To calculate running median given a huge list of numbers

heaps.py

from heapq import heappop, heappush


class MinHeap(object):

    def __init__(self):
        self._items = []

    def push(self, item):
        heappush(self._items, item)

    def peek(self):
        return self._items[0]

    def pop(self):
        return heappop(self._items)

    def length(self):
        return len(self._items)

    def __repr__(self):
        return str(self._items)


class MaxHeap(object):

    def __init__(self):
        self._items = []

    def push(self, item):
        heappush(self._items, -item)

    def peek(self):
        return -self._items[0]

    def pop(self):
        return -heappop(self._items)

    def length(self):
        return len(self._items)

    def __repr__(self):
        return str(self._items)

running_median.py

from heaps import MinHeap, MaxHeap


with open('median_stream.txt', 'r') as f:
    numbers = [int(line) for line in f.readlines()]


def balance_heaps(left_heap, right_heap):
    if left_heap.length() > right_heap.length():
        right_heap.push(left_heap.pop())
    elif right_heap.length() > left_heap.length():
        left_heap.push(right_heap.pop())

    return (left_heap, right_heap)


def number_stream():
    for n in numbers:
        yield n

left_heap = MaxHeap()
right_heap = MinHeap()

ctr = 0
current = None
medians = []
for num in number_stream():
    ctr += 1
    if ctr == 1:
        left_heap.push(num)
        medians.append(num)
        current = num
        continue

    if abs(left_heap.length() - right_heap.length()) == 2:
        left_heap, right_heap = balance_heaps(left_heap, right_heap)
        # change the current to new median
        current = left_heap.peek()

    if num <= current:
        left_heap.push(num)
    if num > current:
        right_heap.push(num)

    if abs(left_heap.length() - right_heap.length()) == 2:
        left_heap, right_heap = balance_heaps(left_heap, right_heap)

    if ctr % 2 == 0:
        current = left_heap.peek()
    else:
        if right_heap.length() > left_heap.length():
            current = right_heap.peek()
        else:
            current = left_heap.peek()

    medians.append(current)


print sum(medians) % 10000