Is it possible to count exact characters entered by a user inside lexcial RTE, ignoring all the html tags and json meta data?

[arsalankhan19]

I am working on a project where we are extensively using lexical RTE, we support formatting text, code blocks, tables, links, images inside our RTE.
We don't want to overwhelm backend with a lot of data hence we are trying to put a check on character limit for RTE, but this doesn't seems to be straightforward.
Looking for an effective solution which can help us precisely count the actual characters within RTE whilst ignoring other html tags or json meta data.