题目大意是传入一条字符串,计算出这样的这样一条子字符串,要求子字符串是原字符串的连续的某一段,且子字符串内不包含两个或两个以上的重复字符。求符合上面条件的字符串中最长的那一条的长度。


首先注意到任意一条无重复子字符串的任意子字符串应该也满足无重复这一特性。因此这个问题可以用动态规划解决。

需要先计算出字符串s的每个元素对应的首个重复字符的下标,字符串中下标为i的元素的首个重复字符的下标j应该满足j > i && s[i] == s[j],且是满足这些条件中的最小值。比如asaa中下标为0的元素对应的首个重复字符的下标为2,尽管3也满足条件,但不是最小值。

计算的方式如下:

registries = int[256]

nextOccurance = int[s.length]

initialize elements in registries with -1 //将所有registries 中的元素都初始化为-1

initialize elements in nextOccurance with s.length //将所有registries 中的元素都初始化为s.length

for(i = 0; i < s.length; i++)

  c = s[i]

  if registries[c] != -1 then

    nextOccurance [registries[c]] = i

  registries[c] = i

这里用注册表registries的注册项registries[c]记录当前对字符c的查找请求,而nextOccurance的元素nextOccurance[i]表示s[i]的首个重复字符子s中的下标。每次循环时,都会检测注册表项registries[c]是否以及被注册,如果注册则为nextOccurance [registries[c]]记录当前下标i。最后则将之前的注册项清除,而用当前下标作为新的请求项。

之后使用动态规划解决问题。利用数组longestLengthes,longestLengthes[i]表示所有以下标i作为起始的无重复子字符串的最长长度。显然longestLengthes[s.length - 1]应该为1。而对于任意i < s.length - 1,应该有longestLengthes[i] = min(longestLengthes[i + 1] + 1, nextOccurance [i] - i)。其中longestLengthes[i + 1] + 1表示理想状态下(不考虑出现多个longestLengthes[i]对应的字符)的最优长度,而nextOccurance [i] - i表示可能的以下标i作为起始的无重复子字符串的最长长度,因为s[i] == s[nextOccurance [i]],这里已经发生了重复。之所以敢保证longestLengthes[i] <= longestLengthes[i + 1] + 1是因为假如longestLengthes[i]>longestLengthes[i + 1] + 1,那么显然longestLengthes[i + 1] >= longestLengthes[i] - 1 >longestLengthes[i + 1](整体不重复自然局部不会重复),这是不可能发生的。(如果对这部分不理解,可以枚举可能的情况,总共只有三种)转换成代码:

longestLengthes = int[s.length]

longestLengthes[s.length - 1] = 1

for(i = s.length - 2; i >= 0; i--)

  longestLengthes[i] = min(longestLengthes[i + 1] + 1, nextOccurance [i] - i)

之后遍历整个longestLengthes数组就可以找到最长的无重复子字符串的长度。

把上面两部分的代码整合,可以轻易得出整体的复杂度为O(n),其中n为传入字符串的长度。


最后给出整个实现代码,给有需要的人:

package cn.dalt.leetcode;

/**
 * Created by Administrator on 2017/6/4.
 */
public class LongestSubstringWithoutRepeatingCharacters {
    public static void main(String[] args) {
        String s = "yiwgczzovxdrrgeebkqliobitcjgqxeqhbxkcyaxvdqplxtmhmarcbzwekewkknrnmdpmfohlfyweujlgjf";
        System.out.println(new LongestSubstringWithoutRepeatingCharacters().lengthOfLongestSubstring(s));
        for(char c = 0; c < 256; c++)
        {
            System.out.println((int)c + ":" + c);
        }
    }

    int[] nextOccurIndexes = null;
    int[] registries = new int[1 << 8];

    public int lengthOfLongestSubstring(String s) {
        //Calculate all next occur indexes
        //nextOccurIndexes[i] is the minimun index of s which has properties that index > i && s[i] = s[index]
        int slength = s.length();
        if (slength == 0) {
            return 0;
        }
        nextOccurIndexes = new int[s.length()];
        for (int i = 0, bound = registries.length; i < bound; i++) {
            registries[i] = -1;
        }
        for (int i = 0, bound = s.length(); i < bound; i++) {
            int c = s.charAt(i);
            int registry = registries[c];
            if (registry != -1) {
                nextOccurIndexes[registry] = i;
            }
            registries[c] = i;
        }
        for (int registry : registries) {
            if (registry != -1) {
                nextOccurIndexes[registry] = slength;
            }
        }

        int[] longestNoneRepetitionSubstringLengthes = new int[s.length()];
        longestNoneRepetitionSubstringLengthes[s.length() - 1] = 1;
        int longestNoneRepetitionSubstringIndex = s.length() - 1;
        for (int i = s.length() - 2; i >= 0; i--) {
            int probablyMaxLength1 = longestNoneRepetitionSubstringLengthes[i + 1] + 1;
            int probablyMaxLength2 = nextOccurIndexes[i] - i;
            longestNoneRepetitionSubstringLengthes[i] = Math.min(probablyMaxLength1, probablyMaxLength2);
            longestNoneRepetitionSubstringIndex =
                    longestNoneRepetitionSubstringLengthes[i] > longestNoneRepetitionSubstringLengthes[longestNoneRepetitionSubstringIndex] ?
                            i : longestNoneRepetitionSubstringIndex;
        }
        return longestNoneRepetitionSubstringLengthes[longestNoneRepetitionSubstringIndex];
    }
}

Leetcode:Longest Substring Without Repeating Characters分析和实现的更多相关文章

  1. [LeetCode] Longest Substring Without Repeating Characters 最长无重复子串

    Given a string, find the length of the longest substring without repeating characters. For example, ...

  2. leetcode: longest substring without repeating characters

    July 16, 2015 Problem statement: Longest Substring Without Repeating Characters Read the blog: http: ...

  3. LeetCode:Longest Substring Without Repeating Characters(最长不重复子串)

    题目链接 Given a string, find the length of the longest substring without repeating characters. For exam ...

  4. [LeetCode] Longest Substring Without Repeating Characters 最长无重复字符的子串

    Given a string, find the length of the longest substring without repeating characters. Example 1: In ...

  5. C++ leetcode Longest Substring Without Repeating Characters

    要开学了,不开森.键盘声音有点大,担心会吵到舍友.今年要当个可爱的技术宅呀~ 题目:Given a string, find the length of the longest substring w ...

  6. [LeetCode]Longest Substring Without Repeating Characters题解

    Longest Substring Without Repeating Characters: Given a string, find the length of the longest subst ...

  7. [LeetCode] Longest Substring Without Repeating Characters 最长无重复字符的子串 C++实现java实现

    最长无重复字符的子串 Given a string, find the length of the longest substring without repeating characters. Ex ...

  8. LeetCode——Longest Substring Without Repeating Characters

    Given a string, find the length of the longest substring without repeating characters. For example, ...

  9. [Leetcode] Longest Substring Without Repeating Characters (C++)

    题目: Given a string, find the length of the longest substring without repeating characters. For examp ...

随机推荐

  1. 再读《Java编程思想 》

    前段时间在豆瓣上无意间看到一个帖子"我为什么把thinking in java 读了10遍",是11年的帖子,下面评论至今,各种声音都有,不过大多数还是佩服和支持的.我个人来讲也是 ...

  2. c++ 霍夫变换检测直线

    通常这是一幅边缘图像,比如来自 Canny算子.cv:: Houghlines函数的输出是cV::Vec2f向量,每个元素都是一对代表检测到的直线的浮点数(p,0).在下例中我们首先应用 Canny算 ...

  3. Java 对称加密

    最近在做一个用户 token 功能,学习了加密相关 AES/DES.RSA 等.其中涉及一个对称和非对称加密问题.对称加密虽然没有非对称加密那样安全性高,但好处是加密速度快,但某些场合还是可以选择使用 ...

  4. I.MX6 View长宽大于屏的分辨率

    /******************************************************************************** * I.MX6 View长宽大于屏的 ...

  5. Python之functools库

    functools库用于高阶函数,指那些作用于函数或者返回其他函数的函数 functools提供方法如下: cmp_to_key 将一个比较函数转换关键字函数 partial 针对函数起作用,并且是部 ...

  6. Python--csv文件处理

    CSV(Comma-Separator Values)逗号分割值,由于是纯文本文件,任何编辑器都可以打开.下面用csv和pandas两种方式进行csv文件操作 原始csv文件内容 Supplier N ...

  7. flask的request的用法

    其中在头部取值是这样的,request.headers,得到的是一个字典 参考链接:http://blog.csdn.net/yannanxiu/article/details/53116652

  8. Cocos2d-x -自己定义动作 圆周运动

    原文地址:http://blog.csdn.net/u012945598/article/details/17605409 在之前的文章中我们以前讲过Cocos2d-x中的各种动作的用法,我们先来简单 ...

  9. C#网络编程(同步传输字符串) - Part.2

    服务端客户端通信 在与服务端的连接建立以后,我们就可以通过此连接来发送和接收数据.端口与端口之间以流(Stream)的形式传输数据,因为几乎任何对象都可以保存到流中,所以实际上可以在客户端与服务端之间 ...

  10. 11g R2 rac linstener 监听配置

    两个节点host,ipvip ,scan的信息 #eth0-Public IP 162.12.0.1    cqltjcpt1 162.12.0.3    cqltjcpt2 #eth1 PRIVAT ...