chore: use identifiers used in platforms package for arch and os names and get rid of uname dep

Author: aignas

python/private/python_register_toolchains.bzl

@@ -160,7 +160,10 @@ def python_register_toolchains(
                 platform = platform,
             ))
 
-    host_toolchain(name = name + "_host")
+    host_toolchain(
+        name = name + "_host",
+        platforms = loaded_platforms,
+    )
 
     toolchain_aliases(
         name = name,

python/private/toolchains_repo.bzl

@@ -25,8 +25,6 @@ platform-specific repositories.
 
 load(
     "//python:versions.bzl",
-    "LINUX_NAME",
-    "MACOS_NAME",
     "PLATFORMS",
     "WINDOWS_NAME",
 )
@@ -220,15 +218,15 @@ actions.""",
 )
 
 def _host_toolchain_impl(rctx):
-    logger = repo_utils.logger(rctx)
     rctx.file("BUILD.bazel", """\
 # Generated by python/private/toolchains_repo.bzl
 
 exports_files(["python"], visibility = ["//visibility:public"])
 """)
 
-    (os_name, arch) = _get_host_os_arch(rctx, logger)
-    host_platform = _get_host_platform(os_name, arch)
+    os_name = repo_utils.get_platforms_os_name(rctx)
+    cpu_name = repo_utils.get_platforms_cpu_name(rctx)
+    host_platform = _get_host_platform(rctx, os_name, cpu_name, rctx.attr.platforms)
     repo = "@@{py_repository}_{host_platform}".format(
         py_repository = rctx.attr.name[:-len("_host")],
         host_platform = host_platform,
@@ -297,6 +295,7 @@ toolchain_aliases repo because referencing the `python` interpreter target from
 this repo causes an eager fetch of the toolchain for the host platform.
     """,
     attrs = {
+        "platforms": attr.string_list(mandatory = True),
         "_rule_name": attr.string(default = "host_toolchain"),
         "_rules_python_workspace": attr.label(default = Label("//:WORKSPACE")),
     },
@@ -365,57 +364,47 @@ multi_toolchain_aliases = repository_rule(
 def sanitize_platform_name(platform):
     return platform.replace("-", "_")
 
-def _get_host_platform(os_name, arch):
+def _get_host_platform(rctx, os_name, cpu_name, platforms):
     """Gets the host platform.
 
     Args:
-        os_name: the host OS name.
-        arch: the host arch.
+        rctx: {type}`repository_ctx`.
+        os_name: {type}`str` the host OS name.
+        cpu_name: {type}`str` the host CPU name.
+        platforms: {type}`list[str]` the list of loaded platforms.
     Returns:
         The host platform.
     """
-    host_platform = None
-    for platform, meta in PLATFORMS.items():
-        if "freethreaded" in platform:
-            continue
-
-        if meta.os_name == os_name and meta.arch == arch:
-            host_platform = platform
-    if not host_platform:
-        fail("No platform declared for host OS {} on arch {}".format(os_name, arch))
-    return host_platform
+    candidates = []
+    for platform in platforms:
+        meta = PLATFORMS[platform]
 
-def _get_host_os_arch(rctx, logger):
-    """Infer the host OS name and arch from a repository context.
+        if meta.os_name == os_name and meta.arch == cpu_name:
+            candidates.append(platform)
 
-    Args:
-        rctx: Bazel's repository_ctx.
-        logger: Logger to use for operations.
+    if len(candidates) == 1:
+        return candidates[0]
 
-    Returns:
-        A tuple with the host OS name and arch.
-    """
-    os_name = rctx.os.name
-
-    # We assume the arch for Windows is always x86_64.
-    if "windows" in os_name.lower():
-        arch = "x86_64"
-
-        # Normalize the os_name. E.g. os_name could be "OS windows server 2019".
-        os_name = WINDOWS_NAME
-    else:
-        # This is not ideal, but bazel doesn't directly expose arch.
-        arch = repo_utils.execute_unchecked(
+    if candidates:
+        preference = repo_utils.getenv(
             rctx,
-            op = "GetUname",
-            arguments = [repo_utils.which_checked(rctx, "uname"), "-m"],
-            logger = logger,
-        ).stdout.strip()
-
-        # Normalize the os_name.
-        if "mac" in os_name.lower():
-            os_name = MACOS_NAME
-        elif "linux" in os_name.lower():
-            os_name = LINUX_NAME
-
-    return (os_name, arch)
+            "RULES_PYTHON_{}_{}_USE_HOST_PYTHON".format(
+                os_name.upper(),
+                cpu_name.upper(),
+            ),
+        )
+        if preference == None:
+            candidates = sorted(candidates, lambda k: ("freethreaded" in k, k))
+        elif preference not in candidates:
+            fail("Please choose a prefrered interpreter out of the following platforms: {}".format(candidates))
+        else:
+            candidates = [preference]
+
+    if candidates:
+        return candidates[0]
+
+    fail("Could not find a compatible 'host' python for '{os_name}', '{cpu_name}' from the loaded platforms: {platforms}".format(
+        os_name = os_name,
+        cpu_name = cpu_name,
+        platforms = platforms,
+    ))

python/versions.bzl

@@ -15,8 +15,8 @@
 """The Python versions we use for the toolchains.
 """
 
-# Values returned by https://bazel.build/rules/lib/repository_os.
-MACOS_NAME = "mac os"
+# Values present in the @platforms//os package
+MACOS_NAME = "osx"
 LINUX_NAME = "linux"
 WINDOWS_NAME = "windows"
 FREETHREADED = "freethreaded"
@@ -604,9 +604,8 @@ def _generate_platforms():
             ],
             flag_values = {},
             os_name = MACOS_NAME,
-            # Matches the value returned from:
-            # repository_ctx.execute(["uname", "-m"]).stdout.strip()
-            arch = "arm64",
+            # Matches the value in @platforms//cpu package
+            arch = "aarch64",
         ),
         "aarch64-unknown-linux-gnu": struct(
             compatible_with = [
@@ -617,9 +616,7 @@ def _generate_platforms():
                 libc: "glibc",
             },
             os_name = LINUX_NAME,
-            # Note: this string differs between OSX and Linux
-            # Matches the value returned from:
-            # repository_ctx.execute(["uname", "-m"]).stdout.strip()
+            # Matches the value in @platforms//cpu package
             arch = "aarch64",
         ),
         "armv7-unknown-linux-gnu": struct(
@@ -631,7 +628,8 @@ def _generate_platforms():
                 libc: "glibc",
             },
             os_name = LINUX_NAME,
-            arch = "armv7",
+            # Matches the value in @platforms//cpu package
+            arch = "arm",
         ),
         "i386-unknown-linux-gnu": struct(
             compatible_with = [
@@ -642,7 +640,8 @@ def _generate_platforms():
                 libc: "glibc",
             },
             os_name = LINUX_NAME,
-            arch = "i386",
+            # Matches the value in @platforms//cpu package
+            arch = "x86_32",
         ),
         "ppc64le-unknown-linux-gnu": struct(
             compatible_with = [
@@ -653,10 +652,8 @@ def _generate_platforms():
                 libc: "glibc",
             },
             os_name = LINUX_NAME,
-            # Note: this string differs between OSX and Linux
-            # Matches the value returned from:
-            # repository_ctx.execute(["uname", "-m"]).stdout.strip()
-            arch = "ppc64le",
+            # Matches the value in @platforms//cpu package
+            arch = "ppc",
         ),
         "riscv64-unknown-linux-gnu": struct(
             compatible_with = [
@@ -667,6 +664,7 @@ def _generate_platforms():
                 Label("//python/config_settings:py_linux_libc"): "glibc",
             },
             os_name = LINUX_NAME,
+            # Matches the value in @platforms//cpu package
             arch = "riscv64",
         ),
         "s390x-unknown-linux-gnu": struct(
@@ -678,9 +676,7 @@ def _generate_platforms():
                 Label("//python/config_settings:py_linux_libc"): "glibc",
             },
             os_name = LINUX_NAME,
-            # Note: this string differs between OSX and Linux
-            # Matches the value returned from:
-            # repository_ctx.execute(["uname", "-m"]).stdout.strip()
+            # Matches the value in @platforms//cpu package
             arch = "s390x",
         ),
         "x86_64-apple-darwin": struct(
@@ -690,6 +686,7 @@ def _generate_platforms():
             ],
             flag_values = {},
             os_name = MACOS_NAME,
+            # Matches the value in @platforms//cpu package
             arch = "x86_64",
         ),
         "x86_64-pc-windows-msvc": struct(
@@ -699,6 +696,7 @@ def _generate_platforms():
             ],
             flag_values = {},
             os_name = WINDOWS_NAME,
+            # Matches the value in @platforms//cpu package
             arch = "x86_64",
         ),
         "x86_64-unknown-linux-gnu": struct(
@@ -710,6 +708,7 @@ def _generate_platforms():
                 libc: "glibc",
             },
             os_name = LINUX_NAME,
+            # Matches the value in @platforms//cpu package
             arch = "x86_64",
         ),
     }