How can I add a column that depends on another columns value, but also involves other rows?

As you can see, the TransactionNo's are not unique, in other words not every entry is its own order. So I want to make a column that have a list of all items for a particular TransactionNo when Coffee was bought with that TransactionNo.

For example, on row 7(TransactionNo 5) you see coffee is bought. So I want a column that called 'extras' that puts the other items bought with TransactionNo 5 on it. So under the column 'extras', on row 7, you would see a list ['Pastry', 'Bread']. I've tried using np.where but I cannot figure this out.

TransactionNo	Items
0	1	Bread
1	2	Scandinavian
2	2	Scandinavian
3	3	Hot chocolate
4	3	Jam
5	3	Cookies
6	4	Muffin
7	5	Coffee
8	5	Pastry
9	5	Bread

I tried df['extras'] = np.where(df['Items'] == 'Coffee', x, y) but couldn't figure out what to put for x or y.

You can group the items except Coffee into list then assign the list to the Coffee

m = df['Items'].eq('Coffee')

dct = df.loc[~m, 'Items'].groupby(df['TransactionNo']).agg(list)

df['extras'] = df['TransactionNo'].map(dct).where(m, '')

$ print(dct)

TransactionNo
1                          [Bread]
2     [Scandinavian, Scandinavian]
3    [Hot chocolate, Jam, Cookies]
4                         [Muffin]
5                  [Pastry, Bread]
Name: Items, dtype: object

$ print(df)

   TransactionNo          Items           extras
0              1          Bread
1              2   Scandinavian
2              2   Scandinavian
3              3  Hot chocolate
4              3            Jam
5              3        Cookies
6              4         Muffin
7              5         Coffee  [Pastry, Bread]
8              5         Pastry
9              5          Bread