openai 替换为 通义千问

app/service/prompt_generation/chatgpt_for_translation.py

@@ -1,8 +1,11 @@
 import logging
 
-from langchain.chains import LLMChain
-from langchain.chat_models import ChatOpenAI
+from dashscope import Generation
+# from langchain.chains import LLMChain
+from langchain_community.chat_models import QianfanChatEndpoint, ChatTongyi
+# from langchain.chat_models import ChatOpenAI
 from langchain_core.prompts import SystemMessagePromptTemplate, HumanMessagePromptTemplate, ChatPromptTemplate
+from langchain_core.runnables import RunnableSequence
 
 from app.core.config import OPENAI_MODEL, OPENAI_API_KEY
 
@@ -10,9 +13,9 @@ from app.core.config import OPENAI_MODEL, OPENAI_API_KEY
 # os.environ["https_proxy"] = "http://127.0.0.1:7890"
 
 
-llm = ChatOpenAI(model_name=OPENAI_MODEL,
-                 openai_api_key=OPENAI_API_KEY,
-                 temperature=0)
+# llm = ChatOpenAI(model_name=OPENAI_MODEL,
+#                  openai_api_key=OPENAI_API_KEY,
+#                  temperature=0)
 
 
 def translate_to_en(text):
@@ -24,48 +27,34 @@ def translate_to_en(text):
         output the input text exactly as it is without any modifications or additions.
         If there are grammatical errors, correct them and then output the sentence."""
     )
-    system_message_prompt = SystemMessagePromptTemplate.from_template(template)
+    messages = [
+        {
+            "content": template,  # 系统message
+            "role": "system"
+        },
+        {
+            # "content": input('请输入：'),  # 用户message
+            "content": text,  # 用户message
+            "role": "user"
+        }
+    ]
+    first_response = get_response(messages)
+    assistant_output = first_response.output.choices[0].message
+    print("translate result : {}".format(assistant_output))
+    return assistant_output.content
 
-    # 待翻译文本由 Human 角色输入
-    human_template = "User input : {text}"
-    human_message_prompt = HumanMessagePromptTemplate.from_template(input_variables=["text"], template=human_template)
 
-    # 使用 System 和 Human 角色的提示模板构造 ChatPromptTemplate
-    chat_prompt_template = ChatPromptTemplate.from_messages(
-        [system_message_prompt, human_message_prompt]
+
+def get_response(messages):
+    response = Generation.call(
+        model='qwen-max',
+        api_key='sk-7658298c6b99443c98184a5e634fe6ab',
+        messages=messages,
+        # seed=random.randint(1, 10000),  # 设置随机数种子seed，如果没有设置，则随机数种子默认为1234
+        result_format='message',  # 将输出设置为message形式
+        enable_search='True'
     )
-    translate_chain = LLMChain(llm=llm, prompt=chat_prompt_template)
-
-    result = translate_chain.invoke(text)
-
-    logging.info("translate result : " + result.get('text'))
-    # print("translate result : " + result.get('text'))
-    return result.get('text')
-
-    # template = (
-    #     """
-    #     Input sentence:
-    #     {translate}
-    #     1. Based on the input,adjust the input sentence to make it more suitable for prompts for generating images,
-    #     ensuring all key nouns or adjectives related to the image are retained.
-    #     2. Simplify complex sentence structures and clarify ambiguous expressions.
-    #     3. Only Output the adjusted English sentence.
-    #
-    #     Output :
-    #     """
-    # )
-    # # "Based on the input sentence, extract key adjectives and nouns.Only Output extracted key words."
-    # # 1. Check if the input sentence contains any grammatical errors. If there are errors, please correct them before proceeding.
-    #
-    # prompt_template = PromptTemplate(input_variables=["translate"], template=template)
-    # prompt_chain = LLMChain(llm=llm, prompt=prompt_template)
-    #
-    # from langchain.chains import SimpleSequentialChain
-    # overall_chain = SimpleSequentialChain(chains=[translate_chain, prompt_chain], verbose=True)
-    #
-    # response = overall_chain.run(text)
-    # return response
-
+    return response
 
 def main():
     """Main function"""