Right now sources are being downloaded in a sequential manner. This could be improved by using threads, multiprocessing or asyncio.