ignore content-type encoding when choosing a parser for topic data