third_party/rules_graalvm_unicode.patch

From 5c67c84a014a1b0c165b8f1ff47243973fb58b6e Mon Sep 17 00:00:00 2001
From: Fabian Meumertzheim <fabian@meumertzhe.im>
Date: Wed, 4 Dec 2024 21:05:49 +0100
Subject: [PATCH] fix: Force `sun.jnu.encoding` to be UTF-8 in images (#514)

Images pick up `sun.jnu.encoding` from the host build environment, which thus needs to be UTF-8 to be able to access paths with non-ASCII characters in them on Linux. Since Bazel does not inherit `LC_CTYPE` from the host, the JVM would otherwise default to an ASCII locale.

Signed-off-by: Fabian Meumertzheim <fabian@meumertzhe.im>
---
 internal/native_image/rules.bzl | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

diff --git a/internal/native_image/rules.bzl b/internal/native_image/rules.bzl
index f21566e..1d07278 100644
--- a/internal/native_image/rules.bzl
+++ b/internal/native_image/rules.bzl
@@ -64,6 +64,9 @@ def _graal_binary_implementation(ctx):
             or install a GraalVM `native-image` toolchain.
         """)
 
+    is_linux = ctx.target_platform_has_constraint(
+        ctx.attr._linux_constraint[platform_common.ConstraintValueInfo],
+    )
     is_macos = ctx.target_platform_has_constraint(
         ctx.attr._macos_constraint[platform_common.ConstraintValueInfo],
     )
@@ -100,6 +103,15 @@ def _graal_binary_implementation(ctx):
         bin_postfix = bin_postfix,
     )
 
+    env = native_toolchain.env
+    # The native image will use the same native encoding (as determined by "sun.jnu.encoding")
+    # as the build environment, so we need to force a UTF-8 locale. On other platforms, the
+    # encoding is always UTF-8 (on macOS since JEP 400) or determined by the active code page
+    # on Windows.
+    # TODO: Match on the exec platform instead once Graal supports cross-compilation.
+    if is_linux:
+        env["LC_CTYPE"] = "C.UTF-8"
+
     # assemble final inputs
     inputs = depset(
         direct_inputs,
@@ -110,7 +122,7 @@ def _graal_binary_implementation(ctx):
         "executable": graal,
         "inputs": inputs,
         "mnemonic": "NativeImage",
-        "env": native_toolchain.env,
+        "env": env,
         "execution_requirements": {k: "" for k in native_toolchain.execution_requirements},
         "progress_message": "Native Image __target__ (__mode__) %{label}"
             .replace("__mode__", _build_action_message(ctx))