127. Word Ladder via java

A classic BFS problem, that is, we use a Queue to store temporary solution and record the size of the current Queue. 

For each node in this BFS tree, we try k times of transformation, k is the length of the word.

We record every word which already has been seen in case of a loop. Thus we need a HashSet.

public class Solution {
  // use BFS to solve
  // in every level of BFS searching, 
  // we find the word that replaced by one char and also in the dict without have seen before
  // data structure: seen(HashSet), cur(Queue)
  // assume that the wordList is not null, case insensitive, no duplicate, every word is non-empty
  public int ladderLength(String beginWord, String endWord, List<String> wordList) {
    // Write your solution here
    HashSet<String> seen = new HashSet<>();
    Queue<String> cur = new LinkedList<>();
    seen.add(beginWord);
    cur.offer(beginWord);
    int cnt = 1;
    while(cur.size()>0){
      cnt++;
      int size = cur.size();
      for(int i=0; i<size; i++){
        String todo = cur.poll();
        for(int j=0; j<todo.length(); j++){
          List<String> changed= getQualified(todo, j, wordList, seen);
          for(String s: changed){
            if(String.valueOf(s)==String.valueOf(endWord)){
              return cnt;
            }else{
              cur.offer(s);
              seen.add(s);
            }
          }
        }
      }
    }
    return 0;
  }

  private List<String> getQualified(String from, int ignore, List<String> wordList, HashSet<String> seen){
    List<String> res = new ArrayList<>();
    for(String s: wordList){
      int flag = 0;
      for(int i=0; i<from.length(); i++){
        if(i!=ignore && from.charAt(i)!=s.charAt(i)){
          flag=1;
        }
      }
      if(flag==0 && !seen.contains(s)){
        res.add(s);
      }
    }
    return res;
  }
}