Clean Persistence with Sugar ORM

Sugar ORM is a database persistence library that provides a simple and concise way to integrate your application models into SQLite. In contrast to ActiveAndroid, which is mature, powerful, and flexible, Sugar ORM is:

  • Less verbose
  • Quicker to set up
  • More hands-free

Install

Sugar ORM has no dependencies, so installation is as simple as downloading the .jar file and putting it in your libs folder.
It's also available on Maven central. Installing with gradle is as simple as adding compile 'com.github.satyan:sugar:1.3' to your build.gradle file.

The current stable release is v1.3, but the beta release v1.4_beta is highly recommended. Once you have the .jar file in your libs folder, finishing the installation requires just setting the android:name attribute of the application tag in your AndroidManifest.xml:

<manifest xmlns:android="http://schemas.android.com/apk/res/android"
    package="com.example"
    android:versionCode="1"
    android:versionName="1.0" >

    <uses-sdk ... />
    <uses-permission ... />

    <application
        ...
        android:name="com.orm.SugarApp" >  <!--Use this attribute verbatim-->
        <meta-data android:name="DATABASE" android:value="example.db" />
        <meta-data android:name="VERSION" android:value="1" />
        <meta-data android:name="QUERY_LOG" android:value="true" />
        <meta-data android:name="DOMAIN_PACKAGE_NAME" android:value="com.example" />
        <activity ... />
        <activity ... />
    </application>
</manifest>

There are four additional parameters you can set. DATABASE is the name of the database file that will be created by Sugar ORM. VERSION is the version of the database schema. This is used for schema migrations, which are described in more detail below. QUERY_LOG can be set to "true" or "false" and determines whether to log debug messages for the queries made to the underlying SQLite database. Finally, DOMAIN_PACKAGE_NAME narrows down the packages that Sugar ORM scans for Classes to persist.

Integrate

v1.2

Persisting your models is dead simple. Just extend SugarRecord<YourModel>, and change your constructors to take in a Context object as its first argument, and call super(context) right after the constructor declaration (examples adapted from Designing Your Entities):

public class Author extends SugarRecord<Author> {
    String fullName;
    int age;
    
    public Author(Context context) {
        super(context);
    }

    public Author(Context context, String fullName, int age) {
        super(context);
        this.fullName = fullName;
        this.age = age;
    }
}

And that's it! For camel cased field names, the underlying SQL database creates a column using underscores. For instance, fullName becomes full_name (this is important for querying).

Creating relationships between models also just works:

public class Book extends SugarRecord<Book> {
    Author author;
}

will create the appropriate column indicating a relationship between Book and Author.

To prevent a member field from being persisted in the SQLite database, you can ignore it with an annotation:

public class Author extends SugarRecord<Author> {
    String fullName;
    int age;

    @Ignore
    String password;
}

This will prevent password from having a column created for it in the underlying database.

v1.3

While the above integration is simple, clean, and quick, integrating with v1.3 is strictly better. The only difference between the two versions in terms of their integration is that Context is no longer a required argument in v1.3. A no argument constructor is the only remaining requirement. The above Author model would become:

public class Author extends SugarRecord<Author> {
    String fullName;
    int age;

    public Author() {
    }

    public Author(String fullName, int age) {
        this.fullName = fullName;
        this.age = age;
    }
}

This makes it even easier to drop in Sugar ORM into an existing project and have things just work right out of the box.

Operations

The typical family of database operations has an interface in Sugar ORM. Most follow the format of ModelName.operation(ModelName.class, arguments).

Static methods

These operations can be performed on your Model class as a whole, requiring no instance of the Model itself.

Find by ID
Author author = Author.findById(Author.class, 4L);

The L at the end of that number is necessary because findById expects a Long (an overloaded version that takes an Integer is in a pull request).

Find all
List<Author> allAuthors = Author.listAll(Author.class);

This query performs no filtering at all, so for tables with a lot of records, this operation may take a long time to complete.

Find with conditions
List<Author> authors = Author.find(Author.class, "full_name = ?", "Nathan");
List<Book> books = Book.findWithQuery(Book.class, "Select * from Book limit ?", "4");

The first query embeds a "where" clause for convenience. The second query is a generic query that more closely proxies raw SQL. Both take a variable number of String arguments equal to the number of ?'s in the query string.

Bulk delete
Book.deleteAll(Book.class);
Author.deleteAll(Author.class, "age = ?", "31");

The first operation will, unsurprisingly, delete every Book record. The second operation only deletes Author records that have an age of 31. Note that for inserting a value into the where clause, it's necessary to make it a String, even though the underlying column is an int.

Count (available in v1.3+ only)

If the table is not initialized return value will be -1.

long numberOfBooks = Book.count(Book.class, null, null);
long numberOfAuthors = Author.count(Author.class, "full_name = ?", "Timothy");

Member methods

These operations are performed on a particular instance of the model.

Save
Author author = new Author("J.K. Rowling", 48);
author.save();

This produces a record in the Author table with columns full_name = "J.K. Rowling" and age = 48.

Delete
Author author = Author.findById(Author.class, 23L);
author.delete()

This removes the Author with ID = 23 from the database. A common gotcha here is when there is no such Author with this ID, so a null pointer check here is often advisable.

Query Builder

In addition to the above operations, Sugar ORM also comes with a query building interface. Using it allows you to separate the query step from the execution step. In this way, you can save a query and execute it multiple times, where each execution is done against the current database state.

Conditions

The way queries are built using this system is through chaining. That is, an adding another condition or filter returns the Select object back, meaning one can chain together conditions or filters easily. For instance, to create a query to get all Author's whose age is 20 or below, one can make:

Select youngAuthorQuery = Select.from(Author.class)
                                .where(Condition.prop("age").lt(20));

For a query of the the Author's between 20 and 50, limited to 5 results:

Select specificAuthorQuery = Select.from(Author.class)
                                   .where(Condition.prop("age").gt(20),
                                          Condition.prop("age").lt(50))
                                   .limit(5);

This is by no means an exhaustive tutorial on the possible ways of chaining conditions and statements. A more thorough list of possible options can be found here (pending a proper Java doc, this is the best resource for the interface).

Execution

All queries (instances of the Select object) can be saved to be executed later. There are three ways of executing a query, and all have their own purposes:

// Get number of items that would be returned by a query with .count()
long numberYoungAuthors = youngAuthorQuery.count();

// Get all of the items corresponding to the query with .list()
List<Author> specificAuthors = specificAuthorQuery.list()

// Get just the first item that corresponds to the query with .first()
Author firstAuthor = specificAuthorQuery.first();

Migrations

Migrations have light support via raw SQL statements. The meta-data tag used earlier in AndroidManifest.xml:

<meta-data android:name="VERSION" android:value="1" />

contains the version information. This number is used to apply migrations if it is different from the version of the current database.

To create a migration, make a new folder in assets called sugar_upgrades and populate it with <version>.sql files. These files should contain raw SQL statements separated by semicolons. Let's illustrate how this all works with an example. Using the above Author model, add a new field:

public class Author extends SugarRecord<Author> {
    String fullName;
    int age;
    int income;  // This is a new field

    public Author() {
    }

    public Author(String fullName, int age, int income) {
        this.fullName = fullName;
        this.age = age;
        this.income = income;
    }
}

In assets/sugar_upgrades/, create two files, 1.sql and 2.sql. Leave 1.sql empty because it represents the initial database state at version 1. In 2.sql, write:

alter table AUTHOR add INCOME INTEGER;

Finally, change the meta-data tag in AndroidManifest.xml to:

<meta-data android:name="VERSION" android:value="2" />

When the app runs, it will check the version of the current database, and compare that to the version defined in this tag. If they differ, it will apply all of the migrations necessary to reach version 2, in the natural order. In this case, all that needs to be done is to apply 2.sql, which adds an INCOME column to the AUTHOR table.

References

Fork me on GitHub