Github の言語人気順位をとってくる

getpopularity.rb

#!/usr/local/bin/ruby
# encoding: utf-8

require 'rest_client'
require 'nokogiri'

def get_popularity(link)
  doc = Nokogiri::HTML(RestClient.get "https://github.com#{link.attr(:href)}")
  title = doc.css(".pagehead h1").first.content
  unless title =~ /(.+) is the(?: #(\d+))? most popular language on GitHub/
    raise "parse error #{link}"
  end
  [$2.to_i, $1].tap { |rslt| p rslt }
end

def main
  doc = Nokogiri::HTML(RestClient.get 'https://github.com/languages')
  langs = doc.css(".all_languages a").map { |link| get_popularity(link) }
  langs.sort_by(&:first).each do |popularity, lang|
    popularity = 1 if popularity == 0
    if popularity < 51
      puts "<tr><td>#{popularity}</td><td>#{lang}</td></tr>"
    else
      print ", #{lang}"
    end
  end
  print "\n"
end

main