app/tasks/celery_app.py

"""Author RAG Chatbot SaaS — Celery Application Factory.

All background tasks are registered here.
Scheduled tasks (beat) are also configured here.
"""

from celery import Celery
from celery.schedules import crontab

from app.config import get_settings

cfg = get_settings()

celery_app = Celery(
    "authorbot",
    broker=cfg.CELERY_BROKER_URL,
    backend=cfg.CELERY_RESULT_BACKEND,
    include=[
        "app.tasks.ingestion_task",
        "app.tasks.analytics_task",
        "app.tasks.email_task",
        "app.tasks.expiry_check_task",
        "app.tasks.link_health_task",
        "app.tasks.geo_update_task",
        "app.tasks.backup_task",  # R-131: Backup task must be in include list
    ],
)

celery_app.conf.update(
    task_serializer="json",
    result_serializer="json",
    accept_content=["json"],
    timezone="UTC",
    enable_utc=True,
    task_track_started=True,
    task_acks_late=True,           # Re-queue if worker crashes
    worker_prefetch_multiplier=1,  # One task at a time per worker slot
)

# ── Scheduled Tasks (Celery Beat) ─────────────────────────
celery_app.conf.beat_schedule = {
    # Analytics rollup: runs every hour
    "analytics-hourly-rollup": {
        "task": "app.tasks.analytics_task.aggregate_hourly",
        "schedule": crontab(minute=0),
    },
    # Subscription expiry check: runs every day at 8am UTC
    "subscription-expiry-check": {
        "task": "app.tasks.expiry_check_task.check_expiring_subscriptions",
        "schedule": crontab(hour=8, minute=0),
    },
    # Link health check: runs every day at 3am UTC
    "link-health-check": {
        "task": "app.tasks.link_health_task.check_all_links",
        "schedule": crontab(hour=3, minute=0),
    },
    # MaxMind GeoIP DB update: runs every Sunday at 2am UTC
    "geoip-db-update": {
        "task": "app.tasks.geo_update_task.update_geoip_database",
        "schedule": crontab(hour=2, minute=0, day_of_week=0),
    },
    # Weekly digest emails: every Monday at 9am UTC
    "weekly-digest-emails": {
        "task": "app.tasks.email_task.send_weekly_digests",
        "schedule": crontab(hour=9, minute=0, day_of_week=1),
    },
    # R-106/R-131: Database backup: daily at 4am UTC
    "daily-database-backup": {
        "task": "app.tasks.backup_task.run_backup",
        "schedule": crontab(hour=4, minute=0),
    },
}