Applied a workaround to Firefox's bug which allowed sending search queries with Cyrillic characters encoded in cp1251. Since multitran.com understands only utf8, that made such queries return useless responses complaining that the search terms have been not recognized.