scnner.cpp

简单源程序的词法分析程序的C语言实现

#include<stdio.h>
#include<ctype.h>
#include<string.h>
#include<stdlib.h>

typedef struct KeyWord
{
	int num;
	char *word;
}KeyWordTable;

KeyWordTable kewwordtable[12];// 保存保留字表的全局数组

char TOKEN[20];
FILE *ifp;
FILE *ofp;
int begin_note = 0;//注释开始

long position;

void init_KeyWordTable();
bool isequal(char *s1,char *s2);
int lookup(char *token );
void out (int m,char *token);
void report_error();
void scnner(FILE *fp);


int main(int argc, char* argv[])
{
	if(argc!=3)
	{
		printf("You forgot to enter a filename\n");
		exit(0);
	}

	if((ifp=fopen(argv[1],"r"))==NULL)
	{
		printf("Open file: %s error!\n",argv[1]);
		exit(0);
	}

	if((ofp=fopen(argv[2],"a"))==NULL)
	{
		printf("Open file: %s error!\n",argv[2]);
		exit(0);
	}

	init_KeyWordTable();//初始化保留字表

	char a;
	long position;


			a = fgetc(ifp);
			if(a!=EOF)
			{
				while (isspace (a))//检查是否是空格 制表符或换行符
				{
					a = fgetc(ifp);
				}
			}
			fseek(ifp,-1,1);

	while (!feof(ifp))
	{
			position = ftell(ifp);
			printf("%ld\n",position);
			scnner(ifp);
			position = ftell(ifp);
			printf("%ld\n",position);
			a = fgetc(ifp);
			if(a!=EOF)
			{
				while (isspace (a))//检查是否是空格 制表符或换行符
				{
					a = fgetc(ifp);
				}
				if(a!=EOF)
				{
					fseek(ifp,-1,1);
				}
			}
	}

	
	fclose(ifp);
	fclose(ofp);
	return 0;
}

void init_KeyWordTable()
{
	kewwordtable[0].num = 1;
	kewwordtable[0].word = "begin";
	kewwordtable[1].num = 2;
	kewwordtable[1].word = "end";
	kewwordtable[2].num = 3;
	kewwordtable[2].word = "if";
	kewwordtable[3].num = 4;
	kewwordtable[3].word = "then";
	kewwordtable[4].num = 5;
	kewwordtable[4].word = "else";
	kewwordtable[5].num = 6;
	kewwordtable[5].word = "for";
	kewwordtable[6].num = 7;
	kewwordtable[6].word = "do";
	kewwordtable[7].num = 8;
	kewwordtable[7].word = "while";
	kewwordtable[8].num = 9;
	kewwordtable[8].word = "and";
	kewwordtable[9].num = 10;
	kewwordtable[9].word = "for";
	kewwordtable[10].num = 11;
	kewwordtable[10].word = "not";
	kewwordtable[11].num = 12;
	kewwordtable[11].word = "or";
}

bool isequal(char *s1,char *s2)
{
	int i=0;
	for(;*(s1+i)!='\0'&&*(s2+i)!='\0';)
	{
		if(*(s1+i)==*(s2+i)||*(s1+i)==*(s2+i)+32||s1[i]==s2[i]-32)//大小写不敏感
		{
			i++;
		}
		else 
		{
			return 0;
			break;
		}
	}
	if(s1[i]=='\0'&&s2[i]=='\0')
		return 1;
	else 
		return 0;
}

int lookup(char *token )
{
	for(int i=0;i<12;i++)
	if(isequal(token,kewwordtable[i].word))
	{
		return kewwordtable[i].num;
		break;
	}	
		return 0;
}

void out (int m,char *token)
{

	fprintf(ofp,"%c%d,%s%c",'(',m,token,')');
}

void report_error()
{
	printf("Error:undeclared identifier!\n");
}

void scnner(FILE *fp)
{
	char ch;
	int i;
	int c;
	ch = fgetc(fp);
	
	if(ch=='/')
	{
		ch = fgetc(fp);
		if(ch=='*' ) 
		{
			if(begin_note % 2 == 1)
			{
				printf("Error!嵌套的\" /* \"!\n");
			}
			begin_note ++;
			return;
		}
		else 
		{
			fseek(fp,-1,1);
			out (22,"  ");
			return;
		}
	}

	if(ch=='*')
	{
		ch = fgetc(fp);
		if(ch=='/' ) 
		{
			if(begin_note % 2 == 0)
			{
				printf("Error!未匹配的\" */ \"!\n");
				return;
			}
			begin_note++;
			return;
		}
		else 
		{
			fseek(fp,-1,1);
			out (27,"  ");
			return;
		}
	}

	if(begin_note % 2==1)
	{
		return;
	}

	if(isalpha (ch))
	{
		TOKEN[0] = ch;
		ch = fgetc(fp);

		i = 1;
		while (isalnum(ch))
		{
			TOKEN[i] = ch;
			i++;
			ch = fgetc(fp);
		}
		TOKEN[i] = '\0';
        if (ch!=EOF)  //////
		{
			fseek(fp,-1,1);
		}

		c = lookup(TOKEN);
		if (0==c)
			out (13,TOKEN);
		else out (c,"  ");
	}
	else 
		if(isdigit(ch))
	{
		TOKEN[0] = ch;
		ch = fgetc(fp);
		i = 1;
		while(isdigit(ch))
		{
			TOKEN[i] = ch;
			i++;
			ch = fgetc(fp);
		}
		TOKEN[i] = '\0';
		fseek(fp,-1,1);



		out(14,TOKEN);
	}
	else
		switch(ch)
	{
		case'<':
			ch = fgetc(fp);
			if(ch=='=') 
				out(16,"  ");
			else if(ch=='>')
				out(19,"  ");
			else 
			{
				fseek(fp,-1,1);
				out (15,"  ");
			}
			break;
		case'=':
			out(17,"  ");
			break;
		case'>':
			ch = fgetc(fp);
			if(ch=='=') 
				out(20,"  ");
			else 
			{
				fseek(fp,-1,1);
				out (18,"  ");
			}
			break;
		case':':
			ch = fgetc(fp);
			if(ch=='=') 
				out(23,"  ");
			else 
			{
				fseek(fp,-1,1);
				out (21,"  ");
			}
			break;
		case'+':
			out(25,"  ");
			break;
		case'-':
			out(26,"  ");
			break;
		case'|':
			out(28,"  ");
			break;
/*		case'/':
			ch = fgetc(fp);
			if(ch=='*') 
				out(24,"  ");
			else 
			{
				fseek(fp,-1,1);
				out (22,"  ");
			}
			break;
		case EOF:
			break;
		case ' ':
			break;
		case '\t'://制表符
			break;
		case '\n': //换行
			break;
		case '\r'://回车
			break;           */
		default:
			report_error();
			break;
	}
	return;
}