iso2022_jp2.h

libiconv是一个很不错的字符集转换库。程序接口也很简单

          conv->ostate = state;
          return 0;
        }
        break;
      case 'k':
        if (state3 == STATE_TAG_LANGUAGE) {
          state3 = STATE_TAG_LANGUAGE_k;
          COMBINE_STATE;
          conv->ostate = state;
          return 0;
        }
        break;
      case 'o':
        if (state3 == STATE_TAG_LANGUAGE_k) {
          state3 = STATE_TAG_LANGUAGE_ko;
          COMBINE_STATE;
          conv->ostate = state;
          return 0;
        }
        break;
      case 'z':
        if (state3 == STATE_TAG_LANGUAGE) {
          state3 = STATE_TAG_LANGUAGE_z;
          COMBINE_STATE;
          conv->ostate = state;
          return 0;
        }
        break;
      case 'h':
        if (state3 == STATE_TAG_LANGUAGE_z) {
          state3 = STATE_TAG_LANGUAGE_zh;
          COMBINE_STATE;
          conv->ostate = state;
          return 0;
        }
        break;
      case 0x7f:
        state3 = STATE_TAG_NONE;
        COMBINE_STATE;
        conv->ostate = state;
        return 0;
      default:
        break;
    }
    /* Other tag characters reset the tag parsing state or are ignored. */
    if (state3 >= STATE_TAG_LANGUAGE)
      state3 = STATE_TAG_NONE;
    COMBINE_STATE;
    conv->ostate = state;
    return 0;
  }
  if (state3 >= STATE_TAG_LANGUAGE)
    state3 = STATE_TAG_NONE;

  /* Try ASCII. */
  ret = ascii_wctomb(conv,buf,wc,1);
  if (ret != RET_ILUNI) {
    if (ret != 1) abort();
    if (buf[0] < 0x80) {
      int count = (state1 == STATE_ASCII ? 1 : 4);
      if (n < count)
        return RET_TOOSMALL;
      if (state1 != STATE_ASCII) {
        r[0] = ESC;
        r[1] = '(';
        r[2] = 'B';
        r += 3;
        state1 = STATE_ASCII;
      }
      r[0] = buf[0];
      if (wc == 0x000a || wc == 0x000d)
        state2 = STATE_G2_NONE;
      COMBINE_STATE;
      conv->ostate = state;
      return count;
    }
  }

  conversion_list = conversion_lists[state3];

  do {
    switch (conversion_list & ((1 << 3) - 1)) {

      case european:

        /* Try ISO-8859-1. */
        ret = iso8859_1_wctomb(conv,buf,wc,1);
        if (ret != RET_ILUNI) {
          if (ret != 1) abort();
          if (buf[0] >= 0x80) {
            int count = (state2 == STATE_G2_ISO8859_1 ? 3 : 6);
            if (n < count)
              return RET_TOOSMALL;
            if (state2 != STATE_G2_ISO8859_1) {
              r[0] = ESC;
              r[1] = '.';
              r[2] = 'A';
              r += 3;
              state2 = STATE_G2_ISO8859_1;
            }
            r[0] = ESC;
            r[1] = 'N';
            r[2] = buf[0]-0x80;
            COMBINE_STATE;
            conv->ostate = state;
            return count;
          }
        }

        /* Try ISO-8859-7. */
        ret = iso8859_7_wctomb(conv,buf,wc,1);
        if (ret != RET_ILUNI) {
          if (ret != 1) abort();
          if (buf[0] >= 0x80) {
            int count = (state2 == STATE_G2_ISO8859_7 ? 3 : 6);
            if (n < count)
              return RET_TOOSMALL;
            if (state2 != STATE_G2_ISO8859_7) {
              r[0] = ESC;
              r[1] = '.';
              r[2] = 'F';
              r += 3;
              state2 = STATE_G2_ISO8859_7;
            }
            r[0] = ESC;
            r[1] = 'N';
            r[2] = buf[0]-0x80;
            COMBINE_STATE;
            conv->ostate = state;
            return count;
          }
        }

        break;

      case japanese:

        /* Try JIS X 0201-1976 Roman. */
        ret = jisx0201_wctomb(conv,buf,wc,1);
        if (ret != RET_ILUNI) {
          if (ret != 1) abort();
          if (buf[0] < 0x80) {
            int count = (state1 == STATE_JISX0201ROMAN ? 1 : 4);
            if (n < count)
              return RET_TOOSMALL;
            if (state1 != STATE_JISX0201ROMAN) {
              r[0] = ESC;
              r[1] = '(';
              r[2] = 'J';
              r += 3;
              state1 = STATE_JISX0201ROMAN;
            }
            r[0] = buf[0];
            if (wc == 0x000a || wc == 0x000d)
              state2 = STATE_G2_NONE;
            COMBINE_STATE;
            conv->ostate = state;
            return count;
          }
        }

        /* Try JIS X 0208-1990 in place of JIS X 0208-1978 and
           JIS X 0208-1983. */
        ret = jisx0208_wctomb(conv,buf,wc,2);
        if (ret != RET_ILUNI) {
          if (ret != 2) abort();
          if (buf[0] < 0x80 && buf[1] < 0x80) {
            int count = (state1 == STATE_JISX0208 ? 2 : 5);
            if (n < count)
              return RET_TOOSMALL;
            if (state1 != STATE_JISX0208) {
              r[0] = ESC;
              r[1] = '$';
              r[2] = 'B';
              r += 3;
              state1 = STATE_JISX0208;
            }
            r[0] = buf[0];
            r[1] = buf[1];
            COMBINE_STATE;
            conv->ostate = state;
            return count;
          }
        }

        /* Try JIS X 0212-1990. */
        ret = jisx0212_wctomb(conv,buf,wc,2);
        if (ret != RET_ILUNI) {
          if (ret != 2) abort();
          if (buf[0] < 0x80 && buf[1] < 0x80) {
            int count = (state1 == STATE_JISX0212 ? 2 : 6);
            if (n < count)
              return RET_TOOSMALL;
            if (state1 != STATE_JISX0212) {
              r[0] = ESC;
              r[1] = '$';
              r[2] = '(';
              r[3] = 'D';
              r += 4;
              state1 = STATE_JISX0212;
            }
            r[0] = buf[0];
            r[1] = buf[1];
            COMBINE_STATE;
            conv->ostate = state;
            return count;
          }
        }

        break;

      case chinese:

        /* Try GB 2312-1980. */
        ret = gb2312_wctomb(conv,buf,wc,2);
        if (ret != RET_ILUNI) {
          if (ret != 2) abort();
          if (buf[0] < 0x80 && buf[1] < 0x80) {
            int count = (state1 == STATE_GB2312 ? 2 : 5);
            if (n < count)
              return RET_TOOSMALL;
            if (state1 != STATE_GB2312) {
              r[0] = ESC;
              r[1] = '$';
              r[2] = 'A';
              r += 3;
              state1 = STATE_GB2312;
            }
            r[0] = buf[0];
            r[1] = buf[1];
            COMBINE_STATE;
            conv->ostate = state;
            return count;
          }
        }

        break;

      case korean:

        /* Try KS C 5601-1992. */
        ret = ksc5601_wctomb(conv,buf,wc,2);
        if (ret != RET_ILUNI) {
          if (ret != 2) abort();
          if (buf[0] < 0x80 && buf[1] < 0x80) {
            int count = (state1 == STATE_KSC5601 ? 2 : 6);
            if (n < count)
              return RET_TOOSMALL;
            if (state1 != STATE_KSC5601) {
              r[0] = ESC;
              r[1] = '$';
              r[2] = '(';
              r[3] = 'C';
              r += 4;
              state1 = STATE_KSC5601;
            }
            r[0] = buf[0];
            r[1] = buf[1];
            COMBINE_STATE;
            conv->ostate = state;
            return count;
          }
        }

        break;

      case other:

        /* Try JIS X 0201-1976 Kana. This is not officially part of
           ISO-2022-JP-2, according to RFC 1554. Therefore we try this
           only after all other attempts. */
        ret = jisx0201_wctomb(conv,buf,wc,1);
        if (ret != RET_ILUNI) {
          if (ret != 1) abort();
          if (buf[0] >= 0x80) {
            int count = (state1 == STATE_JISX0201KATAKANA ? 1 : 4);
            if (n < count)
              return RET_TOOSMALL;
            if (state1 != STATE_JISX0201KATAKANA) {
              r[0] = ESC;
              r[1] = '(';
              r[2] = 'I';
              r += 3;
              state1 = STATE_JISX0201KATAKANA;
            }
            r[0] = buf[0]-0x80;
            COMBINE_STATE;
            conv->ostate = state;
            return count;
          }
        }

        break;

      default:
        abort();
    }

    conversion_list = conversion_list >> 3;
  } while (conversion_list != 0);

  return RET_ILUNI;
}

static int
iso2022_jp2_reset (conv_t conv, unsigned char *r, int n)
{
  state_t state = conv->ostate;
  SPLIT_STATE;
  (void)state2;
  (void)state3;
  if (state1 != STATE_ASCII) {
    if (n < 3)
      return RET_TOOSMALL;
    r[0] = ESC;
    r[1] = '(';
    r[2] = 'B';
    /* conv->ostate = 0; will be done by the caller */
    return 3;
  } else
    return 0;
}

#undef COMBINE_STATE
#undef SPLIT_STATE
#undef STATE_TAG_LANGUAGE_zh
#undef STATE_TAG_LANGUAGE_z
#undef STATE_TAG_LANGUAGE_ko
#undef STATE_TAG_LANGUAGE_k
#undef STATE_TAG_LANGUAGE_ja
#undef STATE_TAG_LANGUAGE_j
#undef STATE_TAG_LANGUAGE
#undef STATE_TAG_NONE
#undef STATE_G2_ISO8859_7
#undef STATE_G2_ISO8859_1
#undef STATE_G2_NONE
#undef STATE_KSC5601
#undef STATE_GB2312
#undef STATE_JISX0212
#undef STATE_JISX0208
#undef STATE_JISX0201KATAKANA
#undef STATE_JISX0201ROMAN
#undef STATE_ASCII