elgrapho

2.4.0 • Public • Published

El Grapho

What is El Grapho?

El Grapho is a high performance WebGL graph data visualization engine. El Grapho can support millions of interactive nodes and edges in any modern browser.

https://www.elgrapho.com/

Why Would I Use This?

If you need to build a graph visualization for the web of any kind, such as a tree, force directed graph, network graph, etc., and scale and performance are important to you, then El Grapho is a great option. El Grapho was built to push the limits of modern browsers.

Live Examples

Getting Started

To get started, you can install El Grapho from npm like this

npm install --save elgrapho

or you can download the latest El Grapho distribution file found here

https://github.com/ericdrowell/ElGrapho/blob/master/engine/dist/ElGrapho.min.js

API

To create a simple El Grapho data visualization, you can instantiate a new graph like this

let graph = new ElGrapho({
  container: document.getElementById('container'),
  model: {
    nodes: [
      {x: 0,    y: 0.6,  group: 0, label: 0},
      {x: -0.4, y: 0,    group: 1, label: 1},
      {x: 0.4,  y: 0,    group: 1, label: 2},
      {x: -0.6, y: -0.6, group: 2, label: 3},
      {x: -0.2, y: -0.6, group: 2, label: 4},
      {x: 0.2,  y: -0.6, group: 2, label: 5},
      {x: 0.6,  y: -0.6, group: 2, label: 6}
    ],
    edges: [
      {from: 0, to: 1},
      {from: 0, to: 2},
      {from: 1, to: 3},
      {from: 1, to: 4},
      {from: 2, to: 5},
      {from: 2, to: 6}
    ]
  },
  width: 500,
  height: 500
});
  • container - DOM element that will contain the El Grapho graph.

  • model.nodes - object that defines the nodes in the graph. Each node is defined by a position (x and y), and also a group. El Grapho x and y ranges are between -1 and 1. If x is -1, then the node position is on the very left of the viewport. If x is 0 it is in the center, and if x is 1 it is on the very right of the viewport. Groups are integer values, and El Grapho automatically assigns colors to each unique grouping. To add labels, simply add a label key and value pair to each node.

  • model.edges - object that defines the edges between nodes based on their indices. Each edge is defined by a from-node-index and a to-node-index. In the example above, the first edge begins at node 0 and ends at node 1. For non directed graphs, or bi-directional graphs, from and to are interchangeable.

  • width - number that defines the width of the El Grapho viewport in pixels. The default is 500.

  • height - number defines the height of the El Grapho viewport in pixels. The default is 500.

  • nodeSize - number between 0 and 1 which defines the node size. The default is 1.

  • nodeOutline - boolean that enables or disables node outlines. The default is true.

  • edgeSize - number between 0 and 1 which defines the edge size. Edge sizes are relative to the connecting node size. The default is 0.25.

  • darkMode - boolean that enables or disables dark mode. The default is false.

  • glowBlend - number between 0 and 1 that defines the glow blending of nodes and edges. A value of 0 has no glow blending, and a value of 1 has full glow blending. Glow blending can be used as a visual treatment to emphasize node clustering or edge bundling. It is most effective when used in conjunction with dark mode. The default is 0.

  • fillContainer - boolean that enables or disables auto filling the container. When true, El Grapho will automatically detect anytime its container has changed shape, and will auto resize itself. The default is false.

  • tooltips - boolean that enables or disables tooltips. The default is true.

  • arrows - boolean that enables or disables edge arrows. For non directed or bi-directional graphs, you should keep arrows as false. The default is false.

  • animations - boolean that defines animation strategy. When animations is true, zoom and pan transitions will be animated. Otherwise the transitions will be immediate. Although animations utilize requestAnimationFrame for dynamic frame rates, in some situations you may prefer to set animations to false to improve transition performance for very high cardinality graphs with millions of nodes and edges. The default is true.

  • debug - boolean that can be used to enable debug mode. Debug mode will show the node and edge count in the bottom right corner of the visualization. The default is false.

Layouts

Determining the positions of the nodes for your graph can be a lot of work! While it's nice to have the power to construct custom graph layouts, most El Grapho users will want to leverage the provided El Grapho layouts which will generate node positions for you. Currently, El Grapho supports ForceDirected, Tree, RadialTree, Hairball, Chord, and Cluster

ForceDirected Layout

let model = {
  nodes: [
    {group: 0},
    {group: 1},
    {group: 1},
    {group: 2},
    {group: 2},
    {group: 3}
  ],
  edges: [
    {from: 0, to: 1},
    {from: 0, to: 2},
    {from: 1, to: 3},
    {from: 1, to: 4},
    {from: 2, to: 5},
  ],
  steps: 30
};

graph = new ElGrapho({
  container: document.getElementById('container'),
  model: ElGrapho.layouts.ForceDirected(model),
  width: 500,
  height: 500
});

The ForceDirected layout uses d3-force to reduce edge crossing and optimize node spacing. Although the algorithm is fairly efficient with a runtime of O(nlog(n)) via the Barnes-Hut approximation which uses quad trees, this can still be slow for very large graphs of 50k nodes or more. If it's possible to build your models in advance, it's a good idea to build the force directed graph model on the server and then cache it. If you require your models to be constructed at execution time, and the number of nodes is very high, you may consider using an O(n) model substitude such as HairBall or Cluster

The ForceDirected layout accepts a steps property from the model config which can be used to define the accuracy of the result. This is because force directed graphs require multiple passes to obtain a final result. The default is 30. Lowering this number will result in faster model construction but less accurate results. Increasing this number will result in slower model construction but more accurate results.

Tree Layout

let model = {
  nodes: [
    {group: 0},
    {group: 1},
    {group: 1},
    {group: 2},
    {group: 2},
    {group: 3},
    {group: 3}
  ],
  edges: [
    {from: 0, to: 1},
    {from: 0, to: 2},
    {from: 1, to: 3},
    {from: 1, to: 4},
    {from: 2, to: 5},
    {from: 2, to: 6}
  ]
};

let graph = new ElGrapho({
  container: document.getElementById('container'),
  model: ElGrapho.layouts.Tree(model),
  width: 500,
  height: 500
});

RadialTree Layout

let model = {
  nodes: [
    {group: 0},
    {group: 1},
    {group: 1},
    {group: 2},
    {group: 2},
    {group: 3},
    {group: 3}
  ],
  edges: [
    {from: 0, to: 1},
    {from: 0, to: 2},
    {from: 1, to: 3},
    {from: 1, to: 4},
    {from: 2, to: 5},
    {from: 2, to: 6}
  ]
};

let graph = new ElGrapho({
  container: document.getElementById('container'),
  model: ElGrapho.layouts.RadialTree(model),
  width: 500,
  height: 500
});

Hairball Layout

let model = {
  nodes: [
    {group: 0},
    {group: 1},
    {group: 1},
    {group: 2},
    {group: 2},
    {group: 3},
    {group: 3}
  ],
  edges: [
    {from: 0, to: 1},
    {from: 0, to: 2},
    {from: 1, to: 3},
    {from: 1, to: 4},
    {from: 2, to: 5},
    {from: 2, to: 6}
  ]
};

let graph = new ElGrapho({
  container: document.getElementById('container'),
  model: ElGrapho.layouts.Hairball(model),
  width: 500,
  height: 500
});

The Hairball layout is an excellent alternative to ForceDirected layouts when the number of nodes is very high, for example in the millions. This is because it runs in O(n) time, i.e. linear time. Essentially, nodes with lots of edges are forced towards the center, and nodes with few edges are pushed out to the perimeter, creating a hairball effect.

Cluster Layout

let model = {
  nodes: [
    {group: 0},
    {group: 1},
    {group: 1},
    {group: 2},
    {group: 2},
    {group: 3},
    {group: 3}
  ],
  edges: [
    {from: 0, to: 1},
    {from: 0, to: 2},
    {from: 1, to: 3},
    {from: 1, to: 4},
    {from: 2, to: 5},
    {from: 2, to: 6}
  ]
};

let graph = new ElGrapho({
  container: document.getElementById('container'),
  model: ElGrapho.layouts.Cluster(model),
  arrows: true,
  width: 500,
  height: 500
});

The Cluster layout clusters nodes by group. If a single group is used for all of the nodes, El Grapho will generate a single centered cluster. If there are several groups used, El Grapho will render distinct clusters. Because Cluster layouts can be generated in O(n) time, i.e. linear time, they are another great alternative to ForceDirected if performance becomes an issue.

Chord Layout

let model = {
  nodes: [
    {group: 0},
    {group: 1},
    {group: 1},
    {group: 2},
    {group: 2},
    {group: 3},
    {group: 3}
  ],
  edges: [
    {from: 0, to: 1},
    {from: 0, to: 2},
    {from: 1, to: 3},
    {from: 1, to: 4},
    {from: 2, to: 5},
    {from: 2, to: 6}
  ]
};

let graph = new ElGrapho({
  container: document.getElementById('container'),
  model: ElGrapho.layouts.Chord(model),
  width: 500,
  height: 500
});

Layout Polymorphism

You may have noticed that the layout config schema is identical for all layouts. As a result, El Grapho visualizations are polymorphic, meaning you can pass the same data structure into different layouts and get different graphs. Pretty cool!

Server Side Model Generation

Because the El Grapho layouts are fully decoupled from the rendering engine itself, they can be executed on the client or on the server, depending on your needs. For really complex layouts, it may be best to build the model on the server, deliver the output over http to the browser, and pass it directly into the El Grapho config.

Controls

El Grapho has controls in the upper right corner of the visualization that enable users to navigate large and complex graphs. These controls appear when you mouseover the visualization. There are three modes:

  • select - use this mode to select nodes
  • zoom - use this mode to draw zoom boxes around areas of interest or to zoom into a particular region of the graph
  • pan - use this mode to pan the visualization around

And there are three action buttons:

  • reset
  • zoom in
  • zoom out

Tooltips

El Grapho ships with a default template and default content. It is assumed however that you will be providing your own tooltip content (at the end of the day, most people want something custom anyways). To set the tooltip template, simply do the following:

let graph = new ElGrapho(config);

graph.tooltipTemplate = function(index, el) {
  el.innerHTML = 'node index: ' + index;
};

This means that your tooltips can be anything! You can show lots of information about the node, insert images, etc. The tooltip template is decoupled from the El Grapho config in order to ensure that the config is serializable and thus transferable over http.

To disable tooltips entirely, you can set the tooltips property to false in the graph config.

Events

El Grapho also has a built in event bus. Currently, the supported events are:

  • idle
  • node-mouseover
  • node-mouseout
  • node-click

and you can use these events like this:

let graph = new ElGrapho(config);

graph.on('idle', function() {
  console.log('idle');
});

graph.on('node-mouseover', function(evt) {
  console.log('node-mouseover: ' + evt.dataIndex);
});

Methods

  • graph.selectGroup(index) - select a group by index, where index is an integer

  • graph.deselectGroup() - deselect the currently selected group

  • graph.setSize(width, height) - set the graph size in pixels, where width and height is a number

  • graph.setDarkMode(mode) - enable or disable dark mode, where mode is a boolean

  • graph.selectNode(index) - select node by index, where index is an integer

  • graph.deselectNode() - deselect a selected node

About the Name and Logo

Why is this called El Grapho? - Why not?

Okay... and why is the logo a skeleton with a rose in his mouth? - Why not?

Package Sidebar

Install

npm i elgrapho

Weekly Downloads

0

Version

2.4.0

License

MIT

Unpacked Size

18.3 MB

Total Files

93

Last publish

Collaborators

  • ericdrowell