url-match.clj

;;
;; Matcher should recognize and destruct URL by:
;; host: domain
;; path: parts, splitted with "/"
;; queryparam: name/value pairs of query
;; (see examples below)
;;
;; Each string that is started from "?" is a "bind"
;; (recognize matcher) should return nil or seq of binds
;;

(import java.net.URL)


(defn new-pattern [pattern]
  "parse input pattern into list of (command, param) tuples"
  (let [command-re #"(host|path|queryparam)\(([^)]*)\)"
        command-matches (->> (re-seq command-re pattern) (map rest))]
    command-matches))


(defn parse-binds [pattern url-part]
  "parse binds values from any part of a url"
  (let [bind-re #"\?([^?/=&]+)"
        parsed-pattern (clojure.string/replace pattern bind-re "([^?/=&]+)")
        parsed-re (re-pattern parsed-pattern)
        parsed-bind-names (->> (re-seq bind-re pattern) (map last))
        parsed-bind-values (->> (re-find parsed-re url-part) rest)
        binds-list (map array-map parsed-bind-names parsed-bind-values)
        binds-map (apply (partial merge-with into) binds-list)]
    (clojure.walk/keywordize-keys binds-map)))


(defn recognize-one [[command pattern] parsed-url]
  "process one input pattern only"
  (if-let [url-part (case command
                      "host" (. parsed-url getHost)
                      "path" (. parsed-url getPath)
                      "queryparam" (. parsed-url getQuery))]
    (if (clojure.string/includes? pattern "?")
      (parse-binds pattern url-part)
      (if (= pattern url-part) url-part))))


(defn merge-recognized [res match]
  "reduce function to generate final result"
  (if (or (nil? res) (nil? match)) nil
      (if (string? match) res
          (into res (seq match)))))


(defn recognize [pattern url]
  "recognize all patterns in a url"
  (let [url (URL. url)
        all-recognized (map #(recognize-one % url) pattern)]
    (reduce merge-recognized [] all-recognized)))


;; Tests
(let [youtube (new-pattern "host(?host); path(/watch/?type); queryparam(id=?id)")]
  (clojure.test/is
   (= (recognize youtube "http://youtube.com/watch/video/?id=1")
      [[:host "youtube.com"] [:type "video"] [:id "1"]])))

(let [twitter (new-pattern "host(twitter.com); path(?user/status/?id);")]
  (clojure.test/is
   (= (recognize twitter "http://twitter.com/bradfitz/status/562360748727611392")
      [[:user "bradfitz"] [:id "562360748727611392"]])))

(let [dribbble (new-pattern "host(dribbble.com); path(shots/?id); queryparam(offset=?offset);")]
  (clojure.test/is
   (= (recognize dribbble "https://dribbble.com/shots/1905065-Travel-Icons-pack?list=users&offset=1")
      [[:id "1905065-Travel-Icons-pack"] [:offset "1"]]))
  (clojure.test/is
   (nil? (recognize dribbble "https://twitter.com/shots/1905065-Travel-Icons-pack?list=users&offset=1")))
  (clojure.test/is
   (nil? (recognize dribbble "https://dribbble.com/shots/1905065-Travel-Icons-pack?list=users"))))