!7

/* Task: Search a short string in 45 MB of source codes

Results:
  simple contains() takes <10 ms when optimized
    (String.indexOf is a HotSpot-intrinsic)
  containsIC is really slow (700 ms)
  Boyer-Moore is 150 ms (case-insensitive) or 50 ms (case-sensitive)
*/

cprint {
  start-thread {
    SS map = dm_snippetIDToTextMap();
    time "Load texts" {
      map = mapValues upper(cloneAsHashMap(map));
    }
    BoyerMooreStringSearch_upper search = new("Boyer");
    print(nChars(totalStringLength(values(map))));
    LS result = null;
    repeat 10 {
      time "Search" {
        //result = keysWithValueThat(map, text -> containsIC(text, "Boyer"));
        result = keysWithValueThat(map, text -> contains(text, "BOYER"));
        //result = keysWithValueThat(map, text -> search.containedIn(text));
      }
    }
    print(result);
  }
}