Implementing sum() for custom datatype

Question

As I understand it, for it to be possible to cast sum() on an object, it must be an iterable, and must be "addable", i.e. it must implement methods __iter__ and __add__. However, when I do this for my class Point (just an example), this doesn't work.

class Point:
    def __init__(self, x, y):
        self.x = x
        self.y = y

    def __add__(self, other):
        return Point(self.x + other.x, self.y + other.y)

    def __iter__(self):
        return self

    def __str__(self):
        return str((self.x, self.y))


print(Point(2, 2) + Point(1, 1))
>>> (3, 3)    # As expected; good!

points = [Point(0, 0), Point(2, 0), Point(0, 2), Point(2, 2)]
print(sum(points))    # Expect (4, 4)
>>> TypeError: unsupported operand type(s) for +: 'int' and 'Point'

If I implement __radd__ the same as __add__ I then get an attribute error when I try sum():

AttributeError: 'int' object has no attribute 'x'

Based on the errors my Points are somewhere being separated to just ints but I'm not sure where. Thanks for any help.

Note that sum is not required to work for non-numeric types. It's not entirely clear, though, what sum will consider a numeric type. (For example, it works for lists, but explicitly rejects strs.) — chepner
– chepner, Commented Nov 18, 2019 at 14:15
Isn't that just because ''.join() is a better alternative for string concatenation than sum()? Edit, see this answer — QuantumChris
– QuantumChris, Commented Nov 18, 2019 at 14:18
It absolutely is, though I'm not entirely sure why sum can't be implemented to call join itself, rather than raise an exception. — chepner
– chepner, Commented Nov 18, 2019 at 14:24
I'm inclined to agree, a few others on the above thread agree too. — QuantumChris
– QuantumChris, Commented Nov 18, 2019 at 14:27

han solo · Accepted Answer · 2019-11-18 14:17:28Z

13

This happens because sum starts with a default value of int and when doing sum(points) what really is happening is sum is first trying to add 0 + Point(0, 0) and hence the error. If you check the help on sum, this will become pretty evident,

Help on built-in function sum in module builtins:

sum(iterable, start=0, /) Return the sum of a 'start' value (default: 0) plus an iterable of numbers
When the iterable is empty, return the start value.
This function is intended specifically for use with numeric values and may
reject non-numeric types.

Change the line from,

>>> print(sum(points))

to

>>> print(sum(points, Point(0, 0)))

edited Nov 18, 2019 at 14:17

answered Nov 18, 2019 at 14:10

han solo

6,6501 gold badge20 silver badges22 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

QuantumChris Over a year ago

Interesting, thanks. I wonder why sum(sequence) was chosen to start from 0 rather than just add the elements left to right (or right to left). Maybe to handle cases where len(sequence) <= 1.

han solo Over a year ago

I am not too sure why the implementation chose that. I'd have to check : )

chepner Over a year ago

0 is just the default starting value, both to handle empty sequences and the normal use of simply summing the values. You can pass a different starting value if you like. sum(values, x) == x + sum(values). For example, sum([[1], [2]], [3]) == [3] + sum([[1], [2]], []) == [3, 1, 2].

han solo Over a year ago

Yeah, make sense to have the starting value as 0. It's easier to have a starting value, and since usually the iterable’s items are normally numbers, and the start value is not allowed to be a string.

Collectives™ on Stack Overflow

Implementing sum() for custom datatype

1 Answer 1

4 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related