vxodbc/win_unicode.cc

   1 /*
   2  * Description: This module contains utf8 <-> ucs2 conversion routines
   3  *              under Windows
   4  */
   5 #include "psqlodbc.h"
   6 #ifdef  WIN32
   7 /* gcc <malloc.h> has been replaced by <stdlib.h> */
   8 #include <malloc.h>
   9 #ifdef  _DEBUG
  10 #ifndef _MEMORY_DEBUG_
  11 #include <stdlib.h>
  12 #define _CRTDBG_MAP_ALLOC
  13 #include <crtdbg.h>
  14 #endif                          /* _MEMORY_DEBUG_ */
  15 #endif                          /* _DEBUG */
  16 #else
  17 #include <stdlib.h>
  18 #endif                          /* WIN32 */
  19 #include <string.h>
  20
  21 #define byte3check      0xfffff800
  22 #define byte2_base      0x80c0
  23 #define byte2_mask1     0x07c0
  24 #define byte2_mask2     0x003f
  25 #define byte3_base      0x8080e0
  26 #define byte3_mask1     0xf000
  27 #define byte3_mask2     0x0fc0
  28 #define byte3_mask3     0x003f
  29
  30 #include <ctype.h>
  31 #ifndef WIN32
  32 #ifdef HAVE_ISWASCII
  33 #include <wctype.h>
  34 #else
  35 #include <wchar.h>
  36 int iswascii(wchar_t c)
  37 {
  38     return isascii(wctob(c));
  39 }
  40 #endif                          /* HAVE_ISWASCII */
  41 #endif                          /* WIN32 */
  42
  43 SQLULEN ucs2strlen(const SQLWCHAR * ucs2str)
  44 {
  45     SQLULEN len;
  46     for (len = 0; ucs2str[len]; len++)
  47         ;
  48     return len;
  49 }
  50 char *ucs2_to_utf8(const SQLWCHAR * ucs2str, SQLLEN ilen, SQLLEN * olen,
  51                    BOOL lower_identifier)
  52 {
  53     char *utf8str;
  54 /*mylog("ucs2_to_utf8 %p ilen=%d ", ucs2str, ilen);*/
  55
  56     if (!ucs2str)
  57     {
  58         *olen = SQL_NULL_DATA;
  59         return NULL;
  60     }
  61     if (SQL_NTS == ilen)
  62         ilen = ucs2strlen(ucs2str);
  63 /*mylog(" newlen=%d", ilen);*/
  64     utf8str = (char *) malloc(ilen * 3 + 1);
  65     if (utf8str)
  66     {
  67         int i, len = 0;
  68         UInt2 byte2code;
  69         Int4 byte4code;
  70         const SQLWCHAR *wstr;
  71
  72         for (i = 0, wstr = ucs2str; i < ilen; i++, wstr++)
  73         {
  74             if (!*wstr)
  75                 break;
  76             else if (0 == (*wstr & 0xffffff80)) /* ASCII */
  77             {
  78                 if (lower_identifier)
  79                     utf8str[len++] = (char) tolower(*wstr);
  80                 else
  81                     utf8str[len++] = (char) *wstr;
  82             } else if ((*wstr & byte3check) == 0)
  83             {
  84                 byte2code = byte2_base |
  85                     ((byte2_mask1 & *wstr) >> 6) |
  86                     ((byte2_mask2 & *wstr) << 8);
  87                 memcpy(utf8str + len, (char *) &byte2code,
  88                        sizeof(byte2code));
  89                 len += sizeof(byte2code);
  90             } else
  91             {
  92                 byte4code = byte3_base |
  93                     ((byte3_mask1 & *wstr) >> 12) |
  94                     ((byte3_mask2 & *wstr) << 2) |
  95                     ((byte3_mask3 & *wstr) << 16);
  96                 memcpy(utf8str + len, (char *) &byte4code, 3);
  97                 len += 3;
  98             }
  99         }
 100         utf8str[len] = '\0';
 101         if (olen)
 102             *olen = len;
 103     }
 104 /*mylog(" olen=%d %s\n", *olen, utf8str ? utf8str : "");*/
 105     return utf8str;
 106 }
 107
 108 #define byte3_m1        0x0f
 109 #define byte3_m2        0x3f
 110 #define byte3_m3        0x3f
 111 #define byte2_m1        0x1f
 112 #define byte2_m2        0x3f
 113 SQLULEN utf8_to_ucs2_lf(const char *utf8str, SQLLEN ilen, BOOL lfconv,
 114                         SQLWCHAR * ucs2str, SQLULEN bufcount)
 115 {
 116     int i;
 117     SQLULEN ocount, wcode;
 118     const UCHAR *str;
 119
 120 /*mylog("utf8_to_ucs2 ilen=%d bufcount=%d", ilen, bufcount);*/
 121     if (!utf8str)
 122         return 0;
 123 /*mylog(" string=%s\n", utf8str);*/
 124     if (!bufcount)
 125         ucs2str = NULL;
 126     else if (!ucs2str)
 127         bufcount = 0;
 128     if (ilen < 0)
 129         ilen = strlen(utf8str);
 130     for (i = 0, ocount = 0, str = (const UCHAR *)utf8str; i < ilen && *str;)
 131     {
 132         /* if (iswascii(*str)) */
 133         if (isascii(*str))
 134         {
 135             if (lfconv && PG_LINEFEED == *str &&
 136                 (i == 0 || PG_CARRIAGE_RETURN != str[-1]))
 137             {
 138                 if (ocount < bufcount)
 139                     ucs2str[ocount] = PG_CARRIAGE_RETURN;
 140                 ocount++;
 141             }
 142             if (ocount < bufcount)
 143                 ucs2str[ocount] = *str;
 144             ocount++;
 145             i++;
 146             str++;
 147         } else if (0xe0 == (*str & 0xe0))       /* 3 byte code */
 148         {
 149             if (ocount < bufcount)
 150             {
 151                 wcode = ((((UInt4) * str) & byte3_m1) << 12) |
 152                     ((((UInt4) str[1]) & byte3_m2) << 6) |
 153                     (((UInt4) str[2]) & byte3_m3);
 154                 ucs2str[ocount] = (SQLWCHAR) wcode;
 155             }
 156             ocount++;
 157             i += 3;
 158             str += 3;
 159         } else
 160         {
 161             if (ocount < bufcount)
 162             {
 163                 wcode = ((((UInt4) * str) & byte2_m1) << 6) |
 164                     (((UInt4) str[1]) & byte2_m2);
 165                 ucs2str[ocount] = (SQLWCHAR) wcode;
 166             }
 167             ocount++;
 168             i += 2;
 169             str += 2;
 170         }
 171     }
 172     if (ocount < bufcount && ucs2str)
 173         ucs2str[ocount] = 0;
 174 /*mylog(" ocount=%d\n", ocount);*/
 175     return ocount;
 176 }