From fe53a4100a553a17f08a0f6f83af90c481877833 Mon Sep 17 00:00:00 2001
From: Manoj Ampalam <manojampalam@live.com>
Date: Sun, 9 Oct 2016 23:29:52 -0700
Subject: [PATCH] fopen Windows wrapper to support reading UTF-8 encoded files

---
 contrib/win32/win32compat/inc/unistd.h   |  2 +
 contrib/win32/win32compat/inc/w32posix.h |  1 +
 contrib/win32/win32compat/misc.c         | 48 ++++++++++++++++++++++++
 3 files changed, 51 insertions(+)

diff --git a/contrib/win32/win32compat/inc/unistd.h b/contrib/win32/win32compat/inc/unistd.h
index 3e1a055..d65150c 100644
--- a/contrib/win32/win32compat/inc/unistd.h
+++ b/contrib/win32/win32compat/inc/unistd.h
@@ -35,6 +35,8 @@
 
 #define getdtablesize() MAX_FDS
 
+#define fopen w32_fopen_utf8
+
 /* Compatibility header to avoid lots of #ifdefs in includes.h on Win32 */
 
 #include <conio.h>
diff --git a/contrib/win32/win32compat/inc/w32posix.h b/contrib/win32/win32compat/inc/w32posix.h
index 4fc91df..78a31fd 100644
--- a/contrib/win32/win32compat/inc/w32posix.h
+++ b/contrib/win32/win32compat/inc/w32posix.h
@@ -69,6 +69,7 @@ sighandler_t w32_signal(int signum, sighandler_t handler);
 int w32_sigprocmask(int how, const sigset_t *set, sigset_t *oldset);
 int w32_raise(int sig);
 int w32_kill(int pid, int sig);
+FILE* w32_fopen_utf8(const char *, const char *);
 
 
 /* Shutdown constants */
diff --git a/contrib/win32/win32compat/misc.c b/contrib/win32/win32compat/misc.c
index 038bf80..ec6c109 100644
--- a/contrib/win32/win32compat/misc.c
+++ b/contrib/win32/win32compat/misc.c
@@ -1,4 +1,5 @@
 #include <Windows.h>
+#include <stdio.h>
 #include "inc\defs.h"
 #include "inc\sys\statvfs.h"
 
@@ -69,4 +70,51 @@ int dlclose(HMODULE handle) {
 
 FARPROC dlsym(HMODULE handle, const char *symbol) {
 	return GetProcAddress(handle, symbol);
+}
+
+
+/*fopen on Windows to mimic https://linux.die.net/man/3/fopen
+* only r, w, a are supported for now
+*/
+FILE*
+w32_fopen_utf8(const char *path, const char *mode) {
+	wchar_t wpath[MAX_PATH], wmode[5];
+	FILE* f;
+	char utf8_bom[] = { 0xEF,0xBB,0xBF };
+	char first3_bytes[3];
+
+	if (mode[1] != '\0') {
+		errno = ENOTSUP;
+		return NULL;
+	}
+
+	if (MultiByteToWideChar(CP_UTF8, 0, path, -1, wpath, MAX_PATH) == 0 ||
+		MultiByteToWideChar(CP_UTF8, 0, mode, -1, wmode, 5) == 0) {
+		errno = EFAULT;
+		debug("WideCharToMultiByte failed for %c - ERROR:%d", path, GetLastError());
+		return NULL;
+	}
+
+	f = _wfopen(wpath, wmode);
+
+	if (f) {
+		/* BOM adjustments for file streams*/
+		if (mode[0] == 'w' && fseek(f, 0, SEEK_SET) != EBADF) {
+			/* write UTF-8 BOM - should we ?*/
+			/*if (fwrite(utf8_bom, sizeof(utf8_bom), 1, f) != 1) {
+				fclose(f);
+				return NULL;
+			}*/
+
+		}
+		else if (mode[0] == 'r' && fseek(f, 0, SEEK_SET) != EBADF) {
+			/* read out UTF-8 BOM if present*/
+			if (fread(first3_bytes, 3, 1, f) != 1 ||
+				memcmp(first3_bytes, utf8_bom, 3) != 0) {
+				fseek(f, 0, SEEK_SET);
+			}
+		}
+	}
+
+	return f;
 }
\ No newline at end of file