oxc/wasm/parser/README.md

64 lines
1.8 KiB
Markdown

## About
Experimental wasm package for the oxc parser, with full TypeScript typings support.
This package is built with different [wasm-pack's target](https://rustwasm.github.io/docs/wasm-bindgen/reference/deployment.html) builds:
- `wasm-pack build --target web` for bundler (webpack / vite) consumption.
- `wasm-pack build --target nodejs` for node.js
And exports the files as
```json
"main": "./node/oxc_parser_wasm.js",
"browser": "./web/oxc_parser_wasm.js",
"types": "./node/oxc_parser_wasm.d.ts",
```
Checkout [oxc-parser](https://www.npmjs.com/package/oxc-parser) for usage in node.js via napi bindings.
Source code: https://github.com/oxc-project/oxc/tree/main/wasm/parser
## Usage
```js
import initWasm, { parseSync } from '@oxc-parser/wasm';
async function main() {
await initWasm();
const code = 'let foo';
const result = parseSync(code, { sourceFilename: 'test.ts' });
console.log(result);
}
main();
```
## Notes
### UTF8 vs UTF16 byte offsets
The `span` value returned from the ASTs and diagnostics is in UTF8 byte offsets. Converting to UTF16 byte offsets:
```js
let sourceTextUtf8 = new TextEncoder().encode(sourceText);
const convertToUtf8 = (sourceTextUtf8, d) => {
return new TextDecoder().decode(sourceTextUtf8.slice(0, d)).length;
};
const diagnostics = result.errors.map((d) => ({
from: convertToUtf8(sourceTextUtf8, d.start),
to: convertToUtf8(sourceTextUtf8, d.end),
severity: d.severity.toLowerCase(),
message: d.message,
}));
```
### Vite
`wasm-pack build --target web` is used for the wasm build.
You may need something like https://github.com/nshen/vite-plugin-wasm-pack to get it working with vite,
otherwise vite will load the wasm file as a HTML file causing a `CompileError: WebAssembly.instantiate(): expected magic word` error.