Automagic extraction of Resource data breaks when unexpected data format is received

[peta]

Vocabulary/ontology documents must be in one of the formats application/rdf+xml or application/owl+xml. When the server ignores the Accept header and instead quietly serves a document in another format (like XHTML with embedded RDF data) the current extraction method chokes and can't recover gracefully.