I want to replace special symbol to another text in python with pandas-CodePudding

I want to change the characters at once, but it doesn't change when I use the special symbol like [ or ( or : or - . What should I do?

my sample datatable is below

df


   col1
0  ( red ) apple
1  [ 20220901 ] autumn
2  - gotohome
3  sample : salt bread

and I want to get this below

df

   col1
0  red
   apple
1  20220901 
   autumn
2  gotohome
3  sample
   salt bread

my trial is below but it's not working.

change_word = {
    '( red )'        : 'red\n', 
    '[ 20220901 ]'   : '20220901\n',
    '- '             : '', 
    ':'              : '\n'
}

regex = r'\b(?:'   r'|'.join(change_word.keys())   r')\b'
df["col1"] = df["col1"].str.replace(regex, lambda m: change_word[m.group()], regex=True)

CodePudding user response：

You can maybe use something like:

import re

badchars = '()[]\t-:'
df2  = (df['col1']
        .str.strip(badchars ' ') # strip unwanted chars at extremities
        .str.split(f'\s*[{re.escape(badchars)}] \s*') # split on badchars   spaces
        .explode().to_frame() # explode as new rows
       )

Output:

         col1
0         red
0       apple
1    20220901
1      autumn
2    gotohome
3      sample
3  salt bread