AIRegulation-DocAnalysis/backend/app/infrastructure/tasks/celery_app.py

"""Shared Celery application instance for background task processing.

All workers and enqueueing call sites import `celery_app` from this module
so the broker/backend configuration stays in one place.
"""

from __future__ import annotations

from celery import Celery

from app.config.settings import settings


def _redis_url() -> str:
    """Return a Redis connection URL from application settings."""
    if settings.redis_password:
        return (
            f"redis://:{settings.redis_password}@"
            f"{settings.redis_host}:{settings.redis_port}/{settings.redis_db}"
        )
    return f"redis://{settings.redis_host}:{settings.redis_port}/{settings.redis_db}"


_BROKER = _redis_url()
_BACKEND = _redis_url()

celery_app = Celery(
    "compliance_hub",
    broker=_BROKER,
    backend=_BACKEND,
    include=[
        "app.infrastructure.tasks.document_tasks",
        "app.infrastructure.tasks.perception_tasks",
    ],
)

celery_app.conf.update(
    task_serializer="json",
    result_serializer="json",
    accept_content=["json"],
    timezone="UTC",
    enable_utc=True,
    # Acknowledge task only after successful execution to avoid data loss.
    task_acks_late=True,
    task_reject_on_worker_lost=True,
    # Keep results for 1 hour for status polling.
    result_expires=3600,
    # Scheduled counterpart to the Perception page's manual "Refresh" button.
    # Only takes effect while a Beat process is running (./dev.sh start beat).
    beat_schedule={
        "crawl-regulations-periodic": {
            "task": "app.infrastructure.tasks.perception_tasks.crawl_regulations_task",
            "schedule": settings.perception_crawl_interval_seconds,
        },
    },
)