Async boto3 with Autogenerated JIT Data Classes


This library is forked from an internal project that works with a lot of backend AWS APIs, and I got tired of having to constantly parse the returned responses before being able to work with them for (most of the time), a few seconds. Any API-driven application that uses boto3 tends to suffer from being syncronous only. So this library solves a few major problems:

  • Enables both sync and async from the same client.

  • Client can be called implicitly without needing initialization, i.e. you can directly use AwsClient without needing to initialize.

  • Dynamically generates class functions based on the boto3.client child functions, and allows you to call them. This is useful in repl or ipython environments where type-hints are always helpful, especially when the names are so long.

  • Automatically initializes the aws client if its not initialized from the defaults, simply by calling it.

  • Translates all results into Automatically Generated Dataclasses through lazycls.

    • Can be disabled by setting aws.config.AutoCls = False


pip install --upgrade awspydk

from aws import AwsClient

# Sync Method
buckets = AwsClient.v1.s3_list_buckets(as_cls=True)

# Async Method
buckets = await AwsClient.v1.async_s3_list_buckets(as_cls=True)

Both yield the same results.
The underlying classes are auto-generated from Pydantic BaseModels, so anything you can do with Pydantic Models, you can do with these.

    'Buckets': [
        AwsS3Bucket(CreationDate=datetime.datetime(2021, 8, 25, 16, 42, 46, tzinfo=tzutc()), Name='...'),
        AwsS3Bucket(CreationDate=datetime.datetime(2021, 9, 2, 17, 54, 56, tzinfo=tzutc()), Name='...',
        AwsS3Bucket(CreationDate=datetime.datetime(2021, 9, 3, 4, 20, 10, tzinfo=tzutc()), Name='...'),
        AwsS3Bucket(CreationDate=datetime.datetime(2021, 9, 1, 20, 50, 33, tzinfo=tzutc()), Name='...'),
        AwsS3Bucket(CreationDate=datetime.datetime(2021, 9, 2, 4, 2, 28, tzinfo=tzutc()), Name='...')
    'Owner': AwsS3Owner(DisplayName='...', ID='...')

## Change Regions

## Change the defaut clients created
from aws.config import DefaultClients

## Modify to only create ec2 client
DefaultClients = {
    'ec2': 'ec2'

## Reset implicitly

BotoKwargs = {
    'AWS_PROFILE': ...,

## Reset Explicitly
AwsClient.reset(clients=DefaultClients, boto_kwargs=BotoKwargs)

Client Defaults

These are found in aws.config

AwsRegion = envToStr('AWS_REGION', 'us-east-1')
AutoCls = envToBool('AWSSDK_AUTOCLS', 'true')

## These are the default clients that will be autogenerated.
## Key is the shorthand, value is the actual AWS API Name in boto3
DefaultClients = {
    'ec2': 'ec2', 
    'ecr': 'ecr', 
    'r53' :'route53', 
    'acm': 'acm',
    'elb': 'elb',
    'elbv2': 'elbv2',
    'asg': 'autoscaling',
    's3': 's3'

## These are the default resources that will be autogenerated.
## Key is the shorthand, value is the actual AWS Resource API Name in boto3
DefaultResources = {
    'Ec2': 'ec2',
    'S3': 's3',
    'Iam': 'iam'

# These are the default filter args for querying
DefaultFilterArgs = {
    'string_only': True,
    'remove_null': True


View Github