feat(venv): support data, include, and scripts schemes

MODULE.bazel

@@ -246,6 +246,8 @@ use_repo(
     "somepkg_with_build_files",
     "whl_library_extras_direct_dep",
     "whl_with_build_files",
+    "whl_with_data1",
+    "whl_with_data2",
 )
 
 dev_rules_python_config = use_extension(

examples/pip_parse/pip_parse_test.py

@@ -52,16 +52,16 @@ def test_data(self):
         self.assertIsNotNone(actual)
         actual = self._remove_leading_dirs(actual.split(" "))
 
-        self.assertListEqual(
-            actual,
-            [
-                "data/share/doc/packages/s3cmd/INSTALL.md",
-                "data/share/doc/packages/s3cmd/LICENSE",
-                "data/share/doc/packages/s3cmd/NEWS",
-                "data/share/doc/packages/s3cmd/README.md",
-                "data/share/man/man1/s3cmd.1",
-            ],
-        )
+        expected = [
+            "bin/s3cmd",
+            "data/share/doc/packages/s3cmd/INSTALL.md",
+            "data/share/doc/packages/s3cmd/LICENSE",
+            "data/share/doc/packages/s3cmd/NEWS",
+            "data/share/doc/packages/s3cmd/README.md",
+            "data/share/man/man1/s3cmd.1",
+        ]
+
+        self.assertListEqual(actual, expected)
 
     def test_dist_info(self):
         actual = os.environ.get("WHEEL_DIST_INFO_CONTENTS")

internal_dev_setup.bzl

@@ -26,6 +26,8 @@ load("//python:versions.bzl", "MINOR_MAPPING", "TOOL_VERSIONS")
 load("//python/private:pythons_hub.bzl", "hub_repo")  # buildifier: disable=bzl-visibility
 load("//python/private:runtime_env_repo.bzl", "runtime_env_repo")  # buildifier: disable=bzl-visibility
 load("//python/private/pypi:deps.bzl", "pypi_deps")  # buildifier: disable=bzl-visibility
+load("//python/private/pypi:whl_library.bzl", "whl_library")  # buildifier: disable=bzl-visibility
+load("//tests/support/whl_from_dir:whl_from_dir_repo.bzl", "whl_from_dir_repo")  # buildifier: disable=bzl-visibility
 
 def rules_python_internal_setup():
     """Setup for development and testing of rules_python itself."""
@@ -59,3 +61,25 @@ def rules_python_internal_setup():
     bazel_features_deps()
     rules_shell_dependencies()
     rules_shell_toolchains()
+
+    whl_from_dir_repo(
+        name = "whl_with_data1_whl",
+        root = "//tests/repos/whl_with_data1:BUILD.bazel",
+        output = "whl_with_data1-1.0-any-none-any.whl",
+    )
+    whl_library(
+        name = "whl_with_data1",
+        whl_file = "@whl_with_data1_whl//:whl_with_data1-1.0-any-none-any.whl",
+        requirement = "whl-with-data1",
+    )
+
+    whl_from_dir_repo(
+        name = "whl_with_data2_whl",
+        root = "//tests/repos/whl_with_data2:BUILD.bazel",
+        output = "whl_with_data2-1.0-any-none-any.whl",
+    )
+    whl_library(
+        name = "whl_with_data2",
+        whl_file = "@whl_with_data2_whl//:whl_with_data2-1.0-any-none-any.whl",
+        requirement = "whl-with-data2",
+    )

python/config_settings/BUILD.bazel

@@ -229,7 +229,7 @@ string_flag(
 )
 
 config_setting(
-    name = "is_venvs_site_packages",
+    name = "_is_venvs_site_packages_yes",
     flag_values = {
         ":venvs_site_packages": VenvsSitePackages.YES,
     },

python/private/internal_dev_deps.bzl

@@ -91,6 +91,28 @@ def _internal_dev_deps_impl(mctx):
         enable_implicit_namespace_pkgs = False,
     )
 
+    whl_from_dir_repo(
+        name = "whl_with_data1_whl",
+        root = "//tests/repos/whl_with_data1:BUILD.bazel",
+        output = "whl_with_data1-1.0-any-none-any.whl",
+    )
+    whl_library(
+        name = "whl_with_data1",
+        whl_file = "@whl_with_data1_whl//:whl_with_data1-1.0-any-none-any.whl",
+        requirement = "whl-with-data1",
+    )
+
+    whl_from_dir_repo(
+        name = "whl_with_data2_whl",
+        root = "//tests/repos/whl_with_data2:BUILD.bazel",
+        output = "whl_with_data2-1.0-any-none-any.whl",
+    )
+    whl_library(
+        name = "whl_with_data2",
+        whl_file = "@whl_with_data2_whl//:whl_with_data2-1.0-any-none-any.whl",
+        requirement = "whl-with-data2",
+    )
+
     _whl_library_from_dir(
         name = "whl_library_extras_direct_dep",
         root = "//tests/pypi/whl_library/testdata/pkg:BUILD.bazel",

python/private/py_executable.bzl

@@ -575,9 +575,13 @@ def _create_venv(ctx, output_prefix, imports, runtime_details, add_runfiles_root
         computed_substitutions = computed_subs,
     )
 
+    # See https://docs.python.org/3/library/sysconfig.html#posix-prefix
+    # for how schemes map under the venv.
     venv_dir_map = {
-        VenvSymlinkKind.BIN: venv_details.bin_dir,
+        VenvSymlinkKind.BIN: "{}/{}".format(venv_ctx_rel_root, venv_details.bin_dir),
         VenvSymlinkKind.LIB: site_packages,
+        VenvSymlinkKind.INCLUDE: "{}/{}".format(venv_ctx_rel_root, venv_details.include_dir),
+        VenvSymlinkKind.DATA: venv_ctx_rel_root,
     }
     venv_app_files = create_venv_app_files(
         ctx,
@@ -659,7 +663,7 @@ def _create_venv_unixy(ctx, *, venv_ctx_rel_root, runtime, interpreter_actual_pa
 
     recreate_venv_at_runtime = False
 
-    bin_dir = "{}/bin".format(venv_ctx_rel_root)
+    venv_bin_ctx_rel_path = "{}/bin".format(venv_ctx_rel_root)
     if create_full_venv:
         # Some wrappers around the interpreter (e.g. pyenv) use the program
         # name to decide what to do, so preserve the name.
@@ -671,15 +675,15 @@ def _create_venv_unixy(ctx, *, venv_ctx_rel_root, runtime, interpreter_actual_pa
             # When the venv symlinks are disabled, the $venv/bin/python3 file isn't
             # needed or used at runtime. However, the zip code uses the interpreter
             # File object to figure out some paths.
-            interpreter = ctx.actions.declare_file("{}/{}".format(bin_dir, py_exe_basename))
+            interpreter = ctx.actions.declare_file("{}/{}".format(venv_bin_ctx_rel_path, py_exe_basename))
             ctx.actions.write(interpreter, "actual:{}".format(interpreter_actual_path))
 
         elif runtime.interpreter:
             # Even though ctx.actions.symlink() is used, using
             # declare_symlink() is required to ensure that the resulting file
             # in runfiles is always a symlink. An RBE implementation, for example,
             # may choose to write what symlink() points to instead.
-            interpreter = ctx.actions.declare_symlink("{}/{}".format(bin_dir, py_exe_basename))
+            interpreter = ctx.actions.declare_symlink("{}/{}".format(venv_bin_ctx_rel_path, py_exe_basename))
             interpreter_runfiles.add(interpreter)
 
             rel_path = relative_path(
@@ -690,7 +694,7 @@ def _create_venv_unixy(ctx, *, venv_ctx_rel_root, runtime, interpreter_actual_pa
             )
             ctx.actions.symlink(output = interpreter, target_path = rel_path)
         else:
-            interpreter = ctx.actions.declare_symlink("{}/{}".format(bin_dir, py_exe_basename))
+            interpreter = ctx.actions.declare_symlink("{}/{}".format(venv_bin_ctx_rel_path, py_exe_basename))
             interpreter_runfiles.add(interpreter)
             ctx.actions.symlink(output = interpreter, target_path = runtime.interpreter_path)
     else:
@@ -715,7 +719,8 @@ def _create_venv_unixy(ctx, *, venv_ctx_rel_root, runtime, interpreter_actual_pa
         interpreter = interpreter,
         pyvenv_cfg = pyvenv_cfg,
         site_packages = site_packages,
-        bin_dir = bin_dir,
+        bin_dir = "bin",
+        include_dir = "include",
         recreate_venv_at_runtime = recreate_venv_at_runtime,
         interpreter_runfiles = interpreter_runfiles.build(ctx),
         interpreter_symlinks = depset(),
@@ -777,7 +782,8 @@ def _create_venv_windows(ctx, *, venv_ctx_rel_root, runtime, interpreter_actual_
         interpreter = interpreter,
         pyvenv_cfg = None,
         site_packages = site_packages,
-        bin_dir = venv_bin_ctx_rel_path,
+        bin_dir = venv_bin_rel_path,
+        include_dir = "Include",
         recreate_venv_at_runtime = True,
         interpreter_runfiles = interpreter_runfiles.build(ctx),
         interpreter_symlinks = interpreter_symlinks.build(),
@@ -789,6 +795,7 @@ def _venv_details(
         pyvenv_cfg,
         site_packages,
         bin_dir,
+        include_dir,
         recreate_venv_at_runtime,
         interpreter_runfiles,
         interpreter_symlinks):
@@ -801,8 +808,10 @@ def _venv_details(
         pyvenv_cfg = pyvenv_cfg,
         # str; venv-relative path to the site-packages directory
         site_packages = site_packages,
-        # str; ctx-relative path to the venv's bin directory.
+        # str; venv-relative path to the venv's bin directory.
         bin_dir = bin_dir,
+        # str; venv-relative-path to the venv's include directory.
+        include_dir = include_dir,
         # bool; True if the venv needs to be recreated at runtime (because the
         # build-time construction isn't sufficient). False if the build-time
         # constructed venv is sufficient.

python/private/py_info.bzl

@@ -37,6 +37,12 @@ def _VenvSymlinkKind_typedef():
 
     Indicates to create paths under the venv's include directory.
     :::
+
+    :::{field} DATA
+    :type: object
+
+    Indicates to create paths under the venv's data directory.
+    :::
     """
 
 # buildifier: disable=name-conventions
@@ -45,6 +51,7 @@ VenvSymlinkKind = struct(
     BIN = "BIN",
     LIB = "LIB",
     INCLUDE = "INCLUDE",
+    DATA = "DATA",
 )
 
 def _VenvSymlinkEntry_init(**kwargs):

python/private/pypi/whl_library_targets.bzl

@@ -43,6 +43,8 @@ _BAZEL_REPO_FILE_GLOBS = [
     "WORKSPACE.bazel",
 ]
 
+_IS_VENV_SITE_PACKAGES_YES = Label("//python/config_settings:_is_venvs_site_packages_yes")
+
 def whl_library_targets_from_requires(
         *,
         name,
@@ -195,7 +197,7 @@ def whl_library_targets(
                 include = ["site-packages/*.dist-info/**"],
             ),
             DATA_LABEL: dict(
-                include = ["data/**"],
+                include = ["data/**", "bin/**", "include/**"],
             ),
         }
 
@@ -369,7 +371,7 @@ def whl_library_targets(
 
         if not enable_implicit_namespace_pkgs:
             generated_namespace_package_files = select({
-                Label("//python/config_settings:is_venvs_site_packages"): [],
+                _IS_VENV_SITE_PACKAGES_YES: [],
                 "//conditions:default": rules.create_inits(
                     srcs = srcs + data + pyi_srcs,
                     ignored_dirnames = [],  # If you need to ignore certain folders, you can patch rules_python here to do so.
@@ -379,6 +381,13 @@ def whl_library_targets(
             namespace_package_files += generated_namespace_package_files
             srcs = srcs + generated_namespace_package_files
 
+        # This must be doe after the above because create_inits() is macro-phase,
+        # so can't handle select() values.
+        data = data + select({
+            _IS_VENV_SITE_PACKAGES_YES: [DATA_LABEL],
+            "//conditions:default": [],
+        })
+
         rules.py_library(
             name = py_library_label,
             srcs = srcs,

python/private/venv_runfiles.bzl

@@ -69,17 +69,35 @@ def create_venv_app_files(ctx, deps, venv_dir_map):
         ctx.label.package,
     )
 
+    seen_bin_venv_paths = {}
+
     for kind, kind_map in link_map.items():
         base = venv_dir_map[kind]
         for venv_path, link_to in kind_map.items():
             bin_venv_path = paths.join(base, venv_path)
+            if bin_venv_path in seen_bin_venv_paths:
+                continue
+            seen_bin_venv_paths[bin_venv_path] = True
+
             if is_file(link_to):
                 # use paths.join to handle ctx.label.package = ""
                 # runfile_prefix should be prepended as we use runfiles.root_symlinks
                 runfile_prefix = ctx.label.repo_name or ctx.workspace_name
                 symlink_from = paths.join(runfile_prefix, ctx.label.package, bin_venv_path)
 
                 runfiles_symlinks[symlink_from] = link_to
+
+                # On Windows, we need to explicitly create the symlink in the venv
+                # because the bootstrap script won't otherwise know about it.
+                if is_windows:
+                    rf_path = paths.join(ctx_rf_path, bin_venv_path)
+                    _, _, venv_path = bin_venv_path.partition(".venv/")
+                    explicit_symlinks.append(ExplicitSymlink(
+                        runfiles_path = rf_path,
+                        venv_path = venv_path,
+                        link_to_path = runfiles_root_path(ctx, link_to.short_path),
+                        files = depset([link_to]),
+                    ))
             elif not is_windows:
                 venv_link = ctx.actions.declare_symlink(bin_venv_path)
                 venv_link_rf_path = runfiles_root_path(ctx, venv_link.short_path)
@@ -275,10 +293,16 @@ def _get_file_venv_path(ctx, f, site_packages_root):
 
     Returns:
         A tuple `(venv_path, rf_root_path)` if the file is under
-        `site_packages_root`, otherwise `(None, None)`.
+        `site_packages_root` or data/, bin/, include/ otherwise `(None, None)`.
     """
     rf_root_path = runfiles_root_path(ctx, f.short_path)
     _, _, repo_rel_path = rf_root_path.partition("/")
+
+    # Check for wheel data/bin/include folders first
+    for prefix in ["data/", "bin/", "include/"]:
+        if repo_rel_path.startswith(prefix):
+            return (repo_rel_path, rf_root_path)
+
     head, found_sp_root, venv_path = repo_rel_path.partition(site_packages_root)
     if head or not found_sp_root:
         # If head is set, then the path didn't start with site_packages_root
@@ -324,17 +348,37 @@ def get_venv_symlinks(
 
     all_files = sorted(files, key = lambda f: f.short_path)
 
+    cannot_be_linked_directly = {}
+    for dirname in [
+        # The venv directories that bin, include, and data get put into are
+        # shared across wheels, are also shared across wheels, so we cannot link
+        # them directly
+        "bin",
+        "include",
+        "data",
+        # The data scheme is overlaid on the venv root, so the files under it
+        # could, in theory, get installed into e.g. bin/ or similar. Explicitly
+        # mark them as non-directly linkable to avoid issues.
+        "data/bin",
+        "data/include",
+        "data/lib",
+        "data/Scripts",
+        "data/Include",
+        "data/Lib",
+    ]:
+        cannot_be_linked_directly[dirname] = True
+
     # dict[str venv-relative dirname, bool is_namespace_package]
     namespace_package_dirs = {
         ns: True
         for ns in _WELL_KNOWN_NAMESPACE_PACKAGES
     }
 
     # venv paths that cannot be directly linked. Dict acting as set.
-    cannot_be_linked_directly = {
+    cannot_be_linked_directly.update({
         dirname: True
         for dirname in namespace_package_dirs.keys()
-    }
+    })
     for f in namespace_package_files:
         venv_path, _ = _get_file_venv_path(ctx, f, site_packages_root)
         if venv_path == None:
@@ -452,19 +496,36 @@ def get_venv_symlinks(
 
     # Finally, for each group, we create the VenvSymlinkEntry objects
     for venv_path, files in optimized_groups.items():
+        if venv_path.startswith("data/"):
+            out_venv_path = venv_path[len("data/"):]
+            kind = VenvSymlinkKind.DATA
+            prefix = ""
+        elif venv_path.startswith("include/"):
+            out_venv_path = venv_path[len("include/"):]
+            kind = VenvSymlinkKind.INCLUDE
+            prefix = ""
+        elif venv_path.startswith("bin/"):
+            out_venv_path = venv_path[len("bin/"):]
+            kind = VenvSymlinkKind.BIN
+            prefix = ""
+        else:
+            out_venv_path = venv_path
+            kind = VenvSymlinkKind.LIB
+            prefix = site_packages_root
+
         link_to_path = (
             _get_label_runfiles_repo(ctx, files[0].owner) +
             "/" +
-            site_packages_root +
+            prefix +
             venv_path
         )
         venv_symlinks[venv_path] = VenvSymlinkEntry(
-            kind = VenvSymlinkKind.LIB,
+            kind = kind,
             link_to_path = link_to_path,
             link_to_file = None,
             package = package,
             version = version_str,
-            venv_path = venv_path,
+            venv_path = out_venv_path,
             files = depset(files),
         )