#include "MySQLWrapper.h"
#include <iostream>
#include <sstream>
#include <vector>
#include <map>
#include <set>
#include <cstdlib>
#include <cstdio>
#include <cstring>
#include <algorithm>
#include <fstream>
using namespace std;

#include "local_db_account.h"
#include "util.hpp"

int main()
{
    /// 3.3 大数据岗位需求量前十个

    MySQLConn conn;
    conn.connect("127.0.0.1",db_user,db_passwd,"bigdata3",3306);
    cout<<"Connected to DB"<<endl;
    conn.exec("set names gbk",nullptr);

    auto fn1=[](map<string,vector<int>>& mp)
    {
        for(auto iter=mp.begin(); iter!=mp.end(); )
        {
            bool updated=false;
            for(auto xiter=mp.begin(); xiter!=iter; )
            {
                if(iter->first.find(xiter->first)!=string::npos)
                {
                    /// Found something?
                    /// targetvec.push_back(thisvec) : Add current vector to the previous vector.
                    xiter->second.insert(xiter->second.end(),
                    iter->second.begin(),iter->second.end());


                    /// iter should now be deleted. Get next iter first
                    auto nowiter=iter;
                    ++iter;
                    mp.erase(nowiter);

                    /// Stop internal loop.
                    updated=true;
                    break;
                }
                else ++xiter;
            }

            if(!updated) ++iter;
        }

        for(auto& pr:mp)
        {
            int sum=0;
            int n=pr.second.size();
            for(int i=0; i<n; i++)
            {
                sum+=pr.second[i];
            }
            pr.second[0]=sum;
        }

    };

    /// title, vector<int>
    map<string,vector<int>> dmp;

    conn.exec("use bigdata3",nullptr);
    conn.exec("select title,count(*) as want_num from lagou_job where ((title like '%大数据%' ) or ( tags like '%大数据%')) group by title order by want_num desc",[&](MySQLResult& res)
              {
                  res.stepRow([&](char** val,unsigned long* len)
                              {
                                  string title(val[0]);
                                  int cc=ParseInt(val[1]);
                                  dmp[Trim(title)].push_back(cc);
                              });
              });

    conn.exec("use bigdata5",nullptr);
    conn.exec("select title,count(*) as want_num from pages where ((title like '%大数据%' ) or ( tags like '%大数据%')) group by title order by want_num desc",[&](MySQLResult& res)
              {
                  res.stepRow([&](char** val,unsigned long* len)
                              {
                                  string city(val[0]);
                                  int cc=ParseInt(val[1]);
                                  dmp[Trim(city)].push_back(cc);
                              });
              });

    conn.exec("use bigdata2",nullptr);
    conn.exec("select title,count(*) as want_num from pages where ( title like '%大数据%' ) group by title order by want_num desc",[&](MySQLResult& res)
              {
                  res.stepRow([&](char** val,unsigned long* len)
                              {
                                  string city(val[0]);
                                  int cc=ParseInt(val[1]);
                                  dmp[Trim(city)].push_back(cc);
                              });
              });

    fn1(dmp);

    vector<pair<string,int>> vec;
    for(const auto& pr:dmp)
    {
        vec.push_back(make_pair(pr.first,pr.second[0]));
    }
    sort(vec.begin(),vec.end(),[](const pair<string,int>& a,const pair<string,int>& b)
         {
             return a.second > b.second;
         });

    ofstream ofs("result/3.3.csv");
    #define cout ofs
    cout<<"岗位名称,需求量"<<endl;
    int szLoop=min(10u,vec.size());
    for(int i=0;i<szLoop;i++)
    {
        cout<<vec[i].first<<","<<vec[i].second<<endl;
    }
    #undef cout

    return 0;
}