Simple Textdata Converter from CESU-8 to UTF-8

CESU8toUTF8.c

/*####################################################################
#
# CESU8toUTF8.c - Simple Textdata Converter from CESU-8 to UTF-8
#
# Usage : cat your_textfile.txt | ./CESU8toUTF8 > converted_text.txt
#
# How to compile me : cc -O3 -o CESU8toUTF8 CESU8toUTF8.c
#
#
# Written by Colonel Richie (@colrichie) on 2023-11-21
#
####################################################################*/


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

#define MAXBUF 1048576


int find_cesu8(char* pszLine) {

  int               i;
  unsigned char*    puc;

  puc = (unsigned char*)pszLine;
  for (i=0; i<strlen(pszLine); ) {
    if (puc[i  ]!=0xED                ){i++ ;continue;}
    if (puc[i+1]<0xA0 || puc[i+1]>0xAF){i+=2;continue;}
    if (puc[i+2]<0x80 || puc[i+2]>0xBF){i+=3;continue;}
    if (puc[i+3]!=0xED                ){i+=4;continue;}
    if (puc[i+4]<0xB0 || puc[i+4]>0xBF){i+=5;continue;}
    if (puc[i+5]<0x80 || puc[i+5]>0xBF){i+=6;continue;}
    return i;
  }

  return -1;
}


int main(void) {

  FILE  *fp;
  char  szBuf[MAXBUF], *pszBuf;
  int   i, iOfs;
  char  szUTF8[4];

  setlocale(LC_CTYPE, "");

  fp = stdin;
  while(fgets(szBuf, MAXBUF, fp) != NULL) {
    pszBuf=szBuf;
    while ((iOfs=find_cesu8(pszBuf))>=0) {
      for (i=0; i<iOfs; i++) {putchar(*pszBuf);pszBuf++;}
      pszBuf[1]++;
      szUTF8[0] = 0xF0 | ((pszBuf[1] & 0x1C)>>2)                          ;
      szUTF8[1] = 0x80 | ((pszBuf[1] & 0x03)<<4) | ((pszBuf[2] & 0x3C)>>2);
      szUTF8[2] = 0x80 | ((pszBuf[2] & 0x03)<<4) | ( pszBuf[4] & 0x0F    );
      szUTF8[3] = pszBuf[5];                                              ;
      putchar(szUTF8[0]);putchar(szUTF8[1]);putchar(szUTF8[2]);putchar(szUTF8[3]);
      pszBuf+=6;
    }
    while (*pszBuf!=0) {putchar(*pszBuf);pszBuf++;}
  }

  return 0;
}