s3-buckets-analysis.py

import json
import boto3
import os
import logging
import time
import re

logger = logging.getLogger()
logger.setLevel(logging.INFO)

sts = boto3.client('sts')

role_name = 'crossaccount_role'
accounts = ['123456789012', '123456789012']
sourceaccount = sts.get_caller_identity()['Account']

def aws_session(account_id):
    if str(account_id) == str(sourceaccount):
        session = boto3.Session()
        return session
    else:
        try:
            role_arn = 'arn:aws:iam::' + account_id + ':role/' + role_name
            logger.info('Trying to assume role: ' + str(role_arn))
            if account_id:
                try:
                    response = sts.assume_role(RoleArn=role_arn, RoleSessionName="AssumedRoleSession")
                    session = boto3.Session(
                        aws_access_key_id=response['Credentials']['AccessKeyId'],
                        aws_secret_access_key=response['Credentials']['SecretAccessKey'],
                        aws_session_token=response['Credentials']['SessionToken'])
                    return session
                except Exception as e:
                    pass
        except:
            pass

def main():
    # Run across all stated
    empty_buckets = 0
    near_empty_buckets = 0
    used_buckets = 0
    for account in accounts:
        session = aws_session(str(account))

        if session:
            print("Account: {}".format(account))
            s3_client = session.client('s3')
            buckets = s3_client.list_buckets()
            for bucket in buckets['Buckets']:
                s3Bucket = session.resource('s3').Bucket(bucket['Name'])
                objects = s3Bucket.objects.all()
                obj_count = 0
                obj_size = 0
                try:
                    for s3object in objects:
                        obj_count += 1
                        if obj_count > 5:
                            break
                        obj_size += s3object.size
                    obj_size_mbs = obj_size // 1048576
                    if obj_count == 0:
                        empty_buckets += 1
                        print("\t{}: Empty Bucket".format(bucket['Name']))
                    elif obj_count <= 5:
                        near_empty_buckets += 1
                        print("\t{}: {} files taking {}MB".format(bucket['Name'], obj_count, obj_size_mbs))
                    else:
                        used_buckets += 1
                except:
                    print("Unable to access bucket: {}".format(bucket['Name']))
        else:
            print("Unable To Access Account: {}".format(account))
    print()
    print("Empty Buckets: {}".format(empty_buckets))
    print("Near Empty Buckets: {}".format(near_empty_buckets))
    print("Used Buckets: {}".format(used_buckets))

if __name__ == "__main__":
    main()