Trie树

Trie树，即字典树或单词查找树，主要用于大量字符串的检索、去重、排序等操作。

主要原理就是利用字符串的公共前缀建立一棵多叉树，牺牲空间换取时间。

//Trie树
#include <iostream>
#include <string>
using std::cin;
using std::cout;
using std::endl;
using std::string;

const int SIZE = 26;
const char BASE = 'a';

class TrieNode {
public:
    //is_terminal means that this char is the last one of this string
    //we can get this string by going from the root to this terminal
    bool is_terminal;
    TrieNode **childs;
    TrieNode() {
        is_terminal = false;
        childs = new TrieNode*[SIZE];
        for (int i = 0; i < SIZE; ++i) {
            childs[i] = nullptr;
        }
    }
    ~TrieNode() {
        for (int i = 0; i < SIZE; ++i) {
            delete childs[i];
        }
        delete[] childs;
    }
};

class StringPatterns {
private:
    TrieNode *root;
public:
    StringPatterns() {
        root = new TrieNode();
    }
    ~StringPatterns() {
        delete root;
    }
    void add_pattern(const string &pattern) {
        TrieNode *now=root;
        for(int i=0;i<pattern.length();i++){
            //this char is new in current level, means that this char is from a new different string
            if(now->childs[pattern[i]-BASE]==nullptr){
                //add this char
                now->childs[pattern[i]-BASE]=new TrieNode();
            }
            //go to next char
            now=now->childs[pattern[i]-BASE];
        }
        //now points to the last char of this added string
        now->is_terminal=true;
    }
    bool match(const string &buffer) const {
        TrieNode *now=root;
        for(int i=0;i<buffer.length();i++){
            if(now->childs[buffer[i]-BASE]==nullptr){
                return false;
            }
            now=now->childs[buffer[i]-BASE];
        }
        return now->is_terminal;
    }
};

int main() {
    StringPatterns trie;
    int n;
    cin >> n;
    for (int i = 0; i < n; ++i) {
        string pattern;
        cin >> pattern;
        trie.add_pattern(pattern);
    }
    string str_buffer;
    cin >> str_buffer;
    if (trie.match(str_buffer)) {
        cout << "match success" << endl;
    } else {
        cout << "match failed" << endl;
    }
    return 0;
}

 

 用Trie树求一个string的所有不相同的子串：

//find all different substrings of a string
#include <iostream>
#include <string>
using std::cin;
using std::cout;
using std::endl;
using std::string;

const int SIZE = 26;
const char BASE = 'a';

class TrieNode {
public:
    bool is_terminal;
    TrieNode **childs;
    TrieNode() {
        is_terminal=false;
        childs=new TrieNode*[SIZE];
        for(int i=0;i<SIZE;i++){
            childs[i]=nullptr;
        }
    }
    ~TrieNode() {
        for(int i=0;i<SIZE;i++){
            delete childs[i];
        }
        delete[] childs;
    }
};

class StringPatterns {
private:
    TrieNode *root;
    //number of substrings
    int subs;
public:
    StringPatterns() {
        subs=0;
        root=new TrieNode();
    }
    ~StringPatterns() {
        delete root;
    }
    void add_pattern(const string &pattern) {
        TrieNode *now=root;
        for(int i=0;i<pattern.length();i++){
            now=root;
            //add in all substrings starting with pattern[j]
            for(int j=i;j<pattern.length();j++){
                if(now->childs[pattern[j]-BASE]==nullptr){
                    //means this is a new different substring
                    now->childs[pattern[j]-BASE]=new TrieNode();
                    subs++;
                }
                now=now->childs[pattern[j]-BASE];
            }
        }
        cout<<subs<<endl;
    }
    
    
};

int main() {
    StringPatterns strp;
    string it;
    cin>>it;
    strp.add_pattern(it);
    return 0;
}