chartjs-plugin-streaming

Chart.js plugin for live streaming data

The latest version of chartjs-plugin-streaming requires Chart.js 3.0.0 or later. If you need Chart.js 2.x support, use the following versions.

• For Chart.js 2.9.x, 2.8.x or 2.7.x, use version 1.9.0
• For Chart.js 2.6.x, use version 1.2.0

Installation

You can download the latest version of chartjs-plugin-streaming from the GitHub releases.

To install via npm:

npm install chartjs-plugin-streaming --save

To use CDN:

<script src="https://cdn.jsdelivr.net/npm/chartjs-plugin-streaming@latest"></script>

Usage

chartjs-plugin-streaming can be used with ES6 modules, plain JavaScript and module loaders.

chartjs-plugin-streaming requires both a date library and a corresponding adapter to display time scales. Please choose from the available adapters.

Version 2.0 supports the line and bar chart types with both Primitive[] and Object[] data (each data point is specified an array of objects containing x and y properties) as well as the bubble and scatter chart types with Object[] data. In case of Object[] data, either x or y must be in any of the date formats that the date library accepts (Formatting in Luxon for example), and the corresponding axis must have a 'realtime' scale that has the same options as time scale. Once the realtime scale is specified, the chart will auto-scroll along with that axis. Old data will be automatically deleted after the time specified by the ttl option, or as it disappears off the chart.

Usage in ES6 as module

Nothing else than importing the module should be enough.

import 'chartjs-plugin-streaming';

Tutorial and Samples

You can find a tutorial and samples at nagix.github.io/chartjs-plugin-streaming.

Configuration

The plugin options can be changed at 3 different levels and with the following priority:

• per axis: options.scales[scaleId].realtime.*
• per chart: options.plugins.streaming.*
• globally: Chart.defaults.plugins.streaming.*

All available options are listed below. This example shows how each option affects the appearance of a chart.

Name	Type	Default	Description
duration	number	10000	Duration of the chart in milliseconds (how much time of data it will show).
ttl	number		Duration of the data to be kept in milliseconds. If not set, old data will be automatically deleted as it disappears off the chart.
delay	number	0	Delay added to the chart in milliseconds so that upcoming values are known before lines are plotted. This makes the chart look like a continual stream rather than very jumpy on the right hand side. Specify the maximum expected delay.
refresh	number	1000	Refresh interval of data in milliseconds. onRefresh callback function will be called at this interval.
onRefresh	function	null	Callback function that will be called at a regular interval. The callback takes one argument, a reference to the chart object. You can update your datasets here. The chart will be automatically updated after returning.
frameRate	number	30	Frequency at which the chart is drawn on a display (frames per second). This option can be set at chart level but not at axis level. Decrease this value to save CPU power. more...
pause	boolean	false	If set to true, scrolling stops. Note that onRefresh callback is called even when this is set to true.

Note that the following axis options are ignored for the 'realtime' scale.

• bounds
• max
• min
• offset (always false)
• ticks.autoSkip (always false)
• ticks.major.enabled (always true)

Data Feed Models

This plugin supports both pull and push based data feed.

Pull Model (Polling Based) - Synchronous

In the pull model, the user code needs to ask for new data and pull it from a data source. To enable this, the plugin provides two options: onRefresh which is the callback function that is called at a regular interval to check the data source and refresh which specifies the interval. In this callback function, you can add data into the existing data array as usual, but you don't need to call the update function as it is called internally.

For example:

{
  type: 'line',             // 'line', 'bar', 'bubble' and 'scatter' types are supported
  data: {
    datasets: [{
      data: []              // empty at the beginning
    }]
  },
  options: {
    scales: {
      x: {
        type: 'realtime',   // x axis will auto-scroll from right to left
        realtime: {         // per-axis options
          duration: 20000,  // data in the past 20000 ms will be displayed
          refresh: 1000,    // onRefresh callback will be called every 1000 ms
          delay: 1000,      // delay of 1000 ms, so upcoming values are known before plotting a line
          pause: false,     // chart is not paused
          ttl: undefined,   // data will be automatically deleted as it disappears off the chart

          // a callback to update datasets
          onRefresh: function(chart) {

            // query your data source and get the array of {x: timestamp, y: value} objects
            var data = getLatestData();

            // append the new data array to the existing chart data
            Array.prototype.push.apply(chart.data.datasets[0].data, data);
          }
        }
      }
    },
    plugins: {
      streaming: {          // per-chart option
        frameRate: 30       // chart is drawn 30 times every second
      }
    }
  }
}

Pull Model (Polling Based) - Asynchronous

If your data source responds to requests asynchronously, you will probably receive the results in a callback function in which you can add data into the existing data array. The update function needs to be called after adding new data.

This model is suitable for data sources such as web servers, Kafka (REST Proxy), Kinesis (Data Streams API) and other time series databases with REST API support including Elasticsearch, OpenTSDB and Graphite.

For example:

{
  options: {
    scales: {
      x: {
        realtime: {
          onRefresh: function(chart) {
            // request data so that it can be received in a callback function
            var xhr = new XMLHttpRequest();
            xhr.open('GET', YOUR_DATA_SOURCE_URL);
            xhr.onload = function () {
              if (xhr.readyState === 4 && xhr.status === 200) {

                // assume the response is an array of {x: timestamp, y: value} objects
                var data = JSON.parse(xhr.responseText);

                // append the new data array to the existing chart data
                Array.prototype.push.apply(chart.data.datasets[0].data, data);

                // update chart datasets without animation
                chart.update('none');
              }
            };
            xhr.send();
          }
        }
      }
    }
  }
}

Push Model (Listening Based)

In the push model, the user code registers a listener that waits for new data, and data can be picked up immediately after it arrives. Usually, data source connector libraries that support the push model provide a listener callback function in which you can add data into the existing data array. onRefresh is unnecessary in this model, but the update function needs to be called after adding new data like asynchronous pull model.

This model is suitable for data sources such as WebSocket, MQTT, Kinesis (Client Library) and other realtime messaging services including Socket.IO, Pusher and Firebase.

Here is an example of a listener function:

// save the chart instance to a variable
var myChart = new Chart(ctx, config);

// your event listener code - assuming the event object has the timestamp and value properties
function onReceive(event) {

  // append the new data to the existing chart data
  myChart.data.datasets[0].data.push({
    x: event.timestamp,
    y: event.value
  });

  // update chart datasets without animation
  myChart.update('none');
}

Support for Zooming and panning

By using together with chartjs-plugin-zoom, zooming and panning of a streaming chart can be done via the mouse or finger gestures. Unlike other scale types, the min, max and minRange in the limits options are not used. Instead, the minDelay, maxDelay, minDuration and maxDuration limit the range of the delay and duration option values.

  options: {
    // Assume x axis is the realtime scale
    pan: {
      enabled: true,        // Enable panning
      mode: 'x'             // Allow panning in the x direction
    },
    zoom: {
      enabled: true,        // Enable zooming
      mode: 'x'             // Allow zooming in the x direction
    },
    limits: {
      x: {
        minDelay: null,     // Min value of the delay option
        maxDelay: null,     // Max value of the delay option
        minDuration: null,  // Min value of the duration option
        maxDuration: null   // Max value of the duration option
      }
    }
  }

Note that chartjs-plugin-zoom.js needs to be included before chartjs-plugin-streaming.js.

Lowering CPU Usage

If you are using this plugin on resource constrained devices or drawing multiple charts on a large screen, it might be a good idea to decrease the frame rate to lower CPU usage. The following settings also reduce CPU usage by disabling animations, and improve general page performance.

  options: {
    animation: false,
    plugins: {
      streaming: {
        frameRate: 5  // chart is drawn 5 times every second
      }
    }
  }

Building

You first need to install node dependencies (requires Node.js):

npm install

The following commands will then be available from the repository root:

npm run build      # build dist files
npm run build:dev  # build and watch for changes
npm run docs       # build docs
npm run docs:dev   # build docs and run the doc server locally
npm run lint       # perform code linting
npm run package    # create an archive with dist files and samples

License

chartjs-plugin-streaming is available under the MIT license.