The Po.et Node allows you to timestamp documents in a decentralized manner.
It's built on top of the Bitcoin's blockchain and IPFS.
For any questions about developing an application that integrates with the Po.et Node or contributing to Po.et that aren't answered here check out our Gitter community at https://gitter.im/poetapp.
To run the Po.et Node, you need to clone this repo, make sure you have NodeJS installed and just npm start
.
You also need to have RabbitMQ, IPFS and MongoDB installed. See Dependencies down below.
# Install NVM
curl -o- https://raw.githubusercontent.com/creationix/nvm/v0.33.8/install.sh | bash
# Activate NVM
. ~/.nvm/nvm.sh
# Clone The Po.et Node
git clone https://github.com/poetapp/node.git
# Build The Po.et Node
cd node
# Install NodeJS (will install node version in .nvmrc)
nvm install
npm i
npm run build
# Run The Po.et Node
npm start
You need to have Docker and docker-compose installed
git clone https://github.com/poetapp/node.git
cd node
docker-compose up
# Clone The Po.et Node
git clone https://github.com/poetapp/node.git
cd node
make all
# to clean up and stop processes:
make stop
make clean
Make Commands available:
make all # makes all dependancies
make stop # stops the docker containers
make clean # removes node_modules and all stopped containers
make setup # setups the api nodejs deps
make containers # creates the dependant docker containers for mongodb, rabbitmq and ipfs
The Po.et Node depends on RabbitMQ, IPFS, MongoDB and InsightAPI. By default, it looks for all these things in localhost, except for the InsightAPI, which defaults to https://test-insight.bitpay.com/.
For a quick startup, we provide make
commands that build and run these dependencies in Docker containers.
You just need to sudo make mongo rabbit ipfs
once to create the Docker images, and sudo make start-all
to start them when they shut down.
You can also sudo make sh-mongo
and sudo make sh-ipfs
to run the mongo shell or ssh into the IPFS container.
You'll need to have Docker installed and running for this. See How to Install Docker CE.
No Docker image is provided for InsightAPI since Bitpay offers a usable InsightAPI publicly.
In a production environment, you may want to run these applications natively installed on the OS rather than dockerized. If you choose to run IPFS dockerized, make sure it's able to communicate with other IPFS nodes outside your network.
The Po.et Node comes with a default configuration that works out of the box.
By default, timestamping to the blockchain is disabled, and RabbitMQ, IPFS and MongoDB are expected to be running in localhost with their default ports.
You can change any configuration by placing a json file in ~/.po.et/configuration.json
. Po.et will look for this file upon startup and, if found, merge its contents with the default configuration.
Note: Po.et will NOT reload the configuration while it's running if you change it. You'll need to restart the Node for configuration changes to apply.
This is what the default configuration looks like:
{
rabbitmqUrl: 'amqp://localhost',
mongodbUrl: 'mongodb://localhost:27017/poet',
ipfsUrl: 'http://localhost:5001',
insightUrl: 'https://test-insight.bitpay.com/api',
apiPort: 18080,
poetNetwork: 'BARD',
poetVersion: [0, 0, 0, 2],
minimumBlockHeight: 1253828,
blockchainReaderIntervalInSeconds: 5,
enableTimestamping: false,
bitcoinAddress: '',
bitcoinAddressPrivateKey: '',
timestampIntervalInSeconds: 30
}
To enable timestamping to the Bitcoin blockchain, you need to set enableTimestamping
to true
and provide a valid bitcoinAddress
and the bitcoinAddressPrivateKey
that owns it.
You can create a valid bitcoin address using Ian Coleman's Mnemonic Code Converter, creating and exporting the private keys and addresses from bitcoin-cli or wallets that support this or use any other means that work for you.
You'll also need some bitcoins in that address. In testnet, you can get some for free using a testnet faucet. We've found flyingkiwi's one particularly nice.
Right now, Po.et is timestamping to Testnet, so just make sure your address is a valid Testnet address.
Currently, the Node exposes three endpoints.
Returns a single claim by its Id.
For simplicity, this endpoint adds a .timestamp
in the response, which is not a real part of the claim, but provides valuable information such as the id of the transaction in which this claim has been timestamped, the IPFS hash by which it can be found, etc.
Returns 404 if the claim isn't found in this Node's database. This doesn't strictly mean the claim doesn't exist in the Po.et network — it just doesn't exist in this Node.
Returns an array of claims — all the claims belonging to the passed public key.
Retrieving all works isn't supported yet. The Node will assumme you intended to call GET /works?publicKey=undefined
, which will normally return an empty array. Support for this endpoint will be added in the future.
Publish a work.
This endpoint is async — unless an immediate error can be detected (such as a malformed claim), the endpoint will return an ACK. There's no guarantee that the work has actually been processed, timestamped an sent to IPFS. To check that, you'll need to GET /works/:id
and check the .timestamp
attribute.
This endpoint expects a fully constructed claim — with the correct .id
, .publicKey
, .signature
and .dateCreated
. See Building Claims for information on how to correctly create these attributes.
A Po.et Claim is a JSON object that holds arbitrary information plus a few attributes that allow the network to verify that the claim has actually been created by a certain person, that the claim has not been modified since its creation, and a special field type
which will allow more features in the future.
For example, a claim could look like this:
{
id: '15867401b92567b4f7ea83e39a646ab9eb581b560bc78488b7a0c1b586c70215',
publicKey: '02badf4650ba545608242c2d303d587cf4f778ae3cf2b3ef99fbda37555a400fd2',
signature: '304402201824b78d3703162eb7f240341968ebfecad1f002f988dbc9ec80c1317e49d6290220470124c7425a5d8024778991863f0a25931a7e45fb72223bea81728a08e30b50',
type: ClaimType.Work,
dateCreated: new Date('2017-12-11T22:58:11.375Z'),
attributes: {
name: 'The Murders in the Rue Morgue',
author: 'Edgar Allan Poe',
tags: 'short story, detective story, detective',
dateCreated: '1841-01-01T00:00:00.000Z',
datePublished: '1841-01-01T00:00:00.000Z',
content: 'The mental features discoursed of as the analytical, are, in themselves, but little susceptible of analysis...'
}
}
The publicKey
field must be set to the public part of a key pair you own. You'll need the corresponding private key to prove this claim was generated by you.
The signature
must be set to the result of cryptographically signing the id
field with the private key you own using the elliptic curve DSA signature scheme. This signature is currently being validated with bitcore.Crypto.ECDSA.verify, and we're using bitcore.Crypto.ECDSA.sign to sign our claims.
The id
field is the sha256
of the claim, excluding the id
and signature
fields, so getId(claim) == getId(getId(claim))
. We're using decodeIO's implementation of Google's Protobuf library in order to serialize the claims to a byte buffer deterministically and hashing this byte buffer. The .proto
file we're using can be found in src/Serialization/PoetProto.json. There's a poet.proto file that you can use in any other programming language.
All this logic is abstracted away in poet-js, so if you're working with JavaScript or TypeScript you can simply use the createClaim(privateKey, claimType, attributes)
function like so:
import { ClaimType, createClaim } from '@po.et/poet-js'
const privateKey = 'L1mptZyB6aWkiJU7dvAK4UUjLSaqzcRNYJn3KuAA7oEVyiNn3ZPF'
const claim = createClaim(privateKey, ClaimType.Work, {
name: 'The Murders in the Rue Morgue',
author: 'Edgar Allan Poe',
tags: 'short story, detective story, detective',
dateCreated: '1841-01-01T00:00:00.000Z',
datePublished: '1841-01-01T00:00:00.000Z',
content: 'The mental features discoursed of as the analytical, are, in themselves, but little susceptible of analysis...'
})
You can find more examples on how to build and publish claims in the integration tests in test/Integration/PostWork.
Create a file with the following contents and place it in ~/.config/systemd/user/poet-node.service
:
[Unit]
Description=Po.et Node Daemon
After=network.target
[Service]
ExecStart=/home/ubuntu/.nvm/versions/node/v9.3.0/bin/node /home/ubuntu/node/dist/babel/src/index.js daemon
WorkingDirectory=/home/ubuntu/node/dist/babel/src/
Restart=always
StandardOutput=syslog
StandardError=syslog
SyslogIdentifier=poet-node
[Install]
WantedBy=default.target
Make sure the paths to NodeJS and Po.et in the ExecStart
and WorkingDirectory
entries are correct.
You can then use the following commands to handle the daemon:
systemctl --user start poet-node
systemctl --user stop poet-node
systemctl --user about poet-node
systemctl --user restart poet-node
And journalctl -f --user-unit poet-node
to tail the logs, or without the -f
to just less
them.
The Po.et Node has been tested in Ubuntu, Linux Mint and Mac OS.
The npm run build
command depends on npm run build-clear
and npm run copy-json
, which use bash' rm
and cp
, so building under Windows may require some tweaking.
Run npm run build
to compile the source. This will run TypeScript on the source files and place the output in dist/ts
, and then it'll run Babel and place the output in dist/babel
.
Currently, we're only using Babel to support absolute import paths.
During development, you can also run npm run watch
to automatically watch for file changes, build the changed files and restart the application on the fly.
Both unit and integration tests live in this same repo. You can run both with npm test
or separately with npm run test-unit
and npm run test-integration
.
The integration tests are hard-coded to hit the http://localhost:18080
. In the future, this will be picked up from an environment variable and defaulted to that same url.
Warning: Running the integration tests wipes out the entire
db.poet.works
collection and inserts testing data. This is done by thetest/integration/PrepareDB.ts
file. In the future, a less invasive mechanism will be developed. Meanwhile, make sure you're comfortable with this before running the integration tests!
Currently, Po.et Node is lacking some tests. The most critical paths that aren't being tested right now are:
- Broadcasting of transactions in a single Node (submit work, wait a bit, get work and expect transactionId to be set and valid)
- Replication across nodes (submit WORK to Node A, get WORK in Node B)
See issues [#21][i21], [#22][i22], [#25][i25] and [#27][i27] for more info on this topic.
Coverage is generated with Istanbul whenever tests are run.
A more complete report can be generated by running npm run coverage
.
This area needs some reviewing — the reports look a bit off sometimes.
Also, we'll want the Travis job to check that coverage doesn't go down with pull requests. See [#11][i11].
The master branch is blocked - no one can commit to it directly. To contribute changes, branch off from master and make a PR back to it.
TravisCI will run all tests automatically for all pull requests submitted.
Please run npm run lint
. The linting configuration still needs some tweaking, and it'll be added to Travis in the future.