def __init__(self, hq_hotel_id_set, logger, ota_hotels=None):
self.hq_hotel_id_set = hq_hotel_id_set
self.logger = logger
self.download_suppliers = Tools.get_dls_suppliers()
self.ota_hotels = ota_hotels
self.hotel_city_map = dict()
"""记录ota各个端的酒店信息"""
def pick_interested_hotels(self, limit=2000):
"""
选取用户感兴趣好巧酒店, 30天以内穿透量最多的limit家好巧酒店
:param limit: 吐出酒店数量上限
:return: set
"""
try:
interested_hotels_info = dict()
download_suppliers = Tools.get_dls_suppliers()
if not download_suppliers:
return set()
day = datetime.datetime.today() - datetime.timedelta(days=30)
for supplier_id in download_suppliers:
try:
sql = "SELECT hq_hotel_id, price_num FROM summary%s WHERE stat_date >= '%s' " \
"ORDER BY price_num DESC limit %s;" % (supplier_id, day, limit)
ret = pymysql_query(sql, Const.summary)
if ret == DB_EXCEPTION:
self.logger.error(traceback.format_exc())
continue
for row in ret:
hq_hotel_id = row[Const.hq_hotel_id]
price_num = row[Const.price_num]
if price_num == 0:
break
if hq_hotel_id not in interested_hotels_info:
interested_hotels_info[hq_hotel_id] = price_num
else:
interested_hotels_info[hq_hotel_id] += price_num
except:
self.logger.error(traceback.format_exc())
# 根据最大请求数降序排序
sorted_hotels = sorted(interested_hotels_info.items(),
lambda x, y: cmp(x[1], y[1]),
reverse=True)
interested_hotels = set()
for item in sorted_hotels[:limit]:
hq_hotel_id, num = item
interested_hotels.add(hq_hotel_id)
return interested_hotels
except:
self.logger.error(traceback.format_exc())
return set()