// This code reads standard in, coded in utf-8, and computes integer code points.
// Each call to g yields one 32 bit 'character'.

#include <stdio.h>
#include <stdlib.h>
static int cil = 1;
static int lc = 1;
static int bc = 0;
static int ad = 0, oldc;
typedef unsigned char uchar;
void loc(){printf("at character %d of line %d,"
    "(or 0x%x in file):\n", cil, lc, bc);}

static char gc(){char c = getchar(); ++bc;
  if(c == EOF && feof(stdin)) exit(printf(
     "end of file at char: %d in line %d\n", cil, lc));
  if(c == 10 || c == 13) {cil = 1; ++lc;}
  return c;}

static char gx(){char c = gc(); if ((c&0xc0) != 0x80)
   {loc(); printf("Bad utf8 extension byte: %02X\n", c);}
   return c;}

static int gp(){uchar c = gc(); if(c < 128) return c;
     {int v=0, sc=0, C=c;  
     while(C&0x40){C <<=1; v = (v<<6) | gx() & 0x3f; ++sc;}
     return (v | (0x3f>>sc & (int)c) << 6*sc);}}

void back(){ad=1;}

int g(){if(ad) {ad=0; ++cil;} else oldc = gp(); return oldc;}

// int main(){while(1) printf(" %x", g());}