전국 어린이집 좌표 가져오기

gistfile1.rb

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

TIMEOUT_CNT = 42
#cd = '29200000641'
#cd = '11620000061'

def print_nursery_geo(nursery_code) 
  cd = nursery_code
  retries = TIMEOUT_CNT
  req = "/cpis2gi/nursery/NurseryMap.jsp?flag=NA&stcode=#{cd}&ctprvn1=41000&ctprvn2=41000&signgu1=40100&signgu2=40100&dong1=&dong2=&road1=&road2="
  begin
    Timeout::timeout(5) {
      doc = Net::HTTP.get('www.childcare.go.kr', req)
      if doc.nil?
        puts "PRINT,#{cd},ERROR"
      else
        aa = doc.scan(/position: new daum.maps.LatLng\((.*), (.*)\)/)
        if aa.nil?
          puts "GEO,#{cd},null,null"
        else
          puts "GEO,#{cd},#{aa[0][0]},#{aa[0][1]}"
        end
      end
    }
  rescue OpenURI::HTTPError
    puts "ERROR,PRINT,#{nursery_code},HTTP"
  rescue EOFError
    puts "ERROR,PRINT,#{nursery_code},EOF"
    retries -= 1
    if retries > 0
      puts "sleep"
      sleep TIMEOUT_CNT - retries
      retry
    else
      puts "raise"
      raise
    end
  rescue Timeout::Error
    retries -= 1
    if retries > 0
      puts "sleep"
      sleep TIMEOUT_CNT - retries
      retry
    else
      puts "raise"
      raise
    end
  end # TIMEOUT
end

province = ['11', '26', '27', '28',  '29', '30', '36', '41', '42', '43', '44', '45','46', '47', '48','49']

province.each do |prvn|
  puts "FETCH,#{prvn}"
  #for offset in 1..10000
  offset = 1 #5191 #prvn=41 #1 #6710
  while true
    #break if offset > 2
    cnt = 0
    codes = []
    names = []

    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))
        codes = []
        names = []
        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 { |hr| hr.scan(/'(.*)'/) { |x| codes <<  x[0]} }
        jj.each { |nm| names << nm.content }
        puts "NEXT,#{prvn},#{offset},#{codes.length}"
        has_next = true  if codes.length <= 0
        if codes.length > 0 
          cnt = codes.length
        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
    for i in 0..codes.length-1
      puts "PRINT,#{codes[i]}"
      puts "INFO,#{codes[i]},이름,#{names[i]}"
      print_nursery_geo(codes[i])
    end # for i
    break if cnt <= 0
    offset += cnt
  end # offset 1 to 10000 step by codes.length
end # prvn
puts "END"