from fastapi import FastAPI, HTTPException, Request, Depends from fastapi.staticfiles import StaticFiles from pytube import YouTube import uuid import logging import aiohttp import base64 import io import os import random import string import json app = FastAPI() whisper_origin = os.getenv("WHISPER_ORIGIN") whisper_base_url = os.getenv("WHISPER_BASE_URL") img2location_name = os.getenv("IMG2LOCATION_NAME") img2location_origin = os.getenv("IMG2LOCATION_ORIGIN") img2location_base_url = os.getenv("IMG2LOCATION_BASE_URL") pixart_sigma_base_url = os.getenv("PIXART_SIGMA_BASE_URL") allowed_user_agent = os.getenv("ALLOWED_USER_AGENT") def generate_hash(length=12): # Characters that can appear in the hash characters = string.ascii_lowercase + string.digits # Generate a random string of the specified length hash_string = ''.join(random.choice(characters) for _ in range(length)) return hash_string @app.get("/") async def read_root(): return {"message": "Saqib's API"} # Create a directory to store MP3 files if it doesn't exist AUDIO_DIR = "audio_files" os.makedirs(AUDIO_DIR, exist_ok=True) # Mount the audio directory app.mount("/audio", StaticFiles(directory=AUDIO_DIR), name="audio") @app.get("/get_audio") async def get_audio(url: str): if not url: raise HTTPException(status_code=400, detail="URL is required") try: yt = YouTube(url) video = yt.streams.filter(only_audio=True).first() # Generate a unique filename unique_filename = f"{uuid.uuid4().hex}.mp3" out_file = os.path.join(AUDIO_DIR, unique_filename) # Download the audio video.download(output_path=AUDIO_DIR, filename=unique_filename) file_stats = os.stat(out_file) logging.info(f'Size of audio file in Bytes: {file_stats.st_size}') if file_stats.st_size <= 30000000: # Construct the URL for the MP3 file mp3_url = f"/audio/{unique_filename}" return mp3_url else: os.remove(out_file) raise HTTPException(status_code=413, detail="Audio file is too large. Limited to about 1.5 hours.") except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/whisper") async def whisper(request: Request): data = await request.json() # Extracting JSON data from request if "audio_url" not in data: raise HTTPException(status_code=400, detail="audio_url not found in request") url = data["audio_url"] headers = { 'Accept': 'application/json, text/plain, */*', 'Accept-Language': 'en-US,en;q=0.9', 'Cache-Control': 'no-cache', 'Connection': 'keep-alive', 'Content-Type': 'application/json', 'DNT': '1', 'Origin': whisper_origin, 'Pragma': 'no-cache', 'Referer': f'{whisper_origin}/', 'Sec-Fetch-Dest': 'empty', 'Sec-Fetch-Mode': 'cors', 'Sec-Fetch-Site': 'same-site', 'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36 Edg/124.0.0.0', 'sec-ch-ua': '"Chromium";v="124", "Microsoft Edge";v="124", "Not-A.Brand";v="99"', 'sec-ch-ua-mobile': '?0', 'sec-ch-ua-platform': '"Windows"', } # Async HTTP request to get the audio file async with aiohttp.ClientSession() as session: async with session.get(url) as resp: if resp.status != 200: return f"Failed to download audio: {resp.status}" audio_data = await resp.read() # Encode the audio data to base64 audio_base64 = base64.b64encode(audio_data).decode("utf-8") json_data = '{"audio": "' + audio_base64 + '"}' # Post request to the API async with session.post(f'{whisper_base_url}/v1/inference/openai/whisper-large', headers=headers, data=json_data) as post_resp: if post_resp.status != 200: return f"API request failed: {post_resp.status}" return await post_resp.json() @app.post("/img2location") async def img2location(request: Request): request_json = await request.json() image_url = request_json.get("image_url", None) if not image_url: raise HTTPException(status_code=400, detail="image_url not found in request") headers = { 'accept': '*/*', 'accept-language': 'en-US,en;q=0.9', 'cache-control': 'no-cache', 'dnt': '1', 'origin': img2location_origin, 'pragma': 'no-cache', 'priority': 'u=1, i', 'referer': f'{img2location_origin}/', 'sec-ch-ua': '"Chromium";v="124", "Microsoft Edge";v="124", "Not-A.Brand";v="99"', 'sec-ch-ua-mobile': '?0', 'sec-ch-ua-platform': '"Windows"', 'sec-fetch-dest': 'empty', 'sec-fetch-mode': 'cors', 'sec-fetch-site': 'cross-site', 'user-agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36 Edg/124.0.0.0', } async with aiohttp.ClientSession() as session: # Fetch the image from the URL async with session.get(image_url) as img_response: if img_response.status != 200: return f"Failed to fetch image: HTTP {img_response.status}" image_data = await img_response.read() # Using BytesIO to handle the byte content data = aiohttp.FormData() data.add_field('image', io.BytesIO(image_data), filename="image.png", content_type='image/png') # Sending the POST request async with session.post(img2location_base_url, headers=headers, data=data) as response: if response.status != 200: return f"Failed to upload image: HTTP {response.status}" json_response = await response.json() return json_response @app.post("/pixart-sigma") async def pixart_sigma(request: Request): request_json = await request.json() prompt = request_json.get("prompt", None) negative_prompt = request_json.get("negative_prompt", "") style = request_json.get("style", "(No style)") use_negative_prompt = request_json.get("use_negative_prompt", True) num_imgs = request_json.get("num_imgs", 1) seed = request_json.get("seed", 0) width = request_json.get("width", 1024) height = request_json.get("height", 1024) schedule = request_json.get("schedule", "DPM-Solver") dpms_guidance_scale = request_json.get("dpms_guidance_scale", 4.5) sas_guidance_scale = request_json.get("sas_guidance_scale", 3) dpms_inference_steps = request_json.get("dpms_inference_steps", 14) sas_inference_steps = request_json.get("sas_inference_steps", 25) randomize_seed = request_json.get("randomize_seed", True) hash = generate_hash() headers = { 'accept': '*/*' } params = { '__theme': 'light', } json_data = { 'data': [ prompt, negative_prompt, style, use_negative_prompt, num_imgs, seed, width, height, schedule, dpms_guidance_scale, sas_guidance_scale, dpms_inference_steps, sas_inference_steps, randomize_seed, ], 'event_data': None, 'fn_index': 3, 'trigger_id': 7, 'session_hash': hash, } async with aiohttp.ClientSession() as session: async with session.post(f'{pixart_sigma_base_url}/queue/join', params=params, headers=headers, json=json_data, ssl=False) as response: print(response.status) params = { 'session_hash': hash, } async with session.get(f'{pixart_sigma_base_url}/queue/data', params=params, headers=headers, ssl=False) as response: async for line in response.content: try: if line: line = line.decode('utf-8') line = line.replace('data: ', '') line_json = json.loads(line) if line_json["msg"] == "process_completed": image_url = line_json["output"]["data"][0][0]["image"]["url"] return {"image_url": image_url} except: pass # if __name__ == "__main__": # import uvicorn # uvicorn.run(app, host="0.0.0.0", port=8000)