libs/wine/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., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
  19  */
  20
  21 #include <string.h>
  22
  23 #include "wine/asm.h"
  24
  25 #ifdef __ASM_OBSOLETE
  26
  27 #include "unicode.h"
  28
  29 extern WCHAR wine_compose( const WCHAR *str ) DECLSPEC_HIDDEN;
  30
  31 /****************************************************************/
  32 /* sbcs support */
  33
  34 /* check if 'ch' is an acceptable sbcs mapping for 'wch' */
  35 static inline int is_valid_sbcs_mapping( const struct sbcs_table *table, int flags,
  36                                          WCHAR wch, unsigned char ch )
  37 {
  38     if ((flags & WC_NO_BEST_FIT_CHARS) || ch == (unsigned char)table->info.def_char)
  39         return (table->cp2uni[ch] == wch);
  40     return 1;
  41 }
  42
  43 /* query necessary dst length for src string */
  44 static int get_length_sbcs( const struct sbcs_table *table, int flags,
  45                             const WCHAR *src, unsigned int srclen, int *used )
  46 {
  47     const unsigned char  * const uni2cp_low = table->uni2cp_low;
  48     const unsigned short * const uni2cp_high = table->uni2cp_high;
  49     int ret, tmp;
  50     WCHAR composed;
  51
  52     if (!used) used = &tmp;  /* avoid checking on every char */
  53     *used = 0;
  54
  55     for (ret = 0; srclen; ret++, src++, srclen--)
  56     {
  57         WCHAR wch = *src;
  58         unsigned char ch;
  59
  60         if ((flags & WC_COMPOSITECHECK) && (srclen > 1) && (composed = wine_compose(src)))
  61         {
  62             /* now check if we can use the composed char */
  63             ch = uni2cp_low[uni2cp_high[composed >> 8] + (composed & 0xff)];
  64             if (is_valid_sbcs_mapping( table, flags, composed, ch ))
  65             {
  66                 /* we have a good mapping, use it */
  67                 src++;
  68                 srclen--;
  69                 continue;
  70             }
  71             /* no mapping for the composed char, check the other flags */
  72             if (flags & WC_DEFAULTCHAR) /* use the default char instead */
  73             {
  74                 *used = 1;
  75                 src++;  /* skip the non-spacing char */
  76                 srclen--;
  77                 continue;
  78             }
  79             if (flags & WC_DISCARDNS) /* skip the second char of the composition */
  80             {
  81                 src++;
  82                 srclen--;
  83             }
  84             /* WC_SEPCHARS is the default */
  85         }
  86         if (!*used)
  87         {
  88             ch = uni2cp_low[uni2cp_high[wch >> 8] + (wch & 0xff)];
  89             *used = !is_valid_sbcs_mapping( table, flags, wch, ch );
  90         }
  91     }
  92     return ret;
  93 }
  94
  95 /* wcstombs for single-byte code page */
  96 static inline int wcstombs_sbcs( const struct sbcs_table *table,
  97                                  const WCHAR *src, unsigned int srclen,
  98                                  char *dst, unsigned int dstlen )
  99 {
 100     const unsigned char  * const uni2cp_low = table->uni2cp_low;
 101     const unsigned short * const uni2cp_high = table->uni2cp_high;
 102     int ret = srclen;
 103
 104     if (dstlen < srclen)
 105     {
 106         /* buffer too small: fill it up to dstlen and return error */
 107         srclen = dstlen;
 108         ret = -1;
 109     }
 110
 111     while (srclen >= 16)
 112     {
 113         dst[0]  = uni2cp_low[uni2cp_high[src[0]  >> 8] + (src[0]  & 0xff)];
 114         dst[1]  = uni2cp_low[uni2cp_high[src[1]  >> 8] + (src[1]  & 0xff)];
 115         dst[2]  = uni2cp_low[uni2cp_high[src[2]  >> 8] + (src[2]  & 0xff)];
 116         dst[3]  = uni2cp_low[uni2cp_high[src[3]  >> 8] + (src[3]  & 0xff)];
 117         dst[4]  = uni2cp_low[uni2cp_high[src[4]  >> 8] + (src[4]  & 0xff)];
 118         dst[5]  = uni2cp_low[uni2cp_high[src[5]  >> 8] + (src[5]  & 0xff)];
 119         dst[6]  = uni2cp_low[uni2cp_high[src[6]  >> 8] + (src[6]  & 0xff)];
 120         dst[7]  = uni2cp_low[uni2cp_high[src[7]  >> 8] + (src[7]  & 0xff)];
 121         dst[8]  = uni2cp_low[uni2cp_high[src[8]  >> 8] + (src[8]  & 0xff)];
 122         dst[9]  = uni2cp_low[uni2cp_high[src[9]  >> 8] + (src[9]  & 0xff)];
 123         dst[10] = uni2cp_low[uni2cp_high[src[10] >> 8] + (src[10] & 0xff)];
 124         dst[11] = uni2cp_low[uni2cp_high[src[11] >> 8] + (src[11] & 0xff)];
 125         dst[12] = uni2cp_low[uni2cp_high[src[12] >> 8] + (src[12] & 0xff)];
 126         dst[13] = uni2cp_low[uni2cp_high[src[13] >> 8] + (src[13] & 0xff)];
 127         dst[14] = uni2cp_low[uni2cp_high[src[14] >> 8] + (src[14] & 0xff)];
 128         dst[15] = uni2cp_low[uni2cp_high[src[15] >> 8] + (src[15] & 0xff)];
 129         src += 16;
 130         dst += 16;
 131         srclen -= 16;
 132     }
 133
 134     /* now handle remaining characters */
 135     src += srclen;
 136     dst += srclen;
 137     switch(srclen)
 138     {
 139     case 15: dst[-15] = uni2cp_low[uni2cp_high[src[-15] >> 8] + (src[-15] & 0xff)];
 140     case 14: dst[-14] = uni2cp_low[uni2cp_high[src[-14] >> 8] + (src[-14] & 0xff)];
 141     case 13: dst[-13] = uni2cp_low[uni2cp_high[src[-13] >> 8] + (src[-13] & 0xff)];
 142     case 12: dst[-12] = uni2cp_low[uni2cp_high[src[-12] >> 8] + (src[-12] & 0xff)];
 143     case 11: dst[-11] = uni2cp_low[uni2cp_high[src[-11] >> 8] + (src[-11] & 0xff)];
 144     case 10: dst[-10] = uni2cp_low[uni2cp_high[src[-10] >> 8] + (src[-10] & 0xff)];
 145     case 9:  dst[-9]  = uni2cp_low[uni2cp_high[src[-9]  >> 8] + (src[-9]  & 0xff)];
 146     case 8:  dst[-8]  = uni2cp_low[uni2cp_high[src[-8]  >> 8] + (src[-8]  & 0xff)];
 147     case 7:  dst[-7]  = uni2cp_low[uni2cp_high[src[-7]  >> 8] + (src[-7]  & 0xff)];
 148     case 6:  dst[-6]  = uni2cp_low[uni2cp_high[src[-6]  >> 8] + (src[-6]  & 0xff)];
 149     case 5:  dst[-5]  = uni2cp_low[uni2cp_high[src[-5]  >> 8] + (src[-5]  & 0xff)];
 150     case 4:  dst[-4]  = uni2cp_low[uni2cp_high[src[-4]  >> 8] + (src[-4]  & 0xff)];
 151     case 3:  dst[-3]  = uni2cp_low[uni2cp_high[src[-3]  >> 8] + (src[-3]  & 0xff)];
 152     case 2:  dst[-2]  = uni2cp_low[uni2cp_high[src[-2]  >> 8] + (src[-2]  & 0xff)];
 153     case 1:  dst[-1]  = uni2cp_low[uni2cp_high[src[-1]  >> 8] + (src[-1]  & 0xff)];
 154     case 0: break;
 155     }
 156     return ret;
 157 }
 158
 159 /* slow version of wcstombs_sbcs that handles the various flags */
 160 static int wcstombs_sbcs_slow( const struct sbcs_table *table, int flags,
 161                                const WCHAR *src, unsigned int srclen,
 162                                char *dst, unsigned int dstlen,
 163                                const char *defchar, int *used )
 164 {
 165     const unsigned char  * const uni2cp_low = table->uni2cp_low;
 166     const unsigned short * const uni2cp_high = table->uni2cp_high;
 167     unsigned char def;
 168     unsigned int len;
 169     int tmp;
 170     WCHAR composed;
 171
 172     if (!defchar)
 173         def = table->info.def_char & 0xff;
 174     else
 175         def = *defchar;
 176
 177     if (!used) used = &tmp;  /* avoid checking on every char */
 178     *used = 0;
 179
 180     for (len = dstlen; srclen && len; dst++, len--, src++, srclen--)
 181     {
 182         WCHAR wch = *src;
 183
 184         if ((flags & WC_COMPOSITECHECK) && (srclen > 1) && (composed = wine_compose(src)))
 185         {
 186             /* now check if we can use the composed char */
 187             *dst = uni2cp_low[uni2cp_high[composed >> 8] + (composed & 0xff)];
 188             if (is_valid_sbcs_mapping( table, flags, composed, *dst ))
 189             {
 190                 /* we have a good mapping, use it */
 191                 src++;
 192                 srclen--;
 193                 continue;
 194             }
 195             /* no mapping for the composed char, check the other flags */
 196             if (flags & WC_DEFAULTCHAR) /* use the default char instead */
 197             {
 198                 *dst = def;
 199                 *used = 1;
 200                 src++;  /* skip the non-spacing char */
 201                 srclen--;
 202                 continue;
 203             }
 204             if (flags & WC_DISCARDNS) /* skip the second char of the composition */
 205             {
 206                 src++;
 207                 srclen--;
 208             }
 209             /* WC_SEPCHARS is the default */
 210         }
 211
 212         *dst = uni2cp_low[uni2cp_high[wch >> 8] + (wch & 0xff)];
 213         if (!is_valid_sbcs_mapping( table, flags, wch, *dst ))
 214         {
 215             *dst = def;
 216             *used = 1;
 217         }
 218     }
 219     if (srclen) return -1;  /* overflow */
 220     return dstlen - len;
 221 }
 222
 223
 224 /****************************************************************/
 225 /* dbcs support */
 226
 227 /* check if 'ch' is an acceptable dbcs mapping for 'wch' */
 228 static inline int is_valid_dbcs_mapping( const struct dbcs_table *table, int flags,
 229                                          WCHAR wch, unsigned short ch )
 230 {
 231     if ((flags & WC_NO_BEST_FIT_CHARS) || ch == table->info.def_char)
 232     {
 233         /* check if char maps back to the same Unicode value */
 234         if (ch & 0xff00)
 235         {
 236             unsigned char off = table->cp2uni_leadbytes[ch >> 8];
 237             return (table->cp2uni[(off << 8) + (ch & 0xff)] == wch);
 238         }
 239         return (table->cp2uni[ch & 0xff] == wch);
 240     }
 241     return 1;
 242 }
 243
 244 /* compute the default char for the dbcs case */
 245 static inline WCHAR get_defchar_dbcs( const struct dbcs_table *table, const char *defchar )
 246 {
 247     if (!defchar) return table->info.def_char;
 248     if (!defchar[1]) return (unsigned char)defchar[0];
 249     return ((unsigned char)defchar[0] << 8) | (unsigned char)defchar[1];
 250 }
 251
 252 /* query necessary dst length for src string */
 253 static int get_length_dbcs( const struct dbcs_table *table, int flags,
 254                             const WCHAR *src, unsigned int srclen,
 255                             const char *defchar, int *used )
 256 {
 257     const unsigned short * const uni2cp_low = table->uni2cp_low;
 258     const unsigned short * const uni2cp_high = table->uni2cp_high;
 259     WCHAR defchar_value, composed;
 260     int len, tmp;
 261
 262     if (!defchar && !used && !(flags & WC_COMPOSITECHECK))
 263     {
 264         for (len = 0; srclen; srclen--, src++, len++)
 265         {
 266             if (uni2cp_low[uni2cp_high[*src >> 8] + (*src & 0xff)] & 0xff00) len++;
 267         }
 268         return len;
 269     }
 270
 271     defchar_value = get_defchar_dbcs( table, defchar );
 272     if (!used) used = &tmp;  /* avoid checking on every char */
 273     *used = 0;
 274     for (len = 0; srclen; len++, srclen--, src++)
 275     {
 276         unsigned short res;
 277         WCHAR wch = *src;
 278
 279         if ((flags & WC_COMPOSITECHECK) && (srclen > 1) && (composed = wine_compose(src)))
 280         {
 281             /* now check if we can use the composed char */
 282             res = uni2cp_low[uni2cp_high[composed >> 8] + (composed & 0xff)];
 283
 284             if (is_valid_dbcs_mapping( table, flags, composed, res ))
 285             {
 286                 /* we have a good mapping for the composed char, use it */
 287                 if (res & 0xff00) len++;
 288                 src++;
 289                 srclen--;
 290                 continue;
 291             }
 292             /* no mapping for the composed char, check the other flags */
 293             if (flags & WC_DEFAULTCHAR) /* use the default char instead */
 294             {
 295                 if (defchar_value & 0xff00) len++;
 296                 *used = 1;
 297                 src++;  /* skip the non-spacing char */
 298                 srclen--;
 299                 continue;
 300             }
 301             if (flags & WC_DISCARDNS) /* skip the second char of the composition */
 302             {
 303                 src++;
 304                 srclen--;
 305             }
 306             /* WC_SEPCHARS is the default */
 307         }
 308
 309         res = uni2cp_low[uni2cp_high[wch >> 8] + (wch & 0xff)];
 310         if (!is_valid_dbcs_mapping( table, flags, wch, res ))
 311         {
 312             res = defchar_value;
 313             *used = 1;
 314         }
 315         if (res & 0xff00) len++;
 316     }
 317     return len;
 318 }
 319
 320 /* wcstombs for double-byte code page */
 321 static inline int wcstombs_dbcs( const struct dbcs_table *table,
 322                                  const WCHAR *src, unsigned int srclen,
 323                                  char *dst, unsigned int dstlen )
 324 {
 325     const unsigned short * const uni2cp_low = table->uni2cp_low;
 326     const unsigned short * const uni2cp_high = table->uni2cp_high;
 327     int len;
 328
 329     for (len = dstlen; srclen && len; len--, srclen--, src++)
 330     {
 331         unsigned short res = uni2cp_low[uni2cp_high[*src >> 8] + (*src & 0xff)];
 332         if (res & 0xff00)
 333         {
 334             if (len == 1) break;  /* do not output a partial char */
 335             len--;
 336             *dst++ = res >> 8;
 337         }
 338         *dst++ = (char)res;
 339     }
 340     if (srclen) return -1;  /* overflow */
 341     return dstlen - len;
 342 }
 343
 344 /* slow version of wcstombs_dbcs that handles the various flags */
 345 static int wcstombs_dbcs_slow( const struct dbcs_table *table, int flags,
 346                                const WCHAR *src, unsigned int srclen,
 347                                char *dst, unsigned int dstlen,
 348                                const char *defchar, int *used )
 349 {
 350     const unsigned short * const uni2cp_low = table->uni2cp_low;
 351     const unsigned short * const uni2cp_high = table->uni2cp_high;
 352     WCHAR defchar_value = get_defchar_dbcs( table, defchar );
 353     WCHAR composed;
 354     int len, tmp;
 355
 356     if (!used) used = &tmp;  /* avoid checking on every char */
 357     *used = 0;
 358
 359     for (len = dstlen; srclen && len; len--, srclen--, src++)
 360     {
 361         unsigned short res;
 362         WCHAR wch = *src;
 363
 364         if ((flags & WC_COMPOSITECHECK) && (srclen > 1) && (composed = wine_compose(src)))
 365         {
 366             /* now check if we can use the composed char */
 367             res = uni2cp_low[uni2cp_high[composed >> 8] + (composed & 0xff)];
 368
 369             if (is_valid_dbcs_mapping( table, flags, composed, res ))
 370             {
 371                 /* we have a good mapping for the composed char, use it */
 372                 src++;
 373                 srclen--;
 374                 goto output_char;
 375             }
 376             /* no mapping for the composed char, check the other flags */
 377             if (flags & WC_DEFAULTCHAR) /* use the default char instead */
 378             {
 379                 res = defchar_value;
 380                 *used = 1;
 381                 src++;  /* skip the non-spacing char */
 382                 srclen--;
 383                 goto output_char;
 384             }
 385             if (flags & WC_DISCARDNS) /* skip the second char of the composition */
 386             {
 387                 src++;
 388                 srclen--;
 389             }
 390             /* WC_SEPCHARS is the default */
 391         }
 392
 393         res = uni2cp_low[uni2cp_high[wch >> 8] + (wch & 0xff)];
 394         if (!is_valid_dbcs_mapping( table, flags, wch, res ))
 395         {
 396             res = defchar_value;
 397             *used = 1;
 398         }
 399
 400     output_char:
 401         if (res & 0xff00)
 402         {
 403             if (len == 1) break;  /* do not output a partial char */
 404             len--;
 405             *dst++ = res >> 8;
 406         }
 407         *dst++ = (char)res;
 408     }
 409     if (srclen) return -1;  /* overflow */
 410     return dstlen - len;
 411 }
 412
 413 /* wide char to multi byte string conversion */
 414 /* return -1 on dst buffer overflow */
 415 int wine_cp_wcstombs_obsolete( const union cptable *table, int flags,
 416                                const WCHAR *src, int srclen,
 417                                char *dst, int dstlen, const char *defchar, int *used )
 418 {
 419     if (table->info.char_size == 1)
 420     {
 421         if (flags || defchar || used)
 422         {
 423             if (!dstlen) return get_length_sbcs( &table->sbcs, flags, src, srclen, used );
 424             return wcstombs_sbcs_slow( &table->sbcs, flags, src, srclen,
 425                                        dst, dstlen, defchar, used );
 426         }
 427         if (!dstlen) return srclen;
 428         return wcstombs_sbcs( &table->sbcs, src, srclen, dst, dstlen );
 429     }
 430     else /* mbcs */
 431     {
 432         if (!dstlen) return get_length_dbcs( &table->dbcs, flags, src, srclen, defchar, used );
 433         if (flags || defchar || used)
 434             return wcstombs_dbcs_slow( &table->dbcs, flags, src, srclen,
 435                                        dst, dstlen, defchar, used );
 436         return wcstombs_dbcs( &table->dbcs, src, srclen, dst, dstlen );
 437     }
 438 }
 439
 440 __ASM_OBSOLETE(wine_cp_wcstombs);
 441
 442 #endif /* __ASM_OBSOLETE */