/**
 * @brief Function for generating set of First(a)
 * @author 立秋小豬
 * @time: 2021/10/13
 * @notice: 要求產(chǎn)生體句型不得有空格
 *          左遞歸的產(chǎn)生體中必須有空串（必須能夠終結(jié)）
 *          char '#' act as varepsilon 
 * **/

#include <iostream>
#include <unordered_map>
#include <vector>
#include <string>
#include <fstream>
#include <unordered_set>

using namespace std;

unordered_map<string, vector<string>> P; //產(chǎn)生式P的集合

void scan(){
    //scan函數(shù)實(shí)現(xiàn)從文件掃描文法，將對(duì)應(yīng)的產(chǎn)生式加入到映射P中
    fstream fs;
    string input;
    fs.open("lan.txt");
    if(!fs.is_open()){ // 文件打開(kāi)失敗
        cout << "Error: Could not open the file" << endl;
        exit(-1);
    }
    fs >> input;
    while(input != "end"){
        string VN = input; // 產(chǎn)生式的非終結(jié)符

        fs >> input; //跳過(guò)推導(dǎo)符號(hào)
        if (input != "->" && input != "→"){
            cout << "Error: undefined symbol [" << input << "]" << endl;
            exit(-2);
        }

        fs >> input; //產(chǎn)生體拆開(kāi)后加入到set集合中，默認(rèn)推導(dǎo)符號(hào)后必有一個(gè)產(chǎn)生體
        P[VN].emplace_back(input);
        while( fs >> input && input == "|"){
                fs >> input;
                P[VN].emplace_back(input);
        }
    }
}

// void generate(){
// }

unordered_set<char> First(const string& str){
    // 終結(jié)符以及空串情況下, whether has the VN or not
    if(str == "" || str == "#" || P.find(str) == P.end())
        return {};
    if(!(str[0] >= 'A' && str[0] <= 'Z'))
        return {str[0]};

    vector<string> bodys = P[str]; // str -> bodys
    unordered_set<char> res = {};
    for(auto &s: bodys){
        bool hasBlank = true;//是否含有空串，是否繼續(xù)讀產(chǎn)生體
        for (int i = 0; i < s.size() && hasBlank; ++i){
            if(s[i] >= 'A' && s[i] <= 'Z'){//是否為終結(jié)符
                unordered_set<char> temp = {};//遞歸的臨時(shí)集
                string next;
                if(i < s.size() - 1 && s[i + 1] == '\''){ // 大寫字母 + ' 的非終結(jié)符
                    next = s.substr(i, 2);
                    ++i;
                }else{ //僅僅是大寫字母的終結(jié)符
                    next = s[i];
                }
                if(next != str){ //避免無(wú)限遞歸，默認(rèn)自身是含有空串（hasBlank為True）
                    temp = First(next); //遞歸求解
                    hasBlank = false; //先默認(rèn)temp中沒(méi)有空串
                    for(auto &c : temp)
                        if(c == '#')
                            hasBlank = true;//temp中發(fā)現(xiàn)了空串
                        else
                            res.emplace(c);
                }
            }else{
                res.emplace(s[i]);
                hasBlank = false;//默認(rèn)連接的終結(jié)符不為空，故此終結(jié)符后不會(huì)再有新元素加入First集
            }
        }
        if(hasBlank) //產(chǎn)生體中所有非終結(jié)符都包含空串，則將空串加入first集中
            res.emplace('#');
    }
    return res;
}

 

int main(){
    // unordered_map<string, vector<char>> First; //First集合
    scan();
    cout << "輸入的產(chǎn)生式如下：\n"
         << "********************************\n";
    for(auto &[vn, bodys]: P){
        cout << vn << " -> " << bodys[0];
        for (int i = 1; i < bodys.size(); ++i)
            cout << " | " << bodys[i];
        cout << endl;
    }
    cout << "********************************\n";

    for(auto &[vn,_]: P){
        unordered_set<char> f = First(vn);
        cout << "First(" << vn << ") : ";
        auto iter = f.begin();

        if(iter != f.end()){
            cout << *iter;
            while(++iter != f.end()){
                cout << " , " << *iter;
            }
        }
        cout << endl;
    }

    return 0;
}

E -> TE'
E' -> +TE' | #
T -> FT'
T' -> *FT' | #
F -> (E) | id
end

S -> SaRb | #
R -> RSQ | #
Q -> e
end