用AntlR4实现简单的汇编编译器

最近在学习《计算机系统要素——从零构建现代计算机》这本书，花了两天时间用antlr4把书中第6章的汇编编译器实现了.

输出的机器指令和参考答案完全一致.

下面把antlr4的代码贴出来，供后来者参考

/**
 * Define a grammar called Hello
 */
grammar HackAsm;
@parser::header{
import java.util.Map;
import java.util.HashMap;
import java.util.List;
import java.util.LinkedList;
}
@parser::members{
	class Command{
		String code;  //指令的编码
		final static boolean DEBUG=false;  //是否为调试模式
	}
	
	class ACommand extends Command
	{
		Integer addr;
		String ID;
		void update()
		{
			
			
			String str = Integer.toBinaryString(addr);
			
			if (Command.DEBUG) {
				System.out.println("addr="+addr);
				System.out.println("str="+str);
			}
			
			int offset = 16-str.length();
			if(offset>0)
			{
				byte[] b= new byte[offset];
				for(int i=0;i<offset;++i)
					b[i]='0';
				str = new String(b)+str;	//前面补0
				
				if (Command.DEBUG)
					System.out.println("str="+str);
			}
			
			code = str.substring(0,16);
		}
	}
}

prog
locals[
	List<Command> cmds;
	Map<String,Integer> sym_table
	
]
@init{
	$ctx.cmds= new LinkedList<Command>();
	$ctx.sym_table = new HashMap<String,Integer>();
	Map<String,Integer> sym_table =$ctx.sym_table;
	//建立预定义符号表
	sym_table.put("SP",0);
	sym_table.put("LCL",1);
	sym_table.put("ARG",2);
	sym_table.put("THIS",3);
	sym_table.put("THAT",4);
	sym_table.put("SCREEN",16384);
	sym_table.put("KBD",24576);
	for(int i=0;i<16;++i)
		sym_table.put("R"+i,i);
}

@after{
	if (Command.DEBUG) {
		System.out.println("after prog");
		System.out.println("processed "+$ctx.cmds.size()+" commands");
		System.out.println($ctx.sym_table.size()+" syms");
	}
	
	//遍历每个A command,把里面的ID变成数字
	Map<String,Integer> vars = new HashMap<String,Integer>();
	Map<String,Integer> syms =$ctx.sym_table;
	List<Command> cmds = $ctx.cmds;
	int var_addr = 16;
	for(Command c:cmds)
	{
		if (c instanceof ACommand){
			ACommand cmd = (ACommand) c;
			String id = cmd.ID; 
			if (id != null)
			{
				cmd.addr = syms.get(id);
				if (cmd.addr == null){
					cmd.addr = vars.get(id);
					if (cmd.addr == null)
					{
						cmd.addr = var_addr;
						vars.put(id,var_addr);
						++var_addr;
					}
				}
			}
			cmd.update();
		}
	}
	
	for(Command c:cmds)
	{
		System.out.println(c.code);	
	}
	
}
 
: (command? EL)+
{
	
}
;        
	
command 
	@after{
		ProgContext prog = (ProgContext)($ctx.getParent());
	}
:
	a_command 
	{
		ProgContext prog = (ProgContext)($ctx.getParent());
		List<Command> cmds = prog.cmds;
		cmds.add($a_command.cmd);
	}
	
	|c_command
	{
		ProgContext prog = (ProgContext)($ctx.getParent());
		List<Command> cmds = prog.cmds;
		cmds.add($c_command.cmd);
	}
	|label
	{
		ProgContext prog = (ProgContext)($ctx.getParent());
		Map<String,Integer> sym_table = prog.sym_table;
		String str =$label.label_name; 
		if (sym_table.containsKey(str))
			throw new RuntimeException("label "+ str +"has already used!");	
		else
			sym_table.put(str,prog.cmds.size());
	}
	;
	
label returns[String label_name] 
:
	LP ID RP
	{
		String str =$ID.text;
		$label_name= str;
		
		if (Command.DEBUG)
			System.out.println("label="+$label_name);	
	}
    ;
	
a_command returns[ACommand cmd]
@init{
	$cmd = new ACommand();
}
@after{
	if (Command.DEBUG)
		System.out.println("ACommand: addr= "
			+$cmd.addr+", ID="+$cmd.ID
		);
}
:
	AT NUM
	{
		$cmd.addr = $NUM.int;
		$cmd.ID = null;
	}
	|AT ID
	{
		$cmd.ID = $ID.text;
		//System.out.println("ACommand: ID= "	+", ID="+$ID.text);
	}
	//下面两个分支是因为0 1也属于NUM,不写的话运行时会出问题!
	|AT ZERO
	{
		$cmd.addr = 0;
		$cmd.ID = null;
	}
	|AT ONE
	{
		$cmd.addr = 1;
		$cmd.ID = null;
	}
	;
	
c_command returns[Command cmd]
locals[
	StringBuilder destcode;
	String compcode;
	String jmpcode
]
@init{
	$cmd = new Command();
	$ctx.destcode = new StringBuilder("000");
	$ctx.jmpcode ="000";
	$ctx.compcode ="0000000";
}
@after{
	$cmd.code = "111"+$ctx.compcode+$ctx.destcode+$ctx.jmpcode;			
	
	if (Command.DEBUG){
		System.out.println($ctx.getText()+" is a c_command.");
		System.out.println("dest:"+$ctx.destcode);
		System.out.println("comp:"+$ctx.compcode);
		System.out.println("jmp:"+$ctx.jmpcode);
		System.out.println("code:"+$cmd.code);
	}
}
:

	(dests=(ID|DEST) ASSIGN)? comp (SEMICOLON jmp)?
	{
		String dest = $dests.text;
		
		if (Command.DEBUG)
			System.out.println("comp.dests.text"+dest);
		
		if (dest!=null)
		{	
			if(dest.indexOf("A")!=-1)
				$ctx.destcode.setCharAt(0,'1');
			if(dest.indexOf("D")!=-1)
				$ctx.destcode.setCharAt(1,'1');
			if(dest.indexOf("M")!=-1)
				$ctx.destcode.setCharAt(2,'1');		
		}
		
		if (Command.DEBUG)
			System.out.println("comp_part:"+$comp.text);
		
		$ctx.compcode = $comp.afield + $comp.code;
		
		String JMPstr=$jmp.text;
		if (JMPstr != null)
			$ctx.jmpcode = $jmp.code;
	}
	//|comp (SEMICOLON jmp)?
	;	
	
comp returns[String code,String afield]
@init{
	$code ="000000";
	$afield="0";
}
@after{
	if (Command.DEBUG)
		System.out.println("comp_code:"+$code);
}
:
	bin_expr
	{
		String str = $bin_expr.text;
		if (str.indexOf("M")!= -1)
			$afield="1";
		
		if (str.equals("D+1"))
			$code="011111";
		else if(str.equals("D-1"))
		 	$code="001110";
		else if (str.equals("A+1") ||str.equals("M+1"))
			$code="110111";
		else if (str.equals("A-1") ||str.equals("M-1"))
			$code="110010";	
		else if (str.equals("D+A") ||str.equals("D+M"))
			$code="000010";
		else if (str.equals("D-A") ||str.equals("D-M"))
			$code="010011";
		else if (str.equals("A-D") ||str.equals("M-D"))
			$code="000111";
		else if (str.equals("D&A") ||str.equals("D&M"))
			$code="000000";
		else if (str.equals("D|A") ||str.equals("D|M"))
			$code="010101";	
	}
	|un_expr
	{
		String str = $un_expr.text;
		if (str.indexOf("M")!= -1)
			$afield="1";
		
		if (str.equals("-1"))
			$code="111010";
		else if	(str.equals("-D"))
			$code="001111";
		else if	(str.equals("-A")||str.equals("-M"))
			$code="110011";
		else if	(str.equals("!D"))
			$code="001101";			
		else if	(str.equals("!A") ||str.equals("!M"))
			$code="110001";				
		
	}
	|ZERO
	{$code="101010";}
	|ONE
	{$code="111111";}
	|DEST
	{
		String str =$DEST.text;
		
		if (Command.DEBUG)
			System.out.println("comp.dest.text:"+str);
			
		if (str.equals("D"))
			$code="001100";
		else if (str.equals("A")||str.equals("M"))
			$code="110000";
			
		if (str.equals("M"))
			$afield="1";	
	}
;

bin_expr: 
	DEST bin_op (ONE|DEST)
	;
bin_op:PLUS|MINUS|BIT_AND|BIT_OR
	; 
	
un_expr:
	neg_expr|not_expr
	;
	
neg_expr:MINUS (ONE|DEST);
not_expr:BIT_NOT DEST;  

jmp returns[String code]
	:
	 'JGT'{$code ="001";}
	|'JEQ'{$code ="010";}
	|'JGE'{$code ="011";}
	|'JLT'{$code ="100";}
	|'JNE'{$code ="101";}
	|'JLE'{$code ="110";}
	|'JMP'{$code ="111";}
	;


BLOCK_COMMENT
	: '/*' .*? '*/' -> channel(HIDDEN)
	;

SL_COMMENT
    : '//' ~[\r\n]* -> channel(HIDDEN)   //过滤行注释,本技巧见课本中文版的第181页的12.1节,但是要保留最后的回车符!
    ;

DEST:'A'|'M'|'D';

ZERO:'0';
ONE:'1';

	
ID : [A-Za-z$:.][A-Za-z0-9_$:.]* ;      
NUM:[0-9][0-9]*
	;


PLUS: '+';
MINUS: '-';
BIT_AND:'&';
BIT_OR:'|';
BIT_NOT:'!';

ASSIGN: '=';
SEMICOLON : ';' ;
LP: '('
	;
RP:')'
	;
AT: '@';

EL : '\n';
WS : [ \t\r]+ -> skip ; // skip spaces, tabs, newlines

运行办法：把上面的代码保存到HackAsm.g4文件中,然后和

antlr-4.7.1-complete.jar放到一个目录下,设置好java的环境变量之后,

进入到文件所在的命令行,执行下面的命令

java -cp .\antlr-4.7.1-complete.jar;%CLASSPATH% org.antlr.v4.Tool  HackAsm.g4
javac -cp .\antlr-4.7.1-complete.jar;%CLASSPATH% *.java

如果没有错误的话,把书中第6章的所有测试代码文件也复制到本目录下,

然后执行

@java -cp .\antlr-4.7.1-complete.jar;%CLASSPATH% org.antlr.v4.gui.TestRig HackAsm prog Add.asm

输出结果为

0000000000000010
1110110000010000
0000000000000011
1110000010010000
0000000000000000
1110001100001000

经过比对和书中自带的汇编编译器产生的结果完全一致,说明测试通过