Custom Vega-Lite

Use a custom visualization when the built-in visual types cannot express the analytical view. Custom programs run in a sandbox and receive only the compiled in-memory dataset.

[!EXPERIMENTAL] Custom Vega-Lite is experimental. Its authoring contract, sandbox capabilities, and rendering behavior may change before it is promoted to stable. Prefer a built-in visual type whenever it can express the analytical view.

Every preview on this page is generated from the YAML shown below it using a fixed documentation dataset.

Basic

Reference primary as the named dataset and use query aliases in encodings. LeapView rejects network data, inline values, transforms, parameters, and executable expressions.

Key fields

visuals:
  custom_monthly_revenue:
    title: Monthly revenue
    description: A constrained Vega-Lite bar chart over governed query data.
    type: custom
    query:
      dimensions:
        purchase_month: orders.purchase_month
      measures:
        revenue: null
      sort:
        - field: purchase_month
          direction: asc
      limit: 30
    custom:
      engine: vega_lite
      program:
        data:
          name: primary
        mark: bar
        encoding:
          x:
            field: purchase_month
            type: ordinal
          y:
            field: revenue
            type: quantitative

API reference

Kind: chart. Renderer: vega-lite-sandbox. Supported result shapes: custom.

Field	Type	Default	Allowed values	Description
`query.dimensions`	`field mapping`	—	—	Groups query results and supplies category, hierarchy, matrix, graph, or geographic labels.
`query.limit`	`integer`	`no limit`	`positive integer`	Caps the number of rows returned to the renderer.
`query.measures`	`measure mapping`	—	—	Selects the governed semantic measures consumed by the visual shape.
`query.sort`	`sort list`	—	—	Orders query results by a returned field or measure alias.
`custom.engine`	`string`	—	`vega_lite`	Selects the constrained custom visualization engine.
`custom.program`	`Vega-Lite object`	—	—	Defines a validated declarative program over the in-memory primary dataset; network data, transforms, expressions, and inline values are prohibited.

Accessibility. Use a descriptive title and unit, and do not rely on color alone to distinguish series or values.