Files
prowler/api/src/backend/tasks/jobs/muting.py
T

65 lines
2.1 KiB
Python

from celery.utils.log import get_task_logger
from config.django.base import DJANGO_FINDINGS_BATCH_SIZE
from tasks.utils import batched
from api.db_utils import rls_transaction
from api.models import Finding, MuteRule
logger = get_task_logger(__name__)
def mute_historical_findings(tenant_id: str, mute_rule_id: str):
"""
Mute historical findings that match the given mute rule.
This function processes findings in batches, updating their muted status
and adding the mute reason.
Args:
tenant_id (str): The tenant ID for RLS context
mute_rule_id (str): The ID of the mute rule to apply
Returns:
dict: Summary of the muting operation with findings_muted count
"""
findings_muted_count = 0
# Get the list of UIDs to mute and the reason
with rls_transaction(tenant_id):
mute_rule = MuteRule.objects.get(id=mute_rule_id, tenant_id=tenant_id)
finding_uids = mute_rule.finding_uids
mute_reason = mute_rule.reason
muted_at = mute_rule.inserted_at
# Query findings that match the UIDs and are not already muted
with rls_transaction(tenant_id):
findings_to_mute = Finding.objects.filter(
tenant_id=tenant_id, uid__in=finding_uids, muted=False
)
total_findings = findings_to_mute.count()
logger.info(
f"Processing {total_findings} findings for mute rule {mute_rule_id}"
)
if total_findings > 0:
for batch, is_last in batched(
findings_to_mute.iterator(), DJANGO_FINDINGS_BATCH_SIZE
):
batch_ids = [f.id for f in batch]
updated_count = Finding.all_objects.filter(
id__in=batch_ids, tenant_id=tenant_id
).update(
muted=True,
muted_at=muted_at,
muted_reason=mute_reason,
)
findings_muted_count += updated_count
logger.info(f"Muted {findings_muted_count} findings for rule {mute_rule_id}")
return {
"findings_muted": findings_muted_count,
"rule_id": mute_rule_id,
}