qunar 0.0.1 → 0.0.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: abb1c8b487e496de74a3d1aba1011297d09ec936
-  data.tar.gz: c31420590fc4344ff961faa8cf5a74c57dd66b49
+  metadata.gz: e4ba1d679946abe2f3016259dc8e0b4e7ee0379c
+  data.tar.gz: c55780c6f8d0986e17d8886dcf4b650076117b8e
 SHA512:
-  metadata.gz: 7595b89f23e5424c60c5f4d640d51bdc9b31da7a27802390ee2716df5b3909ffccf5d48b4f164cdad164d5d99a04e43ffac0d4abf05c912258b822f96ff66052
-  data.tar.gz: c0805458111c8147aff0a921abf7f58b1d9f28fc755f7311e00cbe574f8197af1062945db20e55222cb2daa798cef0acef16ec0fa4ee2289715f1eabe4d007b9
+  metadata.gz: e938a846a8799b42590edfca4e09b777dbe04094210b3830128b3727d735d492f7a3721783e2f75e97a61115754b4ab852fda934df45df5c8411ed243956ba65
+  data.tar.gz: 5df5d5d986d06d5249bd76891cc5ff253bf1949563b275bc031865f88ed2d51a7c305ee70eb97036919c97db459476de5b34f3530ade39ffe2d8e2f7c77b403f

data/.rspec

@@ -0,0 +1,2 @@
+--color
+--require spec_helper

data/lib/qunar/hotel.rb

@@ -4,7 +4,7 @@ module Qunar
   class Hotel
     BASEURI = "http://hotel.qunar.com"
     def initialize(hotelId)
-      #super   
+      #super
       raise "id should be string" if !hotelId.instance_of?(String)
       @HotelSEQ = hotelId             #hotelId is "cityurl_hotelcode"
       @cityurl = hotelId[0..hotelId.rindex("_",-1)-1]
@@ -14,24 +14,11 @@ module Qunar
       @hotel = Nokogiri::HTML page
     end
 
-    def parse_mixkey                            #get some kind of code unkown
-      uri = URI(File.join(BASEURI, 'city', @cityurl, "dt-#{@hotelcode}"))
-      http_res = HTTParty.get uri.to_s
-      res = Nokogiri::HTML http_res
-      node = res.search('//span[@id="eyKxim"]').first
-      return nil if node.nil?
-      return node.text
-    end
-
     def name          #get the name of the hotel
       @hotel.search("//div[@class='htl-info fr']//span").first.text.strip
-      # name_tag = @hotel.search("//div[@class='htl-info fr']/h2/span").first
-      # return nil if name_tag.nil?
-      # name_tag.text[0..name_tag.text.index("\n",1)].strip
     end
 
     def star         #get the level of the hotel
-      #name_tag = @hotel.search('//div[@class="eriefinfo"]//dt').first
       name_tag = @hotel.search('//div[@class="htl-info fr"]//em[@class="dangci"]').first
       return 0 if name_tag.nil?
       if name_tag.text.include?'经济型'
@@ -67,22 +54,18 @@ module Qunar
       @hotelcode = @hotelcode
     end
 
-    def address               
+    def address
       @hotel.search("//p[@class='adress']/span").first['title'].strip
       #return "" if address.nil?
       #return address['title'].strip
     end
 
     def phone
-      #phone ||= @hotel.search("//li[contains(text(),'电话：')]").text.sub('电话：','').strip
       @hotel.search("//dl[@class='inform-list']//cite").first.text.sub('电话','').strip
     end
 
     #the description of the hotel
-    def desc     
-      #desc = @hotel.search("p[@class='summery less_summery']").text
-      #desc = @hotel.search("p[@class='summery expan_summery']").text if desc.empty?
-      #desc.gsub(/[[:space:]]/, '')
+    def desc
       @hotel.search("div[@class='introduce']/p").text
     end
 
@@ -92,20 +75,8 @@ module Qunar
       #date.strip
       @hotel.search("dl[@class='inform-list']//cite")[-2].text.sub("年开业",'')
     end
-=begin
-    #简略描述
-    def abstract_desc
-      abstract_desc = @hotel.search("p[@class='h_desc']").first
-      return "" if abstract_desc.nil?
-      return abstract_desc.text.sub('描述：','').strip
-    end
 
-    def brand
-      brand ||= @hotel.search("//li[contains(text(),'所属品牌：')]").text.sub('所属品牌：','').strip
-      brand.strip
-    end
-=end    
-    def reference_price      
+    def reference_price
       match_data = @hotel.to_s.match(/var\s*miniRetailPrice\s*=\s*\'\d+\';/)
       return 0 if match_data.nil?
       match_data = match_data[0].match(/[\d]+/)
@@ -115,98 +86,12 @@ module Qunar
 
     def coordinate
       xy = @hotel.to_s.match(/hotelPoint=\[(\d+\.\d+),(\d+\.\d+)\];/)
-      puts xy
       if xy.nil?
         return [0.0,0.0]
       else
         return [xy[1],xy[2]]
       end
     end
-    #设施
-    def facilities
-      facility = @hotel.search("//div[@class=\"roundmilieu htintroborder\"]").first
-      facilities = Array.new()
-      unless facility.nil?
-        facility.search("dl").each do |dl|
-          ele = Hash.new
-          facility_cata = dl.search("dt").first.text.delete('：')
-          ele1 = Array.new
-          dl.search("dd/ul/li").each do |facility_type|
-            ele1 << facility_type.text
-          end
-          ele['name'] = facility_cata
-          ele['info'] = ele1
-          facilities << ele
-        end
-      end
-      facilities
-    end
 
-    #交通信息   not runable
-=begin
-    def traffic
-      traffic = "http://hotel.qunar.com/detail/detailMapData.jsp?seq=#{@hotelcode}&type=traffic,canguan,jingdian,ent"
-      traffic = HTTParty.get traffic
-      traffic_data = Array.new()
-      if traffic['ret'] == true
-        traffic['data']['ent'].each do |line|
-          ele = Hash.new
-          ele['name'] = line['name']
-          ele['distance'] = line['distance']
-          traffic_data << ele
-        end
-      end
-    end
-
-    def decorate_date    #not found
-      match_data = @hotel.to_s.match(/最后装修时间：\d+年/)
-      return "" if match_data.nil?
-      match_data[0].strip
-    end
-
-
-    def comments
-      url = "http://review.qunar.com/api/h/#{@hotelcode}/detail/rank/v1/page/1"
-      comments = HTTParty.get url
-      comments = JSON.parse comments
-      comments['data']['list'].map{|comment| JSON.parse(comment['content'])['feedContent'] }
-    end
-
-    #评论总数
-    def total_comment
-      url = "http://review.qunar.com/api/h/#{@hotelcode}/detail/rank/page/1?rate=all"
-      body = HTTParty.get url
-      match_data = body.match(/"count":\d+/)
-      return "0" if match_data.nil?
-      match_data[0].sub(/\"count\":/,"").strip
-    end
-    #好评
-    def good_comment
-      url = "http://review.qunar.com/api/h/#{@hotelcode}/detail/rank/page/1?rate=positive"
-      body = HTTParty.get url
-      match_data = body.match(/"count":\d+/)
-      return "0" if match_data.nil?
-      match_data[0].sub(/\"count\":/,"")
-    end
-
-
-    def images
-      url = "http://hotel.qunar.com/render/hotelDetailAllImage.jsp?hotelseq=#{@hotelcode}"
-      body = HTTParty.get url
-      body = JSON.parse(body)
-      return body['data']["all"] if(body['ret'])
-    end
-
-    def score
-      url = "http://review.qunar.com/api/h/#{@hotelcode}/detail"
-      body = HTTParty.get url
-      begin
-        body = JSON.parse body
-        return body["data"]["score"].strip
-      rescue Exception => e
-        return "0.0"
-      end
-    end
-=end
   end
 end

data/lib/qunar/rank.rb

@@ -1,11 +1,7 @@
-#require 'rubygems'
 require 'capybara'
-#require 'selenium-webdriver'
 require 'nokogiri'
 require 'capybara-webkit'
 
-#headless = Headless.new
-#headless.start
 Capybara.current_driver = :webkit
 Capybara.default_selector = :xpath
 Capybara.app_host = 'http://hotel.qunar.com'
@@ -32,21 +28,26 @@ module Qunar
         file.puts @page
       end
     end
-=begin
-      File.open("test1.html","w") do |file|
-        file.puts @page
-      end
-=end
 
-    def room
-      room = {}
+
+    def rooms
+      rooms = []
+      # room
       @hotel.search("//ul[@class='htl-type-list clrfix']/li").each do |li|
+        room = {}
         type = li.search("span[@class='type-name']").first.text.strip
+        room['room_type'] = type
+        room['otas'] = []
+        # ota
 
         li.search("div[@class='similar-type-agent-list']/div/div").each do |row|
-          ota_pic = row.search("div[@class='agent-pic']/img").first['src'].strip
+          otas = {}
+
+          ota = row.search("div[@class='profile-tit']").first.text.strip
+          otas['ota'] = ota
 
           bookable = row.search("a[@class='btn-book']/span").first.text.strip
+          otas['bookable'] = bookable
 
           insurance = row.search("p[@class='dbt-ct']").first
           if insurance == nil
@@ -54,6 +55,7 @@ module Qunar
           else
             insurance = true
           end
+          otas['insurance'] = insurance
 
           coupon = row.search("span[@class='fan']/em[@class='pr'] |span[@class='fan laps']/em[@class='pr']").first
           if coupon == nil
@@ -61,6 +63,7 @@ module Qunar
           else
             coupon = coupon.text.sub(/¥/,'').strip
           end
+          otas['coupon'] = coupon
 
           prepay = row.search("span[@title='需要预先支付房款']").first
           if prepay == nil
@@ -68,106 +71,22 @@ module Qunar
           else
             prepay = true
           end
+          otas['prepay'] = prepay
 
           price = row.search("p[@class='final-price ']/b").first.text.strip
-          neat_price = row.search("p[@class='count_pr']").first.text
-
-          puts "#{type} #{ota_pic} #{price} #{neat_price} #{coupon} #{prepay} #{insurance}  #{bookable}" 
-        end
-
-      end
-      
-      return 0
-    end
-
-
-  end
-end
-
-
-
-=begin
-require 'nokogiri'
-require 'headless'
-require 'capybara'
-require 'capybara/dsl'
-module Qunar
-  class Rank
-    include Capybara::DSL
-    BASEURI = "http://hotel.qunar.com"
-    def initialize(hotelId)
-      raise "id should be string" if !hotelId.instance_of?(String)  #ckeck the whether hotelId is a string
-      @HotelSEQ = hotelId                #hotelId is "cityurl_hotelcode"
-      @cityurl = hotelId[0..hotelId.rindex("_",-1)-1]
-      @hotelcode = hotelId[hotelId.rindex("_",-1)+1..hotelId.length]
-    end
-
-    def parse_mixkey
-      uri = URI(File.join(BASEURI, 'city', @cityurl, "dt-#{@hotelcode}"))    #ex:http://hotel.qunar.com/city/beijing_city/dt-2933/
-      http_res = HTTParty.get uri.to_s
-      res = Nokogiri::HTML http_res
-      node = res.search('//span[@id="eyKxim"]').first
-      return nil if node.nil?
-      return node.text
-    end
+          otas['price'] = price
 
-    def set_cookies
-      headless = Headless.new
-      headless.start
-      Capybara.current_driver = :selenium
-      Capybara.app_host = BASEURI
-      Capybara.default_wait_time = 10
-      begin
-        visit "/city/#{@cityurl}/dt-#{@hotelcode}/"
-      rescue Errno::ECONNREFUSED => e
-        raise
-      end
-      @cookies = {}
-      Capybara.current_session.driver.browser.manage.all_cookies.each { |cookie| @cookies[cookie[:name]] = cookie[:value] }
-    end
+          neat_price = row.search("p[@class='count_pr']").first.children[1].to_s
+          otas['neat_price'] = neat_price
 
-    def load_price_ajax
-      mixKey = parse_mixkey
-      set_cookies
-      raise QunarMixKeyParsedException if mixKey.nil?
-      uri = URI(File.join(BASEURI, 'render', 'detailV2.jsp'))
-      params = {:fromDate=>Date.today+2, :toDate=>Date.today+3, :cityurl=>@cityurl, :HotelSEQ=>@HotelSEQ, :mixKey=>mixKey}
-      headers = Hash.new
-      headers['User-Agent'] = 'Mozilla/5.0 (Windows NT 5.1; rv:26.0) Gecko/20100101 Firefox/26.0'
-      headers['Connection'] = 'keep-alive'
-      headers['Accept-Encoding'] = 'gzip, deflate'
-      headers['Cookie'] = @cookies.map{|k, v| k+'='+v}.join('; ')
-      uri.query = URI.encode_www_form(params)
-      res = HTTParty.get(uri.to_s, :headers=>headers)
-      File.open("test1.html","w") do |file|
-        puts res
-      end
+          room['otas'] << otas
+        end
+        rooms << room
 
-      json_ajax = parse_json_from_response_body(res.body)
-      if !json_ajax['errcode'].nil? and json_ajax['errcode'].eql?110
-        raise QunarIpBlockException,json_ajax
       end
-      raise QunarIpBlockException,json_ajax if !json_ajax['errcode'].nil? and json_ajax['errcode'].eql?110
-      @ranks = json_ajax['result'].values
-    end
-
-    def rank(ota, room)
-      @ranks.each { |rank| return rank[15]+1 if rank[3].eql?room and rank[6].eql?ota and rank[9].eql?1 }
-      return nil
+      return rooms
     end
 
-    def rooms
-      @rooms ||= @ranks.map { |e|  e[3] }.uniq
-    end
 
-    def otas
-      @otas ||= @ranks.map{|e| e[6] }.uniq
-    end
-
-    def parse_json_from_response_body(res)
-      JSON.parse(res.gsub(/[()]/,""))
-    end
-    
   end
 end
-=end

data/lib/qunar/version.rb

@@ -1,3 +1,3 @@
 module Qunar
-  VERSION = "0.0.1"
+  VERSION = "0.0.2"
 end

data/lib/qunar.rb

@@ -1,6 +1,6 @@
-require_relative 'qunar/version'
-require_relative 'qunar/hotel'
-require_relative 'qunar/rank'
+require 'qunar/version'
+require 'qunar/hotel'
+require 'qunar/rank'
 
 module Qunar
   # Your code goes here...

data/spec/hotel_spec.rb

@@ -0,0 +1,88 @@
+require 'qunar/hotel'
+
+describe Qunar::Hotel do 
+	bj_sample = Qunar::Hotel.new('beijing_city_2933')
+
+	it "has name '北京诺富特三元酒店'" do 
+		# subject.name.should == '北京诺富特三元酒店'
+		expect(bj_sample.name).to eq '北京诺富特三元酒店'
+	end
+
+	it "has star 4" do
+		expect(bj_sample.star).to eq 4
+	end
+
+	it "has citynameEn 'beijing'" do
+		expect(bj_sample.citynameEn).to eq "beijing"
+	end
+
+	it "has citynameCn '北京' " do
+		expect(bj_sample.citynameCn).to eq "北京"
+	end
+
+	it "has id '2933' " do
+		expect(bj_sample.id).to eq "2933"
+	end
+
+	it "has address '北京朝阳区曙光西里A5号18座' " do
+		expect(bj_sample.address).to eq '北京朝阳区曙光西里A5号18座'
+	end
+
+	it "has phone '010-58296666'" do
+		expect(bj_sample.phone).to eq "010-58296666"
+	end
+
+	
+	it "has desc '北京诺富特三元酒店坐落于三元桥东北角，紧靠地铁十号线以及机场快轨线，周边老国展中心和商业中心繁盛，交通十分便捷。 距机场仅 约16分钟车程。酒店距使馆区、燕莎购物中心、秀水街和3个展览中心仅几分钟车程。' " do
+		expect(bj_sample.desc).to eq "北京诺富特三元酒店坐落于三元桥东北角，紧靠地铁十号线以及机场快轨线，周边老国展中心和商业中心繁盛，交通十分便捷。 距机场仅 约16分钟车程。酒店距使馆区、燕莎购物中心、秀水街和3个展览中心仅几分钟车程。"
+	end
+
+
+	it "has opentime '2008' " do
+		expect(bj_sample.opentime).to eq "2008"
+	end
+
+	it "has reference_price '498'" do
+		expect(bj_sample.reference_price).to eq '498'
+	end	
+
+
+	it "has coordinate ['39.961277','116.45537'] " do
+		expect(bj_sample.coordinate).to eq ['39.961277','116.45537']
+	end
+
+	#=====================================================================================================
+
+	
+	gz_sample = Qunar::Hotel.new('guangzhou_49')
+
+	it "has name '广州长隆酒店'" do 
+		# subject.name.should == '北京诺富特三元酒店'
+		expect(gz_sample.name).to eq '广州长隆酒店'
+	end
+
+	it "has star 5" do
+		expect(gz_sample.star).to eq 5
+	end
+
+	it "has citynameEn 'guangzhou'" do
+		expect(gz_sample.citynameEn).to eq "guangzhou"
+	end
+
+	it "has citynameCn '广州' " do
+		expect(gz_sample.citynameCn).to eq "广州"
+	end
+
+	it "has id '49' " do
+		expect(gz_sample.id).to eq "49"
+	end
+
+	it "has address '广州市番禺区番禺大道' " do
+		expect(gz_sample.address).to eq '广州市番禺区番禺大道'
+	end
+
+	it "has phone '020-84786838'" do
+		expect(gz_sample.phone).to eq "020-84786838"
+	end
+
+end

data/spec/rank_spec.rb

@@ -0,0 +1,87 @@
+require 'qunar/rank'
+
+describe Qunar::Rank do
+  sample = Qunar::Rank.new('beijing_city_2933')
+ 
+  it "has 7 rooms" do
+    # when directly check the hash object the Chinese charactervalue is saved as format like "\u6807\u51C6\u53CC\u5E8A\u623F", it is not been parsed here, so it is not equal to chinese charactor, it's just a sequnce of coding,it needs to be parse,then it could show as Chinese character.
+    # room = {"room_type" => "\u6807\u51C6\u53CC\u5E8A\u623F", "ota" => "\u827A\u9F99\u65C5\u884C\u7F51", "bookable" => "\u9884\u8BA2", "insurance" => true, "coupon" => "0", "prepay" => false, "price" => "918", "neat_price" => "918"}
+    expect(sample.rooms.size).to eq 7
+  end
+
+  it "first room names '标准双床房'" do 
+    expect(sample.rooms.first['room_type']).to eq '标准双床房'
+  end
+
+  it "second room names '标准大床房'" do 
+    expect(sample.rooms[1]['room_type']).to eq '标准大床房'
+  end
+
+  it "has 22 otas for the first room type '标准双床房'" do 
+    expect(sample.rooms.first['otas'].size).to eq 22
+  end
+
+  it "has 20 otas for the second room type '标准大床房'" do 
+    expect(sample.rooms[1]['otas'].size).to eq 20
+  end 
+
+  it "first ota of first roomtype names '艺龙旅行网'" do
+    #puts subject.rooms.first['otas'].first['ota']
+    expect(sample.rooms.first['otas'].first['ota']).to eq '艺龙旅行网'
+  end
+
+  it "first ota of second roomtype names Expedia" do
+    #puts subject.rooms.first['otas'].first['ota']
+    expect(sample.rooms[1]['otas'].first['ota']).to eq 'Booking.com'
+  end
+
+  it "first ota of first roomtype is bookable" do
+    #puts subject.rooms.first['otas'].first['bookable']
+    expect(sample.rooms.first['otas'].first['bookable']).to eq '预订'
+  end
+
+  it "last ota of first roomtype is bookable" do
+    #puts subject.rooms[-1]['otas'].first['bookable']
+    expect(sample.rooms.first['otas'][-1]['bookable']).to eq '订完'
+  end
+
+  it "first ota of first roomtype need insurance" do
+    expect(sample.rooms.first['otas'].first['insurance']).to eq true
+  end
+
+  it "third ota of first roomtype need insurance" do
+    #puts subject.rooms.first['otas'][2]['insurance']
+    expect(sample.rooms.first['otas'][2]['insurance']).to eq false
+  end
+
+  it "first ota of first roomtype coupon is 0" do
+    expect(sample.rooms.first['otas'].first['coupon']).to eq '0'
+  end
+
+
+  it "first ota of first roomtype need prepay" do
+    expect(sample.rooms.first['otas'].first['prepay']).to eq false
+  end
+
+  it "second ota of first roomtype need prepay" do
+    expect(sample.rooms.first['otas'][1]['prepay']).to eq true
+  end
+
+  it "first ota of first roomtype price" do
+    expect(sample.rooms.first['otas'].first['price']).to eq '918'
+  end
+
+  it "sixth ota of first roomtype price" do
+    expect(sample.rooms.first['otas'][5]['price']).to eq '638'
+  end
+
+  it "first ota of first roomtype neat_price" do
+    expect(sample.rooms.first['otas'].first['neat_price']).to eq '918'
+  end
+
+  it "sixth ota of first roomtype neat_price" do
+    #puts subject.rooms.first['otas'][5]['neat_price']
+    expect(sample.rooms.first['otas'][5]['neat_price']).to eq '734'
+  end
+
+end