return rooms
other_info = ''
try:
other_info = other_info_pat.findall(each_room_info)[0]
if '免费取消' in other_info:
room.is_cancel_free = 'Yes'
else:
room.is_cancel_free = 'No'
except:
room.is_cancel_free = 'No'
try:
room_desc_temp = room_desc_pat.findall(each_room_info)[0]
room_desc = fan_to_jian(room_desc_temp)
room.room_desc = room_desc
if '含早餐' in other_info or 'Breakfast' in room_desc:
room.has_breakfast = 'Yes'
room.is_breakfast_free = 'Yes'
else:
logger.info('Cannot parse breakfast info!')
room.has_breakfast = 'No'
room.is_breakfast_free = 'No'
type_num01 = room_desc.find('(')
type_num02 = room_desc.find('Room')
if type_num01 > 0:
room.room_type = room_desc[:type_num01]
elif type_num02 > 0:
room.room_type = room_desc[:type_num02 + 4]
room.room_type = room_type
except Exception, e:
logger.error('Cannot paese room type of this hotel!' + str(e))
try:
room.source_roomid = each_hotel['RoomId']
except Exception, e:
logger.info('Cannot parse this room id with error: ' + str(e))
try:
room_desc = ''
if each_hotel['rp'] != 'none':
room_desc += each_hotel['rp']
if each_hotel['roomStaticDescription'] != '':
room_desc += each_hotel['roomStaticDescription']
room.room_desc = re.sub('<.*?>', '', room_desc)
except Exception, e:
logger.info('Cannot parse this room description with error: ' +
str(e))
try:
if '可以免费取消' in each_hotel['cancellationPolicyNew'] or '可以免费取消' \
in each_hotel['cancellationPolicy']:
room.is_cancel_free = 'Yes'
except Exception, e:
logger.info('Cannot parse cancel info with error: ' + str(e))
try:
bed_type = each_hotel['bedTypes'][0]['description']
room.bed_type = str(bed_type)
except Exception, e:
room.room_type = room_type
except Exception,e:
logger.error('Cannot paese room type of this hotel!' + str(e))
try:
room.source_roomid = each_hotel['RoomId']
except Exception, e:
logger.info('Cannot parse this room id with error: ' + str(e))
try:
room_desc = ''
if each_hotel['rp'] != 'none':
room_desc += each_hotel['rp']
if each_hotel['roomStaticDescription'] != '':
room_desc += each_hotel['roomStaticDescription']
room.room_desc = re.sub('<.*?>','', room_desc)
except Exception, e:
logger.info('Cannot parse this room description with error: ' + str(e))
try:
if '可以免费取消' in each_hotel['cancellationPolicyNew'] or '可以免费取消' \
in each_hotel['cancellationPolicy']:
room.is_cancel_free = 'Yes'
except Exception,e:
logger.info('Cannot parse cancel info with error: ' + str(e))
try:
bed_type = each_hotel['bedTypes'][0]['description']
room.bed_type = str(bed_type)
except Exception, e:
logger.info('Cannot parse bed type of this room with error: ' + str(e))
if price_page == None or price_page == '':
invalid_proxy(proxy=p, source='youzhanHotel')
result['error'] = PROXY_INVALID
return result
#print price_page
price_list = price_parser(price_page,hotel_id)
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;':
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
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
return rooms
other_info = ''
try:
other_info = other_info_pat.findall(each_room_info)[0]
if '免费取消' in other_info:
room.is_cancel_free = 'Yes'
else:
room.is_cancel_free = 'No'
except:
room.is_cancel_free = 'No'
try:
room_desc_temp = room_desc_pat.findall(each_room_info)[0]
room_desc = fan_to_jian(room_desc_temp)
room.room_desc = room_desc
if '含早餐' in other_info or 'Breakfast' in room_desc:
room.has_breakfast = 'Yes'
room.is_breakfast_free = 'Yes'
else:
logger.info('Cannot parse breakfast info!')
room.has_breakfast = 'No'
room.is_breakfast_free = 'No'
type_num01 = room_desc.find('(')
type_num02 = room_desc.find('Room')
if type_num01 > 0:
room.room_type = room_desc[:type_num01]
elif type_num02 > 0:
room.room_type = room_desc[:type_num02 + 4]
_ERROR('ctripHotel::parseRoom',
['Parse this room failed',
str(e)])
return room_list
try:
#print 'okkkkkkkkkkkkkkkkkkkkkkkkkkkkkk'
room.room_type = room_type_pat.findall(each_room_content)[0]
except:
#print 'faillllllllllllllllleddddddddddddddddddd'
try:
room_type_info = room_type_pat2.findall(
each_room_content)[0]
#print '^^--^^'
room.room_type = room_type_info[0]
room.room_desc = room_type_info[1]
except Exception, e:
_ERROR('ctripHotel::parseRoom',
['Parse room type failed!',
str(e)])
return room_list
each_night_content_list = rate_pat.findall(each_room_content)
if len(each_night_content_list) == 0:
_ERROR('ctripHotel::parseRoom',
['Parse this day content failed'])
continue
for each_night_content in each_night_content_list:
#print each_night_content
try:
for each_type_room_content in room_type_list:
room = Room()
room.hotel_name = hotel_name_real
room.city = city_name_zh
room.source = 'hotels'
room.source_hotelid = hotel_id
room.real_source = 'hotels'
room.currency = 'CNY'
room.check_in = check_in
room.check_out = check_out
try:
room_desc_temp = room_desc_pat.findall(each_type_room_content)[0].strip()
room_desc_temp = '<' + room_desc_temp
room.room_desc = re.sub('<.*?>','',room_desc_temp).replace('\n','').replace(' ',',')
room.room_desc = room.room_desc.replace(',,','').replace(' ','').replace('。,','。')
except Exception, e:
#logger.info('haodingHotel::Cannot parse room desc of this type room')
#logger.info('haodingHotel::' + str(e))
pass
#print room.room_desc
try:
room_type_temp = room_type_pat.findall(each_type_room_content)[0]
bed_type_temp = bed_type_pat.findall(each_type_room_content)[0]
room_type_temp = '<' + room_type_temp
room_type = re.sub('<.*?.>','',room_type_temp).strip()
bed_type = re.sub('<.*?>','',bed_type_temp).strip()
#print bed_type
#if '加床' in bed_type: