We faced this problem a long time ago, and ended up just extracting all the matching documents, re-analyzing and counting the terms using a MultiSet<http://guava-libraries.googlecode.com/svn/tags/release02/javadoc/com/google/common/collect/Multiset.html>. It was very slow, but it worked. You might get better perfomance if you index with term vectors and use IndexReader#getTermFreqVector()<http://lucene.apache.org/java/3_0_3/api/core/org/apache/lucene/index/IndexReader.html#getTermFreqVector(int, java.lang.String)> to count the number of appearances of each term in a document.
Either way, I don't know of a fast way to do this.
