Table Docs
Basic Examples
Feature Examples
- Cell Selection
- Cell Spanning
- Column Filters
- Column Filters (Faceted)
- Bucketed Faceted Filters
- Column Ordering
- Column Pinning
- Column Pinning (Split)
- Sticky Column Pinning
- Column Sizing
- Column Resizing
- Performant Column Resizing
- Column Visibility
- Expanding
- Expanding Sub Components
- Grouping
- Aggregation
- Grouped Aggregation
- Pagination
- Row Pinning
- Row Selection
- Sorting
- Sorting (Dynamic Data)
Specialized Examples
Alpine Example: Basic Create Table
import Alpine from 'alpinejs'
import {
FlexRender,
createColumnHelper,
createTable,
tableFeatures,
} from '@tanstack/alpine-table'
import { makeData } from './makeData'
import './index.css'
import type { Person } from './makeData'
// This example uses the standalone `createTable` function to create a table without the `createTableHook` util.
// 1. New in V9! Tell the table which features and row models we want to use. In this case, this will be a basic table with no additional features
const features = tableFeatures({}) // util method to create sharable TFeatures object/type
// 2. Create a column helper with the table features and row type
const columnHelper = createColumnHelper<typeof features, Person>()
// 3. Define the columns for your table with the column helper
const columns = columnHelper.columns([
columnHelper.accessor('firstName', {
header: 'First Name',
cell: (info) => info.getValue(),
}),
columnHelper.accessor((row) => row.lastName, {
id: 'lastName',
header: () => 'Last Name',
cell: (info) => info.getValue(),
}),
columnHelper.accessor((row) => Number(row.age), {
id: 'age',
header: () => 'Age',
cell: (info) => info.renderValue(),
}),
columnHelper.accessor('visits', {
header: () => 'Visits',
}),
columnHelper.accessor('status', {
header: 'Status',
}),
columnHelper.accessor('progress', {
header: 'Profile Progress',
}),
])
// 4. Register the Alpine component. Store data in Alpine-reactive state so the
// buttons can swap it out and the table re-renders.
Alpine.data('table', () => {
const local = Alpine.reactive({ data: makeData(20) })
// 5. Create the table instance with required features, columns, and data
const table = createTable({
debugTable: true, // optionally, enable console logging debug messages
features, // new required option in V9. Tell the table which features you are importing and using (better tree-shaking)
columns,
get data() {
return local.data
},
// add additional table options here
})
return {
table,
FlexRender, // exposed so the template can call it inside `x-html`
refreshData() {
local.data = makeData(20)
},
stressTest() {
local.data = makeData(1_000)
},
}
})
window.Alpine = Alpine
Alpine.start()