Exemplo n.º 1
0
    def get_build_info(self, build_lis, co_id):
        for build_ in build_lis:
            build_url = "http://xx.yyfdcw.com" + build_
            try:
                build_res = requests.get(build_url, headers=self.headers)
            except Exception as e:
                print("co_index={},楼栋信息错误".format(co_index), e)
                continue
            con = build_res.text
            bu = Building(co_index)
            bu.co_id = co_id
            bu.bu_id = re.search('Bid=(\d+)', build_).group(1)
            bu.bu_num = re.search('名称.*?">(.*?)</spa', con).group(1)
            bu.bu_pre_sale = re.search("编.*?red'>(.*?)</a", con).group(1)
            bu.bu_pre_sale_date = re.search('颁发日期.*?Date">(.*?)</span',
                                            con).group(1)
            bu.bo_build_start_time = re.search('开工日期.*?">(.*?)</span',
                                               con).group(1)
            bu.bo_build_end_time = re.search('竣工日期.*?">(.*?)</span',
                                             con).group(1)
            bu.bo_develops = re.search('单位.*?">(.*?)</span', con).group(1)
            bu.bu_floor = re.search('层数.*?">(.*?)</span', con).group(1)
            bu.bu_live_size = re.search('住宅面积.*?">(.*?)</span', con).group(1)
            bu.size = re.search('总面积.*?">(.*?)</span', con).group(1)

            bu.insert_db()

            id = re.search('测量号.*?">(.*?)</span', con).group(1)
            self.get_house_info(co_id, bu.bu_id, id)
Exemplo n.º 2
0
 def get_build_info(self, build_url_list):
     for i in build_url_list:
         try:
             build = Building(co_index)
             build_url = 'http://222.223.160.199:8088/website/buildquery/selectBuild.jsp?buildID=' + i[0]
             response = requests.get(build_url, headers=self.headers)
             html = response.text
             build.bu_id = i[0]
             build.co_build_structural = re.search('结构类型.*?<td.*?>(.*?)<', html, re.S | re.M).group(1)
             build.bo_build_end_time = re.search('建成年份.*?<td.*?>(.*?)<', html, re.S | re.M).group(1)
             build.bu_build_size = re.search('总建筑面积.*?<td.*?>(.*?)<', html, re.S | re.M).group(1)
             build.bu_num = re.search('幢号.*?<td.*?>(.*?)<', html, re.S | re.M).group(1)
             build.size = re.search('占地面积.*?<td>(.*?)<', html, re.S | re.M).group(1)
             build.bu_floor = re.search('房屋层数.*?<td>(.*?)<', html, re.S | re.M).group(1)
             build.bu_all_house = re.search('房屋套数.*?<td>(.*?)<', html, re.S | re.M).group(1)
             build.area = re.search('坐落区.*?<td>(.*?)<', html, re.S | re.M).group(1)
             build.insert_db()
             self.get_house_info(build.bu_id)
         except Exception as e:
             print('请求错误,url={}'.format(build_url),e)