Improve require/File.expand_path performance on Windows

* configure.in (mingw): add shlwapi to the list of dependency
  libs for Windows.
* win32/Makefile.sub (EXTSOLIBS): ditto.

* internal.h: declare internal functions rb_w32_init_file,
  rb_file_expand_path_internal and rb_file_expand_path_fast.

* file.c (Init_File): invoke Windows initialization rb_w32_init_file

* win32/file.c (rb_file_load_path_internal): new function.
  Windows-specific implementation that replaces file_expand_path.
  [Bug #6836][ruby-core:46996]

* win32/file.c (rb_w32_init_file): new function. Initialize codepage
  cache for faster conversion encodings lookup.

* file.c (file_expand_path): rename to rb_file_expand_path_internal.
  Conditionally exclude from Windows.

* file.c (rb_file_expand_path_fast): new function. delegates to
  rb_file_expand_path_internal without performing a hit to the
  filesystem.

* file.c (file_expand_path_1): use rb_file_expand_path_internal without
  path expansion (used by require).
* file.c (rb_find_file_ext_safe): ditto.
* file.c (rb_find_file_safe): ditto.

* load.c (rb_get_expanded_load_path): use rb_file_expand_path_fast.
* load.c (rb_feature_provided): ditto.

* file.c (rb_file_expand_path): use rb_file_expand_path_internal with
  path expansion.
* file.c (rb_file_absolute_path): ditto.

* test/ruby/test_file_exhaustive.rb: new tests to exercise
  rb_file_expand_path_internal implementation and compliance with
  existing behaviors.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@36811 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

9 files changed, 909 insertions, 13 deletions

diff --git a/ChangeLog b/ChangeLog
index b710091d25..a9987cc284 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,44 @@
+Fri Aug 24 12:40:15 2012  Luis Lavena  <luislavena@gmail.com>
+
+	* configure.in (mingw): add shlwapi to the list of dependency
+	  libs for Windows.
+	* win32/Makefile.sub (EXTSOLIBS): ditto.
+
+	* internal.h: declare internal functions rb_w32_init_file,
+	  rb_file_expand_path_internal and rb_file_expand_path_fast.
+
+	* file.c (Init_File): invoke Windows initialization rb_w32_init_file
+
+	* win32/file.c (rb_file_load_path_internal): new function.
+	  Windows-specific implementation that replaces file_expand_path.
+	  [Bug #6836][ruby-core:46996]
+
+	* win32/file.c (rb_w32_init_file): new function. Initialize codepage
+	  cache for faster conversion encodings lookup.
+
+	* file.c (file_expand_path): rename to rb_file_expand_path_internal.
+	  Conditionally exclude from Windows.
+
+	* file.c (rb_file_expand_path_fast): new function. delegates to
+	  rb_file_expand_path_internal without performing a hit to the
+	  filesystem.
+
+	* file.c (file_expand_path_1): use rb_file_expand_path_internal without
+	  path expansion (used by require).
+	* file.c (rb_find_file_ext_safe): ditto.
+	* file.c (rb_find_file_safe): ditto.
+
+	* load.c (rb_get_expanded_load_path): use rb_file_expand_path_fast.
+	* load.c (rb_feature_provided): ditto.
+
+	* file.c (rb_file_expand_path): use rb_file_expand_path_internal with
+	  path expansion.
+	* file.c (rb_file_absolute_path): ditto.
+
+	* test/ruby/test_file_exhaustive.rb: new tests to exercise
+	  rb_file_expand_path_internal implementation and compliance with
+	  existing behaviors.
+
 Fri Aug 24 07:35:24 2012  Eric Hodel  <drbrain@segment7.net>
 
 	* lib/net/http/backward.rb (class Net):  Restored Net::HTTPSession to
diff --git a/configure.in b/configure.in
index 4a915456f0..b490c45219 100644
--- a/configure.in
+++ b/configure.in
@@ -1186,7 +1186,7 @@ main()
 		AC_CHECK_FUNCS(cygwin_conv_path)
 		AC_LIBOBJ([langinfo])
 		],
-[mingw*], [	LIBS="-lshell32 -lws2_32 -limagehlp $LIBS"
+[mingw*], [	LIBS="-lshell32 -lws2_32 -limagehlp -lshlwapi $LIBS"
 		ac_cv_header_a_out_h=no
 		ac_cv_header_pwd_h=no
 		ac_cv_header_utime_h=no
diff --git a/file.c b/file.c
index 24aa1115fe..8c63f56d01 100644
--- a/file.c
+++ b/file.c
@@ -2882,8 +2882,9 @@ append_fspath(VALUE result, VALUE fname, char *dir, rb_encoding **enc, rb_encodi
     return buf + dirlen;
 }
 
-static VALUE
-file_expand_path(VALUE fname, VALUE dname, int abs_mode, VALUE result)
+#ifndef _WIN32
+VALUE
+rb_file_expand_path_internal(VALUE fname, VALUE dname, int abs_mode, int long_name, VALUE result)
 {
     const char *s, *b, *fend;
     char *buf, *p, *pend, *root;
@@ -2945,7 +2946,7 @@ file_expand_path(VALUE fname, VALUE dname, int abs_mode, VALUE result)
 	    /* specified drive, but not full path */
 	    int same = 0;
 	    if (!NIL_P(dname) && !not_same_drive(dname, s[0])) {
-		file_expand_path(dname, Qnil, abs_mode, result);
+		rb_file_expand_path_internal(dname, Qnil, abs_mode, long_name, result);
 		BUFINIT();
 		if (has_drive_letter(p) && TOLOWER(p[0]) == TOLOWER(s[0])) {
 		    /* ok, same drive */
@@ -2969,7 +2970,7 @@ file_expand_path(VALUE fname, VALUE dname, int abs_mode, VALUE result)
 #endif
     else if (!rb_is_absolute_path(s)) {
 	if (!NIL_P(dname)) {
-	    file_expand_path(dname, Qnil, abs_mode, result);
+	    rb_file_expand_path_internal(dname, Qnil, abs_mode, long_name, result);
 	    rb_enc_associate(result, rb_enc_check(result, fname));
 	    BUFINIT();
 	    p = pend;
@@ -3222,6 +3223,7 @@ file_expand_path(VALUE fname, VALUE dname, int abs_mode, VALUE result)
     ENC_CODERANGE_CLEAR(result);
     return result;
 }
+#endif /* _WIN32 */
 
 #define EXPAND_PATH_BUFFER() rb_usascii_str_new(0, MAXPATHLEN + 2)
 
@@ -3232,14 +3234,21 @@ file_expand_path(VALUE fname, VALUE dname, int abs_mode, VALUE result)
 static VALUE
 file_expand_path_1(VALUE fname)
 {
-    return file_expand_path(fname, Qnil, 0, EXPAND_PATH_BUFFER());
+    return rb_file_expand_path_internal(fname, Qnil, 0, 0, EXPAND_PATH_BUFFER());
 }
 
 VALUE
 rb_file_expand_path(VALUE fname, VALUE dname)
 {
     check_expand_path_args(fname, dname);
-    return file_expand_path(fname, dname, 0, EXPAND_PATH_BUFFER());
+    return rb_file_expand_path_internal(fname, dname, 0, 1, EXPAND_PATH_BUFFER());
+}
+
+VALUE
+rb_file_expand_path_fast(VALUE fname, VALUE dname)
+{
+    check_expand_path_args(fname, dname);
+    return rb_file_expand_path_internal(fname, dname, 0, 0, EXPAND_PATH_BUFFER());
 }
 
 /*
@@ -3276,7 +3285,7 @@ VALUE
 rb_file_absolute_path(VALUE fname, VALUE dname)
 {
     check_expand_path_args(fname, dname);
-    return file_expand_path(fname, dname, 1, EXPAND_PATH_BUFFER());
+    return rb_file_expand_path_internal(fname, dname, 1, 1, EXPAND_PATH_BUFFER());
 }
 
 /*
@@ -5251,7 +5260,7 @@ rb_find_file_ext_safe(VALUE *filep, const char *const *ext, int safe_level)
 
 	    RB_GC_GUARD(str) = rb_get_path_check(str, safe_level);
 	    if (RSTRING_LEN(str) == 0) continue;
-	    file_expand_path(fname, str, 0, tmp);
+	    rb_file_expand_path_internal(fname, str, 0, 0, tmp);
 	    if (rb_file_load_ok(RSTRING_PTR(tmp))) {
 		*filep = copy_path_class(tmp, *filep);
 		return (int)(j+1);
@@ -5311,7 +5320,7 @@ rb_find_file_safe(VALUE path, int safe_level)
 	    VALUE str = RARRAY_PTR(load_path)[i];
 	    RB_GC_GUARD(str) = rb_get_path_check(str, safe_level);
 	    if (RSTRING_LEN(str) > 0) {
-		file_expand_path(path, str, 0, tmp);
+		rb_file_expand_path_internal(path, str, 0, 0, tmp);
 		f = RSTRING_PTR(tmp);
 		if (rb_file_load_ok(f)) goto found;
 	    }
@@ -5546,4 +5555,8 @@ Init_File(void)
     rb_define_method(rb_cStat, "setuid?",  rb_stat_suid, 0);
     rb_define_method(rb_cStat, "setgid?",  rb_stat_sgid, 0);
     rb_define_method(rb_cStat, "sticky?",  rb_stat_sticky, 0);
+
+#ifdef _WIN32
+    rb_w32_init_file();
+#endif
 }
diff --git a/internal.h b/internal.h
index c091a1c49e..ed4864c1eb 100644
--- a/internal.h
+++ b/internal.h
@@ -105,8 +105,15 @@ VALUE rb_home_dir(const char *user, VALUE result);
 VALUE rb_realpath_internal(VALUE basedir, VALUE path, int strict);
 void rb_file_const(const char*, VALUE);
 int rb_file_load_ok(const char *);
+VALUE rb_file_expand_path_fast(VALUE, VALUE);
+VALUE rb_file_expand_path_internal(VALUE, VALUE, int, int, VALUE);
 void Init_File(void);
 
+#ifdef _WIN32
+/* file.c, win32/file.c */
+void rb_w32_init_file(void);
+#endif
+
 /* gc.c */
 void Init_heap(void);
 void *ruby_mimmalloc(size_t size);
diff --git a/load.c b/load.c
index 7f1b3f65b8..b0e6f16e4f 100644
--- a/load.c
+++ b/load.c
@@ -43,7 +43,7 @@ rb_get_expanded_load_path(void)
 
     ary = rb_ary_new2(RARRAY_LEN(load_path));
     for (i = 0; i < RARRAY_LEN(load_path); ++i) {
-	VALUE path = rb_file_expand_path(RARRAY_PTR(load_path)[i], Qnil);
+	VALUE path = rb_file_expand_path_fast(RARRAY_PTR(load_path)[i], Qnil);
 	rb_str_freeze(path);
 	rb_ary_push(ary, path);
     }
@@ -233,7 +233,7 @@ rb_feature_provided(const char *feature, const char **loading)
 
     if (*feature == '.' &&
 	(feature[1] == '/' || strncmp(feature+1, "./", 2) == 0)) {
-	fullpath = rb_file_expand_path(rb_str_new2(feature), Qnil);
+	fullpath = rb_file_expand_path_fast(rb_str_new2(feature), Qnil);
 	feature = RSTRING_PTR(fullpath);
     }
     if (ext && !strchr(ext, '/')) {
diff --git a/test/ruby/test_file_exhaustive.rb b/test/ruby/test_file_exhaustive.rb
index 7467bffdfe..1523823bb7 100644
--- a/test/ruby/test_file_exhaustive.rb
+++ b/test/ruby/test_file_exhaustive.rb
@@ -14,6 +14,7 @@ class TestFileExhaustive < Test::Unit::TestCase
 
   def setup
     @dir = Dir.mktmpdir("rubytest-file")
+    @rootdir = "#{DRIVE}/"
     File.chown(-1, Process.gid, @dir)
     @file = make_tmp_filename("file")
     @zerofile = make_tmp_filename("zerofile")
@@ -450,9 +451,24 @@ class TestFileExhaustive < Test::Unit::TestCase
       assert_equal(expected.force_encoding(cp), File.expand_path(a.dup.force_encoding(cp)), cp)
     end
 
+    path = "\u3042\u3044\u3046\u3048\u304a".encode("EUC-JP")
+    assert_equal("#{Dir.pwd}/#{path}".encode("CP932"), File.expand_path(path).encode("CP932"))
+
+    path = "\u3042\u3044\u3046\u3048\u304a".encode("CP51932")
+    assert_equal("#{Dir.pwd}/#{path}", File.expand_path(path))
+
     assert_incompatible_encoding {|d| File.expand_path(d)}
   end
 
+  def test_expand_path_encoding_filesystem
+    path = "~".encode("US-ASCII")
+    dir = "C:/".encode("IBM437")
+    fs = Encoding.find("filesystem")
+
+    assert_equal fs, File.expand_path(path).encoding
+    assert_equal fs, File.expand_path(path, dir).encoding
+  end
+
   def test_expand_path_home
     assert_kind_of(String, File.expand_path("~")) if ENV["HOME"]
     assert_raise(ArgumentError) { File.expand_path("~foo_bar_baz_unknown_user_wahaha") }
@@ -460,7 +476,13 @@ class TestFileExhaustive < Test::Unit::TestCase
     begin
       bug3630 = '[ruby-core:31537]'
       home = ENV["HOME"]
+      home_drive = ENV["HOMEDRIVE"]
+      home_path = ENV["HOMEPATH"]
+      user_profile = ENV["USERPROFILE"]
       ENV["HOME"] = nil
+      ENV["HOMEDRIVE"] = nil
+      ENV["HOMEPATH"] = nil
+      ENV["USERPROFILE"] = nil
       assert_raise(ArgumentError) { File.expand_path("~") }
       ENV["HOME"] = "~"
       assert_raise(ArgumentError, bug3630) { File.expand_path("~") }
@@ -468,9 +490,193 @@ class TestFileExhaustive < Test::Unit::TestCase
       assert_raise(ArgumentError, bug3630) { File.expand_path("~") }
     ensure
       ENV["HOME"] = home
+      ENV["HOMEDRIVE"] = home_drive
+      ENV["HOMEPATH"] = home_path
+      ENV["USERPROFILE"] = user_profile
+    end
+  end
+
+  def test_expand_path_remove_trailing_alternative_data
+    assert_equal File.join(@rootdir, "aaa"), File.expand_path("#{@rootdir}/aaa::$DATA")
+    assert_equal File.join(@rootdir, "aa:a"), File.expand_path("#{@rootdir}/aa:a:$DATA")
+    assert_equal File.join(@rootdir, "aaa:$DATA"), File.expand_path("#{@rootdir}/aaa:$DATA")
+  end if DRIVE
+
+  def test_expand_path_resolve_empty_string_current_directory
+    assert_equal(Dir.pwd, File.expand_path(""))
+  end
+
+  def test_expand_path_resolve_dot_current_directory
+    assert_equal(Dir.pwd, File.expand_path("."))
+  end
+
+  def test_expand_path_resolve_file_name_relative_current_directory
+    assert_equal(File.join(Dir.pwd, "foo"), File.expand_path("foo"))
+  end
+
+  def test_ignore_nil_dir_string
+    assert_equal(File.join(Dir.pwd, "foo"), File.expand_path("foo", nil))
+  end
+
+  def test_expand_path_resolve_file_name_and_dir_string_relative
+    assert_equal(File.join(Dir.pwd, "bar", "foo"),
+      File.expand_path("foo", "bar"))
+  end
+
+  def test_expand_path_cleanup_dots_file_name
+    bug = "[ruby-talk:18512]"
+
+    assert_equal(File.join(Dir.pwd, ".a"), File.expand_path(".a"), bug)
+    assert_equal(File.join(Dir.pwd, "..a"), File.expand_path("..a"), bug)
+
+    if DRIVE
+      # cleanup dots only on Windows
+      assert_equal(File.join(Dir.pwd, "a"), File.expand_path("a."), bug)
+      skip "FIXME"
+      assert_equal(File.join(Dir.pwd, "a"), File.expand_path("a.."), bug)
+    else
+      assert_equal(File.join(Dir.pwd, "a."), File.expand_path("a."), bug)
+      assert_equal(File.join(Dir.pwd, "a.."), File.expand_path("a.."), bug)
+    end
+  end
+
+  def test_expand_path_converts_a_pathname_to_an_absolute_pathname_using_a_complete_path
+    assert_equal(@dir, File.expand_path("", "#{@dir}"))
+    assert_equal(File.join(@dir, "a"), File.expand_path("a", "#{@dir}"))
+    assert_equal(File.join(@dir, "a"), File.expand_path("../a", "#{@dir}/xxx"))
+    assert_equal(@rootdir, File.expand_path(".", "#{@rootdir}"))
+  end
+
+  def test_expand_path_ignores_supplied_dir_if_path_contains_a_drive_letter
+    assert_equal(@rootdir, File.expand_path(@rootdir, "D:/"))
+  end if DRIVE
+
+  def test_expand_path_removes_trailing_slashes_from_absolute_path
+    assert_equal(File.join(@rootdir, "foo"), File.expand_path("#{@rootdir}foo/"))
+    assert_equal(File.join(@rootdir, "foo.rb"), File.expand_path("#{@rootdir}foo.rb/"))
+  end
+
+  def test_expand_path_removes_trailing_spaces_from_absolute_path
+    assert_equal(File.join(@rootdir, "a"), File.expand_path("#{@rootdir}a "))
+  end if DRIVE
+
+  def test_expand_path_converts_a_pathname_which_starts_with_a_slash_using_dir_s_drive
+    assert_match(%r"\Az:/foo\z"i, File.expand_path('/foo', "z:/bar"))
+  end if DRIVE
+
+  def test_expand_path_converts_a_pathname_which_starts_with_a_slash_and_unc_pathname
+    assert_equal("//foo", File.expand_path('//foo', "//bar"))
+    assert_equal("//bar/foo", File.expand_path('/foo', "//bar"))
+    assert_equal("//foo", File.expand_path('//foo', "/bar"))
+  end if DRIVE
+
+  def test_expand_path_converts_a_dot_with_unc_dir
+    assert_equal("//", File.expand_path('.', "//"))
+  end
+
+  def test_expand_path_preserves_unc_path_root
+    assert_equal("//", File.expand_path("//"))
+    assert_equal("//", File.expand_path("//."))
+    assert_equal("//", File.expand_path("//.."))
+  end
+
+  def test_expand_path_converts_a_pathname_which_starts_with_a_slash_using_host_share
+    assert_match(%r"\A//host/share/foo\z"i, File.expand_path('/foo', "//host/share/bar"))
+  end if DRIVE
+
+  def test_expand_path_converts_a_pathname_which_starts_with_a_slash_using_a_current_drive
+    assert_match(%r"\A#{DRIVE}/foo\z"i, File.expand_path('/foo'))
+  end
+
+  def test_expand_path_returns_tainted_strings_or_not
+    assert_equal(true, File.expand_path('foo').tainted?)
+    assert_equal(true, File.expand_path('foo'.taint).tainted?)
+    assert_equal(true, File.expand_path('/foo'.taint).tainted?)
+    assert_equal(true, File.expand_path('foo', 'bar').tainted?)
+    assert_equal(true, File.expand_path('foo', '/bar'.taint).tainted?)
+    assert_equal(true, File.expand_path('foo'.taint, '/bar').tainted?)
+    assert_equal(true, File.expand_path('~').tainted?)
+
+    if DRIVE
+      assert_equal(true, File.expand_path('/foo').tainted?)
+      assert_equal(false, File.expand_path('//foo').tainted?)
+      assert_equal(true, File.expand_path('C:/foo'.taint).tainted?)
+      assert_equal(false, File.expand_path('C:/foo').tainted?)
+      assert_equal(true, File.expand_path('foo', '/bar').tainted?)
+      assert_equal(true, File.expand_path('foo', 'C:/bar'.taint).tainted?)
+      assert_equal(true, File.expand_path('foo'.taint, 'C:/bar').tainted?)
+      assert_equal(false, File.expand_path('foo', 'C:/bar').tainted?)
+      assert_equal(false, File.expand_path('C:/foo/../bar').tainted?)
+      assert_equal(false, File.expand_path('foo', '//bar').tainted?)
+    else
+      assert_equal(false, File.expand_path('/foo').tainted?)
+      assert_equal(false, File.expand_path('foo', '/bar').tainted?)
     end
   end
 
+  def test_expand_path_converts_a_pathname_to_an_absolute_pathname_using_home_as_base
+    old_home = ENV["HOME"]
+    home = ENV["HOME"] = "#{DRIVE}/UserHome"
+    assert_equal(home, File.expand_path("~"))
+    assert_equal(home, File.expand_path("~", "C:/FooBar"))
+    assert_equal(File.join(home, "a"), File.expand_path("~/a", "C:/FooBar"))
+  ensure
+    ENV["HOME"] = old_home
+  end
+
+  def test_expand_path_converts_a_pathname_to_an_absolute_pathname_using_unc_home
+    old_home = ENV["HOME"]
+    unc_home = ENV["HOME"] = "//UserHome"
+    assert_equal(unc_home, File.expand_path("~"))
+  ensure
+    ENV["HOME"] = old_home
+  end if DRIVE
+
+  def test_expand_path_does_not_modify_a_home_string_argument
+    old_home = ENV["HOME"]
+    home = ENV["HOME"] = "#{DRIVE}/UserHome"
+    str = "~/a"
+    assert_equal("#{home}/a", File.expand_path(str))
+    assert_equal("~/a", str)
+  ensure
+    ENV["HOME"] = old_home
+  end
+
+  def test_expand_path_raises_argument_error_for_any_supplied_username
+    bug = '[ruby-core:39597]'
+    assert_raise(ArgumentError, bug) { File.expand_path("~anything") }
+  end if DRIVE
+
+  def test_expand_path_raises_a_type_error_if_not_passed_a_string_type
+    assert_raise(TypeError) { File.expand_path(1) }
+    assert_raise(TypeError) { File.expand_path(nil) }
+    assert_raise(TypeError) { File.expand_path(true) }
+  end
+
+  def test_expand_path_expands_dot_dir
+    assert_equal("#{DRIVE}/dir", File.expand_path("#{DRIVE}/./dir"))
+  end
+
+  def test_expand_path_does_not_modify_the_string_argument
+    str = "./a/b/../c"
+    assert_equal("#{Dir.pwd}/a/c", File.expand_path(str, Dir.pwd))
+    assert_equal("./a/b/../c", str)
+  end
+
+  def test_expand_path_returns_a_string_when_passed_a_string_subclass
+    sub = Class.new(String)
+    str = sub.new "./a/b/../c"
+    path = File.expand_path(str, Dir.pwd)
+    assert_equal("#{Dir.pwd}/a/c", path)
+    assert_instance_of(String, path)
+  end
+
+  def test_expand_path_accepts_objects_that_have_a_to_path_method
+    klass = Class.new { def to_path; "a/b/c"; end }
+    obj = klass.new
+    assert_equal("#{Dir.pwd}/a/b/c", File.expand_path(obj))
+  end
+
   def test_basename
     assert_equal(File.basename(@file).sub(/\.test$/, ""), File.basename(@file, ".test"))
     assert_equal("", s = File.basename(""))
diff --git a/win32/.time b/win32/.time
new file mode 100644
index 0000000000..e69de29bb2
--- /dev/null
+++ b/win32/.time
diff --git a/win32/Makefile.sub b/win32/Makefile.sub
index 13e6471f4b..fe33455ecf 100644
--- a/win32/Makefile.sub
+++ b/win32/Makefile.sub
@@ -226,7 +226,7 @@ EXTLIBS =
 EXTSOLIBS =
 !endif
 !if !defined(LIBS)
-LIBS = oldnames.lib user32.lib advapi32.lib shell32.lib ws2_32.lib imagehlp.lib $(EXTLIBS)
+LIBS = oldnames.lib user32.lib advapi32.lib shell32.lib ws2_32.lib imagehlp.lib shlwapi.lib $(EXTLIBS)
 !endif
 !if !defined(MISSING)
 MISSING = acosh.obj cbrt.obj crypt.obj erf.obj ffs.obj langinfo.obj lgamma_r.obj strlcat.obj strlcpy.obj tgamma.obj win32/win32.obj win32/file.obj setproctitle.obj
diff --git a/win32/file.c b/win32/file.c
index 955f91bc67..c8bb445c82 100644
--- a/win32/file.c
+++ b/win32/file.c
@@ -1,10 +1,630 @@
 #include "ruby/ruby.h"
+#include "ruby/encoding.h"
 #include <winbase.h>
+#include <wchar.h>
+#include <shlwapi.h>
 
 #ifndef INVALID_FILE_ATTRIBUTES
 # define INVALID_FILE_ATTRIBUTES ((DWORD)-1)
 #endif
 
+/* cache 'encoding name' => 'code page' into a hash */
+static VALUE rb_code_page;
+
+#define IS_DIR_SEPARATOR_P(c) (c == L'\\' || c == L'/')
+#define IS_DIR_UNC_P(c) (IS_DIR_SEPARATOR_P(c[0]) && IS_DIR_SEPARATOR_P(c[1]))
+
+/* MultiByteToWideChar() doesn't work with code page 51932 */
+#define INVALID_CODE_PAGE 51932
+#define PATH_BUFFER_SIZE MAX_PATH * 2
+
+#define insecure_obj_p(obj, level) ((level) >= 4 || ((level) > 0 && OBJ_TAINTED(obj)))
+
+static inline void
+replace_wchar(wchar_t *s, int find, int replace)
+{
+    while (*s != 0) {
+	if (*s == find)
+	    *s = replace;
+	s++;
+    }
+}
+
+/* Convert str from multibyte char to wchar with specified code page */
+static inline void
+convert_mb_to_wchar(VALUE str, wchar_t **wstr, wchar_t **wstr_pos, size_t *wstr_len, UINT code_page)
+{
+    size_t len;
+
+    if (NIL_P(str))
+	return;
+
+    len = MultiByteToWideChar(code_page, 0, RSTRING_PTR(str), -1, NULL, 0) + 1;
+    *wstr = (wchar_t *)xmalloc(len * sizeof(wchar_t));
+    if (wstr_pos)
+	*wstr_pos = *wstr;
+
+    MultiByteToWideChar(code_page, 0, RSTRING_PTR(str), -1, *wstr, len);
+    *wstr_len = len - 2;
+}
+
+static inline void
+convert_wchar_to_mb(const wchar_t *wstr, char **str, size_t *str_len, UINT code_page)
+{
+    size_t len;
+
+    len = WideCharToMultiByte(code_page, 0, wstr, -1, NULL, 0, NULL, NULL);
+    *str = (char *)xmalloc(len * sizeof(char));
+    WideCharToMultiByte(code_page, 0, wstr, -1, *str, len, NULL, NULL);
+
+    /* do not count terminator as part of the string length */
+    *str_len = len - 1;
+}
+
+/*
+  Return user's home directory using environment variables combinations.
+  Memory allocated by this function should be manually freeded afterwards.
+
+  Try:
+  HOME, HOMEDRIVE + HOMEPATH and USERPROFILE environment variables
+  TODO: Special Folders - Profile and Personal
+*/
+static wchar_t *
+home_dir()
+{
+    wchar_t *buffer = NULL;
+    size_t buffer_len = 0, len = 0;
+    size_t home_env = 0;
+
+    /*
+      GetEnvironmentVariableW when used with NULL will return the required
+      buffer size and its terminating character.
+      http://msdn.microsoft.com/en-us/library/windows/desktop/ms683188(v=vs.85).aspx
+    */
+
+    if (len = GetEnvironmentVariableW(L"HOME", NULL, 0)) {
+	buffer_len = len;
+	home_env = 1;
+    }
+    else if (len = GetEnvironmentVariableW(L"HOMEDRIVE", NULL, 0)) {
+	buffer_len = len;
+	if (len = GetEnvironmentVariableW(L"HOMEPATH", NULL, 0)) {
+	    buffer_len += len;
+	    home_env = 2;
+	}
+	else {
+	    buffer_len = 0;
+	}
+    }
+    else if (len = GetEnvironmentVariableW(L"USERPROFILE", NULL, 0)) {
+	buffer_len = len;
+	home_env = 3;
+    }
+
+    /* allocate buffer */
+    if (home_env)
+	buffer = (wchar_t *)xmalloc(buffer_len * sizeof(wchar_t));
+
+    switch (home_env) {
+	case 1:
+	    /* HOME */
+	    GetEnvironmentVariableW(L"HOME", buffer, buffer_len);
+	    break;
+	case 2:
+	    /* HOMEDRIVE + HOMEPATH */
+	    len = GetEnvironmentVariableW(L"HOMEDRIVE", buffer, buffer_len);
+	    GetEnvironmentVariableW(L"HOMEPATH", buffer + len, buffer_len - len);
+	    break;
+	case 3:
+	    /* USERPROFILE */
+	    GetEnvironmentVariableW(L"USERPROFILE", buffer, buffer_len);
+	    break;
+	default:
+	    break;
+	}
+
+    if (home_env) {
+	/* sanitize backslashes with forwardslashes */
+	replace_wchar(buffer, L'\\', L'/');
+
+	return buffer;
+    }
+
+    return NULL;
+}
+
+/* Remove trailing invalid ':$DATA' of the path. */
+static inline size_t
+remove_invalid_alternative_data(wchar_t *wfullpath, size_t size) {
+    static const wchar_t prime[] = L":$DATA";
+    enum { prime_len = (sizeof(prime) / sizeof(wchar_t)) -1 };
+
+    if (size <= prime_len || _wcsnicmp(wfullpath + size - prime_len, prime, prime_len) != 0)
+	return size;
+
+    /* alias of stream */
+    /* get rid of a bug of x64 VC++ */
+    if (wfullpath[size - (prime_len + 1)] == ':') {
+	/* remove trailing '::$DATA' */
+	size -= prime_len + 1; /* prime */
+	wfullpath[size] = L'\0';
+    }
+    else {
+	/* remove trailing ':$DATA' of paths like '/aa:a:$DATA' */
+	wchar_t *pos = wfullpath + size - (prime_len + 1);
+	while (!IS_DIR_SEPARATOR_P(*pos) && pos != wfullpath) {
+	    if (*pos == L':') {
+		size -= prime_len; /* alternative */
+		wfullpath[size] = L'\0';
+		break;
+	    }
+	    pos--;
+	}
+    }
+    return size;
+}
+
+/* Return system code page. */
+static inline UINT
+system_code_page() {
+    return AreFileApisANSI() ? CP_ACP : CP_OEMCP;
+}
+
+/*
+  Return code page number of the encoding.
+  Cache code page into a hash for performance since finding the code page in
+  Encoding#names is slow.
+*/
+static UINT
+code_page(rb_encoding *enc)
+{
+    VALUE code_page_value, name_key;
+    VALUE encoding, names_ary = Qundef, name;
+    char *enc_name;
+    struct RString fake_str;
+    ID names;
+    long i;
+
+    if (!enc)
+	return system_code_page();
+
+    enc_name = (char *)rb_enc_name(enc);
+
+    fake_str.basic.flags = T_STRING|RSTRING_NOEMBED;
+    fake_str.basic.klass = rb_cString;
+    fake_str.as.heap.len = strlen(enc_name);
+    fake_str.as.heap.ptr = enc_name;
+    fake_str.as.heap.aux.capa = fake_str.as.heap.len;
+    name_key = (VALUE)&fake_str;
+    ENCODING_CODERANGE_SET(name_key, rb_usascii_encindex(), ENC_CODERANGE_7BIT);
+
+    code_page_value = rb_hash_lookup(rb_code_page, name_key);
+    if (code_page_value != Qnil)
+	return (UINT)FIX2INT(code_page_value);
+
+    name_key = rb_usascii_str_new2(enc_name);
+
+    encoding = rb_enc_from_encoding(enc);
+    if (!NIL_P(encoding)) {
+	CONST_ID(names, "names");
+	names_ary = rb_funcall(encoding, names, 0);
+    }
+
+    /* map US-ASCII and ASCII-8bit as code page 20127 (us-ascii) */
+    if (enc == rb_usascii_encoding() || enc == rb_ascii8bit_encoding()) {
+	UINT code_page = 20127;
+	rb_hash_aset(rb_code_page, name_key, INT2FIX(code_page));
+	return code_page;
+    }
+
+    if (names_ary != Qundef) {
+	for (i = 0; i < RARRAY_LEN(names_ary); i++) {
+	    name = RARRAY_PTR(names_ary)[i];
+	    if (strncmp("CP", RSTRING_PTR(name), 2) == 0) {
+		int code_page = atoi(RSTRING_PTR(name) + 2);
+		if (code_page != 0) {
+		    rb_hash_aset(rb_code_page, name_key, INT2FIX(code_page));
+		    return (UINT)code_page;
+		}
+	    }
+	}
+    }
+
+    rb_hash_aset(rb_code_page, name_key, INT2FIX(INVALID_CODE_PAGE));
+    return INVALID_CODE_PAGE;
+}
+
+static inline VALUE
+fix_string_encoding(VALUE str, rb_encoding *encoding)
+{
+    VALUE result, tmp;
+
+    tmp = rb_enc_str_new(RSTRING_PTR(str), RSTRING_LEN(str), encoding);
+    result = rb_str_encode(tmp, rb_enc_from_encoding(rb_utf8_encoding()), 0, Qnil);
+
+    return result;
+}
+
+/*
+  Replace the last part of the path to long name.
+  We try to avoid to call FindFirstFileW() since it takes long time.
+*/
+static inline size_t
+replace_to_long_name(wchar_t **wfullpath, size_t size, int heap) {
+    WIN32_FIND_DATAW find_data;
+    HANDLE find_handle;
+
+    /*
+      Skip long name conversion if the path is already long name.
+      Short name is 8.3 format.
+      http://en.wikipedia.org/wiki/8.3_filename
+      This check can be skipped for directory components that have file
+      extensions longer than 3 characters, or total lengths longer than
+      12 characters.
+      http://msdn.microsoft.com/en-us/library/windows/desktop/aa364980(v=vs.85).aspx
+    */
+    size_t const max_short_name_size = 8 + 1 + 3;
+    size_t const max_extension_size = 3;
+    size_t path_len = 1, extension_len = 0;
+    wchar_t *pos = *wfullpath;
+
+    if (size == 3 && pos[1] == L':' && pos[2] == L'\\' && pos[3] == L'\0') {
+	/* root path doesn't need short name expansion */
+	return size;
+    }
+
+    pos = *wfullpath + size - 1;
+    while (!IS_DIR_SEPARATOR_P(*pos) && pos != *wfullpath) {
+	if (!extension_len && *pos == L'.') {
+	    extension_len = path_len - 1;
+	}
+	if (path_len > max_short_name_size || extension_len > max_extension_size) {
+	    return size;
+	}
+	path_len++;
+	pos--;
+    }
+
+    find_handle = FindFirstFileW(*wfullpath, &find_data);
+    if (find_handle != INVALID_HANDLE_VALUE) {
+	size_t trail_pos = wcslen(*wfullpath);
+	size_t file_len = wcslen(find_data.cFileName);
+
+	FindClose(find_handle);
+	while (trail_pos > 0) {
+	    if (IS_DIR_SEPARATOR_P((*wfullpath)[trail_pos]))
+		break;
+	    trail_pos--;
+	}
+	size = trail_pos + 1 + file_len;
+	if ((size + 1) > sizeof(*wfullpath) / sizeof((*wfullpath)[0])) {
+	    wchar_t *buf = (wchar_t *)xmalloc((size + 1) * sizeof(wchar_t));
+	    wcsncpy(buf, *wfullpath, trail_pos + 1);
+	    if (heap)
+		xfree(*wfullpath);
+	    *wfullpath = buf;
+	}
+	wcsncpy(*wfullpath + trail_pos + 1, find_data.cFileName, file_len + 1);
+    }
+    return size;
+}
+
+VALUE
+rb_file_expand_path_internal(VALUE fname, VALUE dname, int abs_mode, int long_name, VALUE result)
+{
+    size_t size = 0, wpath_len = 0, wdir_len = 0, whome_len = 0;
+    size_t buffer_len = 0;
+    char *fullpath = NULL;
+    wchar_t *wfullpath = NULL, *wpath = NULL, *wpath_pos = NULL, *wdir = NULL;
+    wchar_t *whome = NULL, *buffer = NULL, *buffer_pos = NULL;
+    UINT path_cp, cp;
+    VALUE path = fname, dir = dname;
+    wchar_t wfullpath_buffer[PATH_BUFFER_SIZE];
+    wchar_t path_drive = L'\0', dir_drive = L'\0';
+    int ignore_dir = 0;
+    rb_encoding *path_encoding;
+    int tainted = 0;
+
+    /* tainted if path is tainted */
+    tainted = OBJ_TAINTED(path);
+
+    /* get path encoding */
+    if (NIL_P(dir)) {
+	path_encoding = rb_enc_get(path);
+    }
+    else {
+	path_encoding = rb_enc_check(path, dir);
+    }
+
+    cp = path_cp = code_page(path_encoding);
+
+    /* workaround invalid codepage */
+    if (path_cp == INVALID_CODE_PAGE) {
+	cp = CP_UTF8;
+	if (!NIL_P(path)) {
+	    path = fix_string_encoding(path, path_encoding);
+	}
+    }
+
+    /* convert char * to wchar_t */
+    convert_mb_to_wchar(path, &wpath, &wpath_pos, &wpath_len, cp);
+
+    /* determine if we need the user's home directory */
+    /* expand '~' only if NOT rb_file_absolute_path() where `abs_mode` is 1 */
+    if (abs_mode == 0 && ((wpath_len == 1 && wpath_pos[0] == L'~') ||
+	(wpath_len >= 2 && wpath_pos[0] == L'~' && IS_DIR_SEPARATOR_P(wpath_pos[1])))) {
+	/* tainted if expanding '~' */
+	tainted = 1;
+
+	whome = home_dir();
+	if (whome == NULL) {
+	    xfree(wpath);
+	    rb_raise(rb_eArgError, "couldn't find HOME environment -- expanding `~'");
+	}
+	whome_len = wcslen(whome);
+
+	if (PathIsRelativeW(whome) && !(whome_len >= 2 && IS_DIR_UNC_P(whome))) {
+	    xfree(wpath);
+	    rb_raise(rb_eArgError, "non-absolute home");
+	}
+
+	/* use filesystem encoding if expanding home dir */
+	path_encoding = rb_filesystem_encoding();
+	cp = path_cp = system_code_page();
+
+	/* ignores dir since we are expading home */
+	ignore_dir = 1;
+
+	/* exclude ~ from the result */
+	wpath_pos++;
+	wpath_len--;
+
+	/* exclude separator if present */
+	if (wpath_len && IS_DIR_SEPARATOR_P(wpath_pos[0])) {
+	    wpath_pos++;
+	    wpath_len--;
+	}
+    }
+    else if (wpath_len >= 2 && wpath_pos[1] == L':') {
+	if (wpath_len >= 3 && IS_DIR_SEPARATOR_P(wpath_pos[2])) {
+	    /* ignore dir since path contains a drive letter and a root slash */
+	    ignore_dir = 1;
+	}
+	else {
+	    /* determine if we ignore dir or not later */
+	    path_drive = wpath_pos[0];
+	}
+    }
+    else if (abs_mode == 0 && wpath_len >= 2 && wpath_pos[0] == L'~') {
+	wchar_t *wuser = wpath_pos + 1;
+	wchar_t *pos = wuser;
+	char *user;
+
+	/* tainted if expanding '~' */
+	tainted = 1;
+
+	while (!IS_DIR_SEPARATOR_P(*pos) && *pos != '\0')
+	    pos++;
+
+	*pos = '\0';
+	convert_wchar_to_mb(wuser, &user, &size, cp);
+
+	/* convert to VALUE and set the path encoding */
+	if (path_cp == INVALID_CODE_PAGE) {
+	    VALUE tmp = rb_enc_str_new(user, size, rb_utf8_encoding());
+	    result = rb_str_encode(tmp, rb_enc_from_encoding(path_encoding), 0, Qnil);
+	    rb_str_resize(tmp, 0);
+	}
+	else {
+	    result = rb_enc_str_new(user, size, path_encoding);
+	}
+
+	xfree(wpath);
+	if (user)
+	    xfree(user);
+
+	rb_raise(rb_eArgError, "can't find user %s", StringValuePtr(result));
+    }
+
+    /* convert dir */
+    if (!ignore_dir && !NIL_P(dir)) {
+	/* fix string encoding */
+	if (path_cp == INVALID_CODE_PAGE) {
+	    dir = fix_string_encoding(dir, path_encoding);
+	}
+
+	/* convert char * to wchar_t */
+	convert_mb_to_wchar(dir, &wdir, NULL, &wdir_len, cp);
+
+	if (wdir_len >= 2 && wdir[1] == L':') {
+	    dir_drive = wdir[0];
+	    if (wpath_len && IS_DIR_SEPARATOR_P(wpath_pos[0])) {
+		wdir_len = 2;
+	    }
+	}
+	else if (wdir_len >= 2 && IS_DIR_UNC_P(wdir)) {
+	    /* UNC path */
+	    if (wpath_len && IS_DIR_SEPARATOR_P(wpath_pos[0])) {
+		/* cut the UNC path tail to '//host/share' */
+		size_t separators = 0;
+		size_t pos = 2;
+		while (pos < wdir_len && separators < 2) {
+		    if (IS_DIR_SEPARATOR_P(wdir[pos])) {
+			separators++;
+		    }
+		    pos++;
+		}
+		if (separators == 2)
+		    wdir_len = pos - 1;
+	    }
+	}
+    }
+
+    /* determine if we ignore dir or not */
+    if (!ignore_dir && path_drive && dir_drive) {
+	if (towupper(path_drive) == towupper(dir_drive)) {
+	    /* exclude path drive letter to use dir */
+	    wpath_pos += 2;
+	    wpath_len -= 2;
+	}
+	else {
+	    /* ignore dir since path drive is different from dir drive */
+	    ignore_dir = 1;
+	    wdir_len = 0;
+	}
+    }
+
+    if (!ignore_dir && wpath_len >= 2 && IS_DIR_UNC_P(wpath)) {
+	/* ignore dir since path has UNC root */
+	ignore_dir = 1;
+	wdir_len = 0;
+    }
+    else if (!ignore_dir && wpath_len >= 1 && IS_DIR_SEPARATOR_P(wpath[0]) &&
+	!dir_drive && !(wdir_len >= 2 && IS_DIR_UNC_P(wdir))) {
+	/* ignore dir since path has root slash and dir doesn't have drive or UNC root */
+	ignore_dir = 1;
+	wdir_len = 0;
+    }
+
+    buffer_len = wpath_len + 1 + wdir_len + 1 + whome_len + 1;
+
+    buffer = buffer_pos = (wchar_t *)xmalloc((buffer_len + 1) * sizeof(wchar_t));
+
+    /* add home */
+    if (whome_len) {
+	wcsncpy(buffer_pos, whome, whome_len);
+	buffer_pos += whome_len;
+    }
+
+    /* Add separator if required */
+    if (whome_len && wcsrchr(L"\\/:", buffer_pos[-1]) == NULL) {
+	buffer_pos[0] = L'\\';
+	buffer_pos++;
+    }
+
+    if (wdir_len) {
+	/* tainted if dir is used and dir is tainted */
+	if (!tainted && OBJ_TAINTED(dir))
+	    tainted = 1;
+
+	wcsncpy(buffer_pos, wdir, wdir_len);
+	buffer_pos += wdir_len;
+    }
+
+    /* add separator if required */
+    if (wdir_len && wcsrchr(L"\\/:", buffer_pos[-1]) == NULL) {
+	buffer_pos[0] = L'\\';
+	buffer_pos++;
+    }
+
+    /* now deal with path */
+    if (wpath_len) {
+	wcsncpy(buffer_pos, wpath_pos, wpath_len);
+	buffer_pos += wpath_len;
+    }
+
+    /* GetFullPathNameW requires at least "." to determine current directory */
+    if (wpath_len == 0) {
+	buffer_pos[0] = L'.';
+	buffer_pos++;
+    }
+
+    /* Ensure buffer is NULL terminated */
+    buffer_pos[0] = L'\0';
+
+    /* tainted if path is relative */
+    if (!tainted && PathIsRelativeW(buffer) && !(buffer_len >= 2 && IS_DIR_UNC_P(buffer)))
+	tainted = 1;
+
+    /* FIXME: Make this more robust */
+    /* Determine require buffer size */
+    size = GetFullPathNameW(buffer, PATH_BUFFER_SIZE, wfullpath_buffer, NULL);
+    if (size > PATH_BUFFER_SIZE) {
+	/* allocate more memory than alloted originally by PATH_BUFFER_SIZE */
+	wfullpath = (wchar_t *)xmalloc(size * sizeof(wchar_t));
+	size = GetFullPathNameW(buffer, size, wfullpath, NULL);
+    }
+    else {
+	wfullpath = wfullpath_buffer;
+    }
+
+    /* Remove any trailing slashes */
+    if (IS_DIR_SEPARATOR_P(wfullpath[size - 1]) &&
+	wfullpath[size - 2] != L':' &&
+	!(size == 2 && IS_DIR_UNC_P(wfullpath))) {
+	size -= 1;
+	wfullpath[size] = L'\0';
+    }
+
+    /* Remove any trailing dot */
+    if (wfullpath[size - 1] == L'.') {
+	size -= 1;
+	wfullpath[size] = L'\0';
+    }
+
+    /* removes trailing invalid ':$DATA' */
+    size = remove_invalid_alternative_data(wfullpath, size);
+
+    /* Replace the trailing path to long name */
+    if (long_name)
+	size = replace_to_long_name(&wfullpath, size, (wfullpath != wfullpath_buffer));
+
+    /* sanitize backslashes with forwardslashes */
+    replace_wchar(wfullpath, L'\\', L'/');
+
+    /* convert to char * */
+    size = WideCharToMultiByte(cp, 0, wfullpath, size, NULL, 0, NULL, NULL);
+    if (size > (size_t)RSTRING_LEN(result)) {
+	rb_str_modify(result);
+	rb_str_resize(result, size);
+    }
+
+    WideCharToMultiByte(cp, 0, wfullpath, size, RSTRING_PTR(result), size, NULL, NULL);
+    rb_str_set_len(result, size);
+
+    /* convert to VALUE and set the path encoding */
+    if (path_cp == INVALID_CODE_PAGE) {
+	VALUE tmp;
+	size_t len;
+
+	rb_enc_associate(result, rb_utf8_encoding());
+	ENC_CODERANGE_CLEAR(result);
+	tmp = rb_str_encode(result, rb_enc_from_encoding(path_encoding), 0, Qnil);
+	len = RSTRING_LEN(tmp);
+	rb_str_modify(result);
+	rb_str_resize(result, len);
+	memcpy(RSTRING_PTR(result), RSTRING_PTR(tmp), len);
+	rb_str_resize(tmp, 0);
+    }
+    rb_enc_associate(result, path_encoding);
+    ENC_CODERANGE_CLEAR(result);
+
+    /* makes the result object tainted if expanding tainted strings or returning modified path */
+    if (tainted)
+	OBJ_TAINT(result);
+
+    /* TODO: better cleanup */
+    if (buffer)
+	xfree(buffer);
+
+    if (wpath)
+	xfree(wpath);
+
+    if (wdir)
+	xfree(wdir);
+
+    if (whome)
+	xfree(whome);
+
+    if (wfullpath && wfullpath != wfullpath_buffer)
+	xfree(wfullpath);
+
+    if (fullpath)
+	xfree(fullpath);
+
+    return result;
+}
+
 int
 rb_file_load_ok(const char *path)
 {
@@ -27,3 +647,12 @@ rb_file_load_ok(const char *path)
     }
     return ret;
 }
+
+void
+rb_w32_init_file(void)
+{
+    rb_code_page = rb_hash_new();
+
+    /* prevent GC removing rb_code_page */
+    rb_gc_register_mark_object(rb_code_page);
+}