I would like to create an index with Lucene to a document collections of text files.
The index should be created in such a way, that for the search I can enforce that query term A and query term B are contained within the same sentence.

How should implement the index? Should I have for every sentence a different field (but make sure that it is not a multi-valued field because they would get merged which is exactly what I do not want)?
Would it be problematic that different documents would then end up having different numbes of fields?

Thank you in advance!