'labels' parameter of visualize_ner()

[dmesquita]

Hi! The default labels parameter of visualize_ner() is an empty tuple and I think we can make it more explicit because today I was debugging to figure out why my custom entities were not being displayed :P. Maybe I'm missing something, what does keyword-only mean? (docs table)

Back to the labels parameter, I'm not sure if it's best to:

• modify the method to check if the tuple is empty and if so get all labels from the doc
• simply add a required label in the docs

I modified the docs because it's more straightforward, let me know if you think it's not a good path. Thanks!

[svlandeg]

Hi!

Apologies for the late follow-up.

what does keyword-only mean? (docs table)

This means that an argument can only be specified with its keyword, e.g. visualize_ner(doc, labels=my_labels) works, but visualize_ner(doc, my_labels) doesn't. This is triggered by using the * in the definition:

def visualize_ner(
    doc: spacy.tokens.Doc,
    *,
    labels: Sequence[str] = tuple(),
    attrs: List[str] = NER_ATTRS,
    show_table: bool = True,
    title: Optional[str] = "Named Entities",
    colors: Dict[str, str] = {},
    key: Optional[str] = None,
)

It's an approach we often take for optional arguments to ensure that code is robust to changing order of the parameters & future extensions/changes of the function definition.

You're right that the empty tuple default is confusing though. Perhaps we could raise a warning if labels is empty? That will also catch cases where the user accidently passes on an empty list or such.

[svlandeg]

Closing this one in favour of the more general solution presented in https://github.com/explosion/spacy-streamlit/pull/26, but thanks again for the report!