Awesome

WebSharper.Charting

WebSharper.Charting is a library for creating and rendering client-side charts in the browser. The library is primarily meant for use on the client and by default comes with a client-side renderer implementation, but one could quite easily add different rendering engines.

Design

The two main building blocks of the library are the charts themselves and the renderers. A chart only specifies its data-source and defines some properties (only which make sense) for each chart. There is a clear distinction between live and static charts: static ones show a non-changing set of data while live ones depend on streams and are thus very convenient to display live data with.

There is currently one renderer implementation, which uses Chart.js with our ChartJs bindings. Because of that the current chart types are limited to those that can be rendered by Chart.js but the API is easily extensible with new ones.

The API was designed to try to stay as close to FSharp.Charting as possible so that people who used that wouldn't have to learn a whole new way of doing things. With that said, since we don't roll our own rendering implementation but leverage Chart.js by default instead, we had to align some parts to how the library expects to be handed the data. Because of this we cannot render accurate arbitrary point data like FSharp.Charting can — we need explicit string labels on the x axis — but in turn we can show some fancy animated charts.

Chart configurations are immutable in a sense that if you change the chart through it's public interface you get a new copy of it, but they are mutable in their data-sets. It's very common to want to change the data of the chart after it has been rendered without having to create a new chart so therefor there are methods for modifying the data. Do note though, that these modifications do not affect the initial-data, they are only present as update events, so that rendering implementations can choose to subscribe to them and update the canvas.

Chart Types

Currently there are two kinds of charts each with their own set of types differentiated by what sort of data they show:

• Serial data:
  • line chart
  • bar chart
  • radar chart
• Polar area data:
  • polar area chart
  • pie chart
  • doughnut chart

and there's a special type: combined charts. Combined charts can be created from series charts (as it wouldn't make much sense to combine polar area charts). You can only combine charts of the same static type at the moment but this may change in the future.

All the chart types may have different properties based on which ones make sense for them. For example there's a IColorChart interface which defines methods for changing various color properties of charts. This can only be called on charts which implement the interface, and thus know what the color properties mean for them.

Usage

All the snippets assume the WebSharper, WebSharper.Charting and WebSharper.UI.Next.Client namespaces are open.

Static charts

Defining a new chart:

open WebSharper.Charting

let data = [for x in 1.0 .. 9.0 -> (string x, x * x)]
let chart = Chart.Line data

Modifying properties of the chart:

let data = [for x in 1.0 .. 9.0 -> (string x, x * x)]
let chart =
    Chart.Line(data)
      .WithTitle("Square numbers")
      .WithFillColor(Color.Rgba(120, 120, 120, 0.2))
let blueLine = chart.WithStrokeColor(Color.Name "blue")
let redLine = chart.WithStrokeColor(Color.Name "red")

Rendering charts:

let data = [for x in 1.0 .. 20.0 -> (string x, x * x)]
let chart =
    Chart.Radar(data)
        .WithFillColor(Color.Rgba(30, 30, 120, 0.2))
Renderers.ChartJs.Render(chart, Size = Size(500, 350))
|> Doc.RunById "entry" // where "entry" is the id of some element in the DOM

Updating data:

let months = [|"January"; "February"; "March"; "April"; "May"; "June";
               "July"; "August"; "September"; "October"; "November"; "December"|]
let data = Array.zip months [| for _ in 1 .. 12 -> 0.0 |]
let chart = Chart.Bar data
Renderers.ChartJs.Render(chart, Size = Size(500, 350))
|> Doc.RunById "entry" // where "entry" is the id of some element in the DOM

let rnd = System.Random()
async {
    while true do
        do! Async.Sleep 300
        let i = rnd.Next(0,12)
        let a = rnd.NextDouble() * 10.
        chart.UpdateData(i, fun e -> e + a)
}
|> Async.Start

Combining charts and specifying render parameters:

Chart.Combine [
    Chart.Line([for x in 1.0 .. 9.0 -> string x, x * x])
    Chart.Line([for x in 1.0 .. 9.0 -> string x, x])
        .WithFillColor(Color.Rgba(10, 10, 70, 0.5))
]
|> fun ch ->
    Renderers.ChartJs.Render(ch,
        Config =
            ChartJs.LineChartConfiguration(
                DatasetFill = true,
                BezierCurve = true)
)
|> Doc.RunById "entry"

Live charts

Defining a new live chart:

let source = Event<float>()
let chart = LiveChart.Line source.Publish

Pushing data into the stream and rendering it:

let source = Event<float>()
LiveChart.Line source.Publish
|> fun ch -> Renderers.ChartJs.Render(ch, Window = 10) // only display 10 points of data max
|> Doc.RunById "entry"


let rnd = System.Random()
async {
    while true do
        do! Async.Sleep 600
        let d = rnd.NextDouble() * 300.
        source.Trigger d
}
|> Async.Start

Combining live charts (there's a catch here: the chart will only update if there's new value on all streams):

open IntelliFactory.Reactive

let source = Event<float>()

let averageByPoint =
    Reactive.Select
    <| Reactive.Aggregate source.Publish (0., 0)
        (fun (p, i) c ->
            let nc = (p * float i + c) / float (i + 1)
            (nc, i + 1))
    <| fst

Chart.Combine [
    LiveChart.Line source.Publish
    LiveChart.Line(averageByPoint)
        .WithStrokeColor(Color.Name "blue")
]
|> fun ch -> Renderers.ChartJs.Render(ch, Window = 10) // only display 10 points of data max
|> Doc.RunById "entry"


let rnd = System.Random()
async {
    while true do
        do! Async.Sleep 600
        let d = rnd.NextDouble() * 300.
        source.Trigger d
}
|> Async.Start

We expect the averageByPoint stream to converge to 150 here since we are generating uniform random data.

Rendering

The default renderers can be accessed from the WebSharper.Charting.Renderers.ChartJs namespace. As you can see in the Usage section there are some arguments you can pass here to the renderer. Size defines how large the rendering canvas should be, Window specifies how many data points can be at max on the chart (especially useful with live charts) and there's a Config parameter which depends on what kind of chart you are rendering, and specifies how different charts should be displayed. For a detailed explanation on these please consult the Chart.js documentation.

Data Sources

There are two types of main data sources you can build charts from: static and live. You don't have to manually deal with them just tell the library if you want to instantiate a LiveChart with a stream or a static Chart with a seq and the renderer will take care of using the appropriate rendering function.

The series charts expect to be given label and value pairs in the format of string * float. Currently we do not support feeding them any other type of data, you will have to convert it manually. You can construct them with only the float values too, in which case the label will be the index starting from 0.

The polar area charts expect to be given PolarData which — along with the label and value — specifies colors for the segment in the chart. If you only specify the label and the value in the form of string * float the colors will be auto-generated. The random color will be the same for both the fill color and the highlight color, only their opacity will differ.

Type Hierarchy

The base type of all charts is the IChart interface. All charts (expet from CompositeChart). There are a couple other interfaces defined with different traits of charts:

• ISeriesChart is a series chart to which fill and stroke coloring can be applied for instance,
• IMutableChart indicates that the chart can be mutated with special methods, and others can subscribe to these update events,
• IColorChart is a chart which you can apply different colorings to,
• IPolarAreaChart is a polar are chart (pie and doughnut charts e.g.).

In order for this to play nicely with immutability IChart — and thus the interfaces inerhiting from it — have a "self-bound" type parameter which indicates the type of the implementing class so that we can require modification methods to return the same element.

The idea is:

[<Interface>]
type IChart<'Self when 'Self :> IChart<'Self>> =
    abstract member WithTitle : string -> 'Self

type EmptyChart(title : string) =
    interface IChart<EmptyChart> with
        override x.WithTitle t = EmptyChart(t)

So if you want to define a new chart type you should implement these interfaces.