def build_crawler(self, co_id, co_name, comm_con): bu = Building(co_index, co_id=co_id, co_name=co_name) build_list = re.search('查看楼盘表.*?<tr>(.*?)</table>', comm_con, re.S | re.M).group(1) build = re.findall('<tr>(.*?)</tr>', build_list, re.S | re.M) for bul in build: try: bul_html = etree.HTML(bul) buli = bul_html.xpath("//td/text()") bu.bu_num = bu_num = buli[1] bu.bu_all_house = buli[2] bu.size = buli[3] house_url = re.search( r'"(.*?)" t', bul, ).group(1) bu.bu_id = bu_id = re.search('-(\d+)', house_url).group(1) bu.insert_db() except: continue self.house_crawler(house_url, bu_num, co_id, bu_id)
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)
def comm_info( self, con, ): # 小区及楼栋 comm = Comm(co_index) comm.co_name = con.xpath( "//span[@id='ctl00_ContentPlaceHolder2_web_item_retail1_lb_item_name']/text()" )[0] # 小区名称 co_id_str = con.xpath("//form[@id='aspnetForm']/@action")[0] # 小区id comm.co_id = re.search(r"\d+", co_id_str).group(0) comm.co_address = con.xpath( "//span[@id='ctl00_ContentPlaceHolder2_lb_item_seat']/text()")[ 0] # 小区地址 comm.co_develops = con.xpath( "//span[@id='ctl00_ContentPlaceHolder2_lb_enter_name']/text()")[ 0] # 开发商 comm.co_size = con.xpath( "//span[@id='ctl00_ContentPlaceHolder2_lb_area']/text()")[0] # 总面积 comm.co_build_size = con.xpath( "//span[@id='ctl00_ContentPlaceHolder2_lb_item_area']/text()")[ 0] # 建筑面积 comm.co_build_end_time = con.xpath( "//span[@id='ctl00_ContentPlaceHolder2_lb_item_ew_date']/text()")[ 0] # 竣工时间 comm.co_plan_pro = con.xpath( "//span[@id='ctl00_ContentPlaceHolder2_lb_program_pcode']/text()")[ 0] # 用地规划许可 comm.co_work_pro = con.xpath( "//span[@id='ctl00_ContentPlaceHolder2_lb_jg']/text()")[0] # 施工许可 comm.co_green = con.xpath( "//span[@id='ctl00_ContentPlaceHolder2_lb_item_green_rate']/text()" )[0] # 绿地百分比 comm.co_land_use = con.xpath( "//span[@id='ctl00_ContentPlaceHolder2_lb_td']/text()")[0] # 土地使用证 comm.insert_db() build = Building(co_index) build_table = con.xpath("//tr[@style='color:#000066;']") room_list = [] for build_list in build_table: build.co_id = comm.co_id build.co_name = comm.co_name build_info = build_list.xpath("./td/text()") build.bu_id = build_info[0] build.bu_num = build_info[1] build.bu_all_house = build_info[2] build.size = build_info[3] build.bu_floor = build_info[4] build.bu_pre_sale = build_info[5] build.insert_db() room_url = build_list.xpath("./td/a/@href")[0] room_list.append(room_url) return room_list
def build_info(self, bu_info_list, co_id): for bu_info in bu_info_list: try: bu = Building(co_index) url = bu_info.xpath("./@onclick")[0] bu.bu_id = re.search('dbh=(\d+)', url).group(1) bu.co_id = co_id bu.bu_num = bu_info.xpath("./td[@class='org']/text()")[0] bu.bu_all_house = bu_info.xpath("./td[3]/text()")[0] bu.size = bu_info.xpath("./td[2]/text()")[0] bu.insert_db() except Exception as e: log.error('楼栋信息错误', e)
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)