print("MMC assembler")
import sys
args=sys.argv
if(len(args)==1):
    args.append(input("input filename:"))
file=open(args[1])
filedat=file.read()
file.close()

split=False
tmp=""
tokens=[]
for i in filedat:
    if i=='"':
        split=not split
        if(tmp!=""):
            tokens.append(tmp)
        tmp=""
        if split:
            tokens.append('"')
    elif (i.isspace() or i==',') and not split:
        if(tmp!=""):
            tokens.append(tmp)
        tmp=""
    else:
        tmp+=i
tokens.append(tmp)
oldtokens=tokens
names={}
location=0
progstart=0
i=0
registers=["pc","ax","bx","cx","dx","sp","si","di"]
reg8=["al","ah","bl","bh","cl","ch","dl","dh"]
while i<len(tokens):
    if tokens[i].endswith(':'):
        names[tokens[i].replace(':','')]=location
        tokens.remove(tokens[i])
        i-=1
    else:
        tokens[i]=tokens[i].lower()
        if tokens[i]=="org":
            i=i+1
            location=int(tokens[i])
            progstart=location
        elif tokens[i]=='"':
            i+=1
            location+=len(tokens[i])+1
            """elif tokens[i].startswith('%'):
                if(tokens[i].find("%bits")>=0):
                    codesize=int(tokens[i].replace("%bits",''))"""
        elif tokens[i]=="jmp":
            i=i+1
            if(tokens[i] in registers):
                location+=1
            else:
                location+=3
        elif tokens[i]=="jpz":
            i=i+1
            if(tokens[i] in reg8):
                location+=3
            else:
                print("'"+tokens[i]+"' is not a valid 8-bit register for JPZ opcode.")
            i+=1
        elif tokens[i]=="sub":
            i=i+2
            if(tokens[i] in reg8)or tokens[i] in registers:
                location+=2
            else:
                location+=3
        elif tokens[i]=="call":
            location+=3
            i+=1
        elif tokens[i]=='hlt':
            location+=1
        elif tokens[i]=='ldr':
            location+=2
            i+=2
        elif tokens[i]=='ldw':
            location+=2
            i+=2
        elif tokens[i]=='kbd':
            i+=1
            location+=1
        elif tokens[i]=='wrt':
            i+=1
            location+=1
        elif tokens[i]=='ret':
            location+=1
        elif tokens[i]=='rtz':
            location+=1
        elif tokens[i]=='lod':
            location+=1
        elif tokens[i]=='sto':
            location+=1
        elif tokens[i]=='ld':
            i+=2
            location+=3
        else:
            print("error: invalid:"+tokens[i])
            tokens.remove(tokens[i])
            i-=1
    i+=1
i=0
location=0
out=[]
while i<len(tokens):
    if tokens[i]=="org":
        i=i+1
        location=int(tokens[i])
    elif tokens[i]=='"':
        i+=1
        location+=len(tokens[i])+1
        tmp=list(tokens[i])
        for x in tmp:
            out.append(ord(x))
        out.append(0)
    elif tokens[i]=="jmp":
        i=i+1
        if(tokens[i] in registers):
            location+=1
            out.append(0xC8+registers.index(tokens[i]))
        elif(tokens[i]=='$'):
            out.append(0xC3)
            out.append(location>>8)
            out.append(location&255)
            location+=3
        elif(tokens[i] in names.keys()):
            out.append(0xC3)
            out.append(names[tokens[i]]>>8)
            out.append(names[tokens[i]]&255)
            location+=3
        else:
            out.append(0xC3)
            out.append(int(tokens[i])>>8)
            out.append(int(tokens[i])&255)
            location+=3
    elif tokens[i]=="jpz":
        i=i+1
        if(tokens[i] in reg8):
            out.append(0xD8+reg8.index(tokens[i]))
            i+=1
            if(tokens[i] in registers):
                print("'jpz' does not support registers as addresses.")
            elif(tokens[i]=='$'):
                out.append(location>>8)
                out.append(location&255)
            elif(tokens[i] in names.keys()):
                out.append(names[tokens[i]]>>8)
                out.append(names[tokens[i]]&255)
            else:
                out.append(int(tokens[i])>>8)
                out.append(int(tokens[i])&255)
            location+=3
    elif tokens[i]=="call":
        i=i+1
        if(tokens[i] in registers):
            print("'call' does not support registers. use 'push "+tokens[i]+"' and 'ret'")
        elif(tokens[i]=='$'):
            out.append(0xC4)
            out.append(location>>8)
            out.append(location&255)
            location+=3
        elif(tokens[i] in names.keys()):
            out.append(0xC4)
            out.append(names[tokens[i]]>>8)
            out.append(names[tokens[i]]&255)
            location+=3
        else:
            out.append(0xC4)
            out.append(int(tokens[i])>>8)
            out.append(int(tokens[i])&255)
            location+=3
    elif tokens[i]=="sub":
        i=i+1
        if(tokens[i] in reg8):
            i+=1
            if(tokens[i] in reg8):
                out.append(0x4E)
                out.append((reg8.index(tokens[i-1])<<3)|reg8.index(tokens[i]))
                location-=1
            elif(tokens[i]=='$'):
                print("'$' used for 16-bit value in an 8-bit instruction.")
            elif(tokens[i] in names.keys()):
                out.append(2)
                out.append(reg8.index(tokens[i-1]))
                out.append(names[tokens[i]]&255)
            else:
                out.append(2)
                out.append(reg8.index(tokens[i-1]))
                out.append(int(tokens[i])&255)
            location+=3
        elif(tokens[i] in registers):
            i+=1
            if(tokens[i] in registers):
                out.append(0x4F)
                out.append((registers.index(tokens[i-1])<<3)|registers.index(tokens[i]))
                location-=1
            elif(tokens[i]=='$'):
                out.append(0x40+registers.index(tokens[i]))
                out.append(location>>8)
                out.append(location&255)
            elif(tokens[i] in names.keys()):
                out.append(0x40+registers.index(tokens[i]))
                out.append(names[tokens[i]]>>8)
                out.append(names[tokens[i]]&255)
            else:
                out.append(0x40+registers.index(tokens[i-1]))
                out.append(int(tokens[i])>>8)
                out.append(int(tokens[i])&255)
    elif tokens[i]=="ld":
        i=i+1
        if(tokens[i] in reg8):
            reg=reg8.index(tokens[i])
            i+=1
            if(tokens[i]=='$'):
                print("'$' used for 16-bit value in an 8-bit instruction.")
                location+=3
            elif(tokens[i] in names.keys()):
                out.append(0x06)
                out.append(reg)
                out.append(names[tokens[i]]&255)
                location+=3
            elif(tokens[i] in reg8):
                out.append(0x04)
                out.append(reg8.index(tokens[i])+8*reg8.index(tokens[i-1]))
                location+=2
            else:
                out.append(0x06)
                out.append(reg)
                out.append(int(tokens[i])&255)
                location+=3
        elif(tokens[i] in registers):
            reg=registers.index(tokens[i])
            i+=1
            out.append(0x08+reg)
            if(tokens[i]=='$'):
                out.append(int(location)>>8)
                out.append(int(location)&255)
                location+=3
            elif(tokens[i] in names.keys()):
                out.append(names[tokens[i]]>>8)
                out.append(names[tokens[i]]&255)
                location+=3
            elif(tokens[i] in registers):
                out.append(16)
                out.append(registers.index(tokens[i])+8*registers.index(tokens[i-1]))
                location+=2
            else:
                out.append(int(tokens[i])>>8)
                out.append(int(tokens[i])&255)
                location+=3
    elif tokens[i]=='hlt':
        location+=1
        out.append(0x07)
    elif tokens[i]=='kbd':
        location+=1
        i+=1
        out.append(reg8.index(tokens[i])+0xF0)
    elif tokens[i]=='wrt':
        location+=1
        i+=1
        out.append(reg8.index(tokens[i])+0x80)
    elif tokens[i]=='ldw':
        location+=2
        i+=2
        out.append(0x1F)
        out.append(reg8.index(tokens[i-1])+8*registers.index(tokens[i]))
    elif tokens[i]=='ldr':
        location+=2
        i+=2
        out.append(0x2F)
        out.append(reg8.index(tokens[i-1])+8*registers.index(tokens[i]))
    elif tokens[i]=='ret':
        location+=1
        out.append(0x03)
    elif tokens[i]=='rtz':
        location+=1
        out.append(0x19)
    elif tokens[i]=='lod':
        location+=1
        out.append(0x2B)
    elif tokens[i]=='sto':
        location+=1
        out.append(0x2A)
    else:
        print("error: invalid:"+tokens[i])
    i+=1
stringout="#define BOOTLEN "+str(len(out))+"\nchar boot[BOOTLEN]={"
for i in range(len(out)):
    stringout+=hex(out[i])+','
stringout+='};\n'
print(stringout)