openai 接口返回token数量来修剪会话长度

bot/chatgpt/chat_gpt_bot.py

@@ -6,7 +6,6 @@ from common.log import logger
 from common.expired_dict import ExpiredDict
 import openai
 import time
-import json
 
 if conf().get('expires_in_seconds'):
     user_session = ExpiredDict(conf().get('expires_in_seconds'))
@@ -44,12 +43,19 @@ class ChatGPTBot(Bot):
             logger.debug("[OPEN_AI] new_query={}, user={}, reply_cont={}".format(new_query, from_user_id, reply_content))
             if reply_content:
                 Session.save_session(query, reply_content, from_user_id)
-            return reply_content
+            return reply_content[1]
 
         elif context.get('type', None) == 'IMAGE_CREATE':
             return self.create_img(query, 0)
 
     def reply_text(self, query, user_id, retry_count=0):
+        '''
+        call openai's ChatCompletion to get the answer
+        :param query: query content
+        :param user_id: from user id
+        :param retry_count: retry count
+        :return: [0]-tokens used and [1]-answer
+        '''
         try:
             response = openai.ChatCompletion.create(
                 model="gpt-3.5-turbo",  # 对话模型的名称
@@ -62,8 +68,8 @@ class ChatGPTBot(Bot):
             )
             # res_content = response.choices[0]['text'].strip().replace('<|endoftext|>', '')
             logger.info(response.choices[0]['message']['content'])
-            # log.info("[OPEN_AI] reply={}".format(res_content))
-            return response.choices[0]['message']['content']
+
+            return response["usage"]["prompt_tokens"],response.choices[0]['message']['content']
         except openai.error.RateLimitError as e:
             # rate limit exception
             logger.warn(e)
@@ -72,21 +78,21 @@ class ChatGPTBot(Bot):
                 logger.warn("[OPEN_AI] RateLimit exceed, 第{}次重试".format(retry_count+1))
                 return self.reply_text(query, user_id, retry_count+1)
             else:
-                return "提问太快啦，请休息一下再问我吧"
+                return 0,"提问太快啦，请休息一下再问我吧"
         except openai.error.APIConnectionError as e:
             # api connection exception
             logger.warn(e)
             logger.warn("[OPEN_AI] APIConnection failed")
-            return "我连接不到你的网络"
+            return 0,"我连接不到你的网络"
         except openai.error.Timeout as e:
             logger.warn(e)
             logger.warn("[OPEN_AI] Timeout")
-            return "我没有收到你的消息"
+            return 0,"我没有收到你的消息"
         except Exception as e:
             # unknown exception
             logger.exception(e)
             Session.clear_session(user_id)
-            return "请再问我一次吧"
+            return 0,"请再问我一次吧"
 
     def create_img(self, query, retry_count=0):
         try:
@@ -142,31 +148,27 @@ class Session(object):
         if not max_tokens:
             # default 3000
             max_tokens = 1000
+        max_tokens=int(max_tokens)
 
         session = user_session.get(user_id)
         if session:
             # append conversation
-            gpt_item = {'role': 'assistant', 'content': answer}
+            gpt_item = {'role': 'assistant', 'content': answer[1]}
             session.append(gpt_item)
 
         # discard exceed limit conversation
-        Session.discard_exceed_conversation(user_session[user_id], max_tokens) 
+        used_tokens=int(answer[0])
+        # logger.info("prompt tokens used={},max_tokens={}".format(used_tokens,max_tokens))
+
+        while used_tokens > max_tokens:
+            # pop first conversation
+            if len(session) > 0:
+                session.pop(0) 
+            else:
+                break    
+
+            used_tokens=used_tokens-max_tokens
 
-    @staticmethod
-    def discard_exceed_conversation(session, max_tokens):
-        count = 0
-        count_list = list()
-        for i in range(len(session)-1, -1, -1):
-        # count tokens of conversation list
-            history_conv = session[i]
-            tokens=json.dumps(history_conv).split()
-            count += len(tokens)
-            count_list.append(count)
-
-        for c in count_list:
-            if c > max_tokens:
-                # pop first conversation
-                session.pop(0)
 
     @staticmethod
     def clear_session(user_id):