Python: How to separate list based on numerical values and strings from a given list

Question

Consider I have a list like

listed = [1, 't', 'ret', 89, 95, 'man65', 67, 'rr']

Now I need to write a script which will be like 2 lists; one stringList and numberList where,

stringList = ['t', 'ret', 'man65', 'rr']
numberList = [1, 89, 95, 67]

@wahwahwah: I was going to say that, but then the answers started pouring in. So it's probably a little late now to ask for an MCVE. :) — PM 2Ring, May 14 '15 at 13:50

Uyghur Lives Matter · Answer 1 · 2015-05-14T14:42:45.707

The standard way to check what kind of object you have is by using the isinstance(object, classinfo) function. Using isinstance(...) is preferred over type(object) because type(...) returns the exact type of an object, and does not account for sub-classes when checking types (this is significant in more complex scenarios).

You can check if you have a number (int, float, etc.) by comparing each class individually or with numbers.Number:

# Check for any type of number in Python 2 or 3.
import numbers
isinstance(value, numbers.Number)

# Check for floating-point number in Python 2 or 3.
isinstance(value, float)

# Check for integers in Python 3.
isinstance(value, int)

# Check for integers in Python 2.
isinstance(value, (int, long))

You can check if you have a string (str, bytes, etc.) by comparing with the individual classes which vary depending on whether you're using Python 2 or 3:

# Check for unicode string in Python 3.
isinstance(value, str)

# Check for binary string in Python 3.
isinstance(value, bytes)

# Check for any type of string in Python 3.
isinstance(value, (str, bytes))

# Check for unicode string in Python 2.
isinstance(value, unicode)

# Check for binary string in Python 2.
isinstance(value, str)

# Check for any type of string in Python 2.
isinstance(value, basestring)

So, putting that all together you'd have:

import numbers

stringList = []
numberList = []
for value in mixedList:
    if isinstance(value, str):
        stringList.append(value)
    elif isinstance(value, numbers.Number):
        numberList.append(value)

note in Python 2 you should use `basestring` instead of `str` ...and in Py 3 I think you should use a tuple of `(str, bytes)` http://www.rfk.id.au/blog/entry/preparing-pyenchant-for-python-3/ — Anentropic, May 14 '15 at 13:32

Stefan Pochmann · Answer 2 · 2015-05-14T14:59:08.517

2

numbers, strings = both = [], []
for x in listed:
    both[isinstance(x, str)].append(x)

An alternative as an answer to Ajay's challenge in the comments to do it with just one list comprehension (note: this is a bad solution):

strings = [listed.pop(i) for i, x in list(enumerate(listed))[::-1] if isinstance(x, str)][::-1]
numbers = listed

Another:

numbers, strings = [[x for x in listed if isinstance(x, t)] for t in (int, str)]

Yet another, with the advantage of reading listed only once (inspired by PM 2Ring):

numbers, strings = [[x for x in l if isinstance(x, t)]
                    for l, t in zip(itertools.tee(listed), (int, str))]

edited May 14 '15 at 14:59

answered May 14 '15 at 13:34

Stefan Pochmann

27,593
8
44
107

Cute, but potentially confusing. :) – PM 2Ring May 14 '15 at 13:37
Hey @Stefan is it possible with list comprehension.I mean iterating only once – Ajay May 14 '15 at 13:57
@Ajay Depends on what you mean. You could do `[both[isinstance(x, str)].append(x) for x in listed]`, of course :-P – Stefan Pochmann May 14 '15 at 14:02
but the immediate output is None,None,None...although numbers and strings will be populated – Ajay May 14 '15 at 14:05
@Ajay: Exactly. Note the smilie on Stefan's comment: that code is intentionally silly. – PM 2Ring May 14 '15 at 14:06
@Ajay What 2Ring said. I gave a technically correct solution, but it's silly. I don't think it's possible what you have in mind, since one list comprehension gives you one list, and we want two. – Stefan Pochmann May 14 '15 at 14:10
[[i][j]] something like that – Ajay May 14 '15 at 14:12
@Ajay I thought about something like that, but didn't get far. I added mine to my answer now, have a look. – Stefan Pochmann May 14 '15 at 14:20
@PM2Ring It's less silly than the one I added to my answer now, though. – Stefan Pochmann May 14 '15 at 14:20
i thought of pop too ...but dropped that idea – Ajay May 14 '15 at 14:23
plus 1 for the second answer...but you come up with complicated solutions which need some serious thinking – Ajay May 14 '15 at 14:26
@Ajay I don't find my first solution complicated. Also, thinking is good :-) – Stefan Pochmann May 14 '15 at 14:28
@StefanPochmann: No, the first solution is fine, but it'd be a little more readable IMHO if its first line was split into two. But if you want to see an alternative silly list comp, pop over to sopython chat. :) – PM 2Ring May 14 '15 at 14:40
1

@Ajay I added two more versions, using the [[][]] idea. The latter only goes through `listed` once. – Stefan Pochmann May 14 '15 at 15:04

Tom · Answer 3 · 2015-05-14T13:38:26.847

1

You can use the built in type function to test each element in the list, and append it to a list based in its type.

numberList = []
stringList = []

for x in listed:
    if type(x) == int: numberList.append(x)
    else: stringList.append(x)

edited May 14 '15 at 13:38

answered May 14 '15 at 13:31

Tom

918
6
19

1

Using `isinstance()` is perferred over checking with `type()` because `isinstance()` can handle subclasses while `type()` checks the exact type. – Uyghur Lives Matter May 14 '15 at 13:33
@cpburnz True but in the OP's example he only has ints and strings. But I will note isinstance for the future, as I did not know about isinstance. – Tom May 14 '15 at 13:38
For further info on `isinstance()` vs type`()` please see http://stackoverflow.com/questions/1549801/differences-between-isinstance-and-type-in-python – PM 2Ring May 14 '15 at 13:47
@PM2Ring Thanks i'll have a look – Tom May 14 '15 at 13:52

Ajay · Answer 4 · 2015-05-14T14:11:38.643

1

In [16]: sl=[i for i in a if isinstance(i,str)]

In [17]: nl=[i for i in a if isinstance(i,int)]

In [18]: sl
Out[18]: ['t', 'ret', 'man65', 'rr']

In [19]: nl
Out[19]: [1, 89, 95, 67]

edited May 14 '15 at 14:11

answered May 14 '15 at 13:40

Ajay

5,267
2
23
30

One can use filter as well - but this one is faster and hence a preferred one – gabhijit May 14 '15 at 14:02

score 1 · Answer 5 · answered May 14 '15 at 14:17

Generic solution: grouping by exact object type.

multityped_objects = [1, 't', 'ret', 89, 95, 'man65', 67, 'rr', 12.8]
grouped = {}
for o in multityped_objects:
    try:
        grouped[type(o)].append(o)
    except KeyError:
        grouped[type(o)] = [o]

assert grouped[str] == ['t', 'ret', 'man65', 'rr']
assert grouped[int] == [1, 89, 95, 67]
assert grouped[float] == [12.8]

Python: How to separate list based on numerical values and strings from a given list

5 Answers5

Linked