/* autor: Milan Ruzic 394/2000 */
#include <stdio.h>

#define BufferSize 1000
#define MaxStrLen  20           /* majoranta duzina stringova koji ce biti
                                 pretrazivani u tekstu */

typedef struct {char srcStr[5], destStr[3];} Treplace_pair;
 /* ime samo kaze - ako se na odredjenom mestu u tekstu nalazi source string, 
    bice zamenjen destination stringom */

int trenPos=BufferSize;
 /*trenutna pozicija u bufferu; globalna promenjiva omogucava korisnickim
   procedurama da povecaju poziciju za vise mesta ako je odgovarajuci string
   prepoznat ili da se eventualno vrate unazad */


char *Buffer_Read(void);
int myStrCmp(char *s1,char *s2);      /* ovde s1 ne mora da se zavrsava NULLom */
int myStriCmp(char *s1, char *s2);    /* gleda se da li je neki prefiks od s1==s2 */
int toUpCase(int ch);

/**************************************************************************/
char *Buffer_Read(void)
/* vraca pointer na tren. poziciju u character bufferu i povecava trenPos za 1*/
{static char chBuffer[BufferSize];

 if (trenPos<BufferSize-MaxStrLen) return &chBuffer[trenPos++];
 else { int i,c=0;
	for (i=0;i+trenPos<BufferSize && c!=EOF;chBuffer[i++]=c=chBuffer[i+trenPos]);
	while (c!=EOF && i<BufferSize) chBuffer[i++]=c=getc(stdin);
	trenPos=1;
	return chBuffer;}
}

/**************************************************************************/
int myStrCmp(char *s1,char *s2)   /* ovde s1 ne mora da se zavrsava NULLom */
{for (;*s1==*s2 && *s2;s1++,s2++);

 if (!*s2) return 0;
  else if (*s1>*s2) return 1;
	   else    return -1;
}

int myStriCmp(char *s1,char *s2)
{for (;toUpCase(*s1)==toUpCase(*s2) && *s2;s1++,s2++);

 if (!*s2) return 0;
  else if (toUpCase(*s1)>toUpCase(*s2)) return 1;
	   else    return -1;
}

/**************************************************************************/
#define LcodesNum 16

void iso8859_2(void)
{Treplace_pair codes[LcodesNum]={"&Cx;","\xC8","&cx;","\xE8","&Cy;","\xC6","&cy;","\xE6",
  "&Dx;","\xD0","&dx;","\xF0","&Dy;","Dz","&dy;","dz","&Zx;","\xAE","&zx;","\xBE",
  "&Sx;","\xA9","&sx;","\xB9","&Lx;","Lj","&lx;","lj","&Nx;","Nj","&nx;","nj"};
 char *trenStr, inTag=0;

 for (trenStr=Buffer_Read();*trenStr!=EOF;trenStr=Buffer_Read())
 {char c=*trenStr;
   if (inTag) { putc(c,stdout); if (c=='>'|| c==']') inTag=0;}
   else if (c=='<'||c=='[') { putc(c,stdout); inTag=1;}
	else {int i;
	      for (i=0;i<LcodesNum;i++) 
                  if (!myStrCmp(trenStr,codes[i].srcStr)) break;
	      if (i==LcodesNum) putc(c,stdout);
	      else { printf("%s",codes[i].destStr); trenPos+=4-1;} 
              /* svi izvorni stringovi su duzine 4 pa se radi trenPos+=4-1 */         
	      }
   }
}

/**************************************************************************/
int strlen(char s[])
{int i;
 for (i=0;s[i]!=0;i++);
 return i;
}

int strCmp(char *s1,char *s2)       /* normalni strcmp koji se koristi u qsort() */
{ for (;*s1==*s2 && *s2;s1++,s2++);
 if (*s1==*s2) return 0;
 else if (*s1>*s2) return 1;
      else return -1;
}

/* posto u iso8859-5() treba ispitati malo vise stringova primenjuje binarna
   pretraga koja zahteva sortiran niz */

void qsort(Treplace_pair a[],int l,int r)
{Treplace_pair temp;
 char mid[5];
 int i,j=(l+r)>>1;
 for (i=0;i<=5;i++) mid[i]=a[j].srcStr[i];
 i=l;j=r;
 do { for(;strCmp(a[i].srcStr,mid)<0;i++);
      for(;strCmp(a[j].srcStr,mid)>0;j--);
      if (i<=j)
       {temp=a[i];
        a[i]=a[j];
        a[j]=temp;
        i++; j--; }
   } while (i<=j);
 if (i<r) qsort(a,i,r);
 if (j>l) qsort(a,l,j);
}

#define CcodesNum 60

void iso8859_5(void)
{Treplace_pair codes[CcodesNum]={"&Cx;","\xC7","&cx;","\xE7","&Cy;","\xAB","&cy;","\xFB",
  "&Dx;","\xA2","&dx;","\xF2","&Dy;","\xAF","&dy;","\xFF","&Zx;","\xB6","&zx;","\xD6",
  "&Sx;","\xC8","&sx;","\xE8","&Lx;","\xA9","&lx;","\xF9","&Nx;","\xAA","&nx;","\xFA",
   "A","\xB0","B","\xB1","V","\xB2","G","\xB3","D","\xB4","E","\xB5","Z","\xB7","I","\xB8",
   "J","\xA8","K","\xBA","L","\xBB","M","\xBC","N","\xBD","O","\xBE","P","\xBF","R","\xC0",
   "S","\xC1","T","\xC2","U","\xC3","F","\xC4","H","\xC5","C","\xC6",
   "a","\xD0","b","\xD1","v","\xD2","g","\xD3","d","\xD4","e","\xD5","z","\xD7","i","\xD8",
   "j","\xF8","k","\xDA","l","\xDB","m","\xDC","n","\xDD","o","\xDE","p","\xDF","r","\xE0",
   "s","\xE1","t","\xE2","u","\xE3","f","\xE4","h","\xE5","c","\xE6"};
 char *trenStr, inTag=0;
 qsort(codes,0,CcodesNum-1);

 for (trenStr=Buffer_Read();*trenStr!=EOF;trenStr=Buffer_Read())
 {char c=*trenStr;
   if (inTag) { putc(c,stdout); if (c=='>'|| c==']') inTag=0;}
   else if (c=='<'||c=='[') { putc(c,stdout); inTag=1;}
        else { /* sada ide binarna pretraga */
              int i=0,j=CcodesNum-1;
              do {int k=(i+j)>>1;
		  if (myStrCmp(trenStr,codes[k].srcStr)>0) i=k+1;
                  else j=k;
                  } while (i<j);

              if (myStrCmp(trenStr,codes[i].srcStr)) putc(c,stdout); /* ako nisu jednaki */
              else { printf("%s",codes[i].destStr); trenPos+=strlen(codes[i].srcStr)-1;}
	      }
   }
}

/**************************************************************************/
#define beginTagStr "<META"
#define chStr "charset"

int skipMarkers()
{char c;
 for(c=*Buffer_Read();c==' '|| c==13 || c==10;putc(c,stdout),c=*Buffer_Read());
 trenPos--;
 return 1;
}

/* shema() unutar META tagova trazi nisku oblika "charsetXXX=XXXiso8859-1" gde
   je XXX sekvenca karaktera ' ' i '\n' koji se ignorisu; potom se kodna
   shema zamenjuje trazenom */

int shema(int option)
{char *destStr[2]={"iso-8859-2","iso-8859-5"};
 char *trenStr, inTag=0;

 for (trenStr=Buffer_Read();*trenStr!=EOF;trenStr=Buffer_Read())
 { putc(*trenStr,stdout);
   if (inTag) {if (*trenStr=='>') inTag=0;
	       else if (!myStriCmp(trenStr,chStr))
		       {trenPos+=sizeof(chStr)-2;printf("%s",chStr+1);
			skipMarkers();
			trenStr=Buffer_Read(); putc(*trenStr,stdout);
			if (*trenStr=='=')
			  {skipMarkers();
			   if (!myStriCmp(trenStr=Buffer_Read(),"iso-8859-1"))
			      {printf("%s",destStr[option]);trenPos+=10-1;break;}
			   else putc(*trenStr,stdout);
			  }
			}
	       }
   else if (!myStriCmp(trenStr,beginTagStr)) {printf("%s",beginTagStr+1);   /* da li ulazimo u meta tag */
					      trenPos+=sizeof(beginTagStr)-2;
					      inTag=1;}
  }

 if (*trenStr==EOF) return 0;  /* neuspeh - oznaka kodne sheme nije nadjena */
 else {do putc(*(trenStr=Buffer_Read()),stdout); while (*trenStr!='>'); /* dodji do kraja taga */
       return 1;}
}
/**************************************************************************/

int toUpCase(int ch)
{ if (ch>=97 && ch<=122) return ch-32;
  else return ch;}

/**************************************************************************/
int main(int argc, char *argv[])
{char cparam=toUpCase(*argv[1]);

 if (cparam=='L') if (shema(0)) iso8859_2();
                  else { printf(" \n Pogresan format!"); return 1;}
 else if (cparam=='C') if (shema(1)) iso8859_5();
		  else { printf("\n Pogresan format!"); return 1;}
 else { printf(" Greska u parametru! \n"); return 1;}

 return 0; /* USPEH */

}