components/iconv/ucs2_to_gb2312.c

   1 /**************************************************************************
   2  *              Copyright (C), AirM2M Tech. Co., Ltd.
   3  *
   4  * Name:    ucs2_to_gb2312.c
   5  * Author:  liweiqiang
   6  * Version: V0.1
   7  * Date:    2013/7/15
   8  *
   9  * Description:
  10  *          ucs2 ×ª»» gb2312
  11  **************************************************************************/
  12
  13 #include <stdint.h>
  14 #include "stdio.h"
  15 #include "errno.h"
  16
  17 #include "ucs2_to_gb2312_table.h"
  18 #include "ucs2_to_gb2312_offset.h"
  19
  20 const uint8_t number_of_bit_1[256] =
  21 {
  22     0x00, 0x01, 0x01, 0x02, 0x01, 0x02, 0x02, 0x03,
  23     0x01, 0x02, 0x02, 0x03, 0x02, 0x03, 0x03, 0x04,
  24     0x01, 0x02, 0x02, 0x03, 0x02, 0x03, 0x03, 0x04,
  25     0x02, 0x03, 0x03, 0x04, 0x03, 0x04, 0x04, 0x05,
  26     0x01, 0x02, 0x02, 0x03, 0x02, 0x03, 0x03, 0x04,
  27     0x02, 0x03, 0x03, 0x04, 0x03, 0x04, 0x04, 0x05,
  28     0x02, 0x03, 0x03, 0x04, 0x03, 0x04, 0x04, 0x05,
  29     0x03, 0x04, 0x04, 0x05, 0x04, 0x05, 0x05, 0x06,
  30     0x01, 0x02, 0x02, 0x03, 0x02, 0x03, 0x03, 0x04,
  31     0x02, 0x03, 0x03, 0x04, 0x03, 0x04, 0x04, 0x05,
  32     0x02, 0x03, 0x03, 0x04, 0x03, 0x04, 0x04, 0x05,
  33     0x03, 0x04, 0x04, 0x05, 0x04, 0x05, 0x05, 0x06,
  34     0x02, 0x03, 0x03, 0x04, 0x03, 0x04, 0x04, 0x05,
  35     0x03, 0x04, 0x04, 0x05, 0x04, 0x05, 0x05, 0x06,
  36     0x03, 0x04, 0x04, 0x05, 0x04, 0x05, 0x05, 0x06,
  37     0x04, 0x05, 0x05, 0x06, 0x05, 0x06, 0x06, 0x07,
  38     0x01, 0x02, 0x02, 0x03, 0x02, 0x03, 0x03, 0x04,
  39     0x02, 0x03, 0x03, 0x04, 0x03, 0x04, 0x04, 0x05,
  40     0x02, 0x03, 0x03, 0x04, 0x03, 0x04, 0x04, 0x05,
  41     0x03, 0x04, 0x04, 0x05, 0x04, 0x05, 0x05, 0x06,
  42     0x02, 0x03, 0x03, 0x04, 0x03, 0x04, 0x04, 0x05,
  43     0x03, 0x04, 0x04, 0x05, 0x04, 0x05, 0x05, 0x06,
  44     0x03, 0x04, 0x04, 0x05, 0x04, 0x05, 0x05, 0x06,
  45     0x04, 0x05, 0x05, 0x06, 0x05, 0x06, 0x06, 0x07,
  46     0x02, 0x03, 0x03, 0x04, 0x03, 0x04, 0x04, 0x05,
  47     0x03, 0x04, 0x04, 0x05, 0x04, 0x05, 0x05, 0x06,
  48     0x03, 0x04, 0x04, 0x05, 0x04, 0x05, 0x05, 0x06,
  49     0x04, 0x05, 0x05, 0x06, 0x05, 0x06, 0x06, 0x07,
  50     0x03, 0x04, 0x04, 0x05, 0x04, 0x05, 0x05, 0x06,
  51     0x04, 0x05, 0x05, 0x06, 0x05, 0x06, 0x06, 0x07,
  52     0x04, 0x05, 0x05, 0x06, 0x05, 0x06, 0x06, 0x07,
  53     0x05, 0x06, 0x06, 0x07, 0x06, 0x07, 0x07, 0x08,
  54 };
  55
  56 /* 0x4E00 <= ucs2 < 0xA000 */
  57 static uint16_t get_ucs2_offset(uint16_t ucs2)
  58 {
  59     uint16_t   offset, page, tmp;
  60     uint8_t    *mirror_ptr, ch;
  61
  62     page = (ucs2>>8) - 0x4E;
  63     ucs2 &= 0xFF;
  64
  65     tmp        = ucs2>>6; /* now 0 <= tmp < 4  */
  66     offset     = ucs2_index_table_4E00_9FFF[page][tmp];
  67     mirror_ptr = (uint8_t*)&ucs2_mirror_4E00_9FFF[page][tmp<<3]; /* [0, 8, 16, 24] */
  68
  69     tmp = ucs2&0x3F; /* mod 64 */
  70
  71     while(tmp >= 8)
  72     {
  73         offset += number_of_bit_1[*mirror_ptr];
  74         mirror_ptr++;
  75         tmp -= 8;
  76     }
  77
  78     ch = *mirror_ptr;
  79     if(ch&(0x1<<tmp))
  80     {   /* Ok , this ucs2 can be covert to GB2312. */
  81         while(tmp)
  82         {
  83             if(ch&0x1)
  84             offset++;
  85             ch>>=1;
  86             tmp--;
  87         }
  88         return offset;
  89     }
  90
  91     return (uint16_t)(-1);
  92 }
  93
  94 uint16_t unicode_to_gb2312(uint16_t ucs2, uint8_t marks)
  95 {
  96         uint16_t gb = 0xA1A1;
  97         if(0x80 > ucs2)
  98     {
  99         // can be convert to ASCII char
 100         gb = ucs2;
 101     }
 102     else
 103     {
 104         if((0x4E00 <= ucs2) && (0xA000 > ucs2))
 105         {
 106             uint16_t offset = get_ucs2_offset(ucs2);
 107             if((uint16_t)(-1) != offset)
 108             {
 109                 gb = ucs2_to_gb2312_table[offset];
 110             }
 111         }
 112         else if (marks == 0)
 113         {
 114             uint16_t u16count = sizeof(tab_UCS2_to_GBK)/4;
 115             for(uint16_t ui=0; ui < u16count; ui++)
 116             {
 117                 if(ucs2 == tab_UCS2_to_GBK[ui][0])
 118                 {
 119                     gb = tab_UCS2_to_GBK[ui][1];
 120                 }
 121             }
 122
 123         }
 124     }
 125         return gb;
 126 }
 127
 128 /*+\NEW\liweiqiang\2013.11.26\ÍêÉÆgb2312<->ucs2(ucs2be)±àÂë×ª»»*/
 129 static size_t iconv_ucs2_to_gb2312_endian(char **_inbuf, size_t *inbytesleft, char **_outbuf, size_t *outbytesleft, int endian)
 130 {
 131     uint16_t gb2312 = 0xA1A1;
 132     uint16_t ucs2;
 133     size_t gb_length = 0;
 134     uint16_t *ucs2buf = (uint16_t*)*_inbuf;
 135     char *outbuf = (char *)*_outbuf;
 136     size_t inlen = *inbytesleft/2;
 137     size_t outlen = *outbytesleft;
 138     size_t ret = 0;
 139
 140     while(inlen > 0)
 141     {
 142         if(gb_length+2 > outlen)
 143         {
 144             errno = E2BIG;
 145             ret = -1;
 146             goto ucs2_to_gb2312_exit;
 147         }
 148
 149         ucs2 = *ucs2buf++;
 150
 151         if(endian == 1)
 152             ucs2 = (ucs2<<8)|(ucs2>>8);
 153
 154         gb2312 = unicode_to_gb2312(ucs2, 0);
 155         //End 7205
 156
 157         if(0x80 > gb2312)
 158         {
 159             // can be convert to ASCII char
 160             *outbuf++ = (uint8_t)gb2312;
 161             gb_length++;
 162         }
 163         else
 164         {
 165             *outbuf++ = (uint8_t)(gb2312>>8);
 166             *outbuf++ = (uint8_t)(gb2312);
 167             gb_length += 2;
 168         }
 169
 170         inlen--;
 171     }
 172
 173     if(inlen > 0)
 174     {
 175         errno = EINVAL;
 176         ret = -1;
 177     }
 178
 179 ucs2_to_gb2312_exit:
 180     *inbytesleft = inlen;
 181     *outbytesleft -= gb_length;
 182
 183     return ret;
 184 }
 185
 186 size_t iconv_ucs2_to_gb2312(char **_inbuf, size_t *inbytesleft, char **_outbuf, size_t *outbytesleft)
 187 {
 188     return iconv_ucs2_to_gb2312_endian(_inbuf, inbytesleft, _outbuf, outbytesleft, 0);
 189 }
 190
 191 size_t iconv_ucs2be_to_gb2312(char **_inbuf, size_t *inbytesleft, char **_outbuf, size_t *outbytesleft)
 192 {
 193     return iconv_ucs2_to_gb2312_endian(_inbuf, inbytesleft, _outbuf, outbytesleft, 1);
 194 }
 195 /*-\NEW\liweiqiang\2013.11.26\ÍêÉÆgb2312<->ucs2(ucs2be)±àÂë×ª»»*/
 196