Extract "phrase token" like one word by setting "user dictionary"

[lamrongol]

For example, "good morning!" is tokenized to separating "good" and "morning" as following:

good	ADJ
morning	NOUN
!	!	PUNCT

But I want to get result like good morning INTJ.

I think this feature doesn't exist now, PhraseMatcher and Rule-based matching only extract specified phrases.

I want to set phrases by myself like following "dictionary file"

good morning	INTJ

This feature exists in Kuromoji(Japanese NLP library) as following Japanese page: https://qiita.com/masato_ka/items/2bae7f1e8ce245443947