c语言词法分析器

/*程序全部在TC2.o下跳过的，在其它的编译器下可能调不过。这是因为有些非标准的函数，其它的编译器并没有*/

#include <stdio.h>

#include <string.h>

#include <conio.h> /*屏幕操作函数*/

#include <bios.h>

#define Num 256/*缓冲区大小，当分析Mb大小的文件时候，可以让缓冲区设置到大一些，可以缩短分析的时间，提高整体的效率*/

char q1[Num],temp1[Num],token[Num];

char *pp,*qq;/*两个缓冲区的指针 */

char ch;

int syn,p;

int tabflag=1,spaceflag=1,enterflag=1;/*预处理标识符 */

int count=0,flag2=1;/*count记录分析出的结果总数，flag2结束分析的标志，当为0时即由@出现时候分析结束*/

long ft=0;/*记录读到的文件位置 */

char *rwtab[]={"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"};/*33个关键字*/

char *limit[]={"(",")","[","]","}","{","<",">","!","*","/",

"+","-","=","&","->",".","|","#","%","~",",",";","<<","<=",

">>",">=","!=","*=","/=","+=","++","-=","-","==","&&","||","^"};/*运算、限界符38*/

/*功能：清理缓冲区

para：q 缓冲区开始指针，n缓冲区大小 */

void cleanBuf(char *q,int n)

{

int i;

for(i=0;i<n;i++)

q[i]='\0';

}

/*帮助菜单界面1*/

void help1()

{

int i;

printf("--------------------------------------------------------------------------------\n");

printf(" C Lexical analyzer\n");

printf("--------------------------------------------------------------------------------");

printf(" 1.Analysis of the beginning || 2.Results || 3.Set up || 4.Quit || 5.Help\n");

printf("--------------------------------------------------------------------------------");

printf("-------Word symbols-------types of code-------Word symbols------types of code\n");

for(i=0;i<14;i++)

printf("||-------------%-8s|--------%2d|-----------------%2s-----------%2d|---------||\n",rwtab[i],i+1,limit[i],i+36);

/* for(i=0;i<6;i=i+2)

printf("||-------------%-8s|--------%2d|-----------------%2s-----------%2d|---------||\n",limit[i+32],i+68,limit[i+33],i+69);

*/

gotoxy(1,22);

printf("--------------------------------------------------------------------------------");

gotoxy(35,23);

printf("up Next");

gotoxy(1,24);

printf("--------------------------------------------------------------------------------");

gotoxy(1,25);

textcolor(YELLOW);

cprintf(" Version 1.0 All Rights Reserved:KONG Email:bilplink@gmail.com");

textcolor(YELLOW);

textbackground(2);

gotoxy(5,5);

cprintf("A");

gotoxy(34,5);

cprintf("R");

gotoxy(47,5);

cprintf("S");

gotoxy(59,5);

cprintf("Q");

textcolor(RED);

gotoxy(69,5);

cprintf("H");

}

/*帮助菜单界面2*/

void help2()

{

int i;

printf("--------------------------------------------------------------------------------\n");

printf(" C Lexical analyzer\n");

printf("--------------------------------------------------------------------------------");

printf(" 1.Analysis of the beginning || 2.Results || 3.Set up || 4.Quit || 5.Help\n");

printf("--------------------------------------------------------------------------------");

printf("-------Word symbols-------types of code-------Word symbols------types of code\n");

for(i=14;i<28;i++)

printf("||-------------%-8s|--------%2d|-----------------%2s-----------%2d|---------||\n",rwtab[i],i+1,limit[i],i+36);

/* for(i=0;i<6;i=i+2)

printf("||-------------%-8s|--------%2d|-----------------%2s-----------%2d|---------||\n",limit[i+32],i+68,limit[i+33],i+69);

*/

gotoxy(1,22);

printf("--------------------------------------------------------------------------------");

gotoxy(35,23);

printf("up Next");

gotoxy(1,24);

printf("--------------------------------------------------------------------------------");

gotoxy(1,25);

textcolor(YELLOW);

cprintf(" Version 1.0 All Rights Reserved:KONG Email:bilplink@gmail.com");

textcolor(YELLOW);

textbackground(2);

gotoxy(5,5);

cprintf("A");

gotoxy(34,5);

cprintf("R");

gotoxy(47,5);

cprintf("S");

gotoxy(59,5);

cprintf("Q");

textcolor(RED);

gotoxy(69,5);

cprintf("H");

}

/*帮助菜单界面3*/

void help3()

{

int i;

printf("--------------------------------------------------------------------------------\n");

printf(" C Lexical analyzer\n");

printf("--------------------------------------------------------------------------------");

printf(" 1.Analysis of the beginning || 2.Results || 3.Set up || 4.Quit || 5.Help\n");

printf("--------------------------------------------------------------------------------");

printf("-------Word symbols-------types of code-------Word symbols------types of code\n");

for(i=28;i<33;i++)

printf("||-------------%-8s|--------%2d|-----------------%2s-----------%2d|---------||\n",rwtab[i],i+1,limit[i],i+36);

printf("||-------------Identifier|------34|------------Digital-----------35|---------||\n");

for(i=0;i<6;i=i+2)

printf("||-------------%-8s|--------%2d|-----------------%2s-----------%2d|---------||\n",limit[i+32],i+68,limit[i+33],i+69);

gotoxy(1,22);

printf("--------------------------------------------------------------------------------");

gotoxy(35,23);

printf("up Next");

gotoxy(1,24);

printf("--------------------------------------------------------------------------------");

gotoxy(1,25);

textcolor(YELLOW);

cprintf(" Version 1.0 All Rights Reserved:KONG Email:bilplink@gmail.com");

textcolor(YELLOW);

textbackground(2);

gotoxy(5,5);

cprintf("A");

gotoxy(34,5);

cprintf("R");

gotoxy(47,5);

cprintf("S");

gotoxy(59,5);

cprintf("Q");

textcolor(RED);

gotoxy(69,5);

cprintf("H");

}

/*主菜单界面*/

void menu()

{

textcolor(YELLOW);

textbackground(2);

printf("--------------------------------------------------------------------------------\n");

printf(" C Lexical analyzer\n");

printf("--------------------------------------------------------------------------------");

printf(" 1.Analysis of the beginning || 2.Results || 3.Set up || 4.Quit || 5.Help\n");

printf("--------------------------------------------------------------------------------");

gotoxy(1,24);

printf("--------------------------------------------------------------------------------");

gotoxy(1,25);

cprintf(" Version 1.0 All Rights Reserved:KONG Email:bilplink@gmail.com");

gotoxy(5,5);

cprintf("A");

gotoxy(34,5);

cprintf("R");

gotoxy(47,5);

cprintf("S");

gotoxy(59,5);

cprintf("Q");

gotoxy(69,5);

cprintf("H");

}

/*输入文件地址界面（开始分析程序界面)*/

void Readmenu(char *fileAddress)

{

printf("--------------------------------------------------------------------------------\n");

printf(" C Lexical analyzer\n");

printf("--------------------------------------------------------------------------------");

printf(" 1.Analysis of the beginning || 2.Results || 3.Set up || 4.Quit || 5.Help\n");

printf("--------------------------------------------------------------------------------");

printf("Please input your address to open the file:\n");

gotoxy(1,24);

printf("--------------------------------------------------------------------------------");

gotoxy(1,25);

textcolor(YELLOW);

cprintf(" Version 1.0 All Rights Reserved:KONG Email:bilplink@gmail.com");

textcolor(RED);

textbackground(2);

gotoxy(5,5);

cprintf("A");

textcolor(YELLOW);

gotoxy(34,5);

cprintf("R");

gotoxy(47,5);

cprintf("S");

gotoxy(59,5);

cprintf("Q");

gotoxy(69,5);

cprintf("H");

gotoxy(1,8);

scanf("%s",fileAddress);

}

/*词法分析器开始分析等待界面*/

void resultWait()

{

printf("--------------------------------------------------------------------------------\n");

printf(" C Lexical analyzer\n");

printf("--------------------------------------------------------------------------------");

printf(" 1.Analysis of the beginning || 2.Results || 3.Set up || 4.Quit || 5.Help\n");

printf("--------------------------------------------------------------------------------");

gotoxy(35,14);

textcolor(7);

cprintf("Please wait......\n");

gotoxy(1,24);

printf("--------------------------------------------------------------------------------");

gotoxy(1,25);

textcolor(YELLOW);

cprintf(" Version 1.0 All Rights Reserved:KONG Email:bilplink@gmail.com");

textcolor(RED);

textbackground(2);

gotoxy(5,5);

cprintf("A");

textcolor(YELLOW);

gotoxy(34,5);

cprintf("R");

gotoxy(47,5);

cprintf("S");

gotoxy(59,5);

cprintf("Q");

gotoxy(69,5);

cprintf("H");

}

/*分析完成界面*/

void resultFinish()

{

printf("--------------------------------------------------------------------------------\n");

printf(" C Lexical analyzer\n");

printf("--------------------------------------------------------------------------------");

printf(" 1.Analysis of the beginning || 2.Results || 3.Set up || 4.Quit || 5.Help\n");

printf("--------------------------------------------------------------------------------");

gotoxy(30,14);

textcolor(7);

cprintf("Analysis has been completed\n");

gotoxy(1,24);

printf("--------------------------------------------------------------------------------");

gotoxy(1,25);

textcolor(YELLOW);

cprintf(" Version 1.0 All Rights Reserved:KONG Email:bilplink@gmail.com");

textcolor(RED);

textbackground(2);

gotoxy(5,5);

cprintf("A");

textcolor(YELLOW);

gotoxy(34,5);

cprintf("R");

gotoxy(47,5);

cprintf("S");

gotoxy(59,5);

cprintf("Q");

gotoxy(69,5);

cprintf("H");

}

/*打开文件失败界面*/

void OpenError()

{

printf("--------------------------------------------------------------------------------\n");

printf(" C Lexical analyzer\n");

printf("--------------------------------------------------------------------------------");

printf(" 1.Analysis of the beginning || 2.Results || 3.Set up || 4.Quit || 5.Help\n");

printf("--------------------------------------------------------------------------------");

gotoxy(35,14);

textcolor(7);

cprintf("Failure to open files\n");

gotoxy(1,24);

printf("--------------------------------------------------------------------------------");

gotoxy(1,25);

textcolor(YELLOW);

cprintf(" Version 1.0 All Rights Reserved:KONG Email:bilplink@gmail.com");

textcolor(RED);

textbackground(2);

gotoxy(5,5);

cprintf("A");

textcolor(YELLOW);

gotoxy(34,5);

cprintf("R");

gotoxy(47,5);

cprintf("S");

gotoxy(59,5);

cprintf("Q");

gotoxy(69,5);

cprintf("H");

}

/*功能：分析字符 词法分析器核心 程序最终以@字符结束整个分析

para: q第一缓冲区开始指针的指针，n 缓冲区大小 ，p当前指针所在缓冲区位置，temp第二缓冲区开始指针的指针，fileAddress打开文件的地址*/

void scaner(char **q,int n,int *p,char **temp,char *fileAddress)

{

int i=0,j=0,m=0,h;

char *t;

for(i=0;i<Num;i++)

token[i]='\0';

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

while(ch==' ')

{

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

}

if((ch>='A'&&ch<='Z')||(ch>='a'&&ch<='z')||ch=='_')

{

while((ch>='0'&&ch<='9')||(ch>='A'&&ch<='Z')||(ch>='a'&&ch<='z')||ch=='_')

{

token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

}

token[m++]='\0';

(*p)--;

syn=34;/*34为标示符

//1-33为c语言关键字 */

for(j=0;j<32;j++)

if(strcmp(token,rwtab[j])==0)

{

syn=j+1;

break;

}

}

else

if((ch>='0'&&ch<='9'))

{

while((ch>='0'&&ch<='9'))

{

token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

}

(*p)--;

syn=35;/*为数字 */

}

else

switch(ch)

{

case'<':token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

if(ch=='<')

{

syn=59;

token[m++]=ch;

}

else

if(ch=='=')

{

syn=60;

token[m++]=ch;

}

else

{

syn=42;(*p)--;

}

break;

case'>':token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

if(ch=='>')

{

syn=61;

token[m++]=ch;

}

else if(ch=='=')

{

syn=62;

token[m++]=ch;

}

else

{

syn=43;

(*p)--;

}

break;

case'!':token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

if(ch=='=')

{

syn=63;

token[m++]=ch;

}

else

{

syn=44;

(*p)--;

}

break;

case'*':token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

if(ch=='=')

{

syn=64;

token[m++]=ch;

}

else

{

syn=45;

(*p)--;

}

break;

case'/':token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

if(ch=='=')

{

syn=65;

token[m++]=ch;

}

else

{

syn=46;

(*p)--;

}

break;

case'+':token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

if(ch=='=')

{

syn=66;

token[m++]=ch;

}

else if(ch=='+')

{

syn=67;

token[m++]=ch;

}

else

{

syn=47;

(*p)--;

}

break;

case'-':token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

if(ch=='=')

{

syn=68;

token[m++]=ch;

}

else if(ch=='-')

{

syn=69;

token[m++]=ch;

}

else if(ch=='>')

{

syn=51;

token[m++]=ch;

}

else

{

syn=48;

(*p)--;

}

break;

case'=':token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

if(ch=='=')

{

syn=70;

token[m++]=ch;

}

else

{

syn=49;

(*p)--;

}

break;

case'|':token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

if(ch=='|')

{

syn=72;

token[m++]=ch;

}

else

{

syn=53;

(*p)--;

}

break;

case'&':token[m++]=ch;

if(*p>=n)

{

t=*q;

*q=*temp;

*temp=t;

*p=0;

cleanBuf(*temp,Num);

h=readqramm(*temp,Num,fileAddress );

if(h==-1) {system("cls");OpenError();}

}

ch=(*q)[(*p)++];

if(ch=='*')

{

syn=71;

token[m++]=ch;

}

else

{

syn=50;

(*p)--;

}

break;

case'.':syn=52;token[0]=ch;break;

case'\'':syn=54;token[0]=ch;break;

case'%':syn=55;token[0]=ch;break;

case'~':syn=56;token[0]=ch;break;

case'^':syn=73;token[0]=ch;break;

case':':syn=74;token[0]=ch;break;

case',':syn=57;token[0]=ch;break;

case';':syn=58;token[0]=ch;break;

case'{':syn=41;token[0]=ch;break;

case'}':syn=40;token[0]=ch;break;

case'[':syn=39;token[0]=ch;break;

case']':syn=38;token[0]=ch;break;

case'(':syn=36;token[0]=ch;break;

case')':syn=37;token[0]=ch;break;

case'#':syn=59;token[0]=ch;break;

case'@':flag2=0;break;

default:syn=-1;token[0]=ch;break;

}

}

/*功能：从磁盘中读去程序内容

para:q 缓冲区指针，n缓冲区大小，fileAddress 读文件的地址（c:\2.txt)

return: 返回1 文件中还有可读字符 0文件没有可读字符 -1 打开文件失败 */

int readqramm(char *q,int n,char *fileAddress)

{

FILE *fp;

char ch;

int p=0;

if((fp=fopen(fileAddress,"r"))==NULL)

{

return -1;

}

fseek(fp,ft,0);

ch=fgetc(fp);

while(ch!=EOF&&n>p)

{

if(ch==' '&&spaceflag==1)

{

q[p++]=ch;

spaceflag=0;

}

if(ch=='\t'&&tabflag==1)

{

if(q[p-1]!=' ') q[p++]=' ';

tabflag=0;

}

if(ch!=' '&&ch!='\t'&&ch!='\r'&&ch!='\n')

{

q[p++]=ch;

spaceflag=1;

tabflag=1;

}

ch=fgetc(fp);

}

ft=ftell(fp)-1;

fclose(fp);

if(ch==EOF) return 0;

else return 1;

}

int main()

{

char fileAddress[Num];

char ch;

int flag1=1;

FILE *fp;

int jj=0,help_count=0,key;

pp=q1;

qq=temp1;

system("cls");

menu();

while(flag1)

{

key=bioskey(0);

switch(key)

{

case 7777:

case 7745:system("cls");

Readmenu(fileAddress);

flag2=1;

if(fileAddress!=NULL)

{

system("cls");

resultWait();

if(readqramm(qq,Num,fileAddress )==-1||readqramm(pp,Num,fileAddress )==-1) {system("cls");OpenError();}

else

do

{

scaner(&qq,Num,&p,&pp,fileAddress);

if(jj==0)

{

jj++;

if((fp=fopen("result.txt","w"))==NULL)

{

system("cls");OpenError();

}

}

else

{

jj++;

if((fp=fopen("result.txt","a"))==NULL)

{

system("cls");OpenError();

}

}

switch(syn)

{

case -1:fprintf(fp,"%d. (%d,%s) ",++count,syn,token);fclose(fp);continue;/*写入到result.txt文件中，这里最好使用缓冲区可以提高程序的效率，频繁打开文件写，效率低*/

default:fprintf(fp,"%d. (%d,%s) ",++count,syn,token);fclose(fp);continue;/*做法先让这些分析结果写到缓冲区中等缓冲满了，在一次写入到文件中，读取文件就使用了双缓冲效果很好*/

}

}while(flag2);

}

if(flag2==0)

{

system("cls");

resultFinish();

}

break;

case 4978:

case 4946:break;

case 8051:

case 8019:break;

case 4209:

case 4177:flag1=0;break;

case 9064:

case 9032:help_count=0;system("cls");help1();

while((key=bioskey(0))!=283)

{

switch(key)

{

case 20480:if(help_count==0) {system("cls");help2();help_count++;}

else if(help_count==1) {system("cls");help3();help_count++;}

break;

case 18432:if(help_count==2) {system("cls");help2();help_count--;}

else if(help_count==1) {system("cls");help1();help_count--;}

break;

}

}

system("cls");

menu();

break;

}

}

}