* unicode.c: Unicode <-> locale functions
*****************************************************************************
* Copyright (C) 2005-2006 the VideoLAN team
- * Copyright © 2005-2006 Rémi Denis-Courmont
- * $Id$
+ * Copyright © 2005-2008 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 "libvlc.h" /* utf8_mkdir */
#include <assert.h>
# 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
+#if defined (ASSUME_UTF8)
+/* Cool */
+#elif defined (WIN32) || defined (UNDER_CE)
+# define USE_MB2MB 1
+#elif defined (HAVE_ICONV)
+# define USE_ICONV 1
+#else
+# error No UTF8 charset conversion implemented on this platform!
#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;
+# include <langinfo.h>
+static char charset[sizeof ("CSISO11SWEDISHFORNAMES")] = "";
-
-void LocaleInit( vlc_object_t *p_this )
+static void find_charset_once (void)
{
-#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
+ strlcpy (charset, nl_langinfo (CODESET), sizeof (charset));
+ if (!strcasecmp (charset, "ASCII")
+ || !strcasecmp (charset, "ANSI_X3.4-1968"))
+ strcpy (charset, "UTF-8"); /* superset... */
}
-void LocaleDeinit( void )
+static int find_charset (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 pthread_once_t once = PTHREAD_ONCE_INIT;
+ pthread_once (&once, find_charset_once);
+ return !strcasecmp (charset, "UTF-8");
}
+#endif
+
-static char *locale_fast (const char *string, locale_data_t *p)
+static char *locale_fast (const char *string, bool from)
{
#if defined (USE_ICONV)
- vlc_iconv_t hd = p->hd;
+ if (find_charset ())
+ return (char *)string;
+ vlc_iconv_t hd = vlc_iconv_open (from ? "UTF-8" : charset,
+ from ? charset : "UTF-8");
if (hd == (vlc_iconv_t)(-1))
- return (char *)string;
+ return NULL; /* Uho! */
const char *iptr = string;
size_t inb = strlen (string);
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_iconv (hd, NULL, NULL, NULL, NULL); /* reset */
}
- vlc_mutex_unlock (&p->lock);
*optr = '\0';
+ vlc_iconv_close (hd);
assert (inb == 0);
assert (*iptr == '\0');
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;
-
+ len = 1 + MultiByteToWideChar (from ? CP_ACP : CP_UTF8,
+ 0, string, -1, NULL, 0);
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;
+ MultiByteToWideChar (from ? CP_ACP : CP_UTF8, 0, string, -1, wide, len);
+ len = 1 + WideCharToMultiByte (from ? CP_UTF8 : CP_ACP, 0, wide, -1,
+ NULL, 0, NULL, NULL);
out = malloc (len);
+ if (out == NULL)
+ return NULL;
- WideCharToMultiByte (p->toCP, 0, wide, -1, out, len, NULL, NULL);
+ WideCharToMultiByte (from ? CP_UTF8 : CP_ACP, 0, wide, -1, out, len,
+ NULL, NULL);
return out;
#else
+ (void)from;
return (char *)string;
#endif
}
-static inline char *locale_dup (const char *string, locale_data_t *p)
+static inline char *locale_dup (const char *string, bool from)
{
+ assert( string );
+
#if defined (USE_ICONV)
- return (p->hd == (vlc_iconv_t)(-1))
- ? strdup (string)
- : locale_fast (string, p);
+ if (find_charset ())
+ return strdup (string);
+ return locale_fast (string, from);
#elif defined (USE_MB2MB)
- return locale_fast (string, p);
+ return locale_fast (string, from);
#else
+ (void)from;
return strdup (string);
#endif
}
-
+/**
+ * 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) )
+ if (!find_charset ())
free ((char *)str);
#elif defined (USE_MB2MB)
free ((char *)str);
+#else
+ (void)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);
+ return locale_fast (locale, true);
}
+/**
+ * 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);
+ return locale_dup (locale, true);
}
/**
- * 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);
+ return locale_fast (utf8, false);
}
-static char *ToLocaleDup (const char *utf8)
+/**
+ * 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.
+ */
+char *ToLocaleDup (const char *utf8)
{
- return locale_dup (utf8, &to_locale);
+ return locale_dup (utf8, false);
}
/**
- * utf8_open: open() wrapper for UTF-8 filenames
+ * Opens a system file handle using UTF-8 paths.
+ *
+ * @param filename file path to open (with UTF-8 encoding)
+ * @param flags open() flags, see the C library open() documentation
+ * @param mode file permissions if creating a new file
+ * @return a file handle on success, -1 on error (see errno).
*/
int utf8_open (const char *filename, int flags, mode_t mode)
{
}
/**
- * utf8_fopen: fopen() wrapper for UTF-8 filenames
+ * Opens a FILE pointer using UTF-8 filenames.
+ * @param filename file path, using UTF-8 encoding
+ * @param mode fopen file open mode
+ * @return NULL on error, an open FILE pointer on success.
*/
FILE *utf8_fopen (const char *filename, const char *mode)
{
int rwflags = 0, oflags = 0;
- vlc_bool_t append = VLC_FALSE;
+ bool append = false;
for (const char *ptr = mode; *ptr; ptr++)
{
case 'a':
rwflags = O_WRONLY;
oflags |= O_CREAT;
- append = VLC_TRUE;
+ append = true;
break;
case 'w':
case '+':
rwflags = O_RDWR;
break;
+
+#ifdef O_TEXT
+ case 't':
+ oflags |= O_TEXT;
+ break;
+#endif
}
}
}
/**
- * utf8_mkdir: Calls mkdir() after conversion of file name to OS locale
+ * Creates a directory using UTF-8 paths.
+ *
+ * @param dirname a UTF-8 string with the name of the directory that you
+ * want to create.
+ * @param mode directory permissions
+ * @return 0 on success, -1 on error (see errno).
*/
-int utf8_mkdir( const char *dirname )
+int utf8_mkdir( const char *dirname, mode_t mode )
{
#if defined (UNDER_CE) || defined (WIN32)
+ VLC_UNUSED( mode );
+
wchar_t wname[MAX_PATH + 1];
char mod[MAX_PATH + 1];
int i;
errno = ENOENT;
return -1;
}
- res = mkdir( locname, 0755 );
+ res = mkdir( locname, mode );
LocaleFree( locname );
return res;
#endif
}
-
+/**
+ * Opens a DIR pointer using UTF-8 paths
+ *
+ * @param dirname UTF-8 representation of the directory name
+ * @return a pointer to the DIR struct, or NULL in case of error.
+ * Release with standard closedir().
+ */
DIR *utf8_opendir( const char *dirname )
{
- /* TODO: support for WinNT non-ACP filenames */
+#ifdef WIN32
+ wchar_t wname[MAX_PATH + 1];
+
+ if (MultiByteToWideChar (CP_UTF8, 0, dirname, -1, wname, MAX_PATH))
+ {
+ wname[MAX_PATH] = L'\0';
+ return (DIR *)vlc_wopendir (wname);
+ }
+#else
const char *local_name = ToLocale( dirname );
if( local_name != NULL )
{
- DIR *dir = vlc_opendir_wrapper( local_name );
+ DIR *dir = opendir( local_name );
LocaleFree( local_name );
return dir;
}
- else
- errno = ENOENT;
+#endif
+
+ errno = ENOENT;
return NULL;
}
-
-char *utf8_readdir( void *dir )
+/**
+ * Reads the next file name from an open directory.
+ *
+ * @param dir The directory that is being read
+ *
+ * @return a UTF-8 string of the directory entry.
+ * Use free() to free this memory.
+ */
+char *utf8_readdir( DIR *dir )
{
+#ifdef WIN32
+ struct _wdirent *ent = vlc_wreaddir (dir);
+ if (ent == NULL)
+ return NULL;
+
+ return FromWide (ent->d_name);
+#else
struct dirent *ent;
- ent = vlc_readdir_wrapper( (DIR *)dir );
+ ent = readdir( (DIR *)dir );
if( ent == NULL )
return NULL;
return vlc_fix_readdir( ent->d_name );
+#endif
}
static int dummy_select( const char *str )
return 1;
}
-int utf8_scandir( const char *dirname, char ***namelist,
+/**
+ * Does the same as utf8_scandir(), but takes an open directory pointer
+ * instead of a directory path.
+ */
+int utf8_loaddir( DIR *dir, char ***namelist,
int (*select)( const char * ),
int (*compar)( const char **, const char ** ) )
{
- DIR *dir = utf8_opendir( dirname );
-
if( select == NULL )
select = dummy_select;
char *entry;
unsigned num = 0;
+ rewinddir( dir );
+
while( ( entry = utf8_readdir( dir ) ) != NULL )
{
char **newtab;
tab = newtab;
tab[num++] = entry;
}
- vlc_closedir_wrapper( dir );
if( compar != NULL )
qsort( tab, num, sizeof( tab[0] ),
free( tab[i] );
if( tab != NULL )
free( tab );
- return -1;}
+ }
}
+ return -1;
}
+/**
+ * Selects file entries from a directory, as GNU C scandir(), yet using
+ * UTF-8 file names.
+ *
+ * @param dirname UTF-8 diretory path
+ * @param pointer [OUT] pointer set, on succesful completion, to the address
+ * of a table of UTF-8 filenames. All filenames must be freed with free().
+ * The table itself must be freed with free() as well.
+ *
+ * @return How many file names were selected (possibly 0),
+ * or -1 in case of error.
+ */
+int utf8_scandir( const char *dirname, char ***namelist,
+ int (*select)( const char * ),
+ int (*compar)( const char **, const char ** ) )
+{
+ DIR *dir = utf8_opendir (dirname);
+ int val = -1;
+
+ if (dir != NULL)
+ {
+ val = utf8_loaddir (dir, namelist, select, compar);
+ closedir (dir);
+ }
+ return val;
+}
static int utf8_statEx( const char *filename, struct stat *buf,
- vlc_bool_t deref )
+ bool deref )
{
#if defined (WIN32) || defined (UNDER_CE)
/* retrieve Windows OS version */
return -1;
}
-
+/**
+ * Finds file/inode informations, as stat().
+ * Consider usign fstat() instead, if possible.
+ *
+ * @param filename UTF-8 file path
+ */
int utf8_stat( const char *filename, struct stat *buf)
{
- return utf8_statEx( filename, buf, VLC_TRUE );
+ return utf8_statEx( filename, buf, true );
}
+/**
+ * Finds file/inode informations, as lstat().
+ * Consider usign fstat() instead, if possible.
+ *
+ * @param filename UTF-8 file path
+ */
int utf8_lstat( const char *filename, struct stat *buf)
{
- return utf8_statEx( filename, buf, VLC_FALSE );
+ return utf8_statEx( filename, buf, false );
}
/**
- * utf8_*printf: *printf with conversion from UTF-8 to local encoding
+ * Removes a file.
+ *
+ * @param filename a UTF-8 string with the name of the file you want to delete.
+ * @return A 0 return value indicates success. A -1 return value indicates an
+ * error, and an error code is stored in errno
+ */
+int utf8_unlink( const char *filename )
+{
+#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';
+
+ /*
+ * unlink() cannot open files with non-“ANSI” characters on Windows.
+ * We use _wunlink() instead.
+ */
+ return _wunlink( wpath );
+ }
+#endif
+ const char *local_name = ToLocale( filename );
+
+ if( local_name == NULL )
+ {
+ errno = ENOENT;
+ return -1;
+ }
+
+ int ret = unlink( local_name );
+ LocaleFree( local_name );
+ return ret;
+}
+
+
+
+/**
+ * Formats an UTF-8 string as vasprintf(), then print it to stdout, with
+ * appropriate conversion to local encoding.
*/
static int utf8_vasprintf( char **str, const char *fmt, va_list ap )
{
return res;
}
+/**
+ * Formats an UTF-8 string as vfprintf(), then print it, with
+ * appropriate conversion to local encoding.
+ */
int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap )
{
char *str;
return res;
}
+/**
+ * Formats an UTF-8 string as fprintf(), then print it, with
+ * appropriate conversion to local encoding.
+ */
int utf8_fprintf( FILE *stream, const char *fmt, ... )
{
va_list ap;
static char *CheckUTF8( char *str, char rep )
-#define isutf8cont( c ) (((c) >= 0x80) && ((c) <= 0xBF))
{
- unsigned char *ptr, c;
-
+ uint8_t *ptr = (uint8_t *)str;
assert (str != NULL);
- ptr = (unsigned char *)str;
- while( (c = *ptr) != '\0' )
+ for (;;)
{
- /* 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 ) )
+ uint8_t c = ptr[0];
+ int charlen = -1;
+
+ if (c == '\0')
+ break;
+
+ for (int i = 0; i < 7; i++)
+ if ((c >> (7 - i)) == ((0xff >> (7 - i)) ^ 1))
{
- c = ptr[2];
- if( isutf8cont( c ) )
- {
- c = ptr[3];
- if( isutf8cont( c ) )
- ptr += 4; /* OK */
- else
- goto error;
- }
- else
- goto error;
+ charlen = i;
+ break;
}
- else
- goto error;
- }
- else
- if( ( c >= 0xF1 ) && ( c <= 0xF3 ) )
+
+ switch (charlen)
{
- 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
+ case 0: // 7-bit ASCII character -> OK
+ ptr++;
+ continue;
+
+ case -1: // 1111111x -> error
+ case 1: // continuation byte -> error
goto error;
}
- else
- if( c == 0xF4 )
+
+ assert (charlen >= 2);
+
+ uint32_t cp = c & ~((0xff >> (7 - charlen)) << (7 - charlen));
+ for (int i = 1; i < charlen; i++)
{
- 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
+ assert (cp < (1 << 26));
+ c = ptr[i];
+
+ if ((c == '\0') // unexpected end of string
+ || ((c >> 6) != 2)) // not a continuation byte
goto error;
+
+ cp = (cp << 6) | (ptr[i] & 0x3f);
}
- else
+
+ if (cp < 128) // overlong (special case for ASCII)
+ goto error;
+ if (cp < (1u << (5 * charlen - 3))) // overlong
goto error;
+ ptr += charlen;
continue;
-error:
- if( rep == 0 )
+ error:
+ if (rep == 0)
return NULL;
- *ptr++ = '?';
+ *ptr++ = rep;
str = 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.
*
/**
- * 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
*
projects / vlc / blobdiff