/*
Mycket enkel assembler för Am2910-chipet och lab 6 i kursen Digitalteknik

av Fredrik Roubert 1996
*/

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

#define	MAXSYMBOL	256
#define SYMBOLLEN	32

#define iswhitespace(q)	(q==' ' || q=='\t' || q=='\n')

typedef unsigned int	symbol;

const char	*mnemonics[16]={	"jz",
					"cjs",
					"jmap",
					"cjp",
					"push",
					"jsrp",
					"cjv",
					"jrp",
					"rfct",
					"rpct",
					"crtn",
					"cjpp",
					"ldct",
					"loop",
					"cont",
					"twb"	};

char		symbol_names[MAXSYMBOL+1][SYMBOLLEN+1];
symbol		symbol_values[MAXSYMBOL+1];
int		source_line,max_symbol;

char		print_symboltable,print_mnemonics;

FILE		*source;


/* Läser in nästa, av whitespace avgränsade, sträng. Returnerar strängens längd eller EOF */

int next_token(char *t)
{
	int	q,x;

	do
	{
		q=fgetc(source);

		if(q==';')
			while(q!=EOF && q!='\n')
				q=fgetc(source);

		if(q=='\n')
			source_line++;
	}
	while(q!=EOF && iswhitespace(q));

	t[0]=tolower(q);
	x=0;

	while(t[x]!=EOF && !iswhitespace(t[x]))
	{
		t[++x]=tolower(fgetc(source));

		if(t[x]==';')
			while(t[x]!=EOF && t[x]!='\n')
				t[x]=fgetc(source);

		if(t[x]=='\n')
			source_line++;
	}

	q=t[x];
	t[x]='\0';

	return (q==EOF?EOF:x);
}


int match_symbol(char *t)
{
	char	tt[SYMBOLLEN+1];
	int	x;

	strncpy(tt,t,SYMBOLLEN);
	tt[SYMBOLLEN]='\0';

	if(tt[strlen(tt)-1]==':')
		tt[strlen(tt)-1]='\0';

	for(x=0;x<max_symbol;x++)
		if(strcmp(tt,symbol_names[x])==0)
			return x;

/*
	fprintf(stderr,"Warning on line %u:\n\tUnable to relocate the symbol '%s'.\n\n",source_line,t);
*/
	return -1;
}


int match_token(char *t)
{
	int	x;

	for(x=0;x<16;x++)
		if(strcmp(t,mnemonics[x])==0)
			return x;

	return -1;
}


symbol parse_value(char *t)
{
	symbol	v;
	int	x,l,p;

	v=0;
	l=strlen(t)-1;

	switch(t[0])
	{
		case '$':	/* hex */
			for(x=l,p=1;x>0;x--,p*=16)
			{
				if(!((t[x]>='0' && t[x]<='9') || (t[x]>='a' && t[x]<='f')))
				{
					fprintf(stderr,"Parse error on line %u:\n\tInvalid character '%c' in hexadecimal constant.\n\n",source_line,t[x]);
					return 0;
				}

				v+=p*(  (t[x]>='0' && t[x]<='9')?(t[x]-'0'):(t[x]-'a'+10)  );
			}
			break;

		case '#':	/* dec */
			for(x=l,p=1;x>0;x--,p*=10)
			{
				if(t[x]<'0' || t[x]>'9')
				{
					fprintf(stderr,"Parse error on line %u:\n\tInvalid character '%c' in decimal constant.\n\n",source_line,t[x]);
					return 0;
				}

				v+=p*(t[x]-'0');
			}
			break;

		default:	/* bin */
			for(x=l,p=1;x>=0;x--,p*=2)
			{
				if(t[x]!='0' && t[x]!='1')
				{
					fprintf(stderr,"Parse error on line %u:\n\tInvalid character '%c' in binary constant.\n\n",source_line,t[x]);
					return 0;
				}

				v+=p*(t[x]-'0');
			}
	}

	return v;
}


void build_symboltable(void)
{
	char	buf[255],lastbuf[255];
	int	l,line;

	max_symbol=0;
	lastbuf[0]='\0';
	line=0;

	while(next_token(buf)!=EOF)
	{
		if(match_token(buf)!=-1)
			line++;

		l=strlen(buf)-1;

		if(buf[l]==':')
		{
			strncpy(symbol_names[max_symbol],buf,SYMBOLLEN);
			symbol_names[max_symbol][l>SYMBOLLEN?SYMBOLLEN:l]='\0';
			if(l>SYMBOLLEN)
				fprintf(stderr,"Warning on line %u:\n\tSymbol name truncated to '%s'.\n\n",source_line,symbol_names[max_symbol]);
			symbol_values[max_symbol]=line;
			if(max_symbol==MAXSYMBOL)
				fprintf(stderr,"Internal error on line %u:\n\tToo many symbols used.\n\n",source_line);
			else
				max_symbol++;
		}
		else
			if(strcmp(buf,"equ")==0)
			{
				if(strlen(lastbuf)==0)
				{
					fprintf(stderr,"Parse error on line %u:\n\tToken <equ> is not preceeded by a valid symbol name.\n\n",source_line);
					buf[0]='\0';
				}
				else
				{
					strncpy(symbol_names[max_symbol],lastbuf,SYMBOLLEN);
					if(strlen(lastbuf)>SYMBOLLEN)
					{
						symbol_names[max_symbol][SYMBOLLEN]='\0';
						fprintf(stderr,"Warning on line %u:\n\tSymbol name truncated to '%s'.\n\n",source_line,symbol_names[max_symbol]);
					}
					if(max_symbol==MAXSYMBOL)
						fprintf(stderr,"Internal error on line %u:\n\tToo many symbols used.\n\n",source_line);
					else
						max_symbol++;
				}
			}
			else
				if(strcmp(lastbuf,"equ")==0)
					symbol_values[max_symbol-1]=parse_value(buf);

		strcpy(lastbuf,buf);
	}
}

void print_binary(int line,char instr,symbol test,symbol data,symbol plus)
{
	int	x;

	printf("%02x: ",line);

	for(x=8;x;x/=2)
		putchar((instr & x)?'1':'0');

	putchar(' ');

	for(x=8;x;x/=2)
		putchar((test & x)?'1':'0');

	putchar(' ');

	for(x=16;x;x/=2)
		putchar((data & x)?'1':'0');

	putchar(' ');

	putchar((plus & 1024)?'1':'0');

	putchar(' ');

	for(x=512;x>4;x/=2)
		putchar((plus & x)?'1':'0');

	putchar(' ');

	for(x=4;x;x/=2)
		putchar((plus & x)?'1':'0');

	putchar('\n');
}

int assemble(void)
{
	char	buf[255],line,quit;
	int	s,instr;
	symbol	test,data,plus;


	if(print_symboltable)
	{
		for(s=0;s<max_symbol;s++)
			printf("[%02x][%s] : %x\n",s,symbol_names[s],symbol_values[s]);
		putchar('\n');
	}

	line=0;
	quit=0;
	while(!quit)
	{
		do
		{
			if((s=next_token(buf))==EOF)
				return 1;
			instr=match_token(buf);
/*
			fprintf(stderr,"--> (%i) \"%s\"\n",instr,buf);
*/
		}
		while(instr==-1);

		if((s=next_token(buf))==EOF)
		{
			fprintf(stderr,"Parse error on line %u:\n\tUnexpected end of file.\n\n",source_line);
			return -1;
		}

		s=match_symbol(buf);
		test=(s==-1?parse_value(buf):symbol_values[s]);

		if((s=next_token(buf))==EOF)
		{
			fprintf(stderr,"Parse error on line %u:\n\tUnexpected end of file.\n\n",source_line);
			return -1;
		}

		s=match_symbol(buf);
		data=(s==-1?parse_value(buf):symbol_values[s]);

		if((s=next_token(buf))==EOF)
		{
			fprintf(stderr,"Parse error on line %u:\n\tUnexpected end of file.\n\n",source_line);
			return -1;
		}

		s=match_symbol(buf);
		plus=(s==-1?parse_value(buf):symbol_values[s]);

		if(print_mnemonics)
			printf("%02x\t%s\t%x\t%x\t%x\n",line,mnemonics[instr],test,data,plus);
		else
			print_binary(line,instr,test,data,plus);

		line++;
	}

	return 1;

}


int main(int argc,char *argv[])
{
	int	status,n;
	char	*filename;

	if(argc<2)
	{
		fputs("You must specify a file name!\n\"am2910 -help\" for help.\n",stderr);
		return -1;
	}

	print_symboltable=0;
	print_mnemonics=0;
	filename=NULL;

	for(n=1;n<argc;n++)
	{
		if(strcmp("-h",argv[n])==0 || strcmp("-help",argv[n])==0 || strcmp("--help",argv[n])==0)
		{
			puts("Am2910, a very simple symbolic assembler for the am2910 chipset.\n\251 Fredrik Roubert 1996 <roubert@df.lth.se>\n\nUsage: am2910 [OPTION] [FILE]\n\nThe following options are supported:\n\n\t-h  Display this help text\n\t-s  Print out the symbol table\n\t-m  Show mnemonics instead of binary digits\n");
			return 1;
		} 
		else
			if(strcmp("-s",argv[n])==0)
				print_symboltable=1;
			else
				if(strcmp("-m",argv[n])==0)
					print_mnemonics=1;
				else
					filename=argv[n];
	}

	if((source=fopen(filename,"r"))==NULL)
	{
		fprintf(stderr,"Unable to open file '%s'.\n",filename);
		return -1;
	}
	source_line=1;

	build_symboltable();
	rewind(source);
	source_line=1;
	status=assemble();

	fclose(source);
	return status;
}