lexer.java

编译器中的词法分析

					tk = new Token (INT, charBuffer, tokenColumn, tokenLine);
				}
				//如果后一位是.
				else if (LA(1) == '.')
				{
					charBuffer += nextChar();
					
					//如果后一位是终结符
					if (isEndChar(LA(1)))
					{
						tk = new Token (INVALIDNUM, charBuffer, tokenColumn, tokenLine);
					}
					//如果后一位是数字
					else if (Character.isDigit(LA(1)))
					{
						charBuffer += nextChar();
						boolean key = true;
						boolean keepNum = true;
						
						//持续读取字符，直到结束符为止
						while(key) 
						{
							//如果下一个字符是结束符
							if (isEndChar(LA(1)))
							{
								//如果是合法数字串
								if (keepNum)
								{
									tk = new Token (REAL, charBuffer, tokenColumn, tokenLine);
									key = false;
								}
								//出现其他字符
								else 
								{
									tk = new Token (INVALIDNUM, charBuffer, tokenColumn, tokenLine);
									key = false;
								}
							}
							//如果下个字符是数字
							else if (Character.isDigit(LA(1)))
							{
								charBuffer += nextChar();
							}
							//如果下个字符是其他字符
							else 
							{
								charBuffer += nextChar();
								keepNum = false;
							}
						}
					}
					//如果后一位是其他
					else 
					{					
						charBuffer += nextChar();
						boolean key = true;	
						
						//持续读取其他符号，直至遇到终结符停止，输出字符串
						while(key){
							if (isEndChar(LA(1))) 
							{							
								tk = new Token (INVALIDNUM, charBuffer, tokenColumn, tokenLine);
								key = false;
							}
							else  
							{
								charBuffer += nextChar();
							}
						}							
					}
				}
				//如果后一位是数字
				else if (Character.isDigit(LA(1)))
				{
					charBuffer += nextChar();					
					boolean keyIsDigit = true;					
					
					//持续读入数字字符
					while (keyIsDigit)
					{
						if (Character.isDigit(LA(1)))
						{
							charBuffer += nextChar();	
						}
						else
						{
							keyIsDigit = false;
						}
					}
					
					//如果下一个字符是终结符
					if (isEndChar(LA(1)))
					{
						tk = new Token (INT, charBuffer, tokenColumn, tokenLine);
					}
					//如果下一个字符是.
					else if (LA(1)== '.')
					{
						charBuffer += nextChar();
						
						//如果下一个字符为终结符
						if (isEndChar(LA(1))) 
						{
							tk = new Token (INVALIDNUM, charBuffer, tokenColumn, tokenLine);
						}
						else if (Character.isDigit(LA(1)))
						{
							charBuffer += nextChar();
							boolean key = true;
							boolean keepNum = true;
							
							while(key) 
							{
								if (isEndChar(LA(1))) 
								{	
									if (keepNum)
									{
										tk = new Token (REAL, charBuffer, tokenColumn, tokenLine);
										key = false;
									}
									else 
									{
										tk = new Token (INVALIDNUM, charBuffer, tokenColumn, tokenLine);
										key = false; 
									}
								}
								else if (Character.isDigit(LA(1)))
								{
									charBuffer += nextChar();
								}
								else 
								{
									keepNum = false;
									charBuffer += nextChar();
								}
							}
						}
						else
						{				
							charBuffer += nextChar();
							boolean key = true;	
							
							//持续读取其他符号，直至遇到终结符停止，输出字符串
							while(key){
								if (isEndChar(LA(1))) 
								{							
									tk = new Token (INVALIDNUM, charBuffer, tokenColumn, tokenLine);
									key = false;
								}
								else  
								{
									charBuffer += nextChar();
								}
							}						
						}
					}
					//如果下一个字符其他字符
					else  {			
						charBuffer += nextChar();
						boolean key = true;
						
						//持续读取其他符号，直至遇到终结符停止，输出字符串
						while(key){
							if (isEndChar(LA(1))) 
							{							
								tk = new Token (INVALIDNUM, charBuffer, tokenColumn, tokenLine);
								key = false;
							}
							else  
							{
								charBuffer += nextChar();
							}
						}
					}
					
				}
				//如果后一位是其他
				else 
				{
					charBuffer += nextChar();
					boolean key = true;	
					
					//持续读取其他符号，直至遇到终结符停止，输出字符串
					while(key){
						if (isEndChar(LA(1))) 
						{							
							tk = new Token (INVALIDNUM, charBuffer, tokenColumn, tokenLine);
							key = false;
						}
						else  
						{
							charBuffer += nextChar();
						}
					}			
				}
				
				break;				
		}
		return tk;		
	}
	
	/**
	 * 判断是否是标识符或是关键字
	 * @param startPlace
	 * @param charCh
	 * @return
	 */
	public Token isIdentifier(int startPlace,char curCh) {
		Token tk = null;
		String charBuffer = Character.toString(curCh);
		//下一个字符是终结符
		if(isEndChar(LA(1)))
		{
			tk = new Token(ID,charBuffer,tokenColumn,tokenLine);
		}
		//下一个字符是数字，字母或下划线
		else if(Character.isDigit(LA(1)) || ('a' <= LA(1) && 'z' >= LA(1)) || ('A' <= LA(1) && 'Z' >= LA(1))
				|| LA(1) == '_')
		{
			charBuffer += nextChar();
			boolean isID = true;
			boolean key = true;
			
			//持续读取下一个字符直到终结符
			while(key)
			{
				//下一字符依然是数字，字母，下划线
				if(Character.isDigit(LA(1)) || ('a' <= LA(1) && 'z' >= LA(1)) || ('A' <= LA(1) && 'Z' >= LA(1))
						|| LA(1) == '_')
				{
					charBuffer += nextChar();
				}
				//下一个字符是终结符
				else if(isEndChar(LA(1)))
				{
					if(isID)
					{
						if(LA(0) == '_')
						{
							tk = new Token(INVALIDID,charBuffer,tokenColumn,tokenLine);
							key = false;
						}
						else
						{
							//判断是否是关键字
							if(charBuffer.equals("char"))
							{
								tk = new Token(CHAR,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("string"))
							{
								tk = new Token(STRING,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("bool"))
							{
								tk = new Token(BOOL,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("int"))
							{
								tk = new Token(INT,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("real"))
							{
								tk = new Token(REAL,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("if"))
							{
								tk = new Token(IF,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("else"))
							{
								tk = new Token(ELSE,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("while"))
							{
								tk = new Token(WHILE,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("break"))
							{
								tk = new Token(BREAK,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("continue"))
							{
								tk = new Token(CONTINUE,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("return"))
							{
								tk = new Token(RETURN,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("write"))
							{
								tk = new Token(WRITE,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("read"))
							{
								tk = new Token(READ,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("writeln"))
							{
								tk = new Token(WRITELN,charBuffer,tokenColumn,tokenLine);
							}
							else if(charBuffer.equals("void"))
							{
								tk = new Token(VOID,charBuffer,tokenColumn,tokenLine);
							}
							//为标识符
							else
							{
								tk = new Token(ID,charBuffer,tokenColumn,tokenLine);
							}
							key = false;
						}
					}
					else
					{
						tk = new Token(INVALIDID,charBuffer,tokenColumn,tokenLine);
						key = false;
					}
				}
				//下一个字符是其他符号
				else
				{
					charBuffer += nextChar();
					isID = false;
				}
			}
			
		}
		//下一个字符是其他字符
		else
		{
			charBuffer += nextChar();
			boolean key = true;	
			
			//持续读取其他符号，直至遇到终结符停止，输出字符串
			while(key){
				if (isEndChar(LA(1))) 
				{							
					tk = new Token (INVALIDNUM, charBuffer, tokenColumn, tokenLine);
					key = false;
				}
				else  
				{
					charBuffer += nextChar();
				}
			}	
		}
		return tk;
	}
	
	/**
	 * 判断是否是非法字符串
	 * @param startPlace
	 * @param charCh
	 * @return
	 */
	public Token isSpecialSymbol(int startPlace,char curCh) {
		
		Token tk = null;
		String charBuffer = Character.toString(curCh);
		boolean key = true;
		
		//持续读取其他符号，直至遇到终结符停止，输出字符串
		while(key){
			if (isEndChar(LA(1))) 
			{							
				tk = new Token (INVALIDID, charBuffer, tokenColumn, tokenLine);
				key = false;
			}
			else  
			{
				charBuffer += nextChar();
			}
		}	
		return tk;
	}
	
	/**
	 * 判断当前字符是否为终结符
	 * @param ch
	 * @return
	 */
	public boolean isEndChar(char ch){
		
		if (ch == ',' || ch == ';' || ch == '\'' || ch == '"' || ch == '(' 
			|| ch == ')' || ch == '[' || ch == ']' || ch == '{' || ch == '}'
			|| ch == '+' || ch == '-' || ch == '*' || ch == '/' || ch == '='
			|| ch == '!' || ch == '>' || ch == '<' || ch == '&' || ch == '|'
			|| ch == '#' || ch == ' ' || ch == '\r' || ch == '\n' || ch == '\t'
			|| ch == 'ん') 
		{
			
			return true;
		} 
		else 
		{
			return false;
		}
	}

}