app.py

import gradio as gr
import logging, os, sys, threading, time

from agent_langchain import agent_langchain
from agent_llamaindex import agent_llamaindex
from openai import OpenAI
from trace import trace_wandb

from dotenv import load_dotenv, find_dotenv
_ = load_dotenv(find_dotenv())

import nltk
nltk.download('wordnet')

lock = threading.Lock()

AGENT_OFF = "Off"
AGENT_LANGCHAIN  = "LangChain"
AGENT_LLAMAINDEX = "LlamaIndex"

config = {
    "model": "gpt-4o",
    "temperature": 0
}

logging.basicConfig(stream = sys.stdout, level = logging.DEBUG)
logging.getLogger().addHandler(logging.StreamHandler(stream = sys.stdout))

def invoke(openai_api_key, prompt, agent_option):
    if not openai_api_key:
        raise gr.Error("OpenAI API Key is required.")
    if not prompt:
        raise gr.Error("Prompt is required.")
    if not agent_option:
        raise gr.Error("Use Agent is required.")

    with lock:
        os.environ["OPENAI_API_KEY"] = openai_api_key
    
        completion = ""
        result = ""
        callback = ""
        err_msg = ""
        
        try:
            start_time_ms = round(time.time() * 1000)
            
            if (agent_option == AGENT_LANGCHAIN):
                completion, callback = agent_langchain(
                    config,
                    prompt
                )
        
                result = completion["output"]
            elif (agent_option == AGENT_LLAMAINDEX):
                result = agent_llamaindex(
                    config,
                    prompt
                )
            else:
                client = OpenAI()
        
                completion = client.chat.completions.create(
                    messages = [{"role": "user", "content": prompt}],
                    model = config["model"],
                    temperature = config["temperature"]
                )
    
                callback = completion.usage
                result = completion.choices[0].message.content
        except Exception as e:
            err_msg = e
    
            raise gr.Error(e)
        finally:
            end_time_ms = round(time.time() * 1000)
            
            trace_wandb(
                config,
                agent_option,
                prompt, 
                completion, 
                result,
                callback,
                err_msg, 
                start_time_ms, 
                end_time_ms
            )

            del os.environ["OPENAI_API_KEY"]

        #print("===")
        #print(result)
        #print("===")
        
        return result

gr.close_all()

demo = gr.Interface(
    fn = invoke, 
    inputs = [gr.Textbox(label = "OpenAI API Key", type = "password", lines = 1),
              gr.Textbox(label = "Prompt", lines = 1, 
                         value = "How does current weather in San Francisco and Paris compare in metric and imperial system? Answer in JSON format and include today's date."),
              gr.Radio([AGENT_OFF, AGENT_LANGCHAIN, AGENT_LLAMAINDEX], label = "Use Agent", value = AGENT_LANGCHAIN)],
    outputs = [gr.Markdown(label = "Completion", value=os.environ["OUTPUT"])],
    title = "Agentic Reasoning Application",
    description = os.environ["DESCRIPTION"]
)

demo.launch()