/*
汇编优化工具
使汇编代码更具可读性，并提高汇编速度
优化15000行大约需要30秒
*/
import rfile.h
main
{
name=getsl
s=rfile.read_all_n(name)
s=s.replace('\n','\r\n')
s=s.replace('dword esp','esp')
s=s.replace('dword ebp','ebp')
s=s.replace('dword edi','edi')
s=s.replace('dword esi','esi')
s=s.replace('dword ebx','ebx')
s=s.replace(',ecx',', ecx')
s=s.replace('ebx,0','ebx , 0')
s=s.replace('ecx,[','ecx , [')
s=s.replace('mov dword [ edi ] , esi','mov [ edi ] , esi')
s=opt_sym(s)
s=line_del(s)
s=opt_jmp(s)
s=line_del(s)
s=opt_two(s)
s=line_del(s)
s=sym_replace(s)
s=line_del(s)
rfile.write_all_n(name,s)
}
rstr line_del(rstr& s)
{
return s.replace('\r\n\r\n\r\n','\r\n\r\n')
}
//符号替换
rstr sym_replace(rstr s)
{
rbuf<rstr> arr
r_split_e<rstr>(s,'\r\n',arr)
rstr ret
for i in arr
temp=arr[i]
if temp.get_top==`:&&is_symbol(temp.sub(0,temp.count-1))
ret+=sym_trans(temp.sub(0,temp.count-1))+':'
ret+='\r\n'
continue
if (temp.sub(0,5)==' jmp '||
temp.sub(0,6)==' call '||
temp.sub(0,5)==' jnz '||
temp.sub(0,4)==' jg '||
temp.sub(0,4)==' jl '||
temp.sub(0,5)==' jae '||
temp.sub(0,4)==' jz ')
index=temp.find(' ')
symbol=temp.sub(index+1)
if is_symbol(symbol)
ret+=temp.sub(0,index+1)+sym_trans(symbol)
ret+='\r\n'
continue
ret+=temp
ret+='\r\n'
return ret
}
rstr sym_trans(rstr s)
{
rstr ret
for i in s
two=s.sub(i,i+2)
if two=='28'||two=='29'//()
i++
ret+='_'
elif two=='2E'||two=='2C'||two=='5F'//.,_
i++
ret+='_'
elif two=='26'
i++
ret+='q'
elif two=='7E'
i++
ret+='d'
elif two=='7C'
i++
ret+='h'
elif two=='5B'
i++
ret+='i'
elif two=='5D'
i++
ret+='j'
elif two=='3D'//=
i++
ret+='a'
elif two=='2B'
i++
ret+='b'
elif two=='2D'//-
i++
ret+='s'
elif two=='2A'//*
i++
ret+='t'
elif two=='2F'
i++
ret+='f'
elif two=='25'
i++
ret+='e'
elif two=='21'//!
i++
ret+='o'
elif two=='3E'//>
i++
ret+='x'
elif two=='3C'//<
i++
ret+='y'
else
ret+=s[i]
return seq_replace(ret)
}
rstr seq_replace(rstr s)
{
//s=s.replace('__','_')
if s.get_top==`_
s.pop
if s.get_top==`_
s.pop
return s
}
//无引用符号删除
rstr opt_sym(rstr s)
{
rbuf<rstr> arr
r_split_e<rstr>(s,'\r\n',arr)
rstr ret
for i in arr
temp=arr[i]
full=temp.sub(0,temp.count-1)
if temp.get_top==`:&&is_symbol(full)
symbol=get_symbol(full)
ifn symbol.empty
if get_count(arr,i,symbol,full)==1
continue
ret+=temp
ret+='\r\n'
return ret
}
//jmp后接一个跳转符号
rstr opt_jmp(rstr s)
{
rbuf<rstr> arr
r_split_e<rstr>(s,'\r\n',arr)
rstr ret
for i in arr
temp=arr[i]
a=arr.get(i+1)
full=a.sub(0,a.count-1)
if temp.sub(0,5)==' jmp '&&temp.sub(5)==full&&a.get_top==`:
symbol=get_symbol(full)
ifn symbol.empty
if get_count(arr,i,symbol,full)==2
i++
continue
else
continue
ret+=temp
ret+='\r\n'
return ret
}
//连续2个jmp，后一个jmp不可能被执行
rstr opt_two(rstr s)
{
rbuf<rstr> arr
r_split_e<rstr>(s,'\r\n',arr)
rstr ret
for i in arr
temp=arr[i]
a=arr.get(i+1)
if temp.sub(0,5)==' jmp '&&a.sub(0,5)==' jmp '
ret+=temp
ret+='\r\n'
i++
continue
ret+=temp
ret+='\r\n'
return ret
}
int get_count(rbuf<rstr>& arr,int i,rstr& symbol,rstr& full)
{
for j=i-1;j>=0;j--
if arr[j].sub(0,arr[j].count-1)==symbol&&`:==arr[j].get_top
break
for k=i+1;k<arr.count;k++
if arr[k].empty
break
return find_count(arr,j,k,full)
}
rstr get_symbol(rstr s)
{
for i=s.count-1;i>=0;i--
if s[i]==`_&&s.sub(i+1).is_number
return s.sub(0,i)
return ''
}
int find_count(rbuf<rstr>& arr,int left,int right,rstr& symbol)
{
count=0
for i=left;i<=right;i++
count+=find_count(arr[i],symbol)
return count
}
int find_count(rstr src,rstr dst)
{
count=0
for i in src
if src.sub(i,i+dst.count)==dst
count++
return count
}
bool is_symbol(rstr s)
{
for i in s
ifn is_symbol(s[i])
return false
return true
}
bool is_symbol(char ch)
{
return ch>=`A&&ch<=`Z||ch>=`a&&ch<=`z||ch>=`0&&ch<=`9||ch==`_
}
原文地址：http://blog.csdn.net/roundsheep/article/details/28249085
