commit	9150caa9d857e3768a4cf5ef6c3e88668b7ec84f	[log] [tgz]
author	Jon Brandvein <brandjon@google.com>	Tue Oct 08 23:54:04 2019 -0400
committer	GitHub <noreply@github.com>	Tue Oct 08 23:54:04 2019 -0400
tree	bed4e6ce8c95387472963391237f3d4c070472c3
parent	e953b0ad875b6b5dc786b71d431775a7daf75607 [diff]

tree: bed4e6ce8c95387472963391237f3d4c070472c3

README.md

Python Rules for Bazel

Recent updates

2019-07-26: The canonical name of this repo has been changed from @io_bazel_rules_python to just @rules_python, in accordance with convention. Please update your WORKSPACE file and labels that reference this repo accordingly.

Overview

This repository is the home of the core Python rules -- py_library, py_binary, py_test, and related symbols that provide the basis for Python support in Bazel. It also contains packaging rules for integrating with PyPI (pip). Documentation lives in the docs/ directory and in the Bazel Build Encyclopedia.

Currently the core rules are bundled with Bazel itself, and the symbols in this repository are simple aliases. However, in the future the rules will be migrated to Starlark and debundled from Bazel. Therefore, the future-proof way to depend on Python rules is via this repository. SeeMigrating from the Bundled Rules below.

The core rules are stable. Their implementation in Bazel is subject to Bazel's backward compatibility policy. Once they are fully migrated to rules_python, they may evolve at a different rate, but this repository will still follow semantic versioning.

The packaging rules (pip_import, etc.) are less stable. We may make breaking changes as they evolve. There are no guarantees for rules underneath the experimental/ directory.

Getting started

To import rules_python in your project, you first need to add it to your WORKSPACE file. If you are using the Bazel Federation, you will want to copy the boilerplate in the rules_python release's notes, under the “WORKSPACE setup” heading. This will look something like the following:

load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")
http_archive(
    name = "rules_python",
    # NOT VALID: Replace with actual version and SHA.
    url = "https://github.com/bazelbuild/rules_python/releases/download/<RELEASE>/rules_python-<RELEASE>.tar.gz",
    sha256 = "<SHA>",
)
load("@rules_python//python:repositories.bzl", "py_repositories", "rules_python_toolchains")
py_repositories()
rules_python_toolchains()

Otherwise, you may import rules_python in a standalone way by copying the following:

load("@bazel_tools//tools/build_defs/repo:git.bzl", "git_repository")
git_repository(
    name = "rules_python",
    remote = "https://github.com/bazelbuild/rules_python.git",
    # NOT VALID: Replace with actual Git commit SHA.
    commit = "{HEAD}",
)
# This call should always be present.
load("@rules_python//python:repositories.bzl", "py_repositories")
py_repositories()
# This one is only needed if you're using the packaging rules.
load("@rules_python//python:pip.bzl", "pip_repositories")
pip_repositories()

Either way, you can then load the core rules in your BUILD files with:

load("@rules_python//python:defs.bzl", "py_binary")

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

Using the packaging rules

Importing `pip` dependencies

The packaging 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("@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).

Migrating from the bundled rules

The core rules are currently available in Bazel as built-in symbols, but this form is deprecated. Instead, you should depend on rules_python in your WORKSPACE file and load the Python rules from @rules_python//python:defs.bzl.

A buildifier fix is available to automatically migrate BUILD and .bzl files to add the appropriate load() statements and rewrite uses of native.py_*.

# Also consider using the -r flag to modify an entire workspace.
buildifier --lint=fix --warnings=native-py <files>

Currently the WORKSPACE file needs to be updated manually as per Getting started above.

Note that Starlark-defined bundled symbols underneath @bazel_tools//tools/python are also deprecated. These are not yet rewritten by buildifier.

Development

Documentation

The content underneath docs/ is generated. To update the documentation, simply run this in the root of the repository:

./update_docs.sh

Precompiled .par files

The piptool.par and whltool.par files underneath tools/ are compiled versions of the Python scripts under the packaging/ directory. We need to check in built artifacts because they are executed during WORKSPACE evaluation, before Bazel itself is able to build anything from source.

The .par files need to be regenerated whenever their sources are updated. This can be done by running

# You can pass --nodocker if Docker is not available on your system.
./update_tools.sh

from the repository root. However, since these files contain compiled code, we do not accept commits that modify them from untrusted sources.1 If you submit a pull request that modifies the sources and we accept the changes, we will regenerate these files for you before merging.

1 See “Reflections on Trusting Trust”.