CODEWITHSUNDEEP
python-3.xpandas
Paul
Paul

Reputation: 326

Rename columns regex, keep name if no match

data = {'First_Column': [1,2,3], 'Second_Column': [1,2,3], 
        '\First\Mid\LAST.Ending': [1,2,3], 'First1\Mid1\LAST1.Ending': [1,2,3]}

df = pd.DataFrame(data)

     First_Column   Second_Column   \First\Mid\LAST.Ending  First1\Mid1\LAST1.Ending
0    1              1               1                       1
1    2              2               2                       2
2    3              3               3                       3

I want to rename the columns as follows:

     First_Column   Second_Column   LAST    LAST1
0    1              1               1        1
1    2              2               2        2
2    3              3               3        3

So i tried:

df.columns.str.extract(r'([^\\]+)\.Ending')   

    0
0   NaN
1   NaN
2   LAST
3   LAST1

and

col = df.columns.tolist()
for i in col[2:]:
    print(re.search(r'([^\\]+)\.Ending', i).group())

LAST.Ending
LAST1.Ending

First thing i noticed the different outputs of the regex argument. Why is that? Second, i prefer the version with extract. But how to keep the original name if there is no match?

THX

Upvotes: 2

Views: 113

Answers (2)

Umar.H
Umar.H

Reputation: 23099

another method is to use df.filter to find your target columns then a dict with rename after using your regex

s = df.filter(like='\\',axis=1).columns

s1 = s.str.extract(r'([^\\]+)\.Ending')[0].tolist()

df.rename(columns=dict(zip(s,s1)))

print(df)

   First_Column  Second_Column  LAST  LAST1
0             1              1     1      1
1             2              2     2      2
2             3              3     3      3

Upvotes: 2

Quang Hoang
Quang Hoang

Reputation: 150765

You can use np.where to fill where it doesn't match:

s = df.columns.str.extract(r'([^\\]+)\.Ending')[0]

df.columns = np.where(s.isna(), df.columns, s)
# equivalently
# df.columns = s.mask(s.isna(), df.columns.values)

Output:

   First_Column  Second_Column  LAST  LAST1
0             1              1     1      1
1             2              2     2      2
2             3              3     3      3

Upvotes: 3

Related Questions