A file system parser and validator with full type inference. Inspired by Zod. Currently under development.
This package is mainly being used as a library for file-based content management systems on my other projects.
//Matches either an image inside the "public/" folder or a .url file
const videoOrImageSchema = typefs.union(typefs.image("public/"), typefs.url());
async function example() {
const parseResult = await videoOrImageSchema.parse(safePath("video.url"));
if (parseResult.wasResultSuccessful) {
//Inferred as a discriminated union of Image and Url
const videoOrImage = parseResult.okValue;
if (videoOrImage.option === 0) {
/* Both are inferred as numbers because
videoOrImage.value is inferred as an Image */
const [width, height] = [
videoOrImage.value.width,
videoOrImage.value.height,
];
console.log(width, height);
} else {
//videoOrImage.value is inferred as a Url
const videoUrl = videoOrImage.value.url;
console.log(videoUrl);
}
}
}//matches a folder named "videos" that contains .url files
const videoUrls = typefs.array(typefs.url()).withName("videos");
async function example() {
const video = await videoUrls.parse(safePath("videos"));
if (video.wasResultSuccessful) {
//Inferred as Url[]
const videoUrls = video.okValue.parsed;
videoUrls.map((url) => console.log(url.url));
}
}I wanted to have a file system parser that I could easily use to define a single source of truth for my file system entities. I wanted schemas and validators to be inferred from this source of truth. That way, I could quickly respond to requirements changes by changing the file system schema and the validators would automatically update. This is similar to how Zod works. I could've used something like io-ts, but I wanted full control over the schema primitives and their parsing logic.
This package uses TypeScript's type inference to infer a schema's output type from the schema itself. This is done by recursively traversing the schema and inferring its properties' types.
Probably not. This package is still in development and is not ready for production use. I'm primarily using it for my own projects, so expect breaking changes, bugs, as well as missing features. Also, most new features will most likely only be added via dogfooding.