Getting Started¶
Python Requirements¶
PyOxidizer currently targets Python 3.8, 3.9, and 3.10. Your Python application will need to already be compatible with 1 of these versions for it to work with PyOxidizer. See Why is Python 3.8 Required? for more on the minimum Python requirement.
Operating System Requirements¶
PyOxidizer is officially supported on the following operating systems:
Windows x86 (32-bit)
Windows x86_64/amd64 (64-bit)
macOS x86_64 (Intel processors)
macOS aarch64 (ARM/Apple processors)
Linux i686 (32-bit)
Linux x86_64 (64-bit)
It is likely possible to run PyOxidizer on unsupported operating systems and architectures. However, PyOxidizer needs to run Python interpreters on the machine performing build/packaging actions and the built binary needs to run a Python interpreter for the target architecture and operating system. These Python interpreters need to be built/packaged in a specific way so PyOxidizer can interact with them.
See Available Python Distributions for the full list of available Python distributions. The supported operating systems and architectures of the built-in Python distributions are:
Linux x86_64 (glibc 2.19 or musl linked)
Windows 8+ / Server 2012+ i686 and x86_64
macOS 10.9+ Intel x86_64 or 11.0+ ARM
Other System Dependencies¶
You will need a working C compiler/toolchain in order to build binaries. If a C compiler cannot be found, you should see an error message with instructions on how to install one.
On macOS, you will need an Apple SDK that is at least as new as the SDK used to build the Python distribution embedded in the binary. PyOxidizer will automatically attempt to locate, validate, and use an appropriate SDK. See Build Machine Requirements for more.
There is a known issue with PyOxidizer on Fedora 30+ that will require you
to install the libxcrypt-compat package to avoid an error due to a missing
libcrypt.so.1 file. See https://github.com/indygreg/PyOxidizer/issues/89
for more info.
While PyOxidizer is implemented in Rust and invokes the Rust compiler and build tooling to build binaries, PyOxidizer manages a Rust installation for you. This means Rust is not an explicit install dependency for PyOxidizer unless you are building PyOxidizer from source code.
Installing¶
Pre-Built Installers and Executables¶
PyOxidizer provides pre-built installers and executables as part of its release process. The following should be made available:
Linux x86-64 statically linked binary.
macOS universal binary.
Windows x86 (32-bit) MSI installer.
Windows amd64 (64-bit) MSI installer.
Windows universal (x86+amd64) EXE installer.
Python wheels.
These installers can generally be found at https://github.com/indygreg/PyOxidizer/releases/latest.
If this URL does not redirect to a PyOxidizer release, go to
https://github.com/indygreg/PyOxidizer/releases and look for a release with
PyOxidizer release artifacts. You should see giant text that reads
PyOxidizer <version> that looks different from other entries in the
list. You may have to click through multiple next links at the bottom of
the release list until you find a PyOxidizer release.
If pre-built artifacts are not available for your machine, you will need to compile PyOxidizer from source code.
Python Wheels¶
PyOxidizer is made available as a binary Python wheel (.whl) and
releases are published on PyPI. So you can install PyOxidizer like any
other Python package:
$ python3 -m pip install pyoxidizer
# To upgrade an existing install
$ python3 -m pip install --upgrade pyoxidizer
Installing PyOxidizer from Source¶
Installing Rust¶
PyOxidizer is a Rust application and requires Rust (1.60 or newer) to be installed in order to build PyOxidizer.
You can verify your installed version of Rust by running:
$ rustc --version
rustc 1.61.0 (fe5b13d68 2022-05-18)
If you don’t have Rust installed, https://www.rust-lang.org/ has very detailed instructions on how to install it.
Rust releases a new version every 6 weeks and language development moves
faster than other programming languages. It is common for the Rust packages
provided by common package managers to lag behind the latest Rust release by
several releases. For that reason, use of the rustup tool for managing
Rust is highly recommended.
If you are a security paranoid individual and don’t want to follow the
official rustup install instructions involving a curl | sh (your
paranoia is understood), you can find instructions for alternative installation
methods at https://github.com/rust-lang/rustup.rs/#other-installation-methods.
Installing PyOxidizer¶
Once Rust is installed, PyOxidizer can be installed from its latest published crate on Rust’s official/default package repository:
$ cargo install pyoxidizer
From PyOxidizer’s canonical Git repository using cargo:
# The latest commit in source control.
$ cargo install --git https://github.com/indygreg/PyOxidizer.git --branch main pyoxidizer
$ A specific release
$ cargo install --git https://github.com/indygreg/PyOxidizer.git --tag <TAG> pyoxidizer
Or by cloning the canonical Git repository and building the project locally:
$ git clone https://github.com/indygreg/PyOxidizer.git
$ cd PyOxidizer
$ cargo install --path pyoxidizer
Note
PyOxidizer’s project policy is for the main branch to be stable. So it
should always be relatively safe to use main instead of a released
version.
Danger
A cargo build from the repository root directory will likely fail due
to how some of the Rust crates are configured.
See Using Cargo with PyOxidizer Source Checkouts for instructions on how to invoke
cargo.
Once the pyoxidizer executable is installed, try to run it:
$ pyoxidizer
PyOxidizer 0.14.0-pre
Gregory Szorc <gregory.szorc@gmail.com>
Build and distribute Python applications
USAGE:
pyoxidizer [FLAGS] [SUBCOMMAND]
...
Congratulations, PyOxidizer is installed! Now let’s move on to using it.
High-Level Project Lifecycle¶
PyOxidizer exposes various functionality through the interaction
of pyoxidizer commands and configuration files.
The first step of any project is to create it. This is achieved
with a pyoxidizer init-* command to create files required by
PyOxidizer.
After that, various pyoxidizer commands can be used to evaluate
configuration files and perform actions from the evaluated file.
PyOxidizer provides functionality for building binaries, installing
files into a directory tree, and running the results of build actions.
Your First PyOxidizer Project¶
The pyoxidizer init-config-file command will create a new PyOxidizer
configuration file in a directory of your choosing:
$ pyoxidizer init-config-file 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 configuration created by pyoxidizer init-config-file will
produce an executable that embeds Python and starts a Python REPL by default.
Let’s test that:
$ cd pyapp
$ pyoxidizer run
resolving 1 targets
resolving target exe
...
Compiling pyapp v0.1.0 (/tmp/pyoxidizer.nv7QvpNPRgL5/pyapp)
Finished dev [unoptimized + debuginfo] target(s) in 26.07s
writing executable to /home/gps/src/pyapp/build/x86_64-unknown-linux-gnu/debug/exe/pyapp
>>>
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.)
Continue reading The pyoxidizer Command Line Tool to learn more about the
pyoxidizer tool. Or read on for a preview of how to customize your
application’s behavior.
The pyoxidizer.bzl Configuration File¶
The most important file for a PyOxidizer project is the pyoxidizer.bzl
configuration file. This is a Starlark file evaluated in a context that
provides special functionality for PyOxidizer.
Starlark is a Python-like interpreted language and its syntax and semantics should be familiar to any Python programmer.
From a high-level, PyOxidizer’s configuration files define named
targets, which are callable functions associated with a name - the
target - that resolve to an entity. For example, a configuration file
may define a build_exe() function which returns an object representing
a standalone executable file embedding Python. The pyoxidizer build
command can be used to evaluate just that target/function.
Target functions can call out to other target functions. For example, there
may be an install target that creates a set of files composing a full
application. Its function may evaluate the exe target to produce an
executable file.
See Configuration Files for comprehensive documentation of pyoxidizer.bzl
files and their semantics.
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.
The autogenerated pyoxidizer.bzl file created as part of running
pyoxidizer init-config-file defines how your application is configured
and built. It controls everything from what Python distribution to use,
which Python packages to install, how the embedded Python interpreter is
configured, and what code to run in that interpreter.
Open pyoxidizer.bzl in your favorite editor and find the commented lines
assigning to python_config.run_*. Let’s uncomment or add a line
to match the following:
python_config.run_command = "import uuid; print(uuid.uuid4())"
We’re now telling the interpreter to run the Python statement
eval(import uuid; print(uuid.uuid4()) when it starts. Test that out:
$ pyoxidizer run
...
Compiling pyapp v0.1.0 (/home/gps/src/pyapp)
Finished dev [unoptimized + debuginfo] target(s) in 3.92s
Running `target/debug/pyapp`
writing executable to /home/gps/src/pyapp/build/x86_64-unknown-linux-gnu/debug/exe/pyapp
96f776c8-c32d-48d8-8c1c-aef8a735f535
It works!
This is still pretty trivial. But it demonstrates how the pyoxidizer.bzl
is used to influence the behavior of built executables.
Let’s do something a little bit more complicated, like package an existing Python application!
Find the exe = dist.to_python_executable( line in the
pyoxidizer.bzl file. Let’s add a new line to make_exe() just
below where exe is assigned:
for resource in exe.pip_install(["pyflakes==2.2.0"]):
resource.add_location = "in-memory"
exe.add_python_resource(resource)
In addition, set the python_config.run_command attribute to execute pyflakes:
python_config.run_command = "from pyflakes.api import main; main()"
Now let’s try building and running the new configuration:
$ pyoxidizer run -- --help
...
Compiling pyapp v0.1.0 (/home/gps/src/pyapp)
Finished dev [unoptimized + debuginfo] target(s) in 5.49s
writing executable to /home/gps/src/pyapp/build/x86_64-unknown-linux-gnu/debug/exe/pyapp
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!
Note
pyflakes with no command arguments will read from stdin and will
effectively hang until stdin is closed (typically via CTRL + D). So
the -- --help in the above example is important, as it forces
the command to produce output.
There are far more powerful packaging and configuration settings available.
Read all about them at Configuration Files and Packaging User Guide. Or continue
on to The pyoxidizer Command Line Tool to learn more about the pyoxidizer tool.