[From Nand to Tetris] 第8章 虚拟机项目 python 实现

[From Nand to Tetris] 第8章 虚拟机项目 python 实现

为防闲逛至此的看官不知所云： From Nand to Tetris 是一个在线课程，目标是指导学生从 Nand 逻辑门开始从头到尾完成一整套计算机系统。

好像撸串一样的爽快：------芯片--硬件---编译原理---操作系统---应用程序------>

这里提供的是第八章的作业，以供半路摔进坑里的同学们扶一下。。。

人家老师确实是不希望扩散答案，不过我做的过程中遇到很多坑，搞半天后发现全是些脑残原因，实在是浪费时间，希望卡壳的同学们能以增进效率为目的适当参考答案。毕竟学习这种东西，有没有学到手只有自己知道。。。

注释不多，因为代码相当 self-explanatory ，就是书上那些，没自由发挥什么。

如果你是闲逛进来，而且对这块内容有兴趣的话，强烈建议点开上面的课程链接试试，我是真心非常喜欢这门课，请收下我的安利。。。

另外还有第六章的作业答案：第6章 汇编器项目 python 实现

# _*_ coding: utf-8 _*_

import sys
import os
import glob


class C_TYPE:
    '''Command Type'''
    C_ARITHMETIC, C_PUSH, C_POP, C_LABEL, C_GOTO, C_IF, C_FUNCTION, C_RETURN, C_CALL = range(9)


class Parser:

    def __init__(self, fname):
        self.finput = open(fname, 'rU')
        self.current = None
        self.commandPart = []
        self.cType = -1

    def __exit__(self, exc_type, exc_val, exc_tb):
        self.finput.close()

    def has_more_commands(self):

        self.current = self.finput.readline()

        while self.current == '\n' or self.current[:2] == '//':
            self.current = self.finput.readline()

        return self.current != ''

    def advance(self):
        self.commandPart = self.current.strip().split(' ')[:3]

        if self.commandPart[0] == 'add' or\
                self.commandPart[0] == 'sub' or\
                self.commandPart[0] == 'neg' or\
                self.commandPart[0] == 'eq' or\
                self.commandPart[0] == 'gt' or\
                self.commandPart[0] == 'lt' or\
                self.commandPart[0] == 'and' or\
                self.commandPart[0] == 'or' or\
                self.commandPart[0] == 'not':
            self.cType = C_TYPE.C_ARITHMETIC
        elif self.commandPart[0] == 'push':
            self.cType = C_TYPE.C_PUSH
        elif self.commandPart[0] == 'pop':
            self.cType = C_TYPE.C_POP
        elif self.commandPart[0] == 'label':
            self.cType = C_TYPE.C_LABEL
        elif self.commandPart[0] == 'goto':
            self.cType = C_TYPE.C_GOTO
        elif self.commandPart[0] == 'if-goto':
            self.cType = C_TYPE.C_IF
        elif self.commandPart[0] == 'function':
            self.cType = C_TYPE.C_FUNCTION
        elif self.commandPart[0] == 'call':
            self.cType = C_TYPE.C_CALL
        elif self.commandPart[0] == 'return':
            self.cType = C_TYPE.C_RETURN

    def command_type(self):
        return self.cType

    def arg1(self):
        if self.command_type() == C_TYPE.C_ARITHMETIC:
            return self.commandPart[0]
        return self.commandPart[1]

    def arg2(self):
        return self.commandPart[2]


class CodeWriter:

    def __init__(self, fname):
        self.foutput = open(fname, 'w')
        self.uniqueFlag = 0  # 用于构造唯一的标签，每次使用后加1
        self.currentFile = '' # 当前处理的文件的名字
        self.currentFunc = '' # 当前处理的函数的名字

    def set_file_name(self, fname):
        self.currentFile = fname

    def write_arithmetic(self, command):
        if command == 'add':
            self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@SP\nM=M-1\nA=M\nM=M+D\n@SP\nM=M+1\n')
        elif command == 'sub':
            self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@SP\nM=M-1\nA=M\nM=M-D\n@SP\nM=M+1\n')
        elif command == 'neg':
            self.foutput.write('@SP\nM=M-1\nA=M\nM=-M\n@SP\nM=M+1\n')
        elif command == 'eq':
            self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@SP\nM=M-1\nA=M\nD=M-D\n@RET_TRUE'+str(self.uniqueFlag)+'\n'
                               'D;JEQ\nD=0\n@END'+str(self.uniqueFlag)+'\n0;JMP\n(RET_TRUE'+str(self.uniqueFlag)+')\n'
                               'D=-1\n(END'+str(self.uniqueFlag)+')\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
            self.uniqueFlag += 1
        elif command == 'gt':
            self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@SP\nM=M-1\nA=M\nD=M-D\n@RET_TRUE'+str(self.uniqueFlag)+'\n'
                               'D;JGT\nD=0\n@END'+str(self.uniqueFlag)+'\n0;JMP\n(RET_TRUE'+str(self.uniqueFlag)+')\n'
                               'D=-1\n(END'+str(self.uniqueFlag)+')\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
            self.uniqueFlag += 1
        elif command == 'lt':
            self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@SP\nM=M-1\nA=M\nD=M-D\n@RET_TRUE'+str(self.uniqueFlag)+'\n'
                               'D;JLT\nD=0\n@END'+str(self.uniqueFlag)+'\n0;JMP\n(RET_TRUE'+str(self.uniqueFlag)+')\n'
                               'D=-1\n(END'+str(self.uniqueFlag)+')\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
            self.uniqueFlag += 1
        elif command == 'and':
            self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@SP\nM=M-1\nA=M\nM=M&D\n@SP\nM=M+1\n')
        elif command == 'or':
            self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@SP\nM=M-1\nA=M\nM=M|D\n@SP\nM=M+1\n')
        elif command == 'not':
            self.foutput.write('@SP\nM=M-1\nA=M\nM=!M\n@SP\nM=M+1\n')

    def write_push_pop(self, command, segment, index):
        if command == C_TYPE.C_PUSH:
            if segment == 'constant':
                self.foutput.write('@'+index+'\nD=A\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
            elif segment == 'local':
                self.foutput.write('@LCL\nD=M\n@'+index+'\nA=A+D\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
            elif segment == 'argument':
                self.foutput.write('@ARG\nD=M\n@'+index+'\nA=A+D\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
            elif segment == 'this':
                self.foutput.write('@THIS\nD=M\n@'+index+'\nA=A+D\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
            elif segment == 'that':
                self.foutput.write('@THAT\nD=M\n@'+index+'\nA=A+D\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
            elif segment == 'pointer':
                self.foutput.write('@'+str(int(index)+3)+'\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
            elif segment == 'temp':
                self.foutput.write('@'+str(int(index)+5)+'\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
            elif segment == 'static':
                self.foutput.write('@'+self.currentFile+'.'+index+'\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n')
        elif command == C_TYPE.C_POP:
            if segment == 'local':
                self.foutput.write('@LCL\nD=M\n@'+index+'\nD=A+D\n@R13\nM=D\n@SP\nM=M-1\nA=M\nD=M\n@R13\nA=M\nM=D\n')
            elif segment == 'argument':
                self.foutput.write('@ARG\nD=M\n@'+index+'\nD=A+D\n@R13\nM=D\n@SP\nM=M-1\nA=M\nD=M\n@R13\nA=M\nM=D\n')
            elif segment == 'this':
                self.foutput.write('@THIS\nD=M\n@'+index+'\nD=A+D\n@R13\nM=D\n@SP\nM=M-1\nA=M\nD=M\n@R13\nA=M\nM=D\n')
            elif segment == 'that':
                self.foutput.write('@THAT\nD=M\n@'+index+'\nD=A+D\n@R13\nM=D\n@SP\nM=M-1\nA=M\nD=M\n@R13\nA=M\nM=D\n')
            elif segment == 'pointer':
                self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@'+str(int(index)+3)+'\nM=D\n')
            elif segment == 'temp':
                self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@'+str(int(index)+5)+'\nM=D\n')
            elif segment == 'static':
                self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@'+self.currentFile+'.'+index+'\nM=D\n')

    def write_label(self, label):
        self.foutput.write('('+self.currentFunc+'$'+label+')\n') # 构造 (funcName$label) 格式的标记，网站上没提，但书里有

    def write_init(self):
        self.foutput.write('@256\nD=A\n@SP\nM=D\n')
        self.write_call('Sys.init', 0)

    def write_goto(self, label):
        self.foutput.write('@'+self.currentFunc+'$'+label+'\n0;JMP\n')

    def write_if(self, label):
        self.foutput.write('@SP\nM=M-1\nA=M\nD=M\n@'+self.currentFunc+'$'+label+'\nD;JNE\n')

    def write_call(self, function_name, num_args):
        self.foutput.write('@RETURN_ADDRESS'+str(self.uniqueFlag)+'\nD=A\n@SP\nA=M\nM=D\n@SP\nM=M+1\n'
                           '@LCL\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n'
                           '@ARG\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n'
                           '@THIS\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n'
                           '@THAT\nD=M\n@SP\nA=M\nM=D\n@SP\nM=M+1\n'
                           '@SP\nD=M\n@'+str(int(num_args)+5)+'\nD=D-A\n@ARG\nM=D\n'
                           '@SP\nD=M\n@LCL\nM=D\n'
                           '@'+function_name+'\n0;JMP\n'
                           '(RETURN_ADDRESS'+str(self.uniqueFlag)+')\n')
        self.uniqueFlag += 1

    def write_return(self):
        self.foutput.write('@LCL\nD=M\n@R13\nM=D\n'
                           '@5\nA=D-A\nD=M\n@R14\nM=D\n'
                           '@SP\nM=M-1\nA=M\nD=M\n@ARG\nA=M\nM=D\n'
                           '@ARG\nD=M+1\n@SP\nM=D\n'
                           '@R13\nA=M-1\nD=M\n@THAT\nM=D\n'
                           '@R13\nA=M-1\nA=A-1\nD=M\n@THIS\nM=D\n'  # 蠢萌蠢萌的减n次1，比下一句还省一条指令。。。
                           '@R13\nD=M\n@3\nA=D-A\nD=M\n@ARG\nM=D\n'
                           '@R13\nD=M\n@4\nA=D-A\nD=M\n@LCL\nM=D\n'
                           '@R14\nA=M\n0;JMP\n')

    def write_function(self, function_name, num_locals):
        self.currentFunc = function_name
        commandsInitLocals = ''
        for i in range(int(num_locals)):
            commandsInitLocals += '@LCL\nD=M\n@'+str(i)+'\nA=A+D\nM=0\n'
        self.foutput.write('('+function_name+')\n'+commandsInitLocals)

    def close(self):
        self.foutput.close()


def process_vm_file(fpath):
    '''
    处理一个 .vm 文件

    fpath: 待处理 .vm 文件的绝对路径
    '''
    parser = Parser(fpath)
    writer.set_file_name(os.path.basename(fpath.strip('.vm')))
    while parser.has_more_commands():
        parser.advance()
        if parser.command_type() == C_TYPE.C_PUSH or parser.command_type() == C_TYPE.C_POP:
            writer.write_push_pop(parser.command_type(), parser.arg1(), parser.arg2())
        elif parser.command_type() == C_TYPE.C_ARITHMETIC:
            writer.write_arithmetic(parser.arg1())
        elif parser.command_type() == C_TYPE.C_LABEL:
            writer.write_label(parser.arg1())
        elif parser.command_type() == C_TYPE.C_GOTO:
            writer.write_goto(parser.arg1())
        elif parser.command_type() == C_TYPE.C_IF:
            writer.write_if(parser.arg1())
        elif parser.command_type() == C_TYPE.C_CALL:
            writer.write_call(parser.arg1(), parser.arg2())
        elif parser.command_type() == C_TYPE.C_FUNCTION:
            writer.write_function(parser.arg1(), parser.arg2())
        elif parser.command_type() == C_TYPE.C_RETURN:
            writer.write_return()

# main program
if os.path.isfile(sys.argv[1]) and sys.argv[1].endswith('.vm'):  # 参数为 .vm 文件，只翻译一个文件
    writer = CodeWriter(os.path.splitext(sys.argv[1])[0]+'.asm')
    writer.write_init()
    process_vm_file(sys.argv[1])
elif os.path.isdir(sys.argv[1]): # 参数为文件夹，将文件夹中所有文件翻译为 hack 汇编
    writer = CodeWriter(sys.argv[1]+'/'+os.path.basename(sys.argv[1])+'.asm')
    writer.write_init()
    for f in glob.glob(sys.argv[1] + '/*.vm'):
        process_vm_file(f)