如何在run_binary规则中使用pip安装的hatch工具？

问题背景 #

通过WORKSPACE中的pip_parse安装了hatch依赖，希望将其作为工具在run_binary规则中执行hatch build，但直接指定tool = "hatch"时出现找不到目标的错误，需求是找到正确引用pip安装二进制文件的方法，或替代实现方案。

原配置文件 #

WORKSPACE

load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")

http_archive(
    name = "bazel_skylib",
    sha256 = "b8a1527901774180afc798aeb28c4634bdccf19c4d98e7bdd1ce79d1fe9aaad7",
    urls = [
        "https://mirror.bazel.build/github.com/bazelbuild/bazel-skylib/releases/download/1.4.1/bazel-skylib-1.4.1.tar.gz",
        "https://github.com/bazelbuild/bazel-skylib/releases/download/1.4.1/bazel-skylib-1.4.1.tar.gz",
    ]
)
http_archive(
    name = "rules_python",
    sha256 = "ffc7b877c95413c82bfd5482c017edcf759a6250d8b24e82f41f3c8b8d9e287e",
    strip_prefix = "rules_python-0.19.0",
    url = "https://github.com/bazelbuild/rules_python/releases/download/0.19.0/rules_python-0.19.0.tar.gz",
)

http_archive(
    name = "aspect_bazel_lib",
    sha256 = "2518c757715d4f5fc7cc7e0a68742dd1155eaafc78fb9196b8a18e13a738cea2",
    strip_prefix = "bazel-lib-1.28.0",
    url = "https://github.com/aspect-build/bazel-lib/releases/download/v1.28.0/bazel-lib-v1.28.0.tar.gz",
)

# for aspect's run_binary rule
load("@aspect_bazel_lib//lib:repositories.bzl", "aspect_bazel_lib_dependencies")
aspect_bazel_lib_dependencies()

# install Python requirements
load("@rules_python//python:pip.bzl", "pip_parse")
pip_parse(
    name = "py_mod_deps",
    requirements_lock = "//py-mod:requirements.txt"
)
load("@py_mod_deps//:requirements.bzl", "install_deps")
install_deps()

py-mod/requirements.txt

hatch==1.6.3
# other dependencies from pip-compile requirements.in

原py-mod/BUILD

load("@aspect_bazel_lib//lib:run_binary.bzl", "run_binary")

run_binary(
    name = "mymodule",
    tool = "hatch",
    args = ["build"],
    srcs = glob(["**/*.py"]),
    out_dirs = ["dist/"]
)

报错信息 #

ERROR: .../py-mod/BUILD: RunBinary py-mod/dist failed: missing input file '//py-mod:hatch'
ERROR: .../py-mod/BUILD: RunBinary py-mod/dist failed: 1 input file(s) do not exist

解决方案 #

方案一：直接引用pip_parse生成的Hatch目标 #

pip_parse会为每个包含可执行入口的Python包生成对应的Bazel目标，格式为@<pip_parse_name>//:<package_name>，这里pip_parse的name是py_mod_deps，因此Hatch的目标是@py_mod_deps//:hatch。

修改后的py-mod/BUILD：

load("@aspect_bazel_lib//lib:run_binary.bzl", "run_binary")

run_binary(
    name = "mymodule",
    tool = "@py_mod_deps//:hatch",
    args = ["build"],
    # 包含hatch构建所需的所有配置文件，比如pyproject.toml
    srcs = glob(["**/*.py", "pyproject.toml"]),
    out_dirs = ["dist/"],
)

关键提示

• 必须确保srcs包含Hatch构建依赖的所有文件，比如pyproject.toml，否则Hatch会因找不到配置文件而失败。

方案二：用py_binary包装Hatch调用（替代方案） #

如果需要更灵活的控制，可以编写Python脚本包装Hatch的调用，通过py_binary生成可执行目标后供run_binary使用。

1. 在py-mod目录下创建run_hatch.py：

from hatch.cli.main import main
import sys

if __name__ == "__main__":
    sys.exit(main())

2. 修改py-mod/BUILD：

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

py_binary(
    name = "hatch_wrapper",
    srcs = ["run_hatch.py"],
    deps = ["@py_mod_deps//:hatch"],
    main = "run_hatch.py",
)

run_binary(
    name = "mymodule",
    tool = ":hatch_wrapper",
    args = ["build"],
    srcs = glob(["**/*.py", "pyproject.toml"]),
    out_dirs = ["dist/"],
)

验证执行 #

修改完成后，执行以下命令即可正常构建：

bazel build //py-mod:mymodule

内容的提问来源于stack exchange，提问作者tomocafe