Can I pass a custom definition context to the parser?

[kkshinkai]

In Markdown, inline [label] elements are resolved based on their corresponding definitions. Micromark retains these links when it finds a matching [label]: ... definition.

However, when building an editor using Micromark, I only want to re-parse a single block at a time. I can't find a way to recognize [label] correctly.

This is a [label].
^^^^^^^^^^^^^^^^^^only re-parse this while typing

[label]: www.example.com

What I'm looking for is something like this, is there any workaround?

const chunks = micromark.preprocess()("This is a [label]", undefined, true)

const parseContext = micromark.parse({
  definitions:[
    "label", // To parse `[label]` correctly.
    ...
  ],
});

const events = micromark.postprocess(
  parseContext
    .document()
    .write(chunks))
// → content "This is a [label]" (0..17)
// → paragraph "This is a [label]" (0..17)
// → data "This is a " (0..10)
// ← data "This is a " (0..10)
// → link "[label]" (10..17)  <-----. I want to get these events
// → label "[label]" (10..17)       |
// → labelMarker "[" (10..11)       |
// ← labelMarker "[" (10..11)       |
// → labelText "label" (11..16)     |
// → data "label" (11..16)          |
// ← data "label" (11..16)          |
// ← labelText "label" (11..16)     |
// → labelMarker "]" (16..17)       |
// ← labelMarker "]" (16..17)       |
// ← label "[label]" (10..17)       |
// ← link "[label]" (10..17)  <-----'
// ← paragraph "This is a [label]" (0..17)
// ← content "This is a [label]" (0..17)

Thanks for the great works!

[kkshinkai]

I figured it out myself ... I misunderstood normalizeIdentifier, it works now:

const parseContext = micromark.parse();
parseContext.defined = ["LABEL"]
//                      ^^^^^^^ use normalized string "LABEL" instead of "label"
const tokenizeConetext = parseContext.document();
const events = micromark.postprocess(tokenizeConetext.write(chunks))

[wooorm]

Perhaps you can access it with an extension too. And, perhaps use normalizeIdentifier to map your input IDs to the expected canonical format?