word.cpp

1、编译原理的词法分析 2、根据输入的文件

#include ".\word.h"
#using <mscorlib.dll>

Word::Word(void)
{inword="void main int short char float double long unsigned struct sizeof auto static register extern typedef union enum if else  goto while do switch case  return break continue for";
 sign="+ - * / % ++ -- += -= *= /= %= = [ ] & ^ | ~ << >> ! ? : < >  . && || == >= <=  ->";
 marco="#define #undefine #endif #include #ifdef #else #ifndef";
 bound="; { } ( ) \" \' , ";
 word[0]='\0';
 flag=0;//默认不写入程序源码注释
 flagto_txt=0;//注释形式 /* */  的开关
}

bool Word::setFlag()
{
	flag=!flag;return 1;
}
bool Word::isSign(char frist,char second)
//功能:检查字符是不是属于运算符集  不是返回0  是返回1
{	
	char arry[3];
	arry[0]=frist;
	arry[1]=second;
	arry[2]='\0';
	//cout<<arry<<endl;getch();
	if(sign.isIn(arry)) return 1;
	else return 0;
}
bool Word::isBound(char data)
//功能:判断字符是否是界符,是返回1 否则返回0 
{
	char arry[2];
	arry[0]=data;
	arry[1]='\0';
	if(bound.isIn(arry)) return 1;
	else return 0;
}
bool Word::isMarco(char *arry)
//功能:判断是否是预编译命令 是返回1 否则返回 0 
{	
	if(marco.isIn(arry) ) return 1;
	else return 0;
}
bool Word::isInWord(char *arry)
//检查字符串是否是保留字 是返回1  不是返回 0
{
	if(inword.isIn(arry) ) return 1;
	else return 0;
}
bool Word::isNum(char data)
{//判断是否是数字字符
	if(data>='0'&&data<='9') return 1;
	else return 0;
}
bool Word::isWord(char data)
{//判断是否是字母
	if( (data>='a'&&data<='z')||(data>='A'&&data<='Z') ) return 1;
	else return 0;
}

bool Word::outPut(char *word,char *strings)
{
	//功能:将word和strings指向的字符串写入文件
	outfile.write(word,strlen(word));// write to file
	outfile.put('\t');//写入一个制表符
	if(strings)outfile.write(strings,strlen(strings));
	outfile.put(10);//写入一个回车
	return 1;
}
bool Word::outPut(char data,char *strings)
{	//功能:将字符data写进文件
	outfile.put(data);
	outfile.put('\t');
	outfile.write(strings,strlen(strings));
	outfile.put(10);
	return 1;
}
bool Word::outPut(char data1,char data2,char *strings)
{	
	//将字符data1和字符data2和属性文字strings写进文件
	outfile.put(data1);
	outfile.put(data2);
	outfile.put('\t');
	outfile.write(strings,strlen(strings));
	outfile.put(10);
	return 1;
}

bool Word::infIn()
{	outputfilename[0]='\0';
	ifstream infile;//输入文件流对象
	char filename[30];//存放文件名
	cout<<"[输入一个供分析的C语言源文件:]"<<endl;
	cout<<"	[源程序文件名]:";
	cin>>filename;
	infile.open(filename,ios_base::in);
	if(infile==NULL)
	{
		cout<<"	[打开源文件失败]:请检查文件名是否输入正确.";getch();return 0;
	}
	cout<<"[输入保存分析结果的文件名:]"<<endl;
	while(1)
	{
		cout<<"	[保存到文件]:";
		cin>>outputfilename;
		ifstream outtext(outputfilename,ios_base::in);
		if(outtext)
		{
			cout<<"	[警告]:文件已经存在,要覆盖吗?[y/n]";
			char c;
			c=getch();
			if(c=='y'){outtext.close();break;}//同意覆盖
			continue;//重新输入文件名
		}
		outtext.close();
		break;
	}//while(1)
	//将输入的文件名与私有数据 ofstream outfile 挂钩
	outfile.open(outputfilename,ios_base::out);
	if(!outfile)
	{
		cout<<"[错误]:输出文件创建失败."<<endl;getch();exit(0);
	}
	outPut("[词法分析的结果]");
	outPut("[分析 的 文 件:]",filename);
	outPut("[当  前  文 件:]",outputfilename);
	outPut("[分析结 果如下:]");
	char linebuffer[81];//行缓冲区
	
	system("cls");
	cout<<"[开始对文件进行分析:]"<<endl;
	while(infile.good() )
	{
		infile.getline(linebuffer,80,10);	//读一行到行缓冲区
		cout<<"//"<<linebuffer<<endl;
		if(flag) outPut("//",linebuffer);
		//cout<<endl;
		wordAnalyze(linebuffer);			//对这一行做词法分析
	}
	//getch();
	infile.close();
	outfile.close();
	return 1;
}

bool Word::wordAnalyze(char *line)
{//功能:对一行做词法分析
	if( strlen(line) == 0) return 1;//不对空行做词法分析

	char *p;
	bool sign=0;
	p=line;
	int i;

	while(*p)
	{
		if(*p==' '||*p==10||*p=='\t'){p++;continue;}//过滤空格和回车符
		if(*p=='/'&& *(p+1)=='/') return 1;//过滤掉注释
		if(*p=='/'&& *(p+1)=='*') 
		{
			flagto_txt=!flagto_txt;p=p+2;
		}
		if(flagto_txt)
		{
			while(*p&& !( *p=='*' && *(p+1)=='/') ) p++;
			if(! *p) return 0;//还没有出现与/* 匹配的 */ 返回
			flagto_txt=!flagto_txt;//出现与/*匹配的*/ 继续
			p=p+2;//跳过 */
			if(! *p) return 1;
		}
		if(*p=='#')
		{
			i=0;
			while(*p!=' '&&*p && *p!='<')
			{
				word[i]=*p;
				p++;i++;
			}
			word[i]='\0';
			if(marco.isIn(word)) 
			{
				cout<<word<<"			"<<"预编译命令"<<endl;
				outPut(word,"预编译命令");
			}
			else 
			{
				cout<<word<<"			"<<"非法形式"<<endl;
				outPut(word,"非法形式");
			}
			if(! *p) return 1;
			else continue;
		}
		else if( isWord(*p) ) //是字母
		{	i=0;			
			while(isWord(*p) && *p )//是保留字或者是标识符
			{
				word[i]=*p;
				p++;i++;
			}
			if(*p=='_')
				while(*p!=' '&&*p!=10&& *p&&!isBound(*p) )
				{
					word[i]=*p;
					p++;i++;
				}
			word[i]='\0';//分离出一个词
			//cout<<word<<endl;getch();
			if(isInWord(word))
			{
				cout<<word<<"			"<<"保留字"<<endl;
				outPut(word,"保留字");
			}
			else	
			{
				outPut(word,"标识符或者过程名");
				cout<<word<<"			"<<"标识符或过程名"<<endl;
			}
				continue;
		}//else if
		else if( isNum(*p) )//是数字
		{	i=0;
			while(*p=='.'||isNum(*p) )
			{	
				word[i]=*p;
				i++;
				p++;
			}
			if(*p=='_')
			{
				while(*p!=' '&&*p!=10 && *p&& !isSign(*p) )
				{
					word[i]=*p;
					p++;i++;
				}
				word[i]='\0';
				cout<<word<<"			"<<"字符常量或标识符"<<endl;//分离出数
				outPut(word,"字符常量或标识符");
			}
			else
			{
				word[i]='\0';
				cout<<word<<"			数据常量"<<endl;
				outPut(word,"数据常量");
			}
			continue;
		}//else if
		else if(isSign(*p))				//是运算符
		{	if( *(p+1) && isSign( *p,*(p+1))) //双字符运算符
			{
				cout<<*p<<*(p+1)<<"			"<<"运算符"<<endl;
				outPut(*p,*(p+1),"运算符");
				p=p+2;
				continue;
			}
			else//单字符运算符
			{
				cout<<*p<<"			"<<"运算符"<<endl;
				outPut(*p,"运算符");
				p++;
				//getch();
			}//else
			if(!*p) return 1;//完成
			continue;
		}//if
		else if(isBound(*p)) //如果是分界符
		{
			if(*p=='\"')
				{	cout<<*p<<"			分界符"<<endl;
					outPut(*p,"分界符");
					sign=!sign;//引号开关 1:前引号  0 后引号
					if(sign)
					{	i=0;
						p++;//跳过前引号
						while(*p!='\"'&&*p!=10)
						{
							word[i]=*p;
							p++;i++;
						}
						word[i]='\0';
						cout<<word<<"			字符串常量"<<endl;
						outPut(word,"字符串常量");
					}//if
					if(*p=='\"')
					{	cout<<*p<<"			分界符"<<endl;
						outPut(*p,"分界符");
						p++;//跳过后引号
					}
				}//if
			else 
			{
				cout<<*p<<"			分界符"<<endl;
				outPut(*p,"分界符");
				p++;
			}
			if(! *p) return 1;
			else continue;
		}//else if
		else if(*p=='_')
		{	i=0;
			while(*p!=' '&& *p!=10 && *p &&!isSign(*p) )
			{
				word[i]=*p;
				p++;i++;
			}
			word[i]='\0';
			cout<<word<<"			字符常量或标识符"<<endl;
			outPut(word,"字符常量或标识符");
			continue;
		}
		else 
		{	if(*p!=10) cout<<*p<<"			"<<"非法字符"<<endl;
			outPut(*p,"非法字符");
			p++;
		}//else
	}//while(*p);
	return 1;
}

bool Word::forMain()
{
	char c[10];
	while(1)
	{
		system("cls");
		cout<<"		[C语言词法分析程序]"<<endl;
		cout<<"		[孤单北半球@2005_12_28]"<<endl;
		cout<<"		[go]	词法分析."<<endl;
		cout<<"		[exit]	退出."<<endl;
		cout<<"		[detel]	是否向输出文件写入注释形式的源码."<<endl;	
		cout<<"□□□□□□□□□□□□□□□□□□□□□□□□□□□□□□□"<<endl;
		if(flag)cout<<"		[细节显示]"<<endl;
		else cout<<endl;
		cout<<"		[输入命令]:";
		cin>>c;
		if(strcmp(c,"go")==0)  
		{	system("cls");
			cout<<"[词法分析:]"<<endl;
			infIn();
			cout<<"[任意键返回到菜单...]";
			getch();
		} 
		else if(strcmp(c,"exit")==0)break;
		else if(strcmp(c,"detel")==0) setFlag();
		else 
		{
			cout<<"[错误]:"<<c<<"不是一个预定的命令";
			getch();
		}
	}//while(1)
	return 1;
}//forMain();