pysize.py

import sys

def get_size(obj, seen=None):
    """Recursively finds size of objects"""
    size = sys.getsizeof(obj)
    if seen is None:
        seen = set()
    obj_id = id(obj)
    if obj_id in seen:
        return 0
    # Important mark as seen *before* entering recursion to gracefully handle
    # self-referential objects
    seen.add(obj_id)
    if isinstance(obj, dict):
        size += sum([get_size(v, seen) for v in obj.values()])
        size += sum([get_size(k, seen) for k in obj.keys()])
    elif hasattr(obj, '__dict__'):
        size += get_size(obj.__dict__, seen)
    elif hasattr(obj, '__iter__') and not isinstance(obj, (str, bytes, bytearray)):
        size += sum([get_size(i, seen) for i in obj])
    return size

Amazing code ! thanks

@bosswissam

Is this not weird?

>>> from pysize import get_size
>>> get_size(b)
152
>>> get_size(b[0])
28
>>> get_size(b[1])
28
>>> len(b)
300

I've been happily using this code for a long time, but I just encountered a use case where this breaks down: a class built over a simple namedtuple data core. This pattern is desirable for certain multi-processing/cloud computing contexts.

from __future__ import print_function
from collections import namedtuple
import sys
import numpy as np

my_tup = namedtuple('MyNamedTuple', ['Array','Name'])
class my_class(my_tup):
    def __init__(self, *kwargs):
        super(my_class, self).__init__(*kwargs)
    # Add workhorse functions...

dat_tuple = my_tup(np.zeros([1000,1000]), 'long name'*10)
dat_obj = my_class(np.zeros([1000,1000]), 'long name'*10)
print(get_size(dat_tuple), get_size(dat_obj))

These sizes should be almost the same, but they are not.

8000946 360

The problem is caused because dat_obj has an empty __dict__ and data stored in __iter__.

Here is the fix I made. It doesn't come out exactly the same, but it's a lot closer than before:

def get_size2(obj, seen=None):
    """Recursively finds size of objects"""
    size = sys.getsizeof(obj)
    if seen is None:
        seen = set()
    obj_id = id(obj)
    if obj_id in seen:
        return 0
    # Important mark as seen *before* entering recursion to gracefully handle
    # self-referential objects
    seen.add(obj_id)
    if isinstance(obj, dict):
        size += sum([get_size(v, seen) for v in obj.values()])
        size += sum([get_size(k, seen) for k in obj.keys()])
    elif hasattr(obj, '__iter__') and not isinstance(obj, (str, bytes, bytearray)):
        size += sum([get_size(i, seen) for i in obj])
        if hasattr(obj, '__dict__'):
            size += get_size(obj.__dict__.values(), seen)
    elif hasattr(obj, '__dict__'):
        size += get_size(obj.__dict__, seen)
    return size

print(get_size2(dat_tuple), get_size2(dat_obj))

8000671 8000647

I implemented a truly generic solution here.
It uses the gc module instead of trying to guess the object children and avoid using recursive calls.
You can simply install it via pip if you want: pip install objsize.

There is python module that provides similar functionality and other things as well such as tracking the memory consumption of the instances of a specific class, etc. called Pympler.
https://pympler.readthedocs.io/en/latest/