fields exposes the function select-keys-by-fields which offers the
same functionality as
select-keys but
for complex nested maps. Instead of specifying paths in a vector, we write a query
describing the structure of the desired output map.
fieldsis inspired by the same-named HTTP request parameter as
described by Google
here. In the
background it parses the query into a
zipper and traverses the zipper
filtering the map along the way.
(def data {:a 1
:b [{:aa 2 :bb 3} {:aa 4 :bb 5}]
:c {:cc 6 :dd {:fff 7 :ggg 8}}})
(select-keys-by-fields data "(a)")
=> {:a 1}
(select-keys-by-fields data "(a,b(aa))")
=> {:a 1, :b ({:aa 2} {:aa 4})}
(select-keys-by-fields data "(b,c(cc,dd(ggg)))")
=> {:b ({:aa 2, ::bb 3} {:aa 4, :bb 5}), :c {:cc 6, :dd {:ggg 8}}}The parser which parses a query string and outputs a zipper is a pure function that is worth caching whenever you have many differnt maps but only a small set of queries. For such use cases you can just memozie the parser and use select-keys-by-zipper instead of select-keys-by-fields.
(def parse-memo (memoize parse))
(select-keys-by-zipper data (parse-memo "(a,b(aa))"))
=> {:a 1, :b ({:aa 2} {:aa 4})}
MIT