C++ 對一段英文進行詞頻統計

/**
 * 對一段英文的詞頻統計
 * @Author [email protected]
 * date 2010/06/17
 */
#include <iostream>
#include <string>

using namespace std;
/**
 * 單詞對象
 */
struct Word
{
    Word() : Str(""), Count(0)
    {}
    string Str;
    int Count;

    /**
     * 交換單詞(用於排序)
     * @param word 交換的單詞對象
     */
    void exchange(Word &word)
    {
        string tStr = word.Str;
        int tCount = word.Count;
        word.Str = Str;
        word.Count = Count;
        Str = tStr;
        Count = tCount;
    }
};

/**
 * 統計詞頻
 * @param words 單詞數組
 * @param newWord 單詞內容
 * @param size 單詞總數
 */
void CalcCount(Word * words, string &newWord, int size)
{
    int i = 0;
    for(; i < size; i++)
    {
        if(words[i].Str == newWord)
        {
            words[i].Count++;
            return;
        }
        else if(words[i].Str == "")
            break;
    }
    words[i].Str = newWord;
    words[i].Count = 1;
}

/**
 * 以單詞出現頻率降序排列單詞
 * @param words 單詞數組
 * @param size 單詞數量
 */
void SortWordDown(Word * words, int size)
{
    for(int i = 0; i < size; i++)
    {
        for(int j = 0; j < size-1; j++)
        {
            if(words[j].Count <  words[j+1].Count)
            {
                words[j].exchange(words[j+1]);
            }
        }
    }
}

int main()
{
    Word * words;
    string content;
    cout << "輸入一段英文:";
    getline(cin, content);

    //計算單詞總數
    int wCount = 1;
    for(unsigned int i = 0; i < content.length(); i++)
    {
        if(content[i] == ' ')
            wCount++;
    }
    words = new Word[wCount];

    string::size_type offset = content.find(' ');//單詞都是以空格隔開
    while(offset != string::npos)
    {
        string wStr = content.substr(0, offset);
        content.erase(0, offset+1);
        CalcCount(words, wStr, wCount);
        offset = content.find(' ');
    }
    CalcCount(words, content, wCount);//計算最後一個單詞

    SortWordDown(words, wCount);
    int printCount = wCount < 5 ? wCount : 5;
    cout << "出現頻率最高的前" << printCount << "個單詞是:" << endl;
    for(int i = 0; i < printCount; i++)
    {
        cout << words[i].Str << "/t頻率:" << words[i].Count << "次" << endl;
    }

    delete [] words;
    return 0;
}

 

效果: