feat: 高级编排支持文件上传(WIP)

apps/application/flow/step_node/image_understand_step_node/impl/base_image_understand_node.py

@@ -14,7 +14,7 @@
 
 def _write_context(node_variable: Dict, workflow_variable: Dict, node: INode, workflow, answer: str):
     chat_model = node_variable.get('chat_model')
-    message_tokens = chat_model.get_num_tokens_from_messages(node_variable.get('message_list'))
+    message_tokens = node_variable['usage_metadata']['output_tokens'] if 'usage_metadata' in node_variable else 0
     answer_tokens = chat_model.get_num_tokens(answer)
     node.context['message_tokens'] = message_tokens
     node.context['answer_tokens'] = answer_tokens

apps/setting/models_provider/impl/openai_model_provider/model/image.py

@@ -23,6 +23,7 @@ def new_instance(model_type, model_name, model_credential: Dict[str, object], **
             model=model_name,
             openai_api_base=model_credential.get('api_base'),
             openai_api_key=model_credential.get('api_key'),
+            stream_options={"include_usage": True},
             **optional_params,
         )
 

apps/setting/models_provider/impl/qwen_model_provider/model/image.py

@@ -2,22 +2,15 @@
 
 import base64
 import os
-from typing import List, Dict, Optional, Iterator, Any, cast
+from typing import Dict
 
 from langchain_community.chat_models import ChatOpenAI
-from langchain_community.llms.tongyi import generate_with_last_element_mark
-from langchain_core.callbacks import CallbackManagerForLLMRun
-from langchain_core.language_models import LanguageModelInput
-from langchain_core.messages import BaseMessage
 from langchain_core.messages import HumanMessage
-from langchain_core.outputs import ChatGenerationChunk, ChatGeneration
-from langchain_core.runnables import RunnableConfig, ensure_config
 
 from setting.models_provider.base_model_provider import MaxKBBaseModel
 
 
 class QwenVLChatModel(MaxKBBaseModel, ChatOpenAI):
-    usage_metadata: dict = {}
 
     @staticmethod
     def new_instance(model_type, model_name, model_credential: Dict[str, object], **model_kwargs):
@@ -44,12 +37,5 @@ def generate_message(prompt: str, image) -> list[HumanMessage]:
                     ])]
         return [HumanMessage(prompt)]
 
-    def get_last_generation_info(self) -> Optional[Dict[str, Any]]:
-        return self.usage_metadata
 
-    def get_num_tokens_from_messages(self, messages: List[BaseMessage]) -> int:
-        return self.usage_metadata.get('input_tokens', 0)
-
-    def get_num_tokens(self, text: str) -> int:
-        return self.usage_metadata.get('output_tokens', 0)