JavaScript

A quick-start guide to connecting with the Trino JavaScript client

Doorda data is served through Trino, so you can query your tables directly with SQL instead of calling per-table API endpoints. The official Trino client for Node.js handles the connection, paging and results for you.

Connection details

npm install @trinodb/trino-js-client

import { Trino, BasicAuth } from '@trinodb/trino-js-client';
const trino = Trino.create({
server: 'https://host.doorda.com',
catalog: '<catalog>',
schema: '<schema>',
source: 'your-app-name',
auth: new BasicAuth(
process.env.DOORDA_USER,
process.env.DOORDA_PASSWORD
),
});

Create one client and reuse it. Keep your credentials in environment variables or a secrets manager, never in source code.

catalog and schema set the defaults for unqualified table names. You can leave them out and use full names in your SQL instead: catalog.schema.table.

About source. This is a free-text label sent with every query (the X-Trino-Source header). If you leave it out, the client sends trino-js-client. Please set it to the name of your application or service, so that we can identify your traffic and help quickly if you raise a support query.

async function runQuery(sql) {
const pages = await trino.query(sql);
let columns = [];
const rows = [];
for await (const page of pages) {
if (page.error) {
throw new Error(`${page.error.errorName}: ${page.error.message}`);
}
if (page.columns) columns = page.columns.map(c => c.name);
for (const values of page.data ?? []) {
rows.push(Object.fromEntries(values.map((v, i) => [columns[i], v])));
}
}
return rows;
}
const rows = await runQuery('SELECT * FROM <table> LIMIT 10');
console.log(rows);

Results arrive in pages. Each page holds rows as arrays of values, with the column names alongside. This helper collects every page and returns plain objects.

Errors

Problems with the query itself (a syntax error, an unknown table, a table outside your subscription) are returned on the result page as page.error rather than thrown, which is why the helper checks for it. Connection and login problems, such as a wrong password, are thrown as ordinary HTTP errors by trino.query().

SHOW CATALOGS;
SHOW SCHEMAS FROM <catalog>;
SHOW TABLES FROM <catalog>.<schema>;
DESCRIBE <catalog>.<schema>.<table>;

Good practice

• Run it on your server. The client is built for Node.js. Do not call Doorda from browser code, as that would expose your credentials.
• Ask only for what you need. Name the columns you want, filter with WHERE, and use LIMIT while exploring. Smaller queries return faster.
• Expect analytical response times. Queries typically take from under a second to several seconds. For lookups inside a live user journey, fetch the data ahead of time and store it on your side.
• Large results. The helper above holds every row in memory. For big extracts, process each page inside the loop instead of collecting them all.
• Stopping a query. Every page carries the query id; call trino.cancel(id) to stop a query you no longer need.