JSON and the designer

Everything in a schema except its hooks is data, so a schema can be stored, versioned and edited without touching code.

The designer

The editor’s toolbar has a schema button that opens the schema designer. With a workbook open, it can:

• Generate a schema from the workbook: it finds the tables and their headings, and guesses each field’s type from the values under them.
• Edit it: the sheets, tables, fields, their types and checks, and the workbook’s settings, with the workbook beside it.
• Export it as JSON, or import one.

People who know the workbooks best can describe them, and you keep the result with your app. Turn the button off with features: { schemas: false } where people only fill workbooks in.

To and from JSON

import { schemaFromJSON, schemaToJSON } from '@ruled/sheet';

const json = schemaToJSON(priceList);          // a string
const schema = schemaFromJSON(json);           // a WorkbookSchema again

Regular expressions in labels are written as { "pattern": "…", "flags": "i" }, and date bounds as ISO dates.

Hooks stay in code

JSON can’t hold functions, so a schema’s hooks (its validate, rules, onChange and beforeSave) live in your app. Add them to a schema loaded from JSON before registering it:

const priceList = schemaFromJSON(await fetch('/schemas/price-list.json').then((r) => r.text()));
priceList.onChange = fillFromCatalogue;
priceList.sheets[0].tables![0].fields.find((f) => f.key === 'sku')!.validate = checkSku;
editor.registerSchema(priceList);

When someone edits a schema you registered in the designer, the hooks stay with it: they’re matched to the edited schema by the workbook’s sheet names, and the tables’, fields’ and cells’ keys.