fgets and utf8 functions for Windows (#87)

contrib/win32/openssh/libssh.vcxproj

@@ -283,7 +283,6 @@
     <ClCompile Include="$(OpenSSH-Src-Path)platform-tracing.c" />
     <ClCompile Include="$(OpenSSH-Src-Path)platform.c" />
     <ClCompile Include="$(OpenSSH-Src-Path)sandbox-pledge.c" />
-    <ClCompile Include="$(OpenSSH-Src-Path)utf8.c" />
     <ClCompile Include="$(OpenSSH-Src-Path)contrib\win32\win32compat\ttymodes_windows.c" />
     <ClCompile Include="$(OpenSSH-Src-Path)digest-openssl.c">
       <ExcludedFromBuild Condition="$(UseOpenSSL)==false">true</ExcludedFromBuild>

contrib/win32/openssh/libssh.vcxproj.filters

@@ -282,9 +282,6 @@
     <ClCompile Include="$(OpenSSH-Src-Path)sandbox-pledge.c">
       <Filter>Source Files</Filter>
     </ClCompile>
-    <ClCompile Include="$(OpenSSH-Src-Path)utf8.c">
-      <Filter>Source Files</Filter>
-    </ClCompile>
     <ClCompile Include="$(OpenSSH-Src-Path)contrib\win32\win32compat\ttymodes_windows.c">
       <Filter>Source Files</Filter>
     </ClCompile>

contrib/win32/openssh/win32iocompat.vcxproj

@@ -160,6 +160,7 @@
     <ClCompile Include="$(OpenSSH-Src-Path)\contrib\win32\win32compat\tncon.c" />
     <ClCompile Include="$(OpenSSH-Src-Path)\contrib\win32\win32compat\tnnet.c" />
     <ClCompile Include="$(OpenSSH-Src-Path)\contrib\win32\win32compat\utf.c" />
+    <ClCompile Include="$(OpenSSH-Src-Path)\contrib\win32\win32compat\win32-utf8.c" />
   </ItemGroup>
   <ItemGroup>
     <ClInclude Include="$(OpenSSH-Src-Path)\contrib\win32\win32compat\w32fd.h" />

contrib/win32/openssh/win32iocompat.vcxproj.filters

@@ -19,6 +19,7 @@
     <ClCompile Include="$(OpenSSH-Src-Path)\contrib\win32\win32compat\tncon.c" />
     <ClCompile Include="$(OpenSSH-Src-Path)\contrib\win32\win32compat\tnnet.c" />
     <ClCompile Include="$(OpenSSH-Src-Path)\contrib\win32\win32compat\utf.c" />
+    <ClCompile Include="$(OpenSSH-Src-Path)\contrib\win32\win32compat\win32-utf8.c" />
   </ItemGroup>
   <ItemGroup>
     <ClInclude Include="$(OpenSSH-Src-Path)\contrib\win32\win32compat\w32fd.h" />

contrib/win32/win32compat/inc/stdio.h

@@ -5,6 +5,12 @@
 FILE* w32_fopen_utf8(const char *, const char *);
 #define fopen w32_fopen_utf8
 
+char* w32_fgets(char *str, int n, FILE *stream);
+#define fgets w32_fgets
+
+int w32_setvbuf(FILE *stream,char *buffer, int mode, size_t size);
+#define setvbuf w32_setvbuf
+
 /* stdio.h additional definitions */
 #define popen _popen
 #define pclose _pclose
@@ -14,4 +20,3 @@ FILE* w32_fdopen(int fd, const char *mode);
 
 int w32_rename(const char *old_name, const char *new_name);
 #define rename w32_rename
-

contrib/win32/win32compat/misc.c

@@ -276,6 +276,75 @@ w32_fopen_utf8(const char *path, const char *mode)
 	return f;
 }
 
+/* fgets to support Unicode input */
+char*
+ w32_fgets(char *str, int n, FILE *stream) {
+	HANDLE h = (HANDLE)_get_osfhandle(_fileno(stream));
+	wchar_t* str_w = NULL;
+	char *ret = NULL, *str_tmp = NULL;
+
+	if (h != NULL && h != INVALID_HANDLE_VALUE
+	    && GetFileType(h) == FILE_TYPE_CHAR) {
+		/* 
+		 * read only n/4 wide chars from console 
+		 * each UTF-16 char may bloat upto 4 utf-8 chars when converted to utf-8 
+		 * so we can fit in str[n] provided as input
+		 */
+		if ((str_w = malloc((n/4) * sizeof(wchar_t))) == NULL) {
+			errno = ENOMEM;
+			goto cleanup;
+		}
+		/* prepare for Unicode input */
+		_setmode(_fileno(stream), O_U16TEXT); 
+		if (fgetws(str_w, n/4, stream) == NULL)
+			goto cleanup;
+		if ((str_tmp = utf16_to_utf8(str_w)) == NULL) {
+			errno = ENOMEM;
+			goto cleanup;
+		}
+		if (strlen(str_tmp) > n - 1) {
+			/* shouldn't happen. but handling in case */
+			errno = EINVAL;
+			goto cleanup;
+		}
+		memcpy(str, str_tmp, strlen(str_tmp) + 1);
+		ret = str;
+	}
+	else
+		ret = fgets(str, n, stream);
+cleanup:
+	if (str_w)
+		free(str_w);
+	if (str_tmp)
+		free(str_tmp);
+	return ret;
+}
+
+/* Account for differences between Unix's and Windows versions of setvbuf */
+int 
+w32_setvbuf(FILE *stream, char *buffer, int mode, size_t size) {
+	
+	/* BUG: setvbuf on console stream interferes with Unicode I/O	*/
+	HANDLE h = (HANDLE)_get_osfhandle(_fileno(stream));
+	
+	if (h != NULL && h != INVALID_HANDLE_VALUE
+	    && GetFileType(h) == FILE_TYPE_CHAR)
+		return 0;
+
+	/* BUG: setvbuf on file stream is interfering with w32_fopen */
+	/* short circuit for now*/
+	return 0;
+
+	/*
+	 * if size is 0, set no buffering. 
+	 * Windows does not differentiate __IOLBF and _IOFBF
+	 */
+	if (size == 0)
+		return setvbuf(stream, NULL, _IONBF, 0);
+	else
+		return setvbuf(stream, buffer, mode, size);
+}
+
 char *
 w32_programdir()
 {

contrib/win32/win32compat/win32-utf8.c

@@ -0,0 +1,52 @@
+/* 
+ * Temporary Windows versions of functions implemented in utf8.c
+ */
+ #include <stdio.h>
+#include <stdarg.h>
+
+int
+vfmprintf(FILE *f, const char *fmt, va_list list) 
+{
+	return vfprintf(f, fmt, list);
+}
+
+int
+mprintf(const char *fmt, ...) 
+{
+	int ret = 0;
+	va_list valist;
+	va_start(valist, fmt);
+	ret = vfmprintf(stdout, fmt, valist);
+	va_end(valist);
+	return ret;
+}
+
+int
+fmprintf(FILE *f, const char *fmt, ...) 
+{
+	int ret = 0;
+	va_list valist;
+	va_start(valist, fmt);
+	ret = vfmprintf(f, fmt, valist);
+	va_end(valist);
+	return ret;
+}
+
+int
+snmprintf(char *buf, size_t len, int *written, const char *fmt, ...) 
+{
+	int num;
+	va_list valist;
+	va_start(valist, fmt);
+	num = vsnprintf(buf, len, fmt, valist);
+	va_end(valist);
+	*written = num;
+	return 0;
+}
+
+void
+msetlocale(void) 
+{
+	return;
+}
+

scp.c

@@ -1034,17 +1034,7 @@ rsource(char *name, struct stat *statp)
 	(void) snprintf(path, sizeof path, "D%04o %d %.1024s\n",
 	    (u_int) (statp->st_mode & FILEMODEMASK), 0, last);
 	if (verbose_mode)
-#ifdef WINDOWS
-	/* TODO - make fmprintf work for Windows  */
-	{
-		printf("Entering directory: ");
-		wchar_t* wtmp = utf8_to_utf16(path);
-		WriteConsoleW(GetStdHandle(STD_ERROR_HANDLE), wtmp, wcslen(wtmp), 0, 0);
-		free(wtmp);
-	}
-#else /* !WINDOWS */
 		fmprintf(stderr, "Entering directory: %s", path);
-#endif /* !WINDOWS */
 	(void) atomicio(vwrite, remout, path, strlen(path));
 	if (response() < 0) {
 		closedir(dirp);
@@ -1119,17 +1109,7 @@ sink(int argc, char **argv)
 		} while (cp < &buf[sizeof(buf) - 1] && ch != '\n');
 		*cp = 0;
 		if (verbose_mode)
-#ifdef WINDOWS
-		/* TODO - make fmprintf work for Windows  */
-		{
-			printf("Sink: ");
-			wchar_t* wtmp = utf8_to_utf16(buf);
-			WriteConsoleW(GetStdHandle(STD_ERROR_HANDLE), wtmp, wcslen(wtmp), 0, 0);
-			free(wtmp);
-		}
-#else /* !WINDOWS */
 			fmprintf(stderr, "Sink: %s", buf);
-#endif /* !WINDOWS */
 
 		if (buf[0] == '\01' || buf[0] == '\02') {
 			if (iamremote == 0) {

sftp.c

@@ -292,27 +292,6 @@ help(void)
 	    "?                                  Synonym for help\n");
 }
 
-#ifdef WINDOWS
-/* printf version to account for utf-8 input */
-/* TODO - merge this with vfmprint */
-static void printf_utf8(char *fmt,  ... ) {
-	/* TODO - is 1024 sufficient */
-	char buf[1024];
-	int length = 0;
-	
-	va_list valist;
-	va_start(valist, fmt);
-	length = vsnprintf(buf, 1024, fmt, valist);
-	va_end(valist);
-
-	write(STDOUT_FILENO, buf, length);
-}
-
-/* override mprintf */
-#define mprintf(a,...)		printf_utf8((a), __VA_ARGS__)
-#define printf(a,...)		printf_utf8((a), __VA_ARGS__)
-#endif   /* WINDOWS */
-
 static void
 local_do_shell(const char *args)
 {
@@ -420,7 +399,7 @@ make_absolute(char *p, const char *pwd)
 		p = abs_str;
 	}
 
-	/* convert '\\' tp '/' */
+	/* convert '\\' to '/' */
 	convertToForwardslash(p);
 
 	/* Append "/" if needed to the absolute windows path */	
@@ -925,23 +904,7 @@ do_ls_dir(struct sftp_conn *conn, const char *path,
 			} else
 				mprintf("%s\n", d[n]->longname);
 		} else {
-#ifdef WINDOWS
-			/* cannot use printf_utf8 becuase of width specification */
-			/* printf_utf8 does not account for utf-16 based argument widths */
-			char *p = NULL;
-			wchar_t buf[1024]; 
-			wchar_t* wtmp = utf8_to_utf16(fname);
-			swprintf(buf, 1024, L"%-*s", colspace, wtmp);
-			
-			if ((p = utf16_to_utf8(buf)) == NULL)
-				continue;
-			
-			write(STDOUT_FILENO, p, strlen(p));			
-			free(wtmp);
-			free(p);
-#else
 			mprintf("%-*s", colspace, fname);
-#endif
 			if (c >= columns) {
 				printf("\n");
 				c = 1;
@@ -1025,23 +988,7 @@ do_globbed_ls(struct sftp_conn *conn, const char *path,
 			mprintf("%s\n", lname);
 			free(lname);
 		} else {
-#ifdef WINDOWS
-			/* cannot use printf_utf8 becuase of width specification */
-			/* printf_utf8 does not account for utf-16 based argument widths */
-			char *p = NULL;
-			wchar_t buf[1024];
-			wchar_t* wtmp = utf8_to_utf16(fname);
-			swprintf(buf, 1024, L"%-*s", colspace, wtmp);
-			
-			if ((p = utf16_to_utf8(buf)) == NULL)
-				continue;
-
-			write(STDOUT_FILENO, p, strlen(p));			
-			free(wtmp);
-			free(p);
-#else
 			mprintf("%-*s", colspace, fname);
-#endif
 			if (c >= columns) {
 				printf("\n");
 				c = 1;
@@ -2211,20 +2158,8 @@ interactive_loop(struct sftp_conn *conn, char *file1, char *file2)
 	interactive = !batchmode && isatty(STDIN_FILENO);
 	err = 0;
 
-#ifdef WINDOWS
-	/* Min buffer size allowed in Windows is 2*/
-	setvbuf(stdout, NULL, _IOLBF, 2);
-	
-	/* We do this only in interactive mode as we are unable to read files with UTF8 BOM */
-	if (interactive) {
-		setvbuf(infile, NULL, _IOLBF, 2);
-		_setmode(_fileno(stdin), O_U16TEXT); /* prepare for Unicode input */
-	}
-#else   /* !WINDOWS */
 	setvbuf(stdout, NULL, _IOLBF, 0);
 	setvbuf(infile, NULL, _IOLBF, 0);
-#endif   /* !WINDOWS */
-
 
 	for (;;) {
 		char *cp;
@@ -2232,25 +2167,6 @@ interactive_loop(struct sftp_conn *conn, char *file1, char *file2)
 		signal(SIGINT, SIG_IGN);
 
 		if (el == NULL) {
-#ifdef WINDOWS
-			/* fgets on Windows does not support Unicode input*/
-			if (interactive) {
-				wchar_t wcmd[2048];
-				printf("sftp> ");
-				if (fgetws(wcmd, sizeof(cmd)/sizeof(wchar_t), infile) == NULL) {
-					printf("\n");
-					break;
-				}
-				else {
-					char *pcmd = NULL;
-					if ((pcmd = utf16_to_utf8(wcmd)) == NULL)
-						fatal("failed to convert input arguments");
-					strcpy(cmd, pcmd);
-					free(pcmd);                    
-				}
-			} else if (fgets(cmd, sizeof(cmd), infile) == NULL) 
-				break;
-#else /* !WINDOWS */
 			if (interactive)
 				printf("sftp> ");
 			if (fgets(cmd, sizeof(cmd), infile) == NULL) {
@@ -2258,7 +2174,6 @@ interactive_loop(struct sftp_conn *conn, char *file1, char *file2)
 					printf("\n");
 				break;
 			}
-#endif/* !WINDOWS */
 			if (!interactive) { /* Echo command */
 				mprintf("sftp> %s", cmd);
 				if (strlen(cmd) > 0 &&

utf8.c

@@ -1,4 +1,4 @@
-/* $OpenBSD: utf8.c,v 1.4 2017/02/02 10:54:25 jsg Exp $ */
+/* $OpenBSD: utf8.c,v 1.5 2017/02/19 00:10:57 djm Exp $ */
 /*
  * Copyright (c) 2016 Ingo Schwarze <schwarze@openbsd.org>
  *
@@ -57,16 +57,11 @@ static int	 vasnmprintf(char **, size_t, int *, const char *, va_list);
 
 static int
 dangerous_locale(void) {
-#ifdef WINDOWS
-	wchar_t loc[LOCALE_NAME_MAX_LENGTH];
-	GetSystemDefaultLocaleName(loc, LOCALE_NAME_MAX_LENGTH);
-	return wcscmp(loc, L"US-ASCII") && wcscmp(loc, L"UTF-8");
-#else   /* !WINDOWS */
 	char	*loc;
 
 	loc = nl_langinfo(CODESET);
-	return strcmp(loc, "US-ASCII") && strcmp(loc, "UTF-8");
+	return strcmp(loc, "US-ASCII") != 0 && strcmp(loc, "UTF-8") != 0 &&
-#endif    /* !WINDOWS */
+	    strcmp(loc, "ANSI_X3.4-1968") != 0;
 }
 
 static int
@@ -337,3 +332,4 @@ msetlocale(void)
 	/* We can handle this locale */
 	setlocale(LC_CTYPE, "");
 }
+