Problem when using stemmer

Hi,

I used a stemmer tokenizer with light_french langage, and when I test to
search for example : "masag" → it works and I receive the doc that contain
the word 'massage'
But the problem is when I search with 'masage' → 0 hits in result

I used this QueryDSL : text, text_phrase_prefix, field
When I used Fuzzy Query, it solves the problem, however I think that it
hasn't sens to use Fuzzy Query with stemmers !

Is there any solution ?
Thanks