feathers-objection

Feathers database adapter for Objection.js, an ORM based on KnexJS SQL query builder for Postgres, Redshift, MSSQL, MySQL, MariaDB, SQLite3, and Oracle.

Installation

npm install --save feathers-objection
npm install --save objection
npm install --save knex

Then add one of the following:

npm install --save pg
npm install --save sqlite3
npm install --save mysql
npm install --save mysql2
npm install --save oracle
npm install --save mssql

If you want to use a MariaDB instance, you can use the mysql driver.

Feathers CLI

Use feathers generate service command to generate a new Objection service.

Documentation

Please refer to the Feathers database adapter documentation for more details or directly at:

• Querying - The common adapter querying mechanism
• Pagination and Sorting - How to use pagination and sorting for the database adapter
• Extending - How to extend a database adapter

Refer to the official Objection.js documention.

It works like the Knex service adapter, except it has all the benefits of the Objection ORM.

Initializing the Library

config/defaults.json

{
  "mysql": {
    "client": "mysql2",
    "connection": {
      "host": "mysql.example.com",
      "user": "root",
      "password": "secret",
      "database": "example"
    }
  }
}

objection.js

const { Model } = require('objection');

module.exports = function(app) {
  const { client, connection } = app.get('mysql');
  const knex = require('knex')({ client, connection, useNullAsDefault: false });

  Model.knex(knex);

  app.set('knex', knex);
};

Service Options

• model (required) - The Objection model definition.

• id (optional, default: model.idColumn or 'id') - The name of the id field property. Use array of strings for composite primary keys.

• events (optional) - List of custom service events sent by this service.

• paginate (optional) - Pagination object containing a default and max page size.

• multi (optional) - Allow create with arrays and update and remove with id null to change multiple items. Can be true for all methods or an array of allowed methods (e.g. [ 'remove', 'create' ]).

• whitelist (optional) - List of additional query operators to allow (e.g. [ '$eager', '$joinRelation' ]).

• schema (optional) - Database schema to use with all the service queries (e.g. public). See withSchema documentation.

Default Query Operators

Starting at version 2.0.0 feathers-objection converts queries securely. If you want to support additional Objection operators, the whitelist service option can contain an array of additional allowed operators. By default, supported operators are:

'$eq',
'$ne',
'$gte',
'$gt',
'$lte',
'$lt',
'$in',
'$nin',
'$like',
'$notLike',
'$ilike',
'$notILike',
'$or',
'$and',
'$sort',
'$not'

Eager Queries

Eager queries is one way of solving the SQL database relational model in Feathers services, instead of relying on hooks.

Service Options

Note that all this eager related options are optional.

• allowedEager - relation expression to limit the allowed eager queries in the service. Defaults to '[]', meaning no eager queries allowed. See allowGraph documentation.

• eagerOptions - options object to use with $eager and $joinEager query operators. See GraphOptions documentation.

• eagerFilters - option to impose compulsory eager filter. It takes an object or array of objects with the following properties:

  • expression - the relation expression that the filter will be applied.
  • filter - the filter function. It uses modifyGraph internally.

Query Operators

• $modify - modifiers allow you to easily reuse snippets of query logic. you can pass arguments and use multiple modifiers. value can be one of the following:

  • String with comma-separated modifier names. e.g. modifier1,modifier2
  • Array or serialized array with modifier name or array of modifier names as the first item. The rest of the array items would be the modifier/s arguments. e.g. ['modifier1', arg1, arg2] or [['modifier1', 'modifier2'], arg1, arg2]
  • Object or serialized object with modifiers as keys and their arguments as values. Set modifier's value to true when it has no arguments. e.g. { modifier1: [arg1, arg2], modifier2: [arg3, arg4], modifier3: true }

  See modify documentation.

• $eager - eager load relations defined in models' relationMappings getter methods. See withGraphFetched documentation.

• $joinRelation - filter based on a relation's field. use with $eager to also fetch the relation. See joinRelated documentation.

• $leftJoinRelation - filter based on a relation's field using LEFT JOIN. use with $eager to also fetch the relation. See leftJoinRelated documentation.

• $joinEager - filter based on a relation's field. See withGraphJoined documentation.

• $modifyEager - filter relation based on a relation's field. does not support JSON fields. e.g. companies.find({ query: { $eager: 'employees', $modifyEager: { employees: { name: 'John' } } } })

• $mergeEager - merge an eager expression to $eager, e.g. companies.find({ query: { $eager: 'employees', $mergeEager: 'ceos' } })

• $allowRefs - allow the usage of ref keyword to reference another field. Reference a relation's field using $joinEager or $joinRelation, e.g. companies.find({ query: { name: 'ref(size)', $allowRefs: true } }), employees.find({ query: { $joinEager: 'company', 'company.name': 'ref(employees.name)', $allowRefs: true } }). See ref documentation.

• $select - add SELECT statement with given array of column names, e.g. ['name', 'ref(jsonb:a)', 'ref(jsonb:a) as a']. See select and FieldExpression documentation.

• $sort - add an order by clause to the query, e.g. query: { $sort: { a: 1, 'b.c': -1, 'ref(jsonb:a)': 1 } }. See FieldExpression documentation.

• $noSelect - skips SELECT queries in create, patch & remove requests. response data will be based on the input data.

• $null - filter based on if a column is NULL with REST support, e.g. companies.find({ query: { ceo: { $null: false } } }), companies.find({ query: { ceo: { $null: 'false' } } })

• $not - filter based on if a query is NOT true. It can be used with an object $not: { name: { $in: ['craig', 'tim'] } } or array $not: [ { $id: 1 }, { $id: 2 } ]

• $between - filter based on if a column value is between range of values

• $notBetween - filter based on if a column value is not between range of values

• $like - filter column value based on a LIKE pattern

• $notLike - filter column value based on a NOT LIKE pattern

• $regexp - filter column value based on a REGEXP pattern

• $notRegexp - filter column value based on a NOT REGEXP pattern

• $ilike - (Postgres) filter column value based on a case-insensitive LIKE pattern

• $notILike - (Postgres) filter column value based on a case-insensitive NOT LIKE pattern

• $iRegexp - (Postgres) filter column value based on a case-insensitive REGEXP pattern

• $notIRegexp - (Postgres) filter column value based on a case-insensitive NOT REGEXP pattern

• $containsKey (Postgres) - filter based on if a column contains a key

• $any (Postgres) - filter based on if a column contains any key from array of strings

• $all (Postgres) - filter based on if a column contains all keys from array of strings

• $contains (Postgres) - filter based on if a column contains all values from array of values

• $contained (Postgres) - filter based on if a column is contained within a serialized object

Params Operators

• transaction - A transaction object. See transaction documentation.

• atomic - when true ensure that multi create or graph insert/upsert success or fail all at once. Under the hood, automatically create a transaction and commit on success or rollback on partial or total failure. Ignored if you added your own transaction object in params.

• mergeAllowEager - Will merge the given expression to the existing expression from the allowEager service option. See allowGraph documentation.

• eagerOptions - Options object to use with $eager and $joinEager query operators. merges on top of the eagerOptions service option. See GraphOptions documentation.

• schema - Database schema to use with the query (e.g. public) See withSchema documentation.

• modifierFiltersResults - when false the total count of a find() query is calculated from the original result set, ignoring the count of any $modify query. The default behaviour is to apply the count of the modifier to the result total, assuming that the modifier may influence the result total by filtering the result set. This can be used to workaround issues with groupBy and the result count. See this issue for a detailed explanation.

Composite primary keys

Composite primary keys can be passed as the id argument using the following methods:

• String with values separated by the idSeparator property (order matter, recommended for REST)
• JSON array (order matter, recommended for internal service calls)
• JSON object (more readable, recommended for internal service calls)

When calling a service method with the id argument, all primary keys are required to be passed.

Service Options

• idSeparator - (optional) separator char to separate composite primary keys in the id argument of get/patch/update/remove external service calls. Defaults to ','.

app.use('/user-todos', service({
  id: ['userId', 'todoId'],
  idSeparator: ','
})

app.service('/user-todos').get('1,2')
app.service('/user-todos').get([1, 2])
app.service('/user-todos').get({ userId: 1, todoId: 2 })

JSON column

JSON column will be automatically converted from and to JS object/array and will be saved as text in unsupported databases. it must be defined in the model class.

Query against a JSON column in PostgresSQL:

app.service('companies').find({
  query: {
    obj: { stringField: 'string' }
  }
});

app.service('companies').find({
  query: {
    obj: { numberField: 1.5 }
  }
});

app.service('companies').find({ 
  query: {
    obj: { numberField: { $gt: 1.5 } }
  }
});

app.service('companies').find({
  query: {
    obj: { 'objectField.object': 'string in obj.objectField.object' }
  }
});

app.service('companies').find({
  query: {
    obj: { 'arrayField(0).object': 'string in obj.arrayField[0].object' }
  }
});

app.service('companies').find({
  query: {
    arr: { '(0).objectField.object': 'string in arr[0].objectField.object' }
  }
});

app.service('companies').find({
  query: {
    obj: { "(field.WithDot)": 'string' }
  }
});

Graph upsert

Arbitrary relation graphs can be upserted (insert + update + delete) using the upsertGraph method. See examples for a better explanation.

Runs on update and patch service methods when id is set. When the data object also contains id, then both must be the same or an error is thrown.

Service Options

• allowedUpsert - relation expression to allow relations to be upserted along with update. Defaults to null, meaning relations will not be automatically upserted unless specified here. See allowGraph documentation.
• upsertGraphOptions - See upsertGraphOptions documentation.
• createUseUpsertGraph - If set to true, Graph Upsert will also be used for .create(data, params) method instead of Graph Insert.

app.use('/companies', service({
  model: Company,
  allowedEager: 'clients',
  allowedUpsert: 'clients'
})

app.service('/companies').update(1, {
  name: 'New Name',
  clients: [{
    id: 100,
    name: 'Existing Client'
  }, {
    name: 'New Client'
  }]
})

In the example above, we are updating the name of an existing company, along with adding a new client which is a relationship for companies. The client without the ID would be inserted and related. The client with the ID will just be updated (if there are any changes at all).

Params Operators

• mergeAllowUpsert - Merge given expression into allowedUpsert.
• mergeUpsertGraphOptions - Merge given options into upsertGraphOptions.

Graph insert

Arbitrary relation graphs can be inserted using the insertGraph method. Provides the ability to relate the inserted object with its associations.

Runs on the .create(data, params) service method.

Service Options

• allowedInsert - relation expression to allow relations to be created along with insert. Defaults to null, meaning relations will not be automatically created unless specified here. See allowGraph documentation.
• insertGraphOptions - See insertGraphOptions documentation.

Params Operators

• mergeAllowInsert - Merge given expression into allowedInsert.
• mergeInsertGraphOptions - Merge given options into insertGraphOptions.

Transaction

Create a transaction object and pass it to series of service calls using the transaction params operator.
Commit the transaction by calling await transaction.trx.commit().
Rollback by calling await transaction.trx.rollback().

Service

users.service.js

const createService = require('feathers-objection');
const createModel = require('../../models/users.model');
const hooks = require('./users.hooks');

module.exports = function(app) {
  const Model = createModel(app);
  const paginate = app.get('paginate');

  const options = {
    model: Model,
    paginate,
    whitelist: ['$eager', '$joinRelation'],
    allowedEager: 'todos'
  };

  app.use('/users', createService(options));

  const service = app.service('users');

  service.hooks(hooks);
};

todos.service.js

const createService = require('feathers-objection');
const createModel = require('../../models/todos.model');
const hooks = require('./todos.hooks');

module.exports = function(app) {
  const Model = createModel(app);
  const paginate = app.get('paginate');

  const options = {
    model: Model,
    paginate,
    whitelist: ['$eager', '$joinRelation'],
    allowedEager: '[user, subtask]',
    eagerFilters: [
      {
        expression: 'subtask',
        filter: function(builder) {
          builder.where('archived', true);
        }
      }
    ]
  };

  app.use('/todos', createService(options));

  const service = app.service('todos');

  service.hooks(hooks);
};

Use eager queries as follows:

// Get all todos and their unfinished tasks
app.service('/todos').find({
  query: {
    $eager: 'subtask(unDone)'
  }
});

// Get all todos of an active user with firstName 'John'
app.service('/todos').find({
  query: {
    'user.firstName': 'John',
    $eager: 'user(active)',
    $joinRelation: 'user(active)'
  }
});

See this article for more information.

Models

Objection requires you to define Models with JSON Schema format for your tables:

users.model.js

const { Model } = require('objection');

class User extends Model {
  static get tableName() {
    return 'user';
  }

  static get jsonSchema() {
    return {
      type: 'object',
      required: ['firstName', 'lastName'],

      properties: {
        id: { type: 'integer' },
        firstName: { type: 'string', maxLength: 45 },
        lastName: { type: 'string', maxLength: 45 },
        status: {
          type: 'string',
          enum: ['active', 'disabled'],
          default: 'active'
        },
        address: {
          type: 'object',
          properties: {
            street: { type: 'string' },
            city: { type: 'string' },
            zipCode: { type: 'string' }
          }
        },
        list: {
          type: 'array',
          maxItems: 3,
          items: { type: 'string' }
        }
      }
    };
  }

  static get relationMappings() {
    const Todo = require('./todos.model')();

    return {
      todos: {
        relation: Model.HasManyRelation,
        modelClass: Todo,
        join: {
          from: 'user.id',
          to: 'todo.userId'
        }
      }
    };
  }

  static get modifiers() {
    return {
      active: builder => {
        builder.where('status', 'active');
      }
    };
  }

  $beforeInsert() {
    this.createdAt = this.updatedAt = new Date().toISOString();
  }

  $beforeUpdate() {
    this.updatedAt = new Date().toISOString();
  }
}

module.exports = function(app) {
  if (app) {
    const db = app.get('knex');

    db.schema
      .hasTable('user')
      .then(exists => {
        if (!exists) {
          db.schema
            .createTable('user', table => {
              table.increments('id');
              table.string('firstName', 45);
              table.string('lastName', 45);
              table.enum('status', ['active', 'disabled']).defaultTo('active');
              table.timestamp('createdAt');
              table.timestamp('updatedAt');
            })
            .then(() => console.log('Created user table'))
            .catch(e => console.error('Error creating user table', e));
        }
      })
      .catch(e => console.error('Error creating user table', e));
  }

  return User;
};

module.exports = User;

todos.model.js

const { Model } = require('objection');

class Todo extends Model {
  static setup(app) {
    this.app = app;
  }

  static get tableName() {
    return 'todo';
  }

  static get jsonSchema() {
    return {
      type: 'object',
      required: ['userId', 'text'],

      properties: {
        id: { type: 'integer' },
        userId: { type: 'integer' },
        text: { type: 'string', maxLength: 500 },
        complete: { type: 'boolean', default: false },
        dueDate: { type: 'string', format: 'date-time' }
      }
    };
  }

  static get relationMappings() {
    const User = require('./users.model')();

    return {
      user: {
        relation: Model.BelongsToOneRelation,
        modelClass: User,
        join: {
          from: 'todo.userId',
          to: 'user.id'
        }
      }
    };
  }

  static get modifiers() {
    const knex = this.app.get('knex');

    return {
      unDone: function(builder) {
        builder.where('complete', false);
      },
      overdue: builder => {
        builder
          .where('complete', false)
          .where('dueDate', '<', knex.fn.now());
      }
    };
  }

  $beforeInsert() {
    this.createdAt = this.updatedAt = new Date().toISOString();
  }

  $beforeUpdate() {
    this.updatedAt = new Date().toISOString();
  }
}

module.exports = function(app) {
  if (app) {
    Todo.setup(app);

    const db = app.get('knex');

    db.schema
      .hasTable('todo')
      .then(exists => {
        if (!exists) {
          db.schema
            .createTable('todo', table => {
              table.increments('id');
              table.integer('userId');
              table.string('text', 500);
              table.boolean('complete');
              table.timestamp('dueDate');
              table.timestamp('createdAt');
              table.timestamp('updatedAt');
            })
            .then(() => console.log('Created todo table'))
            .catch(e => console.error('Error creating todo table', e));
        }
      })
      .catch(e => console.error('Error creating todo table', e));
  }

  return Todo;
};

Complete Example

Here's a complete example of a Feathers server with a todos SQLite service:

$ npm install @feathersjs/feathers @feathersjs/express body-parser feathers-objection objection knex sqlite3

app.js

const feathers = require('@feathersjs/feathers');
const express = require('@feathersjs/express');
const rest = require('@feathersjs/express/rest');
const errorHandler = require('@feathersjs/express/errors');
const bodyParser = require('body-parser');
const createService = require('feathers-objection');
const { Model } = require('objection');

const knex = require('knex')({
  client: 'sqlite3',
  connection: {
    filename: './db.sqlite'
  },
  useNullAsDefault: false
});

// Bind Objection.js
Model.knex(knex);

// Clean up our data. This is optional and is here
// because of our integration tests
knex.schema.dropTableIfExists('todo').then(function() {
  console.log('Dropped todo table');

  // Initialize your table
  return knex.schema.createTable('todo', function(table) {
    console.log('Creating todo table');
    table.increments('id');
    table.string('text');
    table.boolean('complete');
    table.timestamp('createdAt');
    table.timestamp('updatedAt');
  });
});

// Create a feathers instance.
const app = express(feathers())
  // Enable REST services
  .configure(rest())
  // Turn on JSON parser for REST services
  .use(bodyParser.json())
  // Turn on URL-encoded parser for REST services
  .use(bodyParser.urlencoded({ extended: true }));

// Create an Objection Model
class Todo extends Model {
  static get tableName() {
    return 'todo';
  }

  static get jsonSchema() {
    return {
      type: 'object',
      required: ['text'],

      properties: {
        id: { type: 'integer' },
        text: { type: 'string' },
        complete: { type: 'boolean', default: false }
      }
    };
  }

  $beforeInsert() {
    this.createdAt = this.updatedAt = new Date().toISOString();
  }

  $beforeUpdate() {
    this.updatedAt = new Date().toISOString();
  }
}

// Create Objection Feathers service with a default page size of 2 items
// and a maximum size of 4
app.use(
  '/todos',
  createService({
    model: Todo,
    id: 'id',
    paginate: {
      default: 2,
      max: 4
    }
  })
);

// Handle Errors
app.use(errorHandler());

// Start the server
module.exports = app.listen(3030);

console.log('Feathers Todo Objection service running on 127.0.0.1:3030');

Run the example app with npm run example and go to localhost:3030/todos.

You should see an empty array. That's because you don't have any Todos yet, but you now have full CRUD for your new todos service!

DB migrations

Knex Migration CLI can be used to manage DB migrations and to seed a table with mock data.

Error handling

As of version 4.8.0, feathers-objection only throws Feathers Errors with the message.
On the server, the original error can be retrieved through a secure symbol via error[require('feathers-objection').ERROR].

const { ERROR } = require('feathers-objection');

try {
  await objectionService.doSomething();
} catch (error) {
  // error is a FeathersError with just the message
  // Safely retrieve the original error
  const originalError = error[ERROR];
}

As of version 7.0.0, feathers-objection has normalized errors accross all databases supported by Objection, and makes a best-effort attempt to provide reasonable error messages that can be returned directly to the client.

If these error messages do not work for your needs, the original error is still available using the symbol described above.

Migrating to feathers-objection v2

feathers-objection 2.0.0 comes with important security and usability updates

Important: For general migration information to the new database adapter functionality see
crow.docs.feathersjs.com/migrating.html#database-adapters

The following breaking changes have been introduced:

• All methods allow additional query parameters
• Multiple updates are disabled by default (see the multi option)
• Objection related operators are disabled by default (see the whitelist option)

Migrating to feathers-objection v5

feathers-objection 5.0.0 comes with usability updates and was migrated to use Objection v2

Important: For general migration information to Objection v2 see
https://vincit.github.io/objection.js/release-notes/migration.html

The following breaking changes have been introduced:

• $pick query operator was removed
• namedEagerFilters service option was removed. use Model's modifiers instead
• Model's namedFilters property was renamed to modifiers

Migrating to feathers-objection v6

feathers-objection 6.0.0 comes with usability and security updates

• $not operator is now available. It can be used with an object $not: { name: { $in: ["craig", "tim"] } } or array $not: [ { $id: 1 }, { $id: 2 } ]
• $eager is no longer needed with upsert operations

The following breaking changes have been introduced:

• Graph upsert now requires that id fields in the data object will match the id argument
• $noSelect now always return the input data
• $select is now honored with upsert methods
• patch method now enforce params.query with upsert
• NotFound error will be thrown when get & update methods are called with different values in id & params.query.id

Migrating to feathers-objection v7

feathers-objection 7.0.0 comes with improved error handling.

The following breaking changes have been introduced:

• All Databases will return the same types of errors based on the underlying Objection error
• SQL Driver error text is no longer used as the Feathers error message
• Objection errors are mapped more accurately to Feathers errors, e.g.
  • Objection's UniqueViolationError -> Feathers' Conflict error type

License

Copyright © 2020

Licensed under the MIT license.