编译原理（1）简单词法分析程序（C++实现）

1. 程序流程图

简单词法分析程序流程图

 2. C++代码实现

#include <bits/stdc++.h>
using namespace std;

/// 函数声明
void getInput();
void continueOrExit();
string trim(string str);
void controller(string input);
bool isLegalStr(string str);
void checkController(string input);
void k_v_pair(string single);
void k_v_check(string key, string value);
int key_check(string key);
void value_check(string value);

/// 全局变量
int int_num;
int string_num;
int float_num;
int char_num;
int over_flag = 0;  // 非法语句标志
const int DupError = -1; // 重名错误返回
const int True = 1;
const int False = 0;
vector<string> keys;


/// 开始了
int main()
{
    getInput();
    return 0;
}

/*
* 初始化并取得输入函数
*/
void getInput()
{
    // 初始化
    over_flag = 0;
    int_num = 0;
    string_num = 0;
    float_num = 0;
    char_num = 0;
    vector <string>().swap(keys);
    cout<<"Please input a string: "<<endl;
    string input;
    getline(cin, input);
    controller(input);
}
/*
* 继续输入/结束函数
*/
void continueOrExit()
{
    cout<<"-------------------------------------------------------"<<endl;
    cout<<endl;
    cout<<"Do you want to continue or quit?  1. Continue | 2. Exit"<<endl;
    cout<<"Your choice: "<<endl;;
    int op;
    cin>>op;
    cin.ignore();
    switch(op)
    {
    case 1:
        getInput();
        break;
    case 2:
        exit(0);
        break;
    default:
        cout<<"Error input! Please try again!"<<endl;
        continueOrExit();
    }
}
/*
* 滤首尾空格函数
*/
string trim(string str)
{
    if(str.empty())
        return str;
    str.erase(0, str.find_first_not_of(" "));
    str.erase(str.find_last_not_of(" ")+1);
    return str;
}
/*
* 判断字符串是否为合法的常量说明语句
* 串首const  串尾;
*/
bool isLegalStr(string str)
{
    bool legal = str.find("const")!=string::npos && str.find_last_of(';')==str.size()-1;
    return legal;
}
/*
* 控制器-->处理字符串是否合法两个分支
*/
void controller(string input)
{
    input = trim(input);
    if(isLegalStr(input))
    {
        // 字符串合法
        // 去const
        input.erase(0, 5);
        checkController(input);
    }
    else
    {
        // 字符串不合法
        cout<<"It is not a constant declaration satement!"<<endl;
        continueOrExit();
    }
}
/*
* 控制器-->拆分单个赋值语句，并进行分析，统计各类型值的数量
*/
void checkController(string input)
{
    string single = "";
    for(int i=0; i<input.size(); i++)
    {
        if(over_flag ==1 )
        {
            cout<<"Illegal input!"<<endl;
            break;
        }
        if(input[i]==',' || input[i]==';')
        {
            if(input[i]==';')
                over_flag = 1;
            k_v_pair(single);
            single="";
            continue;
        }
        single += input[i];
    }
    cout<<"int_num = "<<int_num<<"; ";
    cout<<"float_num = "<<float_num<<"; ";
    cout<<"char_num = "<<char_num<<"; ";
    cout<<"string_num = "<<string_num<<"."<<endl;
    continueOrExit();
}
/*
* key，value拆分
*/
void k_v_pair(string single)
{
    cout<<"single是对的吗？"<<single<<endl;
    string key = "";
    string value = "";
    int flag = 1;
    for(int i=0; i<single.size(); i++)
    {
        if(flag==1 && single[i]=='=')
        {
            flag = 0;
            continue;
        }
        if(flag==1)
            key += single[i];
        else
            value += single[i];
    }
    k_v_check(key, value);
}
/*
* key，value检查
*/
void k_v_check(string key, string value)
{
    key = trim(key);
    value = trim(value);
    int k_flag = key_check(key);
    if(k_flag == False)
    {
        cout<<key<<"(Wrong! It is not a identifier!)"<<endl;
    }
    else if(k_flag == True)
    {
        value_check(value);
    }
}
/*
* key检查
*/
int key_check(string key)
{
    cout<<key;
    regex r_key("[a-zA-Z_][a-zA-Z0-9_]*");
    if(regex_match(key, r_key))
    {
        // key重名检查
        vector<string>::iterator it = find(keys.begin(), keys.end(), key);
        if(it!=keys.end())
        {
            cout<<"(Wrong!Duplication of name.)"<<endl;
            return DupError;
        }
        else
        {
            keys.push_back(key);
            return True;
        }
    }
    return False;
}
/*
* value检查
*/
void value_check(string value)
{
    regex r_num1("[+|-]?[0-9]*");
    regex r_num2("[[1-9]{1}[e|E]{1}[+]?[0-9]*");
    regex r_float1("[+|-]?[0-9]*[.][0-9]+");
    regex r_float2("[+|-]?[0-9]*[.][0-9]+[e|E]{1}[+|-]?[0-9]*");

    // 判断int
    if(regex_match(value,r_num1) || regex_match(value, r_num2))
    {
        int_num++;
        cout<<"(integer,"<<value<<")"<<endl;
    }

    // 判断float
    else if(regex_match(value, r_float1) || regex_match(value, r_float2))
    {
        float_num++;
        cout<<"(float,"<<value<<")"<<endl;
    }

    // 判断char
    else if(value.find("'")==0 && value.find_last_of("'")==value.size()-1)
    {
        if(value.size()==3)
        {
            char_num++;
            value.erase(0,1);
            value.erase(value.size()-1, 1);
            cout<<"(char,"<<value<<")"<<endl;
        }
        else
        {
            cout<<"(Wrong!There are more than one char in ''.)"<<endl;
        }
    }

    // 判断string
    else if(value.find("\"")==0 && value.find_last_of("\"")==value.size()-1)
    {
        string_num++;
        value.erase(0,1);
        value.erase(value.size()-1, 1);
        cout<<"(string,"<<value<<")"<<endl;
    }

    // 其他错误情况
    else
    {
        cout<<"(Wrong!The format of the value string is not correct!)"<<endl;
    }
}

3. Bug总结

（1）判断由科学计数法表示的整数的正则表达式不完整，存在缺漏。

（2）以逗号分割单句表达式，当char或string语句中存在逗号时，会出错。
        改进思路：当逗号存在于char或string中时，进行优先级判断。