How to Use Python's Import Feature - From X Import Y Examples

If you have read the import statements wiki then you’d have seen me use this statement in one of the examples. Today, we’ll try to understand what it does

So picking up the same example:

>>> from math import ceil, sqrt
>>> # here it would be
>>> sqrt(36)
<<< 6

Run Code

Or we could use this:

>>> import math
>>> # here it would be
>>> math.sqrt(36)
<<< 6

Run Code

Then our code would look like math.sqrt(x) instead of sqrt(x) . This happens because when we use import x , a namespace x is itself created to avoid name conflicts. You have to access every single object of the module as x.<name> . But when we use from x import y we agree to add y to the main global namespace. So while using this we have to make sure that we don’t have an object with same name in our program.

Never use from x import y if an object named y already exists

For example, in os module there’s a method open . But we even have a built-in function called open . So, here we should avoid using from os import open .

We can even use form x import * , this would import all the methods, classes of that module to the global namespace of the program. This is a bad programming practice. Please avoid it.

In general you should avoid from x import y simply because of the problems it may cause in large scale programs. For example, you never know if a fellow programmer might want to make a new function that happens to be the name of one of the existing functions. You also do not know whether Python will change the library that you are importing functions from. While these problems won’t exist as often for solo projects, as stated before, it is bad programming practice and should be avoided.

You may use from x import y as z , this would import y from x module as z . Using as you can rename the method imported according to your wish. At times, the method name may be too large to spell again and again. By using as , you can rename it as you wish. For example, from matplotlib import pyplot as plt - here pyplot has been renamed to plt and can be used normally.