Fargate Processor

1# Fargate file processor for Step Functions workflow
2import json
3import boto3
4import os
5import sys
6
7s3_client = boto3.client('s3')
8
9def process_file(bucket, key):
10    # Download the file
11    local_path = f"/tmp/{os.path.basename(key)}"
12    s3_client.download_file(bucket, key, local_path)
13    print(f"Downloaded s3://{bucket}/{key} to {local_path}")
14
15    # Log metadata
16    file_size = os.path.getsize(local_path)
17    metadata = {
18        'bucket': bucket,
19        'key': key,
20        'local_path': local_path,
21        'file_size_bytes': file_size
22    }
23    print(f"File metadata: {json.dumps(metadata)}")
24
25    return metadata
26
27if __name__ == '__main__':
28    event = json.loads(os.environ.get('TASK_EVENT', '{}'))
29
30    detail = event.get('detail', {})
31    bucket = detail.get('bucket', {}).get('name')
32    key = detail.get('object', {}).get('key')
33
34    if not bucket or not key:
35        print("Error: No bucket or key in event")
36        sys.exit(1)
37
38    # Validate inputs
39    if not isinstance(key, str) or '/..' in key or len(key) > 1024:
40        print("Error: Invalid S3 key")
41        sys.exit(1)
42
43    metadata = process_file(bucket, key)
44    print(f"Done. File size: {metadata['file_size_bytes']} bytes")