def get_comm_info(self, comm_list, all_page_url):
for i in comm_list:
try:
comm = Comm('中安房')
comm.city = '合肥'
comm.district_name = re.search('zaf-nowrap.*?>(.*?)<', i, re.S | re.M).group(1).strip()
trade_date = re.search('zaf-fblue">(.*?)<', i, re.S | re.M).group(1).strip()
if trade_date:
t = time.strptime(trade_date, "%Y-%m-%d")
y = t.tm_year
m = t.tm_mon
d = t.tm_mday
comm.trade_date = datetime.datetime(y, m, d)
total_price = re.search('list-right-data.*?<span.*?>(.*?)<', i, re.S | re.M).group(1).strip()
comm.total_price = int(re.search('(\d+)', total_price, re.S | re.M).group(1)) * 10000
info = re.search('list-details-area.*?<span.*?>(.*?)<', i, re.S | re.M).group(1).strip()
area = info.split(' ')[0].replace('㎡', '')
if area:
area = float(area)
comm.area = round(area, 2)
try:
room_type = info.split(' ')[1]
except Exception as e:
room_type = None
try:
comm.room = int(re.search('(\d)室', room_type, re.S | re.M).group(1))
except Exception as e:
comm.room = 0
try:
comm.hall = int(re.search('(\d)厅', room_type, re.S | re.M).group(1))
except Exception as e:
comm.hall = None
try:
comm.toilet = int(re.search('(\d)卫', room_type, re.S | re.M).group(1))
except Exception as e:
comm.toilet = None
try:
avg_price = info.split(' ')[2]
comm.avg_price = int(re.search('(\d+)', avg_price, re.S | re.M).group(1))
except Exception as e:
comm.avg_price = None
info_2 = re.search('list-details-area.*?<span.*?<span>(.*?)<', i, re.S | re.M).group(1).strip()
comm.direction = info_2.split(' ')[0]
try:
comm.fitment = info_2.split(' ')[1]
except Exception as e:
comm.fitment = None
info_3 = re.search('list-details-address1.*?<span>(.*?)<', i, re.S | re.M).group(1).strip()
comm.region = info_3.split(' ')[0].strip()
comm.insert_db()
except Exception as e:
log.error('解析错误，source={},url="{}",e="{}"'.format('中安房', all_page_url, e))
def into_mongo(coll):
com = Comm('澜斯')
results = coll.find(no_cursor_timeout=True)
for result in results:
# 这个地方写一个try是因为我再测试的时候发现有的木有fj_city
try:
com.city = result['fj_city'] # 城市
com.region = result['fj_region'] # 区域
except Exception as e:
log.error('城市或者区域没有')
com.m_date = result['updatedate'] # 更新日期
com.create_date = datetime.datetime.now() # 创建时间
com.fitment = result['newdiskdecoration'] # 装修
com.floor = result['flevel'] # 所在楼层
# try是因为在插入数据库中这几个如果不符合，就不会插入
try:
com.district_name = result['fj_name'] # 小区名称
com.avg_price = result['unitprice'] # 单价
com.total_price = result['usd'] # 总价
com.area = result['acreage'] # 面积=建筑面积
t = time.strptime(result['signingdate'].split('T')[0], "%Y-%m-%d")
y = t.tm_year
m = t.tm_mon
d = t.tm_mday
com.trade_date = datetime.datetime(y, m, d)
except Exception as e:
log.error(e)
# 这一部分我写了正则从地址中匹配单元号和室号，如果组长感觉不对，，直接注释掉就好
houseaddress = result['houseaddress']
try:
res = re.search('(\d+)号(\d+)', houseaddress)
com.unit_num = res.group(1) # 单元号
com.room_num = res.group(2) # 室号
except Exception as e:
print('无法匹配大盘单元号和室号，houseaddress={}'.find(houseaddress))
# 以下数据库确定无法匹配,写上是为了让您看看
# com.direction = None # 朝向
# com.room = None # 室数
# com.hall = None # 厅数
# com.toilet = None # 卫数
# com.height = None # 总楼层
# com.house_num = None # 楼栋号
# 执行插入操作
com.insert_db()
def room(self, co_list, city_name):
for co in co_list:
try:
co_name = co.xpath("./div[1]/a/text()")[0]
co_url = "http:" + co.xpath("./div[1]/a/@href")[0]
region = co.xpath("./div[3]/span[1]/a[1]/text()")[0]
addr = co.xpath("./div[3]/span[3]/@title")[0]
detail = requests.get(co_url, headers=self.headers)
html = etree.HTML(detail.text)
room_url = "http:" + html.xpath("//div[@class='tab-toolbar pr']//li/a/@href")[-1]
page_index = requests.get(room_url, headers=self.headers)
except:
continue
if re.search('共(\d+)页', page_index.text):
page_num = re.search('共(\d+)页', page_index.text).group(1)
else:
log.info('小区无相关数据')
continue
for i in range(1, int(page_num) + 1):
url = re.sub('#.*', 'n', room_url) + str(i)
while True:
try:
res = requests.get(url, headers=self.headers)
break
except:
continue
con = res.text
room_html = etree.HTML(con)
room_list = room_html.xpath("//div[@class='right-information']")
for m in room_list:
try:
room = Comm(source)
room.district_name = co_name
room.city = city_name
room.region = region
room_type = m.xpath("./h3/span[2]/text()")[0]
try:
room.room = int(re.search('(\d)室', room_type, re.S | re.M).group(1))
except Exception as e:
room.room = None
try:
room.hall = int(re.search('(\d)厅', room_type, re.S | re.M).group(1))
except Exception as e:
room.hall = None
size = m.xpath("./h3/span[3]/text()")[0]
area = size.replace('平米', '')
if area:
area = float(area)
room.area = round(area, 2)
total_price = m.xpath(".//div[@class='price fs14 ']/em/text()")[0]
room.total_price = int(re.search('(\d+)', total_price, re.S | re.M).group(1))*10000
avg_price = m.xpath(".//div[@class='size fs14']/text()")[0]
room.avg_price = int(re.search('(\d+)', avg_price, re.S | re.M).group(1))
try:
room.fitment = m.xpath(".//div[@class='t1 fs14']/text()[3]")[0]
room.direction = m.xpath(".//div[@class='t1 fs14']/text()[2]")[0]
# room.use = m.xpath(".//div[@class='t1 fs14']/text()[1]")[0]
except:
room.fitment = None
room.direction = None
# room.use = None
floor_info = m.xpath(".//div[@class='fs14']/text()[1]")[0]
try:
floor = re.search('(.*?)/', floor_info).group(1)
room.floor = int(re.search('\d+',floor).group(0))
except Exception as e:
room.floor = None
try:
room.height = int(re.search('.*?/(\d+)层', floor_info).group(1))
except:
room.height = None
trade_date = m.xpath(".//div[@class='date']/text()")[0]
if trade_date:
t = time.strptime(trade_date, "%Y-%m-%d")
y = t.tm_year
m = t.tm_mon
d = t.tm_mday
room.trade_date = datetime.datetime(y, m, d)
room.insert_db()
except Exception as e:
log.error('房屋信息提取失败{}'.format(e))
def get_comm_detail(self, comm_url, region, city):
comm = Comm('购房网')
comm.url = comm_url
comm.region = region.strip()
comm.city = city
try:
response = requests.get(url=comm_url,
headers=self.headers,
proxies=next(p))
except Exception as e:
log.error('请求错误,source="{}"，url="{}",e="{}"'.format(
'购房网', comm_url, e))
return
html = response.text
comm.district_name = re.search('title fl.*?<h1>(.*?)</h1>', html,
re.S | re.M).group(1).strip()
comm_info_html = re.search('<ul class="lscjlist">.*?</ul>', html,
re.S | re.M).group()
comm_info_list = re.findall('<li>(.*?)</li>', comm_info_html,
re.S | re.M)
if not comm_info_list:
log.info('source={}, 此小区没有数据，url="{}"'.format('购房网', comm_url))
for i in comm_info_list:
trade_date = re.search('<span>(.*?)</span>', i,
re.S | re.M).group(1).strip()
if trade_date:
t = time.strptime(trade_date, "%Y-%m-%d")
y = t.tm_year
m = t.tm_mon
d = t.tm_mday
comm.trade_date = datetime.datetime(y, m, d)
room_type = re.search('<span>.*?<span>(.*?)</span>', i,
re.S | re.M).group(1).strip()
try:
comm.room = int(
re.search('(\d)室', room_type, re.S | re.M).group(1))
comm.hall = int(
re.search('(\d)厅', room_type, re.S | re.M).group(1))
except Exception as e:
comm.room = None
comm.hall = None
area = re.search('<span>.*?<span>.*?<span>(.*?)</span>', i,
re.S | re.M).group(1).strip().replace('㎡',
'').replace(
'平', '')
if area:
area = float(area)
comm.area = round(area, 2)
try:
height = re.search(
'<span>.*?<span>.*?<span>.*?<span>.*?/(.*?)</span>', i,
re.S | re.M).group(1).strip()
comm.height = int(re.search('(\d+)', height).group(1))
except Exception as e:
comm.height = None
comm.fitment = re.search(
'<span>.*?<span>.*?<span>.*?<span>.*?<span>(.*?)</span>', i,
re.S | re.M).group(1).strip()
comm.direction = re.search(
'<span>.*?<span>.*?<span>.*?<span>.*?<span>.*?<span>(.*?)</span>',
i, re.S | re.M).group(1).strip()
avg_price = re.search(
'<span>.*?<span>.*?<span>.*?<span>.*?<span>.*?<span>.*?<span>(.*?)</span>',
i, re.S | re.M).group(1)
comm.avg_price = int(
re.search('(\d+)', avg_price, re.S | re.M).group(1))
total_price = re.search(
'<span>.*?<span>.*?<span>.*?<span>.*?<span>.*?<span>.*?<span>.*?<span.*?>(.*?)</span>',
i, re.S | re.M).group(1)
comm.total_price = int(
re.search('(\d+)', total_price, re.S | re.M).group(1)) * 10000
comm.insert_db()
def get_city_info(self, city_dict):
for city in city_dict:
city_url = city_dict[city] + 'chengjiao/'
try:
response = requests.get(city_url, headers=self.headers)
html = response.text
area_html = re.search('data-role="ershoufang".*?地铁', html,
re.S | re.M).group()
area_list_str = re.findall('<a.*?</a>', area_html, re.S | re.M)
for area_i in area_list_str:
if 'ershoufang' in area_i:
continue
area_url = re.search('href="(.*?)"', area_i,
re.S | re.M).group(1)
area = re.search('<a.*?>(.*?)<', area_i,
re.S | re.M).group(1)
for i in range(1, 101):
city_url_ = city_url.replace(
'/chengjiao/', '') + area_url + 'pg' + str(i)
try:
result = requests.get(city_url_,
headers=self.headers)
content = result.text
comm_str_list = re.findall(
'class="info".*?</div></div></li>', content,
re.S | re.M)
for i in comm_str_list:
comm = Comm('链家在线')
comm.region = area.strip()
comm.city = city.strip()
comm.district_name = re.search(
'target="_blank">(.*?)<', i,
re.S | re.M).group(1).strip()
comm.direction = re.search(
'class="houseIcon"></span>(.*?) \|', i,
re.S | re.M).group(1).strip()
try:
comm.fitment = re.search(
'class="houseIcon"></span>.*? \|(.*?)\| ',
i, re.S | re.M).group(1).strip()
except Exception as e:
comm.fitment = None
try:
height = re.search(
'class="positionIcon"></span>.*?\((.*?)\)',
i, re.S | re.M).group(1).strip()
comm.height = int(
re.search('(\d+)', height,
re.S | re.M).group(1))
except Exception as e:
comm.height = None
total_price = re.search(
"class='number'>(.*?)<", i,
re.S | re.M).group(1).strip()
if "*" in total_price:
continue
comm.total_price = int(
re.search('(\d+)', total_price,
re.S | re.M).group(1)) * 10000
room_type = re.search(
'arget="_blank">.*? (.*?) ', i,
re.S | re.M).group(1).strip()
try:
comm.room = int(
re.search('(\d)室', room_type,
re.S | re.M).group(1))
except Exception as e:
comm.room = 0
try:
comm.hall = int(
re.search('(\d)厅', room_type,
re.S | re.M).group(1))
except Exception as e:
comm.hall = None
area_ = re.search(
'target="_blank">.*? .*? (.*?平米)', i,
re.S | re.M).group(1).strip()
if area_:
area_ = area_.replace('㎡', '').replace(
'平米', '')
try:
area_ = float(area_)
comm.area = round(area_, 2)
except Exception as e:
comm.area = None
trade_date = re.search(
'dealDate">(.*?)<', i,
re.S | re.M).group(1).strip()
if trade_date:
t = time.strptime(trade_date, "%Y.%m.%d")
y = t.tm_year
m = t.tm_mon
d = t.tm_mday
comm.trade_date = datetime.datetime(
y, m, d)
try:
comm.avg_price = int(i['total_price'] /
i['area'])
except Exception as e:
comm.avg_price = None
comm.insert_db()
except Exception as e:
log.error(
'解析错误，source="{}",html="{}",e="{}"'.format(
'链家在线', html, e))
except Exception as e:
log.error('请求错误，source="{}",url="{}",e="{}"'.format(
'链家在线', city_url, e))