lex.cpp

词法分析程序

			result_tuple.flag=o_comma;

			break;

		case ';':

			rc=LEX_SUCCESS;

			result_tuple.flag=o_semicolon;

			break;

		case '*':

			rc=LEX_SUCCESS;

			result_tuple.flag=o_time;

			break;

		case '/':

			rc=LEX_SUCCESS;

			result_tuple.flag=o_div;

			break;

		case ')':

			rc=LEX_SUCCESS;

			result_tuple.flag=o_rbracket;

			break;

		case '[':

			rc=LEX_SUCCESS;

			result_tuple.flag=o_lparent;

			break;

		case ']':

			rc=LEX_SUCCESS;

			result_tuple.flag=o_rparent;

			break;

		case '+':

			rc=LEX_SUCCESS;

			result_tuple.flag=o_plus;

			break;

		case '-':

			rc=LEX_SUCCESS;

			result_tuple.flag=o_minus;

			break;

		case ':':

			c=yyinput();

			if(c!='='){

				result_tuple.flag=o_colon;

				yyunput(c,yytext);

			}else result_tuple.flag=o_becomes;

			rc=LEX_SUCCESS;

			break;

		case '<':

			c=yyinput();

			if(c=='=')result_tuple.flag=o_le;

			else if(c=='>')result_tuple.flag=o_ne;

			else{

				result_tuple.flag=o_ls;

				yyunput(c,yytext);

			}

			rc=LEX_SUCCESS;

			break;

		case '>':

			c=yyinput();

			if(c!='='){

				result_tuple.flag=o_gt;

				yyunput(c,yytext);

			}else result_tuple.flag=o_ge;

			rc=LEX_SUCCESS;

			break;

		case '.':

			c=yyinput();

			if(!isdigit(c)){

				if(c=='.'){

					result_tuple.flag=o_range;

				}else{

					result_tuple.flag=o_period;

					yyunput(c,yytext);

				}

				rc=LEX_SUCCESS;

			}else{

				rc=act_number(c,3);

			}

			break;

		case '\r':

		case '\n':

		case '\t':

		case ' ':

			break;

		default:

			fprintf(stderr,"%-8d非法字符%c\n",yy_linenum,c);

		}

	}

	return rc;

}

/////////////////////////////////

static int frw[11]={

	0,	0,	0,	5,	16,

	25,	31,	35,	39,	40,

	41};

struct s_keyword{char * s;keyword f;} 

	t_keyword[]={

		{"if",k_if},//0

		{"of",k_of},

		{"do",k_do},

		{"to",k_to},

		{"or",k_or},

		{"var",k_var},//5

		{"for",k_for},

		{"end",k_end},

		{"not",k_not},

		{"and",k_and},

		{"div",k_div},//10

		{"mod",k_mod},

		{"abs",f_abs},

		{"sqr",f_sqr},

		{"ord",f_ord},

		{"chr",f_chr},//15

		{"odd",f_odd},

		{"type",k_type},

		{"then",k_then},

		{"else",k_else},

		{"true",c_true},//20

		{"char",t_char},

		{"succ",f_succ},

		{"pred",f_pred},

		{"read",f_read},

		{"const",k_const},//25

		{"begin",k_begin},

		{"while",k_while},

		{"array",k_array},

		{"false",c_false},

		{"write",f_write},//30

		{"downto",k_downto},

		{"record",k_record},

		{"maxint",c_maxint},

		{"readln",f_readln},

		{"program",k_program},//35

		{"integer",t_integer},

		{"boolean",t_boolean},

		{"writeln",f_writeln},

		{"function",k_function},

		{"procedure",k_procedure}};//40

char * t_identity[LEX_MAX_IDENTITY+1];

int    t_id_index;

void local_init(){

	char stringfile[FILENAME_LENGTH+2];

	strcpy(stringfile,sourcefile);

	strcat(stringfile,".str");

	yystr=fopen(stringfile,"w");

	yy_stringpos=0;

	memset(t_identity,0,sizeof(t_identity));

	t_id_index=1;

}

void local_end(){

	int i;

	if(yystr!=NULL)fclose(yystr);

	fprintf(yyout,"\n下面是标志符表\n\n");

	for(i=1;i<=LEX_MAX_IDENTITY;i++){

		if(t_identity[i]!=0){

			fprintf(yyout,"%-8d%s\n",i,t_identity[i]);

			delete[] t_identity[i];

		}

	}

}

static int _stdcall act_identity(char&c){

	do{

		c=yyinput();

		if(c==EOF)break;

	}while(isalpha(c)||(c=='_')||isdigit(c));

	yyunput(c,yytext);

	//现在yytext就是所发现的标志符

	int i,j,k;

	k=strlen(yytext);

	if(k<10){

		i=frw[k];j=frw[k+1]-1;

		while((i<=j)&&(strcmp(yytext,t_keyword[i].s)!=0))i++;

		if(i<=j){

			result_tuple.flag=t_keyword[i].f;

			return LEX_SUCCESS;

		}

	}

	t_identity[0]=yytext;

	i=t_id_index-1;

	while(strcmp(yytext,t_identity[i])!=0)i--;

	if(i==0){

		if(t_id_index>LEX_MAX_IDENTITY)

			yy_fatal_error("标志符表溢出");

		t_identity[i=t_id_index]=new char[k+1];

		strcpy(t_identity[t_id_index++],yytext);

	}

	result_tuple.flag=identity;

	result_tuple.data.i_val.int_val1=i;

	return LEX_SUCCESS;

}

static int _stdcall act_comment2(char&c){

	do{

		c=yyinput();

		if(c==EOF)break;

	}while(c!='}');

	//此处为注释2

	return LEX_NOTUPLE;

}

static int _stdcall act_string(char&c){

	char temp[STRLENGTH+2];

	register int  i=0;

	bool once=true;

loop:

	c=yyinput();

	temp[i++]=c;

	if((c==10)||(c==EOF)){

		fprintf(stderr,"%-8d缺少右单引号,定义字符串:%s\n",yy_linenum-1,yytext);

		once=false;

		goto exit;

	}

	if((i<STRLENGTH)&&(c!='\''))goto loop;

	if(c=='\''){

		c=yyinput();

		if(c=='\''){

			goto loop;

		}else{

			goto exit;

		}

	}

	if(i-->=STRLENGTH){

		if(once){

			fprintf(stderr,"%-8d字符串常量太长:%s\n",yy_linenum-1,yytext);

			once=false;

		}

	}

	goto loop;

exit:

	yyunput(c,yytext);

	temp[i-1]=0;

	//temp中为字符串

	i=strlen(temp);

	if(i==1){//为字符常量

		result_tuple.flag=c_char;

		result_tuple.data.c_val=temp[0];

	}else{

		result_tuple.flag=c_string;

		result_tuple.data.i_val.int_val1=yy_stringpos;

		result_tuple.data.i_val.int_val2=i;

		yy_stringpos+=i+2;

		fprintf(yystr,"%s\n",temp);

	}

	return once?LEX_SUCCESS:LEX_WITH_ERROR;

}

static int _stdcall act_comment1(char&c){

	char c1;

	c=yyinput();

	do{

		c1=c;

		c=yyinput();

		if((c==')')&&(c1=='*')){

			return LEX_NOTUPLE;

		}

	}while(c!=EOF);

	yyunput(c,yytext);

	return LEX_NOTUPLE;

}

static int _stdcall act_chinese(char&c){

	do{

		c=yyinput();

	}while(c<0);

	yyunput(c,yytext);

	fprintf(stderr,"%-8d非法的中文字符:%s\n",yy_linenum,yytext);

	return LEX_NOTUPLE;

}

static int _stdcall act_number(char&c,int state){

	register s=state;

	do{

		switch(s){

		case 1:

			s=2;

			break;

		case 2:

			if(isdigit(c)){

			}else if(c=='.')s=4;

			else if(c=='E'||c=='e')s=6;

			else goto exit;

			break;

		case 3:

			s=5;

			break;

		case 4:

			if(isdigit(c))s=5;

			else if(c=='E'||c=='e')s=6;

			else goto exit;

			break;

		case 5:

			if(isdigit(c));

			else if(c=='E'||c=='e')s=6;

			else goto exit;

		case 6:

			if(isdigit(c))s=8;

			else if(c=='+'||c=='-')s=7;

			else goto exit;

			break;

		case 7:

			if(isdigit(c))s=8;

			else goto exit;

			break;

		case 8:

			if(!isdigit(c))goto exit;

		}

	}while((c=yyinput())!=EOF);

exit:

	yyunput(c,yytext);

	if(s==2||s==4){

		result_tuple.flag=c_integer;

		if(s==4)yyunput('.',yytext);

		result_tuple.data.i_val.int_val1=atoi(yytext);

	}else{

		result_tuple.flag=c_real;

		result_tuple.data.d_val=atof(yytext);

	}

	if(s==7||s==6){

		fprintf(stderr,"%-8d错误的实数定义:%s\n",yy_linenum,yytext);

		return LEX_WITH_ERROR;

	}else 

		return LEX_SUCCESS;

}

#ifdef _DEBUG

void print_result(FILE*o){

	fprintf(o,"%-8d%-8d%s\n",yy_linenum,result_tuple.flag,yytext);

}

#endif