Fix Windows unicode paths

Makefile.am

@@ -128,6 +128,7 @@ noinst_HEADERS = \
        src/bin/write/mod_xlsx.h \
        src/bin/write/module.h \
        src/bin/write/module_util.h \
+       src/bin/util/main.h \
        src/bin/util/file_format.h \
        src/bin/util/quote_and_escape.h \
        src/bin/util/readstat_dta_days.h \

VS17/ReadStat_App/ReadStat_App.vcxproj

@@ -57,6 +57,7 @@
     <ClInclude Include="..\..\src\bin\write\module.h" />
     <ClInclude Include="..\..\src\bin\write\module_util.h" />
     <ClInclude Include="..\..\src\bin\util\file_format.h" />
+    <ClInclude Include="..\..\src\bin\util\main.h" />
     <ClInclude Include="..\..\src\bin\util\quote_and_escape.h" />
   </ItemGroup>
   <PropertyGroup Label="Globals">

VS17/ReadStat_App/ReadStat_App.vcxproj.filters

@@ -131,6 +131,9 @@
     <ClInclude Include="..\..\src\bin\util\file_format.h">
       <Filter>Header Files</Filter>
     </ClInclude>
+    <ClInclude Include="..\..\src\bin\util\main.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
     <ClInclude Include="..\..\src\bin\util\quote_and_escape.h">
       <Filter>Header Files</Filter>
     </ClInclude>

src/bin/extract_metadata.c

@@ -9,6 +9,7 @@
 #include "util/readstat_dta_days.h"
 #include "util/quote_and_escape.h"
 #include "util/file_format.h"
+#include "util/main.h"
 #include "extract_metadata.h"
 #include "write/json/write_missing_values.h"
 #include "write/json/write_value_labels.h"
@@ -303,7 +304,7 @@ cleanup: readstat_parser_free(parser);
     return ret;
 }
 
-int main(int argc, char *argv[]) {
+int portable_main(int argc, char *argv[]) {
     if (argc != 3) {
         printf("Usage: %s <input-filename.(dta|sav)> <output-metadata.json>\n", argv[0]);
         return 1;

src/bin/readstat.c

@@ -44,6 +44,7 @@ int gettimeofday(struct timeval* t, void* timezone)
 #endif
 
 #include "util/file_format.h"
+#include "util/main.h"
 
 #if defined _MSC_VER
 #define unlink _unlink
@@ -480,7 +481,7 @@ static int dump_file(const char *input_filename) {
     return 0;
 }
 
-int main(int argc, char** argv) {
+int portable_main(int argc, char** argv) {
     char *input_filename = NULL;
     char *catalog_filename = NULL;
     char *output_filename = NULL;

src/bin/util/main.h

@@ -0,0 +1,70 @@
+#include <stdlib.h>
+#include <stdio.h>
+
+
+// True main for all platforms
+int portable_main(int argc, char *argv[]);
+
+
+#if defined _WIN32
+#include <windows.h>
+    // Standard way of decoding wide-string command-line arguments on Windows.
+    // Call portable_main with UTF-8 strings.
+    int main(int unused_argc, char *unused_argv[]) {
+        int argc;
+        int ret = 1;
+        wchar_t** utf16_argv = NULL;
+        char** utf8_argv = NULL;
+
+        // Manual standard argument decoding needed since wmain is not supported by MinGW by default.
+        utf16_argv = CommandLineToArgvW(GetCommandLineW(), &argc);
+
+        if(utf16_argv == NULL) {
+            fprintf(stderr, "Fatal error: command line argument extraction failure\n");
+            goto cleanup;
+        }
+
+        utf8_argv = calloc(argc, sizeof(char*));
+
+        for (int i=0; i<argc; ++i) {
+            const int len = WideCharToMultiByte(CP_UTF8, 0, utf16_argv[i], -1, NULL, 0, NULL, NULL);
+
+            if (len <= 0) {
+                fprintf(stderr, "Fatal error: command line encoding failure (argument %d)\n", i+1);
+                goto cleanup;
+            }
+
+            utf8_argv[i] = malloc(len + 1);
+            const size_t ret = WideCharToMultiByte(CP_UTF8, 0, utf16_argv[i], -1, utf8_argv[i], len, NULL, NULL);
+
+            if (ret <= 0) {
+                fprintf(stderr, "Fatal error: command line encoding failure (argument %d)\n", i+1);
+                goto cleanup;
+            }
+
+            utf8_argv[i][len] = 0;
+        }
+
+        ret = portable_main(argc, utf8_argv);
+
+    cleanup:
+        if(utf8_argv != NULL)
+            for(int i=0; i<argc; ++i)
+                free(utf8_argv[i]);
+
+        free(utf8_argv);
+        LocalFree(utf16_argv);
+        return ret;
+    }
+#else
+    // Proxy main.
+    // On Mac, argv encoding is the current local encoding.
+    // On Linux, argv encoding is difficult to know, but it 
+    // should often be the current local encoding (generally UTF-8).
+    int main(int argc, char *argv[])
+    {
+        portable_main(argc, argv);
+        return 0;
+    }
+#endif
+

src/readstat_io_unistd.c

@@ -1,9 +1,13 @@
 
 #include <fcntl.h>
 #include <stdlib.h>
+#include <wchar.h>
 #if !defined(_MSC_VER)
 #   include <unistd.h>
 #endif
+#if defined _WIN32
+#   include <windows.h>
+#endif
 
 #include "readstat.h"
 #include "readstat_io_unistd.h"
@@ -24,8 +28,35 @@
 #endif
 
 
+int open_with_unicode(const char *path, int options)
+{
+#if defined _WIN32
+    const int buffer_size = MultiByteToWideChar(CP_UTF8, 0, path, -1, NULL, 0);
+
+    if(buffer_size <= 0)
+        return -1;
+
+    wchar_t* wpath = malloc((buffer_size + 1) * sizeof(wchar_t));
+    const int res = MultiByteToWideChar(CP_UTF8, 0, path, -1, wpath, buffer_size);
+    wpath[buffer_size] = 0;
+
+    if(res <= 0)
+    {
+        free(wpath);
+        return -1;
+    }
+
+    int fd = _wopen(wpath, options);
+
+    free(wpath);
+    return fd;
+#else
+    return open(path, options);
+#endif
+}
+
 int unistd_open_handler(const char *path, void *io_ctx) {
-    int fd = open(path, UNISTD_OPEN_OPTIONS);
+    int fd = open_with_unicode(path, UNISTD_OPEN_OPTIONS);
     ((unistd_io_ctx_t*) io_ctx)->fd = fd;
     return fd;
 }