Index schema not validated in SeriesSchema

[Lucas-Cerqueira]

Similar to pandas.DataFrame, the Series also has an index and thought that doing like the code below I would be able to validate both the Series' values and index. However, it seems like the index schema is not validated.

input_schema = pa.SeriesSchema(
    pa.Float,
    nullable=False,
    coerce=True,
    checks=[pa.Check.in_range(0, 1)],
    index=pa.Index(
        pa.Int, coerce=False, allow_duplicates=False, checks=[pa.Check.ge(1)]
    ),
)

input_values = pd.Series({1.1: 0.1, "2": 0.2, 3: 0.3})
input_schema.validate(input_values)

On the other hand, if I validate the values using the Index schema separately, it throws the expected error. Is that the expected behavior?

index_schema = pa.Index(pa.Int, coerce=False, allow_duplicates=False, checks=[pa.Check.ge(1)])
index_schema.validate(input_values)
# pandera.errors.SchemaError: expected series 'None' to have type 'int64', got 'object'

[cosmicBboy]

hi @Lucas-Cerqueira, currently coerce at the schema-level will override any settings at the schema component level (in this case the pa.Index). This is why the index schema itself raises an error but not the series schema.

What's your use case for coercing the series values but not the index?

[Lucas-Cerqueira]

I see. It's because I'd like to allow the series value to be coerced to float automatically but the index I want to be sure it is an integer.