From 0b141fb952fe3996ffcba72b10100a4aa6d212b3 Mon Sep 17 00:00:00 2001
From: mengyonghao <1533512157@qq.com>
Date: Thu, 25 Dec 2025 14:47:22 +0800
Subject: [PATCH] fix(workflow): fix missing re-ranking for hybrid retrieval
 results in knowledge base node, add config example

---
 api/app/core/workflow/nodes/if_else/node.py     | 13 ++++++++++++-
 api/app/core/workflow/nodes/knowledge/config.py | 16 ++++++++++++++++
 api/app/core/workflow/nodes/knowledge/node.py   |  3 ++-
 3 files changed, 30 insertions(+), 2 deletions(-)

diff --git a/api/app/core/workflow/nodes/if_else/node.py b/api/app/core/workflow/nodes/if_else/node.py
index aedf0727..579c2840 100644
--- a/api/app/core/workflow/nodes/if_else/node.py
+++ b/api/app/core/workflow/nodes/if_else/node.py
@@ -158,10 +158,21 @@ class IfElseNode(BaseNode):
 
     async def execute(self, state: WorkflowState) -> Any:
         """
+        Execute the conditional branching logic of the node.
+
+        Evaluates the node's configured conditional expressions (expressions) in order.
+        Once a condition is satisfied, it returns the corresponding CASE identifier.
+        If none of the conditions match, it returns the default last CASE.
+
+        Args:
+            state (WorkflowState): The current workflow state, containing variables, messages, node outputs, etc.
+
+        Returns:
+            str: The matched branch identifier, e.g., 'CASE1', 'CASE2', ..., used for node transitions.
         """
         expressions = self.build_conditional_edge_expressions()
         for i in range(len(expressions)):
             logger.info(expressions[i])
             if self._evaluate_condition(expressions[i], state):
-                return f'CASE{i+1}'
+                return f'CASE{i + 1}'
         return f'CASE{len(expressions)}'
diff --git a/api/app/core/workflow/nodes/knowledge/config.py b/api/app/core/workflow/nodes/knowledge/config.py
index 530116ff..09c23855 100644
--- a/api/app/core/workflow/nodes/knowledge/config.py
+++ b/api/app/core/workflow/nodes/knowledge/config.py
@@ -36,3 +36,19 @@ class KnowledgeRetrievalNodeConfig(BaseNodeConfig):
         default=RetrieveType.PARTICIPLE,
         description="Retrieve type"
     )
+
+    class Config:
+        json_schema_extra = {
+            "examples": [
+                {
+                    "query": "{{sys.message}}",
+                    "kb_ids": [
+                        "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
+                    ],
+                    "similarity_threshold": 0.2,
+                    "vector_similarity_weight": 0.3,
+                    "top_k": 1,
+                    "retrieve_type": "hybrid"
+                }
+            ]
+        }
diff --git a/api/app/core/workflow/nodes/knowledge/node.py b/api/app/core/workflow/nodes/knowledge/node.py
index 10b877d8..319a0b88 100644
--- a/api/app/core/workflow/nodes/knowledge/node.py
+++ b/api/app/core/workflow/nodes/knowledge/node.py
@@ -163,5 +163,6 @@ class KnowledgeRetrievalNode(BaseNode):
                                                              indices=indices,
                                                              score_threshold=self.typed_config.similarity_threshold)
                     # Deduplicate hybrid retrieval results
-                    rs = self._deduplicate_docs(rs1, rs2)
+                    unique_rs = self._deduplicate_docs(rs1, rs2)
+                    rs = vector_service.rerank(query=query, docs=unique_rs, top_k=self.typed_config.top_k)
             return [chunk.model_dump() for chunk in rs]