/react-native-skia

High-performance React Native Graphics using Skia

Primary LanguageTypeScriptMIT LicenseMIT

React Native Skia

High-performance 2d Graphics for React Native using Skia

Tests npm version issues

skia

Checkout the full documentation here.

React Native Skia brings the Skia Graphics Library to React Native. Skia serves as the graphics engine for Google Chrome and Chrome OS, Android, Flutter, Mozilla Firefox and Firefox OS, and many other products.

Getting Started

Installation instructions

Library Development

To develop react-native-skia, you need to build the skia libraries on your computer.

Make sure to check out the sub modules:

git submodule update --init --recursive

You also need to install some tools for the build scripts to work. Run yarn in the root of the project to install them.

Make sure you have all the tools required for building the skia libraries (XCode, Ninja, CMake, Android NDK / build tools).

On MacOS you can install Ninja via homebrew:

brew install ninja

If you have Android Studio installed, make sure $ANDROID_NDK is available. ANDROID_NDK=/Users/username/Library/Android/sdk/ndk/<version> for instance.

If the NDK is not installed, you can install it via Android Studio by going to the menu File > Project Structure.

And then the SDK Location section. It will show you the NDK path, or the option to Download it if you don't have it installed.

Building

  • Install dependencies yarn bootstrap
  • Build the Skia libraries with yarn build-skia (this can take a while)
  • Copy Skia headers yarn copy-skia-headers
  • run pod install in example/ios and fabricexample/ios

Upgrading

If a new version of Skia is included in an upgrade of this library, you need to perform a few extra steps before continuing:

  1. Update submodules: git submodule update --recursive --remote
  2. Clean Skia: yarn clean-skia
  3. Build Skia: yarn build-skia
  4. Copy Skia Headers: yarn copy-skia-headers
  5. Run pod install in the example project

Publishing

  • Run the commands in the Building section
  • Build the Android binaries with yarn build-skia-android
  • Build the NPM package with yarn build-npm

Publish the NPM package manually. The output is found in the dist folder.

  • Install Cocoapods in the example/ios folder cd example/ios && pod install && cd ..

Contributing

When making contributions to the project, an important part is testing. In the package folder, we have several scripts set up to help you maintain the quality of the codebase and test your changes:

  • yarn lint — Lints the code for potential errors and to ensure consistency with our coding standards.
  • yarn tsc — Runs the TypeScript compiler to check for typing issues.
  • yarn test — Executes the unit tests to ensure existing features work as expected after changes.
  • yarn e2e — Runs end-to-end tests. For these tests to run properly, you need to have the example app running. Use yarn ios or yarn android in the example folder and navigate to the Tests screen within the app.

Running End-to-End Tests

To ensure the best reliability, we encourage running end-to-end tests before submitting your changes:

  1. Start the example app:
cd example
yarn ios # or yarn android for Android testing

Once the app is open in your simulator or device, press the "Tests" item at the bottom of the list.

  1. With the example app running and the Tests screen open, run the following command in the package folder:
yarn e2e

This will run through the automated tests and verify that your changes have not introduced any regressions. You can also run a particular using the following command:

E2E=true yarn test -i e2e/Colors

Writing End-to-End Tests

Contributing end-to-end tests to React Native Skia is extremely useful. Below you'll find guidelines for writing tests using the eval, draw, and drawOffscreen commands.

e2e tests are located in the package/__tests__/e2e/ directory. You can create a file there or add a new test to an existing file depending on what is most sensible. When looking to contribute a new test, you can refer to existing tests to see how these can be built. The eval command is used to test Skia's imperative API. It requires a pure function that invokes Skia operations and returns a serialized result.

it("should generate commands properly", async () => {
  const result = await surface.eval((Skia) => {
    const path = Skia.Path.Make();
    path.lineTo(30, 30);
    return path.toCmds();
  });
  expect(result).toEqual([[0, 0, 0], [1, 30, 30]]);
});

Both the eval and draw commands require a function that will be executed in an isolated context, so the functions must be pure (without external dependencies) and serializable. You can use the second parameter to provide extra data to that function.

it("should generate commands properly", async () => {
  // Referencing the SVG variable directly in the tests would fail
  // as the function wouldn't be able to run in an isolated context
  const svg = "M 0 0, L 30 30";
  const result = await surface.eval((Skia, ctx) => {
    const path = Skia.Path.MakeFromSVGString(ctx.svg);
    return path.toCmds();
  }, { svg });
  expect(result).toEqual([[0, 0, 0], [1, 30, 30]]);
});

A second option is to use the draw command where you can test the Skia components and get the resulting image:

it("Path with default fillType", async () => {
  const { Skia } = importSkia();
  const path = star(Skia);
  const img = await surface.draw(
    <>
      <Fill color="white" />
      <Path path={path} style="stroke" strokeWidth={4} color="#3EB489" />
      <Path path={path} color="lightblue" />
    </>
  );
  checkImage(image, "snapshots/drawings/path.png");
});

Finally, you can use drawOffscreen to receive a canvas object as parameter. You will also get the resulting image:

  it("Should draw cyan", async () => {
    const image = await surface.drawOffscreen(
      (Skia, canvas, { size }) => {
        canvas.drawColor(Skia.Color("cyan"));
      }
    );
    checkImage(image, "snapshots/cyan.png");
  });

Again, since eval, draw, and drawOffscreen serialize the function's content, avoid any external dependencies that can't be serialized.