size_num_temp = room.room_desc.find('平方米')
if size_num_temp > 0:
size_content = room.room_desc[:size_num_temp]
size_num_temp2 = size_content.rfind('（')
if size_num_temp2 > 0:
room_size = size_content[size_num_temp2 + 1:]
room.size = room_size.strip().replace(' ', '')
except Exception, e:
logger.info('Cannot parse size of this room with error: ' + str(e))
except Exception, e:
logger.info('Cannot parse room size info with error: ' + str(e))
try:
if '不含早餐' not in room.room_desc and '早餐' in room.room_desc:
room.has_breakfast = 'Yes'
if '免费' in room.room_desc and '早餐' in room.room_desc:
room.is_breakfast_free = 'Yes'
except Exception, e:
logger.info('Cannot parse breakfast info with error: ' + str(e))
try:
if '加床' in room.room_desc:
room.is_extrabed = 'Yes'
if '免费加床' in room.room_desc or '加床免费' in room.room_desc:
room.is_extrabed_free = 'Yes'
except Exception, e:
logger.info('Cannor parse extrabed info with error: ' + str(e))
try:
try:
num = each_room[3].find('-')
if num > 0:
if len(each_room[3][:num]) < 20:
room.room_type = each_room[3][:num]
else:
room.room_type = 'NULL'
else:
if len(each_room[3]) < 20:
room.room_type = each_room[3]
else:
room.room_type = 'NULL'
if each_room[0] != u'nbsp;':
room.price = each_room[0]
room.has_breakfast = each_room[1]
room.room_desc = each_room[3]
if '免费WiFi' in room.room_desc:
room.is_wifi_free = 'Yes'
if '免费取消' in room.room_desc:
room.is_cancel_free = 'Yes'
room.currency = 'CNY'
room.source = 'youzhan'
room.source_hotelid = hotel_id
room.check_in = from_date
room.check_out = to_date
room_tuple = (room.hotel_name,room.city,room.source,room.source_hotelid,\
def youzhan_task_parser(taskcontent):
all_info = []
room_list = []
taskcontent = taskcontent.encode('utf-8').strip()
hotel_id = taskcontent.split('&')[0]
star = taskcontent.split('&')[2]
ipathid = taskcontent.split('&')[1]
city = taskcontent.split('&')[3]
country = taskcontent.split('&')[4]
#room_type = taskcontent.split('&')[3]
from_date_temp = taskcontent.split('&')[5]
from_date = from_date_temp[:4] + '-' + from_date_temp[4:6] + '-' \
+ from_date_temp[6:]
to_date_temp = datetime.datetime(int(from_date_temp[:4]), int(from_date_temp[4:6]), \
int(from_date_temp[6:]))
to_date = str(to_date_temp + datetime.timedelta(days = 1))[:10]
#获取代理
p = get_proxy()
#if p == "":
#logger.error("get proxy failed")
#return None
hotel = Hotel()
room = Room()
rating_url = get_rating_url(hotel_id)
rating_page = crawl_single_page(rating_url, proxy=p)
grade_str = grade_parser(rating_page)
if grade_str != '':
hotel.grade = grade_str[:-1]
else:
pass
#logger.error('Error: No grade_str found!')
map_url = get_map_url(hotel_id)
map_page = crawl_single_page(map_url, proxy=p)
#print map_page
map_info_list = staticmap_parser(map_page)
if map_info_list != []:
hotel.hotel_name = map_info_list[1]
if is_alphabet(hotel.hotel_name.decode('utf-8')) == True:
hotel.hotel_name_en = hotel.hotel_name
else:
hotel.hotel_name_en = 'NULL'
hotel.map_info = map_info_list[0]
else:
logger.error('youzhanHotel: Map info do not have hotel name and map_info')
return []
info_url = get_info_url(hotel_id,from_date,to_date)
info_page = crawl_single_page(info_url,proxy=p)
if info_page == '':
#invalid_proxy(p)
return []
info_list = info_parser(info_page)
if info_list != []:
hotel.country = country
hotel.city = city
hotel.address = info_list[1]
hotel_desc_temp = info_list[3].replace('<br/>','').replace(''','')
if hotel_desc_temp != '':
hotel.description = hotel_desc_temp
else:
hotel.description = 'NULL'
hotel.service = info_list[4]
if '停车场' in hotel.service:
hotel.has_parking = 'Yes'
if '无线网络' in hotel.service or 'wifi' in hotel.service:
hotel.has_wifi = 'Yes'
else:
return []
hotel.source = 'youzhan'
hotel.source_id = hotel_id
hotel.star = star
price_url = get_price_url(hotel_id,ipathid,from_date,to_date)
price_page = crawl_single_page(price_url,proxy=p)
price_list = price_parser(price_page,hotel_id)
#print '********'
#print price_list
if price_list != []:
for each_room in price_list:
if len(each_room) > 3:
room.city = city
room.occupancy = 2
room.hotel_name = hotel.hotel_name
#print '******'
#print each_room
room.room_desc = each_room[3]
room.real_source = each_room[2]
num = each_room[3].find('-')
if num > 0:
if len(each_room[3][:num]) < 20:
room.room_type = each_room[3][:num]
else:
room.room_type = 'NULL'
else:
if len(each_room[3]) < 20:
room.room_type = each_room[3]
else:
room.room_type = 'NULL'
if each_room[0] != u'nbsp;':
room.price = each_room[0]
room.has_breakfast = each_room[1]
room.room_desc = each_room[3]
if '免费WiFi' in room.room_desc:
hotel.is_wifi_free = 'Yes'
if '免费取消' in room.room_desc:
hotel.is_cancel_free = 'Yes'
room.currency = 'CNY'
room.source = 'youzhan'
room.source_hotelid = hotel_id
room.check_in = from_date
room.check_out = to_date
room_tuple = (room.hotel_name,room.city,room.source,room.source_hotelid,\
room.source_roomid,room.real_source,room.room_type,room.occupancy,\
room.bed_type,room.size,room.floor,room.check_in,room.check_out,room.price,\
room.tax,room.currency,room.is_extrabed,room.is_extrabed_free,room.has_breakfast,\
room.is_breakfast_free,room.is_cancel_free,room.room_desc)
room_list.append(room_tuple)
hotel_tuple = (hotel.hotel_name, hotel.hotel_name_en,hotel.source,hotel.source_id,hotel.brand_name,\
hotel.map_info,hotel.address,hotel.city,hotel.country,hotel.postal_code, \
hotel.star,hotel.grade,hotel.has_wifi,hotel.is_wifi_free,hotel.has_parking,\
hotel.is_parking_free,hotel.service,hotel.img_items,hotel.description)
hotel_list = []
hotel_list.append(hotel_tuple)
all_info.append(hotel_list)
all_info.append(room_list)
return all_info
size_num_temp = room.room_desc.find('平方米')
if size_num_temp > 0:
size_content = room.room_desc[:size_num_temp]
size_num_temp2 = size_content.rfind('（')
if size_num_temp2 > 0:
room_size = size_content[size_num_temp2 + 1:]
room.size = room_size.strip().replace(' ','')
except Exception, e:
logger.info('Cannot parse size of this room with error: ' + str(e))
except Exception,e:
logger.info('Cannot parse room size info with error: ' + str(e))
try:
if '不含早餐' not in room.room_desc and '早餐' in room.room_desc:
room.has_breakfast = 'Yes'
if '免费' in room.room_desc and '早餐' in room.room_desc:
room.is_breakfast_free = 'Yes'
except Exception,e:
logger.info('Cannot parse breakfast info with error: ' + str(e))
try:
if '加床' in room.room_desc:
room.is_extrabed = 'Yes'
if '免费加床' in room.room_desc or '加床免费' in room.room_desc:
room.is_extrabed_free = 'Yes'
except Exception, e:
logger.info('Cannor parse extrabed info with error: ' + str(e))
try:
try:
def youzhan_task_parser(taskcontent):
all_info = []
room_list = []
taskcontent = taskcontent.encode('utf-8').strip()
hotel_id = taskcontent.split('&')[0]
star = taskcontent.split('&')[2]
ipathid = taskcontent.split('&')[1]
city = taskcontent.split('&')[3]
country = taskcontent.split('&')[4]
#room_type = taskcontent.split('&')[3]
from_date_temp = taskcontent.split('&')[5]
from_date = from_date_temp[:4] + '-' + from_date_temp[4:6] + '-' \
+ from_date_temp[6:]
to_date_temp = datetime.datetime(int(from_date_temp[:4]), int(from_date_temp[4:6]), \
int(from_date_temp[6:]))
to_date = str(to_date_temp + datetime.timedelta(days=1))[:10]
#获取代理
p = get_proxy()
#if p == "":
#logger.error("get proxy failed")
#return None
hotel = Hotel()
room = Room()
rating_url = get_rating_url(hotel_id)
rating_page = crawl_single_page(rating_url, proxy=p)
grade_str = grade_parser(rating_page)
if grade_str != '':
hotel.grade = grade_str[:-1]
else:
pass
#logger.error('Error: No grade_str found!')
map_url = get_map_url(hotel_id)
map_page = crawl_single_page(map_url, proxy=p)
#print map_page
map_info_list = staticmap_parser(map_page)
if map_info_list != []:
hotel.hotel_name = map_info_list[1]
if is_alphabet(hotel.hotel_name.decode('utf-8')) == True:
hotel.hotel_name_en = hotel.hotel_name
else:
hotel.hotel_name_en = 'NULL'
hotel.map_info = map_info_list[0]
else:
logger.error(
'youzhanHotel: Map info do not have hotel name and map_info')
return []
info_url = get_info_url(hotel_id, from_date, to_date)
info_page = crawl_single_page(info_url, proxy=p)
if info_page == '':
#invalid_proxy(p)
return []
info_list = info_parser(info_page)
if info_list != []:
hotel.country = country
hotel.city = city
hotel.address = info_list[1]
hotel_desc_temp = info_list[3].replace('<br/>',
'').replace(''', '')
if hotel_desc_temp != '':
hotel.description = hotel_desc_temp
else:
hotel.description = 'NULL'
hotel.service = info_list[4]
if '停车场' in hotel.service:
hotel.has_parking = 'Yes'
if '无线网络' in hotel.service or 'wifi' in hotel.service:
hotel.has_wifi = 'Yes'
else:
return []
hotel.source = 'youzhan'
hotel.source_id = hotel_id
hotel.star = star
price_url = get_price_url(hotel_id, ipathid, from_date, to_date)
price_page = crawl_single_page(price_url, proxy=p)
price_list = price_parser(price_page, hotel_id)
#print '********'
#print price_list
if price_list != []:
for each_room in price_list:
if len(each_room) > 3:
room.city = city
room.occupancy = 2
room.hotel_name = hotel.hotel_name
#print '******'
#print each_room
room.room_desc = each_room[3]
room.real_source = each_room[2]
num = each_room[3].find('-')
if num > 0:
if len(each_room[3][:num]) < 20:
room.room_type = each_room[3][:num]
else:
room.room_type = 'NULL'
else:
if len(each_room[3]) < 20:
room.room_type = each_room[3]
else:
room.room_type = 'NULL'
if each_room[0] != u'nbsp;':
room.price = each_room[0]
room.has_breakfast = each_room[1]
room.room_desc = each_room[3]
if '免费WiFi' in room.room_desc:
hotel.is_wifi_free = 'Yes'
if '免费取消' in room.room_desc:
hotel.is_cancel_free = 'Yes'
room.currency = 'CNY'
room.source = 'youzhan'
room.source_hotelid = hotel_id
room.check_in = from_date
room.check_out = to_date
room_tuple = (room.hotel_name,room.city,room.source,room.source_hotelid,\
room.source_roomid,room.real_source,room.room_type,room.occupancy,\
room.bed_type,room.size,room.floor,room.check_in,room.check_out,room.price,\
room.tax,room.currency,room.is_extrabed,room.is_extrabed_free,room.has_breakfast,\
room.is_breakfast_free,room.is_cancel_free,room.room_desc)
room_list.append(room_tuple)
hotel_tuple = (hotel.hotel_name, hotel.hotel_name_en,hotel.source,hotel.source_id,hotel.brand_name,\
hotel.map_info,hotel.address,hotel.city,hotel.country,hotel.postal_code, \
hotel.star,hotel.grade,hotel.has_wifi,hotel.is_wifi_free,hotel.has_parking,\
hotel.is_parking_free,hotel.service,hotel.img_items,hotel.description)
hotel_list = []
hotel_list.append(hotel_tuple)
all_info.append(hotel_list)
all_info.append(room_list)
return all_info