NoSQL Databases - MongoDB first cut

A few days ago I tried MongoDB, my first encounter with what is called a NOSQL Database. No, it does not mean 'no' 'sql', rather Not only SQL, anyway I thought I would try mongodb, out of the many available there. It has both the document-oriented database design and ability to be accessed from Javascript, both features I really like. However, since I spend most time in Java, I thought I would start there.

First, let me tell you you should try their web console before writing a line of code or downloading anything because it's a really good self-run tutorial, that lets you type real commands and get real output all from the browser! pretty awesome...

Anyway, here is what I did to get started.
1. Download mongodb from http://www.mongodb.org/
2. Create working dir for MongoDB to use as data directory
3. Open favorite IDE or editor, and start writing test, yes Test, not source, in the spirit of Test Driven Development.
4. In test package, create test such as:

    @BeforeClass
    public static void setup() {
        db = new MongoDB(); db.connect(); // connect to mongodb
        conn = db.connectToDB("test"); // connect to a db called 'test', will be auto created
    }

    @AfterClass
    public static void close() {
        db.close();
    }

    @Before
    public void createdb() {
        coll = conn.getCollection("artifacts"); // fetch a collection called 'artifacts', will be autocreated
    }

    @After
    public void cleanup() {
        coll.drop(); // remove table
    }

    @Test
    public void testConn() {
        assertThat(conn.getName(), is("artifacts"));
        assertThat(conn.getCollectionNames().size(), is(0));
    }

Self explanatory, but shows how easy it is to get started with it.

Run this test, it will fail of course! Now on to the green of TDD...

Now, to make this test pass, we need to do two things : Create the objects we use in the tests, and start MongoDB database server.

First, in a source package, add the class MongoDB which would be like:

import com.mongodb.Mongo;
import com.mongodb.DB;

public class MongoDB {
    private Mongo m = null;

    public void connect() {
        try {
            m = new Mongo();
        } catch (Exception e) {
        }
    }

    public DB connectToDB(String name) {
        DB db = m.getDB(name);
        return db;
    }

    public void close() {
        m.close();
    }
}


Then start MongoDB as:
With a custom data directory (otherwise it assumes a weird location like /data), such as
   >     mongod --dbpath=/local/test/mongodb


Voila, run the test now and it should pass.

You noticed that I tested for the size of collections to be zero, not 1, because the collection 'artifacts' is not created until something is inserted into it.
Also, note that /local/test/mongodb acts as the data directory. for automated tests, you can create this dir in a temp dir, then delete it at the end of testing, which will require shutdown of mongodb. But we won't worry about that now.
Also inspect this directory as you run these tests.

To test this auto create feature of MongoDB, add a test to add a row to the collection, and see if you can get it back, the collection should be auto created.

@Test
    public void testMultipleEntries() {
        for (int i = 0; i < 10000; i++) {
            coll.insert(new BasicDBObject().append("i", i).append("j", i / 2)
                    .append("k", i * 2));
        }
        assertEquals(10000, coll.count());
    }

Pretty straightforward! The BasicDBObject is a Document object. In MongoDB a document is a record, much like a row is in a relational database. The insert() line basically inserts a document with fields 'i' 'j', 'k', into the collection 'artifacts', and with the use of builder pattern you can add as many append()s to keep inserting columns

So what are inserting in each loop iteration looks like:
{ "_id" : { "$oid" : "4dd076881368c78a6cfbdf96"} , "i" : 0 , "j" : 0 , "k" : 0}

If you want to see this formatted document entry, add something like
         System.out.print(coll.findOne());
which displays the first document record (or row) of the collection in JSON format