b-fuze / deno-dom

Browser DOM & HTML parser in Deno
https://jsr.io/@b-fuze/deno-dom
MIT License
423 stars 47 forks source link
browser-dom deno dom html-parser rust typescript wasm

Deno DOM

An implementation of the browser DOM—primarily for SSR—in Deno. Implemented with Rust, WASM, and obviously, Deno/TypeScript.

Example

import { DOMParser, Element } from "jsr:@b-fuze/deno-dom";

//   non-JSR wasm url import: https://deno.land/x/deno_dom/deno-dom-wasm.ts
// non-JSR native url import: https://deno.land/x/deno_dom/deno-dom-native.ts

const doc = new DOMParser().parseFromString(
  `
    <h1>Hello World!</h1>
    <p>Hello from <a href="https://deno.land/">Deno!</a></p>
  `,
  "text/html",
);

const p = doc.querySelector("p")!;

console.log(p.textContent); // "Hello from Deno!"
console.log(p.childNodes[1].textContent); // "Deno!"

p.innerHTML = "DOM in <b>Deno</b> is pretty cool";
console.log(p.children[0].outerHTML); // "<b>Deno</b>"

Deno DOM has two backends, WASM and native using Deno native plugins. Both APIs are identical, the difference being only in performance. The WASM backend works with all Deno restrictions, but the native backend requires the --unstable-ffi --allow-ffi --allow-env --allow-read --allow-net=deno.land flags. A shorter version could be --unstable-ffi -A, but that allows all permissions so you'd have to assess your risk and requirements. You can switch between them by importing either jsr:@b-fuze/deno-dom for WASM or jsr:@b-fuze/deno-dom/native for the native binary.

Deno DOM is still under development, but is fairly usable for basic HTML manipulation needs.

WebAssembly Startup Penalty

Deno suffers an initial startup penalty in Deno DOM WASM due to Top Level Await (TLA) preparing the WASM parser. As an alternative to running the initiation on startup, you can initialize Deno DOM's parser on-demand yourself when you need it by importing from jsr:@b-fuze/deno-dom/wasm-noinit. Example:

import { DOMParser, initParser } from "jsr:@b-fuze/deno-dom/wasm-noinit";

// ...and when you need Deno DOM make sure you initialize the parser...
await initParser();

// Then you can use Deno DOM as you would normally
const doc = new DOMParser().parseFromString(
  `
    <h1>Lorem ipsum dolor...</h1>
  `,
  "text/html",
);

Documentation

Refer to MDN (Mozilla Developer Network) for documentation. If there are inconsistencies (that aren't a result of legacy APIs) file an issue.

Goals

Non-Goals

Running tests

To run tests (excluding WPT tests) use the following for WASM

deno test --allow-read --allow-net wasm.test.ts

Or the following for native (native requires more permissions)

deno test --unstable -A native.test.ts

To run WPT tests update the WPT submodule

git submodule update --progress --depth 1

Then append -- --wpt to the test command before running it, e.g. for WASM

deno test --allow-read --allow-net wasm.test.ts -- --wpt

WPT tests are still a WIP, passed tests likely haven't actually passed.

Building Deno DOM Native

Deno DOM native is a faster backend for Deno DOM (check benchmarks), however, the WASM backend is sufficient for almost all use-cases.

Note: If you're running an x86_64 system with either Windows, Linux, or macOS, then you probably don't need to build the plugin. Deno DOM native downloads a prebuilt binary in those cases.

To build Deno DOM's native backend, install Rust if you haven't already, then run

cargo build --release

which produces a binary located at target/release/libplugin.{so,dll,dylib} (extension depends on your system).

To use the new binary you need to set the DENO_DOM_PLUGIN env var to the path of the binary produced in the previous step. Don't forget to run Deno with --allow-env.

Credits