annotate.extract_json_response

annotate.extract_json_response(
    response,
    annotation_type,
    min_value=None,
    max_value=None,
    options=None,
    label_options=None,
    text=None,
)

Extract and validate JSON response based on annotation type

Parameters

Name	Type	Description	Default
response		Raw model response text that should contain a JSON object.	required
annotation_type		Annotation type string such as "dropdown" or "likert".	required
min_value		Optional integer lower bound for Likert annotations.	None
max_value		Optional integer upper bound for Likert annotations.	None
options		Optional dropdown option list used to normalize categorical labels.	None
label_options		Allowed labels for span annotations.	None
text		Source text for span annotations (used to validate offsets).	None

Returns

Name	Type	Description
		Parsed response value coerced into the expected annotation format. For
		annotation_type == "span" this is a list of span dicts.