#  File:    download_oews.py
#  Path:    apps/etl/src/airiskindex_etl/download_oews.py
#  Project: AI Risk Index — airiskindex.io
#  Author:  Simon-Pierre Boucher
#  Contact: contact@spboucher.ai
#  Copyright © 2026 Simon-Pierre Boucher. All rights reserved.
#
#  Description: Fetch BLS OEWS national wage data into data/raw/oews/.

"""Fetch BLS OEWS national wage data (May 2025) into data/raw/oews/.

BLS blocks non-browser user agents — a descriptive UA with contact info works.
Source: https://www.bls.gov/oes/tables.htm (docs/research/02-data-sources.md).
"""

from __future__ import annotations

import sys
import zipfile
from pathlib import Path

import requests

OEWS_URL = "https://www.bls.gov/oes/special-requests/oesm25nat.zip"
USER_AGENT = "Mozilla/5.0 (compatible; airiskindex-etl/0.1; +https://www.airiskindex.io)"

REPO_ROOT = Path(__file__).resolve().parents[4]
RAW_DIR = REPO_ROOT / "data" / "raw" / "oews"


def download() -> Path:
    RAW_DIR.mkdir(parents=True, exist_ok=True)
    archive = RAW_DIR / "oesm25nat.zip"
    if not archive.exists():
        print(f"downloading {OEWS_URL}")
        response = requests.get(OEWS_URL, headers={"User-Agent": USER_AGENT}, timeout=300)
        response.raise_for_status()
        archive.write_bytes(response.content)
    with zipfile.ZipFile(archive) as zf:
        zf.extractall(RAW_DIR)
    print(f"extracted to {RAW_DIR}")
    return archive


if __name__ == "__main__":
    try:
        download()
    except requests.RequestException as error:
        print(f"download failed: {error}", file=sys.stderr)
        sys.exit(1)