Filter html

html-o-matic

(use 'pl.danieljanus.tagsoup)
(use 'hiccup.core)

(defn parser [input]
  (pl.danieljanus.tagsoup/parse
   (java.io.ByteArrayInputStream.
    (.getBytes
     (slurp input)))))

(defn dissoc-all [dmap keys]
  (apply dissoc dmap keys))

(defn soap [element keys]
  (cond
    (map? element) (dissoc-all element  keys)
    (= (type []) (type element)) (mapv identity (map #(soap % keys) element))
    :else element))

(defn sanitzer [parsed_input keys]
  (mapv identity (map #(soap % keys) parsed_input)))

(defn html-o-matic
  "Input: A file containing html data, output: another file containing filtered html data & optionally keys to filter"
  ([input output]
   (html-o-matic input output :class :style :id))
  ([input output & keys]
   (let [veckeys (mapv identity keys)]
     (with-open [w
                 (clojure.java.io/writer output)]
       (.write w
               (hiccup.core/html (sanitzer (parser input) veckeys)))))))