# -*- coding: utf8 -*-
'''
__author__ = 'dabay.wang@gmail.com'
https://oj.leetcode.com/problems/longest-palindromic-substring/
Given a string S, find the longest palindromic substring in S. You may assume that the maximum length of S is 1000,
and there exists one unique longest palindromic substring.
===Comments by Dabay===
从最长的字符串开始判断,直到长度为2。但是这样的算法会超时Time Limit Exceeded。
使用一个著名的算法,叫Manacher's ALGORITHM:
每一个字符之间加上一个#来隔开,遍历的时候,从每个字符往两边延生,记录最长子串的长度。
这里加入#,避免了回文字符串是奇数或偶数的问题。
最后,在生成的记录长度的数组中,找到最大回文子串即可。
S # a # b # b # a # b # c # b # a #
P 1 2 1 2 5 2 1 4 1 2 1 6 1 2 1 2 1
mid记录包含最大右边界的回文子串的中心位置,max_right记录最大子串的右边界。
初始化位置0
对于每一个位置i
如果i不超过max_right:
判断基于mid对称点的回文长度是否超过边界
如果没有超过边界:
不用继续判断了,直接赋值为这个对称点的值,继续判断下一个i
如果超过边界:
在边界内的部分不用判断了(因为对称性,肯定是回文),继续寻找最长回文,记录长度
如果i超过了max_right:
寻找最长回文,记录长度
如果max_right得到了拓展,更新mid和max_right
'''
class Solution:
# @return a string
def longestPalindrome(self, s):
s_list = ['#']
for c in s:
s_list.append(c)
s_list.append('#')
max_right = 0
mid = 0
p = [0 for _ in s_list]
p[0] = 1
for i in xrange(1, len(p)):
if i <= max_right:
if p[2*mid-i] < max_right-i:
p[i] = p[2*mid-i]
continue
else:
j = max_right - i
while i + j < len(s_list) and i - j >= 0 and s_list[i+j] == s_list[i-j]:
j = j + 1
p[i] = j
else:
j = 1
while i + j < len(s_list) and i - j >= 0 and s_list[i+j] == s_list[i-j]:
j = j + 1
p[i] = j
if max_right < p[i] + i - 1:
max_right = p[i] + i - 1
mid = i
max_p = max(p)
max_index = p.index(max_p)
longest = ""
for i in range(max_index-max_p+1, max_index+max_p):
if s_list[i] == '#':
continue
longest = longest + s_list[i]
return longest
def main():
s = Solution()
string = "cubwqvhxdammpkwkycrqtegepyxtohspeasrdtinjhbesilsvffmnzznmltsspjwuogdyzvanalohmzrywdwqqcukjceothydlgtocukc"
#string = "abbabcba"
print s.longestPalindrome(string)
if __name__ == "__main__":
import time
start = time.clock()
main()
print "%s sec" % (time.clock() - start)