CODEWITHSUNDEEP
pythonalgorithmtypescomparison
ThisIsNotAnId
ThisIsNotAnId

Reputation: 197

"Type comparisons" in Python

I'm writing a program where I have to test if elements of a list are of a certain type. The list (z, found below) can get large and I could test the standard way:

for l in z:
  typeL = type(l)
  if typeL == complex or typeL == float:
    #evaluate suite
  elif typeL == list:
    #evaluate different suite

Or, I can do the following:

for l in z:
  lTemp = l*0
  if lTemp == 0:
    #evaluate suite
  elif lTemp == []:
    #evaluate different suite

My question is, which method is faster, more efficient, or preferred? Thanks to all in advance.

Upvotes: 1

Views: 959

Answers (2)

Odomontois
Odomontois

Reputation: 16328

According to Zen of Python

  • Explicit is better than implicit.
  • Simple is better than complex

So if you want to compare types, you should get types and compare them. Clearer method is using isinstance method that is accepting tuple-of-types argument. So your code may look like this:

for l in z:
  if isinstance(l,(complex,float)):
    #evaluate suite
  elif isinstance(l,list):
    #evaluate different suite

Also isinstance is compatible with ABC metaclasses, so you could be more straight and clear in your choises:

from numbers import Number
from collections import Iterable

for l in z:
  if isinstance(l,Number):
    #evaluate suite
  elif isinstance(l,Iterable):
    #evaluate different suite

UPD: Answering to a question in commentary about perfomance, you should remember, that premature optimization is the root of all evil. When your program will work, you could analize it and find where the perfomance actually falls. Next you could optimize it in python, rewrite in cython or C, could use another implementation, etc. But while you are coding in python you must respect what is better for python code.

Upvotes: 12

Jörg Beyer
Jörg Beyer

Reputation: 3671

I am not sure, what you try to do, but since you have the two implementations (nearly) ready, you should benchmark them with your data, to decide which implementation is faster in your case.

I would be surprised to see a big difference in performance, only if #evaluate suite is a very cheap operation.

From a readability or maintainability point of view I would prefer the first one. The trick to do lTemp = l*0 is not very intuitive to me.

Upvotes: 2

Related Questions