As per this tweet: https://twitter.com/_tallison/status/1501584655597850632?s=21 an alternative integration pattern is to register this properly as a Detector, but return null and add the results to the Metadata object instead. This puts the results where you can get them, but leaves Tika in charge of the ID-then-parse flow.
This has been partially implemented, as PRONOM-related results have been added as Metadata, but it still returns the combined MIME type (as that's how I'm using it in webarchive-discovery for now).
As per this tweet: https://twitter.com/_tallison/status/1501584655597850632?s=21 an alternative integration pattern is to register this properly as a Detector, but return
null
and add the results to theMetadata
object instead. This puts the results where you can get them, but leaves Tika in charge of the ID-then-parse flow.