Support KANJI file name in (e)upTeX and XeTeX on Windows (from T. Tanaka)

git-svn-id: svn://tug.org/texlive/trunk@32308 c570f23f-e606-0410-a88d-b1316a301751

9 files changed, 187 insertions, 12 deletions

diff --git a/Build/source/texk/kpathsea/ChangeLog b/Build/source/texk/kpathsea/ChangeLog
index c6ec4f39ed1..9ffda1014e1 100644
--- a/Build/source/texk/kpathsea/ChangeLog
+++ b/Build/source/texk/kpathsea/ChangeLog
@@ -1,3 +1,8 @@
+2013-12-03  TANAKA Takuji <KXD02663@nifty.ne.jp>
+
+	* elt-dirs.c, knj.c, knj.h, progname.c, readable.c: Support KANJI
+	file name in (e)upTeX and XeTeX on Windows.
+
 2013-09-22  Karl Berry  <karl@tug.org>
 
 	* mktexlsr: output normal msgs to stdout, not stderr.
diff --git a/Build/source/texk/kpathsea/elt-dirs.c b/Build/source/texk/kpathsea/elt-dirs.c
index 7d61d850086..69544946928 100644
--- a/Build/source/texk/kpathsea/elt-dirs.c
+++ b/Build/source/texk/kpathsea/elt-dirs.c
@@ -104,7 +104,9 @@ static void expand_elt (kpathsea, str_llist_type *, string, unsigned);
 
 #ifdef WIN32
 /* Shared across recursive calls, it acts like a stack. */
-static char dirname[MAX_PATH];
+static char dirname[MAX_PATH*2];
+static wchar_t dirnamew[MAX_PATH];
+static char *potname;
 #endif
 
 static void
@@ -112,7 +114,7 @@ do_subdir (kpathsea kpse, str_llist_type *str_list_ptr, string elt,
               unsigned elt_length, string post)
 {
 #ifdef WIN32
-  WIN32_FIND_DATA find_file_data;
+  WIN32_FIND_DATAW find_file_data;
   HANDLE hnd;
   int proceed;
   int nlinks = 2;
@@ -131,7 +133,8 @@ do_subdir (kpathsea kpse, str_llist_type *str_list_ptr, string elt,
 #if defined (WIN32)
   strcpy(dirname, FN_STRING(name));
   strcat(dirname, "/*.*");         /* "*.*" or "*" -- seems equivalent. */
-  hnd = FindFirstFile(dirname, &find_file_data);
+  get_wstring_from_fsyscp(dirname, dirnamew);
+  hnd = FindFirstFileW(dirnamew, &find_file_data);
 
   if (hnd == INVALID_HANDLE_VALUE) {
     fn_free(&name);
@@ -151,11 +154,13 @@ do_subdir (kpathsea kpse, str_llist_type *str_list_ptr, string elt,
   }
   proceed = 1;
   while (proceed) {
-    if (find_file_data.cFileName[0] != '.') {
+    if (find_file_data.cFileName[0] != L'.') {
       int links;
 
       /* Construct the potential subdirectory name.  */
-      fn_str_grow (&name, find_file_data.cFileName);
+      potname = get_fsyscp_from_wstring(find_file_data.cFileName, potname=NULL);
+      fn_str_grow (&name, potname);
+      free(potname);
 
       /* Maybe we have cached the leafness of this directory.
                  The function will return 0 if unknown,
@@ -197,7 +202,7 @@ do_subdir (kpathsea kpse, str_llist_type *str_list_ptr, string elt,
       }
       fn_shrink_to (&name, elt_length);
     }
-    proceed = FindNextFile (hnd, &find_file_data);
+    proceed = FindNextFileW (hnd, &find_file_data);
   }
   /* Update the leafness of name. */
   kpathsea_dir_links(kpse, FN_STRING(name), nlinks);
diff --git a/Build/source/texk/kpathsea/knj.c b/Build/source/texk/kpathsea/knj.c
index bdf7d5cbf75..eeb1120e913 100644
--- a/Build/source/texk/kpathsea/knj.c
+++ b/Build/source/texk/kpathsea/knj.c
@@ -1,6 +1,7 @@
 /* knj.c: check for 2-Byte Kanji (CP 932, SJIS) codes.
 
    Copyright 2010, 2011 Akira Kakuto.
+   Copyright 2013 TANAKA Takuji.
 
    This library is free software; you can redistribute it and/or
    modify it under the terms of the GNU Lesser General Public
@@ -16,8 +17,9 @@
    along with this library; if not, see <http://www.gnu.org/licenses/>.  */
 
 #include <kpathsea/config.h>
+#include <kpathsea/debug.h>
 
-int is_cp932_system;
+int is_cp932_system, file_system_codepage;
 
 int isknj(int c)
 {
@@ -48,3 +50,80 @@ int isknj2(int c)
     return(0);
   }
 }
+
+/*
+  Get wide string from multibyte string.
+*/
+wchar_t *
+get_wstring_from_mbstring(int cp, const char *mbstr, wchar_t *wstr)
+{
+  int len;
+
+  len = MultiByteToWideChar(cp, 0, mbstr, -1, wstr, 0);
+  if (len==0) {
+    FATAL("cannot convert string to wide string");
+  }
+  if (wstr==NULL) {
+    wstr = xmalloc(sizeof(wchar_t)*(len+1));
+  }
+  len = MultiByteToWideChar(cp, 0, mbstr, -1, wstr, len+1);
+  if (len==0) {
+    FATAL("cannot convert multibyte string to wide string");
+  }
+  return wstr;
+}
+
+/*
+  Get multibyte string from wide string.
+*/
+char *
+get_mbstring_from_wstring(int cp, const wchar_t *wstr, char *mbstr)
+{
+  int len;
+
+  len = WideCharToMultiByte(cp, 0, wstr, -1, mbstr, 0, NULL, NULL);
+  if (len==0) {
+    FATAL("cannot convert string to wide string");
+  }
+  if (mbstr==NULL) {
+    mbstr = xmalloc(len+1);
+  }
+  len = WideCharToMultiByte(cp, 0, wstr, -1, mbstr, len+1, NULL, NULL);
+  if (len==0) {
+    FATAL("cannot convert wide string to multibyte string");
+  }
+  return mbstr;
+}
+
+/*
+  xfopen by file system codepage
+*/
+FILE *
+fsyscp_xfopen (const char *filename, const char *mode)
+{
+    FILE *f;
+    wchar_t *fnamew, modew[4];
+#if defined (KPSE_COMPAT_API)
+    kpathsea kpse;
+#endif
+    assert(filename && mode);
+
+    fnamew = get_wstring_from_fsyscp(filename, fnamew=NULL);
+    get_wstring_from_fsyscp(mode, modew);
+    f = _wfopen(fnamew, modew);
+    if (f == NULL)
+        FATAL_PERROR(filename);
+#if defined (KPSE_COMPAT_API)
+    kpse = kpse_def;
+    if (KPATHSEA_DEBUG_P (KPSE_DEBUG_FOPEN)) {
+        DEBUGF_START ();
+        fprintf (stderr, "fsyscp_xfopen(%s [", filename);
+        WriteConsoleW( GetStdHandle( STD_ERROR_HANDLE ), fnamew, wcslen( fnamew ), NULL, NULL );
+        fprintf (stderr, "], %s) => 0x%lx\n", mode, (unsigned long) f);
+        DEBUGF_END ();
+    }
+#endif
+    free(fnamew);
+
+    return f;
+}
diff --git a/Build/source/texk/kpathsea/knj.h b/Build/source/texk/kpathsea/knj.h
index 8f1cd919687..0629213a309 100644
--- a/Build/source/texk/kpathsea/knj.h
+++ b/Build/source/texk/kpathsea/knj.h
@@ -1,6 +1,7 @@
 /* knj.h: check for 2-Byte Kanji (CP 932, SJIS) codes.
 
    Copyright 2010, 2011 Akira Kakuto.
+   Copyright 2013 TANAKA Takuji.
 
    This library is free software; you can redistribute it and/or
    modify it under the terms of the GNU Lesser General Public
@@ -23,10 +24,15 @@ extern "C" {
 #endif
 
 extern KPSEDLL int is_cp932_system;
+extern KPSEDLL int file_system_codepage;
 
 extern KPSEDLL int isknj(int c);
 extern KPSEDLL int isknj2(int c);
 
+extern KPSEDLL wchar_t* get_wstring_from_mbstring(int cp, const char *mbstr, wchar_t *wstr);
+extern KPSEDLL char* get_mbstring_from_wstring(int cp, const wchar_t *wstr, char *mbstr);
+extern KPSEDLL FILE* fsyscp_xfopen(const char *filename, const char *mode);
+
 #ifdef __cplusplus
 }
 #endif
@@ -34,4 +40,13 @@ extern KPSEDLL int isknj2(int c);
 /* True if P points to a 2-Byte Kanji (CP 932, SJIS) code.  */
 #define IS_KANJI(p) is_cp932_system && isknj(*(p)) && isknj2(*(p+1))
 
+/* Get wide string from multibyte string in UTF-8 */
+#define get_wstring_from_utf8(mb,w) get_wstring_from_mbstring(CP_UTF8,mb,w)
+/* Get multibyte string in UTF-8 from wide string */
+#define get_utf8_from_wstring(w,mb) get_mbstring_from_wstring(CP_UTF8,w,mb)
+/* Get wide string from multibyte string in file system codepage */
+#define get_wstring_from_fsyscp(mb,w) get_wstring_from_mbstring(file_system_codepage,mb,w)
+/* Get multibyte string in file system codepage from wide string */
+#define get_fsyscp_from_wstring(w,mb) get_mbstring_from_wstring(file_system_codepage,w,mb)
+
 #endif /* not KPATHSEA_KNJ_H */
diff --git a/Build/source/texk/kpathsea/progname.c b/Build/source/texk/kpathsea/progname.c
index ba43bd8679f..5952936579b 100644
--- a/Build/source/texk/kpathsea/progname.c
+++ b/Build/source/texk/kpathsea/progname.c
@@ -501,12 +501,13 @@ kpathsea_set_program_name (kpathsea kpse,  const_string argv0,
   }
 
 #if defined(WIN32)
-  cp = GetOEMCP();
+  cp = GetACP();
   if (cp == 932 || cp == 936 || cp == 950) {
     is_cp932_system = cp;
   }
   else
     is_cp932_system = 0;
+  file_system_codepage = AreFileApisANSI() ? GetACP() : GetOEMCP();
 
 #if defined(__MINGW32__)
   /* Set various info about user. Among many things,
diff --git a/Build/source/texk/kpathsea/readable.c b/Build/source/texk/kpathsea/readable.c
index d62957e5ba0..3e604cd9c05 100644
--- a/Build/source/texk/kpathsea/readable.c
+++ b/Build/source/texk/kpathsea/readable.c
@@ -40,7 +40,9 @@
 static boolean
 READABLE(const_string fn, unsigned int st)
 {
-  if ((st = GetFileAttributes(fn)) != 0xFFFFFFFF) {
+  wchar_t *fnw;
+  fnw = get_wstring_from_fsyscp(fn, fnw=NULL);
+  if ((st = GetFileAttributesW(fnw)) != 0xFFFFFFFF) {
       /* succeeded */
       errno = 0;
   } else {
@@ -56,6 +58,7 @@ READABLE(const_string fn, unsigned int st)
           break;
       }
   }
+  if(fnw) free(fnw);
   return ((st != 0xFFFFFFFF) &&
                   !(st & FILE_ATTRIBUTE_DIRECTORY));
 }
diff --git a/Build/source/texk/web2c/lib/ChangeLog b/Build/source/texk/web2c/lib/ChangeLog
index 3d19d8073b4..c81beaad816 100644
--- a/Build/source/texk/web2c/lib/ChangeLog
+++ b/Build/source/texk/web2c/lib/ChangeLog
@@ -1,3 +1,8 @@
+2013-12-03  TANAKA Takuji <KXD02663@nifty.ne.jp>
+
+	* openclose.c, texmfmp.c: Support KANJI file name in (e)upTeX and
+	XeTeX on Windows.
+
 2013-06-11  Peter Breitenlohner  <peb@mppmu.mpg.de>
 
 	Update to e-pTeX and e-upTeX 130605 from Hironori Kitagawa.
diff --git a/Build/source/texk/web2c/lib/openclose.c b/Build/source/texk/web2c/lib/openclose.c
index 2620b8894bd..35301b5187c 100644
--- a/Build/source/texk/web2c/lib/openclose.c
+++ b/Build/source/texk/web2c/lib/openclose.c
@@ -57,7 +57,11 @@ recorder_start(void)
       recorder_name = temp;
     }
     
+#if defined(WIN32)
+    recorder_file = fsyscp_xfopen(recorder_name, FOPEN_W_MODE);
+#else
     recorder_file = xfopen(recorder_name, FOPEN_W_MODE);
+#endif
     
     cwd = xgetcwd();
     fprintf(recorder_file, "PWD %s\n", cwd);
@@ -101,7 +105,7 @@ recorder_change_filename (string new_name)
 
    /* reopen the recorder file by FOPEN_A_MODE. */
 #if defined(WIN32)
-   recorder_file = fopen (recorder_name, FOPEN_A_MODE);
+   recorder_file = fsyscp_xfopen (recorder_name, FOPEN_A_MODE);
 #endif
 
    if (temp)
@@ -165,7 +169,11 @@ open_input (FILE **f_ptr, int filefmt, const_string fopen_mode)
        them from there.  We only look for the name as-is.  */
     if (output_directory && !kpse_absolute_p (nameoffile+1, false)) {
         fname = concat3 (output_directory, DIR_SEP_STRING, nameoffile + 1);
+#if defined(WIN32)
+        *f_ptr = fsyscp_xfopen (fname, fopen_mode);
+#else
         *f_ptr = fopen (fname, fopen_mode);
+#endif
         if (*f_ptr) {
             free (nameoffile);
             namelength = strlen (fname);
@@ -182,7 +190,11 @@ open_input (FILE **f_ptr, int filefmt, const_string fopen_mode)
         /* A negative FILEFMT means don't use a path.  */
         if (filefmt < 0) {
             /* no_file_path, for BibTeX .aux files and MetaPost things.  */
+#if defined(WIN32)
+            *f_ptr = fsyscp_xfopen(nameoffile + 1, fopen_mode);
+#else
             *f_ptr = fopen(nameoffile + 1, fopen_mode);
+#endif
             /* FIXME... fullnameoffile = xstrdup(nameoffile + 1); */
         } else {
             /* The only exception to `must_exist' being true is \openin, for
@@ -227,7 +239,11 @@ open_input (FILE **f_ptr, int filefmt, const_string fopen_mode)
                     *f_ptr = nkf_open (nameoffile + 1, fopen_mode);
                 } else
 #endif
+#if defined(WIN32)
+                *f_ptr = fsyscp_xfopen (nameoffile + 1, fopen_mode);
+#else
                 *f_ptr = xfopen (nameoffile + 1, fopen_mode);
+#endif
             }
         }
     }
@@ -275,7 +291,11 @@ open_output (FILE **f_ptr, const_string fopen_mode)
     }
 
     /* Is the filename openable as given?  */
+#if defined(WIN32)
+    *f_ptr = fsyscp_xfopen (fname, fopen_mode);
+#else
     *f_ptr = fopen (fname, fopen_mode);
+#endif
 
     if (!*f_ptr) {
         /* Can't open as given.  Try the envvar.  */
diff --git a/Build/source/texk/web2c/lib/texmfmp.c b/Build/source/texk/web2c/lib/texmfmp.c
index 515d68b0b50..548012365e2 100644
--- a/Build/source/texk/web2c/lib/texmfmp.c
+++ b/Build/source/texk/web2c/lib/texmfmp.c
@@ -628,7 +628,9 @@ void
 maininit (int ac, string *av)
 {
   string main_input_file;
-
+#if (IS_upTeX || defined(XeTeX)) && defined(WIN32)
+  string enc;
+#endif
   /* Save to pass along to topenin.  */
   argc = ac;
   argv = av;
@@ -655,11 +657,51 @@ maininit (int ac, string *av)
   kpse_set_program_name (argv[0], NULL);
   initkanji ();
 #endif
+#if defined(XeTeX) && defined(WIN32)
+  kpse_set_program_name (argv[0], NULL);
+#endif
+#if (IS_upTeX || defined(XeTeX)) && defined(WIN32)
+  enc = kpse_var_value("command_line_encoding");
+  if (enc) {
+#ifdef DEBUG
+    fprintf(stderr, "command_line_encoding (%s)\n", enc);
+#endif /* DEBUG */
+    if (!(strncmp(enc,"utf8",5) && strncmp(enc,"utf-8",6))) {
+      DWORD ret;
+      LPWSTR *argvw;
+      INT argcw, i;
+      string s;
+#ifdef DEBUG
+      HANDLE hStderr;
+      hStderr = GetStdHandle( STD_ERROR_HANDLE );
+#endif /* DEBUG */
+      file_system_codepage = CP_UTF8;
+      is_cp932_system = 0;
+      argvw = CommandLineToArgvW(GetCommandLineW(), &argcw);
+      argc = argcw;
+      argv = xmalloc(sizeof(char *)*(argcw+1));
+      for (i=0; i<argcw; i++) {
+        s = get_utf8_from_wstring(argvw[i], s=NULL);
+        argv[i] = s;
+#ifdef DEBUG
+        fprintf(stderr, "Commandline arguments %d:(%s) [", i, argv[i]);
+        WriteConsoleW( hStderr, argvw[i], wcslen(argvw[i]), &ret, NULL);
+        fprintf(stderr, "]\n");
+#endif /* DEBUG */
+      }
+      argv[argcw] = NULL;
+    }
+  }
+#endif
 
   /* If the user says --help or --version, we need to notice early.  And
      since we want the --ini option, have to do it before getting into
      the web (which would read the base file, etc.).  */
+#if (IS_upTeX || defined(XeTeX)) && defined(WIN32)
+  parse_options (argc, argv);
+#else
   parse_options (ac, av);
+#endif
 
 #if IS_pTeX
   /* In pTeX and friends, texmf.cnf is not recorded in the case of --recorder,
@@ -679,7 +721,7 @@ maininit (int ac, string *av)
   /* Do this early so we can inspect kpse_invocation_name and
      kpse_program_name below, and because we have to do this before
      any path searching.  */
-#if IS_pTeX
+#if IS_pTeX || (defined(XeTeX) && defined(WIN32))
   if (user_progname)
     kpse_reset_program_name (user_progname);
 #else