Data structures

Complexity
Reference: https://www.bigocheatsheet.com/

Array

Usage: Using same type of data efficiently
Pro: Fast access (index accessing)
Con: add/del hard (must pre-define max length)

Queue

Usage: Multi-tasking process scheduling (OS)
- Key feature: FIFO (first-in, first out)
Pro: -
Con: -
Enqueue @ TAIL
Dequeue @ HEAD

q = []
def enqueue(data):
  q.append(data)

def dequeue(data):
  return q.pop(0)

# priority q
import queue
q = queue.PriorityQueue()
q.put((10, "asd"))
q.put((1, "231"))
q.put((5, "12sdf"))
print(q.qsize()) # 3
print(q.get()) # (5, "12sdf")
print(q.get()) # (10, "asd")

Stack

Usage: Computer internal process’ function call
- Key feature: FIFO (first-in, first out)
Pro: Simple structure -> easy to implement, fast read/write
Con: maximum # of data must be set. If max # is larger than usage -> waste of memory.
Push @ HEAD
Pop @ TAIL

stck = []

def push(data):
    stck.append(data)

def pop():
    return stck.pop()

Singly-Linked List

Pro: Don’t need to pre-allocate memory: Can connect data far away from src
Con:
1. Need extra space for connecting: memory inefficient.
2. Slow finding connecting info.
3. Need extra connecting step when add/delete data in middle.

class Node:
    def __init__(self, data):
        self.data = data
        self.next = None

class NodeMgmt:
    def __init__(self, data):
        self.head = Node(data)
    
    def add(self, data):
        if self.head == '':
            self.head = Node(data)
        else:
            node = self.head
            while node.next:
                node = node.next
            node.next = Node(data)

    def desc(self):
        node = self.head
        while node:
            print (node.data)
            node = node.next

    def delete(self, data):
        if self.head == '':
            print ('해당 값을 가진 노드가 없습니다.')
            return
        if self.head.data == data: # 경우의 수1: self.head를 삭제해야할 경우 - self.head를 바꿔줘야 함
            temp = self.head # self.head 객체를 삭제하기 위해, 임시로 temp에 담아서 객체를 삭제했음
            self.head = self.head.next # 만약 self.head 객체를 삭제하면, 이 코드가 실행이 안되기 때문!
            del temp
        else:
            node = self.head
            while node.next: # 경우의 수2: self.head가 아닌 노드를 삭제해야할 경우
                if node.next.data == data:
                    temp = node.next
                    node.next = node.next.next       
                    del temp                         
                    pass                             
                else:
                    node = node.next
                    
    def search_node(self, data):
        node = self.head
        while node:
            if node.data == data:
                return node
            else:
                node = node.next

# 테스트
node_mgmt = NodeMgmt(0)
for data in range(1, 10):
    node_mgmt.add(data)

node = node_mgmt.search_node(4)
print (node.data)

Doubly-Linked List

Pro: Better accessability (both way)

class Node:
    def __init__(self, data, prev=None, next=None):
        self.prev = prev
        self.data = data
        self.next = next

class NodeMgmt:
    def __init__(self, data):
        self.head = Node(data)
        self.tail = self.head
    
    def insert_before(self, data, before_data):
        if self.head == None:
            self.head = Node(data)
            return True            
        else:
            node = self.tail
            while node.data != before_data:
                node = node.prev
                if node == None:
                    return False
            new = Node(data)
            before_new = node.prev
            before_new.next = new
            new.next = node
            return True

    def insert_after(self, data, after_data):
        if self.head == None:
            self.head = Node(data)
            return True            
        else:
            node = self.head
            while node.data != after_data:
                node = node.next
                if node == None:
                    return False
            new = Node(data)
            after_new = node.next
            new.next = after_new
            new.prev = node
            node.next = new
            if new.next == None:
                self.tail = new
            return True

    def insert(self, data):
        if self.head == None:
            self.head = Node(data)
        else:
            node = self.head
            while node.next:
                node = node.next
            new = Node(data)
            node.next = new
            new.prev = node
            self.tail = new

    def desc(self):
        node = self.head
        while node:
            print (node.data)
            node = node.next

node_mgmt = NodeMgmt(0)
for data in range(1, 10):
    node_mgmt.insert(data)

node_mgmt.desc()

node_mgmt.insert_after(1.5, 1)
node_mgmt.desc()

Hash Table

Python: dict()
Usage:
1. When need lot of searching.
2. When need lot of write/del/read.
3. When need cache (Easy to check duplicate).
Pro:
1. O(1) in data read/write
2. Easy to check duplicate
Con:
1. Need lot of memory
2. Need extra data structure to avoid same address collision for multiple keys. (e.g. Chaining, Linear Probling)
  - Chaining
  - Open hashing method: Use space outside of hash table
  - Use linked-list to connect data
- Linear Probling
  - Close hashing method: Solve collision inside of hash table
  - If collide in certain address, search next available space and put it in.

# Chaining
hash_table = list([0 for i in range(8)])

def get_key(data):
    return hash(data)

def hash_function(key):
    return key % 8

def save_data(data, value):
    index_key = get_key(data)
    hash_address = hash_function(index_key)
    if hash_table[hash_address] != 0:
        for index in range(len(hash_table[hash_address])):
            if hash_table[hash_address][index][0] == index_key:
                hash_table[hash_address][index][1] = value
                return
        hash_table[hash_address].append([index_key, value])
    else:
        hash_table[hash_address] = [[index_key, value]]
    
def read_data(data):
    index_key = get_key(data)
    hash_address = hash_function(index_key)

    if hash_table[hash_address] != 0:
        for index in range(len(hash_table[hash_address])):
            if hash_table[hash_address][index][0] == index_key:
                return hash_table[hash_address][index][1]
        return None
    else:
        return None

print (hash('Dave') % 8)
print (hash('Dd') % 8)
print (hash('Data') % 8)
save_data('Dd', '1201023010')
save_data('Data', '3301023010')
read_data('Dd')
print(hash_table)

# Linear probling
hash_table = list([0 for i in range(8)])

def get_key(data):
    return hash(data)

def hash_function(key):
    return key % 8

def save_data(data, value):
    index_key = get_key(data)
    hash_address = hash_function(index_key)
    if hash_table[hash_address] != 0:
        for index in range(hash_address, len(hash_table)):
            if hash_table[index] == 0:
                hash_table[index] = [index_key, value]
                return
            elif hash_table[index][0] == index_key:
                hash_table[index][1] = value
                return
    else:
        hash_table[hash_address] = [index_key, value]

def read_data(data):
    index_key = get_key(data)
    hash_address = hash_function(index_key)
    
    if hash_table[hash_address] != 0:
        for index in range(hash_address, len(hash_table)):
            if hash_table[index] == 0:
                return None
            elif hash_table[index][0] == index_key:
                return hash_table[index][1]
    else:
        return None

print (hash('dk') % 8)
print (hash('da') % 8)
print (hash('dc') % 8)
save_data('dk', '01200123123')
save_data('da', '3333333333')
read_data('dc')

hash() may return different values in every run.
SHA (secure hash algorithm) is well-known
- Any data has unique return value

# using hashlib and chaining
import hashlib

hash_table = list([0 for i in range(8)])

def get_key(data):
        hash_object = hashlib.sha256()
        hash_object.update(data.encode())
        hex_dig = hash_object.hexdigest()
        return int(hex_dig, 16)

def hash_function(key):
    return key % 8

def save_data(data, value):
    index_key = get_key(data)
    hash_address = hash_function(index_key)
    if hash_table[hash_address] != 0:
        for index in range(hash_address, len(hash_table)):
            if hash_table[index] == 0:
                hash_table[index] = [index_key, value]
                return
            elif hash_table[index][0] == index_key:
                hash_table[index][1] = value
                return
    else:
        hash_table[hash_address] = [index_key, value]

def read_data(data):
    index_key = get_key(data)
    hash_address = hash_function(index_key)
    
    if hash_table[hash_address] != 0:
        for index in range(hash_address, len(hash_table)):
            if hash_table[index] == 0:
                return None
            elif hash_table[index][0] == index_key:
                return hash_table[index][1]
    else:
        return None

print (get_key('db') % 8)
print (get_key('da') % 8)
print (get_key('dh') % 8)
save_data('da', '01200123123')
save_data('dh', '3333333333')
read_data('dh')

Tree

Data structure that uses node and branch
Usage: Searching

class Node:
    def __init__(self, value):
        self.value = value
        self.left = None
        self.right = None

        
class NodeMgmt:
    def __init__(self, head):
        self.head = head
    
    def insert(self, value):
        self.current_node = self.head
        while True:
            if value < self.current_node.value:
                if self.current_node.left != None:
                    self.current_node = self.current_node.left
                else:
                    self.current_node.left = Node(value)
                    break
            else:
                if self.current_node.right != None:
                    self.current_node = self.current_node.right
                else:
                    self.current_node.right = Node(value)
                    break
    
    def search(self, value):
        self.current_node = self.head
        while self.current_node:
            if self.current_node.value == value:
                return True
            elif value < self.current_node.value:
                self.current_node = self.current_node.left
            else:
                self.current_node = self.current_node.right
        return False        
    
    def delete(self, value):
        # 삭제할 노드 탐색
        searched = False
        self.current_node = self.head
        self.parent = self.head
        while self.current_node:
            if self.current_node.value == value:
                searched = True
                break
            elif value < self.current_node.value:
                self.parent = self.current_node
                self.current_node = self.current_node.left
            else:
                self.parent = self.current_node
                self.current_node = self.current_node.right

        if searched == False:
            return False    

        # case1
        if  self.current_node.left == None and self.current_node.right == None:
            if value < self.parent.value:
                self.parent.left = None
            else:
                self.parent.right = None
        
        # case2
        elif self.current_node.left != None and self.current_node.right == None:
            if value < self.parent.value:
                self.parent.left = self.current_node.left
            else:
                self.parent.right = self.current_node.left
        elif self.current_node.left == None and self.current_node.right != None:
            if value < self.parent.value:
                self.parent.left = self.current_node.right
            else:
                self.parent.right = self.current_node.right        
        
        # case 3
        elif self.current_node.left != None and self.current_node.right != None:
            # case3-1
            if value < self.parent.value:
                self.change_node = self.current_node.right
                self.change_node_parent = self.current_node.right
                while self.change_node.left != None:
                    self.change_node_parent = self.change_node
                    self.change_node = self.change_node.left
                if self.change_node.right != None:
                    self.change_node_parent.left = self.change_node.right
                else:
                    self.change_node_parent.left = None
                self.parent.left = self.change_node
                self.change_node.right = self.current_node.right
                self.change_node.left = self.change_node.left
            # case 3-2
            else:
                self.change_node = self.current_node.right
                self.change_node_parent = self.current_node.right
                while self.change_node.left != None:
                    self.change_node_parent = self.change_node
                    self.change_node = self.change_node.left
                if self.change_node.right != None:
                    self.change_node_parent.left = self.change_node.right
                else:
                    self.change_node_parent.left = None
                self.parent.right = self.change_node
                self.change_node.right = self.current_node.right
                self.change_node.left = self.current_node.left

        return True

# test
# 0 ~ 999 숫자 중에서 임의로 100개를 추출해서, 이진 탐색 트리에 입력, 검색, 삭제
import random

# 0 ~ 999 중, 100 개의 숫자 랜덤 선택
bst_nums = set()
while len(bst_nums) != 100:
    bst_nums.add(random.randint(0, 999))
#print (bst_nums)

# 선택된 100개의 숫자를 이진 탐색 트리에 입력, 임의로 루트노드는 500을 넣기로 함
head = Node(500)
binary_tree = NodeMgmt(head)
for num in bst_nums:
    binary_tree.insert(num)
    
# 입력한 100개의 숫자 검색 (검색 기능 확인)
for num in bst_nums:
    if binary_tree.search(num) == False:
        print ('search failed', num)

# 입력한 100개의 숫자 중 10개의 숫자를 랜덤 선택
delete_nums = set()
bst_nums = list(bst_nums)
while len(delete_nums) != 10:
    delete_nums.add(bst_nums[random.randint(0, 99)])

# 선택한 10개의 숫자를 삭제 (삭제 기능 확인)
for del_num in delete_nums:
    if binary_tree.delete(del_num) == False:
        print('delete failed', del_num)

Heap

Complete Binary Tree that can fast search min and max value
Usage: When algo need faster min/max return such as priority queue (regular min/max run in O(n) while heap run in O(n log n))

Heap vs Binary Tree

Same: Both are binary tree
Difference:

Heap	Binary Tree
Each nodes’ value is greater than or equal to child node (for __MAX heap)__ (Heap doesn’t have rule about which child is large or small)	left child < parent < right child
For finding max/min	For Searching

# heap
class Heap:
    def __init__(self, data):
        self.heap_array = list()
        self.heap_array.append(None)
        self.heap_array.append(data)
    
    def move_down(self, popped_idx):
        left_child_popped_idx = popped_idx * 2
        right_child_popped_idx = popped_idx * 2 + 1
        
        # case1: 왼쪽 자식 노드도 없을 때
        if left_child_popped_idx >= len(self.heap_array):
            return False
        # case2: 오른쪽 자식 노드만 없을 때
        elif right_child_popped_idx >= len(self.heap_array):
            if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
                return True
            else:
                return False
        # case3: 왼쪽, 오른쪽 자식 노드 모두 있을 때
        else:
            if self.heap_array[left_child_popped_idx] > self.heap_array[right_child_popped_idx]:
                if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
                    return True
                else:
                    return False
            else:
                if self.heap_array[popped_idx] < self.heap_array[right_child_popped_idx]:
                    return True
                else:
                    return False
    
    def pop(self):
        if len(self.heap_array) <= 1:
            return None
        
        returned_data = self.heap_array[1]
        self.heap_array[1] = self.heap_array[-1]
        del self.heap_array[-1]
        popped_idx = 1
        
        while self.move_down(popped_idx):
            left_child_popped_idx = popped_idx * 2
            right_child_popped_idx = popped_idx * 2 + 1

            # case2: 오른쪽 자식 노드만 없을 때
            if right_child_popped_idx >= len(self.heap_array):
                if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
                    self.heap_array[popped_idx], self.heap_array[left_child_popped_idx] = self.heap_array[left_child_popped_idx], self.heap_array[popped_idx]
                    popped_idx = left_child_popped_idx
            # case3: 왼쪽, 오른쪽 자식 노드 모두 있을 때
            else:
                if self.heap_array[left_child_popped_idx] > self.heap_array[right_child_popped_idx]:
                    if self.heap_array[popped_idx] < self.heap_array[left_child_popped_idx]:
                        self.heap_array[popped_idx], self.heap_array[left_child_popped_idx] = self.heap_array[left_child_popped_idx], self.heap_array[popped_idx]
                        popped_idx = left_child_popped_idx
                else:
                    if self.heap_array[popped_idx] < self.heap_array[right_child_popped_idx]:
                        self.heap_array[popped_idx], self.heap_array[right_child_popped_idx] = self.heap_array[right_child_popped_idx], self.heap_array[popped_idx]
                        popped_idx = right_child_popped_idx
        
        return returned_data
    
    def move_up(self, inserted_idx):
        if inserted_idx <= 1:
            return False
        parent_idx = inserted_idx // 2
        if self.heap_array[inserted_idx] > self.heap_array[parent_idx]:
            return True
        else:
            return False

    def insert(self, data):
        if len(self.heap_array) == 1:
            self.heap_array.append(data)
            return True
        
        self.heap_array.append(data)
        inserted_idx = len(self.heap_array) - 1
        
        while self.move_up(inserted_idx):
            parent_idx = inserted_idx // 2
            self.heap_array[inserted_idx], self.heap_array[parent_idx] = self.heap_array[parent_idx], self.heap_array[inserted_idx]
            inserted_idx = parent_idx
        return True    

Reference: https://www.fun-coding.org/daveblog.html