src/text/unicode.c

   1 /*****************************************************************************
   2  * unicode.c: Unicode <-> locale functions
   3  *****************************************************************************
   4  * Copyright (C) 2005-2006 the VideoLAN team
   5  * Copyright © 2005-2006 Rémi Denis-Courmont
   6  * $Id$
   7  *
   8  * Authors: Rémi Denis-Courmont <rem # videolan.org>
   9  *
  10  * This program is free software; you can redistribute it and/or modify
  11  * it under the terms of the GNU General Public License as published by
  12  * the Free Software Foundation; either version 2 of the License, or
  13  * (at your option) any later version.
  14  *
  15  * This program is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18  * GNU General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU General Public License
  21  * along with this program; if not, write to the Free Software
  22  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  23  *****************************************************************************/
  24
  25 /*****************************************************************************
  26  * Preamble
  27  *****************************************************************************/
  28 #ifdef HAVE_CONFIG_H
  29 # include "config.h"
  30 #endif
  31
  32 #include <vlc/vlc.h>
  33 #include <vlc_charset.h>
  34 #include "libvlc.h" /* utf8_mkdir */
  35
  36 #include <assert.h>
  37
  38 #include <stdio.h>
  39 #include <stdarg.h>
  40 #include <stdlib.h>
  41 #include <errno.h>
  42 #include <sys/types.h>
  43 #ifdef HAVE_DIRENT_H
  44 #  include <dirent.h>
  45 #endif
  46 #ifdef UNDER_CE
  47 #  include <tchar.h>
  48 #endif
  49 #ifdef HAVE_SYS_STAT_H
  50 # include <sys/stat.h>
  51 #endif
  52 #ifdef HAVE_FCNTL_H
  53 # include <fcntl.h>
  54 #endif
  55 #ifdef WIN32
  56 # include <io.h>
  57 #else
  58 # include <unistd.h>
  59 #endif
  60
  61 #ifndef HAVE_LSTAT
  62 # define lstat( a, b ) stat(a, b)
  63 #endif
  64
  65 #ifdef __APPLE__
  66 /* Define this if the OS always use UTF-8 internally */
  67 # define ASSUME_UTF8 1
  68 #endif
  69
  70 #if defined (ASSUME_UTF8)
  71 /* Cool */
  72 #elif defined (WIN32) || defined (UNDER_CE)
  73 # define USE_MB2MB 1
  74 #elif defined (HAVE_ICONV)
  75 # define USE_ICONV 1
  76 #else
  77 # error No UTF8 charset conversion implemented on this platform!
  78 #endif
  79
  80 #if defined (USE_ICONV)
  81 static char charset[sizeof ("CSISO11SWEDISHFORNAMES//translit")] = "";
  82
  83 static void find_charset_once (void)
  84 {
  85     char *psz_charset;
  86     if (vlc_current_charset (&psz_charset)
  87      || (psz_charset == NULL)
  88      || (strcmp (psz_charset, "ASCII") == 0)
  89      || ((size_t)snprintf (charset, sizeof (charset), "%s//translit",
  90                            psz_charset) >= sizeof (charset)))
  91         strcpy (charset, "UTF-8");
  92
  93     free (psz_charset);
  94 }
  95
  96 static int find_charset (void)
  97 {
  98     static pthread_once_t once = PTHREAD_ONCE_INIT;
  99     pthread_once (&once, find_charset_once);
 100     return !strcmp (charset, "UTF-8");
 101 }
 102 #endif
 103
 104
 105 static char *locale_fast (const char *string, vlc_bool_t from)
 106 {
 107 #if defined (USE_ICONV)
 108     if (find_charset ())
 109         return (char *)string;
 110
 111     vlc_iconv_t hd = vlc_iconv_open (from ? "UTF-8" : charset,
 112                                      from ? charset : "UTF-8");
 113     if (hd == (vlc_iconv_t)(-1))
 114         return strdup (string); /* Uho! */
 115
 116     const char *iptr = string;
 117     size_t inb = strlen (string);
 118     size_t outb = inb * 6 + 1;
 119     char output[outb], *optr = output;
 120
 121     if (string == NULL)
 122         return NULL;
 123
 124     while (vlc_iconv (hd, &iptr, &inb, &optr, &outb) == (size_t)(-1))
 125     {
 126         *optr++ = '?';
 127         outb--;
 128         iptr++;
 129         inb--;
 130         vlc_iconv (hd, NULL, NULL, NULL, NULL);
 131     }
 132     *optr = '\0';
 133     vlc_iconv_close (hd);
 134
 135     assert (inb == 0);
 136     assert (*iptr == '\0');
 137     assert (*optr == '\0');
 138     assert (strlen (output) == (size_t)(optr - output));
 139     return strdup (output);
 140 #elif defined (USE_MB2MB)
 141     char *out;
 142     int len;
 143
 144     if (string == NULL)
 145         return NULL;
 146
 147     len = 1 + MultiByteToWideChar (from ? CP_ACP : CP_UTF8,
 148                                    0, string, -1, NULL, 0);
 149     wchar_t wide[len];
 150
 151     MultiByteToWideChar (from ? CP_ACP : CP_UTF8, 0, string, -1, wide, len);
 152     len = 1 + WideCharToMultiByte (from ? CP_UTF8 : CP_ACP, 0, wide, -1, NULL, 0, NULL, NULL);
 153     out = malloc (len);
 154     if (out == NULL)
 155         return NULL;
 156
 157     WideCharToMultiByte (from ? CP_UTF8 : CP_ACP, 0, wide, -1, out, len, NULL, NULL);
 158     return out;
 159 #else
 160     return (char *)string;
 161 #endif
 162 }
 163
 164
 165 static inline char *locale_dup (const char *string, vlc_bool_t from)
 166 {
 167 #if defined (USE_ICONV)
 168     if (find_charset ())
 169         return strdup (string);
 170     return locale_fast (string, from);
 171 #elif defined (USE_MB2MB)
 172     return locale_fast (string, from);
 173 #else
 174     return strdup (string);
 175 #endif
 176 }
 177
 178
 179 void LocaleFree (const char *str)
 180 {
 181 #if defined (USE_ICONV)
 182     if (!find_charset ())
 183         free ((char *)str);
 184 #elif defined (USE_MB2MB)
 185     free ((char *)str);
 186 #endif
 187 }
 188
 189
 190 /**
 191  * FromLocale: converts a locale string to UTF-8
 192  *
 193  * @param locale nul-terminated string to be converted
 194  *
 195  * @return a nul-terminated UTF-8 string, or NULL in case of error.
 196  * To avoid memory leak, you have to pass the result to LocaleFree()
 197  * when it is no longer needed.
 198  */
 199 char *FromLocale (const char *locale)
 200 {
 201     return locale_fast (locale, VLC_TRUE);
 202 }
 203
 204 char *FromLocaleDup (const char *locale)
 205 {
 206     return locale_dup (locale, VLC_TRUE);
 207 }
 208
 209
 210 /**
 211  * ToLocale: converts a UTF-8 string to local system encoding.
 212  *
 213  * @param utf8 nul-terminated string to be converted
 214  *
 215  * @return a nul-terminated string, or NULL in case of error.
 216  * To avoid memory leak, you have to pass the result to LocaleFree()
 217  * when it is no longer needed.
 218  */
 219 char *ToLocale (const char *utf8)
 220 {
 221     return locale_fast (utf8, VLC_FALSE);
 222 }
 223
 224
 225 static char *ToLocaleDup (const char *utf8)
 226 {
 227     return locale_dup (utf8, VLC_FALSE);
 228 }
 229
 230
 231 /**
 232  * utf8_open: open() wrapper for UTF-8 filenames
 233  */
 234 int utf8_open (const char *filename, int flags, mode_t mode)
 235 {
 236 #if defined (WIN32) || defined (UNDER_CE)
 237     if (GetVersion() < 0x80000000)
 238     {
 239         /* for Windows NT and above */
 240         wchar_t wpath[MAX_PATH + 1];
 241
 242         if (!MultiByteToWideChar (CP_UTF8, 0, filename, -1, wpath, MAX_PATH))
 243         {
 244             errno = ENOENT;
 245             return -1;
 246         }
 247         wpath[MAX_PATH] = L'\0';
 248
 249         /*
 250          * open() cannot open files with non-“ANSI” characters on Windows.
 251          * We use _wopen() instead. Same thing for mkdir() and stat().
 252          */
 253         return _wopen (wpath, flags, mode);
 254     }
 255 #endif
 256     const char *local_name = ToLocale (filename);
 257
 258     if (local_name == NULL)
 259     {
 260         errno = ENOENT;
 261         return -1;
 262     }
 263
 264     int fd = open (local_name, flags, mode);
 265     LocaleFree (local_name);
 266     return fd;
 267 }
 268
 269 /**
 270  * utf8_fopen: fopen() wrapper for UTF-8 filenames
 271  */
 272 FILE *utf8_fopen (const char *filename, const char *mode)
 273 {
 274     int rwflags = 0, oflags = 0;
 275     vlc_bool_t append = VLC_FALSE;
 276
 277     for (const char *ptr = mode; *ptr; ptr++)
 278     {
 279         switch (*ptr)
 280         {
 281             case 'r':
 282                 rwflags = O_RDONLY;
 283                 break;
 284
 285             case 'a':
 286                 rwflags = O_WRONLY;
 287                 oflags |= O_CREAT;
 288                 append = VLC_TRUE;
 289                 break;
 290
 291             case 'w':
 292                 rwflags = O_WRONLY;
 293                 oflags |= O_CREAT | O_TRUNC;
 294                 break;
 295
 296             case '+':
 297                 rwflags = O_RDWR;
 298                 break;
 299
 300 #ifdef O_TEXT
 301             case 't':
 302                 oflags |= O_TEXT;
 303                 break;
 304 #endif
 305         }
 306     }
 307
 308     int fd = utf8_open (filename, rwflags | oflags, 0666);
 309     if (fd == -1)
 310         return NULL;
 311
 312     if (append && (lseek (fd, 0, SEEK_END) == -1))
 313     {
 314         close (fd);
 315         return NULL;
 316     }
 317
 318     FILE *stream = fdopen (fd, mode);
 319     if (stream == NULL)
 320         close (fd);
 321
 322     return stream;
 323 }
 324
 325 /**
 326  * utf8_mkdir: Calls mkdir() after conversion of file name to OS locale
 327  *
 328  * @param dirname a UTF-8 string with the name of the directory that you
 329  *        want to create.
 330  * @return A 0 return value indicates success. A -1 return value indicates an
 331  *        error, and an error code is stored in errno
 332  */
 333 int utf8_mkdir( const char *dirname, mode_t mode )
 334 {
 335 #if defined (UNDER_CE) || defined (WIN32)
 336     wchar_t wname[MAX_PATH + 1];
 337     char mod[MAX_PATH + 1];
 338     int i;
 339
 340     /* Convert '/' into '\' */
 341     for( i = 0; *dirname; i++ )
 342     {
 343         if( i == MAX_PATH )
 344             return -1; /* overflow */
 345
 346         if( *dirname == '/' )
 347             mod[i] = '\\';
 348         else
 349             mod[i] = *dirname;
 350         dirname++;
 351
 352     }
 353     mod[i] = 0;
 354
 355     if( MultiByteToWideChar( CP_UTF8, 0, mod, -1, wname, MAX_PATH ) == 0 )
 356     {
 357         errno = ENOENT;
 358         return -1;
 359     }
 360     wname[MAX_PATH] = L'\0';
 361
 362     if( CreateDirectoryW( wname, NULL ) == 0 )
 363     {
 364         if( GetLastError( ) == ERROR_ALREADY_EXISTS )
 365             errno = EEXIST;
 366         else
 367             errno = ENOENT;
 368         return -1;
 369     }
 370     return 0;
 371 #else
 372     char *locname = ToLocale( dirname );
 373     int res;
 374
 375     if( locname == NULL )
 376     {
 377         errno = ENOENT;
 378         return -1;
 379     }
 380     res = mkdir( locname, mode );
 381
 382     LocaleFree( locname );
 383     return res;
 384 #endif
 385 }
 386
 387 /**
 388  * utf8_opendir: wrapper that converts dirname to the locale in use by the OS
 389  *
 390  * @param dirname UTF-8 representation of the directory name
 391  *
 392  * @return a pointer to the DIR struct. Release with closedir().
 393  */
 394 DIR *utf8_opendir( const char *dirname )
 395 {
 396 #ifdef WIN32
 397     wchar_t wname[MAX_PATH + 1];
 398
 399     if (MultiByteToWideChar (CP_UTF8, 0, dirname, -1, wname, MAX_PATH))
 400     {
 401         wname[MAX_PATH] = L'\0';
 402         return (DIR *)vlc_wopendir (wname);
 403     }
 404 #else
 405     const char *local_name = ToLocale( dirname );
 406
 407     if( local_name != NULL )
 408     {
 409         DIR *dir = opendir( local_name );
 410         LocaleFree( local_name );
 411         return dir;
 412     }
 413 #endif
 414
 415     errno = ENOENT;
 416     return NULL;
 417 }
 418
 419 /**
 420  * utf8_readdir: a readdir wrapper that returns the name of the next entry
 421  *     in the directory as a UTF-8 string.
 422  *
 423  * @param dir The directory that is being read
 424  *
 425  * @return a UTF-8 string of the directory entry. Use free() to free this memory.
 426  */
 427 char *utf8_readdir( DIR *dir )
 428 {
 429 #ifdef WIN32
 430     struct _wdirent *ent = vlc_wreaddir (dir);
 431     if (ent == NULL)
 432         return NULL;
 433
 434     return FromWide (ent->d_name);
 435 #else
 436     struct dirent *ent;
 437
 438     ent = readdir( (DIR *)dir );
 439     if( ent == NULL )
 440         return NULL;
 441
 442     return vlc_fix_readdir( ent->d_name );
 443 #endif
 444 }
 445
 446 static int dummy_select( const char *str )
 447 {
 448     (void)str;
 449     return 1;
 450 }
 451
 452 int utf8_loaddir( DIR *dir, char ***namelist,
 453                   int (*select)( const char * ),
 454                   int (*compar)( const char **, const char ** ) )
 455 {
 456     if( select == NULL )
 457         select = dummy_select;
 458
 459     if( dir == NULL )
 460         return -1;
 461     else
 462     {
 463         char **tab = NULL;
 464         char *entry;
 465         unsigned num = 0;
 466
 467         rewinddir( dir );
 468
 469         while( ( entry = utf8_readdir( dir ) ) != NULL )
 470         {
 471             char **newtab;
 472
 473             if( !select( entry ) )
 474             {
 475                 free( entry );
 476                 continue;
 477             }
 478
 479             newtab = realloc( tab, sizeof( char * ) * (num + 1) );
 480             if( newtab == NULL )
 481             {
 482                 free( entry );
 483                 goto error;
 484             }
 485             tab = newtab;
 486             tab[num++] = entry;
 487         }
 488
 489         if( compar != NULL )
 490             qsort( tab, num, sizeof( tab[0] ),
 491                    (int (*)( const void *, const void *))compar );
 492
 493         *namelist = tab;
 494         return num;
 495
 496     error:{
 497         unsigned i;
 498
 499         for( i = 0; i < num; i++ )
 500             free( tab[i] );
 501         if( tab != NULL )
 502             free( tab );
 503         }
 504     }
 505     return -1;
 506 }
 507
 508 int utf8_scandir( const char *dirname, char ***namelist,
 509                   int (*select)( const char * ),
 510                   int (*compar)( const char **, const char ** ) )
 511 {
 512     DIR *dir = utf8_opendir (dirname);
 513     int val = -1;
 514
 515     if (dir != NULL)
 516     {
 517         val = utf8_loaddir (dir, namelist, select, compar);
 518         closedir (dir);
 519     }
 520     return val;
 521 }
 522
 523 static int utf8_statEx( const char *filename, struct stat *buf,
 524                         vlc_bool_t deref )
 525 {
 526 #if defined (WIN32) || defined (UNDER_CE)
 527     /* retrieve Windows OS version */
 528     if( GetVersion() < 0x80000000 )
 529     {
 530         /* for Windows NT and above */
 531         wchar_t wpath[MAX_PATH + 1];
 532
 533         if( !MultiByteToWideChar( CP_UTF8, 0, filename, -1, wpath, MAX_PATH ) )
 534         {
 535             errno = ENOENT;
 536             return -1;
 537         }
 538         wpath[MAX_PATH] = L'\0';
 539
 540         return _wstati64( wpath, buf );
 541     }
 542 #endif
 543 #ifdef HAVE_SYS_STAT_H
 544     const char *local_name = ToLocale( filename );
 545
 546     if( local_name != NULL )
 547     {
 548         int res = deref ? stat( local_name, buf )
 549                        : lstat( local_name, buf );
 550         LocaleFree( local_name );
 551         return res;
 552     }
 553     errno = ENOENT;
 554 #endif
 555     return -1;
 556 }
 557
 558
 559 int utf8_stat( const char *filename, struct stat *buf)
 560 {
 561     return utf8_statEx( filename, buf, VLC_TRUE );
 562 }
 563
 564 int utf8_lstat( const char *filename, struct stat *buf)
 565 {
 566     return utf8_statEx( filename, buf, VLC_FALSE );
 567 }
 568
 569 /**
 570  * utf8_unlink: Calls unlink() after conversion of file name to OS locale
 571  *
 572  * @param filename a UTF-8 string with the name of the file you want to delete.
 573  * @return A 0 return value indicates success. A -1 return value indicates an
 574  *        error, and an error code is stored in errno
 575  */
 576 int utf8_unlink( const char *filename )
 577 {
 578 #if defined (WIN32) || defined (UNDER_CE)
 579     if( GetVersion() < 0x80000000 )
 580     {
 581         /* for Windows NT and above */
 582         wchar_t wpath[MAX_PATH + 1];
 583
 584         if( !MultiByteToWideChar( CP_UTF8, 0, filename, -1, wpath, MAX_PATH ) )
 585         {
 586             errno = ENOENT;
 587             return -1;
 588         }
 589         wpath[MAX_PATH] = L'\0';
 590
 591         /*
 592          * unlink() cannot open files with non-“ANSI” characters on Windows.
 593          * We use _wunlink() instead.
 594          */
 595         return _wunlink( wpath );
 596     }
 597 #endif
 598     const char *local_name = ToLocale( filename );
 599
 600     if( local_name == NULL )
 601     {
 602         errno = ENOENT;
 603         return -1;
 604     }
 605
 606     int ret = unlink( local_name );
 607     LocaleFree( local_name );
 608     return ret;
 609 }
 610
 611
 612
 613 /**
 614  * utf8_*printf: *printf with conversion from UTF-8 to local encoding
 615  */
 616 static int utf8_vasprintf( char **str, const char *fmt, va_list ap )
 617 {
 618     char *utf8;
 619     int res = vasprintf( &utf8, fmt, ap );
 620     if( res == -1 )
 621         return -1;
 622
 623     *str = ToLocaleDup( utf8 );
 624     free( utf8 );
 625     return res;
 626 }
 627
 628 int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap )
 629 {
 630     char *str;
 631     int res = utf8_vasprintf( &str, fmt, ap );
 632     if( res == -1 )
 633         return -1;
 634
 635     fputs( str, stream );
 636     free( str );
 637     return res;
 638 }
 639
 640 int utf8_fprintf( FILE *stream, const char *fmt, ... )
 641 {
 642     va_list ap;
 643     int res;
 644
 645     va_start( ap, fmt );
 646     res = utf8_vfprintf( stream, fmt, ap );
 647     va_end( ap );
 648     return res;
 649 }
 650
 651
 652 static char *CheckUTF8( char *str, char rep )
 653 {
 654     uint8_t *ptr = (uint8_t *)str;
 655     assert (str != NULL);
 656
 657     for (;;)
 658     {
 659         uint8_t c = ptr[0];
 660         int charlen = -1;
 661
 662         if (c == '\0')
 663             break;
 664
 665         for (int i = 0; i < 7; i++)
 666             if ((c >> (7 - i)) == ((0xff >> (7 - i)) ^ 1))
 667             {
 668                 charlen = i;
 669                 break;
 670             }
 671
 672         switch (charlen)
 673         {
 674             case 0: // 7-bit ASCII character -> OK
 675                 ptr++;
 676                 continue;
 677
 678             case -1: // 1111111x -> error
 679             case 1: // continuation byte -> error
 680                 goto error;
 681         }
 682
 683         assert (charlen >= 2);
 684
 685         uint32_t cp = c & ~((0xff >> (7 - charlen)) << (7 - charlen));
 686         for (int i = 1; i < charlen; i++)
 687         {
 688             assert (cp < (1 << 26));
 689             c = ptr[i];
 690
 691             if ((c == '\0') // unexpected end of string
 692              || ((c >> 6) != 2)) // not a continuation byte
 693                 goto error;
 694
 695             cp = (cp << 6) | (ptr[i] & 0x3f);
 696         }
 697
 698         if (cp < 128) // overlong (special case for ASCII)
 699             goto error;
 700         if (cp < (1u << (5 * charlen - 3))) // overlong
 701             goto error;
 702
 703         ptr += charlen;
 704         continue;
 705
 706     error:
 707         if (rep == 0)
 708             return NULL;
 709         *ptr++ = rep;
 710         str = NULL;
 711     }
 712
 713     return str;
 714 }
 715
 716 /**
 717  * EnsureUTF8: replaces invalid/overlong UTF-8 sequences with question marks
 718  * Note that it is not possible to convert from Latin-1 to UTF-8 on the fly,
 719  * so we don't try that, even though it would be less disruptive.
 720  *
 721  * @return str if it was valid UTF-8, NULL if not.
 722  */
 723 char *EnsureUTF8( char *str )
 724 {
 725     return CheckUTF8( str, '?' );
 726 }
 727
 728
 729 /**
 730  * IsUTF8: checks whether a string is a valid UTF-8 byte sequence.
 731  *
 732  * @param str nul-terminated string to be checked
 733  *
 734  * @return str if it was valid UTF-8, NULL if not.
 735  */
 736 const char *IsUTF8( const char *str )
 737 {
 738     return CheckUTF8( (char *)str, 0 );
 739 }