Lego.js

A lightweight SQL (string) builder using ES6 template strings. Lego embraces SQL instead of adding yet another abstraction layer.

Lego.sql `SELECT * FROM users WHERE name = ${name}`;

Lego does not do simple string concatenation. Instead, Lego creates parameterized queries. For example, the following query is created from the previous call:

SELECT * FROM users WHERE name = $1

Quick start

Lego uses ES6 template strings. From the template string, a parameterized query is created and passed to the driver. The driver creates a pool of connections with the url from process.env.DATABASE_URL.

Lego.sql `INSERT INTO projects (name) VALUES (${name}) RETURNING *`
	.then((projects) => {
		return Lego.sql `INSERT INTO project_settings (project_id) VALUES (${projects[0].id})`;
	})
	.then(() => {
		// Ready! :-)
	})

You can also nest arrays of Lego instances:

Lego.sql `INSERT INTO projects (name) VALUES ${projects.map((project) => {
	return Lego.sql `(${project.name})`;
})}`;

Which creates and executes the query INSERT INTO projects (name) VALUES ($1), ($2).

Lego#append

In some cases, you want to append a statement:

const lego = Lego.sql `SELECT * FROM tests`;

if (shouldOrderBy) {
	lego.append `ORDER BY value`;
}

Lego#raw

You cannot pass raw values to your queries, unless you use Lego#raw. Be very careful not to use this with user input.

const column = 'name';
Lego.sql `UPDATE users SET ${Lego.raw(column)} = ${value}`;

Return value

Lego.sql returns a Promise-like object and the query is executed when .then(..) is invoked. The Promise is resolved with the query's result.

In DELETE, UPDATE and INSERT queries, when not using a RETURNING clause, the number of affected rows is resolved. Otherwise, the row data is resolved.

Rows to objects

Lego makes it easy to parse rows and transform them to objects. Consider the following rows:

const rows = [{
	id: 1,
	test_id: 1,
	test_name: 'Test 1'
}, {
	id: 1,
	test_id: 2,
	test_name: 'Test 2'
}];

These rows are flat but do contain 1 root object and 2 child objects. Something like the below:

const objects = [{
	id: 1,
	tests: [{
		id: 1,
		name: 'Test 1'
	}, {
		id: 2,
		name: 'Test 2'
	}]
}]

You can transform the rows by simply passing the rows to Lego's parse system and providing a definition object:

Lego.parse(rows, [{
	id: 'id',
	tests: [{
		id: 'test_id',
		name: 'test_name'
	}]
}]);

The definition object describes how to map columns and rows to objects. Every property refers to a column name. You can also call .parse(..) on a Lego object directly.

const project = await Lego.sql `SELECT
	projects.id,
	projects.created_at,
	project_members.id member_id,
	project_members.name member_name,
	project_members.email member_email
	project_members.joined_at member_joined_at
FROM projects
INNER JOIN project_members ON projects.id = project_members.project_id
WHERE
	projects.id = ${projectID}`
	.parse(rows, {
		id: 'id',
		createdAt: 'created_at',
		members: [{
			id: 'member_id',
			name: 'member_name',
			email: 'member_email',
			joinedAt: 'member_joined_at'
		}]
	});

Please have a look at the parse test cases to learn more about the different ways to transform rows to objects.

Lego#first

Or if you just want the first result from a query (or null if there were no results):

Lego.sql `SELECT * FROM accounts LIMIT 1`
	.first()
	.then((account) => {
		// account is the first row from the query, or null if no rows were found.
	});

Transactions

Transactions are also supported. You can either chain the calls manually by returning a promise in the transaction's callback:

Lego.transaction((transaction) => {
	return transaction.sql `UPDATE money SET value = value + 100`
		.then(() => {
			return transaction.sql `UPDATE money SET value = value - 100`;
		});
});

Or use the transaction's queue which invokes the queries in series:

Lego.transaction((transaction) => {
	transaction.sql `UPDATE money SET value = value + 100`;
	transaction.sql `UPDATE money SET value = value - 100`;
});

Alternatively, you can construct regular Lego instances and assign them to the transaction:

Lego.transaction((transaction) => {
	return Lego
		.sql `UPDATE money SET value = value + 100`
		.transacting(transaction)
		.then(() => {
			return Lego
				.sql `UPDATE money SET value = value - 100`
				.transacting(transaction);
		});
});

Lego#transaction returns a promise with the result of the transaction's callback.

Migrations

To create a migration you can simply invoke lego migrate:make. This creates an empty migration with an up and a down function in which you can write your queries to alter your database.

To execute migrations, simply invoke lego migrate:latest. Migrations are executed in a transaction. The transaction is passed as argument in the migrate functions.

export function up(transaction) {
	transaction.sql `CREATE TABLE tests (name TEXT UNIQUE, value INTEGER)`;
	transaction.sql `INSERT INTO tests VALUES ('Martijn', 123)`;
}

export function down(transaction) {
	transaction.sql `DROP TABLE tests`;
}

Lego creates a migrations table to keep track of all the migrations. This migrations table is created in it's own schema called lego, so don't worry about any collisions.

To execute your migrations when you call npm run release you should add a run script to your package.json:

"release": "node -r babel-register ./node_modules/.bin/lego migrate:latest",

CLI

The command line interface supports the following commands:

lego version                         Prints the version of Lego.
lego migrate:make                    Creates a new migration file.
lego migrate:latest                  Migrates to the latest migration.
lego migrate:rollback                Rolls back the previous migration.
lego migrate:<version>               Migrates or rolls back to the target migration <version>.

Environment variables