python AWS boto3 create presigned url for file upload

Question:

I’m writing a django backend for an application in which the client will upload a video file to s3. I want to use presigned urls, so the django server will sign a url and pass it back to the client, who will then upload their video to s3. The problem is, the generate_presigned_url method does not seem to know about the s3 client upload_file method…

Following this example, I use the following code to generate the url for upload:

s3_client = boto3.client('s3')
try:
    s3_object_name = str(uuid4()) + file_extension
    params = {
        "file_name": local_filename,
        "bucket": settings.VIDEO_UPLOAD_BUCKET_NAME,
        "object_name": s3_object_name,
    }
    response = s3_client.generate_presigned_url(ClientMethod="upload_file",
                                                Params=params,
                                                ExpiresIn=500)
except ClientError as e:
    logging.error(e)
    return HttpResponse(503, reason="Could not retrieve upload url.")

When running it I get the error:

File “/Users/bridgedudley/.local/share/virtualenvs/ShoMe/lib/python3.6/site-packages/botocore/signers.py”, line 574, in generate_presigned_url
operation_name = self._PY_TO_OP_NAME[client_method]
KeyError: ‘upload_file’

which triggers the exception:

botocore.exceptions.UnknownClientMethodError: Client does not have method: upload_file

Afer debugging I found that the self._PY_TO_OP_NAME dictionary only contains a subset of the s3 client commands offered here:
enter image description here

scrolling down to “upload”…
enter image description here

No upload_file method! I tried the same code using “list_buckets” and it worked perfectly, giving me a presigned url that listed the buckets under the signer’s credentials.

So without the upload_file method available in the generate_presigned_url function, how can I achieve my desired functionality?

Thanks!

Asked By: Bridge Dudley

||

Answers:

You should be able to use the put_object method here. It is a pure client object, rather than a meta client object like upload_file. That is the reason that upload_file is not appearing in client._PY_TO_OP_NAME. The two functions do take different inputs, which may necessitate a slight refactor in your code.

put_object: https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/s3.html#S3.Client.put_object

Answered By: eatsfood

In addition to the already mentioned usage of:

boto3.client('s3').generate_presigned_url('put_object', Params={'Bucket':'your-bucket-name', 'Key':'your-object-name'})

You can also use:

boto3.client('s3').generate_presigned_post('your-bucket_name', 'your-object_name')

Reference: https://boto3.amazonaws.com/v1/documentation/api/latest/guide/s3-presigned-urls.html#generating-a-presigned-url-to-upload-a-file

Sample generation of URL:

import boto3

bucket_name = 'my-bucket'
key_name = 'any-name.txt'

s3_client = boto3.client('s3')
upload_details = s3_client.generate_presigned_post(bucket_name, key_name)

print(upload_details)

Output:

{'url': 'https://my-bucket.s3.amazonaws.com/', 'fields': {'key': 'any-name.txt', 'AWSAccessKeyId': 'QWERTYUOP123', 'x-amz-security-token': 'a1s2d3f4g5h6j7k8l9', 'policy': 'z0x9c8v7b6n5m4', 'signature': 'qaz123wsx456edc'}}

Sample uploading of file:

import requests

filename_to_upload = './some-file.txt'

with open(filename_to_upload, 'rb') as file_to_upload:
    files = {'file': (filename_to_upload, file_to_upload)}
    upload_response = requests.post(upload_details['url'], data=upload_details['fields'], files=files)

print(f"Upload response: {upload_response.status_code}")

Output:

Upload response: 204

Additional notes:

As documented:

The credentials used by the presigned URL are those of the AWS user
who generated the URL.

Thus, make sure that the entity that would execute this generation of a presigned URL allows the policy s3:PutObject to be able to upload a file to S3 using the signed URL. Once created, it can be configured through different ways. Some of them are:

As an allowed policy for a Lambda function

Or through boto3:

s3_client = boto3.client('s3',
        aws_access_key_id="your-access-key-id",
        aws_secret_access_key="your-secret-access-key",
        aws_session_token="your-session-token",  # Only for credentials that has it
    )

Or on the working environment:

# Run in the Linux environment
export AWS_ACCESS_KEY_ID="your-access-key-id"
export AWS_SECRET_ACCESS_KEY="your-secret-access-key"
export AWS_SESSION_TOKEN="your-session-token",  # Only for credentials that has it

Or through libraries e.g. django-storages for Django

The accepted answer doesn’t let you post your data to S3 from your client. This will:

import boto3

s3_client = boto3.client('s3',
  aws_access_key_id="AKIA....",
  aws_secret_access_key="G789...",
)

s3_client.generate_presigned_url('put_object', Params={
  'Bucket': 'cat-pictures',
  'Key': 'whiskers.png',
  'ContentType': 'image/png', # required!
})

Send that to your front end, then in JavaScript on the frontend:

fetch(url, {
  method: "PUT",
  body: file,
})

Where file is a File object.

Answered By: duhaime