pl0.cpp

编译原理的------语法分析。有兴趣的欢迎下载。

#include "pl0.h"
/**************************************************************************
*	PL/0 语言编译程序，由江汉石油学院计算机系周云才根据清华大学出版社出版，
*	吕映芝，张素琴，蒋维杜编写的教材《编译原理》中第二章(PL/0编译程序的实现)
*	以及附录A中的代码改编而成。
*	代码版权由周云才拥有，使用者必须遵循以下约定：
*	可以免费使用此文件，但必须包含此声明。
*	可以修改、传播、打印、出版这里的源代码。
*	可以在任何软件工程中使用这里的源代码。
*	周云才对于由此源代码的使用而引起的任何问题没有任何责任。
*	周云才地址：湖北省荆州市江汉石油学院计算机系
*		  邮编：434023
*	      信箱：[e-mail]
*		  电话: [phone]
**********************************************************************************************/

string PL0::errStr[]={" ","error 0001:  常数说明中“=”写成“：=”",
	"error 0002:  常数说明中的“=”后应为数字",
	"error 0003:  常数说明中的标识符后应是“=”",
	"error 0004:  const,var,procedure后应为标识符",
	"error 0005:  漏掉了‘,’或‘;’",
	"error 0006:  过程说明后的符号不正确（应是语句开始符或过程开始符）",
	"error 0007:  应是语句开始符",
	"error 0008:  过程体内语句部分的后跟符不正确",
	"error 0009:  程序皆为丢了句号‘.’",
	"error 0010:  语句之间漏了‘;’",
	"error 0011:  标识符没说明",
	"error 0012:  赋值语句中，赋值号左部标识符属性应是变量",
	"error 0013:  赋值语句左部标识符应是赋值号:=",
	"error 0014:  call后应为标识符",
	"error 0015:  call后标识符属性应为过程",
	"error 0016:  条件语句中丢了then",
	"error 0017:  丢了end或;",
	"error 0018:  while型循环语句中丢了do",
	"error 0019:  语句后的标识符不正确",
	"error 0020:  应为关系运算符",
	"error 0021:  表达式内标识符属性不能是过程",
	"error 0022:  表达式中漏掉了右括号‘)’",
	"error 0023:  因子后的非法符号",
	"error 0024:  表达式开始符不能是此符号",
	"error 0025:  文件在不该结束的地方结束了",
	"error 0026:  结束符出现在不该结束的地方",
	"error 0027:  ","error 0028:  ","error 0029:  ","error 0030:  ",
	"error 0031:  数越界",
	"error 0032:  read语句括号中标识符不是变量"};

PL0::PL0(char* source,char*destination)
{
	listswitch=true,sourceEnd=false;
	strcpy(word[1],"begin");
	strcpy(word[2],"call");
	strcpy(word[3],"const"); 
	strcpy(word[4],"do");
	strcpy(word[5],"end");
	strcpy(word[6],"if");
	strcpy(word[7],"odd");
	strcpy(word[8],"procedure");
	strcpy(word[9],"read");
	strcpy(word[10],"then");
	strcpy(word[11],"var");
	strcpy(word[12],"while");
	strcpy(word[13],"write");
	wsym[1]= BEGINSYM;   wsym[2]=CALLSYM;
	wsym[3]= CONSTSYM;   wsym[4]=DOSYM;
	wsym[5]=ENDSYM;     wsym[6]=IFSYM;
	wsym[7]=ODDSYM;     wsym[8]=PROCSYM;
	wsym[9]=READSYM;	wsym[10]=THENSYM;
	wsym[11]=VARSYM;    wsym[12]=WHILESYM;
	wsym[13]=WRITESYM;
	memset(code,0,sizeof(code));
	memset(ssym,0,100*sizeof(symbol));
	memset(table,0,sizeof(table));
	memset(line,0,sizeof(line));
	ssym['+']= PLUS;
	ssym['-']= MINUS;
	ssym['*']= TIMES;
	ssym['/']= SLASH;
	ssym['(']= LPAREN;
	ssym[')']= RPAREN;
	ssym['=']= EQL;
	ssym[',']= COMMA;
	ssym['.']= PERIOD;
	ssym['#']= NEQ;
	ssym['<']= LSS;
	ssym['>']= GTR;
	ssym[';']= SEMICOLON;
	strcpy(mnemonic[LIT]," lit ");
	strcpy(mnemonic[OPR]," opr ");
	strcpy(mnemonic[LOD]," lod ");
	strcpy(mnemonic[STO]," sto ");
	strcpy(mnemonic[CAL]," cal ");
	strcpy(mnemonic[INT]," int ");
	strcpy(mnemonic[JMP]," jmp ");
	strcpy(mnemonic[JPC]," jpc ");
	declbegsys.insert(CONSTSYM),declbegsys.insert(VARSYM),declbegsys.insert(PROCSYM);
	statbegsys.insert(BEGINSYM),statbegsys.insert(CALLSYM),statbegsys.insert(IFSYM),statbegsys.insert(WHILESYM);
	facbegsys.insert(IDENT),facbegsys.insert(NUMBER),facbegsys.insert(LPAREN);
	err= 0;
	cc= 0;
	cx= 0;
	ll= 0;
	ch= ' ';
	kk= al;
	codeNo=0;
	fin=fopen(source,"r");
	fout=fopen(destination,"w");
}
void PL0::error(int n)
{
	char s[10];
	sprintf(s,"第 %d 行:",codeNo);
	errorString.push_back(s+errStr[n]);
	err= err+1;//error count
}//error end
void PL0::getch()
{
	if(cc==ll)
	{
		if(feof(fin))
		{
			if(sym!=PERIOD)
				error(25);
			sourceEnd=true;
			return;
		}
		cc= 0;
		fgets(line,lineLength,fin);
		codeNo++;
		ll=strlen(line);
		if(line[ll-1]==10)ll--;
	}
	ch= line[cc];
	cc= cc+1;
}
void PL0::getsym()
{
	if(sourceEnd)
		return;
	int i,j,k;
	while (ch ==' '||ch==9)
		getch();//cls space and tab
	if(isalpha(ch))//id or reserved word
	{
		k=0;
		memset(a,0,al+1);
		do{
			if (k < al)
			{
				a[k]= ch;
				k= k+1;
			}
			getch();
			if(sourceEnd)
				return;
		}while(isalpha(ch)||isdigit(ch));
		if(k >= kk)
			kk= k;
		else
		{
			do{
				a[kk]= ' ';
				kk= kk-1;
			}while(kk > k);
		}
		strcpy(id,a);
		i= 1;
		j= norw;
		do{
			k= (i+j) / 2;
			if(strcmp(id, word[k])<=0)
				j= k-1;
			if(strcmp(id,word[k])>=0)
				i= k+1;
		}while(i<=j);
		if(i-1 > j)
			sym= wsym[k];
		else
			sym= IDENT;
	}
	else if(isdigit(ch))//number
	{
		k= 0;
		num= 0;
		sym= NUMBER;
		do{
			num= 10 * num+ch-'0';
			k= k+1;
			getch();
		}while(isdigit(ch));
		if(k > nmax)
			error(30);
	}
	else if (ch == ':')
	{
		getch();
		if( ch== '=')
		{
			sym= BECOMES;
			getch();
		}
		else
			sym= NUL;
	}
	else if(ch == '<')//extra stuff added to support <=
	{
		getch();
		if (ch== '=')
		{
			sym= LEQ;
			getch();
		}
		else
			sym= LSS;
	}
	else if (ch == '>')
	{
		getch();
		if( ch == '=')
		{
			sym= GEQ;
			getch();
		}
		else
			sym= GTR;
	}
	else//end of extra stuff
	{
		sym= ssym[ch];
		getch();
	}
}

void PL0::gen(fct x,int y,int z)
{
	if (cx > cxmax)
	{
		cout<<"Program too long\n";
		return;
	}
	code[cx].f= x;
	code[cx].l= y;
	code[cx].a= z;
    cx= cx+1;
}//gen end

void PL0::test(symset s1,symset s2,int n)
{
	if(sourceEnd)return;
	if (s1.find(sym)==s1.end())
	{
		error(n);
		symset::iterator it;
		for(it=s2.begin();it!=s2.end();it++)
			s1.insert(*it);//s1=s1+s2
		while (s1.find(sym)==s1.end())
			getsym();
	}
}//test end
void PL0::enter(obj0 k,int &tx,int &dx,int lev)
{
	tx= tx+1;
	strcpy(table[tx].name,id);
	table[tx].kind=k;
	switch(k)
	{
	case CONSTANT:
		if(num>amax)
		{
			error(31);
			num=0;
		}
		table[tx].other.val=num;
		break;
	case VARIABLE:
		table[tx].other.inOther.level=lev;
		table[tx].other.inOther.adr=dx;
		dx++;
		break;
	case PROCEDURE:
		table[tx].other.inOther.level=lev;
		break;
	}
}//enter end

int PL0::position(alfa id,int tx)//find identifier id in table
{
	int i;
	strcpy(table[0].name, id);
	i= tx;
	while (strcmp(table[i].name,id)!=0)i--;
	return i;
}//position end
void PL0::constdeclaration(int&tx,int&dx,int lev)
{
	if(sym = IDENT)
	{
	  getsym();
	  if(sym>=EQL&&sym<=BECOMES)
	  {
		if( sym ==BECOMES)
			  error(1);
		getsym();
		if( sym == NUMBER)
		{
			enter(CONSTANT,tx,dx,lev);
			getsym();
		}
		else
		  error(2);
	  }
	  else
		error(3);
	}
	else
	  error(4);
}// constdeclaration end

void PL0::vardeclaration(int&tx,int&dx,int lev)
{
	if( sym == IDENT)
	{
		enter(VARIABLE,tx,dx,lev);
		getsym();
	}
	else
		error(4);
}//vardeclaration end

void PL0::listcode(int cx0)//list code generated for this block
{
	int i;
	if(listswitch)
		for (i= cx0;i<cx;i++)
		   cout<<" "<<i<<"  "<<mnemonic[code[i].f]
			<<"  "<<code[i].l<<"  "<<code[i].a<<endl;
}// listcode end
void PL0:: factor(symset fsys,int tx,int lev)
{
	int i;
	test(facbegsys,fsys,24);
	while(facbegsys.find(sym)!=facbegsys.end())
	{
		if( sym ==IDENT)
		{
			i= position(id,tx);
			if( i == 0)
				error(11);
			else
				switch(table[i].kind)
			{
			 case CONSTANT:
				 gen(LIT,0,table[i].other.val);
				 break;
			 case VARIABLE:
				 gen(LOD,lev-table[i].other.inOther.level,table[i].other.inOther.adr);
				 break;
			 case PROCEDURE:
				 error(21);
				 break;
			}
			getsym();
		}
		else  if (sym ==NUMBER)
		{
			if (num>amax)
			{
				error(31);
				num= 0;
			}
			gen(LIT,0,num);
			getsym();
		}
		else if( sym ==LPAREN)
		{
			getsym();
			symset tmp=fsys;
			tmp.insert(RPAREN);
			expression(tmp,tx,lev);
			if (sym == RPAREN)
				getsym();
			else
				error(22);
		}
		test(fsys,facbegsys,23);
	}
}//factor end
void PL0::term(symset fsys,int tx,int lev)
{
	if(sourceEnd)
		return;
	symbol  mulop;
	symset tmp=fsys;
	for(int t=TIMES;t<=SLASH;t++)
		tmp.insert((symbol)t);
	factor(tmp,tx,lev);
	while( sym>=TIMES && sym<=SLASH)
	{
		mulop= sym;
		getsym();
		factor(tmp,tx,lev);
		if (mulop ==TIMES)
			gen(OPR,0,4);
		else
			gen(OPR,0,5);
	}
}// term end
void PL0::expression(symset fsys,int tx,int lev)
{
	symbol addop;
	symset tmp=fsys;
	for(int t=PLUS;t<=MINUS;t++)
		tmp.insert((symbol)t);
	if( sym>=PLUS&&sym<=MINUS)
	{
		addop= sym;
		getsym();
		term(tmp,tx,lev);
		if( addop ==MINUS)
			gen(OPR,0,1);
	}
	else
		term(tmp,tx,lev);
	while (sym >=PLUS&&sym<=MINUS)
	{
		addop= sym;
		getsym();
		term(tmp,tx,lev);
		if (addop ==PLUS)
			gen(OPR,0,2);
		else
			gen(OPR,0,3);
	}
}// expression end
void PL0::condition(symset fsys,int tx,int lev)
{
	symbol relop;
	symset tmp=fsys;
	tmp.insert(EQL),tmp.insert(NEQ),tmp.insert(LSS),tmp.insert(LEQ),tmp.insert(GTR),tmp.insert(GEQ);
    if( sym ==ODDSYM)
	{
		getsym();
		expression(fsys,tx,lev);