CODEWITHSUNDEEP
pythonregexpandas
MEhsan
MEhsan

Reputation: 2324

How to match regex pattern and replace it with a matching group using Pandas?

I have the following Pandas Series:

SC_S193_M7.CONTROLDAY10.EPI.P1_Stem
SC_S194_M7.CONTROLDAY10.EPI.P1_Goblet
SC_S102_M1.CONTROLDAY3.EPI2_Enterocyte
SC_S106_M1.CONTROLDAY3.EPI2_Goblet

I want to use regex to extract the string after the last underscore in each row of this series. I was able to come up with regex that match with the last string but note sure how to implement it in a pandas series method.

The regex I used to match the pattern and replace with the first matching group \1:

SC_S\d{3}_M\d\.CONTROLDAY\d{1,2}\.EPI\d?(?:\.P\d_|_)

I tried using .replace() as follows but that did not work out:

.replace('SC_S\d{3}_M\d\.CONTROLDAY\d{1,2}\.EPI\d?(?:\.P\d_|_)(\w+)')

Any idea how to use Pandas series method to extract the last string before the underscore or find the matching pattern and replace it with the first group?

Upvotes: 1

Views: 690

Answers (2)

Davide Fiocco
Davide Fiocco

Reputation: 5914

Another variant (assuming that s is your series) that should work is something along the lines of

s.apply(lambda r : re.sub('.*_([^_]*)$', '\\1', r))

Upvotes: 2

MaxU - stand with Ukraine
MaxU - stand with Ukraine

Reputation: 210842

I think you can split it instead of using RegEx:

In [170]: s
Out[170]:
0       SC_S193_M7.CONTROLDAY10.EPI.P1_Stem
1     SC_S194_M7.CONTROLDAY10.EPI.P1_Goblet
2    SC_S102_M1.CONTROLDAY3.EPI2_Enterocyte
3        SC_S106_M1.CONTROLDAY3.EPI2_Goblet
Name: 0, dtype: object

In [171]: s.str.split('_').str[-1]
Out[171]:
0          Stem
1        Goblet
2    Enterocyte
3        Goblet
Name: 0, dtype: object

or better using rsplit(..., n=1):

In [174]: s.str.rsplit('_', n=1).str[-1]
Out[174]:
0          Stem
1        Goblet
2    Enterocyte
3        Goblet
Name: 0, dtype: object

alternatively you can use .str.extract():

In [177]: s.str.extract(r'.*_([^_]*)$', expand=False)
Out[177]:
0          Stem
1        Goblet
2    Enterocyte
3        Goblet
Name: 0, dtype: object

Upvotes: 4

Related Questions