cf.py - recursively count the files within a directory tree, displaying directories with the highest file count.

cf.py

#!/usr/bin/env python3

import os
import os.path
from contextlib import suppress
import argparse


def visit(path, state):
    if os.path.isdir(path):
        count, size = 0, 0
        for entry in os.listdir(path):
            with suppress(Exception):
                entry_abspath = os.path.join(path, entry)
                c_e, s_e = visit(entry_abspath, state)
                count += c_e
                size += s_e
        state[path] = (count, size)
        return (count, size)
    else:
        return (1, os.path.getsize(path))


def to_readable_size(n_bytes):
    n_bytes = float(n_bytes)
    unit = ['bytes', 'KiB', 'MiB', 'GiB', 'TiB', 'PiB']
    idx = 0
    while n_bytes > 1000 and idx < len(unit) - 1:
        n_bytes /= 1024.0
        idx += 1
    return f"{n_bytes:.2f} {unit[idx]}"    
    


def print_report(state : dict, limit : int):
    top = sorted(state.keys(), reverse=True, key=lambda x: state[x][0])[:limit]
    print(f"\nTop {limit} directories by file count:")
    for k in top:
        c, s = state[k]
        print(f"Directory '{k}' - count: {c}, size: {to_readable_size(s)}")



if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument('path', type=str, default=os.getcwd(), help="Path to the directory to analyse.")
    parser.add_argument('-l', '--limit', default=10, type=int, help="Limit the directories with the highest file count being displayed to a certain number.")

    args = vars(parser.parse_args())

    start = os.path.abspath(args['path'])
    state = {}
    visit(start, state)
    count, size = state[start]
    print(f"Root directory '{start}' contains {count} files and occupies a total of {to_readable_size(size)}")
    print_report(state, args['limit'])