pgsql-test

pgsql-test gives you instant, isolated PostgreSQL databases for each test — with automatic transaction rollbacks, context switching, and clean seeding. Forget flaky tests and brittle environments. Write real SQL. Get real coverage. Stay fast.

Install

npm install pgsql-test

Features

• ⚡ Instant test DBs — each one seeded, isolated, and UUID-named
• 🔄 Per-test rollback — every test runs in its own transaction or savepoint
• 🛡️ RLS-friendly — test with role-based auth via .setContext()
• 🌱 Flexible seeding — run .sql files, programmatic seeds, or even load fixtures
• 🧪 Compatible with any async runner — works with Jest, Mocha, etc.
• 🧹 Auto teardown — no residue, no reboots, just clean exits

LaunchQL migrations

Part of the LaunchQL ecosystem, pgsql-test is built to pair seamlessly with our TypeScript-based Sqitch engine rewrite:

• 🚀 Lightning-fast migrations — powered by LaunchQL’s native deployer (10x faster than legacy Sqitch)
• 🔧 Composable test scaffolds — integrate with full LaunchQL stacks or use standalone

Table of Contents

1. Install
2. Features
3. Quick Start
4. getConnections() Overview
5. PgTestClient API Overview
6. Usage Examples
   • Basic Setup
   • Role-Based Context
   • Seeding System
   • SQL File Seeding
   • Programmatic Seeding
   • CSV Seeding
   • JSON Seeding
   • Sqitch Seeding
   • LaunchQL Seeding
7. getConnections() Options
8. Disclaimer

✨ Quick Start

import { getConnections } from 'pgsql-test';

let db, teardown;

beforeAll(async () => {
  ({ db, teardown } = await getConnections());
  await db.query(`SELECT 1`); // ✅ Ready to run queries
});

afterAll(() => teardown());

getConnections() Overview

import { getConnections } from 'pgsql-test';

// Complete object destructuring
const { pg, db, admin, teardown, manager } = await getConnections();

// Most common pattern
const { db, teardown } = await getConnections();

The getConnections() helper sets up a fresh PostgreSQL test database and returns a structured object with:

• pg: a PgTestClient connected as the root or superuser — useful for administrative setup or introspection
• db: a PgTestClient connected as the app-level user — used for running tests with RLS and granted permissions
• admin: a DbAdmin utility for managing database state, extensions, roles, and templates
• teardown(): a function that shuts down the test environment and database pool
• manager: a shared connection pool manager (PgTestConnector) behind both clients

Together, these allow fast, isolated, role-aware test environments with per-test rollback and full control over setup and teardown.

The PgTestClient returned by getConnections() is a fully-featured wrapper around pg.Pool. It provides:

• Automatic transaction and savepoint management for test isolation
• Easy switching of role-based contexts for RLS testing
• A clean, high-level API for integration testing PostgreSQL systems

PgTestClient API Overview

let pg: PgTestClient;
let teardown: () => Promise<void>;

beforeAll(async () => {
  ({ pg, teardown } = await getConnections());
});

beforeEach(() => pg.beforeEach());
afterEach(() => pg.afterEach());
afterAll(() => teardown());

The PgTestClient returned by getConnections() wraps a pg.Client and provides convenient helpers for query execution, test isolation, and context switching.

Common Methods

• query(sql, values?) – Run a raw SQL query and get the QueryResult
• beforeEach() – Begins a transaction and sets a savepoint (called at the start of each test)
• afterEach() – Rolls back to the savepoint and commits the outer transaction (cleans up test state)
• setContext({ key: value }) – Sets PostgreSQL config variables (like role) to simulate RLS contexts
• any, one, oneOrNone, many, manyOrNone, none, result – Typed query helpers for specific result expectations

These methods make it easier to build expressive and isolated integration tests with strong typing and error handling.

The PgTestClient returned by getConnections() is a fully-featured wrapper around pg.Pool. It provides:

• Automatic transaction and savepoint management for test isolation
• Easy switching of role-based contexts for RLS testing
• A clean, high-level API for integration testing PostgreSQL systems

Usage Examples

⚡ Basic Setup

import { getConnections } from 'pgsql-test';

let db; // A fully wrapped PgTestClient using pg.Pool with savepoint-based rollback per test
let teardown;

beforeAll(async () => {
  ({ db, teardown } = await getConnections());

  await db.query(`
    CREATE TABLE users (id SERIAL PRIMARY KEY, name TEXT);
    CREATE TABLE posts (id SERIAL PRIMARY KEY, user_id INT REFERENCES users(id), content TEXT);

    INSERT INTO users (name) VALUES ('Alice'), ('Bob');
    INSERT INTO posts (user_id, content) VALUES (1, 'Hello world!'), (2, 'Graphile is cool!');
  `);
});

afterAll(() => teardown());

beforeEach(() => db.beforeEach());
afterEach(() => db.afterEach());

test('user count starts at 2', async () => {
  const res = await db.query('SELECT COUNT(*) FROM users');
  expect(res.rows[0].count).toBe('2');
});

🔐 Role-Based Context

The pgsql-test framework provides powerful tools to simulate authentication contexts during tests, which is particularly useful when testing Row-Level Security (RLS) policies.

Setting Test Context

Use setContext() to simulate different user roles and JWT claims:

db.setContext({
  role: 'authenticated',
  'jwt.claims.user_id': '123',
  'jwt.claims.org_id': 'acme'
});

This applies the settings using SET LOCAL statements, ensuring they persist only for the current transaction and maintain proper isolation between tests.

Testing Role-Based Access

describe('authenticated role', () => {
  beforeEach(async () => {
    db.setContext({ role: 'authenticated' });
    await db.beforeEach();
  });

  afterEach(() => db.afterEach());

  it('runs as authenticated', async () => {
    const res = await db.query(`SELECT current_setting('role', true) AS role`);
    expect(res.rows[0].role).toBe('authenticated');
  });
});

Database Connection Options

For non-superuser testing, use the connection options described in the options section. The db.connection property allows you to customize the non-privileged user account for your tests.

Use setContext() to simulate Role-Based Access Control (RBAC) during tests. This is useful when testing Row-Level Security (RLS) policies. Your actual server should manage role/user claims via secure tokens (e.g., setting current_setting('jwt.claims.user_id')), but this interface helps emulate those behaviors in test environments.

Common Testing Scenarios

This approach enables testing various access patterns:

• Authenticated vs. anonymous user access
• Per-user data filtering
• Admin privilege bypass behavior
• Custom claim-based restrictions (organization membership, admin status)

Note: While this interface helps simulate RBAC for testing, your production server should manage user/role claims via secure authentication tokens, typically by setting values like current_setting('jwt.claims.user_id') through proper authentication middleware.

🌱 Seeding System

The second argument to getConnections() is an optional array of SeedAdapter objects:

const { db, teardown } = await getConnections(getConnectionOptions, seedAdapters);

This array lets you fully customize how your test database is seeded. You can compose multiple strategies:

• seed.sqlfile() – Execute raw .sql files from disk
• seed.fn() – Run JavaScript/TypeScript logic to programmatically insert data
• seed.csv() – Load tabular data from CSV files
• seed.json() – Use in-memory objects as seed data
• seed.sqitch() – Deploy a Sqitch-compatible migration project
• seed.launchql() – Apply a LaunchQL module using deployFast() (compatible with sqitch)

✨ Default Behavior: If no SeedAdapter[] is passed, LaunchQL seeding is assumed. This makes pgsql-test zero-config for LaunchQL-based projects.

This composable system allows you to mix-and-match data setup strategies for flexible, realistic, and fast database tests.

🔌 SQL File Seeding

Use .sql files to set up your database state before tests:

import path from 'path';
import { getConnections, seed } from 'pgsql-test';

const sql = (f: string) => path.join(__dirname, 'sql', f);

let db;
let teardown;

beforeAll(async () => {
  ({ db, teardown } = await getConnections({}, [
      seed.sqlfile([
        sql('schema.sql'),
        sql('fixtures.sql')
      ])
  ]));
});

afterAll(async () => {
  await teardown();
});

🧠 Programmatic Seeding

Use JavaScript functions to insert seed data:

import { getConnections, seed } from 'pgsql-test';

let db;
let teardown;

beforeAll(async () => {
  ({ db, teardown } = await getConnections({}, [
    seed.fn(async ({ pg }) => {
      await pg.query(`
        INSERT INTO users (name) VALUES ('Seeded User');
      `);
    })
  ]));
});

🗃️ CSV Seeding

You can load tables from CSV files using seed.csv({ ... }). CSV headers must match the table column names exactly. This is useful for loading stable fixture data for integration tests or CI environments.

import path from 'path';
import { getConnections, seed } from 'pgsql-test';

const csv = (file: string) => path.resolve(__dirname, '../csv', file);

let db;
let teardown;

beforeAll(async () => {
  ({ db, teardown } = await getConnections({}, [
    // Create schema
    seed.fn(async ({ pg }) => {
      await pg.query(`
        CREATE TABLE users (
          id SERIAL PRIMARY KEY,
          name TEXT NOT NULL
        );

        CREATE TABLE posts (
          id SERIAL PRIMARY KEY,
          user_id INT REFERENCES users(id),
          content TEXT NOT NULL
        );
      `);
    }),
    // Load from CSV
    seed.csv({
      users: csv('users.csv'),
      posts: csv('posts.csv')
    }),
    // Adjust SERIAL sequences to avoid conflicts
    seed.fn(async ({ pg }) => {
      await pg.query(`SELECT setval(pg_get_serial_sequence('users', 'id'), (SELECT MAX(id) FROM users));`);
      await pg.query(`SELECT setval(pg_get_serial_sequence('posts', 'id'), (SELECT MAX(id) FROM posts));`);
    })
  ]));
});

afterAll(() => teardown());

it('has loaded rows', async () => {
  const res = await db.query('SELECT COUNT(*) FROM users');
  expect(+res.rows[0].count).toBeGreaterThan(0);
});

🗃️ JSON Seeding

You can seed tables using in-memory JSON objects. This is useful when you want fast, inline fixtures without managing external files.

import { getConnections, seed } from 'pgsql-test';

let db;
let teardown;

beforeAll(async () => {
  ({ db, teardown } = await getConnections({}, [
    // Create schema
    seed.fn(async ({ pg }) => {
      await pg.query(`
        CREATE SCHEMA custom;
        CREATE TABLE custom.users (
          id SERIAL PRIMARY KEY,
          name TEXT NOT NULL
        );

        CREATE TABLE custom.posts (
          id SERIAL PRIMARY KEY,
          user_id INT REFERENCES custom.users(id),
          content TEXT NOT NULL
        );
      `);
    }),
    // Seed with in-memory JSON
    seed.json({
      'custom.users': [
        { id: 1, name: 'Alice' },
        { id: 2, name: 'Bob' }
      ],
      'custom.posts': [
        { id: 1, user_id: 1, content: 'Hello world!' },
        { id: 2, user_id: 2, content: 'Graphile is cool!' }
      ]
    }),
    // Fix SERIAL sequences
    seed.fn(async ({ pg }) => {
      await pg.query(`SELECT setval(pg_get_serial_sequence('custom.users', 'id'), (SELECT MAX(id) FROM custom.users));`);
      await pg.query(`SELECT setval(pg_get_serial_sequence('custom.posts', 'id'), (SELECT MAX(id) FROM custom.posts));`);
    })
  ]));
});

afterAll(() => teardown());

it('has loaded rows', async () => {
  const res = await db.query('SELECT COUNT(*) FROM custom.users');
  expect(+res.rows[0].count).toBeGreaterThan(0);
});

🏗️ Sqitch Seeding

Note: While compatible with Sqitch syntax, LaunchQL uses its own high-performance TypeScript-based deploy engine. that we encourage using for sqitch projects

You can seed your test database using a Sqitch project but with significantly improved performance by leveraging LaunchQL's TypeScript deployment engine:

import path from 'path';
import { getConnections, seed } from 'pgsql-test';

const cwd = path.resolve(__dirname, '../path/to/sqitch');

beforeAll(async () => {
  ({ db, teardown } = await getConnections({}, [
    seed.sqitch(cwd)
  ]));
});

This works for any Sqitch-compatible module, now accelerated by LaunchQL's deployment tooling.

🚀 LaunchQL Seeding

If your project uses LaunchQL modules with a precompiled sqitch.plan, you can use pgsql-test with zero configuration. Just call getConnections() — and it just works:

import { getConnections } from 'pgsql-test';

let db, teardown;

beforeAll(async () => {
  ({ db, teardown } = await getConnections()); // 🚀 LaunchQL deployFast() is used automatically - up to 10x faster than traditional Sqitch!
});

This works out of the box because pgsql-test uses the high-speed deployFast() function by default, applying any compiled LaunchQL schema located in the current working directory (process.cwd()).

If you want to specify a custom path to your LaunchQL module, use seed.launchql() explicitly:

import path from 'path';
import { getConnections, seed } from 'pgsql-test';

const cwd = path.resolve(__dirname, '../path/to/launchql');

beforeAll(async () => {
  ({ db, teardown } = await getConnections({}, [
    seed.launchql(cwd) // uses deployFast() - up to 10x faster than traditional Sqitch!
  ]));
});

Why LaunchQL's Approach?

LaunchQL provides the best of both worlds:

1. Sqitch Compatibility: Keep your familiar Sqitch syntax and migration approach
2. TypeScript Performance: Our TS-rewritten deployment engine delivers up to 10x faster schema deployments
3. Developer Experience: Tight feedback loops with near-instant schema setup for tests
4. CI Optimization: Dramatically reduced test suite run times with optimized deployment

By maintaining Sqitch compatibility while supercharging performance, LaunchQL enables you to keep your existing migration patterns while enjoying the speed benefits of our TypeScript engine.

getConnections Options

This table documents the available options for the getConnections function. The options are passed as a combination of pg and db configuration objects.

db Options (PgTestConnectionOptions)

pg Options (PgConfig)

Environment variables will override these options when available:

• PGHOST, PGPORT, PGUSER, PGPASSWORD, PGDATABASE

Usage

const { conn, db, teardown } = await getConnections({
  pg: { user: 'postgres', password: 'secret' },
  db: {
    extensions: ['uuid-ossp'],
    cwd: '/path/to/project',
    connection: { user: 'test_user', password: 'secret', role: 'authenticated' },
    template: 'test_template',
    prefix: 'test_',
    rootDb: 'postgres'
  }
});

Related LaunchQL Tooling

🧪 Testing

• launchql/pgsql-test: 📊 Isolated testing environments with per-test transaction rollbacks—ideal for integration tests, complex migrations, and RLS simulation.
• launchql/graphile-test: 🔐 Authentication mocking for Graphile-focused test helpers and emulating row-level security contexts.
• launchql/pg-query-context: 🔒 Session context injection to add session-local context (e.g., SET LOCAL) into queries—ideal for setting role, jwt.claims, and other session settings.

🧠 Parsing & AST

• launchql/pgsql-parser: 🔄 SQL conversion engine that interprets and converts PostgreSQL syntax.
• launchql/libpg-query-node: 🌉 Node.js bindings for libpg_query, converting SQL into parse trees.
• launchql/pg-proto-parser: 📦 Protobuf parser for parsing PostgreSQL Protocol Buffers definitions to generate TypeScript interfaces, utility functions, and JSON mappings for enums.
• @pgsql/enums: 🏷️ TypeScript enums for PostgreSQL AST for safe and ergonomic parsing logic.
• @pgsql/types: 📝 Type definitions for PostgreSQL AST nodes in TypeScript.
• @pgsql/utils: 🛠️ AST utilities for constructing and transforming PostgreSQL syntax trees.
• launchql/pg-ast: 🔍 Low-level AST tools and transformations for Postgres query structures.

🚀 API & Dev Tools

• launchql/server: ⚡ Express-based API server powered by PostGraphile to expose a secure, scalable GraphQL API over your Postgres database.
• launchql/explorer: 🔎 Visual API explorer with GraphiQL for browsing across all databases and schemas—useful for debugging, documentation, and API prototyping.

🔁 Streaming & Uploads

• launchql/s3-streamer: 📤 Direct S3 streaming for large files with support for metadata injection and content validation.
• launchql/etag-hash: 🏷️ S3-compatible ETags created by streaming and hashing file uploads in chunks.
• launchql/etag-stream: 🔄 ETag computation via Node stream transformer during upload or transfer.
• launchql/uuid-hash: 🆔 Deterministic UUIDs generated from hashed content, great for deduplication and asset referencing.
• launchql/uuid-stream: 🌊 Streaming UUID generation based on piped file content—ideal for upload pipelines.
• launchql/upload-names: 📂 Collision-resistant filenames utility for structured and unique file names for uploads.

🧰 CLI & Codegen

• @launchql/cli: 🖥️ Command-line toolkit for managing LaunchQL projects—supports database scaffolding, migrations, seeding, code generation, and automation.
• launchql/launchql-gen: ✨ Auto-generated GraphQL mutations and queries dynamically built from introspected schema data.
• @launchql/query-builder: 🏗️ SQL constructor providing a robust TypeScript-based query builder for dynamic generation of SELECT, INSERT, UPDATE, DELETE, and stored procedure calls—supports advanced SQL features like JOIN, GROUP BY, and schema-qualified queries.
• @launchql/query: 🧩 Fluent GraphQL builder for PostGraphile schemas. ⚡ Schema-aware via introspection, 🧩 composable and ergonomic for building deeply nested queries.

Disclaimer

AS DESCRIBED IN THE LICENSES, THE SOFTWARE IS PROVIDED "AS IS", AT YOUR OWN RISK, AND WITHOUT WARRANTIES OF ANY KIND.

No developer or entity involved in creating this software will be liable for any claims or damages whatsoever associated with your use, inability to use, or your interaction with other users of the code, including any direct, indirect, incidental, special, exemplary, punitive or consequential damages, or loss of profits, cryptocurrencies, tokens, or anything else of value.