Extractors should run on source files only.
They should not depend on an acitive web server or database connection.