计算机系统要素：第十一章 编译器：代码生成

标签：编译器   语义分析   计算机系统要素   

一，项目介绍

终于来到了编译器部分的最后一个章节——代码生成阶段。本章的目标就是将Jack语言转化为VM语言，完成Jack编译器的构建。

刚刚接触这章的内容时，会比较难上手，最主要的问题就在于，这章的内容看起来和第十章没有什么关系。刚开始做这个项目时，我就很疑惑，第十章输出的不是一个结构化的xml文件吗？这个文件在第十一章根本不需要输出，那么这章的内容从何开始呢？

的确，这个xml文件是不需要输出的，但是第十章的目的并不单纯是输出这个xml文件，它更重要的目的是为了让我们了解如何对jack程序文件进行语法分析，以完成CompilationEngine的构建。所以，我们需要关注的是CompilationEngine的函数结构，这个函数结构才是第十一章内容的基础。

二，操作步骤

总体而言，作者为我们设计的操作顺序是非常合理的。在此，我再提出几点预备步骤，这些步骤并不是必要的，但是通过这些操作，能够使得整个项目的实现更加流畅。

1，先给命令行加上-x选项，如果命令行中出现-x，则表示输出xml文件和VM文件，不加-x，则表示只输出VM文件。这样子就将两个不同“写入文件流”区分开来。

2，构建符号表模块。存储符号表时，我所用的数据结构是Python语言中的二维列表。这一阶段的任务是把每一个遇到的Identifier都加以标注并且输出相关信息。

3，进入输出VM语言的阶段。首先可以使用内置的JackCompiler将Jack语言转换为VM语言，（Windows上的JackCompiler需要自己设置配置文件才能够使用，具体教程在这儿）从简单的文件开始转换，自己认真分析代码的转换过程。例如，最简单的Seven函数的Jack代码和VM代码分别如下。

class Main {
  function void main() {
      do Output.printInt(1 + (2 * 3));
      return;
   }
}

function Main.main 0
push constant 1
push constant 2
push constant 3
call Math.multiply 2
add
call Output.printInt 1
pop temp 0
push constant 0
return

之后你便可以对照二者，分析转换规则了，例如第一句function Main.main 0肯定是在读取完了所有的ClassVarDev，知道了函数名之后才写入的，于是，写入语句必然就是在compileStatements之前。照这个步骤，逐步完善你的编译器。

三，注意点

我的建议是，先回过头去复习VM代码和Jack语言，了解高级代码转化为VM代码的具体过程，你可以通过看图11.6，图7.9来了解其中的逻辑。

在写编译器的过程中，注意点非常多，这一方面，书中11.2节阐述的非常清楚，在此我重申几个比较关键的问题：

1，constructor, method和function参数配置不同，method方法会默认带一个this的参数，需要加以区分。而讨论参数时，VM代码中function xxx n与call function m中的n与m也是不同的，前者指的是函数中的局部变量数(local)，后者指的是调用函数时引入的参数(argument)。

2，Function和Method的调用方式不同，Function只需调用类名ClassName.Function就可以使用，但是method需要调用具体的类实例如abc.Method才可以调用，如果方法就在类中的话，也可直接使用method()。

3，数组只可能在两个地方出现，一是term中，用于引用，另外是Let语句的左边，用于数组赋值。要注意的是，这两处调用的VM代码是不同的，需要加以区分。

4，constructor是构造函数，在编译时，需要先分析Class中有多少个field变量，然后使用Memory.alloc(size)来给他们分配空间，最后再将其基地址存入this指针中。

上述这些注意点的具体代码都可以通过JackCompiler编译现有文件而得到，我就不再赘述了。

最后，debug的过程是痛苦的，也是无可避免的。如果代码出现问题，可以比对JackCompiler的输出文件与你的编译器输出文件的不同。这个过程能够是你对编译有更深的理解。

JackCompiler.py

#!/usr/bin/python
import CompilationEngine
import SymbolTable
import sys,os

'''
The command line of this module is : JackCompiler.py (-x) sourcename
The first option is -x, which decides whether to run xmlWriter() and to output the constructive xml fileputting forward by CompilationEngine.
'''

option=sys.argv[1]
if option == '-x':
	filename=sys.argv[2]
else:
	filename=sys.argv[1]

#clear all the // /* ... notes, create a new file to save the result
readfile = open(filename,'r')
copyfile = open('copyfile','w')
line=readfile.readline()
while line:
	while line == '\n' or line.startswith('//'):
		line=readfile.readline()
	if '//' in line:
		line=line[:line.find('//')]
	if '/*' in line:
		aline=line[:line.find('/*')]
		while line.find('*/')<0:
			line=readfile.readline()
		bline=line[line.find('*/')+2:]
		line=aline+bline
	copyfile.write(line)
	line=readfile.readline()
copyfile.close()
readfile.close()

#Main Function
readCopyFile=open('copyfile','r')
writeXmlFile=open(filename.strip('.jack')+'.xml','w')
writeVmFile=open(filename.strip('.jack')+'.vm','w')

outputCompile=CompilationEngine.Compile(readCopyFile,writeXmlFile,writeVmFile)
outputCompile.compileClass()

readCopyFile.close()
writeXmlFile.close()
writeVmFile.close()
os.remove('copyfile')

if option != '-x':
	os.remove(filename.strip('.jack')+'.xml')

#!/usr/bin/python
import JackTokenizer
import SymbolTable
import VMWriter

class Compile():
	def __init__(self,rfile,wfile,wVmFile):
		self.rfile=rfile
		self.wfile=wfile #Write XML file
		self.vmWriter=VMWriter.VMwriter(wVmFile) #Write VM file
		self.tokenizer=JackTokenizer.Tokenizer(self.rfile)
		self.class_symbol=SymbolTable.SymbolTable()
		self.sub_symbol=SymbolTable.SymbolTable()
		self.Stype=''  #Stype records the type of the identifier.
		self.Skind=''
		#ClassName records the name of the class, used to make the sub_functionName
		self.ClassName=''
		self.expressionListNum=0	#Record the number of expression in ExpressionList.
		self.WHILEFLAG=0	#the index of while_loop in case of tautonomy
		self.IFFLAG=0
		

	def writeXmlTag(self,token):
		self.wfile.write(token)

	def writeXml(self,tType,token):
		if tType == 'symbol':
			if self.tokenizer.token=='>':
				self.writeXmlTag('<'+tType+'> '+'>'+' </'+tType+'>\n')
			elif self.tokenizer.token=='<':
				self.writeXmlTag('<'+tType+'> '+'<'+' </'+tType+'>\n')
			elif self.tokenizer.token=='&':
				self.writeXmlTag('<'+tType+'> '+'&'+' </'+tType+'>\n')
			else:
				self.writeXmlTag('<'+tType+'> '+token+' </'+tType+'>\n')
		else:
			self.writeXmlTag('<'+tType+'> '+token+' </'+tType+'>\n')

	def NextToken(self):
		if self.tokenizer.hasMoreTokens():
			self.tokenizer.advance()

	def moveBack(self):
		#Move back to the last token.
		lennum=-len(self.tokenizer.token)
		self.rfile.seek(lennum,1)

	def writeArrayPush(self,symbolName):
		#This function is used in 'Push' Array Terms.
		SubTag=self.sub_symbol.FoundName(symbolName)
		if SubTag==-1:
			ClassTag=self.class_symbol.FoundName(symbolName)
			if ClassTag==-1:
				print 'Error Term!'
				exit()
			else:
				self.vmWriter.writePush('this',self.class_symbol.Scope[ClassTag][3])
		else:
			KINDFLAG=self.sub_symbol.Scope[SubTag][2]
			self.vmWriter.writePush(KINDFLAG,self.sub_symbol.Scope[SubTag][3])

	def defineSymbol(self,symbolName,_symbol):
		#This function adds symbolName into SymbolTable.
		_symbol.Define(symbolName,self.Stype,self.Skind)

	def checkSymbol(self,symbolName):
		#Check the index of the Identifier
		SubTag=self.sub_symbol.FoundName(symbolName)
		if SubTag==-1:
			ClassTag=self.class_symbol.FoundName(symbolName)
			if ClassTag==-1:
				return -1
			else:
				return self.class_symbol.Scope[ClassTag]
		else:
			return self.sub_symbol.Scope[SubTag]

	def compileType(self):
		tType=self.tokenizer.tokenType()
		if tType == 'KEYWORD':
			self.Stype=self.tokenizer.token
			self.writeXml('keyword',self.tokenizer.token)
		elif tType == 'IDENTIFIER':
			self.Stype=self.tokenizer.token
			self.writeXml('identifier',self.tokenizer.token)

	def compileTermType(self):
		tType=self.tokenizer.tokenType()
		if tType == 'KEYWORD':
			kWord=self.tokenizer.token
			if kWord=='true':
				self.vmWriter.writePush('constant',1)
				self.vmWriter.writeArithmetic('neg')
			elif kWord=='false' or kWord=='null':
				self.vmWriter.writePush('constant',0)
			elif kWord=='this':
				self.vmWriter.writePush('pointer',0)
			self.writeXml('keyword',self.tokenizer.token)
		elif tType == 'INT_CONSTANT':
			self.writeXml('integerConstant',self.tokenizer.token)
			self.vmWriter.writePush('constant',int(self.tokenizer.token))
		elif tType == 'STRING_CONSTANT':	
			string_copy=self.tokenizer.token.strip('"')
			self.writeXml('stringConstant',string_copy)
			string_length=len(string_copy)
			self.vmWriter.writePush('constant',string_length)
			self.vmWriter.writeCall('String.new',1)
			for i in range(0,string_length):
				self.vmWriter.writePush('constant',ord(string_copy[i]))
				self.vmWriter.writeCall('String.appendChar',2)

	def compileVarDec(self):
		'''
		var type varName(,'varName')*;
		'''
		self.writeXmlTag('<varDec>\n')
		self.writeXml('keyword','var')
		self.Skind='var'
		#type
		self.NextToken()
		self.compileType()
		#varName
		self.NextToken()
		self.writeXml('identifier',self.tokenizer.token)
		self.defineSymbol(self.tokenizer.token,self.sub_symbol)
		#(,varName)*
		self.NextToken()
		while self.tokenizer.token != ';':
			self.writeXml('symbol',self.tokenizer.token)
			self.NextToken()
			self.writeXml('identifier',self.tokenizer.token)
			self.defineSymbol(self.tokenizer.token,self.sub_symbol)
			self.NextToken()
		self.writeXml('symbol',self.tokenizer.token)
		self.writeXmlTag('</varDec>\n')
		
	def compileParameterList(self):
		'''
		((type varName)(, type varName)*)?
		'''
		self.writeXmlTag('<parameterList>\n')
		self.NextToken()
		while self.tokenizer.token != ')':
			self.Skind='argument'
			if self.tokenizer.token != ',':
				self.compileType()
				self.NextToken()
				self.writeXml('identifier',self.tokenizer.token)
				self.defineSymbol(self.tokenizer.token,self.sub_symbol)
				self.NextToken()
			else:
				self.writeXml('symbol',self.tokenizer.token)
				self.NextToken()
				self.compileType()
				self.NextToken()
				self.writeXml('identifier',self.tokenizer.token)
				self.defineSymbol(self.tokenizer.token,self.sub_symbol)
				self.NextToken()				
		self.writeXmlTag('</parameterList>\n')

	def compileClassVarDec(self):
		'''
		('static'|'field') type varName(, varName)*;
		'''
		self.writeXmlTag('<classVarDec>\n')
		self.writeXml('keyword',self.tokenizer.token)
		self.Skind=self.tokenizer.token

		self.NextToken()
		self.compileType()
		#varName
		self.NextToken()
		self.writeXml('identifier',self.tokenizer.token)
		self.defineSymbol(self.tokenizer.token,self.class_symbol)
		#(,varName)*
		self.NextToken()
		while self.tokenizer.token != ';':
			self.writeXml('symbol',self.tokenizer.token)
			self.NextToken()
			self.writeXml('identifier',self.tokenizer.token)
			self.defineSymbol(self.tokenizer.token,self.class_symbol)
			self.NextToken()
		self.writeXml('symbol',self.tokenizer.token)
		self.writeXmlTag('</classVarDec>\n')

	def compileTerm(self):
		self.writeXmlTag('<term>\n')
		self.NextToken()
		tType=self.tokenizer.tokenType()
		if tType == 'IDENTIFIER':
			temp=self.rfile.read(1)
			if temp=='.':
				lennum=-len(self.tokenizer.token)-1
				self.rfile.seek(lennum,1)
				self.subroutineCall()
			elif temp=='[':
				self.writeXml('identifier',self.tokenizer.token)
				self.writeArrayPush(self.tokenizer.token)
				self.writeXml('symbol','[')
				self.compileExpression()
				self.vmWriter.writeArithmetic('add')
				self.vmWriter.writePop('pointer',1)
				self.vmWriter.writePush('that',0)
				self.writeXml('symbol',']')
			else:
				self.rfile.seek(-1,1)
				self.writeXml('identifier',self.tokenizer.token)
				ListSeg=self.checkSymbol(self.tokenizer.token)
				self.vmWriter.writePush(ListSeg[2],ListSeg[3])
		elif self.tokenizer.token in ('-','~'):
			UnaryOp=self.tokenizer.token
			self.writeXml('symbol',self.tokenizer.token)
			self.compileTerm()
			if UnaryOp == '-':
				self.vmWriter.writeArithmetic('neg')
			else:
				self.vmWriter.writeArithmetic('not')
		elif self.tokenizer.token == '(':
			self.writeXml('symbol',self.tokenizer.token)
			self.compileExpression()
			self.writeXml('symbol',')')
		else:
			self.compileTermType()
		self.writeXmlTag('</term>\n')

	def compileExpression(self):
		'''
		term (op term)*
		'''
		self.writeXmlTag('<expression>\n')
		self.compileTerm()
		self.NextToken()
		while (self.tokenizer.tokenType() == 'SYMBOL' and 			self.tokenizer.Symbol() in '+-*/&|<>='):
			operator = self.tokenizer.Symbol()
			self.writeXml('symbol', self.tokenizer.token)
			self.compileTerm()
			if operator == '+':
				self.vmWriter.writeArithmetic('add')
			elif operator == '-':
				self.vmWriter.writeArithmetic('sub')
			elif operator == '*':
				self.vmWriter.writeCall('Math.multiply', 2)
			elif operator == '/':
				self.vmWriter.writeCall('Math.divide', 2)
			elif operator == '&':
				self.vmWriter.writeArithmetic('and')
			elif operator == '|':
				self.vmWriter.writeArithmetic('or')
			elif operator == '<':
				self.vmWriter.writeArithmetic('lt')
			elif operator == '>':
				self.vmWriter.writeArithmetic('gt')
			elif operator == '=':
				self.vmWriter.writeArithmetic('eq')
			self.NextToken()

		self.writeXmlTag('</expression>\n')	

	def compileExpressionList(self):
		self.writeXmlTag('<expressionList>\n')
		self.expressionListNum=0
		self.NextToken()
		while self.tokenizer.token != ')':
			if self.tokenizer.token != ',':
				self.moveBack()
				self.compileExpression()
				self.expressionListNum+=1
			else:
				self.writeXml('symbol',self.tokenizer.token)
				self.compileExpression()
				self.expressionListNum+=1
		self.writeXmlTag('</expressionList>\n')

	def subroutineCall(self):
		sub_MethodFlag=False 
		self.NextToken()
		self.writeXml('identifier',self.tokenizer.token)
		sub_className=self.tokenizer.token
		self.NextToken()
		if self.tokenizer.token=='.':
			self.writeXml('symbol',self.tokenizer.token)
			self.NextToken()
			self.writeXml('identifier',self.tokenizer.token)
			sub_funcName=self.tokenizer.token
			#To check if sub_className is a ClassName or an instance
			SubCallTag=self.sub_symbol.FoundName(sub_className)
			if SubCallTag==-1:
				ClassCallTag=self.class_symbol.FoundName(sub_className)
				if ClassCallTag==-1:
					sub_Name=sub_className+'.'+sub_funcName
				else:
					sub_MethodFlag=True
					sub_className=self.class_symbol.Scope[ClassCallTag][1]
					sub_index=self.class_symbol.Scope[ClassCallTag][3]
					self.vmWriter.writePush('this',sub_index)
					sub_Name=sub_className+'.'+sub_funcName
			else:
				sub_MethodFlag=True
				sub_className=self.sub_symbol.Scope[SubCallTag][1]
				sub_index=self.sub_symbol.Scope[SubCallTag][3]
				self.vmWriter.writePush('local',sub_index)
				sub_Name=sub_className+'.'+sub_funcName
			self.rfile.read(1)
			self.writeXml('symbol','(')
			self.compileExpressionList()
			self.writeXml('symbol',')')
			if sub_MethodFlag:
				self.vmWriter.writeCall(sub_Name,self.expressionListNum+1)
			else:
				self.vmWriter.writeCall(sub_Name,self.expressionListNum)
		elif self.tokenizer.token=='(':
			sub_Name=self.ClassName+'.'+sub_className
			self.writeXml('symbol','(')
			self.vmWriter.writePush('pointer',0)
			self.compileExpressionList()
			self.vmWriter.writeCall(sub_Name,self.expressionListNum+1)
			self.writeXml('symbol',')')

	def compileDo(self):
		self.writeXmlTag('<doStatement>\n')
		self.writeXml('keyword',self.tokenizer.token)
		self.subroutineCall()
		self.vmWriter.writePop('temp',0)
		self.NextToken()
		self.writeXml('symbol',self.tokenizer.token)
		self.writeXmlTag('</doStatement>\n')	

	def compileLet(self):
		'''
		If the term on the left of '=' is Array, the order of the VM code is 
		totally different from other conditions. 
		'''
		self.writeXmlTag('<letStatement>\n')
		self.writeXml('keyword',self.tokenizer.token)
		self.NextToken()
		self.writeXml('identifier',self.tokenizer.token)
		LetVarName=self.tokenizer.token
		ListSeg=self.checkSymbol(LetVarName)
		self.NextToken()
		temp=self.tokenizer.token
		if temp=='[':
			self.writeArrayPush(LetVarName)
			self.writeXml('symbol',self.tokenizer.token)
			self.compileExpression()
			self.writeXml('symbol',']')
			self.vmWriter.writeArithmetic('add')
			self.NextToken()
			self.writeXml('symbol',self.tokenizer.token)
			self.compileExpression()
			self.vmWriter.writePop('temp',0)
			self.vmWriter.writePop('pointer',1)
			self.vmWriter.writePush('temp',0)
			self.vmWriter.writePop('that',0)
			self.writeXml('symbol',';')
			self.writeXmlTag('</letStatement>\n')	
		elif temp == '=':
			self.writeXml('symbol',self.tokenizer.token)
			self.compileExpression()
			self.vmWriter.writePop(ListSeg[2],ListSeg[3])
			self.writeXml('symbol',';')
			self.writeXmlTag('</letStatement>\n')

	def compileWhile(self):
		self.writeXmlTag('<whileStatement>\n')
		self.writeXml('keyword',self.tokenizer.token)
		sub_WHILEFLAG=self.WHILEFLAG
		self.WHILEFLAG+=1
		self.vmWriter.writeLabel('WHILE_START'+str(sub_WHILEFLAG))
		#(expression)
		self.NextToken()
		self.writeXml('symbol',self.tokenizer.token)
		self.compileExpression()
		self.writeXml('symbol',')')
		self.vmWriter.writeArithmetic('not')
		self.vmWriter.writeIf('WHILE_OVER'+str(sub_WHILEFLAG))
		#{statements}
		self.NextToken()
		self.writeXml('symbol',self.tokenizer.token)
		self.compileStatements()
		self.vmWriter.writeGoto('WHILE_START'+str(sub_WHILEFLAG))
		self.vmWriter.writeLabel('WHILE_OVER'+str(sub_WHILEFLAG))
		self.writeXml('symbol',self.tokenizer.token)	
		self.writeXmlTag('</whileStatement>\n')

	def compileReturn(self):
		self.writeXmlTag('<returnStatement>\n')
		self.writeXml('keyword',self.tokenizer.token)
		#expression?
		self.NextToken()
		if self.tokenizer.token == ';':
			self.writeXml('symbol',self.tokenizer.token)
			self.vmWriter.writePush('constant',0)
			self.vmWriter.writeReturn()
		else:
			self.moveBack()
			self.compileExpression()
			self.vmWriter.writeReturn()
			self.writeXml('symbol',';')
		self.writeXmlTag('</returnStatement>\n')

	def compileStatements(self):
		self.writeXmlTag('<statements>\n')
		self.NextToken()
		while self.tokenizer.token != '}':
			if self.tokenizer.token =='let':
				self.compileLet()
			elif self.tokenizer.token == 'if':
				self.compileIf()
			elif self.tokenizer.token == 'while':
				self.compileWhile()
			elif self.tokenizer.token == 'do':
				self.compileDo()
			elif self.tokenizer.token == 'return':
				self.compileReturn()
			else:
				print 'Error!'+self.tokenizer.token
				exit()
			self.NextToken()
		self.writeXmlTag('</statements>\n')

	def compileIf(self):
		self.writeXmlTag('<ifStatement>\n')
		sub_IFFLAG=self.IFFLAG
		self.IFFLAG+=1
		self.writeXml('keyword',self.tokenizer.token)
		#(expression)
		self.NextToken()
		self.writeXml('symbol',self.tokenizer.token)
		self.compileExpression()
		self.writeXml('symbol',')')
		self.vmWriter.writeArithmetic('not')
		self.vmWriter.writeIf('IF_RIGHT'+str(sub_IFFLAG))	
		#{statements}
		self.NextToken()
		self.writeXml('symbol',self.tokenizer.token)
		self.compileStatements()
		self.writeXml('symbol',self.tokenizer.token)
		#(else {statements})?	
		self.NextToken()
		if self.tokenizer.token=='else':
			self.vmWriter.writeGoto('IF_WRONG'+str(sub_IFFLAG))
			self.vmWriter.writeLabel('IF_RIGHT'+str(sub_IFFLAG))
			self.writeXml('keyword',self.tokenizer.token)
			self.NextToken()
			self.writeXml('symbol',self.tokenizer.token)
			self.compileStatements()
			self.vmWriter.writeLabel('IF_WRONG'+str(sub_IFFLAG))
			self.writeXml('symbol',self.tokenizer.token)
		else:
			self.vmWriter.writeLabel('IF_RIGHT'+str(sub_IFFLAG))
			self.moveBack()
		self.writeXmlTag('</ifStatement>\n')	

	def compileClass(self):
		self.writeXmlTag('<class>\n')
		self.NextToken()
		self.writeXml('keyword',self.tokenizer.token)
		self.NextToken()
		self.writeXml('identifier',self.tokenizer.token)
		self.ClassName=self.tokenizer.token
		self.NextToken()
		self.writeXml('keyword',self.tokenizer.token)
		#classVarDec*
		self.NextToken()
		while self.tokenizer.token in ('static','field'):		
			self.compileClassVarDec()
			self.NextToken()
		#subroutineDec*
		while self.tokenizer.token in ('constructor','function','method'):
			self.compileSubroutine()
			self.NextToken()
		self.writeXml('symbol',self.tokenizer.token)
		self.writeXmlTag('</class>\n')

	def compileSubroutine(self):
		Subroutine_Flag=''
		self.WHILEFLAG=0
		self.IFFLAG=0
		self.writeXmlTag('<subroutineDec>\n')
		self.writeXml('keyword',self.tokenizer.token)
		self.sub_symbol.startSubroutine()
		if self.tokenizer.token =='method':
			self.sub_symbol.Define('this',self.ClassName,'argument')
			Subroutine_Flag='METHOD'
		elif self.tokenizer.token == 'constructor':
			Subroutine_Flag='CONSTRUCTOR'
		else:
			Subroutine_Flag='FUNCTION'
		#(void|type) subroutineName (parameterList)
		self.NextToken()
		self.compileType()
		self.NextToken()
		self.writeXml('identifier',self.tokenizer.token)
		#special, to be xxx.yyy
		FunctionName=self.ClassName+'.'+self.tokenizer.token
		self.NextToken()
		self.writeXml('symbol',self.tokenizer.token)
		self.compileParameterList()
		self.writeXml('symbol',self.tokenizer.token)
		#subroutinBody
		self.writeXmlTag('<subroutineBody>\n')
		#{varDec* statements}
		self.NextToken()
		self.writeXml('symbol',self.tokenizer.token)
		self.NextToken()
		while self.tokenizer.token == 'var':
			self.compileVarDec()
			self.NextToken()
		self.moveBack()
		LclNum=self.sub_symbol.VarCount('var')
		self.vmWriter.writeFunction(FunctionName,LclNum)
		if Subroutine_Flag == 'METHOD':
			self.vmWriter.writePush('argument',0)
			self.vmWriter.writePop('pointer',0)
		elif Subroutine_Flag=='CONSTRUCTOR':
			FieldNum=self.class_symbol.VarCount('field')
			self.vmWriter.writePush('constant',FieldNum)
			self.vmWriter.writeCall('Memory.alloc',1)
			self.vmWriter.writePop('pointer',0)
		self.compileStatements()
		self.writeXml('symbol',self.tokenizer.token)
		self.writeXmlTag('</subroutineBody>\n')
		self.writeXmlTag('</subroutineDec>\n')

#!/usr/bin/python
class SymbolTable:
	'''
	SymbolTable is a two-dimensional list.
	The first list contains all the names of the symbols. And Each name 
	is also a single list, containing the [name,type,kind,index] of the 
	symbol.
	'''
	def __init__(self):
		self.Scope=[]

	def Constructor(self):
		self.Scope=[]

	def startSubroutine(self):
		self.Scope=[]

	def FoundName(self,name):
		#Search the funcName in SymbolTable
		for i in range(0,len(self.Scope)):
			if name == self.Scope[i][0]:
				return i
		return -1

	def Define(self,name,segType,kind):
		#Add new elements into the List.
		index=self.VarCount(kind)
		if kind == 'field':
			kind='this'		
		elif kind == 'var':
			kind='local'
		name=[name,segType,kind,index]
		self.Scope.append(name)

	def VarCount(self,kind):
		#count the number of existed elements with 'kind'.
		#It is used to count the index of the elements.
		if kind == 'field':
			kind='this'		
		elif kind == 'var':
			kind='local'
		lengthKind=0
		for i in range(0,len(self.Scope)):
			if self.Scope[i][2]==kind:
				lengthKind+=1
		return lengthKind

	def KindOf(self,name):
		for i in range(0,len(self.Scope)):
			if name == self.Scope[i][0]:
				return self.Scope[i][2]
		return 'NONE'

	def TypeOf(self,name):
		for i in range(0,len(self.Scope)):
			if name == self.Scope[i][0]:
				return self.Scope[i][1]
		return 'NONE'

	def IndexOf(self,name):
		for i in range(0,len(self.Scope)):
			if name == self.Scope[i][0]:
				return self.Scope[i][3]
		return 'NONE'

#!/usr/bin/python

class VMwriter:
	def __init__(self,wfile):
		self.wfile=wfile

	def writePush(self,segment,index):
		self.wfile.write('push '+segment+' '+str(index)+'\n')

	def writePop(self,segment,index):
		self.wfile.write('pop '+segment+' '+str(index)+'\n')

	def writeArithmetic(self,command):
		self.wfile.write(command+'\n')

	def writeFunction(self,functionName,LclNum):
		self.wfile.write('function '+functionName+' '+str(LclNum)+'\n')

	def writeReturn(self):
		self.wfile.write('return\n')

	def writeCall(self,functionName,ELNum):
		self.wfile.write('call '+functionName+' '+str(ELNum)+'\n')

	def writeLabel(self,label):
		self.wfile.write('label '+label+'\n')

	def writeGoto(self,label):
		self.wfile.write('goto '+label+'\n')

	def writeIf(self,label):
		self.wfile.write('if-goto '+label+'\n')

#!/usr/bin/python
STable=('{','}','(',')','[',']','.',',',';','+','-','*','/','&','|','<','>','=','~')
KWtable=('class','constructor','function','method','field','static','var','int','char','boolean',	'void','true','false','null','this','let','do','if','else','while','return')

class Tokenizer():
	def __init__(self,rfile):
		self.rfile=rfile
		self.token=''

	def hasMoreTokens(self):
		temp=self.rfile.read(1)
		while temp in ' \n\t' and temp != '':
			temp=self.rfile.read(1)
		if not temp:
			return 0
		else:
			self.rfile.seek(-1,1)
			return 1

	def advance(self):
		self.token=''
		temp=self.rfile.read(1)

		if temp.isalpha() or temp.isdigit() or temp == '_':
			while temp.isalpha() or temp.isdigit() or temp == '_':
				self.token+=temp
				temp=self.rfile.read(1)
			if temp in STable or temp =='"':
				self.rfile.seek(-1,1)
			elif temp == ' ' or temp == '\n':
				self.rfile.seek(-1,1)
		elif temp in STable:
			self.token=temp
		elif temp =='"':
			self.token += '"'
			temp=self.rfile.read(1)
			while temp != '"':
				self.token+=temp
				temp=self.rfile.read(1)
			self.token+='"'

	def tokenType(self):
		if self.token in KWtable:
			return 'KEYWORD'
		elif self.token in STable:
			return 'SYMBOL'
		elif self.token.isdigit():
			return 'INT_CONSTANT'
		elif self.token.startswith('"'):
			return 'STRING_CONSTANT'
		else:
			return 'IDENTIFIER'

	def Keyword(self):
		return self.token

	def Symbol(self):
		return self.token

	def Identifier(self):
		return self.token

	def intVal(self):
		return int(self.token)

	def stringVal(self):
		return self.token

计算机系统要素：第十一章 编译器：代码生成

标签：编译器   语义分析   计算机系统要素   

原文地址：http://blog.csdn.net/thomas_in_june/article/details/39344965