/* convert.c
 * This file is split off to contain the kana<->romaji routines.
 */

#include <X11/Xlib.h>
#include <X11/Intrinsic.h>
#include <X11/keysym.h>

#include <stdio.h>

#include "defs.h"
#include "convert.h"
#include "searchwidgets.h"

/* 
 * Note.. these state indicators, are for parsing
 * romaji from RIGHT to LEFT
 * "state" stores the state to the right of where we are.
 */
enum { STATE_NONE, STATE_A, STATE_I, STATE_U, STATE_E, STATE_O,
		STATE_YA, STATE_YU, STATE_YO,
		/* we actualy use "y_state" instead of the above _YX states*/
		 STATE_HI, /* for chi or shi */
		 STATE_SU, /* for tsu */
		 STATE_SPACE,  /* for lone 'n' at end */
		 STATE_OTHER
} ;


/* Order of these is very important. DO NOT CHANGE */
enum  {ADD_NONE, ADD_A, ADD_I, ADD_U, ADD_E, ADD_O};
/*enum  {ADD_NONE, ADD_YA, ADD_FILL1, ADD_YU, ADD_FILL2, ADD_YO};*/


/* map for converting kana to romaji */
static char * kanamap[128]=
{
	"",   "",   "",   "",   "",   "",   "",   "",
	"",   "",   "",   "",   "",   "",   "",   "",


	"",   "",   "",   "",   "",   "",   "",   "",
	"",   "",   "",   "",   "",   "",   "",   "",  

	"",   "",   "a",  "",   "i",  "",   "u",  "",
	"e",  "",   "o",  "ka", "ga", "ki", "gi", "ku",

	"gu", "ke", "ge", "ko", "go", "sa", "za", "shi",
	"zhi", "su", "zu", "se", "ze", "so","zo", "ta",

	"da", "chi","ji", "",   "tsu","du", "te", "de",
	"to", "do", "na", "ni", "nu", "ne", "no", "ha",
	
	"ba", "pa", "hi", "bi", "pi", "fu", "bu", "pu",
	"he", "be", "pe", "ho", "bo", "po", "ma", "mi",
	
	"mu",  "me",  "mo","",  "ya", "",   "yu", "",
	"yo",  "ra",  "ri","ru","re", "ro", "",   "wa",

	"",    "",    "o", "n", "",   "",   "",   "",
	"",    "",    "",  "",  "",   "",   "",   "",

};




/* Take a kana string as input. Skip along, converting kana to
 * romaji. Assume 0x24 or 0x25 means kana or kanji start.
 * Assume estring buffer is "long enough".
 * Note that "long enough" should probably be
 *   kstrlen(kstring) *2 * 2;
 *
 * Sigh. better put a limit on it: MAXROMAJI
 *
 * Note that we also have to special-case punctuation. sigh.
 *  "(",     ")"
 *  0x214a,  0x214b
 *
 * All this, is so we can print out the "reading" entries, for folks
 * who can't read kana. WIMPS! :-)
 *
 * Currently, this is a state engine.
 * It would be much more consistent if I just used romajimap1/2[] and worked
 * backwards. Oh well.
 */
void kanatoromaji(XChar2b *kstring, char *estring)
{
	char *eptr = estring;
	XChar2b *kptr = kstring;
	int lastchar=0;	/* for state machine */

	eptr[0]='\0';
	
	do{
		if(strlen(estring)> (MAXROMAJI-3) ) {
			puts("WARNING: MAXROMAJI overflow");
			estring[MAXROMAJI-2]='X';
			estring[MAXROMAJI-1]='\0';
			return;
		}

		/* how boring.. punctuation */
		if(kptr->byte1 == 0x21)	{
			int val=kptr->byte2;
			/* print character before this punc */
			strcat(eptr, kanamap[lastchar]);

			lastchar=0;
			kptr++;
			switch(val) {
			   case 0x21:
				strcat(eptr," ");
				continue;
			   case 0x3c:
			   case 0x41:
				strcat(eptr,"-");
				continue;
			   case 0x4a:
			   case 0x5a:
				strcat(eptr,"(");
				continue;
			   case 0x4b:
			   case 0x5b:
				strcat(eptr,")");
				continue;
			}
			printf("warning.. kanatoromaji found 0x21,0x%x\n",
			       val);
			continue;
			
		}
		if((kptr->byte1 != 0x24 ) && (kptr->byte1 != 0x25 )
		   && (kptr->byte1 != 0x0))
		{
			printf("warning... 0x%x%x found in kanatoromaji\n",
			       kptr->byte1,kptr->byte2);
			kptr++;
			continue;
		}

		/******************************************************/


		/* This is a wierd parsing routine.. Try to ignore it!*/
		if(kptr->byte2>128){
			strcat(eptr, "[bad byte2 found]");
			kptr++;
			continue;
		}
		switch(kptr->byte2){
		   case 0x63:
		   case 0x65:
		   case 0x67:
			/* small ya,yu,yo. Need to backtrack */
			/* fall out of switch, and do so, below */
			/* This is a "goto", in essence */
			break;

		   case 0:
		   default:
			/* print previous char!... */
			/* then save THIS char,
			 * and continue through loop..
			 */
			strcat(eptr, kanamap[lastchar]);
			lastchar = kptr->byte2;
			kptr++;
			continue;
			/* hopefully, will continue through while loop */

		}

		/* Must be on ya,yu,yo */
		/* Put the appropriate prefix */
		/* Later, will append "a", "u", or "o" */

		switch(lastchar){
		   case 0x2d:	/*ki*/
			strcat(eptr,"ky");
			break;
		   case 0x2e:	/*gi*/
			strcat(eptr,"gy");
			break;
		   case 0x37:	/*shi*/
			strcat(eptr,"sh");
			break;
		   case 0x38:	/* shi''*/
			strcat(eptr,"j");
			break;
		   case 0x41:	/*chi*/
			strcat(eptr,"ch");
			break;
		   case 0x4b:	/*ni*/
			strcat(eptr,"ny");
			break;
		   case 0x52:	/*hi*/
			strcat(eptr,"hy");
			break;
		   case 0x53:	/*bi*/
			strcat(eptr,"by");
			break;
		   case 0x54:	/*pi*/
			strcat(eptr,"py");
			break;
		   case 0x5f:	/*mi*/
			strcat(eptr,"my");
			break;
		   case 0x6a:	/*ri*/
			strcat(eptr,"ry");
			break;
		   default:
			/*strcat(eptr, "[bad placement of ya/yu/yo]");*/
			/* oh well.. just print it as-is */
			/* but mark it as small */
			strcat(eptr, "_y");
			break;
		}
		switch(kptr->byte2){
		   case 0x63:
			strcat(eptr, "a");
			break;
		   case 0x65:
			strcat(eptr, "u");
			break;
		   case 0x67:
			strcat(eptr, "o");
			break;
			/* must be ya,yu,yo */
		}
		lastchar = kptr->byte2;
		kptr++;

	} while (kptr->byte1 != 0);
	strcat(eptr, kanamap[lastchar]);

}


#define ROMAJIMAPSIZE 150 /* there is some empty space for growth */

/* romajimap1 has to match romajimap2, in exact order!! */
/* If we were nice, this would also be XChar2b encoded, but..*/
static char *romajimap1[ROMAJIMAPSIZE]=
{
	"shya", "shyu", "shyo",
	"chya","chyu","chyo",
	"sya","syu","syo",
	"sha","shi", "shu","sho",
	"jya","jyu","jyo",
	"cha","chu","cho",
	"gya","gyu","gyo",
	"dzu","tzu",             /* same as "du" */

	"hya","hyu","hyo",
	"bya","byu","byo",
	"pya","pyu","pyo",

	"mya","myu","myo",
	"nya","nyu","nyo",

	"kya","kyu","kyo",
	"rya","ryu","ryo",
	"chi","tsu",
	"ka","ga","ki","gi","ku","gu","ke","ge","ko","go",
	"sa","za","si","zi","su","zu","se","ze","so","zo",
	"ja", "ji", "ju", "jo",
	"ta","da","ti","di","tu","du","te","de","to","do",

	"na","ni","nu","ne","no",
	"n ","nn",
	"ha","ba","pa",
	"hi","bi","pi",
	"hu","fu","bu","pu",
	"he","be","pe",
	"ho","bo","po",

	
	"ma","mi","mu","me","mo",
	"ra","ri","ru","re","ro",
	"wa","wo",
	"ya","yu","yo",
	"_ya","_yu","_yo", /* Small letters, the hard way. For romaji input*/
	"xya","xyu","xyo", /* Small letters, the hard way. For romaji input*/
	"a","i","u","e","o",

	/* the 5 faces of smalltsu */
	"'","_tu","_tsu","xtu","xtsu",


	"-", "(",")", /* These are here not so much for romaji input,
	                   * as to match dictionary search patterns,
			   * and translate kana to romaji.
			   */
	""
	/* See also hardcodes for special keys in Handle_romajikana()
	 * and matchromaji()
	 */
};

/* must match order of romajimap1[]!!*/
static XChar2b romajimap2[ROMAJIMAPSIZE][3]=
{
	{{0x24, 0x37}, {0x24, 0x63}, {0, 0}},/*shya*/
	{{0x24, 0x37}, {0x24, 0x65}, {0, 0}},/*shyu*/
	{{0x24, 0x37}, {0x24, 0x67}, {0, 0}},/*shyo*/

	{{0x24, 0x41}, {0x24, 0x63}, {0, 0}},/*chya*/
	{{0x24, 0x41}, {0x24, 0x65}, {0, 0}},/*chyu*/
	{{0x24, 0x41}, {0x24, 0x67}, {0, 0}},/*chyo*/

	{{0x24, 0x37}, {0x24, 0x63}, {0, 0}},/*sya*/
	{{0x24, 0x37}, {0x24, 0x65}, {0, 0}},/*syu*/
	{{0x24, 0x37}, {0x24, 0x67}, {0, 0}},/*syo*/

	{{0x24, 0x37}, {0x24, 0x63}, {0, 0}},/*sha*/
	{{0x24, 0x37}, {0, 0}},		   /*shi*/
	{{0x24, 0x37}, {0x24, 0x65}, {0, 0}},/*shu*/
	/*{{0x24, 0x37}, {0x24, 0x67}, {0, 0}}, ??she??*/
	{{0x24, 0x37}, {0x24, 0x67}, {0, 0}},/*sho*/

	{{0x24, 0x38}, {0x24, 0x63}, {0, 0}},/*jya*/
	{{0x24, 0x38}, {0x24, 0x65}, {0, 0}},/*jyu*/
	{{0x24, 0x38}, {0x24, 0x67}, {0, 0}},/*jyo*/

	{{0x24, 0x41}, {0x24, 0x63}, {0, 0}},/*cha*/
	{{0x24, 0x41}, {0x24, 0x65}, {0, 0}},/*chu*/
	{{0x24, 0x41}, {0x24, 0x67}, {0, 0}},/*cho*/

	{{0x24, 0x2e}, {0x24, 0x63}, {0, 0}},/*gya*/
	{{0x24, 0x2e}, {0x24, 0x65}, {0, 0}},/*gyu*/
	{{0x24, 0x2e}, {0x24, 0x67}, {0, 0}},/*gyo*/

	{{0x24, 0x45}, {0, 0}},/*dzu*/
	{{0x24, 0x45}, {0, 0}},/*tzu*/

	{{0x24, 0x52}, {0x24, 0x63}, {0, 0}},/*hya*/
	{{0x24, 0x52}, {0x24, 0x65}, {0, 0}},/*hyu*/
	{{0x24, 0x52}, {0x24, 0x67}, {0, 0}},/*hyo*/

	{{0x24, 0x53}, {0x24, 0x63}, {0, 0}},/*bya*/
	{{0x24, 0x53}, {0x24, 0x65}, {0, 0}},/*byu*/
	{{0x24, 0x53}, {0x24, 0x67}, {0, 0}},/*byo*/

	{{0x24, 0x54}, {0x24, 0x63}, {0, 0}},/*pya*/
	{{0x24, 0x54}, {0x24, 0x65}, {0, 0}},/*pyu*/
	{{0x24, 0x54}, {0x24, 0x67}, {0, 0}},/*pyo*/

	{{0x24, 0x5f}, {0x24, 0x63}, {0, 0}},/*mya*/
	{{0x24, 0x5f}, {0x24, 0x65}, {0, 0}},/*myu*/
	{{0x24, 0x5f}, {0x24, 0x67}, {0, 0}},/*myo*/

	{{0x24, 0x4b}, {0x24, 0x63}, {0, 0}},/*nya*/
	{{0x24, 0x4b}, {0x24, 0x65}, {0, 0}},/*nyu*/
	{{0x24, 0x4b}, {0x24, 0x67}, {0, 0}},/*nyo*/

	{{0x24, 0x2d}, {0x24, 0x63}, {0, 0}},/*kya*/
	{{0x24, 0x2d}, {0x24, 0x65}, {0, 0}},/*kyu*/
	{{0x24, 0x2d}, {0x24, 0x67}, {0, 0}},/*kyo*/

	{{0x24, 0x6a}, {0x24, 0x63}, {0, 0}},/*rya*/
	{{0x24, 0x6a}, {0x24, 0x65}, {0, 0}},/*ryu*/
	{{0x24, 0x6a}, {0x24, 0x67}, {0, 0}},/*ryo*/


	{{0x24, 0x41}, {0, 0}},/*chi*/
	{{0x24, 0x44}, {0, 0}},/*tsu*/

	{{0x24, 0x2b}, {0, 0}},/*ka*/
	{{0x24, 0x2c}, {0, 0}},/*ga*/
	{{0x24, 0x2d}, {0, 0}},/*ki*/
	{{0x24, 0x2e}, {0, 0}},/*gi*/
	{{0x24, 0x2f}, {0, 0}},/*ku*/
	{{0x24, 0x30}, {0, 0}},/*gu*/
	{{0x24, 0x31}, {0, 0}},/*ke*/
	{{0x24, 0x32}, {0, 0}},/*ge*/
	{{0x24, 0x33}, {0, 0}},/*ko*/
	{{0x24, 0x34}, {0, 0}},/*go*/

	{{0x24, 0x35}, {0, 0}},/*sa*/
	{{0x24, 0x36}, {0, 0}},/*za*/
	{{0x24, 0x37}, {0, 0}},/*si*/
	{{0x24, 0x38}, {0, 0}},/*zi*/
	{{0x24, 0x39}, {0, 0}},/*su*/
	{{0x24, 0x3a}, {0, 0}},/*zu*/
	{{0x24, 0x3b}, {0, 0}},/*se*/
	{{0x24, 0x3c}, {0, 0}},/*ze*/
	{{0x24, 0x3d}, {0, 0}},/*so*/
	{{0x24, 0x3e}, {0, 0}},/*zo*/

	{{0x24, 0x38}, {0x24, 0x63}, {0, 0}},/*ja*/
	{{0x24, 0x38}, {0, 0}},/*ji*/
	{{0x24, 0x38}, {0x24, 0x65}, {0, 0}},/*ju*/
	{{0x24, 0x38}, {0x24, 0x67}, {0, 0}},/*jo*/


	{{0x24, 0x3f}, {0, 0}},/*ta*/
	{{0x24, 0x40}, {0, 0}},/*da*/
	{{0x24, 0x41}, {0, 0}},/*ti*/
	{{0x24, 0x42}, {0, 0}},/*di*/
	{{0x24, 0x44}, {0, 0}},/*tu*/
	{{0x24, 0x45}, {0, 0}},/*du*/
	{{0x24, 0x46}, {0, 0}},/*te*/
	{{0x24, 0x47}, {0, 0}},/*de*/
	{{0x24, 0x48}, {0, 0}},/*to*/
	{{0x24, 0x49}, {0, 0}},/*do*/

	{{0x24, 0x4a}, {0, 0}},/*na*/
	{{0x24, 0x4b}, {0, 0}},/*ni*/
	{{0x24, 0x4c}, {0, 0}},/*nu*/
	{{0x24, 0x4d}, {0, 0}},/*ne*/
	{{0x24, 0x4e}, {0, 0}},/*no*/

	{{0x24, 0x73}, {0, 0}},/*'n '*/
	{{0x24, 0x73}, {0, 0}},/*'nn'*/

	{{0x24, 0x4f}, {0, 0}},/*ha*/
	{{0x24, 0x50}, {0, 0}},/*ba*/
	{{0x24, 0x51}, {0, 0}},/*pa*/

	{{0x24, 0x52}, {0, 0}},/*hi*/
	{{0x24, 0x53}, {0, 0}},/*bi*/
	{{0x24, 0x54}, {0, 0}},/*pi*/

	{{0x24, 0x55}, {0, 0}},/*hu*/
	{{0x24, 0x55}, {0, 0}},/*fu*/
	{{0x24, 0x56}, {0, 0}},/*bu*/
	{{0x24, 0x57}, {0, 0}},/*pu*/

	{{0x24, 0x58}, {0, 0}},/*he*/
	{{0x24, 0x59}, {0, 0}},/*be*/
	{{0x24, 0x5a}, {0, 0}},/*pe*/

	{{0x24, 0x5b}, {0, 0}},/*ho*/
	{{0x24, 0x5c}, {0, 0}},/*bo*/
	{{0x24, 0x5d}, {0, 0}},/*po*/

	{{0x24, 0x5e}, {0, 0}},/*ma*/
	{{0x24, 0x5f}, {0, 0}},/*mi*/
	{{0x24, 0x60}, {0, 0}},/*mu*/
	{{0x24, 0x61}, {0, 0}},/*me*/
	{{0x24, 0x62}, {0, 0}},/*mo*/

	{{0x24, 0x69}, {0, 0}},/*ra*/
	{{0x24, 0x6a}, {0, 0}},/*ri*/
	{{0x24, 0x6b}, {0, 0}},/*ru*/
	{{0x24, 0x6c}, {0, 0}},/*re*/
	{{0x24, 0x6d}, {0, 0}},/*ro*/


	{{0x24, 0x6f}, {0, 0}},/*wa*/
	{{0x24, 0x72}, {0, 0}},/*wo*/


	{{0x24, 0x64}, {0, 0}},/*ya*/
	{{0x24, 0x66}, {0, 0}},/*yu*/
	{{0x24, 0x68}, {0, 0}},/*yo*/

	{{0x24, 0x63}, {0, 0}},/*_ya*/
	{{0x24, 0x65}, {0, 0}},/*_yu*/
	{{0x24, 0x67}, {0, 0}},/*_yo*/

	{{0x24, 0x63}, {0, 0}},/*xya*/
	{{0x24, 0x65}, {0, 0}},/*xyu*/
	{{0x24, 0x67}, {0, 0}},/*xyo*/

	{{0x24, 0x22}, {0, 0}},/*a*/
	{{0x24, 0x24}, {0, 0}},/*i*/
	{{0x24, 0x26}, {0, 0}},/*u*/
	{{0x24, 0x28}, {0, 0}},/*e*/
	{{0x24, 0x2a}, {0, 0}},/*o*/

	{{0x24, 0x43}, {0, 0}},/*"'" for small-tsu. a bunch */
	{{0x24, 0x43}, {0, 0}},
	{{0x24, 0x43}, {0, 0}},
	{{0x24, 0x43}, {0, 0}},
	{{0x24, 0x43}, {0, 0}},

	/* These are here really for kana-to-romaji only */
	/* But make sure Handle_romajikana uses same values! */
	{{0x21, 0x3c}, {0, 0}},/*- for elongation*/	
	{{0x21, 0x4a}, {0, 0}},/*"(" */
	{{0x21, 0x4b}, {0, 0}},/*")" */

	{{0}}
	
};


/* subroutine for romajitokana()
 * romaji is one of the standard romaji tokens.
 * kstring is user input.
 *
 * return 1 if kstring matches the known romaji combination.
 */
static int matchromaji(char *romaji, XChar2b *kstring)
{
	while(*romaji!='\0'){
		if(kstring->byte1==0){
			return 0;
		}

		/*durn non-ascii punctuation codes...*/
		/* IFF first char is JIS '_', then try to match
		 * with "_ya" romaji combinations.
		 * If we dont HAVE a _yXX combo to compare with,
		 * then we dont have a match!
		 */
		if(kstring->byte1==0x21){
			if(kstring->byte2 == 0x32){
				if(*romaji == '_'){
					/* force match */
					romaji++;
					kstring++;
					continue;
				} else {
					return 0;
				}
			}
		}

		if(*romaji != kstring->byte2){
			return 0;
		}

		if(kstring->byte1!=0x23) {
			/* Theoretically, this should be redundant. */
			/* All matches should be ASCII to pseudo-ASCII now */

			printf("WARNING: matchromaji hit nonascii:%x:%x\n",
			       kstring->byte1,kstring->byte2);
			return 0;
		}


		romaji++;
		kstring++;
	}

	/* match found! */
	return 1;
}
/*
 * We get called by process_kinput, on the "raw" string,
 * before the string actually gets displayed.
 *
 * We assume kstart points to an array that we can modify in-place.
 * We also assume array is fully XChar2b encoded, as a "mix" of
 * kana and romaji. But we assume that it always starts with 0 or more
 * kana, and followed by 0 or more romaji. (encoded as 0x23,0x??)
 * We never should be more mixed than that. If we are, we don't
 * handle it well :-)
 *
 * Expected range of characters is determined by whatever
 * Handle_romajikana puts in the string, minus whatever process_kinput
 * has filtered out.
 */
void romajitokana(XChar2b *kstart)
{
	XChar2b *kparse;
	int matchcount;

	/* start off by looking for pseudo-8-bit chars, aka romaji.
	 * look for first possible romaji string match, or end-of-string.
	 */
	while(kstart->byte1!=0) {
		if(kstart->byte1==0x23){
			break;
		}
		/* special handling just for _ya combo */
		if(kstart->byte1==0x21){
			if(kstart->byte2==0x32){
#ifdef DEBUG
				puts("romajitokana: found __");
#endif
				break;
			}
		}
		kstart++;
	}
	if(kstart->byte1==0){
		/* no match! */
		return;
	}
	
	/* kstart now has first ascii-like 8-bit char
	 * embedded in the XChar2b string
	 */
	kparse = kstart;


	for(matchcount=0; romajimap1[matchcount][0]!='\0' ; matchcount++){
		if(matchromaji(romajimap1[matchcount], kstart)==1){
			kparse=&romajimap2[matchcount][0];
			while(kparse->byte1 != 0){
				kstart->byte1=kparse->byte1;
				kstart->byte2=kparse->byte2;
				kstart++;
				kparse++;
			}
			/* yes, force truncation */
			kstart->byte1=kstart->byte2=0;
			return;
		}
	}

	/* Special case for 'n' in the middle of things.
	 * Convert it ourselves to standalone 'n', unless potentially part
	 * of "nya" or similar known compound
	 */
	if((kstart[0].byte1==0x23) && (kstart[0].byte2=='n'))
		if((kstart[1].byte1!=0) &&  (kstart[1].byte2!='y') &&
		   (kstart[1].byte2!='n'))
		{
			kstart[0].byte1=0x24;
			kstart[0].byte2=0x73;
		}
	return;
}


	

	

/************************************************************/

/* This is a KeyPress event handler 
 * Gets called every time a key is pressed in the romajikana window.
 * ALso handles "go search now".
 */
void Handle_romajikana(Widget w, XtPointer closure, XEvent *e, Boolean *cont)
{
	XKeyEvent *key_event;
	KeySym inbetweenK;
	char *charpressed;
	XChar2b addchar;
	
	if(e->type != KeyPress){
		if(e->type == KeyRelease){
#ifdef DEBUG
			puts("key released");
#endif
			return;
		}
		printf("Some other strange event found in for romaji: %d\n",
		       e->type);
		return;
	}
	key_event = (XKeyEvent *) e;
	
	/*inbetweenK = XKeycodeToKeysym(XtDisplay(w), key_event->keycode,0);*/
	/* need XtGetActionkeysym to detect Shift-9=='(' */
	inbetweenK = XtGetActionKeysym(e, NULL);

	if(inbetweenK == (KeySym)NULL){
		puts("NULL keysym on kana input???");
		return;
	}

	/* we switch based on what character has just been
	 * pressed on an ASCII-based keyboard.
	 * <return> gets translated to "go search now"
	 * Making byte1== 0x23, is a special signal to romajitokana,
	 * that says "Hey, this is romaji"
	 */
	switch(inbetweenK){
	   case XK_BackSpace:
	   case XK_Delete:
		addchar.byte1 = 0x22;
		addchar.byte2 = 0x2b;
		process_kinput(addchar);
		return;
	   case XK_Return:
	   /* pass our strange "accept" char, that means "do search now"*/
		addchar.byte1 = paragraphglyph[0].byte1;
		addchar.byte2 = paragraphglyph[0].byte2;
		process_kinput(addchar);
		return;
	   case XK_space:
		/* add " ", but in kana range */
		/* This is a nasty hack to get "n" right */
		addchar.byte1 = 0x23;
		addchar.byte2 = ' ';
		process_kinput(addchar);
		return;
	/* Special handling for punctuation, because JIS punctuation
	 * conflicts with 8-bit numbers of ASCII punctuation
	 */
	   case XK_minus:
		addchar.byte1 = 0x21;
		addchar.byte2 = 0x3c;	/* -- in JIS */
		process_kinput(addchar);
		return;
	   case XK_apostrophe:
		addchar.byte1 = 0x24;
		addchar.byte2 = 0x43;  /* "'" in JIS */
		process_kinput(addchar);
		return;

	   case XK_parenleft:
		/*
		 * Note that this "char" is invisible in JIS space
		 * It is pseudo-ASCII.  See note on XK_minus
		 */
		addchar.byte1 = 0x21;
		addchar.byte2 = 0x4a;
			/* in JIS it would be 214a */
		process_kinput(addchar);
		return;
	   case XK_parenright:
		/*
		 * directly convert to JIS ourselves.
		 * These must match whatever ReadPronunciation()
		 * converts '()' in the dictionary to.
		 */
		addchar.byte1 = 0x21;
		addchar.byte2 = 0x4b;
			/* in JIS it would be 214b */
		process_kinput(addchar);
		return;
		
	   case XK_underscore:
		addchar.byte1 = 0x21;
		addchar.byte2 = 0x32; /* 0x2132 == '_' in JIS */
		process_kinput(addchar);
		return;
	}
	charpressed = XKeysymToString(inbetweenK);
	if(charpressed == NULL)
		return;

#ifdef DEBUG
	printf("got string \"%s\"\n", charpressed);
#endif		
	/* now use process_kinput, 222b is erase */
	if((*charpressed <0x61) || (*charpressed >0x7a)){
		/* outside range of ascii chars we like */
		if((*charpressed!='(') && *charpressed != ')'){
#ifdef DEBUG
			puts("ignoring.. not in normal ascii range");
#endif
			return;
		}
	}
	
	addchar.byte1 = 0x23;
	addchar.byte2 = *charpressed;
	process_kinput(addchar);

}