feat: 适配completion和chat_completions

pkg/openai/api/chat_completion.py

@@ -1,4 +1,5 @@
 import openai
+from openai.types.chat import chat_completion_message
 import json
 import logging
 
@@ -13,13 +14,14 @@ class ChatCompletionRequest(RequestBase):
     此类保证每一次返回的角色为assistant的信息的finish_reason一定为stop。
     若有函数调用响应，本类的返回瀑布是：函数调用请求->函数调用结果->...->assistant的信息->stop。
     """
+
     model: str
     messages: list[dict[str, str]]
     kwargs: dict
 
     stopped: bool = False
 
-    pending_func_call: dict = None
+    pending_func_call: chat_completion_message.FunctionCall = None
 
     pending_msg: str
 
@@ -46,16 +48,18 @@ class ChatCompletionRequest(RequestBase):
 
     def __init__(
         self,
+        client: openai.Client,
         model: str,
         messages: list[dict[str, str]],
         **kwargs
     ):
+        self.client = client
         self.model = model
         self.messages = messages.copy()
 
         self.kwargs = kwargs
 
-        self.req_func = openai.ChatCompletion.acreate
+        self.req_func = self.client.chat.completions.create
 
         self.pending_func_call = None
 
@@ -84,39 +88,48 @@ class ChatCompletionRequest(RequestBase):
 
             # 拼接kwargs
             args = {**args, **self.kwargs}
+            
+            from openai.types.chat import chat_completion
 
-            resp = self._req(**args)
+            resp: chat_completion.ChatCompletion = self._req(**args)
 
-            choice0 = resp["choices"][0]
+            choice0 = resp.choices[0]
 
             # 如果不是函数调用，且finish_reason为stop，则停止迭代
-            if choice0['finish_reason'] == 'stop':  #  and choice0["finish_reason"] == "stop"
+            if choice0.finish_reason == 'stop':  #  and choice0["finish_reason"] == "stop"
                 self.stopped = True
             
-            if 'function_call' in choice0['message']:
+            if hasattr(choice0.message, 'function_call') and choice0.message.function_call is not None:
-                self.pending_func_call = choice0['message']['function_call']
+                self.pending_func_call = choice0.message.function_call
 
                 self.append_message(
                     role="assistant",
-                    content=choice0['message']['content'],
+                    content=choice0.message.content,
-                    function_call=choice0['message']['function_call']
+                    function_call=choice0.message.function_call
                 )
 
                 return {
-                    "id": resp["id"],
+                    "id": resp.id,
                     "choices": [
                         {
-                            "index": choice0["index"],
+                            "index": choice0.index,
                             "message": {
                                 "role": "assistant",
                                 "type": "function_call",
-                                "content": choice0['message']['content'],
+                                "content": choice0.message.content,
-                                "function_call": choice0['message']['function_call']
+                                "function_call": {
+                                    "name": choice0.message.function_call.name,
+                                    "arguments": choice0.message.function_call.arguments
+                                }
                             },
                             "finish_reason": "function_call"
                         }
                     ],
-                    "usage": resp["usage"]
+                    "usage": {
+                        "prompt_tokens": resp.usage.prompt_tokens,
+                        "completion_tokens": resp.usage.completion_tokens,
+                        "total_tokens": resp.usage.total_tokens
+                    }
                 }
             else:
 
@@ -124,19 +137,23 @@ class ChatCompletionRequest(RequestBase):
                 # 普通回复一定处于最后方，故不用再追加进内部messages
 
                 return {
-                    "id": resp["id"],
+                    "id": resp.id,
                     "choices": [
                         {
-                            "index": choice0["index"],
+                            "index": choice0.index,
                             "message": {
                                 "role": "assistant",
                                 "type": "text",
-                                "content": choice0['message']['content']
+                                "content": choice0.message.content
                             },
-                            "finish_reason": choice0["finish_reason"]
+                            "finish_reason": choice0.finish_reason
                         }
                     ],
-                    "usage": resp["usage"]
+                    "usage": {
+                        "prompt_tokens": resp.usage.prompt_tokens,
+                        "completion_tokens": resp.usage.completion_tokens,
+                        "total_tokens": resp.usage.total_tokens
+                    }
                 }
         else:  # 处理函数调用请求
 
@@ -144,20 +161,20 @@ class ChatCompletionRequest(RequestBase):
             
             self.pending_func_call = None
 
-            func_name = cp_pending_func_call['name']
+            func_name = cp_pending_func_call.name
             arguments = {}
 
             try:
 
                 try:
-                    arguments = json.loads(cp_pending_func_call['arguments'])
+                    arguments = json.loads(cp_pending_func_call.arguments)
                 # 若不是json格式的异常处理
                 except json.decoder.JSONDecodeError:
                     # 获取函数的参数列表
                     func_schema = get_func_schema(func_name)
 
                     arguments = {
-                        func_schema['parameters']['required'][0]: cp_pending_func_call['arguments']
+                        func_schema['parameters']['required'][0]: cp_pending_func_call.arguments
                     }
                 
                 logging.info("执行函数调用: name={}, arguments={}".format(func_name, arguments))

pkg/openai/api/completion.py

@@ -1,4 +1,5 @@
 import openai
+from openai.types import completion, completion_choice
 
 from .model import RequestBase
 
@@ -17,10 +18,12 @@ class CompletionRequest(RequestBase):
 
     def __init__(
         self,
+        client: openai.Client,
         model: str,
         messages: list[dict[str, str]],
         **kwargs
     ):
+        self.client = client
         self.model = model
         self.prompt = ""
         
@@ -31,7 +34,7 @@ class CompletionRequest(RequestBase):
 
         self.kwargs = kwargs
 
-        self.req_func = openai.Completion.acreate
+        self.req_func = self.client.completions.create
 
     def __iter__(self):
         return self
@@ -63,49 +66,35 @@ class CompletionRequest(RequestBase):
         if self.stopped:
             raise StopIteration()
 
-        resp = self._req(
+        resp: completion.Completion = self._req(
             model=self.model,
             prompt=self.prompt,
             **self.kwargs
         )
 
-        if resp["choices"][0]["finish_reason"] == "stop":
+        if resp.choices[0].finish_reason == "stop":
             self.stopped = True
 
-        choice0 = resp["choices"][0]
+        choice0: completion_choice.CompletionChoice = resp.choices[0]
 
-        self.prompt += choice0["text"]
+        self.prompt += choice0.text
 
         return {
-            "id": resp["id"],
+            "id": resp.id,
             "choices": [
                 {
-                    "index": choice0["index"],
+                    "index": choice0.index,
                     "message": {
                         "role": "assistant",
                         "type": "text",
-                        "content": choice0["text"]
+                        "content": choice0.text
                     },
-                    "finish_reason": choice0["finish_reason"]
+                    "finish_reason": choice0.finish_reason
                 }
             ],
-            "usage": resp["usage"]
+            "usage": {
-        }
+                "prompt_tokens": resp.usage.prompt_tokens,
-        
+                "completion_tokens": resp.usage.completion_tokens,
-if __name__ == "__main__":
+                "total_tokens": resp.usage.total_tokens
-    import os
-
-    openai.api_key = os.environ["OPENAI_API_KEY"]
-
-    for resp in CompletionRequest(
-        model="text-davinci-003",
-        messages=[
-            {
-                "role": "user",
-                "content": "Hello, who are you?"
             }
-        ]
+        }
-    ):
-        print(resp)
-        if resp["choices"][0]["finish_reason"] == "stop":
-            break

pkg/openai/api/model.py

@@ -8,6 +8,8 @@ import openai
 
 class RequestBase:
 
+    client: openai.Client
+
     req_func: callable
 
     def __init__(self, *args, **kwargs):
@@ -17,41 +19,17 @@ class RequestBase:
         import pkg.utils.context as context
         switched, name = context.get_openai_manager().key_mgr.auto_switch()
         logging.debug("切换api-key: switched={}, name={}".format(switched, name))
-        openai.api_key = context.get_openai_manager().key_mgr.get_using_key()
+        self.client.api_key = context.get_openai_manager().key_mgr.get_using_key()
 
     def _req(self, **kwargs):
         """处理代理问题"""
         import config
 
-        ret: dict = {}
+        ret = self.req_func(**kwargs)
-        exception: Exception = None
+        logging.debug("接口请求返回：%s", str(ret))
 
-        async def awrapper(**kwargs):
+        if config.switch_strategy == 'active':
-            nonlocal ret, exception
+            self._next_key()
-
-            try:
-                ret = await self.req_func(**kwargs)
-                logging.debug("接口请求返回：%s", str(ret))
-                
-                if config.switch_strategy == 'active':
-                    self._next_key()
-
-                return ret
-            except Exception as e:
-                exception = e
-            
-        loop = asyncio.new_event_loop()
-
-        thr = threading.Thread(
-            target=loop.run_until_complete,
-            args=(awrapper(**kwargs),)
-        )
-
-        thr.start()
-        thr.join()
-
-        if exception is not None:
-            raise exception
 
         return ret
 

pkg/openai/manager.py

@@ -24,6 +24,8 @@ class OpenAIInteract:
         "size": "256x256",
     }
 
+    client: openai.Client = None
+
     def __init__(self, api_key: str):
 
         self.key_mgr = pkg.openai.keymgr.KeysManager(api_key)
@@ -31,7 +33,9 @@ class OpenAIInteract:
 
         # logging.info("文字总使用量：%d", self.audit_mgr.get_total_text_length())
 
-        openai.api_key = self.key_mgr.get_using_key()
+        self.client = openai.Client(
+            api_key=self.key_mgr.get_using_key()
+        )
 
         pkg.utils.context.set_openai_manager(self)
 
@@ -48,7 +52,7 @@ class OpenAIInteract:
         cp_parmas = config.completion_api_params.copy()
         del cp_parmas['model']
 
-        request = select_request_cls(model, messages, cp_parmas)
+        request = select_request_cls(self.client, model, messages, cp_parmas)
 
         # 请求接口
         for resp in request:

pkg/openai/modelmgr.py

@@ -6,6 +6,7 @@ Completion - text-davinci-003 等模型
 此模块封装此两个接口的请求实现，为上层提供统一的调用方式
 """
 import tiktoken
+import openai
 
 from pkg.openai.api.model import RequestBase
 from pkg.openai.api.completion import CompletionRequest
@@ -51,11 +52,11 @@ IMAGE_MODELS = {
 }
 
 
-def select_request_cls(model_name: str, messages: list, args: dict) -> RequestBase:
+def select_request_cls(client: openai.Client, model_name: str, messages: list, args: dict) -> RequestBase:
     if model_name in CHAT_COMPLETION_MODELS:
-        return ChatCompletionRequest(model_name, messages, **args)
+        return ChatCompletionRequest(client, model_name, messages, **args)
     elif model_name in COMPLETION_MODELS:
-        return CompletionRequest(model_name, messages, **args)
+        return CompletionRequest(client, model_name, messages, **args)
     raise ValueError("不支持模型[{}]，请检查配置文件".format(model_name))
 
 

pkg/openai/session.py

@@ -278,7 +278,7 @@ class Session:
             if resp['choices'][0]['message']['role'] == "assistant" and resp['choices'][0]['message']['content'] != None:  # 包含纯文本响应
 
                 if not trace_func_calls:
-                    res_text += resp['choices'][0]['message']['content'] + "\n"
+                    res_text += resp['choices'][0]['message']['content']
                 else:
                     res_text = resp['choices'][0]['message']['content']
                     pending_res_text = resp['choices'][0]['message']['content']