JavaScript

Execute JavaScript against an in-memory dataset in Node.js or the browser.

Configure

import { AVA } from '@antv/ava/browser';
 
const ava = new AVA({
  llm: { model: 'YOUR_MODEL' },
  engine: { type: 'javascript' },
});

The JavaScript engine supports inline CSV, JSON, and text. Text loading requires the model. It keeps data as one in-memory array.

Query bindings

Queries receive data and the stat helper, and must assign the output to result.

await ava.source({
  type: 'json',
  options: { data: [{ revenue: 120 }, { revenue: 90 }] },
});
const result = await ava.query('const result = [{ revenue: data.reduce((sum, row) => sum + row.revenue, 0) }];');

Input data is cloned before execution. Non-array results are normalized to a row with a value field, then result limits are applied.

Execution environments

In Node.js, code runs in a node:vm context with a 10-second timeout. In browsers, it runs in a sandboxed iframe with a restrictive CSP and a 10-second response timeout.

These mechanisms have different isolation properties. A Node.js VM is not a security boundary for hostile code; deploy an appropriate isolation boundary if your application accepts untrusted executable code.