src/text/unicode.c

   1 /*****************************************************************************
   2  * unicode.c: Unicode <-> locale functions
   3  *****************************************************************************
   4  * Copyright (C) 2005-2006 VLC authors and VideoLAN
   5  * Copyright © 2005-2010 Rémi Denis-Courmont
   6  *
   7  * Authors: Rémi Denis-Courmont <rem # videolan.org>
   8  *
   9  * This program is free software; you can redistribute it and/or modify it
  10  * under the terms of the GNU Lesser General Public License as published by
  11  * the Free Software Foundation; either version 2.1 of the License, or
  12  * (at your option) any later version.
  13  *
  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 Lesser General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU Lesser General Public License
  20  * along with this program; if not, write to the Free Software Foundation,
  21  * Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  22  *****************************************************************************/
  23
  24 /*****************************************************************************
  25  * Preamble
  26  *****************************************************************************/
  27 #ifdef HAVE_CONFIG_H
  28 # include "config.h"
  29 #endif
  30
  31 #include <vlc_common.h>
  32
  33 #include "libvlc.h"
  34 #include <vlc_charset.h>
  35
  36 #include <assert.h>
  37
  38 #include <stdio.h>
  39 #include <stdarg.h>
  40 #include <stdlib.h>
  41 #include <sys/types.h>
  42 #if defined(_WIN32)
  43 #  include <io.h>
  44 #endif
  45 #include <errno.h>
  46 #include <wctype.h>
  47
  48 /**
  49  * Formats an UTF-8 string as vfprintf(), then print it, with
  50  * appropriate conversion to local encoding.
  51  */
  52 int utf8_vfprintf( FILE *stream, const char *fmt, va_list ap )
  53 {
  54 #ifndef _WIN32
  55     return vfprintf (stream, fmt, ap);
  56 #else
  57     char *str;
  58     int res = vasprintf (&str, fmt, ap);
  59     if (unlikely(res == -1))
  60         return -1;
  61
  62 #if !VLC_WINSTORE_APP
  63     /* Writing to the console is a lot of fun on Microsoft Windows.
  64      * If you use the standard I/O functions, you must use the OEM code page,
  65      * which is different from the usual ANSI code page. Or maybe not, if the
  66      * user called "chcp". Anyway, we prefer Unicode. */
  67     int fd = _fileno (stream);
  68     if (likely(fd != -1) && _isatty (fd))
  69     {
  70         wchar_t *wide = ToWide (str);
  71         if (likely(wide != NULL))
  72         {
  73             HANDLE h = (HANDLE)((uintptr_t)_get_osfhandle (fd));
  74             DWORD out;
  75             /* XXX: It is not clear whether WriteConsole() wants the number of
  76              * Unicode characters or the size of the wchar_t array. */
  77             BOOL ok = WriteConsoleW (h, wide, wcslen (wide), &out, NULL);
  78             free (wide);
  79             if (ok)
  80                 goto out;
  81         }
  82     }
  83 #endif
  84     wchar_t *wide = ToWide(str);
  85     if (likely(wide != NULL))
  86     {
  87         res = fputws(wide, stream);
  88         free(wide);
  89     }
  90     else
  91         res = -1;
  92 out:
  93     free (str);
  94     return res;
  95 #endif
  96 }
  97
  98 /**
  99  * Formats an UTF-8 string as fprintf(), then print it, with
 100  * appropriate conversion to local encoding.
 101  */
 102 int utf8_fprintf( FILE *stream, const char *fmt, ... )
 103 {
 104     va_list ap;
 105     int res;
 106
 107     va_start( ap, fmt );
 108     res = utf8_vfprintf( stream, fmt, ap );
 109     va_end( ap );
 110     return res;
 111 }
 112
 113 size_t vlc_towc (const char *str, uint32_t *restrict pwc)
 114 {
 115     uint8_t *ptr = (uint8_t *)str, c;
 116     uint32_t cp;
 117
 118     assert (str != NULL);
 119
 120     c = *ptr;
 121     if (unlikely(c > 0xF4))
 122         return -1;
 123
 124     int charlen = clz((unsigned char)(c ^ 0xFF));
 125     switch (charlen)
 126     {
 127         case 0: // 7-bit ASCII character -> short cut
 128             *pwc = c;
 129             return c != '\0';
 130
 131         case 1: // continuation byte -> error
 132             return -1;
 133
 134         case 2:
 135             if (unlikely(c < 0xC2)) // ASCII overlong
 136                 return -1;
 137             cp = (c & 0x1F) << 6;
 138             break;
 139
 140         case 3:
 141             cp = (c & 0x0F) << 12;
 142             break;
 143
 144         case 4:
 145             cp = (c & 0x07) << 18;
 146             break;
 147
 148         default:
 149             vlc_assert_unreachable ();
 150     }
 151
 152     /* Unrolled continuation bytes decoding */
 153     switch (charlen)
 154     {
 155         case 4:
 156             c = *++ptr;
 157             if (unlikely((c & 0xC0) != 0x80)) // not a continuation byte
 158                 return -1;
 159             cp |= (c & 0x3F) << 12;
 160
 161             if (unlikely(cp >= 0x110000)) // beyond Unicode range
 162                 return -1;
 163             /* fall through */
 164         case 3:
 165             c = *++ptr;
 166             if (unlikely((c & 0xC0) != 0x80)) // not a continuation byte
 167                 return -1;
 168             cp |= (c & 0x3F) << 6;
 169
 170             if (unlikely(cp >= 0xD800 && cp < 0xE000)) // UTF-16 surrogate
 171                 return -1;
 172             if (unlikely(cp < (1u << (5 * charlen - 4)))) // non-ASCII overlong
 173                 return -1;
 174             /* fall through */
 175         case 2:
 176             c = *++ptr;
 177             if (unlikely((c & 0xC0) != 0x80)) // not a continuation byte
 178                 return -1;
 179             cp |= (c & 0x3F);
 180             break;
 181     }
 182
 183     *pwc = cp;
 184     return charlen;
 185 }
 186
 187 /**
 188  * Look for an UTF-8 string within another one in a case-insensitive fashion.
 189  * Beware that this is quite slow. Contrary to strcasestr(), this function
 190  * works regardless of the system character encoding, and handles multibyte
 191  * code points correctly.
 192
 193  * @param haystack string to look into
 194  * @param needle string to look for
 195  * @return a pointer to the first occurrence of the needle within the haystack,
 196  * or NULL if no occurrence were found.
 197  */
 198 char *vlc_strcasestr (const char *haystack, const char *needle)
 199 {
 200     ssize_t s;
 201
 202     do
 203     {
 204         const char *h = haystack, *n = needle;
 205
 206         for (;;)
 207         {
 208             uint32_t cph, cpn;
 209
 210             s = vlc_towc (n, &cpn);
 211             if (s == 0)
 212                 return (char *)haystack;
 213             if (unlikely(s < 0))
 214                 return NULL;
 215             n += s;
 216
 217             s = vlc_towc (h, &cph);
 218             if (s <= 0 || towlower (cph) != towlower (cpn))
 219                 break;
 220             h += s;
 221         }
 222
 223         s = vlc_towc (haystack, &(uint32_t) { 0 });
 224         haystack += s;
 225     }
 226     while (s > 0);
 227
 228     return NULL;
 229 }
 230
 231 /**
 232  * Converts a string from the given character encoding to utf-8.
 233  *
 234  * @return a nul-terminated utf-8 string, or null in case of error.
 235  * The result must be freed using free().
 236  */
 237 char *FromCharset(const char *charset, const void *data, size_t data_size)
 238 {
 239     vlc_iconv_t handle = vlc_iconv_open ("UTF-8", charset);
 240     if (handle == (vlc_iconv_t)(-1))
 241         return NULL;
 242
 243     char *out = NULL;
 244     for(unsigned mul = 4; mul < 8; mul++ )
 245     {
 246         size_t in_size = data_size;
 247         const char *in = data;
 248         size_t out_max = mul * data_size;
 249         char *tmp = out = malloc (1 + out_max);
 250         if (!out)
 251             break;
 252
 253         if (vlc_iconv (handle, &in, &in_size, &tmp, &out_max) != (size_t)(-1)) {
 254             *tmp = '\0';
 255             break;
 256         }
 257         free(out);
 258         out = NULL;
 259
 260         if (errno != E2BIG)
 261             break;
 262     }
 263     vlc_iconv_close(handle);
 264     return out;
 265 }
 266
 267 /**
 268  * Converts a nul-terminated UTF-8 string to a given character encoding.
 269  * @param charset iconv name of the character set
 270  * @param in nul-terminated UTF-8 string
 271  * @param outsize pointer to hold the byte size of result
 272  *
 273  * @return A pointer to the result, which must be released using free().
 274  * The UTF-8 nul terminator is included in the conversion if the target
 275  * character encoding supports it. However it is not included in the returned
 276  * byte size.
 277  * In case of error, NULL is returned and the byte size is undefined.
 278  */
 279 void *ToCharset(const char *charset, const char *in, size_t *outsize)
 280 {
 281     vlc_iconv_t hd = vlc_iconv_open (charset, "UTF-8");
 282     if (hd == (vlc_iconv_t)(-1))
 283         return NULL;
 284
 285     const size_t inlen = strlen (in);
 286     void *res;
 287
 288     for (unsigned mul = 4; mul < 16; mul++)
 289     {
 290         size_t outlen = mul * (inlen + 1);
 291         res = malloc (outlen);
 292         if (unlikely(res == NULL))
 293             break;
 294
 295         const char *inp = in;
 296         char *outp = res;
 297         size_t inb = inlen;
 298         size_t outb = outlen - mul;
 299
 300         if (vlc_iconv (hd, &inp, &inb, &outp, &outb) != (size_t)(-1))
 301         {
 302             *outsize = outlen - mul - outb;
 303             outb += mul;
 304             inb = 1; /* append nul terminator if possible */
 305             if (vlc_iconv (hd, &inp, &inb, &outp, &outb) != (size_t)(-1))
 306                 break;
 307             if (errno == EILSEQ) /* cannot translate nul terminator!? */
 308                 break;
 309         }
 310
 311         free (res);
 312         res = NULL;
 313         if (errno != E2BIG) /* conversion failure */
 314             break;
 315     }
 316     vlc_iconv_close (hd);
 317     return res;
 318 }
 319