Plotting a stacked Bar Chart in pandas for multiple x-axis attributes

Question

l have four bars in my histogram which represent the frequency of letter, digit, special characters and alphnumeric in my file.

l did the following to get the frequency of each family :

def  group_data():
    def find_group(val):
        val = str(val)
        val = val.lower()

        if val.isalpha():
            return 'alphabet'
        elif val.isdigit():
            return 'digit'
        elif val.isalnum and any(c.isalpha() for c in val) and any(c.isdigit() for c in val):

            return 'alphanum'
        else:
            return 'spec-char'


    df = pd.read_csv('file1.csv', sep=',')
    df = df.astype(str)
    df.manual_raw_value = df.manual_raw_value.str.lower()
    df.manual_raw_value.apply(find_group)
    df.manual_raw_value.apply(find_group).value_counts().plot(kind='bar')

and l got the following for the first file and second file

Now l would like to get a stacked bar so l did the following

normal = df.manual_raw_value.apply(find_group).value_counts()
aug=df2.manual_raw_value.apply(find_group).value_counts()
sub_df = pd.concat([normal, aug], keys=["alphabet", "digit","spec-char","alpha-num"]).unstack()
sub_df.plot(kind='bar', stacked=True, rot=1)
plt.show()

However, l got a wrong plot

EDIT-1 I'm supposed to get plot like the following: red bars for

aug=df2.manual_raw_value.apply(find_group).value_counts()

and blue bars for

normal = df.manual_raw_value.apply(find_group).value_counts()