I have the below example DataFrame

import pandas as pd
import pandera as pa

schema = pa.DataFrameSchema({
    'Invoice Type': pa.Column(str, pa.Check(lambda s: s.isin(['Invoice', 'Credit']))),
    'Quantity': pa.Column(
        int, 
        checks=[
            pa.Check(
                lambda df: df[True] < 0,
                groupby=lambda df: (
                    df.assign(is_credit=lambda d: d["Invoice Type"] == 'Credit').groupby(['is_credit'])
                ),
            name = 'negative credits quantities'),
            pa.Check(
                lambda df: df[True] >= 0,
                groupby=lambda df: (
                    df.assign(is_invoice=lambda d: d['Invoice Type'] == 'Invoice').groupby(['is_invoice'])
                ),
            name = 'postitive invoices quantities'),
        ]),
    'Customer Number': pa.Column(str)
})

df = pandas.DataFrame(
    {'Quantity': [1, 2, -3],
    'Invoice Type': ['Invoice', 'Invoice', 'Credit'],
    'Customer Number': ['ABC', 'ABC', 'DEF']}
)

Which correctly identifies all credits that are negative and invoices that are positive. Where I’m running into trouble is when I don’t have any values for that particular series. I would commonly have situations where there are no credits, example below:

df = pandas.DataFrame(
    {'Quantity': [1, 2, 3],
    'Invoice Type': ['Invoice', 'Invoice', 'Invoice'],
    'Customer Number': ['ABC', 'ABC', 'DEF']}
)

which means that the DataFrame with all invoices fails with a key error. How can I ensure that this passes validation if only one invoice type is in the DataFrame