analyze_lots_of_replays.py

from pathlib import Path
from multiprocessing import Pool
from sc2_tournament_analysis import recursive_parse
from zephyrus_sc2_parser import parse_replay

def handle_replay(path, player_names, identifiers):
    replay = parse_replay(path)
    # do stuff and then return the data you want


# required for multiprocessing
if __name__ == '__main__':
    results = []
    replays = Path('replays')
    replay_paths = recursive_parse(
        sub_dir=replays,
        # data_function is unused with multiprocessing, but required
        data_function=handle_replay,
        # this means you get a list of paths instead of synchronous parsing
        multi=True,
    )

    # Pool multiprocessing for better throughput of parsing
    with Pool(10) as p:
        results = p.starmap(handle_replay, replay_paths)
        
    # do stuff with aggregated data