Dear reader, if you got someting out of my blog and if it helped you, it would be great if you supported me by clicking on the Google ads-banner at the right. This will not cost you anything but time but will generate a little bit of revenue for my time invested ;)
I'm as well happy if you LIKE or 1+ it and about comments that help to improve my articles.

Wednesday, November 25, 2015

Building Data Driven Applications: Combining Vis.js with Paper.js


Currently I am working on a visualization project where I want display real world entities with the help of a network graph. Additional to the classical network graph and graphics I want to display information like groups of network nodes, or I want to highlight groups of nodes based on node attributes. Here an example of combined information:
The network graph displays the connection of "things"/entities and the colored areas show groups of these entities.
All the data I am using is stored in a database, so I was looking for a way to visualize these data with the least effort and media breaks. Furthermore, one factor for the choice of tools was the possibility to deploy visualized data with the least possible effort and in best case as view on "real time" data in the database.

Frameworks for Network Graphs

To achieve my task I evaluated a couple of tools and frameworks. My first choice was GEPHI, which is available in version 0.8.2. GEPHI provides means to layout quite large numbers of nodes and edges into nicely looking network graphs and it is quite popular among scientists and practitioners using networks. However, I found a couple of drawbacks which made me to leave the. GEPHI path: first of all GEPHI is a closed, standalone program. Hence, it is hard to enhance a readily layouted graph with additional information and this requires additional external tools. Furthermore it is not possible to provide interactive, real time graphs which can be presented to an end user. The GEPHI workflow requires the following steps: draw data from database (GEPHI has a direct DB interface), layouting of the network with some algorythm, export visualization to a format which allows post processing - e.g. SVG - and do the actual post processing.
Finaly I decided to abandon the GEPHI path as it involved too many steps with human intervention and turned to web-based tools. Quite quickly I found javascript frameworks as my weapon of choice and evaluated the following frameworks for network layouting: D3.js, cytoscape.js,  sigma.js and vis.js. To keep things short, I ended up with Vis.js, mostly because it was the only framework which allowed me to pin down nodes via configuration file (a JSON I export from database) and because it provides a "force atlas 2"-like layouting algorithm which yields networks layouts I was aiming for.

Frameworks for HTML5 Graphics

For the "additional" graphics frameworks I ended up quickly with the "big three": Paper.js, Processing.js and Raphael.js. I started with Raphael.js which I knew from a project I was involved before, but it quickly turned out that Raphael's SVG cannot or is hard to combine with vis.js as the latter one uses the HTML5 canvas. I didn't get warm with Processing.js as it uses its own scripting language, so finally I ended up with Paper.js, which for my taste yields satisfactory results.

Vis.js' Door to other Frameworks

The first steps of combining vis.js and paper.js were not easy. I managed to overcome the first obstacles with the help of the vis.js team (thanks for that). Based on one of the vis.js network examples I was able to hook Paper.js into Vis.js.

As the first step you have to tell Paper.js to paint on the Vis.js canvas. This can be done in the main program by the following lines:
[...] // Things required to setup network, etc.
// "mynetwork" is defined in html <body>
var container = document.getElementById('mynetwork');
var network = new vis.Network(container, data, options);
[...] // More network stuff here if needed
paperCanvas = container.firstChild.canvas;

The main entry points to inject other graphics into the Vis .js canvas are the Vis.js network events
  • network.on("beforeDrawing", function(ctx) { [...] }) this will draw graphics behind the network graph
  • network.on("afterDrawing", function(ctx) { [...] }) this will draw in front of th network graph
With these hook events one can start to draw Paper.js elements and display them together with the network graph. You can put any Paper.js related code replacing the square brackets. I use Vis.js Network/DataSet methods to extract geometric data on positions, etc. and pass them to Paper.js to draw elements related to the network nodes, etc.

I hope this posts helps you with your own project. Keep on geekin', your