gistfile1.txt

(use 'clojure.java.io)

(def string-url "http://java.sun.com/j2se/1.5.0/docs/api/java/lang/String.html")

(defonce text (let [writer (java.io.StringWriter.)] 
	(copy (-> (java.net.URL. url) .openStream) writer) writer))

(defonce jdoc-pattern #"(?s)<A NAME=\"(\w*)\((.*?)\).*?</A>.*?<PRE>(.*?)</PRE>(.*?)<HR>")

(def html-entities
  {#"<.*?>" ""
   #"&nbsp;" " "
   #"&lt;" "<"
   #"&gt;" ">"})

(def doc-matches
  {#"(?s)<TR.*?<CODE>.*?#(.*?)\".*?</CODE>.*?<CODE>(.*?)</CODE>(.*?)</TR>" [:ignored1 :returnType :name :summary]
   #"(?s)<A NAME=\"(\w*)\((.*?)\).*?</A>.*?<PRE>(.*?)</PRE>(.*?)<HR>" [:ignored2 :name :args :declaration :docs]})

(defn- plain-text [html]
  (reduce 
    (fn [str entry] 
      (let [matcher (re-matcher (key entry) str)]
        (if matcher (.replaceAll matcher (val entry)) str))) html html-entities))

(defn re-zipmap[text re bindings] 
  (let [matches (re-seq re text)]
    (map #(zipmap bindings (map plain-text %)) matches)))

(defn scrape [html]
  (apply join (map #(re-zipmap html (key %) (val %)) doc-matches)))

(defn jdoc [url func]
  "Takes the URL to a javadoc and a function name and returns plaintext-formatted documentation meant to be used by the doc macro"
  (let [docs (re-seq jdoc-pattern (str text))
        matches (filter #(= func (nth % 1)) docs)]
  (for [item matches, index [3 4], :let [html (item index)]] (plain-text html))))

(jdoc string-url "replaceAll")