Natural sorting for python.

• Source Code: https://github.com/SethMMorton/natsort
• Downloads: https://pypi.python.org/pypi/natsort
• Documentation: https://pythonhosted.org/natsort

Please see Moving from older Natsort versions to see if this update requires you to modify your natsort calls in your code (99% of users will not).

When you try to sort a list of strings that contain numbers, the normal python sort algorithm sorts lexicographically, so you might not get the results that you expect:

>>> a = ['a2', 'a9', 'a1', 'a4', 'a10']
>>> sorted(a)
['a1', 'a10', 'a2', 'a4', 'a9']

Notice that it has the order ('1', '10', '2') - this is because the list is being sorted in lexicographical order, which sorts numbers like you would letters (i.e. 'b', 'ba', 'c').

natsort provides a function natsorted that helps sort lists "naturally", either as real numbers (i.e. signed/unsigned floats or ints), or as versions. Using natsorted is simple:

>>> from natsort import natsorted
>>> a = ['a2', 'a9', 'a1', 'a4', 'a10']
>>> natsorted(a)
['a1', 'a2', 'a4', 'a9', 'a10']

natsorted identifies real numbers anywhere in a string and sorts them naturally.

Sorting versions is handled properly by default (as of natsort version >= 4.0.0):

>>> a = ['version-1.9', 'version-2.0', 'version-1.11', 'version-1.10']
>>> natsorted(a)
['version-1.9', 'version-1.10', 'version-1.11', 'version-2.0']

If you need to sort release candidates, please see this useful hack .

You can also perform locale-aware sorting (or "human sorting"), where the non-numeric characters are ordered based on their meaning, not on their ordinal value; this can be achieved with the humansorted function:

>>> a = ['Apple', 'Banana', 'apple', 'banana']
>>> natsorted(a)
['Apple', 'Banana', 'apple', 'banana']
>>> import locale
>>> locale.setlocale(locale.LC_ALL, 'en_US.UTF-8')
'en_US.UTF-8'
>>> from natsort import humansorted
>>> humansorted(a)
['apple', 'Apple', 'banana', 'Banana']

You may find you need to explicitly set the locale to get this to work (as shown in the example). Please see the following caveat and the Optional Dependencies section below before using the humansorted function, especially if you are on a BSD-based system (like Mac OS X).

You can sort signed floats (i.e. real numbers) using the realsorted; this is useful in scientific data analysis. This was the default behavior of natsorted for natsort version < 4.0.0:

>>> from natsort import realsorted
>>> a = ['num5.10', 'num-3', 'num5.3', 'num2']
>>> natsorted(a)
['num2', 'num5.3', 'num5.10', 'num-3']
>>> realsorted(a)
['num-3', 'num2', 'num5.10', 'num5.3']

You can mix and match int, float, and str (or unicode) types when you sort:

>>> a = ['4.5', 6, 2.0, '5', 'a']
>>> natsorted(a)
[2.0, '4.5', '5', 6, 'a']
>>> # On Python 2, sorted(a) would return [2.0, 6, '4.5', '5', 'a']
>>> # On Python 3, sorted(a) would raise an "unorderable types" TypeError

natsort does not officially support the bytes type on Python 3, but convenience functions are provided that help you decode to str first:

>>> from natsort import as_utf8
>>> a = [b'a', 14.0, 'b']
>>> # On Python 2, natsorted(a) would would work as expected.
>>> # On Python 3, natsorted(a) would raise a TypeError (bytes() < str())
>>> natsorted(a, key=as_utf8) == [14.0, b'a', 'b']
True
>>> a = [b'a56', b'a5', b'a6', b'a40']
>>> # On Python 2, natsorted(a) would would work as expected.
>>> # On Python 3, natsorted(a) would return the same results as sorted(a)
>>> natsorted(a, key=as_utf8) == [b'a5', b'a6', b'a40', b'a56']
True

The natsort algorithm does other fancy things like

• recursively descend into lists of lists
• control the case-sensitivity
• sort file paths correctly
• allow custom sorting keys
• exposes a natsort_key generator to pass to list.sort

Please see the package documentation for more details, including examples and recipes.

natsort comes with a shell script called natsort, or can also be called from the command line with python -m natsort.

natsort requires Python version 2.7 or greater or Python 3.3 or greater. Python 2.6 and 3.2 are no longer officially supported (no unit tests are performed) but it should work.

The most efficient sorting can occur if you install the fastnumbers package (it helps with the string to number conversions.) natsort will still run (efficiently) without the package, but if you need to squeeze out that extra juice it is recommended you include this as a dependency. natsort will not require (or check) that fastnumbers is installed at installation.

On BSD-based systems (this includes Mac OS X), the underlying locale library can be buggy (please see https://bugs.python.org/issue23195); locale is used for the ns.LOCALE option and humansorted function.. To remedy this, one can

1. Use "*.ISO8859-1" locale (i.e. 'en_US.ISO8859-1') rather than "*.UTF-8" locale. These locales do not suffer from as many problems as "UTF-8" and thus should give expected results.
2. Use PyICU. If PyICU is installed, natsort will use it under the hood; this will give more reliable cross-platform results in the long run. natsort will not require (or check) that PyICU is installed at installation. Please visit SethMMorton#21 for more details and how to install on Mac OS X. Please note that using PyICU is the only way to guarantee correct results for all input on BSD-based systems, since every other suggestion is a workaround.
3. Do nothing. As of natsort version 4.0.0, natsort is configured to compensate for a broken locale library in terms of case-handling; if you do not need to be able to properly handle non-ASCII characters then this may be the best option for you.

Note that the above solutions should not be required for Windows or Linux since in Linux-based systems and Windows systems locale should work just fine.

• The default sorting algorithm for natsort has changed in version 4.0.0 from signed floats (with exponents) to unsigned integers. The motivation for this change is that it will cause natsort to return results that pass the "least astonishment" test for the most common use case, which is sorting version numbers. If you relied on the default behavior to be signed floats, add alg=ns.F | ns.S to your natsort calls or switch to the new realsorted function which behaves identically to the older natsorted with default values. For 99% of users this change will not effect their code... it is only expected that this will effect users using natsort for science and engineering. This will also affect the default behavior of the natsort shell script.
• In natsort version 4.0.0, the number_type, signed, exp, as_path, and py3_safe options have be removed from the (documented) API in favor of the alg option and ns enum.
• In natsort version 4.0.0, the natsort_key function has been removed from the public API.

Seth M. Morton

These are the last three entries of the changelog. See the package documentation for the complete changelog.

• Made default behavior of 'natsort' search for unsigned ints, rather than signed floats. This is a backwards-incompatible change but in 99% of use cases it should not require any end-user changes.
• Improved handling of locale-aware sorting on systems where the underlying locale library is broken.
• Greatly improved all unit tests by adding the hypothesis library.

• Added 'UNGROUPLETTERS' algorithm to get the case-grouping behavior of an ordinal sort when using 'LOCALE'.
• Added convenience functions 'decoder', 'as_ascii', and 'as_utf8' for dealing with bytes types.

• Added 'realsorted' and 'index_realsorted' functions for forward-compatibility with >= 4.0.0.
• Made explanation of when to use "TYPESAFE" more clear in the docs.