app.py

from fastapi import FastAPI, HTTPException
import base64
import os
from contextlib import asynccontextmanager

# Define the lifespan context manager
@asynccontextmanager
async def lifespan(app: FastAPI):
    lora_weights = None
    lora_path = "./lora_file.pth"
    if os.path.exists(lora_path):
        with open(lora_path, "rb") as f:
            lora_weights = base64.b64encode(f.read()).decode("utf-8")
        print("LoRA weights loaded and preprocessed successfully.")
    else:
        raise HTTPException(status_code=500, detail="LoRA file not found.")
    
    # Yield control back to the FastAPI app during runtime
    yield
    
    # Cleanup or shutdown logic here if necessary
    print("App is shutting down.")

# Create the FastAPI app with the lifespan context manager
app = FastAPI(lifespan=lifespan)

@app.post("/modify-prompt")
async def modify_prompt(prompt: str):
    global lora_weights
    if lora_weights is None:
        raise HTTPException(status_code=500, detail="LoRA weights not loaded.")
    # Combine prompt with preprocessed LoRA data
    extended_prompt = {
        "prompt": prompt,
        "lora": lora_weights
    }
    return extended_prompt