scan.c~

简单的词法分析器在Linux上用C语言实现

/*
 * scan.c
 *
 *  Created on: 2009-3-11
 *      Author: xiaohe
 */
#include<string.h>
#include<ctype.h>
#include<stdlib.h>

#include"readfile.h"
#include"scan.h"
#include"option.h"
#include"tokens.h"

struct kwords keywords[]={
		{"int",INT},
		{"char",CHAR},
		{"float",FLOAT},
		{"void",VOID},
		{"const",CONST},
		{"for",FOR},
		{"if",IF},
		{"else",ELSE},
		{"then",THEN},
		{"while",WHILE},
		{"switch",SWITCH},
		{"break",BREAK},
		{"begin",BEGIN},
		{"end",END},
		{NULL,0}
};
char oper[]={'+','-','*','/','%','<','>','=','!','0'};
char operator[][3]={"+","-","*","/","%","(",")","[","]","<",">","<=",">=","=","!=",";",",","/=","+=","-=","*=","%=","||","&&","!","0"};

int scanner_init()
{
	pchar=line;

    fline=0;
    nerror=0;

	save=NULL;
	source=NULL;

	type=0;
	count=0;

	return 0;
}

int scanner()
{
	char str[20];
	int n;
    char ch=' ';
	while(isspace(ch))
		ch=readchar();

	if(isalpha(ch)||ch=='_')
	{
		n=0;

		do
		{
			if(n<20)
				str[n++]=ch;
			ch=readchar();
		}while(isalpha(ch)||isalnum(ch));
		str[n]='\0';
		pchar--;
		if(!Iskeywords(str))
		{
			Isidentfier(str);
		}
	}
	else if(isalnum(ch))
	{
		n=0;
		do
		{
			if(n<20)
				str[n++]=ch;
			ch=readchar();
		}while(isalpha(ch)||isalnum(ch));
		str[n]='\0';

		Isnumber(str);
	}
	else
	{
		str[0]=ch;
		str[1]='\0';
		Isoperator(str);
	}

	return 0;
}

int Iskeywords(char *str)
{
	int i=0;
	while(keywords[i].sp)
	{
		if(!strcmp(str,keywords[i].sp))
		{
			type=keywords[i].sy;
			strcpy(identfier[count].name,str);
			identfier[count].code=i+1;
			count ++;
			return 1;
		}
		i++;
	}

	return 0;
}

int Isidentfier(char *str)
{
	int i;
	for(i=0;i<count;i++)
	{
		if(!strcmp(str,identfier[i].name))
		{
			return 1;
		}
	}
	if(identfier[count].type!=type)
	{
		strcpy(identfier[count].name,str);
		identfier[count].type=type;
		identfier[count].code=ID;
		count++;
		return 1;
	}

	return 0;
}

int Isnumber(char *str)
{
	int i;
	char *temp=str;
	for(i=0;i<strlen(str);i++)
	{
		if(!isdigit(*temp))
		{
			error(str,1);
			return 0;
		}
		temp++;
	}

	strcpy(identfier[count].name,str);
	identfier[count].code=NUM;
	count++;
	return 1;
}

int Isoperator(char *str)
{
	int i;
	char ch;
	if(Isoper(str)) //+ - * / % < > = !
	{
		ch=readchar();
		if(ch=='=') //+= -= *= /= %= <= >= == !=
		{
			strcat(str,"=");
			strcpy(identfier[count].name,str);
			i=0;
			while(strcmp(str,operator[i]))
			{
				i++;
			}
			identfier[count].code=i+19;
			count++;
			return 1;
		}
		else
		{
			i=0;
			pchar--;
			strcpy(identfier[count].name,str);
			while(strcmp(str,operator[i]))
			{
				i++;
			}
			identfier[count].code=i+19;
			count++;
			return 1;
		}
	}
	else
	{
		if(*str=='|')
		{
			ch=readchar();
			if(ch=='|')
			{
				i=0;
				strcat(str,"|");
				strcpy(identfier[count].name,str);
				while(strcmp(str,operator[i]))
				{
					i++;
				}
				identfier[count].code=i+19;
				count++;
				return 1;
			}
			else
			{
				pchar--;
				error(str,2);
				return 0;
			}
		}
		if(*str=='&')
		{
			ch=readchar();
			if(ch=='&')
			{
				i=0;
				strcat(str,"&");
				strcpy(identfier[count].name,str);
				while(strcmp(str,operator[i]))
				{
					i++;
				}
				identfier[count].code=i+19;
				count++;
				return 1;
			}
			else
			{
				pchar--;
				error(str,2);
				return 0;
			}
		}
		i=0;
		while(strcmp(str,operator[i]))
		{
			if(!strcmp(operator[i],"0"))
				return 0;
			i++;
		}
		strcpy(identfier[count].name,str);
		identfier[count].code=i+19;
		count++;
		return 1;
	}

	return 1;
}

int Isoper(char *str)
{
	int i=0;
	while(oper[i]!='0')
	{
		if(*str==oper[i])
			return 1;
		i++;
	}
	return 0;
}

void error(char *str,int num)
{
	nerror++;
	switch(num)
	{
	case 1:
		fprintf(stderr,"\"%s\" is unknow identfier line : %d\n",str,fline);
		break;
	case 2:
		fprintf(stderr,"\"%s\" is unknow operator line : %d\n",str,fline);
		break;
	case 3:
		break;
	}
}

void show_ident()
{
	int i;
	printf("The total errors is %d\n",nerror);
	printf("Identfier\n");

	printf("Name\t\t\tType\t\t\n");
	for(i=0;i<count;i++)
	{
		if(identfier[i].code==ID)
			printf("%s\t\t\t%d\n",identfier[i].name,identfier[i].type);
	}
}
void show_keywords()
{
	int i;
	printf("Keywords\n");

	printf("Name\t\t\tCode\t\t\n");
	for(i=0;i<count;i++)
	{
		if(identfier[i].code<ID)
			printf("%s\t\t\t%d\n",identfier[i].name,identfier[i].code);
	}
}
void show_number()
{
	int i;
	printf("Number\n");

	printf("Name\t\t\tCode\t\t\n");
	for(i=0;i<count;i++)
	{
		if(identfier[i].code==NUM)
			printf("%s\t\t\t%d\n",identfier[i].name,identfier[i].code);
	}
}
void show_operator()
{
	int i;
	printf("Operator\n");

	printf("Name\t\t\tCode\t\t\n");
	for(i=0;i<count;i++)
	{
		if(identfier[i].code>NUM)
			printf("%s\t\t\t%d\n",identfier[i].name,identfier[i].code);
	}
}