analyzer.java

CMM编译器词法分析及语法分析代码

package try_sw;
import java.util.*;
import java.util.regex.*;
import java.io.*;

class FormatException extends Exception{
	private String msg="";
	public FormatException(String msg){
		this.msg= msg;
	}
	public String toString(){
		return "FormatException  : "+msg;
	}
}

class lexicalError extends Exception{
	private String msg="";
	public lexicalError(String msg){
		this.msg= msg;
	}
	public String toString(){
		return "lexicalError  : "+msg;
	}
}

public class analyzer {

	/**
	 * @param args
	 */
	public  Mystack analyze(Newhash hash,String contest){
		
		Mystack sta = new Mystack();
	//	Mystack stb = new Mystack();//what is that for?
		
		
		
		Pattern pn = Pattern.compile("([a-zA-Z0-9]|[\\.])+");//number,variable,illegic variable
		Pattern pn1= Pattern.compile("[a-zA-Z]([a-zA-Z0-9]*)");//variable
		Pattern pn2= Pattern.compile("([0]|([1-9]([0-9]*)))[\\.]([0-9]+)|[0]|([1-9]([0-9]*))");//number
		Pattern pn3= Pattern.compile("[^\\.a-zA-Z0-9][^\\.a-zA-Z0-9]");//两个非词
		Pattern pn4= Pattern.compile("[^\\.a-zA-Z0-9]");//a non-word
		Pattern pn5= Pattern.compile("/\\*.+\\*/");
/////////////innotation
		Matcher m6 = pn5.matcher(contest);
		while(m6.find())contest=m6.replaceAll("");
	////////////innotation
		String[] test = contest.split("\\s+");
		
		
		//非字母字符：先查两再查一
		for(String str : test){
		   
			
			String work = str;	
			
			while(work.length()>0){
				try{
				
				Matcher m = pn.matcher(work);
			if(m.lookingAt()){//是数字或variable
				Matcher m1= pn1.matcher(m.group());
				Matcher m2 = pn2.matcher(m.group());
				
				if(m1.matches()){//是合法变量组合
					if(hash.containsKey(m.group()))
					sta.push(m.group(), 2);//在表里找得到的，是保留字
					else sta.push(m.group(), 1);//找不到的，是变量
				}else if(m2.matches()){//是数字组合
					sta.push(m.group(), 3);//是数字
				}else{//是字母数字，或数字，但不是合法组合
					work=m.replaceFirst("");//将匹配部分去掉
					throw new FormatException(m.group());
				}
				work=m.replaceFirst("");//将匹配部分去掉
			}else if(work.length()>1){//1.不是数字字母2。还有两个以上剩余
				Matcher m3 = pn3.matcher(work);
				Matcher m4 = pn4.matcher(work);
				if(m3.lookingAt()){//若两个非字母数字相连
					if (hash.containsKey(m3.group())){//若他们合在一起是个东西
						sta.push(m3.group(), hash.get(m3.group()));
						work=m3.replaceFirst("");//将匹配部分去掉
					}//end if
					else if(m4.lookingAt()){//在只有一个是non字母数字时
						if(hash.containsKey(m4.group())){//若他是可以找到的
							sta.push(m4.group(), hash.get(m4.group()));//就压栈
						}else{work=m4.replaceFirst("");//将匹配部分去掉 
							throw new FormatException(m4.group());}//若找不到，就报错
						work=m4.replaceFirst("");//将匹配部分去掉
					
				}//end else if
					
				}else if(m4.lookingAt()){//////添加
					if(hash.containsKey(m4.group())){
						sta.push(m4.group(), hash.get(m4.group()));
						work=m4.replaceFirst("");//将匹配部分去掉
					}else{work=m4.replaceFirst("");
					throw new FormatException(m4.group());}//若找不到，就报错
					
				}
			}else if (work.length()==1){//1.不是数字字母2。只有一个剩余
				Matcher m5 = pn4.matcher(work);
				if(m5.lookingAt()){
					if(hash.containsKey(m5.group())){
						sta.push(m5.group(), hash.get(m5.group()));
						work=m5.replaceFirst("");//将匹配部分去掉
					}else{//1.only one left which is not a num 2.it can't be found
						work=m5.replaceFirst("");//将匹配部分去掉
						throw new FormatException(m5.group());
					}
				}
			}else{//这里其实不可达
				
				throw new lexicalError(work.substring(0,10)+"...");
				
			}//end else
			
			
			}catch(FormatException fe){System.out.println(fe);}
		catch(lexicalError le){System.out.println(le);}
				
				//end try
			
		}//end while
		}//end for
		
		sta.push("#", 0);//#-->0
			
		
		return sta;
		
		
	}
	/////////////////////////////
	
	static Newhash nh = new Newhash();//用例测试
	
	static void hashinit(){
		
		/*
		 * 1--variable
		 * 2--preserved
		 * 3--number
		 * 4--logicsymbol
		 * 5--mathsymbol
		 * 6--othersymbol
		 */
		String[] preservedarr = {"if","else","while","read","write","int","real"};
		String[] logicop= {"<","==","<>"};
		String[] mathop= {"+","-","*","/","(",")"};
		String[] otherop = {"=",";","{","}","/*","*/","[","]"};
		nh.put(preservedarr, 2);
		nh.put(logicop,4);
		nh.put(mathop, 5);
		nh.put(otherop, 6);
		
		
	}
	
	public static void main(String[] args) throws IOException {
		// TODO Auto-generated method stub

		/*
		 * 
		 * 
		BufferedReader rd = new BufferedReader(new InputStreamReader(System.in));
		String test2 = "";
		test2 = rd.readLine();
		
		
		
		
		String[] test = test2.split("\\s+");
		
		Pattern pn = Pattern.compile("([a-zA-Z0-9]|[.])+");//number,variable,illegic variable
		Pattern pn1= Pattern.compile("[a-zA-Z]([a-zA-Z0-9]*)");//variable
		Pattern pn2= Pattern.compile("([0]|([1-9]([0-9]*)))[.]([0-9]+)|[0]|([1-9]([0-9]*))");//number
		Pattern pn3= Pattern.compile("[^\\.a-zA-Z0-9][^\\.a-zA-Z0-9]");//两个非词
		Pattern pn4= Pattern.compile("[^\\.a-zA-Z0-9]");//a non-word
		
		
		//非字母字符：先查两再查一
		for(String str : test){
		   
			String work = str;	
			
			while(work.length()>0){
				try{
				Matcher m = pn.matcher(work);
			if(m.lookingAt()){//是数字或variable
				Matcher m1= pn1.matcher(m.group());
				Matcher m2 = pn2.matcher(m.group());
				
				if(m1.matches()){//是合法变量组合
					
					System.out.println(m.group());//在表里找得到的，是保留字
					
				}else if(m2.matches()){//是数字组合
					System.out.println(m2.group());
				}else{//是字母数字，或数字，但不是合法组合
					work=m.replaceFirst("");//将匹配部分去掉
					throw new FormatException(m.group());
				}
				work=m.replaceFirst("");//将匹配部分去掉
			}else if(work.length()>1){//1.不是数字字母2。还有两个以上剩余
				Matcher m3 = pn3.matcher(work);
				Matcher m4 = pn4.matcher(work);//一个non
				if(m3.lookingAt()){//若两个非字母数字相连
					if (m3.group().equals("{{")){//若他们合在一起是个东西
						System.out.println(m3.group());
						work=m3.replaceFirst("");//将匹配部分去掉
					}//end if
					else if(m4.lookingAt()){//在只有一个是non字母数字时
						
						if(m4.group().equals("*")){//若他是可以找到的
							System.out.println(m4.group());
							work=m4.replaceFirst("");//将匹配部分去掉
						}else{work=m4.replaceFirst("");//将匹配部分去掉 
							throw new FormatException(m4.group());}//若找不到，就报错
						
					
				}//end else if
					
				}else if(m4.lookingAt()){//////添加
					if(m4.group().equals("*")){
						System.out.println(m4.group());
						work=m4.replaceFirst("");//将匹配部分去掉
					}else{work=m4.replaceFirst("");
					throw new FormatException(m4.group());}//若找不到，就报错
					
				}
			}else if (work.length()==1){//1.不是数字字母2。只有一个剩余
				Matcher m5 = pn4.matcher(work);
				if(m5.lookingAt()){
					if(m5.group().equals("*")){
						System.out.println(m5.group());
						work=m5.replaceFirst("");//将匹配部分去掉
					}else{//1.only one left which is not a num 2.it can't be found
						work=m5.replaceFirst("");//将匹配部分去掉
						throw new FormatException(m5.group());
					}
				}
			}else{//这里其实不可达
				
				throw new lexicalError(work.substring(0,10)+"...");
				
			}//end else
			
			
			}catch(FormatException fe){System.out.println(fe);}
		catch(lexicalError le){System.out.println(le);}
				
				//end try
			
		}//end while
		}//end for
    	 
		*///test it
		Pattern pn7= Pattern.compile("/\\*.+\\*/");
		BufferedReader rd = new BufferedReader(new InputStreamReader(System.in));
		Matcher m7 = pn7.matcher(rd.readLine());
		if(m7.find())
		System.out.println(m7.group());
		
		}
		
		
	}