cut large file, faster than system's cut by seeking to the target position instead of by chars or lines

cut.py

#!/usr/bin/env python3
import os, sys
import argparse

try:
    from tqdm import tqdm
except ModuleNotFoundError:
    def tqdm(it):
        return it

BLOCK_SIZE = 1024 * 64

def convert_percent(arg, total):
    if arg is None:
        return 0
    return arg if arg >= 1 else int(total * arg)

def main(args):
    ret = validate_args(args)
    if ret:
        return ret

    file_length = os.path.getsize(args.input)
    cur = start = convert_percent(args.start, file_length)
    length = convert_percent(args.length, file_length)
    end = args.end or args.length and min(start + length, file_length) or file_length
    output = args.output or f'{args.input}.cut'
    count = args.count or 1
    size = max(end - start, BLOCK_SIZE) // count
    size = (size + BLOCK_SIZE - 1) // BLOCK_SIZE * BLOCK_SIZE
    #print(f'params: {file_length=} {start=} {length=} {end=} {output=} {count=} {size=}')

    with open(args.input, 'b+r') as inf:
        inf.seek(start)
        for cnt in range(1, count + 1):
            output_file = f"{output}.{cnt}" if count > 1 else output
            next = min(cur + size, file_length)
            next_size = next - cur
            print(f'Saving "{args.input}" to "{output_file}" from position {cur} to {next} (size={next_size:,})')
            with open(output_file, 'b+w') as outf:
                for i in tqdm(range((next_size + BLOCK_SIZE - 1) // BLOCK_SIZE)):
                    r = inf.read(BLOCK_SIZE)
                    if len(r) == 0:
                        break
                    outf.write(r)
                else:
                    cur = inf.tell()
                    continue
                break

def validate_args(args):
    if args.end and args.end < args.start:
        print(f'{args.end=} is smaller than {args.start=}')
        return -1
    return 0

def convert_args(args):
    for k, v in args._get_kwargs():
        if k in ('input', 'output'):
            continue
        if v is None:
            setattr(args, k, 0)
        elif v.endswith('G'):
            setattr(args, k, int(v[:-1]) * 1000 * 1000 * 1000)
        elif v.endswith('Gi'):
            setattr(args, k, int(v[:-2]) * 1024 * 1024 * 1024)
        elif v.endswith('M'):
            setattr(args, k, int(v[:-1]) * 1000 * 1000)
        elif v.endswith('Mi'):
            setattr(args, k, int(v[:-2]) * 1024 * 1024)
        elif v.endswith('K'):
            setattr(args, k, int(v[:-1]) * 1000)
        elif v.endswith('Ki'):
            setattr(args, k, int(v[:-2]) * 1024)
        elif v.endswith('%'):
            setattr(args, k, int(v[:-1]) / 100)
        else:
            setattr(args, k, int(v))

if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('input')
    parser.add_argument('-s', '--start')
    parser.add_argument('-e', '--end')
    parser.add_argument('-l', '--length')
    parser.add_argument('-c', '--count', help="the count of split files with the same size")
    parser.add_argument('-o', '--output')

    args = parser.parse_args()

    convert_args(args)

    sys.exit(main(args))