wordcheck.c

针对几种语言的词法分析的一点东东! 望接纳!

#include <stdio.h>
#include <string.h>
//#include <process.h>
//#include <windows.h>
#define MAX 50

/* if,else,while,do,char,int,float,long,for,main,void return,exit,static,struct,typdef,
   goto,extern,swich,case,continue,break*/
void main()
{
    FILE *in,*out;/*存放输入字符串和输出单词串的文件*/
    char arr[MAX];/*arr数组存放单词符号*/
	char currentchar;/*currentchar存放当前输入字符*/
	
    int i=0;/*arr数组的一个指针*/
    /*以读方式打开输入文件*/
    if((in=fopen("infile.txt","r"))==NULL)
    {
        printf("can not open file\n");
        return;
		//exit(0);
    }
    /*以写方式打开输出文件*/
    if((out=fopen("outfile.txt","w"))==NULL)
    {
        printf("cannout open outfile\n");
	    return;
		//exit(1);
    }
    //去掉开头空格
    currentchar=fgetc(in);
    while(currentchar==' ')
    { 
        currentchar=fgetc(in);
    }
//    int ch;
    
    while(currentchar!='@')
    {
        /*常数部分判断*/
        if((currentchar>='a'&&currentchar<='z')||(currentchar>='A'&&currentchar<='Z')||
            (currentchar>='0'&&currentchar<='9')||currentchar=='('||currentchar==')'||
            currentchar=='{'||currentchar=='}'||currentchar=='>'||currentchar=='<'||
            currentchar==';'||currentchar==':'||currentchar=='+'||currentchar=='-'||
            currentchar=='*'||currentchar=='/'||currentchar=='='||currentchar=='!'||
            currentchar=='&'||currentchar=='\n'||currentchar==' '||currentchar=='['||
            currentchar==']'||currentchar=='"'||currentchar=='\''||currentchar=='\\'
			||currentchar==','||currentchar=='%'||currentchar=='	')
			// 字母，数字，() {} <> : ; + _ * / = ! & \n [] " ' \
			
        {
             
			i=0;//清空arr字符数组
            if(currentchar>='0'&&currentchar<='9')
            {
                arr[i++]=currentchar;
                currentchar=fgetc(in);//再读入下一个字符
                while(currentchar>='0'&&currentchar<='9')
                {
                    arr[i++]=currentchar;
                    currentchar=fgetc(in);
                }
                if(currentchar!='.')//如果当前读入的符号不为小数点，则常数读完毕，输出此常数
                {
                    arr[i++]='\0';
                    fprintf(out,"%s%d,%c%s%c%s\n","(",100,'"',arr,'"',")");//数字种码为100
					
				}
                else//当前字符为小数点则继续向下读入
                {
                    arr[i++]=currentchar;
                    currentchar=fgetc(in);
                    while(currentchar>='0'&&currentchar<='9')
                    {
                        arr[i++]=currentchar;
                        currentchar=fgetc(in);
                    }
                    arr[i++]='\0';
                    fprintf(out,"%s%d,%c%s%c%s\n","(",100,'"',arr,'"',")");
                }
                while(currentchar==' '||currentchar=='	')
                { 
                    currentchar=fgetc(in);
                }
			
            }
            /*判断标识符和关键字部分*/
            if((currentchar>='a'&&currentchar<='z')||(currentchar>='A'&&currentchar<='Z'))
            {
                i=0;//清空arr字符数组
                while((currentchar>='a'&&currentchar<='z')||(currentchar>='0'&&currentchar<='9')
					||(currentchar>='A'&&currentchar<='Z'))
                {
                    arr[i++]=currentchar;
                    currentchar=fgetc(in);
                }
                //把字符数组arr和关键字表比较，判断单词串是关键字还是标识符
                arr[i++]='\0';
                if(strcmp(arr,"if")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",0,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
                else if(strcmp(arr,"else")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",1,'"',arr,'"',")");
				
				//currentchar=fgetc(in);
				}
                else if(strcmp(arr,"while")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",2,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
                else if(strcmp(arr,"do")==0)
				{ fprintf(out,"%s%d,%c%s%c%s\n","(",3,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
                else if(strcmp(arr,"for")==0)
				{ fprintf(out,"%s%d,%c%s%c%s\n","(",4,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
                else if(strcmp(arr,"int")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",5,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
                else if(strcmp(arr,"char")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",6,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
			    else if(strcmp(arr,"float")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",7,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
				else if(strcmp(arr,"struct")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",8,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
				else if(strcmp(arr,"typedef")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",9,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
				else if(strcmp(arr,"static")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",10,'"',arr,'"',")");
				//urrentchar=fgetc(in);
				}
				else if(strcmp(arr,"void")==0)
				{ fprintf(out,"%s%d,%c%s%c%s\n","(",11,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
				else if(strcmp(arr,"main")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",12,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
				else if(strcmp(arr,"return")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",13,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
				else if(strcmp(arr,"exit")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",14,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
				else if(strcmp(arr,"extern")==0)
				{ fprintf(out,"%s%d,%c%s%c%s\n","(",15,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
				else if(strcmp(arr,"goto")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",16,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
				else if(strcmp(arr,"switch")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",17,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
                else if(strcmp(arr,"case")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",18,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
				else if(strcmp(arr,"continue")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",19,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
			else if(strcmp(arr,"break")==0)
				{fprintf(out,"%s%d,%c%s%c%s\n","(",20,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}	
		    
			else                    
				{fprintf(out,"%s%d,%c%s%c%s\n","(",60,'"',arr,'"',")");
				//currentchar=fgetc(in);
				}
            }//0到59是关键字，60是变量名，100 是数字，60到99可以设为运算符等
            while(currentchar==' '||currentchar=='	')
            { 
                currentchar=fgetc(in);
            }
            //判断标识符包括：+, -, *, /, (, ), {, }, <, >, ;, :, !=, >=, <=, ==, =.
            if(currentchar=='+')
            {
                fprintf(out,"%s%d,%c%c%c%s\n","(",61,'"',currentchar,'"',")");
                currentchar=fgetc(in);
            }
            while(currentchar==' '||currentchar=='	')
            { 
                currentchar=fgetc(in);
            }
            if(currentchar=='-')
            {
                fprintf(out,"%s%d,%c%c%c%s\n","(",62,'"',currentchar,'"',")");
                currentchar=fgetc(in);
            }
            while(currentchar==' '||currentchar=='	')
            { 
                currentchar=fgetc(in);
            }
            if(currentchar=='*')
            {
                fprintf(out,"%s%d,%c%c%c%s\n","(",63,'"',currentchar,'"',")");
                currentchar=fgetc(in);
            }
            while(currentchar==' '||currentchar=='	')
            { 
                currentchar=fgetc(in);
            }
            if(currentchar=='/')
            {
                i=0;
                arr[i++]=currentchar;
                currentchar=fgetc(in);
                if(currentchar=='/')
                {
                    arr[i++]=currentchar;
                    arr[i++]='\0';
                    fprintf(out,"%s%d,%c%s%c%s\n","(",65,'"',arr,'"',")");
                    currentchar=fgetc(in);
                    while(currentchar!='\n')
                    {
                        currentchar=fgetc(in);
                    }
                }
                else if(currentchar=='*')
                {
                    arr[i++]=currentchar;
                    arr[i++]='\0';
                    fprintf(out,"%s%d,%c%s%c%s\n","(",66,'"',arr,'"',")");
                    currentchar=fgetc(in);
                    while(1){
                        while(currentchar!='*')
                        {
                            currentchar=fgetc(in);
                        }
                        i=0;
                        arr[i++]=currentchar;
                        currentchar=fgetc(in);
                        if(currentchar=='/')
                        {
                            arr[i++]=currentchar;
                            arr[i++]='\0';
                            fprintf(out,"%s%d,%c%s%c%s\n","(",67,'"',arr,'"',")");
                            currentchar=fgetc(in);
                            break;
                        }
                    }
                }
                else 
                {
                    arr[i++]='\0';
                    fprintf(out,"%s%d,%c%s%c%s\n","(",64,'"',arr,'"',")");
                    //currentchar=fgetc(in);