nand2tetris/asm/asm02.py

'''
1. Kreieren der Symboltabelle
2. Einlesen der Datei
    a. Entfernen der Kommentare -- erledigt
    b. Suchen der (Label) und Ablegen dieser in die Symboltabelle -- erledigt
    c. Entfernen der (Label) aus dem ASM-File
    d. HACK-Code erzeugen
'''

import sys, os
from pathlib import Path
import re
from icecream import ic

symboltable = {}
asm = []

def appendTable(table, id, item):
    print("appendTable ",id, " ",  item )
    
    table["@"+ str(item)] = id
    symboltable = table


def createSymboltable():
    return {
	key: (value)
	for key, value in {
	    **{'@SP': 0,
	       '@LCL': 1,
	       '@ARG': 2,
	       '@THIS': 3,
	       '@THAT': 4,
	       '@SCREEN': 0x4000,
	       '@KBD': 0x6000,},
	     **{f'@R{i}': i
		for i in range(16)}
    }.items()}

def load_asm_file(filename):
    fp = open(filename, 'r')
    comment = r"^/"
    eol = r"^\n"
    tmp = []
    t = fp.readlines()
    for l in t:
        if re.match(comment, l):
            pass
        elif re.match(eol, l):
            pass
        else:
            tmp.append(re.sub('\s', '', l).split('//')[0])  # Wenn noch ein Kommentar hinter der Anweisung steht dann Abtrenen
    fp.close()
     #ic("tmp = ", tmp)
    return tmp

def label_to_symboltable(data):
    #ic(len(data))
    i = 0
    tmp = {}
    for l in data:
        if (l.startswith('(') and l.endswith(')')):
            l = l[:-1]
            l = l.lstrip('(')
            tmp['@'+str(l)] = i
            #ic(l)
        else:
            i=i+1
    
    if len(tmp) != 0:
        symboltable.update(tmp)
        #ic(symboltable)
        #ic (data)
       
        # c. Löschen der (Label) aus der Liste
        count = 0
        for i in tmp:
            #ic(tmp[i]) # Zeilennummern
            data.pop(tmp[i]-count)
            
        #ic(len(data))
        searchSymbols(data)
            


def searchSymbols(asm):
    '''
    wenn Zeile mit @ beginnt schauen ob Variable in Symboltable
    sonst
    ist Variable @x ein int, dann ohne @ in die Symboltable übernehmen
    '''
    #ic(asm)
    i = 16
    for line in asm:
        if line.startswith('@'):
            if line[1:].isdigit():
                    if line not in symboltable:
                        addtoSymboltable(line, int(line[1:]))
            elif line not in symboltable:
                addtoSymboltable(line, i)
                i = i+1
                    
def ciinstruction(ci):
    c = "111"
    if ci.__contains__('='):
        c2 = ci.split('=')
        comp = c2[1]
        dest = c2[0]
        c = c + c_comp(comp) + c_dest(dest) +"000"
    elif ci.__contains__(";"):
        jmp = ci.split(';')
        comp = jmp[0]
        jmpinst = jmp[1]
        c = c + c_comp(comp) + "000" + c_jump(jmpinst)
        
        
    return c +"\n"
        
def c_jump(jmp):
    match jmp:
        case "JGT":
            return "001"
        case "JEQ":
            return "010"
        case "JGE":
            return "011"
        case "JLT":
            return "100"
        case "JNE":
            return "101"
        case "JLE":
            return "110"
        case "JMP":
            return "111"
        case _:
            return "000"
        
    

    
def c_comp(comp):
    match comp: 
        # A = 0 
        case "D":
            return "0001100"
        case "0":
            return "0001100"
        case "1":
            return "0111111"
        case "-1":
            return "0111010"
        case "A":
            return "0110000"
        case "!D":
            return "0001101"
        case "!A":
            return "0110001"
        case "-D":
            return "0001111"
        case "-A":
            return "0110011"
        case "D+1":
            return "0011111"
        case "A+1":
            return "0110111"
        case "D-1":
            return "0001110"
        case "A-1":
            return "0110010"
        case "D+A":
            return "0000010"
        case "D-A":
            return "0010011"
        case "A-D":
            return "0000111"
        case "D&A":
            return "0000000"
        case "D|A":
            return "0010101"
        # a = 1
        case "M":
            return "1110000"
        case "!M":
            return "1110001"
        case "-M":
            return "1110011"
        case "M+1":
            return "1110111"
        case "M-1":
            return "1110010"
        case "D+M":
            return "1000010"
        case "D-M":
            return "1010011"
        case "M-D":
            return "1000111"
        case "D&M":
            return "1000000"
        case "D|M":
            return "1010101"
        case _:
            return ""

def c_dest(dest):
    match dest:
        case "M":
            return "001"
        case "D":
            return "010"
        case "MD":
            return "011"
        case "A":
            return "100"
        case "AM":
            return "101"
        case "AD":
            return "110"
        case "AMD":
            return "111"
        case _:
            return "000"
    
  
        
    #return '{0:016b}'.format(c)+"\n"
    #return  bin(c)[2:]+"\n" # 111 in binär --> falscher Weg
    # dest = comp; jump
    
                    
            
def addtoSymboltable(key,value):
    symboltable[key]=value
        

def createAsmFile():
    datename = os.path.splitext(os.path.basename(filename))[0]+".hack"
    f = open(datename,"w")
    for line in asm:
        if line in symboltable:
            f.write('{0:016b}'.format(symboltable[line]) + "\n")
        else:
            f.write(ciinstruction(line))
    f.close()


if __name__ == "__main__":
    filename = sys.argv[1]
    #filename = "Pong.asm"
    symboltable = createSymboltable()
    asm = load_asm_file(filename)
    label_to_symboltable(asm)
    searchSymbols(asm)
    createAsmFile()