So, I have a Pandas dataframe with food and cuisine some people like. I have to break those into columns, so each food or cuisine should become a column. Each food/cuisine comes afer a comma, but if i break my string only by commas, I'll lost the content inside the parenthesis, which should be there, close to the dish. I think I should use '),' as a separator, right? But I don't know how to do that. This is my DF:

>>> PD_FOODS
  USER_ID  | FOODS_I_LIKE                                                      |
_______________________________________________________________________________
0   100    |   Pizza(without garlic, tomatos and onion),pasta                  |
1   101    |   Seafood,veggies                                                 |
2   102    |   Indian food (no pepper, no curry),mexican food(no pepper)       |
3   103    |   Texmex, african food, japanese food,italian food                |
4   104    |   Seafood(no shrimps, no lobster),italian food(no gluten, no milk)|

Is it possible to get a result like this bellow?

>>> PD_FOODS
  USER_ID  |   FOODS_I_LIKE_1                          |  FOODS_I_LIKE_2       |
_______________________________________________________________________________
0   100    |   Pizza(without garlic, tomatos and onion)|  pasta                |

Thank you!

CodePudding user response：

You could use a regex with a negative lookahead:

(df['FOODS_I_LIKE'].str.split(',\s*(?![^()]*\))', expand=True)
   .rename(columns=lambda x: int(x) 1)
   .add_prefix('FOODS_I_LIKE_')
)

output:

                             FOODS_I_LIKE_1                    FOODS_I_LIKE_2 FOODS_I_LIKE_3 FOODS_I_LIKE_4
0  Pizza(without garlic, tomatos and onion)                             pasta           None           None
1                                   Seafood                           veggies           None           None
2         Indian food (no pepper, no curry)           mexican food(no pepper)           None           None
3                                    Texmex                      african food  japanese food   italian food
4           Seafood(no shrimps, no lobster)  italian food(no gluten, no milk)           None           None

You can test the regex here

NB. this won't work on nested parenthesis, you would need to use a parser

CodePudding user response：

Try this:

df=pd.DataFrame({"User_ID":[1000,1001,1002,1003,1004],
               "FOODS_I_LIKE":['Pizza(without garlic, tomatos and onion),pasta',
                              'Seafood,veggies',
                              'Indian food (no pepper, no curry),mexican food(no pepper)',
                              'Texmex, african food, japanese food,italian food',
                              'Seafood(no shrimps, no lobster),italian food(no gluten, no milk)']})

def my_func(my_string, item_num):
    try:
        if ')' in my_string:
            if item_num == 0:
                return my_string.split('),')[item_num] ')'
            else:
                return my_string.split('),')[item_num]
        else:
            return my_string.split(',')[item_num]
    except IndexError:
        return np.nan
    
for k in range(0,4):
    K=str(k 1)
    df[f'FOODS_I_LIKE_{K}']=df.FOODS_I_LIKE.apply(lambda x: my_func(x, k))

df.drop(columns='FOODS_I_LIKE')

Output: