Skip to Main Content

Reactive-Functional Fun on the Blockchain with web3j

Yolande Poirier-OracleFeb 21 2017 — edited Feb 28 2017

by Conor Svensson

Reactive Java on the blockchain with web3j

In a recent issue of Java Magazine, I provide a primer on working with the blockchain technology Ethereum using web3j to build Java applications on it.

In this article, we're going to use web3j's reactive-functional API to shed some light on the events taking place in the public Ethereum blockchain.

The code for all examples is available at


During the past year, the technology and financial press has been full of talk about blockchain and its disruptive potential. A blockchain is a decentralized, immutable data store. Because it is immutable, data can only be appended to the blockchain. This is achieved with transactions that are grouped together into blocks, which are added to the end of the blockchain.


Figure 1. The structure of a blockchain

The state of the data residing in the blockchain is built by replaying the transactions or events that have gone before. You can think of it as a distributed event log.

Blockchains are completely decentralized, stored across unrelated nodes potentially in an untrusted public environment such as the internet. There are a number of different blockchain technologies in existence. Of those, Ethereum has emerged as the dominant public blockchain technology.

For more background on blockchain technology, please refer to the Java Magazine article).

Getting Started with Ethereum

The decentralized Ethereum network is made of clients that form the peers of the Ethereum network. To talk to the network, you need to have access to one of these clients. The easiest way to do this is to run a client yourself. The two main clients are Geth and Parity.

Once you have installed a client, you can start it as follows:


$ geth --fast --cache=512 --rpcapi personal,db,eth,net,web3 --rpc --testnet


$ parity --chain testnet

The client will then find other nodes to connect to and start syncing a local copy of the testnet blockchain. There are two public Ethereum blockchains—mainnet and testnet—reflecting the production and test environments, respectively.

The web3j Library

Now that you have a client running, it's easy to start talking to the Ethereum blockchain, thanks to web3j, which is a lightweight Java library for working with clients on the Ethereum blockchain. You can learn more about web3j and Ethereum at


Figure 2. web3j provides an integration layer to the Ethereum blockchain for Java applications

To incorporate web3j into your project, add the following Maven dependency:


If you're working with Android, use core-android instead.

New Block Subscriptions

Using a couple of lines of code, you can now hook up to the Ethereum blockchain and be notified of new blocks being added to the blockchain:

Web3j web3 = HttpService());  // defaults to http://localhost:8545/ 
Subscription subscription = web3j.blockObservable(false).subscribe(block -> {
    System.out.println("Sweet, block number " + block.getBlock().getNumber() + " has just been created");
}, Throwable::printStackTrace);

This blockObservable will now emit a block object to the subscriber each time a new block is appended to the Ethereum blockchain.

We need to include the sleep statement, because the subscription takes place asynchronously in a different thread of execution to the rest of our program.

We'll fill this example out so that it provides details about the number of transactions in each block, it provides the block hash that uniquely identifies the block, and it provides the parent hash of the previous block.

We'll restrict this example to only 10 blocks, and we'll use a latch so that we wait for the 10 blocks to be emitted.

CountDownLatch countDownLatch = new CountDownLatch(1);
System.out.println("Waiting for " + COUNT + " transactions...");
Subscription subscription = web3j.blockObservable(true)
        .subscribe(ethBlock -> {
            EthBlock.Block block = ethBlock.getBlock();
            LocalDateTime timestamp = Instant.ofEpochSecond(
            int transactionCount = block.getTransactions().size();
            String hash = block.getHash();
            String parentHash = block.getParentHash();
                    timestamp + " " +
                            "Tx count: " + transactionCount + ", " +
                            "Hash: " + hash + ", " +
                            "Parent hash: " + parentHash
        }, Throwable::printStackTrace);

By running the code, we can see details for the last 10 blocks added to the Ethereum blockchain:

2016-12-22T00:27:11 Tx count: 3, Hash: 0xbed93a59bcd30c0f11e155109a5dbff3e56a5354e90e514108917c1f54c4182d, Parent hash: 0x74f8bffd8734833a2fbaad0d9e48eb6206ee25dc9c55425cf1edca9dfe45aa06
2016-12-22T00:27:44 Tx count: 2, Hash: 0xb147a4ca79bc7c5e767ee92be648a9fd23ee15ec64f764b3f2dc8a4cb7229a50, Parent hash: 0xbed93a59bcd30c0f11e155109a5dbff3e56a5354e90e514108917c1f54c4182d
2016-12-22T00:27:51 Tx count: 1, Hash: 0xa2d8d3572592470e30b7fb504ce504444b1ea1208035f1545c85fc824a882e1b, Parent hash: 0xb147a4ca79bc7c5e767ee92be648a9fd23ee15ec64f764b3f2dc8a4cb7229a50
2016-12-22T00:28:09 Tx count: 2, Hash: 0xf4d4a7d7a2202f8c64e72b47947adb5ae419f2bdadabec9386cb9d4a27cd7ef8, Parent hash: 0xa2d8d3572592470e30b7fb504ce504444b1ea1208035f1545c85fc824a882e1b
2016-12-22T00:28:17 Tx count: 2, Hash: 0x43dc1da772b3f5aea3388fb4ddf14c188753646de4971b62e90b903774c5c2bc, Parent hash: 0xf4d4a7d7a2202f8c64e72b47947adb5ae419f2bdadabec9386cb9d4a27cd7ef8
2016-12-22T00:28:22 Tx count: 1, Hash: 0x80d03ac26a3aba27d01beecb530b9055dd28b199a52bcc53704c8e418ba437fa, Parent hash: 0x43dc1da772b3f5aea3388fb4ddf14c188753646de4971b62e90b903774c5c2bc
2016-12-22T00:28:27 Tx count: 0, Hash: 0x897af69eafa47c2951379780a40ee47fc3383959197419a4feadc59d1d2b2c13, Parent hash: 0x80d03ac26a3aba27d01beecb530b9055dd28b199a52bcc53704c8e418ba437fa
2016-12-22T00:28:28 Tx count: 0, Hash: 0xf12762e30caa3d0f1c6c7deb7fff93cd00727565575b8b35a7466ca75ba112e6, Parent hash: 0x897af69eafa47c2951379780a40ee47fc3383959197419a4feadc59d1d2b2c13
2016-12-22T00:28:43 Tx count: 1, Hash: 0x3754841b596ad15b17b37dd6293db77ebb672743918877dcf92e47407a2153e5, Parent hash: 0xf12762e30caa3d0f1c6c7deb7fff93cd00727565575b8b35a7466ca75ba112e6
2016-12-22T00:28:49 Tx count: 2, Hash: 0x6613c62dfd05761b3794658d341c1dab95f19db1e2bebd8cf091861c1ae88cd4, Parent hash: 0x3754841b596ad15b17b37dd6293db77ebb672743918877dcf92e47407a2153e5

Functional Composition

Because we're using RxJava's Observables, we can easily compose them to add additional functionality.

For instance, if we wish to create an Observable

that emits every new transaction that is written to the blockchain, we can use our blockObservable()

again, and extract the list of transactions contained in the block, and then use the flatMapIterable()

method to emit these transactions individually:

         .flatMapIterable(ethBlock -> (List) ethBlock.getBlock().getTransactions());

This time, we pass the true parameter, which requests blocks and full details of all the transactions contained in those blocks.

web3j has already implemented this composition for us via web3j.transactionObservable().

Counting Ether

Ethereum comes with its own cryptocurrency, named Ether, which you can think of as similar to Bitcoin. Ether is used to pay for transactions that take place on the network. These transactions can also transfer Ether from one person to another. Details about the Ether associated with a transaction are contained in the value field of Ethereum transactions.

Using the transaction Observable we created in the previous section, we can easily start pulling information out of the blockchain in real time. For instance, to obtain the total value of transactions taking place during a given number of blocks, we can use the following:

CountDownLatch countDownLatch = new CountDownLatch(COUNT);
System.out.println("Waiting for " + COUNT + " transactions...");
Observable<BigInteger> transactionValue = web3j.transactionObservable()
        .reduce(BigInteger.ZERO, BigInteger::add);
Subscription subscription = transactionValue.subscribe(total -> {
    System.out.println("Transaction value: " +
            Convert.fromWei(new BigDecimal(total), Convert.Unit.ETHER) + " Ether");
}, Throwable::printStackTrace);
countDownLatch.await(10, TimeUnit.MINUTES);

The code above produces the following output:

Transaction value: 5.011 Ether (5011000000000000000 Wei)

Here, we obtain the value of each emitted transaction and sum them together via the reduce() method. Remember that you need to be dealing with a finite stream in order for the reduce function to provide a final result. Also, not all transactions will be transfers of Ether, so don't be surprised if you receive a transaction value of 0.

Ether comes in a number of different denominations, Wei being the most granular. One Wei equals 10e-18 Ether! The Convert method converts the value in Wei that we get back from our transactions into Ether.

Further Observables

Other observables available. For instance, to obtain details of pending transactions—those that have not yet been grouped together into a block and appended to the blockchain—we can use this:

Subscription subscription = web3j.pendingTransactionObservable().subscribe(tx -> {

Also, web3j provides observables for all Ethereum API calls, such as the following trivial example, which provides the client version of your Ethereum client:

Web3j web3 = HttpService());  // defaults to http://localhost:8545/
web3j.web3ClientVersion().observable().subscribe(x -> {

The code above produces the following output:

Client is running version:     Geth/v1.5.4-stable-b70acf3c/darwin/go1.7.3

For the full list of available API calls, refer to the Ethereum interface.

There are additional event types you can subscribe to in web3j, which you can read about at


I have provided a brief overview of the capabilities of web3j for querying the Ethereum blockchain via its reactive-functional API. If you want to learn more about web3j and Ethereum, please head over to the web3j home page at, which contains further resources.

About the Author

Conor Svensson (@conors10) is the author of web3j, the Java library for integrating applications with the Ethereum blockchain. He previously cofounded the startups Cohome and Huffle before becoming the CTO at Othera and building Othera's blockchain lending platform and exchange. He is also a regular speaker and one of the organizers of the Sydney Java user group. He blogs about tech and finance stuff, and when not in front a screen, he likes to make the most of surfing at his local beach, Maroubra in Sydney.

Join the Conversation

Join the Java community conversation on Facebook, Twitter, and the Oracle Java Blog!

Post Details
Added on Feb 21 2017
1 comment