transient sclass DeepBitSetWordIndex<A> {
  S regexp = "\\w+";
  new Map<A, SingleTextWordIndex> singleTextIndices;
  new ElementInstanceMatrix<A, S> mainIndex;

  void add(A a, S text) {
    singleTextIndices.put(a, new SingleTextWordIndex(regexp, text));
    mainIndex.add(a, mapToSet upper(regexpExtractAll(regexp, text));
  }
  
  void doneAdding {
    mainIndex.doneAdding();
  }
  
  LPair<S, Int> wordsAndOffsets(S text) {
    ret map(regexpFindRanges(regexp, text),
      r -> pair(upper(substring(text, r)), r.start));
  }

  // assumes word boundaries left and right of query
  Cl<A> preSearch(S query, O... _) {
    optPar bool debug;
    LPair<S, Int> l = wordsAndOffsets(query);
    Cl<A> candidates = mainIndex.instancesContainingAllElements(pairsA(l));
    if (debug) {
      L<Int> lengths = map(candidates, a -> singleTextIndices.get(a).length);
      print(nCandidates(candidates) + ", total length: " + n2(intSum(lengths)) + ", lengths: " + lengths);
    }
    ret filter(candidates, a -> nempty(singleTextIndices.get(a).indicesOfWordCombination(l)));
  }
  
  int numWords() { ret mainIndex.numElements(); }
}

Began life as a copy of #1029078