refactor: remove streaming structured output from invoke_llm_with_pydantic_model

Signed-off-by: Stream <Stream_2@qq.com>

api/core/llm_generator/llm_generator.py

@@ -471,7 +471,6 @@ class LLMGenerator:
                 prompt_messages=complete_messages,
                 output_model=CodeNodeStructuredOutput,
                 model_parameters=model_parameters,
-                stream=True,
                 tenant_id=tenant_id,
             )
 
@@ -553,16 +552,10 @@ class LLMGenerator:
 
         completion_params = model_config.get("completion_params", {}) if model_config else {}
         try:
-            response = invoke_llm_with_pydantic_model(
-                provider=model_instance.provider,
-                model_schema=model_schema,
-                model_instance=model_instance,
-                prompt_messages=prompt_messages,
-                output_model=SuggestedQuestionsOutput,
-                model_parameters=completion_params,
-                stream=True,
-                tenant_id=tenant_id,
-            )
+            response = invoke_llm_with_pydantic_model(provider=model_instance.provider, model_schema=model_schema,
+                                                      model_instance=model_instance, prompt_messages=prompt_messages,
+                                                      output_model=SuggestedQuestionsOutput,
+                                                      model_parameters=completion_params, tenant_id=tenant_id)
 
             return {"questions": response.questions, "error": ""}
 
@@ -842,15 +835,11 @@ Generate {language} code to extract/transform available variables for the target
         try:
             from core.llm_generator.output_parser.structured_output import invoke_llm_with_pydantic_model
 
-            response = invoke_llm_with_pydantic_model(
-                provider=model_instance.provider,
-                model_schema=model_schema,
-                model_instance=model_instance,
-                prompt_messages=list(prompt_messages),
-                output_model=InstructionModifyOutput,
-                model_parameters=model_parameters,
-                stream=True,
-            )
+            response = invoke_llm_with_pydantic_model(provider=model_instance.provider, model_schema=model_schema,
+                                                      model_instance=model_instance,
+                                                      prompt_messages=list(prompt_messages),
+                                                      output_model=InstructionModifyOutput,
+                                                      model_parameters=model_parameters)
             return response.model_dump(mode="python")
         except InvokeError as e:
             error = str(e)

api/core/llm_generator/output_parser/structured_output.py

@@ -262,7 +262,6 @@ def invoke_llm_with_pydantic_model(
     model_parameters: Mapping | None = None,
     tools: Sequence[PromptMessageTool] | None = None,
     stop: list[str] | None = None,
-    stream: bool = True,  # Some model plugin implementations don't support stream=False
     user: str | None = None,
     callbacks: list[Callback] | None = None,
     tenant_id: str | None = None,
@@ -281,36 +280,6 @@ def invoke_llm_with_pydantic_model(
     """
     json_schema = _schema_from_pydantic(output_model)
 
-    if stream:
-        result_generator = invoke_llm_with_structured_output(
-            provider=provider,
-            model_schema=model_schema,
-            model_instance=model_instance,
-            prompt_messages=prompt_messages,
-            json_schema=json_schema,
-            model_parameters=model_parameters,
-            tools=tools,
-            stop=stop,
-            stream=True,
-            user=user,
-            callbacks=callbacks,
-            tenant_id=tenant_id,
-        )
-
-        # Consume the generator to get the final chunk with structured_output
-        last_chunk: LLMResultChunkWithStructuredOutput | None = None
-        for chunk in result_generator:
-            last_chunk = chunk
-
-        if last_chunk is None:
-            raise OutputParserError("No chunks received from LLM")
-
-        structured_output = last_chunk.structured_output
-        if structured_output is None:
-            raise OutputParserError("Structured output is empty")
-
-        return _validate_structured_output(output_model, structured_output)
-
     result = invoke_llm_with_structured_output(
         provider=provider,
         model_schema=model_schema,

api/tests/unit_tests/utils/structured_output_parser/test_structured_output_parser.py

@@ -492,7 +492,6 @@ def test_structured_output_with_pydantic_model_non_streaming():
         model_instance=model_instance,
         prompt_messages=prompt_messages,
         output_model=ExampleOutput,
-        stream=False,
     )
 
     assert isinstance(result, ExampleOutput)
@@ -532,8 +531,7 @@ def test_structured_output_with_pydantic_model_streaming():
         model_schema=model_schema,
         model_instance=model_instance,
         prompt_messages=[UserPromptMessage(content="Return a JSON object with name.")],
-        output_model=ExampleOutput,
-        stream=True,
+        output_model=ExampleOutput
     )
 
     assert isinstance(result, ExampleOutput)
@@ -555,8 +553,7 @@ def test_structured_output_with_pydantic_model_validation_error():
             model_schema=model_schema,
             model_instance=model_instance,
             prompt_messages=[UserPromptMessage(content="test")],
-            output_model=ExampleOutput,
-            stream=False,
+            output_model=ExampleOutput
         )