How to check the types contained in a list?

Question

I want to do the following without defining a function:

if isinstance(x,(list,tuple)) and every_element_isinstance(x,basestring):
   foobar

ie: implementing type checking

Is there a shorthand/builtin for this?

Answer 1

I think this is the best solution (if I understand the question)

if isinstance(x, (list, tuple)) and all(isinstance(i, basestring) for i in x):
    #do whatever

Answer 2

if isinstance(x, (list, tuple)) and all([isinstance(i, basestring) for i in x]):
    foobar

Surprisingly, a list comprehension with [ ... ] is faster here than one without, both with short and long lists:

short lists:

>>> timeit('isinstance(x, (list, tuple)) and all(isinstance(i, basestring) for i in x)', "x=['a','b','c']")
2.7594685942680144
>>> timeit('isinstance(x, (list, tuple)) and all(isinstance(i, basestring) for i in x)', "x=['a','b','c']")
2.8013695153947538
>>> timeit('isinstance(x, (list, tuple)) and all([isinstance(i, basestring) for i in x])', "x=['a','b','c']")
2.4351678506033068
>>> timeit('isinstance(x, (list, tuple)) and all([isinstance(i, basestring) for i in x])', "x=['a','b','c']")
2.4491469896721583

long lists:

>>> timeit('isinstance(x, (list, tuple)) and all(isinstance(i, basestring) for i in x)', "x=['a','b','c'] * 1000", number=1000)
1.3357901657891489
>>> timeit('isinstance(x, (list, tuple)) and all(isinstance(i, basestring) for i in x)', "x=['a','b','c'] * 1000", number=1000)
1.3305278872818462
>>> timeit('isinstance(x, (list, tuple)) and all([isinstance(i, basestring) for i in x])', "x=['a','b','c'] * 1000", number=1000)
1.2626525921055531
>>> timeit('isinstance(x, (list, tuple)) and all([isinstance(i, basestring) for i in x])', "x=['a','b','c'] * 1000", number=1000)
1.2881240045551863

Answer 3

There is not built-in to define generic types. But there are a lot of validation libraries, which can mimic this functionality.

Example using https://github.com/alecthomas/voluptuous:

>>> from voluptuous import Schema
>>> s_list = Schema([basestring]) # only strings in a list are allowed
>>> s_list("hello") 
...
voluptuous.InvalidList: expected a list
>>> s_list([123])
...
voluptuous.InvalidList: invalid list value @ data[0]
>>> s_list(["correct"])
["correct"] # returns the object, if validation was successful

A few days ago, support for tuple was added to this library:

>>> s_tuple = voluptuous.Schema((basestring, ))

Now combine the two to get your result:

>>> from voluptuous import any

# - this is now equivalent to your code
# - raises Exceptions on invalid input
>>> schema = Schema(any(s_list, s_tuple))

It's even slightly faster that the double-isinstance:

>>> from timeit import timeit

>>> timeit('(schema(i) for i in x)', "x=['a','b','c']")
0.679318904876709

>>> timeit("""
        (isinstance(x, (list, tuple)) 
         and all(isinstance(i, basestring)) for i in x)""", "x=['a','b','c']")
0.7801780700683594

Answer 4

With voluptuous '0.8.7' you can update answer from miku and skip the "tuple part":

>>> from voluptuous import Schema
>>> from timeit import timeit
>>> s_list = Schema([basestring]) # only strings in a list are allowed
>>> timeit('(s_list(i) for i in x)', "x=['a','b','c']")
0.503572940826416
>>> timeit("(isinstance(x, (list, tuple)) and all(isinstance(i, basestring)) for i in x)", "x=['a','b','c']")
0.5400209426879883