@fmtk/pgdb

pgdatabase

What?

An UNFINISHED micro-ORM for Postgres.

Why?

For when you want to write code for the platform you're targeting, rather than fighting an abstraction that lets you target all the database engines you will never use while trading off on flexibility and access to database features.

How?

Install with yarn:

yarn add @fmtk/pgdb

Or, install with npm:

npm install --save @fmtk/pgdb

Types definitions?

Included.

Details

Database

Create an instance:

const db = new Database(
  connectionStringOrConfig, // null to get from env
  'migrationNamespace', // e.g. name of app
  migrations, // array of Migration objects
);

Initialise:

// update to latest (run migrations)
await db.init(true);

Make a query:

const results = db.connection.query<Foo>('SELECT * FROM foo');

Column Maps

Create a model interface:

interface Customer {
  id: int;
  name: string;
  email: string;
  mobile: string;
}

Create a column map:

const customerMap = new ColumnMap<keyof Customer>({
  id: 'id',
  name: 'name',
  email: 'email',
  mobile: 'mobile',
});

Auto-generate SQL:

const customer = await db.connection.Select<Customer>(
  'customers',
  customerMap,
  { id: 4 },
  true, // single
  true, // throwIfEmpty
);
// typeof customer is Customer.

Migrations

The migration system will create a table as follows, where NAMESPACE is provided when the Database class is initialised:

CREATE TABLE ___NAMESPACE_migrations (
  id int NOT NULL PRIMARY KEY,
  at timestamp NOT NULL
    DEFAULT (NOW() AT TIME ZONE 'utc'),
  hash text NOT NULL
)

This table stores migrations that have been applied, along with time of migration and a hash of all the DDL that was executed. When the Database class is first initialised, it will check the migrations table and throw an error if the migrations have changed.

A migration looks like this:

const migration = new Migration(1, [
  `CREATE TABLE foo (
    id int primary key,
    value text not null
  )`,
  `CREATE IX_value INDEX ON foo(value)`,
  // ... more DDL
]);

For convenience, DDL can be generated using the DDL methods:

const migration = new Migration(1, [
  createTable(
    'foo',
    column('id', 'int', { primaryKey: true }),
    column('value', 'text', { nullable: true }),
  ),
  createIndex('foo', 'value'),
  // ... more DDL
]);

Constraints and indices generated using the DDL methods will have auto-generated names:

Type	Format
Primary key	PK:table:column
Unique constraint	UQ:table:column
Foreign key	FK:table:column:target
Index	IX:table:column1:column2:...:columnN

These names can be generated using the following functions:

• makePrimaryKeyName
• makeUniqueConstraintName
• makeForeignKeyName
• makeIndexName