app.py

import os
import gradio as gr
import warnings
import json
from dotenv import load_dotenv
from typing import Dict, Any, List, Optional
import time
from functools import lru_cache
import logging
 
from langchain.agents import Tool, AgentExecutor
from langchain.tools.retriever import create_retriever_tool
from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain_community.vectorstores import FAISS
from langchain_community.embeddings import AzureOpenAIEmbeddings
from langchain_core.messages import HumanMessage, AIMessage, SystemMessage
from openai import AzureOpenAI


# Patch Gradio bug
import gradio_client.utils
gradio_client.utils.json_schema_to_python_type = lambda schema, defs=None: "string"


# Load environment variables
load_dotenv()
AZURE_OPENAI_API_KEY = os.getenv("AZURE_OPENAI_API_KEY")
AZURE_OPENAI_ENDPOINT = os.getenv("AZURE_OPENAI_ENDPOINT")
AZURE_OPENAI_LLM_DEPLOYMENT = os.getenv("AZURE_OPENAI_LLM_DEPLOYMENT")
AZURE_OPENAI_EMBEDDING_DEPLOYMENT = os.getenv("AZURE_OPENAI_EMBEDDING_DEPLOYMENT")
 
if not all([AZURE_OPENAI_API_KEY, AZURE_OPENAI_ENDPOINT, AZURE_OPENAI_LLM_DEPLOYMENT, AZURE_OPENAI_EMBEDDING_DEPLOYMENT]):
    raise ValueError("Missing one or more Azure OpenAI environment variables.")
 
warnings.filterwarnings("ignore")
 
# Embeddings for retriever
embeddings = AzureOpenAIEmbeddings(
    azure_deployment=AZURE_OPENAI_EMBEDDING_DEPLOYMENT,
    azure_endpoint=AZURE_OPENAI_ENDPOINT,
    openai_api_key=AZURE_OPENAI_API_KEY,
    openai_api_version="2025-01-01-preview",
    chunk_size=1000
)
 
# Get the directory where this script is located
SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
 
# Build the absolute path to the faiss_index_sysml directory relative to this script
FAISS_INDEX_PATH = os.path.join(SCRIPT_DIR, "faiss_index_sysml")
# Load FAISS vectorstore
vectorstore = FAISS.load_local(FAISS_INDEX_PATH, embeddings, allow_dangerous_deserialization=True)
 
# Initialize Azure OpenAI client directly
client = AzureOpenAI(
    api_key=AZURE_OPENAI_API_KEY,
    api_version="2025-01-01-preview",
    azure_endpoint=AZURE_OPENAI_ENDPOINT
)
 
 
logger = logging.getLogger(__name__)
# SysML retriever function
@lru_cache(maxsize=100)
def sysml_retriever(query: str) -> str:
    start_time = time.time()
    try:
        results = vectorstore.similarity_search(query, k=100)
        contexts = [doc.page_content for doc in results]
        response = "\n\n".join(contexts)
       
        # Log performance metrics
        duration = time.time() - start_time
        print(f"Retrieval completed in {duration:.2f}s for query: {query[:50]}...")
       
        return response
    except Exception as e:
        logger.error(f"Retrieval error: {str(e)}")
        return "Unable to retrieve information at this time."
 
 
# sysml_retriever = create_retriever_tool(
#     retriever=vectorstore.as_retriever(),
#     name="SysMLRetriever",
#     description="Use this to answer questions about SysML diagrams and modeling."
# )

# Dummy functions
def dummy_weather_lookup(location: str = "London") -> str:
    return f"The weather in {location} is sunny and 25°C."
 
def dummy_time_lookup(timezone: str = "UTC") -> str:
    return f"The current time in {timezone} is 3:00 PM."
 
# Tools definition for OpenAI function calling
tools_definition = [
    {
        "type": "function",
        "function": {
            "name": "SysMLRetriever",
            "description": "Use this to answer questions about SysML diagrams and modeling.",
            "parameters": {
                "type": "object",
                "properties": {
                    "query": {
                        "type": "string",
                        "description": "The search query to find information about SysML"
                    }
                },
                "required": ["query"]
            }
        }
    },
    {
        "type": "function",
        "function": {
            "name": "WeatherLookup",
            "description": "Use this to look up the current weather in a specified location.",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {
                        "type": "string",
                        "description": "The location to look up the weather for"
                    }
                },
                "required": ["location"]
            }
        },
    },
    {
        "type": "function",
        "function": {
            "name": "TimeLookup",
            "description": "Use this to look up the current time in a specified timezone.",
            "parameters": {
                "type": "object",
                "properties": {
                    "timezone": {
                        "type": "string",
                        "description": "The timezone to look up the current time for"
                    }
                },
                "required": ["timezone"]
            }
        }
    }
]
 
# Tool execution mapping
tool_mapping = {
    "SysMLRetriever": sysml_retriever,
    "WeatherLookup": dummy_weather_lookup,
    "TimeLookup": dummy_time_lookup
}
 
# Convert chat history
def convert_history_to_messages(history):
    messages = []
    for user, bot in history:
        messages.append({"role": "user", "content": user})
        messages.append({"role": "assistant", "content": bot})
    return messages


history = []


# Main chatbot function with direct function calling
def sysml_chatbot(message, history):
    # Convert history to messages format
    chat_messages = convert_history_to_messages(history)
   
    # Add system message at beginning
    full_messages = [
        {"role": "system", "content": "You are a helpful SysML modeling assistant and also a capable smart Assistant "}
    ]
    full_messages.extend(chat_messages)
   
    # Add current user message
    full_messages.append({"role": "user", "content": message})
   
    try:
        # First call to get either a direct answer or a function call
        response = client.chat.completions.create(
            model=AZURE_OPENAI_LLM_DEPLOYMENT,
            messages=full_messages,
            tools=tools_definition,
            tool_choice={"type": "function", "function": {"name": "SysMLRetriever"}}
        )
       
        assistant_message = response.choices[0].message
       
        # Check if the model wants to call a function
        if assistant_message.tool_calls:
            # Get the function call details
            tool_call = assistant_message.tool_calls[0]
            function_name = tool_call.function.name
            function_args = json.loads(tool_call.function.arguments)
            print("Attempting function calling...")
            # Execute the function
            if function_name in tool_mapping:
                function_response = tool_mapping[function_name](**function_args)
               
                # Append the assistant's request and the function response to messages
                full_messages.append({"role": "assistant", "content": None, "tool_calls": [
                    {"id": tool_call.id, "type": "function", "function": {"name": function_name, "arguments": tool_call.function.arguments}}
                ]})
               
                full_messages.append({
                    "role": "tool",
                    "tool_call_id": tool_call.id,
                    "content": function_response
                })
               
                # Second call to get the final answer based on the function result
                second_response = client.chat.completions.create(
                    model=AZURE_OPENAI_LLM_DEPLOYMENT,
                    messages=full_messages
                )
               
                answer = second_response.choices[0].message.content
                print("Getting final response after function execution...")
                #print(f"Function '{function_name}' executed successfully. Response: {answer}")
            else:
                answer = f"I tried to use a function '{function_name}' that's not available. Let me try again with general knowledge: SysML is a modeling language for systems engineering that helps visualize and analyze complex systems."
        else:
            # Model provided a direct answer
            answer = assistant_message.content
       
        history.append((message, answer))
        return answer, history
   
    except Exception as e:
        print(f"Error in function calling: {str(e)}")
       
        # Fallback to a direct response without function calling
        try:
            simple_messages = [
                {"role": "system", "content": "You are a helpful SysML modeling assistant."}
            ]
            simple_messages.extend(chat_messages)
            simple_messages.append({"role": "user", "content": message})
           
            fallback_response = client.chat.completions.create(
                model=AZURE_OPENAI_LLM_DEPLOYMENT,
                messages=simple_messages
            )
           
            answer = fallback_response.choices[0].message.content
        except Exception as fallback_error:
            print(f"Error in fallback: {str(fallback_error)}")
            answer = "I'm having trouble accessing my tools right now. SysML is a modeling language used in systems engineering to visualize and analyze complex systems through various diagram types."
       
        history.append((message, answer))
        return answer, history

# Gradio UI
with gr.Blocks(css="""
    #submit-btn {
        height: 100%;
        background-color: #48CAE4;
        color: white;
        font-size: 1.5em;
    }
""") as demo:
    
    gr.Markdown("## SysModeler Chatbot")

    chatbot = gr.Chatbot(height=600)
    
    with gr.Row():
        with gr.Column(scale=5):
            msg = gr.Textbox(
                placeholder="Ask me about SysML diagrams or concepts...",
                lines=3,
                show_label=False
            )
        with gr.Column(scale=1, min_width=50):
            submit_btn = gr.Button("➤")

    clear = gr.Button("Clear")

    state = gr.State(history)

    submit_btn.click(fn=sysml_chatbot, inputs=[msg, state], outputs=[msg, chatbot])
    msg.submit(fn=sysml_chatbot, inputs=[msg, state], outputs=[msg, chatbot])  # still supports enter key
    clear.click(fn=lambda: ([], ""), inputs=None, outputs=[chatbot, msg])

if __name__ == "__main__":
    demo.launch()