optimize

#!/usr/bin/env python3
import sys
import re

def isImmed(var):
    if var[0] == '$':
        return True
    return False

def toOpr(var):
    if var == 'sub':
        return -1
    if var == 'add':
        return 1
    print('Unexpected value', var)
    exit(1)

fname = sys.argv[1]
c = [k for k in open(fname).read().split('\n') if  len(k.strip())>0 and k.strip()[0] != '#']
print("New code size: ", len(c))

# Optimization pass 1; JUMP over JUMP
while True:
    l = len(c)
    flag = False
    for i in range(0, l-1):
        s1, s2 = c[i], c[i+1]
        if s1.strip()[-1] == ':' and s2.strip().split()[0].lower() == 'jmp':
            l1name = s1.strip()[:-1]
            l2name = s2.strip().split()[1]
            flag = True
            break

    if flag:
        c = [k.replace(l1name, l2name) for k in c if k.strip().replace(' ', '') != l1name + ':']
    else:
        break

print("New code size: ", len(c))

# Optimization pass 2; WRITE before WRITE
while True:
    writeOp = ['mov', 'lea']
    l = len(c)
    flag = False
    for i in range(0, l-1):
        s1 = [k.strip() for k in re.findall(r"((?:-?\d+)?\(.*?\)|[^ ,]+)", c[i].strip().replace('\t', ' '))]
        s2 = [k.strip() for k in re.findall(r"((?:-?\d+)?\(.*?\)|[^ ,]+)", c[i+1].strip().replace('\t', ' '))]
        if s1[0] not in writeOp or s2[0] not in writeOp:
            continue
        dst1 = s1[2]
        dst2 = s2[2]
        if dst1 == dst2:
            if s2[2] not in s2[1]:
                c[i] = ''
                flag = True
                break
    if flag:
        c = [k for k in c if len(k.strip()) > 0]
    else:
        break

# Optimization pass 3; CONSTANT FOLDING
while True:
    arithOp = ['add', 'sub']
    l = len(c)
    flag = False
    for i in range(0, l-1):
        s1 = [k.strip() for k in re.findall(r"((?:-?\d+)?\(.*?\)|[^ ,]+)", c[i].strip().replace('\t', ' '))]
        s2 = [k.strip() for k in re.findall(r"((?:-?\d+)?\(.*?\)|[^ ,]+)", c[i+1].strip().replace('\t', ' '))]
        if s1[0] not in arithOp or s2[0] not in arithOp:
            continue
        if not isImmed(s1[1]) or not isImmed(s2[1]):
            continue
        if not s1[2] == s2[2]:
            continue
        val1 = toOpr(s1[0]) * int(s1[1][1:])
        val2 = toOpr(s2[0]) * int(s2[1][1:])
        c[i+1] = ''
        c[i] = '\tadd ${}, {}'.format(val1+val2, s1[2])
        flag = True
        break
    if flag:
        c = [k for k in c if len(k.strip()) > 0]
    else:
        break

print("New code size: ", len(c))

open(fname, 'w').write('\n'.join(c) + "\n")