字典树又叫前缀树、单词查找树,树形结构,是哈希树的变种。能够统计、排序和保存大量的字符串,经常被搜索引擎系统用于文本词频统计。优点是利用字符串的公共前缀来减少查询时间,最大程度减少无谓字符串的比较,查询效率高于哈希树。
最近恶补数据结构,在leetcode刷题遇到了一种比较有意思的数据结构——字典树。具体实现不难,写篇短文记录一下。
字典树的原理是这样的,假设我现在有一些字符串["a","ab","abc","abcd"],它们的特点是具有公共前缀,适合使用树形数据结构保存,从根节点到叶子结点的一条路径就能保存所有具有公共前缀的单词。
使用Python实现非常简单,考虑到删除操作涉及较少,暂不实现,简洁版代码如下:
class TreeNode(object):
def __init__(self):
self.nodes = {} # 记录当前结点的子结点
self.is_leaf = False # 当前结点是否表示一个单词
self.count = 0 # 单词树中单词的总量
def insert(self,word):
curr = self
for c in word:
if not curr.nodes.get(c,None):
new_node = TreeNode()
curr.nodes[c] = new_node
curr = curr.nodes[c]
curr.is_leaf = True
self.count += 1
return
def insert_many(self,words):
for word in words:
self.insert(word)
return
def search(self,word):
curr = self
try:
for c in word:
curr = curr.nodes[c]
except:
return False
return curr.is_leaf
简单测试一下:
实战1:720. 词典中最长的单词
解决本题要点如下:
- 对字符串数组排序,构建字典树
- 遍历数组,将每个单词插入字典树,并记录满足题目要求的最长单词
- 在遍历的过程中,如果已经出现前缀缺失,及时剪枝
class Solution:
def longestWord(self, words: List[str]) -> str:
class TreeNode:
def __init__(self):
self.nodes = {}
self.best = 0
self.best_word = ""
def add(self,word):
count = 0
curr = self
for c in word[:-1]:
# 当前结点无子结点,发生前缀缺失
if not curr.nodes.get(c,None):
# 剪枝
return
new_node = TreeNode()
curr.nodes[c] = new_node
curr = curr.nodes[c]
count += 1
curr.nodes[word[-1]] = TreeNode()
count += 1
if count > self.best:
self.best = count
self.best_word = word
return
def add_many(self,words):
for word in words:
self.add(word)
return
words.sort()
head = TreeNode()
head.add_many(words)
return head.best_word
实战2:648. 单词替换
首先想到的当然是切词之后遍历,每个词都find一下词根,查到就替换 。但我们现在实现了字典树,就要用高级的方法了 :)
本题要点如下:
- 使用字典树保存词根
- 将 sentence 切割为单词,保存在数组中
- 遍历数组中的单词,如果访问至叶子结点(is_leaf == True),证明单词具有该词根,将其替换
class Solution:
def replaceWords(self, dictionary: List[str], sentence: str) -> str:
class TreeNode:
def __init__(self):
self.nodes = {}
self.is_leaf = False
def add(self,word):
curr = self
for c in word:
if not curr.nodes.get(c,None):
new_node = TreeNode()
curr.nodes[c] = new_node
curr = curr.nodes[c]
curr.is_leaf = True
def add_many(self,words):
for word in words:
self.add(word)
def replace(self,word):
curr = self
s = ""
for c in word:
if curr.is_leaf == True:
return s
if not curr.nodes.get(c,None):
return word
s += c
curr = curr.nodes[c]
return word
head = TreeNode()
head.add_many(dictionary)
return " ".join(map(head.replace,sentence.split(" ")))