Fix unicode errors on Windows in path_is_dir, path_exists, getcwd and rust_localtime.

This make these functions use wchar_t version of APIs, instead of char version.

Author: nitric1

src/libstd/libc.rs

@@ -1208,7 +1208,7 @@ pub mod consts {
             pub static ERROR_ALREADY_EXISTS : c_int = 183;
             pub static ERROR_INVALID_ADDRESS : c_int = 487;
             pub static ERROR_FILE_INVALID : c_int = 1006;
-            pub static INVALID_HANDLE_VALUE: c_int = -1;
+            pub static INVALID_HANDLE_VALUE : c_int = -1;
 
             pub static DELETE : DWORD = 0x00010000;
             pub static READ_CONTROL : DWORD = 0x00020000;
@@ -3352,11 +3352,14 @@ pub mod funcs {
                                         LPSECURITY_ATTRIBUTES)
                                         -> BOOL;
                 pub fn CopyFileW(lpExistingFileName: LPCWSTR,
-                                        lpNewFileName: LPCWSTR,
-                                        bFailIfExists: BOOL)
-                                        -> BOOL;
+                                 lpNewFileName: LPCWSTR,
+                                 bFailIfExists: BOOL)
+                                 -> BOOL;
                 pub fn DeleteFileW(lpPathName: LPCWSTR) -> BOOL;
                 pub fn RemoveDirectoryW(lpPathName: LPCWSTR) -> BOOL;
+                pub fn GetCurrentDirectoryW(nBufferLength: DWORD,
+                                            lpBuffer: LPWSTR)
+                                            -> DWORD;
                 pub fn SetCurrentDirectoryW(lpPathName: LPCWSTR) -> BOOL;
                 pub fn GetLastError() -> DWORD;
                 pub fn FindFirstFileW(fileName: *u16, findFileData: HANDLE)
@@ -3462,6 +3465,9 @@ pub mod funcs {
                                         -> BOOL;
                 pub fn DeleteFileW(lpPathName: LPCWSTR) -> BOOL;
                 pub fn RemoveDirectoryW(lpPathName: LPCWSTR) -> BOOL;
+                pub fn GetCurrentDirectoryW(nBufferLength: DWORD,
+                                            lpBuffer: LPWSTR)
+                                            -> DWORD;
                 pub fn SetCurrentDirectoryW(lpPathName: LPCWSTR) -> BOOL;
                 pub fn GetLastError() -> DWORD;
                 pub fn FindFirstFileW(fileName: *u16, findFileData: HANDLE)

src/libstd/os.rs

@@ -28,7 +28,8 @@
 
 #[allow(missing_doc)];
 
-use c_str::{CString, ToCStr};
+#[cfg(unix)]
+use c_str::CString;
 use clone::Clone;
 use container::Container;
 use io;
@@ -56,6 +57,11 @@ pub fn close(fd: c_int) -> c_int {
     }
 }
 
+// On Windows, wide character version of function must be used to support
+// unicode, so functions should be split into at least two versions,
+// which are for Windows and for non-Windows, if necessary.
+// See https://github.com/mozilla/rust/issues/9822 for more information.
+
 pub mod rustrt {
     use libc::{c_char, c_int};
     use libc;
@@ -64,11 +70,19 @@ pub mod rustrt {
         pub fn rust_path_is_dir(path: *libc::c_char) -> c_int;
         pub fn rust_path_exists(path: *libc::c_char) -> c_int;
     }
+
+    // Uses _wstat instead of stat.
+    #[cfg(windows)]
+    extern {
+        pub fn rust_path_is_dir_u16(path: *u16) -> c_int;
+        pub fn rust_path_exists_u16(path: *u16) -> c_int;
+    }
 }
 
 pub static TMPBUF_SZ : uint = 1000u;
 static BUF_BYTES : uint = 2048u;
 
+#[cfg(unix)]
 pub fn getcwd() -> Path {
     #[fixed_stack_segment]; #[inline(never)];
     let mut buf = [0 as libc::c_char, ..BUF_BYTES];
@@ -83,6 +97,22 @@ pub fn getcwd() -> Path {
     }
 }
 
+#[cfg(windows)]
+pub fn getcwd() -> Path {
+    #[fixed_stack_segment]; #[inline(never)];
+    use libc::DWORD;
+    use libc::GetCurrentDirectoryW;
+    let mut buf = [0 as u16, ..BUF_BYTES];
+    do buf.as_mut_buf |buf, len| {
+        unsafe {
+            if libc::GetCurrentDirectoryW(len as DWORD, buf) == 0 as DWORD {
+                fail2!();
+            }
+        }
+    }
+    Path::new(str::from_utf16(buf))
+}
+
 #[cfg(windows)]
 pub mod win32 {
     use libc;
@@ -613,6 +643,7 @@ pub fn walk_dir(p: &Path, f: &fn(&Path) -> bool) -> bool {
     })
 }
 
+#[cfg(unix)]
 /// Indicates whether a path represents a directory
 pub fn path_is_dir(p: &Path) -> bool {
     #[fixed_stack_segment]; #[inline(never)];
@@ -623,6 +654,18 @@ pub fn path_is_dir(p: &Path) -> bool {
     }
 }
 
+
+#[cfg(windows)]
+pub fn path_is_dir(p: &Path) -> bool {
+    #[fixed_stack_segment]; #[inline(never)];
+    unsafe {
+        do os::win32::as_utf16_p(p.as_str().unwrap()) |buf| {
+            rustrt::rust_path_is_dir_u16(buf) != 0 as c_int
+        }
+    }
+}
+
+#[cfg(unix)]
 /// Indicates whether a path exists
 pub fn path_exists(p: &Path) -> bool {
     #[fixed_stack_segment]; #[inline(never)];
@@ -633,6 +676,16 @@ pub fn path_exists(p: &Path) -> bool {
     }
 }
 
+#[cfg(windows)]
+pub fn path_exists(p: &Path) -> bool {
+    #[fixed_stack_segment]; #[inline(never)];
+    unsafe {
+        do os::win32::as_utf16_p(p.as_str().unwrap()) |buf| {
+            rustrt::rust_path_exists_u16(buf) != 0 as c_int
+        }
+    }
+}
+
 /**
  * Convert a relative path to an absolute path
  *
@@ -1922,15 +1975,47 @@ mod tests {
 
     #[test]
     fn path_is_dir() {
+        use rt::io::file::open;
+        use rt::io::{OpenOrCreate, Read};
+
         assert!((os::path_is_dir(&Path::new("."))));
         assert!((!os::path_is_dir(&Path::new("test/stdtest/fs.rs"))));
+
+        let mut dirpath = os::tmpdir();
+        dirpath.push(format!("rust-test-{}/test-\uac00\u4e00\u30fc\u4f60\u597d",
+            rand::random::<u32>())); // 가一ー你好
+        debug2!("path_is_dir dirpath: {}", dirpath.display());
+
+        let mkdir_result = os::mkdir_recursive(&dirpath, (S_IRUSR | S_IWUSR | S_IXUSR) as i32);
+        debug2!("path_is_dir mkdir_result: {}", mkdir_result);
+
+        assert!((os::path_is_dir(&dirpath)));
+
+        let mut filepath = dirpath;
+        filepath.push("unicode-file-\uac00\u4e00\u30fc\u4f60\u597d.rs");
+        debug2!("path_is_dir filepath: {}", filepath.display());
+
+        open(&filepath, OpenOrCreate, Read); // ignore return; touch only
+        assert!((!os::path_is_dir(&filepath)));
+
+        assert!((!os::path_is_dir(&Path::new(
+                     "test/unicode-bogus-dir-\uac00\u4e00\u30fc\u4f60\u597d"))));
     }
 
     #[test]
     fn path_exists() {
         assert!((os::path_exists(&Path::new("."))));
         assert!((!os::path_exists(&Path::new(
                      "test/nonexistent-bogus-path"))));
+
+        let mut dirpath = os::tmpdir();
+        dirpath.push(format!("rust-test-{}/test-\uac01\u4e01\u30fc\u518d\u89c1",
+            rand::random::<u32>())); // 각丁ー再见
+
+        os::mkdir_recursive(&dirpath, (S_IRUSR | S_IWUSR | S_IXUSR) as i32);
+        assert!((os::path_exists(&dirpath)));
+        assert!((!os::path_exists(&Path::new(
+                     "test/unicode-bogus-path-\uac01\u4e01\u30fc\u518d\u89c1"))));
     }
 
     #[test]

src/rt/rust_builtin.cpp

@@ -107,7 +107,7 @@ rust_list_dir_wfd_fp_buf(void* wfd) {
 #endif
 
 extern "C" CDECL int
-rust_path_is_dir(char *path) {
+rust_path_is_dir(const char *path) {
     struct stat buf;
     if (stat(path, &buf)) {
         return 0;
@@ -116,14 +116,48 @@ rust_path_is_dir(char *path) {
 }
 
 extern "C" CDECL int
-rust_path_exists(char *path) {
+#if defined(__WIN32__)
+rust_path_is_dir_u16(const wchar_t *path) {
+    struct _stat buf;
+    // Don't use GetFileAttributesW, it cannot get attributes of
+    // some system files (e.g. pagefile.sys).
+    if (_wstat(path, &buf)) {
+        return 0;
+    }
+    return S_ISDIR(buf.st_mode);
+}
+#else
+rust_path_is_dir_u16(const void *path) {
+    // Wide version of function is only used on Windows.
+    return 0;
+}
+#endif
+
+extern "C" CDECL int
+rust_path_exists(const char *path) {
     struct stat buf;
     if (stat(path, &buf)) {
         return 0;
     }
     return 1;
 }
 
+extern "C" CDECL int
+#if defined(__WIN32__)
+rust_path_exists_u16(const wchar_t *path) {
+    struct _stat buf;
+    if (_wstat(path, &buf)) {
+        return 0;
+    }
+    return 1;
+}
+#else
+rust_path_exists_u16(const void *path) {
+    // Wide version of function is only used on Windows.
+    return 0;
+}
+#endif
+
 extern "C" CDECL FILE* rust_get_stdin() {return stdin;}
 extern "C" CDECL FILE* rust_get_stdout() {return stdout;}
 extern "C" CDECL FILE* rust_get_stderr() {return stderr;}
@@ -294,8 +328,12 @@ rust_localtime(int64_t sec, int32_t nsec, rust_tm *timeptr) {
     const char* zone = NULL;
 #if defined(__WIN32__)
     int32_t gmtoff = -timezone;
-    char buffer[64];
-    if (strftime(buffer, sizeof(buffer), "%Z", &tm) > 0) {
+    wchar_t wbuffer[64];
+    char buffer[256];
+    // strftime("%Z") can contain non-UTF-8 characters on non-English locale (issue #9418),
+    // so time zone should be converted from UTF-16 string set by wcsftime.
+    if (wcsftime(wbuffer, sizeof(wbuffer) / sizeof(wchar_t), L"%Z", &tm) > 0) {
+        WideCharToMultiByte(CP_UTF8, 0, wbuffer, -1, buffer, sizeof(buffer), NULL, NULL);
         zone = buffer;
     }
 #else

src/rt/rust_globals.h

@@ -54,6 +54,14 @@
 #include <assert.h>
 
 #if defined(__WIN32__)
+// Prevent unnecessary #include's from <windows.h>
+#ifndef WIN32_LEAN_AND_MEAN
+#define WIN32_LEAN_AND_MEAN
+#endif
+// Prevent defining min and max macro
+#ifndef NOMINMAX
+#define NOMINMAX
+#endif
 extern "C" {
 #include <windows.h>
 #include <tchar.h>

src/rt/rustrt.def.in

@@ -10,7 +10,9 @@ rust_timegm
 rust_mktime
 precise_time_ns
 rust_path_is_dir
+rust_path_is_dir_u16
 rust_path_exists
+rust_path_exists_u16
 rust_get_stdin
 rust_get_stdout
 rust_get_stderr