lexical.cs

syntax check 语法分析器 简单

using System;
using System.IO;
using System.Collections;

namespace Syntax_Analyzer
{
	/// <summary>
	/// Lexical 的摘要说明。
	/// </summary>
	public class Lexical
	{

		
   		
		private int rnum;//行号
		private int rpos;//列号
		private static ArrayList symbolTable;
		private StreamReader br;
		private char c;
		private char bC;
		private int state;
		private String str;

		public Lexical(StreamReader br)
		{	
			symbolTable = new ArrayList();
			this.br=br;
			c=' ';
			bC=' ';
			state = 0;
			rnum=1;
			rpos=0;
			str="";
		}

		public ArrayList getSymbolTable()
		{
			c = nextChar();
			state = 0;
			Console.WriteLine("-----------------------------词法分析开始---------------------------------\n");
			analyzer();
			Console.WriteLine("-----------------------------词法分析结束---------------------------------\n");
			return symbolTable;
		}

		
		private void analyzer()
		{
			bool isRunning=true;

			while(isRunning)
			{
				switch (state)
				{
					case 0:
					
						if(c==' ')            //空格
						{
                    
						}
						else if(c=='\t')      //tab
						{
							rpos+=3;
						}
						else if((int)c==13)   //空格
						{
                    
						}
						else if((int)c==10)   //回车
						{
							rpos = 0;
							rnum++;
						}
						else if(c=='~')       //结束符
						{
							return;
						}
						else if(c=='+')       //操作符+
						{
							install("操作符","+",rpos,rnum);
						} 
						else if(c=='-')       //操作符-
						{
							install("操作符","-",rpos,rnum);
						}
						else if(c=='/')       //操作符/
						{
							state = 1;
						}
						else if(c=='*')       //操作符*
						{
							install("操作符","*",rpos,rnum);
						}
						else if(c=='=')       //操作符=
						{ 
							state = 2;
						}
						else if(c=='<')       //操作符<
						{
							state = 3;
						}
						else if(c=='>')       //操作符>
						{
							state = 4;
						}
						else if(c=='!')       //操作符!
						{
							state = 5;
						}
						else if(c=='{')       //分隔符{
						{
							install("分隔符","{",rpos,rnum);
						}
						else if(c=='}')       //分隔符}
						{
							install("分隔符","}",rpos,rnum);
						}
						else if(c=='(')		  //分隔符(
						{
							install("分隔符","(",rpos,rnum);
						}
						else if(c==')')		  //分隔符)
						{
							install("分隔符",")",rpos,rnum);
						}
						else if(c==';')       //分隔符;
						{
							install("分隔符",";",rpos,rnum);
						}
						else if((int)c==65535)
						{
							isRunning=false;
						}
						else if(isLetter(c))  //读到了字符
						{
							bC = c;
							state = 6;
						}
						else if(isDigit(c))   //读到了数字符
						{
							bC = c;
							str="";
							state = 7;
						}
						else if(c=='#')
						{
							state=13;
						}
						else 
						{
							state=0;
							isRunning=false;
							fail(0);
						}
                       
						c = nextChar();
						break;
					case 1:
						if(c=='/')             //表示注释后面的东西,不读它
						{
							while((int)c!=10)
							{
								c = nextChar();
							}
							rpos=0;           //从下一行开始
							rnum++;
							c = nextChar();
							state = 0;
						}
						else 
						{
							state = 0;        // 下一个不是/, 后退一步到state=0
							install("操作符","/",rpos-1,rnum);
						}

						break;
					case 2:
                
						if(c=='=')           //表示操作符==
						{
							state = 0;
							install("操作符","==",rpos-1,rnum);
							c = nextChar();
						}
						else 
						{
							state = 0;
							install("操作符","=",rpos-1,rnum);
						}

						break;
					case 3:
						if(c=='=')          //表示操作符<=
						{
							state = 0;
							install("操作符","<=",rpos-1,rnum);
							c = nextChar();
						}
						else 
						{
							state = 0;
							install("操作符","<",rpos-1,rnum);
						}
                
						break;
					case 4:
						if(c=='=')          //表示操作符>=
						{
							state = 0;
							install("操作符",">=",rpos-1,rnum);
							c = nextChar();
						}
						else 
						{
							state = 0;
							install("操作符",">",rpos-1,rnum);
						}

						break;
					case 5:
						if(c=='=')         // 表示操作符!=
						{
							state = 0;
							install("操作符","!=",rpos-1,rnum);
							c = nextChar();
						}
						else 
						{
							state = 0;
							isRunning=false;
							fail(1);
						}
						break;
				
						//***********************************
						//读入标识符,后面是 数字符和字母才是对的
						//***********************************
					case 6:                      
						String id = ""+bC;
						while(isLetter(c)||isDigit(c))
						{
							id+=c;
							c = nextChar();
						}
						if(isKey(id))
							install("关键字",id,rpos-id.Length,rnum);
						else
							install("标识符",id,rpos-id.Length,rnum);
						state = 0;
						break;
						//*********************************************
						//读入数字符,后面是 数字符(循环); .(转状态);E(转状态)
						//**********************************************
					case 7:                         
						str+=bC; 
						while(isDigit(c))            
						{
							str+=c;
							c=nextChar();
						}
						if(c=='.')                  
						{
							state=8;
							c=nextChar();         
						}
						else if(c=='E')              
						{
							state=10;
							c=nextChar();
						}
						
							//需要判断是否是  11w等
						else if(isLetter(c))
						{
							state = 0;
							isRunning=false;
							fail(4);
                        
						}
						else                        
						{
							state = 0;
							install("数字符",str,rpos-str.Length,rnum);
						}
						break;
						//***********************************
						//  " ." 后面跟的字符,除了数字符其它都出错
						//***********************************
					case 8:                         
						str+='.';
						if(isDigit(c))
						{
							state = 9;
							str+=c;
							c=nextChar();
						}
						else
						{
							state = 0;
							isRunning=false;
							fail(2);
						}
						break;
						//***********************************************************************************
						//    ".数字符" 后面字符,可以是数字符或者是E(需要加错误提示,当不是空格而是其他字母时候)
						//************************************************************************************
					case 9:
                
						while(isDigit(c))
						{
							str+=c;
							c=nextChar();
						}
						if(c=='E')
						{
							state = 10;
							c=nextChar();
						}

							//需要判断是否是  11w 等
						else if(isLetter(c))
						{
							state = 0;
							isRunning=false;
							fail(2);
                        
						}
						else
						{
							state = 0;
							install("数字符",str,rpos-str.Length,rnum);
						}
						break;

						//*********************************************
						// "E"  后面字符,  "+";"-";"数字符" 都是对的,其他都是错的
						//********************************************
					case 10:
						str+='E';
						if(c=='+'||c=='-')
						{
							state=11;
							str+=c;
							c=nextChar();
						}
						else if(isDigit(c))
						{
							state=12;
							str+=c;
							c=nextChar();
						}
						else
						{
							state=0;
							isRunning=false;
							fail(2);
						}
						break;
						//*****************************************
						//   "+" "-" 后面字符, 数字符是对的.其他都是错的
						//*****************************************
					case 11:
						if(isDigit(c))
						{
							state=12;
							str+=c;
							c=nextChar();
						}
						else
						{
							state = 0;
							isRunning=false;
							fail(2);
						}
						break;
						//***********************************
						// "E数字符"  后面字符,只有数字符是对的
						//*********************************
					case 12:
						while(isDigit(c))
						{
							str+=c;
							c=nextChar();
						}

						if(isLetter(c))
						{
							state = 0;
							isRunning=false;
							fail(2);
                        
						}
						install("数字符",str,rpos-str.Length,rnum);
						state=0;
						break;


					case 13: //声明部分
 
						id ="#";
						while(isLetter(c))
						{
							id+=c;
							c = nextChar();
						}
						if(id=="#include"&&c=='<')
						{
										    
							id+=c;
							c = nextChar();
							while(isLetter(c)||this.isDigit(c))
							{
							
								id+=c;
								c = nextChar();
							}
							if(c=='>')
							{
								id+=c;
								install("文件引入",id,rpos-id.Length,rnum);
								c = nextChar();
							}
							else
								fail(3);
                                         
						}
						else
							fail(0);
						state = 0;
						break;
					
				}  
			}
		}
 
		private void fail(int i) 
		{
			string typeerror="";;
			switch(i)
			{
				case 0:
					typeerror="输入非法字符错误";
					break;
				case 1:
					typeerror="操作符错误";
					break;

				case 2:
					typeerror="数字符错误";
					break;

				case 3:
					typeerror="声明错误";
					break;
				case 4:
					typeerror="标识符错误";
					break;

			}
			
			Console.Write(typeerror+",位置<"+rnum+","+rpos+">");
			Console.WriteLine();
		}

		private char nextChar()
		{
			char t;
			rpos++;
			if(br.Peek() == -1)
			{
				br.Close();
				return '~';
			}

			t= (char) br.Read();
			return t;
		}
    
		private void install(String tType,String aValue,int rpos,int rnum)
		{
			ArrayList tokenArray = new ArrayList();
			tokenArray.Add(tType);
			tokenArray.Add(aValue);
			tokenArray.Add("<"+rnum.ToString()+","+rpos.ToString()+">");
			symbolTable.Add(tokenArray);

			ArrayList tempArray=tokenArray;

			for(int j=0;j<tempArray.Count;j++)
			{
				Console.Write(tempArray[j].ToString()+"     ");
			}
			Console.WriteLine();

			str="";
		}
    
		private bool isLetter(char checkChar)
		{
			int checkInt = (int)checkChar;
			if((checkInt<=122&&checkInt>=97)||(checkInt>=65&&checkInt<=97))
			{
				return true;
			}
			return false;
		}
    
		private bool isDigit(char checkChar)
		{
			int checkInt = (int)checkChar;
			if(checkInt<=57&&checkInt>=48)
			{
				return true;
			}
			return false;
		}
    
		private bool isKey(String checkKey)
		{
			if(checkKey.Equals("if")||checkKey.Equals("then")||
				checkKey.Equals("else")||checkKey.Equals("int")||
				checkKey.Equals("real")||checkKey.Equals("while"))
			{
				return true;
			}
			return false;
		}
	}
}