python natsort sort strings recursively

Question

I find that with natsort.natsorted the sorting order changes part-way through a string:

In [31]: import natsort as ns
In [32]: ns.natsorted(["01-08", "02-07", "01-06", "02-09"])
Out[32]: ['01-08', '01-06', '02-09', '02-07']

In this case, the behaviour I want is:

In [33]: sorted(["01-08", "02-07", "01-06", "02-09"])
Out[33]: ['01-06', '01-08', '02-07', '02-09']

@SethMMorton, is there something I'm missing in the documentation? — user2561747
ns.versorted seems to have the behaviour I'm looking for, but I wish to use it on paths, and other strings, not versions. Is that recommended use? — user2561747
Yes, you can use versorted with ns.PATH without any issue at all. — SethMMorton
Or, as mentioned in the accepted answer, you can use ns.INT | ns.UNSIGNED | ns.PATH, or ns.I | ns.U | ns.P. — SethMMorton
You no longer need to deal with the shenanigans of modifying the algorithm when using natsort anymore for all but special cases. Simply using ns.natsorted(["01-08", "02-07", "01-06", "02-09"]) works as expected now. — SethMMorton

cge cge · Accepted Answer · 2015-04-09T20:59:48

Try this:

ns.natsorted(["01-08", "02-07", "01-06", "02-09"], alg=ns.ns.INT | ns.ns.UNSIGNED)

The problem is that natsorted is interpreting your strings incorrectly. This manually sets the algorithm to look for unsigned ints. Otherwise, it searches for signed ints, and that "-" causes problems (if you interpret "-08", for example, as -8, then the sorting makes sense).

This is actually equivalent to versorted, which is just a shortcut for this algorithm, but I think it's better to explicitly write what you're doing, especially as versorted could change to be more applicable to versions in the future.

python natsort sort strings recursively

1 Answers