Schema field reference

This page is a reference for the schema field types and options available in Bundata. Use it when defining schemas in Schema Studio or via the API for schema-aware extraction and smart bites.

Primitive types

Type	API/key	Description	Example use
text	`string`	Free-form text	`title`, `body`, `vendor_name`
number	`number`	Numeric value	`total`, `quantity`, `page_number`
date	`date` / `string` (ISO)	Date or datetime	`effective_date`, `invoice_date`
boolean	`boolean`	True/false	`is_signed`, `has_amendments`

text — No format constraint by default; use for any string. Optional max length where supported.
number — Integer or decimal depending on configuration. Use for amounts, counts, page numbers.
date — Prefer ISO 8601 (e.g. YYYY-MM-DD) for consistency and filtering. See product docs for timezone handling.
boolean — Use for flags; extraction may derive from presence/absence of a clause or field.

Complex types

Type	Description	Example use
object	Nested key-value structure	`address`, `party` (name, role, contact)
array	List of values or objects	`line_items[]`, `sections[]`, `signatories[]`

object — Define child fields. Use for a single structured block (e.g. one address per document).
array — Element type can be primitive (e.g. array of strings) or object (e.g. array of { description, quantity, amount }). Use for repeated blocks like invoice line items or contract sections.

Required vs. optional

required — Field must be present for the run to be considered successful for that document. Use sparingly; missing data can cause failures. See Schema design.
optional — Field may be absent. Prefer optional for fields that don’t appear in every document (e.g. optional clauses, attachments).

Validation and constraints

format — For dates, specify expected format (e.g. YYYY-MM-DD) where configurable.
min/max — For numbers, optional min/max bounds. For text, optional max length.
Product-specific options may exist (e.g. enum, pattern). Check Schema Studio or the API schema definition for your deployment.

API representation

Schemas are typically represented as JSON: an object with a fields array (or equivalent). Each field has name, type, required, and type-specific options. Example (conceptual):

{
  "fields": [
    { "name": "title", "type": "text", "required": false },
    { "name": "effective_date", "type": "date", "required": false },
    { "name": "line_items", "type": "array", "items": { "type": "object", "fields": [...] } }
  ]
}

Exact schema is defined in the API reference for your plan.

Schema Studio: Field types — When to use each type.
Schema design — Design principles.
Extraction overview — How extraction uses schemas.