main.py

from fastapi import FastAPI, File, UploadFile
from fastapi.responses import FileResponse
import os
import shutil
from modules.whisper.whisper_factory import WhisperFactory

app = FastAPI()

# Initialize Whisper inference engine
whisper_inf = WhisperFactory.create_whisper_inference(
    whisper_type="faster-whisper",
    whisper_model_dir=os.path.join("models", "Whisper"),
    faster_whisper_model_dir=os.path.join("models", "Whisper", "faster-whisper"),
    insanely_fast_whisper_model_dir=os.path.join("models", "Whisper", "insanely-fast-whisper"),
    output_dir=os.path.join("outputs"),
)

@app.post("/upload-video/")
async def upload_video(file: UploadFile = File(...)):
    """
    Upload a video file and get the generated SRT file as a response.
    """
    # Save the uploaded video file temporarily
    input_video_path = os.path.join("temp", file.filename)
    os.makedirs("temp", exist_ok=True)
    
    with open(input_video_path, "wb") as buffer:
        shutil.copyfileobj(file.file, buffer)

    # Generate the subtitle file
    output_srt_path = whisper_inf.transcribe_file(
        input_video_path,
        file_format="SRT",
        add_timestamp=True
    )

    # Return the SRT file as a response
    return FileResponse(path=output_srt_path, filename=os.path.basename(output_srt_path))