* 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>
*
# include "config.h"
#endif
-#include <vlc/vlc.h>
+#include <vlc_common.h>
#include <vlc_charset.h>
-#include "libvlc.h" /* utf8_mkdir */
#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
-
-#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
-
-#if defined (USE_ICONV)
-static char charset[sizeof ("CSISO11SWEDISHFORNAMES//translit")] = "";
-
-static void find_charset_once (void)
-{
- char *psz_charset;
- if (vlc_current_charset (&psz_charset)
- || (psz_charset == NULL)
- || (strcmp (psz_charset, "ASCII") == 0)
- || ((size_t)snprintf (charset, sizeof (charset), "%s//translit",
- psz_charset) >= sizeof (charset)))
- strcpy (charset, "UTF-8");
-
- free (psz_charset);
-}
-
-static int find_charset (void)
-{
- static pthread_once_t once = PTHREAD_ONCE_INIT;
- pthread_once (&once, find_charset_once);
- return !strcmp (charset, "UTF-8");
-}
-#endif
-
-
-static char *locale_fast (const char *string, bool from)
-{
-#if defined (USE_ICONV)
- 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 strdup (string); /* Uho! */
-
- 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;
-
- while (vlc_iconv (hd, &iptr, &inb, &optr, &outb) == (size_t)(-1))
- {
- *optr++ = '?';
- outb--;
- iptr++;
- inb--;
- vlc_iconv (hd, NULL, NULL, NULL, NULL);
- }
- *optr = '\0';
- vlc_iconv_close (hd);
-
- 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;
- int len;
-
- if (string == NULL)
- return NULL;
-
- len = 1 + MultiByteToWideChar (from ? CP_ACP : CP_UTF8,
- 0, string, -1, NULL, 0);
- wchar_t wide[len];
-
- 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 (from ? CP_UTF8 : CP_ACP, 0, wide, -1, out, len, NULL, NULL);
- return out;
-#else
- VLC_UNUSED(from);
- return (char *)string;
-#endif
-}
-
-
-static inline char *locale_dup (const char *string, bool from)
-{
-#if defined (USE_ICONV)
- if (find_charset ())
- return strdup (string);
- return locale_fast (string, from);
-#elif defined (USE_MB2MB)
- return locale_fast (string, from);
-#else
- VLC_UNUSED(from);
- 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)
- if (!find_charset ())
- free ((char *)str);
-#elif defined (USE_MB2MB)
- free ((char *)str);
+#ifdef ASSUME_UTF8
+ (void) str;
#else
- VLC_UNUSED(str);
+ 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, true);
+#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, true);
+#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, false);
+#ifdef ASSUME_UTF8
+ return (char *)utf8;
+#else
+ size_t outsize;
+ return utf8 ? ToCharset ("", utf8, &outsize) : NULL;
+#endif
}
-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, false);
+#ifdef ASSUME_UTF8
+ return strdup (utf8);
+#else
+ size_t outsize;
+ return ToCharset ("", utf8, &outsize);
+#endif
}
-
/**
- * utf8_open: open() wrapper for UTF-8 filenames
+ * Formats an UTF-8 string as vfprintf(), then print it, with
+ * appropriate conversion to local encoding.
*/
-int utf8_open (const char *filename, int flags, mode_t mode)
+int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap )
{
-#if defined (WIN32) || defined (UNDER_CE)
- if (GetVersion() < 0x80000000)
+#ifdef ASSUME_UTF8
+ return vfprintf (stream, fmt, ap);
+#else
+ char *str;
+ int res;
+
+# 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))
{
- /* for Windows NT and above */
- wchar_t wpath[MAX_PATH + 1];
+ res = vasprintf (&str, fmt, ap);
+ if (unlikely(res == -1))
+ return -1;
- if (!MultiByteToWideChar (CP_UTF8, 0, filename, -1, wpath, MAX_PATH))
+ size_t wlen = 2 * (res + 1);
+ wchar_t *wide = malloc (wlen);
+ if (likely(wide != NULL))
{
- errno = ENOENT;
- return -1;
- }
- wpath[MAX_PATH] = L'\0';
+ wlen = MultiByteToWideChar (CP_UTF8, 0, str, res + 1, wide, wlen);
+ if (wlen > 0)
+ {
+ HANDLE h = (HANDLE)(intptr_t)_get_osfhandle (fd);
+ DWORD out;
- /*
- * 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);
+ WriteConsoleW (h, wide, wlen - 1, &out, NULL);
+ }
+ else
+ res = -1;
+ free (wide);
+ }
+ else
+ res = -1;
+ free (str);
+ return res;
}
-#endif
- const char *local_name = ToLocale (filename);
+# endif
- if (local_name == NULL)
- {
- errno = ENOENT;
+ res = vasprintf (&str, fmt, ap);
+ if (unlikely(res == -1))
return -1;
- }
- int fd = open (local_name, flags, mode);
- LocaleFree (local_name);
- return fd;
+ char *ansi = ToLocaleDup (str);
+ free (str);
+
+ if (ansi == NULL)
+ return -1;
+ fputs (ansi, stream);
+ free (ansi);
+ return res;
+#endif
}
/**
- * utf8_fopen: fopen() wrapper for UTF-8 filenames
+ * Formats an UTF-8 string as fprintf(), then print it, with
+ * appropriate conversion to local encoding.
*/
-FILE *utf8_fopen (const char *filename, const char *mode)
+int utf8_fprintf( FILE *stream, const char *fmt, ... )
{
- int rwflags = 0, oflags = 0;
- bool append = false;
-
- for (const char *ptr = mode; *ptr; ptr++)
- {
- switch (*ptr)
- {
- case 'r':
- rwflags = O_RDONLY;
- break;
-
- case 'a':
- rwflags = O_WRONLY;
- oflags |= O_CREAT;
- append = true;
- break;
-
- case 'w':
- rwflags = O_WRONLY;
- oflags |= O_CREAT | O_TRUNC;
- break;
-
- case '+':
- rwflags = O_RDWR;
- break;
-
-#ifdef O_TEXT
- case 't':
- oflags |= O_TEXT;
- break;
-#endif
- }
- }
-
- 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);
+ va_list ap;
+ int res;
- return stream;
+ va_start( ap, fmt );
+ res = utf8_vfprintf( stream, fmt, ap );
+ va_end( ap );
+ return res;
}
+
/**
- * utf8_mkdir: Calls mkdir() after conversion of file name to OS locale
+ * Converts the first character from a UTF-8 sequence into a code point.
*
- * @param dirname a UTF-8 string with the name of the directory that you
- * want to create.
- * @return A 0 return value indicates success. A -1 return value indicates an
- * error, and an error code is stored in errno
+ * @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.
*/
-int utf8_mkdir( const char *dirname, mode_t mode )
+static size_t vlc_towc (const char *str, uint32_t *restrict pwc)
{
-#if defined (UNDER_CE) || defined (WIN32)
- VLC_UNUSED( mode );
+ uint8_t *ptr = (uint8_t *)str;
+ assert (str != NULL);
- wchar_t wname[MAX_PATH + 1];
- char mod[MAX_PATH + 1];
- int i;
+ uint8_t c = ptr[0];
- /* Convert '/' into '\' */
- for( i = 0; *dirname; i++ )
+ if (unlikely(c == '\0'))
{
- if( i == MAX_PATH )
- return -1; /* overflow */
-
- if( *dirname == '/' )
- mod[i] = '\\';
- else
- mod[i] = *dirname;
- dirname++;
-
+ *pwc = 0;
+ return 0;
}
- mod[i] = 0;
- if( MultiByteToWideChar( CP_UTF8, 0, mod, -1, wname, MAX_PATH ) == 0 )
- {
- errno = ENOENT;
+ if (unlikely(c > 0xF4))
return -1;
- }
- wname[MAX_PATH] = L'\0';
- if( CreateDirectoryW( wname, NULL ) == 0 )
+ int charlen = clz8 (c ^ 0xFF);
+ switch (charlen)
{
- if( GetLastError( ) == ERROR_ALREADY_EXISTS )
- errno = EEXIST;
- else
- errno = ENOENT;
- return -1;
- }
- return 0;
-#else
- char *locname = ToLocale( dirname );
- int res;
+ case 0: // 7-bit ASCII character -> OK
+ *pwc = c;
+ return 1;
- if( locname == NULL )
- {
- errno = ENOENT;
- return -1;
+ case 1: // continuation byte -> error
+ return -1;
}
- res = mkdir( locname, mode );
- LocaleFree( locname );
- return res;
-#endif
-}
+ assert (charlen >= 2 && charlen <= 4);
-/**
- * utf8_opendir: wrapper that converts dirname to the locale in use by the OS
- *
- * @param dirname UTF-8 representation of the directory name
- *
- * @return a pointer to the DIR struct. Release with closedir().
- */
-DIR *utf8_opendir( const char *dirname )
-{
-#ifdef WIN32
- wchar_t wname[MAX_PATH + 1];
-
- if (MultiByteToWideChar (CP_UTF8, 0, dirname, -1, wname, MAX_PATH))
+ uint32_t cp = c & ~((0xff >> (7 - charlen)) << (7 - charlen));
+ for (int i = 1; i < charlen; i++)
{
- wname[MAX_PATH] = L'\0';
- return (DIR *)vlc_wopendir (wname);
+ assert (cp < (1 << 26));
+ c = ptr[i];
+
+ if (unlikely((c >> 6) != 2)) // not a continuation byte
+ return -1;
+
+ cp = (cp << 6) | (ptr[i] & 0x3f);
}
-#else
- const char *local_name = ToLocale( dirname );
- if( local_name != NULL )
+ switch (charlen)
{
- DIR *dir = opendir( local_name );
- LocaleFree( local_name );
- return dir;
+ 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;
}
-#endif
-
- errno = ENOENT;
- return NULL;
+ *pwc = cp;
+ return charlen;
}
/**
- * utf8_readdir: a readdir wrapper that returns the name of the next entry
- * in the directory as a UTF-8 string.
- *
- * @param dir The directory that is being read
- *
- * @return a UTF-8 string of the directory entry. Use free() to free this memory.
+ * 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.
*/
-char *utf8_readdir( DIR *dir )
+char *vlc_strcasestr (const char *haystack, const char *needle)
{
-#ifdef WIN32
- struct _wdirent *ent = vlc_wreaddir (dir);
- if (ent == NULL)
- return NULL;
-
- return FromWide (ent->d_name);
-#else
- struct dirent *ent;
-
- ent = readdir( (DIR *)dir );
- if( ent == NULL )
- return NULL;
+ ssize_t s;
- return vlc_fix_readdir( ent->d_name );
-#endif
-}
-
-static int dummy_select( const char *str )
-{
- (void)str;
- return 1;
-}
-
-int utf8_loaddir( DIR *dir, char ***namelist,
- int (*select)( const char * ),
- int (*compar)( const char **, const char ** ) )
-{
- if( select == NULL )
- select = dummy_select;
-
- if( dir == NULL )
- return -1;
- else
+ do
{
- char **tab = NULL;
- char *entry;
- unsigned num = 0;
+ const char *h = haystack, *n = needle;
- rewinddir( dir );
-
- while( ( entry = utf8_readdir( dir ) ) != NULL )
+ for (;;)
{
- char **newtab;
+ uint32_t cph, cpn;
- if( !select( entry ) )
- {
- free( entry );
- continue;
- }
+ s = vlc_towc (n, &cpn);
+ if (s == 0)
+ return (char *)haystack;
+ if (unlikely(s < 0))
+ return NULL;
+ n += s;
- newtab = realloc( tab, sizeof( char * ) * (num + 1) );
- if( newtab == NULL )
- {
- free( entry );
- goto error;
- }
- tab = newtab;
- tab[num++] = entry;
+ s = vlc_towc (h, &cph);
+ if (s <= 0 || towlower (cph) != towlower (cpn))
+ break;
+ h += s;
}
- if( compar != NULL )
- qsort( tab, num, sizeof( tab[0] ),
- (int (*)( const void *, const void *))compar );
-
- *namelist = tab;
- return num;
-
- error:{
- unsigned i;
-
- for( i = 0; i < num; i++ )
- free( tab[i] );
- if( tab != NULL )
- free( tab );
- }
+ s = vlc_towc (haystack, &(uint32_t) { 0 });
+ haystack += s;
}
- return -1;
-}
+ while (s != 0);
-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;
+ return NULL;
}
-static int utf8_statEx( const char *filename, struct stat *buf,
- bool deref )
+/**
+ * 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.
+ *
+ * @return str if it was valid UTF-8, NULL if not.
+ */
+char *EnsureUTF8( char *str )
{
-#if defined (WIN32) || defined (UNDER_CE)
- /* retrieve Windows OS version */
- if( GetVersion() < 0x80000000 )
- {
- /* for Windows NT and above */
- wchar_t wpath[MAX_PATH + 1];
+ char *ret = str;
+ size_t n;
+ uint32_t cp;
- if( !MultiByteToWideChar( CP_UTF8, 0, filename, -1, wpath, MAX_PATH ) )
+ while ((n = vlc_towc (str, &cp)) != 0)
+ if (likely(n != (size_t)-1))
+ str += n;
+ else
{
- errno = ENOENT;
- return -1;
+ *str++ = '?';
+ ret = NULL;
}
- wpath[MAX_PATH] = L'\0';
-
- return _wstati64( wpath, buf );
- }
-#endif
-#ifdef HAVE_SYS_STAT_H
- const char *local_name = ToLocale( filename );
-
- if( local_name != NULL )
- {
- int res = deref ? stat( local_name, buf )
- : lstat( local_name, buf );
- LocaleFree( local_name );
- return res;
- }
- errno = ENOENT;
-#endif
- return -1;
+ return ret;
}
-int utf8_stat( const char *filename, struct stat *buf)
+/**
+ * Checks whether a string is a valid UTF-8 byte sequence.
+ *
+ * @param str nul-terminated string to be checked
+ *
+ * @return str if it was valid UTF-8, NULL if not.
+ */
+const char *IsUTF8( const char *str )
{
- return utf8_statEx( filename, buf, true );
-}
+ size_t n;
+ uint32_t cp;
-int utf8_lstat( const char *filename, struct stat *buf)
-{
- return utf8_statEx( filename, buf, false );
+ while ((n = vlc_towc (str, &cp)) != 0)
+ if (likely(n != (size_t)-1))
+ str += n;
+ else
+ return NULL;
+ return str;
}
/**
- * utf8_unlink: Calls unlink() after conversion of file name to OS locale
+ * Converts a string from the given character encoding to utf-8.
*
- * @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
+ * @return a nul-terminated utf-8 string, or null in case of error.
+ * The result must be freed using free().
*/
-int utf8_unlink( const char *filename )
+char *FromCharset(const char *charset, const void *data, size_t data_size)
{
-#if defined (WIN32) || defined (UNDER_CE)
- if( GetVersion() < 0x80000000 )
+ 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++ )
{
- /* for Windows NT and above */
- wchar_t wpath[MAX_PATH + 1];
+ 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( !MultiByteToWideChar( CP_UTF8, 0, filename, -1, wpath, MAX_PATH ) )
- {
- errno = ENOENT;
- return -1;
+ if (vlc_iconv (handle, &in, &in_size, &tmp, &out_max) != (size_t)(-1)) {
+ *tmp = '\0';
+ break;
}
- wpath[MAX_PATH] = L'\0';
+ free(out);
+ out = NULL;
- /*
- * 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;
+ if (errno != E2BIG)
+ break;
}
-
- int ret = unlink( local_name );
- LocaleFree( local_name );
- return ret;
+ vlc_iconv_close(handle);
+ return out;
}
-
-
/**
- * utf8_*printf: *printf with conversion from UTF-8 to local encoding
+ * 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.
*/
-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, ... )
+void *ToCharset(const char *charset, const char *in, size_t *outsize)
{
- va_list ap;
- int res;
-
- va_start( ap, fmt );
- res = utf8_vfprintf( stream, fmt, ap );
- va_end( ap );
- return res;
-}
+ 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;
-static char *CheckUTF8( char *str, char rep )
-{
- uint8_t *ptr = (uint8_t *)str;
- assert (str != NULL);
-
- for (;;)
+ for (unsigned mul = 4; mul < 16; mul++)
{
- uint8_t c = ptr[0];
- int charlen = -1;
-
- if (c == '\0')
+ size_t outlen = mul * (inlen + 1);
+ res = malloc (outlen);
+ if (unlikely(res == NULL))
break;
- for (int i = 0; i < 7; i++)
- if ((c >> (7 - i)) == ((0xff >> (7 - i)) ^ 1))
- {
- charlen = i;
- break;
- }
-
- switch (charlen)
- {
- case 0: // 7-bit ASCII character -> OK
- ptr++;
- continue;
-
- case -1: // 1111111x -> error
- case 1: // continuation byte -> error
- goto error;
- }
-
- assert (charlen >= 2);
+ const char *inp = in;
+ char *outp = res;
+ size_t inb = inlen;
+ size_t outb = outlen;
- uint32_t cp = c & ~((0xff >> (7 - charlen)) << (7 - charlen));
- for (int i = 1; i < charlen; i++)
+ if (vlc_iconv (hd, &inp, &inb, &outp, &outb) != (size_t)(-1))
{
- 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);
+ *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;
}
- 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)
- return NULL;
- *ptr++ = rep;
- str = NULL;
+ free (res);
+ res = NULL;
+ if (errno != E2BIG) /* conversion failure */
+ break;
}
-
- return str;
-}
-
-/**
- * EnsureUTF8: 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.
- *
- * @return str if it was valid UTF-8, NULL if not.
- */
-char *EnsureUTF8( char *str )
-{
- return CheckUTF8( str, '?' );
+ vlc_iconv_close (hd);
+ return res;
}
-
-/**
- * IsUTF8: checks whether a string is a valid UTF-8 byte sequence.
- *
- * @param str nul-terminated string to be checked
- *
- * @return str if it was valid UTF-8, NULL if not.
- */
-const char *IsUTF8( const char *str )
-{
- return CheckUTF8( (char *)str, 0 );
-}