건강보험심사평가원 병원 위치 등 일반현황 가져오기

hir.getinfo.rb

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

TIMEOUT_CNT = 42

def print_info code
  retries = TIMEOUT_CNT
  begin
    Timeout::timeout(5) {
      info_dt = []
      info_dd = []
      doc = Nokogiri::HTML(open("http://m.hira.or.kr/eva/data.do?view=hos&code=#{code}"))
      doc.xpath('//h3').each do |a|
        info_dt << "병원이름"
        info_dd << a.inner_text
      end
      doc.xpath('//dl/dt').each do |a|
        info_dt << a.inner_text.scan(/(.*) :/).join
      end
      doc.xpath('//dl/dd').each do |a|
        info_dd << a.inner_text
      end
      
      doc.xpath('//table/tbody/tr/th').each do |a|
        info_dt << a.inner_text
      end
      doc.xpath('//table/tbody/tr/td').each do |a|
        info_dd << a.inner_text
      end

      while info_dt.length < info_dd.length
        info_dd.pop
      end
      
      while !info_dt.empty?
        puts  code + "," + info_dt.pop + "," + info_dd.pop
      end
      
      ch = doc.css('.lst_clinic').children
      ch.each do |a|
        puts code + "," + a.inner_text.scan(/(.*) \((.*)\/(.*)\)/).join(',') if a.type == 1
      end
    }
  rescue Timeout::Error
    retries -= 1
    if retries > 0 
      puts "sleep"
      sleep 1.42
      retry
    else
      puts "raise"
      raise
    end
  end
end # def print_info

def get_info begin_n, thru_n # [ )
  n = begin_n
  begin 
    retries = TIMEOUT_CNT
    code_list = Array.new
    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
            code = code.join()
            puts "#{n},#{cnt},#{code},'list',#{retries}"
            code_list.push(code)
          end
        end #doc.xpath
      }
    rescue Timeout::Error
      retries -= 1
      if retries > 0 
        puts "sleep"
        sleep 1.42
        retry
      else
        puts "raise"
        raise
      end
    end
    retries = TIMEOUT_CNT
    cl_cnt = 0
    code_list.each do |cl|
      cl_cnt += 1
      puts "#{n},#{cnt},#{cl},FETCH,-"
      begin
        print_info cl  
      rescue OpenURI::HTTPError
        puts "#{n},#{cl_cnt},#{cl},NOT_FOUND,-"
        next
      end
    end # code_list
    n = n + 1
    sleep 0.42
  end while 0 < cnt && n < thru_n
  puts "#{n},#{cnt},0_LT_CNT,-,-"
end # def
  
if ARGV.length == 2
  bn, tn = ARGV
  get_info bn.to_i, tn.to_i
end