From ad2102c1a81e1c5296530613b6c456aae98525cd Mon Sep 17 00:00:00 2001
From: qinling0210 <88864212+qinling0210@users.noreply.github.com>
Date: Tue, 28 Jul 2026 15:26:40 +0800
Subject: [PATCH] fix(infinity): parent_kwd incorrectly split as keyword list
 causing orphaned clusters (#17475)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

…rphaned clusters

### Summary

**Problem**
When deleting a document from the knowledge base, the parent cluster
could never be located and was therefore never cleaned up, leaving
orphaned clusters in Infinity indefinitely.

**Root Cause**
InfinityConnection.field_keyword() treats all _kwd-suffixed columns as
keyword lists and applies .split("###") to them in get_fields().
parent_kwd was missing from the exclusion list. This caused a parent
cluster name like "Toronto Transit & Authority Control 5e8d5eb6" to
become ['Toronto Transit & Authority Control 5e8d5eb6'] (a
single-element list). When that list was fed into _nav_cluster_id(), it
produced a different hash than the actual cluster's hash (computed from
the plain string), so every deletion-time cluster lookup silently
failed.

**Fix**
Added "parent_kwd" to the exclusion list in field_keyword(), alongside
docnm_kwd, important_kwd, question_kwd, etc.
---
 rag/utils/infinity_conn.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/rag/utils/infinity_conn.py b/rag/utils/infinity_conn.py
index f69324e07f..e2980f19e2 100644
--- a/rag/utils/infinity_conn.py
+++ b/rag/utils/infinity_conn.py
@@ -72,11 +72,11 @@ class InfinityConnection(InfinityConnectionBase):
 
     @staticmethod
     def field_keyword(field_name: str):
-        # Treat "*_kwd" tag-like columns as keyword lists except knowledge_graph_kwd; source_id is also keyword-like.
+        # Treat "*_kwd" tag-like columns as keyword lists except for the fields in the exclusion list; source_id is also keyword-like.
         # source_doc_ids / source_chunk_ids are multi-valued provenance lists (artifact/wiki rows) and must be
         # stored/read/updated as keyword lists so the delete-time ref-count (remove one id, drop row when empty) works.
         if field_name in ("source_id", "source_doc_ids", "source_chunk_ids") or (
-            field_name.endswith("_kwd") and field_name not in ["knowledge_graph_kwd", "docnm_kwd", "important_kwd", "question_kwd"]
+            field_name.endswith("_kwd") and field_name not in ["knowledge_graph_kwd", "docnm_kwd", "important_kwd", "question_kwd", "parent_kwd"]
         ):
             return True
         return False