Default to a UTF-8 locale in Java stub template

On non-macOS Unix, without any locale variable set, the OpenJDK defaults
to using ASCII rather than UTF-8 as the encoding for file system paths
(i.e., the value of the `sun.jnu.encoding` property).

Fixes bazelbuild#15106

Closes bazelbuild#15159.

PiperOrigin-RevId: 445520597

Author: fmeum

src/main/java/com/google/devtools/build/lib/bazel/rules/java/java_stub_template.txt

@@ -94,6 +94,16 @@ function is_macos() {
   [[ "${OSTYPE}" =~ darwin* ]]
 }
 
+function available_utf8_locale() {
+  # Both C.UTF-8 and en_US.UTF-8 do not cause any language-specific effects
+  # when set as LC_CTYPE, but neither is certain to exist on all systems.
+  if [[ $(LC_CTYPE=C.UTF-8 locale charmap 2>/dev/null) == "UTF-8" ]]; then
+    echo "C.UTF-8"
+  elif [[ $(LC_CTYPE=en_US.UTF-8 locale charmap 2>/dev/null) == "UTF-8" ]]; then
+    echo "en_US.UTF-8"
+  fi
+}
+
 # Parse arguments sequentially until the first unrecognized arg is encountered.
 # Scan the remaining args for --wrapper_script_flag=X options and process them.
 ARGS=()
@@ -362,6 +372,17 @@ if [ -z "$CLASSPATH_LIMIT" ]; then
   is_windows && CLASSPATH_LIMIT=7000 || CLASSPATH_LIMIT=120000
 fi
 
+# On non-macOS Unix, without any locale variable set, the JVM would use
+# using ASCII rather than UTF-8 as the encoding for file system paths.
+if ! is_macos; then
+  if [ -z ${LC_CTYPE+x} ] && [ -z ${LC_ALL+x} ] && [ -z ${LANG+x} ]; then
+    UTF8_LOCALE=$(available_utf8_locale)
+    if [[ -n "$UTF8_LOCALE" ]]; then
+      export LC_CTYPE="$UTF8_LOCALE"
+    fi
+  fi
+fi
+
 if (("${#CLASSPATH}" > ${CLASSPATH_LIMIT})); then
   export JACOCO_IS_JAR_WRAPPED=1
   create_and_run_classpath_jar

src/test/shell/bazel/unicode_filenames_test.sh

@@ -206,4 +206,35 @@ function test_utf8_source_artifact_in_bep() {
   expect_log '"name":"pkg/srcs/ünïcödë fïlë.txt"'
 }
 
+function test_utf8_filename_in_java_test() {
+  # Intentionally do not check for available locales: Either C.UTF_8 or
+  # en_US.UTF-8 should exist on all CI machines - if not, we want to learn about
+  # this so that the Java stub template can be adapted accordingly.
+
+  touch WORKSPACE
+  mkdir pkg
+
+  cat >pkg/BUILD <<'EOF'
+java_test(
+    name = "Test",
+    srcs = ["Test.java"],
+    main_class = "Test",
+    use_testrunner = False,
+)
+EOF
+
+  cat >pkg/Test.java <<'EOF'
+import java.nio.file.Files;
+import java.io.IOException;
+
+class Test {
+    public static void main(String[] args) throws IOException {
+        Files.createTempFile("æøå", null);
+    }
+}
+EOF
+
+  bazel test //pkg:Test --test_output=errors 2>$TEST_log || fail "Test should pass"
+}
+
 run_suite "Tests for handling of Unicode filenames"