Copy S3 objects inside one bucket

Part 17 of AWS from Zero. Today we copy one object without downloading it first.

What we are learning

An S3 copy has a source bucket and key, plus a destination bucket and key. Both can be in the same bucket.

Before you run it

REGION="ap-south-1"
BUCKET="aws-zero-copy-inside-12345"
 
aws s3api create-bucket \
  --bucket "$BUCKET" \
  --region "$REGION" \
  --create-bucket-configuration LocationConstraint="$REGION"
 
echo "copy me" > source.txt
aws s3 cp source.txt "s3://$BUCKET/inbox/source.txt"

The command

aws s3api copy-object \
  --bucket "$BUCKET" \
  --copy-source "$BUCKET/inbox/source.txt" \
  --key "archive/source.txt"

S3 performs the copy on the service side. The destination key is a separate object.

Inspect the result

aws s3api list-objects-v2 \
  --bucket "$BUCKET" \
  --query "Contents[].{Key:Key,Size:Size,ETag:ETag}" \
  --output table

You should see both inbox/source.txt and archive/source.txt.

One tiny variation

Copy the object to a new name:

aws s3api copy-object \
  --bucket "$BUCKET" \
  --copy-source "$BUCKET/inbox/source.txt" \
  --key "archive/renamed.txt"

S3 has no standalone rename operation. A rename is normally a copy followed by deletion of the old key.

Common mistake

--copy-source is not an s3:// URI. It uses bucket/key:

bucket-name/path/to/object.txt

Keys containing spaces or special characters must be URL encoded in the copy source.

Cleanup

aws s3 rm "s3://$BUCKET/" --recursive
aws s3api delete-bucket --bucket "$BUCKET" --region "$REGION"
rm source.txt

Next, we will copy an object between two buckets.

AWS CLI copy-object reference