loopback-connector-postgresql

PostgreSQL, is a popular open-source object-relational database. The loopback-connector-postgresql module is the PostgreSQL connector for the LoopBack framework.

Installation

In your application root directory, enter this command to install the connector:

$ npm install loopback-connector-postgresql --save

This installs the module from npm and adds it as a dependency to the application’s package.json file.

If you create a PostgreSQL data source using the data source generator as described below, you don’t have to do this, since the generator will run npm install for you.

Creating a data source

For LoopBack 4 users, use the LoopBack 4 Command-line interface to generate a DataSource with PostgreSQL connector to your LB4 application. Run lb4 datasource, it will prompt for configurations such as host, post, etc. that are required to connect to a PostgreSQL database.

After setting it up, the configuration can be found under src/datasources/<DataSourceName>.datasource.ts, which would look like this:

const config = {
  name: 'db',
  connector: 'postgresql',
  url: '',
  host:'localhost',
  port: 5432,
  user: 'user',
  password: 'pass',
  database: 'testdb',
};

"mydb": {
  "name": "mydb",
  "connector": "postgresql"
  "host": "mydbhost",
  "port": 5432,
  "url": "postgres://admin:admin@mydbhost:5432/db1?ssl=false",
  "database": "db1",
  "password": "admin",
  "user": "admin",
  "ssl": false
}

Connection Pool Settings

You can also specify connection pool settings in <DataSourceName>.datasource.ts ( or datasources.json for LB3 users). For instance you can specify the minimum and the maximum pool size, and the maximum pool client’s idle time before closing the client.

Example of db.datasource.ts:

const config = {
  name: 'db',
  connector: 'postgresql',
  url: '',
  host: 'localhost',
  port: 5432,
  user: 'user',
  password: 'pass',
  database: 'testdb',
  min: 5,
  max: 200,
  idleTimeoutMillis: 60000,
  ssl: false
};

Check out node-pg-pool and node postgres pooling example for more information.

Configuration options

NOTE: By default, the ‘public’ schema is used for all tables.

The PostgreSQL connector uses node-postgres as the driver. For more information about configuration parameters, see node-postgres documentation.

Connecting to UNIX domain socket

A common PostgreSQL configuration is to connect to the UNIX domain socket /var/run/postgresql/.s.PGSQL.5432 instead of using the TCP/IP port. For example:

const config = {
  name: 'db',
  connector: 'postgresql',
  url: '',
  host: '/var/run/postgresql/',
  port: 5432,
  user: 'user',
  password: 'pass',
  database: 'testdb',
  debug: true
};

Defining models

LoopBack allows you to specify some database settings through the model definition and/or the property definition. These definitions would be mapped to the database. Please check out the CLI lb4 model for generating LB4 models. The following is a typical LoopBack 4 model that specifies the schema, table and column details through model definition and property definitions:

@model({
  settings: { postgresql: { schema: 'public', table: 'inventory'} },
})
export class Inventory extends Entity {
  @property({
    type: 'number',
    required: true,
    scale: 0,
    id: 1,
    postgresql: {
      columnName: 'id',
      dataType: 'integer',
      dataLength: null,
      dataPrecision: null,
      dataScale: 0,
      nullable: 'NO',
    },
  })
  id: number;

  @property({
    type: 'string',
    postgresql: {
      columnName: 'name',
      dataType: 'text',
      dataLength: null,
      dataPrecision: null,
      dataScale: null,
      nullable: 'YES',
    },
  })
  name?: string;

  @property({
    type: 'boolean',
    required: true,
    postgresql: {
      columnName: 'available',
      dataType: 'boolean',
      dataLength: null,
      dataPrecision: null,
      dataScale: null,
      nullable: 'NO',
    },
  })
  available: boolean;

  constructor(data?: Partial<User>) {
    super(data);
  }
}

{
  "name": "Inventory",
  "options": {
    "idInjection": false,
    "postgresql": {
      "schema": "strongloop",
      "table": "inventory"
    }
  },
  "properties": {
    "id": {
      "type": "String",
      "required": false,
      "length": 64,
      "precision": null,
      "scale": null,
      "postgresql": {
        "columnName": "id",
        "dataType": "character varying",
        "dataLength": 64,
        "dataPrecision": null,
        "dataScale": null,
        "nullable": "NO"
      }
    },
    "productId": {
      "type": "String",
      "required": false,
      "length": 20,
      "precision": null,
      "scale": null,
      "id": 1,
      "postgresql": {
        "columnName": "product_id",
        "dataType": "character varying",
        "dataLength": 20,
        "dataPrecision": null,
        "dataScale": null,
        "nullable": "YES"
      }
    },
    "locationId": {
      "type": "String",
      "required": false,
      "length": 20,
      "precision": null,
      "scale": null,
      "id": 1,
      "postgresql": {
        "columnName": "location_id",
        "dataType": "character varying",
        "dataLength": 20,
        "dataPrecision": null,
        "dataScale": null,
        "nullable": "YES"
      }
    },
    "available": {
      "type": "Number",
      "required": false,
      "length": null,
      "precision": 32,
      "scale": 0,
      "postgresql": {
        "columnName": "available",
        "dataType": "integer",
        "dataLength": null,
        "dataPrecision": 32,
        "dataScale": 0,
        "nullable": "YES"
      }
    },
    "total": {
      "type": "Number",
      "required": false,
      "length": null,
      "precision": 32,
      "scale": 0,
      "postgresql": {
        "columnName": "total",
        "dataType": "integer",
        "dataLength": null,
        "dataPrecision": 32,
        "dataScale": 0,
        "nullable": "YES"
      }
    }
  }
}

To learn more about specifying database settings, please check the section Data Mapping Properties.

Type mapping

See LoopBack 4 types (or LoopBack 3 types) for details on LoopBack’s data types.

LoopBack to PostgreSQL types

Besides the basic LoopBack types, as we introduced above, you can also specify the database type for model properties. It would be mapped to the database (see Data Mapping Properties). For example, we would like the property price to have database type double precision in the corresponding table in the database, we have specify it as following:

  @property({
    type: 'number',
    postgresql: {
      dataType: 'double precision',
    },
  })
  price?: number;

"properties": {
    // ..
    "price": {
      "type": "Number",
      "postgresql": {
        "dataType": "double precision",
      }
    },

PostgreSQL types to LoopBack

Numeric Data Type

Note: The node.js driver for postgres by default casts Numeric type as a string on GET operation. This is to avoid data precision loss since Numeric types in postgres cannot be safely converted to JavaScript Number.

For details, see the corresponding driver issue.

Querying JSON fields

Note The fields you are querying should be setup to use the JSON postgresql data type - see Defining models

Assuming a model such as this:

  @property({
    type: 'number',
    postgresql: {
      dataType: 'double precision',
    },
  })
  price?: number;

You can query the nested fields with dot notation:

CustomerRepository.find({
  where: {
    address.state: 'California',
  },
  order: 'address.city',
});

Extended operators

PostgreSQL supports the following PostgreSQL-specific operators:

• contains
• containedBy
• containsAny
• match

Please note extended operators are disabled by default, you must enable them at datasource level or model level by setting allowExtendedOperators to true.

Operator contains

The contains operator allow you to query array properties and pick only rows where the stored value contains all of the items specified by the query.

The operator is implemented using PostgreSQL array operator @>.

Note The fields you are querying must be setup to use the postgresql array data type - see Defining models above.

Assuming a model such as this:

@model({
  settings: {
    allowExtendedOperators: true,
  }
})
class Post {
  @property({
    type: ['string'],
    postgresql: {
      dataType: 'varchar[]',
    },
  })
  categories?: string[];
}

You can query the tags fields as follows:

const posts = await postRepository.find({
  where: {
    {
      categories: {'contains': ['AA']},
    }
  }
});

Operator containedBy

Inverse of the contains operator, the containedBy operator allow you to query array properties and pick only rows where the all the items in the stored value are contained by the query.

The operator is implemented using PostgreSQL array operator <@.

Note The fields you are querying must be setup to use the postgresql array data type - see Defining models above.

Assuming a model such as this:

@model({
  settings: {
    allowExtendedOperators: true,
  }
})
class Post {
  @property({
    type: ['string'],
    postgresql: {
      dataType: 'varchar[]',
    },
  })
  categories?: string[];
}

You can query the tags fields as follows:

const posts = await postRepository.find({
  where: {
    {
      categories: {'containedBy': ['AA']},
    }
  }
});

Operator containsAnyOf

The containsAnyOf operator allow you to query array properties and pick only rows where the any of the items in the stored value matches any of the items in the query.

The operator is implemented using PostgreSQL array overlap operator &&.

Note The fields you are querying must be setup to use the postgresql array data type - see Defining models above.

Assuming a model such as this:

@model({
  settings: {
    allowExtendedOperators: true,
  }
})
class Post {
  @property({
    type: ['string'],
    postgresql: {
      dataType: 'varchar[]',
    },
  })
  categories?: string[];
}

You can query the tags fields as follows:

const posts = await postRepository.find({
  where: {
    {
      categories: {'containsAnyOf': ['AA']},
    }
  }
});

Operator match

The match operator allows you to perform a full text search using the @@ operator in PostgreSQL.

Assuming a model such as this:

@model({
  settings: {
    allowExtendedOperators: true,
  }
})
class Post {
  @property({
    type: 'string',
  })
  content: string;
}

You can query the content field as follows:

const posts = await postRepository.find({
  where: {
    {
      content: {match: 'someString'},
    }
  }
});

Discovery and auto-migration

Model discovery

The PostgreSQL connector supports model discovery that enables you to create LoopBack models based on an existing database schema. Once you defined your datasource:

• LoopBack 4 users could use the commend lb4 discover to discover models.
• For LB3 users, please check Discovering models from relational databases.

(See database discovery API for related APIs information)

Auto-migration

The PostgreSQL connector also supports auto-migration that enables you to create a database schema from LoopBack models.

For example, based on the following model, the auto-migration method would create/alter existing customer table under public schema in the database. Table customer would have two columns: name and id, where id is also the primary key and has the default value SERIAL as it has definition of type: 'Number' and generated: true:

@model()
export class Customer extends Entity {
  @property({
    id: true,
    type: 'Number',
    generated: true
  })
  id: number;

  @property({
    type: 'string'
  })
  name: string;
}

By default, tables generated by the auto-migration are under public schema and named in lowercase.

Besides the basic model metadata, LoopBack allows you to specify part of the database schema definition via the property definition, which would be mapped to the database.

For example, based on the following model, after running the auto-migration script, a table named CUSTOMER under schema market will be created. Moreover, you can also have different names for your property and the corresponding column. In the example, by specifying the column name, the property name will be mapped to the customer_name column. This is useful when your database has a different naming convention than LoopBack (camelCase).

@model(
  settings: {
    postgresql: {schema: 'market', table: 'CUSTOMER'},
  }
)
export class Customer extends Entity {
  @property({
    id: true,
    type: 'Number',
    generated: true
  })
  id: number;

  @property({
    type: 'string',
    postgresql: {
      columnName: 'customer_name'
    }
  })
  name: string;
}

For how to run the script and more details:

• For LB4 users, please check Database Migration
• For LB3 users, please check Creating a database schema from models

(See LoopBack auto-migrate method for related APIs information)

Here are some limitations and tips:

• If you defined generated: true in the id property, it generates integers by default. For auto-generated uuid, see Auto-generated id property
• Only the id property supports the auto-generation setting generated: true for now
• Auto-migration doesn’t create foreign key constraints by default. But they can be defined through the model definition. See Auto-migrate with foreign keys
• Destroying models may result in errors due to foreign key integrity. First delete any related models by calling delete on models with relationships.

Auto-migrate/Auto-update models with foreign keys

Foreign key constraints can be defined in the model definition.

Note: The order of table creation is important. A referenced table must exist before creating a foreign key constraint.

Define your models and the foreign key constraints as follows:

customer.model.ts:

@model()
export class Customer extends Entity {
  @property({
    id: true,
    type: 'Number',
    generated: true
  })
  id: number;

  @property({
    type: 'string'
  })
  name: string;
}

order.model.ts:

@model({
  settings: {
    foreignKeys: {
      fk_order_customerId: {
        name: 'fk_order_customerId',
        entity: 'Customer',
        entityKey: 'id',
        foreignKey: 'customerId',
        onDelete: 'CASCADE',
        onUpdate: 'SET NULL'
      },
    },
  })
export class Order extends Entity {
  @property({
    id: true,
    type: 'Number',
    generated: true
  })
  id: number;

  @property({
    type: 'string'
  })
  name: string;

  @property({
    type: 'Number'
  })
  customerId: number;
}

({
  "name": "Customer",
  "options": {
    "idInjection": false
  },
  "properties": {
    "id": {
      "type": "Number",
      "id": 1
    },
    "name": {
      "type": "String",
      "required": false
    }
  }
},
{
  "name": "Order",
  "options": {
    "idInjection": false,
    "foreignKeys": {
      "fk_order_customerId": {
        "name": "fk_order_customerId",
        "entity": "Customer",
        "entityKey": "id",
        "foreignKey": "customerId",
        "onDelete": "CASCADE",
        "onUpdate": "SET NULL"
      }
    }
  },
  "properties": {
    "id": {
      "type": "Number"
      "id": 1
    },
    "customerId": {
      "type": "Number"
    },
    "description": {
      "type": "String",
      "required": false
    }
  }
})

Auto-generated ids

Auto-migrate supports the automatic generation of property values for the id property. For PostgreSQL, the default id type is integer. Thus if you have generated: true in the id property, it generates integers by default:

{
  id: true,
  type: 'Number',
  required: false,
  generated: true // enables auto-generation
}

It is common to use UUIDs as the primary key in PostgreSQL instead of integers. You can enable it with either the following ways:

• use uuid that is generated by your LB application by setting defaultFn: uuid:

  @property({
    id: true,
    type: 'string'
    defaultFn: 'uuid',
    // generated: true,  -> not needed
  })
  id: string;

• use PostgreSQL built-in (extension and) uuid functions:

  @property({
  id: true,
  type: 'String',
  required: false,
  // settings below are needed
  generated: true,
  useDefaultIdType: false,
  postgresql: {
    dataType: 'uuid',
  },
})
  id: string;

The setting uses uuid-ossp extension and uuid_generate_v4() function as default.

If you’d like to use other extensions and functions, you can do:

  @property({
  id: true,
  type: 'String',
  required: false,
  // settings below are needed
  generated: true,
  useDefaultIdType: false,
  postgresql: {
    dataType: 'uuid',
    extension: 'myExtension',
    defaultFn: 'myuuid'
  },
})
  id: string;

WARNING: It is the users’ responsibility to make sure the provided extension and function are valid.

Module Long Term Support Policy

This module adopts the Module Long Term Support (LTS) policy, with the following End Of Life (EOL) dates:

Learn more about our LTS plan in docs.

Running tests

Own instance

If you have a local or remote PostgreSQL instance and would like to use that to run the test suite, use the following command:

• Linux

POSTGRESQL_HOST=<HOST> POSTGRESQL_PORT=<PORT> POSTGRESQL_USER=<USER> POSTGRESQL_PASSWORD=<PASSWORD> POSTGRESQL_DATABASE=<DATABASE> CI=true npm test

• Windows

SET POSTGRESQL_HOST=<HOST> SET POSTGRESQL_PORT=<PORT> SET POSTGRESQL_USER=<USER> SET POSTGRESQL_PASSWORD=<PASSWORD> SET POSTGRESQL_DATABASE=<DATABASE> SET CI=true npm test

Docker

If you do not have a local PostgreSQL instance, you can also run the test suite with very minimal requirements.

• Assuming you have Docker installed, run the following script which would spawn a PostgreSQL instance on your local:

source setup.sh <HOST> <PORT> <USER> <PASSWORD> <DATABASE>

where <HOST>, <PORT>, <USER>, <PASSWORD> and <DATABASE> are optional parameters. The default values are localhost, 5432, root, pass and testdb respectively.

• Run the test:

npm test