Apr 8, 2014
May 7, 2015 (Retired)
Dia Kharrat (dkharrat)
Dia Kharrat (dkharrat)
Source code
APK file



Core Data for Android

NexusData is an object graph and persistence framework for Android. It allows for organizing and managing relational data and serializing it to SQLite or custom stores. The data can be accessed or modified using higher level objects representing entities and their relationships. NexusData manages all the objects in the persistence store, tracks changes, and maintains consistency in relationships.

Essentially, it brings Core Data functionality from iOS to Android. However, the library is not intended to be a straight port of Core Data. Instead, it aims to leverage Core Data's concepts, while having the flexibility to evolve independently.

NexusData is not an ORM in that it's more higher-level and is not tied to a specific storage engine. The query interface is oblivious to the underlying persistence store.

NexusData supports Android API 10+. This library follows semantic versioning. Note that this library is still active in development, and is missing a lot of features. New releases might introduce interface-breaking changes, which will be indicated in the changelog. NexusData 1.0.0 will be the first stable release.


For sample projects that use NexusData, browse the samples directory.


  • Change tracking and management of objects.
  • Relationship maintenance by automatically propagating related changes to maintain consistency.
  • Support for one-to-one and one-to-many relationships.
  • Lazy loading of the object graph to reduce memory overhead.
  • Flexible query interface that is independent of the underlying storage engine.
  • Model generator that generates java classes from the model.
  • Entity inheritence.
  • Support for atomic and incremental persistence stores.
  • Extensible to different persistence storage. Currently, two storage engines are provided out of the box:
    • In-memory
    • SQLite
  • Built-in support for basic attribute types:
    • Short / Integer / Long
    • Boolean
    • Float / Double
    • String
    • Enum
    • Date


The framework is constantly being improved and new features are being implemented. Since it's very early stage, it currently has some limitations:

  • Undo/Redo is not supported.
  • Many-to-many relationships are not supported yet.
  • Schema migrations are not supported yet.
  • Query syntax is currently limited to comparisons and boolean logic. Operations like aggregations and joining are not supported yet.
  • Framework is not yet optimized for large data sets in terms of performance and memory. This is due to the early development of the project and will be improved over time.
  • Custom data types are not supported yet.

Apps Using NexusData

Do you have an app that's utilizing NexusData? Let me know and I'll add a link to it here!

How to Add NexusData to Your Project

There are multiple ways to include your project, depending on your build environment:


Add the following dependency to your build.gradle file for your project:

dependencies {
  compile 'com.github.dkharrat.nexusdata:nexusdata:0.2.1'


Add the following dependency to your pom.xml file for your project:


Android Studio or IntelliJ 13+

Add the appropriate dependency in your build.gradle file and refresh your project.



How to Get Started

For a complete example of how NexusData can be used, please browse through the samples included with the project.

Defining the model

The model is used to provide NexusData with information about the entities and their properties. A model can be defined either programmatically or via JSON file. Here's an example of a JSON-based model file for a ToDo app:


  "metaVersion": 1,
  "model": {
    "name": "Todo",
    "version": 3,
    "packageName": "org.example.todo",
    "entities": [{
      "name": "Task",
      "enums": [{
        "name": "Priority",
        "values": ["HIGH", "MEDIUM", "LOW"]
      "attributes": [{
        "name": "title",
        "type": "String"
      }, {
        "name": "notes",
        "type": "String"
      }, {
        "name": "dueBy",
        "type": "Date"
      }, {
         "name": "completed",
         "type": "Bool",
         "required": true,
         "default": false
       }, {
         "name": "priority",
         "type": "Priority"
      "relationships": [{
        "name": "assignedTo",
        "destinationEntity": "User",
        "inverseName": "tasks",
        "toMany": false
    }, {
      "name": "User",
      "attributes": [{
        "name": "name",
        "type": "String"
      "relationships": [{
        "name": "tasks",
        "destinationEntity": "Task",
        "inverseName": "assignedTo",
        "toMany": true

This model (named "Todo") defines two entities: Task and User. A Task belongs to a User, and a User has many Tasks. Also, each entity has some attributes.

Generating classes from a model file

NexusData comes with a Model Generator that allows you to generate an appropriate class for each entity. Though using the generator is not necessary to use NexusData, the Model Generator reduces the need to write a lot of repetitive and boilerplate code for each entity (getter, setters, etc.). If you choose not to use the Model Generator, you may either create the classes yourself or use ManagedObject directly.

The model generator source code is located under the modelgen directory. To generate the appropriate classes from the above model file, run this command:

gradle :modelgen:run -Pargs="-f /path/to/model/todo.model.json -O /path/to/output/src/main/java"

This will parse the todo.model.json file and generate the corresponding classes in the src/main/java/org/example/todo directory. The output of the generator will look something like this:

03:43:41.970 [main] INFO  c.g.d.n.modelgen.ModelGenerator - Setting up model generator
03:43:42.028 [main] INFO  c.g.d.n.modelgen.ModelGenerator - Parsing model file 'todo.model.json'
03:43:42.098 [main] INFO  c.g.d.n.modelgen.ModelGenerator - Generating class files for 'Todo' model (version 1)
03:43:42.152 [main] INFO  c.g.d.n.modelgen.ModelGenerator - Generating class
03:43:42.161 [main] INFO  c.g.d.n.modelgen.ModelGenerator - Generating class
03:43:42.184 [main] INFO  c.g.d.n.modelgen.ModelGenerator - Generating class
03:43:42.184 [main] INFO  c.g.d.n.modelgen.ModelGenerator - Generating class

For each entity, two classes will be generated. For example, for the Task entity, and are generated. The file contains all the accessors, enums, and relationships based on the model file. The other file,, is an empty class that inherits from _Task. The Task class can be used to define any custom code or derived properties. The reason for creating two classes is to allow you to re-generate the entity classes if the model changes, while maintaining any custom code associated with the entity in a separate file. The generator will not overwrite your custom class (e.g. in this example) if it already exists, but it will overwrite the base class (e.g.

Here's how the generated files look like for the Task entity:


package org.example.todo;

import java.util.Date;
import com.github.dkharrat.nexusdata.core.ManagedObject;

class _Task extends ManagedObject {

    public interface Property {
        String TITLE = "title";
        String NOTES = "notes";
        String DUE_BY = "dueBy";
        String COMPLETED = "completed";
        String PRIORITY = "priority";
        String ASSIGNED_TO = "assignedTo";

    public enum Priority {

    public String getTitle() {
        return (String)getValue(Property.TITLE);

    public void setTitle(String title) {
        setValue(Property.TITLE, title);

    public String getNotes() {
        return (String)getValue(Property.NOTES);

    public void setNotes(String notes) {
        setValue(Property.NOTES, notes);

    public Date getDueBy() {
        return (Date)getValue(Property.DUE_BY);

    public void setDueBy(Date dueBy) {
        setValue(Property.DUE_BY, dueBy);

    public boolean isCompleted() {
        return (Boolean)getValue(Property.COMPLETED);

    public void setCompleted(boolean completed) {
        setValue(Property.COMPLETED, completed);

    public Priority getPriority() {
        return (Priority)getValue(Property.PRIORITY);

    public void setPriority(Priority priority) {
        setValue(Property.PRIORITY, priority);

    public User getAssignedTo() {
        return (User)getValue(Property.ASSIGNED_TO);

    public void setAssignedTo(User assignedTo) {
        setValue(Property.ASSIGNED_TO, assignedTo);

package org.example.todo;

public class Task extends _Task {
    public Task() {

Initializing NexusData Stack

To setup NexusData in your application, you'll need three main parts: ObjectModel, PersistentStoreCoordinator, and a ObjectContext. These can typically be initialized once at startup and used throughout the lifetime of the application.

// create an ObjectModel that describes the meta model
ObjectModel model = new ObjectModel(app.getAssets().open("todo.model.json"));

// create the persistent store coordinator and its associated store
PersistentStoreCoordinator storeCoordinator = new PersistentStoreCoordinator(model);
Context ctx = getApplicationContext(); // the Android context
PersistentStore cacheStore = new AndroidSqlPersistentStore(ctx, ctx.getDatabasePath("todo"));

// create an ObjectContext that will be used to retrieve or save our objects
ObjectContext mainObjectContext = new ObjectContext(storeCoordinator);

Creating/Updating Objects

ObjectContext objCtx = getMainObjectContext();
Task task1 = objCtx.newObject(Task.class);
task1.setTitle("Get groceries");

Task task2 = objCtx.newObject(Task.class);
task2.setTitle("File taxes");;

Deleting Objects


Querying All Objects of Specific Type

List<Task> tasks = objCtx.findAll(Task.class)

Querying Objects Satisfying a Predicate

For example, to query all Tasks that are complete:

FetchRequest<Task> fetchRequest = objCtx.newFetchRequestBuilder(Task.class)
    .predicate("completed == true")
List<Task> tasks = objCtx.executeFetchOperation(fetchRequest);

Use ObjectContext and ManagedObjects in multiple threads

Similar to Core Data, ManagedObject and ObjectContext are not thread-safe, and therefore, should not be used in other threads. Each thread must use its own instance of ObjectContext. To pass objects between multiple threads, pass the object's ObjectID to the other thread, which can then retrieve the object from it's own ObjectContext.

To synchronize multiple ObjectContext with any changes, register a listener and then merge the changes when receiving a notification, as follows:

ObjectContextNotifier.registerListener(new ObjectContextNotifier.DefaultObjectContextListener() {
    @Override public void onPostSave(ObjectContext context, ChangedObjectsSet changedObjects) {
        // ensure that the notification we just got is not from our own context, and that it's from a context using a
        // persistence store that our context is also using.
        if (context != mainObjectContext && context.getPersistentStoreCoordinator() == mainObjectContext.getPersistentStoreCoordinator()) {


See the current Javadoc.


Contributions via pull requests are welcome! For suggestions, feedback, or feature requests, please submit an issue.


Dia Kharrat -


Copyright 2014 Dia Kharrat

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
See the License for the specific language governing permissions and
limitations under the License.