SignURLarity

Fast, lightweight S3 client focused on presigned URL generation. SignURLarity provides a boto3-compatible API with significantly better performance by avoiding the boto3 dependency overhead.

Features

• Fast presigned URL generation (faster than boto3)
• Async support with AsyncClient
• Connection pooling for better performance
• S3-compatible services (AWS S3, MinIO, etc.)

Performance

When creating a Pull Request, performance tests with respects to boto are ran and added as comment. Obviously, due to the nature of github Actions, the numbers may vary (a lot). Just as an example, here are the numbers of one of the later PR (the bigger the better). See below to run it yourself

Test	3.11	3.12	3.13	3.14
create_bucket_aio	0.888	0.953	0.979	1.071
create_bucket_aio_cm	0.939	0.935	0.946	0.976
create_bucket_sync	1.145	1.194	1.142	1.198
create_bucket_sync_cm	1.141	1.144	1.171	1.173
generate_presigned_post_aio	8.516	8.722	8.606	8.804
generate_presigned_post_aio_cm	8.504	8.742	8.465	8.890
generate_presigned_post_sync	6.845	6.468	6.875	6.995
generate_presigned_post_sync_cm	6.713	6.465	6.638	6.768
generate_presigned_url_aio	15.597	15.490	15.075	17.026
generate_presigned_url_aio_cm	15.599	15.137	14.784	17.059
generate_presigned_url_sync	15.361	14.835	13.976	16.008
generate_presigned_url_sync_cm	14.739	14.806	14.014	16.203
head_bucket_aio	0.996	1.080	1.001	1.009
head_bucket_aio_cm	0.996	1.265	0.997	1.071
head_bucket_sync	1.468	1.522	1.456	1.522
head_bucket_sync_cm	1.211	1.147	1.505	1.507
head_object_aio	0.877	0.916	0.834	0.913
head_object_aio_cm	0.966	0.895	0.885	0.900
head_object_sync	1.516	1.443	0.898	1.584
head_object_sync_cm	1.493	1.547	1.511	1.470

Installation

pip install signurlarity

Quick Start

Synchronous Client

from signurlarity import Client

with Client(
    endpoint_url="https://s3.us-west-2.amazonaws.com",
    aws_access_key_id="your-access-key",
    aws_secret_access_key="your-secret-key",
) as client:
    url = client.generate_presigned_url(
        "get_object",
        Params={"Bucket": "mybucket", "Key": "myfile.txt"},
        ExpiresIn=3600,
    )

Async Client

from signurlarity.aio import AsyncClient

async with AsyncClient(
    endpoint_url="https://s3.us-west-2.amazonaws.com",
    aws_access_key_id="your-access-key",
    aws_secret_access_key="your-secret-key",
) as client:
    url = await client.generate_presigned_url(
        "get_object",
        Params={"Bucket": "mybucket", "Key": "myfile.txt"},
        ExpiresIn=3600,
    )

Documentation

For detailed documentation including:

• Complete API reference
• Advanced usage examples
• Error handling
• Additional methods (presigned POST, head operations, etc.)

Please refer to the docstrings in the source code:

• Client - Synchronous client
• AsyncClient - Asynchronous client
• S3Presigner - Low-level presigner

Development

Run tests

installation pixi

This will run functionnal tests. It will spawn docker container to test against rustfs, minio and moto

pixi run unit-test # add any pytest option you want

Any pytest argument can be added

pre-commit

SignURLarity uses pre-commit to format code and check for issues. The easiest way to use pre-commit is to run the following after cloning:

pixi run pre-commit install

This will result in pre-commit being ran automatically each time you run git commit. If you want to explicitly run pre-commit you can use:

pixi run pre-commit # (1)!
pixi run pre-commit --all-files # (2)!

1. Runs pre-commit only for files which are uncommitted or which have been changed.
2. Runs pre-commit for all files even if you haven't changed them.

Benchmark

For a full performance comparison, run

pixi run full-benchmark /whatever/outputdir

This will compare the results of boto and signurlarity against rustfs for python version 3.11, 3.12, 3.13 and 3.14, and generate json files in the output directoty

If you want to run it for a specific version only:

pixi run -e py314 benchmark --test-results-dir=/whatever/you/want

you can then display it with

pixi run -e py314 display-benchmark-comparison --test-results-dir=/whatever/you/want

Profiling tests

A few profiling tests are available

pixi run -e py314 profile-test -s --test-results-dir=/whatever/you/want

This will generate prof files

You can convert it in svg and open it in your web browser like so

pixi shell -e py314
flameprof --format=log profile_generate_presigned_post/presigned_post.prof | flamegraph > profile_generate_presigned_post/presigned_post.svg

: