词法分析.txt

词法分析器

#include<iostream.h>
#include<string.h>
#include<ctype.h>
const  int maxsize=500;//输入程序的总容量
struct table
{
	char keyword[20];
	int	no;	
};
struct table tab[37]={{"and",1},{"begin",2},{"bool",3},{"do",4},{"else",5},{"end",6},{"false",7},{"if",8},
					{"integer",9},{"not",10},{"or",11},{"program",12},{"real",13},{"then",14},{"true",15},
					{"var",16},{"while",17},{"标识符",18},{"整数",19},{"实数",20},{"(",21},{")",22},{"+",23},
					{"-",24},{"*",25},{"/",26},{".",27},{",",28},{":",29},{";",30},{":=",31},{"=",32},
					{"<=",33},{"<",34},{"<>",35},{">",36},{">=",37}
					};//关键字表
struct table1
{
	char word[20];
};
struct table1 stab[50];//符号表
struct table1 ctab[50];//常数表
int slen=0,clen=0;		  //记录符号表长度，和常数表长度
/*-------------------------------------------------*/
char getchar(char s[],int &i)
{
	return s[++i];
}
/*-------------------------------------------------*/
char getbc(char s[],int &i)
{
	char ch;
	ch=s[i];
	while(ch==' '||ch=='\r')
		ch=s[++i];
	return ch;
}
/*-------------------------------------------------*/
void concat(char s[],char ch)
{
	int i;
	
	i=strlen(s);//cout<<i<<endl;
	s[i]=ch;//cout<<s<<endl;
}
/*-------------------------------------------------*/
void concat1(char s[],char ch)
{
	for(int i=0;s[i];i++)
		s[i]=ch;
	s[++i]='\0';
}
/*-------------------------------------------------*/
int reserve(char s[])//返回字符串中关键字编码  否则返回0
{
	for(int i=0;i<37;i++)
	{
		if(strcmp(s,tab[i].keyword)==0)//串比较
			return tab[i].no;
	}
	return 0;
}
/*-------------------------------------------------*/
char retract(char s[],int &i)//回调一个字符位置
{
	return s[--i];
}
/*-------------------------------------------------*/
char pretract(char s[],int &i)//后退一个字符位置
{
	return s[++i];
}
/*-------------------------------------------------*/
int insertid(char s[])
{
	for(int i=0;i<slen;i++)
	{
		if(strcmp(stab[i].word,s)==0) return i+1;//查找标识符在符号表中的位置
	}
	strcpy(stab[i].word,s);//将标识符插到符号表中
	slen++;
	return i+1;
}
/*-------------------------------------------------*/
int insertconst(char s[])//常数表
{
	for(int i=0;i<clen;i++)
	{
		if(strcmp(ctab[i].word,s)==0) return i+1;
	}
	strcpy(ctab[i].word,s);
	clen++;
	return i+1;
}
/*-------------------------------------------------*/
void scan(char s[],int &i)
{
	int code,value;//分别代表机内码和入口地址
	//static  int value=1;
	static int no=0;//代表输入序列
	char strtoken[10]="";
	char ch;
	ch=getchar(s,i);
	ch=getbc(s,i);
	if(isalpha(ch))//当变元为字母表中的字母时，函数返回非0值，否则返回0.
	{
		while(isalpha(ch)||isdigit(ch))//变元为字母或数字均可
		{
			concat(strtoken,ch);
			ch=getchar(s,i);
			//cout<<ch<<endl;
		}
		ch=retract(s,i);//2状态退回一个字符
		code=reserve(strtoken);
		if(code==0)
		{ 
			no++;
			value=insertid(strtoken);
			cout<<"<"<<no<<"\t  "<<strtoken<<"\t    18\t "<<'   '<<value<<"\t   >"<<endl;
		}
		else
		{ 
			no++;
			cout<<"<"<<no<<"\t  "<<strtoken<<"\t    "<<code<<"\t-1\t   >"<<endl;
		}
	}
	else if(isdigit(ch))//当变元为数字时
	{
		while(isdigit(ch))
		{
			concat(strtoken,ch);
			ch=getchar(s,i);
		}
        if(isalpha(ch))
		{
            cout<<"第"<<++no<<"个符号是非法的以数字开头的字母串!"<<endl;
			cout<<"正确的格式如下:"<<endl;
		    ch=retract(s,i);
			value=insertconst(strtoken);
		    cout<<"<"<<no<<"\t  "<<strtoken<<"\t    19\t "<<value<<"\t   >"<<endl;
		}
		else
		{
		   if(ch='.')
		   {
			 concat(strtoken,ch);
			 ch=getchar(s,i);
			 while(isdigit(ch))
			 {
				concat(strtoken,ch);
				ch=getchar(s,i);
			 }
			 if(ch='.')
				ch=retract(s,i);
				value=insertconst(strtoken);
				no++;
				cout<<"<"<<no<<"\t  "<<strtoken<<"\t    20\t "<<value<<"\t   >"<<endl;
		   }
		   else
		   {
			   no++;
		       ch=retract(s,i);
		       value=insertconst(strtoken);
		       cout<<"<"<<no<<"\t  "<<strtoken<<"\t   19\t "<<value<<"\t   >"<<endl;
		   }
		}
	}
	else if(ch=='(')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<"(\t    21\t     -1\t>"<<endl;
	}
	else if(ch==')')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<")\t    22\t-1\t   >"<<endl;
	}
	else if(ch=='+')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<"+\t    23\t-1\t   >"<<endl;
	}
	else if(ch=='-')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<"-\t    24\t-1\t   >"<<endl;
	}
	else if(ch=='*')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<"*\t    25\t-1\t   >"<<endl;
	}
	else if(ch=='/')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<"/\t    26\t-1\t   >"<<endl;
	}
	else if(ch=='.')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<".\t    27\t-1\t   >"<<endl;
	}
	else if(ch==',')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<",\t    28\t-1\t   >"<<endl;
	}
	else if(ch==':')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<":\t    29\t-1\t   >"<<endl;
	}
	else if(ch==';')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<";\t    30\t-1\t   >"<<endl;
	}
	else if(ch==':=')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<":=\t    31\t-1\t   >"<<endl;
	}
	else if(ch=='=')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<"=\t    32\t-1\t   >"<<endl;
	}
	else if(ch=='<=')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<"<=\t    33\t-1\t   >"<<endl;
	}
	else if(ch=='<')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<"<\t    34\t   -1\t>"<<endl;
	}
	else if(ch=='<>')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<"<>\t    35\t-1\t   >"<<endl;
	}
	else if(ch=='>')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<">\t    36\t-1\t   >"<<endl;
	}
	else if(ch=='>=')
	{
		no++;
		cout<<"<"<<no<<"\t  "<<">=\t    37\t-1\t   >"<<endl;
	}
	else
		cout<<"此处输入错误!"<<endl;
}
/*-------------------------------------------------*/
void main()
{
	int i=0,num=0,len;
    char  buf[maxsize];
	cout<<"请输入程序代码(以!结束)："<<endl;
	cin.unsetf(ios::skipws);
		    cin>>buf[num];
	while(buf[num]!='!'&&num<maxsize)
	{
		num++;
        cin>>buf[num];
	}
    len=num;
	cout<<len<<endl;
	cout<<endl<<"词法分析如下："<<endl;
	cout<<"输入序列"<<"  单词"<<"\t   机内码"<<"\t入口地址"<<endl;
	i=-1;
	/*while(i<len-2)
	{
	   scan(buf,i);
	}*/	
	scan(buf,i);
	while(i<len-2)//排除文件最后一位
	{
		if((buf[i]=' ')&&(i<len-2))
		{
		    scan(buf,i);
		}
		else if((buf[i]='\r')&&(i<len-2))//此处回车有问题
		{
			scan(buf,++i);
		}
		else i++;

	}
}