#include <stdio.h>
#include <string.h>
#include <assert.h>
#include <ctype.h>

#define BUF 65535

/***********************************************************************
 *   han2rom  -- Korean hangul romanizer for EUC_KR (KS X 1001:1992)   *
 *								       *
 *         Romanizes according to the ugly but machine-friendly	       *
 *         ISO TR 11941:1996 (E) transliteration scheme		       *
 *								       *
 *   Written by Nick Cipollone, Nov/4/2000			       *
 *   Adapte  Chris Brew, Oct/17/2001,Feb/7/2002        *
 *                                                                     *
 *                                                                     *
 ***********************************************************************/

/* The matrix of hangul syllable romanizations: */

#include "romaca.c"

/* The single hangul characters are off in row 4 of the KS X 1001:1992 table,
   while the hangul syllables are in rows 16-40.  */

char* single_hangul[94] = 
    { "k", "kk", "ks", "n", "nc", "nh", "t", "tt", "r", "rk", "rm", "rp", "rs", "rth", "rph", "rh", "m", "p", "pp", "ps", "s", "ss", "ng", "c", "cc", "ch", "kh", "th", "ph", "h", "a", "ae", "ya", "yae", "eo", "e", "yeo", "ye", "o", "wa", "wae", "oe", "yo", "u", "weo", "we", "wi", "yu", "eu", "yi", "i", " ", "nn", "nt", "ns", "nm", "rks", "rt", "rps", "rm", "rh", "mp", "ms", "mm", "mng", "pk", "pt", "psk", "pst", "pc", "pth", "png", "ppng", "sk", "sn", "st", "sp", "sc", "m", "ngng", "ng", "ngs", "ngm", "phng", "hh", "h", "yoya", "yoyae", "yoe", "yuyeo", "yuye", "yui", "eu", "yi" };

/* For the double consonants, the aspirated stops and ieung (/ng/) 
   we need to add a leading apostrophe word-internally.  This matrix 
   is parallel to romaca and encodes which syllables this rule
   applies to: 

   Comment: this is done in order to disambiguate syllabification. An
   adapted version that explicitly marks syllable boundaries would
   not need these diacritics.
 */

char apostrophe[25][94] = {
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1 },
		{ 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 },
		{ 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 } };




/*
 * A token is a structure that represents the occurrence
 * of a term in a text. It contains
 * -- the text of the term, not necessarily NULL terminated
 * -- start and end offsets into the source text, provided
 *    to allow tokens to be re-associated with their source
 *    for KWIC displays and similar
 * -- a field for type information about the token. Typically
 *    part-of-speech or similar.
 * 
 */


/*
 * In the basic tokenizer token.type takes the following
 * values
 */

typedef enum tokenizer_type_indices {kHangul,
				     kRoman,
				     kWhitespace,
				     kPunct,
				     kMarkup,
				     kNoMoreTypes} tokindex;

typedef const char * Atom;

typedef struct token {
  const unsigned char * text;
  size_t size;
  tokindex type;
  int hyphenate;  /* false if just next to whitespace */
} token,*TOKEN;

typedef struct mapping  {
  const char * nick;
  const char * appendix_n;
} mapping;
 
 

/*
 * Modification by Chris: use knowledge of possible vowels to
 * analyse segment strings. Following is a NULL terminated list
 * of Roman representations of vowels. It matters that the ones
 * that have two letters in the orthography come first, since
 * we use the order of the list to resolve conflicts. 
 *
 * 
 */



static const mapping osu_korean_vowels[] = {
  {"eu","eu"},
  {"eo","eo"},
  {"ae","ae"},  /* not in Kyuchul's dialect */
  {"oe","oe"}, /* not in the phone set, but is in appendix N */
  {"u","u"},
  {"e","e"},
  {"i","i"},
  {"o","o"},
  {"a","a"},
  {NULL,NULL}};


static const mapping osu_korean_consonants[] = {
  {"ph","p"},
  {"pp","bb"},  
  {"th","t"},
  {"tt","dd"},
  {"ch","c"},
  {"cc","jj"}, 
  {"kh","k"},
  {"kk","gg"},
  {"ng","ng"},
  {"ss","ss"},
  {"p","b"},
  {"t","d"},
  {"c","j"},
  {"k","g"},					       
  {"m","m"},
  {"n","n"},
  {"s","s"},
  {"w","w"},
  {"l","l"},
  {"r","l"},
  {"y","y"},
  {"h","h"},
  {NULL,NULL}
};






/*
 * find_vowel - return the first substring starting with a vowel 
 */ 

static const mapping * find_vowel(const char * target) {

  const mapping *p;


  for(p = osu_korean_vowels; p->nick != NULL; p++) {
      if(0 == strncmp(target,p->nick,strlen(p->nick)))
	return p;
  }


  return NULL;

}


static const mapping * find_consonant(const char * target) {

  const mapping *p;


  for(p = osu_korean_consonants; p->nick != NULL; p++) {
      if(0 == strncmp(target,p->nick,strlen(p->nick)))
	return p;
  }


  return NULL;

}


/*
 * single hangul need translation too
 */

static const mapping * find_any(const char * target) {
  const mapping * p;
  p = find_vowel(target);
  if(p) 
    return p;
  else
    return find_consonant(target);
}









static void print_syllable(const char * syllable,int hyphenate) {

  const mapping * m;


  if(hyphenate) 
    printf("-");
 
  while(( m = find_consonant(syllable))) {
    printf("%s",m->appendix_n);
    syllable += strlen(m->nick);
  }
  
  if((m=find_vowel(syllable))) {
    printf("%s",m->appendix_n);
    syllable += strlen(m->nick);
  } else {
    printf("__%s__",syllable);
    exit(-1);
  }


  while((m = find_consonant(syllable))) {
    printf("%s",m->appendix_n);
    syllable += strlen(m->nick);
  }


}


static void print_syllable_nospace(const char * syllable) {

  const mapping * m;


  while((m = find_consonant(syllable))) {
    printf("%s",m->appendix_n);
    syllable += strlen(m->nick);
  }
  if((m=find_vowel(syllable))) {
    printf("%s",m->appendix_n);
    syllable += strlen(m->nick);
  } else {
    printf("__%s__",syllable);
    exit(-1);
  }



  while((m = find_consonant(syllable))) {
    printf("%s",m->appendix_n);
    syllable += strlen(m->nick);
  }

}





typedef void (*Printer)(const token tok);

static void print_single_hangul(const char * fragment) {
    const mapping * m;


  while((m = find_any(fragment))) {
    printf("%s ",m->appendix_n);
    fragment += strlen(m->nick);
  }

}


void print_hangul(const token tok) {
  unsigned char c1 = tok.text[0];
  unsigned char c2 = tok.text[1];
  int hyphenate = tok.hyphenate;
  const int offset = 176;


  if(c1 == 164) {

    /*  
  printf("164-%02d:",c2-offset+16 );
  printf("%c%c=",164,c2);
    */

  print_single_hangul(single_hangul[c2-offset+15]);
  /* printf("\n"); */
  
  } else {

  /*
   * Print the character itself
   */


  /* printf(" %c%c = ",c1,c2); */

  /*
   * Print the romanization of the syllable
   */

  print_syllable(romaca[c1-offset][c2-offset+15],hyphenate);
  /* printf("\n"); */
  }
}


void print_roman(const token tok) {
 
  const char * text = tok.text;
  size_t n = tok.size;
  while(n--){
    printf("%c",*text++);
   
    
  }

  
}


void print_whitespace(const token tok) {

  const char * text = tok.text;
  size_t n = tok.size;
  
  while(n--){
    
    printf("%c",*text++);
  }
  
  

}


void dont_print(const token tok) {

  const char * text = tok.text;
  size_t n = tok.size;
  
  while(n--){
    
    text++;
  }
  
  

}




static Printer printers[kNoMoreTypes] = 
{
    print_hangul,/* Text in Hangul */
    dont_print, /* Text in Roman */
    print_whitespace, /* Text that is whitespace */
    dont_print,  /* Punctuation               */
    dont_print  /* XML markup */

};
    



Printer find_printer(token tok) {

  if (tok.type >= 0 && tok.type < kNoMoreTypes)
    return printers[tok.type];
  else
    return NULL;

}




void print_token(token tok) {
  
  
  Printer p = find_printer(tok);
  if(p!=NULL) {
    (*p)(tok);
  }
  
}


/*
 * Change: there's an "oe" vowel used in the transcription scheme, added it
 * On this,Mary Beckman  sends the following
 *
 * I just asked Soyoung, and Sun-hee and Misun were here to chime in.	
 * Just as the historically high front rounded vowel had broken into a	
 * sequence of rounded glide followed by unrounded vowel /wi/, the	
 * historically mid front rounded vowel that is transliterated as "oe"	
 * in this table is breaking into /we/, so that, e.g.  17-11: goe		* is (for most younger speakers today) homophonous with 17-43 gwe, and	
 * also with 17-05 gwae (=gwe in our system), and I guess 
 * 17-12:  = goeg g oe g can be rewritten as "gweg", and so on.  
 *
 * On the basis of this, we could provide an optional mapping function
 * that maps "oe" into "we".
 * 
 * Other changes:
 *
 * 19-65 is transcribed as "nyeog", Nick had "nyeong" we have to put "nyeok" 
 * into the table.  
 *
 * 19-72 is transcribed as "nyeok", Nick had "nyeog", to get "nyeok", the 
 * entry should be "nyeokh" 
 *
 * 20-39 is transcribed as neub, to get that we need "neup" in the table 
 * 20-43 is transcribed as neup, to get that we need "neuph" in the table 
 * 32-28 is transcribed as eulb, should be eulp, so we put in "eulph" 
 * 40-08 is transcribed as "hom", Nick had "holm". This is presumably just
 * a mistake.
 */

















/*
 * We're trying to get our phone names to look just like those in 
 * appendix N of Lunde, Nick's are a bit different, no doubt because
 * he (and others) know the Yale system. Following is a table of
 * old and new codes.
 * Differences are
 * (1) where Nick writes the fortis stops as
 *   pp, tt, cc, and kk, * the new romanization has bb, dd, jj, and
 *   gg.  
 * (2) where we had been writing the aspirated stops as ph, th,
 *   ch, and kh, the new romanization has p, t, c, and k 
 *
 * We don't need to do anything for vowels, because they are the same
 * in both systems. Because we are simulating Kyuchul's variety of
 * Korean , our phoneset assigns the same phone to "ae" and "e", but 
 * in the present program we maintain the distinction. The mapping
 * table for vowels is the identity, and exists primarily because
 * we want find_vowel and find_consonant to have the same interface (a
 * property that we need in order to translate Nick's romanizations of
 * single hangul.)
 */




static int is_ascii_punctuation(const unsigned char * s) {

  return ispunct(*s);
  
}




/*
 * copy_ascii -- deals with copying a sequence
 * starting with an ASCII character to the
 * current token.
 * 
 * We have a simplistic view of XML markup and
 * copy <.*> in one shot. This will fail if there
 * are embedded '>' in the attributes of some tag.
 */

static const unsigned char * copy_ascii(TOKEN tok,const unsigned char * text) {

  const unsigned char * base = tok->text = text;

  if(*text == '<') {
    for( ; *text && *text != '>'; text++) {
      ;
    }
    if(*text){
      /* then we need to print out the > */
      *text++;
    }

    tok->size = text-base;
    tok->type = kMarkup; 
  } else {
    
    text++;
    tok->size = 1;
    tok->type = kRoman; 

  }
  return text;
}


static const unsigned char * copy_regular_hangul(TOKEN tok,const unsigned char *text,int just_did_whitespace) {
  const unsigned char * base = text;
  unsigned char c1 = *text++;
  unsigned char c2 = *text++;

  tok->text = base;
  tok->size = 2;
  tok->type = kHangul;
  tok->hyphenate = !just_did_whitespace;
  return text;
}


static const unsigned char * copy_single_hangul(TOKEN tok,const unsigned char * text,int just_did_whitespace) {
  const char * base = text;
  int ch = *(++text);  /* Skip the 164 */
  const int offset = 176;
  tok->text = base;
  tok->size = 2;
  tok->type = kHangul;
  tok->hyphenate = !just_did_whitespace;
  return ++text;



}


static const unsigned char *copy_spaces(TOKEN tok,const unsigned char * text) {
   
  tok->text = text;
  while(*text && isspace(*text))
    text++;
  tok->size = text - tok->text;

  tok->type = kWhitespace;
  return text;
}




/*
        fgets_unicode
        
	Similar to 'fgets()', but avoids breaking two byte
	characters.


	Return value: as fgets
 

*/

unsigned char * fgets_unicode(unsigned char *line,int n,FILE *fp)	
{
  unsigned char * result = fgets(line,n,fp);
  int nbytes,ch;
  

  if(result != NULL) {
    nbytes = strlen(result);
    ch = result[nbytes-1];
    if(ch & 0x80) {
      result[nbytes-1] = 0;
      ungetc(ch,fp);

      /* Possibly it was the only byte available */
      if(nbytes == 1)
	result = NULL;
      
    }
  }
  
  return result;

}



/*
 * By default this program prints only the hangul
 * part of each file. 
 */

int main(int argc,char**argv) {
 
  static unsigned char line[BUF];   /* we're dealing with 8-bit chars */
  const unsigned char *temp,* limit = line+BUF;
  static token tok;
  int just_did_whitespace;
  int arg = 0;

  while(++arg < argc) {
    if(strcmp(argv[arg],"--roman") == 0) {
      printers[kRoman] = print_roman;

    }

    if(strcmp(argv[arg],"--markup") == 0) {
      printers[kMarkup] = print_roman;

    }

    if(strcmp(argv[arg],"--punctuation") == 0) {
      printers[kPunct] = print_roman;

    }
    
  }

  while (fgets_unicode(line,BUF,stdin) != NULL) {
    temp = line;
    while(temp < limit && *temp != '\0') {
      if (isspace(*temp)) {     
	temp = copy_spaces(&tok,temp);
	just_did_whitespace = 1;
      } else if (*temp >= 176 && *temp < 201 ) {     
	/* it's regular hangul */
	  temp = copy_regular_hangul(&tok,temp,just_did_whitespace);
	  just_did_whitespace = 0;
      } else if (*temp == 164) {                
	/* 
	 * it's a single hangul 
	 */
	temp = copy_single_hangul(&tok,temp,just_did_whitespace);
	just_did_whitespace = 0;  
      } else {
	temp = copy_ascii(&tok,temp);
	just_did_whitespace = 0;  
	
      }

      print_token(tok);
    }
  }
  return 0;
}