/*******************************************************/
/* copyright (C) Simorgh Software. All rights reserved.*/
/*                                                     */
/* gsz2isi : gerdsooz to ISIRI-3342 encoding converter */
/*                                                     */
/* Compilation:                                        */
/*                                                     */
/*   On a Unix machine or PC with lex compiler utility */
/*   execute the following commands:                   */
/*                                                     */
/*   lex gsz2isi.l                                     */
/*   cc lex.yy.c -o gsz2isi -ll                        */
/*                                                     */
/* Requirements:                                       */
/*                                                     */
/*   input.gsz : a gerdsooz file with =begin= and      */
/*              =end= tags, marking the encoding area. */
/* Usage:                                              */
/*                                                     */
/*   gsz2isi < input.gsz > output.isi                  */
/*                                                     */
/* Options:                                            */
/*                                                     */
/*   -e  Pass through any english text outside of      */
/*       the =begin= and =end= markers.                */
/*                                                     */
/*                                                     */
/*******************************************************/

P		[\!\@\#\$\%\^\&\*\(\)\_\\\+\`\=\-\|\]\}\[\{\'\"\;\:\?\/\>\.\<\,]
D               [0-9]
L               [a-zA-Z ]

%{
#include <stdio.h>
#include <string.h>

#define PERSIAN 1
#define ENGLISH 2
#define TRUE	1
#define FALSE	0

#define ECHO outputString(yytext);

int ch;
int state        = ENGLISH;
int englishOutput= FALSE;
int htmlOutput   = FALSE;
int beginQuote   = TRUE;

void processISIRI( unsigned char ch );
void stringOutput( unsigned char * string );
void charOutput  ( unsigned char ch );

%}

%%

"=begin="	{ state = PERSIAN; }
"=end="		{ state = ENGLISH; }



{L}~		{ if (state == ENGLISH && englishOutput == TRUE ) stringOutput( yytext ); 
		  else if (state == PERSIAN) { 
			processISIRI(yytext[0]); 
			charOutput(0xa1);
		  }
	        }


{P}		{ if (state == ENGLISH && englishOutput == TRUE ) stringOutput( yytext ); 
		  else if ( state == PERSIAN ) processISIRI(yytext[0]); }

{D}		{ if (state == ENGLISH && englishOutput == TRUE) stringOutput( yytext ); 
		  else if ( state == PERSIAN ) processISIRI(yytext[0]); }

{L}		{ if (state == ENGLISH && englishOutput == TRUE) stringOutput( yytext ); 
		  else if ( state == PERSIAN ) processISIRI(yytext[0]); }
[\t\v\n\f]	{ if (state == PERSIAN || englishOutput== TRUE) stringOutput( yytext ); }

"\\\n"		{}
.		{}

%%

void stringOutput ( unsigned char * string ) {

 register int i;

 while ( *string != NULL ) {
	fprintf(yyout,"%c", *string ); 
	string++;
 }		
}

void charOutput ( unsigned char ch ) {

	fprintf(yyout,"%c", ch );

}

void processISIRI(unsigned char ch ) {

	switch (ch) {

	case 'a':
		charOutput(0xc1);
		break;
	case 'A':
		charOutput(0xc0);
		break;
	case 'b':
		charOutput(0xc3);
		break;
	case 'p':
		charOutput(0xc4);
		break;
	case 't':
		charOutput(0xc5);
		break;
	case 'c':
		charOutput(0xc6);
		break;
	case 'j':
		charOutput(0xc7);
		break;
	case 'G':
		charOutput(0xc8);
		break;
	case 'H':
		charOutput(0xc9);
		break;
	case 'K':
		charOutput(0xca);
		break;
	case 'd':
		charOutput(0xcb);
		break;
	case 'Z':
		charOutput(0xcc);
		break;
	case 'r':
		charOutput(0xcd);
		break;
	case 'z':
		charOutput(0xce);
		break;
	case 'J':
		charOutput(0xcf);
		break;
	case 's':
		charOutput(0xd0);
		break;
	case 'S':
		charOutput(0xd1);
		break;
	case 'C':
		charOutput(0xd2);
		break;
	case 'x':
		charOutput(0xd3);
		break;
	case 'T':
		charOutput(0xd4);
		break;
	case 'X':
		charOutput(0xd5);
		break;
	case 'e':
		charOutput(0xd6);
		break;
	case 'Q':
		charOutput(0xd7);
		break;
	case 'f':
		charOutput(0xd8);
		break;
	case 'q':
		charOutput(0xd9);
		break;
	case 'k':
		charOutput(0xda);
		break;
	case 'g':
		charOutput(0xdb);
		break;
	case 'l':
		charOutput(0xdc);
		break;
	case 'm':
		charOutput(0xdd);
		break;
	case 'n':
		charOutput(0xde);
		break;
	case 'v':
		charOutput(0xdf);
		break;
	case 'h':
		charOutput(0xe0);
		break;
	case 'y':
		charOutput(0xe1);
		break;
	case 'i':
		charOutput(0xc2);
		charOutput(0xa2);
		break;
	case '0':
		charOutput(0xb0);
		break;
	case '1':
		charOutput(0xb1);
		break;
	case '2':
		charOutput(0xb2);
		break;
	case '3':
		charOutput(0xb3);
		break;
	case '4':
		charOutput(0xb4);
		break;
	case '5':
		charOutput(0xb5);
		break;
	case '6':
		charOutput(0xb6);
		break;
	case '7':
		charOutput(0xb7);
		break;
	case '8':
		charOutput(0xb8);
		break;
	case '9':
		charOutput(0xb9);
		break;
	case ' ':
		charOutput(0xa0);
		break;
	case '!':
		charOutput(0xa3);
		break;
	case '"':
		if ( beginQuote == TRUE ) {
			beginQuote = FALSE;
			charOutput(0xe7);
		}
		else {
			beginQuote = TRUE;
			charOutput(0xe6);
		}	
		break;
	case '%':
		charOutput(0xa5);
		break;
	case ':':
		charOutput(0xba);
		break;
	case '?':
		charOutput(0xbf);
		break;
	case ')':
		charOutput(0xa9);
		break;
	case '(':
		charOutput(0xa8);
		break;
	case '.':
		charOutput(0xa6);
		break;
	case ',':
		charOutput(0xac);
		break;


	default:
		charOutput(ch);
	}
		
}

void printUsage()
{
	printf("\n");
	printf("*************************************************************\n");
	printf("Copyright(C) 1995 Simorgh Software. All rights reserved.     \n");
 	printf("gsz2isi:      gerdsooz to ISIRI-3342 converter               \n");
	printf("usage:        gsz2isi [-e] < input.gsz > output.isi          \n");
	printf("options:                                                     \n");
	printf("              -e allows English text to pass through filter  \n");
	printf("                                                             \n");
	printf("requirements: Gerdsooz input file have =begin= =end= markings\n");
	printf("              around encoding                                \n");
	printf("*************************************************************\n");
}

main(argc,argv)
int     argc;
char    **argv;

{
 int            result;
 int            errflag=0;
 extern char    *optarg;
 extern int     optind, opterr;

 while (( ch = getopt(argc, argv, "e")) != -1 ) {
     switch (ch) {

        case 'e' : 
             englishOutput = TRUE ;
             break;
	
        default:
             errflag++;
             break;
    }
 }

 if ( errflag ) {
	printUsage();
	exit(-1);
 }

 while ((result=yylex()) > 0 );

}