面试题 16.02. 单词频率(C++)

面试题 16.02. 单词频率(C++)

设计一个方法，找出任意指定单词在一本书中的出现频率。

你的实现应该支持如下操作：

WordsFrequency(book)构造函数，参数为字符串数组构成的一本书
get(word)查询指定单词在书中出现的频率

示例：
WordsFrequency wordsFrequency = new WordsFrequency({“i”, “have”, “an”, “apple”, “he”, “have”, “a”, “pen”});
wordsFrequency.get(“you”); //返回0，"you"没有出现过
wordsFrequency.get(“have”); //返回2，"have"出现2次
wordsFrequency.get(“an”); //返回1
wordsFrequency.get(“apple”); //返回1
wordsFrequency.get(“pen”); //返回1

一开始，我想，这不简单吗，直接vector循环查找即可，暴力法

暴力破解（失败）

class WordsFrequency {
public:
    vector<string> mybook;
    WordsFrequency(vector<string>& book) {
        this->mybook = book;
    }
    
    int get(string word) {
        int times = 0;
        for (vector<string>::iterator it = this->mybook.begin(); it != this->mybook.end(); it++) {
            if(*it == word)times++;
	    }
        return times;
    }
};

/**
 * Your WordsFrequency object will be instantiated and called as such:
 * WordsFrequency* obj = new WordsFrequency(book);
 * int param_1 = obj->get(word);
 */

然后当测试样例堆积的时候，就超出时间限制了…

于是想起来，查找肯定用map啊，于是乎第二份代码

map实现

class WordsFrequency {
public:
    map<string,int> map;
    WordsFrequency(vector<string>& book) {
        for(string str:book)
            map[str]++;
    }
    
    int get(string word) {
        return map[word];
    }
};

/**
 * Your WordsFrequency object will be instantiated and called as such:
 * WordsFrequency* obj = new WordsFrequency(book);
 * int param_1 = obj->get(word);
 */

unordered_map实现

在评论区发现别人并不是用map实现的，更多的是使用unordered_map，无序图？经过查阅才知道，map和unordered_map差不多，但是底层的实现不一样，map是用红黑树实现的，而unordered_map是用哈希表实现的。由于此题并无要求有序，所以理论上查找用哈希表更快。（至于为什么不用hash_map，他们两个的内部结构都是采用哈希表来实现。区别在哪里？unordered_map在C++11的时候被引入标准库了，而hash_map没有，所以建议还是使用unordered_map比较好。）

class WordsFrequency {
public:
    unordered_map<string,int> umap;
    WordsFrequency(vector<string>& book) {
        for(string str:book)
            umap[str]++;
    }
    
    int get(string word) {
        return umap[word];
    }
};

/**
 * Your WordsFrequency object will be instantiated and called as such:
 * WordsFrequency* obj = new WordsFrequency(book);
 * int param_1 = obj->get(word);
 */

速度明显提高，但是内存占用也提高了，原因：两者的内存占有率的问题就转化成 红黑树 VS hash表 , 总体上还是unorder_map占用的内存要高。

字典树实现

参考字典树解法

// 字典树的结点
struct trie {
    int n;
    trie* son[26];
    trie (): n(0) {
        for (int i = 0; i < 26; ++i) {
            son[i] = nullptr;
        }
    }
};

class WordsFrequency {
private:
    trie* root;
public:
    WordsFrequency(vector<string>& book) {
        root = new trie();
        trie* tmp = root;
        for (auto& i : book) {
            tmp = root;
            for (auto& ch : i) {
                int next = ch - 'a';
                if (!tmp->son[next]) {
                    tmp->son[next] = new trie();
                }
                tmp = tmp->son[next];
            }
            // 到达底部后, 将叶子结点++
            ++tmp->n;
        }
    }
    int get(string word) {
        trie* find = root;
        for (auto& i : word) {
            int next = i - 'a';
            if (find->son[next]) find = find->son[next];
            else return 0;
        }
        return find->n;
    }
};

/**
 * Your WordsFrequency object will be instantiated and called as such:
 * WordsFrequency* obj = new WordsFrequency(book);
 * int param_1 = obj->get(word);
 */

虽然时间和空间都很低，但也是一个好思路。

补充说明

auto是C++11的特性，可以自动为变量选择匹配的类型。

• for (auto i : v) ：修改i,v中元素不会改变
• for (auto &i : v)：修改i,v中元素会改变