This repository holds the Gosub browser engine. It will become a standalone library that can be used by other projects but will ultimately be used by the Gosub browser user-agent. See the About section for more information.
Join us at our development Zulip chat!
For more general information you can also join our Discord server.
If you are interested in contributing to Gosub, please check out the contribution guide!
This repository is part of the Gosub browser engine project. This is the main engine that holds the following components:
- HTML5 tokenizer / parser
- CSS3 tokenizer / parser
- Document tree
- Several APIs for connecting to javascript
- Configuration store
- Networking stack
- Rendering engine
- JS bridge
More will follow as the engine grows. The idea is that this engine will receive some kind of stream of bytes (most likely from a socket or file) and parse this into a valid HTML5 document tree and CSS stylesheets. From that point, it can be fed to a renderer engine that will render the document tree into a window, or it can be fed to a more simplistic engine that will render it in a terminal. JS can be executed on the document tree and the document tree can be modified by JS.
This project is in its infancy. There is no usable browser yet. However, you can look at simple html pages and parse them into a document tree and do some initial rendering.
We can parse HTML5 and CSS3 files into a document tree or the respective css tree. This tree can be shown in the terminal or be rendered in a very unfinished renderer. Our renderer cannot render everything yet, but it can render simple html pages, sort of.
We already implemented other parts of the engine, for a JS engine, networking stack, a configuration store and other things however these aren't integrated yet. You can try these out by running the respective binary.
We can render a part for our own site:
Installing dependencies
This project uses cargo and rustup. First
you must install rustup at the link provided. After installing rustup, run:
$ rustup toolchain install 1.73
$ rustc --version
rustc 1.73.0 (cc66ad468 2023-10-03)Once Rust is installed, run this command to pre-build the dependencies:
$ cargo build --releaseYou can run the following binaries:
| Command | Type | Description |
|---|---|---|
cargo run -r --bin config-store |
bin | A simple test application of the config store for testing purposes |
cargo run -r --bin css3-parser |
bin | Show the parsed css tree |
cargo run -r --bin display-text-tree |
bin | A simple parser that will try and return a textual presentation of the website |
cargo run -r --bin gosub-parser |
bin | The actual html5 parser/tokenizer that allows you to convert html5 into a document tree. |
cargo run -r --bin html5-parser-test |
test | A test suite that tests all html5lib tests for the treebuilding |
cargo run -r --bin parser-test |
test | A test suite for the parser that tests specific tests. This will be removed as soon as the parser is completely finished as this tool is for developement only. |
cargo run -r --bin renderer |
bin | Render a html page (WIP) |
cargo run -r --bin run-js |
bin | Run a JS file (Note: console and event loop are not yet implemented) |
For running the binaries, take a look at a quick introduction at /docs/binaries.md
To run the tests and benchmark suite, do:
make test
cargo bench
ls target/criterion/report
index.htmlOur engine can also be compiled to WebAssembly. You need to use WasmPack for this. To build the Wasm version, run:
wasm-pack build --target webAfterwards you need to serve the small useragent around the wasm version in the wasm/ directory. You can do this by
cd wasm
npm run dev # you can also use `bun run dev`To use this demo, you need to enable webgpu in chromium and disable the same origin policy.
chromium --disable-web-security --enable-features=Vulkan --enable-unsafe-webgpu --user-data-dir=/tmp/chromium-temp-profile This command works on Linux only, if someone uses Windows or macOS, please open an PR!
And then you have it! A browser in a browser:
We welcome contributions to this project but the current status makes that we are spending a lot of time researching, building small proof-of-concepts and figuring out what needs to be done next. Much time of a contributor at this stage of the project will be non-coding.
We do like to hear from you if you are interested in contributing to the project and you can join us currently at our Zulip chat!