건강보험심사평가원 병원 코드 가져오기

hira.getcode.rb

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

TIMEOUT_CNT = 42
n = 1 # 500 # 1614
begin 
  retries = TIMEOUT_CNT
  begin
    cnt = 0
    code = ""
    Timeout::timeout(5) {
      doc = Nokogiri::HTML(open("http://m.hira.or.kr/eva/list.do?cateID=&p=#{n}"))
      doc.xpath('//li/a').each do |a|
        code = a['href'].scan(/code=(.*)/)
        if !code.empty?
          cnt = cnt + 1 # if retries >= TIMEOUT_CNT
          code = code.join()
          puts "#{n},#{cnt},#{code},'list',#{retries}"
        end
      end #doc.xpath
    }
  rescue Timeout::Error
    retries -= 1
    if retries > 0 
      puts "sleep"
      sleep 0.42
      retry
    else
      puts "raise"
      raise
    end
  end
  puts "#{n},#{cnt},CHECK_CODE,-,-"
  n = n + 1
end while 0 < cnt
puts "#{n},#{cnt},0_LT_CNT,-,-"