add models to cantrol model version

agents/code_generate_agent.py

@@ -1,8 +1,6 @@
 import re
 from typing import List, Union
 from langchain.chains import LLMChain
-from langchain.chat_models import ChatOpenAI
-from langchain.llms import OpenAI
 from langchain.agents import Tool, LLMSingleActionAgent, AgentExecutor, AgentOutputParser
 from langchain.schema import AgentAction, AgentFinish
 from langchain.agents import initialize_agent
@@ -11,6 +9,7 @@ from agents.promopts import code_generate_agent_template
 from agents.tools.api_layer_code_tool import apiLayerCodeGenerator
 from agents.tools.domain_layer_code_tool import domainLayerCodeGenerator, entityCodeGenerator
 from agents.tools.persistent_layer_code_tool import persistentLayerCodeGenerator
+from models import llm
 
 
 class CustomPromptTemplate(StringPromptTemplate):
@@ -79,7 +78,7 @@ def code_agent_executor() -> AgentExecutor:
         input_variables=["input", "intermediate_steps"]
     )
 
-    code_llm_chain = LLMChain(llm=ChatOpenAI(temperature=0.7), prompt=AGENT_PROMPT)
+    code_llm_chain = LLMChain(llm=llm(temperature=0.7), prompt=AGENT_PROMPT)
 
     tool_names = [tool.name for tool in code_agent_tools]
     code_agent = LLMSingleActionAgent(

agents/tools/api_layer_code_tool.py

@@ -1,7 +1,8 @@
 from langchain import LLMChain, PromptTemplate
-from langchain.chat_models import ChatOpenAI
 from langchain.agents import tool
 
+from models import llm
+
 
 API_LAYER = """You are a software developer. Your task is to generate the api layer tests and product code.
 
@@ -85,7 +86,7 @@ request: {input}"""
 API_LAYER_PROMPT = PromptTemplate(input_variables=["input"], template=API_LAYER,)
 
 
-apiChain = LLMChain(llm = ChatOpenAI(temperature=0.1), prompt=API_LAYER_PROMPT)
+apiChain = LLMChain(llm = llm(temperature=0.1), prompt=API_LAYER_PROMPT)
 
 
 @tool("Generate API Layer Code", return_direct=True)

agents/tools/domain_layer_code_tool.py

@@ -1,7 +1,8 @@
 from langchain import LLMChain, PromptTemplate
-from langchain.chat_models import ChatOpenAI
 from langchain.agents import tool
 
+from models import llm
+
 
 DOMAIN_LAYER = """You are a software developer. Your task is to generate the domain layer tests and product code.
 
@@ -77,7 +78,7 @@ request: {input}"""
 
 DOMAIN_LAYER_PROMPT = PromptTemplate(input_variables=["input"], template=DOMAIN_LAYER,)
 
-domainLayerChain = LLMChain(llm = ChatOpenAI(temperature=0.1), prompt=DOMAIN_LAYER_PROMPT)
+domainLayerChain = LLMChain(llm = llm(temperature=0.1), prompt=DOMAIN_LAYER_PROMPT)
 
 
 @tool("Generate Domain Layer Code", return_direct=True)
@@ -188,7 +189,7 @@ request: {input}"""
 
 ENTITY_PROMPT = PromptTemplate(input_variables=["input"], template=ENTITY,)
 
-entityChain = LLMChain(llm = ChatOpenAI(temperature=0.1), prompt=ENTITY_PROMPT)
+entityChain = LLMChain(llm = llm(temperature=0.1), prompt=ENTITY_PROMPT)
 
 
 @tool("Generate Entity Code", return_direct=True)

agents/tools/persistent_layer_code_tool.py

@@ -1,7 +1,8 @@
 from langchain import LLMChain, PromptTemplate
-from langchain.chat_models import ChatOpenAI
 from langchain.agents import tool
 
+from models import llm
+
 
 PERSISTENT_LAYER = """You are a software developer. Your task is to generate the persistent layer tests and product code.
 
@@ -162,7 +163,7 @@ request: {input}"""
 
 PERSISTENT_LAYER_PROMPT = PromptTemplate(input_variables=["input"], template=PERSISTENT_LAYER,)
 
-persistentChain = LLMChain(llm = ChatOpenAI(temperature=0.1), prompt=PERSISTENT_LAYER_PROMPT)
+persistentChain = LLMChain(llm = llm(temperature=0.1), prompt=PERSISTENT_LAYER_PROMPT)
 
 
 @tool("Generate Persistent Layer Code", return_direct=True)

agents/tools/python_code_tool.py

@@ -3,8 +3,36 @@ from langchain import LLMChain, PromptTemplate
 from langchain.chat_models import ChatOpenAI
 from langchain.llms import OpenAI
 from langchain.agents import tool
-from langchain.utilities import PythonREPL
+# from langchain.utilities import PythonREPL
 
+import sys
+from io import StringIO
+from typing import Dict, Optional
+
+from pydantic import BaseModel, Field
+
+from models import llm
+
+
+class PythonREPL(BaseModel):
+    """Simulates a standalone Python REPL."""
+
+    # globals: Optional[Dict] = Field(default_factory=dict, alias="_globals")
+    # locals: Optional[Dict] = Field(default_factory=dict, alias="_locals")
+
+    def run(self, command: str) -> str:
+        """Run command with own globals/locals and returns anything printed."""
+        old_stdout = sys.stdout
+        sys.stdout = mystdout = StringIO()
+        try:
+            exec(command, globals(), locals())
+            sys.stdout = old_stdout
+            output = mystdout.getvalue()
+        except Exception as e:
+            sys.stdout = old_stdout
+            output = str(e)
+        return output
+    
 
 generate_python_code = """
 Please write Python script to fulfill the following requirement:
@@ -18,7 +46,7 @@ Only output the code section with code block.
 
 generate_python_code_promopt = PromptTemplate(input_variables=["input"], template=generate_python_code,)
 
-generate_code_chain = LLMChain(llm = ChatOpenAI(temperature=0.1), prompt=generate_python_code_promopt, output_key="code")
+generate_code_chain = LLMChain(llm = llm(temperature=0.1), prompt=generate_python_code_promopt, output_key="code")
 
 
 @tool("Generate and Excute Python Code ", return_direct=True)
@@ -37,7 +65,7 @@ def generate_and_excute_python_code(input: str) -> str:
         code_content = answer_code[start:end].strip()
     
     print(code_content)
-    python_repl = PythonREPL()
+    python_repl = PythonREPL(globals={"__name__": "__main__"}, locals={})
     result = python_repl.run(code_content)
     return f"""
 code:

app.py

@@ -1,16 +1,11 @@
 import gradio as gr
-from langchain.chains import LLMChain
-from langchain.chat_models import ChatOpenAI
 from langchain.document_loaders import TextLoader
 from agents.tools.python_code_tool import generate_and_excute_python_code
 from chains import HumanFeedBackChain, contextRewriteChain
 from embedding import CustomEmbedding
 from memories import HumenFeedbackBufferMemory
-from langchain.memory import ConversationBufferMemory
-from promopts import FEEDBACK, FEEDBACK_PROMPT
 from agents.code_generate_agent import code_agent_executor, code_agent_tools
 
-# llm = ChatOpenAI(temperature=0.7)
 
 baMemory = HumenFeedbackBufferMemory(
             input_key="input", human_prefix="Answer", ai_prefix="AI")

chains.py

@@ -1,10 +1,9 @@
 from typing import Any, Optional
 from langchain.chains import LLMChain
 from langchain.base_language import BaseLanguageModel
-from langchain.schema import LLMResult, PromptValue
 from langchain.prompts import PromptTemplate
 from langchain.memory.chat_memory import BaseMemory
-from langchain.chat_models import ChatOpenAI
+from models import llm
 
 from promopts import CONTENT_RE_WRIGHT_PROMPT, FEEDBACK_PROMPT
 
@@ -14,7 +13,7 @@ class HumanFeedBackChain(LLMChain):
 
     memory: Optional[BaseMemory] = None
 
-    def __init__(self, verbose=True, llm: BaseLanguageModel = ChatOpenAI(temperature=0.7),  memory: Optional[BaseMemory] = None, prompt: PromptTemplate = FEEDBACK_PROMPT):
+    def __init__(self, verbose=True, llm: BaseLanguageModel = llm(temperature=0.7),  memory: Optional[BaseMemory] = None, prompt: PromptTemplate = FEEDBACK_PROMPT):
         super().__init__(llm=llm, prompt=prompt, memory=memory, verbose=verbose)
 
     def run(self, *args: Any, **kwargs: Any) -> str:
@@ -40,4 +39,4 @@ class HumanFeedBackChain(LLMChain):
         )
 
 
-contextRewriteChain = LLMChain(llm=ChatOpenAI(temperature=0.7), prompt=CONTENT_RE_WRIGHT_PROMPT)
+contextRewriteChain = LLMChain(llm=llm(temperature=0.7), prompt=CONTENT_RE_WRIGHT_PROMPT)

documents/bussiness_context/business_context.md

@@ -6,4 +6,11 @@ FeatureConfig 用于配置某个 Feature 中控制前端展示效果的配置项
 
 添加 FeatureConfig 的主要目的是为了控制 FeatureConfig 消费方的某个行为。在添加 FeatureConfig 时，应该包含 featureKey、data、saData、status、标题和描述信息。新增的 FeatureConfig 状态为 DRAFT。
 
-客户端用户需要查看 FeatureConfig 中的 data、saData、更新时间和 id。同时，FeatureConfig 可以关联圈人条件，符合圈人条件的配置可以展示给客户端用户。客户端用户仅能查看符合圈人条件的 PUBLISHED 状态的数据。圈人条件包括上传用户白名单、按照比例灰度发布、地理位置和人群标签等。
+客户端用户需要查看 FeatureConfig 中的 data、saData、更新时间和 id。同时，FeatureConfig 可以关联圈人条件，符合圈人条件的配置可以展示给客户端用户。客户端用户仅能查看符合圈人条件的 PUBLISHED 状态的数据。圈人条件包括上传用户白名单、按照比例灰度发布、地理位置和人群标签等。
+
+ 在添加 FeatureConfig 时，应该包含 featureKey、data、saData、status、标题和描述信息。新增的 FeatureConfig 状态为 DRAFT。
+
+ 客户端用户通过客户端访问服务端接口获取FeatureConfig，客户端通过FeatureConfig控制相关Feature展示
+
+ 用户白名单圈人条件需要上传用户id的白名单，仅在白名单里的用户可以获取到相关feature
+地理位置配置端需要设置圈定地区的地理位置编号列表，客户端请求接口是传递地理位置编号参数，位置编号匹配的数据用户可见

embedding.py

@@ -7,10 +7,11 @@ from langchain.chains import RetrievalQA
 from langchain.chains.question_answering import load_qa_chain
 
 from langchain.document_loaders import NotionDirectoryLoader
-from langchain.chat_models import ChatOpenAI
 from langchain.memory import ConversationBufferMemory
 from langchain.chains import ConversationalRetrievalChain
 
+from models import llm
+
 
 class CustomEmbedding:
     notionDirectoryLoader = NotionDirectoryLoader(
@@ -32,7 +33,7 @@ class CustomEmbedding:
 
     
 
-    def getFAQChain(self, llm=ChatOpenAI(temperature=0.7)):
+    def getFAQChain(self, llm=llm(temperature=0.7)):
         memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
         docsearch = FAISS.load_local(
             "./documents/business_context.faiss", self.embeddings)

models.py

@@ -0,0 +1,10 @@
+from langchain.chat_models import ChatOpenAI
+from langchain.base_language import BaseLanguageModel
+
+def llm(temperature=0) -> BaseLanguageModel:
+    # gpt-3.5
+    return ChatOpenAI(temperature=temperature)
+    # gpt-4
+    # return ChatOpenAI(temperature=temperature, model_name="gpt-4")
+    
+