unicode/wctomb.c

   1 /*
   2  * WideCharToMultiByte implementation
   3  *
   4  * Copyright 2000 Alexandre Julliard
   5  */
   6
   7 #include <string.h>
   8
   9 #include "winnls.h"
  10 #include "wine/unicode.h"
  11
  12 /* wcstombs for single-byte code page */
  13 static inline int wcstombs_sbcs( const struct sbcs_table *table,
  14                                  const WCHAR *src, unsigned int srclen,
  15                                  char *dst, unsigned int dstlen )
  16 {
  17     const unsigned char  * const uni2cp_low = table->uni2cp_low;
  18     const unsigned short * const uni2cp_high = table->uni2cp_high;
  19     int ret = srclen;
  20
  21     if (dstlen < srclen)
  22     {
  23         /* buffer too small: fill it up to dstlen and return error */
  24         srclen = dstlen;
  25         ret = -1;
  26     }
  27
  28     for (;;)
  29     {
  30         switch(srclen)
  31         {
  32         default:
  33         case 16: dst[15] = uni2cp_low[uni2cp_high[src[15] >> 8] + (src[15] & 0xff)];
  34         case 15: dst[14] = uni2cp_low[uni2cp_high[src[14] >> 8] + (src[14] & 0xff)];
  35         case 14: dst[13] = uni2cp_low[uni2cp_high[src[13] >> 8] + (src[13] & 0xff)];
  36         case 13: dst[12] = uni2cp_low[uni2cp_high[src[12] >> 8] + (src[12] & 0xff)];
  37         case 12: dst[11] = uni2cp_low[uni2cp_high[src[11] >> 8] + (src[11] & 0xff)];
  38         case 11: dst[10] = uni2cp_low[uni2cp_high[src[10] >> 8] + (src[10] & 0xff)];
  39         case 10: dst[9]  = uni2cp_low[uni2cp_high[src[9]  >> 8] + (src[9]  & 0xff)];
  40         case 9:  dst[8]  = uni2cp_low[uni2cp_high[src[8]  >> 8] + (src[8]  & 0xff)];
  41         case 8:  dst[7]  = uni2cp_low[uni2cp_high[src[7]  >> 8] + (src[7]  & 0xff)];
  42         case 7:  dst[6]  = uni2cp_low[uni2cp_high[src[6]  >> 8] + (src[6]  & 0xff)];
  43         case 6:  dst[5]  = uni2cp_low[uni2cp_high[src[5]  >> 8] + (src[5]  & 0xff)];
  44         case 5:  dst[4]  = uni2cp_low[uni2cp_high[src[4]  >> 8] + (src[4]  & 0xff)];
  45         case 4:  dst[3]  = uni2cp_low[uni2cp_high[src[3]  >> 8] + (src[3]  & 0xff)];
  46         case 3:  dst[2]  = uni2cp_low[uni2cp_high[src[2]  >> 8] + (src[2]  & 0xff)];
  47         case 2:  dst[1]  = uni2cp_low[uni2cp_high[src[1]  >> 8] + (src[1]  & 0xff)];
  48         case 1:  dst[0]  = uni2cp_low[uni2cp_high[src[0]  >> 8] + (src[0]  & 0xff)];
  49         case 0: break;
  50         }
  51         if (srclen < 16) return ret;
  52         dst += 16;
  53         src += 16;
  54         srclen -= 16;
  55     }
  56 }
  57
  58 /* slow version of wcstombs_sbcs that handles the various flags */
  59 static int wcstombs_sbcs_slow( const struct sbcs_table *table, int flags,
  60                                const WCHAR *src, unsigned int srclen,
  61                                char *dst, unsigned int dstlen,
  62                                const char *defchar, int *used )
  63 {
  64     const WCHAR * const cp2uni = table->cp2uni;
  65     const unsigned char  * const uni2cp_low = table->uni2cp_low;
  66     const unsigned short * const uni2cp_high = table->uni2cp_high;
  67     const unsigned char table_default = table->info.def_char & 0xff;
  68     int ret = srclen, tmp;
  69
  70     if (dstlen < srclen)
  71     {
  72         /* buffer too small: fill it up to dstlen and return error */
  73         srclen = dstlen;
  74         ret = -1;
  75     }
  76
  77     if (!defchar) defchar = &table_default;
  78     if (!used) used = &tmp;  /* avoid checking on every char */
  79
  80     while (srclen)
  81     {
  82         unsigned char ch = uni2cp_low[uni2cp_high[*src >> 8] + (*src & 0xff)];
  83         if (((flags & WC_NO_BEST_FIT_CHARS) && (cp2uni[ch] != *src)) ||
  84             (ch == table_default && *src != table->info.def_unicode_char))
  85         {
  86             ch = *defchar;
  87             *used = 1;
  88         }
  89         *dst++ = ch;
  90         src++;
  91         srclen--;
  92     }
  93     return ret;
  94 }
  95
  96 /* query necessary dst length for src string */
  97 static inline int get_length_dbcs( const struct dbcs_table *table,
  98                                    const WCHAR *src, unsigned int srclen )
  99 {
 100     const unsigned short * const uni2cp_low = table->uni2cp_low;
 101     const unsigned short * const uni2cp_high = table->uni2cp_high;
 102     int len;
 103
 104     for (len = 0; srclen; srclen--, src++, len++)
 105     {
 106         if (uni2cp_low[uni2cp_high[*src >> 8] + (*src & 0xff)] & 0xff00) len++;
 107     }
 108     return len;
 109 }
 110
 111 /* wcstombs for double-byte code page */
 112 static inline int wcstombs_dbcs( const struct dbcs_table *table,
 113                                  const WCHAR *src, unsigned int srclen,
 114                                  char *dst, unsigned int dstlen )
 115 {
 116     const unsigned short * const uni2cp_low = table->uni2cp_low;
 117     const unsigned short * const uni2cp_high = table->uni2cp_high;
 118     int len;
 119
 120     for (len = dstlen; srclen && len; len--, srclen--, src++)
 121     {
 122         unsigned short res = uni2cp_low[uni2cp_high[*src >> 8] + (*src & 0xff)];
 123         if (res & 0xff00)
 124         {
 125             if (len == 1) break;  /* do not output a partial char */
 126             len--;
 127             *dst++ = res >> 8;
 128         }
 129         *dst++ = (char)res;
 130     }
 131     if (srclen) return -1;  /* overflow */
 132     return dstlen - len;
 133 }
 134
 135 /* slow version of wcstombs_dbcs that handles the various flags */
 136 static int wcstombs_dbcs_slow( const struct dbcs_table *table, int flags,
 137                                const WCHAR *src, unsigned int srclen,
 138                                char *dst, unsigned int dstlen,
 139                                const char *defchar, int *used )
 140 {
 141     const WCHAR * const cp2uni = table->cp2uni;
 142     const unsigned short * const uni2cp_low = table->uni2cp_low;
 143     const unsigned short * const uni2cp_high = table->uni2cp_high;
 144     const unsigned char  * const cp2uni_lb = table->cp2uni_leadbytes;
 145     WCHAR defchar_value = table->info.def_char;
 146     int len, tmp;
 147
 148     if (defchar) defchar_value = defchar[1] ? ((defchar[0] << 8) | defchar[1]) : defchar[0];
 149     if (!used) used = &tmp;  /* avoid checking on every char */
 150
 151     for (len = dstlen; srclen && len; len--, srclen--, src++)
 152     {
 153         unsigned short res = uni2cp_low[uni2cp_high[*src >> 8] + (*src & 0xff)];
 154
 155         if (res == table->info.def_char && *src != table->info.def_unicode_char)
 156         {
 157             res = defchar_value;
 158             *used = 1;
 159         }
 160         else if (flags & WC_NO_BEST_FIT_CHARS)
 161         {
 162             /* check if char maps back to the same Unicode value */
 163             if (res & 0xff00)
 164             {
 165                 unsigned char off = cp2uni_lb[res >> 8];
 166                 if (cp2uni[(off << 8) + (res & 0xff)] != *src)
 167                 {
 168                     res = defchar_value;
 169                     *used = 1;
 170                 }
 171             }
 172             else if (cp2uni[res & 0xff] != *src)
 173             {
 174                 res = defchar_value;
 175                 *used = 1;
 176             }
 177         }
 178
 179         if (res & 0xff00)
 180         {
 181             if (len == 1) break;  /* do not output a partial char */
 182             len--;
 183             *dst++ = res >> 8;
 184         }
 185         *dst++ = (char)res;
 186     }
 187     if (srclen) return -1;  /* overflow */
 188     return dstlen - len;
 189 }
 190
 191 /* wide char to multi byte string conversion */
 192 /* return -1 on dst buffer overflow */
 193 int cp_wcstombs( const union cptable *table, int flags,
 194                  const WCHAR *src, int srclen,
 195                  char *dst, int dstlen, const char *defchar, int *used )
 196 {
 197     if (table->info.char_size == 1)
 198     {
 199         if (!dstlen) return srclen;
 200         if (flags || defchar || used)
 201             return wcstombs_sbcs_slow( &table->sbcs, flags, src, srclen,
 202                                        dst, dstlen, defchar, used );
 203         return wcstombs_sbcs( &table->sbcs, src, srclen, dst, dstlen );
 204     }
 205     else /* mbcs */
 206     {
 207         if (!dstlen) return get_length_dbcs( &table->dbcs, src, srclen );
 208         if (flags || defchar || used)
 209             return wcstombs_dbcs_slow( &table->dbcs, flags, src, srclen,
 210                                        dst, dstlen, defchar, used );
 211         return wcstombs_dbcs( &table->dbcs, src, srclen, dst, dstlen );
 212     }
 213 }