Spec Driven Transformations

NOTE: this document is partially rewritten as spec coercion.

Like Plumatic Schema, Spec-tools differentiates specs (what) and transformers (how). This enables spec values to be transformed between different formats like JSON and EDN. Core concept is the Transformer protocol:

(defprotocol Transformer
  (-name [this])
  (-options [this])
  (-encoder [this spec value])
  (-decoder [this spec value]))

Spec-tools ships with following transformer implementations:

Name	Description
string-transformer	String-formats like properties files, query- & path-parameters.
json-transformer	JSON format, like string, but numbers and booleans are supported
strip-extra-keys-transformer	Decoding strips out extra keys of s/keys specs.
strip-extra-values-transformer	Decoding strips out extra values of s/tuple specs.
fail-on-extra-keys-transformer	Decoding fails if s/keys specs have extra keys.
nil	No transformations, e.g. EDN & Transit.

Conforming

Functions explain, explain-data, conform and conform! take the transformer an optional third argument and pass it into Specs via dynamic binding. Before CLJ-2116 or CLJ-2251 are fixed, specs need to be wrapped into Spec Records to make this work.

Encoding and Decoding

There are also encode & decode functions that combine the two approaches and considered the best way to transform the values. decode first tries to use coerce and if that doesn't make the value valid against the given spec, falls back to conform & unform which can be used for all specs.

Spec-driven transformations

• :encode/* and :decode/* keys from Specs to declare how the values should be transformed in & out from different formats
• both take a 2-arity function of spec value => value to do the actual transformation

(require '[clojure.string :as str])

(s/def ::spec
  (st/spec
    {:spec #(and (simple-keyword? %) (-> % name str/lower-case keyword (= %)))
     :description "a lowercase keyword, encoded in uppercase in string-mode"
     :decode/string #(-> %2 name str/lower-case keyword)
     :encode/string #(-> %2 name str/upper-case)}))

(st/decode ::spec :kikka)
; :kikka

(as-> "KiKka" $
      (st/decode ::spec $))
; :clojure.spec.alpha/invalid

(as-> "KiKka" $
      (st/decode ::spec $ st/string-transformer))
; :kikka

(as-> "KiKka" $
      (st/decode ::spec $ st/string-transformer)
      (st/encode ::spec $ st/string-transformer))
; "KIKKA"

Spec Bijections?

no, as there can be multiple valid representations for a encoded value. But it can be guaranteed that a decoded values X is always encoded into Y, which can be decoded back into X: y -> X -> Y -> X

(as-> "KikKa" $
      (doto $ prn)
      (st/encode ::spec $ st/string-transformer)
      (doto $ prn)
      (st/decode ::spec $ st/string-transformer)
      (doto $ prn)
      (st/encode ::spec $ st/string-transformer)
      (prn $))
; "KikKa"
; "KIKKA"
; :kikka
; "KIKKA"

Type-driven transformations

• Uses :type information from Specs
  • resolved automatically for most core predicates.
  • top-level spec arguments in encode & decode etc are transformed into Spec Records automatically using IntoSpec protocol.
  • standard types are: :long, :double, :boolean, :string, :keyword, :symbol, :uuid, :uri, :bigdec, :date, :ratio, :map, :set and :vector.

(as-> "2014-02-18T18:25:37Z" $
      (st/decode inst? $))
; :clojure.spec.alpha/invalid

;; decode using string-transformer
(as-> "2014-02-18T18:25:37Z" $
      (st/decode inst? $ st/string-transformer))
; #inst"2014-02-18T18:25:37.000-00:00"

;; encode using string-transformer
(as-> "2014-02-18T18:25:37Z" $
      (st/decode inst? $ st/string-transformer)
      (st/encode inst? $ st/string-transformer))
; "2014-02-18T18:25:37.000+0000"

When creating custom specs, :type gives you encoders & decoders (and docs!) for free, like with Data.Unjson.

(s/def ::kw
  (st/spec
    {:spec #(keyword %) ;; anonymous function
     :type :keyword}))  ;; encode & decode like a keyword

(st/decode ::kw "kikka" st/string-transformer)
;; :kikka

(st/decode ::kw "kikka" st/json-transformer)
;; :kikka

Transforming nested specs

Because of current design of clojure.spec, we need to wrap all non top-level specs into Spec Records manually to enable transformations.

(s/def ::name string?)
(s/def ::birthdate spec/inst?)

(s/def ::languages
  (s/coll-of
    (s/and spec/keyword? #{:clj :cljs})
    :into #{}))

(s/def ::user
  (s/keys
    :req-un [::name ::languages ::age]
    :opt-un [::birthdate]))

(def data
  {:name "Ilona"
   :age "48"
   :languages ["clj" "cljs"]
   :birthdate "1968-01-02T15:04:05Z"})

;; no transformer
(st/decode ::user data)
; ::s/invalid

;; json-transformer doesn't transform numbers
(st/decode ::user data st/json-transformer)
; ::s/invalid

;; string-transformer for the rescue
(st/decode ::user data st/string-transformer)
; {:name "Ilona"
;  :age 48
;  :languages #{:clj :cljs}
;  :birthdate #inst"1968-01-02T15:04:05.000-00:00"}

Transforming Map Specs

To strip out extra keys from a keyset:

(s/def ::name string?)
(s/def ::street string?)
(s/def ::address (st/spec (s/keys :req-un [::street])))
(s/def ::user (st/spec (s/keys :req-un [::name ::address])))

(def inkeri
  {:name "Inkeri"
   :age 102
   :address {:street "Satamakatu"
             :city "Tampere"}})

(st/decode ::user inkeri st/strip-extra-keys-transformer)
; {:name "Inkeri"
;  :address {:street "Satamakatu"}}

There are also a shortcut for this, select-spec:

(st/select-spec ::user inkeri)
; {:name "Inkeri"
;  :address {:street "Satamakatu"}}

Custom Transformers

Transformers should have a simple keyword name and optionally type-based decoders, encoders, default decoder and -encoder set. Currently there is no utility to verify that y -> X -> Y -> X holds for custom transformers.

(require '[clojure.string :as str])
(require '[spec-tools.transform :as stt])

(defn transform [_ value]
  (-> value
      str/upper-case
      str/reverse
      keyword))

;; string-decoding + special keywords
;; encoding writes strings by default
(def my-string-transformer
  (type-transformer
    {:name :custom
     :decoders (merge
                 stt/string-type-decoders
                 {:keyword transform})
     :default-encoder stt/any->string}))

(decode keyword? "kikka")
; :clojure.spec.alpha/invalid

(decode keyword? "kikka" my-string-transformer)
; :AKKIK

; spec-driven transforming
(decode
  (spec
    {:spec #(keyword? %)
     :decode/custom transform})
  "kikka"
  my-string-transformer)
; :AKKIK

;; defaut encoding to strings
(encode int? 1 my-string-transformer)
; "1"

Type-based transformer encoding & decoding mappings are defined as data, so they are easy to compose:

(def strict-json-transformer
  (st/type-transformer
    {:name :custom
     :decoders (merge
                 stt/json-type-decoders
                 stt/strip-extra-keys-type-decoders)
     :encoders stt/json-type-encoders}))

Or using type-transformer directly:

(def strict-json-transformer
  (st/type-transformer
    st/json-transformer
    st/strip-extra-keys-transformer
    st/strip-extra-values-transformer))