391 lines
15 KiB
Python
391 lines
15 KiB
Python
import time
|
||
import random
|
||
import csv
|
||
import os
|
||
import matplotlib.pyplot as plt
|
||
import numpy as np
|
||
|
||
# ===================== 1. Связный список =====================
|
||
def ll_insert(head, name, phone):
|
||
"""Вставка в конец (или обновление), возвращает голову."""
|
||
new_node = {'name': name, 'phone': phone, 'next': None}
|
||
if head is None:
|
||
return new_node
|
||
cur = head
|
||
while True:
|
||
if cur['name'] == name:
|
||
cur['phone'] = phone
|
||
return head
|
||
if cur['next'] is None:
|
||
break
|
||
cur = cur['next']
|
||
cur['next'] = new_node
|
||
return head
|
||
|
||
def ll_find(head, name):
|
||
cur = head
|
||
while cur:
|
||
if cur['name'] == name:
|
||
return cur['phone']
|
||
cur = cur['next']
|
||
return None
|
||
|
||
def ll_delete(head, name):
|
||
if head is None:
|
||
return None
|
||
if head['name'] == name:
|
||
return head['next']
|
||
cur = head
|
||
while cur['next']:
|
||
if cur['next']['name'] == name:
|
||
cur['next'] = cur['next']['next']
|
||
return head
|
||
cur = cur['next']
|
||
return head
|
||
|
||
def ll_list_all(head):
|
||
result = []
|
||
cur = head
|
||
while cur:
|
||
result.append((cur['name'], cur['phone']))
|
||
cur = cur['next']
|
||
result.sort(key=lambda x: x[0])
|
||
return result
|
||
|
||
# ===================== 2. Хеш-таблица =====================
|
||
def ht_hash(name, size):
|
||
h = 0
|
||
for ch in name:
|
||
h = (h * 31 + ord(ch)) % size
|
||
return h
|
||
|
||
def ht_insert(buckets, name, phone):
|
||
idx = ht_hash(name, len(buckets))
|
||
buckets[idx] = ll_insert(buckets[idx], name, phone)
|
||
|
||
def ht_find(buckets, name):
|
||
idx = ht_hash(name, len(buckets))
|
||
return ll_find(buckets[idx], name)
|
||
|
||
def ht_delete(buckets, name):
|
||
idx = ht_hash(name, len(buckets))
|
||
buckets[idx] = ll_delete(buckets[idx], name)
|
||
|
||
def ht_list_all(buckets):
|
||
result = []
|
||
for head in buckets:
|
||
cur = head
|
||
while cur:
|
||
result.append((cur['name'], cur['phone']))
|
||
cur = cur['next']
|
||
result.sort(key=lambda x: x[0])
|
||
return result
|
||
|
||
# ===================== 3. BST =====================
|
||
def bst_insert(root, name, phone):
|
||
"""Итеративная вставка, не вызывает переполнения стека."""
|
||
new_node = {'name': name, 'phone': phone, 'left': None, 'right': None}
|
||
if root is None:
|
||
return new_node
|
||
|
||
cur = root
|
||
while True:
|
||
if name < cur['name']:
|
||
if cur['left'] is None:
|
||
cur['left'] = new_node
|
||
break
|
||
cur = cur['left']
|
||
elif name > cur['name']:
|
||
if cur['right'] is None:
|
||
cur['right'] = new_node
|
||
break
|
||
cur = cur['right']
|
||
else:
|
||
cur['phone'] = phone # обновление
|
||
break
|
||
return root
|
||
|
||
def bst_find(root, name):
|
||
cur = root
|
||
while cur:
|
||
if name == cur['name']:
|
||
return cur['phone']
|
||
elif name < cur['name']:
|
||
cur = cur['left']
|
||
else:
|
||
cur = cur['right']
|
||
return None
|
||
|
||
def bst_delete(root, name):
|
||
# Ищем узел и его родителя
|
||
parent = None
|
||
cur = root
|
||
while cur and cur['name'] != name:
|
||
parent = cur
|
||
if name < cur['name']:
|
||
cur = cur['left']
|
||
else:
|
||
cur = cur['right']
|
||
if cur is None: # не найден
|
||
return root
|
||
|
||
# Случай 1: нет левого потомка
|
||
if cur['left'] is None:
|
||
child = cur['right']
|
||
# Случай 2: нет правого потомка
|
||
elif cur['right'] is None:
|
||
child = cur['left']
|
||
else:
|
||
# Случай 3: два потомка — ищем минимальный в правом поддереве
|
||
succ_parent = cur
|
||
succ = cur['right']
|
||
while succ['left']:
|
||
succ_parent = succ
|
||
succ = succ['left']
|
||
# Копируем данные
|
||
cur['name'] = succ['name']
|
||
cur['phone'] = succ['phone']
|
||
# Удаляем succ (у него нет левого потомка)
|
||
if succ_parent['left'] == succ:
|
||
succ_parent['left'] = succ['right']
|
||
else:
|
||
succ_parent['right'] = succ['right']
|
||
return root
|
||
|
||
# Подключаем child вместо cur
|
||
if parent is None:
|
||
return child
|
||
if parent['left'] == cur:
|
||
parent['left'] = child
|
||
else:
|
||
parent['right'] = child
|
||
return root
|
||
|
||
def bst_list_all(root):
|
||
result = []
|
||
stack = []
|
||
cur = root
|
||
while stack or cur:
|
||
while cur:
|
||
stack.append(cur)
|
||
cur = cur['left']
|
||
cur = stack.pop()
|
||
result.append((cur['name'], cur['phone']))
|
||
cur = cur['right']
|
||
return result
|
||
|
||
# ===================== Генерация данных =====================
|
||
def generate_data(n=10000):
|
||
records = [(f"User_{i:05d}", f"8800{i:07d}") for i in range(n)]
|
||
shuffled = records[:]
|
||
random.shuffle(shuffled)
|
||
sorted_rec = sorted(records, key=lambda x: x[0])
|
||
return shuffled, sorted_rec
|
||
|
||
# ===================== Замеры =====================
|
||
def run_experiment(struct_type, records, n_searches=100, n_missing=10, n_deletes=50, repeats=5):
|
||
"""
|
||
struct_type: 'll', 'ht', 'bst'
|
||
Возвращает словарь с усреднёнными замерами.
|
||
"""
|
||
all_insert_times = []
|
||
all_search_times = []
|
||
all_delete_times = []
|
||
|
||
for _ in range(repeats):
|
||
# --- инициализация структуры ---
|
||
if struct_type == 'll':
|
||
head = None
|
||
elif struct_type == 'ht':
|
||
buckets = [None] * 512 # размер хеш-таблицы
|
||
else: # bst
|
||
root = None
|
||
|
||
# --- вставка ---
|
||
start = time.perf_counter()
|
||
if struct_type == 'll':
|
||
for name, phone in records:
|
||
head = ll_insert(head, name, phone)
|
||
elif struct_type == 'ht':
|
||
for name, phone in records:
|
||
ht_insert(buckets, name, phone)
|
||
else:
|
||
for name, phone in records:
|
||
root = bst_insert(root, name, phone)
|
||
insert_time = time.perf_counter() - start
|
||
all_insert_times.append(insert_time)
|
||
|
||
# --- поиск ---
|
||
existing = random.sample(records, min(n_searches, len(records)))
|
||
missing = [(f"Missing_{i}", "") for i in range(n_missing)]
|
||
test_keys = existing + missing
|
||
random.shuffle(test_keys)
|
||
|
||
start = time.perf_counter()
|
||
if struct_type == 'll':
|
||
for name, _ in test_keys:
|
||
ll_find(head, name)
|
||
elif struct_type == 'ht':
|
||
for name, _ in test_keys:
|
||
ht_find(buckets, name)
|
||
else:
|
||
for name, _ in test_keys:
|
||
bst_find(root, name)
|
||
search_time = time.perf_counter() - start
|
||
all_search_times.append(search_time)
|
||
|
||
# --- удаление ---
|
||
del_sample = random.sample(records, min(n_deletes, len(records)))
|
||
start = time.perf_counter()
|
||
if struct_type == 'll':
|
||
for name, _ in del_sample:
|
||
head = ll_delete(head, name)
|
||
elif struct_type == 'ht':
|
||
for name, _ in del_sample:
|
||
ht_delete(buckets, name)
|
||
else:
|
||
for name, _ in del_sample:
|
||
root = bst_delete(root, name)
|
||
delete_time = time.perf_counter() - start
|
||
all_delete_times.append(delete_time)
|
||
|
||
return {
|
||
'struct': struct_type,
|
||
'insert_avg': sum(all_insert_times) / repeats,
|
||
'search_avg': sum(all_search_times) / repeats,
|
||
'delete_avg': sum(all_delete_times) / repeats,
|
||
'insert_all': all_insert_times,
|
||
'search_all': all_search_times,
|
||
'delete_all': all_delete_times,
|
||
}
|
||
|
||
def main():
|
||
random.seed(42)
|
||
N = 10000
|
||
shuffled, sorted_rec = generate_data(N)
|
||
|
||
results = []
|
||
for struct_name, label in [('ll', 'LinkedList'), ('ht', 'HashTable'), ('bst', 'BST')]:
|
||
for order_name, records in [('shuffled', shuffled), ('sorted', sorted_rec)]:
|
||
print(f"Тестирую {label} на {order_name} данных...")
|
||
res = run_experiment(struct_name, records)
|
||
res['order'] = order_name
|
||
res['label'] = label
|
||
results.append(res)
|
||
print(f"{label:15} | {order_name:10} | insert: {res['insert_avg']:.6f}s | "
|
||
f"search: {res['search_avg']:.6f}s | delete: {res['delete_avg']:.6f}s")
|
||
|
||
# Сохраняем в CSV
|
||
os.makedirs('docs/data', exist_ok=True)
|
||
with open('docs/data/benchmark_results.csv', 'w', newline='') as f:
|
||
writer = csv.writer(f)
|
||
writer.writerow(['structure', 'order', 'run', 'insert', 'search', 'delete'])
|
||
for r in results:
|
||
for i in range(len(r['insert_all'])):
|
||
writer.writerow([r['label'], r['order'], i + 1,
|
||
r['insert_all'][i], r['search_all'][i], r['delete_all'][i]])
|
||
print("\nCSV сохранён в docs/data/benchmark_results.csv")
|
||
|
||
# ===================== ГРАФИКИ =====================
|
||
structures = ['LinkedList', 'HashTable', 'BST']
|
||
orders = ['shuffled', 'sorted']
|
||
metrics = ['insert', 'search', 'delete']
|
||
metric_names = {'insert': 'Вставка (сек)', 'search': 'Поиск (сек)', 'delete': 'Удаление (сек)'}
|
||
colors = {'shuffled': '#4CAF50', 'sorted': '#FF5722'}
|
||
|
||
fig, axes = plt.subplots(1, 3, figsize=(16, 5.5))
|
||
|
||
for idx, metric in enumerate(metrics):
|
||
ax = axes[idx]
|
||
x = np.arange(len(structures))
|
||
width = 0.35
|
||
|
||
# Собираем данные
|
||
shuffled_vals = []
|
||
sorted_vals = []
|
||
for struct in structures:
|
||
for res in results:
|
||
if res['label'] == struct and res['order'] == 'shuffled':
|
||
shuffled_vals.append(res[f'{metric}_avg'])
|
||
elif res['label'] == struct and res['order'] == 'sorted':
|
||
sorted_vals.append(res[f'{metric}_avg'])
|
||
|
||
bars1 = ax.bar(x - width/2, shuffled_vals, width, label='Случайный порядок',
|
||
color=colors['shuffled'], edgecolor='black', linewidth=0.5)
|
||
bars2 = ax.bar(x + width/2, sorted_vals, width, label='Отсортированный порядок',
|
||
color=colors['sorted'], edgecolor='black', linewidth=0.5)
|
||
|
||
# Подписи значений на столбцах
|
||
for bar in bars1:
|
||
height = bar.get_height()
|
||
ax.text(bar.get_x() + bar.get_width()/2., height + max(shuffled_vals)*0.01,
|
||
f'{height:.4f}', ha='center', va='bottom', fontsize=7)
|
||
for bar in bars2:
|
||
height = bar.get_height()
|
||
ax.text(bar.get_x() + bar.get_width()/2., height + max(sorted_vals)*0.01,
|
||
f'{height:.4f}', ha='center', va='bottom', fontsize=7)
|
||
|
||
ax.set_title(metric_names[metric], fontsize=12, fontweight='bold')
|
||
ax.set_xticks(x)
|
||
ax.set_xticklabels(structures, fontsize=10)
|
||
ax.legend(fontsize=9)
|
||
ax.grid(axis='y', alpha=0.3, linestyle='--')
|
||
|
||
# Для поиска — логарифмическая шкала (чтобы было видно разницу)
|
||
if metric == 'search':
|
||
ax.set_yscale('log')
|
||
ax.set_ylabel('Время (сек, лог. шкала)', fontsize=9)
|
||
else:
|
||
ax.set_ylabel('Время (сек)', fontsize=9)
|
||
|
||
plt.suptitle('Сравнение производительности структур данных (N = 10 000 записей)',
|
||
fontsize=14, fontweight='bold', y=1.02)
|
||
plt.tight_layout()
|
||
|
||
# Сохраняем график
|
||
graph_path = 'docs/benchmark_graph.png'
|
||
os.makedirs('docs', exist_ok=True)
|
||
plt.savefig(graph_path, dpi=150, bbox_inches='tight')
|
||
plt.show()
|
||
print(f"График сохранён в {graph_path}")
|
||
|
||
print("АНАЛИЗ РЕЗУЛЬТАТОВ")
|
||
|
||
print("\n1. Влияние порядка данных на BST:")
|
||
bst_shuffled_insert = next(r['insert_avg'] for r in results if r['label']=='BST' and r['order']=='shuffled')
|
||
bst_sorted_insert = next(r['insert_avg'] for r in results if r['label']=='BST' and r['order']=='sorted')
|
||
print(f" - Случайные данные: {bst_shuffled_insert:.6f} сек")
|
||
print(f" - Отсортированные данные: {bst_sorted_insert:.6f} сек")
|
||
print(f" - Замедление в {bst_sorted_insert/bst_shuffled_insert:.1f} раз")
|
||
print(" Причина: на отсортированных данных BST вырождается в связный список (глубина = N)")
|
||
|
||
print("\n2. Стабильность хеш-таблицы:")
|
||
ht_shuffled = next(r['insert_avg'] for r in results if r['label']=='HashTable' and r['order']=='shuffled')
|
||
ht_sorted = next(r['insert_avg'] for r in results if r['label']=='HashTable' and r['order']=='sorted')
|
||
print(f" - Случайные: {ht_shuffled:.6f} сек")
|
||
print(f" - Отсортированные: {ht_sorted:.6f} сек")
|
||
print(" Причина: хеш-функция равномерно распределяет ключи независимо от порядка")
|
||
|
||
print("\n3. Медленный поиск в связном списке:")
|
||
ll_search = next(r['search_avg'] for r in results if r['label']=='LinkedList' and r['order']=='shuffled')
|
||
ht_search = next(r['search_avg'] for r in results if r['label']=='HashTable' and r['order']=='shuffled')
|
||
print(f" - LinkedList: {ll_search:.6f} сек")
|
||
print(f" - HashTable: {ht_search:.6f} сек")
|
||
print(f" - Хеш-таблица быстрее в {ll_search/ht_search:.1f} раз")
|
||
print(" Причина: поиск в списке всегда O(n), в хеш-таблице ~O(1)")
|
||
|
||
print("\n4. Удаление:")
|
||
for label in ['LinkedList', 'HashTable', 'BST']:
|
||
del_shuff = next(r['delete_avg'] for r in results if r['label']==label and r['order']=='shuffled')
|
||
del_sort = next(r['delete_avg'] for r in results if r['label']==label and r['order']=='sorted')
|
||
print(f" - {label:15}: случ.={del_shuff:.6f} сек, отсорт.={del_sort:.6f} сек")
|
||
|
||
print("\n5. Рекомендации:")
|
||
print(" - Частый поиск + вставки → Хеш-таблица")
|
||
print(" - Нужна сортировка «из коробки» → Сбалансированное BST (AVL/Красно-чёрное)")
|
||
print(" - Только добавление в конец → Связный список")
|
||
print(" - Обычный BST опасен на реальных частично упорядоченных данных!")
|
||
print("="*60)
|
||
|
||
if __name__ == '__main__':
|
||
main() |