sort_integers.py

import sys
import tempfile
import heapq

from array import array


def int_array_from_file(int_file, buffer_size=4000):
    int_array = array("i")
    int_array.frombytes(int_file.read(buffer_size))
    return int_array


def read_buffered_ints(int_file):
    while True:
        int_array = int_array_from_file(int_file)

        if not int_array:
            return None

        yield from int_array


with open("random_numbers.bin", "rb") as int_file, open(
    "sorted_numbers.bin", "wb"
) as final_result:
    int_arrays_generators = []

    # We load the numbers in chunk with this loop
    while True:
        int_array = int_array_from_file(int_file, 40000)

        if not int_array:
            break

        # For each chunk, we create a temp file...
        temp_file = tempfile.TemporaryFile()
        # we sort the arrays, convert them back into bytes, 
        # and put them in the file...
        array("i", sorted(int_array)).tofile(temp_file)
        # then save a generator ready to give us back those ints, 
        # in our list
        temp_file.seek(0)
        generator = read_buffered_ints(temp_file)
        int_arrays_generators.append(generator)

    int_buffer = array("i")
    # heapq.merge give us all integer IN ORDER
    for integer in heapq.merge(*int_arrays_generators):
        # We store them all in an array, and once the array is big enough,
        # we append its content into the final result file
        int_buffer.append(integer)
        if len(int_buffer) >= 1000:
            int_buffer.tofile(final_result)
            del int_buffer[:]  # empty the array, it reached the size limit

    # For the last array if it didn't reach 1000
    if int_buffer:
        int_buffer.tofile(final_result)