Build Status

NDDB is a powerful and versatile object database for node.js and the browser.

NDDB (N-Dimensional DataBase) supports indexes, views, hashes, joins, group-by, basic statistics, custom operations, saving and loading from file system and browser localStorage, and much more.

Developer-friendly thanks to an easy api, detailed documentation, and 100% test coverage.

List of features

The complete NDDB api documentation is available here.


Load the library in Node.js:

var NDDB = require('NDDB').NDDB;

or in the browser add a script tag in the page:

<!-- Must load a version of NDDB that includes JSUS (see 'build/' dir) -->
<script src="/path/to/nddb.js"></script>

Create an instance of NDDB:

var db = new NDDB();

Insert an item into the database:

    painter: "Picasso",
    title: "Les Demoiselles d'Avignon",
    year: 1907

Import a collection of items:

var items = [
        painter: "Dali",
        title: "Portrait of Paul Eluard",
        year: 1929,
        portrait: true
        painter: "Dali",
        title: "Barcelonese Mannequin",
        year: 1927
        painter: "Monet",
        title: "Water Lilies",
        year: 1906
        painter: "Monet",
        title: "Wheatstacks (End of Summer)",
        year: 1891
        painter: "Manet",
        title: "Olympia",
        year: 1863


Retrieve the database size:

var db_size = db.size(); // 6

Select Items

Select statements must begin with select, and can be concatened by any number of subsequent and and or statements. The comparison in select statements is performed using three input parameters:

Available operators include standard logical operators:

or advanced comparison operators:

It is possible to access and compare nested properties simply separating them with ..

After a selection is finished, items can be returned using one of the fetch statements.

Select Examples

Select all paintings from Dali:

db.select('painter', '=', 'Dali'); // 2 items

Case sensitive LIKE operator:

db.select('painter', 'LIKE', 'M_net'); // 3 items

Select on multiple properties (*) with case insensitive LIKE:

db.select('*', 'iLIKE', '%e%'); // All items
db.select(['painter', 'portrait'], 'iLIKE', '%e%') // 5 items

Select all portraits:

db.select('portrait'); // 1 item

Fetch all paintings from Dali that are before 1928:

db.select('painter', '=', 'Dali')
  .and('year', '<', 1928);
  .fetch(); // 1 item

Fetch all paintings of the beginning of XX's century:

    db.select('year', '><', [1900, 1910])
      .fetch(); // 2 items

Fetch separately all the painters and all the dates of the paintings:

db.select('year', '><', [1900, 1910])
  .fetchValues(['painter', 'title']);

// { painter: [ 'Jesus', 'Dali', 'Dali', 'Monet', 'Monet', 'Manet' ],
//   year: [ 0, 1929, 1927, 1906, 1891, 1863 ] }


Define a global comparator function that sorts all the entries chronologically:

db.globalCompator = function (o1, o2) {
    if (o1.year < o2.year) return 1;
    if (o1.year < o2.year) return 2;
    return 0;

Sort all the items (global comparator function is automatically used):

db.sort(); // Order: Manet, Monet, Monet, Picasso, Dali, Dali

Reverse the order of the items:

db.reverse(); // Order: Dali, Dali, Picasso, Monet, Monet, Manet

Define a custom comparator function for the name of the painter, which gives highest priorities to the canvases of Picasso;

db.compare('painter', function (o1, o2) {
    if (o1.painter === 'Picasso') return -1;
    if (o2.painter === 'Picasso') return 1;

Sort all the paintings by painter

db.sort('painter'); // Picasso is always listed first


Splits the database in sub-database, each containing semantically consistent set of entries:

// Let us add some cars to our previous database of paintings.
var not_art_items = [
      car: "Ferrari",
      model: "F10",
      speed: 350,
      car: "Fiat",
      model: "500",
      speed: 100,
      car: "BMW",
      model: "Z4",
      speed: 250,

db.view('art', function(o) {
  return o.painter;

db.view('cars', function(o) {
  return o.car;


db.size();          // 9
db.art.size();      // NDDB with 6 art entries
db.cars.size();     // NDDB with 3 car entries


Define a custom hash function that creates a new view on each of the painters in the database:

db.hash('painter', function(o) {
    if (!o) return undefined;
    return o.painter;


db.size();          // 6, unchanged;
db.painter.Picasso; // NDDB with 1 element in db
db.painter.Monet    // NDDB with 2 elements in db
db.painter.Manet    // NDDB with 1 elements in db
db.painter.Dali     // NDDB with 2 elements in db

Listenting to events

Listen to the insert event and modify the inserted items by adding an external index to them:

var id = 0;
function getMyId(){ return id++; };

db.on('insert', function(o) {
    o.painter.id = getMyId();


Define a custom indexing function that gives fast, direct access to the items of the database;

db.index('pid', function(o) {
    return o.id;


db.pid.get(0).name; // Picasso

db.pid.update(0, {
  comment: "Good job Pablo!"

// Counts items in selection.
db.select('comment').count(); // 1

var picasso = db.pid.remove(0);
db.size(); // (0)

// Get all available keys in the index
db.painter.getAllKeys(); // ['0','1', ... ]

// Get all elements indexed by their key in one object

Example of a configuration object

var logFunc = function(txt, level) {
  if (level > 0) {

var options = {
  tags:  {},          // Collection of tags
  update: {           // On every insert, remove and update:
    indexes:  true,   // Updates the indexes, if any
    sort:     true,   // Sorts the items of the database
    pointer:  true,   // Moves the iterator to the last inserted element
  C:  {},             // Collection of comparator functions
  H:  {},             // Collection of hashing functions
  I:  {},             // Collection of indexing functions
  V:  {},             // Collection of view functions
  log: logFunc,       // Default stdout
  logCtx: logCtx      // The context of execution for the log function
  nddb_pointer: 4,    // Set the pointer to element of index 4
  globalCompare: function(o1, o2) {
    // comparing code
  filters: {          // Extends NDDB with new operators for select queries
    '%': function(d, value, comparator){
          return function(elem) {
            if ((elem[d] % value) === 0) {
              return elem;
  share: {           // Contains objects that are copied by reference to
                     // in every new instance of NDDB.
    sharedObj: sharedObj

var nddb = new NDDB(options);

// or

nddb = new NDDB();

Saving and Loading Items

The items in the database can be saved and loaded using the save and load methods, and their synchronous implementations saveSync and loadSync.

Saving and loading to file system (node.js environment)

Two formats are natively supported: .json and .csv, and they are detected by the extension of the filename. If a differ extension is found, NDDB will fall back to the default format (usually json).

It is possible to specify new formats using the addFormat method.

Code Examples

// Saving items in JSON format.
db.save('db.json', function() {
    console.log("Saved db into 'db.json'");

// Saving items in CSV format.
db.save('db.csv', function() {
    console.log("Saved db into db.csv'");

// Saving items synchronously in CSV format.
console.log("Saved db into db.csv'");

// Saving items in the default format (usually json).
db.getDefaultFormat(); // json
db.save('db.out', function() {
    console.log("Saved db into db.out'");

// Specifying the default format and saving into CSV.
db.save('db.out', function() {
    console.log("Saved db into db.out'");

// Transform items before saving them to CSV format.
// Define adapter function that doubles all numbers in column "A".
var options = {};
options.adapter = {
    A: function(item) { return item.A * 2; }
db.save('db2.csv', options, function() {
    console.log("Saved db as csv into 'db2.csv', where numbers in column 'A'" +
                "were doubled");

// Loading items into database.
db.load('db.csv', function() {
                  console.log("Loaded csv file into database");

// Loading items into database synchronously.
console.log("Loaded csv file into database");

// Loading 'adapted' items into database.
db.load('db2.csv', function() {
                   console.log("Loaded csv file into database");

// Transform items before loading them into database.
// Loading items into database.
var options = {};
options.adapter = {
    A: function(item) { return item.A / 2; }

db.load('db2.csv', function() {
                   console.log("Loaded csv file into database");

// Specify a new format.
db.addFormat('asd', {
   save: function(db, file, cb, options) {
         // save file asynchronously.
   load: function(db, file, cb, options) {
         // load file asynchronously.
   saveSync: function(db, file, cb, options) {
         // save file synchronously.
   loadSync: function(db, file, cb, options) {
         // load file synchronously.

// Saving in the new format.

List of all available options


   flags: 'w',                        // The Node.js flag to write to fs.
                                      // Default: 'a' (append).

   encoding: 'utf-8',                 // The encoding of the file.

   mode: 0777,                        // The permission given to the file.
                                      // Default: 0666

   // Options below are processed when the CSV format is detected.

   headers: true,                     // if options.headers === true: use
                                      //   first line of file as headers;
                                      // if !options.headers: use
                                      //   ['X1'...'XN'] as headers;
                                      // if options.headers is an array of
                                      //   strings use it as headers;
                                      // if options.headers is an array
                                      //   containing true/false use entry
                                      //   from file/'Xi' respectively;

   adapter: { A: function(row) {      // An obj containing callbacks for
                  return row['A']-1;  // each header. The callbacks take
                 }                    // an object of strings and
            },                        // return a string. Each entry in
                                      // the file is the result of
                                      // applying the callback of its
                                      // column to its row.

   separator: ',',                    // The character used as separator
                                      // between values. Default ','.

   quote: '"',                        // The character used as quote.
                                      // Default: '"'.

   commentchar: '',                   // The character used for comments.
                                      // Default: ''.

   nestedQuotes: false,               // TRUE, if nested quotes allowed.
                                      // Default FALSE.

   escapeCharacter: '\\',             // The char that should be skipped.
                                      // Default: \.

Saving and loading to the local storage (browser environment)

Items persistance in the browser is available only if NDDB is built with the Shelf.js extension. Alternatively, a custom store function taking as input the name of the local database could be defined.

All items will be saved in the JSON format.

Notice that there exist limitations to maximum number of items that can be saved, depending on the local storage maximum capacity settings of the browser. If the limit is reached an error will be thrown.


NDDB relies on mocha and should.js for testing.

$ npm install # will load all necessary dependencies
$ npm test # will run the test suite against nddb.js


Create your customized build of NDDB using the make file in the bin directory:

In order to run in the browser NDDB needs to have JSUS loaded. You can include it separately, or create a new build that includes it already. See the build help for options.

node make.nddb.js build // Standard build,
node make.nddb.js build -a -o nddb-full // Full build

The build file file will be created inside the build/ directory.


node make.nddb.js --help

API Documentation

Create html API documentation using the make file in the bin directory:

node make.nddb.js doc