/* K D F I L T - filter program for kanjidic

See the "printf" statements below for some simple operating instructions

*/

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

#define TRUE 1
#define FALSE 0

long int lno=0;
FILE *fi,*fo,*fopen();
unsigned char cc,c,instr[512],outstr[512],codes[20];

int i,ip,istate,op,ic;

main(argc,argv)
int argc;
char **argv;
{
	int argk;
	register char **p;
	argk = argc;
	if (argk < 3)
	{
		printf("usage: kdfilt fni -CODES fno\n\n ");
		printf("KDFILT removes unwanted index codes from kanjidic\n\n ");
		printf("  fni - the input file name (e.g. kanjidic)\n ");
		printf("  fno - the output file containing the filtered kanjidic\n ");
		printf("  -CODES the index codes you want removed, e.g. -MQY will \n        result in the Morohashi, Four-Corner & PinYin codes being removed.\n\n ");
		exit(0);
	}
	p=argv;
	p++;
	fi= fopen(*p,"r");
	if (fi == NULL)
	{
		printf("\nCannot open %s\n",*p);
		exit(0);
	}
	p++;
	strcpy(codes,*p);
	if((codes[0] != '-') || (strlen(codes) < 2))
	{
		printf("usage: kdfilt fni -CODES fno\n ");
		exit(0);
	}
	p++;
	fo = fopen(*p,"w");
	if(fo == NULL)
	{
		printf("open of %s failed\n",*p);
		exit(0);
	}
	while (feof(fi) != TRUE)
	{
		fgets (instr,511,fi);
		if(feof(fi)) break;
		if(instr[0] == '#')
		{
			fprintf(fo,"%s",instr);
			continue;
		}
		if((++lno % 10) == 0)printf("Line: %ld\r",lno);
		strcat (instr,"\n");
		for(ip = 0;ip <8;ip++)
		{
			outstr[ip] = instr[ip];
		}
		op = 8;
		ip = 8;
		istate = 0;
		while (TRUE)
		{
			c = instr[ip];
			switch (istate) {

			case 0 :	/* start of field	*/
				if (c == ' ') break;
				if ((c == 0xa4) || (c == 0xa5) || (c == '{'))
				{
					for ( ; instr[ip] >= 0x20; ip++) outstr[op++] = instr[ip];
					c = 0x10;
					break;
				}
				if ( c < 0x20) break;
				ic = c;
				if( strchr(codes,ic) != NULL)
				{
					istate = 1;
					break;
				}
				outstr[op++] = c;
				istate = 2;
				break;

			case 1 :	/* in a field being skipped	*/
				if ( c < 0x20) break;
				if ( c == ' ')
				{
					if(outstr[op-1] != ' ') outstr[op++] = ' ';
					istate = 0;
					break;
				}
				break;
			case 2 :	/* in a field being copied */
				if ( c < 0x20) break;
				if ( c == ' ')
				{
					if(outstr[op-1] != ' ') outstr[op++] = ' ';
					istate = 0;
					break;
				}
				outstr[op++] = c;
				break;
			}
			if ( c < 0x20)
			{
				if(outstr[op-1] != ' ') outstr[op++] = ' ';
				outstr[op] = 0;
				fprintf(fo,"%s\n",outstr);
				break;
			}
			ip++;
		}
	}
	fclose(fi);
	fclose(fo);
}