⚡ Bolt: Optimize dict.setdefault usage to prevent memory allocation overhead

.jules/bolt.md

@@ -10,3 +10,8 @@
 ## 2024-05-24 - Memoizing inline array maps
 **Learning:** Inline mapping of arrays inside JSX in large React components causes O(N) recalculation on every render.
 **Action:** Wrap inline JSX elements that map over arrays (e.g., lists of tasks) in a `useMemo` hook with specific dependencies.
+## 2025-02-12 - Python `dict.setdefault(..., []).append(...)` 메모리 할당 오버헤드
+
+**Learning:** 파이썬에서 반복문 내에 `dict.setdefault(key, []).append(value)`를 사용하면, 키의 존재 여부와 상관없이 매 반복마다 빈 리스트 `[]`를 새로 메모리에 할당하게 됩니다. 큰 컬렉션이나 성능이 중요한 루프에서 이러한 패턴은 심각한 메모리 할당 오버헤드와 속도 저하를 일으킵니다.
+
+**Action:** 요소를 그룹화하거나 리스트를 초기화할 때 `setdefault(key, [])`를 사용하지 말고, 항상 `collections.defaultdict(list)`를 생성한 후 `dict[key].append(value)`를 사용하여 불필요한 메모리 할당을 방지해야 합니다.

backend/api/data.py

@@ -1,3 +1,4 @@
+from collections import defaultdict
 import base64
 import binascii
 from datetime import datetime, timezone
@@ -1692,17 +1693,18 @@ def _risk_matrix_key_part(value: str) -> str:
 def _diligence_risk_matrix(
     snapshot: DataEvidenceSnapshotResponse,
 ) -> list[DataDiligenceRiskMatrixEntry]:
+    # ⚡ Bolt: Use defaultdict(list) instead of dict.setdefault(..., []).append(...) to avoid O(N) empty list allocation overhead inside the loop.
     groups: dict[
         tuple[RemediationPriority, str, str],
         list[DataDiligenceExceptionRegisterEntry],
-    ] = {}
+    ] = defaultdict(list)
     for exception in snapshot.diligence_exception_register:
         key = (
             exception.severity_code,
             exception.owner_area,
             exception.related_artifact,
         )
-        groups.setdefault(key, []).append(exception)
+        groups[key].append(exception)
 
     entries: list[DataDiligenceRiskMatrixEntry] = []
     for (severity, owner_area, related_artifact), exceptions in sorted(
@@ -1844,9 +1846,10 @@ def _diligence_close_decision_summary(
 def _diligence_close_artifact_review_queue(
     snapshot: DataEvidenceSnapshotResponse,
 ) -> list[DataDiligenceCloseArtifactReviewQueueEntry]:
-    groups: dict[str, list[DataDiligenceCloseProofPlanEntry]] = {}
+    # ⚡ Bolt: Use defaultdict(list) instead of dict.setdefault(..., []).append(...) to avoid O(N) empty list allocation overhead inside the loop.
+    groups: dict[str, list[DataDiligenceCloseProofPlanEntry]] = defaultdict(list)
     for proof in snapshot.diligence_close_proof_plan:
-        groups.setdefault(proof.required_proof_artifact, []).append(proof)
+        groups[proof.required_proof_artifact].append(proof)
 
     entries: list[DataDiligenceCloseArtifactReviewQueueEntry] = []
     for artifact, proofs in sorted(groups.items()):
@@ -1889,9 +1892,10 @@ def _diligence_close_artifact_review_queue(
 def _diligence_close_owner_handoff_queue(
     snapshot: DataEvidenceSnapshotResponse,
 ) -> list[DataDiligenceCloseOwnerHandoffQueueEntry]:
-    groups: dict[str, list[DataDiligenceCloseProofPlanEntry]] = {}
+    # ⚡ Bolt: Use defaultdict(list) instead of dict.setdefault(..., []).append(...) to avoid O(N) empty list allocation overhead inside the loop.
+    groups: dict[str, list[DataDiligenceCloseProofPlanEntry]] = defaultdict(list)
     for proof in snapshot.diligence_close_proof_plan:
-        groups.setdefault(proof.owner_area, []).append(proof)
+        groups[proof.owner_area].append(proof)
 
     entries: list[DataDiligenceCloseOwnerHandoffQueueEntry] = []
     for owner_area, proofs in sorted(groups.items()):

backend/services/project_graph/project_registration.py

@@ -1,4 +1,5 @@
 from __future__ import annotations
+from collections import defaultdict
 
 import datetime
 import hashlib
@@ -612,9 +613,10 @@ def _candidate_groups(
     *,
     scope: ProjectGraphQueryScope,
 ) -> tuple[_CandidateGroup, ...]:
-    records_by_email: dict[int, list[ProjectGraphObjectRecord]] = {}
+    # ⚡ Bolt: Use defaultdict(list) instead of dict.setdefault(..., []).append(...) to avoid O(N) empty list allocation overhead inside the loop.
+    records_by_email: dict[int, list[ProjectGraphObjectRecord]] = defaultdict(list)
     for record in records:
-        records_by_email.setdefault(record.email_id, []).append(record)
+        records_by_email[record.email_id].append(record)
 
     groups: list[_CandidateGroup] = []
     for group_records in records_by_email.values():
@@ -842,9 +844,10 @@ def _relation_summary(
     ``relation_count`` descending with a ``relation_type``-ascending tie-break so
     the result is deterministic regardless of relation iteration order.
     """
-    grouped: dict[str, list[ProjectTraceRelation]] = {}
+    # ⚡ Bolt: Use defaultdict(list) instead of dict.setdefault(..., []).append(...) to avoid O(N) empty list allocation overhead inside the loop.
+    grouped: dict[str, list[ProjectTraceRelation]] = defaultdict(list)
     for relation in relations:
-        grouped.setdefault(relation.relation_type, []).append(relation)
+        grouped[relation.relation_type].append(relation)
     type_summaries = [
         ProjectRelationTypeSummary(
             relation_type=relation_type,