Skip to main content

A Django/Django-Storages threaded S3 chunk uploader

Project description

A Django file handler to manage piping uploaded files directly to S3 without passing through the server’s file system. The uploader uses multiple threads to speed up the upload of larger files. This package relies on Django and Django-Storages, allowing the use of the S3 storages FileField but changes the upload behaviour to bypass local file system.

Quick start

  1. Install the package:

    pip install s3chunkuploader

  2. Set the Django FILE_UPLOAD_HANDLERS setting:

    FILE_UPLOAD_HANDLERS = (‘s3chunkuploader.file_handler.S3FileUploadHandler’,)

How it works

The File Handler intercepts the file upload multipart request at the door, and as chunks of the file are received from the browser, they are collectd into an internal queue within custom ThreadPoolWorker. When the queue surpasses a configurable size (by default 5MB which is the minimum Part size for S3 multipart upload), it is submitted to the Thread Pool as a Future which will then resolve. Once all the chunks are uploaded and all the futures are resolved the upload is complete. By default 10 threads are used which means a 100MB file upload can be potentially sent as 20 5MB parts to S3.

The FileHandler ultimately returns a ‘dummy’ django-storages S3Boto3StorageFile which is compatible with the storages S3 File Field, but was not actually used to upload a full file. The file is also enhanced with two additional attributes:

original_name: The original file name uploaded file_size: The actual full file size uploaded

It is recommended to bypass csrf checks on the upload file view as the csrf check will read the POST params before the handler is used. A replacement file field S3FileField is provided in and is satisfied with the S3 object key

By default the S3 key will be generated based on the settings provided. However, it is possible to define a custom function to derive the S3 object key by providing a full dot notated path to the function in the S3_GENERATE_OBJECT_KEY_FUNCTION settings parameter.


The following settings are expected in your Django application (only 2 are required)




Required. Your AWS access key


Required. The AWS secret


Required. The S3 bucket to use


Optional. Region of S3 bucket


Optional. Document root for all uploads (prefix)


Optional [True]. Append the current datetime sring to the uploaded file name


Optional [__prefix]. A query param key name which provides additional prefix for the object key on S3


Optional [5MB]. The part size in bytes to upload to S3


Optional [None]. The maximum file size in bytes for an individual file.


Optional [None]. The s3 endpoint url which overrides the default


Optional [False]. When True, runs the filename through Django’s slugify function to sanitise it.


Optional [None]. A function to generate the S3 key, receiving the request object and filename as arguments.


Optional [None]. A full custom S3 endpoint url (was S3_ENDPOINT_URL in previous version)

Unit Tests

Unit tests can be executing by running python -m unittest from the projects root

Change Log

  • 0.9: The optional setting S3_ENDPOINT_URL was renamed to AWS_S3_ENDPOINT_URL to align with django-storages .

  • 0.10: If content_length is not provided MAX_UPLOAD_SIZE cannot evaluate against it.

  • 0.11: Prefixed settings keys

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

s3chunkuploader-0.13.tar.gz (8.8 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page