main.c

参照清华大学《编译原理》写的cmm语言词法分析程序

/*编译和运行环境
 * gcc4.3.2
 * Archlinux-2.6.26
 * 使用方法
 * 命令行输入 gcc main.c && ./a.out Test.cmm
*/
#include "cmm.h"

int main (int argc, char * argv[]) {
	fname = argv[1];
	fin = fopen(fname,"r");		/* 只读打开文件 */
	if (fin) {
		init();
		while (getsym() != -1) {
			printsym(6);
		}
	}
	else{
		printf("Can't open file! \n");
	}
	return 0;
}

void init() {
	int i;
	/* 初始化单字符的Token值 */
	for (i = 0; i <= 255; i++) {
		ssym[i] = nul;
	}
	ssym['+'] = plus;
	ssym['-'] = minus;
	ssym['*'] = times;
	ssym['/'] = slash;
	ssym['('] = lparen;
	ssym[')'] = rparen;
	ssym['['] = lbracket;
	ssym[']'] = rbracket;
	ssym['{'] = lbrace;
	ssym['}'] = rbrace;
	ssym['='] = assign;
	ssym[','] = comma;
	ssym['.'] = period;
	ssym[';'] = semicolon;
	
	/* 设置保留字的名字,按照字母顺序，便于折半查找 */
	strcpy(&(word[0][0]), "const");
	strcpy(&(word[1][0]), "else");
	strcpy(&(word[2][0]), "if");
	strcpy(&(word[3][0]), "int");
	strcpy(&(word[4][0]), "read");
	strcpy(&(word[5][0]), "real");
	strcpy(&(word[6][0]), "while");
	strcpy(&(word[7][0]), "write");
	
	/* 设置保留字对应的symbol */
	wsym[0] = constsym;
	wsym[1] = elsesym;
	wsym[2] = ifsym;
	wsym[3] = intsym;
	wsym[4] = readsym;
	wsym[5] =  realsym;
	wsym[6] =  whilesym;
	wsym[7] =  writesym;
}

/*
 * 漏掉空格,读取一个字符
 * 每次读一行,存入line缓冲区,line被getsym取空后再读一行
 * 被函数getsym调用
 */
int getch() {
	if ( cc == ll ) {		/* cc == ll说明当前行已经分析完成，或者开始分析第一行 */
		if ( feof(fin) ) {	/* 文件结束 */
			printf("program incomplete");
			return -1;
		}
		ll = 0;
		cc = 0;
		ch = ' ';
		/* 清空上一行余留的字符 */		
		memset(line,0,sizeof(line));
		while (ch != 10) {	/* 碰到换行符停止循环 */
			if (EOF == fscanf(fin,"%c",&ch)) {		/* 文件结束 */
//				printf("文件结束了!此时ch=%d,ll=%d,cc=%d,linenum=%d\n",ch,ll,cc,linenum);
				return -1;
			}
			line[ll] = ch;
			ll++;
		}
		linenum++;
		printline();
//		printf("while循环结束后ch=%d,ll=%d,cc=%d,linenum=%d,line=%s\n",ch,ll,cc,linenum,line);
	}
	ch = line[cc];
	cc++;
//	printf("最后ch=%d,ll=%d,cc=%d,linenum=%d\n",ch,ll,cc,linenum);
	return 0;
}

/* 词法分析，获取下一个Token */
int getsym() {
	int i, j, k;
	while ( ch == ' ' || ch == 10 || ch == 9) {		/* 忽略空格，换行符和TAB */
		getchdo;		/* 忽略后获取下个字符 */
	}
	if (ch >= 'a' && ch <= 'z') {			/* 检测标识符和关键字 */
		k = 0;
		do {
			if (k < al) {
				a[k] = ch;
				k++;
			}
			getchdo;
		}while (ch >= 'a' && ch <= 'z' || ch >= '0' && ch <= '9');
		a[k] = 0;
		strcpy(id,a);
		strcpy(value,a);
		i = 0;
		j = norw - 1;
		while (i <= j) {
			k = (i+j)/2;
			if (strcmp(id,word[k]) < 0) {		/* word[k] 表示 word[k][0]的地址 */
				j = k-1;
			}
			else if (strcmp(id,word[k]) > 0) {
				i = k+1;
			}
			else {
				break;
			}
		}
		if (i > j) {
			sym = ident;		/* 设置当前符号为ident */
		}
		else {
			sym = wsym[k];		/* 设置当前符号为某保留字的symbol */
		}
	}
	else {
		if (ch >= '0' && ch <= '9') {		/* 检测数字 */
			k = 0;					/* 当前number的位数 */
			num = 0;
			sym = number;
			do {					/* 继续检查后面的数字 */
				if (ch == '.') {	/* 处理小数部分 */
					sym = real;
					realvalue = num;
					e = 0.1;
					k++;
					getchdo;
					while (ch >= '0' && ch <= '9') {
						int ich = ch-'0';
						realvalue += e*ich;
						e*=0.1;
						getchdo;
					}
				}
				else {			/* 处理整数部分 */
					num=10*num+ch-'0';
					k++;
					getchdo;
				}
			} while(ch >= '0' && ch <= '9' || ch == '.');
			k--;
			if( k > nmax){
				printf("number too long!\n");
				exit(0);
			}
		}
		else {
			if (ch == '=') {		/* 检测赋值符号和等于号 */
				getchdo;
				if (ch == '=') {
					sym = eql;
					getchdo;
				}
				else {
					sym = assign;
				}
			}
			else {
				if (ch == '<') {	/* 检测小于号和小于等于号*/
					getchdo;
					if (ch == '=') {
						sym = leq;
						getchdo;
					}
					else {
						sym = lss;
					}
				}
				else {
					if (ch == '>') {	/* 检测大于号和大于等于号*/
						getchdo;
						if (ch == '=') {
							sym = geq;
							getchdo;
						}
						else {
							sym = gtr;
						}
					}
					else {
						if (ch == '!') {		/* 检测不等号 */
							getchdo;
							if (ch == '=') {
								sym = neq;
								getchdo;
							}
							else {
								sym = ssym['!'];
							}
						}
						else {		/* 当前符号不满足上述条件时,全部按照单字符号处理 */
							sym = ssym[ch];
							getchdo;
						}
					}
				}
			}
		}
	}
	return 0;	
}

/* 打印当前Token的值 */
void printsym (int offset) {
	int i;
	if (sym != 0) {
		for (i = 0; i< offset; i++) {
			printf(" ");
		}
		switch (sym) {
			case ident : 
				printf("identifier : %s\n",value);break;
			case number : 
				printf("number : %d\n",num);break;
			case real : 
				printf("real : %f\n",realvalue);break;
			case plus : 
				printf("plus : +\n");break;
			case minus : 
				printf("minus : -\n");break;
			case times : 
				printf("times : *\n");break;
			case slash : 
				printf("slash : /\n");break;
			case eql : 
				printf("equal : ==\n");break;
			case neq : 
				printf("not equal : !=\n");break;
			case lss : 
				printf("less than : <\n");break;
			case leq : 
				printf("less equal : <=\n");break;
			case gtr : 
				printf("greater than : >=\n");break;
			case geq : 
				printf("greater equal : >=\n");break;
			case lparen : 
				printf("lparen : (\n");break;
			case assign : 
				printf("assign : =\n");break;
			case rparen : 
				printf("rparen : )\n");break;
			case lbracket : 
				printf("lbracket : [\n");break;
			case rbracket : 
				printf("rbracket : ]\n");break;
			case lbrace : 
				printf("lbrace : {\n");break;
			case rbrace : 
				printf("rbrace : }\n");break;
			case comma : 
				printf("comma : ,\n");break;
			case semicolon : 
				printf("semicolon : ;\n");break;
			case period : 
				printf("period : .\n");break;
			case constsym : 
				printf("keyword : const\n");break;
			case elsesym : 
				printf("keyword : else\n");break;
			case ifsym : 
				printf("keyword : if\n");break;
			case intsym : 
				printf("keyword : int\n");break;
			case readsym : 
				printf("keyword : read\n");break;
			case realsym : 
				printf("keyword : real\n");break;
			case whilesym : 
				printf("keyword : while\n");break;
			case writesym : 
				printf("keyword : write\n");break;	
		}
	}
}

/* 打印当前行 */
void printline () {
	int i;
	printf(" %d: ",linenum);
	for (i = 0; i < sizeof(line); i++) {
		if (line[i] != 32 && line[i] != 10 && line[i] != 9) {
			break;
		}
	}
	for (i;i < sizeof(line) && line[i] != '\n'; i++) {
		printf("%c",line[i]);
	}
	printf("\n");
}