cp932ext.h

libiconv是一个很不错的字符集转换库。程序接口也很简单

  {  293, 0x4000 }, {  294, 0x0000 }, {  294, 0x0010 }, {  295, 0x0400 },
  {  296, 0x0000 }, {  296, 0x0000 }, {  296, 0x0000 }, {  296, 0x0000 },
  {  296, 0x0000 }, {  296, 0x0000 }, {  296, 0x0000 }, {  296, 0x0000 },
  /* 0x7900 */
  {  296, 0x0000 }, {  296, 0x0000 }, {  296, 0x0000 }, {  296, 0x0001 },
  {  297, 0x0000 }, {  297, 0x0000 }, {  297, 0x0000 }, {  297, 0x0000 },
  {  297, 0x0000 }, {  297, 0x0810 }, {  299, 0x0000 }, {  299, 0x0000 },
  {  299, 0x0000 }, {  299, 0x0000 }, {  299, 0x0000 }, {  299, 0x0000 },
  /* 0x7a00 */
  {  299, 0x0000 }, {  299, 0x0000 }, {  299, 0x0000 }, {  299, 0x0000 },
  {  299, 0x0000 }, {  299, 0x0000 }, {  299, 0x0000 }, {  299, 0x0000 },
  {  299, 0x0000 }, {  299, 0x0000 }, {  299, 0x0000 }, {  299, 0x0000 },
  {  299, 0x0000 }, {  299, 0x0002 }, {  300, 0x0880 }, {  302, 0x0000 },
  /* 0x7b00 */
  {  302, 0x0000 }, {  302, 0x0000 }, {  302, 0x0000 }, {  302, 0x0000 },
  {  302, 0x0000 }, {  302, 0x0000 }, {  302, 0x0000 }, {  302, 0x0000 },
  {  302, 0x0000 }, {  302, 0x4000 },
};
static const Summary16 cp932ext_uni2indx_page7d[43] = {
  /* 0x7d00 */
  {  303, 0x0000 }, {  303, 0x0000 }, {  303, 0x0000 }, {  303, 0x0000 },
  {  303, 0x0100 }, {  304, 0x1000 }, {  305, 0x0000 }, {  305, 0x0000 },
  {  305, 0x0000 }, {  305, 0x0000 }, {  305, 0x0001 }, {  306, 0x0080 },
  {  307, 0x0000 }, {  307, 0x0040 }, {  308, 0x0000 }, {  308, 0x0000 },
  /* 0x7e00 */
  {  308, 0x0000 }, {  308, 0x0000 }, {  308, 0x0000 }, {  308, 0x0000 },
  {  308, 0x0000 }, {  308, 0x0004 }, {  309, 0x0000 }, {  309, 0x0000 },
  {  309, 0x0400 }, {  310, 0x0000 }, {  310, 0x0000 }, {  310, 0x0000 },
  {  310, 0x0000 }, {  310, 0x0000 }, {  310, 0x0000 }, {  310, 0x0000 },
  /* 0x7f00 */
  {  310, 0x0000 }, {  310, 0x0000 }, {  310, 0x0000 }, {  310, 0x0000 },
  {  310, 0x0080 }, {  311, 0x0000 }, {  311, 0x0000 }, {  311, 0x0000 },
  {  311, 0x0000 }, {  311, 0x0000 }, {  311, 0x0002 },
};
static const Summary16 cp932ext_uni2indx_page83[44] = {
  /* 0x8300 */
  {  312, 0x0002 }, {  313, 0x0000 }, {  313, 0x0000 }, {  313, 0x0000 },
  {  313, 0x0000 }, {  313, 0x0000 }, {  313, 0x0004 }, {  314, 0x8000 },
  {  315, 0x0000 }, {  315, 0x0000 }, {  315, 0x0000 }, {  315, 0x0000 },
  {  315, 0x0080 }, {  316, 0x0000 }, {  316, 0x0000 }, {  316, 0x0040 },
  /* 0x8400 */
  {  317, 0x0000 }, {  317, 0x0000 }, {  317, 0x0000 }, {  317, 0x0000 },
  {  317, 0x0100 }, {  318, 0x0000 }, {  318, 0x0000 }, {  318, 0x0000 },
  {  318, 0x0000 }, {  318, 0x0000 }, {  318, 0x0000 }, {  318, 0x0010 },
  {  319, 0x0000 }, {  319, 0x1000 }, {  320, 0x0000 }, {  320, 0x0000 },
  /* 0x8500 */
  {  320, 0x0000 }, {  320, 0x0000 }, {  320, 0x0000 }, {  320, 0x0000 },
  {  320, 0x0000 }, {  320, 0x0208 }, {  322, 0x0800 }, {  323, 0x0000 },
  {  323, 0x0000 }, {  323, 0x0000 }, {  323, 0x0000 }, {  323, 0x0001 },
};
static const Summary16 cp932ext_uni2indx_page88[109] = {
  /* 0x8800 */
  {  324, 0x0080 }, {  325, 0x0000 }, {  325, 0x0000 }, {  325, 0x0000 },
  {  325, 0x0000 }, {  325, 0x0000 }, {  325, 0x0000 }, {  325, 0x0000 },
  {  325, 0x0000 }, {  325, 0x0000 }, {  325, 0x0000 }, {  325, 0x0000 },
  {  325, 0x0000 }, {  325, 0x0000 }, {  325, 0x0000 }, {  325, 0x0020 },
  /* 0x8900 */
  {  326, 0x0000 }, {  326, 0x1000 }, {  327, 0x0000 }, {  327, 0x0000 },
  {  327, 0x0000 }, {  327, 0x0000 }, {  327, 0x0000 }, {  327, 0x0000 },
  {  327, 0x0000 }, {  327, 0x0000 }, {  327, 0x0000 }, {  327, 0x0000 },
  {  327, 0x0000 }, {  327, 0x0000 }, {  327, 0x0000 }, {  327, 0x0000 },
  /* 0x8a00 */
  {  327, 0x0000 }, {  327, 0x0004 }, {  328, 0x0000 }, {  328, 0x0080 },
  {  329, 0x0000 }, {  329, 0x0000 }, {  329, 0x0000 }, {  329, 0x0200 },
  {  330, 0x0000 }, {  330, 0x0000 }, {  330, 0x0080 }, {  331, 0x4000 },
  {  332, 0x0000 }, {  332, 0x8000 }, {  333, 0x0000 }, {  333, 0x0040 },
  /* 0x8b00 */
  {  334, 0x0000 }, {  334, 0x0000 }, {  334, 0x0000 }, {  334, 0x0000 },
  {  334, 0x0000 }, {  334, 0x0008 }, {  335, 0x0000 }, {  335, 0x8000 },
  {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 },
  {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 },
  /* 0x8c00 */
  {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 },
  {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 },
  {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 },
  {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0000 }, {  336, 0x0011 },
  /* 0x8d00 */
  {  338, 0x0000 }, {  338, 0x0004 }, {  339, 0x0000 }, {  339, 0x0000 },
  {  339, 0x0000 }, {  339, 0x0000 }, {  339, 0x0000 }, {  339, 0x0040 },
  {  340, 0x0000 }, {  340, 0x0000 }, {  340, 0x0000 }, {  340, 0x0000 },
  {  340, 0x0000 }, {  340, 0x0000 }, {  340, 0x0000 }, {  340, 0x0000 },
  /* 0x8e00 */
  {  340, 0x0000 }, {  340, 0x0000 }, {  340, 0x0000 }, {  340, 0x0000 },
  {  340, 0x0000 }, {  340, 0x0000 }, {  340, 0x0000 }, {  340, 0x0000 },
  {  340, 0x0000 }, {  340, 0x0000 }, {  340, 0x0000 }, {  340, 0x0000 },
  {  340, 0x8000 },
};
static const Summary16 cp932ext_uni2indx_page90[238] = {
  /* 0x9000 */
  {  341, 0x0000 }, {  341, 0x0000 }, {  341, 0x0000 }, {  341, 0x0000 },
  {  341, 0x0000 }, {  341, 0x0000 }, {  341, 0x0080 }, {  342, 0x0000 },
  {  342, 0x0000 }, {  342, 0x0000 }, {  342, 0x0000 }, {  342, 0x0000 },
  {  342, 0x0000 }, {  342, 0x4000 }, {  343, 0x0000 }, {  343, 0x0000 },
  /* 0x9100 */
  {  343, 0x0000 }, {  343, 0x0020 }, {  344, 0x0080 }, {  345, 0x0000 },
  {  345, 0x0000 }, {  345, 0x0000 }, {  345, 0x0000 }, {  345, 0x0000 },
  {  345, 0x0000 }, {  345, 0x0000 }, {  345, 0x0000 }, {  345, 0x0000 },
  {  345, 0x0000 }, {  345, 0x4480 }, {  348, 0x6030 }, {  352, 0x0000 },
  /* 0x9200 */
  {  352, 0x0440 }, {  354, 0x0001 }, {  355, 0x0000 }, {  355, 0x1600 },
  {  358, 0x4001 }, {  360, 0x0202 }, {  362, 0x0080 }, {  363, 0x0180 },
  {  365, 0x0100 }, {  366, 0x0000 }, {  366, 0x0080 }, {  367, 0x0000 },
  {  367, 0x0000 }, {  367, 0x02a9 }, {  372, 0x0081 }, {  374, 0x8a00 },
  /* 0x9300 */
  {  377, 0x0004 }, {  378, 0x6000 }, {  380, 0x0022 }, {  382, 0x0000 },
  {  382, 0x0100 }, {  383, 0x0080 }, {  384, 0x0000 }, {  384, 0x0001 },
  {  385, 0x0000 }, {  385, 0x0000 }, {  385, 0x0010 }, {  386, 0x0000 },
  {  386, 0x0040 }, {  387, 0x4000 }, {  388, 0x0000 }, {  388, 0x0100 },
  /* 0x9400 */
  {  389, 0x0000 }, {  389, 0x0000 }, {  389, 0x0000 }, {  389, 0x0002 },
  {  390, 0x0120 }, {  392, 0x0000 }, {  392, 0x0000 }, {  392, 0x0000 },
  {  392, 0x0000 }, {  392, 0x0000 }, {  392, 0x0000 }, {  392, 0x0000 },
  {  392, 0x0000 }, {  392, 0x0000 }, {  392, 0x0000 }, {  392, 0x0000 },
  /* 0x9500 */
  {  392, 0x0000 }, {  392, 0x0000 }, {  392, 0x0000 }, {  392, 0x0000 },
  {  392, 0x0000 }, {  392, 0x0000 }, {  392, 0x0000 }, {  392, 0x0000 },
  {  392, 0x0000 }, {  392, 0x0004 }, {  393, 0x0000 }, {  393, 0x0000 },
  {  393, 0x0000 }, {  393, 0x0000 }, {  393, 0x0000 }, {  393, 0x0000 },
  /* 0x9600 */
  {  393, 0x0000 }, {  393, 0x0000 }, {  393, 0x0000 }, {  393, 0x0000 },
  {  393, 0x0000 }, {  393, 0x0000 }, {  393, 0x0000 }, {  393, 0x0000 },
  {  393, 0x0000 }, {  393, 0x2000 }, {  394, 0x8000 }, {  395, 0x0000 },
  {  395, 0x0000 }, {  395, 0x0000 }, {  395, 0x0000 }, {  395, 0x0000 },
  /* 0x9700 */
  {  395, 0x0000 }, {  395, 0x0000 }, {  395, 0x0000 }, {  395, 0x0808 },
  {  397, 0xa008 }, {  400, 0x0022 }, {  402, 0x0000 }, {  402, 0x0000 },
  {  402, 0x0000 }, {  402, 0x0000 }, {  402, 0x0000 }, {  402, 0x0000 },
  {  402, 0x0000 }, {  402, 0x0000 }, {  402, 0x0000 }, {  402, 0x0000 },
  /* 0x9800 */
  {  402, 0x0000 }, {  402, 0x0000 }, {  402, 0x0000 }, {  402, 0x0000 },
  {  402, 0x0000 }, {  402, 0x0080 }, {  403, 0x0020 }, {  404, 0x0000 },
  {  404, 0x0000 }, {  404, 0x0000 }, {  404, 0x0000 }, {  404, 0x0000 },
  {  404, 0x0000 }, {  404, 0x0000 }, {  404, 0x0000 }, {  404, 0x0000 },
  /* 0x9900 */
  {  404, 0x0000 }, {  404, 0x0000 }, {  404, 0x0080 }, {  405, 0x0000 },
  {  405, 0x0000 }, {  405, 0x0000 }, {  405, 0x0000 }, {  405, 0x0000 },
  {  405, 0x0000 }, {  405, 0x4000 }, {  406, 0x0000 }, {  406, 0x0000 },
  {  406, 0x0000 }, {  406, 0x0000 }, {  406, 0x0000 }, {  406, 0x0000 },
  /* 0x9a00 */
  {  406, 0x0000 }, {  406, 0x0000 }, {  406, 0x0000 }, {  406, 0x0000 },
  {  406, 0x4000 }, {  407, 0x0000 }, {  407, 0x0000 }, {  407, 0x0000 },
  {  407, 0x0000 }, {  407, 0x0000 }, {  407, 0x0000 }, {  407, 0x0000 },
  {  407, 0x0000 }, {  407, 0x1200 }, {  409, 0x0000 }, {  409, 0x0000 },
  /* 0x9b00 */
  {  409, 0x0000 }, {  409, 0x0000 }, {  409, 0x0000 }, {  409, 0x0000 },
  {  409, 0x0000 }, {  409, 0x0000 }, {  409, 0x0000 }, {  409, 0x0024 },
  {  411, 0x8000 }, {  412, 0x0000 }, {  412, 0x0000 }, {  412, 0x0802 },
  {  414, 0x0000 }, {  414, 0x0000 }, {  414, 0x0000 }, {  414, 0x0000 },
  /* 0x9c00 */
  {  414, 0x0001 }, {  415, 0x0000 }, {  415, 0x0000 }, {  415, 0x0000 },
  {  415, 0x0000 }, {  415, 0x0000 }, {  415, 0x0000 }, {  415, 0x0000 },
  {  415, 0x0000 }, {  415, 0x0000 }, {  415, 0x0000 }, {  415, 0x0000 },
  {  415, 0x0000 }, {  415, 0x0000 }, {  415, 0x0000 }, {  415, 0x0000 },
  /* 0x9d00 */
  {  415, 0x0000 }, {  415, 0x0000 }, {  415, 0x0000 }, {  415, 0x0000 },
  {  415, 0x0000 }, {  415, 0x0000 }, {  415, 0x0800 }, {  416, 0x0001 },
  {  417, 0x0000 }, {  417, 0x0000 }, {  417, 0x0000 }, {  417, 0x0000 },
  {  417, 0x0000 }, {  417, 0x0000 }, {  417, 0x0000 }, {  417, 0x0000 },
  /* 0x9e00 */
  {  417, 0x0000 }, {  417, 0x0200 }, {  418, 0x0000 }, {  418, 0x0000 },
  {  418, 0x0000 }, {  418, 0x0000 }, {  418, 0x0000 }, {  418, 0x0000 },
  {  418, 0x0000 }, {  418, 0x0000 }, {  418, 0x0000 }, {  418, 0x0000 },
  {  418, 0x0000 }, {  418, 0x0002 },
};
static const Summary16 cp932ext_uni2indx_pagef9[19] = {
  /* 0xf900 */
  {  419, 0x0000 }, {  419, 0x0000 }, {  419, 0x0200 }, {  420, 0x0000 },
  {  420, 0x0000 }, {  420, 0x0000 }, {  420, 0x0000 }, {  420, 0x0000 },
  {  420, 0x0000 }, {  420, 0x0000 }, {  420, 0x0000 }, {  420, 0x0000 },
  {  420, 0x0000 }, {  420, 0x1000 }, {  421, 0x0000 }, {  421, 0x0000 },
  /* 0xfa00 */
  {  421, 0xc000 }, {  423, 0xffff }, {  439, 0x3fff },
};
static const Summary16 cp932ext_uni2indx_pageff[15] = {
  /* 0xff00 */
  {  453, 0x0084 }, {  455, 0x0000 }, {  455, 0x0000 }, {  455, 0x0000 },
  {  455, 0x0000 }, {  455, 0x0000 }, {  455, 0x0000 }, {  455, 0x0000 },
  {  455, 0x0000 }, {  455, 0x0000 }, {  455, 0x0000 }, {  455, 0x0000 },
  {  455, 0x0000 }, {  455, 0x0000 }, {  455, 0x0014 },
};

static int
cp932ext_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, int n)
{
  if (n >= 2) {
    const Summary16 *summary = NULL;
    if (wc >= 0x2100 && wc < 0x22c0)
      summary = &cp932ext_uni2indx_page21[(wc>>4)-0x210];
    else if (wc >= 0x2400 && wc < 0x2480)
      summary = &cp932ext_uni2indx_page24[(wc>>4)-0x240];
    else if (wc >= 0x3000 && wc < 0x3020)
      summary = &cp932ext_uni2indx_page30[(wc>>4)-0x300];
    else if (wc >= 0x3200 && wc < 0x33d0)
      summary = &cp932ext_uni2indx_page32[(wc>>4)-0x320];
    else if (wc >= 0x4e00 && wc < 0x5590)
      summary = &cp932ext_uni2indx_page4e[(wc>>4)-0x4e0];
    else if (wc >= 0x5700 && wc < 0x59c0)
      summary = &cp932ext_uni2indx_page57[(wc>>4)-0x570];
    else if (wc >= 0x5b00 && wc < 0x5de0)
      summary = &cp932ext_uni2indx_page5b[(wc>>4)-0x5b0];
    else if (wc >= 0x5f00 && wc < 0x7ba0)
      summary = &cp932ext_uni2indx_page5f[(wc>>4)-0x5f0];
    else if (wc >= 0x7d00 && wc < 0x7fb0)
      summary = &cp932ext_uni2indx_page7d[(wc>>4)-0x7d0];
    else if (wc >= 0x8300 && wc < 0x85c0)
      summary = &cp932ext_uni2indx_page83[(wc>>4)-0x830];
    else if (wc >= 0x8800 && wc < 0x8ed0)
      summary = &cp932ext_uni2indx_page88[(wc>>4)-0x880];
    else if (wc >= 0x9000 && wc < 0x9ee0)
      summary = &cp932ext_uni2indx_page90[(wc>>4)-0x900];
    else if (wc >= 0xf900 && wc < 0xfa30)
      summary = &cp932ext_uni2indx_pagef9[(wc>>4)-0xf90];
    else if (wc >= 0xff00 && wc < 0xfff0)
      summary = &cp932ext_uni2indx_pageff[(wc>>4)-0xff0];
    if (summary) {
      unsigned short used = summary->used;
      unsigned int i = wc & 0x0f;
      if (used & ((unsigned short) 1 << i)) {
        unsigned short c;
        /* Keep in `used' only the bits 0..i-1. */
        used &= ((unsigned short) 1 << i) - 1;
        /* Add `summary->indx' and the number of bits set in `used'. */
        used = (used & 0x5555) + ((used & 0xaaaa) >> 1);
        used = (used & 0x3333) + ((used & 0xcccc) >> 2);
        used = (used & 0x0f0f) + ((used & 0xf0f0) >> 4);
        used = (used & 0x00ff) + (used >> 8);
        c = cp932ext_2charset[summary->indx + used];
        r[0] = (c >> 8); r[1] = (c & 0xff);
        return 2;
      }
    }
    return RET_ILUNI;
  }
  return RET_TOOSMALL;
}