词法扫描器设计

/*词法扫描器设计*/
/*2007-10-24 21:22调试通过*/
#include "iostream.h"
#include "stdlib.h"
#include "fstream.h"
#include "stdio.h"
#include "string.h"
/*the length of token if 9*/
#define Max_len_token 10
#define Max_len_symbol 3
#define keywordnumber 21
#define operatornumber 8

/*
以下为使用的语言子类
保留字
procedure ,var ,begin ,if ,then ,else ,end ,do ,while ,switch ,case ,for ,string ,false ,maxint ,true ,boolean ,char ,real ,integer ,new
运算符
+ - * / ( ) :=
对于其他暂时不做过多考虑
*/
typedef struct token_{
 char str[Max_len_token];
 struct token_ *next;
}mytoken;
typedef struct symbol_{
 char str [Max_len_symbol];
 struct symbol_ *next ;
}symbol;
struct input{//检验输入
 char str [Max_len_token];  //存放字串数据的数组
 int type; //数据类型
};
/*function*/
void read_file(char * );  /*从文件中读出数据*/
void init(void);  /*init*/
int inputchar(char ); /*读入一个字符*/
int isoperator(char ); /*是否为 operator*/
void movestr (void ); /*将字符数组移动到链表中*/
void iskeyword(char * ); /*是否为 keyword*/
int exist(char *, mytoken * ); /*是否已经存在*/
void printall(void); /*全部打印到屏幕*/

/*list*/
//因为不区分大小写，设置两个数组，一个用于比较字符，另一个用于存储输入的字符
mytoken *keywordlist, *idlist ,*constantlist; /*存储字符*/
mytoken *keywordcmp , *idcmp , *constantcmp; /*存储比较字符*/
int operatorflag[operatornumber];
mytoken syskw[keywordnumber];
symbol sysop[operatornumber];
int delimiterflag ;

struct input *n_input =(struct input *)malloc (sizeof(struct input));
struct input *ori_input =(struct input *)malloc (sizeof(struct input));
char operatorbuf[4];
static int bufcount =0;
static int ncount=0;
static int rowcount =0;
static int flag =0;

void main(int argc,char *argv[])
{
 //测试输入参数
/* int count=0;
 while(count <argc)
 {
  cout<<argv[count];
  count ++;
 }*/
 init();
 if(argc >1 )
  read_file(argv[1]);
 else
  read_file("test.txt");
 printall();
};

void read_file(char * filename)
{
 fstream file(filename ,ios::in||ios::nocreate );
 if (! file)
 {
  cout<<"open txt document error!"<<endl;
  return ;
 }
 cout<<"对下列文件进行分析："<< filename<<endl;
 char s[2];
 s[1] = 0 ;
 file.read (s,1);
 int fileflag = 0;
 while(! file.eof())
 {
  /*测试读出的数据*/
/*  cout<<s;
  file.read (s,1);*/
  /*向表中添加数据*/
  if(fileflag == 0){
   fileflag = inputchar( s[0]);
   if(fileflag ==2){
    cout<<"第"<< rowcount <<"行出错"<<endl;
    }
   }
  else if(fileflag == 1){
   movestr();
   fileflag = inputchar( s[0]);
   }
  else if(fileflag ==2){
   if(isoperator (s[0]))
    fileflag = 0;
   else if ( (s[0] == 0x20) ||(s[0] ==0x9))
    fileflag =0 ;
   else if (s[0] == 0x0a || s[0] == 0x3b){
    fileflag = inputchar (s[0]);
    }
   }
  file.read( s , 1 );
 }
 movestr();
};

void init(void ){
 n_input->type = 7;
 delimiterflag = 0;
 for (int i = 0;i < 10 ;i++){
  n_input ->str [i] = 0 ;
  ori_input ->str [i] = 0;
  }
 for( i = 0 ; i < 4 ; i++){
  operatorbuf[i] = 0 ;
  }
 for ( i = 0 ;i < operatornumber ;i++){
  operatorflag[i] = 0;  
  }
 {/*set the pointer*/
  constantlist =(mytoken *)malloc (sizeof (mytoken));
  keywordcmp =(mytoken *)malloc (sizeof (mytoken));
  keywordlist =(mytoken *)malloc (sizeof (mytoken));
  idlist =(mytoken *)malloc (sizeof (mytoken));
  idcmp =(mytoken *)malloc (sizeof (mytoken));
  constantcmp =(mytoken *)malloc (sizeof (mytoken));
  constantlist ->next =NULL;
  constantcmp ->next =NULL;
  keywordlist ->next =NULL;
  keywordcmp ->next =NULL;
  idlist ->next =NULL;
  idcmp ->next =NULL;
 }
 {/*set the keyword*/
  strcpy(syskw[0].str,"string");
  strcpy(syskw[1].str,"procedure");
  strcpy(syskw[2].str,"var");
  strcpy(syskw[3].str,"begin");
  strcpy(syskw[4].str,"if");
  strcpy(syskw[5].str,"then");
  strcpy(syskw[6].str,"else");
  strcpy(syskw[7].str,"end");
  strcpy(syskw[8].str,"do");
  strcpy(syskw[9].str,"while");
  strcpy(syskw[10].str,"switch");
  strcpy(syskw[11].str,"case");
  strcpy(syskw[12].str,"for");
  strcpy(syskw[13].str,"new");
  strcpy(syskw[14].str,"false");
  strcpy(syskw[15].str,"maxint");
  strcpy(syskw[16].str,"true");
  strcpy(syskw[17].str,"boolean");
  strcpy(syskw[18].str,"char");
  strcpy(syskw[19].str,"real");
  strcpy(syskw[20].str,"integer");
 }
 {/*set the operator*/
  strcpy(sysop[0].str,":=");
  strcpy(sysop[1].str,"+");
  strcpy(sysop[2].str,"-");
  strcpy(sysop[3].str,"*");
  strcpy(sysop[4].str,"/");
  strcpy(sysop[5].str,"(");
  strcpy(sysop[6].str,")");
  strcpy(sysop[7].str,"=");
 }
};

/**************************
  0 is continue
  1 is ok
  2 is error
***************************/
int inputchar (char c){
 /*one is origin,the other is compare*/
 char corigin ,ccompare;
 if( (c == 0x20) || (c == 0x9)){
  return 1;
 }
 if(c ==0x0a){//在这里不能设置ncount = 0
  rowcount++;
  return 1;
 }
 if (isoperator (c)){
  if(bufcount >1)
   return 2;
  if( (bufcount == 1)&& (c != '='))
   return 2;
  if( (bufcount == 0 )&&  (c == ':'))
  {
   operatorbuf[bufcount ++] = c;
   return 0;
  }
  operatorbuf[bufcount ++] = c;
  return 1;
 }
 bufcount =0;
 if( (c > 0x40 )&& (c < 0x5b ))
  ccompare = c + 0x20;
 else
  ccompare = c;
 corigin = c;
 if ((ccompare >0x60) && (ccompare < 0x7b)) {
  if( n_input-> type == 4)
  {
   n_input ->type = 0;
   return 2 ;
  }
  if( n_input-> type & 4 != 0 )
   n_input-> type ^= 4;
  n_input-> str[ncount] =ccompare ;
  ori_input-> str[ncount ++] =corigin ;
  return 0;
   }
 else if((ccompare > 0x2f) && (ccompare < 0x3a)){
  if( n_input ->type & 2 != 0)
   n_input-> type ^= 2;
  if( n_input ->type & 1 != 0)
   n_input-> type ^= 1;
  n_input-> str[ncount] =ccompare ;
  ori_input-> str[ncount ++] =corigin ;
  return 0;
   }
 else if(ccompare == '_'){
  if( n_input ->type & 4 != 0)
   n_input-> type ^= 4;
  if( n_input ->type & 2 != 0)
   n_input-> type ^= 2;
  n_input-> str[ncount] =ccompare ;
  ori_input-> str[ncount ++] =corigin ;
  return 0;
   }
 return 2;
};

int isoperator (char c){
 if(c =='*')
  return 1;
 if(c =='-')
  return 1;
 if(c =='+')
  return 1;
 if(c =='/')
  return 1;
 if(c ==':')
  return 1;
 if(c ==';')
  return 1;
 if(c =='=')
  return 1;
 return 0;
};

void movestr (void ){
  if(ncount != 0){
  ncount = 0 ;
  if(n_input ->type ==0)
   return ;
  if(n_input->type <4) //后加的
   iskeyword(n_input->str );
  static mytoken *p = NULL ,* q =NULL;
  if(n_input-> type == 1){
   p = idlist;
   q = idcmp;
  }
  else if(n_input ->type == 2){
   p = keywordlist;
   q = keywordcmp;
  }
  else if(n_input ->type == 4){
   p = constantlist;
   q = constantcmp;
  }
  if(exist (n_input ->str ,q))
   return ;
  //添加数据到list中
  while (p->next != NULL)
   p = p->next ;
  p->next = (mytoken *)malloc (sizeof (mytoken));
  p = p->next ;
  p->next =NULL;
  strcpy(p->str ,ori_input ->str );
  while(q->next != NULL)
   q = q->next ;
  q->next = (mytoken *)malloc (sizeof (mytoken));
  q = q->next ;
  q->next =NULL;
  strcpy(q->str ,n_input ->str );
  //重置
  for (int i = 0;i < 10 ;i++){
   n_input-> str [i] = 0 ;
   ori_input->str [i] = 0 ;
   }
  n_input-> type = 7 ;
  ncount = 0;
  }
 if (bufcount != 0){
  if ( operatorbuf[0] == 0x3b)//此为 界限符 ;
   delimiterflag ++;
  else if (operatorbuf[1] == 0x3b)
   delimiterflag ++;
  else if(bufcount == 1){//此为单个运算符
    for( int i = 1; i < operatornumber ;i++){
     if(operatorbuf[0] == sysop[i].str [0]){
      operatorflag [i]++;
      break ;
      }
     }
   }
  else if (bufcount == 2){//此为 :=
   if( (operatorbuf[0] == ':' ) && (operatorbuf [1] == '=')){
    operatorflag[0]++;
    }
   }
  bufcount = 0;
  operatorbuf[0] = 0 ;
  operatorbuf[1] = 0 ;
  }
 };

void iskeyword(char *key)
{
 for (int i = 0; i < keywordnumber; i++)
 {
  if(strcmp(syskw[i].str ,key) == 0)
  {
   n_input->type =2;
   return ;
  }
 }
 n_input->type = 1;
};

int exist(char *str, mytoken *list)
{
 mytoken *p;
 p=list->next ;
 while(p !=NULL)
 {
  if(strcmp (p->str ,str) == 0)
  {
   return 1;
  }
  p = p->next ;
 }
 return 0;
};

void printall(void ){
 mytoken *p;
 cout<<"*******************"<<endl;
 cout<<"key words are:"<<endl;
 p = keywordlist->next ;
 while(p !=NULL){
  cout<<p ->str<<endl;
  p = p->next;
  }
 cout<<"*******************"<<endl;
 cout<<"ids are:"<<endl;
 p = idlist->next ;
 while(p !=NULL){
  cout<<p ->str<<endl;
  p = p->next;
  }
 cout<<"*******************"<<endl;
 cout<<"operators are:"<<endl;
 for (int i = 0; i < operatornumber ; i++){
  if(operatorflag[i] != 0)
   cout<<sysop[i] .str <<"/t出现了/t"<<operatorflag [i]<<"次"<<endl;
  }
 cout<<"*******************"<<endl;
 cout<<"constants are:"<<endl;
 p = constantlist ->next ;
 while(p !=NULL){
  cout<<p ->str<<endl;
  p = p->next;
  }
 cout<<"*******************"<<endl;
 cout<<"界限符 ; 出现了"<< delimiterflag<<"次"<<endl;
 cout<<"*******************"<<endl;
 };