边工作边刷题:70天一遍leetcode: day 25

Course Schedule I/II

要点:

  • 有向图的topological sort,indegree方法需要图结构吗?需要,因为每层遍历是根据图得到连接关系更新indegree的
  • 图的dfs方法,虽然可以在recursion call之前或者之后更新visited,但是最好统一,这样不会出错。对这题,
    • visited在call外判断,call内更新,
    • recStk在call内开始进,call内结束出
    • 注意res在call内结束入(因为深层的排前面),不pop

错误点:

  • 一个非常tricky的错误是对邻接点先检查visited,然后检查是否在recStk。实际上如果有环,肯定另一端已经visited了,这样不会返回false,正确的方法应该先检查是否在recStk里
class Solution(object):
    def findOrder(self, numCourses, prerequisites):
        """
        :type numCourses: int
        :type prerequisites: List[List[int]]
        :rtype: List[int]
        """
        def dfs(graph, i, recStk, visited, res):
            visited[i]=True
            recStk.add(i)
            for p in graph[i]:
                if p in recStk: 
                    return False
                elif not visited[p]:
                    if not dfs(graph, p, recStk, visited, res):
                        return False
            
            recStk.remove(i)
            res.append(i)
            return True
        
        graph = [[] for i in range(numCourses)]
        for e in prerequisites:
            graph[e[0]].append(e[1])
        
        res = []
        recStk = set()
        visited = [False]*numCourses
        for p in range(numCourses):
            if not visited[p]:
                if not dfs(graph, p, recStk, visited, res):
                    return []
        
        # res.reverse()
        return res
            
                    
posted @ 2016-04-23 10:36  absolute100  阅读(152)  评论(0编辑  收藏  举报