Pandas: findall unique values only-CodePudding

Goal is to extract unique REGEX matches from string. Issue is eliminating duplicate values in the resulting array.

Consider: Given the following dataframe -

df_inventory = pd.DataFrame([[1000, 'A widget for PEX0043 and PEX7657 in Dept. X7842', 'DPT7843','Part PEX7657'], 
                             [1001, 'A widget for PEX7654 in Depts. X7843, X7842', 'DPT7842','Part PEX7658'],
                             [1002, 'A widget for PEX7655 PEX7657 in Dept. X7844, X7844, X7842', 'DPT7845','Part PEX7659'],
                             [1003, 'A widget for PEX7656 PEX7658 in Dept. X7845, X7851, X7850, X7842', 'DPT7844','Part PEX7660'],
                             [1004, 'A widget for PEX7657 in Dept. X7846', 'DPT7847','Part PEX7661'],
                             [1005, 'A widget for PEX7658 in Dept. X7847, X7842', 'DPT7846','Part PEX7655'],
                             [1006, 'A widget for PEX7659 in Dept. X7848, X7842, X7843', 'DPT7849', 'Part PEX7654'],
                             [1007, 'A widget for PEX7660 in Dept. X7849, X7844', 'DPT7848', 'Part PEX0043'],
                             [1008, 'A widget for PEX7661 in Dept. X7850', 'DPT7851', 'Part PEX7656'],
                             [1009, 'A widget for PEX7662 in Dept. X7851,X7842,X7843,X7843,X7845,X7846,X7847', 'DPT7850', 'Part PEX7660']],
                             columns=['Product ID', 'Description', 'Location','Name'])

We extract using something like:

X_pat  = re.compile(r'(?<!PE)(X[0-9]{4})')
df_inventory['X Numbers'] = df_inventory['Description'].str.findall(X_pat).str.join(", ")

When you run this code you find duplicate values in record/row 9:

9   1009    A widget for PEX7662 in Dept. X7851,X7842,X784...   DPT7850     Part PEX7660    X7851, X7842, X7843, X7843, X7845, X7846, X7847

I don't think unique is what I need here. Guessing a lamda function.

CodePudding user response：

you can use str.extractall then a lambda on the groups that are returned to get your unique output.

df_inventory['x_numbers'] = df_inventory["Description"]\
         .str.extractall("(?<!PE)(X[0-9]{4})")\
         .groupby(level=0).agg(
           lambda x: ",".join(x.unique())
     )

print(df_inventory.drop('Description',axis=1)) # just for a nicer print statement.

   Product ID Location          Name                            x_numbers
0        1000  DPT7843  Part PEX7657                                X7842
1        1001  DPT7842  Part PEX7658                          X7843,X7842
2        1002  DPT7845  Part PEX7659                          X7844,X7842
3        1003  DPT7844  Part PEX7660              X7845,X7851,X7850,X7842
4        1004  DPT7847  Part PEX7661                                X7846
5        1005  DPT7846  Part PEX7655                          X7847,X7842
6        1006  DPT7849  Part PEX7654                    X7848,X7842,X7843
7        1007  DPT7848  Part PEX0043                          X7849,X7844
8        1008  DPT7851  Part PEX7656                                X7850
9        1009  DPT7850  Part PEX7660  X7851,X7842,X7843,X7845,X7846,X7847