Getting Started

Installing

PyOxidizer is a Rust application and requires Rust 1.33+ to be installed in order to build binaries. If you don’t have Rust installed, https://www.rust-lang.org/ has very detailed instructions on how to install it.

PyOxidizer can be installed from its latest published crate:

$ cargo install pyoxidizer

From a Git repository using cargo:

# The latest commit in source control.
$ cargo install --git https://github.com/indygreg/PyOxidizer.git --branch master pyoxidizer

$ A specific release
$ cargo install --git https://github.com/indygreg/PyOxidizer.git --tag <TAG> pyoxidizer

Or by cloning the Git repository and building the project locally:

$ git clone https://github.com/indygreg/PyOxidizer.git
$ cd PyOxidizer
$ cargo install --path pyoxidizer

Once the pyoxidizer executable is installed, try to run it:

$ pyoxidizer
PyOxidizer 0.1
Gregory Szorc <gregory.szorc@gmail.com>
Build and distribute Python applications

USAGE:
    pyoxidizer [SUBCOMMAND]

...

Congratulations, PyOxidizer is installed! Now let’s move on to using it.

Creating a PyOxidizer Project

The pyoxidizer init command will create a new [Rust] project which supports embedding Python. Invoke it with the directory you want to create your new project in:

$ pyoxidizer init pyapp

This should have printed out details on what happened and what to do next. If you actually ran this in a terminal, hopefully you don’t need to continue following the directions here as the printed instructions are sufficient! But if you aren’t, keep reading.

The default project created by pyoxidizer init will produce an executable that embeds Python and starts a Python REPL by default. Let’s test that:

$ cd pyapp
$ pyoxidizer run
no existing PyOxidizer artifacts found
processing config file /home/gps/src/pyapp/pyoxidizer.toml
resolving Python distribution...
...
   Compiling pyapp v0.1.0 (/home/gps/src/pyapp)
    Finished dev [unoptimized + debuginfo] target(s) in 53.14s
     Running `target/debug/testapp`
>>>

If all goes according to plan, you just started a Rust executable which started a Python interpreter, which started an interactive Python debugger! Try typing in some Python code:

>>> print("hello, world")
hello, world

It works!

(To exit the REPL, press CTRL+d or CTRL+z.)

Adding PyOxidizer to an Existing Project

Do you have an existing Rust project that you want to add an embedded Python interpreter to? PyOxidizer can help with that too! The pyoxidizer add command can be used to add an embedded Python interpreter to an existing Rust project. Simply give the directory to a project containing a Cargo.toml file:

$ cargo init myrustapp
  Created binary (application) package
$ pyoxidizer add myrustapp

This will add required files and make required modifications to add an embedded Python interpreter to the target project. Most of the modifications are in the form of a new pyembed crate.

Important

It is highly recommended to have the destination project under version control so you can see what changes are made by pyoxidizer add and so you can undo any unwanted changes.

Danger

This command isn’t very well tested. And results have been known to be wrong. If it doesn’t just work, you may want to run pyoxidizer init and incorporate relevant files into your project manually. Sorry for the inconvenience.

Customizing Python and Packaging Behavior

Embedding Python in a Rust executable and starting a REPL is cool and all. But you probably want to do something more exciting.

Inside the project’s root directory is an autogenerated pyoxidizer.toml file. This file configures how the embedded Python interpreter is built as well as defines default run-time behavior for that interpreter.

Open that file in your favorite editor and find the [[python_run]] section. This section configures what to do when the interpreter starts. By default, it should have a mode = "repl" line. Let’s comment that out or delete it and replace it with the following:

[[embedded_python_run]]
mode = "eval"
code = "import uuid; print(uuid.uuid4())"

We’re now telling the interpreter to effectively run the Python statement eval(code) when it starts. Test that out:

$ pyoxidizer run
   Compiling pyembed v0.1.0 (/home/gps/src/pyapp/pyembed)
   Compiling pyapp v0.1.0 (/home/gps/src/pyapp)
    Finished dev [unoptimized + debuginfo] target(s) in 3.92s
     Running `target/debug/pyapp`
96f776c8-c32d-48d8-8c1c-aef8a735f535

It works!

This is still pretty trivial. But it demonstrates how the pyoxidizer.toml is used to influence the behavior of built binaries.

Let’s do something a little bit more complicated, like package an existing Python application!

Find the existing [[python_packages]] section in the pyoxidizer.toml. Now let’s add the following lines after the last of those sections:

[[packaging_rule]]
type = "pip-install-simple"
package = "pyflakes==2.1.1"

And change the [[embedded_python_run]] section to:

[[embedded_python_run]]
mode = "eval"
code = "from pyflakes.api import main; main()"

This tells PyOxidizer that you want to install version 2.1.1 of the pyflakes package. At build time, this will effectively perform a pip install pyflakes==2.1.1 and take all installed files and add them to the produced binary. Let’s try that:

$ pyoxidizer run -- --help
   Compiling pyembed v0.1.0 (/home/gps/tmp/pyapp/pyembed)
   Compiling pyapp v0.1.0 (/home/gps/tmp/pyapp)
    Finished dev [unoptimized + debuginfo] target(s) in 5.49s
     Running `target/debug/pyapp --help`
Usage: pyapp [options]

Options:
  --version   show program's version number and exit
  -h, --help  show this help message and exit

You’ve just produced an executable for pyflakes!

There are far more powerful packaging and configuration settings available. Read all about them at Configuration Files.