Python Performance Optimization

import cProfile
import pstats

# Function to profile
def slow_function():
    total = 0
    for i in range(1000000):
        total += i * i
    return total

# Run profiler
profiler = cProfile.Profile()
profiler.enable()
result = slow_function()
profiler.disable()

# Print stats
stats = pstats.Stats(profiler)
stats.sort_stats('cumtime').print_stats(10)

# Alternative: timeit
import timeit
time_taken = timeit.timeit(slow_function, number=100)
print(f"Average time: {time_taken/100:.4f} seconds")

# Lists vs Tuples
from sys import getsizeof
list_data = [1, 2, 3, 4, 5]
tuple_data = (1, 2, 3, 4, 5)
print(f"List size: {getsizeof(list_data)} bytes")
print(f"Tuple size: {getsizeof(tuple_data)} bytes")

# Sets for membership testing
large_list = [i for i in range(1000000)]
large_set = set(large_list)
print("Testing membership...")
%timeit 999999 in large_list # ~10ms
%timeit 999999 in large_set # ~100ns

# Collections module optimizations
from collections import deque, defaultdict
# Faster pops from both ends
dq = deque(range(1000000))
%timeit dq.popleft() # Much faster than list.pop(0)

# List comprehensions vs loops
def squares_loop(n):
    result = []
    for i in range(n):
        result.append(i * i)
    return result

def squares_comp(n):
    return [i * i for i in range(n)]

%timeit squares_loop(10000) # ~1.5ms
%timeit squares_comp(10000) # ~1.1ms

# Map and filter
data = range(10000)
%timeit [x * x for x in data if x % 2 == 0] # ~600µs
%timeit list(map(lambda x: x * x, filter(lambda x: x % 2 == 0, data))) # ~800µs

# String joining
words = ['hello'] * 10000
%timeit ''.join(words) # ~100µs
%timeit result = ''; for w in words: result += w # ~2ms

import numpy as np
import array

# Native Python list
py_list = [i for i in range(1000000)]
%timeit [x * 2 for x in py_list] # ~100ms

# Python array
py_array = array.array('i', py_list)
%timeit array.array('i', [x * 2 for x in py_array]) # ~90ms

# NumPy array
np_array = np.arange(1000000)
%timeit np_array * 2 # ~1ms

# Cython example (separate .pyx file)
'''
# cython: language_level=3
def cython_sum(arr):
cdef long total = 0
cdef int i
for i in range(len(arr)):
total += arr[i]
return total
'''
# After compiling, can be 100x faster than Python

# I/O-bound: asyncio
import asyncio
async def fetch_data(url):
    # Simulate network request
    await asyncio.sleep(1)
    return f"data from {url}"

async def main():
    tasks = [fetch_data(f"url-{i}") for i in range(10)]
    results = await asyncio.gather(*tasks)
    print(results)

# CPU-bound: multiprocessing
from multiprocessing import Pool
def cpu_intensive(n):
    return sum(i * i for i in range(n))

if __name__ == '__main__':
    with Pool() as pool:
        results = pool.map(cpu_intensive, range(1000, 1100))
        print(results[:5])

# Mixed: concurrent.futures
from concurrent.futures import ThreadPoolExecutor, ProcessPoolExecutor
def mixed_work(task):
    if task['type'] == 'io':
        # I/O work
    else:
        # CPU work
    return result

with ThreadPoolExecutor() as executor:
    results = list(executor.map(mixed_work, tasks))