Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.

cottoncandy logo

Welcome to cottoncandy!

Build Status DOI

sugar for s3

What is cottoncandy?

A python scientific library for storing and accessing numpy array data on S3. This is achieved by reading arrays from memory and downloading arrays directly into memory. This means that you don't have to download your array to disk, and then load it from disk into your python session.

This library relies heavily on boto3

Try it out!

Using cottoncandy to explore the Allen Brain Observatory data: Launch Notebook!

Using cottoncandy to explore OpenNeuro nifti data: Launch notebook!


With pip:

$ pip install cottoncandy

Directly from the repo:

Clone the repo from GitHub and do the usual python install from the command line

$ git clone
$ cd cottoncandy
$ sudo python install

Configuration file

Upon first use, cottoncandy will create a configuration file. This configuration file allows you to enter your S3 and Google Drive credentials and set many other options. See the default configuration file.

The configuration file is created the first time you import cottoncandy and it is stored under:

  • Linux: ~/.config/cottoncandy/options.cfg
  • MAC OS: ~/Library/Application Support/cottoncandy/options.cfg
  • Windows (not supported): C:\Users\<username>\AppData\Local\<AppAuthor>\cottoncandy\options.cfg

By default, cottoncandy sets object and bucket permissions to authenticated-read. If you wish to keep all your objects private, modify your configuration file and set default_acl = private. See AWS ACL overview for more information on S3 permissions.

Advanced (for admins): One can customize the cottoncandy system install by cloning the repo and modifying defaults.cfg. For example, one can set the default encyption key across the system for all users (key = SoMeEncypTionKey). When a user first uses cottoncandy, this deault value will be copied to their personal configuration file. Note however that the user can still overwrite that value.

Getting started

Setup the connection (endpoint, access and secret keys can be specified in the configuration file instead)::

>>> import cottoncandy as cc
>>> cci = cc.get_interface('my_bucket',

Storing numpy arrays

>>> import numpy as np
>>> arr = np.random.randn(100)
>>> s3_response = cci.upload_raw_array('myarray', arr)
>>> arr_down = cci.download_raw_array('myarray')
>>> assert np.allclose(arr, arr_down)

Storing dask arrays

>>> arr = np.random.randn(100,600,1000)
>>> s3_response = cci.upload_dask_array('test_dim', arr, axis=-1)
>>> dask_object = cci.download_dask_array('test_dim')
>>> dask_object
dask.array<array, shape=(100, 600, 1000), dtype=float64, chunksize=(100, 600, 100)>
>>> dask_slice = dask_object[..., :200]
>>> dask_slice
dask.array<getitem..., shape=(100, 600, 1000), dtype=float64, chunksize=(100, 600, 100)>
>>> downloaded_data = np.asarray(dask_slice) # this downloads the array
>>> downloaded_data.shape
(100, 600, 200)

Command-line search

>>> cci.glob('/path/to/*/file01*.grp/image_data')
>>> cci.glob('/path/to/my/file02*.grp/*')

File system-like object browsing

>>> import cottoncandy as cc
>>> browser = cc.get_browser('my_bucket_name',
>>> browser.sweet_project.sub<TAB>
>>> browser.sweet_project.sub01_awesome_analysis_DOT_grp
<cottoncandy-group <bucket:my_bucket_name> (sub01_awesome_analysis.grp: 3 keys)>
>>> browser.sweet_project.sub01_awesome_analysis_DOT_grp.result_model01
<cottoncandy-dataset <bucket:my_bucket_name [1.00MB:shape=(10000)]>

Connection settings (S3 only)

cottoncandy allows users to modify connection settings via botocore. For example, the user can define the connection time out for downloads, and the number of times to retry dropped S3 requests.

from botocore.client import Config
config = Config(connect_timeout=60, read_timeout=60, retries=dict(max_attempts=10))
cci = cc.get_interface('my_bucket_name', config=config)

Google Drive backend

cottoncandy can also use Google Drive as a back-end. This equires a client_secrets.json file in your ~/.config/cottoncandy folder and the pydrive package.

See the Google Drive setup instructions for more details.

>>> import cottoncandy as cc
>>> cci = cc.get_interface(backend='gdrive')


cottoncandyprovides a transparent encryption interface for AWS S3 and Google Drive. This requires the pycrypto package.

WARNING: Encryption is an advance feature. Make sure to create a backup of the encryption keys (stored in ~/.config/cottoncandy/options.cfg). If you lose your encryption keys you will not be able to recover your data!

>>> import cottoncandy as cc
>>> cci = cc.get_encrypted_interface('my_bucket_name',


  • If you find any issues with cottoncandy, please report it by submitting an issue on GitHub.
  • If you wish to contribute, please submit a pull request. Include information as to how you ran the tests and the full output log if possible. Running tests on AWS can incur costs.

Cite as


Nunez-Elizalde AO, Gao JS, Zhang T, Gallant JL (2018). cottoncandy: scientific python package for easy cloud storage. Journal of Open Source Software, 3(28), 890,