135 lines
4.3 KiB
C++
135 lines
4.3 KiB
C++
#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()
|
|
{
|
|
/// 2.4 全国公司的城市分布
|
|
|
|
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);
|
|
|
|
map<string,int> mp;
|
|
conn.exec("select job_city,count(*) from (select job_city from lagou_job group by company_name) as A group by job_city",
|
|
[&](MySQLResult& res)
|
|
{
|
|
res.stepRow([&](char** val,unsigned long* len)
|
|
{
|
|
string city(val[0]);
|
|
int cnt=ParseInt(val[1]);
|
|
|
|
auto iter=mp.find(city);
|
|
if(iter!=mp.end())
|
|
{
|
|
iter->second += cnt;
|
|
}
|
|
else
|
|
{
|
|
mp.insert(make_pair(city,cnt));
|
|
}
|
|
});
|
|
});
|
|
conn.exec("use bigdata5",nullptr);
|
|
conn.exec("select job_city,count(*) from (select job_city from pages group by company_name) as B group by job_city",
|
|
[&](MySQLResult& res)
|
|
{
|
|
res.stepRow([&](char** val,unsigned long* len)
|
|
{
|
|
string city(val[0]);
|
|
int cnt=ParseInt(val[1]);
|
|
|
|
auto iter=mp.find(city);
|
|
if(iter!=mp.end())
|
|
{
|
|
iter->second += cnt;
|
|
}
|
|
else
|
|
{
|
|
mp.insert(make_pair(city,cnt));
|
|
}
|
|
});
|
|
});
|
|
|
|
conn.exec("use bigdata2",nullptr);
|
|
conn.exec("select job_city,count(*) from (select job_city from pages group by company_name) as B group by job_city",
|
|
[&](MySQLResult& res)
|
|
{
|
|
res.stepRow([&](char** val,unsigned long* len)
|
|
{
|
|
string city(val[0]);
|
|
int cnt=ParseInt(val[1]);
|
|
|
|
auto iter=mp.find(city);
|
|
if(iter!=mp.end())
|
|
{
|
|
iter->second += cnt;
|
|
}
|
|
else
|
|
{
|
|
mp.insert(make_pair(city,cnt));
|
|
}
|
|
});
|
|
});
|
|
|
|
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?
|
|
/// target+=this : Add current to the previous
|
|
xiter->second+=iter->second;
|
|
|
|
/// 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;
|
|
}
|
|
|
|
vector<pair<string,int>> vec;
|
|
for(const auto& pr:mp)
|
|
{
|
|
vec.push_back(pr);
|
|
}
|
|
sort(vec.begin(),vec.end(),[](const pair<string,int>& a,const pair<string,int>& b)
|
|
{
|
|
return a.second > b.second;
|
|
});
|
|
|
|
ofstream ofs("result/2.4.csv");
|
|
#define cout ofs
|
|
cout<<"城市,公司数量"<<endl;
|
|
int szVec=vec.size();
|
|
for(int i=0;i<szVec;i++)
|
|
{
|
|
cout<<vec[i].first<<","<<vec[i].second<<endl;
|
|
}
|
|
#undef cout
|
|
|
|
return 0;
|
|
}
|