Red Black Tree
P
p
from __future__ import annotations
from collections.abc import Iterator
class RedBlackTree:
"""
A Red-Black tree, which is a self-balancing BST (binary search
tree).
This tree has similar performance to AVL trees, but the balancing is
less strict, so it will perform faster for writing/deleting nodes
and slower for reading in the average case, though, because they're
both balanced binary search trees, both will get the same asymptotic
performance.
To read more about them, https://en.wikipedia.org/wiki/Red-black_tree
Unless otherwise specified, all asymptotic runtimes are specified in
terms of the size of the tree.
"""
def __init__(
self,
label: int | None = None,
color: int = 0,
parent: RedBlackTree | None = None,
left: RedBlackTree | None = None,
right: RedBlackTree | None = None,
) -> None:
"""Initialize a new Red-Black Tree node with the given values:
label: The value associated with this node
color: 0 if black, 1 if red
parent: The parent to this node
left: This node's left child
right: This node's right child
"""
self.label = label
self.parent = parent
self.left = left
self.right = right
self.color = color
# Here are functions which are specific to red-black trees
def rotate_left(self) -> RedBlackTree:
"""Rotate the subtree rooted at this node to the left and
returns the new root to this subtree.
Performing one rotation can be done in O(1).
"""
parent = self.parent
right = self.right
if right is None:
return self
self.right = right.left
if self.right:
self.right.parent = self
self.parent = right
right.left = self
if parent is not None:
if parent.left == self:
parent.left = right
else:
parent.right = right
right.parent = parent
return right
def rotate_right(self) -> RedBlackTree:
"""Rotate the subtree rooted at this node to the right and
returns the new root to this subtree.
Performing one rotation can be done in O(1).
"""
if self.left is None:
return self
parent = self.parent
left = self.left
self.left = left.right
if self.left:
self.left.parent = self
self.parent = left
left.right = self
if parent is not None:
if parent.right is self:
parent.right = left
else:
parent.left = left
left.parent = parent
return left
def insert(self, label: int) -> RedBlackTree:
"""Inserts label into the subtree rooted at self, performs any
rotations necessary to maintain balance, and then returns the
new root to this subtree (likely self).
This is guaranteed to run in O(log(n)) time.
"""
if self.label is None:
# Only possible with an empty tree
self.label = label
return self
if self.label == label:
return self
elif self.label > label:
if self.left:
self.left.insert(label)
else:
self.left = RedBlackTree(label, 1, self)
self.left._insert_repair()
elif self.right:
self.right.insert(label)
else:
self.right = RedBlackTree(label, 1, self)
self.right._insert_repair()
return self.parent or self
def _insert_repair(self) -> None:
"""Repair the coloring from inserting into a tree."""
if self.parent is None:
# This node is the root, so it just needs to be black
self.color = 0
elif color(self.parent) == 0:
# If the parent is black, then it just needs to be red
self.color = 1
else:
uncle = self.parent.sibling
if color(uncle) == 0:
if self.is_left() and self.parent.is_right():
self.parent.rotate_right()
if self.right:
self.right._insert_repair()
elif self.is_right() and self.parent.is_left():
self.parent.rotate_left()
if self.left:
self.left._insert_repair()
elif self.is_left():
if self.grandparent:
self.grandparent.rotate_right()
self.parent.color = 0
if self.parent.right:
self.parent.right.color = 1
else:
if self.grandparent:
self.grandparent.rotate_left()
self.parent.color = 0
if self.parent.left:
self.parent.left.color = 1
else:
self.parent.color = 0
if uncle and self.grandparent:
uncle.color = 0
self.grandparent.color = 1
self.grandparent._insert_repair()
def remove(self, label: int) -> RedBlackTree:
"""Remove label from this tree."""
if self.label == label:
if self.left and self.right:
# It's easier to balance a node with at most one child,
# so we replace this node with the greatest one less than
# it and remove that.
value = self.left.get_max()
if value is not None:
self.label = value
self.left.remove(value)
else:
# This node has at most one non-None child, so we don't
# need to replace
child = self.left or self.right
if self.color == 1:
# This node is red, and its child is black
# The only way this happens to a node with one child
# is if both children are None leaves.
# We can just remove this node and call it a day.
if self.parent:
if self.is_left():
self.parent.left = None
else:
self.parent.right = None
# The node is black
elif child is None:
# This node and its child are black
if self.parent is None:
# The tree is now empty
return RedBlackTree(None)
else:
self._remove_repair()
if self.is_left():
self.parent.left = None
else:
self.parent.right = None
self.parent = None
else:
# This node is black and its child is red
# Move the child node here and make it black
self.label = child.label
self.left = child.left
self.right = child.right
if self.left:
self.left.parent = self
if self.right:
self.right.parent = self
elif self.label is not None and self.label > label:
if self.left:
self.left.remove(label)
elif self.right:
self.right.remove(label)
return self.parent or self
def _remove_repair(self) -> None:
"""Repair the coloring of the tree that may have been messed up."""
if (
self.parent is None
or self.sibling is None
or self.parent.sibling is None
or self.grandparent is None
):
return
if color(self.sibling) == 1:
self.sibling.color = 0
self.parent.color = 1
if self.is_left():
self.parent.rotate_left()
else:
self.parent.rotate_right()
if (
color(self.parent) == 0
and color(self.sibling) == 0
and color(self.sibling.left) == 0
and color(self.sibling.right) == 0
):
self.sibling.color = 1
self.parent._remove_repair()
return
if (
color(self.parent) == 1
and color(self.sibling) == 0
and color(self.sibling.left) == 0
and color(self.sibling.right) == 0
):
self.sibling.color = 1
self.parent.color = 0
return
if (
self.is_left()
and color(self.sibling) == 0
and color(self.sibling.right) == 0
and color(self.sibling.left) == 1
):
self.sibling.rotate_right()
self.sibling.color = 0
if self.sibling.right:
self.sibling.right.color = 1
if (
self.is_right()
and color(self.sibling) == 0
and color(self.sibling.right) == 1
and color(self.sibling.left) == 0
):
self.sibling.rotate_left()
self.sibling.color = 0
if self.sibling.left:
self.sibling.left.color = 1
if (
self.is_left()
and color(self.sibling) == 0
and color(self.sibling.right) == 1
):
self.parent.rotate_left()
self.grandparent.color = self.parent.color
self.parent.color = 0
self.parent.sibling.color = 0
if (
self.is_right()
and color(self.sibling) == 0
and color(self.sibling.left) == 1
):
self.parent.rotate_right()
self.grandparent.color = self.parent.color
self.parent.color = 0
self.parent.sibling.color = 0
def check_color_properties(self) -> bool:
"""Check the coloring of the tree, and return True iff the tree
is colored in a way which matches these five properties:
(wording stolen from wikipedia article)
1. Each node is either red or black.
2. The root node is black.
3. All leaves are black.
4. If a node is red, then both its children are black.
5. Every path from any node to all of its descendent NIL nodes
has the same number of black nodes.
This function runs in O(n) time, because properties 4 and 5 take
that long to check.
"""
# I assume property 1 to hold because there is nothing that can
# make the color be anything other than 0 or 1.
# Property 2
if self.color:
# The root was red
print("Property 2")
return False
# Property 3 does not need to be checked, because None is assumed
# to be black and is all the leaves.
# Property 4
if not self.check_coloring():
print("Property 4")
return False
# Property 5
if self.black_height() is None:
print("Property 5")
return False
# All properties were met
return True
def check_coloring(self) -> bool:
"""A helper function to recursively check Property 4 of a
Red-Black Tree. See check_color_properties for more info.
"""
if self.color == 1 and 1 in (color(self.left), color(self.right)):
return False
if self.left and not self.left.check_coloring():
return False
return not (self.right and not self.right.check_coloring())
def black_height(self) -> int | None:
"""Returns the number of black nodes from this node to the
leaves of the tree, or None if there isn't one such value (the
tree is color incorrectly).
"""
if self is None or self.left is None or self.right is None:
# If we're already at a leaf, there is no path
return 1
left = RedBlackTree.black_height(self.left)
right = RedBlackTree.black_height(self.right)
if left is None or right is None:
# There are issues with coloring below children nodes
return None
if left != right:
# The two children have unequal depths
return None
# Return the black depth of children, plus one if this node is
# black
return left + (1 - self.color)
# Here are functions which are general to all binary search trees
def __contains__(self, label: int) -> bool:
"""Search through the tree for label, returning True iff it is
found somewhere in the tree.
Guaranteed to run in O(log(n)) time.
"""
return self.search(label) is not None
def search(self, label: int) -> RedBlackTree | None:
"""Search through the tree for label, returning its node if
it's found, and None otherwise.
This method is guaranteed to run in O(log(n)) time.
"""
if self.label == label:
return self
elif self.label is not None and label > self.label:
if self.right is None:
return None
else:
return self.right.search(label)
elif self.left is None:
return None
else:
return self.left.search(label)
def floor(self, label: int) -> int | None:
"""Returns the largest element in this tree which is at most label.
This method is guaranteed to run in O(log(n)) time."""
if self.label == label:
return self.label
elif self.label is not None and self.label > label:
if self.left:
return self.left.floor(label)
else:
return None
else:
if self.right:
attempt = self.right.floor(label)
if attempt is not None:
return attempt
return self.label
def ceil(self, label: int) -> int | None:
"""Returns the smallest element in this tree which is at least label.
This method is guaranteed to run in O(log(n)) time.
"""
if self.label == label:
return self.label
elif self.label is not None and self.label < label:
if self.right:
return self.right.ceil(label)
else:
return None
else:
if self.left:
attempt = self.left.ceil(label)
if attempt is not None:
return attempt
return self.label
def get_max(self) -> int | None:
"""Returns the largest element in this tree.
This method is guaranteed to run in O(log(n)) time.
"""
if self.right:
# Go as far right as possible
return self.right.get_max()
else:
return self.label
def get_min(self) -> int | None:
"""Returns the smallest element in this tree.
This method is guaranteed to run in O(log(n)) time.
"""
if self.left:
# Go as far left as possible
return self.left.get_min()
else:
return self.label
@property
def grandparent(self) -> RedBlackTree | None:
"""Get the current node's grandparent, or None if it doesn't exist."""
if self.parent is None:
return None
else:
return self.parent.parent
@property
def sibling(self) -> RedBlackTree | None:
"""Get the current node's sibling, or None if it doesn't exist."""
if self.parent is None:
return None
elif self.parent.left is self:
return self.parent.right
else:
return self.parent.left
def is_left(self) -> bool:
"""Returns true iff this node is the left child of its parent."""
if self.parent is None:
return False
return self.parent.left is self
def is_right(self) -> bool:
"""Returns true iff this node is the right child of its parent."""
if self.parent is None:
return False
return self.parent.right is self
def __bool__(self) -> bool:
return True
def __len__(self) -> int:
"""
Return the number of nodes in this tree.
"""
ln = 1
if self.left:
ln += len(self.left)
if self.right:
ln += len(self.right)
return ln
def preorder_traverse(self) -> Iterator[int | None]:
yield self.label
if self.left:
yield from self.left.preorder_traverse()
if self.right:
yield from self.right.preorder_traverse()
def inorder_traverse(self) -> Iterator[int | None]:
if self.left:
yield from self.left.inorder_traverse()
yield self.label
if self.right:
yield from self.right.inorder_traverse()
def postorder_traverse(self) -> Iterator[int | None]:
if self.left:
yield from self.left.postorder_traverse()
if self.right:
yield from self.right.postorder_traverse()
yield self.label
def __repr__(self) -> str:
from pprint import pformat
if self.left is None and self.right is None:
return f"'{self.label} {(self.color and 'red') or 'blk'}'"
return pformat(
{
f"{self.label} {(self.color and 'red') or 'blk'}": (
self.left,
self.right,
)
},
indent=1,
)
def __eq__(self, other: object) -> bool:
"""Test if two trees are equal."""
if not isinstance(other, RedBlackTree):
return NotImplemented
if self.label == other.label:
return self.left == other.left and self.right == other.right
else:
return False
def color(node: RedBlackTree | None) -> int:
"""Returns the color of a node, allowing for None leaves."""
if node is None:
return 0
else:
return node.color
"""
Code for testing the various
functions of the red-black tree.
"""
def test_rotations() -> bool:
"""Test that the rotate_left and rotate_right functions work."""
# Make a tree to test on
tree = RedBlackTree(0)
tree.left = RedBlackTree(-10, parent=tree)
tree.right = RedBlackTree(10, parent=tree)
tree.left.left = RedBlackTree(-20, parent=tree.left)
tree.left.right = RedBlackTree(-5, parent=tree.left)
tree.right.left = RedBlackTree(5, parent=tree.right)
tree.right.right = RedBlackTree(20, parent=tree.right)
# Make the right rotation
left_rot = RedBlackTree(10)
left_rot.left = RedBlackTree(0, parent=left_rot)
left_rot.left.left = RedBlackTree(-10, parent=left_rot.left)
left_rot.left.right = RedBlackTree(5, parent=left_rot.left)
left_rot.left.left.left = RedBlackTree(-20, parent=left_rot.left.left)
left_rot.left.left.right = RedBlackTree(-5, parent=left_rot.left.left)
left_rot.right = RedBlackTree(20, parent=left_rot)
tree = tree.rotate_left()
if tree != left_rot:
return False
tree = tree.rotate_right()
tree = tree.rotate_right()
# Make the left rotation
right_rot = RedBlackTree(-10)
right_rot.left = RedBlackTree(-20, parent=right_rot)
right_rot.right = RedBlackTree(0, parent=right_rot)
right_rot.right.left = RedBlackTree(-5, parent=right_rot.right)
right_rot.right.right = RedBlackTree(10, parent=right_rot.right)
right_rot.right.right.left = RedBlackTree(5, parent=right_rot.right.right)
right_rot.right.right.right = RedBlackTree(20, parent=right_rot.right.right)
return tree == right_rot
def test_insertion_speed() -> bool:
"""Test that the tree balances inserts to O(log(n)) by doing a lot
of them.
"""
tree = RedBlackTree(-1)
for i in range(300000):
tree = tree.insert(i)
return True
def test_insert() -> bool:
"""Test the insert() method of the tree correctly balances, colors,
and inserts.
"""
tree = RedBlackTree(0)
tree.insert(8)
tree.insert(-8)
tree.insert(4)
tree.insert(12)
tree.insert(10)
tree.insert(11)
ans = RedBlackTree(0, 0)
ans.left = RedBlackTree(-8, 0, ans)
ans.right = RedBlackTree(8, 1, ans)
ans.right.left = RedBlackTree(4, 0, ans.right)
ans.right.right = RedBlackTree(11, 0, ans.right)
ans.right.right.left = RedBlackTree(10, 1, ans.right.right)
ans.right.right.right = RedBlackTree(12, 1, ans.right.right)
return tree == ans
def test_insert_and_search() -> bool:
"""Tests searching through the tree for values."""
tree = RedBlackTree(0)
tree.insert(8)
tree.insert(-8)
tree.insert(4)
tree.insert(12)
tree.insert(10)
tree.insert(11)
if any(i in tree for i in (5, -6, -10, 13)):
# Found something not in there
return False
# Find all these things in there
return all(i in tree for i in (11, 12, -8, 0))
def test_insert_delete() -> bool:
"""Test the insert() and delete() method of the tree, verifying the
insertion and removal of elements, and the balancing of the tree.
"""
tree = RedBlackTree(0)
tree = tree.insert(-12)
tree = tree.insert(8)
tree = tree.insert(-8)
tree = tree.insert(15)
tree = tree.insert(4)
tree = tree.insert(12)
tree = tree.insert(10)
tree = tree.insert(9)
tree = tree.insert(11)
tree = tree.remove(15)
tree = tree.remove(-12)
tree = tree.remove(9)
if not tree.check_color_properties():
return False
return list(tree.inorder_traverse()) == [-8, 0, 4, 8, 10, 11, 12]
def test_floor_ceil() -> bool:
"""Tests the floor and ceiling functions in the tree."""
tree = RedBlackTree(0)
tree.insert(-16)
tree.insert(16)
tree.insert(8)
tree.insert(24)
tree.insert(20)
tree.insert(22)
tuples = [(-20, None, -16), (-10, -16, 0), (8, 8, 8), (50, 24, None)]
for val, floor, ceil in tuples:
if tree.floor(val) != floor or tree.ceil(val) != ceil:
return False
return True
def test_min_max() -> bool:
"""Tests the min and max functions in the tree."""
tree = RedBlackTree(0)
tree.insert(-16)
tree.insert(16)
tree.insert(8)
tree.insert(24)
tree.insert(20)
tree.insert(22)
return not (tree.get_max() != 22 or tree.get_min() != -16)
def test_tree_traversal() -> bool:
"""Tests the three different tree traversal functions."""
tree = RedBlackTree(0)
tree = tree.insert(-16)
tree.insert(16)
tree.insert(8)
tree.insert(24)
tree.insert(20)
tree.insert(22)
if list(tree.inorder_traverse()) != [-16, 0, 8, 16, 20, 22, 24]:
return False
if list(tree.preorder_traverse()) != [0, -16, 16, 8, 22, 20, 24]:
return False
return list(tree.postorder_traverse()) == [-16, 8, 20, 24, 22, 16, 0]
def test_tree_chaining() -> bool:
"""Tests the three different tree chaining functions."""
tree = RedBlackTree(0)
tree = tree.insert(-16).insert(16).insert(8).insert(24).insert(20).insert(22)
if list(tree.inorder_traverse()) != [-16, 0, 8, 16, 20, 22, 24]:
return False
if list(tree.preorder_traverse()) != [0, -16, 16, 8, 22, 20, 24]:
return False
return list(tree.postorder_traverse()) == [-16, 8, 20, 24, 22, 16, 0]
def print_results(msg: str, passes: bool) -> None:
print(str(msg), "works!" if passes else "doesn't work :(")
def pytests() -> None:
assert test_rotations()
assert test_insert()
assert test_insert_and_search()
assert test_insert_delete()
assert test_floor_ceil()
assert test_tree_traversal()
assert test_tree_chaining()
def main() -> None:
"""
>>> pytests()
"""
print_results("Rotating right and left", test_rotations())
print_results("Inserting", test_insert())
print_results("Searching", test_insert_and_search())
print_results("Deleting", test_insert_delete())
print_results("Floor and ceil", test_floor_ceil())
print_results("Tree traversal", test_tree_traversal())
print_results("Tree traversal", test_tree_chaining())
print("Testing tree balancing...")
print("This should only be a few seconds.")
test_insertion_speed()
print("Done!")
if __name__ == "__main__":
main()
