TypeScript icon, indicating that this package has built-in type declarations

1.2.0 • Public • Published


Build Status Coverage Status install size npm version

A GraphDB and RDF4J data access library written in JavaScript to be used in Node.js.



  • Node >= 8
  • NPM (npm)
npm install graphdb


Library documentation can be found here

The library is written in ES2016. During the build process source files src/ are transpiled to ES2015 and copied to lib/ directory.

A typescript definition file types.d.ts is generated in the lib/ as well.

Documentation is generated in docs/ from the JSDoc annotations in the source code.

Project Structure

  • src : The source code of the library.
  • test : Unit and component level tests written with Jest.
  • lib : Transpiled but files are built here when the library is published to npm.
  • docs : Documentation generated with JSDoc when the library is published to npm.
  • src/index.js : The library external API exporting main functional classes. This is also present in lib.
  • lib/types.d.ts : The typescript definitions generated when the library is published to npm.
  • scripts : Service scripts related to building, publishing and so on.

Setup Environment

  • Checkout or clone the project.
  • Make sure prerequisites are covered: node js and npm must be present and versions should be supported.
  • Enter the project directory and execute
npm install

Running Tests

After any change the tests should be run and check if any existing functionality is not broken in result.

npm run test

or constantly watching for changes in source files and tests and re-execute the test suite

npm run test:watch

Checking the codestyle

The library uses Google style in conjunction with ESLint's recommended ruleset.

npm run lint

Testing the packaging

Library is managed by NPM package manager. During publishing npm consults the .gitignore, .npmignore and package.json#files property in order to decide which resources should be published. If any change in the project structure, .gitignore or .npmignore is made, then publishing must be verified in order to be guaranteed that needed sources will be properly published. The packaging could be verified using the npm pack command which generates a *.tgz archive in the project root. The archive should contain only the needed resources. Furture the archive could be used as a source for npm install where the path pointing the archive is provided.


  • Increase the package version.
npm version patch|minor|major
  • Login in npm.
npm login
  • Publish package in npm.
npm publish



The ServerClient deals with operations on server level like obtaining a list with available repositories, concrete repository or deleting repositories. In order to work with the ServerClient it should be configured ServerClientConfig first.

  • Configure ServerClient
const {ServerClient, ServerClientConfig} = require('graphdb').server;
const {RDFMimeType} = require('graphdb').http;
const serverConfig = new ServerClientConfig('http://rdf4j-compliant-server/', 0, {
const server = new ServerClient(serverConfig);
  • Fetch repository ids
server.getRepositoryIDs().then(ids => {
    // work with ids
}).catch(err => console.log(err));
  • Check if repository with given name exists
server.hasRepository('repository-name').then(exists => {
    if (exists) {
        // repository exists -> delete it for example
}).catch(err => console.log(err));
  • Delete repository with given name
server.deleteRepository('repository-name').then(() => {
    // successfully deleted
}).catch(err => console.log(err));
  • Although a repository instance can be created using a constructor which can be seen in the examples below a client could obtain an instance of RDFRepositoryClient through the server client
server.getRepository('repository-name').then(repository => {
    // repository is a configured RDFRepositoryClient instance
}).catch(err => console.log(err));


  • Instantiating repository client
const readTimeout = 30000;
const writeTimeout = 30000;
const config = new RepositoryClientConfig(['http://GDB/repositories/my-repo'], {
  'Accept': RDFMimeType.TURTLE
}, '', readTimeout, writeTimeout);
const repository = new RDFRepositoryClient(config);
  • Obtaining repository client instance through a ServerClient
const {ServerClient, ServerClientConfig} = require('graphdb').server;
const {RepositoryClientConfig} = require('graphdb').repository;
const config = new ServerClientConfig('http://GDB', 0, {});
const server = new ServerClient(config);
const readTimeout = 30000;
const writeTimeout = 30000;
const repositoryClientConfig = new RepositoryClientConfig(['http://GDB/repositories/my-repo'], {}, '', readTimeout, writeTimeout);
return server.getRepository('automotive', repositoryClientConfig).then((rdfRepositoryClient) => {
// rdfRepositoryClient is a configured instance of RDFRepositoryClient


Statements could be fetched using the RDFRepositoryClient.get, RDFRepositoryClient.query, RDFRepositoryClient.download.

Every reading method can get the response parsed to data objects according to RDFJS data model specification (see Response Parsers).

  • Reading statements
const payload = new GetStatementsPayload()
return repository.get(payload).then((data) => {
    // data contains requested staments in rdf json format
  • Downloading data from repository by consuming a WritableStream
const dest = __dirname + '/statements.ttl';
const output = fs.createWriteStream(dest);
const payload = new GetStatementsPayload()
repository.download(payload).then((response) => {
    response.on('data', (chunk) => {
        output.write(new Buffer(chunk));
    response.on('end', () => {
  • Query evaluation against a sparql endpoint

  • SELECT query returning data objects

repository.registerParser(new SparqlXmlResultParser());
const payload = new GetQueryPayload()
  .setQuery('select * where {?s ?p ?o}')
return repository.query(payload).then((stream) => {
  stream.on('data', (bindings) => {
    // the bindings stream converted to data objects with the registered parser
  stream.on('end', () => {
    // handle end of the stream
  • ASK query returning a boolean result
const payload = new GetQueryPayload()
  .setQuery('ask {?s ?p ?o}')
repository.registerParser(new SparqlJsonResultParser());
return repository.query(payload).then((data) => {
  // data => true|false


  • Uploading data in repository (POST) using ReadStream
const contentType = RDFMimeType.TURTLE;
const turtleFile = __dirname + '/statements.ttl';
fs.readFile(turtleFile, (err, stream) => {
    repository.upload(stream, null, null, contentType).catch((e) => console.log(e));
  • Overwriting data in repository (PUT) using ReadStream
const contentType = RDFMimeType.TURTLE;
const file = __dirname + '/statements-overwrite.ttl';
fs.readFile(file, (err, stream) => {
    repository.overwrite(stream, null, null, contentType).catch((e) => console.log(e));
  • Executing a sparql update query
const payload = new UpdateQueryPayload()
  .setQuery('INSERT {?s ?p ?o} WHERE {?s ?p ?o}')
return repository.update(payload).then(() => {
    // repository should have been updated at this point


  • Delete statement from given context
repository.deleteStatements(subj, pred, obj, contexts).then(() => {


Repository operations can be executed in transaction. In order to work with transactions the TransactionalRepositoryClient is used.

Starting a transaction

RDFRepositoryClient can initiate a transaction via beginTransaction() which produces an instance of TransactionalRepositoryClient.

Each started transaction allows to be committed or rolled back by using respectively commit() and rollback()

The following is a short use example of a transaction:

const turtlePath = __dirname + '/statements.ttl';
let transactionClient;
return repository.beginTransaction().then((transaction) => {
  transactionClient = transaction;
  return transactionClient.addFile(turtlePath);
}).then(() => {
  // File upload was successful, commit the changes
  return transactionClient.commit();
}).catch((e) => {
  if (transactionClient) {
    // Couldn't upload the file, abort the transaction
    return transactionClient.rollback();
  return Promise.reject(e);

For specific isolation level use TransactionIsolationLevel

return repository.beginTransaction(TransactionIsolationLevel.READ_UNCOMMITTED);

The default isolation level is specific for each store implementation.

Important: After commit or rollback, a transaction cannot be reused, any attempts will result in an error. If you are not sure what is the state of the transaction, you can use transaction.isActive()

Working with a transaction

Almost all of the transaction methods for reading & modifying data have the same syntax and parameters as those in RDFRepositoryClient.


TransactionalRepositoryClient supports the following methods for reading data, including any changes that are not yet committed:

  • getSize()
  • get()
  • download()
  • query()
  • add()
  • addQuads()
  • upload()
  • addFile()


Deleting data during a transaction is different than the one in RDFRepositoryClient, it expects RDF data document instead of statements filter parameters.

Currently it supports only Turtle or TriG formatted RDF data:

const turtlePath = __dirname + '/statements.ttl';
const turtleData = fs.readFileSync(turtlePath, 'utf8');
return transaction.deleteData(turtleData);


  • Retrieving all available namespace declarations. The resolved value is an array of Namespace instances.
return repository.getNamespaces().then((namespaces) => {
  namespaces.forEach((namespace) => {
    console.log(namespace.getPrefix() + ' -> ' + namespace.getNamespace());
  • Retrieving specific namespace declaration
return repository.getNamespace('rdf').then((namespace) => {
  • Setting the namespace declaration. This can act as create or update:
return repository.saveNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
  • Deleting specific namespace declaration
return repository.deleteNamespace('rdf');
  • Deleting all namespaces declarations
return repository.deleteNamespaces();

Repository management

Repository operations like create, edit, delete, shutdown are not supported by the library at the moment. Supporting these is planned for next versions. Follow the issue.


This is a Work In Progress feature (See this pull request)

If the library is going to be used agains a secured server, then all API calls must be authenticated by sending an http authorization header with a token which is obtained after a call to rest/login/user_name with a password provided as a specific header. In case the server requires that requests should be authenticated, then in the ServerClientConfig and RepositoryClientConfig must be configured the username and password which to be used for the authentication. If those are provided, then the client assumes that authentication is mandatory and the login with the provided credentials is performed authomatically before the first API call. After a successful login, user details which are received and the auth token are stored in the AuthenticationService. From that moment on, with every API call is sent also an authorization header with the token as value. If the token expires, then the first API call will be rejected with an http error with status 401. The client handles this automatically by re-login the user with the same credentials, updates the stored token and retries the API call. This behavior is the default and can be changed if the ServerClientConfig or RepositoryClientConfig are configured with keepAlive=false.

Response Parsers

Read responses of different content types might be parsed to data objects with parsers registered in the repository instance.

The library provides a way parsers to be implemented and registered with given repository instance which in turn will use them to parse the response before returning it to the client.

Implementing a custom parser

A parser could be implemented by extending the ContentParser and implementing the parse and getSupportedType methods.

class RdfAsJsonParser extends ContentParser {
  getSupportedType() {
    return 'application/rdf+json';
  parse(content) {
    // parse and return the content
    return parsedContent;

The getSupportedType method must return one of the supported RDF and SPARQL MIME types this way defining that the parser is responsible for converting from that type.

Registering parser in the repository

Parsers should be registered in the repository before executing any request.

// Import any of the predefined parsers
const {NTriplesParser} = require('graphdb').parser;
// And register it in the repository
repository.registerParser(new NTriplesParser());

Multiple parsers could be registered for different response types.

Registering a second parser for same content type results in overriding the previously registerted parser!

Predefined parsers

The library provides convenient parser wrappers for the rdf formats using third party libraries:

  • text/turtle: TurtleParser (N3)
  • text/rdf+n3: N3parser (N3)
  • text/x-nquads: NQuadsParser (N3)
  • text/plain (N-Triples): NTriplesParser (N3)
  • application/x-trig: TrigParser (N3)
  • application/ld-json: JsonLDParser (jsonld-streaming-parser)
  • application/rdf+xml: RDFXmlParser (rdfxml-streaming-parser)

For SELECT query results in json and xml formats as well as boolean results from ASK queries following parsers are wrapped and exposed:

  • application/sparql-results+xml, text/boolean: SparqlXmlResultParser (sparqlxml-parse)
  • application/sparql-results+json, text/boolean: SparqlJsonResultParser (sparqljson-parse)


The library provides basic support of extend RDF with a notion of nested triples, also known as reification. Parsers for RDFStar triples are planned for next versions.

When used against server with RDFStar support, for SELECT queries the following Mime-Types are used:

  • application/x-sparqlstar-results+json
  • application/x-sparqlstar-results+tsv

For DESCRIBE and CONSTRUCT queries, the following Mime-Types can be used:

  • application/x-turtlestar
  • application/x-trigstar
const payload = new GetQueryPayload()
    .setQuery('describe <<<http://www.wikidata.org/entity/Q472> <http://www.wikidata.org/prop/direct/P1889> <http://www.wikidata.org/entity/Q202904>>>')
return repository.query(payload).then((stream) => {
    stream.on('data', (data) => {
       // data contains requested statements in trig star format

When RDFStart triple is requested with non supportive Mime-Types, it resolves to an encoded Base64url string. It can be decoded using TermConverter util class.

const payload = new GetQueryPayload()
    .setQuery('describe <<<http://www.wikidata.org/entity/Q472> <http://www.wikidata.org/prop/direct/P1889> <http://www.wikidata.org/entity/Q202904>>>')
repository.registerParser(new RDFXmlParser());
return repository.query(payload).then((stream) => {
    stream.on('data', (data) => {
       // urn:rdf4j:triple:PDw8aHR0cDovL3d3dy53aWtpZGF0YS5vcmcvZW50aXR5L1E0NzI-IDxodHRwOi8vd3d3Lndpa2lkYXRhLm9yZy9wcm9wL2RpcmVjdC9QMTg4OT4gPGh0dHA6Ly93d3cud2lraWRhdGEub3JnL2VudGl0eS9RMjAyOTA0Pj4-
       // <<<http://www.wikidata.org/entity/Q472> <http://www.wikidata.org/prop/direct/P1889> <http://www.wikidata.org/entity/Q202904>>>



Package Sidebar


npm i iffe-graphdb

Weekly Downloads






Unpacked Size

374 kB

Total Files


Last publish


  • tjcaofei