tmccarthy / tmm-scala-plotly

Yet another Scala API for Plotly.js

Version Matrix

tmm-scala-plotly

CircleCI Maven Central

Yet another library for integration between Plotly and Scala.

Getting started

In your build.sbt file.

tmmScalaPlotlyVersion = 0.0.2

libraryDependencies += "au.id.tmm.tmm-scala-plotly" %% "tmm-scala-plotly-core" % tmmScalaPlotlyVersion

Rationale

This library is a representation of the Plotly JavaScript API in Scala. It is a hand-written and almost complete representation of the Plotly API as documented in DefinitelyTyped, and informed by the official documentation.

There are already at least two other libraries that provide Scala integration for Plotly:

My advice would be that if you are looking for a library to use Plotly from Scala, you should use alexarchambault/plotly-scala. It is more mature and better tested than this library, and I've had a lot of success with it.

This library has a couple of key philosophical differences with alexarchambault/plotly-scala, which is why I decided to write it:

  • Completeness: This library is an almost exhaustive representation of the Plotly Javascript API. You should be able to produce every kind of plot supported by the official Javascript library.
  • Self-documenting: This library makes extensive use of value classes and algebraic data types to attempt to clarify the meaning of fields within the Plotly API. Implicit conversions are opt-in and are intended only for ease-of-use (see below).
  • Incorrect: This library does not (yet) use the power of the Scala type system to ensure valid use of Plotly. For example, you can provide z co-ordinates to two-dimensional scatter plots. This is an intentional tradeoff to make it simpler to produce an exhaustive representation of the Javascript API.
  • No ScalaJS support: I don't use ScalaJS, so I haven't provided support for it.
  • Circe as a first-class citizen: This project uses circe to encode the model to Json. The circe integrations are deliberately exposed by the main library to give clients as much flexibility as possible when encoding.

Note that I've hand-written this interface over Plotly but have done almost no thorough testing. Many of the corners of this library are unlikely to have seen any use. Caveat emptor.

Usage

To draw a pie chart:

import au.id.tmm.plotly._
import au.id.tmm.plotly.model._
import au.id.tmm.plotly.model.utilities._

val (labels, values, colours) = List(
  ("Coalition", 5_906_875, Color("blue")),
  ("Labor", 4_752_160, Color("red")),
  ("Greens", 1_482_923, Color("green")),
  ("Other", 2_111_435, Color("grey")),
).unzip3

val plot = Plot(
  data = List(
    Trace(
      `type` = OptArg.Of(Trace.Type.Pie),
      labels = OptArg.Of(DataArray.OfStrings(labels)),
      values = OptArg.Of(DataArray.OfInts(values)),
      marker = OptArg.Of(
        PlotMarker(
          colors = OptArg.Of(colours),
        ),
      ),
    ),
  ),
  layout = OptArg.Of(
    Layout(
      title = OptArg.Of(
        Layout.Title(
          text = OptArg.Of("2019 primary votes"),
        ),
      ),
      width = OptArg.Of(300d),
      height = OptArg.Of(300d),
    ),
  ),
)

Plotting.openInBrowser(plot)

A pie chart showing the primary votes at the 2019 Australian Federal Election

Implicit conversions for ergonomics

In the above you should note that we make use of types like OptArg and DataArray. These allow us to have explicit and type-safe representations of optional parameters and the different types of data array respectively. Using these everywhere gets tiring, so we can import au.id.tmm.plotly.syntax._ to make things a little less verbose while retaining type safety. With this import, the above becomes:

import au.id.tmm.plotly._
import au.id.tmm.plotly.model._
import au.id.tmm.plotly.syntax._

val (labels, values, colours) = List(
  ("Coalition", 5_906_875, Color("blue")),
  ("Labor", 4_752_160, Color("red")),
  ("Greens", 1_482_923, Color("green")),
  ("Other", 2_111_435, Color("grey")),
).unzip3

val plot = Plot(
  data = List(
    Trace(
      `type` = Trace.Type.Pie,
      labels = labels,
      values = values,
      marker = PlotMarker(colors = colours),
    ),
  ),
  layout = Layout(
    title = Layout.Title(text = "2019 primary votes"),
    width = 300,
    height = 300,
  ),
)

Plotting.openInBrowser(plot)

Dedicated trace interfaces

As discussed in the Rationale section above, the Trace class provides an almost-exhaustive representation of the Plotly Javascript interface. While this allows us to use most of the features of Plotly, it also allows us to provide useless or invalid parameters. For example, we could provide the z co-ordinates for a two-dimensional scatter plot.

As a partial solution to this, the project includes some dedicated constructors for traces based on the trace type. These are available in the au.id.tmm.plotly.model.traceinterfaces package. For example, an interface is provided for sunburst plots and is demoed in the examples project.