!7

cprint SnippetsDoubleWordIndex {
  transient DoubleWordIndex<WithIntArray<S>> wordIndex; // inner string = snippet ID, array = word positions in snippet

  start-thread {
    time "Make double word index" {
      print("Making index");
      new DoubleWordIndex<WithIntArray<S>> wordIndex;
      for (virtual CSnippet sn : dm_allSnippets()) {
        S snippetID = (S) rcall snippetID(sn);
        S text = cast rcall text(sn);
        MultiMap<S, Int> positions = ciMultiMap();
        for (IntRange r : wordIndex.wordRanges(text))
          positions.put(substring(text, r), r.start);
        for (S word, L<Int> indices : positions.asMap())
          wordIndex.addWord(new WithIntArray<S>(toIntArray(indices), snippetID), word);
      }
      setField(+wordIndex);
    }
    print("Indexed " + nWords(wordIndex.numWords()));
  }
  
  // API
  
  // may return null (no pre-search result)
  Cl<S> snippetsContainingTextIC_preSearch(S query, O... _) {
    ret deepDoubleWordIndex_lookupString(wordIndex, query, _);
  }
}