使用C++语言统计文件中出现频率排前10的单词

思路：

1. 从文件中读出每一行，将字符串中的标点符号替代为空格；
2. 使用字符串输入流将字符串里的各个单词赋给单个单词变量；
3. 将单词和出现的次数构成映射放入map中；
4. 排序: 复制map到vector中，重载>运算符进行排序。

代码：

#include<iostream>
#include<map>
#include<vector>
#incude<algorithm>
#include<string>
using namespace std;

class CWord
{
	string word;
public:
	CWord(string word)
	{
		this->word = word;

	}

	string GetWord() const { return word; }

	bool operator < (const CWord &w) const // 用于添加集合
	{
		return word < w.GetWord();
	}
	bool operator == (const string &s) const // 用于查询
	{
		return this->word == s;
	}
};


class CWordMap
{
private:
	map<CWord, int> wordmap;
public:
	bool AddString(string s)
	{
		map<CWord, int>::iterator it = wordmap.find(s);
		if (it == wordmap.end())
		{
			pair<CWord, int> p(CWord(s), 1);
			wordmap.insert(p);
		}
		else
		{
			(*it).second += 1;
		}

		return true;
	}

	void Show(ostream &os)
	{
		map<CWord, int>::iterator it = wordmap.begin();
		while (it!=wordmap.end())
		{
			string ss = ((*it).first).GetWord();
			int n = (*it).second;
			os << ((*it).first).GetWord() << "\t" << (*it).second << endl;
			it++;
			
		}
	}



	map<CWord, int>::iterator begin()
	{
		return wordmap.begin();
	}

	map<CWord, int>::iterator end()
	{
		return wordmap.end();
	
	}

	
};

bool compare(pair<CWord,int> &mp1, pair<CWord,int> &mp2)
{
	return mp1.second > mp2.second;
}

int  main()
{
	CWordSet wordset;
	CWordMap wordmap;
	int pos = 0;
	string s = "";
	string deliset = ",.";
	ifstream in("data.txt");
	while (!in.eof())
	{
		getline(in, s);
		if (s == "")
		{
			continue;
		}
		pos = 0;
		while ((pos = s.find_first_of(deliset,pos))!=string::npos)
		{
			s.replace(pos, 1, " ");
		}
		istringstream str(s);
		while (!str.eof())
		{
			str >> s;
			if (s == "")
				continue;
			wordmap.AddString(s);
			//wordset.AddString(s);

		}
		
	}
	in.close();
	cout<<"未排序前："<<endl;
	wordmap.Show(cout);
	vector<pair<CWord, int>>::iterator it= v.begin();
	int n = 0;
	cout<<"排序后："<<endl;
	while (it!= v.end())
	{
		if (n != 10)
		{
			cout << (*it).first.GetWord() << ":" << (*it).second << endl;
			it++;
			n++;
		}
		else
			break;
		
	}
	return 0;
}

程序运行截图：