IAM Builder

A python script to generate an IAM policy based on a yaml or json configuration.

To install:

# Most stable
pip install iam-builder

# OR directly from github
pip install git+git://github.com/moj-analytical-services/iam_builder.git#egg=iam_builder

To use the command line interface:

iam_builder -c examples/iam_config.yaml -o examples/iam_policy.json

• -c is the path to your iam configuration (either a yaml or json file).
• -o is the path to your output iam policy (needs to be a json file).

Or to do the same thing in python:

import yaml
import json
from iam_builder.iam_builder import build_iam_policy

with open('examples/iam_config.yaml') as f:
  config = yaml.load(f, Loader=yaml.FullLoader)

iam_policy = build_iam_policy(config)

with open('examples/iam_policy.json', "w+") as f:
  json.dump(iam_policy, f, indent=4, separators=(',', ': '))

Both scripts will create the output iam_policy seen in the examples folder. You can also see more example configs by looking in the unit tests.

Your config file can be either a yaml or json file.

The example yaml (iam_config.yaml) looks this:

iam_role_name: iam_role_name

athena:
  write: false

glue_job: true

secrets: true

secretsmanager:
  read_only:
    - test_secret

s3:
  read_only:
    - test_bucket_read_only/*

  write_only:
    - test_bucket_write_only/*
    - test_bucket_read_only/write_only_folder/*

  read_write:
    - test_bucket_read_write/*
    - test_bucket_read_only/write_folder/*

  deny:
    - test_bucket_read_write/sensitive_table/*

kms:
  - test_kms_key_arn

bedrock: true

Whilst the example json (iam_config.json) looks like this:

{
  "iam_role_name": "iam_role_name",
  "athena": {
    "write": false
  },
  "glue_job": true,
  "secrets": true,
  "s3": {
    "read_only": [
      "test_bucket_read_only/*"
    ],
    "write_only": [
      "test_bucket_write_only/*",
      "test_bucket_read_only/write_only_folder/*"
    ],
    "read_write": [
      "test_bucket_read_write/*",
      "test_bucket_read_only/write_folder/*"
    ]
  },
  "kms": ["test_kms_key_arn"],
  "bedrock": true,
  "cloudwatch_athena_query_executions": true
}

• iam_role_name: The role name of your airflow job; required if you want to run glue jobs or access secrets.

• athena: Can have two keys.

  • write: Either true or false. If false then only read access to Athena (cannot create, delete or alter tables, databases and partitions). If true then the role will also have the ability to do stuff like CTAS queries, DROP TABLE, CREATE DATABASE, etc.
  • dump_bucket: The location in S3 (either an S3 path or a list of S3 paths) for temporarily storing the results of queries. This defaults to mojap-athena-query-dump and should not normally need changing.

• is_cadet_deployer: Boolean; Gives access to a highly empowered Glue role for Create-A-Derived-Table deployments. Will fail to apply if the iam_role_name doesn't include cadet in the string. Gives the user full control over all glue and athena structures in the named account.

• glue_job: Boolean; must be set to true to allow role to run glue jobs. If false or absent role will not be able to run glue jobs.

• secrets: Boolean or string; must be set to true or "read" to allow role to access secrets from AWS Parameter Store, and readwrite to provide read/write access. If false or absent role will not be able to access secrets.

• s3: Can have up to 4 keys: read_only, write_only, read_write, and deny. Each key describes the level of access you want your iam policy to have with each s3 path. More details below:

  • read_only: A list of s3 paths that the iam_role should be able to access (read only). Each item in the list should either be a path to a object or finish with /* to denote that it can access everything within that directory. Note the S3 paths don't start with s3:// in the config.

  • write_only: A list of s3 paths that the iam_role should be able to access (write only). Each item in the list should either be a path to a object or finish with /* to denote that it can access everything within that directory. Note the S3 paths don't start with s3:// in the config.

  • read_write: A list of s3 paths that the iam_role should be able to access (read and write). Each item in the list should either be a path to a object or finish with /* to denote that it can access everything within that directory. Note the S3 paths don't start with s3:// in the config.

  • deny: A list of s3 paths that the iam_role should not be able to access. This should be used to add exceptions to wildcarded access to folders, for example excluding sensitive tables in order to provide basic access to a database. Each item in the list should either be a path to a object or finish with /* to denote that it can access everything within that directory. Note the S3 paths don't start with s3:// in the config.

• kms: A list of kms arns that the iam_role should be able to access. Can call the DescribeKey, GenerateDataKey, Decrypt, Encrypt and ReEncrypt operations.

• secretsmanager: A secret that the iam_role should be able to access. Can call the GetSecretValue, DescribeSecret and ListSecrets operations.

• bedrock: Boolean; must be set to true to allow role to interact with Amazon Bedrock. If false or absent role will not be able to interact with Amazon Bedrock.

• cloudwatch_athena_query_executions Boolean; must be set to true to allow role to read cloudtrail-athena-events log group. If false or absent role will not be able to read these cloudwatch logs.

How to update

When updating IAM builder, make sure to change the version number in pyproject.toml and describe the change in CHANGELOG.md.

If you have changed any dependencies in pyproject.yaml, run poetry update to update poetry.lock.

Once you have created a release in GitHub, a Github Action will run to publish the release on PyPI automatically.