How do you best map (uniform) RDF to CSV, JSON, INI etc.

RDF has advantages for integration of data, but in the end it is much more convenient to work with simple record structures and many developers prefer simple serialization formats such as CSV, JSON, and INI-Files. Here I don't want to discuss about the reason, but about how do best map RDF to this formats. I don't refer to serialization forms like RDF/JSON but to simple, domain specific, ad-hoc data structures. For instance I'd want a JSON structure like the following (please don't argue whether this structure is good or bad, it is just given as requirement):

{
  "orgs": [
    { "uri": "http://example.org/", "name": "Example Organisation" },
    { "uri": "http://example.com/", "name": "Another Organisation" },
  ],
  "people": [
    { "name": "Alice", "age": 32 }, { "name": "Bob", "age": 24 }
  ]
}

Or I just want a CSV list of people with name, homepage, and age, etc. The general answer would be "SPARQL with some additional post-processing", but I want to manage mappings in a simple, readable form. The best I could find was using a scripting language and a simple API to access RDF like this pseudo-Perl script:

my %result = (
  orgs   => [ map { return { uri => $_->uri, name => $_->getprop('foaf:name') } }    
                  $rdf->get( undef, 'a', 'faof:Organization' ) ],
  people => [ map { return { name => $_->getprop('foaf:name'), 
                             age => $_->getprop('foaf:age') } }
                  $rdf->get( undef, 'a', 'faof:Person' ) ]
);
return toJSON(\%result);

However I do not want to stick to a specific Turing-complete scripting language. Before inventing my own RDF-Mapping language I'd like to know whether there already are good solutions. Maybe the mapping could even be stored in form of RDF, so I could write it down in Turtle?