"Unrecognized lexicon type" for every record in my BlueSky download

[bobg]

Hello! I am trying to parse the contents of the repo.car file that I downloaded from BlueSky just now, using the sample code at https://docs.bsky.app/blog/repo-export#parse-records-from-car-file-as-json. I'm getting a util.ErrUnrecognizedType from Repo.GetRecord for every record in the archive.

Here's my code. Am I missing a step? Thanks in advance.

	...
	r, err := repo.ReadRepoFromCar(ctx, os.Stdin)
	if err != nil {
		return errors.Wrap(err, "reading repo from stdin")
	}

	err = r.ForEach(ctx, "", func(k string, v cid.Cid) error {
		_, rec, err := r.GetRecord(ctx, k)
		if errors.Is(err, util.ErrUnrecognizedType) {
			log.Printf("Skipping record %s: %s", k, err)
			return nil
		}
		...
	}
	...

[bnewbold]

Yup! Many of the default methods in the indigo codebase assume that there are defined structs/types for the records being marshalled/unmarshalled. This "registration" happens when relevant packages (like the api/bsky package) are imported.

It is also possible to pull out data as bytes, and unmarshall to generic map[string]any objects, using the atproto/data package. Check out the goat code for examples, eg:
https://github.com/bluesky-social/indigo/blob/main/cmd/goat/repo.go#L239

definitely need docs around all this, and probably refactor things to default to "schema/type agnostic" with the registered types being more explicit?

[bobg]

Thank you!