This Typst package provides automatic Cantonese segmentation and romanization
(Jyutping (粵拼) and Yale (耶魯)) by wrapping the
rust-canto Rust crate as a WebAssembly
plugin. It integrates seamlessly with the
pycantonese-parser
package to render beautiful Cantonese text with ruby characters.
- Automatic Segmentation: Breaks Cantonese sentences into meaningful words using a dictionary-based trie.
- Multiple Romanizations: Supports both Jyutping and Yale (numeric or diacritics).
- High Performance: Powered by a Rust-compiled WASM plugin for fast processing.
- Typst Integration: Provides a
quick-renderfunction that handles both segmentation and styling in one go.
To use this package, ensure the rust_canto.wasm file is in your project directory.
#import "@preview/auto-canto:0.1.0": quick-render
// 36pt font
// use Libertinus Serif first (for ruby text)
// before falling back to Noto Serif CJK HK (for Chinese characters)
#set text(36pt, font: ("Libertinus Serif", "Noto Serif CJK HK"))
// 1. Basic rendering (defaults to Jyutping)
#quick-render("佢係好學生")
// 2. Rendering with Yale romanization
#quick-render("平時會成日睇書", romanization: "yale")
// 3. Customizing the underlying parser's style
#let my-style = (rb-size: 0.7em, rb-color: blue)
#quick-render("廣東話好難學", style: my-style)
Live demo on YouTube: https://youtu.be/ivUu91eDfvY
The primary high-level function. It fetches data from the WASM plugin and forwards it to the parser.
txt: The Cantonese string to process...args: Named arguments forwarded torender-word-groups(e.g.romanization,style).
Returns the raw segmented data as an array of dictionaries.
- Return format:
arrayof{word: str, jyutping: str, yale: array}.
Utility functions to convert space-delimited Jyutping strings into Yale format.
numeric: "gwong2 dung1 waa2" → "gwong2 dung1 wa2".diacritics: "gwong2 dung1 waa2" → "gwóngdūngwá".
lib.typ: The main entry point containing the Typst wrappers.rust_canto.wasm: The WebAssembly binary compiled from therust-cantocrate.typst.toml: Package metadata and dependencies.
MIT
Contributions are welcome! Please open an issue or submit a pull request.