题目描述
给出 字符串 text
和 字符串列表 words
, 返回所有的索引对 [i, j]
使得在索引对范围内的子字符串 text[i]...text[j]
(包括 i
和 j
)属于字符串列表 words
。
示例 1:
输入: text = "thestoryofleetcodeandme", words = ["story","fleet","leetcode"]
输出: [[3,7],[9,13],[10,17]]
示例 2:
输入: text = "ababa", words = ["aba","ab"]
输出: [[0,1],[0,2],[2,3],[2,4]]
解释:
注意,返回的配对可以有交叉,比如,"aba" 既在 [0,2] 中也在 [2,4] 中
提示:
- 所有字符串都只包含小写字母。
- 保证
words
中的字符串无重复。
1 <= text.length <= 100
1 <= words.length <= 20
1 <= words[i].length <= 50
- 按序返回索引对
[i,j]
(即,按照索引对的第一个索引进行排序,当第一个索引对相同时按照第二个索引对排序)。
解法
方法一:暴力枚举
| class Solution:
def indexPairs(self, text: str, words: List[str]) -> List[List[int]]:
words = set(words)
n = len(text)
return [
[i, j] for i in range(n) for j in range(i, n) if text[i : j + 1] in words
]
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41 | class Trie {
Trie[] children = new Trie[26];
boolean isEnd = false;
void insert(String word) {
Trie node = this;
for (char c : word.toCharArray()) {
c -= 'a';
if (node.children[c] == null) {
node.children[c] = new Trie();
}
node = node.children[c];
}
node.isEnd = true;
}
}
class Solution {
public int[][] indexPairs(String text, String[] words) {
Trie trie = new Trie();
for (String w : words) {
trie.insert(w);
}
int n = text.length();
List<int[]> ans = new ArrayList<>();
for (int i = 0; i < n; ++i) {
Trie node = trie;
for (int j = i; j < n; ++j) {
int idx = text.charAt(j) - 'a';
if (node.children[idx] == null) {
break;
}
node = node.children[idx];
if (node.isEnd) {
ans.add(new int[] {i, j});
}
}
}
return ans.toArray(new int[ans.size()][2]);
}
}
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39 | class Trie {
public:
vector<Trie*> children;
bool isEnd = false;
Trie() {
children.resize(26);
}
void insert(string word) {
Trie* node = this;
for (char c : word) {
c -= 'a';
if (!node->children[c]) node->children[c] = new Trie();
node = node->children[c];
}
node->isEnd = true;
}
};
class Solution {
public:
vector<vector<int>> indexPairs(string text, vector<string>& words) {
Trie* trie = new Trie();
for (auto w : words) trie->insert(w);
int n = text.size();
vector<vector<int>> ans;
for (int i = 0; i < n; ++i) {
Trie* node = trie;
for (int j = i; j < n; ++j) {
int idx = text[j] - 'a';
if (!node->children[idx]) break;
node = node->children[idx];
if (node->isEnd) ans.push_back({i, j});
}
}
return ans;
}
};
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43 | type Trie struct {
children [26]*Trie
isEnd bool
}
func newTrie() *Trie {
return &Trie{}
}
func (this *Trie) insert(word string) {
node := this
for _, c := range word {
idx := int(c - 'a')
if node.children[idx] == nil {
node.children[idx] = newTrie()
}
node = node.children[idx]
}
node.isEnd = true
}
func indexPairs(text string, words []string) [][]int {
trie := newTrie()
for _, w := range words {
trie.insert(w)
}
n := len(text)
var ans [][]int
for i := range text {
node := trie
for j := i; j < n; j++ {
idx := int(text[j] - 'a')
if node.children[idx] == nil {
break
}
node = node.children[idx]
if node.isEnd {
ans = append(ans, []int{i, j})
}
}
}
return ans
}
|
方法二:前缀树
相似题目: