app.py

import json
import uvicorn
import tensorflow as tf
import gpt_2_simple as gpt2
from fastapi import FastAPI, HTTPException, Header
import os
import requests

api_key = os.getenv("api_key")
log_url = os.getenv("log")

app = FastAPI()

# Load GPT-2 model
model_name = "chatbot"
sess = gpt2.start_tf_sess()  # Start a new TensorFlow session
gpt2.load_gpt2(sess, model_name=model_name)  # Load the GPT-2 model

# Define a list of allowed API keys
ALLOWED_API_KEYS = [api_key, "your-api-key-1", "your-api-key-2"]

@app.get("/")
async def read_root():
    return {"Hello": "World!"}

import time

@app.post("/generate")
async def generate_text(data: dict):

    api_key = data.get("api_key")
    if api_key not in ALLOWED_API_KEYS:
        raise HTTPException(status_code=401, detail="Unauthorized")

    prompt = data.get("prompt")

    if not prompt:
        raise HTTPException(status_code=400, detail="Prompt is required.")

    start_time = time.time()   # Get the start time

    # Generate text using GPT-2 model
    generated_text = gpt2.generate(sess, model_name=model_name, prefix=prompt,
                                   length=200,
                                   temperature=0.65,
                                   truncate='<|endoftext|>',
                                   return_as_list=True)[0]

    end_time = time.time()  # Get the end time
    time_taken = end_time - start_time  # Calculate the time taken

    # Log generated text to Python server
    log_server_url = log_url
    log_data = {
        "generated_text": generated_text,
        "time_taken": time_taken
    }
    try:
        response = requests.post(log_server_url, json=log_data)
        response.raise_for_status()
        # Log successful
    except requests.exceptions.RequestException as e:
        # Log error if request fails
        print("Error logging generated text:", e)

    # Return the generated text along with the time taken
    return {"generated_text": generated_text, "time_taken": time_taken}

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=7860)