Skip to content

Instantly share code, notes, and snippets.

@alisade
Created November 21, 2022 19:15
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save alisade/1035102391bc5811ea155e26cb5a70a5 to your computer and use it in GitHub Desktop.
Save alisade/1035102391bc5811ea155e26cb5a70a5 to your computer and use it in GitHub Desktop.
merge existing files on s3 without downloading them first using multi part copy
# Using s3 copy with multipart uploads for efficient file merging
# limitation: min part size is 5MB
import boto3
from datetime import datetime
client = boto3.client('s3')
now = datetime.now()
timestamp = now.strftime("%Y%m%d_%H%M%S")
bucket = "bucket-testing-dev"
key = f'other/FILE_{timestamp}'
files = []
for item in client.list_objects_v2(Bucket=bucket, Prefix="other/test/")['Contents']:
files.append(item['Key'])
r = client.create_multipart_upload(Bucket=bucket, Key=key)
upload_id = r['UploadId']
mp = {}
mp['Parts'] = []
for i, file in enumerate(files):
r = client.upload_part_copy(Bucket=bucket, Key=key, CopySource={'Bucket': bucket, 'Key': file}, PartNumber=i+1, UploadId=upload_id)
mp['Parts'].append(r['CopyPartResult'])
mp['Parts'][i]['PartNumber'] = i + 1
del mp['Parts'][i]['LastModified']
client.complete_multipart_upload(Bucket=bucket, Key=key, UploadId=upload_id, MultipartUpload=mp)
@Anna-Kozhel
Copy link

Please work

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment