From 75f117ae50481e711f41212e89bbe24e880a4723 Mon Sep 17 00:00:00 2001
From: RockChinQ <rockchinq@gmail.com>
Date: Sun, 6 Sep 2026 00:33:20 +0800
Subject: [PATCH] fix(provider): preserve explicit streaming finish reasons

---
 .../modelmgr/requesters/litellmchat.py        |  3 ++
 tests/unit_tests/provider/test_litellmchat.py | 28 +++++++++++++++++++
 2 files changed, 31 insertions(+)

diff --git a/src/langbot/pkg/provider/modelmgr/requesters/litellmchat.py b/src/langbot/pkg/provider/modelmgr/requesters/litellmchat.py
index 1d3adcee5..8800caea2 100644
--- a/src/langbot/pkg/provider/modelmgr/requesters/litellmchat.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/litellmchat.py
@@ -1278,6 +1278,9 @@ class LiteLLMRequester(requester.ProviderAPIRequester):
                 delta_content = delta.get('content') or ''
                 reasoning_content = delta.get('reasoning_content') or ''
                 provider_fields = dict(delta.get('provider_specific_fields') or {})
+                if finish_reason:
+                    # Preserve an explicit provider stop even when the final delta is empty.
+                    provider_fields['finish_reason'] = finish_reason
                 raw_thinking_blocks = delta.get('thinking_blocks')
                 if raw_thinking_blocks:
                     thinking_blocks_state = self._merge_thinking_blocks(thinking_blocks_state, raw_thinking_blocks)
diff --git a/tests/unit_tests/provider/test_litellmchat.py b/tests/unit_tests/provider/test_litellmchat.py
index d684ca9e5..490f7aa12 100644
--- a/tests/unit_tests/provider/test_litellmchat.py
+++ b/tests/unit_tests/provider/test_litellmchat.py
@@ -199,6 +199,34 @@ class TestInvokeLLMStreamUsage:
             chunk.choices = []
         return chunk
 
+    @pytest.mark.asyncio
+    @pytest.mark.parametrize('finish_reason', ['stop', 'length', 'content_filter', 'tool_calls'])
+    async def test_empty_final_delta_preserves_provider_finish_reason(self, finish_reason):
+        import langbot_plugin.api.entities.builtin.provider.message as provider_message
+
+        mock_ap = Mock()
+        mock_ap.tool_mgr.generate_tools_for_openai = AsyncMock(return_value=None)
+        requester = litellmchat.LiteLLMRequester(ap=mock_ap, config={})
+        model = MockRuntimeModel('gpt-4o', 'test-api-key')
+
+        async def stream():
+            yield self._make_chunk(finish_reason=finish_reason)
+
+        with patch.object(litellmchat, 'acompletion', new=AsyncMock(side_effect=lambda **kw: stream())):
+            chunks = [
+                chunk
+                async for chunk in requester.invoke_llm_stream(
+                    query=None,
+                    model=model,
+                    messages=[provider_message.Message(role='user', content='Hi')],
+                )
+            ]
+
+        assert len(chunks) == 1
+        assert chunks[0].is_final
+        assert not chunks[0].content
+        assert chunks[0].provider_specific_fields['finish_reason'] == finish_reason
+
     @pytest.mark.asyncio
     async def test_stream_usage_with_nonempty_choices(self):
         """Usage chunk that still has a choice must populate _stream_usage."""