guess1

/*
 *  call-seq:
 *     NKF.guess1(str)  -> integer
 *
 *  Returns guessed encoding of _str_ as integer.
 *
 *  Algorithm described in:
 *  Ken Lunde. `Understanding Japanese Information Processing'
 *  Sebastopol, CA: O'Reilly & Associates.
 *
 *      case NKF.guess1(input)
 *      when NKF::JIS
 *        "ISO-2022-JP"
 *      when NKF::SJIS
 *        "Shift_JIS"
 *      when NKF::EUC
 *        "EUC-JP"
 *      when NKF::UNKNOWN
 *        "UNKNOWN(ASCII)"
 *      when NKF::BINARY
 *        "BINARY"
 *      end
 */

static VALUE
rb_nkf_guess1(obj, src)
  VALUE obj, src;
{
  unsigned char *p;
  unsigned char *pend;
  int sequence_counter = 0;

  StringValue(src);
  p = (unsigned char *)RSTRING(src)->ptr;
  pend = p + RSTRING(src)->len;
  if (p == pend) return INT2FIX(_UNKNOWN);

#define INCR do {\
      p++;\
      if (p==pend) return INT2FIX(_UNKNOWN);\
      sequence_counter++;\
      if (sequence_counter % 2 == 1 && *p != 0xa4)\
        sequence_counter = 0;\
      if (6 <= sequence_counter) {\
          sequence_counter = 0;\
          return INT2FIX(_EUC);\
      }\
  } while (0)

  if (*p == 0xa4)
    sequence_counter = 1;

  while (p<pend) {
    if (*p == '\033') {
      return INT2FIX(_JIS);
    }
    if (*p < '\006' || *p == 0x7f || *p == 0xff) {
      return INT2FIX(_BINARY);
    }
    if (0x81 <= *p && *p <= 0x8d) {
      return INT2FIX(_SJIS);
    }
    if (0x8f <= *p && *p <= 0x9f) {
      return INT2FIX(_SJIS);
    }
    if (*p == 0x8e) {   /* SS2 */
      INCR;
      if ((0x40 <= *p && *p <= 0x7e) ||
          (0x80 <= *p && *p <= 0xa0) ||
          (0xe0 <= *p && *p <= 0xfc))
        return INT2FIX(_SJIS);
    }
    else if (0xa1 <= *p && *p <= 0xdf) {
      INCR;
      if (0xf0 <= *p && *p <= 0xfe)
        return INT2FIX(_EUC);
      if (0xe0 <= *p && *p <= 0xef) {
        while (p < pend && *p >= 0x40) {
          if (*p >= 0x81) {
            if (*p <= 0x8d || (0x8f <= *p && *p <= 0x9f)) {
              return INT2FIX(_SJIS);
            }
            else if (0xfd <= *p && *p <= 0xfe) {
              return INT2FIX(_EUC);
            }
          }
          INCR;
        }
      }
      else if (*p <= 0x9f) {
        return INT2FIX(_SJIS);
      }
    }
    else if (0xf0 <= *p && *p <= 0xfe) {
      return INT2FIX(_EUC);
    }
    else if (0xe0 <= *p && *p <= 0xef) {
      INCR;
      if ((0x40 <= *p && *p <= 0x7e) ||
          (0x80 <= *p && *p <= 0xa0)) {
        return INT2FIX(_SJIS);
      }
      if (0xfd <= *p && *p <= 0xfe) {
        return INT2FIX(_EUC);
      }
    }
    INCR;
  }
  return INT2FIX(_UNKNOWN);
}