fix(api): bound attack paths normalized-list child IDs (#11960)

api/src/backend/tasks/jobs/attack_paths/sync.py

@@ -19,6 +19,7 @@ import json
 import time
 from collections import defaultdict
 from collections.abc import Iterator
+from hashlib import sha256
 from typing import Any
 
 import neo4j
@@ -392,11 +393,11 @@ def _build_child_props(
 def _build_child_id(provider_id: str, child_label: str, value_key: str) -> str:
     """Deterministic `_provider_element_id` for a list-item child node.
 
-    Dedupes within (tenant, provider): multiple parents referencing the same
-    value share one child node via the existing MERGE-on-_provider_element_id
-    index in both sinks.
+    Hashing the value keeps the ID bounded while preserving deduplication within
+    each provider and child label.
     """
-    return f"{provider_id}::{child_label}::{value_key}"
+    value_digest = sha256(value_key.encode("utf-8")).hexdigest()
+    return f"{provider_id}::{child_label}::{value_digest}"
 
 
 def _build_catalog_index(

api/src/backend/tasks/tests/test_attack_paths_scan.py

@@ -1828,6 +1828,55 @@ def _make_session_ctx(session, call_order=None, name=None):
     return ctx
 
 
+class TestBuildChildId:
+    def test_large_value_is_hashed_and_preserved_as_child_data(self):
+        value = "x" * 22_796
+        spec = sync_module.NormalizedList(
+            "SomeLabel",
+            "values",
+            "SomeLabelValuesItem",
+            "HAS_VALUES",
+        )
+        record = {
+            "element_id": "elem-1",
+            "labels": ["SomeLabel"],
+            "props": {"values": [value]},
+        }
+
+        _, parent, children, relationships = sync_module._node_to_sync_dict(
+            record,
+            "prov-1",
+            sync_module._build_catalog_index([spec]),
+        )
+
+        child = children[0]["row"]
+        child_id = child["provider_element_id"]
+        prefix = "prov-1::SomeLabelValuesItem::"
+        assert parent["provider_element_id"] == "prov-1:elem-1"
+        assert child["props"]["value"] == value
+        assert len(child_id) == len(prefix) + 64
+        assert value not in child_id
+        assert relationships[0]["row"]["end_element_id"] == child_id
+
+    @pytest.mark.parametrize(
+        ("provider_id", "child_label", "value_key"),
+        [
+            ("prov-2", "ChildLabel", "value"),
+            ("prov-1", "OtherChildLabel", "value"),
+            ("prov-1", "ChildLabel", "other-value"),
+        ],
+    )
+    def test_each_identity_component_changes_id(
+        self, provider_id, child_label, value_key
+    ):
+        child_id = sync_module._build_child_id("prov-1", "ChildLabel", "value")
+
+        assert sync_module._build_child_id("prov-1", "ChildLabel", "value") == child_id
+        assert (
+            sync_module._build_child_id(provider_id, child_label, value_key) != child_id
+        )
+
+
 class TestSyncNodes:
     def test_iter_sink_batches_rejects_zero_batch_size(self):
         with pytest.raises(