c语言写的Java词法分析

[代码] [C/C++]代码
#include <stdio.h>
#include <ctype.h>
#include <string.h>
#include <stdlib.h>
#define FILENAME "e:/a.java"
 
enum type
{
 
    blz,//保留字
    bzf,//标识符
    cs,//常数
    zfcl,//字符常量
    zfccl,//字符串常量
    dzf,//单字符
    szf//双字符
};
 
typedef struct output
{
   int type;
   union{
    
       char *_strvalue;
       char _charvalue;
       int _intvalue;
       float _floatvalue;
       double _doublevalue;
 
   }value;
 
   struct output *pnext;
   struct output *ppre;
 
} output;
 
int linenumber=0;
 
char getAChar(FILE *pfile);
char *getString(FILE *pfile);
char isDzf(char c);
output *createOuput(output *srcolutput);
char *getTypeChars(char c,FILE *pfile);
output *process(FILE *pfile);

char *constans_pool[]={
"package",
"class",
"int",
"double",
"float",
"byte",
"boolean",
"char",
"short",
"public",
"private",
"protocted",
"synchronized",
"instanceof",
"extends",
"implements",
"throw",
"throws",
"if",
"else",
"for",
"while",
"return",
"continue",
"break",
"switch",
"case","void","keyword"};

int main()
{
 
 
 
       output *phead=NULL,*pcurt=NULL;
    FILE *pfile=fopen(FILENAME,"r");
    if(pfile)
        phead=process(pfile);
 
 
    if(phead)
    {
            pcurt=phead;
       while(pcurt)
       {
          
           switch(pcurt->type)
           {
           case blz:

                  printf("<b>%s<b>\t",pcurt->value._strvalue);
               break;
           case bzf:
           case zfccl:
           case szf:
               printf("%s\t",pcurt->value._strvalue);
               break;
           case zfcl:
           case dzf:
            printf("%c\t",pcurt->value._charvalue);
            break;
           case cs:
            printf("%d\t",pcurt->value._intvalue);
               break;
  
            
           }
        
           pcurt=pcurt->pnext;
       }
     
    }
 
 
 
 
return 0;
}
 
output *process(FILE *pfile)
{
    char nowchar=0,prechar=0;
    output *phead=NULL,*pcurt=NULL,*pnext=NULL,*ppre=NULL;
    output tmp;
    char *p=NULL;
    char *pnowglz=NULL;
    int i=0;
 
 
  while(!feof(pfile))
  {
      prechar=nowchar;
    nowchar=getAChar(pfile);
 
 
 
     //如果是字幕
     if(isalpha(nowchar))
     {

           p=getTypeChars(nowchar,pfile);

           tmp.type=bzf;
         while(strcmp(constans_pool[i],"keyword"))
         {
         
             if(strcmp(constans_pool[i],p))
             {
               tmp.type=blz;
               break;
             }
         }
              tmp.value._strvalue=p;
          
     }
     //如果是数字
     else if(isdigit(nowchar))
     {
          p=getTypeChars(nowchar,pfile);
            tmp.type=cs;
              tmp.value._intvalue=atoi(p);
              
              if(phead==NULL)            
                   phead=pcurt;
              
      
     }
     else if(nowchar=='/')
     {
        nowchar=fgetc(pfile);
        if(nowchar=='*')//表明是注释；
        {
         
            do
            {
              while(nowchar=getAChar(pfile)!='*'&&nowchar>0);
               nowchar=fgetc(pfile);
            }
               while(nowchar!='/');
                
               continue;
        }
        else if(nowchar=='/')//单行注释
        {
             while((nowchar=fgetc(pfile))!='\n'&&nowchar>0);
             continue;
        }
         else
         {
              tmp.type=dzf;
              tmp.value._charvalue='/';
             ungetc(nowchar,pfile);
         }
     }
     else if(nowchar=='=')
     {
         nowchar=getAChar(pfile);
 
         if(nowchar=='=')
         {
               
             tmp.type=bzf;
              *(tmp.value._strvalue)='=';
              *(tmp.value._strvalue+1)='=';
 
         }
         else
         {
             ungetc(nowchar,pfile);
                continue;
         }
 
     }
     else if(nowchar=='\'')
     {
      
         nowchar=fgetc(pfile);
         if(fgetc(pfile)!='\'')
           printf("此处应该是单引号");
         else
         {
                tmp.type=zfcl;
              tmp.value._charvalue=nowchar;
         }
      
      
     }
     else  if(nowchar=='\"')
     {
            p=getString(pfile);
                tmp.type=zfccl;
              tmp.value._strvalue=p;
      
     }
 
 
     else if(isDzf(nowchar))
     {
                tmp.type=dzf;
              tmp.value._charvalue=nowchar;
     }
     else if(nowchar<0)
 
        // printf("end");
     continue;
 
 
          
    else
    {
         printf("\nerror%c\t%d\n",nowchar,linenumber);
         continue;
    }
 
            phead=createOuput(&tmp);
  }
     return phead;
 
}
//判断是不是单字符
char isDzf(char c)
{
    char dzfarr[]={'+','-','*','(',')','[',']','{','}',';','?','.','!','%'};
 
    unsigned int i=0;
    for(;i<sizeof(dzfarr)/sizeof(char);i++)
    {
       if(dzfarr[i]==c)
           return dzfarr[i];
 
    }
return 0;
 
}
//得到字符串常量
char *getString(FILE *pfile)
{
 
    char nowchar[]={0,0};
    char *p=NULL;
    unsigned int size=4;
 
    while((nowchar[0]=fgetc(pfile))>0&&nowchar[0]!='\"')
    {
     
      if(!p)
      {
        p=(char *)malloc(sizeof(char)*size);
        *p=0;
      }
      strcat(p,nowchar);
 
      if(strlen(p)>=size)
      {
         p=realloc(p,size*=2);
         *(p+size/2)=0;
       
      }
     
    }
    return p;
 
}
 
//读取同一个类型的字符加入到指针只到出现不同类型的字符
char *getTypeChars(char c,FILE *pfile)
{
 
        char nowchar[]={c,'\0'};
        char *p=NULL;
        int (*pchartype)(int)=NULL;
        unsigned int str_size=4;
 
        if(isdigit(c))
             pchartype=isdigit;
        else if(isalnum(c))
                pchartype=isalnum;
 
        do
         {
             if(!p)
             {
                 p=(char *)malloc(sizeof(char)*str_size);
                 *p=0;
             }
 
              strcat(p,nowchar);
 
              if(strlen(p)>=str_size)
              {
                  p=(char *)realloc(p,str_size*=2);
                  *(p+str_size/2)='\0';
              }
             
              nowchar[0]=fgetc(pfile);  
 
         }while(nowchar[0]>0&&pchartype(nowchar[0]));
 
          ungetc(nowchar[0],pfile);
         return p;
 
}
//创建output的链表
output *createOuput(output *srcolutput)
{
 
     static  output *pnext=NULL;
  static  output *ppre=NULL;
  static  output*phead=NULL;
  output *pout=NULL;
   pout=(output *)malloc(sizeof(output));
     pout->pnext=NULL;
  pout->type=srcolutput->type;
    switch(srcolutput->type)
    {
 
    case blz:
    case bzf:
    case zfccl:
    case szf:
        pout->value._strvalue=(char *)malloc(sizeof(char)*strlen(srcolutput->value._strvalue));
        *(pout->value._strvalue)=0;
        strcpy(pout->value._strvalue,srcolutput->value._strvalue);
        free(srcolutput->value._strvalue);
        break;
    case zfcl:
    case dzf:
        pout->value._charvalue=srcolutput->value._charvalue;
        break;
    case cs:
        pout->value._intvalue=srcolutput->value._intvalue;
        break;
     
    }
 
       if(!phead)
            phead=pout;
       if(ppre)
           ppre->pnext=pout;
 
       ppre=pout;
 
    return phead;
 
}
  {jave语言}
//返回一个字符
//这个字符不为空字符。
char getAChar(FILE *pfile)
{
   char c=0;
 
   if(!pfile)
       return EOF;
 
    while((c=fgetc(pfile))=='\t'||c=='\0'||c==' '||c=='\n')
    {
      if(c=='\n')
        linenumber++;
    }
 
    return c;
}

为啥要弄这个，common 做法都是通过 Flex/Bison 来产生 C 代码。