I need to write a custom LemmaTokenFilter, which replaces and indexes the words with their lemmatized(base) form. The problem is, that I get the base forms from an external