Toy grep from Aur's parallelism dojo

parallel_mini_grep.py

import itertools
import os
from concurrent.futures import ThreadPoolExecutor
from pathlib import Path

pool = ThreadPoolExecutor(max_workers=os.cpu_count() * 2)


def find_needle(file, needle):
    with open(file) as f:
        yield from((file, lineno, line) for (lineno, line) in enumerate(f.readlines()) if needle in line)


# needle in a haystack
def grep(path: str, needle: str):
    files = [name for name in sorted(Path(path).glob("*")) if name.is_file()]
    tasks = [pool.submit(find_needle, file, needle) for file in files]
    results = itertools.chain.from_iterable([fut.result() for fut in tasks])
    for result in results:
        file, lineno, line = result
        print(f'{file}:{lineno}: {line}', end='')


if __name__ == '__main__':
    grep(os.getcwd(), 'hay')