[Feature Request] Analyzers on non-text sub fields

[lasiltan]

Is your feature request related to a problem? Please describe

In my domain, Every single type of document has an id that consists of two parts: id space and thei d itself, e.g. 31-111.
When trying to sort by id, I'd actually want to sort by the last part, numerically. Currently I cannot create an analyzer for a sub field (ie.g. d.number) that would index the id part as a long because analyzers work on strings.

Describe the solution you'd like

When the main field is a string (text, keyword) and the sub field is another type, you should be able to set an analyzer to the sub field. Opensearch should try to cast the result to the type of the sub field.

Related component

Indexing

Describe alternatives you've considered

Store the number part in a separate field.

• It's not part of the data so the result is awkward.
• The purpose of sub fields is to avoid this exact thing.

Script sorting

• Performance becomes an issue.

Additional context

No response

[peternied]

[Triage - attendees 1 2 3 4]
Thanks for filing, please feel free to submit a pull request for one of these analyzers that supports your scenario.