Strange 2 dimensional list behaviour in python

Why does python do the following?

>>> bla = [[]] * 5
>>> bla[3].append("blub")
>&g开发者_Go百科t;> print bla
[['blub'], ['blub'], ['blub'], ['blub'], ['blub']]

I would expect

[[], [], [], ['blub'], []]

This is a common point of confusion with Python, to do with the implicit pass-by-reference. When you type

x = []

that binds the name x to point to an empty list in memory. When you do

x = [[]] * 5

that binds the name x to point to a list of five things, each one of which is a reference to the same empty list. This might be clearer if you think of it like this:

>>> y = []
>>> x = [y]*5
>>> x
[[], [], [], [], []]
>>> x[0].append(0)
>>> x
[[0], [0], [0], [0], [0]]
>>> y
[0]

That is, even though [] doesn't bind any variable names, it still creates a reference to an empty list, which is then copied five times.

If you don't want this, you need explicitly to construct five empty lists to store in x. There are various ways of doing this (copy.deepcopy is a general solution), but I think the neatest is:

x = [[] for _ in range(5)]

Alternatively, numpy is an extension module which provides a very powerful multidimensional array.

You're multiplying the contents of the original list five times - that inner list is just a reference, so the same reference is multiplied five times.

In other words, the final list you've got is a list containing the same list multiple times.

An easy way around this for Python 2:

[[] for i in xrange(5)]

Python 3:

[[] for i in range(5)]