gistfile1.rb

# 获取购买记录
def get_trades(crawler, item_id, seller_id, item_sbn, page=1)

  start_at = Date.yesterday.beginning_of_day.to_i * 1000
  end_at   = Date.yesterday.end_of_day.to_i * 1000

  # 购买列表
  crawler.url    = 'http://tbskip.taobao.com/json/show_buyer_list.htm'
  crawler.params = {
    bid_page:      page,
    ends:          end_at,
    starts:        start_at,
    item_id:       item_id,
    seller_num_id: seller_id,
    sbn:           item_sbn,
  }
  item_url = "http://detail.tmall.com/auction/item_detail.htm?item_num_id=#{item_id}&show_review=1&tbpm=1"
  crawler.headers = { 'Host' => 'tbskip.taobao.com', 'Referer' => item_url }
  body = crawler.request.execute
  if body
    body = Zlib::GzipReader.new(StringIO.new(body)).read.force_encoding('GB18030').encode('UTF-8')
    body = Nokogiri::HTML(body)
    return body
  end
end

# 宝贝SBN，用于获取购买记录
def parse_sbn(page_dom)
  sbn_dom  = page_dom.at('button#J_listBuyerOnView') # 找节点
  item_sbn = nil
  if sbn_dom
    sbn_dom  = sbn_dom['detail:params'] # 淘宝自定义属性
    item_sbn = sbn_dom.match(/sbn=(.+)\,showBuyerList/) if sbn_dom
    item_sbn = item_sbn[1] if item_sbn
  else
    logger.warn "无法解析，宝贝购买记录";
  end
  return item_sbn
end
# 拆分交易属性
def props_split(str)
  props = {}
  str.split(';').each do |prop|
    prop = prop.split(':')
    props[prop[0]] = prop[1]
  end
  return props
end

url = "http://jump.taobao.com/jump"
crawler  = Crawler.new(url)
crawler.params  = { target: 'http://detail.tmall.com/auction/item_detail.htm?item_num_id=20485336962&show_review=1&tbpm=1' }
page_dom = crawler.get_dom
if page_dom
  detail_dom  = page_dom.at('div#detail>script')
  if detail_dom
    detail_dom = detail_dom.text.gsub(/\s+/, '').strip
    detail_dom = detail_dom.match(/\"itemDO\"\:(.*)\,\"detail/)
    if detail_dom
      json = detail_dom[1].gsub("'","\"")
      item_info = ActiveSupport::JSON.decode(json)
      item = {}

      item[:id]     = item_info['itemId']
      item[:type]   = item_info['auctionType']
      item[:status] = case
      when item_info['isOnline'] == true
        '在售'
      when item_info['isOnline'] == true
        '售罄'
      else
        '下架'
      end

      item[:category_id] = item_info['categoryId'].to_i
      item[:seller_id]   = item_info['userId'].to_i
      item[:spu_id]      = item_info['spuId'].to_i

      item[:tag_price] = item_info['reservePrice'].to_f

      item[:brand_id] = item_info['brandId'].to_i
      item[:brand]    = item_info['brand']

      item[:quantity] = item_info['quantity'].to_i

      item[:item_sbn] = parse_sbn(page_dom)
      
      trades_dom  = get_trades(crawler, item[:id], item[:seller_id], item[:item_sbn])
      page   = 1
      if trades_dom.at('span.page-end').nil?
        page += 1
      end
      trades_dom  = trades_dom.css('table>tr')
      trades.each do |trade_dom|
        trade_dom = trade_dom.text.split(' ')
        if trade_dom.last == '成交'
          trade = { buyer: trade_dom[0], props: props_split(trade_dom[2]), price: trade_dom[3].to_f, num: trade_dom[4].to_i, date: Time.parse("#{trade_dom[5]} #{trade_dom[6]}") }
          puts trade
        end
      end

    else
      logger.warn "无法解析，宝贝基础信息";
    end
  else
    logger.warn "无法解析，宝贝详情";
  end
else
  logger.error "没什么好说的了~~";
end













json     = str.match(/\"skuMap\"\:(.*)\}\,\"isSevenDaysRefundment/)[1].gsub("'","\"")
item_sku = ActiveSupport::JSON.decode(json)
item_sku.each do |tag, sku|
  props_tag = tag.split(';')
  props_tag.delete("")

  sku_props = {}
  props_tag.each do |prop_tag|
    if props.has_key?(prop_tag)
      prop = props[prop_tag]
      sku_props[prop[:type]] = prop[:name]
    end
  end
  s = { id: sku['skuId'], tag: tag, quantity: sku['stock'].to_i, props: sku_props }
  puts s
end

# 下架宝贝无此节点，宝贝SKU属性
props_dom = page_dom.at('div.tb-sku').css('dl.tb-prop')
props     = {}
props_dom.each do |prop_dom|
  prop_type = prop_dom.at('dt').text
  if prop_type
    prop_dom.css('dd>ul>li').each do |el|  
      props[el['data-value']] = { type: prop_type, name: el.at('a>span').text}
    end
  end
end

# 创建宝贝时，需要解析产品属性
def parse_attributes(page_dom)
  attributes_dom = page_dom.css('div#attributes>div.attributes-list>ul>li')
  attributes     = {}
  if attributes_dom
    attributes_dom.each do |attribute|
      str = attribute.text.strip
      if str.count(" ") > 2
        arr = str.split(' ')
        attr_name = arr[0].gsub(':', '')
        attributes[attr_name] = (arr - [arr[0]]).join(',')
      else
        arr = str.split(':')
        attr_name = arr[0]
        attributes[attr_name] = arr[1].gsub(" ", '')
      end
    end
  else
    logger.warn "无法解析，产品属性";
  end
  return attributes
end