binutils/winduni.c

   1 /* winduni.c -- unicode support for the windres program.
   2    Copyright 1997, 1998, 2000, 2001, 2003, 2005, 2007, 2009
   3    Free Software Foundation, Inc.
   4    Written by Ian Lance Taylor, Cygnus Support.
   5    Rewritten by Kai Tietz, Onevision.
   6
   7    This file is part of GNU Binutils.
   8
   9    This program is free software; you can redistribute it and/or modify
  10    it under the terms of the GNU General Public License as published by
  11    the Free Software Foundation; either version 3 of the License, or
  12    (at your option) any later version.
  13
  14    This program is distributed in the hope that it will be useful,
  15    but WITHOUT ANY WARRANTY; without even the implied warranty of
  16    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17    GNU General Public License for more details.
  18
  19    You should have received a copy of the GNU General Public License
  20    along with this program; if not, write to the Free Software
  21    Foundation, Inc., 51 Franklin Street - Fifth Floor, Boston, MA
  22    02110-1301, USA.  */
  23
  24
  25 /* This file contains unicode support routines for the windres
  26    program.  Ideally, we would have generic unicode support which
  27    would work on all systems.  However, we don't.  Instead, on a
  28    Windows host, we are prepared to call some Windows routines.  This
  29    means that we will generate different output on Windows and Unix
  30    hosts, but that seems better than not really supporting unicode at
  31    all.  */
  32
  33 #include "sysdep.h"
  34 #include "bfd.h"
  35 #include "libiberty.h" /* for xstrdup */
  36 #include "bucomm.h"
  37 /* Must be include before windows.h and winnls.h.  */
  38 #if defined (_WIN32) || defined (__CYGWIN__)
  39 #include <windows.h>
  40 #include <winnls.h>
  41 #endif
  42 #include "winduni.h"
  43 #include "safe-ctype.h"
  44
  45 #if HAVE_ICONV
  46 #include <iconv.h>
  47 #endif
  48
  49 static rc_uint_type wind_WideCharToMultiByte (rc_uint_type, const unichar *, char *, rc_uint_type);
  50 static rc_uint_type wind_MultiByteToWideChar (rc_uint_type, const char *, unichar *, rc_uint_type);
  51 static int unichar_isascii (const unichar *, rc_uint_type);
  52
  53 /* Convert an ASCII string to a unicode string.  We just copy it,
  54    expanding chars to shorts, rather than doing something intelligent.  */
  55
  56 #if !defined (_WIN32) && !defined (__CYGWIN__)
  57
  58 /* Codepages mapped.  */
  59 static local_iconv_map codepages[] =
  60 {
  61   { 0, "MS-ANSI" },
  62   { 1, "WINDOWS-1252" },
  63   { 437, "MS-ANSI" },
  64   { 737, "MS-GREEK" },
  65   { 775, "WINBALTRIM" },
  66   { 850, "MS-ANSI" },
  67   { 852, "MS-EE" },
  68   { 857, "MS-TURK" },
  69   { 862, "CP862" },
  70   { 864, "CP864" },
  71   { 866, "MS-CYRL" },
  72   { 874, "WINDOWS-874" },
  73   { 932, "CP932" },
  74   { 936, "CP936" },
  75   { 949, "CP949" },
  76   { 950, "CP950" },
  77   { 1250, "WINDOWS-1250" },
  78   { 1251, "WINDOWS-1251" },
  79   { 1252, "WINDOWS-1252" },
  80   { 1253, "WINDOWS-1253" },
  81   { 1254, "WINDOWS-1254" },
  82   { 1255, "WINDOWS-1255" },
  83   { 1256, "WINDOWS-1256" },
  84   { 1257, "WINDOWS-1257" },
  85   { 1258, "WINDOWS-1258" },
  86   { CP_UTF7, "UTF-7" },
  87   { CP_UTF8, "UTF-8" },
  88   { CP_UTF16, "UTF-16" },
  89   { (rc_uint_type) -1, NULL }
  90 };
  91
  92 /* Languages supported.  */
  93 static const wind_language_t languages[] =
  94 {
  95   { 0x0000, 437, 1252, "Neutral", "Neutral" },
  96   { 0x0401, 864, 1256, "Arabic", "Saudi Arabia" },    { 0x0402, 866, 1251, "Bulgarian", "Bulgaria" },
  97   { 0x0403, 850, 1252, "Catalan", "Spain" },          { 0x0404, 950,  950, "Chinese", "Taiwan" },
  98   { 0x0405, 852, 1250, "Czech", "Czech Republic" },   { 0x0406, 850, 1252, "Danish", "Denmark" },
  99   { 0x0407, 850, 1252, "German", "Germany" },         { 0x0408, 737, 1253, "Greek", "Greece" },
 100   { 0x0409, 437, 1252, "English", "United States" },  { 0x040A, 850, 1252, "Spanish - Traditional Sort", "Spain" },
 101   { 0x040B, 850, 1252, "Finnish", "Finland" },        { 0x040C, 850, 1252, "French", "France" },
 102   { 0x040D, 862, 1255, "Hebrew", "Israel" },          { 0x040E, 852, 1250, "Hungarian", "Hungary" },
 103   { 0x040F, 850, 1252, "Icelandic", "Iceland" },      { 0x0410, 850, 1252, "Italian", "Italy" },
 104   { 0x0411, 932,  932, "Japanese", "Japan" },         { 0x0412, 949,  949, "Korean", "Korea (south)" },
 105   { 0x0413, 850, 1252, "Dutch", "Netherlands" },      { 0x0414, 850, 1252, "Norwegian (Bokmål)", "Norway" },
 106   { 0x0415, 852, 1250, "Polish", "Poland" },          { 0x0416, 850, 1252, "Portuguese", "Brazil" },
 107   { 0x0418, 852, 1250, "Romanian", "Romania" },       { 0x0419, 866, 1251, "Russian", "Russia" },
 108   { 0x041A, 852, 1250, "Croatian", "Croatia" },       { 0x041B, 852, 1250, "Slovak", "Slovakia" },
 109   { 0x041C, 852, 1250, "Albanian", "Albania" },       { 0x041D, 850, 1252, "Swedish", "Sweden" },
 110   { 0x041E, 874,  874, "Thai", "Thailand" },          { 0x041F, 857, 1254, "Turkish", "Turkey" },
 111   { 0x0421, 850, 1252, "Indonesian", "Indonesia" },   { 0x0422, 866, 1251, "Ukrainian", "Ukraine" },
 112   { 0x0423, 866, 1251, "Belarusian", "Belarus" },     { 0x0424, 852, 1250, "Slovene", "Slovenia" },
 113   { 0x0425, 775, 1257, "Estonian", "Estonia" },       { 0x0426, 775, 1257, "Latvian", "Latvia" },
 114   { 0x0427, 775, 1257, "Lithuanian", "Lithuania" },
 115   { 0x0429, 864, 1256, "Arabic", "Farsi" },           { 0x042A,1258, 1258, "Vietnamese", "Vietnam" },
 116   { 0x042D, 850, 1252, "Basque", "Spain" },
 117   { 0x042F, 866, 1251, "Macedonian", "Former Yugoslav Republic of Macedonia" },
 118   { 0x0436, 850, 1252, "Afrikaans", "South Africa" },
 119   { 0x0438, 850, 1252, "Faroese", "Faroe Islands" },
 120   { 0x043C, 437, 1252, "Irish", "Ireland" },
 121   { 0x043E, 850, 1252, "Malay", "Malaysia" },
 122   { 0x0801, 864, 1256, "Arabic", "Iraq" },
 123   { 0x0804, 936,  936, "Chinese (People's republic of China)", "People's republic of China" },
 124   { 0x0807, 850, 1252, "German", "Switzerland" },
 125   { 0x0809, 850, 1252, "English", "United Kingdom" }, { 0x080A, 850, 1252, "Spanish", "Mexico" },
 126   { 0x080C, 850, 1252, "French", "Belgium" },
 127   { 0x0810, 850, 1252, "Italian", "Switzerland" },
 128   { 0x0813, 850, 1252, "Dutch", "Belgium" },          { 0x0814, 850, 1252, "Norwegian (Nynorsk)", "Norway" },
 129   { 0x0816, 850, 1252, "Portuguese", "Portugal" },
 130   { 0x081A, 852, 1252, "Serbian (latin)", "Yugoslavia" },
 131   { 0x081D, 850, 1252, "Swedish (Finland)", "Finland" },
 132   { 0x0C01, 864, 1256, "Arabic", "Egypt" },
 133   { 0x0C04, 950,  950, "Chinese", "Hong Kong" },
 134   { 0x0C07, 850, 1252, "German", "Austria" },
 135   { 0x0C09, 850, 1252, "English", "Australia" },      { 0x0C0A, 850, 1252, "Spanish - International Sort", "Spain" },
 136   { 0x0C0C, 850, 1252, "French", "Canada"},
 137   { 0x0C1A, 855, 1251, "Serbian (Cyrillic)", "Serbia" },
 138   { 0x1001, 864, 1256, "Arabic", "Libya" },
 139   { 0x1004, 936,  936, "Chinese", "Singapore" },
 140   { 0x1007, 850, 1252, "German", "Luxembourg" },
 141   { 0x1009, 850, 1252, "English", "Canada" },
 142   { 0x100A, 850, 1252, "Spanish", "Guatemala" },
 143   { 0x100C, 850, 1252, "French", "Switzerland" },
 144   { 0x1401, 864, 1256, "Arabic", "Algeria" },
 145   { 0x1407, 850, 1252, "German", "Liechtenstein" },
 146   { 0x1409, 850, 1252, "English", "New Zealand" },    { 0x140A, 850, 1252, "Spanish", "Costa Rica" },
 147   { 0x140C, 850, 1252, "French", "Luxembourg" },
 148   { 0x1801, 864, 1256, "Arabic", "Morocco" },
 149   { 0x1809, 850, 1252, "English", "Ireland" },        { 0x180A, 850, 1252, "Spanish", "Panama" },
 150   { 0x180C, 850, 1252, "French", "Monaco" },
 151   { 0x1C01, 864, 1256, "Arabic", "Tunisia" },
 152   { 0x1C09, 437, 1252, "English", "South Africa" },   { 0x1C0A, 850, 1252, "Spanish", "Dominican Republic" },
 153   { 0x2001, 864, 1256, "Arabic", "Oman" },
 154   { 0x2009, 850, 1252, "English", "Jamaica" },        { 0x200A, 850, 1252, "Spanish", "Venezuela" },
 155   { 0x2401, 864, 1256, "Arabic", "Yemen" },
 156   { 0x2409, 850, 1252, "English", "Caribbean" },      { 0x240A, 850, 1252, "Spanish", "Colombia" },
 157   { 0x2801, 864, 1256, "Arabic", "Syria" },
 158   { 0x2809, 850, 1252, "English", "Belize" },         { 0x280A, 850, 1252, "Spanish", "Peru" },
 159   { 0x2C01, 864, 1256, "Arabic", "Jordan" },
 160   { 0x2C09, 437, 1252, "English", "Trinidad & Tobago" },{ 0x2C0A, 850, 1252, "Spanish", "Argentina" },
 161   { 0x3001, 864, 1256, "Arabic", "Lebanon" },
 162   { 0x3009, 437, 1252, "English", "Zimbabwe" },       { 0x300A, 850, 1252, "Spanish", "Ecuador" },
 163   { 0x3401, 864, 1256, "Arabic", "Kuwait" },
 164   { 0x3409, 437, 1252, "English", "Philippines" },    { 0x340A, 850, 1252, "Spanish", "Chile" },
 165   { 0x3801, 864, 1256, "Arabic", "United Arab Emirates" },
 166   { 0x380A, 850, 1252, "Spanish", "Uruguay" },
 167   { 0x3C01, 864, 1256, "Arabic", "Bahrain" },
 168   { 0x3C0A, 850, 1252, "Spanish", "Paraguay" },
 169   { 0x4001, 864, 1256, "Arabic", "Qatar" },
 170   { 0x400A, 850, 1252, "Spanish", "Bolivia" },
 171   { 0x440A, 850, 1252, "Spanish", "El Salvador" },
 172   { 0x480A, 850, 1252, "Spanish", "Honduras" },
 173   { 0x4C0A, 850, 1252, "Spanish", "Nicaragua" },
 174   { 0x500A, 850, 1252, "Spanish", "Puerto Rico" },
 175   { (unsigned) -1,  0,      0, NULL, NULL }
 176 };
 177
 178 #endif
 179
 180 /* Specifies the default codepage to be used for unicode
 181    transformations.  By default this is CP_ACP.  */
 182 rc_uint_type wind_default_codepage = CP_ACP;
 183
 184 /* Specifies the currently used codepage for unicode
 185    transformations.  By default this is CP_ACP.  */
 186 rc_uint_type wind_current_codepage = CP_ACP;
 187
 188 /* Convert an ASCII string to a unicode string.  We just copy it,
 189    expanding chars to shorts, rather than doing something intelligent.  */
 190
 191 void
 192 unicode_from_ascii (rc_uint_type *length, unichar **unicode, const char *ascii)
 193 {
 194   unicode_from_codepage (length, unicode, ascii, wind_current_codepage);
 195 }
 196
 197 /* Convert an unicode string to an ASCII string.  We just copy it,
 198    shrink shorts to chars, rather than doing something intelligent.
 199    Shorts with not within the char range are replaced by '_'.  */
 200
 201 void
 202 ascii_from_unicode (rc_uint_type *length, const unichar *unicode, char **ascii)
 203 {
 204   codepage_from_unicode (length, unicode, ascii, wind_current_codepage);
 205 }
 206
 207 /* Print the unicode string UNICODE to the file E.  LENGTH is the
 208    number of characters to print, or -1 if we should print until the
 209    end of the string.  FIXME: On a Windows host, we should be calling
 210    some Windows function, probably WideCharToMultiByte.  */
 211
 212 void
 213 unicode_print (FILE *e, const unichar *unicode, rc_uint_type length)
 214 {
 215   while (1)
 216     {
 217       unichar ch;
 218
 219       if (length == 0)
 220         return;
 221       if ((bfd_signed_vma) length > 0)
 222         --length;
 223
 224       ch = *unicode;
 225
 226       if (ch == 0 && (bfd_signed_vma) length < 0)
 227         return;
 228
 229       ++unicode;
 230
 231       if ((ch & 0x7f) == ch)
 232         {
 233           if (ch == '\\')
 234             fputs ("\\\\", e);
 235           else if (ch == '"')
 236             fputs ("\"\"", e);
 237           else if (ISPRINT (ch))
 238             putc (ch, e);
 239           else
 240             {
 241               switch (ch)
 242                 {
 243                 case ESCAPE_A:
 244                   fputs ("\\a", e);
 245                   break;
 246
 247                 case ESCAPE_B:
 248                   fputs ("\\b", e);
 249                   break;
 250
 251                 case ESCAPE_F:
 252                   fputs ("\\f", e);
 253                   break;
 254
 255                 case ESCAPE_N:
 256                   fputs ("\\n", e);
 257                   break;
 258
 259                 case ESCAPE_R:
 260                   fputs ("\\r", e);
 261                   break;
 262
 263                 case ESCAPE_T:
 264                   fputs ("\\t", e);
 265                   break;
 266
 267                 case ESCAPE_V:
 268                   fputs ("\\v", e);
 269                   break;
 270
 271                 default:
 272                   fprintf (e, "\\%03o", (unsigned int) ch);
 273                   break;
 274                 }
 275             }
 276         }
 277       else if ((ch & 0xff) == ch)
 278         fprintf (e, "\\%03o", (unsigned int) ch);
 279       else
 280         fprintf (e, "\\x%04x", (unsigned int) ch);
 281     }
 282 }
 283
 284 /* Print a unicode string to a file.  */
 285
 286 void
 287 ascii_print (FILE *e, const char *s, rc_uint_type length)
 288 {
 289   while (1)
 290     {
 291       char ch;
 292
 293       if (length == 0)
 294         return;
 295       if ((bfd_signed_vma) length > 0)
 296         --length;
 297
 298       ch = *s;
 299
 300       if (ch == 0 && (bfd_signed_vma) length < 0)
 301         return;
 302
 303       ++s;
 304
 305       if ((ch & 0x7f) == ch)
 306         {
 307           if (ch == '\\')
 308             fputs ("\\\\", e);
 309           else if (ch == '"')
 310             fputs ("\"\"", e);
 311           else if (ISPRINT (ch))
 312             putc (ch, e);
 313           else
 314             {
 315               switch (ch)
 316                 {
 317                 case ESCAPE_A:
 318                   fputs ("\\a", e);
 319                   break;
 320
 321                 case ESCAPE_B:
 322                   fputs ("\\b", e);
 323                   break;
 324
 325                 case ESCAPE_F:
 326                   fputs ("\\f", e);
 327                   break;
 328
 329                 case ESCAPE_N:
 330                   fputs ("\\n", e);
 331                   break;
 332
 333                 case ESCAPE_R:
 334                   fputs ("\\r", e);
 335                   break;
 336
 337                 case ESCAPE_T:
 338                   fputs ("\\t", e);
 339                   break;
 340
 341                 case ESCAPE_V:
 342                   fputs ("\\v", e);
 343                   break;
 344
 345                 default:
 346                   fprintf (e, "\\%03o", (unsigned int) ch);
 347                   break;
 348                 }
 349             }
 350         }
 351       else
 352         fprintf (e, "\\%03o", (unsigned int) ch & 0xff);
 353     }
 354 }
 355
 356 rc_uint_type
 357 unichar_len (const unichar *unicode)
 358 {
 359   rc_uint_type r = 0;
 360
 361   if (unicode)
 362     while (unicode[r] != 0)
 363       r++;
 364   else
 365     --r;
 366   return r;
 367 }
 368
 369 unichar *
 370 unichar_dup (const unichar *unicode)
 371 {
 372   unichar *r;
 373   int len;
 374
 375   if (! unicode)
 376     return NULL;
 377   for (len = 0; unicode[len] != 0; ++len)
 378     ;
 379   ++len;
 380   r = ((unichar *) res_alloc (len * sizeof (unichar)));
 381   memcpy (r, unicode, len * sizeof (unichar));
 382   return r;
 383 }
 384
 385 unichar *
 386 unichar_dup_uppercase (const unichar *u)
 387 {
 388   unichar *r = unichar_dup (u);
 389   int i;
 390
 391   if (! r)
 392     return NULL;
 393
 394   for (i = 0; r[i] != 0; ++i)
 395     {
 396       if (r[i] >= 'a' && r[i] <= 'z')
 397         r[i] &= 0xdf;
 398     }
 399   return r;
 400 }
 401
 402 static int
 403 unichar_isascii (const unichar *u, rc_uint_type len)
 404 {
 405   rc_uint_type i;
 406
 407   if ((bfd_signed_vma) len < 0)
 408     {
 409       if (u)
 410         len = (rc_uint_type) unichar_len (u);
 411       else
 412         len = 0;
 413     }
 414
 415   for (i = 0; i < len; i++)
 416     if ((u[i] & 0xff80) != 0)
 417       return 0;
 418   return 1;
 419 }
 420
 421 void
 422 unicode_print_quoted (FILE *e, const unichar *u, rc_uint_type len)
 423 {
 424   if (! unichar_isascii (u, len))
 425     fputc ('L', e);
 426   fputc ('"', e);
 427   unicode_print (e, u, len);
 428   fputc ('"', e);
 429 }
 430
 431 int
 432 unicode_is_valid_codepage (rc_uint_type cp)
 433 {
 434   if ((cp & 0xffff) != cp)
 435     return 0;
 436   if (cp == CP_UTF16 || cp == CP_ACP)
 437     return 1;
 438
 439 #if !defined (_WIN32) && !defined (__CYGWIN__)
 440   if (! wind_find_codepage_info (cp))
 441     return 0;
 442   return 1;
 443 #else
 444   return !! IsValidCodePage ((UINT) cp);
 445 #endif
 446 }
 447
 448 #if defined (_WIN32) || defined (__CYGWIN__)
 449
 450 #define max_cp_string_len 6
 451
 452 static unsigned int
 453 codepage_from_langid (unsigned short langid)
 454 {
 455   char cp_string [max_cp_string_len];
 456   int c;
 457
 458   memset (cp_string, 0, max_cp_string_len);
 459   /* LOCALE_RETURN_NUMBER flag would avoid strtoul conversion,
 460      but is unavailable on Win95.  */
 461   c = GetLocaleInfoA (MAKELCID (langid, SORT_DEFAULT),
 462                       LOCALE_IDEFAULTANSICODEPAGE,
 463                       cp_string, max_cp_string_len);
 464   /* If codepage data for an LCID is not installed on users's system,
 465      GetLocaleInfo returns an empty string.  Fall back to system ANSI
 466      default. */
 467   if (c == 0)
 468     return CP_ACP;
 469   return strtoul (cp_string, 0, 10);
 470 }
 471
 472 static unsigned int
 473 wincodepage_from_langid (unsigned short langid)
 474 {
 475   char cp_string [max_cp_string_len];
 476   int c;
 477
 478   memset (cp_string, 0, max_cp_string_len);
 479   /* LOCALE_RETURN_NUMBER flag would avoid strtoul conversion,
 480      but is unavailable on Win95.  */
 481   c = GetLocaleInfoA (MAKELCID (langid, SORT_DEFAULT),
 482                       LOCALE_IDEFAULTCODEPAGE,
 483                       cp_string, max_cp_string_len);
 484   /* If codepage data for an LCID is not installed on users's system,
 485      GetLocaleInfo returns an empty string.  Fall back to system ANSI
 486      default. */
 487   if (c == 0)
 488     return CP_OEM;
 489   return strtoul (cp_string, 0, 10);
 490 }
 491
 492 static char *
 493 lang_from_langid (unsigned short langid)
 494 {
 495   char cp_string[261];
 496   int c;
 497
 498   memset (cp_string, 0, 261);
 499   c = GetLocaleInfoA (MAKELCID (langid, SORT_DEFAULT),
 500                       LOCALE_SENGLANGUAGE,
 501                       cp_string, 260);
 502   /* If codepage data for an LCID is not installed on users's system,
 503      GetLocaleInfo returns an empty string.  Fall back to system ANSI
 504      default. */
 505   if (c == 0)
 506     strcpy (cp_string, "Neutral");
 507   return xstrdup (cp_string);
 508 }
 509
 510 static char *
 511 country_from_langid (unsigned short langid)
 512 {
 513   char cp_string[261];
 514   int c;
 515
 516   memset (cp_string, 0, 261);
 517   c = GetLocaleInfoA (MAKELCID (langid, SORT_DEFAULT),
 518                       LOCALE_SENGCOUNTRY,
 519                       cp_string, 260);
 520   /* If codepage data for an LCID is not installed on users's system,
 521      GetLocaleInfo returns an empty string.  Fall back to system ANSI
 522      default. */
 523   if (c == 0)
 524     strcpy (cp_string, "Neutral");
 525   return xstrdup (cp_string);
 526 }
 527
 528 #endif
 529
 530 const wind_language_t *
 531 wind_find_language_by_id (unsigned id)
 532 {
 533 #if !defined (_WIN32) && !defined (__CYGWIN__)
 534   int i;
 535
 536   if (! id)
 537     return NULL;
 538   for (i = 0; languages[i].id != (unsigned) -1 && languages[i].id != id; i++)
 539     ;
 540   if (languages[i].id == id)
 541     return &languages[i];
 542   return NULL;
 543 #else
 544   static wind_language_t wl;
 545
 546   wl.id = id;
 547   wl.doscp = codepage_from_langid ((unsigned short) id);
 548   wl.wincp = wincodepage_from_langid ((unsigned short) id);
 549   wl.name = lang_from_langid ((unsigned short) id);
 550   wl.country = country_from_langid ((unsigned short) id);
 551
 552   return & wl;
 553 #endif
 554 }
 555
 556 const local_iconv_map *
 557 wind_find_codepage_info (unsigned cp)
 558 {
 559 #if !defined (_WIN32) && !defined (__CYGWIN__)
 560   int i;
 561
 562   for (i = 0; codepages[i].codepage != (rc_uint_type) -1 && codepages[i].codepage != cp; i++)
 563     ;
 564   if (codepages[i].codepage == (rc_uint_type) -1)
 565     return NULL;
 566   return &codepages[i];
 567 #else
 568   static local_iconv_map lim;
 569   if (!unicode_is_valid_codepage (cp))
 570         return NULL;
 571   lim.codepage = cp;
 572   lim.iconv_name = "";
 573   return & lim;
 574 #endif
 575 }
 576
 577 /* Convert an Codepage string to a unicode string.  */
 578
 579 void
 580 unicode_from_codepage (rc_uint_type *length, unichar **u, const char *src, rc_uint_type cp)
 581 {
 582   rc_uint_type len;
 583
 584   len = wind_MultiByteToWideChar (cp, src, NULL, 0);
 585   if (len)
 586     {
 587       *u = ((unichar *) res_alloc (len));
 588       wind_MultiByteToWideChar (cp, src, *u, len);
 589     }
 590   /* Discount the trailing '/0'.  If MultiByteToWideChar failed,
 591      this will set *length to -1.  */
 592   len -= sizeof (unichar);
 593
 594   if (length != NULL)
 595     *length = len / sizeof (unichar);
 596 }
 597
 598 /* Convert an unicode string to an codepage string.  */
 599
 600 void
 601 codepage_from_unicode (rc_uint_type *length, const unichar *unicode, char **ascii, rc_uint_type cp)
 602 {
 603   rc_uint_type len;
 604
 605   len = wind_WideCharToMultiByte (cp, unicode, NULL, 0);
 606   if (len)
 607     {
 608       *ascii = (char *) res_alloc (len * sizeof (char));
 609       wind_WideCharToMultiByte (cp, unicode, *ascii, len);
 610     }
 611   /* Discount the trailing '/0'.  If MultiByteToWideChar failed,
 612      this will set *length to -1.  */
 613   len--;
 614
 615   if (length != NULL)
 616     *length = len;
 617 }
 618
 619 #if defined (HAVE_ICONV) && !defined (_WIN32) && !defined (__CYGWIN__)
 620 static int
 621 iconv_onechar (iconv_t cd, ICONV_CONST char *s, char *d, int d_len, const char **n_s, char **n_d)
 622 {
 623   int i;
 624
 625   for (i = 1; i <= 32; i++)
 626     {
 627       char *tmp_d = d;
 628       ICONV_CONST char *tmp_s = s;
 629       size_t ret;
 630       size_t s_left = (size_t) i;
 631       size_t d_left = (size_t) d_len;
 632
 633       ret = iconv (cd, & tmp_s, & s_left, & tmp_d, & d_left);
 634
 635       if (ret != (size_t) -1)
 636         {
 637           *n_s = tmp_s;
 638           *n_d = tmp_d;
 639           return 0;
 640         }
 641     }
 642
 643   return 1;
 644 }
 645
 646 static const char *
 647 wind_iconv_cp (rc_uint_type cp)
 648 {
 649   const local_iconv_map *lim = wind_find_codepage_info (cp);
 650
 651   if (!lim)
 652     return NULL;
 653   return lim->iconv_name;
 654 }
 655 #endif /* HAVE_ICONV */
 656
 657 static rc_uint_type
 658 wind_MultiByteToWideChar (rc_uint_type cp, const char *mb,
 659                           unichar *u, rc_uint_type u_len)
 660 {
 661   rc_uint_type ret = 0;
 662
 663 #if defined (_WIN32) || defined (__CYGWIN__)
 664   rc_uint_type conv_flags = MB_PRECOMPOSED;
 665
 666   /* MB_PRECOMPOSED is not allowed for UTF-7 or UTF-8.
 667      MultiByteToWideChar will set the last error to
 668      ERROR_INVALID_FLAGS if we do. */
 669   if (cp == CP_UTF8 || cp == CP_UTF7)
 670     conv_flags = 0;
 671
 672   ret = (rc_uint_type) MultiByteToWideChar (cp, conv_flags,
 673                                             mb, -1, u, u_len);
 674   /* Convert to bytes. */
 675   ret *= sizeof (unichar);
 676
 677 #elif defined (HAVE_ICONV)
 678   int first = 1;
 679   char tmp[32];
 680   char *p_tmp;
 681   const char *iconv_name = wind_iconv_cp (cp);
 682
 683   if (!mb || !iconv_name)
 684     return 0;
 685   iconv_t cd = iconv_open ("UTF-16", iconv_name);
 686
 687   while (1)
 688     {
 689       int iret;
 690       const char *n_mb = "";
 691       char *n_tmp = "";
 692
 693       p_tmp = tmp;
 694       iret = iconv_onechar (cd, (ICONV_CONST char *) mb, p_tmp, 32, & n_mb, & n_tmp);
 695       if (first)
 696         {
 697           first = 0;
 698           continue;
 699         }
 700       if (!iret)
 701         {
 702           size_t l_tmp = (size_t) (n_tmp - p_tmp);
 703
 704           if (u)
 705             {
 706               if ((size_t) u_len < l_tmp)
 707                 break;
 708               memcpy (u, tmp, l_tmp);
 709               u += l_tmp/2;
 710               u_len -= l_tmp;
 711             }
 712           ret += l_tmp;
 713         }
 714       else
 715         break;
 716       if (tmp[0] == 0 && tmp[1] == 0)
 717         break;
 718       mb = n_mb;
 719     }
 720   iconv_close (cd);
 721 #else
 722   if (cp)
 723     ret = 0;
 724   ret = strlen (mb) + 1;
 725   ret *= sizeof (unichar);
 726   if (u != NULL && u_len != 0)
 727     {
 728       do
 729         {
 730           *u++ = ((unichar) *mb) & 0xff;
 731           --u_len; mb++;
 732         }
 733       while (u_len != 0 && mb[-1] != 0);
 734     }
 735   if (u != NULL && u_len != 0)
 736     *u = 0;
 737 #endif
 738   return ret;
 739 }
 740
 741 static rc_uint_type
 742 wind_WideCharToMultiByte (rc_uint_type cp, const unichar *u, char *mb, rc_uint_type mb_len)
 743 {
 744   rc_uint_type ret = 0;
 745 #if defined (_WIN32) || defined (__CYGWIN__)
 746   WINBOOL used_def = FALSE;
 747
 748   ret = (rc_uint_type) WideCharToMultiByte (cp, 0, u, -1, mb, mb_len,
 749                                             NULL, & used_def);
 750 #elif defined (HAVE_ICONV)
 751   int first = 1;
 752   char tmp[32];
 753   char *p_tmp;
 754   const char *iconv_name = wind_iconv_cp (cp);
 755
 756   if (!u || !iconv_name)
 757     return 0;
 758   iconv_t cd = iconv_open (iconv_name, "UTF-16");
 759
 760   while (1)
 761     {
 762       int iret;
 763       const char *n_u = "";
 764       char *n_tmp = "";
 765
 766       p_tmp = tmp;
 767       iret = iconv_onechar (cd, (ICONV_CONST char *) u, p_tmp, 32, &n_u, & n_tmp);
 768       if (first)
 769         {
 770           first = 0;
 771           continue;
 772         }
 773       if (!iret)
 774         {
 775           size_t l_tmp = (size_t) (n_tmp - p_tmp);
 776
 777           if (mb)
 778             {
 779               if ((size_t) mb_len < l_tmp)
 780                 break;
 781               memcpy (mb, tmp, l_tmp);
 782               mb += l_tmp;
 783               mb_len -= l_tmp;
 784             }
 785           ret += l_tmp;
 786         }
 787       else
 788         break;
 789       if (u[0] == 0)
 790         break;
 791       u = (const unichar *) n_u;
 792     }
 793   iconv_close (cd);
 794 #else
 795   if (cp)
 796     ret = 0;
 797
 798   while (u[ret] != 0)
 799     ++ret;
 800
 801   ++ret;
 802
 803   if (mb)
 804     {
 805       while (*u != 0 && mb_len != 0)
 806         {
 807           if (u[0] == (u[0] & 0x7f))
 808             *mb++ = (char) u[0];
 809           else
 810             *mb++ = '_';
 811           ++u; --mb_len;
 812         }
 813       if (mb_len != 0)
 814         *mb = 0;
 815     }
 816 #endif
 817   return ret;
 818 }