Ejemplo n.º 1
0
    def parse_detail(self, response):
        """具体处理获取某一个楼盘的信息"""

        sel = etree.HTML(response.text)
        item = LianjialoupanItem()

        # 楼盘名称
        name = sel.xpath(
            "//div[@class='name-box']/a[@class='clear']/@title")[0]
        item["name"] = name

        # 楼盘状态
        state = sel.xpath(
            "//div[@class='name-box']/div[@class='state-div']/span[@class='state']/text()"
        )[0]
        item['state'] = state

        # 楼盘网址
        href = sel.xpath("//div[@class='name-box']/a[@class='clear']/@href")[0]
        item["url"] = self.base_url + href

        # 楼盘价格
        jiage = sel.xpath(
            "//div[@class='box-left-top']/p[@class='jiage']/span/text()")
        # item["price"] = jiage
        if "均价" in jiage:
            # junjia = sel.xpath("//span[@class='junjia']/text()")
            # yuan = sel.xpath("span[@class='yuan']/text()")
            item["price"] = jiage[1] + jiage[2]
        else:
            item["price"] = jiage[0]

        # 物业
        try:
            wuye = sel.xpath("//p[@class='wu-type ']/span/text()")
            item['wuye'] = wuye[1]
        except Exception as e:
            wuye = sel.xpath("//p[@class='wu-type manager']/span/text()")
            item['wuye'] = wuye[1]
            print("Error:", e)

        # 地址
        try:
            where = sel.xpath("//p[@class='where ']/span/@title")
            item["where"] = where[0]
        except Exception as e:
            where = sel.xpath("//p[@class='where manager']/span/@title")
            item["where"] = where[0]
            print("Error:", e)

        yield item
Ejemplo n.º 2
0
    def parse_detail(self, response):
        """具体处理获取某一个楼盘的信息"""

        url = response.meta['url']
        sel = etree.HTML(response.text)
        item = LianjialoupanItem()

        # 楼盘名称
        name = sel.xpath(
            "//div[@class='title-row']/h1[@class='nameShow']/text()")[0]
        item["name"] = name

        # 楼盘状态
        state = sel.xpath(
            "//div[@class='title-row']/span[@class='status label']/text()")[0]
        item['state'] = state

        # 楼盘网址
        item["url"] = url

        # 楼盘价格
        jiage = sel.xpath("//span[@class='num price']/text()")
        if "已售完" in jiage or "待定" in jiage[0]:
            item["price"] = jiage[0]
        else:
            unit = sel.xpath("//span[@class='unit price']/text()")
            item["price"] = jiage[0] + unit[0]

        # 物业
        wuye = sel.xpath(
            "//div[@class='title-row']/span[@class='type label']/text()")[0]
        item["wuye"] = wuye

        # 地址
        where = sel.xpath("//span[@class='address']/text()")[0]
        item["where"] = where

        yield item