传统的Trie实现简单,但是占用的空间实在是难以接受,特别是当字符集不仅限于英文26个字符的时候,爆炸起来的空间根本无法接受。
网站建设哪家好,找成都创新互联!专注于网页设计、网站建设、微信开发、微信小程序开发、集团企业网站建设等服务项目。为回馈新老客户创新互联还提供了高安免费建站欢迎大家使用!
双数组Trie就是优化了空间的Trie树,原理本文就不讲了,请参考An Efficient Implementation of Trie Structures,本程序的编写也是参考这篇论文的。
关于几点论文没有提及的细节和与论文不一一致的实现:
1.对于插入字符串,如果有一个字符串是另一个字符串的子串的话,我是将结束符也作为一条边,产生一个新的结点,这个结点新节点的Base我置为0
所以一个字符串结束也有2中情况:一个是Base值为负,存储剩余字符(可能只有一个结束符)到Tail数组;另一个是Base为0。
所以在查询的时候要考虑一下这两种情况
2.对于***种冲突(论文中的Case 3),可能要将Tail中的字符串取出一部分,作为边放到索引中。论文是使用将尾串左移的方式,我的方式直接修改Base值,而不是移动尾串。
下面是java实现的代码,可以处理相同字符串插入,子串的插入等情况
- /*
- * Name: Double Array Trie
- * Author: Yaguang Ding
- * Mail: dingyaguang117@gmail.com
- * Blog: blog.csdn.net/dingyaguang117
- * Date: 2012/5/21
- * Note: a word ends may be either of these two case:
- * 1. Base[cur_p] == pos ( pos<0 and Tail[-pos] == 'END_CHAR' )
- * 2. Check[Base[cur_p] + Code('END_CHAR')] == cur_p
- */
- import java.util.ArrayList;
- import java.util.HashMap;
- import java.util.Map;
- import java.util.Arrays;
- public class DoubleArrayTrie {
- final char END_CHAR = '\0';
- final int DEFAULT_LEN = 1024;
- int Base[] = new int [DEFAULT_LEN];
- int Check[] = new int [DEFAULT_LEN];
- char Tail[] = new char [DEFAULT_LEN];
- int Pos = 1;
- Map
CharMap = new HashMap (); - ArrayList
CharList = new ArrayList (); - public DoubleArrayTrie()
- {
- Base[1] = 1;
- CharMap.put(END_CHAR,1);
- CharList.add(END_CHAR);
- CharList.add(END_CHAR);
- for(int i=0;i<26;++i)
- {
- CharMap.put((char)('a'+i),CharMap.size()+1);
- CharList.add((char)('a'+i));
- }
- }
- private void Extend_Array()
- {
- Base = Arrays.copyOf(Base, Base.length*2);
- Check = Arrays.copyOf(Check, Check.length*2);
- }
- private void Extend_Tail()
- {
- Tail = Arrays.copyOf(Tail, Tail.length*2);
- }
- private int GetCharCode(char c)
- {
- if (!CharMap.containsKey(c))
- {
- CharMap.put(c,CharMap.size()+1);
- CharList.add(c);
- }
- return CharMap.get(c);
- }
- private int CopyToTailArray(String s,int p)
- {
- int _Pos = Pos;
- while(s.length()-p+1 > Tail.length-Pos)
- {
- Extend_Tail();
- }
- for(int i=p; i
- {
- Tail[_Pos] = s.charAt(i);
- _Pos++;
- }
- return _Pos;
- }
- private int x_check(Integer []set)
- {
- for(int i=1; ; ++i)
- {
- boolean flag = true;
- for(int j=0;j
- {
- int cur_p = i+set[j];
- if(cur_p>= Base.length) Extend_Array();
- if(Base[cur_p]!= 0 || Check[cur_p]!= 0)
- {
- flag = false;
- break;
- }
- }
- if (flag) return i;
- }
- }
- private ArrayList
GetChildList(int p) - {
- ArrayList
ret = new ArrayList (); - for(int i=1; i<=CharMap.size();++i)
- {
- if(Base[p]+i >= Check.length) break;
- if(Check[Base[p]+i] == p)
- {
- ret.add(i);
- }
- }
- return ret;
- }
- private boolean TailContainString(int start,String s2)
- {
- for(int i=0;i
- {
- if(s2.charAt(i) != Tail[i+start]) return false;
- }
- return true;
- }
- private boolean TailMatchString(int start,String s2)
- {
- s2 += END_CHAR;
- for(int i=0;i
- {
- if(s2.charAt(i) != Tail[i+start]) return false;
- }
- return true;
- }
- public void Insert(String s) throws Exception
- {
- s += END_CHAR;
- int pre_p = 1;
- int cur_p;
- for(int i=0; i
- {
- //获取状态位置
- cur_p = Base[pre_p]+GetCharCode(s.charAt(i));
- //如果长度超过现有,拓展数组
- if (cur_p >= Base.length) Extend_Array();
- //空闲状态
- if(Base[cur_p] == 0 && Check[cur_p] == 0)
- {
- Base[cur_p] = -Pos;
- Check[cur_p] = pre_p;
- Pos = CopyToTailArray(s,i+1);
- break;
- }else
- //已存在状态
- if(Base[cur_p] > 0 && Check[cur_p] == pre_p)
- {
- pre_p = cur_p;
- continue;
- }else
- //冲突 1:遇到 Base[cur_p]小于0的,即遇到一个被压缩存到Tail中的字符串
- if(Base[cur_p] < 0 && Check[cur_p] == pre_p)
- {
- int head = -Base[cur_p];
- if(s.charAt(i+1)== END_CHAR && Tail[head]==END_CHAR) //插入重复字符串
- {
- break;
- }
- //公共字母的情况,因为上一个判断已经排除了结束符,所以一定是2个都不是结束符
- if (Tail[head] == s.charAt(i+1))
- {
- int avail_base = x_check(new Integer[]{GetCharCode(s.charAt(i+1))});
- Base[cur_p] = avail_base;
- Check[avail_base+GetCharCode(s.charAt(i+1))] = cur_p;
- Base[avail_base+GetCharCode(s.charAt(i+1))] = -(head+1);
- pre_p = cur_p;
- continue;
- }
- else
- {
- //2个字母不相同的情况,可能有一个为结束符
- int avail_base ;
- avail_base = x_check(new Integer[]{GetCharCode(s.charAt(i+1)),GetCharCode(Tail[head])});
- Base[cur_p] = avail_base;
- Check[avail_base+GetCharCode(Tail[head])] = cur_p;
- Check[avail_base+GetCharCode(s.charAt(i+1))] = cur_p;
- //Tail 为END_FLAG 的情况
- if(Tail[head] == END_CHAR)
- Base[avail_base+GetCharCode(Tail[head])] = 0;
- else
- Base[avail_base+GetCharCode(Tail[head])] = -(head+1);
- if(s.charAt(i+1) == END_CHAR)
- Base[avail_base+GetCharCode(s.charAt(i+1))] = 0;
- else
- Base[avail_base+GetCharCode(s.charAt(i+1))] = -Pos;
- Pos = CopyToTailArray(s,i+2);
- break;
- }
- }else
- //冲突2:当前结点已经被占用,需要调整pre的base
- if(Check[cur_p] != pre_p)
- {
- ArrayList
list1 = GetChildList(pre_p); - int toBeAdjust;
- ArrayList
list = null; - if(true)
- {
- toBeAdjust = pre_p;
- list = list1;
- }
- int origin_base = Base[toBeAdjust];
- list.add(GetCharCode(s.charAt(i)));
- int avail_base = x_check((Integer[])list.toArray(new Integer[list.size()]));
- list.remove(list.size()-1);
- Base[toBeAdjust] = avail_base;
- for(int j=0; j
- {
- //BUG
- int tmp1 = origin_base + list.get(j);
- int tmp2 = avail_base + list.get(j);
- Base[tmp2] = Base[tmp1];
- Check[tmp2] = Check[tmp1];
- //有后续
- if(Base[tmp1] > 0)
- {
- ArrayList
subsequence = GetChildList(tmp1); - for(int k=0; k
- {
- Check[Base[tmp1]+subsequence.get(k)] = tmp2;
- }
- }
- Base[tmp1] = 0;
- Check[tmp1] = 0;
- }
- //更新新的cur_p
- cur_p = Base[pre_p]+GetCharCode(s.charAt(i));
- if(s.charAt(i) == END_CHAR)
- Base[cur_p] = 0;
- else
- Base[cur_p] = -Pos;
- Check[cur_p] = pre_p;
- Pos = CopyToTailArray(s,i+1);
- break;
- }
- }
- }
- public boolean Exists(String word)
- {
- int pre_p = 1;
- int cur_p = 0;
- for(int i=0;i
- {
- cur_p = Base[pre_p]+GetCharCode(word.charAt(i));
- if(Check[cur_p] != pre_p) return false;
- if(Base[cur_p] < 0)
- {
- if(TailMatchString(-Base[cur_p],word.substring(i+1)))
- return true;
- return false;
- }
- pre_p = cur_p;
- }
- if(Check[Base[cur_p]+GetCharCode(END_CHAR)] == cur_p)
- return true;
- return false;
- }
- //内部函数,返回匹配单词的最靠后的Base index,
- class FindStruct
- {
- int p;
- String prefix="";
- }
- private FindStruct Find(String word)
- {
- int pre_p = 1;
- int cur_p = 0;
- FindStruct fs = new FindStruct();
- for(int i=0;i
- {
- // BUG
- fs.prefix += word.charAt(i);
- cur_p = Base[pre_p]+GetCharCode(word.charAt(i));
- if(Check[cur_p] != pre_p)
- {
- fs.p = -1;
- return fs;
- }
- if(Base[cur_p] < 0)
- {
- if(TailContainString(-Base[cur_p],word.substring(i+1)))
- {
- fs.p = cur_p;
- return fs;
- }
- fs.p = -1;
- return fs;
- }
- pre_p = cur_p;
- }
- fs.p = cur_p;
- return fs;
- }
- public ArrayList
GetAllChildWord(int index) - {
- ArrayList
result = new ArrayList (); - if(Base[index] == 0)
- {
- result.add("");
- return result;
- }
- if(Base[index] < 0)
- {
- String r="";
- for(int i=-Base[index];Tail[i]!=END_CHAR;++i)
- {
- r+= Tail[i];
- }
- result.add(r);
- return result;
- }
- for(int i=1;i<=CharMap.size();++i)
- {
- if(Check[Base[index]+i] == index)
- {
- for(String s:GetAllChildWord(Base[index]+i))
- {
- result.add(CharList.get(i)+s);
- }
- //result.addAll(GetAllChildWord(Base[index]+i));
- }
- }
- return result;
- }
- public ArrayList
FindAllWords(String word) - {
- ArrayList
result = new ArrayList (); - String prefix = "";
- FindStruct fs = Find(word);
- int p = fs.p;
- if (p == -1) return result;
- if(Base[p]<0)
- {
- String r="";
- for(int i=-Base[p];Tail[i]!=END_CHAR;++i)
- {
- r+= Tail[i];
- }
- result.add(fs.prefix+r);
- return result;
- }
- if(Base[p] > 0)
- {
- ArrayList
r = GetAllChildWord(p); - for(int i=0;i
- {
- r.set(i, fs.prefix+r.get(i));
- }
- return r;
- }
- return result;
- }
- }
测 试
- import java.io.BufferedReader;
- import java.io.FileInputStream;
- import java.io.IOException;
- import java.io.InputStream;
- import java.io.InputStreamReader;
- import java.util.ArrayList;
- import java.util.Scanner;
- import javax.xml.crypto.Data;
- public class Main {
- public static void main(String[] args) throws Exception {
- ArrayList
words = new ArrayList (); - BufferedReader reader = new BufferedReader(new InputStreamReader(new FileInputStream("E:/兔子的试验学习中心[课内]/ACM大赛/ACM第四届校赛/E命令提示/words3.dic")));
- String s;
- int num = 0;
- while((s=reader.readLine()) != null)
- {
- words.add(s);
- num ++;
- }
- DoubleArrayTrie dat = new DoubleArrayTrie();
- for(String word: words)
- {
- dat.Insert(word);
- }
- System.out.println(dat.Base.length);
- System.out.println(dat.Tail.length);
- Scanner sc = new Scanner(System.in);
- while(sc.hasNext())
- {
- String word = sc.next();
- System.out.println(dat.Exists(word));
- System.out.println(dat.FindAllWords(word));
- }
- }
- }
下面是测试结果,构造6W英文单词的DAT,大概需要20秒
我增长数组的时候是每次长度增加到2倍,初始1024
Base和Check数组的长度为131072
Tail的长度为262144
原文地址:Java实现双数组Trie树(DoubleArrayTrie,DAT)
当前标题:Java实现双数组Trie树(DoubleArrayTrie,DAT)
本文URL:http://www.shufengxianlan.com/qtweb/news19/238969.html
网站建设、网络推广公司-创新互联,是专注品牌与效果的网站制作,网络营销seo公司;服务项目有等
声明:本网站发布的内容(图片、视频和文字)以用户投稿、用户转载内容为主,如果涉及侵权请尽快告知,我们将会在第一时间删除。文章观点不代表本网站立场,如需处理请联系客服。电话:028-86922220;邮箱:631063699@qq.com。内容未经允许不得转载,或转载时需注明来源: 创新互联