1 /*****************************************************************************
2 * unicode.c: Unicode <-> locale functions
3 *****************************************************************************
4 * Copyright (C) 2005-2006 the VideoLAN team
5 * Copyright © 2005-2008 Rémi Denis-Courmont
7 * Authors: Rémi Denis-Courmont <rem # videolan.org>
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 2 of the License, or
12 * (at your option) any later version.
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.
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 02110-1301, USA.
22 *****************************************************************************/
24 /*****************************************************************************
26 *****************************************************************************/
31 #include <vlc_common.h>
32 #include <vlc_charset.h>
33 #include "libvlc.h" /* utf8_mkdir */
41 #include <sys/types.h>
48 #ifdef HAVE_SYS_STAT_H
49 # include <sys/stat.h>
61 # define lstat( a, b ) stat(a, b)
65 /* Define this if the OS always use UTF-8 internally */
66 # define ASSUME_UTF8 1
69 #if defined (ASSUME_UTF8)
71 #elif defined (WIN32) || defined (UNDER_CE)
73 #elif defined (HAVE_ICONV)
76 # error No UTF8 charset conversion implemented on this platform!
79 #if defined (USE_ICONV)
80 # include <langinfo.h>
81 static char charset[sizeof ("CSISO11SWEDISHFORNAMES")] = "";
83 static void find_charset_once (void)
85 strlcpy (charset, nl_langinfo (CODESET), sizeof (charset));
86 if (!strcasecmp (charset, "ASCII")
87 || !strcasecmp (charset, "ANSI_X3.4-1968"))
88 strcpy (charset, "UTF-8"); /* superset... */
91 static int find_charset (void)
93 static pthread_once_t once = PTHREAD_ONCE_INIT;
94 pthread_once (&once, find_charset_once);
95 return !strcasecmp (charset, "UTF-8");
100 static char *locale_fast (const char *string, bool from)
105 #if defined (USE_ICONV)
107 return (char *)string;
109 vlc_iconv_t hd = vlc_iconv_open (from ? "UTF-8" : charset,
110 from ? charset : "UTF-8");
111 if (hd == (vlc_iconv_t)(-1))
112 return NULL; /* Uho! */
114 const char *iptr = string;
115 size_t inb = strlen (string);
116 size_t outb = inb * 6 + 1;
117 char output[outb], *optr = output;
119 while (vlc_iconv (hd, &iptr, &inb, &optr, &outb) == (size_t)(-1))
125 vlc_iconv (hd, NULL, NULL, NULL, NULL); /* reset */
128 vlc_iconv_close (hd);
131 assert (*iptr == '\0');
132 assert (*optr == '\0');
133 assert (strlen (output) == (size_t)(optr - output));
134 return strdup (output);
135 #elif defined (USE_MB2MB)
139 len = 1 + MultiByteToWideChar (from ? CP_ACP : CP_UTF8,
140 0, string, -1, NULL, 0);
143 MultiByteToWideChar (from ? CP_ACP : CP_UTF8, 0, string, -1, wide, len);
144 len = 1 + WideCharToMultiByte (from ? CP_UTF8 : CP_ACP, 0, wide, -1,
145 NULL, 0, NULL, NULL);
150 WideCharToMultiByte (from ? CP_UTF8 : CP_ACP, 0, wide, -1, out, len,
155 return (char *)string;
160 static inline char *locale_dup (const char *string, bool from)
164 #if defined (USE_ICONV)
166 return strdup (string);
167 return locale_fast (string, from);
168 #elif defined (USE_MB2MB)
169 return locale_fast (string, from);
172 return strdup (string);
177 * Releases (if needed) a localized or uniformized string.
178 * @param str non-NULL return value from FromLocale() or ToLocale().
180 void LocaleFree (const char *str)
182 #if defined (USE_ICONV)
183 if (!find_charset ())
185 #elif defined (USE_MB2MB)
194 * Converts a string from the system locale character encoding to UTF-8.
196 * @param locale nul-terminated string to convert
198 * @return a nul-terminated UTF-8 string, or NULL in case of error.
199 * To avoid memory leak, you have to pass the result to LocaleFree()
200 * when it is no longer needed.
202 char *FromLocale (const char *locale)
204 return locale_fast (locale, true);
208 * converts a string from the system locale character encoding to utf-8,
209 * the result is always allocated on the heap.
211 * @param locale nul-terminated string to convert
213 * @return a nul-terminated utf-8 string, or null in case of error.
214 * The result must be freed using free() - as with the strdup() function.
216 char *FromLocaleDup (const char *locale)
218 return locale_dup (locale, true);
223 * ToLocale: converts an UTF-8 string to local system encoding.
225 * @param utf8 nul-terminated string to be converted
227 * @return a nul-terminated string, or NULL in case of error.
228 * To avoid memory leak, you have to pass the result to LocaleFree()
229 * when it is no longer needed.
231 char *ToLocale (const char *utf8)
233 return locale_fast (utf8, false);
238 * converts a string from UTF-8 to the system locale character encoding,
239 * the result is always allocated on the heap.
241 * @param utf8 nul-terminated string to convert
243 * @return a nul-terminated string, or null in case of error.
244 * The result must be freed using free() - as with the strdup() function.
246 char *ToLocaleDup (const char *utf8)
248 return locale_dup (utf8, false);
253 * Opens a system file handle using UTF-8 paths.
255 * @param filename file path to open (with UTF-8 encoding)
256 * @param flags open() flags, see the C library open() documentation
257 * @param mode file permissions if creating a new file
258 * @return a file handle on success, -1 on error (see errno).
260 int utf8_open (const char *filename, int flags, mode_t mode)
263 /*_open translates to wchar internally on WinCE*/
264 return _open (filename, flags, mode);
265 #elif defined (WIN32)
266 /* for Windows NT and above */
267 wchar_t wpath[MAX_PATH + 1];
269 if (!MultiByteToWideChar (CP_UTF8, 0, filename, -1, wpath, MAX_PATH))
274 wpath[MAX_PATH] = L'\0';
277 * open() cannot open files with non-“ANSI” characters on Windows.
278 * We use _wopen() instead. Same thing for mkdir() and stat().
280 return _wopen (wpath, flags, mode);
283 const char *local_name = ToLocale (filename);
285 if (local_name == NULL)
291 int fd = open (local_name, flags, mode);
292 LocaleFree (local_name);
297 * Opens a FILE pointer using UTF-8 filenames.
298 * @param filename file path, using UTF-8 encoding
299 * @param mode fopen file open mode
300 * @return NULL on error, an open FILE pointer on success.
302 FILE *utf8_fopen (const char *filename, const char *mode)
304 int rwflags = 0, oflags = 0;
307 for (const char *ptr = mode; *ptr; ptr++)
323 oflags |= O_CREAT | O_TRUNC;
338 int fd = utf8_open (filename, rwflags | oflags, 0666);
342 if (append && (lseek (fd, 0, SEEK_END) == -1))
348 FILE *stream = fdopen (fd, mode);
356 * Creates a directory using UTF-8 paths.
358 * @param dirname a UTF-8 string with the name of the directory that you
360 * @param mode directory permissions
361 * @return 0 on success, -1 on error (see errno).
363 int utf8_mkdir( const char *dirname, mode_t mode )
365 #if defined (UNDER_CE) || defined (WIN32)
368 wchar_t wname[MAX_PATH + 1];
369 char mod[MAX_PATH + 1];
372 /* Convert '/' into '\' */
373 for( i = 0; *dirname; i++ )
376 return -1; /* overflow */
378 if( *dirname == '/' )
387 if( MultiByteToWideChar( CP_UTF8, 0, mod, -1, wname, MAX_PATH ) == 0 )
392 wname[MAX_PATH] = L'\0';
394 if( CreateDirectoryW( wname, NULL ) == 0 )
396 if( GetLastError( ) == ERROR_ALREADY_EXISTS )
404 char *locname = ToLocale( dirname );
407 if( locname == NULL )
412 res = mkdir( locname, mode );
414 LocaleFree( locname );
420 * Opens a DIR pointer using UTF-8 paths
422 * @param dirname UTF-8 representation of the directory name
423 * @return a pointer to the DIR struct, or NULL in case of error.
424 * Release with standard closedir().
426 DIR *utf8_opendir( const char *dirname )
429 wchar_t wname[MAX_PATH + 1];
431 if (MultiByteToWideChar (CP_UTF8, 0, dirname, -1, wname, MAX_PATH))
433 wname[MAX_PATH] = L'\0';
434 return (DIR *)vlc_wopendir (wname);
437 const char *local_name = ToLocale( dirname );
439 if( local_name != NULL )
441 DIR *dir = opendir( local_name );
442 LocaleFree( local_name );
452 * Reads the next file name from an open directory.
454 * @param dir The directory that is being read
456 * @return a UTF-8 string of the directory entry.
457 * Use free() to free this memory.
459 char *utf8_readdir( DIR *dir )
462 struct _wdirent *ent = vlc_wreaddir (dir);
466 return FromWide (ent->d_name);
470 ent = readdir( (DIR *)dir );
474 return vlc_fix_readdir( ent->d_name );
478 static int dummy_select( const char *str )
485 * Does the same as utf8_scandir(), but takes an open directory pointer
486 * instead of a directory path.
488 int utf8_loaddir( DIR *dir, char ***namelist,
489 int (*select)( const char * ),
490 int (*compar)( const char **, const char ** ) )
493 select = dummy_select;
505 while( ( entry = utf8_readdir( dir ) ) != NULL )
509 if( !select( entry ) )
515 newtab = realloc( tab, sizeof( char * ) * (num + 1) );
526 qsort( tab, num, sizeof( tab[0] ),
527 (int (*)( const void *, const void *))compar );
535 for( i = 0; i < num; i++ )
545 * Selects file entries from a directory, as GNU C scandir(), yet using
548 * @param dirname UTF-8 diretory path
549 * @param pointer [OUT] pointer set, on succesful completion, to the address
550 * of a table of UTF-8 filenames. All filenames must be freed with free().
551 * The table itself must be freed with free() as well.
553 * @return How many file names were selected (possibly 0),
554 * or -1 in case of error.
556 int utf8_scandir( const char *dirname, char ***namelist,
557 int (*select)( const char * ),
558 int (*compar)( const char **, const char ** ) )
560 DIR *dir = utf8_opendir (dirname);
565 val = utf8_loaddir (dir, namelist, select, compar);
571 static int utf8_statEx( const char *filename, struct stat *buf,
575 /*_stat translates to wchar internally on WinCE*/
576 return _stat( filename, buf );
577 #elif defined (WIN32)
578 /* for Windows NT and above */
579 wchar_t wpath[MAX_PATH + 1];
581 if( !MultiByteToWideChar( CP_UTF8, 0, filename, -1, wpath, MAX_PATH ) )
586 wpath[MAX_PATH] = L'\0';
588 return _wstati64( wpath, buf );
591 #ifdef HAVE_SYS_STAT_H
592 const char *local_name = ToLocale( filename );
594 if( local_name != NULL )
596 int res = deref ? stat( local_name, buf )
597 : lstat( local_name, buf );
598 LocaleFree( local_name );
607 * Finds file/inode informations, as stat().
608 * Consider usign fstat() instead, if possible.
610 * @param filename UTF-8 file path
612 int utf8_stat( const char *filename, struct stat *buf)
614 return utf8_statEx( filename, buf, true );
618 * Finds file/inode informations, as lstat().
619 * Consider usign fstat() instead, if possible.
621 * @param filename UTF-8 file path
623 int utf8_lstat( const char *filename, struct stat *buf)
625 return utf8_statEx( filename, buf, false );
631 * @param filename a UTF-8 string with the name of the file you want to delete.
632 * @return A 0 return value indicates success. A -1 return value indicates an
633 * error, and an error code is stored in errno
635 int utf8_unlink( const char *filename )
638 /*_open translates to wchar internally on WinCE*/
639 return _unlink( filename );
640 #elif defined (WIN32)
641 /* for Windows NT and above */
642 wchar_t wpath[MAX_PATH + 1];
644 if( !MultiByteToWideChar( CP_UTF8, 0, filename, -1, wpath, MAX_PATH ) )
649 wpath[MAX_PATH] = L'\0';
652 * unlink() cannot open files with non-“ANSI” characters on Windows.
653 * We use _wunlink() instead.
655 return _wunlink( wpath );
657 const char *local_name = ToLocale( filename );
659 if( local_name == NULL )
665 int ret = unlink( local_name );
666 LocaleFree( local_name );
673 * Formats an UTF-8 string as vasprintf(), then print it to stdout, with
674 * appropriate conversion to local encoding.
676 static int utf8_vasprintf( char **str, const char *fmt, va_list ap )
679 int res = vasprintf( &utf8, fmt, ap );
683 *str = ToLocaleDup( utf8 );
689 * Formats an UTF-8 string as vfprintf(), then print it, with
690 * appropriate conversion to local encoding.
692 int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap )
695 int res = utf8_vasprintf( &str, fmt, ap );
699 fputs( str, stream );
705 * Formats an UTF-8 string as fprintf(), then print it, with
706 * appropriate conversion to local encoding.
708 int utf8_fprintf( FILE *stream, const char *fmt, ... )
714 res = utf8_vfprintf( stream, fmt, ap );
720 static char *CheckUTF8( char *str, char rep )
722 uint8_t *ptr = (uint8_t *)str;
723 assert (str != NULL);
733 for (int i = 0; i < 7; i++)
734 if ((c >> (7 - i)) == ((0xff >> (7 - i)) ^ 1))
742 case 0: // 7-bit ASCII character -> OK
746 case -1: // 1111111x -> error
747 case 1: // continuation byte -> error
751 assert (charlen >= 2);
753 uint32_t cp = c & ~((0xff >> (7 - charlen)) << (7 - charlen));
754 for (int i = 1; i < charlen; i++)
756 assert (cp < (1 << 26));
759 if ((c == '\0') // unexpected end of string
760 || ((c >> 6) != 2)) // not a continuation byte
763 cp = (cp << 6) | (ptr[i] & 0x3f);
766 if (cp < 128) // overlong (special case for ASCII)
768 if (cp < (1u << (5 * charlen - 3))) // overlong
785 * Replaces invalid/overlong UTF-8 sequences with question marks.
786 * Note that it is not possible to convert from Latin-1 to UTF-8 on the fly,
787 * so we don't try that, even though it would be less disruptive.
789 * @return str if it was valid UTF-8, NULL if not.
791 char *EnsureUTF8( char *str )
793 return CheckUTF8( str, '?' );
798 * Checks whether a string is a valid UTF-8 byte sequence.
800 * @param str nul-terminated string to be checked
802 * @return str if it was valid UTF-8, NULL if not.
804 const char *IsUTF8( const char *str )
806 return CheckUTF8( (char *)str, 0 );