = tgw["TransitGatewayId"]
 
-        neo4j_session.run(
+        run_write_query(
+            neo4j_session,
             ingest_transit_gateway,
             TgwId=tgw_id,
             ARN=tgw["TransitGatewayArn"],
@@ -161,7 +163,8 @@ def _attach_shared_transit_gateway(
     """
 
     if tgw["OwnerId"] != current_aws_account_id:
-        neo4j_session.run(
+        run_write_query(
+            neo4j_session,
             attach_tgw,
             ARN=tgw["TransitGatewayArn"],
             TransitGatewayId=tgw["TransitGatewayId"],
@@ -202,7 +205,8 @@ def load_tgw_attachments(
     for tgwa in data:
         tgwa_id = tgwa["TransitGatewayAttachmentId"]
 
-        neo4j_session.run(
+        run_write_query(
+            neo4j_session,
             ingest_transit_gateway,
             TgwAttachmentId=tgwa_id,
             TransitGatewayId=tgwa["TransitGatewayId"],
@@ -261,7 +265,8 @@ def _attach_tgw_vpc_attachment_to_vpc_subnets(
     SET p.lastupdated = $update_tag
     """
 
-    neo4j_session.run(
+    run_write_query(
+        neo4j_session,
         attach_vpc_tgw_attachment_to_vpc,
         VpcId=tgw_vpc_attachment["VpcId"],
         TgwAttachmentId=tgw_vpc_attachment["TransitGatewayAttachmentId"],
@@ -269,7 +274,8 @@ def _attach_tgw_vpc_attachment_to_vpc_subnets(
     )
 
     for subnet_id in tgw_vpc_attachment["SubnetIds"]:
-        neo4j_session.run(
+        run_write_query(
+            neo4j_session,
             attach_vpc_tgw_attachment_to_subnet,
             SubnetId=subnet_id,
             TgwAttachmentId=tgw_vpc_attachment["TransitGatewayAttachmentId"],

cartography/intel/aws/ec2/volumes.py

@@ -70,7 +70,7 @@ def transform_volumes(
 
         for attachment in active_attachments:
             vol_with_attachment = raw_vol.copy()
-            vol_with_attachment["InstanceId"] = attachment["InstanceId"]
+            vol_with_attachment["InstanceId"] = attachment.get("InstanceId")
             result.append(vol_with_attachment)
 
     return result