/*******************************************
 詞法分析程序
 作者：龔勛
 學號：200131500342
 計科系 13班
********************************************/
#include<stdlib.h>
#include<stdio.h>
#include<string.h>

/*******************************************
 初始化函數
********************************************/
void init()
{             
    char *key[]={" ","auto","break","case","char","const","continue","default","do","double",
                 "else","enum","extern","float","for","goto","if","int","long","register",
                 "return","short","signed","sizeof","static","struct","switch","typedef",
                 "union","unsigned","void","volatile","while"};     /*C語言所有關鍵字，共32個*/
    char *limit[]={" ","(",")","[","]","->",".","!","++","--","&","~",
                   "*","/","%","+","-","<<",">>","<","<=",">",">=","==","!=","&&","||",
                  "=","+=","-=","*=","/=",",",";","{","}","#","_","'"};/*運算、限界符*/
    FILE *fp;
    int i;
    char c;
    fp=fopen("k.txt","w");
    for(i=1;i<=32;i++)
       fprintf(fp,"%s\n",key[i]);
    fclose(fp);               /*初始化關鍵字表*/
    fp=fopen("l.txt","w");
    for(i=1;i<=38;i++)
       fprintf(fp,"%s\n",limit[i]);
    c='"';
    fprintf(fp,"%c\n",c);
    fclose(fp);               /*初始化運算、限界符表*/
    fp=fopen("i.txt","w");
    fclose(fp);               /*初始化標識符表*/
    fp=fopen("c.txt","w");
    fclose(fp);               /*初始化常數表*/
    fp=fopen("output.txt","w");
    fclose(fp);               /*初始化輸出文件*/
}

/*******************************************
 十進制轉二進制函數
********************************************/
char * dtb(char *buf)
{          
    int temp[20];
	char *binary;
	int value=0,i=0,j;
	for(i=0;buf[i]!='\0';i++)
		value=value*10+(buf[i]-48);       /*先將字符轉化為十進制數*/
    if(value==0)
    {
		binary=malloc(2*sizeof(char));
		binary[0]='0';
		binary[1]='\0';
		return(binary);
	}
	i=0;
	while(value!=0)
    {
		temp[i++]=value%2;
		value/=2;
	}
	temp[i]='\0';
	binary=malloc((i+1)*sizeof(char));
	for(j=0;j<=i-1;j++)
       binary[j]=(char)(temp[i-j-1]+48);
    binary[i]='\0';
    return(binary);
}

/*******************************************
 根據不同命令查表或造表函數
********************************************/
int find(char *buf,int type,int command)
{             
    int number=0;
    FILE *fp;
    char c;
    char temp[30];
    int i=0;
    switch(type)
    {
        case 1: fp=fopen("k.txt","r");break;
        case 2: fp=fopen("i.txt","r");break;
        case 3: fp=fopen("c.txt","r");break;
        case 4: fp=fopen("l.txt","r");
    }
    c=fgetc(fp);
    while(c!=EOF)
    {
        while(c!='\n')
        {
            temp[i++]=c;
            c=fgetc(fp);
        }
        temp[i]='\0';
        i=0;
        number++;
        if(strcmp(temp,buf)==0)
        {  
		    fclose(fp);
			return(number);        /*若找到，返回在相應表中的序號*/
        }
        else
           c=fgetc(fp);
     }
     if(command==1)
     {   
	     fclose(fp); 
		 return(0);                /*找不到，當只需查表，返回0，否則還需造表*/
     }
     switch(type)
     {
         case 1: fp=fopen("k.txt","a");break;
         case 2: fp=fopen("i.txt","a");break;
         case 3: fp=fopen("c.txt","a");break;
         case 4: fp=fopen("l.txt","a");
     }
     fprintf(fp,"%s\n",buf);
     fclose(fp);
     return(number+1);             /*造表時，將字符串添加到表尾并返回序號值*/
}

/*******************************************
 數字串處理函數
********************************************/
void cs_manage(char *buffer)
{             
    FILE *fp;
	char *pointer;
	int result;
	pointer=dtb(buffer);
    result=find(pointer,3,2);       /*先查常數表，若找不到則造入常數表并返回序號值*/
    fp=fopen("output.txt","a");
	fprintf(fp,"%s\t\t\t3\t\t\t%d\n",buffer,result);
	fclose(fp);               /*寫入輸出文件*/
}

/*******************************************
 字符串處理函數 
********************************************/
void ch_manage(char *buffer)
{                     
    FILE *fp;
    int result;
    result=find(buffer,1,1);           /*先查關鍵字表*/
    fp=fopen("output.txt","a");
    if(result!=0)
       fprintf(fp,"%s\t\t\t1\t\t\t%d\n",buffer,result);    /*若找到，寫入輸出文件*/
    else
    {
        result=find(buffer,2,2);       /*若找不到，則非關鍵字，查標識符表，還找不到則造入標識符表*/
        fprintf(fp,"%s\t\t\t2\t\t\t%d\n",buffer,result);
    }                                 /*寫入輸出文件*/
    fclose(fp);
}

/*******************************************
 出錯處理函數
********************************************/
void er_manage(char error,int lineno)
{                 
    printf("\nerror: %c ,line %d",error,lineno);    /*報告出錯符號和所在行數*/
}

/*******************************************
 掃描程序
********************************************/
void scanner()
{            
    FILE *fpin,*fpout;
    char filename[20];
    char ch;
    int i=0,line=1;
    int count,result,errorno=0;
	char array[30];
    char *word;
	printf("\nthe file name:");
    scanf("%s",filename);
    if((fpin=fopen(filename,"r"))==NULL)
    {
        printf("cannot open file");
        return;
    }
    ch=fgetc(fpin);
    while(ch!=EOF)
    {                 /*按字符依次掃描源程序，直至結束*/
        i=0;
        if(((ch>='A')&&(ch<='Z'))||((ch>='a')&&(ch<='z'))||(ch=='_'))
        {           /*以字母開頭*/
            while(((ch>='A')&&(ch<='Z'))||((ch>='a')&&(ch<='z'))||(ch=='_')||((ch>='0')&&(ch<='9')))
            {
                array[i++]=ch;
                ch=fgetc(fpin);
            }
            word=(char *)malloc((i+1)*sizeof(char));
			memcpy(word,array,i);
			word[i]='\0';
            ch_manage(word);
            if(ch!=EOF)
			   fseek(fpin,-1L,SEEK_CUR);
        }
        else if(ch>='0'&&ch<='9')
        {          /*以數字開頭*/
            while(ch>='0'&&ch<='9')
            {
                array[i++]=ch;
                ch=fgetc(fpin);
            }
            word=(char *)malloc((i+1)*sizeof(char));
			memcpy(word,array,i);
			word[i]='\0';
            cs_manage(word);
            if(ch!=EOF)
			   fseek(fpin,-1L,SEEK_CUR);
        }
        else if((ch==' ')||(ch=='\t'))
			;           /*消除空格符和水平制表符*/
		else if(ch=='\n')
			line++;           /*消除回車并記錄行數*/
		else if(ch=='/')
        {                 /*消除注釋*/
			ch=fgetc(fpin);
			if(ch=='=')
			{              /*判斷是否為‘/=’符號*/
				fpout=fopen("output.txt","a");
				fprintf(fpout,"/=\t\t\t4\t\t\t32\n");
                fclose(fpout);
			}
			else if(ch!='*')
            {              /*若為除號，寫入輸出文件*/
                fpout=fopen("output.txt","a");
				fprintf(fpout,"/\t\t\t4\t\t\t13\n");
                fclose(fpout);
				fseek(fpin,-1L,SEEK_CUR);
			}
			else if(ch=='*')
			{              /*若為注釋的開始，消除包含在里面的所有字符*/
				count=0;
				ch=fgetc(fpin);
				while(count!=2)
                {          /*當掃描到‘*’且緊接著下一個字符為‘/’才是注釋的結束*/
					count=0;
					while(ch!='*')
						ch=fgetc(fpin);
					count++;
					ch=fgetc(fpin);
					if(ch=='/')
						count++;
					else
						ch=fgetc(fpin);
				}
			}
		}
		else if(ch=='"')
		{                 /*消除包含在雙引號中的字符串常量*/
			fpout=fopen("output.txt","a");
			fprintf(fpout,"%c\t\t\t4\t\t\t37\n",ch);
			ch=fgetc(fpin);
			while(ch!='"')
				ch=fgetc(fpin);
            fprintf(fpout,"%c\t\t\t4\t\t\t37\n",ch);
			fclose(fpout);
		}
		else
        {         /*首字符為其它字符,即運算限界符或非法字符*/
            array[0]=ch;
            ch=fgetc(fpin);       /*再讀入下一個字符，判斷是否為雙字符運算、限界符*/
            if(ch!=EOF)
            {           /*若該字符非文件結束符*/
                array[1]=ch;
				word=(char *)malloc(3*sizeof(char));
				memcpy(word,array,2);
				word[2]='\0';
                result=find(word,4,1);      /*先檢索是否為雙字符運算、限界符*/
				if(result==0)
				{                           /*若不是*/
                    word=(char *)malloc(2*sizeof(char));
					memcpy(word,array,1);
					word[1]='\0';
					result=find(word,4,1);      /*檢索是否為單字符運算、限界符*/
					if(result==0)
                    {                           /*若還不是，則為非法字符*/
						er_manage(array[0],line);
						errorno++;
						fseek(fpin,-1L,SEEK_CUR);
                    }
                    else
					{     /*若為單字符運算、限界符，寫入輸出文件并將掃描文件指針回退一個字符*/
						fpout=fopen("output.txt","a");
						fprintf(fpout,"%s\t\t\t4\t\t\t%d\t\n",word,result);
						fclose(fpout);
						fseek(fpin,-1L,SEEK_CUR);
					}
				}
                else
				{             /*若為雙字符運算、限界符，寫輸出文件*/
					fpout=fopen("output.txt","a");
					fprintf(fpout,"%s\t\t\t4\t\t\t%d\n",word,result);
					fclose(fpout);
				}
            }
            else
			{               /*若讀入的下一個字符為文件結束符*/
				word=(char *)malloc(2*sizeof(char));
				memcpy(word,array,1);
				word[1]='\0';
				result=find(word,4,1);       /*只考慮是否為單字符運算、限界符*/
				if(result==0)                /*若不是，轉出錯處理*/
					er_manage(array[0],line);
				else
				{                            /*若是，寫輸出文件*/
					fpout=fopen("output.txt","a");
					fprintf(fpout,"%s\t\t\t4\t\t\t%d\n",word,result);
					fclose(fpout);
				}
			}
        }
        ch=fgetc(fpin); 
    }
    fclose(fpin);
	printf("\nThere are %d error(s).\n",errorno);         /*報告錯誤字符個數*/
}

/*******************************************
 主函數
********************************************/
main()
{           
    init();            /*初始化*/
    scanner();         /*掃描源程序*/
}