Python: Add include_reference_source_data to agentic search requests

python/packages/azure-ai-search/agent_framework_azure_ai_search/_context_provider.py

@@ -60,6 +60,7 @@
         KnowledgeBaseRetrievalResponse,
         KnowledgeRetrievalIntent,
         KnowledgeRetrievalSemanticIntent,
+        SearchIndexKnowledgeSourceParams,
     )
     from azure.search.documents.knowledgebases.models import (
         KnowledgeRetrievalMinimalReasoningEffort as KBRetrievalMinimalReasoningEffort,
@@ -85,6 +86,7 @@
         KnowledgeBaseRetrievalResponse,
         KnowledgeRetrievalIntent,
         KnowledgeRetrievalSemanticIntent,
+        SearchIndexKnowledgeSourceParams,
     )
     from azure.search.documents.knowledgebases.models import (
         KnowledgeRetrievalMinimalReasoningEffort as KBRetrievalMinimalReasoningEffort,
@@ -576,6 +578,7 @@ def __init__(
             )
 
         self._knowledge_base_initialized = False
+        self._knowledge_source_names: list[str] = []
 
     def _common_client_kwargs(self) -> dict[str, Any]:
         """Build the keyword arguments shared by every Azure AI Search client.
@@ -786,6 +789,13 @@ async def _ensure_knowledge_base(self) -> None:
                     credential=self.credential,
                     **self._common_client_kwargs(),
                 )
+            # Resolve the existing KB's real knowledge source names so agentic
+            # retrieval can request reference source data per source. Without
+            # this, source names were left unset ("None-source").
+            self._knowledge_source_names = []
+            if self._index_client is not None:
+                kb = await self._index_client.get_knowledge_base(knowledge_base_name)
+                self._knowledge_source_names = [ks.name for ks in (kb.knowledge_sources or [])]
             self._knowledge_base_initialized = True
             return
 
@@ -799,6 +809,7 @@ async def _ensure_knowledge_base(self) -> None:
             raise ValueError("index_name is required when creating Knowledge Base from index")
 
         knowledge_source_name = f"{self.index_name}-source"
+        self._knowledge_source_names = [knowledge_source_name]
         try:
             await self._index_client.get_knowledge_source(knowledge_source_name)
         except ResourceNotFoundError:
@@ -882,6 +893,17 @@ async def _agentic_search(self, messages: list[Message]) -> list[Message]:
             # rejects on the stable/GA SDK.
             request_kwargs["messages"] = self._prepare_messages_for_kb_search(messages)
 
+        # Request reference source data per knowledge source so ref.source_data
+        # is populated when the source has source_data_fields configured (#5095).
+        if self._knowledge_source_names:
+            request_kwargs["knowledge_source_params"] = [
+                SearchIndexKnowledgeSourceParams(
+                    knowledge_source_name=name,
+                    include_reference_source_data=True,
+                )
+                for name in self._knowledge_source_names
+            ]
+
         retrieval_request = KnowledgeBaseRetrievalRequest(**request_kwargs)
 
         if not self._retrieval_client:

python/packages/azure-ai-search/tests/test_aisearch_context_provider.py

@@ -1505,6 +1505,57 @@ async def test_none_response_returns_default_message(self) -> None:
         assert len(results) == 1
         assert results[0].text == "No results found from Knowledge Base."
 
+    async def test_requests_reference_source_data_per_source(self) -> None:
+        # Regression for #5095: the retrieval request must carry
+        # knowledge_source_params with include_reference_source_data=True for
+        # each resolved knowledge source, otherwise ref.source_data is always
+        # None even when the source has source_data_fields configured.
+        provider = _make_provider()
+        provider._knowledge_base_initialized = True
+        provider.knowledge_base_name = "kb"
+        provider.retrieval_reasoning_effort = "minimal"
+        provider._knowledge_source_names = ["test-index-source"]
+
+        mock_result = Mock()
+        mock_result.response = []
+        mock_result.references = None
+
+        mock_retrieval = AsyncMock()
+        mock_retrieval.retrieve = AsyncMock(return_value=mock_result)
+        provider._retrieval_client = mock_retrieval
+
+        await provider._agentic_search([Message(role="user", contents=["query"])])
+
+        mock_retrieval.retrieve.assert_awaited_once()
+        request = mock_retrieval.retrieve.call_args.kwargs["retrieval_request"]
+        params = request.knowledge_source_params
+        assert params is not None
+        assert len(params) == 1
+        assert params[0].knowledge_source_name == "test-index-source"
+        assert params[0].include_reference_source_data is True
+
+    async def test_no_source_params_when_no_sources(self) -> None:
+        # When no knowledge source names are resolved, the request must not
+        # carry an empty/placeholder knowledge_source_params list.
+        provider = _make_provider()
+        provider._knowledge_base_initialized = True
+        provider.knowledge_base_name = "kb"
+        provider.retrieval_reasoning_effort = "minimal"
+        provider._knowledge_source_names = []
+
+        mock_result = Mock()
+        mock_result.response = []
+        mock_result.references = None
+
+        mock_retrieval = AsyncMock()
+        mock_retrieval.retrieve = AsyncMock(return_value=mock_result)
+        provider._retrieval_client = mock_retrieval
+
+        await provider._agentic_search([Message(role="user", contents=["query"])])
+
+        request = mock_retrieval.retrieve.call_args.kwargs["retrieval_request"]
+        assert request.knowledge_source_params is None
+
 
 # -- before_run: agentic mode --------------------------------------------------