Applying within group rankings for Pandas DataFrame

The problem statement has people choosing 3 items and ranking them 1 to 3. My Pandas DataFrame contains one row for each item selected, where the first column is the person's name and the second column is the ranking the person has given to the selection.

Something like:

df = pd.DataFrame.from_dict({
  'Name':['Alice','Alice','Alice','Bob','Bob','Bob','Charlie','Charlie','Charlie'], 
  'Item':[...], 
  'Rank':[1,2,3,3,1,2,None,None,None]
})

The issue is some people did not assign rankings to their items, for example Charlie in the DataFrame above. For these people, I want to fill in their rankings with a valid 'random' ranking. AKA, just give each of their items a unique value from 1 to 3. Also, some people only selected 2 items and also forgot to rank their items, so I need to be able a variable amount of items selected.

I was attempting to do a cumsum, where I first filled in each null value with 1, and then run the cumsum within each group of a groupby on the names.

Something like (although I think this is far from correct):

df.groupby('Name').cumsum('Rank')

Also, I understand this may be easy by iterating over the rows. However, this being Pandas I am looking for a more optimal solution.

You can use groupby_cumcount:

df['Rank'] = df['Rank'].fillna(df.groupby('Name').cumcount().add(1))
print(df)

# Output
      Name  Rank
0    Alice   1.0
1    Alice   2.0
2    Alice   3.0
3      Bob   3.0
4      Bob   1.0
5      Bob   2.0
6  Charlie   1.0
7  Charlie   2.0
8  Charlie   3.0

To use random ranking, use sample before:

df['Rank'] = df['Rank'].fillna(df.sample(frac=1).groupby('Name').cumcount().add(1))
print(df)

# Output
      Name  Rank
0    Alice   1.0
1    Alice   2.0
2    Alice   3.0
3      Bob   3.0
4      Bob   1.0
5      Bob   2.0
6  Charlie   3.0
7  Charlie   1.0
8  Charlie   2.0