funcion.cpp

PL/0语言的词法分析

#include "StdAfx.h"

//////////////////////////////////////////////////////////////////////
// gets the order of the given word in the keyword array
int getKeyOrder(char* gword){

	char *word=new char[strlen(gword)+1];
	strcpy(word,gword);
	_strlwr(word);			// convert to low case 
							// same as the array keyword[]

	int head=0, rear=keywordNum;

	int i=0;
	for(i=(head+rear)/2;head<rear;i=(head+rear)/2){
			if( strcmp(word, keyword[i])<0 ) rear=i;
			else if( strcmp(word, keyword[i])>0 ) head=i+1;
			else return i;
	}
	return -1;
}

void getch(void)
{
	while (cc == ll || 0 == ll) // ll: line lenghth	cc: chatacter count
	{
		if (feof(infile))
		{
			printf("\nPROGRAM INCOMPLETE\n");
			exit(1);
		}
		ll = cc = 0;

		while ( !feof(infile) 
			&& ((ch = getc(infile)) != '\n'))
		{
			line[++ll] = ch;
		} // while

		line[++ll] = ' ';
		++ln;
	}
	ch = line[++cc];
} // getch

//////////////////////////////////////////////////////////////////////
// gets a symbol from input stream.
void getsym(void)
{
	static bool tightNumber=false;

	char a[MAXNUMLEN + 1];
	while (ch == ' '|| ch == '\t'){
		getch();	//滤前导空格
		//if(ch == '\t') cc+=3;
	}
	
	int i,k;
	if (isalpha(ch))						//1.读标志符
	{
		cch=cc;

		k = 0;
		do{
			if (k < MAXIDLEN) a[k++] = ch;	// 舍弃大于 MAXIDLEN 部分的标识符 字符
			getch();
		} while (isalpha(ch) || isdigit(ch));
		a[k] = 0;

		strcpy(id, a);

		i = getKeyOrder(id);
		if (-1 != i){
			sym = wsym[i];	// symbol is a reserved word
			tightNumber=false;
		}
		else{
			if(tightNumber){
				cout<<"Illegal identifier \""<<a<<"\" in line "<<ln<<" column "<<cc-k
					<<".Compile Stopped.\n";
				fprintf(outfile, 
					"Illegal identifier \"%s\" in line %d column %d.Compile Stopped.\n", 
					a,ln, cc-k);
				exit(1);
			}
			sym = SYM_IDENTIFIER;   // symbol is an identifier
		}
	}
	else if (isdigit(ch))					//2.读 数字
	{
		cch=cc;

		k = num = 0;
		sym = SYM_NUMBER;
		do
		{
			num = num * 10 + ch - '0';
			a[k++]=ch;
			getch();
		} while (isdigit(ch));
		a[k]=0;
		strcpy(id, a);

		if( isalpha(ch) ) tightNumber=true;
		else tightNumber=false;

		if (k > MAXNUMLEN){
			//error(25);     // The number is too great.
			cout<<"The number in line %d column %d."<<a<<"is too long!"<<endl;
			fprintf(outfile, "The number in line %d column %d is too long.\nCompile Stopped.\n", 
				ln, cc-k);
			exit(1);
		}
	}
	else								//3.读 操作符
	{
		cch=cc;

		switch(ch){
		case ':':
			getch();
			if (ch == '=')
			{
				sym = SYM_BECOMES; // :=
				strcpy(id, ":=");
				getch();
			}
			else
			{
				sym = SYM_NULL;       // illegal? Yes,sir
			}
			break;
		case '>':
			getch();
			if (ch == '=')
			{
				sym = SYM_GEQ;     // >=
				strcpy(id, ">=");
				getch();
			}
			else
			{
				sym = SYM_GTR;     // >
				strcpy(id, ">");
			}
			break;
		case '<':
			getch();
			if (ch == '=')
			{
				sym = SYM_LEQ;     // <=
				strcpy(id, "<=");
				getch();
			}
			else if (ch == '>')
			{
				sym = SYM_NEQ;     // <>
				strcpy(id, "<>");
				getch();
			}
			else
			{
				sym = SYM_LES;     // <
				strcpy(id, "<");
			}
			break;
		case '+':			// +
			sym = SYM_PLUS;
			id[0]='+';
			id[1]=0;
			getch();
			break;
		case '-':			// -
			sym = SYM_MINUS;
			id[0]='-';
			id[1]=0;
			getch();
			break;
		case '*':			// *
			sym = SYM_TIMES;
			id[0]='*';
			id[1]=0;
			getch();
			break;
		case '/':			// /
			sym = SYM_SLASH;
			id[0]='/';
			id[1]=0;
			getch();
			break;
		case '=':			// =
			sym = SYM_EQU;
			id[0]='=';
			id[1]=0;
			getch();
			break;
		case '(':			// (
			sym = SYM_LPAREN;
			id[0]='(';
			id[1]=0;
			getch();
			break;
		case ')':			// )
			sym = SYM_RPAREN;
			id[0]=')';
			id[1]=0;
			getch();
			break;
		case ',':			// ,
			sym = SYM_COMMA;
			id[0] = ',';
			id[1]=0;
			getch();
			break;
		case ';':			// ;
			sym = SYM_SEMICOLON;
			id[0] = ';';
			id[1]=0;
			getch();
			break;
		case '.':			// .
			sym = SYM_PERIOD;
			id[0]='.';
			id[1]=0;
			// no getch()
			break;
		default:	// other tokens
			cout<<"Fatal Error: Unknown character:ACSII value "<<int(ch)<<" \""<<ch<<"\""<<endl;
			fprintf(outfile, "Fatal Error: Unknown character. "__FILE__ " line %d.\n",__LINE__);
			exit(1);
		}
		tightNumber=false;
	}
} // getsym