전국 어린이집 목록(코드, 이름) 출력하는 루비 코드

gistfile1.rb

#encoding: utf-8
require 'net/http'
require 'nokogiri'
require 'open-uri'
require 'timeout'

TIMEOUT_CNT = 42
cd = '29200000641'
   #<ul class="nursry_search_list">
   #http://m.childcare.go.kr/nursery/mAllNurserySlPL.jsp?programId=null&flag=NSSlPL&offset=10000&ctprvn=11000&signgu=&dong=&road=&crtype=&crspec=&crcert=&crpub=&crname=&x=36&y=14
province = ['11', '26', '27', '28',  '29', '30', '36', '41', '42', '43', '44', '45','46', '47', '48','49']
#province = ['45']
province.each do |prvn|
  puts "FETCH,#{prvn}"
  #for offset in 1..10000
  offset = 1 #6710
  while true
    cnt = 0
    url = "http://m.childcare.go.kr/nursery/mAllNurserySlPL.jsp?programId=null&flag=NSSlPL&offset=#{offset}&ctprvn=#{prvn}000&signgu=&dong=&road=&crtype=&crspec=&crcert=&crpub=&crname="
    retries = TIMEOUT_CNT
    begin
      Timeout::timeout(5) {
        puts "FETCH,#{prvn},#{offset}"
        doc = Nokogiri::HTML(open(url))
        names = []
        codes = []
        kk = doc.at_css("ul.nursry_search_list").css("a").map { |link| link['href'] }
        jj = doc.at_css("ul.nursry_search_list").css("li[@class = 'first orange']/strong")
        kk.each do |hr|
          #hr.scan(/'(.*)'/) { |x| puts x}
          hr.scan(/'(.*)'/) { |x| codes <<  x[0]}
        end
        jj.each do |nm|
          #puts nm.content
          names <<  nm.content
        end
        puts "NEXT,#{prvn},#{offset},#{codes.length}"
        has_next = true  if codes.length <= 0
        if codes.length > 0 
          cnt = codes.length
          for i in 0..codes.length-1
            puts "#{codes[i]},#{names[i]}"
          end
        end 
      }
    rescue OpenURI::HTTPError
      puts "FETCH,#{prvn},#{offset},ERROR"
      next
    rescue Timeout::Error
      retries -= 1
      if retries > 0
        puts "sleep"
        sleep 0.42 * (TIMEOUT_CNT - retries)
        retry
      else
        puts "raise"
        raise
      end
    end # TIMEOUT
    break if cnt <= 0
    offset += cnt
  end # offset 1 to 10000 step by codes.length
end # prvn
puts "END"

version 1.9.3 tested