商標検索をもとに読みを取得する

syouhyou_yomi.rb

#!/bin/env ruby
# encoding: utf-8

require 'mechanize'
def syouhyou_yomi(*keywords)
  keywords.flatten!
  wait_time = 0.2

  agent = Mechanize.new
  agent.user_agent_alias = 'Mac Mozilla'
  #agent.log = Logger.new(STDERR)

  results = {}
  keywords.each{ |keyword|
    # キーワード内のスペースは除去します
    keyword.gsub!(" ", "")

    top_url = "https://www.j-platpat.inpit.go.jp/web/all/top/BTmTopPage"
    page = agent.get(top_url)
    #sess_id = page.search("#jplatpatSession").first.text

    page = page.form_with(:name => "searchForm") do |form|
      form.action = '/web/all/top/BTmTopSearchPage.action'
      sess_id = page.search("#jplatpatSession").first.text
      form.field_with(:name => 'bTmFCOMDTO.searchKeyword').value = keyword
      form.field_with(:name => 'bTmFCOMDTO.searchValue').value = "4"
      form.field_with(:name => 'bTmFCOMDTO.enzanShiValue').value = "1"
      form.add_field!('jplatpatSession', sess_id)
    end.submit()

    result_count = page.search(".searchbox-result-count").text.gsub("件", "").to_i
    STDERR.puts "(#{keyword}) result count: #{result_count}"
    STDERR.puts page.body
    if result_count == 0
      # not found yomi
      sleep wait_time
      next
    end

    page = page.form_with(:name => 'syouhyouLink') do |form|
    end.submit()

    tr = page.search(".table-result > tbody > tr").first
    url = tr.search("td > a").first.attributes["href"].value

    page = agent.get(url)
    #yomi = page.search("/html/body/div/div[6]/div/div[2]/div[1]/table[2]/tr[3]/td").text
    yomi = nil
    page.search(".table-information > tr").each{ |tr|
      yomi_index = nil
      tr.search("th").each_with_index{ |th, index|
        if th.text == "(561)称呼（参考情報）"
          yomi_index = index
        end
      }
      if yomi_index
        yomi = tr.search("td")[yomi_index].text
        break;
      end
    }
    # 複数候補がある場合は, で区切られている
    yomis = yomi.split("，")
    results[keyword] = yomis
    sleep wait_time
  }
  results
end

#p syouhyou_yomi(["BeachBunny", "mabitex", "stodja", "KSUBI"])