From 2a8bdc2373531937f03fddf9b77f5f7ada35ae81 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E5=91=86=E8=90=8C=E9=97=B7=E6=B2=B9=E7=93=B6?=
 <lux@njuelectronics.com>
Date: Wed, 3 Jun 2026 15:10:55 +0800
Subject: [PATCH] fix: pydantic_core._pydantic_core.ValidationError: 2
 validation errors for DatasetDetailResponse (#36753)

---
 api/models/dataset.py                         | 11 ++++++-
 .../console/datasets/test_datasets.py         | 33 +++++++++++++++++++
 .../unit_tests/models/test_dataset_models.py  | 25 ++++++++++++++
 3 files changed, 68 insertions(+), 1 deletion(-)

diff --git a/api/models/dataset.py b/api/models/dataset.py
index 0f26752dd3..5c7a58d53f 100644
--- a/api/models/dataset.py
+++ b/api/models/dataset.py
@@ -322,6 +322,12 @@ class Dataset(Base):
 
     @property
     def retrieval_model_dict(self):
+        """Return a normalized retrieval model payload for API responses.
+
+        Older rows may only persist a partial retrieval model dict. Merge the
+        stored value over the current defaults so response validation still sees
+        the required baseline fields.
+        """
         default_retrieval_model = {
             "search_method": RetrievalMethod.SEMANTIC_SEARCH,
             "reranking_enable": False,
@@ -329,7 +335,10 @@ class Dataset(Base):
             "top_k": 2,
             "score_threshold_enabled": False,
         }
-        return self.retrieval_model or default_retrieval_model
+        if not self.retrieval_model:
+            return default_retrieval_model
+
+        return {**default_retrieval_model, **self.retrieval_model}
 
     @property
     def tags(self):
diff --git a/api/tests/unit_tests/controllers/console/datasets/test_datasets.py b/api/tests/unit_tests/controllers/console/datasets/test_datasets.py
index d9f3270bd0..ff1a664515 100644
--- a/api/tests/unit_tests/controllers/console/datasets/test_datasets.py
+++ b/api/tests/unit_tests/controllers/console/datasets/test_datasets.py
@@ -294,6 +294,39 @@ class TestDatasetList:
         assert status == 200
         assert resp["data"][0]["retrieval_model_dict"]["weights"]["weight_type"] is None
 
+    def test_get_merges_partial_retrieval_model_defaults(self, app: Flask):
+        api = DatasetListApi()
+        method = unwrap(api.get)
+
+        current_user = self._mock_user()
+        datasets = [make_dataset(retrieval_model={"top_k": 4, "score_threshold_enabled": False})]
+
+        with app.test_request_context("/datasets"):
+            with (
+                patch(
+                    "controllers.console.datasets.datasets.current_account_with_tenant",
+                    return_value=(current_user, "tenant-1"),
+                ),
+                patch.object(
+                    DatasetService,
+                    "get_datasets",
+                    return_value=(datasets, 1),
+                ),
+                patch.object(
+                    ProviderManager,
+                    "get_configurations",
+                    return_value=MagicMock(get_models=lambda **_: []),
+                ),
+            ):
+                resp, status = method(api)
+
+        assert status == 200
+        retrieval_model = resp["data"][0]["retrieval_model_dict"]
+        assert retrieval_model["search_method"] == "semantic_search"
+        assert retrieval_model["reranking_enable"] is False
+        assert retrieval_model["top_k"] == 4
+        assert retrieval_model["score_threshold_enabled"] is False
+
     def test_embedding_available_false(self, app: Flask):
         api = DatasetListApi()
         method = unwrap(api.get)
diff --git a/api/tests/unit_tests/models/test_dataset_models.py b/api/tests/unit_tests/models/test_dataset_models.py
index f4ccfb4191..44a4e6af98 100644
--- a/api/tests/unit_tests/models/test_dataset_models.py
+++ b/api/tests/unit_tests/models/test_dataset_models.py
@@ -219,6 +219,31 @@ class TestDatasetModelValidation:
         assert result["reranking_enable"] is False
         assert result["score_threshold_enabled"] is False
 
+    def test_dataset_retrieval_model_dict_property_merges_partial_values(self):
+        """Test retrieval_model_dict property fills in missing legacy keys."""
+        # Arrange
+        dataset = Dataset(
+            tenant_id=str(uuid4()),
+            name="Test Dataset",
+            data_source_type=DataSourceType.UPLOAD_FILE,
+            created_by=str(uuid4()),
+            retrieval_model={
+                "top_k": 4,
+                "score_threshold_enabled": True,
+                "score_threshold": 0.42,
+            },
+        )
+
+        # Act
+        result = dataset.retrieval_model_dict
+
+        # Assert
+        assert result["search_method"] == "semantic_search"
+        assert result["reranking_enable"] is False
+        assert result["top_k"] == 4
+        assert result["score_threshold_enabled"] is True
+        assert result["score_threshold"] == 0.42
+
     def test_dataset_gen_collection_name_by_id(self):
         """Test static method for generating collection name."""
         # Arrange