题目：

题解：

class Solution {public List<List<String>> findLadders(String beginWord, String endWord, List<String> wordList) {List<List<String>> res = new ArrayList<>();// 因为需要快速判断扩展出的单词是否在 wordList 里，因此需要将 wordList 存入哈希表，这里命名为「字典」Set<String> dict = new HashSet<>(wordList);// 特殊用例判断if (!dict.contains(endWord)) {return res;}dict.remove(beginWord);// 第 1 步：广度优先搜索建图// 记录扩展出的单词是在第几次扩展的时候得到的，key：单词，value：在广度优先搜索的第几层Map<String, Integer> steps = new HashMap<String, Integer>();steps.put(beginWord, 0);// 记录了单词是从哪些单词扩展而来，key：单词，value：单词列表，这些单词可以变换到 key ，它们是一对多关系Map<String, List<String>> from = new HashMap<String, List<String>>();int step = 1;boolean found = false;int wordLen = beginWord.length();Queue<String> queue = new ArrayDeque<String>();queue.offer(beginWord);while (!queue.isEmpty()) {int size = queue.size();for (int i = 0; i < size; i++) {String currWord = queue.poll();char[] charArray = currWord.toCharArray();// 将每一位替换成 26 个小写英文字母for (int j = 0; j < wordLen; j++) {char origin = charArray[j];for (char c = 'a'; c <= 'z'; c++) {charArray[j] = c;String nextWord = String.valueOf(charArray);if (steps.containsKey(nextWord) && step == steps.get(nextWord)) {from.get(nextWord).add(currWord);}if (!dict.contains(nextWord)) {continue;}// 如果从一个单词扩展出来的单词以前遍历过，距离一定更远，为了避免搜索到已经遍历到，且距离更远的单词，需要将它从 dict 中删除dict.remove(nextWord);// 这一层扩展出的单词进入队列queue.offer(nextWord);// 记录 nextWord 从 currWord 而来from.putIfAbsent(nextWord, new ArrayList<>());from.get(nextWord).add(currWord);// 记录 nextWord 的 stepsteps.put(nextWord, step);if (nextWord.equals(endWord)) {found = true;}}charArray[j] = origin;}}step++;if (found) {break;}}// 第 2 步：回溯找到所有解，从 endWord 恢复到 beginWord ，所以每次尝试操作 path 列表的头部if (found) {Deque<String> path = new ArrayDeque<>();path.add(endWord);backtrack(from, path, beginWord, endWord, res);}return res;}public void backtrack(Map<String, List<String>> from, Deque<String> path, String beginWord, String cur, List<List<String>> res) {if (cur.equals(beginWord)) {res.add(new ArrayList<>(path));return;}for (String precursor : from.get(cur)) {path.addFirst(precursor);backtrack(from, path, beginWord, precursor, res);path.removeFirst();}}
}