Filling rows with conditions in Pandas

Use Series.where with your mask and forward filling missing values:

df['B'] =  df['A'].where(df['A'].str.contains('NBN')).ffill()

#your solution should be changed
#df['B'] =  df['A'].where(df['A'].str.contains(r'^NBN \d|^NBN \d\.\d')).ffill()
print(df)

            A        B
0       NBN 3    NBN 3
1  test text1    NBN 3
2  test text2    NBN 3
3     NBN 3.1  NBN 3.1
4      test 1  NBN 3.1
5      test 2  NBN 3.1

Another solution with Series.str.extract and forward filling missing values:

df['B'] = df['A'].str.extract(r'^(NBN\s+\d\.\d|NBN\s+\d)', expand=False).ffill()