reactos/tools/unicode/wctomb.c

   1 /*
   2  * WideCharToMultiByte implementation
   3  *
   4  * Copyright 2000 Alexandre Julliard
   5  *
   6  * This library is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU Lesser General Public
   8  * License as published by the Free Software Foundation; either
   9  * version 2.1 of the License, or (at your option) any later version.
  10  *
  11  * This library is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * Lesser General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU Lesser General Public
  17  * License along with this library; if not, write to the Free Software
  18  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  19  */
  20
  21 #include <string.h>
  22
  23 #include "wine/unicode.h"
  24
  25 /* search for a character in the unicode_compose_table; helper for compose() */
  26 static inline int binary_search( WCHAR ch, int low, int high )
  27 {
  28     extern const WCHAR unicode_compose_table[];
  29     while (low <= high)
  30     {
  31         int pos = (low + high) / 2;
  32         if (unicode_compose_table[2*pos] < ch)
  33         {
  34             low = pos + 1;
  35             continue;
  36         }
  37         if (unicode_compose_table[2*pos] > ch)
  38         {
  39             high = pos - 1;
  40             continue;
  41         }
  42         return pos;
  43     }
  44     return -1;
  45 }
  46
  47 /* return the result of the composition of two Unicode chars, or 0 if none */
  48 static WCHAR compose( const WCHAR *str )
  49 {
  50     extern const WCHAR unicode_compose_table[];
  51     extern const unsigned int unicode_compose_table_size;
  52
  53     int idx = 1, low = 0, high = unicode_compose_table_size - 1;
  54     for (;;)
  55     {
  56         int pos = binary_search( str[idx], low, high );
  57         if (pos == -1) return 0;
  58         if (!idx--) return unicode_compose_table[2*pos+1];
  59         low = unicode_compose_table[2*pos+1];
  60         high = unicode_compose_table[2*pos+3] - 1;
  61     }
  62 }
  63
  64
  65 /****************************************************************/
  66 /* sbcs support */
  67
  68 /* check if 'ch' is an acceptable sbcs mapping for 'wch' */
  69 static inline int is_valid_sbcs_mapping( const struct sbcs_table *table, int flags,
  70                                          WCHAR wch, unsigned char ch )
  71 {
  72     if (flags & WC_NO_BEST_FIT_CHARS) return (table->cp2uni[ch] == wch);
  73     if (ch != (unsigned char)table->info.def_char) return 1;
  74     return (wch == table->info.def_unicode_char);
  75 }
  76
  77 /* query necessary dst length for src string */
  78 static int get_length_sbcs( const struct sbcs_table *table, int flags,
  79                             const WCHAR *src, unsigned int srclen, int *used )
  80 {
  81     const unsigned char  * const uni2cp_low = table->uni2cp_low;
  82     const unsigned short * const uni2cp_high = table->uni2cp_high;
  83     int ret, tmp;
  84     WCHAR composed;
  85
  86     if (!used) used = &tmp;  /* avoid checking on every char */
  87     *used = 0;
  88
  89     for (ret = 0; srclen; ret++, src++, srclen--)
  90     {
  91         WCHAR wch = *src;
  92         unsigned char ch;
  93
  94         if ((flags & WC_COMPOSITECHECK) && (srclen > 1) && (composed = compose(src)))
  95         {
  96             /* now check if we can use the composed char */
  97             ch = uni2cp_low[uni2cp_high[composed >> 8] + (composed & 0xff)];
  98             if (is_valid_sbcs_mapping( table, flags, composed, ch ))
  99             {
 100                 /* we have a good mapping, use it */
 101                 src++;
 102                 srclen--;
 103                 continue;
 104             }
 105             /* no mapping for the composed char, check the other flags */
 106             if (flags & WC_DEFAULTCHAR) /* use the default char instead */
 107             {
 108                 *used = 1;
 109                 src++;  /* skip the non-spacing char */
 110                 srclen--;
 111                 continue;
 112             }
 113             if (flags & WC_DISCARDNS) /* skip the second char of the composition */
 114             {
 115                 src++;
 116                 srclen--;
 117             }
 118             /* WC_SEPCHARS is the default */
 119         }
 120         if (!*used)
 121         {
 122             ch = uni2cp_low[uni2cp_high[wch >> 8] + (wch & 0xff)];
 123             *used = !is_valid_sbcs_mapping( table, flags, wch, ch );
 124         }
 125     }
 126     return ret;
 127 }
 128
 129 /* wcstombs for single-byte code page */
 130 static inline int wcstombs_sbcs( const struct sbcs_table *table,
 131                                  const WCHAR *src, unsigned int srclen,
 132                                  char *dst, unsigned int dstlen )
 133 {
 134     const unsigned char  * const uni2cp_low = table->uni2cp_low;
 135     const unsigned short * const uni2cp_high = table->uni2cp_high;
 136     int ret = srclen;
 137
 138     if (dstlen < srclen)
 139     {
 140         /* buffer too small: fill it up to dstlen and return error */
 141         srclen = dstlen;
 142         ret = -1;
 143     }
 144
 145     if (dst <= (const char *)src && dst + 16 > (const char *)src)
 146     {
 147         /* overlapping buffers, do it char by char */
 148         while (srclen--)
 149         {
 150             *dst++ = uni2cp_low[uni2cp_high[*src >> 8] + (*src & 0xff)];
 151             src++;
 152         }
 153         return ret;
 154     }
 155
 156     for (;;)
 157     {
 158         switch(srclen)
 159         {
 160         default:
 161         case 16: dst[15] = uni2cp_low[uni2cp_high[src[15] >> 8] + (src[15] & 0xff)];
 162         case 15: dst[14] = uni2cp_low[uni2cp_high[src[14] >> 8] + (src[14] & 0xff)];
 163         case 14: dst[13] = uni2cp_low[uni2cp_high[src[13] >> 8] + (src[13] & 0xff)];
 164         case 13: dst[12] = uni2cp_low[uni2cp_high[src[12] >> 8] + (src[12] & 0xff)];
 165         case 12: dst[11] = uni2cp_low[uni2cp_high[src[11] >> 8] + (src[11] & 0xff)];
 166         case 11: dst[10] = uni2cp_low[uni2cp_high[src[10] >> 8] + (src[10] & 0xff)];
 167         case 10: dst[9]  = uni2cp_low[uni2cp_high[src[9]  >> 8] + (src[9]  & 0xff)];
 168         case 9:  dst[8]  = uni2cp_low[uni2cp_high[src[8]  >> 8] + (src[8]  & 0xff)];
 169         case 8:  dst[7]  = uni2cp_low[uni2cp_high[src[7]  >> 8] + (src[7]  & 0xff)];
 170         case 7:  dst[6]  = uni2cp_low[uni2cp_high[src[6]  >> 8] + (src[6]  & 0xff)];
 171         case 6:  dst[5]  = uni2cp_low[uni2cp_high[src[5]  >> 8] + (src[5]  & 0xff)];
 172         case 5:  dst[4]  = uni2cp_low[uni2cp_high[src[4]  >> 8] + (src[4]  & 0xff)];
 173         case 4:  dst[3]  = uni2cp_low[uni2cp_high[src[3]  >> 8] + (src[3]  & 0xff)];
 174         case 3:  dst[2]  = uni2cp_low[uni2cp_high[src[2]  >> 8] + (src[2]  & 0xff)];
 175         case 2:  dst[1]  = uni2cp_low[uni2cp_high[src[1]  >> 8] + (src[1]  & 0xff)];
 176         case 1:  dst[0]  = uni2cp_low[uni2cp_high[src[0]  >> 8] + (src[0]  & 0xff)];
 177         case 0: break;
 178         }
 179         if (srclen < 16) return ret;
 180         dst += 16;
 181         src += 16;
 182         srclen -= 16;
 183     }
 184 }
 185
 186 /* slow version of wcstombs_sbcs that handles the various flags */
 187 static int wcstombs_sbcs_slow( const struct sbcs_table *table, int flags,
 188                                const WCHAR *src, unsigned int srclen,
 189                                char *dst, unsigned int dstlen,
 190                                const char *defchar, int *used )
 191 {
 192     const unsigned char  * const uni2cp_low = table->uni2cp_low;
 193     const unsigned short * const uni2cp_high = table->uni2cp_high;
 194     const unsigned char table_default = table->info.def_char & 0xff;
 195     unsigned int len;
 196     int tmp;
 197     WCHAR composed;
 198
 199     if (!defchar) defchar = &table_default;
 200     if (!used) used = &tmp;  /* avoid checking on every char */
 201     *used = 0;
 202
 203     for (len = dstlen; srclen && len; dst++, len--, src++, srclen--)
 204     {
 205         WCHAR wch = *src;
 206
 207         if ((flags & WC_COMPOSITECHECK) && (srclen > 1) && (composed = compose(src)))
 208         {
 209             /* now check if we can use the composed char */
 210             *dst = uni2cp_low[uni2cp_high[composed >> 8] + (composed & 0xff)];
 211             if (is_valid_sbcs_mapping( table, flags, composed, *dst ))
 212             {
 213                 /* we have a good mapping, use it */
 214                 src++;
 215                 srclen--;
 216                 continue;
 217             }
 218             /* no mapping for the composed char, check the other flags */
 219             if (flags & WC_DEFAULTCHAR) /* use the default char instead */
 220             {
 221                 *dst = *defchar;
 222                 *used = 1;
 223                 src++;  /* skip the non-spacing char */
 224                 srclen--;
 225                 continue;
 226             }
 227             if (flags & WC_DISCARDNS) /* skip the second char of the composition */
 228             {
 229                 src++;
 230                 srclen--;
 231             }
 232             /* WC_SEPCHARS is the default */
 233         }
 234
 235         *dst = uni2cp_low[uni2cp_high[wch >> 8] + (wch & 0xff)];
 236         if (!is_valid_sbcs_mapping( table, flags, wch, *dst ))
 237         {
 238             *dst = *defchar;
 239             *used = 1;
 240         }
 241     }
 242     if (srclen) return -1;  /* overflow */
 243     return dstlen - len;
 244 }
 245
 246
 247 /****************************************************************/
 248 /* dbcs support */
 249
 250 /* check if 'ch' is an acceptable dbcs mapping for 'wch' */
 251 static inline int is_valid_dbcs_mapping( const struct dbcs_table *table, int flags,
 252                                          WCHAR wch, unsigned short ch )
 253 {
 254     if (ch == table->info.def_char && wch != table->info.def_unicode_char) return 0;
 255     if (flags & WC_NO_BEST_FIT_CHARS)
 256     {
 257         /* check if char maps back to the same Unicode value */
 258         if (ch & 0xff00)
 259         {
 260             unsigned char off = table->cp2uni_leadbytes[ch >> 8];
 261             return (table->cp2uni[(off << 8) + (ch & 0xff)] == wch);
 262         }
 263         return (table->cp2uni[ch & 0xff] == wch);
 264     }
 265     return 1;
 266 }
 267
 268 /* query necessary dst length for src string */
 269 static int get_length_dbcs( const struct dbcs_table *table, int flags,
 270                             const WCHAR *src, unsigned int srclen,
 271                             const char *defchar, int *used )
 272 {
 273     const unsigned short * const uni2cp_low = table->uni2cp_low;
 274     const unsigned short * const uni2cp_high = table->uni2cp_high;
 275     WCHAR defchar_value = table->info.def_char;
 276     WCHAR composed;
 277     int len, tmp;
 278
 279     if (!defchar && !used && !(flags & WC_COMPOSITECHECK))
 280     {
 281         for (len = 0; srclen; srclen--, src++, len++)
 282         {
 283             if (uni2cp_low[uni2cp_high[*src >> 8] + (*src & 0xff)] & 0xff00) len++;
 284         }
 285         return len;
 286     }
 287
 288     if (defchar) defchar_value = defchar[1] ? ((defchar[0] << 8) | defchar[1]) : defchar[0];
 289     if (!used) used = &tmp;  /* avoid checking on every char */
 290     *used = 0;
 291     for (len = 0; srclen; len++, srclen--, src++)
 292     {
 293         unsigned short res;
 294         WCHAR wch = *src;
 295
 296         if ((flags & WC_COMPOSITECHECK) && (srclen > 1) && (composed = compose(src)))
 297         {
 298             /* now check if we can use the composed char */
 299             res = uni2cp_low[uni2cp_high[composed >> 8] + (composed & 0xff)];
 300
 301             if (is_valid_dbcs_mapping( table, flags, composed, res ))
 302             {
 303                 /* we have a good mapping for the composed char, use it */
 304                 if (res & 0xff00) len++;
 305                 src++;
 306                 srclen--;
 307                 continue;
 308             }
 309             /* no mapping for the composed char, check the other flags */
 310             if (flags & WC_DEFAULTCHAR) /* use the default char instead */
 311             {
 312                 if (defchar_value & 0xff00) len++;
 313                 *used = 1;
 314                 src++;  /* skip the non-spacing char */
 315                 srclen--;
 316                 continue;
 317             }
 318             if (flags & WC_DISCARDNS) /* skip the second char of the composition */
 319             {
 320                 src++;
 321                 srclen--;
 322             }
 323             /* WC_SEPCHARS is the default */
 324         }
 325
 326         res = uni2cp_low[uni2cp_high[wch >> 8] + (wch & 0xff)];
 327         if (!is_valid_dbcs_mapping( table, flags, wch, res ))
 328         {
 329             res = defchar_value;
 330             *used = 1;
 331         }
 332         if (res & 0xff00) len++;
 333     }
 334     return len;
 335 }
 336
 337 /* wcstombs for double-byte code page */
 338 static inline int wcstombs_dbcs( const struct dbcs_table *table,
 339                                  const WCHAR *src, unsigned int srclen,
 340                                  char *dst, unsigned int dstlen )
 341 {
 342     const unsigned short * const uni2cp_low = table->uni2cp_low;
 343     const unsigned short * const uni2cp_high = table->uni2cp_high;
 344     int len;
 345
 346     for (len = dstlen; srclen && len; len--, srclen--, src++)
 347     {
 348         unsigned short res = uni2cp_low[uni2cp_high[*src >> 8] + (*src & 0xff)];
 349         if (res & 0xff00)
 350         {
 351             if (len == 1) break;  /* do not output a partial char */
 352             len--;
 353             *dst++ = res >> 8;
 354         }
 355         *dst++ = (char)res;
 356     }
 357     if (srclen) return -1;  /* overflow */
 358     return dstlen - len;
 359 }
 360
 361 /* slow version of wcstombs_dbcs that handles the various flags */
 362 static int wcstombs_dbcs_slow( const struct dbcs_table *table, int flags,
 363                                const WCHAR *src, unsigned int srclen,
 364                                char *dst, unsigned int dstlen,
 365                                const char *defchar, int *used )
 366 {
 367     const unsigned short * const uni2cp_low = table->uni2cp_low;
 368     const unsigned short * const uni2cp_high = table->uni2cp_high;
 369     WCHAR defchar_value = table->info.def_char;
 370     WCHAR composed;
 371     int len, tmp;
 372
 373     if (defchar) defchar_value = defchar[1] ? ((defchar[0] << 8) | defchar[1]) : defchar[0];
 374     if (!used) used = &tmp;  /* avoid checking on every char */
 375     *used = 0;
 376
 377     for (len = dstlen; srclen && len; len--, srclen--, src++)
 378     {
 379         unsigned short res;
 380         WCHAR wch = *src;
 381
 382         if ((flags & WC_COMPOSITECHECK) && (srclen > 1) && (composed = compose(src)))
 383         {
 384             /* now check if we can use the composed char */
 385             res = uni2cp_low[uni2cp_high[composed >> 8] + (composed & 0xff)];
 386
 387             if (is_valid_dbcs_mapping( table, flags, composed, res ))
 388             {
 389                 /* we have a good mapping for the composed char, use it */
 390                 src++;
 391                 srclen--;
 392                 goto output_char;
 393             }
 394             /* no mapping for the composed char, check the other flags */
 395             if (flags & WC_DEFAULTCHAR) /* use the default char instead */
 396             {
 397                 res = defchar_value;
 398                 *used = 1;
 399                 src++;  /* skip the non-spacing char */
 400                 srclen--;
 401                 goto output_char;
 402             }
 403             if (flags & WC_DISCARDNS) /* skip the second char of the composition */
 404             {
 405                 src++;
 406                 srclen--;
 407             }
 408             /* WC_SEPCHARS is the default */
 409         }
 410
 411         res = uni2cp_low[uni2cp_high[wch >> 8] + (wch & 0xff)];
 412         if (!is_valid_dbcs_mapping( table, flags, wch, res ))
 413         {
 414             res = defchar_value;
 415             *used = 1;
 416         }
 417
 418     output_char:
 419         if (res & 0xff00)
 420         {
 421             if (len == 1) break;  /* do not output a partial char */
 422             len--;
 423             *dst++ = res >> 8;
 424         }
 425         *dst++ = (char)res;
 426     }
 427     if (srclen) return -1;  /* overflow */
 428     return dstlen - len;
 429 }
 430
 431 /* wide char to multi byte string conversion */
 432 /* return -1 on dst buffer overflow */
 433 int wine_cp_wcstombs( const union cptable *table, int flags,
 434                       const WCHAR *src, int srclen,
 435                       char *dst, int dstlen, const char *defchar, int *used )
 436 {
 437     if (table->info.char_size == 1)
 438     {
 439         if (flags || defchar || used)
 440         {
 441             if (!dstlen) return get_length_sbcs( &table->sbcs, flags, src, srclen, used );
 442             return wcstombs_sbcs_slow( &table->sbcs, flags, src, srclen,
 443                                        dst, dstlen, defchar, used );
 444         }
 445         if (!dstlen) return srclen;
 446         return wcstombs_sbcs( &table->sbcs, src, srclen, dst, dstlen );
 447     }
 448     else /* mbcs */
 449     {
 450         if (!dstlen) return get_length_dbcs( &table->dbcs, flags, src, srclen, defchar, used );
 451         if (flags || defchar || used)
 452             return wcstombs_dbcs_slow( &table->dbcs, flags, src, srclen,
 453                                        dst, dstlen, defchar, used );
 454         return wcstombs_dbcs( &table->dbcs, src, srclen, dst, dstlen );
 455     }
 456 }
 457
 458 /* CP_SYMBOL implementation */
 459 /* return -1 on dst buffer overflow, -2 on invalid character */
 460 int wine_cpsymbol_wcstombs( const WCHAR *src, int srclen, char *dst, int dstlen)
 461 {
 462     int len, i;
 463     if( dstlen == 0) return srclen;
 464     len = dstlen > srclen ? srclen : dstlen;
 465     for( i = 0; i < len; i++)
 466     {
 467         WCHAR w = src [ i ];
 468         if( w < 0x20 )
 469             dst[i] = w;
 470         else if( w >= 0xf020 && w < 0xf100)
 471             dst[i] = w - 0xf000;
 472         else
 473             return -2;
 474     }
 475     if( srclen > len) return -1;
 476     return len;
 477 }