* unicode.c: Unicode <-> locale functions
*****************************************************************************
* Copyright (C) 2005-2006 the VideoLAN team
- * Copyright © 2005-2006 Rémi Denis-Courmont
- * $Id$
+ * Copyright © 2005-2010 Rémi Denis-Courmont
*
* Authors: Rémi Denis-Courmont <rem # videolan.org>
*
/*****************************************************************************
* Preamble
*****************************************************************************/
-#include <vlc/vlc.h>
-#include "charset.h"
+#ifdef HAVE_CONFIG_H
+# include "config.h"
+#endif
+
+#include <vlc_common.h>
+#include <vlc_charset.h>
#include <assert.h>
#include <stdio.h>
#include <stdarg.h>
#include <stdlib.h>
-#include <errno.h>
#include <sys/types.h>
-#ifdef HAVE_DIRENT_H
-# include <dirent.h>
-#endif
#ifdef UNDER_CE
# include <tchar.h>
+#elif defined(WIN32)
+# include <io.h>
#endif
-#ifdef HAVE_SYS_STAT_H
-# include <sys/stat.h>
-#endif
-#ifdef HAVE_FCNTL_H
-# include <fcntl.h>
-#endif
-#ifdef WIN32
-# include <io.h>
-#else
-# include <unistd.h>
-#endif
-
-#ifndef HAVE_LSTAT
-# define lstat( a, b ) stat(a, b)
-#endif
-
-#ifdef __APPLE__
-/* Define this if the OS always use UTF-8 internally */
-# define ASSUME_UTF8 1
-#endif
-
-#ifndef ASSUME_UTF8
-# if defined (HAVE_ICONV)
-/* libiconv is more powerful than Win32 API (it has translit) */
-# define USE_ICONV 1
-# elif defined (WIN32) || defined (UNDER_CE)
-# define USE_MB2MB 1
-# else
-# error No UTF8 charset conversion implemented on this platform!
-# endif
-#endif
-
-typedef struct locale_data_t
-{
-#if defined (USE_ICONV)
- vlc_iconv_t hd;
- vlc_mutex_t lock;
-#elif defined (USE_MB2MB)
- UINT fromCP;
- UINT toCP;
-#endif
-} locale_data_t;
-
-static locale_data_t from_locale, to_locale;
-
-
-void LocaleInit( vlc_object_t *p_this )
-{
-#if defined USE_ICONV
- char *psz_charset;
-
- if( vlc_current_charset( &psz_charset ) )
- /* UTF-8 */
- from_locale.hd = to_locale.hd = (vlc_iconv_t)(-1);
- else
- {
- /* not UTF-8 */
- char psz_buf[strlen( psz_charset ) + sizeof( "//translit" )];
- const char *psz_conv;
-
- /*
- * Still allow non-ASCII characters when the locale is not set.
- * Western Europeans are being favored for historical reasons.
- */
- if( strcmp( psz_charset, "ASCII" ) )
- {
- sprintf( psz_buf, "%s//translit", psz_charset );
- psz_conv = psz_buf;
- }
- else
- psz_conv = "ISO-8859-1//translit";
-
- vlc_mutex_init( p_this, &from_locale.lock );
- vlc_mutex_init( p_this, &to_locale.lock );
- from_locale.hd = vlc_iconv_open( "UTF-8", psz_conv );
- to_locale.hd = vlc_iconv_open( psz_conv, "UTF-8" );
- }
-
- free( psz_charset );
-
- assert( (from_locale.hd == (vlc_iconv_t)(-1))
- == (to_locale.hd == (vlc_iconv_t)(-1)) );
-
-#elif defined (USE_MB2MB)
- to_locale.toCP = from_locale.fromCP = CP_ACP;
- from_locale.toCP = to_locale.fromCP = CP_UTF8;
-#else
- (void)p_this;
-#endif
-}
-
-void LocaleDeinit( void )
-{
-#ifdef USE_ICONV
- if( to_locale.hd != (vlc_iconv_t)(-1) )
- {
- vlc_iconv_close( to_locale.hd );
- vlc_mutex_destroy( &to_locale.lock );
- }
-
- if( from_locale.hd != (vlc_iconv_t)(-1) )
- {
- vlc_iconv_close( from_locale.hd );
- vlc_mutex_destroy( &from_locale.lock );
- }
-#endif
-}
-
-static char *locale_fast (const char *string, locale_data_t *p)
-{
-#if defined (USE_ICONV)
- vlc_iconv_t hd = p->hd;
-
- if (hd == (vlc_iconv_t)(-1))
- return (char *)string;
-
- const char *iptr = string;
- size_t inb = strlen (string);
- size_t outb = inb * 6 + 1;
- char output[outb], *optr = output;
-
- if (string == NULL)
- return NULL;
-
- vlc_mutex_lock (&p->lock);
- vlc_iconv (hd, NULL, NULL, NULL, NULL);
-
- while (vlc_iconv (hd, &iptr, &inb, &optr, &outb) == (size_t)(-1))
- {
- *optr++ = '?';
- outb--;
- iptr++;
- inb--;
- vlc_iconv (hd, NULL, NULL, NULL, NULL);
- }
- vlc_mutex_unlock (&p->lock);
- *optr = '\0';
-
- assert (inb == 0);
- assert (*iptr == '\0');
- assert (*optr == '\0');
- assert (strlen (output) == (size_t)(optr - output));
- return strdup (output);
-#elif defined (USE_MB2MB)
- char *out;
- wchar_t *wide;
- int len;
-
- if (string == NULL)
- return NULL;
-
- len = MultiByteToWideChar (p->fromCP, 0, string, -1, NULL, 0);
- if (len == 0)
- return NULL;
-
- wchar_t wide[len];
-
- MultiByteToWideChar (p->fromCP, 0, string, -1, wide, len);
- len = WideCharToMultiByte (p->toCP, 0, wide, -1, NULL, 0, NULL, NULL);
- if (len == 0)
- return NULL;
- out = malloc (len);
-
- WideCharToMultiByte (p->toCP, 0, wide, -1, out, len, NULL, NULL);
- return out;
-#else
- return (char *)string;
-#endif
-}
-
-
-static inline char *locale_dup (const char *string, locale_data_t *p)
-{
-#if defined (USE_ICONV)
- return (p->hd == (vlc_iconv_t)(-1))
- ? strdup (string)
- : locale_fast (string, p);
-#elif defined (USE_MB2MB)
- return locale_fast (string, p);
-#else
- return strdup (string);
-#endif
-}
-
+#include <errno.h>
+#include <wctype.h>
+/**
+ * Releases (if needed) a localized or uniformized string.
+ * @param str non-NULL return value from FromLocale() or ToLocale().
+ */
void LocaleFree (const char *str)
{
-#if defined (USE_ICONV)
- assert ((to_locale.hd == (vlc_iconv_t)(-1))
- == (from_locale.hd == (vlc_iconv_t)(-1)));
-
- if( to_locale.hd != (vlc_iconv_t)(-1) )
- free ((char *)str);
-#elif defined (USE_MB2MB)
+#ifdef ASSUME_UTF8
+ (void) str;
+#else
free ((char *)str);
#endif
}
/**
- * FromLocale: converts a locale string to UTF-8
+ * Converts a string from the system locale character encoding to UTF-8.
*
- * @param locale nul-terminated string to be converted
+ * @param locale nul-terminated string to convert
*
* @return a nul-terminated UTF-8 string, or NULL in case of error.
* To avoid memory leak, you have to pass the result to LocaleFree()
*/
char *FromLocale (const char *locale)
{
- return locale_fast (locale, &from_locale);
+#ifdef ASSUME_UTF8
+ return (char *)locale;
+#else
+ return locale ? FromCharset ("", locale, strlen(locale)) : NULL;
+#endif
}
+/**
+ * converts a string from the system locale character encoding to utf-8,
+ * the result is always allocated on the heap.
+ *
+ * @param locale nul-terminated string to convert
+ *
+ * @return a nul-terminated utf-8 string, or null in case of error.
+ * The result must be freed using free() - as with the strdup() function.
+ */
char *FromLocaleDup (const char *locale)
{
- return locale_dup (locale, &from_locale);
+#ifdef ASSUME_UTF8
+ return strdup (locale);
+#else
+ return FromCharset ("", locale, strlen(locale));
+#endif
}
/**
- * ToLocale: converts a UTF-8 string to local system encoding.
+ * ToLocale: converts an UTF-8 string to local system encoding.
*
* @param utf8 nul-terminated string to be converted
*
*/
char *ToLocale (const char *utf8)
{
- return locale_fast (utf8, &to_locale);
-}
-
-
-static char *ToLocaleDup (const char *utf8)
-{
- return locale_dup (utf8, &to_locale);
+#ifdef ASSUME_UTF8
+ return (char *)utf8;
+#else
+ size_t outsize;
+ return utf8 ? ToCharset ("", utf8, &outsize) : NULL;
+#endif
}
/**
- * utf8_open: open() wrapper for UTF-8 filenames
+ * converts a string from UTF-8 to the system locale character encoding,
+ * the result is always allocated on the heap.
+ *
+ * @param utf8 nul-terminated string to convert
+ *
+ * @return a nul-terminated string, or null in case of error.
+ * The result must be freed using free() - as with the strdup() function.
*/
-int utf8_open (const char *filename, int flags, mode_t mode)
+char *ToLocaleDup (const char *utf8)
{
-#if defined (WIN32) || defined (UNDER_CE)
- if (GetVersion() < 0x80000000)
- {
- /* for Windows NT and above */
- wchar_t wpath[MAX_PATH + 1];
-
- if (!MultiByteToWideChar (CP_UTF8, 0, filename, -1, wpath, MAX_PATH))
- {
- errno = ENOENT;
- return -1;
- }
- wpath[MAX_PATH] = L'\0';
-
- /*
- * open() cannot open files with non-“ANSI” characters on Windows.
- * We use _wopen() instead. Same thing for mkdir() and stat().
- */
- return _wopen (wpath, flags, mode);
- }
+#ifdef ASSUME_UTF8
+ return strdup (utf8);
+#else
+ size_t outsize;
+ return ToCharset ("", utf8, &outsize);
#endif
- const char *local_name = ToLocale (filename);
-
- if (local_name == NULL)
- {
- errno = ENOENT;
- return -1;
- }
-
- int fd = open (local_name, flags, mode);
- LocaleFree (local_name);
- return fd;
}
/**
- * utf8_fopen: fopen() wrapper for UTF-8 filenames
+ * Formats an UTF-8 string as vfprintf(), then print it, with
+ * appropriate conversion to local encoding.
*/
-FILE *utf8_fopen (const char *filename, const char *mode)
+int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap )
{
- int rwflags = 0, oflags = 0;
- vlc_bool_t append = VLC_FALSE;
+#ifdef ASSUME_UTF8
+ return vfprintf (stream, fmt, ap);
+#else
+ char *str;
+ int res;
- for (const char *ptr = mode; *ptr; ptr++)
+# if defined( WIN32 ) && !defined( UNDER_CE )
+ /* Writing to the console is a lot of fun on Microsoft Windows.
+ * If you use the standard I/O functions, you must use the OEM code page,
+ * which is different from the usual ANSI code page. Or maybe not, if the
+ * user called "chcp". Anyway, we prefer Unicode. */
+ int fd = _fileno (stream);
+ if (likely(fd != -1) && _isatty (fd))
{
- switch (*ptr)
- {
- case 'r':
- rwflags = O_RDONLY;
- break;
-
- case 'a':
- rwflags = O_WRONLY;
- oflags |= O_CREAT;
- append = VLC_TRUE;
- break;
+ res = vasprintf (&str, fmt, ap);
+ if (unlikely(res == -1))
+ return -1;
- case 'w':
- rwflags = O_WRONLY;
- oflags |= O_CREAT | O_TRUNC;
- break;
+ size_t wlen = 2 * (res + 1);
+ wchar_t *wide = malloc (wlen);
+ if (likely(wide != NULL))
+ {
+ wlen = MultiByteToWideChar (CP_UTF8, 0, str, res + 1, wide, wlen);
+ if (wlen > 0)
+ {
+ HANDLE h = (HANDLE)(intptr_t)_get_osfhandle (fd);
+ DWORD out;
- case '+':
- rwflags = O_RDWR;
- break;
+ WriteConsoleW (h, wide, wlen - 1, &out, NULL);
+ }
+ else
+ res = -1;
+ free (wide);
}
- }
-
- int fd = utf8_open (filename, rwflags | oflags, 0666);
- if (fd == -1)
- return NULL;
-
- if (append && (lseek (fd, 0, SEEK_END) == -1))
- {
- close (fd);
- return NULL;
- }
-
- FILE *stream = fdopen (fd, mode);
- if (stream == NULL)
- close (fd);
-
- return stream;
-}
-
-/**
- * utf8_mkdir: Calls mkdir() after conversion of file name to OS locale
- */
-int utf8_mkdir( const char *dirname )
-{
-#if defined (UNDER_CE) || defined (WIN32)
- wchar_t wname[MAX_PATH + 1];
- char mod[MAX_PATH + 1];
- int i;
-
- /* Convert '/' into '\' */
- for( i = 0; *dirname; i++ )
- {
- if( i == MAX_PATH )
- return -1; /* overflow */
-
- if( *dirname == '/' )
- mod[i] = '\\';
else
- mod[i] = *dirname;
- dirname++;
-
+ res = -1;
+ free (str);
+ return res;
}
- mod[i] = 0;
+# endif
- if( MultiByteToWideChar( CP_UTF8, 0, mod, -1, wname, MAX_PATH ) == 0 )
- {
- errno = ENOENT;
+ res = vasprintf (&str, fmt, ap);
+ if (unlikely(res == -1))
return -1;
- }
- wname[MAX_PATH] = L'\0';
- if( CreateDirectoryW( wname, NULL ) == 0 )
- {
- if( GetLastError( ) == ERROR_ALREADY_EXISTS )
- errno = EEXIST;
- else
- errno = ENOENT;
- return -1;
- }
- return 0;
-#else
- char *locname = ToLocale( dirname );
- int res;
+ char *ansi = ToLocaleDup (str);
+ free (str);
- if( locname == NULL )
- {
- errno = ENOENT;
+ if (ansi == NULL)
return -1;
- }
- res = mkdir( locname, 0755 );
-
- LocaleFree( locname );
+ fputs (ansi, stream);
+ free (ansi);
return res;
#endif
}
-
-DIR *utf8_opendir( const char *dirname )
-{
- /* TODO: support for WinNT non-ACP filenames */
- const char *local_name = ToLocale( dirname );
-
- if( local_name != NULL )
- {
- DIR *dir = opendir( local_name );
- LocaleFree( local_name );
- return dir;
- }
- else
- errno = ENOENT;
- return NULL;
-}
-
-
-char *utf8_readdir( void *dir )
+/**
+ * Formats an UTF-8 string as fprintf(), then print it, with
+ * appropriate conversion to local encoding.
+ */
+int utf8_fprintf( FILE *stream, const char *fmt, ... )
{
- struct dirent *ent;
-
- ent = readdir( (DIR *)dir );
- if( ent == NULL )
- return NULL;
+ va_list ap;
+ int res;
- return vlc_fix_readdir( ent->d_name );
+ va_start( ap, fmt );
+ res = utf8_vfprintf( stream, fmt, ap );
+ va_end( ap );
+ return res;
}
-static int dummy_select( const char *str )
-{
- (void)str;
- return 1;
-}
-int utf8_scandir( const char *dirname, char ***namelist,
- int (*select)( const char * ),
- int (*compar)( const char **, const char ** ) )
+/**
+ * Converts the first character from a UTF-8 sequence into a code point.
+ *
+ * @param str an UTF-8 bytes sequence
+ * @return 0 if str points to an empty string, i.e. the first character is NUL;
+ * number of bytes that the first character occupies (from 1 to 4) otherwise;
+ * -1 if the byte sequence was not a valid UTF-8 sequence.
+ */
+static size_t vlc_towc (const char *str, uint32_t *restrict pwc)
{
- DIR *dir = utf8_opendir( dirname );
+ uint8_t *ptr = (uint8_t *)str;
+ assert (str != NULL);
- if( select == NULL )
- select = dummy_select;
+ uint8_t c = ptr[0];
- if( dir == NULL )
- return -1;
- else
+ if (unlikely(c == '\0'))
{
- char **tab = NULL;
- char *entry;
- unsigned num = 0;
-
- while( ( entry = utf8_readdir( dir ) ) != NULL )
- {
- char **newtab;
-
- if( !select( entry ) )
- {
- free( entry );
- continue;
- }
-
- newtab = realloc( tab, sizeof( char * ) * (num + 1) );
- if( newtab == NULL )
- {
- free( entry );
- goto error;
- }
- tab = newtab;
- tab[num++] = entry;
- }
- closedir( dir );
-
- if( compar != NULL )
- qsort( tab, num, sizeof( tab[0] ),
- (int (*)( const void *, const void *))compar );
+ *pwc = 0;
+ return 0;
+ }
- *namelist = tab;
- return num;
+ if (unlikely(c > 0xF4))
+ return -1;
- error:{
- unsigned i;
+ int charlen = clz8 (c ^ 0xFF);
+ switch (charlen)
+ {
+ case 0: // 7-bit ASCII character -> OK
+ *pwc = c;
+ return 1;
- for( i = 0; i < num; i++ )
- free( tab[i] );
- if( tab != NULL )
- free( tab );
- return -1;}
+ case 1: // continuation byte -> error
+ return -1;
}
-}
+ assert (charlen >= 2 && charlen <= 4);
-static int utf8_statEx( const char *filename, struct stat *buf,
- vlc_bool_t deref )
-{
-#if defined (WIN32) || defined (UNDER_CE)
- /* retrieve Windows OS version */
- if( GetVersion() < 0x80000000 )
+ uint32_t cp = c & ~((0xff >> (7 - charlen)) << (7 - charlen));
+ for (int i = 1; i < charlen; i++)
{
- /* for Windows NT and above */
- wchar_t wpath[MAX_PATH + 1];
+ assert (cp < (1 << 26));
+ c = ptr[i];
- if( !MultiByteToWideChar( CP_UTF8, 0, filename, -1, wpath, MAX_PATH ) )
- {
- errno = ENOENT;
+ if (unlikely((c >> 6) != 2)) // not a continuation byte
return -1;
- }
- wpath[MAX_PATH] = L'\0';
- return _wstati64( wpath, buf );
+ cp = (cp << 6) | (ptr[i] & 0x3f);
}
-#endif
-#ifdef HAVE_SYS_STAT_H
- const char *local_name = ToLocale( filename );
- if( local_name != NULL )
+ switch (charlen)
{
- int res = deref ? stat( local_name, buf )
- : lstat( local_name, buf );
- LocaleFree( local_name );
- return res;
+ case 4:
+ if (unlikely(cp > 0x10FFFF)) // beyond Unicode
+ return -1;
+ case 3:
+ if (unlikely(cp >= 0xD800 && cp < 0xC000)) // UTF-16 surrogate
+ return -1;
+ case 2:
+ if (unlikely(cp < 128)) // ASCII overlong
+ return -1;
+ if (unlikely(cp < (1u << (5 * charlen - 3)))) // overlong
+ return -1;
}
- errno = ENOENT;
-#endif
- return -1;
-}
-
-
-int utf8_stat( const char *filename, struct stat *buf)
-{
- return utf8_statEx( filename, buf, VLC_TRUE );
-}
-
-int utf8_lstat( const char *filename, struct stat *buf)
-{
- return utf8_statEx( filename, buf, VLC_FALSE );
+ *pwc = cp;
+ return charlen;
}
/**
- * utf8_*printf: *printf with conversion from UTF-8 to local encoding
+ * Look for an UTF-8 string within another one in a case-insensitive fashion.
+ * Beware that this is quite slow. Contrary to strcasestr(), this function
+ * works regardless of the system character encoding, and handles multibyte
+ * code points correctly.
+
+ * @param haystack string to look into
+ * @param needle string to look for
+ * @return a pointer to the first occurence of the needle within the haystack,
+ * or NULL if no occurence were found.
*/
-static int utf8_vasprintf( char **str, const char *fmt, va_list ap )
-{
- char *utf8;
- int res = vasprintf( &utf8, fmt, ap );
- if( res == -1 )
- return -1;
-
- *str = ToLocaleDup( utf8 );
- free( utf8 );
- return res;
-}
-
-int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap )
-{
- char *str;
- int res = utf8_vasprintf( &str, fmt, ap );
- if( res == -1 )
- return -1;
-
- fputs( str, stream );
- free( str );
- return res;
-}
-
-int utf8_fprintf( FILE *stream, const char *fmt, ... )
+char *vlc_strcasestr (const char *haystack, const char *needle)
{
- va_list ap;
- int res;
-
- va_start( ap, fmt );
- res = utf8_vfprintf( stream, fmt, ap );
- va_end( ap );
- return res;
-}
+ ssize_t s;
+ do
+ {
+ const char *h = haystack, *n = needle;
-static char *CheckUTF8( char *str, char rep )
-#define isutf8cont( c ) (((c) >= 0x80) && ((c) <= 0xBF))
-{
- unsigned char *ptr, c;
+ for (;;)
+ {
+ uint32_t cph, cpn;
- assert (str != NULL);
+ s = vlc_towc (n, &cpn);
+ if (s == 0)
+ return (char *)haystack;
+ if (unlikely(s < 0))
+ return NULL;
+ n += s;
- ptr = (unsigned char *)str;
- while( (c = *ptr) != '\0' )
- {
- /* US-ASCII, 1 byte */
- if( c <= 0x7F )
- ptr++; /* OK */
- else
- /* 2 bytes */
- if( ( c >= 0xC2 ) && ( c <= 0xDF ) )
- {
- c = ptr[1];
- if( isutf8cont( c ) )
- ptr += 2; /* OK */
- else
- goto error;
- }
- else
- /* 3 bytes */
- if( c == 0xE0 )
- {
- c = ptr[1];
- if( ( c >= 0xA0 ) && ( c <= 0xBF ) )
- {
- c = ptr[2];
- if( isutf8cont( c ) )
- ptr += 3; /* OK */
- else
- goto error;
- }
- else
- goto error;
- }
- else
- if( ( ( c >= 0xE1 ) && ( c <= 0xEC ) ) || ( c == 0xEC )
- || ( c == 0xEE ) || ( c == 0xEF ) )
- {
- c = ptr[1];
- if( isutf8cont( c ) )
- {
- c = ptr[2];
- if( isutf8cont( c ) )
- ptr += 3; /* OK */
- else
- goto error;
- }
- else
- goto error;
- }
- else
- if( c == 0xED )
- {
- c = ptr[1];
- if( ( c >= 0x80 ) && ( c <= 0x9F ) )
- {
- c = ptr[2];
- if( isutf8cont( c ) )
- ptr += 3; /* OK */
- else
- goto error;
- }
- else
- goto error;
- }
- else
- /* 4 bytes */
- if( c == 0xF0 )
- {
- c = ptr[1];
- if( ( c >= 0x90 ) && ( c <= 0xBF ) )
- {
- c = ptr[2];
- if( isutf8cont( c ) )
- {
- c = ptr[3];
- if( isutf8cont( c ) )
- ptr += 4; /* OK */
- else
- goto error;
- }
- else
- goto error;
- }
- else
- goto error;
- }
- else
- if( ( c >= 0xF1 ) && ( c <= 0xF3 ) )
- {
- c = ptr[1];
- if( isutf8cont( c ) )
- {
- c = ptr[2];
- if( isutf8cont( c ) )
- {
- c = ptr[3];
- if( isutf8cont( c ) )
- ptr += 4; /* OK */
- goto error;
- }
- else
- goto error;
- }
- else
- goto error;
- }
- else
- if( c == 0xF4 )
- {
- c = ptr[1];
- if( ( c >= 0x80 ) && ( c <= 0x8F ) )
- {
- c = ptr[2];
- if( isutf8cont( c ) )
- {
- c = ptr[3];
- if( isutf8cont( c ) )
- ptr += 4; /* OK */
- else
- goto error;
- }
- else
- goto error;
- }
- else
- goto error;
+ s = vlc_towc (h, &cph);
+ if (s <= 0 || towlower (cph) != towlower (cpn))
+ break;
+ h += s;
}
- else
- goto error;
-
- continue;
-error:
- if( rep == 0 )
- return NULL;
- *ptr++ = '?';
- str = NULL;
+ s = vlc_towc (haystack, &(uint32_t) { 0 });
+ haystack += s;
}
+ while (s != 0);
- return str;
+ return NULL;
}
/**
- * EnsureUTF8: replaces invalid/overlong UTF-8 sequences with question marks
+ * Replaces invalid/overlong UTF-8 sequences with question marks.
* Note that it is not possible to convert from Latin-1 to UTF-8 on the fly,
* so we don't try that, even though it would be less disruptive.
*
*/
char *EnsureUTF8( char *str )
{
- return CheckUTF8( str, '?' );
+ char *ret = str;
+ size_t n;
+ uint32_t cp;
+
+ while ((n = vlc_towc (str, &cp)) != 0)
+ if (likely(n != (size_t)-1))
+ str += n;
+ else
+ {
+ *str++ = '?';
+ ret = NULL;
+ }
+ return ret;
}
/**
- * IsUTF8: checks whether a string is a valid UTF-8 byte sequence.
+ * Checks whether a string is a valid UTF-8 byte sequence.
*
* @param str nul-terminated string to be checked
*
*/
const char *IsUTF8( const char *str )
{
- return CheckUTF8( (char *)str, 0 );
+ size_t n;
+ uint32_t cp;
+
+ while ((n = vlc_towc (str, &cp)) != 0)
+ if (likely(n != (size_t)-1))
+ str += n;
+ else
+ return NULL;
+ return str;
+}
+
+/**
+ * Converts a string from the given character encoding to utf-8.
+ *
+ * @return a nul-terminated utf-8 string, or null in case of error.
+ * The result must be freed using free().
+ */
+char *FromCharset(const char *charset, const void *data, size_t data_size)
+{
+ vlc_iconv_t handle = vlc_iconv_open ("UTF-8", charset);
+ if (handle == (vlc_iconv_t)(-1))
+ return NULL;
+
+ char *out = NULL;
+ for(unsigned mul = 4; mul < 8; mul++ )
+ {
+ size_t in_size = data_size;
+ const char *in = data;
+ size_t out_max = mul * data_size;
+ char *tmp = out = malloc (1 + out_max);
+ if (!out)
+ break;
+
+ if (vlc_iconv (handle, &in, &in_size, &tmp, &out_max) != (size_t)(-1)) {
+ *tmp = '\0';
+ break;
+ }
+ free(out);
+ out = NULL;
+
+ if (errno != E2BIG)
+ break;
+ }
+ vlc_iconv_close(handle);
+ return out;
+}
+
+/**
+ * Converts a nul-terminated UTF-8 string to a given character encoding.
+ * @param charset iconv name of the character set
+ * @param in nul-terminated UTF-8 string
+ * @param outsize pointer to hold the byte size of result
+ *
+ * @return A pointer to the result, which must be released using free().
+ * The UTF-8 nul terminator is included in the conversion if the target
+ * character encoding supports it. However it is not included in the returned
+ * byte size.
+ * In case of error, NULL is returned and the byte size is undefined.
+ */
+void *ToCharset(const char *charset, const char *in, size_t *outsize)
+{
+ vlc_iconv_t hd = vlc_iconv_open (charset, "UTF-8");
+ if (hd == (vlc_iconv_t)(-1))
+ return NULL;
+
+ const size_t inlen = strlen (in);
+ void *res;
+
+ for (unsigned mul = 4; mul < 16; mul++)
+ {
+ size_t outlen = mul * (inlen + 1);
+ res = malloc (outlen);
+ if (unlikely(res == NULL))
+ break;
+
+ const char *inp = in;
+ char *outp = res;
+ size_t inb = inlen;
+ size_t outb = outlen;
+
+ if (vlc_iconv (hd, &inp, &inb, &outp, &outb) != (size_t)(-1))
+ {
+ *outsize = outlen - outb;
+ inb = 1; /* append nul terminator if possible */
+ if (vlc_iconv (hd, &inp, &inb, &outp, &outb) != (size_t)(-1))
+ break;
+ if (errno == EILSEQ) /* cannot translate nul terminator!? */
+ break;
+ }
+
+ free (res);
+ res = NULL;
+ if (errno != E2BIG) /* conversion failure */
+ break;
+ }
+ vlc_iconv_close (hd);
+ return res;
}
+