Pulumi & Python

Pulumi supports writing your infrastructure as code in the Python language running on any supported version.

Install Python. To reduce potential issues with setting up your Python environment on Windows or macOS, you should install Python through the official Python installer.

Pulumi Programming Model

The Pulumi programming model defines the core concepts you will use when creating infrastructure as code programs using Pulumi. Concepts describes these concepts with examples available in Python. These concepts are made available to you in the Pulumi SDK.

The Pulumi SDK is available to Python developers as a package distributed on PyPI. To learn more, refer to the Pulumi SDK Reference Guide.

Inputs and Outputs

The Pulumi programming model includes a core concept of Input and Output values, which are used to track how outputs of one resource flow in as inputs to another resource. This concept is important to understand when getting started with Python and Pulumi, and the Inputs and Outputs documentation is recommended to get a feel for how to work with this core part of Pulumi in common cases.

In Python, inputs that are objects, that is inputs that group multiple values together, can be represented either as classes or as dictionary literals. The types for the argument classes have the suffix Args, whereas the types for the dictionaries have the suffix ArgsDict. Both types take the same arguments, but the dictionary types are often more concise.

This example shows two ways to create an s3.Bucket resource in Python, once using a dictionary literal and once using a class:

import pulumi_aws as aws

bucket1 = aws.s3.Bucket(
    "my-bucket-with-dictionary-literals",
    website={
        "error_document": "error.html",
        "index_document": "index.html",
    },
)

bucket2 = aws.s3.Bucket(
    "my-bucket-with-args",
    website=aws.s3.BucketWebsiteArgs(
        error_document="error.html",
        index_document="index.html",
    ),
)

Blocking and Asynchronous Code

A Python Pulumi program is single threaded and the Pulumi runtime creates an event loop to enable the runtime to be asynchronous.

Given these constraints, Blocking and Async Python with Pulumi gives some recommendations on using blocking and asynchronous code within Python Pulumi programs.

Using Pulumi PyPI Packages

Virtual Environments

It is not required, but we recommend using a virtual environment to isolate the dependencies of your projects and ensure reproducibility between machines.

When creating a new Python project with pulumi new, you are offered the choice between pip (default) and poetry to manage your dependencies. If you choose pip, Pulumi will create a virtual environment in a venv directory with required dependencies from requirements.txt installed in it. If you choose poetry, Pulumi will create a pyproject.toml file and run Poetry to create a virtual environment in its default location. Pulumi requires Poetry version 1.8.0 or later. Pulumi will automatically use this virtual environment when running the program.

This behavior is controlled by the toolchain and virtualenv runtime options in Pulumi.yaml.

To use pip, set the toolchain option to pip, along with the virtualenv option:

runtime:
  name: python
  options:
    toolchain: pip
    virtualenv: venv

virtualenv is the path to a virtual environment to use.

To use Poetry, set the toolchain option to poetry:

runtime:
  name: python
  options:
    toolchain: poetry

To further configure poetry, you can provide a poetry.toml configuration file in the project directory.

Existing Python projects that do not use a virtual environment can opt-in to using the built-in virtual environment support by setting the above options. After updating the options, run pulumi install to create the virtual environment and install dependencies.

Self managed virtual environments

If you prefer to manage the virtual environment on your own (for example, using a tool like Pipenv), you can delete the local venv directory and unset the virtualenv option in Pulumi.yaml:

runtime: python

When managing the virtual environment on your own, you’ll need to run any pulumi commands (such as pulumi up) from an activated virtual environment shell (or, if using a tool like Pipenv, prefix any pulumi commands with pipenv run pulumi ...).

Type Checking

Pulumi Python libraries ship with type hints. You can manually run a type checking system of your choice when developing with Pulumi to make use of these type hints. As of 3.113.0 Pulumi also has first class support for mypy and pyright when running your program.

This behavior is controlled by the following typechecker runtime option in Pulumi.yaml:

runtime:
  name: python
  options:
    typechecker: mypy

When set, Pulumi will invoke the type checker before running your program. This can be used to ensure your program is always type safe when running pulumi up without you having to remember to run a separate checking command beforehand.

Adding a new dependency

There are many Pulumi Python packages available.

Pip

To install a new dependency in the virtual environment when using pip, add an entry to requirements.txt, and run the following in your project directory:

$ venv/bin/pip install -r requirements.txt

$ venv/bin/pip install -r requirements.txt

> venv\Scripts\pip install -r requirements.txt

Poetry

To add a new dependency when using poetry, run the poetry add command in your project directory:

$ poetry add ${PACKAGE_NAME}

$ poetry add ${PACKAGE_NAME}

> poetry add ${PACKAGE_NAME}

Dev Versions

Pulumi SDKs also publish pre-release versions that include all the latest changes from the main development branch. If you would like to install a pre-release version, you can use the --pre flag with pip or the --allow-prereleaases flag with poetry. For example:

pip install --pre -r requirements.txt
poetry add --allow-prereleases ${PACKAGE_NAME}

Package Documentation

In addition to the standard packages the Pulumi Registry houses 100+ Python packages.

Standard Packages

Pulumi SDK

pulumi

Pulumi Policy

pulumi_policy

Pulumi Terraform

pulumi_terraform