TCF Agent: added UNICODE support in file names on Windows

3 files changed, 186 insertions, 39 deletions

diff --git a/filesystem.c b/filesystem.c
index b7928bb4..d14af4a9 100644
--- a/filesystem.c
+++ b/filesystem.c
@@ -27,7 +27,7 @@
 #include <assert.h>
 #include <fcntl.h>
 #include <sys/stat.h>
-#ifndef _MSC_VER
+#if !defined(WIN32) || defined(__CYGWIN__)
 #  include <utime.h>
 #  include <dirent.h>
 #endif
@@ -948,7 +948,7 @@ static void command_mkdir(char * token, Channel * c) {
     if (attrs.flags & ATTR_PERMISSIONS) {
         mode = attrs.permissions;
     }
-#if defined(_MSC_VER) || defined(_WRS_KERNEL) || defined(__MINGW32__)
+#if defined(_WRS_KERNEL)
     if (mkdir(path) < 0) err = errno;
 #else
     if (mkdir(path, mode) < 0) err = errno;
diff --git a/mdep.c b/mdep.c
index dd43f627..d164e1b4 100644
--- a/mdep.c
+++ b/mdep.c
@@ -445,7 +445,7 @@ int inet_pton(int af, const char * src, void * dst) {
 
 #endif /* WIN32 */
 
-#if defined(_MSC_VER) || defined(__MINGW32__)
+#if defined(WIN32) && !defined(__CYGWIN__)
 
 static __int64 file_time_to_unix_time(const FILETIME * ft) {
     __int64 res = (__int64)ft->dwHighDateTime << 32;
@@ -478,7 +478,7 @@ void usleep(useconds_t useconds) {
 
 int truncate(const char * path, int64_t size) {
     int res = 0;
-    int f = _open(path, _O_RDWR | _O_BINARY);
+    int f = open(path, _O_RDWR | _O_BINARY, 0);
     if (f < 0) return -1;
     res = ftruncate(f, size);
     _close(f);
@@ -545,10 +545,105 @@ ssize_t pwrite(int fd, const void * buf, size_t size, off_t offset) {
     return wr;
 }
 
-#endif /* defined(_MSC_VER) || defined(__MINGW32__) */
+int utf8_stat(const char * name, struct utf8_stat * buf) {
+    struct _stati64 tmp;
+    wchar_t path[FILE_PATH_SIZE];
+    if (!MultiByteToWideChar(CP_UTF8, 0, name, -1, path, sizeof(path))) {
+        set_win32_errno(GetLastError());
+        return -1;
+    }
+    if (_wstati64(path, &tmp)) return -1;
+    buf->st_dev = tmp.st_dev;
+    buf->st_ino = tmp.st_ino;
+    buf->st_mode = tmp.st_mode;
+    buf->st_nlink = tmp.st_nlink;
+    buf->st_uid = tmp.st_uid;
+    buf->st_gid = tmp.st_gid;
+    buf->st_rdev = tmp.st_rdev;
+    buf->st_size = tmp.st_size;
+    buf->st_atime = tmp.st_atime;
+    buf->st_mtime = tmp.st_mtime;
+    buf->st_ctime = tmp.st_ctime;
+    return 0;
+}
+
+int utf8_fstat(int fd, struct utf8_stat * buf) {
+    struct _stati64 tmp;
+    if (_fstati64(fd, &tmp)) return -1;
+    buf->st_dev = tmp.st_dev;
+    buf->st_ino = tmp.st_ino;
+    buf->st_mode = tmp.st_mode;
+    buf->st_nlink = tmp.st_nlink;
+    buf->st_uid = tmp.st_uid;
+    buf->st_gid = tmp.st_gid;
+    buf->st_rdev = tmp.st_rdev;
+    buf->st_size = tmp.st_size;
+    buf->st_atime = tmp.st_atime;
+    buf->st_mtime = tmp.st_mtime;
+    buf->st_ctime = tmp.st_ctime;
+    return 0;
+}
+
+int utf8_open(const char * name, int flags, int perms) {
+    wchar_t path[FILE_PATH_SIZE];
+    if (!MultiByteToWideChar(CP_UTF8, 0, name, -1, path, sizeof(path))) {
+        set_win32_errno(GetLastError());
+        return -1;
+    }
+    return _wopen(path, flags, perms);
+}
 
-#if defined(_MSC_VER)
-DIR * opendir(const char *path) {
+int utf8_chmod(const char * name, int mode) {
+    wchar_t path[FILE_PATH_SIZE];
+    if (!MultiByteToWideChar(CP_UTF8, 0, name, -1, path, sizeof(path))) {
+        set_win32_errno(GetLastError());
+        return -1;
+    }
+    return _wchmod(path, mode);
+}
+
+int utf8_remove(const char * name) {
+    wchar_t path[FILE_PATH_SIZE];
+    if (!MultiByteToWideChar(CP_UTF8, 0, name, -1, path, sizeof(path))) {
+        set_win32_errno(GetLastError());
+        return -1;
+    }
+    return _wremove(path);
+}
+
+int utf8_rmdir(const char * name) {
+    wchar_t path[FILE_PATH_SIZE];
+    if (!MultiByteToWideChar(CP_UTF8, 0, name, -1, path, sizeof(path))) {
+        set_win32_errno(GetLastError());
+        return -1;
+    }
+    return _wrmdir(path);
+}
+
+int utf8_mkdir(const char * name, int mode) {
+    wchar_t path[FILE_PATH_SIZE];
+    if (!MultiByteToWideChar(CP_UTF8, 0, name, -1, path, sizeof(path))) {
+        set_win32_errno(GetLastError());
+        return -1;
+    }
+    return _wmkdir(path);
+}
+
+int utf8_rename(const char * name1, const char * name2) {
+    wchar_t path1[FILE_PATH_SIZE];
+    wchar_t path2[FILE_PATH_SIZE];
+    if (!MultiByteToWideChar(CP_UTF8, 0, name1, -1, path1, sizeof(path1))) {
+        set_win32_errno(GetLastError());
+        return -1;
+    }
+    if (!MultiByteToWideChar(CP_UTF8, 0, name2, -1, path2, sizeof(path2))) {
+        set_win32_errno(GetLastError());
+        return -1;
+    }
+    return _wrename(path1, path2);
+}
+
+DIR * utf8_opendir(const char * path) {
     DIR * d = (DIR *)loc_alloc(sizeof(DIR));
     if (!d) { errno = ENOMEM; return 0; }
     strcpy(d->path, path);
@@ -557,23 +652,32 @@ DIR * opendir(const char *path) {
     return d;
 }
 
-struct dirent * readdir(DIR *d) {
+struct dirent * utf8_readdir(DIR * d) {
     static struct dirent de;
+
     if (d->hdl < 0) {
-        d->hdl = _findfirsti64(d->path, &d->blk);
+        wchar_t path[FILE_PATH_SIZE];
+        if (!MultiByteToWideChar(CP_UTF8, 0, d->path, -1, path, sizeof(path))) {
+            set_win32_errno(GetLastError());
+            return 0;
+        }
+        d->hdl = _wfindfirsti64(path, &d->blk);
         if (d->hdl < 0) {
             if (errno == ENOENT) errno = 0;
             return 0;
         }
     }
     else {
-        int r = _findnexti64(d->hdl, &d->blk);
+        int r = _wfindnexti64(d->hdl, &d->blk);
         if (r < 0) {
             if (errno == ENOENT) errno = 0;
             return 0;
         }
     }
-    strcpy(de.d_name, d->blk.name);
+    if (!WideCharToMultiByte(CP_UTF8, 0, d->blk.name, -1, de.d_name, sizeof(de.d_name), NULL, NULL)) {
+        set_win32_errno(GetLastError());
+        return 0;
+    }
     de.d_size = d->blk.size;
     de.d_atime = d->blk.time_access;
     de.d_ctime = d->blk.time_create;
@@ -581,7 +685,7 @@ struct dirent * readdir(DIR *d) {
     return &de;
 }
 
-int closedir(DIR * d) {
+int utf8_closedir(DIR * d) {
     int r = 0;
     if (!d) {
         errno = EBADF;
@@ -592,7 +696,7 @@ int closedir(DIR * d) {
     return r;
 }
 
-#endif /* defined(_MSC_VER) */
+#endif /* defined(WIN32) && !defined(__CYGWIN__) */
 
 #if defined(WIN32)
 
@@ -804,16 +908,23 @@ void ini_mdep(void) {
 
 #if defined(WIN32)
 
-char * canonicalize_file_name(const char * path) {
-    char buf[MAX_PATH];
-    char * basename;
+char * canonicalize_file_name(const char * name) {
+    DWORD len;
     int i = 0;
-    DWORD len = GetFullPathName(path, sizeof(buf), buf, &basename);
+    wchar_t buf[FILE_PATH_SIZE];
+    wchar_t * basename;
+    wchar_t path[FILE_PATH_SIZE];
+    char res[FILE_PATH_SIZE];
+    if (!MultiByteToWideChar(CP_UTF8, 0, name, -1, path, sizeof(path))) {
+        set_win32_errno(GetLastError());
+        return NULL;
+    }
+    len = GetFullPathNameW(path, sizeof(buf), buf, &basename);
     if (len == 0) {
         errno = ENOENT;
         return NULL;
     }
-    if (len > MAX_PATH - 1) {
+    if (len > FILE_PATH_SIZE - 1) {
         errno = ENAMETOOLONG;
         return NULL;
     }
@@ -821,7 +932,11 @@ char * canonicalize_file_name(const char * path) {
         if (buf[i] == '\\') buf[i] = '/';
         i++;
     }
-    return strdup(buf);
+    if (!WideCharToMultiByte(CP_UTF8, 0, buf, -1, res, sizeof(res), NULL, NULL)) {
+        set_win32_errno(GetLastError());
+        return NULL;
+    }
+    return strdup(res);
 }
 
 #elif defined(_WRS_KERNEL)
diff --git a/mdep.h b/mdep.h
index bfd00022..68c65e09 100644
--- a/mdep.h
+++ b/mdep.h
@@ -126,9 +126,6 @@ extern void usleep(useconds_t useconds);
 
 #define off_t __int64
 #define lseek _lseeki64
-#define stat  _stati64
-#define lstat _stati64
-#define fstat _fstati64
 extern int truncate(const char * path, int64_t size);
 extern int ftruncate(int f, int64_t size);
 #define utimbuf _utimbuf
@@ -146,24 +143,48 @@ extern int getegid(void);
 extern ssize_t pread(int fd, const void * buf, size_t size, off_t offset);
 extern ssize_t pwrite(int fd, const void * buf, size_t size, off_t offset);
 
-#endif /* __CYGWIN__ */
-
-#define MSG_MORE 0
-
-extern const char * inet_ntop(int af, const void * src, char * dst, socklen_t size);
-extern int inet_pton(int af, const char * src, void * dst);
+/* UTF-8 support */
+struct utf8_stat {
+    dev_t      st_dev;
+    ino_t      st_ino;
+    unsigned short st_mode;
+    short      st_nlink;
+    short      st_uid;
+    short      st_gid;
+    dev_t      st_rdev;
+    int64_t    st_size;
+    int64_t    st_atime;
+    int64_t    st_mtime;
+    int64_t    st_ctime;
+};
+#define stat   utf8_stat
+#define lstat  utf8_stat
+#define fstat  utf8_fstat
+#define open   utf8_open
+#define chmod  utf8_chmod
+#define remove utf8_remove
+#define rmdir  utf8_rmdir
+#define mkdir  utf8_mkdir
+#define rename utf8_rename
+extern int utf8_stat(const char * name, struct utf8_stat * buf);
+extern int utf8_fstat(int fd, struct utf8_stat * buf);
+extern int utf8_open(const char * name, int flags, int perms);
+extern int utf8_chmod(const char * name, int mode);
+extern int utf8_remove(const char * path);
+extern int utf8_rmdir(const char * path);
+extern int utf8_mkdir(const char * path, int mode);
+extern int utf8_rename(const char * path1, const char * path2);
 
-#if defined(_MSC_VER)
 /*
- * readdir() emulation
+ * readdir() emulation with UTF-8 support
  */
-struct DIR {
+struct UTF8_DIR {
   long hdl;
-  struct _finddatai64_t blk;
+  struct _wfinddatai64_t blk;
   char path[FILE_PATH_SIZE];
 };
 
-struct dirent {
+struct utf8_dirent {
   char d_name[FILE_PATH_SIZE];
   int64_t d_size;
   time_t d_atime;
@@ -171,13 +192,24 @@ struct dirent {
   time_t d_wtime;
 };
 
-typedef struct DIR DIR;
+typedef struct UTF8_DIR UTF8_DIR;
 
-extern DIR * opendir(const char * path);
-extern int closedir(DIR * dir);
-extern struct dirent * readdir(DIR * dir);
+#define DIR UTF8_DIR
+#define dirent   utf8_dirent
+#define opendir  utf8_opendir
+#define closedir utf8_closedir
+#define readdir  utf8_readdir
 
-#endif /* _MSC_VER */
+extern DIR * utf8_opendir(const char * path);
+extern int utf8_closedir(DIR * dir);
+extern struct utf8_dirent * readdir(DIR * dir);
+
+#endif /* __CYGWIN__ */
+
+#define MSG_MORE 0
+
+extern const char * inet_ntop(int af, const void * src, char * dst, socklen_t size);
+extern int inet_pton(int af, const char * src, void * dst);
 
 /*
  * PThreads emulation.
@@ -235,7 +267,7 @@ extern int wsa_sendto(int socket, const void * buf, size_t size, int flags,
 
 #ifndef SHUT_RDWR
 #define SHUT_RDWR SD_BOTH
-#endif SHUT_RDWR
+#endif
 
 extern char * canonicalize_file_name(const char * path);