编译原理词法分析器

直接上代码：

#include <cstdio>
#include <cstring>
#include <string>
#include <iostream>
#include <map>
#include <algorithm>
using namespace std;

map<string,int>mp_res;
map<string,int>mp_oper;
map<char,int>mp_dili;
string pre="";

int flag1=false,flag2=1;//flag1用作判断/*或*/是否在引号内，flag2判断内容是否处于注释状态 
string mstr,flagstr,ansstr="";

string double_slash(string s)//双斜杠注释处理 
{
    string ans="";
    bool flag=false;//判断//是否在双引号内作为字符串 
    for(int i=0;i<s.length();i++)
    {
        if(s[i]=='/'&&s[i+1]=='/'&&i+1<s.length()&&!flag)
            break;
        if(s[i]==34)
            flag=!flag;
        ans+=s[i];
    }
    return ans;    
}

string block_comments(string s)//去除/*  */的注释 
{
    string ans;
    for(int i=0;i<s.length();i++)
    {
        if(s[i]==34)
            flag1=!flag1; 
        if(s[i]=='/'&&s[i+1]=='*'&&i+1<s.length()&&!flag1)
        {
            flag2=0;
            i+=2;
        }
        if(s[i]=='*'&&s[i+1]=='/'&&i+1<s.length()&&!flag1)
        {
            flag2=1;
            i+=2;
        }
        if(flag2)
        ans+=s[i];
    }
    return ans;
 } 

string space(string s)//多个连续空格并为一个，去除tab 
{
    bool flag=1;
    string ans="";
    for(int i=0;i<s.length();i++)
    {
        if(s[i]=='\t')
            continue;
        if(s[i]!=' ')
            flag=0;
        if(flag&&s[i]==' ')
            continue;
        if(s[i]==' ')
            flag=1;
            ans+=s[i];    
    }
    return ans;
}

void init_table()//初始化对应值 
{
    string resword[35]={"int","long","short","float","double","char","bool","unsigned","signed","const","void","volatile",
        "enum","struct","union","if","else","goto","switch","case","do","while","for","continue","break","return",
        "default","typedef","auto","register","extern","static","sizeof","string","namespace"}; 
    string operchar[33]={"+","-","*","/","%","++","--",
        ">","<","==",">=","<=","!=",
        "&&","||","!",
        "&","|","~","^","<<",">>",
        "=","+=","-=","*=","/=","&=","|=","^=",">>=","<<=","."};
    char dilimiters[11]={'(',')','{','}','[',']',';','\'','"','#',','};    
    //标识符为1，数字为2 
    for(int i=0;i<35;i++)
        mp_res[resword[i]]=i+2; 
    for(int i=0;i<33;i++)
        mp_oper[operchar[i]]=i+37;
    for(int i=0;i<11;i++)
        mp_dili[dilimiters[i]]=i+70;
}

int is_digit(string s)//判断是否是数字 
{
    if(s[0]>='0'&&s[0]<='9')
    {
        for(int i=0;i<s.length();i++)
        {
            if(!(s[i]>='0'&&s[i]<='9'))
                return -1;
        }
        return 1;
    }
    return 0;
}

void print()//处理输出 
{
    string temp="",oper="";
    int x;
    for(int i=0;i<ansstr.length();i++)
    {
        string ts=""; ts+=ansstr[i];
        if(ansstr[i]!=' '&&mp_dili[ansstr[i]]==0&&mp_oper[ts]==0)
            temp+=ansstr[i];
        else
        {
            if(temp.size()>0)
            {
                cout<<temp;
                x=30-temp.length();
                for(int j=1;j<=x;j++)
                printf(" ");
                printf("|");
                if(mp_res[temp]!=0)
                {    
                    printf("%10d",mp_res[temp]);
                    printf("          |");
                    printf("          keyword\n");
                    pre="keyword";
                }
                else if(is_digit(temp)==1)
                {
                    printf("%10d",2);
                    printf("          |");
                    printf("          digit");
                    if(pre=="digit")
                        cout<<"(error)"<<endl;
                    else
                        cout<<endl;
                    pre="digit";
                }    
                else if(is_digit(temp)==0)
                {
                    printf("%10d",1);
                    printf("          |");
                    printf("          identifier");
                    if(pre=="identifier")
                        cout<<"(error)"<<endl;
                    else
                        cout<<endl;
                    pre="identifier";  
                 }
                 else
                 {
                     printf("%10d",-1);
                     printf("          |");
                     printf("          ERROR\n");
                 } 
                temp="";
                
            }
            if(mp_dili[ansstr[i]]!=0)
            {
                cout<<ansstr[i];
                for(int j=1;j<=29;j++)
                printf(" ");
                printf("|");
                printf("%10d",mp_dili[ansstr[i]]);
                printf("          |");
                printf("          delimiter\n");
                pre="delimiter";
            }
            if(mp_oper[ts]!=0)
            {
                while(mp_oper[ts]!=0)
                {
                    oper=ts;
                    i++;
                    ts+=ansstr[i];
                }
                cout<<oper;
                x=30-oper.length();
                for(int j=1;j<=x;j++)
                printf(" ");
                printf("|");
                printf("%10d",mp_oper[oper]);
                oper="";
                i--;
                printf("          |");
                printf("          operator\n");
                pre="operator";
            }
            
        }
            
     }
 } 
 
int main()
{
    freopen("datain.txt","r",stdin);
    freopen("dataout.txt","w+",stdout);
    cout<<"单词符号                      |       种别码       |          种类"<<endl; 
    while(getline(cin,mstr))//边读边去注释 
    {
        mstr=double_slash(mstr);
        ansstr+=block_comments(mstr);
        ansstr+=' ';
    }    
    ansstr=space(ansstr);//合并空格 
    init_table();     
    print(); 
}