app/tasks/geo_update_task.py

"""Author RAG Chatbot SaaS — GeoIP Database Update Task.

Runs weekly. Downloads the latest MaxMind GeoLite2 City database.
Requires MAXMIND_LICENSE_KEY env var (free MaxMind account).
"""

import os
import structlog
from app.tasks.celery_app import celery_app

logger = structlog.get_logger(__name__)


@celery_app.task(name="app.tasks.geo_update_task.update_geoip_database")
def update_geoip_database():
    """Download and update the MaxMind GeoLite2 City database."""
    import httpx
    import tarfile
    import shutil
    from app.config import get_settings

    cfg = get_settings()
    license_key = os.environ.get("MAXMIND_LICENSE_KEY", "")
    if not license_key:
        logger.warning("MAXMIND_LICENSE_KEY not set — skipping GeoIP update")
        return

    url = (
        f"https://download.maxmind.com/app/geoip_download"
        f"?edition_id=GeoLite2-City&license_key={license_key}&suffix=tar.gz"
    )

    try:
        with httpx.Client(timeout=60) as client:
            response = client.get(url)
            response.raise_for_status()

        # Write to temp file
        tmp_path = "/tmp/geolite2.tar.gz"
        with open(tmp_path, "wb") as f:
            f.write(response.content)

        # Extract .mmdb file
        with tarfile.open(tmp_path, "r:gz") as tar:
            for member in tar.getmembers():
                if member.name.endswith(".mmdb"):
                    member.name = os.path.basename(member.name)
                    tar.extract(member, path=os.path.dirname(cfg.GEO_DB_PATH))
                    break

        os.remove(tmp_path)
        logger.info("GeoIP database updated successfully")

    except Exception as e:
        logger.error("GeoIP update failed", error=str(e))