missy

Slim & sexy, universal ORM

npm install missy
9 downloads in the last month

Missy

Missy is a slim database-agnostic data mapper for NodeJS with pluggable drivers.

Whenever you need a truly flexible, lightweight DB tool - she's here for you.

Quick overview:

  • Database-agnostic. Currently, PostgreSQL and MongoDB are supported
  • Allows to customize the DB client object
  • Full CRUD operations support
  • Custom update/remove operations
  • Easy custom data types
  • Absolutely no limitations on the underlying schema and keys
  • Model events & hooks for full control
  • Rich data selection control: projections, limit/offset, sorting
  • Model relations, even across databases
  • Supports schema-less NoSQL documents with custom fields
  • MongoDB-style API
  • Reliable DB connection handling
  • Promise-based: uses the q package
  • Amazingly simple and well-structured
  • Documented and rich on comments
  • 100% tests coverage

Table Of Contents

Glossary

Commonly used terms:

Driver
Database driver that handles low-level Missy queries like find, insert, save, etc.
Schema
A collection of Models bound to a database Driver. Also, Type Handlers are defined on it.
Type Handler
A class that converts Field values to/from the database
Entity
An entity is a document persisted in the database
Model
A Model defines fields on some DB namespace (table, collection, etc) and provides methods to access Entities.
Field
Fields are defined on a model and specify its name and type. Other options are available
Relation
A relation is a way of accessing the associated entities

Tutorial

Let's peek at how Missy works:

var missy = require('missy'),
    MongodbDriver = require('missy-mongodb')
    ;

// Driver
var mongoDriver = new MongodbDriver(function(){
    // Connecter function

}, {
    // driver options
    journal: true
});

// Schema
var schema = new missy.Schema(mongoDriver, {
    queryWhenConnected: false // when disconnected, queries will wait for the driver to connect
});

// Models
var User = schema.define('User', {
    // Fields definition
    _id: Number, // simple form
    name: { type: 'string', required: true }, // full form
    ctime: { type: 'date', def: function(){ return new Date(); } }, // with a default value
    sex: String,
    age: Number,
    description: String
}, {
    // Model options
    pk: '_id', // Primary key
    // table: 'users' // using the default table name: 'users'
});

var Post = schema.define('Post', {
    _id: Number,
    uid: Number, // author
    title: String,
    body: String,
    tags: Array
}, {
    pk: 'id',
    table: 'user_posts', // the default table name would have been 'posts'. Override this.
    entityPrototype: { // add methods to all entities
        addTag: function(tag){
            this.tags.push(tag);
        }
    }
});

var Comment = schema.define('Comment', {
    _id: Number,
    post_id: Number,
    name: String, // arbitrary user name
    body: String
});

// Relations
User.hasMany('posts', Post, { 'id': 'uid' }); // name, model, fields mapping
Post.hasOne('author', User, { 'uid': 'id' });
Comment.hasOne('post', Post, { 'post_id': '_id' });
Post.hasMany('comments', Comment, { '_id': 'post_id' });

// Model hooks
User.hooks.beforeExport = function(entity){ // adds a synchronous hook
    // validate sex
    if (['m','f'].indexOf(entity.sex) === -1)
        entity.sex = '?';
};

User.hooks.on('afterInsert', function(entities){ // event hook
    // When a new user is inserted, detect teenagers
    if (user.age < 18)
        console.log('teens here!');
});

// Connect & action!
schema.connect()
    .catch(function(){ // DB error
        console.error('DB connection failed');
    })

    // Saving entities
    .then(function(){
        return User
            .withRelated('posts') // save related entities as well
            .save([
                { _id: 1, name: 'Lily', sex: 'f', age: 19, // `ctime` gets the default
                  posts: [ // her posts
                        { // no `uid` field: it automatically gets a value
                            _id: 1, title: 'Help me', body: 'I broke my nail!', tags: ['help']
                        }
                    ]
                 },
                { _id: 2, name: 'Ivy', sex: 'f', age: 21 }, // no posts
                { _id: 3, name: 'Carrie', sex: 'f', age: '23' } // `age` converted to Number on save
            ]);
    })

    // Fetching entities
    .then(function(){ // promise-based success handler callback
        // Now the schema is connected and we can use the models
        return User
            .withRelated('posts') // eagerly load posts
            .withRelated('posts.comments') // deep eager load: also load all comments
            .find(
                { sex: 'f', age: { $gt: 18, $lt: 25 } }, // find girls aged from 18 to 25
                { description: 0 }, // exclude the description field from the result set
                { age: +1 } // sort by `age` ascending
            ); // the result is passed to the next `then()`
    })

    // Removing entities
    .then(function(girls){
        return User
            .with('posts') // with posts
            .remove(girls); // don't try this at home
    })

    // Updating an entity without loading it (!)
    .then(function(){
        return User.updateQuery(
            { sex: 'f', name: 'Lily' }, // match all girls named 'Lily'
            { $set: { description: 'Cute!' } } // set the `description` field to a compliment
        ); // the full updated entities are returned
    })

    // Removing an entity without loading it (!)
    .then(function(){
        return User.removeQuery( // remove all matching entities
            { sex: 'f', name: 'John' } // match all Johns pretending to be a girl
        ); // the removed entities are returned
    });

Core Classes

Most Missy methods use these so you'll need to know how to work with them. This includes:

  • Converting values to/from the database format
  • Specifying field projections
  • Search criteria format
  • Sorting entities
  • Specifying the update expressions

Though most of them are designed after MongoDB, they're applicable to all DB drivers with no limitations.

Converter

Source: lib/util/model.js#Converter

Converter transparently transparently converts field values to/from the DB format.

NOTE: Converter does not touch fields that are not defined in the model! Keep this in mind when working with NoSQL databases as in general documents can contain arbitrary fields.

Consider the example:

var User = schema.define('User', {
    id: Number,
    name: String,
    login: { type: 'string', required: true },
    tags: Array,
    ctime: { type: 'date', required: true, def: function(){ return new Date(); })
};

User.save({ id: '1', name: 111, login: 'first', tags:'events' });
User.findOne({ id: '1' })

On save, the 'id' field is converted to a number, name and login - to string, tags - to array. Also, 'ctime' is set to the current time as no value was not provided.

On find, the query criteria 'id' field is converted to a number, and the resulting entity is converted back.

In general, Converter does the following:

  • Sets default values on fields that are undefined and have def field property set to either a value or a function.
  • Applies type handlers to entity fields defined in the model.

Type Handlers

Converter is driven by type handlers: classes which can convert a JS value to the DB format and back.

Each type handler has a name to be referenced in the model fields definition. Alternatively, you can use a JS built-in object as a shortcut.

Missy defines some built-in type handlers in lib/types/:

Name Shortcut JS Type Default Comment
any undefined * undefined No-op converter to use the value as is
string String String,null '', null Ensure a string, or null
number Number Number,null 0, null Ensure a number, or null
boolean Boolean Boolean,null false, null Ensure a boolean, or null
date Date Date,null null Convert to JS Date, or null
object Object Object,null {}, null Use a JS Object, or null.
array Array Array,null [], null Ensure an array, or null. Creates arrays from scalar values
json - String,null null Un/serializes JSON, or null. Throws MissyTypeError on parse error.

Note: most built-in types allow null value only when the field is not defined as required (see below).

Note: DB drivers may define own type handlers and even redefine standard types for all models handled by the driver.

Also, the following aliases are defined:

Alias Definition
bool { type: 'boolean' }
int { type: 'number' }
float { type: 'number' }

Custom Type Handlers

A type handler is a class that implements the IMissyTypeHandler interface (lib/interfaces.js):

  • Constructor receives 2 arguments: the schema to bind to, and the type handler name.
  • Method load(value, field) which converts a value loaded from the database
  • Method save(value, field) which converts a value to be stored to the database
  • Method norm(value, field) which normalizes the value
  • Has 2 properties: schema and name

Once the type handler is defined, you register in on a Schema:

var stringTypeHandler = require('missy').types.String;
schema.registerType('string', stringTypeHandler);

// Now you can use it on a model:
var User = schema.define('User', {
    login: { type: 'string' }
});

Note: built-in types are registered automatically, there's no need to follow this example.

MissyProjection

Source: lib/util/model.js#MissyProjection

When selecting entities from the DB, you may want to fetch a subset of fields. This is what MissyProjection does: allows you to specify the fields to include or exclude from the resulting entities.

MissyProjection closely follows MongoDB Projections specification.

A projection can be used in one of the 3 available modes:

  • all mode. All available fields are fetched.
  • inclusion mode. Fetch the named fields only.
  • exclusion mode. Fetch all fields except the named ones.

The following projection input formats are supported:

  • String syntax. Projection mode + Comma-separated field names.

    • '*': include all fields
    • '+a,b,c': include only fields a, b, c
    • '-a,b,c': exclude fields a, b, c
  • Array syntax. Array of field names to include.

    • []: include all fields
    • ['a','b','c']: include only fields a, b, c
    • Exclusion mode is not supported
  • Object syntax. MongoDB-style projection object.

    • {}: include all fields
    • { a:1, b:1, c:1 }: include only fields a, b, c
    • { a:0, b:0, c:0 }: exclude fields a, b, c
  • MissyCriteria. A MissyCriteria object.

Usage example:

var User = schema.define('User', {
    id: Number,
    login: String,
    //...
});

User.find({}, { id:1, login: 1 }) // only fetch 2 fields: id, login
    .then(function(users){ ... });

MissyCriteria

Source: lib/util/model.js#MissyCriteria

Specifies the search conditions.

Implements a simplified version of MongoDB collection.find criteria document:

  • Use { field: value, .. } syntax to match entities by equality ;
  • Use {field: { $operator: value, .. }, .. } syntax to compare with an operator.

Example:

{
    id: 1, // equality
    login: { $eq: 'kolypto' }, // equality with an operator
    role: { $in: ['admin','user'] }, // $in operator example
    age: { $gt: 18, $lt: 22 } // 2 operators on a single field
}

Note: To keep the implementation simple and effective, Missy does not support complex queries and logical operators. If you need them, see Using The Driver Directly.

The following operators are supported:

Operator Definition Example Comment
$gt > { money: { $gt: 20000 } } Greater than
$gte <= { height: { $gte: 180 } } Greater than or equal
$eq == { login: { $eq: 'kolypto' } } Equal to
$ne != { name: { $ne: 'Lucy' } } Not equal to
$lt < { age: { $lt: 18 } } Lower than
$lte <= { weight: { $lte: 50 } } Lower than or equal
$in IN { role: { $in: ['adm', 'usr' ] } } In array of values. Scalar operand is converted to array.
$nin NOT IN { state: { $nin: ['init','error'] } } Not in array of values. Scalar operand is converted to array.

Before querying, MissyCriteria uses Converter to convert the given field values to the DB types. For instance, the { id: '1' } criteria will be converted to { id: { $eq: 1 } }.

Example:

var User = schema.define('User', {
    age: Number,
    //...
});

User.find({ age: { $gt: 18, $lt: 22 } }) // 18 <= age <= 22
    .then(function(users){ ... });

MissySort

Source: lib/util/model.js#MissySort

Defines the sort order of the result set.

MissySort closely follows the MongoDB sort specification.

The following sort input formats are supported:

  • String syntax. Comma-separated list of fields suffixed by the sorting operator: + or -.

    • a,b+.c-: sort by a asc, b asc, c desc
  • Array syntax. Same as String syntax, but split into an array.

    • ['a', 'b+', 'c-' ]: sort by a asc, b asc, c desc
  • Object syntax. MongoDB-style object which maps field names to sorting operator: 1 or -1.

    • { a: 1, b: 1, c: -1 }: sort by a asc, b asc, c desc

Example:

var User = schema.define('User', {
    id: Number,
    age: Number,
    //...
});

User.find({}, {}, { age: -1 }) // sort by `age` descending
    .then(function(users){ ... });

MissyUpdate

Source: lib/util/model.js#MissyUpdate

Declares the update operations to perform on matching entities.

MissyUpdate implements the simplified form of MongoDB update document.

  • Use { field: value, .. } syntax to set a field's value ;
  • Use { $operator: { field: value } } syntax to apply a more complex action defined by the operator.

Example:

{
    mtime: new Date(),                      // assign a value
    $set: { mtime: new Date() },            // assign a value
    $inc: { hits: +1 },                     // increment a field
    $unset: { error: '' },                  // unset a field
    $setOnInsert: { ctime: new Date() },    // set on insert, not update
    $rename: { 'current': 'previous' }      // rename a field
}

The following operators are supported:

Operator Comment
$set Set the value of a field
$inc Increment the value of a field by the specified amount. To decrement, use negative amounts
$unset Remove the field (with some drivers: set it to null)
$setOnInsert Set the value of a field only when a new entity is inserted (see upsert with Model.updateQuery)
$rename Rename a field

Before querying, MissyUpdate uses Converter to convert the given field values to the DB types. For instance, the { id: '1' } criteria will be converted to { $set: { id: 1 } }.

Example:

var User = schema.define('User', {
    id: Number,
    mtime: Date,
    //...
});
User.updateQuery({ id: 1 }, { $set: { mtime: new Date(); } }) // update mtime without fetching
    .then(function(user){ ... });

Driver

Missy is a database abstraction and has no DB-specific logic, while Drivers implement the missing part. A Driver is a class that implements the IMissyDriver interface (lib/interfaces.js).

Each driver is created with a connecter function: a function that returns a database client through a promise. Missy does not handle it internally so you can specify all the necessary options and tune the client to your taste.

The first thing you start with is instantiating a driver.

var missy = require('missy');

// For demo, we'll use `MemoryDriver`: it does not require any connecter function at all.
var memory = new missy.drivers.MemoryDriver();

Note: MemoryDriver is built into Missy, but is extremely slow: it's designed for unit-tests and not for production!

At the user level, you don't use the driver directly. However, it has two handy events:

memory.on('connect', function(){
    console.log('Driver connected');
});

memory.on('disconnect', function(){
    console.log('Driver disconnected');
});

Each driver has (at least) the following properties:

  • client:*: The DB client
  • connected:Boolean: Whether the client is currently connected

Supported Drivers

Missy drivers are pluggable: just require another package, and you'll get a new entry under missy.drivers.

Driver Database Package name Github
MemoryDriver in-memory missy built-in
PostgresDriver PostgreSQL missy-postgres https://github.com/kolypto/nodejs-missy-postgres
MongodbDriver MongoDB missy-mongodb https://github.com/kolypto/nodejs-missy-mongodb

Contributions are welcome, provided your driver is covered with unit-tests :)

Schema

Source: lib/Schema.js#Schema

The instantiated driver is useless on its own: you just pass it to the Schema object which is the bridge that connects your Models with the Driver of your choice.

You're free to use multiple schemas with different drivers: Missy is smart enough to handle them all, with no limitations.

Note: a single driver can only be used with a single schema!

var missy = require('missy');

// Create: Driver, Schema
var driver = new missy.drivers.MemoryDriver(),
    schema = new missy.Schema(driver, {});

// Initially, the schema is not connected

schema.connect()
    .then(function(){
        console.log('DB connected!');
    });

Schema(driver, settings?)

Constructor. Creates a schema bound to a driver.

  • driver:IMissyDriver: The driver to work with
  • settings:Object?: Schema settings. An object:

    • queryWhenConnected: Boolean

      Determines how to handle queries on models of a disconnected schema.

      When false (default), querying on a disconnected schema throws MissyDriverError

      When true, the query is delayed until the driver connects.

      Source: lib/options.js#SchemaSettings

Initially, the schema is not connected. Use Schema.connect() to make the driver connect.

A Schema instance has the following properties:

  • driver:IMissyDriver: The driver the schema is bound to
  • settings:Object: Schema settings object
  • types:Object.<String, IMissyTypeHandler>: Custom type handlers defined on the schema
  • models:Object.<String, Model>: Models defined on the schema

Schema.define(name, fields, options?):Model

Defines a model on the schema. The model uses the driver bound to the schema.

Note: you can freely define models on a schema that is not connected.

  • name:String: Model name
  • fields:Object: Model fields definition
  • options:Object?: Model options

See: Model Definition.

schema.define('User', {
    id: Number,
    login: String
}, { pk: 'id' });

Schema.registerType(name, TypeHandler):Schema

Register a custom Type Handler on this schema. This type becomes available to all models defined on the schema.

  • name: String: The type handler name. Use it in model fields: { type: 'name' }.
  • TypeHandler:IMissyTypeHandler: The type handler class to use. Must implement IMissyTypeHandler

See: Custom Type Handlers

Schema.connect():promise

Ask the driver to connect to the database.

Returns a promise.

schema.connect()
    .then(function(){
        // DB connected
    })
    .catch(function(err){
        // DB connection error
    });

Note: once connected, the Schema automatically reconnects when the connection is lost.

Schema.disconnect():promise

Ask the driver to disconnect from the database.

Returns: a promise

schema.disconnect()
    .then(function(){
        // Disconnected
    });

Note: this disables automatic reconnects until you explicitly connect().

Schema.getClient():*

Convenience method to get the vanilla DB client of the underlying driver.

Handy when you need to make some complex query which is not supported by Missy.

See: Using The Driver Directly

Model

Source: lib/Model.js#Model

A Model is the representation of some database namespace: a table, a collection, whatever. It defines the rules for a certain type of entity, including its fields and business-logic.

Once defined, a model has the following properties:

  • schema:Schema: The Schema the Model is defined on
  • name:String: Model name
  • fields:Object.<String, IModelFieldDefinition>: Field definitions
  • options:Object: Model options
  • relations:Object.<String, IMissyRelation>: Relation handlers

Model Definition

To define a Model on a Schema, you use schema.define():

var missy = require('missy');

var driver = new missy.drivers.MemoryDriver(),
    schema = new missy.Schema(driver, {});

var User = schema.define('User', {
    id: Number,
    login: { type: String, required: true }
}, { pk: 'id' });

Schema.define accepts the following arguments:

  • name:String: Model name.
  • fields:Object: Model fields definition
  • options:Object?: Model options

Fields Definition

Source: lib/interfaces.js#IModelFieldDefinition

As stated in the type handlers section, a field definition can be:

  • A native JavaScript type (String, Number, Date, Object, Array)
  • A Field type handler string name: 'string', 'any', etc.
  • An object with the following fields:

    • type:String: Field type handler
    • required:Boolean?: Is the field required?

      A required field is handled differently by type handlers: it is now allowed to contain null and gets some default value determined by the type handler.

      Default value: uses required from model options.

    • def:*|function(this:Model): The default value to use when the field is undefined.

      Can be either a value, or a function that returns a value.

      The default value is used when an entity is being saved to the DB, and the field value is either undefined, or null and having the required=true attribute.

Model Options

Source: lib/options.js#ModelOptions

  • table:String?: the database table name to use.

    Default value: Missy casts the model name to lowercase and adds 's'. For instance, 'User' model will get the default table name of 'users'.

  • pk:String|Array.<String>?: the primary key field name, or an array of fields for compound PK.

    Every model in Missy needs to have a primary key.

    Default value: 'id'

  • required:Boolean?: The default value for fields' required attribute.

    Default value: false

  • entityPrototype: Object?: The prototype of all entities fetched from the database.

    This is the Missy way of adding custom methods to individual entities:

      var Wallet = schema.define('Wallet', {
          uid: Number,
          amount: Number,
          currency: String
      }, {
          pk: 'uid',
          entityPrototype: {
              toString: function(){
                  return this.amount + ' ' + this.currency;
              }
          }
      });
    
      Wallet.findOne(1) // assuming there exists { uid: 1, amount: 100, currency: 'USD' }
          .then(function(wallet){
              console.log(wallet + ''); // -> 100 USD
          });
    

Helpers

The following low-level model methods are available in case you need to manually handle entities that were fetched or are going to be stored by using the driver directly:

Model.getClient():Object

Returns the vanilla DB client from the Schema.

Model.entityImport(entity):Q

Process an entity (or an array of them) loaded from the DB, just like Missy does it:

  • Invoke model hooks (see: Model Hooks)
  • Use Converter to apply field types
  • Assign entity prototypes (if configured)

Returns a promise for an entity (or an array of them).

var Post = schema.define('Post', {
    id: Number,
    title: String,
    tags: Array
});

Post.entityImport({ id: '1', title: 1111, tags: 'test' })
    .then(function(entity){
        console.log(entity); // { id: 1, title: '1111', tags: ['test'] }
    });

Model.entityExport(entity):Q

Process an entity (or an array of them) before saving it to the DB, just like Missy does it:

  • Invoke model hooks (see: Model Hooks)
  • Use Converter to apply field types

Returns a promise for an entity (or an array of them).

Operations

All examples use the following schema:

var User = schema.define('User', {
    id: Number,
    name: String,
    age: Number,
    sex: String
});

All promise-returning methods can return the following error:

  • MissyDriverError: driver-specific error

Read Operations

Model.get(pk, fields?):Q

Get a single entity by its primary key.

Arguments:

  • pk: *|Array|Object: The Primary Key value. For compound primary keys, use array or object of values.

    Accepts the following values:

    • pk: *: Any scalar primary key value. Only applicable for single-column Primary Keys.
    • pk: Array: An array of PK values. Use with compound Primary Keys.
    • pk: Object: PK values as an object.
  • fields: String|Array|Object|MissyProjection?: Fields projection.

Returns: a promise for an entity, or null when the entity is not found.

Errors:

  • MissyModelError: invalid primary key: empty or incomplete.
// single-column PK
var User = schema.define('User', {id: Number }, { pk: 'id' });

User.get(1) // get User(id=1)
    .then(function(entity){ /*...*/ });

User.get([1], ['id', 'login']) // User(id=1) ; only fetch fields `id`, `login`
    .then(function(entity){ /*...*/ });

User.get({ id: 1 }, { id: 1, login: 1 }) // User(id=1) ; only fetch fields `id`, `login`
    .then(function(entity){ /*...*/ });

// multi-column PK
var Post = schema.define('Post', { uid: Number, id: Number }, { pk: ['uid','id'] });

User.get([1, 15]) // get Post(uid=1,id=15)
    .then(function(entity){ /* ... */ });

User.get({ uid: 1, id: 15 }, { roles: 0 }) // get Post(uid=1,id=15) ; omit field `roles`
    .then(function(entity){ /* ... */ });

User.get({ uid: 1 }) // incomplete PK
    .catch(function(e){ /* e=MissyModelError: incomplete PK */ });

Model.findOne(criteria?, fields?, sort?, options?):Q

Find a single entity matching the specified criteria. When multiple entities are found - only the first one is returned.

Arguments:

Returns: a promise for an entity, or null when no entity is found.

User.findOne(
    { age: { $gt: 18, $lt: 21 }, sex: 'f' }, // criteria
    { id: 1, name: 1 }, // projection
    { age: -1 } // sort
).then(function(user){
    //...
});

Model.find(criteria?, fields?, sort?, options?):Q

Find all entities matching the specified criteria.

Arguments:

  • criteria: Object|MissyCriteria?: Search criteria
  • fields: String|Object|MissyProjection?: Fields projection
  • sort: String|Object|Array|MissySort?: Sort specification
  • options: Object?: Driver-specific options, if supported.

    Options supported by all drivers:

    • skip: Number?: The number of entities to skip. Default: 0, no skip
    • limit: Number?: Limit the returned number of entities. Default: 0, no limit

Returns: a promise for an array or entities.

User.find(
    { age: { $gt: 18, $lt: 21 }, sex: 'f' }, // criteria
    { id: 1, name: 1 }, // projection
    { age: -1 }, // sort
    { skip: 0, limit: 10 } // 10 per page, first page
).then(function(users){
    //...
});

Model.count(criteria?, options?):Q

Count entities that match the criteria

Arguments:

  • criteria: Object|MissyCriteria?: Search criteria
  • options: Object?: Driver-specific options, if supported.

Returns: a promise for a number.

User.count({ sex: 'f' })
    .then(function(girlsCount){ /* ... */ });.
});

Write Operations

Model.insert(entities, options?):Q

Insert a new entity (or an array of them).

Arguments:

  • entities: Object|Array.<Object>: a single entity, or an array of them.
  • options: Object?: Driver-specific options, if supported.

Returns: a promise for the inserted entity (or an array of them).

Errors:

  • EntityExists: the entity already exists

Notes:

  • The drivers are required to return entities as saved by the DB, and in the order matching the input.
  • If any entity already exists, the driver throws an exception and stops. Entities that were already inserted are not removed.
// Insert a single entity
User.insert({ login: 'Carrie', age: 23})
    .then(function(entity){
        entity.id; // was set by the DB
    })
    .catch(function(err){
        if (err instanceof missy.errors.EntityExists){
            // ...
        } else throw err; // throw it to the upper chain
    }).done()
    ;

// Insert an array of entities
User.insert([
    { login: 'Carrie', age: 23},
    { login: 'Lily', age: 19},

]).then(function(entities){ /* ... */ });

Model.update(entities, options?):Q

Update (replace) an existing entity (or an array of them).

Arguments:

  • entities: Object|Array.<Object>: a single entity, or an array of them.
  • options: Object?: Driver-specific options, if supported.

Returns: a promise for the updated entity (or an array of them).

Errors:

  • EntityNotFound: the entity does not exist
// Insert a single entity
User.update({ _id: 1, login: 'Carrie', age: 23})
    .then(function(entity){ /* ... */ });

Model.save(entities, options?):Q

Save an arbitrary entity (or an array of them). This inserts the missing entities and updates the existing ones.

Arguments:

  • entities: Object|Array.<Object>: a single entity, or an array of them.
  • options: Object?: Driver-specific options, if supported.

Returns: a promise for the entity (or an array of them).

User.save({ login: 'Carrie', age: 23})
    .then(function(entity){
        entity.id; // was set by the DB
    })

Model.remove(entities, options?):Q

Remove an existing entity (or an array of them) from the database.

Arguments:

  • entities: Object|Array.<Object>: a single entity, or an array of them.
  • options: Object?: Driver-specific options, if supported.

Returns: a promise for the entity (or an array of them).

Errors:

  • EntityNotFound: the entity does not exist
User.remove({ _id: 1 }) // PK is enough
    .then(function(entity){
        entity; // the full removed entity
    });

Queries

Model.updateQuery(criteria, update, options?):Q

Update entities that match a criteria using update operators. This tool allows to update entities without actually fetching them. DB drivers do this atomically, if possible.

Arguments:

  • criteria: Object|MissyCriteria: Search criteria
  • update: Object|MissyUpdate: Update operations
  • options: Object?: Driver-specific options, if supported.

    Options supported by all drivers:

    • upsert: Boolean?: Do an upsert: when no entity matches the criteria, a new entity is built from the criteria (equality operators converted to values) and update operators applied to it. Default: false
    • multi: Boolean?: Allow updating multiple entities. In multi=true mode, the method returns an array. Default: false

Returns: a promise for a full updated entity or null when no matching entity is found. When multi=true, returns an array of entities.

// update a single entity
User.updateQuery({ name: 'Ivy' }, { $inc: { age: 1 } ) // find 'Ivy' and add a year
    .then(function(user){ // 1 user
        user; // the full entity, or `null` when none found
    });

// update multiple entities
User.updateQuery({ age: { $lt: 18 } }, { banned: true }, { multi: true } ) // find teens and ban them
    .then(function(users){ // multiple users
        users; // array of matching users. Empty array when none found
    });

// upsert an entity
Users.updateQuery({ name: 'Dolly' }, { age: 22 }, { upsert: true })
    .then(function(user){ // 1 user
        user; // { id: 1, name: 'Dolly', age: 22 } - either updated or inserted
    });

Model.removeQuery(criteria, options?):Q

Remove entities that match a criteria. This tool allows to remove entities without actually fetching them. DB drivers do this atomically, if possible.

Arguments:

  • criteria: Object|MissyCriteria: Search criteria
  • options: Object?: Driver-specific options, if supported.

    Options supported by all drivers:

    • multi: Boolean?: Allow removing multiple entities. In multi=true mode, the method returns an array. Default: true

Returns: a promise for a removed entity or null when no matching entity is found. When multi=true, returns an array of entities.

// remove a single entity
User.removeQuery({ id: 1 }, { multi: false })
    .then(function(user){
        user; // the removed user or `null` when none found
    });

// remove multiple entities
User.removeQuery({ age: { $lt: 18 } }) // remove teens
    .then(function(users){ // multiple users
        users; // array of removed users. Empty array when none found
    });

Chaining

Some Missy methods support chaining: the specified arguments are stashed for the future and used by the target method.

Currently, the following methods are supported: Model.get(), Model.findOne(), Model.find().

Model.pick(fields)

Stash the fields argument for the next query.

User.pick({ id: 1, name: 1}).find();

Model.sort(sort)

Stash the sort argument for the next query.

User.sort({ id: -1 }).find();

Model.skip(n)

Stash the skip option for the next query.

User.skip(10).find();

Model.limit(n)

Stash the limit option for the next query.

User.limit(10).find();

Using The Driver Directly

Missy search criteria is limited in order to keep the implementation simple. To make complex queries, you'll need to use the Driver directly, and optionally pass the returned entities through Missy.

In order to mimic Missy behavior, you need to do the following:

  1. Get the vanilla DB client object from the Schema or Model.
  2. Execute a query using the client
  3. Use Missy methods to preprocess the data. This includes the hooks!

MongoDB Example:

var Q = require('q');

var client = User.getClient(); // get the vanilla DB client

// Load an entity
Q.nmcall(
    client.collection(User.options.table),
    'findOne', // method, wrapped in a promise
    {
        $or: [
            // find any root user
            { role: 'root' },
            { id: 0 }
        ]
    }
    ).then(function(entity){
        return User.entityImport(entity); // process the loaded value with Missy
    })
    .then(function(entity){
        entity; // converted from DB format!
    });

// Save an entity
var entity = { id: '0', age: '23' }; // wrong field types!

User.entityExport(entity)
    .then(function(entity){
        entity; // { id: 0, age: 23 } - converted!
        return Q.nmcall(
            client.collection(User.options.table),
            'save',
            entity
        );
    });

Model Hooks

Missy Models allow you to hook into the internal processes, which allow you to preprocess the data, tune the internal behavior, or just tap the data.

Hooks are implemented with MissyHooks which is instantiated under the Model.hooks property of every mode. It supports both synchronous hooks and event hooks.

You can assign synchronous hooks which are executed within the Missy and can interfere with the process:

// Add a hook
User.hooks.beforeInsert = function(entities, ctx){
    _.each(entities, function(entity){
        entity.ctime = new Date(); // set the creation date
    });
};

Also, you can subscribe to events named after the available hooks:

User.hooks.on('afterInsert', function(entities, ctx){
    _.each(entities, function(entity){
        console.log('New user created: ', entity);
    });
});

Almost every Missy method is integrated with the hook system.

Converter Hooks

Allow you to hook into entityImport() and entityExport() functions and alter the way entities are preprocessed while Missy talks to database driver.

Hook name Arguments Called in Usage example
beforeImport entity:Object In entityImport() right after fetching the entity from the DB. Prepare values fetched from the DB
afterImport entity:Object In entityImport() after the field type convertions are applied to the entity. Final tuning before the entity is returned to the user
beforeExport entity:Object In entityExport() right after the entity is provided by the Missy user Sanitize the data
afterExport entity:Object In entityExport() after the field type convertions are applied to the entity. Prepare the data before storing it to the DB ; Validation
User.afterExport = function(entity){
    delete entity.password; // never expose the password
};

Query Hooks

Allow you to hook into Missy query methods and alter the way these are executed, including all the input/output values.

All hooks receive the ctx: IModelContext argument: an object that holds the current query context. In hooks, you can modify its fields to alter the Missy behavior. ctx fields set depends on the query, but in general:

  • ctx.model: Model: The model the query is executed on
  • ctx.criteria: MissyCriteria?: Search criteria
  • ctx.fields: MissyProjection?: Fields projection
  • ctx.sort: MissySort?: Sorting
  • ctx.update: MissyUpdate?: Update operations
  • ctx.options: Object?: Driver-dependent options
  • ctx.entities: Array.<Object>?: The entities being handled (fetched or returned)

Each missy method invokes its own pair of before* and after* hooks. The after* hook is not invoked when an error occurs.

In addition, methods that accept/return entities invoke entityImport() and entityExport() on every entity, which in turn triggers the converter hooks described above.

Hook name Arguments Model method
beforeFindOne entity:undefined, ctx: IModelContext findOne()
afterFindOne entity:Object, ctx: IModelContext findOne()
beforeFind entities:undefined, ctx: IModelContext find()
afterFind entities:Array.<Object>, ctx: IModelContext find()
beforeInsert entities:Array.<Object>, ctx: IModelContext insert()
afterInsert entities:Array.<Object>, ctx: IModelContext insert()
beforeUpdate entities:Array.<Object>, ctx: IModelContext update()
afterUpdate entities:Array.<Object>, ctx: IModelContext update()
beforeSave entities:Array.<Object>, ctx: IModelContext save()
afterSave entities:Array.<Object>, ctx: IModelContext save()
beforeRemove entities:Array.<Object>, ctx: IModelContext remove()
afterRemove entities:Array.<Object>, ctx: IModelContext remove()
beforeUpdateQuery entities:undefined, ctx: IModelContext updateQuery()
afterUpdateQuery entities:Array.<Object>, ctx: IModelContext updateQuery()
beforeRemoveQuery entities:undefined, ctx: IModelContext removeQuery()
afterRemoveQuery entities:Array.<Object>, ctx: IModelContext removeQuery()

Relations

Missy supports automatic loading & saving of related entities assigned to the host entities.

Defining Relations

Model.hasOne(prop, foreign, fields)

Define a 1-1 or N-1 relation to a foreign Model foreign, stored in the local field prop.

Arguments:

  • prop: String: Name of the local property to handle the related entity in. This also becomes the name of the relation.
  • foreign: Model: The foreign Model
  • fields: String|Array.<String>|Object: Name of the common Primary Key field, or an array of common fields, or an object with the fields' mapping:

     Article.hasOne('author', User, 'user_id');
     Article.hasOne('author', User, { 'user_id': 'id' });
    

After a relation was defined, the local model's prop field will be used for loading & saving the related entity.

Model.hasMany(prop, foreign, fields)

Define a 1-N relation to a foreign Model.

Same as hasOne, but handles an array of related entities.

Model.loadRelated(entities, prop, fields?, sort?, options?):Q

For the given entities, load their related entities as defined by the prop relation.

Arguments:

  • entities: Object|Array.<Object>: Entity of the current model, or an array of them
  • prop: String|Array.<String>|undefined: The relation name to load, or multiple relation names as an array.

    When undefined is given, all available relations are loaded.

    You can also load nested relations using the '.'-notation: 'articles.comments' (see Model.withRelated).

  • fields: String|Object|MissyProjection?: Fields projection for the related entities. Optional.

    With the help of this field, you can load partial related entities.

  • sort: String|Object|Array|MissySort?: Sort specification for the related entities. Optional.

  • options: Object?: Driver-dependent options for the related Model.find() method. Optional.

Relations are effectively loaded with a single query per relation, irrespective to the number of host entities.

After the method is executed, all entities will have the prop property populated with the related entities:

  • For hasOne, this is a single entity, or undefined when no related entity exists.
  • For hasMany, this is always an array, possibly - empty.

Model.saveRelated(entities, prop, options?):Q

For the given entities, save their related entities as defined by the prop relation.

Arguments:

  • entities: Object|Array.<Object>: Entity of the current model, or an array of them
  • prop: String|Array.<String>|undefined: The relation name to save, or multiple relation names as an array.

    When undefined is given, all available relations are saved.

    You can also save nested relations using the '.'-notation: 'articles.comments' (see Model.withRelated).

  • options: Object?: Driver-dependent options for the related Model.save() method. Optional.

This method automatically sets the foreign keys on the related entities and saves them to the DB.

Model.removeRelated(entities, prop, options?):Q

For the given entities, remove their related entities as defined by the prop relation.

Arguments:

  • entities: Object|Array.<Object>: Entity of the current model, or an array of them
  • prop: String|Array.<String>|undefined: (same as above)
  • options: Object?: Driver-dependent options for the related Model.removeQuery() method. Optional.

This method removes all entities that are related to this one with the specified relation.

Model.withRelated(prop, ...):Model

Automatically process the related entities with the next query:

    • find(), findOne(): load related entities
    • insert(), update(), save(): save related entities (replaces them & removes the missing ones)
    • remove(): remove related entities

In fact, this method just stashes the arguments for loadRelated(), saveRelated(), removeRelated(), and calls the corresponding method in the subsequent query:

  • Model.withRelated(prop, fields, sort, options) when going to load entities
  • Model.withRelated(prop, options) when going to save entities
  • Model.withRelated(prop, options) when going to remove entities

See examples below.

Example

For instance, having the following schema:

var User = schema.define('User', {
    id: Number,
    login: String
});

var Article = schema.define('Article', {
    id: Number,
    user_id: Number,
    title: String
    text: String
});

var Comment = schema.define('Comment', {
    id: Number,
    user_id: Number,
    article_id: Number,
    ctime: Date,
    text: String,
});

// Define relations
User.hasMany('articles', Article, {'id': 'user_id'});
Article.hasMany('comments', Comment, { 'id' : 'article_id' });

Article.hasOne('author', User, {'user_id': 'id'});
Comment.hasOne('article', Article, {'article_id': 'id'});
Comment.hasOne('author', User, {'user_id': 'id'});
User
    .withRelated('articles') // process the named relation in the subsequent query
    .save([
        {
            login: 'dizzy',
            articles: [
                // When Dizzy gets an id, the related entities will use it
                { title: 'First post', text: 'Welcome to my page' },
                { title: 'Second post', text: 'Welcome to my page' },
            ]
        }
    ])
    .then(function(){
        // 'dizzy' saved, with 2 posts
    });
User
    .withRelated('articles') // load articles into the `articles` field
    .withRelated('articles.comments', {}, { ctime: -1 }) // load comments for each article, sorted (nested relation)
    .find({ age: { $gt: 18 } })
    .then(function(users){
        users[0]; // user
        users[0].articles; // her articles
        users[0].articles[0].comments; // comments for each article
    });
User
    .withRelated('articles')
    .remove({ id: 1 })
    .then(function(user){
        // User with id=1 removed, as well as her articles
    });

Recipes

Validation

Missy does not support any validation out of the box, but you're free to choose any external validation engine.

The current approach is to install the validation procedure into the beforeExport Model hook and check entities that are saved or updated. Note that this approach won't validate entities modified with Model.updateQuery, as this method does not handle full entities!

An example is on its way.

npm loves you