B+ Tree in Python
Last Updated :
18 Apr, 2024
In computer science, data structures are crucial in efficiently managing and organizing data. Among these, the B+ tree is a powerful and important data structure, widely used in databases and file systems. In this article, we will discuss the concept of B+ trees, exploring their structure, operations, and implementation in Python.
B+ Tree in Python
A B+ tree is a self-balancing tree data structure designed for efficient storage and retrieval of data in secondary memory such as disk storage. It is a variant of the B-tree, characterized by its ability to store multiple keys in each node, with only the leaf nodes containing actual data pointers. The internal nodes act as index nodes, facilitating fast searching and traversal.
Key Characteristics of B+ Trees:
- Ordered Structure: Keys in a B+ tree are stored in sorted order, enabling efficient searching through binary search.
- Balanced Tree: B+ trees are self-balancing, ensuring that operations such as insertion and deletion maintain a balanced tree structure, which results in optimal performance.
- Leaf Node Linked List: All leaf nodes are linked together, forming a linked list. This feature facilitates a range of queries and sequential access.
Operations on B+ Trees:
- Search: Searching in a B+ tree involves traversing the tree from the root node to the leaf node, and performing a binary search to locate the desired key.
- Insertion: Inserting a new key-value pair into a B+ tree begins with a search operation to find the appropriate leaf node. If the leaf node has space, the key-value pair is inserted. Otherwise, the node is split, and the insertion is propagated upwards.
- Deletion: Deleting a key-value pair from a B+ tree follows a similar process to insertion. After locating the leaf node containing the key to be deleted, the key is removed. If the node becomes underflowed, it may borrow keys from sibling nodes or merge with them to maintain balance.
Searching in B+ Tree:
Searching in a B-tree involves traversing the tree from the root to find the node that might contain the desired key.
Step-by-step algorithm:
- Start at the root of the B+ tree.
- Check if the current node is a leaf node.
- If not a leaf node, compare the search key with the keys in the node to determine the appropriate child node to traverse.
- If a leaf node, search for the key within the keys stored in the leaf node.
- If the key is found, return the corresponding value.
- If the key is not found in any leaf node, return a failure indication (e.g., null or false).
- Repeat steps 3-6 until the key is found or until there are no more child nodes to traverse.
Below is the implementation of the above idea:
Python3
import math
# Node creation
class Node:
def __init__(self, order):
self.order = order
self.values = []
self.keys = []
self.nextKey = None
self.parent = None
self.check_leaf = False
# Search operation
def search(self, value):
current_node = self
while(not current_node.check_leaf):
temp2 = current_node.values
for i in range(len(temp2)):
if (value == temp2[i]):
current_node = current_node.keys[i + 1]
break
elif (value < temp2[i]):
current_node = current_node.keys[i]
break
elif (i + 1 == len(current_node.values)):
current_node = current_node.keys[i + 1]
break
return current_node
# B plus tree
class BplusTree:
def __init__(self, order):
self.root = Node(order)
self.root.check_leaf = True
# Search operation
def search(self, value):
current_node = self.root
while(not current_node.check_leaf):
temp2 = current_node.values
for i in range(len(temp2)):
if (value == temp2[i]):
current_node = current_node.keys[i + 1]
break
elif (value < temp2[i]):
current_node = current_node.keys[i]
break
elif (i + 1 == len(current_node.values)):
current_node = current_node.keys[i + 1]
break
return current_node
# Print the tree
def printTree(tree):
lst = [tree.root]
level = [0]
leaf = None
flag = 0
lev_leaf = 0
node1 = Node(str(level[0]) + str(tree.root.values))
while (len(lst) != 0):
x = lst.pop(0)
lev = level.pop(0)
if (x.check_leaf == False):
for i, item in enumerate(x.keys):
print(item.values)
else:
for i, item in enumerate(x.keys):
print(item.values)
if (flag == 0):
lev_leaf = lev
leaf = x
flag = 1
# Main code
record_len = 3
bplustree = BplusTree(record_len)
bplustree.search('5')
printTree(bplustree)
if(bplustree.search('5')):
print("Found")
else:
print("Not found")
Time Complexity: O(log n)
Auxiliary Space: O(1)
Insertion in B+ Tree:
Insertion in a B+ tree involves placing a new key-value pair into the appropriate leaf node and ensuring tree balance through splits and updates.
Step-by-step algorithm:
- Insertion:
- Search for the leaf node to insert the key-value pair.
- If the leaf node is full, split it.
- Propagate median key to parent if necessary.
- Split parent if full.
- Repeat until balanced.
- Search:
- Traverse down the tree to find leaf node.
- Search for key in leaf node.
- Return value if found, else indicate failure.
- Print Tree:
- Traverse tree to print its structure.
- Example Operations:
- Insert key-value pairs.
- Search for specific keys.
- Print the tree structure.
Below is the implementation of the above idea:
Python3
import math
# Node creation
class Node:
def __init__(self, order):
self.order = order
self.values = []
self.keys = []
self.nextKey = None
self.parent = None
self.check_leaf = False
# Insert at the leaf
def insert_at_leaf(self, leaf, value, key):
if (self.values):
temp1 = self.values
for i in range(len(temp1)):
if (value == temp1[i]):
self.keys[i].append(key)
break
elif (value < temp1[i]):
self.values = self.values[:i] + [value] + self.values[i:]
self.keys = self.keys[:i] + [[key]] + self.keys[i:]
break
elif (i + 1 == len(temp1)):
self.values.append(value)
self.keys.append([key])
break
else:
self.values = [value]
self.keys = [[key]]
# B plus tree
class BplusTree:
def __init__(self, order):
self.root = Node(order)
self.root.check_leaf = True
# Insert operation
def insert(self, value, key):
value = str(value)
old_node = self.search(value)
old_node.insert_at_leaf(old_node, value, key)
if (len(old_node.values) == old_node.order):
node1 = Node(old_node.order)
node1.check_leaf = True
node1.parent = old_node.parent
mid = int(math.ceil(old_node.order / 2)) - 1
node1.values = old_node.values[mid + 1:]
node1.keys = old_node.keys[mid + 1:]
node1.nextKey = old_node.nextKey
old_node.values = old_node.values[:mid + 1]
old_node.keys = old_node.keys[:mid + 1]
old_node.nextKey = node1
self.insert_in_parent(old_node, node1.values[0], node1)
# Search operation for different operations
def search(self, value):
current_node = self.root
while(current_node.check_leaf == False):
temp2 = current_node.values
for i in range(len(temp2)):
if (value == temp2[i]):
current_node = current_node.keys[i + 1]
break
elif (value < temp2[i]):
current_node = current_node.keys[i]
break
elif (i + 1 == len(current_node.values)):
current_node = current_node.keys[i + 1]
break
return current_node
# Find the node
def find(self, value, key):
l = self.search(value)
for i, item in enumerate(l.values):
if item == value:
if key in l.keys[i]:
return True
else:
return False
return False
# Inserting at the parent
def insert_in_parent(self, n, value, ndash):
if (self.root == n):
rootNode = Node(n.order)
rootNode.values = [value]
rootNode.keys = [n, ndash]
self.root = rootNode
n.parent = rootNode
ndash.parent = rootNode
return
parentNode = n.parent
temp3 = parentNode.keys
for i in range(len(temp3)):
if (temp3[i] == n):
parentNode.values = parentNode.values[:i] + \
[value] + parentNode.values[i:]
parentNode.keys = parentNode.keys[:i +
1] + [ndash] + parentNode.keys[i + 1:]
if (len(parentNode.keys) > parentNode.order):
parentdash = Node(parentNode.order)
parentdash.parent = parentNode.parent
mid = int(math.ceil(parentNode.order / 2)) - 1
parentdash.values = parentNode.values[mid + 1:]
parentdash.keys = parentNode.keys[mid + 1:]
value_ = parentNode.values[mid]
if (mid == 0):
parentNode.values = parentNode.values[:mid + 1]
else:
parentNode.values = parentNode.values[:mid]
parentNode.keys = parentNode.keys[:mid + 1]
for j in parentNode.keys:
j.parent = parentNode
for j in parentdash.keys:
j.parent = parentdash
self.insert_in_parent(parentNode, value_, parentdash)
# Print the tree
def printTree(tree):
lst = [tree.root]
level = [0]
leaf = None
flag = 0
lev_leaf = 0
node1 = Node(str(level[0]) + str(tree.root.values))
while (len(lst) != 0):
x = lst.pop(0)
lev = level.pop(0)
if (x.check_leaf == False):
for i, item in enumerate(x.keys):
print(item.values)
else:
for i, item in enumerate(x.keys):
print(item.values)
if (flag == 0):
lev_leaf = lev
leaf = x
flag = 1
record_len = 3
bplustree = BplusTree(record_len)
bplustree.insert('5', '33')
bplustree.insert('15', '21')
bplustree.insert('25', '31')
bplustree.insert('35', '41')
bplustree.insert('45', '10')
printTree(bplustree)
if(bplustree.find('5', '34')):
print("Found")
else:
print("Not found")
Output['15', '25']
['35', '45']
['5']
Not found
Time Complexity: O(log n)
Auxiliary Space: O(1)
Deletion in B+ Tree:
Deletion in a B+ tree involves removing a key-value pair, adjusting the structure of the tree, and potentially merging or redistributing nodes to maintain B+ tree properties.
Step-by-step algorithm:
- Define Node with order, values, keys, nextKey, parent, and check_leaf.
- Implement insert_at_leaf in Node to insert values and keys.
- Define BplusTree with root as leaf.
- Implement insert to add key-value, splitting leaf nodes if needed.
- Implement search to find leaf node with given value.
- Implement find to locate specific key-value pair.
- Implement insert_in_parent to handle parent node splitting.
- Implement delete to remove key-value, adjusting structure.
- Implement deleteEntry for deletion at various levels.
- Define printTree to visualize tree.
- Initialize B+ tree, insert key-value pairs, print tree structure, and search.
Below is the implementation of the above idea:
Python3
# B+ tee in python
import math
# Node creation
class Node:
def __init__(self, order):
self.order = order
self.values = []
self.keys = []
self.nextKey = None
self.parent = None
self.check_leaf = False
# Insert at the leaf
def insert_at_leaf(self, leaf, value, key):
if (self.values):
temp1 = self.values
for i in range(len(temp1)):
if (value == temp1[i]):
self.keys[i].append(key)
break
elif (value < temp1[i]):
self.values = self.values[:i] + [value] + self.values[i:]
self.keys = self.keys[:i] + [[key]] + self.keys[i:]
break
elif (i + 1 == len(temp1)):
self.values.append(value)
self.keys.append([key])
break
else:
self.values = [value]
self.keys = [[key]]
# B plus tree
class BplusTree:
def __init__(self, order):
self.root = Node(order)
self.root.check_leaf = True
# Insert operation
def insert(self, value, key):
value = str(value)
old_node = self.search(value)
old_node.insert_at_leaf(old_node, value, key)
if (len(old_node.values) == old_node.order):
node1 = Node(old_node.order)
node1.check_leaf = True
node1.parent = old_node.parent
mid = int(math.ceil(old_node.order / 2)) - 1
node1.values = old_node.values[mid + 1:]
node1.keys = old_node.keys[mid + 1:]
node1.nextKey = old_node.nextKey
old_node.values = old_node.values[:mid + 1]
old_node.keys = old_node.keys[:mid + 1]
old_node.nextKey = node1
self.insert_in_parent(old_node, node1.values[0], node1)
# Search operation for different operations
def search(self, value):
current_node = self.root
while(current_node.check_leaf == False):
temp2 = current_node.values
for i in range(len(temp2)):
if (value == temp2[i]):
current_node = current_node.keys[i + 1]
break
elif (value < temp2[i]):
current_node = current_node.keys[i]
break
elif (i + 1 == len(current_node.values)):
current_node = current_node.keys[i + 1]
break
return current_node
# Find the node
def find(self, value, key):
l = self.search(value)
for i, item in enumerate(l.values):
if item == value:
if key in l.keys[i]:
return True
else:
return False
return False
# Inserting at the parent
def insert_in_parent(self, n, value, ndash):
if (self.root == n):
rootNode = Node(n.order)
rootNode.values = [value]
rootNode.keys = [n, ndash]
self.root = rootNode
n.parent = rootNode
ndash.parent = rootNode
return
parentNode = n.parent
temp3 = parentNode.keys
for i in range(len(temp3)):
if (temp3[i] == n):
parentNode.values = parentNode.values[:i] + \
[value] + parentNode.values[i:]
parentNode.keys = parentNode.keys[:i +
1] + [ndash] + parentNode.keys[i + 1:]
if (len(parentNode.keys) > parentNode.order):
parentdash = Node(parentNode.order)
parentdash.parent = parentNode.parent
mid = int(math.ceil(parentNode.order / 2)) - 1
parentdash.values = parentNode.values[mid + 1:]
parentdash.keys = parentNode.keys[mid + 1:]
value_ = parentNode.values[mid]
if (mid == 0):
parentNode.values = parentNode.values[:mid + 1]
else:
parentNode.values = parentNode.values[:mid]
parentNode.keys = parentNode.keys[:mid + 1]
for j in parentNode.keys:
j.parent = parentNode
for j in parentdash.keys:
j.parent = parentdash
self.insert_in_parent(parentNode, value_, parentdash)
# Delete a node
def delete(self, value, key):
node_ = self.search(value)
temp = 0
for i, item in enumerate(node_.values):
if item == value:
temp = 1
if key in node_.keys[i]:
if len(node_.keys[i]) > 1:
node_.keys[i].pop(node_.keys[i].index(key))
elif node_ == self.root:
node_.values.pop(i)
node_.keys.pop(i)
else:
node_.keys[i].pop(node_.keys[i].index(key))
del node_.keys[i]
node_.values.pop(node_.values.index(value))
self.deleteEntry(node_, value, key)
else:
print("Value not in Key")
return
if temp == 0:
print("Value not in Tree")
return
# Delete an entry
def deleteEntry(self, node_, value, key):
if not node_.check_leaf:
for i, item in enumerate(node_.keys):
if item == key:
node_.keys.pop(i)
break
for i, item in enumerate(node_.values):
if item == value:
node_.values.pop(i)
break
if self.root == node_ and len(node_.keys) == 1:
self.root = node_.keys[0]
node_.keys[0].parent = None
del node_
return
elif (len(node_.keys) < int(math.ceil(node_.order / 2)) and node_.check_leaf == False) or (len(node_.values) < int(math.ceil((node_.order - 1) / 2)) and node_.check_leaf == True):
is_predecessor = 0
parentNode = node_.parent
PrevNode = -1
NextNode = -1
PrevK = -1
PostK = -1
for i, item in enumerate(parentNode.keys):
if item == node_:
if i > 0:
PrevNode = parentNode.keys[i - 1]
PrevK = parentNode.values[i - 1]
if i < len(parentNode.keys) - 1:
NextNode = parentNode.keys[i + 1]
PostK = parentNode.values[i]
if PrevNode == -1:
ndash = NextNode
value_ = PostK
elif NextNode == -1:
is_predecessor = 1
ndash = PrevNode
value_ = PrevK
else:
if len(node_.values) + len(NextNode.values) < node_.order:
ndash = NextNode
value_ = PostK
else:
is_predecessor = 1
ndash = PrevNode
value_ = PrevK
if len(node_.values) + len(ndash.values) < node_.order:
if is_predecessor == 0:
node_, ndash = ndash, node_
ndash.keys += node_.keys
if not node_.check_leaf:
ndash.values.append(value_)
else:
ndash.nextKey = node_.nextKey
ndash.values += node_.values
if not ndash.check_leaf:
for j in ndash.keys:
j.parent = ndash
self.deleteEntry(node_.parent, value_, node_)
del node_
else:
if is_predecessor == 1:
if not node_.check_leaf:
ndashpm = ndash.keys.pop(-1)
ndashkm_1 = ndash.values.pop(-1)
node_.keys = [ndashpm] + node_.keys
node_.values = [value_] + node_.values
parentNode = node_.parent
for i, item in enumerate(parentNode.values):
if item == value_:
p.values[i] = ndashkm_1
break
else:
ndashpm = ndash.keys.pop(-1)
ndashkm = ndash.values.pop(-1)
node_.keys = [ndashpm] + node_.keys
node_.values = [ndashkm] + node_.values
parentNode = node_.parent
for i, item in enumerate(p.values):
if item == value_:
parentNode.values[i] = ndashkm
break
else:
if not node_.check_leaf:
ndashp0 = ndash.keys.pop(0)
ndashk0 = ndash.values.pop(0)
node_.keys = node_.keys + [ndashp0]
node_.values = node_.values + [value_]
parentNode = node_.parent
for i, item in enumerate(parentNode.values):
if item == value_:
parentNode.values[i] = ndashk0
break
else:
ndashp0 = ndash.keys.pop(0)
ndashk0 = ndash.values.pop(0)
node_.keys = node_.keys + [ndashp0]
node_.values = node_.values + [ndashk0]
parentNode = node_.parent
for i, item in enumerate(parentNode.values):
if item == value_:
parentNode.values[i] = ndash.values[0]
break
if not ndash.check_leaf:
for j in ndash.keys:
j.parent = ndash
if not node_.check_leaf:
for j in node_.keys:
j.parent = node_
if not parentNode.check_leaf:
for j in parentNode.keys:
j.parent = parentNode
# Print the tree
def printTree(tree):
lst = [tree.root]
level = [0]
leaf = None
flag = 0
lev_leaf = 0
node1 = Node(str(level[0]) + str(tree.root.values))
while (len(lst) != 0):
x = lst.pop(0)
lev = level.pop(0)
if (x.check_leaf == False):
for i, item in enumerate(x.keys):
print(item.values)
else:
for i, item in enumerate(x.keys):
print(item.values)
if (flag == 0):
lev_leaf = lev
leaf = x
flag = 1
record_len = 3
bplustree = BplusTree(record_len)
bplustree.insert('5', '33')
bplustree.insert('15', '21')
bplustree.insert('25', '31')
bplustree.insert('35', '41')
bplustree.insert('45', '10')
printTree(bplustree)
if(bplustree.find('5', '34')):
print("Found")
else:
print("Not found")
Output['15', '25']
['35', '45']
['5']
Not found
Time Complexity: O(log n)
Auxiliary Space: O(1)
Share your thoughts in the comments
Please Login to comment...