Changing the process working directory at import time (os.chdir(...)) creates a global side effect that can break callers (e.g., if this script is invoked from another Python process or imported for reuse). Prefer keeping the original CWD and writing downloads to an explicit output_dir derived from Path(__file__).resolve().parent (use that path when opening the output file); if you keep chdir, at least move it (and the download loop) under an if __name__ == "__main__" guard.