Duality

[WARNING: This package is still very alpha. This mesasage will be removed when this thing is actually kinda ready for real work.]

Duality is a framework for writing and running data transformation pipelines. Pipelines are built from components that use the Node.js Streams system to move data down the pipeline, enriching or transforming it as it goes.

Duality makes it easy to create, run, and manage pipeline scripts. It offers sensible conventions to make it easy to manage and maintain a large library of scripts.

Example

This simple pipeline queries contact records from a database, and splits each record into separate 'person' and 'company' records, then inserts the results into a second database.

p = new Pipeline();
p.use(new database.Query('select * from contacts'));
p.use(function(contact) {
    var companyId = GUID();
    return [
        {
            _headers: {type: 'person'},
            email: contact.email,
            name: contact.firstname + ' ' + contact.lastname
            company_id: companyId
        },
        {
            _headers: {type: 'company'},
            name: contact.compnay_name,
            guid: companyId
        }
    ]
});
p.use(util.Logger());
p.use(new database.insert(null, 'database1'));

p.run()

Features

• Efficient data flow based on Node Streams
• Toolbox of useful input, output, and transform components
• Easy configuration system
• Simple syntax for creating pipelines
• Easily create new, re-usuable components
• Built-in logging system
• Flexible runtime system

Messages

Duality pipelines operate in objectMode, passing discrete JS objects as messsages. Objects can use any format. By convention objects may contain a key _headers which holds a dictionary of metadata to describe the message.

Components

Components are the basic unit of logic in a pipeline. Components can either generate new messages, transform existing messages, or write messages to some output. Duality includes a set of standard components, and makes it very easy to create new ad-hoc components from simple functions. It's also possible to construct new re-usuable components by subclassing the standard classes in the Node stream package.

Ad-hoc components

To create an ad-hoc component, simply pass a function to the use method on a Pipeline. Synchronous functions are passed each message in the pipeline, and the return value from the function is passed on down the pipeline:

p.use(function(msg, context) {
    msg.newkey = 'newvalue';
    return msg;
});

If you need to return results asynchronously just define your function to take a callback second argument:

p.use(function(msg, context, callback) {
    resource.get(function(rows) {
        rows.forEach(function(row) {
            callback(row, true);
        });
        callback(null);
    });
});

Invoke the callback with output messages and pass true as the second argument until there are no more messages to generate.

The context

The context is a global object available in the pipeline. It is typically used to pass configuration data into a component. The configuration for the pipeline is passed to the pipeline constructor, and becomes available under the config key in the context:

p = new Pipeline('test max', {max:10});
p.use(function(msg, context) {
    console.log("Max is: ", context.config.max);  <-- prints: Max is 10
});

Components may add data to the context in order to share that data with other components. For example, the database.Connection components adds the database connection object to the context.

Running your pipeline

Any pipeline can be executed by calling run:

p = new Pipeline();
...
p.run();

This works for ad-hoc scripts, but Duality includes support for multiple types of triggers. Triggers activate the pipeline, and may supply 0 or more initial messages. The timer trigger can run a pipeline on an interval. The http_get and http_post triggers are activated by HTTP calls and will provide their inputs as message(s) to the pipeline.