Регулярные выражения в Python от простого к сложному. Подробности, примеры, картинки, упражнения

from time import perf_counter
import re
import random
from string import ascii_lowercase, ascii_uppercase
chars = ''.join(chr(i) for i in range(33, 127))
chars += ascii_uppercase * 1 + ascii_lowercase * 7
chars += ' ' * 30

NUM_RUNS = 10
NUM_TEXTS = 10000
TEXT_LENS = 1000

texts = []
for __ in range(NUM_TEXTS):
    texts.append(''.join(random.choices(chars, k=TEXT_LENS)))

regexps = [
    r'\b[a-z]+\b',  # слова только из маленьких букв
    r'\b[A-Z]\w+\b',  # слова с заглавной
    r'\b(\w{10})\b',  # слова из 10 символов с сохранением
    r'te\w*st',  # Ищем тест
    r'a\w*b\w*c',  # a*b*c
    r'\(([^)]*)\)',  # (...) с сохранением
    r'\W{3,}',  # Длинные не-слова
    r'[aeiouy]+',  # Только гласные
    r'(?:[aeiouy][bcdfghjklmnpqrstvwxz])+',  # Читаем по слогам
    r'[\s,.!?;]+',  # Для сплит'а
]

def test_raw():
    tot = 0
    st = perf_counter()
    for text in texts:
        for regex in regexps:
            tot += sum(1 for m in re.finditer(regex, text))
    en = perf_counter()
    print(f'{tot} matches found in {en-st:0.4} seconds (without compiling)')
    return en-st


def test_compiled():
    tot = 0
    st = perf_counter()
    regexps_compiled = [re.compile(r) for r in regexps]
    for text in texts:
        for regex in regexps_compiled:
            tot += sum(1 for m in regex.finditer(text))
    en = perf_counter()
    print(f'{tot} matches found in {en-st:0.4} seconds (with compiling)')
    return en-st


raw_durs = [test_raw() for __ in range(NUM_RUNS)]
compiled_durs = [test_compiled() for __ in range(NUM_RUNS)]
tot_runs = NUM_RUNS*NUM_TEXTS*len(regexps)
raw_per_regex = sum(raw_durs) / tot_runs
comp_per_regex = sum(compiled_durs) / tot_runs

print(f'{tot_runs} finditer runs total. {sum(raw_durs):.2f} sec for raw VS {sum(compiled_durs):.2f} sec for compiled')
print(f'Raw      regexp run: {raw_per_regex:.3} seconds per regexp, x{comp_per_regex/raw_per_regex:.3f} faster')
print(f'Compiled regexp run: {comp_per_regex:.3} seconds per regexp, x{raw_per_regex/comp_per_regex:.3f} faster')