simples3 - Simple, quick Amazon AWS S3 interface

simples3 is a fairly simple, decently quick interface to Amazon’s S3 storage service written in pure Python.

It grew out of frustration with other libraries that were either written too pragmatically (slow), too bloatedly, or just half-done.

The aims and goals are, in order of importance,

  1. simplicity,
  2. decent speed,
  3. non-intrusiveness.

It really is designed to fit into programmer memory. The three basic operations are as easy as with dictionaries.

Out of simplicity comes no dependencies - the code relies solely on Python standard libraries.

simples3 requires Python 2.5 or a newer 2.x release.


A simple Amazon AWS S3 interface

Access to a bucket is done via the S3Bucket class. It has three required arguments:

>>> s = S3Bucket(bucket,
...              access_key=access_key,
...              secret_key=secret_key)
>>> print s  
<S3Bucket ... at ''>

or if you’d like to use the use-any-domain-you-want stuff, set base_url to something like

>>> s = S3Bucket(bucket,
...              access_key=access_key,
...              secret_key=secret_key,
...              base_url=base_url)
>>> print s  
<S3Bucket ... at 'http...'>

Note that missing slash above, it’s important. Think of it as “The prefix to which all calls are made.” Also the scheme can be https or regular http, or any other urllib2-compatible scheme (as in you could register your own scheme.)

Now, let’s start doing something useful. Start out by putting a simple file onto there:

>>> s.put("my file", "my content")

Alright, and fetch it back:

>>> f = s.get("my file")
'my content'

Nice and tidy, but what if we want to know more about our fetched file? Easy:

>>> f.s3_info["modify"]  
>>> f.s3_info["mimetype"]
>>> f.s3_info.keys()
['mimetype', 'modify', 'headers', 'date', 'size', 'metadata']
>>> f.close()

Note that the type was octet stream. That’s simply because we didn’t specify anything else. Do that using the mimetype keyword argument:

>>> s.put("my new file!", "Improved content!\nMultiple lines!",
...       mimetype="text/plain")

Let’s be cool and use the very Pythonic API to do fetch:

>>> f = s["my new file!"]
>>> print
Improved content!
Multiple lines!
>>> f.s3_info["mimetype"]
>>> f.close()

Great job, huh. Now, let’s delete it:

>>> del s["my new file!"]

Could’ve used the delete method instead, but we didn’t.

If you just want to know about a key, ask and ye shall receive:

>>> from pprint import pprint
>>> s["This is a testfile."] = S3File("Hi!", metadata={"hairdo": "Secret"})
>>> pprint("This is a testfile."))  
{'date': datetime.datetime(...),
 'headers': {'content-length': '3',
             'content-type': 'application/octet-stream',
             'date': '...',
             'etag': '"..."',
             'last-modified': '...',
             'server': 'AmazonS3',
             'x-amz-id-2': '...',
             'x-amz-meta-hairdo': 'Secret',
             'x-amz-request-id': '...'},
 'metadata': {'hairdo': 'Secret'},
 'mimetype': 'application/octet-stream',
 'modify': datetime.datetime(...),
 'size': 3}

Notable is that you got the metadata parsed out in the metadata key. You might also have noticed how the file was uploaded, using an S3File object like that. That’s a nicer way to do it, in a way.

The S3File simply takes its keyword arguments, and passes them on to put later. Other than that, it’s a str subclass.

And the last dict-like behavior is in tests:

>>> "This is a testfile." in s
>>> del s["This is a testfile."]
>>> "This is a testfile." in s

You can also set a canned ACL using put, which is very simple:

>>> s.put("test/foo", "test", acl="public-read")
>>> s.put("test/bar", "rawr", acl="public-read")

Boom. What’s more? Listing the bucket:

>>> for (key, modify, etag, size) in s.listdir(prefix="test/"):
...     print "%r (%r) is size %r, modified %r" % (key, etag, size, modify)
'test/bar' ('"..."') is size 4, modified datetime.datetime(...)
'test/foo' ('"..."') is size 4, modified datetime.datetime(...)

That about sums the basics up.

Learn more at Advanced usage.


If you’re an IRC geek, stop by at #sendapatch on Know that the channel is idle sometimes. If you want a more direct channel, I suggest prepending ludvig@ to the current hostname (which is and use that as an e-mail address to write to.