ochingin.clj

(ns ochingin
  (:require [clojure.java.io :as io])
  (:use [incanter core io charts]
        [net.cgrand enlive-html]))

(defn salary-range-us [lang]
  (with-open [rdr (io/reader (str "http://www.indeed.com/jobs?l=&q=" lang))]
    (let [ds (->>
              (-> (select (html-resource rdr) [:div (attr= :id "rb_Salary Estimate")])
                  (select [:li :a]))
              (map #(re-matches #"\$([\d,]+)\+ \((\d+)\)" (get-in % [:attrs :title])))
              (map (fn [_] (-> _
                               (update-in [1] #(-> % (.replace "," "") (Long/parseLong)))
                               (update-in [2] #(-> % (.replace "," "") (Long/parseLong))))))
              (dataset [:label :salary :count]))
          total (apply + (sel ds :cols :count))]
      (add-column :percentage ($map #(/ % total) :count ds) ds))))

(defn salary-range [lang]
  (with-open [rdr (io/reader (str "http://jp.indeed.com/%E6%B1%82%E4%BA%BA?l=&q=" lang))]
    (let [ds (->>
              (-> (select (html-resource rdr) [:div (attr= :id "rb_Salary Estimate")])
                  (select [:li :a]))
              (map #(re-matches #"([\d,]+)万円~ \((\d+)\)" (get-in % [:attrs :title])))
              (map (fn [_] (-> _
                               (update-in [1] #(-> % (.replace "," "") (Long/parseLong)))
                               (update-in [2] #(-> % (.replace "," "") (Long/parseLong))))))
              (dataset [:label :salary :count]))
          total (apply + (sel ds :cols :count))]
      (add-column :percentage ($map #(/ % total) :count ds) ds))))

(defn draw-salary-distribution [words]
  (let [ds (salary-range (first words))
        chart (xy-plot (sel ds :cols :salary)
                       (sel ds :cols :percentage)
                       :legend true
                       :series-label (first words)
                       :x-label "Salary"
                       :y-label "Percentage")]
    (doseq [lang (rest words)]
      (let [ds (salary-range lang)]
        (add-lines chart
                   (sel ds :cols :salary)
                   (sel ds :cols :percentage)
                   :series-label lang)))
    (view chart)))