Python: copying objects with copy.deepcopy or writing my own copy

Question

Currently I am using the copy module to create a copy of some object that I have.

Under certain conditions (that occur frequently), I will need to create possibly several copies of the original object, and then modify each of them individually (hence why I am using copy).

The object has several instances of various datatypes (lists, ints, strings, other classes), and the copies need to have the same values, but I am not sure whether it would faster to call copy.deepcopy(), or do something like

def copy(self, other):
 other.prop1 = self.prop1
 other.prop2 = self.prop2
 other.prop3 = self.prop3

Has anyone run into this problem and then decided it was better to use the copy module because it would be faster than anything most people could come up with?

PS: the code above wouldn't properly copy mutable objects and those "other classes" that I mentioned. Perhaps that suggests deepcopy is the safest (and fastest) route?

• lambacck: If any of the prop1, prop2 or prop3 are mutable (i.e. a list, dict, set, object etc as opposed to int, tuple, float, frozenset) then what you have above will result in a shared object and possibly result in bugs. a = [1,2,3]; b = a; b.append(4); print a results in [1,2,3,4]
• Karl Knechtel: What kind of object is it? What are the copies for? In Python, like other reference-semantic languages, it's usual to work around the value-copying problem by writing algorithms in a way that avoids the need for value-copying.
• Maxim Egorushkin: The correct name for the above function is shallow_copy.

Answer 1

Not trying to be glib, but have you considered profiling your code? Documentation here. Compare copy.deepcopy with with your own custom copy method. That way you'll have hard data on your particular situation and can make the best decision--and back it up if anyone asks why you did it that way!

• MxLDevs: Yes, I've been thinking of profiling my code for awhile. I will try it out.