熊猫：从另一列修改数据框中的值

Question

While splitting data into columns, there was some glitch, due to which I have got some noisy data.

    site          code
    ---           ---
0   apple_123     45
1   apple_456     xy_33
2   facebook_123  24
3   google_123    NaN
4   google_123    pq_51

I need to clean the data, such that I get the following result:

    site            code
    ---             ---
0   apple_123       45
1   apple_456_xy    33
2   facebook_123    24
3   google_123      NaN
4   google_123_pq   51

I have been able to obtain the rows that need to be modified, but am unable to progress further:

import numpy as np
import pandas as pd

site = ['apple_123','apple_456','facebook_123','google_123','google_123']
code = [45,'xy_33',24,np.nan,'pq_51']
df = pd.DataFrame(list(zip(site,code)), columns=['site','code'])

df[(~df.code.astype(str).str.isdigit())&(~df.code.isna())] 

Answer 1

Use Series.str.extract for get non numeric and numeric values to helper DataFrame and then processing each column separately - remove _ by Series.str.strip , add from right side by Series.radd and convert missing values to emty string, last add to code column, for second use Series.fillna for replace not mached values from 1 column to original:

df1 = df.code.str.extract('(\D+)(\d+)')

df['site'] += df1[0].str.strip('_').radd('_').fillna('')
df['code'] = df1[1].fillna(df['code'])
print (df)
            site code
0      apple_123   45
1   apple_456_xy   33
2   facebook_123   24
3     google_123  NaN
4  google_123_pq   51