【C++】统计文本词频程序

#include <iostream>
#include <fstream>
#include <string>
#include <iomanip>
#include <vector>
#include <map>
#include <cctype>
#include <algorithm>
bool cmp(std::pair<std::string, size_t>& a,
         std::pair<std::string, size_t>& b){
    return a.second > b.second;
}//词频从大到小排序

int main(){
    std::ifstream ifs{};
    ifs.open("./input.txt", std::ifstream::in);//从input.txt读入文本
    std::string text( (std::istreambuf_iterator<char>(ifs) ),
                       (std::istreambuf_iterator<char>()    ) );
    //std::istreambuf_iterator<char>()表示文件结尾
    ifs.close();
    const std::string separators{" ():;.[],\n"};//分隔符，可自定义
    std::map<std::string, size_t> mp;
    std::string word{};
    size_t start{text.find_first_not_of(separators)}, end{};
    while(start != std::string::npos){
        end = text.find_first_of(separators, start+1);
        word = text.substr(start, end-start);
        std::transform(word.begin(), word.end(), word.begin(),
            [](unsigned char c){ return std::tolower(c); });//将单词统一转换为小写字母
        ++mp[word];
        start =text.find_first_not_of(separators, end+1);
    }
    size_t cnt{};
    const size_t perline{6};//每行输出6个单词和词频
    std::vector<std::pair<std::string, size_t> > v;
    copy(mp.begin(),
            mp.end(),
            std::back_inserter<std::vector<std::pair<std::string, size_t> > >(v));
    sort(v.begin(), v.end(), cmp);
    std::stringstream ss{};
    for(const auto& item : v){
        ss << std::right << std::setw(15) << item.first << " : " << std::left << std::setw(5) << item.second;
        if(++cnt % perline == 0){
            ss << std::endl;
        } 
    }
    ss << std::endl;
    std::ofstream ofs{};
    ofs.open("./res.txt", std::ofstream::out | std::ofstream::trunc );//结果输出到res.txt中
    ofs << ss.str();
    ofs.close();
    return 0;
}