Preserve Unicode Windows arguments when loading native models

crates/searchi-semantic/native/CMakeLists.txt

@@ -67,6 +67,10 @@ set(MTMD_VIDEO OFF CACHE BOOL "Frames decoded by bounded media adapter" FORCE)
 set(LLAMA_OPENSSL OFF CACHE BOOL "No network runtime dependencies" FORCE)
 add_subdirectory("${LLAMA_CPP_DIR}" llama-build)
 add_executable(searchi-gemma-worker worker.cpp)
+if(MINGW)
+    # Windows argument conversion uses wmain; keep the same wide CRT entry.
+    target_link_options(searchi-gemma-worker PRIVATE -municode)
+endif()
 target_include_directories(searchi-gemma-worker PRIVATE "${LLAMA_CPP_DIR}/vendor")
 target_link_libraries(searchi-gemma-worker PRIVATE llama mtmd ${CMAKE_DL_LIBS})
 if(SEARCHI_CPU_DISPATCH)

crates/searchi-semantic/native/worker.cpp

@@ -15,16 +15,17 @@
 #include <stdexcept>
 #include <string>
 #include <vector>
-#ifdef SEARCHI_CPU_DISPATCH
-#include <filesystem>
 #ifdef _WIN32
 #define WIN32_LEAN_AND_MEAN
 #define NOMINMAX
 #include <windows.h>
-#elif defined(__linux__)
+#endif
+#ifdef SEARCHI_CPU_DISPATCH
+#include <filesystem>
+#if defined(__linux__)
 #include <dlfcn.h>
 #include <unistd.h>
-#else
+#elif !defined(_WIN32)
 #error Unsupported platform for packaged CPU dispatch
 #endif
 #endif
@@ -384,7 +385,7 @@ public:
     }
 };
 
-int main(int argc, char ** argv) {
+static int worker_main(int argc, const char * const * argv) {
     try {
         std::string path, device = "auto", role = "indexing"; int threads = 1;
         for (int i = 1; i < argc; ++i) {
@@ -447,3 +448,35 @@ int main(int argc, char ** argv) {
         return 1;
     }
 }
+
+#ifdef _WIN32
+// The narrow CRT argv uses the system ANSI code page (commonly1252), whereas
+// ggml_fopen requires UTF-8. Preserve Windows UTF-16 arguments before parsing;
+// model/runtime directories may contain any valid Unicode user/folder name.
+int wmain(int argc, wchar_t ** argv) {
+    try {
+        std::vector<std::string> utf8;
+        utf8.reserve(size_t(argc));
+        for (int i = 0; i < argc; ++i) {
+            int bytes = WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, argv[i], -1, nullptr, 0, nullptr, nullptr);
+            require(bytes > 0, "Worker argument contains invalid Windows Unicode");
+            std::string argument(size_t(bytes), '\0');
+            require(WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, argv[i], -1, argument.data(), bytes, nullptr, nullptr) == bytes,
+                "Cannot convert worker argument to UTF-8");
+            argument.pop_back(); // Conversion includes the terminating NUL.
+            utf8.push_back(std::move(argument));
+        }
+        std::vector<const char *> arguments;
+        arguments.reserve(utf8.size());
+        for (const auto & argument : utf8) arguments.push_back(argument.c_str());
+        return worker_main(argc, arguments.data());
+    } catch (const std::exception & error) {
+        std::cerr << "Searchi Gemma worker: " << error.what() << '\n';
+        return 1;
+    }
+}
+#else
+int main(int argc, char ** argv) {
+    return worker_main(argc, argv);
+}
+#endif