transient sclass DeepWordIndex<A> {
  S regexp = "\\w+";
  
  new Map<A, Entry<A>> entries;
  MultiSetMap<S, Entry<A>> entriesByWord = ciMultiSetMap();
  
  sclass Entry<A> {
    A id;
    Map<S, int[]> wordPositions = ciMap();
    
    *(A *id) {}
  }
  
  L<IntRange> wordRanges(S text) {
    ret regexpFindRanges(regexp, text);
  }
   
  void add(A a, S text) {
    Entry<A> e = new Entry<A>(a);
    if (entries.put(a, e) != null) fail("Double insertion");
    MultiMap<S, Int> wordPositions = ciMultiMap();
    for (IntRange r : wordRanges(text)) {
      S word = substring(text, r);
      wordPositions.put(word, r.start);
      entriesByWord.put(word, e);
    }
    for (S word : keys(wordPositions))
      e.wordPositions.put(word, toIntArray(wordPositions.get(word)));
  }
  
  Set<Entry<A>> get(S word) { ret entriesByWord.get(word); }
  
  int numWords() { ret entriesByWord.keysSize(); }
}