[BugFix]Fix finish reason (#4543)

* fix finish reason

* add unit test

* add unit test

* fix unie test

* fix unit test

tests/entrypoints/openai/test_max_streaming_tokens.py

@@ -391,6 +391,7 @@ class TestMaxStreamingResponseTokens(IsolatedAsyncioTestCase):
                         "raw_prediction": "raw_answer_0",
                     },
                     "finished": True,
+                    "previous_num_tokens": 2,
                 },
                 "mock_request": ChatCompletionRequest(
                     model="test", messages=[], return_token_ids=True, max_tokens=10, n=2
@@ -417,6 +418,7 @@ class TestMaxStreamingResponseTokens(IsolatedAsyncioTestCase):
                         "raw_prediction": None,
                     },
                     "finished": True,
+                    "previous_num_tokens": 1,
                 },
                 "mock_request": ChatCompletionRequest(
                     model="test", messages=[], return_token_ids=True, max_tokens=5, n=2
@@ -435,7 +437,7 @@ class TestMaxStreamingResponseTokens(IsolatedAsyncioTestCase):
 
         prompt_token_ids = [1, 2]
         prompt_tokens = "test_prompt"
-        logprob_contents = [[], []]
+        logprob_contents = [[{"token": "hello", "logprob": 0.1}], [{"token": "hello", "logprob": 0.1}]]
         mock_response_processor = Mock()
         mock_response_processor.enable_multimodal_content.return_value = False
         completion_token_ids = [[], []]
@@ -443,8 +445,10 @@ class TestMaxStreamingResponseTokens(IsolatedAsyncioTestCase):
 
         for idx, case in enumerate(test_cases):
             actual_choice = await self.chat_serving._create_chat_completion_choice(
-                data=case["test_data"],
+                output=case["test_data"]["outputs"],
+                index=idx,
                 request=case["mock_request"],
+                previous_num_tokens=case["test_data"]["previous_num_tokens"],
                 prompt_token_ids=prompt_token_ids,
                 prompt_tokens=prompt_tokens,
                 completion_token_ids=completion_token_ids[idx],
@@ -465,6 +469,7 @@ class TestMaxStreamingResponseTokens(IsolatedAsyncioTestCase):
 
             self.assertEqual(num_cached_tokens[expected["index"]], expected["num_cached_tokens"])
             self.assertEqual(actual_choice.finish_reason, expected["finish_reason"])
+            assert actual_choice.logprobs is not None
 
 
 if __name__ == "__main__":