5656.txt

它输入描述构词规则的一系列正规式,然后构建有穷自动机和这个有穷自动机的一个驱动程序,进而生成一个词法分析程序.

//词法分析器V0.4 by 1T0F
//请在命令行中执行
#include <iostream>
#include <fstream>
using namespace std;
//函数定义 
string InputSource(string);     //输入源代码 
string DeleteExtra(string);     //删除额外字符
void OutputCode(string);     //输出结果 
void SaveResult(int,int,string);   //保存结果
void SaveError(int,int,string);    //保存错误 
bool IsDigit(char);       //判断是否数字 
bool IsLetter(char);      //判断是否字母 
bool IsOperate(char);      //判断是否运算符 
bool IsSymbol(char);       //判断是否符号 
bool IsReserved(string);     //判断是否保留字 
void IsNumber(int,string&,int,int&); //数字类型
void IsIdenti(int,string&,int,int&); //标识符
void IsString(int,string&,int,int&); //字符串 
void IsMarco(int,string&,int,int&);   //宏 
void IsOpt(int,string&,int,int&);   //操作粗 
void IsSym(int,string&,int,int&);   //分隔符 
//存储容器
struct save{
int line;
int type;
string content;
}Error[1000],Result[1000];
int Ei=0,Ri=0;
//主函数 
int main(int argc,char* argv[]){ 
//输入代码 
if(argc==1){
   cout<<"词法分析器V0.4 BY 1T0F\n"<<endl;
   exit(-1);
}
string TempFile=InputSource(argv[1]);
TempFile=DeleteExtra(TempFile);
//词法分析部分 
int i=0,LineNum=1;
while(TempFile[i]){
   if(TempFile[i]=='\n')
    {i++;LineNum++;}      //行号 
   else if(TempFile[i]=='"')
    IsString(LineNum,TempFile,i++,i); //字符串 
   else if(TempFile[i]=='#')
    IsMarco(LineNum,TempFile,i++,i); //宏 
   else if(IsDigit(TempFile[i]))
    IsNumber(LineNum,TempFile,i++,i); //数字(+/-/0-9)
   else if(IsLetter(TempFile[i])||TempFile[i]=='_')
    IsIdenti(LineNum,TempFile,i++,i); //标识符(字母或下划线)
   else if(IsOperate(TempFile[i]))
    IsOpt(LineNum,TempFile,i++,i);   //运算符
   else if(IsSymbol(TempFile[i]))
    IsSym(LineNum,TempFile,i++,i);   //分隔符 
   else
    i++;         //其他 
} 
OutputCode(argv[1]);
cout<<"运行结束"<<endl;
return 0;
}
//输入源代码
string InputSource(string path){
ifstream infile(path.c_str(),ios::in);
if (!infile){
   cout<<"找不到文件，请检查文件路径是否正确"<<endl;
   exit(-1);
}
char temp[3000];
infile.getline(temp,3000,EOF);
infile.close();
return temp;
}
//删除额外字符
string DeleteExtra(string code){
string temp;
int i=0;
while(code[i]){
   if(code[i]=='/'&&code[i+1]=='/')while(code[i]!='\n')i++;//过滤注释 
   while(code[i]==' '&&code[i+1]==' ')i++;      //过滤连续空格 
   if(code[i]!='\t')temp=temp+code[i];       //过滤TAB
   i++;
}
return temp;
}
//输出结果 
void OutputCode(string path){
cout<<"错误数:"<<Ei<<endl;
if(Ei>0){
   cout<<"出错了!"<<endl;
   for(int j=0;j<Ei;j++){
    cout<<"行号:"<<Error[j].line<<"\t\t"<<Error[j].content<<"\t\t原因:";
    if(Error[j].type==0)
     cout<<"未知字符";
    else if(Error[j].type==1)
     cout<<"错误的数字表示";
    else if(Error[j].type==2)
     cout<<"字符串内错误";
    cout<<endl;
   }
}
else{
   string filename=path;
   filename=filename+".txt";
   ofstream outfile(filename.c_str(),ios::out);
   if (!outfile){
    cout<<"分析结果无法保存"<<endl;
    exit(-1);
   }
   for(int j=0;j<Ri;j++){
    outfile<<"L:"<<Result[j].line<<"\t\t";
    if(Result[j].type==0)
     outfile<<"Mar";
    else if(Result[j].type==1)
     outfile<<"Num";
    else if(Result[j].type==2)
     outfile<<"Key";
    else if(Result[j].type==3)
     outfile<<"ID";
    else if(Result[j].type==4)
     outfile<<"Opt";
    else if(Result[j].type==5)
     outfile<<"Sym";
    else if(Result[j].type==6)
     outfile<<"Str";
    outfile<<"\t\t"<<Result[j].content<<endl;
   }
   outfile.close();
}
}
//保存结果
void SaveResult(int line,int type,string content){
Result[Ri].line=line;
Result[Ri].type=type;
Result[Ri].content=content;
Ri++;;
}
//保存错误
void SaveError(int line,int err,string content){
Error[Ei].line=line;
Error[Ei].type=err;
Error[Ei].content=content;
Ei++;
}
//是否是数字 
bool IsDigit(char letter){
if(letter>='0'&&letter<='9')
   return true;
return false;
}
//是否是字母 
bool IsLetter(char letter){
if(letter>='a'&&letter<='z')
   return true;
if(letter>='A'&&letter<='Z')
   return true;
return false;
}
//是否是运算符 
bool IsOperate(char letter){
char Array[7]={'*','/','=','<','>','+','-'};
for(int i=0;i<7;i++)
   if(Array[i]==letter)return true;
return false;
}
//是否是符号 
bool IsSymbol(char letter){
char Array[8]={',',';','{','}','(',')','[',']'};
for(int i=0;i<8;i++)
   if(Array[i]==letter)return true;
return false;
}
//是否是保留字 
bool IsReserved(string str){
string Array[17]={"char","int","float","void","const","break","case","continue","default","do","else","for","goto","if","return","switch","while"};
for(int i=0;i<17;i++)
   if(Array[i]==str)return true;
return false;
}
//数字类型
void IsNumber(int line,string& src,int begin,int& end){
int err=-1;
while(IsDigit(src[end]))end++;
if(src[end]=='.'){
   end++;
   while(IsDigit(src[end]))end++;
}
if(src[end]=='e'||src[end]=='E'){
   end++;
   if(src[end]=='+'||src[end]=='-'||IsDigit(src[end])){
    end++;
    while(IsDigit(src[end]))end++;
   }
   else
    err=1;
}
string TmpWord(src,begin,end-begin);
if(err!=-1)
   SaveError(line,1,TmpWord);
else if(TmpWord=="+"||TmpWord=="-")
   SaveResult(line,4,TmpWord);
else
   SaveResult(line,1,TmpWord);
}
//标识符（含关键字的区分）
void IsIdenti(int line,string& src,int begin,int& end){
while(IsLetter(src[end])||IsDigit(src[end])||src[end]=='_')end++;
string TmpWord(src,begin,end-begin);
if(IsReserved(TmpWord))
   SaveResult(line,2,TmpWord);
else
   SaveResult(line,3,TmpWord);
}
//字符串
void IsString(int line,string& src,int begin,int& end){
while(src[end]!='"'){
   if(src[end]=='\\')end++;
   else if(src[end]=='\n'){
    SaveError(line,2,"");
    break;
   }   
   end++;
}
end++;
string TmpWord(src,begin,end-begin);
SaveResult(line,6,TmpWord);
}
//宏 
void IsMarco(int line,string& src,int begin,int& end){
while(src[end]!='\n')end++;
string TmpWord(src,begin,end-begin);
SaveResult(line,0,TmpWord);
}
//运算符
void IsOpt(int line,string& src,int begin,int& end){
while(IsOperate(src[end]))end++;
string TmpWord(src,begin,end-begin);
SaveResult(line,4,TmpWord);
}
//分隔符
void IsSym(int line,string& src,int begin,int& end){
string TmpWord(src,begin,end-begin);
SaveResult(line,5,TmpWord);
}