@opendata-ai/openchart-core

Types, spec builders, formatters, and palettes for the OpenChart visualization library.

This is the foundation package. It has no DOM dependencies and runs in any JavaScript environment. The framework packages (openchart-react, openchart-vue, openchart-svelte) all depend on it.

Install

npm install @opendata-ai/openchart-core

Spec builders

Convenience functions that create chart and table specs from data. They infer field types automatically (quantitative, temporal, nominal) by sampling your data, so you don't have to specify types manually.

import { lineChart, columnChart, barChart } from '@opendata-ai/openchart-core';

const line = lineChart(data, 'date', 'value');
const columns = columnChart(data, 'month', 'revenue');
const bars = barChart(data, 'category', 'count');

All chart builders accept an optional options argument for color encoding, annotations, chrome (title/subtitle/source), theme overrides, and dark mode.

const spec = lineChart(data, 'date', 'value', {
  color: 'region',
  chrome: { title: 'Sales by Region', subtitle: 'Q1 2025' },
});

Available builders

Function	Chart type	Params
`lineChart(data, x, y, options?)`	Line	x/y position fields
`barChart(data, category, value, options?)`	Horizontal bar	category on y-axis, value on x-axis
`columnChart(data, x, y, options?)`	Vertical column	x/y position fields
`pieChart(data, category, value, options?)`	Pie	category for color, value for size
`donutChart(data, category, value, options?)`	Donut	category for color, value for size
`areaChart(data, x, y, options?)`	Area	x/y position fields
`dotChart(data, x, y, options?)`	Dot/strip	x/y position fields
`scatterChart(data, x, y, options?)`	Scatter	x/y position fields
`dataTable(data, options?)`	Data table	columns auto-generated if omitted

Field refs can be a plain string (field name) or a full EncodingChannel object when you need control over type, axis, scale, or aggregation.

dataTable()

The table builder is worth calling out specifically. If you pass data without a columns config, it auto-generates column definitions from your data keys. It infers types from the values: numbers get right-aligned, text gets left-aligned.

import { dataTable } from '@opendata-ai/openchart-core';

// Columns auto-generated from data keys
const spec = dataTable([
  { name: 'Alice', age: 30, city: 'Portland' },
  { name: 'Bob', age: 25, city: 'Seattle' },
]);
// age column auto-detected as quantitative, right-aligned
// name and city columns detected as nominal, left-aligned

For more control, pass explicit column configs:

const spec = dataTable(data, {
  columns: [
    { key: 'name', label: 'Name' },
    { key: 'age', label: 'Age', align: 'right' },
  ],
  search: true,
  pagination: { pageSize: 20 },
});

StoredVizSpec types

When you need to persist specs without bulky data arrays (for storage, serialization, or database records), use the data-stripped type variants:

import type {
  StoredVizSpec,
  ChartSpecWithoutData,
  TableSpecWithoutData,
  GraphSpecWithoutData,
} from '@opendata-ai/openchart-core';

// Store the spec shape without data
const stored: ChartSpecWithoutData = {
  type: 'line',
  encoding: {
    x: { field: 'date', type: 'temporal' },
    y: { field: 'value', type: 'quantitative' },
  },
};

// StoredVizSpec is the union of all three
function saveSpec(spec: StoredVizSpec) {
  localStorage.setItem('viz', JSON.stringify(spec));
}

Type guards

Runtime narrowing for VizSpec values. Useful when you have a generic spec and need to branch on its type:

import { isChartSpec, isTableSpec, isGraphSpec } from '@opendata-ai/openchart-core';
import type { VizSpec } from '@opendata-ai/openchart-core';

function describe(spec: VizSpec): string {
  if (isChartSpec(spec)) return `Chart: ${spec.type}`;
  if (isTableSpec(spec)) return 'Data table';
  if (isGraphSpec(spec)) return 'Network graph';
}

Other exports

Beyond spec builders, this package provides:

Theme system - DEFAULT_THEME, resolveTheme(), adaptTheme() for dark mode adaptation
Color palettes - CATEGORICAL_PALETTE, SEQUENTIAL_PALETTES, DIVERGING_PALETTES
Accessibility - generateAltText(), generateAriaLabels(), contrast ratio utilities
Formatting - formatNumber(), formatDate(), abbreviateNumber()
Layout - computeChrome(), estimateTextWidth(), responsive breakpoints
Label collision - resolveCollisions() for automatic label placement