Skip to content

jkinkead/rules_python

 
 

Repository files navigation

Experimental Bazel Python Rules

Status: This is ALPHA software.

Build status

Rules

Overview

This repository provides Python rules for Bazel. Currently, support for rules that are available from Bazel core are simple aliases to that bundled functionality. On top of that, this repository provides support for installing dependencies typically managed via pip.

Setup

Add the following to your WORKSPACE file to add the external repositories:

git_repository(
    name = "io_bazel_rules_python",
    remote = "https://github.com/bazelbuild/rules_python.git",
    # NOT VALID!  Replace this with a Git commit SHA.
    commit = "{HEAD}",
)

# Only needed for PIP support:
load("@io_bazel_rules_python//python:pip.bzl", "pip_repositories")

pip_repositories()

Then in your BUILD files load the python rules with:

load(
  "@io_bazel_rules_python//python:python.bzl",
  "py_binary", "py_library", "py_test",
)

py_binary(
  name = "main",
  ...
)

Importing pip dependencies

These rules are designed to have developers continue using requirements.txt to express their dependencies in a Python idiomatic manner. These dependencies are imported into the Bazel dependency graph via a two-phased process in WORKSPACE:

load("@io_bazel_rules_python//python:pip.bzl", "pip_import")

# This rule translates the specified requirements.txt into
# @my_deps//:requirements.bzl, which itself exposes a pip_install method.
pip_import(
   name = "my_deps",
   requirements = "//path/to:requirements.txt",
)

# Load the pip_install symbol for my_deps, and create the dependencies'
# repositories.
load("@my_deps//:requirements.bzl", "pip_install")
pip_install()

Consuming pip dependencies

Once a set of dependencies has been imported via pip_import and pip_install we can start consuming them in our py_{binary,library,test} rules. In support of this, the generated requirements.bzl also contains a requirement method, which can be used directly in deps=[] to reference an imported py_library.

load("@my_deps//:requirements.bzl", "requirement")

py_library(
    name = "mylib",
    srcs = ["mylib.py"],
    deps = [
        ":myotherlib",
	# This takes the name as specified in requirements.txt
	requirement("importeddep"),
    ]
)

Canonical whl_library naming

It is notable that whl_library rules imported via pip_import are canonically named, following the pattern: pypi__{distribution}_{version}. Characters in these components that are illegal in Bazel label names (e.g. -, .) are replaced with _.

This canonical naming helps avoid redundant work to import the same library multiple times. It is expected that this naming will remain stable, so folks should be able to reliably depend directly on e.g. @pypi__futures_3_1_1//:pkg for dependencies, however, it is recommended that folks stick with the requirement pattern in case the need arises for us to make changes to this format in the future.

"Extras" will have a target of the extra name (in place of pkg above).

Updating docs/

All of the content (except BUILD) under docs/ is generated. To update the documentation simply run this in the root of the repository:

./update_docs.sh

Updating tools/

All of the content (except BUILD) under tools/ is generated. To update the documentation simply run this in the root of the repository:

./update_tools.sh

About

Experimental Bazel Python Rules

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 95.7%
  • Shell 4.3%