django-s3-storage
django-s3-storage provides a Django Amazon S3 file storage.
Features
- Django file storage for Amazon S3.
- Django static file storage for Amazon S3.
- Works in Python 3!
Installation
- Install using
pip install django-s3-storage
. - Add
'django_s3_storage'
to yourINSTALLED_APPS
setting. - Set your
DEFAULT_FILE_STORAGE
setting to"django_s3_storage.storage.S3Storage"
. - Set your
STATICFILES_STORAGE
setting to"django_s3_storage.storage.StaticS3Storage"
or"django_s3_storage.storage.ManifestStaticS3Storage"
. - Configure your Amazon S3 settings (see Available settings, below).
Authentication settings
Use the following settings to authenticate with Amazon AWS.
# The AWS region to connect to.
AWS_REGION = "us-east-1"
# The AWS access key to use.
AWS_ACCESS_KEY_ID = ""
# The AWS secret access key to use.
AWS_SECRET_ACCESS_KEY = ""
# The optional AWS session token to use.
AWS_SESSION_TOKEN = ""
File storage settings
Use the following settings to configure the S3 file storage. You must
provide at least AWS_S3_BUCKET_NAME
.
# The name of the bucket to store files in.
AWS_S3_BUCKET_NAME = ""
# How to construct S3 URLs ("auto", "path", "virtual").
AWS_S3_ADDRESSING_STYLE = "auto"
# The full URL to the S3 endpoint. Leave blank to use the default region URL.
AWS_S3_ENDPOINT_URL = ""
# A prefix to be applied to every stored file. This will be joined to every filename using the "/" separator.
AWS_S3_KEY_PREFIX = ""
# Whether to enable authentication for stored files. If True, then generated URLs will include an authentication
# token valid for `AWS_S3_MAX_AGE_SECONDS`. If False, then generated URLs will not include an authentication token,
# and their permissions will be set to "public-read".
AWS_S3_BUCKET_AUTH = True
# How long generated URLs are valid for. This affects the expiry of authentication tokens if `AWS_S3_BUCKET_AUTH`
# is True. It also affects the "Cache-Control" header of the files.
# Important: Changing this setting will not affect existing files.
AWS_S3_MAX_AGE_SECONDS = 60 * 60 # 1 hours.
# A URL prefix to be used for generated URLs. This is useful if your bucket is served through a CDN.
AWS_S3_PUBLIC_URL = ""
# If True, then files will be stored with reduced redundancy. Check the S3 documentation and make sure you
# understand the consequences before enabling.
# Important: Changing this setting will not affect existing files.
AWS_S3_REDUCED_REDUNDANCY = False
# The Content-Disposition header used when the file is downloaded. This can be a string, or a function taking a
# single `name` argument.
# Important: Changing this setting will not affect existing files.
AWS_S3_CONTENT_DISPOSITION = ""
# The Content-Language header used when the file is downloaded. This can be a string, or a function taking a
# single `name` argument.
# Important: Changing this setting will not affect existing files.
AWS_S3_CONTENT_LANGUAGE = ""
# A mapping of custom metadata for each file. Each value can be a string, or a function taking a
# single `name` argument.
# Important: Changing this setting will not affect existing files.
AWS_S3_METADATA = {}
# If True, then files will be stored using AES256 server-side encryption.
# If this is a string value (e.g., "aws:kms"), that encryption type will be used.
# Otherwise, server-side encryption is not be enabled.
# Important: Changing this setting will not affect existing files.
AWS_S3_ENCRYPT_KEY = False
# The AWS S3 KMS encryption key ID (the `SSEKMSKeyId` parameter) is set from this string if present.
# This is only relevant if AWS S3 KMS server-side encryption is enabled (above).
AWS_S3_KMS_ENCRYPTION_KEY_ID = ""
# If True, then text files will be stored using gzip content encoding. Files will only be gzipped if their
# compressed size is smaller than their uncompressed size.
# Important: Changing this setting will not affect existing files.
AWS_S3_GZIP = True
# The signature version to use for S3 requests.
AWS_S3_SIGNATURE_VERSION = None
# If True, then files with the same name will overwrite each other. By default it's set to False to have
# extra characters appended.
AWS_S3_FILE_OVERWRITE = False
# If True, use default behaviour for boto3 of using threads when doing S3 operations. If gevent or similar
# is used it must be disabled
AWS_S3_USE_THREADS = True
# Max pool of connections for massive S3 interactions
AWS_S3_MAX_POOL_CONNECTIONS = 10
# Time to raise timeout when submitting a new file
AWS_S3_CONNECT_TIMEOUT = 60
Important: Several of these settings (noted above) will not affect
existing files. To sync the new settings to existing files, run
./manage.py s3_sync_meta django.core.files.storage.default_storage
.
These settings can be provided in field storage definition like this:
from django.db import models
from django_s3_storage.storage import S3Storage
storage = S3Storage(aws_s3_bucket_name='test_bucket')
class Car(models.Model):
name = models.CharField(max_length=255)
photo = models.ImageField(storage=storage)
Note: settings key in storage definition should be [lowercase]{.title-ref}.
Staticfiles storage settings
All of the file storage settings are available for the staticfiles
storage, suffixed with _STATIC
. You must provide at least
AWS_S3_BUCKET_NAME_STATIC
. Remember to run ./manage.py collectstatic
after changing your staticfiles storage backend.
The following staticfiles storage settings have different default values to their file storage counterparts.
AWS_S3_BUCKET_AUTH_STATIC = False
The following additional staticfiles storage settings also exist:
# For ManifestStaticS3Storage, how long the browser should cache md5-hashed filenames. This affects the expiry of
# authentication tokens if `AWS_S3_BUCKET_AUTH` is True. It also affects the "Cache-Control" header of the files.
# Important: Changing this setting will not affect existing files.
AWS_S3_MAX_AGE_SECONDS_CACHED_STATIC = 60 * 60 * 24 * 365 # 1 year.
Important: Several of these settings (noted above) will not affect
existing files. To sync the new settings to existing files, run
./manage.py s3_sync_meta django.contrib.staticfiles.storage.staticfiles_storage
.
Optimizing media file caching
The default settings assume that media file are private. This means that they are only accessible via S3 authenticated URLs, which is bad for browser caching.
To make media files public, and enable aggressive caching, make the
following changes to your settings.py
.
AWS_S3_BUCKET_AUTH = False
AWS_S3_MAX_AGE_SECONDS = 60 * 60 * 24 * 365 # 1 year.
Important: By making these changes, all media files will be public. Ensure they do not contain confidential information.
The default settings for staticfiles storage are already optimizing for aggressive caching.
Custom URLs
Sometimes the default settings aren\'t flexible enough and custom
handling of object is needed. For example, the Content-Disposition
might be set to force download of a file instead of opening it:
url = storage.url("foo/bar.pdf", extra_params={"ResponseContentDisposition": "attachment"})
Another example is a link to a specific version of the file (within a bucket that has versioning enabled):
url = storage.url("foo/bar.pdf", extra_params={"VersionId": "FRy3fTduRtqHsRAoNp0REzPJj_WunDfl"})
The extra_params
dict accepts the same parameters as
get_object().
Please note, however, that custom URLs will not work with
AWS_S3_PUBLIC_URL where the URL doesn\'t accept extra parameters, and
it will raise ValueError
.
Pre-signed URL uploads
Pre-signed URLs allow temporary access to S3 objects without AWS credentials. A pre-signed URL allows HTTP clients to upload files directly, improving performance and reducing the load on your server.
To generate a presigned URL allowing a file upload with HTTP PUT
:
url = storage.url("foo/bar.pdf", client_method="put_object")
Management commands
s3_sync_meta
Syncronizes the meta information on S3 files.
Several settings (noted above) will not affect existing files. Run this command to sync the new settings to existing files.
Example usage:
./manage.py s3_sync_meta django.core.files.storage.default_storage
IAM permissions
In order to use all features of django-s3-storages, either authenticate with your AWS root credentials (not recommended), or create a dedicated IAM role. The minimum set of permissions required by django-s3-storage is:
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Action": [
"s3:ListBucket"
],
"Resource": [
"arn:aws:s3:::my-bucket"
]
},
{
"Effect": "Allow",
"Action": [
"s3:PutObject",
"s3:PutObjectAcl",
"s3:GetObject",
"s3:GetObjectAcl",
"s3:DeleteObject"
],
"Resource": [
"arn:aws:s3:::my-bucket/*"
]
}
]
}
How does django-s3-storage compare with django-storages?
django-storages supports a variety of other storage backends, whereas django-s3-storage provides similar features, but only supports S3. It was originally written to support Python 3 at a time when the future of django-storages was unclear. It\'s a small, well-tested and self-contained library that aims to do one thing very well.
The author of django-s3-storage is not aware of significant differences in functionality with django-storages. If you notice some differences, please file an issue!
Migration from django-storages
If your are updating a project that used django-storages just for S3 file storage, migration is trivial.
Follow the installation instructions, replacing \'storages\' in
INSTALLED_APPS
. Be sure to scrutinize the rest of your settings file
for changes, most notably AWS_S3_BUCKET_NAME
for
AWS_STORAGE_BUCKET_NAME
.
Build status
This project is built on every push using the Travis-CI service.
Support and announcements
Downloads and bug tracking can be found at the main project website.
More information
The django-s3-storage project was developed by Dave Hall. You can get the code from the django-s3-storage project site.
Dave Hall is a freelance web developer, based in Cambridge, UK. You can usually find him on the Internet in a number of different places: