Extending the SpanRuler to support more than just label and id

[NixBiks]

I wonder if there is an easy way to extend the SpanRuler to allow for more fields than just label and id.

I have a use case where I want to extract amounts into structured data. E.g. a phrase like 5,123 mEUR can be matched using a pattern but then I'd like to apply some function to it as well where I provide a regex showing where the number is, where the multiplier is and where the currency is.

I solved this in the past with a solution with my own SecondOpinonRuler where I extend SpanRuler component and override add_patterns and match methods, basically to add an additional key in self._match_label_id_map dict values. But I feel like there might be a smarter way with less to maintain. Any thoughts?

[NixBiks]

A helpful extension would be the ability to set extension attributes to the span when proving a pattern

ruler.add_patterns([
    {
        "label": "GPE",
        "pattern": "Apple",
        "_": {
            "custom_attribute": {"name": "John Doe"},
        },
    }
])

Then I could access that via doc.spans['ruler'][0]._.get("custom_attribute"). Would that be interesting as a PR?