/* Last edited on 2024-12-25 10:15:26 by stolfi */
/*
  Reads a GB-encoded Chinese text, replaces all ideograms by a decimal
  number enclosed in guillemots «».

  Assumes the ideograms are encoded in the GB 2312-80 EUC-compatible format,
  where each ideogram consists of a `row' byte in the range 0xA1 to 0xF7
  and a `column' byte in the range 0xA1 to 0xFE.  Assumes also that 
  Any byte that is not in GB encoding is in the range 0x00 to 0xA0.
*/

#include <stdio.h>
#include <stdlib.h>

#include <gb-to-decimal_doc.h>

#define NCOLS (0xFE - 0xA1 + 1)
#define NROWS (0xF7 - 0xA1 + 1)

void doputchar(int c);
void error(char *msg);

int main(int argc, char **argv)
{
  int r, s, rs;
  while(1)
    {
      r = getchar();
      if (r == EOF)
        { fclose(stdout); exit(0); }
      else if (r < 0xA1) 
        { doputchar(r); } 
      else if (r > 0xF7)
        { doputchar(r); error("row byte over 0xF7"); }
      else
        { s = getchar();
          if (s == EOF)
            { doputchar(r); error("unexpected EOF in GB code"); }
          else if (s < 0xA1)
            { doputchar(r); doputchar(s); error("incomplete GB code"); }
          else
            { rs = NCOLS*(r - 0xA1) + (s - 0xA1) + 1;
              printf("«%d»", rs);
            }
        }
    }
}

void doputchar(int c)
{
  int res = putchar(c);
  if (res != c) { error("write error"); }
}

void error(char *msg)
{
  fprintf(stderr, "%s\n", msg);
  fclose(stdout); exit(1); 
}