[Leetcode] 98. Validate Binary Search Tree 验证二叉搜索树

方法1: Traverse

corner case: what if there are equal numbers;
in-order traverse. we need to keep a lastNode, which compares to current node, if lastNode != None and lastNode.val >= currentNode.val, return False;
time complexity is O(n), space complexity O(n) since we keep up to the entire tree;
when we use traverse, there is no return value in the dfs method, so we need to keep a global variable by either self.variable or keep in the parameter. here we use global variable;
set initial value as True, change to False and return when it visit a invalid node. if it is False at beginning, it’s hard to determine when to set it to True;

class Solution:
    def isValidBST(self, root):
        self.isValid = True      # set initial value as True, change to False when it visit a invalid node
        self.lastNode = None
        self.dfs(root)
        return self.isValid
    
    def dfs(self, root):
        if root is None:
            return
        self.dfs(root.left)
        if self.lastNode is not None and self.lastNode.val >= root.val:
            self.isValid = False
            return
        self.lastNode = root
        self.dfs(root.right)

方法2：divide-and-conquer

keep in mind that we need to compare the root with maxleft and minright, not root.left and root.right

class Solution:
    def isValidBST(self, root):
        _, _, isValid = self.dfs(root)
        return isValid

    def dfs(self, root):
        if root is None:
            return None, None, True
        
        leftmin, leftmax, validLeft = self.dfs(root.left)
        rightmin, rightmax, validRight = self.dfs(root.right)
        if not validLeft or not validRight:
            return None, None, False
        if leftmax and leftmax.val >= root.val:
            return None, None, False
        if rightmin and rightmin.val <= root.val:
            return None, None, False
        minNode = leftmin if leftmin else root    # in case no left or no right
        maxNode = rightmax if rightmax else root
        return minNode, maxNode, True