Pandas findall by pattern but not duplicated ones

I need to have a list of all non- duplicated regex matches.

Consider below dataframe:

Letter      Actions
r1          a30,a30
r2          a30,a12-rf,a15,a15
r3          0
r4          a10,a93
r5          a13

I expect:

Letter      Actions
r1          ['a30']
r2          ['a30','a12','a15']
r3          0
r4          ['a10','a93']
r5          ['a13']

I have below but it returns all pattern matches, while I need don’t need the duplicated ones:

import pandas as pd

df = pd.DataFrame(
    [['r1', 'a30,a30'],
     ['r2', 'a30,a12-rf,a15,a15'],
     ['r3', '0'],
     ['r4', 'a10,a93'],
     ['r5', 'a13']],
    columns=['Letter', 'Actions'])

df['Action_list'] = df['Actions'].str.findall(r'([a]\d{2})')

>Solution :

You can use set to remove duplicates:

mask = df["Actions"].str.contains(r"a\d+", regex=True)

df["new_Actions"] = np.where(
    mask, df["Actions"].str.findall(r"a\d+").apply(set).apply(list), df["Actions"]
)
print(df)

Prints:

  Letter             Actions      new_Actions
0     r1             a30,a30            [a30]
1     r2  a30,a12-rf,a15,a15  [a30, a15, a12]
2     r3                   0                0
3     r4             a10,a93       [a93, a10]
4     r5                 a13            [a13]