Flatten HTML

flatten_html.clj

;;include [clj-tagsoup "0.3.0"]

(require '[pl.danieljanus.tagsoup :as tags])

;; each node's structure is as such
;; [:tag attr-map node node ....]
(defn flatten [x]
 (tree-seq (comp keyword? first) (partial drop 2) x))

(defn get-url
  [url]
  (let [x (tags/parse url)]
    (-> x
        xml-seq
        first))) 


#find all anchor tags
(filter (comp (partial = :a) first)
        (flatten (get-url "http://www.google.com")))