Fix singlejar's handling of unicode filenames on windows

Unicode command line arguments on windows require using the `wmain` entry point with the `wchar_t**` argument type. This is then converted into multi-byte `char**` arguments.

Work towards bazelbuild/rules_java#295

PiperOrigin-RevId: 772959857
Change-Id: I8aa1cf54aa0291c40c7e5e62d14decc56f3780bd

Author: hvadehra

src/main/cpp/util/strings.cc

@@ -381,6 +381,18 @@ std::wstring CstringToWstring(const std::string &input) {
   return result;
 }
 
+char** WArgsToCArgs(int argc, wchar_t **wargv) {
+  char **argv = new char*[argc];
+  for (int i = 0; i < argc; i++) {
+    std::string arg = WstringToCstring(std::wstring(wargv[i]));
+    argv[i] = new char[arg.length()+1];
+    for (int j = 0; j < arg.length(); j++) {
+      argv[i][j] = arg[j];
+    }
+    argv[i][arg.length()] = 0;
+  }
+  return argv;
+}
 #endif  // defined(_WIN32) || defined(__CYGWIN__)
 
 }  // namespace blaze_util

src/main/cpp/util/strings.h

@@ -135,6 +135,8 @@ std::string WstringToCstring(const std::wstring &input);
 
 // Deprecated. Use AcpToWcs or Utf8ToWcs.
 std::wstring CstringToWstring(const std::string &input);
+
+char** WArgsToCArgs(int argc, wchar_t **wargv);
 #endif  // defined(_WIN32) || defined(__CYGWIN__)
 
 }  // namespace blaze_util

src/tools/singlejar/BUILD

@@ -1,3 +1,4 @@
+load("@bazel_skylib//rules:build_test.bzl", "build_test")
 load("@rules_cc//cc:cc_binary.bzl", "cc_binary")
 load("@rules_cc//cc:cc_library.bzl", "cc_library")
 load("@rules_cc//cc:cc_test.bzl", "cc_test")
@@ -588,3 +589,24 @@ genrule(
     toolchains = ["@rules_java//toolchains:current_java_runtime"],
     tools = ["@rules_java//toolchains:current_java_runtime"],
 )
+
+build_test(
+    name = "singlejar_unicode_filenames_test",
+    targets = [
+        ":gen_singlejar_unicode_filenames",
+    ],
+)
+
+genrule(
+    name = "gen_singlejar_unicode_filenames",
+    outs = [
+        "unicode_filenames.jar",
+    ],
+    cmd = """
+          touch रैन्डम.txt らんどみ.txt 艾诶艾迪勒开.txt
+          $(location :singlejar) --resources रैन्डम.txt らんどみ.txt 艾诶艾迪勒开.txt --output $@
+          """,
+    tools = [
+        ":singlejar",
+    ],
+)

src/tools/singlejar/singlejar_local_main.cc

@@ -19,7 +19,13 @@
 #include "src/tools/singlejar/options.h"
 #include "src/tools/singlejar/output_jar.h"
 
+#ifdef _WIN32
+#include "src/main/cpp/util/strings.h"
+int wmain(int argc, wchar_t *wargv[]) {
+  char **argv = blaze_util::WArgsToCArgs(argc, wargv);
+#else
 int main(int argc, char *argv[]) {
+#endif
   Options options;
   options.ParseCommandLine(argc - 1, argv + 1);
   OutputJar output_jar;

src/tools/singlejar/singlejar_main.cc

@@ -18,7 +18,13 @@
 #include "src/tools/singlejar/options.h"
 #include "src/tools/singlejar/output_jar.h"
 
+#ifdef _WIN32
+#include "src/main/cpp/util/strings.h"
+int wmain(int argc, wchar_t *wargv[]) {
+  char **argv = blaze_util::WArgsToCArgs(argc, wargv);
+#else
 int main(int argc, char *argv[]) {
+#endif
   Options options;
   options.ParseCommandLine(argc - 1, argv + 1);
   OutputJar output_jar;