/* shindou6.c - Don Yang (uguu.org)

   02/25/07
                                                                           */

#include<stdio.h>
#include<stdlib.h>
#ifdef _WIN32
   #include<io.h>
   #include<fcntl.h>
#endif

#define KANA_RANGE (0x3093-0x3041)

static FILE *infile;
static unsigned char *buffer, *p, *const_str =
   "!!\0  !\xff\1"
   "UTF-8\0"
   "!\x81IOCCC"
   "!\1\xff\xfe\xff""can not open!\0! \0"
   "\xef\xbb\xbf"
   "EUC-JP\0"
   " 'PoPoPo\4""0?PoPo\3\20\17Po\2\1"
   "\"\xef\xbc\x81"
   "Shift_JIS\0"
   "!\xa1\xaa"
   "QO\220,\2""0?\220,\2\1"
   "UCS-2LE\0"
   "qxq.\2}}q.\2`$q.\2),q.\2^^q.\2__q.q.\3\1"
   "UCS-2BE\0"
   ;
static int
   buffer_size, read_size,
   i, c, encoding,
   last_char_start, i0, next_offset,
   t,
   byte_offset,
   bom[] = {23, 24, 43};

#define DONT_CARE 0xdeadbeef
static int data[5][19] =
{
   {
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      96,
      0,
      21,
      40,
      103,
      DONT_CARE,
      0
   },
   {
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      96,
      38,
      5,
      1,
      144,
      DONT_CARE,
      0
   },
   {
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      DONT_CARE,
      53,
      4,
      75,
      3,
      8,
      DONT_CARE,
      0
   },
   {
      DONT_CARE,
      8,
      DONT_CARE,
      DONT_CARE,
      3,
      7,
      161,
      1,
      DONT_CARE,
      2,
      3,
      33089,
      92,
      4,
      14,
      3,
      79,
      350,
      0
   },
   {
      DONT_CARE,
      8,
      DONT_CARE,
      DONT_CARE,
      3,
      7,
      36513,
      1,
      DONT_CARE,
      2,
      3,
      41378,
      111,
      4,
      89,
      3,
      46,
      767,
      0
   }
};

#define F(func) static void func(void)
#define D(index) data[encoding][index]

F(AllocBuffer)
{
   buffer = (unsigned char*)malloc(buffer_size *= 2);
}
F(FreeBuffer)
{
   free(p);
}

F(LoadInput)
{
   buffer_size = 0x800;
   AllocBuffer();
   for(read_size = 0; !feof(infile);
       read_size += fread(buffer + read_size,
                          1, buffer_size - read_size, infile))
   {
      if( read_size == buffer_size )
      {
         p = buffer;
         AllocBuffer();
         for(byte_offset = 0; byte_offset < buffer_size; byte_offset++)
            buffer[byte_offset] = p[byte_offset];
         free(p);
      }
   }
}

F(NextChar)
{
   c = byte_offset = 0;
   for(p = const_str + D(12); *p > 4 && c + 1; p++)
   {
      if( buffer[i + byte_offset] < (*p ^ 208) ||
          buffer[i + byte_offset] > (p[1] ^ 208) )
      {
         if( byte_offset )
         {
            c = -1;
         }
         else
         {
            for(; *p > 4; p++);
         }
      }
      else
      {
         byte_offset++;
         p++;
      }
   }
   for(byte_offset = 0; c + 1 && byte_offset < *p; byte_offset++)
      c = (c << 8) | buffer[i++];

   c =
      (encoding == 2 && c > 7<<21)
         ? ((c >> 4) & 0xf000) | ((c >> 2) & 4032) | (c & 63)
         : encoding
            ? c
            : (c >> 8) | ((c & 255) << 8);
}

F(NextCharUnconditional)
{
   NextChar();
   if( c < 0 )
      i++;
}

F(DetectEncoding)
{
   for(i = 0; i < 3; i++)
   {
      if( *buffer == const_str[bom[i]] &&
          buffer[1] == const_str[bom[i]+1] &&
          (i < 2 || buffer[2] == const_str[bom[i]+2]) )
      {
         encoding = i;
         return;
      }
   }

   for(i = 0; i < read_size - 1; i += 2)
   {
      if( !buffer[i] && buffer[i + 1] == 10 )
      {
         encoding = 1;
         return;
      }
      if( buffer[i] == 10 && !buffer[i + 1] )
      {
         encoding = 0;
         return;
      }
   }

   for(encoding = 2; encoding < 5; encoding++)
   {
      for(i = D(18) = 0;
          i < read_size - 3 && D(18) + 1;)
      {
         NextChar();
         if( c + 1 )
         {
            if( c >= D(17) &&
                c <= D(17) + KANA_RANGE )
               D(18)++;
         }
         else
         {
            D(18) = -1;
         }
      }
   }

   for(i = encoding = 2; i < 5; i++)
   {
      if( data[i][18] > D(18) )
         encoding = i;
   }
}

int IsPunct(int range)
{
   for(t = 0; t < range && c - D(t); t++);
   return t < range;
}

void Output(void *bin, int size)
{
   if( size > 0 )
      fwrite(bin, size, 1, stdout);
}

F(WriteExclamationMark)
{
   p = const_str +
      D(c > 255 ? 14 : 13);
   Output(p + 1, *p - 31);
}


int main(int argc, char **argv)
{
   for(encoding = 5; encoding-- > 0;)
   {
      D(0) = 33;
      D(2) = 46;
      D(3) = 63;
      D(8) = 44;
   }
   for(; ++encoding < 3;)
   {
      D(10) = 3 +    /* 0xff64 */
      (D(6) = 66 +   /* 0xff61 */
      (D(5) = 17 +   /* 0xff1f */
      (D(4) = 2 +    /* 0xff0e */
      (D(9) = 11 +   /* 0xff0c */
      (D(1) = 0xff01)))));
      D(17) = 145 +  /* 0x3093 */
      (D(7) = 1 +    /* 0x3002 */
      (D(11) = 0x3001));
   }
   for(; encoding < 5; encoding++)
   {
      D(1) += (i = D(11));
      D(4) += i;
      D(5) += i;
      D(7) += i;
      D(9) += i;
      D(17) += i;
      D(6) += D(10);
   }

   if( *++argv )
   {
      for(; *argv; ++argv)
      {
         printf("%s: ", *argv);
         i0 = 26;
         if( infile = fopen(*argv, "rb") )
         {
            LoadInput();
            fclose(infile);
            DetectEncoding();
            i0 = D(16);
            p = buffer;
            FreeBuffer();
         }
         puts(const_str + i0);
      }
   }
   else
   {
      #ifdef _WIN32
         setmode(fileno(stdin), O_BINARY);
         setmode(fileno(stdout), O_BINARY);
      #endif

      infile = stdin;
      LoadInput();
      DetectEncoding();
      for(i = i0 = 0; i < read_size;)
      {
         last_char_start = i;
         NextCharUnconditional();

         if( IsPunct(12) || c == 10 )
         {
            if( IsPunct(8) )
            {
               next_offset = i;
               for(i = i0; i < last_char_start;)
               {
                  NextCharUnconditional();
                  Output(buffer + i0, i - i0);
                  p = const_str + D(15);
                  Output(p + 1, *p - 31);
                  i0 = i;
               }

               NextCharUnconditional();
               if( IsPunct(2) )
                  WriteExclamationMark();
               WriteExclamationMark();
               i0 = i;
            }
            else
            {
               Output(buffer + i0, last_char_start - i0);

               if( c - 10 )
                  WriteExclamationMark();
               else
                  Output(buffer + last_char_start, i - last_char_start);
               i0 = i;
            }
         }
      }
      p = buffer;
      FreeBuffer();
   }
   return 0;
}