Sort a column containing strings representing software versions in Pandas

Question

I have a dataframe where one of the columns is a string with a software name and different versions. When trying to sort by this column the ordering is not respecting the versioning.

The column to sort has a format of this kind

>>> import pandas as pd
>>> df = pd.DataFrame({'versions': ['cd-2.8.10', 'cd-3.10.3', 'cd-3.3.1', 'cd-3.10.10', 'cd-3.12.0', 'ab-5.2.1', 'cd-3.1.3', 'cd-3.5.2', 'ab-3.0.2', 'cd-3.10.1', 'cd-3.20.1', 'cd-3.11.4']})
>>> df
      versions
0    cd-2.8.10
1    cd-3.10.3
2     cd-3.3.1
3   cd-3.10.10
4    cd-3.12.0
5     ab-5.2.1
6     cd-3.1.3
7     cd-3.5.2
8     ab-3.0.2
9    cd-3.10.1
10   cd-3.20.1
11   cd-3.11.4

when using sort_values() the string part with characters before the dash is perfectly sorted alphabetically, but for a given software the version number sorting is wrong by taking 3.10.1 as smaller to 3.3.1, or 3.10.10 as smaller than 3.10.3

>>> df.sort_values('versions')
      versions
8     ab-3.0.2
5     ab-5.2.1
0    cd-2.8.10
6     cd-3.1.3
9    cd-3.10.1
3   cd-3.10.10
1    cd-3.10.3
11   cd-3.11.4
4    cd-3.12.0
10   cd-3.20.1
2     cd-3.3.1
7     cd-3.5.2

I would like to get the correct version ordering as

      versions
8     ab-3.0.2
5     ab-5.2.1
0    cd-2.8.10
6     cd-3.1.3
2     cd-3.3.1
7     cd-3.5.2
9    cd-3.10.1
1    cd-3.10.3
3   cd-3.10.10
11   cd-3.11.4
4    cd-3.12.0
10   cd-3.20.1

Would it be possible here to use a method similar to stackoverflow.com/questions/43787091/…? — Green Cloak Guy
– Green Cloak Guy, Commented May 21, 2019 at 17:03

cs95 · Accepted Answer · 2019-05-21 17:04:03Z

1

This is a complicated issue because pandas does not directly support natural sorting. Thankfully, using the natsort module, this should be easy and also handle most version formats.

from natsort import natsorted
df.iloc[natsorted(df.index, key=lambda x: df.loc[x, 'versions'])]

      versions
8     ab-3.0.2
5     ab-5.2.1
0    cd-2.8.10
6     cd-3.1.3
2     cd-3.3.1
7     cd-3.5.2
9    cd-3.10.1
1    cd-3.10.3
3   cd-3.10.10
11   cd-3.11.4
4    cd-3.12.0
10   cd-3.20.1

Here's another way of sorting this data (it is slightly faster because we avoid the lambda),

d = df.versions.to_dict()
df.iloc[natsorted(d, key=d.get)]

      versions
8     ab-3.0.2
5     ab-5.2.1
0    cd-2.8.10
6     cd-3.1.3
2     cd-3.3.1
7     cd-3.5.2
9    cd-3.10.1
1    cd-3.10.3
3   cd-3.10.10
11   cd-3.11.4
4    cd-3.12.0
10   cd-3.20.1

answered May 21, 2019 at 17:04

cs95

406k106 gold badges744 silver badges797 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Sort a column containing strings representing software versions in Pandas

1 Answer 1

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related