从零开始的CPP（36）——操作Excel

现有一个Excel A1.csv。

其表格第一列为：生物样本的名称；其他列为：生物样本的含量。表格第一行第一列是空格，第一行其他列为:受试者名称。

需求

如下：设计一个程序，可以指定受试者名称（某列），对该列所有生物样本含量进行排序，保留含量前X名的生物样本，其余生物样本含量合并，命名为others。

设计思路：

1.如何根据受试者名称（第一行），读取csv中指定某列数据？

cpp 复制代码

vector<string> getRownames() {
    ifstream file("A1.csv");
    string line;
    string rowname;
    vector<string> rownames;
    getline(file, line);
    for (int i = 0; i < line.length();) {
        if (line[i] == ',') {
            rowname = line.substr(0, i);
            rownames.push_back(rowname);
            line = line.substr(i + 1, line.length());
            i = 0;
        }
        else i++;
    }
    for (auto i : rownames) {
        cout << i << " ";
    }
    return rownames;
}

通过这段代码，就可以得到一个储存每列名称的vector了。

cpp 复制代码

int findRowname(vector<string> rownames,string aim) {
    for (int i = 0; i < rownames.size(); i++) {
        if (rownames[i] == aim) {
            return i;
        }
    }
}

通过这段代码，就可以得到目标受试者所在列了。

2.如何获得目标受试者所在列的数据及其对应样本名称？

cpp 复制代码

vector<pair<string, double>> getNameAndData(int row) {
    vector<pair<string, double>> NameAndData;
    ifstream file("A1.csv");
    string line;
    string rowdata;
    double value;
    getline(file, line);
    while (getline(file, line)) {
        vector<string> rowdatas;
        for (int i = 0; i < line.length();) {
            if (line[i] == ',') {
                rowdata = line.substr(0, i);
                rowdatas.push_back(rowdata);
                line = line.substr(i + 1, line.length());
                i = 0;
            }
            else i++;
        }            
        value = stod(rowdatas[row]);
        NameAndData.push_back({ rowdatas[0], value });
    }
    for (auto i : NameAndData) {
        cout << i.first << " " << i.second << endl;
    }
    return NameAndData;
}

循环读取文件每一行，把每一行都存进vector，找到vector的0列（样本名）与row列（受试者所在列），将其pair，存入新vector。

这里有代码重复了，就是从line转换为vector的代码，可以将其封装。

cpp 复制代码

void split(string line, vector<string>& linedatas) {
    string linedata;
    for (int i = 0; i < line.length();) {
        if (line[i] == ',') {
            linedata = line.substr(0, i);
            linedatas.push_back(linedata);
            line = line.substr(i + 1, line.length());
            i = 0;
        }
        else i++;
    }
    return;
}

需要时直接传入line和vector就可以啦。

3. 排序，并保留前X行，剩余行合并为others

复习了lambda表达式与sort函数结合的用法

cpp 复制代码

vector<pair<string, double>> getNewNameAndData(vector<pair<string, double>> NameAndData,int reserve) {
    sort(NameAndData.begin(), NameAndData.end(), [](const auto& a, const auto& b) {
        return a.second > b.second;
        });
    vector<pair<string, double>> NewNameAndData(reserve + 1);
    for (int i = 0; i < reserve; i++) {
        NewNameAndData[i] = NameAndData[i];
    }
    double reservesum = 0;
    for (int i = reserve; i < NameAndData.size(); i++) {
        reservesum = reservesum + NameAndData[i].second;
    }
    NewNameAndData[reserve] = { "Others",reservesum };
    for (auto i : NewNameAndData) {
        cout << i.first << " " << i.second << endl;
    }
    return NewNameAndData;
}

4.将new vector写入新文件

cpp 复制代码

void getNewCsv(vector<pair<string, double>> NewNameAndData,int row) {
    ofstream file("newA.csv");
    ifstream file1("A1.csv");
    string line;
    getline(file1, line);
    vector<string> data0;
    split(line, data0);
    file << data0[0] << "," << data0[row] << endl;
    for (int i = 0; i < NewNameAndData.size(); i++) {
        file << NewNameAndData[i].first << "," << NewNameAndData[i].second << endl;
    }
    return;
}

5.完整代码

cpp 复制代码

#include <iostream>
#include <fstream>
#include <sstream>
#include <iostream>
#include <fstream>
#include <sstream>
#include <string>
#include <vector>
#include <algorithm>
#include <cassert>
using namespace std;
class Solution {
public:
    vector<string> getRownames() {
        ifstream file("A1.csv");
        string line;
        string rowname;
        vector<string> rownames;
        getline(file, line);
        split(line, rownames);
        /*
        for (auto i : rownames) {
            cout << i << " ";
        }
        */
        return rownames;
    }

    int findRowname(vector<string> rownames,string aim) {
        for (int i = 0; i < rownames.size(); i++) {
            if (rownames[i] == aim) {
                return i;
            }
        }
        return 0;
    }

    vector<pair<string, double>> getNameAndData(int row) {
        vector<pair<string, double>> NameAndData;
        ifstream file("A1.csv");
        string line;
        string rowdata;
        double value;
        getline(file, line);
        while (getline(file, line)) {
            vector<string> rowdatas;
            split(line, rowdatas);
            value = stod(rowdatas[row]);
            NameAndData.push_back({ rowdatas[0], value });
        }
        /*
        for (auto i : NameAndData) {
            cout << i.first << " " << i.second << endl;
        }
        */
        return NameAndData;
    }
    vector<pair<string, double>> getNewNameAndData(vector<pair<string, double>> NameAndData,int reserve) {
        sort(NameAndData.begin(), NameAndData.end(), [](const auto& a, const auto& b) {
            return a.second > b.second;
            });
        vector<pair<string, double>> NewNameAndData(reserve + 1);
        for (int i = 0; i < reserve; i++) {
            NewNameAndData[i] = NameAndData[i];
        }
        double reservesum = 0;
        for (int i = reserve; i < NameAndData.size(); i++) {
            reservesum = reservesum + NameAndData[i].second;
        }
        NewNameAndData[reserve] = { "Others",reservesum };
        /*
        for (auto i : NewNameAndData) {
            cout << i.first << " " << i.second << endl;
        }
        */
        return NewNameAndData;
    }

    void getNewCsv(vector<pair<string, double>> NewNameAndData,int row,string aim, int reverse) {
        stringstream ss;
        ss << reverse;
        string s1 = ss.str();
        string str = aim + "_"+s1+"_newA.csv";
        ofstream file(str);
        ifstream file1("A1.csv");
        string line;
        getline(file1, line);
        vector<string> data0;
        split(line, data0);
        file << data0[0] << "," << data0[row] << endl;
        for (int i = 0; i < NewNameAndData.size(); i++) {
            file << NewNameAndData[i].first << "," << NewNameAndData[i].second << endl;
        }
        cout << "文件已经保存在"<< str <<"中。";
        return;
    }
    
private:
    void split(string line, vector<string>& linedatas) {
        string linedata;
        for (int i = 0; i < line.length();) {
            if (line[i] == ',') {
                linedata = line.substr(0, i);
                linedatas.push_back(linedata);
                line = line.substr(i + 1, line.length());
                i = 0;
            }
            else i++;
        }
        return;
    }
};

int main()
{
    Solution solution;
    vector<string> rownames=solution.getRownames();
    // 用户输入目标名
    string aim;
    cout << "请输入目标名: ";
    cin >> aim;
    int row=solution.findRowname(rownames, aim);
    //cout << row;
    vector<pair<string, double>> NameAndData = solution.getNameAndData(row);
    // 用户输入保留行数
    int reverse;
    cout << "请输入保留行数: ";
    cin >> reverse;
    vector<pair<string, double>> NewNameAndData = solution.getNewNameAndData(NameAndData, reverse);
    solution.getNewCsv(NewNameAndData,row,aim,reverse);
    
    return 0;
}