Vega-Lite V5 for Rust

A Rust API for Vega-Lite V4 to build chart with a rusty API.

Similar to the Altair project in python, this crate build upon Vega-Lite specifications. Vega-Lite is a high-level grammar of interactive graphics. It provides a concise JSON syntax for rapidly generating visualizations to support analysis. Vega-Lite specifications can be compiled to Vega specifications. Those specifications are then parsed by Vega’s JavaScript runtime to generate both static images or interactive web-based views. This crate has a complete mapping of Vega-Lite 3.4 specification and can be found in src/schema.rs. With all the types and structs, it's possible to create your Rust Vegalite graph that will be serialize into a Vega-Lite JSON. Thanks to Showata the resulting visualization can be display in your Web-Browser or in a Rust Jupyter Notebook. It's also possible to use an existing Vega-Lite json and plug your data source seamlessly. This way you can leverage existing vizualisation and adapt it to your design.

Examples

In order to have a complete mapping of the Vega-Lite V4 specification the code for the schema was automaticlly generated. To help describe all the possible features a gallery of example is provided on github

To launch all examples

cargo install cargo-make
cargo make run-all-examples

Simple chart using ndarray generated data

let values: Array2<f64> = Array::random((100, 2), StandardNormal);

let chart = VegaliteBuilder::default()
    .title("Random points")
    .data(values)
    .mark(Mark::Point)
    .encoding(
        EncodingBuilder::default()
            .x(XClassBuilder::default()
                .field("data.0")
                .def_type(StandardType::Quantitative)
                .build()?)
            .y(YClassBuilder::default()
                .field("data.1")
                .def_type(StandardType::Quantitative)
                .build()?)
            .build()?,
    )
    .build()?;
chart.show()?;

Simple chart using existing json definition with new data

// Use existing vega-lite json specification
let spec = r##"{
    "$schema": "https://vega.github.io/schema/vega-lite/v5.json",
    "encoding": {
        "x": {
            "field": "data.0",
            "type": "quantitative"
        },
        "y": {
            "field": "data.1",
            "type": "quantitative"
        }
    },
    "mark": "point",
    "title": "Random points"
}"##;

// Use you own data to populate the chart
let values: Array2<f64> = Array::random((100, 2), StandardNormal);
let mut chart: Vegalite = serde_json::from_str(spec)?;
chart.data = values.into();

// display the chart using `showata`
chart.show()?;

Features

Links

• Wiki - AGuideToRustGraphicsLibraries2019
• A Dramatic Tour through Python’s Data Visualization Landscape (including ggplot and Altair) – Regress to Impress
• Specifying Data in Altair — Altair 3.0.0 documentation
• Visualization — list of Rust libraries/crates // Lib.rs
• Quicktype (got issue with the alternative) was used to bootstrap src/schema.rs from the vega-lite's json schema

Troubleshoot

Stack size

The vegalite json schema is large with lot of alternative, so the typed rust version create a large set of struct and enum (the generated source file before macro expension is 28K lines). So the size of a model in stack could be large (it's also why Box is used in the struct).

On wasm32, with the default stack size (~ 1 MB), using vegalite_5 can raise error like:

• crash tab with SIGSEVG (on chromium based browser)
• Uncaught (in promise) RuntimeError: memory access out of bounds or simply Uncaught (in promise) RuntimeError

The current work arround is to increase the stacksize (eg ~ 1.5 MB). For cargo based project you can add into .cargo/config file of the project:

[target.wasm32-unknown-unknown]
rustflags = [
  "-C", "link-args=-z stack-size=1500000",
]