!7

srecord noeq Entry(
  S snippetID,
  int[] indices // where in the text does the word occur
) {}}

cprint SnippetsDoubleWordIndex {
  transient DoubleWordIndex<Entry> wordIndex;

  start-thread {
    time "Make double word index" {
      new DoubleWordIndex<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.wordRange(text))
          positions.put(substring(text, r), r.start);
        for (S word, L<Int> indices : asMap())
          wordIndex.addWord(word, new Entry(text, toIntArray(indices)));
      }
      setField(+wordIndex);
    }
    print("Indexed " + nWords(wordIndex.numWords()));
  }
  
  // API
  
  // may return null (no pre-search result)
  Cl<S> snippetsContainingTextIC_preSearch(S query) {
    ret doubleWordIndex_lookupString(wordIndex, query);
  }
}