如何设置一个 Django 项目与 Django 存储和 Amazon S3,但不同的文件夹静态文件和媒体文件?

我正在配置一个 Django 项目,它使用服务器文件系统来存储应用程序静态文件(STATIC_ROOT)和用户上传的文件(MEDIA_ROOT)。

我现在需要在 Amazon 的 S3上托管所有这些内容,所以我为此创建了一个 bucket。使用带有 boto存储后端的 django-storages,我设法将收集到的静态数据上传到 S3存储桶:

MEDIA_ROOT = '/media/'
STATIC_ROOT = '/static/'


DEFAULT_FILE_STORAGE = 'storages.backends.s3boto.S3BotoStorage'
AWS_ACCESS_KEY_ID = 'KEY_ID...'
AWS_SECRET_ACCESS_KEY = 'ACCESS_KEY...'
AWS_STORAGE_BUCKET_NAME = 'bucket-name'
STATICFILES_STORAGE = 'storages.backends.s3boto.S3BotoStorage'

然后,我遇到了一个问题: 在 bucket 中没有使用 MEDIA_ROOTSTATIC_ROOT,因此 bucket 根包含静态文件和用户上传路径。

然后我就可以设定:

S3_URL = 'http://s3.amazonaws.com/%s' % AWS_STORAGE_BUCKET_NAME
STATIC_URL = S3_URL + STATIC_ROOT
MEDIA_URL = 'S3_URL + MEDIA_ROOT

并在模板中使用这些设置,但在使用 django-storages存储在 S3中时,静态/媒体文件没有区别。

怎样才能做到这一点?

谢谢!

33122 次浏览

I'm currently using this code in a separated s3utils module:

from django.core.exceptions import SuspiciousOperation
from django.utils.encoding import force_unicode


from storages.backends.s3boto import S3BotoStorage




def safe_join(base, *paths):
"""
A version of django.utils._os.safe_join for S3 paths.


Joins one or more path components to the base path component intelligently.
Returns a normalized version of the final path.


The final path must be located inside of the base path component (otherwise
a ValueError is raised).


Paths outside the base path indicate a possible security sensitive operation.
"""
from urlparse import urljoin
base_path = force_unicode(base)
paths = map(lambda p: force_unicode(p), paths)
final_path = urljoin(base_path + ("/" if not base_path.endswith("/") else ""), *paths)
# Ensure final_path starts with base_path and that the next character after
# the final path is '/' (or nothing, in which case final_path must be
# equal to base_path).
base_path_len = len(base_path) - 1
if not final_path.startswith(base_path) \
or final_path[base_path_len:base_path_len + 1] not in ('', '/'):
raise ValueError('the joined path is located outside of the base path'
' component')
return final_path




class StaticRootS3BotoStorage(S3BotoStorage):
def __init__(self, *args, **kwargs):
super(StaticRootS3BotoStorage, self).__init__(*args, **kwargs)
self.location = kwargs.get('location', '')
self.location = 'static/' + self.location.lstrip('/')


def _normalize_name(self, name):
try:
return safe_join(self.location, name).lstrip('/')
except ValueError:
raise SuspiciousOperation("Attempted access to '%s' denied." % name)




class MediaRootS3BotoStorage(S3BotoStorage):
def __init__(self, *args, **kwargs):
super(MediaRootS3BotoStorage, self).__init__(*args, **kwargs)
self.location = kwargs.get('location', '')
self.location = 'media/' + self.location.lstrip('/')


def _normalize_name(self, name):
try:
return safe_join(self.location, name).lstrip('/')
except ValueError:
raise SuspiciousOperation("Attempted access to '%s' denied." % name)

Then, in my settings module:

DEFAULT_FILE_STORAGE = 'myproyect.s3utils.MediaRootS3BotoStorage'
STATICFILES_STORAGE = 'myproyect.s3utils.StaticRootS3BotoStorage'

I got to redefine the _normalize_name() private method to use a "fixed" version of the safe_join() function, since the original code is giving me SuspiciousOperation exceptions for legal paths.

I'm posting this for consideration, if anyone can give a better answer or improve this one, it will be very welcome.

I think the following should work, and be simpler than Mandx's method, although it's very similar:

Create a s3utils.py file:

from storages.backends.s3boto import S3BotoStorage


StaticRootS3BotoStorage = lambda: S3BotoStorage(location='static')
MediaRootS3BotoStorage  = lambda: S3BotoStorage(location='media')

Then in your settings.py:

DEFAULT_FILE_STORAGE = 'myproject.s3utils.MediaRootS3BotoStorage'
STATICFILES_STORAGE = 'myproject.s3utils.StaticRootS3BotoStorage'

A different but related example (that I've actually tested) can be seen in the two example_ files here.

I think the answer is pretty simple and done by default. This is working for me on AWS Elastic Beanstalk with Django 1.6.5 and Boto 2.28.0:

STATICFILES_FINDERS = (
'django.contrib.staticfiles.finders.FileSystemFinder',
'django.contrib.staticfiles.finders.AppDirectoriesFinder',
)


TEMPLATE_LOADERS = (
'django.template.loaders.filesystem.Loader',
'django.template.loaders.app_directories.Loader',
)


DEFAULT_FILE_STORAGE = 'storages.backends.s3boto.S3BotoStorage'
STATICFILES_STORAGE = 'storages.backends.s3boto.S3BotoStorage'
AWS_ACCESS_KEY_ID = os.environ['AWS_ACCESS_KEY_ID']
AWS_SECRET_ACCESS_KEY = os.environ['AWS_SECRET_KEY']

The AWS keys are passed in from the container config file and I have no STATIC_ROOT or STATIC_URL set at all. Also, no need for the s3utils.py file. These details are handled by the storage system automatically. The trick here is that I needed to reference this unknown path in my templates correctly and dynamically. For example:

<link rel="icon" href="{% static "img/favicon.ico" %}">

That is how I address my favicon which lives locally (pre-deployment) in ~/Projects/my_app/project/my_app/static/img/favicon.ico.

Of course I have a separate local_settings.py file for accessing this stuff locally in dev environment and it does have STATIC and MEDIA settings. I had to do a lot of experimenting and reading to find this solution and it works consistently with no errors.

I understand that you need the static and root separation and considering that you can only provide one bucket I would point out that this method takes all the folders in my local environment under ~/Projects/my_app/project/my_app/static/and creates a folder in the bucket root (ie: S3bucket/img/ as in the example above). So you do get separation of files. For example you could have a media folder in the static folder and access it via templating with this:

{% static "media/" %}

I hope this helps. I came here looking for the answer and pushed a bit harder to find a simpler solution than to extend the storage system. Instead, I read the documentation about the intended use of Boto and I found that a lot of what I needed was built-in by default. Cheers!


File: PROJECT_NAME/custom_storages.py

from django.conf import settings
from storages.backends.s3boto import S3BotoStorage


class StaticStorage(S3BotoStorage):
location = settings.STATICFILES_LOCATION


class MediaStorage(S3BotoStorage):
location = settings.MEDIAFILES_LOCATION

File: PROJECT_NAME/settings.py

STATICFILES_LOCATION = 'static'
MEDIAFILES_LOCATION = 'media'


if not DEBUG:
STATICFILES_STORAGE = 'PROJECT_NAME.custom_storages.StaticStorage'
DEFAULT_FILE_STORAGE = 'PROJECT_NAME.custom_storages.MediaStorage'
AWS_ACCESS_KEY_ID = 'KEY_XXXXXXX'
AWS_SECRET_ACCESS_KEY = 'SECRET_XXXXXXXXX'
AWS_STORAGE_BUCKET_NAME = 'BUCKET_NAME'
AWS_HEADERS = {'Cache-Control': 'max-age=86400',}
AWS_QUERYSTRING_AUTH = False

And run: python manage.py collectstatic

If you want to have subfolders even before media or static seperations, you can use AWS_LOCATION on top of bradenm answer. Reference: https://django-storages.readthedocs.io/en/latest/backends/amazon-S3.html#usage

AWS_STORAGE_BUCKET_NAME = 'bucket_name'
AWS_LOCATION = 'path1/path2/'

Bradenm's answer is outdated and doesn't work so I updated it in March 2021.

Updated One:


Create a s3utils.py in the same folder of "settings.py":

from storages.backends.s3boto3 import S3Boto3Storage


StaticRootS3Boto3Storage = lambda: S3Boto3Storage(location='static')
MediaRootS3Boto3Storage  = lambda: S3Boto3Storage(location='media')

Then, add 2 lines of code to settings.py and change "myproject" to your folder name:

DEFAULT_FILE_STORAGE = 'myproject.s3utils.MediaRootS3Boto3Storage'
STATICFILES_STORAGE = 'myproject.s3utils.StaticRootS3Boto3Storage'

The updated one has multiple "3s" as I emphasize below.

s3utils.py:

from storages.backends.s3boto"3" import S3Boto"3"Storage
    

StaticRootS3Boto"3"Storage = lambda: S3Boto"3"Storage(location='static')
MediaRootS3Boto"3"Storage  = lambda: S3Boto"3"Storage(location='media')

settings.py:

DEFAULT_FILE_STORAGE = 'myproject.s3utils.MediaRootS3Boto"3"Storage'
STATICFILES_STORAGE = 'myproject.s3utils.StaticRootS3Boto"3"Storage'

Check and compare with Bradenm's (outdated) answer.

"I respect Bradenm's answer."