Using Temporal.Instant for dates

[mgood]

This is mostly just informative rather than a feature request at the moment, could be interesting to consider how to offer representing dateTime with the proposed Temporal.Instant instead of Date.

I had started using a Temporal polyfill mainly for working with Durations, and found it useful to represent the dates as an Instant.

So, I tried this quick-and-dirty mapping to substitute the ldkit handling for xsd.dateTime:

import { registerDataHandler } from 'ldkit';
import { xsd } from 'ldkit/namespaces';
import { Temporal } from '@js-temporal/polyfill';

declare module 'ldkit' {
	interface CustomDataTypes {
		[xsd.dateTime]: Temporal.Instant;
	}
}

registerDataHandler(
	xsd.dateTime,
	(literalValue: string) => {
		const m = literalValue.match(/^"([^"]+)"\^\^/);
		if (m) return Temporal.Instant.from(m[1]);
		return Temporal.Instant.from(literalValue);
	},
	(value: Temporal.Instant) => {
		return `"${value.toString()}"^^${xsd.dateTime}`;
	}
);

I saw ldkit has some helpers internally for converting the value^^type format, but they didn't seem to be available through the exported APIs. Do you have a suggestion for a better way to handle the string encoding & decoding?

[karelklima]

Thank you for the suggestion and for the code snippet. Supporting the Temporal API is a good idea and it is definitely something that LDkit should offer out of the box in the near future.

Regarding your code snippet - I believe that the following should be enough (I haven't tested it though):

registerDataHandler(
	xsd.dateTime,
	(literalValue: string) => {
		return Temporal.Instant.from(literalValue);
	},
	(value: Temporal.Instant) => {
		return value.toString();
	}
);

The data handler basically just works with the stringified value of the literal. It then creates an actual RDF literal using a DataFactory and supplies the proper data type internally, so no need to attach the type there (unless there is a bug somewhere I am not aware of).

LDkit uses the rdf-literal NPM package to convert to and from RDF - so if you are interested at how to actually process raw data, I suggest to take a look at that library.

export { fromRdf, toRdf } from "npm:rdf-literal@^2";