The HttpRecognizer receives streamed HTTP, and when it accumulates enough to identify that it's seen an HTTP header, it attempts to extract some basic information from it. This is not a parser, and it is woefully incomplete, but for many tasks it understands just enough HTTP to be useful.