chore: always use starlark-based extraction for wheels (bazel-contrib#3748)

Currently, the Python `installer` library is used for extracting a wheel
when Bazel 7 is used. This library only outputs platform-specific
extractions, and it was hard-coded to use a unixy style install. This
ends up hiding Windows `pythonw`-based scripts and entry points because
those are normalized to simply `python` on unix.

To fix, use the Starlark-based extraction logic instead. This is a
simpler implementation that extracts the wheels mostly as-is, which
allows build-phase logic to handle platform-specific differences.

This also makes the Starlark based extraction work for Bazel 8.3 and
earlier, which previously didn't support it.

Author: rickeylev

python/private/pypi/whl_installer/BUILD.bazel

@@ -5,7 +5,6 @@ py_library(
     name = "lib",
     srcs = [
         "arguments.py",
-        "wheel.py",
         "wheel_installer.py",
     ],
     visibility = [

python/private/pypi/whl_installer/arguments.py

@@ -55,11 +55,6 @@ def parser(**kwargs: Any) -> argparse.ArgumentParser:
         help="Use 'pip download' instead of 'pip wheel'. Disables building wheels from source, but allows use of "
         "--platform, --python-version, --implementation, and --abi in --extra_pip_args.",
     )
-    parser.add_argument(
-        "--whl-file",
-        type=pathlib.Path,
-        help="Extract a whl file to be used within Bazel.",
-    )
     return parser
 
 

python/private/pypi/whl_installer/wheel.py

@@ -25,9 +25,7 @@
 from tempfile import NamedTemporaryFile
 from typing import Dict, List, Optional, Set, Tuple
 
-from pip._vendor.packaging.utils import canonicalize_name
-
-from python.private.pypi.whl_installer import arguments, wheel
+from python.private.pypi.whl_installer import arguments
 
 
 def _configure_reproducible_wheels() -> None:
@@ -55,63 +53,13 @@ def _configure_reproducible_wheels() -> None:
         os.environ["PYTHONHASHSEED"] = "0"
 
 
-def _parse_requirement_for_extra(
-    requirement: str,
-) -> Tuple[Optional[str], Optional[Set[str]]]:
-    """Given a requirement string, returns the requirement name and set of extras, if extras specified.
-    Else, returns (None, None)
-    """
-
-    # https://www.python.org/dev/peps/pep-0508/#grammar
-    extras_pattern = re.compile(
-        r"^\s*([0-9A-Za-z][0-9A-Za-z_.\-]*)\s*\[\s*([0-9A-Za-z][0-9A-Za-z_.\-]*(?:\s*,\s*[0-9A-Za-z][0-9A-Za-z_.\-]*)*)\s*\]"
-    )
-
-    matches = extras_pattern.match(requirement)
-    if matches:
-        return (
-            canonicalize_name(matches.group(1)),
-            {extra.strip() for extra in matches.group(2).split(",")},
-        )
-
-    return None, None
-
-
-def _extract_wheel(
-    wheel_file: str,
-    extras: Dict[str, Set[str]],
-    installation_dir: Path = Path("."),
-) -> None:
-    """Extracts wheel into given directory and creates py_library and filegroup targets.
-
-    Args:
-        wheel_file: the filepath of the .whl
-        installation_dir: the destination directory for installation of the wheel.
-        extras: a list of extras to add as dependencies for the installed wheel
-    """
-
-    whl = wheel.Wheel(wheel_file)
-    whl.unzip(installation_dir)
-
-
 def main() -> None:
     args = arguments.parser(description=__doc__).parse_args()
     deserialized_args = dict(vars(args))
     arguments.deserialize_structured_args(deserialized_args)
 
     _configure_reproducible_wheels()
 
-    if args.whl_file:
-        whl = Path(args.whl_file)
-
-        name, extras_for_pkg = _parse_requirement_for_extra(args.requirement)
-        extras = {name: extras_for_pkg} if extras_for_pkg and name else dict()
-        _extract_wheel(
-            wheel_file=whl,
-            extras=extras,
-        )
-        return
-
     pip_args = (
         [sys.executable, "-m", "pip"]
         + (["--isolated"] if args.isolated else [])

python/private/pypi/whl_installer/wheel_installer.py

@@ -14,7 +14,6 @@
 
 ""
 
-load("@rules_python_internal//:rules_python_config.bzl", rp_config = "config")
 load("//python/private:auth.bzl", "AUTH_ATTRS", "get_auth")
 load("//python/private:envsubst.bzl", "envsubst")
 load("//python/private:is_standalone_interpreter.bzl", "is_standalone_interpreter")
@@ -261,22 +260,6 @@ def _create_repository_execution_environment(rctx, python_interpreter, logger =
         env[_CPPFLAGS] = " ".join(cppflags)
     return env
 
-def _extract_whl_py(rctx, *, python_interpreter, args, whl_path, environment, logger):
-    pypi_repo_utils.execute_checked(
-        rctx,
-        op = "whl_library.ExtractWheel({}, {})".format(rctx.attr.name, whl_path),
-        python = python_interpreter,
-        arguments = args + [
-            "--whl-file",
-            whl_path,
-        ],
-        srcs = rctx.attr._python_srcs,
-        environment = environment,
-        quiet = rctx.attr.quiet,
-        timeout = rctx.attr.timeout,
-        logger = logger,
-    )
-
 def _get_entry_points(rctx, install_dir_path, metadata):
     dist_info_dir = "{}-{}.dist-info".format(
         metadata.name.replace("-", "_"),
@@ -376,11 +359,10 @@ def _whl_library_impl(rctx):
             # build deps from PyPI (e.g. `flit_core`) if they are missing.
             extra_pip_args.extend(["--find-links", "."])
 
-    enable_pipstar_extract = rp_config.bazel_8_or_later
-
-    # When pipstar is enabled, Python isn't used, so there's no need
-    # to setup env vars to run Python, unless we need to build an sdist
-    if enable_pipstar_extract and whl_path and not rctx.attr.whl_patches:
+    # When we already have a wheel and there are no patches, Python isn't used,
+    # so there's no need to setup env vars to run Python, unless we need to
+    # build an sdist or resolve a requirement.
+    if whl_path and not rctx.attr.whl_patches:
         environment = {}
         args = []
         python_interpreter = None
@@ -446,17 +428,7 @@ def _whl_library_impl(rctx):
                 timeout = rctx.attr.timeout,
             )
 
-    if enable_pipstar_extract:
-        whl_extract(rctx, whl_path = whl_path, logger = logger)
-    else:
-        _extract_whl_py(
-            rctx,
-            python_interpreter = python_interpreter,
-            args = args,
-            whl_path = whl_path,
-            environment = environment,
-            logger = logger,
-        )
+    whl_extract(rctx, whl_path = whl_path, logger = logger)
 
     install_dir_path = whl_path.dirname.get_child("site-packages")
     metadata = whl_metadata(
@@ -513,9 +485,8 @@ repo(
     _remove_files(rctx, "BUILD", "BUILD.bazel")
     rctx.file("BUILD.bazel", build_file_contents)
 
-    if enable_pipstar_extract:
-        if hasattr(rctx, "repo_metadata"):
-            return rctx.repo_metadata(reproducible = True)
+    if hasattr(rctx, "repo_metadata"):
+        return rctx.repo_metadata(reproducible = True)
 
     return None
 
@@ -632,7 +603,6 @@ way to define whl_library and move whl patching to a separate place. INTERNAL US
     "_python_srcs": attr.label_list(
         # Used as a default value in a rule to ensure we fetch the dependencies.
         default = [
-            Label("//python/private/pypi/whl_installer:wheel.py"),
             Label("//python/private/pypi/whl_installer:wheel_installer.py"),
             Label("//python/private/pypi/whl_installer:arguments.py"),
         ] + record_files.values(),

python/private/pypi/whl_library.bzl

@@ -1,29 +1,12 @@
 load("//python:py_test.bzl", "py_test")
 
-alias(
-    name = "lib",
-    actual = "//python/private/pypi/whl_installer:lib",
-)
-
 py_test(
     name = "arguments_test",
     size = "small",
     srcs = [
         "arguments_test.py",
     ],
     deps = [
-        ":lib",
-    ],
-)
-
-py_test(
-    name = "wheel_installer_test",
-    size = "small",
-    srcs = [
-        "wheel_installer_test.py",
-    ],
-    data = ["//examples/wheel:minimal_with_py_package"],
-    deps = [
-        ":lib",
+        "//python/private/pypi/whl_installer:lib",
     ],
 )

tests/pypi/whl_installer/BUILD.bazel

@@ -15,7 +15,7 @@
 import json
 import unittest
 
-from python.private.pypi.whl_installer import arguments, wheel
+from python.private.pypi.whl_installer import arguments
 
 
 class ArgumentsTestCase(unittest.TestCase):