orm

NodeJS Object-relational mapping

npm install orm

Supported: 0.8, 0.10, 0.12, iojs-1.5

Tests are run on Travis CI for node versions 0.10.x, 0.12.x and iojs-1.5. If you want you can run tests locally:

npm test
  • MySQL & MariaDB
  • PostgreSQL
  • Amazon Redshift
  • SQLite
  • MongoDB (beta, missing aggregation for now)
  • Create Models, sync, drop, bulk create, get, find, remove, count, aggregated functions
  • Create Model associations, find, check, create and remove
  • Define custom validations (several builtin validations, check instance properties before saving - see enforce for details)
  • Model instance caching and integrity (table rows fetched twice are the same object, changes to one change all)
  • Plugins: MySQL FTS , Pagination , Transaction, Timestamps, Migrations

This is a node.js object relational mapping module.

An example:

var orm = require("orm");
 
orm.connect("mysql://username:password@host/database", function (errdb) {
  if (err) throw err;
 
    var Person = db.define("person", {
        name      : String,
        surname   : String,
        age       : Number, // FLOAT 
        male      : Boolean,
        continent : [ "Europe", "America", "Asia", "Africa", "Australia", "Antartica" ], // ENUM type 
        photo     : Buffer, // BLOB/BINARY 
        data      : Object // JSON encoded 
    }, {
        methods: {
            fullNamefunction () {
                return this.name + ' ' + this.surname;
            }
        },
        validations: {
            age: orm.enforce.ranges.number(18, undefined, "under-age")
        }
    });
 
    Person.find({ surname: "Doe" }, function (errpeople) {
        // SQL: "SELECT * FROM person WHERE surname = 'Doe'" 
 
        console.log("People found: %d", people.length);
        console.log("First person: %s, age %d", people[0].fullName(), people[0].age);
 
        people[0].age = 16;
        people[0].save(function (err) {
            // err.msg = "under-age"; 
        });
    });
});

You can use the promise enabled wrapper library.

If you're using Express, you might want to use the simple middleware to integrate more easily.

var express = require('express');
var orm = require('orm');
var app = express();
 
app.use(orm.express("mysql://username:password@host/database", {
    definefunction (dbmodelsnext) {
        models.person = db.define("person", { ... });
        next();
    }
}));
app.listen(80);
 
app.get("/", function (reqres) {
    // req.models is a reference to models used above in define() 
    req.models.person.find(...);
});

You can call orm.express more than once to have multiple database connections. Models defined across connections will be joined together in req.models. Don't forget to use it before app.use(app.router), preferably right after your assets public folder(s).

See examples/anontxt for an example express based app.

Documentation is moving to the wiki.

See information in the wiki.

See information in the wiki.

A Model is an abstraction over one or more database tables. Models support associations (more below). The name of the model is assumed to match the table name.

Models support behaviours for accessing and manipulating table data.

See information in the wiki.

See information in the wiki.

Are passed in during model definition.

var Person = db.define('person', {
    name    : String,
    surname : String
}, {
    methods: {
        fullNamefunction () {
            return this.name + ' ' + this.surname;
        }
    }
});
 
Person.get(4, function(errperson) {
    console.log( person.fullName() );
})

Are defined directly on the model.

var Person = db.define('person', {
    name    : String,
    height  : { type: 'integer' }
});
Person.tallerThan = function(heightcallback) {
    this.find({ height: orm.gt(height) }, callback);
};
 
Person.tallerThan( 192, function(errtallPeople) { ... } );

Models can be in separate modules. Simply ensure that the module holding the models uses module.exports to publish a function that accepts the database connection, then load your models however you like.

Note - using this technique you can have cascading loads.

// your main file (after connecting) 
db.load("./models", function (err) {
    // loaded! 
    var Person = db.models.person;
    var Pet    = db.models.pet;
});
 
// models.js 
module.exports = function (dbcb) {
    db.load("./models-extra", function (err) {
        if (err) {
            return cb(err);
        }
 
        db.define('person', {
            name : String
        });
 
        return cb();
    });
};
 
// models-extra.js 
module.exports = function (dbcb) {
    db.define('pet', {
        name : String
    });
 
    return cb();
};

See information in the wiki.

See information in the wiki.

ORM2 allows you some advanced tweaks on your Model definitions. You can configure these via settings or in the call to define when you setup the Model.

For example, each Model instance has a unique ID in the database. This table column is added automatically, and called "id" by default.
If you define your own key: true column, "id" will not be added:

var Person = db.define("person", {
    personId : { type: 'serial', key: true },
    name     : String
});
 
// You can also change the default "id" property name globally: 
db.settings.set("properties.primary_key", "UID");
 
// ..and then define your Models 
var Pet = db.define("pet", {
    name : String
});

Pet model will have 2 columns, an UID and a name.

It's also possible to have composite keys:

var Person = db.define("person", {
    firstname : { type: 'string', key: true },
    lastname  : { type: 'string', key: true }
});

Other options:

  • cache : (default: true) Set it to false to disable Instance cache (Singletons) or set a timeout value (in seconds);
  • autoSave : (default: false) Set it to true to save an Instance right after changing any property;
  • autoFetch : (default: false) Set it to true to fetch associations when fetching an instance from the database;
  • autoFetchLimit : (default: 1) If autoFetch is enabled this defines how many hoops (associations of associations) you want it to automatically fetch.

See information in the wiki.

To get a specific element from the database use Model.get.

Person.get(123, function (errperson) {
    // finds person with id = 123 
});

Finding one or more elements has more options, each one can be given in no specific parameter order. Only options has to be after conditions (even if it's an empty object).

Person.find({ name: "John", surname: "Doe" }, 3, function (errpeople) {
    // finds people with name='John' AND surname='Doe' and returns the first 3 
});

If you need to sort the results because you're limiting or just because you want them sorted do:

Person.find({ surname: "Doe" }, "name", function (errpeople) {
    // finds people with surname='Doe' and returns sorted by name ascending 
});
Person.find({ surname: "Doe" }, [ "name", "Z" ], function (errpeople) {
    // finds people with surname='Doe' and returns sorted by name descending 
    // ('Z' means DESC; 'A' means ASC - default) 
});

There are more options that you can pass to find something. These options are passed in a second object:

Person.find({ surname: "Doe" }, { offset: 2 }, function (errpeople) {
    // finds people with surname='Doe', skips the first 2 and returns the others 
});

You can also use raw SQL when searching. It's documented in the Chaining section below.

If you just want to count the number of items that match a condition you can just use .count() instead of finding all of them and counting. This will actually tell the database server to do a count (it won't be done in the node process itself).

Person.count({ surname: "Doe" }, function (errcount) {
    console.log("We have %d Does in our db", count);
});

Similar to .count(), this method just checks if the count is greater than zero or not.

Person.exists({ surname: "Doe" }, function (errexists) {
    console.log("We %s Does in our db", exists ? "have" : "don't have");
});

If you need to get some aggregated values from a Model, you can use Model.aggregate(). Here's an example to better illustrate:

Person.aggregate({ surname: "Doe" }).min("age").max("age").get(function (errminmax) {
    console.log("The youngest Doe guy has %d years, while the oldest is %d", min, max);
});

An Array of properties can be passed to select only a few properties. An Object is also accepted to define conditions.

Here's an example to illustrate how to use .groupBy():

//The same as "select avg(weight), age from person where country='someCountry' group by age;" 
Person.aggregate(["age"], { country: "someCountry" }).avg("weight").groupBy("age").get(function (errstats) {
    // stats is an Array, each item should have 'age' and 'avg_weight' 
});
  • .limit(): you can pass a number as a limit, or two numbers as offset and limit respectively
  • .order(): same as Model.find().order()
  • min
  • max
  • avg
  • sum
  • count (there's a shortcut to this - Model.count)

There are more aggregate functions depending on the driver (Math functions for example).

If you prefer less complicated syntax you can chain .find() by not giving a callback parameter.

Person.find({ surname: "Doe" }).limit(3).offset(2).only("name", "surname").run(function (errpeople) {
    // finds people with surname='Doe', skips first 2 and limits to 3 elements, 
    // returning only 'name' and 'surname' properties 
});

If you want to skip just one or two properties, you can call .omit() instead of .only.

Chaining allows for more complicated queries. For example, we can search by specifying custom SQL:

Person.find({ age: 18 }).where("LOWER(surname) LIKE ?", ['dea%']).all( ... );

It's bad practice to manually escape SQL parameters as it's error prone and exposes your application to SQL injection. The ? syntax takes care of escaping for you, by safely substituting the question mark in the query with the parameters provided. You can also chain multiple where clauses as needed.

You can also order or orderRaw:

Person.find({ age: 18 }).order('-name').all( ... );
// see the 'Raw queries' section below for more details 
Person.find({ age: 18 }).orderRaw("?? DESC", ['age']).all( ... );

You can also chain and just get the count in the end. In this case, offset, limit and order are ignored.

Person.find({ surname: "Doe" }).count(function (errpeople) {
    // people = number of people with surname="Doe" 
});

Also available is the option to remove the selected items. Note that a chained remove will not run any hooks.

Person.find({ surname: "Doe" }).remove(function (err) {
    // Does gone.. 
});

You can also make modifications to your instances using common Array traversal methods and save everything in the end.

Person.find({ surname: "Doe" }).each(function (person) {
    person.surname = "Dean";
}).save(function (err) {
    // done! 
});
 
Person.find({ surname: "Doe" }).each().filter(function (person) {
    return person.age >= 18;
}).sort(function (person1person2) {
    return person1.age < person2.age;
}).get(function (people) {
    // get all people with at least 18 years, sorted by age 
});

Of course you could do this directly on .find(), but for some more complicated tasks this can be very usefull.

Model.find() does not return an Array so you can't just chain directly. To start chaining you have to call .each() (with an optional callback if you want to traverse the list). You can then use the common functions .filter(), .sort() and .forEach() more than once.

In the end (or during the process..) you can call:

  • .count() if you just want to know how many items there are;
  • .get() to retrieve the list;
  • .save() to save all item changes.

Conditions are defined as an object where every key is a property (table column). All keys are supposed to be concatenated by the logical AND. Values are considered to match exactly, unless you're passing an Array. In this case it is considered a list to compare the property with.

{ col1: 123, col2: "foo" } // `col1` = 123 AND `col2` = 'foo' 
{ col1: [ 1, 3, 5 ] } // `col1` IN (1, 3, 5) 

If you need other comparisons, you have to use a special object created by some helper functions. Here are a few examples to describe it:

{ col1: orm.eq(123) } // `col1` = 123 (default) 
{ col1: orm.ne(123) } // `col1` <> 123 
{ col1: orm.gt(123) } // `col1` > 123 
{ col1: orm.gte(123) } // `col1` >= 123 
{ col1: orm.lt(123) } // `col1` < 123 
{ col1: orm.lte(123) } // `col1` <= 123 
{ col1: orm.between(123, 456) } // `col1` BETWEEN 123 AND 456 
{ col1: orm.not_between(123, 456) } // `col1` NOT BETWEEN 123 AND 456 
{ col1: orm.like(12 + "%") } // `col1` LIKE '12%' 
{ col1: orm.not_like(12 + "%") } // `col1` NOT LIKE '12%' 
{ col1: orm.not_in([1, 4, 8]) } // `col1` NOT IN (1, 4, 8) 
db.driver.execQuery("SELECT id, email FROM user", function (errdata) { ... })
 
// You can escape identifiers and values. 
// For identifier substitution use: ?? 
// For value substitution use: ? 
db.driver.execQuery(
  "SELECT user.??, user.?? FROM user WHERE user.?? LIKE ? AND user.?? > ?",
  ['id', 'name', 'name', 'john', 'id', 55],
  function (errdata) { ... }
)
 
// Identifiers don't need to be scaped most of the time 
db.driver.execQuery(
  "SELECT user.id, user.name FROM user WHERE user.name LIKE ? AND user.id > ?",
  ['john', 55],
  function (errdata) { ... }
)

Model instances are cached. If multiple different queries will result in the same result, you will get the same object. If you have other systems that can change your database (or you're developing and need to make some manual changes) you should remove this feature by disabling cache. This can be done when you're defining the Model.

var Person = db.define('person', {
    name    : String
}, {
    cache   : false
});

and also globally:

orm.connect('...', function(errdb) {
  db.settings.set('instance.cache', false);
});

The cache can be configured to expire after a period of time by passing in a number instead of a boolean. The number will be considered the cache timeout in seconds (you can use floating point).

Note: One exception about Caching is that it won't be used if an instance is not saved. For example, if you fetch a Person and then change it, while it doesn't get saved it won't be passed from Cache.

To insert new elements to the database use Model.create.

Person.create([
    {
        name: "John",
        surname: "Doe",
        age: 25,
        male: true
    },
    {
        name: "Liza",
        surname: "Kollan",
        age: 19,
        male: false
    }
], function (erritems) {
    // err - description of the error or null 
    // items - array of inserted items 
});

Every item returned has the properties that were defined to the Model and also a couple of methods you can use to change each item.

Person.get(1, function (errJohn) {
    John.name = "Joe";
    John.surname = "Doe";
    John.save(function (err) {
        console.log("saved!");
    });
});

Updating and then saving an instance can be done in a single call:

Person.get(1, function (errJohn) {
    John.save({ name: "Joe", surname: "Doe" }, function (err) {
        console.log("saved!");
    });
});

If you want to remove an instance, just do:

// you could do this without even fetching it, look at Chaining section above 
Person.get(1, function (errJohn) {
    John.remove(function (err) {
        console.log("removed!");
    });
});

See information in the wiki.

An association is a relation between one or more tables.

Is a many to one relationship. It's the same as belongs to.
Eg: Animal.hasOne('owner', Person).
Animal can only have one owner, but Person can have many animals.
Animal will have the owner_id property automatically added.

The following functions will become available:

animal.getOwner(function..)         // Gets owner 
animal.setOwner(person, function..) // Sets owner_id 
animal.hasOwner(function..)         // Checks if owner exists 
animal.removeOwner()                // Sets owner_id to 0 

Chain Find

The hasOne association is also chain find compatible. Using the example above, we can do this to access a new instance of a ChainFind object:

Animal.findByOwner({ /* options */ })

Reverse access

Animal.hasOne('owner', Person, {reverse: 'pets'})

will add the following:

// Instance methods 
person.getPets(function..)
person.setPets(cat, function..)
 
// Model methods 
Person.findByPets({ /* options */ }) // returns ChainFind object 

Is a many to many relationship (includes join table).
Eg: Patient.hasMany('doctors', Doctor, { why: String }, { reverse: 'patients', key: true }).
Patient can have many different doctors. Each doctor can have many different patients.

This will create a join table patient_doctors when you call Patient.sync():

column nametype
patient_idInteger (composite key)
doctor_idInteger (composite key)
whyvarchar(255)

The following functions will be available:

patient.getDoctors(function..)           // List of doctors 
patient.addDoctors(docs, function...)    // Adds entries to join table 
patient.setDoctors(docs, function...)    // Removes existing entries in join table, adds new ones 
patient.hasDoctors(docs, function...)    // Checks if patient is associated to specified doctors 
patient.removeDoctors(docs, function...) // Removes specified doctors from join table 
 
doctor.getPatients(function..)
etc...
 
// You can also do: 
patient.doctors = [doc1, doc2];
patient.save(...)

To associate a doctor to a patient:

patient.addDoctor(surgeon, {why: "remove appendix"}, function(err) { ... } )

which will add {patient_id: 4, doctor_id: 6, why: "remove appendix"} to the join table.

This accessor in this type of association returns a ChainFind if not passing a callback. This means you can do things like:

patient.getDoctors().order("name").offset(1).run(function (errdoctors), {
    // ... all doctors, ordered by name, excluding first one 
});

If you want to split maybe optional properties into different tables or collections. Every extension will be in a new table, where the unique identifier of each row is the main model instance id. For example:

var Person = db.define("person", {
    name : String
});
var PersonAddress = Person.extendsTo("address", {
    street : String,
    number : Number
});

This will create a table person with columns id and name. The extension will create a table person_address with columns person_id, street and number. The methods available in the Person model are similar to an hasOne association. In this example you would be able to call .getAddress(cb), .setAddress(Address, cb), ..

Note: you don't have to save the result from Person.extendsTo. It returns an extended model. You can use it to query directly this extended table (and even find the related model) but that's up to you. If you only want to access it using the original model you can just discard the return.

If you have a relation of 1 to n, you should use hasOne (belongs to) association.

var Person = db.define('person', {
    name : String
});
var Animal = db.define('animal', {
    name : String
});
Animal.hasOne("owner", Person); // creates column 'owner_id' in 'animal' table 
 
// get animal with id = 123 
Animal.get(123, function (erranimal) {
    // animal is the animal model instance, if found 
    animal.getOwner(function (errperson) {
        // if animal has really an owner, person points to it 
    });
});

You can mark the owner_id field as required in the database by specifying the required option:

Animal.hasOne("owner", Person, { required: true });

If a field is not required, but should be validated even if it is not present, then specify the alwaysValidate option. (this can happen, for example when validation of a null field depends on other fields in the record)

Animal.hasOne("owner", Person, { required: false, alwaysValidate: true });

If you prefer to use another name for the field (owner_id) you can change this parameter in the settings.

db.settings.set("properties.association_key", "{field}_{name}"); // {name} will be replaced by 'owner' and {field} will be replaced by 'id' in this case 

Note: This has to be done before the association is specified.

The hasMany associations can have additional properties in the association table.

var Person = db.define('person', {
    name : String
});
Person.hasMany("friends", {
    rate : Number
}, {}, { key: true });
 
Person.get(123, function (errJohn) {
    John.getFriends(function (errfriends) {
        // assumes rate is another column on table person_friends 
        // you can access it by going to friends[N].extra.rate 
    });
});

If you prefer you can activate autoFetch. This way associations are automatically fetched when you get or find instances of a model.

var Person = db.define('person', {
  name : String
});
Person.hasMany("friends", {
    rate : Number
}, {
    key       : true, // Turns the foreign keys in the join table into a composite key 
    autoFetch : true
});
 
Person.get(123, function (errJohn) {
    // no need to do John.getFriends() , John already has John.friends Array 
});

You can also define this option globally instead of a per association basis.

var Person = db.define('person', {
    name : String
}, {
    autoFetch : true
});
Person.hasMany("friends", {
    rate : Number
}, {
  key: true
});

Associations can make calls to the associated Model by using the reverse option. For example, if you have an association from ModelA to ModelB, you can create an accessor in ModelB to get instances from ModelA. Confusing? Look at the next example.

var Pet = db.define('pet', {
    name : String
});
var Person = db.define('person', {
    name : String
});
Pet.hasOne("owner", Person, {
    reverse : "pets"
});
 
Person(4).getPets(function (errpets) {
    // although the association was made on Pet, 
    // Person will have an accessor (getPets) 
    // 
    // In this example, ORM will fetch all pets 
    // whose owner_id = 4 
});

This makes even more sense when having hasMany associations since you can manage the many to many associations from both sides.

var Pet = db.define('pet', {
    name : String
});
var Person = db.define('person', {
    name : String
});
Person.hasMany("pets", Pet, {
    bought  : Date
}, {
    key     : true,
    reverse : "owners"
});
 
Person(1).getPets(...);
Pet(2).getOwners(...);

To add an external database adapter to orm, call the addAdapter method, passing in the alias to use for connecting with this adapter, along with the constructor for the adapter:

require('orm').addAdapter('cassandra', CassandraAdapter);

See the documentation for creating adapters for more details.