lambda_function.py

import urllib.parse
import boto3
import json
from datetime import datetime

print('Loading function')

s3 = boto3.resource('s3')
s3_client = boto3.client('s3')

def lambda_handler(event, context):
    sns_message = json.loads(event['Records'][0]['Sns']['Message'])
    record = sns_message['Records'][0]
    
    target_bucket = context.function_name[:-1]
    source_bucket = str(record['s3']['bucket']['name'])
    key = str(urllib.parse.unquote_plus(record['s3']['object']['key']))
    
    # Stop cascading deletes: ignore S3 delete events triggered by a delete Lambda (e.g. "...:bahaimedia-sgd")
    principal_id = record.get('userIdentity', {}).get('principalId', '')
    if principal_id.endswith(f":{source_bucket}d"):
        print(f"Skipping cascading delete of {key} from {source_bucket}: triggered by Lambda ({principal_id})")
        return

    # Parse S3 eventTime (e.g. "2026-10-15T02:41:38.123Z") as timezone-aware UTC datetime
    event_time = datetime.fromisoformat(record['eventTime'].replace('Z', '+00:00'))
    
    obj = s3.Object(target_bucket, key)
    try:
        obj.load()
    except Exception as e:
        print(f"Did not delete {key} from {target_bucket} because target object does not exist: {e}")
        return

    # 1. Only delete if the object in target_bucket is older than the delete event
    if obj.last_modified >= event_time:
        print(f"Skipping delete of {key} in {target_bucket}: target object ({obj.last_modified}) is newer than or equal to delete event ({event_time})")
        return

    # 2. Check if source_bucket now has a live/newer file (e.g. quick re-upload after delete)
    try:
        src_head = s3_client.head_object(Bucket=source_bucket, Key=key)
        if src_head['LastModified'] >= event_time:
            print(f"Skipping delete of {key} in {target_bucket}: source bucket {source_bucket} has a newer object ({src_head['LastModified']})")
            return
    except Exception:
        # 404 NotFound (or 403 Forbidden without s3:ListBucket) in source_bucket is expected when the file is genuinely deleted
        pass

    print(f"Deleting {key} from bucket {target_bucket}")
    try:
        obj.delete()
    except Exception as e:
        print("Did not delete because of error", e)