# asm - assembler for hack computer
#   usage: awk -f asm.awk assembly-program-file.asm

BEGIN {
	src_file = ARGV[1]
	tmp_file = "asm.tmp"
	out_file = substr(src_file, 1, index(src_file, ".")) "hack"
	split("0 M D DM A AM AD ADM", x)
	split("0 JGT JEQ JGE JLT JNE JLE JMP", y)
	split("000 001 010 011 100 101 110 111", b)
	for (i = 1; i <= 8; i++) {
		destc[x[i]] = b[i]
		jumpc[y[i]] = b[i]
	}
	destc["MD"] = "011"
	destc["MA"] = "101"
	destc["DA"] = "110"
	destc["DAM"] = "111"
	destc["MAD"] = "111"
	destc["MDA"] = "111"
	destc["DMA"] = "111"
	destc["AMD"] = "111"
	n = split("0 1 -1 D A !D !A -D -A D+1 A+1 D-1 A-1 D+A D-A A-D D&A D|A", z)
	split("0101010 0111111 0111010 0001100 0110000 0001101 0110001 0001111 0110011 0011111 0110111 0001110 0110010 0000010 0010011 0000111 0000000 0010101", w)
	for (i = 1; i <= n; i++)
		opc[z[i]] = w[i]
	opc["M"]   = "1110000"
	opc["!M"]  = "1110001"
	opc["-M"]  = "1110011"
	opc["M+1"] = "1110111"
	opc["M-1"] = "1110010"
	opc["D+M"] = "1000010"
	opc["D-M"] = "1010011"
	opc["M-D"] = "1000111"
	opc["D&M"] = "1000000"
	opc["D|M"] = "1010101"
	symtab[R0] = 0
	symtab[R1] = 1
	symtab[R2] = 2
	symtab[R3] = 3
	symtab[R4] = 4
	symtab[R5] = 5
	symtab[R6] = 6
	symtab[R7] = 7
	symtab[R8] = 8
	symtab[R9] = 9
	symtab[R10] = 10
	symtab[R11] = 11
	symtab[R12] = 12
	symtab[R13] = 13
	symtab[R14] = 14
	symtab[R15] = 15
	symtab[SP] = 0
	symtab[LCL] = 1
	symtab[ARG] = 2
	symtab[THIS] = 3
	symtab[THAT] = 4
	symtab[SCREEN] = 16384
	symtab[KBD] = 24576

# ASSEMBLER PASS 1
	while (getline <src_file > 0) {
		if (/^[[:space:]]*(\/\/.*)*$/)
			continue
		else if (/\(.*\)/) {
			symtab[substr($0, 2, length($0)-2)] = nl-ns+1
			ns++; nl++
		} else {
			# print sub(/[[:space:]]*/, "")
			print $0 >tmp_file
			nl++
		}
	}
	close(tmp_file)

# ASSEMBLER PASS 2
	nextram = 16
	while (getline <tmp_file > 0) {
		if (/^@/) {            # A-instruction
			con = substr($0, 2)
			if (con !~ /^[:digit:]+$/)
				if (con in symtab) {
					con = symtab[con]
				} else {
					symtab[con] = nextram
					con = nextram
					nextram++
				}
			binstring = ""
			for (i = 0; i < 15; i++)
				if (and(con, lshift(1,i))) {
					binstring = "1" binstring
				} else {
					binstring = "0" binstring
				}
			print "0" binstring > out_file
		} else {               # C-instruction
			m = index($0, "=")
			n = index($0, ";")
			dest = m ? substr($0, 1, m-1) : 0
			comm = n ? substr($0, m+1, n-m-1) : substr($0, m+1)
			jump = n ? substr($0, n+1) : 0
			print comm
			printf("comm %s; jump %s\n", comm, jump)
			printf("111%s%s%s\n", opc[op], destc[dest], jumpc[jump]) > out_file
		}
	}
}