S3
The s3 driver downloads an object from Amazon S3 or an S3-compatible endpoint and opens it with
the detected data driver.
URL format
s3://<bucket>/<object>[?region=<region>]
s3://[<access_key_id>:<secret_access_key>@]<host>[:<port>]/<bucket>/<object>?scheme=<http|https>
Options
| Option | Behavior / default |
|---|---|
region | AWS region; defaults to the AWS SDK configuration chain. |
session_token | Session token accompanying explicit URL access-key credentials. |
scheme | Set http or https to use the URL host and port as a custom endpoint. Without it, the AWS SDK chooses the service endpoint. |
force_path_style | Force path-style bucket addressing when true; defaults to the SDK setting. |
Examples
rsql --url 's3://example-bucket/people.parquet?region=us-east-1' -- 'SELECT * FROM people LIMIT 10;'
rsql --url 's3://test:test@localhost:9000/bucket/people.csv?scheme=http&force_path_style=true®ion=us-east-1'
Usage notes
Without scheme, the URL host is the bucket name. With scheme, the host is a custom endpoint and
the first path segment is the bucket. Custom endpoints default to port 443 unless specified.
Credentials default to the AWS SDK configuration chain. The object is downloaded into a temporary directory, then opened by an enabled format driver. Query parameters are forwarded to that driver. Queries do not upload changes back to S3. See Parquet and CSV for format behavior.