def _parse(self): # pylint: disable=R0912,R0915
result = self._order
stations = self._etree.xpath(plf_xp.STATION)
if stations:
result['stations'] = [remove_space(i) for i in stations]
else:
self.logger.error('stations is empty %s', self._message_id)
flight_date = strip_str(take_first(self._etree, plf_xp.DATE))
if flight_date:
flight_date = flight_date.split(' - ')
if len(flight_date) == 2:
result['depart_date'], result['arrive_date'] = flight_date
else:
self.logger.error('flight_date is empty %s', self._message_id)
flight_time = strip_str(take_first(self._etree, plf_xp.TIME))
if flight_time:
flight_time = flight_time.split(' - ')
if len(flight_time) == 2:
result['depart_time'], result['arrive_time'] = flight_time
else:
self.logger.error('flight_time is empty %s', self._message_id)
de_date, ar_date = result.get('depart_date'), result.get('arrive_date')
de_time, ar_time = result.get('depart_time'), result.get('arrive_time')
de_datetime = de_date and de_time and ' '.join([de_date, de_time])
ar_datetime = ar_date and ar_time and ' '.join([ar_date, ar_time])
depart, arrive = unpack(stations)
tz_depart, tz_arrive = to_timezone(depart), to_timezone(arrive)
if tz_arrive and tz_depart and de_datetime and ar_datetime:
depart_datetime_formatted = DateTime(
de_datetime,
'MMM DD YYYY hh:mm A').tz_to_datetime(tz_depart) # D
if depart_datetime_formatted:
result['depart_datetime_formatted'] = depart_datetime_formatted
else:
self.logger.error('depart_datetime_formatted is empty %s',
self._message_id)
arrive_datetime_formatted = DateTime(
ar_datetime, 'MMM DD YYYY hh:mm A').tz_to_datetime(tz_arrive)
if arrive_datetime_formatted:
result['arrive_datetime_formatted'] = arrive_datetime_formatted
else:
self.logger.error('arrive_datetime_formatted is empty %s',
self._message_id)
else:
self.logger.error('datetime formatted is empty %s',
self._message_id)
plane_model = strip_str(take_last(self._etree, plf_xp.TIME))
if plane_model:
result['plane_model'] = plane_model
else:
self.logger.warning('plane_model is empty %s', self._message_id)
mileage = strip_str(take_first(self._etree, plf_xp.MILES))
if mileage:
result['flight_duration'] = mileage
else:
self.logger.warning('mileage is empty %s', mileage)
confirm_num = take_last(self._etree, plf_xp.CONFIRM_NUM)
if confirm_num:
confirm_num = confirm_num.split(':')[-1].strip()
result['confirm_code'] = confirm_num
else:
self.logger.error('confirm_number is empty %s', self._message_id)
ticket = take_first(self._etree, plf_xp.TICKET)
if ticket:
ticket = ticket.split(':')[-1].strip()
result['ticket_number'] = ticket
else:
self.logger.error('ticket_number is empty %s', self._message_id)
depart_station = strip_str(take_first(self._etree, plf_xp.DEPART))
if depart_station:
result['depart_city'] = depart_station.split(', ')[-1]
result['depart_station'] = depart_station
else:
self.logger.error('depart_station is empty %s', self._message_id)
arrive_station = strip_str(take_first(self._etree, plf_xp.ARRIVE))
if arrive_station:
result['arrive_city'] = arrive_station.split(', ')[-1]
result['arrive_station'] = arrive_station
else:
self.logger.error('arrive_station is empty %s', self._message_id)
tel = take_first(self._etree, plf_xp.PHONE_PRICELINE)
if tel:
result['telephone'] = tel.split(':')[-1].strip()
else:
self.logger.error('service telephone is empty %s',
self._message_id)
help_keys = self._etree.xpath(plf_xp.HELP_KEYS)
help_values = self._etree.xpath(plf_xp.HELP_VALUES)
contact = filter_dict_value(to_dict(help_keys, help_values))
if contact:
result['contact_information'] = contact
else:
self.logger.warning('contact_information is empty %s',
self._message_id)
user_tel = take_first(self._etree, plf_xp.PHONE_USER)
if user_tel:
result['guest_telephone'] = user_tel.split(':')[-1].strip()
else:
self.logger.warning('guest_telephone is empty %s',
self._message_id)
price_keys = strip_list(self._etree.xpath(plf_xp.PRICE_KEYS))
price_values = strip_list(self._etree.xpath(plf_xp.PRICE_VALUES))
name_dict = {
'Billing Name': 'guest_name',
'Ticket Cost': 'ticket_cost',
'Taxes & Fees': 'taxes_fee',
'Tickets': 'number_of_tickets',
'Total Price': 'total_cost',
'Bonus': 'bonus'
}
if price_keys and price_values:
for i, j in zip(price_keys, price_values):
if i in name_dict.keys():
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('price_details is empty %s', self._message_id)
notice = self._etree.xpath(plf_xp.NOTICE)
if notice:
notice = remove_tags(etree.tostring(notice[0])) \
.replace(' \n', ' ')
result['notice'] = strip_list(notice.split('\n\n'))
else:
self.logger.error('notice is empty %s', self._message_id)
related_link_text = self._etree.xpath(plf_xp.RELATED_LINK_TEXT)
related_link = self._etree.xpath(plf_xp.RELATED_LINK)
related_links = filter_dict_value(
to_dict(related_link_text, related_link))
if related_links:
result['related_links'] = related_links
else:
self.logger.warning('related_links is empty %s', self._message_id)
return result
def _parse(self):
result = self._order
confirm_code = take_first(self._etree, delta_xp.CONFIRM_CODE)
if not confirm_code:
raise MissingColumnError('confirm_code', self._message_id)
confirm_code = remove_tags(etree.tostring(confirm_code))
result['confirm_code'] = ''.join(confirm_code)
result.set('modify_link', take_first(self._etree,
delta_xp.CHANGE_LINK))
# TODO(LensHo): 起飞降落只有一个日期
flight_date = strip_list(self._etree.xpath(delta_xp.FLIGHT_DATE))
if not flight_date:
raise MissingColumnError('flight_date', self._message_id)
result['depart_date'] = result['arrive_date'] = flight_date[0]
if len(flight_date) >= 2:
result['trans_depart_date'] = flight_date[1]
result['trans_arrive_date'] = flight_date[1]
flight_num = strip_list(self._etree.xpath(delta_xp.FLIGHT_NUM))
if not flight_num:
raise MissingColumnError('flight_number', self._message_id)
result['flight_number'] = flight_num[0]
if len(flight_num) >= 2:
result['trans_flight_number'] = flight_num[1]
cabin = strip_list(self._etree.xpath(delta_xp.CABIN))
if cabin:
result['class'] = cabin[0]
if len(cabin) >= 2:
result['trans_class'] = cabin[1]
else:
self.logger.error('class is empty %s', self._message_id)
depart_station = strip_list(self._etree.xpath(delta_xp.DEPART_STATION))
if not depart_station:
raise MissingColumnError('depart_station', self._message_id)
result['depart_city'] = depart_station[0]
if len(depart_station) >= 2:
result['trans_depart_city'] = depart_station[1]
depart_time = strip_list(self._etree.xpath(delta_xp.DEPART_TIME))
if not depart_time:
raise MissingColumnError('depart_time', self._message_id)
result['depart_time'] = depart_time[0]
if len(depart_time) >= 2:
result['trans_depart_time'] = depart_time[1]
arrive_time = strip_list(self._etree.xpath(delta_xp.ARRIVE_TIME))
if not arrive_time:
raise MissingColumnError('arrive_time', self._message_id)
result['arrive_time'] = arrive_time[0]
if len(arrive_time) >= 2:
result['trans_arrive_time'] = arrive_time[1]
elif len(arrive_time) > 2 and 'm' in arrive_time[2]:
self.logger.error('换乘超过1次 %s', self._message_id)
arrive_station = strip_list(self._etree.xpath(delta_xp.ARRIVE_STATION))
if not arrive_station:
raise MissingColumnError('arrive_station', self._message_id)
result['arrive_city'] = arrive_station[0]
if len(arrive_station) >= 2:
result['trans_arrive_city'] = arrive_station[1]
restricted_title = strip_str(
take_first(self._etree, delta_xp.RESTRICTED_TITLE))
restricted_text = take_first(self._etree, delta_xp.RESTRICTED_TEXT)
if len(restricted_text) and restricted_title:
text = strip_str(remove_tags(etree.tostring(restricted_text)))
if 'RESTRICTED HAZARDOUS ITEMS' in restricted_title:
result['restricted_hazardous_items'] = text
else:
# 有可能是其他条目
self.logger.warning('%s is %s %s', restricted_title, text,
self._message_id)
else:
self.logger.error('restricted_hazardous_items is empty %s',
self._message_id)
result['guest_name'] = strip_str(
take_first(self._etree, delta_xp.GUEST_NAME))
if not result['guest_name']:
result.pop('guest_name')
self.logger.error('guest_name is empty %s', self._message_id)
seat = strip_list(self._etree.xpath(delta_xp.SEAT))
if seat:
result['seat'] = seat[0]
if len(seat) >= 2:
result['trans_seat'] = seat[1]
else:
self.logger.error('seat is empty %s', self._message_id)
result['ticket_number'] = strip_str(
take_last(self._etree, delta_xp.TICKET_NUM))
if not result['ticket_number']:
raise MissingColumnError('ticket_number', self._message_id)
try:
result['issue_date'], result['expire_date'] = strip_list(
self._etree.xpath(delta_xp.ISSUE_EXPIRE_DATE))
except ValueError:
self.logger.error('issue_expire_date is empty %s',
self._message_id)
year = int('20' + result.get('issue_date')[-2:]) \
if result.get('issue_date') else 0
tz_dp = to_timezone(result.get('depart_city'))
tz_ar = to_timezone(result.get('arrive_city'))
dp_dt = ' '.join([result['depart_date'], result['depart_time']])
ar_dt = ' '.join([result['arrive_date'], result['arrive_time']])
dt_format = 'DDMMM h:mmA'
dt_dp = DateTime(dp_dt, dt_format)
year = year or dt_dp.received_time_to_year(self.received_time)
result['depart_datetime_formatted'] = dt_dp.year_to_datetime(
year, tz_dp)
if not result['depart_datetime_formatted']:
result.pop('depart_datetime_formatted')
self.logger.error('depart datetime formatted is empty %s',
self._message_id)
result['arrive_datetime_formatted'] = DateTime(
ar_dt, dt_format).year_to_datetime(year, tz_ar)
if not result['arrive_datetime_formatted']:
result.pop('arrive_datetime_formatted')
self.logger.error('arrive datetime formatted is empty %s',
self._message_id)
# TODO(LensHo): 没考虑转机时过年的情况, 只考虑换乘一次
trans_dp_date = result.get('trans_depart_date')
trans_dp_time = result.get('trans_depart_time')
if trans_dp_date and trans_dp_time:
trans_dp_dt = trans_dp_date + ' ' + trans_dp_time
result['trans_depart_datetime_formatted'] = DateTime(
trans_dp_dt, dt_format).year_to_datetime(year, tz_ar)
if not result['trans_depart_datetime_formatted']:
result.pop('trans_depart_datetime_formatted')
self.logger.error(
'trans_depart_datetime_formatted is empty %s',
self._message_id)
trans_ar_date = result.get('trans_arrive_date')
trans_ar_time = result.get('trans_arrive_time')
tz_trans_ar = to_timezone(result.get('trans_arrive_city'))
if trans_ar_date and trans_ar_time:
trans_ar_dt = trans_ar_date + ' ' + trans_ar_time
result['trans_arrive_datetime_formatted'] = DateTime(
trans_ar_dt, dt_format).year_to_datetime(year, tz_trans_ar)
if not result['trans_arrive_datetime_formatted']:
result.pop('trans_arrive_datetime_formatted')
self.logger.error(
'trans_arrive_datetime_formatted is empty %s',
self._message_id)
result['payment_card_number'] = strip_str(
take_first(self._etree, delta_xp.PAYMENT_METHOD_CARD))
if not result['payment_card_number']:
result.pop('payment_card_number')
self.logger.error('payment_card is empty %s', self._message_id)
result['payment'] = strip_str(
take_first(self._etree, delta_xp.PAYMENT_METHOD_MONEY))
if not result['payment']:
result.pop('payment')
self.logger.error('payment is empty %s', self._message_id)
result['duration'] = strip_str(
take_first(self._etree, delta_xp.DURATION))
if not result['duration']:
result.pop('duration')
self.logger.error('duration is empty %s', self._message_id)
result['transportation_fare'] = strip_str(
take_first(self._etree, delta_xp.BASE_FARE))
if not result['transportation_fare']:
result.pop('transportation_fare')
self.logger.error('transportation_fare is empty %s',
self._message_id)
result['price'] = strip_str(take_first(self._etree,
delta_xp.TAXES_FEE))
if not result['price']:
result.pop('price')
self.logger.error('price is empty %s', self._message_id)
result['total_cost'] = strip_str(
take_first(self._etree, delta_xp.TOTAL))
if not result['total_cost']:
result.pop('total_cost')
self.logger.error('total_cost is empty %s', self._message_id)
baggage_info = take_first(self._etree, delta_xp.BAGGAGE_INFO)
baggage_text = take_first(self._etree, delta_xp.BAGGAGE_TEXT)
baggage_link = take_first(self._etree, delta_xp.BAGGAGE_LINK)
if baggage_info:
result['baggage_allowance'] = {
'text': remove_tags(etree.tostring(baggage_info)).strip()
}
if baggage_link and baggage_text:
result['baggage_allowance'].update(
{'links': {
'name': baggage_text,
'value': baggage_link
}})
else:
self.logger.error('baggage_allowance is empty %s',
self._message_id)
hazardous_title = strip_str(
take_first(self._etree, delta_xp.HAZARDOUS_MATERIALS_TITLE))
hazardous_text_1 = strip_str(
take_first(self._etree, delta_xp.HAZARDOUS_MATERIALS_TEXT_1))
hazardous_text_2 = take_first(self._etree,
delta_xp.HAZARDOUS_MATERIALS_TEXT_2)
link = take_first(self._etree, delta_xp.HAZARDOUS_MATERIALS_LINK)
text = strip_str(
take_first(self._etree, delta_xp.HAZARDOUS_MATERIALS_TEXT))
if hazardous_title == 'Transportation of Hazardous Materials':
if hazardous_text_2:
hazardous_text_2 = remove_tags(
etree.tostring(hazardous_text_2))
else:
hazardous_text_2 = ''
self.logger.error(
'part of hazardous_material text is empty %s',
self._message_id)
if hazardous_text_1 or hazardous_text_2:
hazardous_text = '\n'.join(
[hazardous_text_1, hazardous_text_2]).strip()
result['transportation_of_hazardous_materials'] = {
'text': hazardous_text
}
if link and text and \
result.get('transportation_of_hazardous_materials'):
result['transportation_of_hazardous_materials'].update(
{'links': {
'name': text,
'value': link
}})
else:
self.logger.warning(
'Transportation of Hazardous Materials not found %s',
self._message_id)
return result
def _parse(self):
result = self._order
result['hotel_name'] = strip_str(
take_first(self._etree, tz_xp.HOTEL_NAME))
if not result['hotel_name']:
raise MissingColumnError('hotel_name', self._message_id)
result['address'] = ', '.join(
[i.strip(',') for i in self._etree.xpath(tz_xp.ADDRESS)])
if not result['address']:
raise MissingColumnError('address', self._message_id)
result['telephone'] = take_first(self._etree, tz_xp.PHONE)
if not result['telephone']:
raise MissingColumnError('telephone', self._message_id)
result.set('hotel_link', take_first(self._etree, tz_xp.WEBSITE))
name_code = strip_list(self._etree.xpath(tz_xp.NAME_CODE))
if len(name_code) < 3:
raise MissingColumnError('guest_name or confirm_code',
self._message_id)
result['guest_name'] = name_code.pop(0)
result['travelzoo_reference_code'] = name_code.pop(0)
result['confirm_code'] = name_code.pop(0)
# 有可能有其他额外确认信息
if name_code:
self.logger.warning('other confirmation: %s %s', name_code.pop(),
self._message_id)
date = strip_str(take_first(self._etree, tz_xp.DATE))
if not date:
raise MissingColumnError('date is empty', self._message_id)
check_in_date, check_out_date = unpack(date.split(' - '))
if not check_in_date or not check_out_date:
raise MissingColumnError('check_in_date ot checkout_date',
self._message_id)
tz = to_timezone(result.get('address'))
date_format = 'MMM DD, YYYY'
result['check_in_date'] = check_in_date
result['check_in_date_formatted'] = DateTime(
check_in_date, date_format).tz_to_datetime(tz)
if not result['check_in_date_formatted']:
result.pop('check_in_date_formatted')
self.logger.error('check_in_date_formatted is empty %s',
self._message_id)
result['check_out_date'] = check_out_date
result['check_out_date_formatted'] = DateTime(
check_out_date, date_format).tz_to_datetime(tz)
if not result['check_out_date_formatted']:
result.pop('check_out_date_formatted')
self.logger.error('check_out_date_formatted is empty %s',
self._message_id)
result['number_of_guests'] = strip_str(
take_first(self._etree, tz_xp.GUEST_NUM))
if not result['number_of_guests']:
result.pop('number_of_guests')
self.logger.error('number of guests is empty %s', self._message_id)
roomtype_price = strip_list(self._etree.xpath(tz_xp.ROOMTYPE_PRICE))
if len(roomtype_price) >= 2:
result['room_type'] = roomtype_price.pop(0)
price = roomtype_price.pop(0)
result['price'] = price.split('- ')[-1]
# 可能会有其他信息
if roomtype_price:
self.logger.warning('other price: %s %s', roomtype_price.pop(),
self._message_id)
else:
self.logger.error('price is empty %s', self._message_id)
payment = self._etree.xpath(tz_xp.PAYMENT)
if payment:
for item in payment:
name = strip_str(take_first(item, './td[1]/text()'))
cost = strip_str(take_first(item, './td[2]/text()'))
if not name or not cost:
continue
if 'Night' in name:
result['number_of_nights'] = name
result['total_without_taxes'] = cost
elif 'tax' in name:
result['taxes_name'] = name
result['taxes_fee'] = cost
elif 'Total' in name:
result['total_cost'] = cost
# 可能会有其他价格信息
else:
self.logger.warning('payment: %s is %s %s', name, cost,
self._message_id)
else:
self.logger.error('payment is empty %s', self._message_id)
policies = self._etree.xpath('//b')
if policies:
for policy in policies:
name = strip_str(take_first(policy, './text()'))
value = strip_list(policy.xpath('../text()'))
if name in self.name_dict:
result[self.name_dict[name]] = value
elif result['hotel_name'] in name:
pass
else:
self.logger.warning('policy: %s is %s %s', name, value,
self._message_id)
else:
self.logger.error('policies is empty %s', self._message_id)
result['cancellation_link'] = take_first(self._etree,
tz_xp.CANCEL_LINK)
if not result['cancellation_link']:
result.pop('cancellation_link')
self.logger.error('cancellation link is empty %s',
self._message_id)
contact = self._etree.xpath(tz_xp.CONTACT)
if contact:
result['contact_information'] = []
for item in contact:
link = take_first(item, './a/@href')
if 'tel' in link:
region = take_first(item, './text()')
tel = take_first(item, './a/text()')
if region and tel:
result['contact_information'].append({
'name': region,
'value': tel
})
if not result['contact_information']:
result.pop('contact_information')
self.logger.warning('contact information is empty %s',
self._message_id)
else:
self.logger.warning('contact is empty %s', self._message_id)
return result
def _parse(self): # pylint: disable=R0912,R0915
"""Override"""
result = self._order
confirm_num_values = self._etree.xpath(
booking_xp.CONFIRMATION_NUMBER_VALUES)
confirm_num_values = strip_list(confirm_num_values)
if len(confirm_num_values) == 2:
result['confirm_code'] = confirm_num_values[0]
result['pin_code'] = confirm_num_values[1]
else:
self.logger.error('confirm_number is empty %s', self._message_id)
name = strip_str(take_first(self._etree, booking_xp.HOTEL_NAME))
if name:
result['hotel_name'] = name
else:
self.logger.error('hotel_name is empty %s', self._message_id)
address = slice_list(
strip_str(take_first(self._etree, booking_xp.ADDRESS)), -3)[0]
if address:
result['address'] = address
else:
self.logger.error('address is empty %s', self._message_id)
telephone = strip_str(take_first(self._etree, booking_xp.PHONE))
if telephone:
result['telephone'] = telephone
else:
self.logger.error('telephone is empty %s', self._message_id)
check_in_date_formatted = take_first(
self._etree, booking_xp.CHECK_IN_DATE_FORMATTED)
if check_in_date_formatted:
result['check_in_date_formatted'] = \
arrow.get(check_in_date_formatted).datetime
check_out_date_formatted = take_first(
self._etree, booking_xp.CHECK_OUT_DATE_FORMATTED)
if check_out_date_formatted:
result['check_out_date_formatted'] = \
arrow.get(check_out_date_formatted).datetime
check_in_date = strip_str(
take_first(self._etree, booking_xp.CHECK_IN_DATE))
if check_in_date:
result['check_in_date'] = check_in_date
else:
self.logger.error('check_in_date is empty %s', self._message_id)
check_in_time = strip_str(
take_first(self._etree, booking_xp.CHECK_IN_TIME)
) \
.strip('(').strip(')')
if check_in_time:
result['check_in_time'] = check_in_time
else:
self.logger.error('check_in_time is empty %s', self._message_id)
check_out_date = strip_str(
take_first(self._etree, booking_xp.CHECK_OUT_DATE))
if check_out_date:
result['check_out_date'] = check_out_date
else:
self.logger.error('check_out_date is empty %s', self._message_id)
check_out_time = strip_str(
take_first(self._etree, booking_xp.CHECK_OUT_TIME)
) \
.strip('(').strip(')')
if check_out_time:
result['check_out_time'] = check_out_time
else:
self.logger.error('check_out_time is empty %s', self._message_id)
price_details = strip_list(self._etree.xpath(booking_xp.PRICE_DETAILS))
if price_details:
result['bed_type'] = price_details[0]
result['price'] = price_details[1]
result['total_cost'] = price_details[-1]
tmp = zip(price_details[::2][1:-1], price_details[1::2][1:-1])
for i, j in tmp:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('price_details is empty %s', self._message_id)
requests = strip_list(
self._etree.xpath(booking_xp.SPECIAL_REQUEST_1)
or self._etree.xpath(booking_xp.SPECIAL_REQUEST_2))
your_special_requirements = [
i for i in requests
if i and i != "\u2022" and i != "Special Requests"
]
if your_special_requirements:
result['your_special_requirements'] = your_special_requirements
else:
self.logger.warning('your_special_requirements is empty %s',
self._message_id)
free_cancellation = strip_str(
take_first(self._etree, booking_xp.CANCELLATION))
free_cancellation = free_cancellation.split('.')[0].replace('\n', ' ')
if free_cancellation:
result['free_cancellation'] = free_cancellation
else:
self.logger.error('free_cancellation_policy is empty %s',
self._message_id)
payment_forms = strip_str(
take_first(self._etree, booking_xp.PAYMENT_FORMS))
if payment_forms:
result['payment_forms'] = payment_forms
else:
self.logger.warning('payment_forms is empty %s', self._message_id)
if free_cancellation:
tz = check_out_date_formatted[-5:] \
if check_out_date_formatted else to_timezone(address)
free_cancellation_time = \
DateTime(free_cancellation,
'MMMM DD, YYYY hh:mm A').tz_to_datetime(tz) # D, H
if str(free_cancellation_time):
result['free_cancellation_time'] = free_cancellation_time
notice = strip_list(self._etree.xpath(
booking_xp.IMPORTANT_INFORMATION))
price_extra = strip_list(self._etree.xpath(booking_xp.PRICE_EXTRA))
if price_extra:
result['price_tips'] = [
i for i in price_extra if 'Book.com' not in i
]
else:
self.logger.error('price_extra is empty %s', self._message_id)
if notice:
result['notice'] = notice
else:
self.logger.error('notice is empty %s', self._message_id)
stay = strip_str(take_first(self._etree, booking_xp.STAY))
if stay:
result['number_of_nights'] = stay.split(',')[0]
result['number_of_rooms'] = stay.split(',')[-1].strip()
else:
self.logger.error('stay is empty %s', self._message_id)
booking_summary_keys = strip_list(
self._etree.xpath(booking_xp.BOOKING_SUMMERY_KEYS))
booking_summary_values = self._etree.xpath(
booking_xp.BOOKING_SUMMERY_VALUES)
if booking_summary_values and booking_summary_keys:
booking_summary_values = strip_list([
remove_tags(etree.tostring(i)) for i in booking_summary_values
])
for k, v in zip(booking_summary_keys, booking_summary_values):
if 'Cancellation cost' in k:
cost = self._etree.xpath(booking_xp.CANCELLATION_COST)
if cost:
v = [remove_tags(etree.tostring(i)) for i in cost]
v = [i.replace('\n', ' ').strip() for i in v]
result['cancellation_cost'] = v
elif 'Prepayment' in k:
result['prepayment'] = v
elif k in ['Check-in', 'Check-out', 'Your reservation']:
pass
else:
self.logger.warning('%s is %s %s', k, v, self._message_id)
else:
self.logger.error('booking_summary is empty %s', self._message_id)
room_tip = slice_list(
strip_list(self._etree.xpath(booking_xp.ROOM_AREA)), 1)[1]
if room_tip:
result['room_tips'] = room_tip
else:
self.logger.warning('room_tip is empty %s', self._message_id)
order_condition_keys = self._etree.xpath(
booking_xp.BOOKING_CONDITIONS_KEYS)
order_condition_keys = slice_list(order_condition_keys, 3)[0]
order_condition_values = self._etree.xpath(
booking_xp.BOOKING_CONDITIONS_VALUES)
order_condition_values = slice_list(order_condition_values, 3)[0]
if order_condition_values and order_condition_keys:
order_condition_values = map(
lambda i: remove_tags(etree.tostring(i)),
order_condition_values)
order_condition_values = strip_list(order_condition_values)
name_dict = {
'Guest parking': 'guest_parking',
'Internet': 'internet'
}
for k, v in zip(order_condition_keys, order_condition_values):
if k in order_condition_keys:
result[name_dict[k]] = v
elif 'Cancellation policy' in k:
result['cancellation_policies'] = [v]
else:
self.logger.warning('%s is %s %s', k, v, self._message_id)
else:
self.logger.error('room_details is empty %s', self._message_id)
hotel_link = take_first(self._etree, booking_xp.HOTEL_URL)
if hotel_link:
result['hotel_link'] = hotel_link
else:
self.logger.error('hotel_link is empty %s', self._message_id)
address_link = take_first(self._etree, booking_xp.ADDRESS_URL)
if address_link:
result['map_link'] = address_link
else:
self.logger.error('map_link is empty %s', self._message_id)
cancellation_link = take_first(self._etree,
booking_xp.CANCELLATION_URL)
if cancellation_link:
result['cancellation_link'] = cancellation_link
else:
self.logger.error('cancellation_link is empty %s',
self._message_id)
change_keys = strip_list(
self._etree.xpath(booking_xp.CHANGE_LINKS_KEYS))
change_values = self._etree.xpath(booking_xp.CHANGE_LINKS_VALUES)
related_links = filter_dict_value(to_dict(change_keys, change_values))
if related_links:
result['related_links'] = related_links
else:
self.logger.error('related_links is empty %s', self._message_id)
return result
def _parse(self): # pylint: disable=R0912,R0915
result = self._order
hotel_name = strip_str(take_first(self._etree, hilton_xp.HOTEL_NAME))
if hotel_name:
result['hotel_name'] = hotel_name
else:
self.logger.error('hotel_name is empty %s', self._message_id)
confirm_num = take_first(self._etree, hilton_xp.CONFIRM_NUM)
if confirm_num:
confirm_num = confirm_num.split(':')[-1].strip()
result['confirm_code'] = confirm_num
else:
self.logger.error('confirm_number is empty %s', self._message_id)
address = strip_str(take_first(self._etree, hilton_xp.ADDRESS))
if address:
result['address'] = remove_space(address)
else:
self.logger.error('address is empty %s', self._message_id)
phone = strip_list(
strip_str(take_first(self._etree, hilton_xp.PHONE)).split('\n'))
if phone:
result['telephone'] = ' '.join(phone)
else:
self.logger.error('telephone is empty %s', self._message_id)
json_data = take_first(self._etree, hilton_xp.JSON)
if json_data:
data = json.loads(json_data)
check_in_datetime = data.get('checkinTime')
check_out_datetime = data.get('checkoutTime')
else:
check_in_datetime = check_out_datetime = ''
self.logger.error(
'check_in_datetime and check_out_datetime is empty %s',
self._message_id)
if check_in_datetime and check_out_datetime:
result['check_in_date_formatted'] = \
arrow.get(check_in_datetime).datetime
result['check_out_date_formatted'] = \
arrow.get(check_out_datetime).datetime
result['check_in_date'] = check_in_datetime.split('T')[0]
result['check_in_time'] = check_in_datetime.split('T')[-1].split(
'-')[0]
result['check_out_date'] = check_out_datetime.split('T')[0]
result['check_out_time'] = \
check_out_datetime.split('T')[-1].split('-')[0]
update_link = take_first(self._etree, hilton_xp.UPDATE_LINK)
if update_link:
result['modify_link'] = update_link
else:
self.logger.error('modify link is empty %s', self._message_id)
related_links = self._etree.xpath(hilton_xp.RELATED_LINK)
names = ['Explore', 'Maps', 'Dining', 'Convenience']
name_dict = {
'Explore': 'explore_link',
'Maps': 'map_link',
'Dining': 'dining_link',
'Convenience': 'convenience_link'
}
if related_links:
for i, j in zip(names, related_links):
if i in name_dict.keys():
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('related_links is empty %s', self._message_id)
reservation_info = self._etree.xpath(hilton_xp.ROOM_INFO)
if reservation_info:
room_type = reservation_info.pop(0)
room_type = ' '.join(strip_list(room_type.xpath('./td/text()')))
if room_type:
result['room_type'] = room_type
else:
self.logger.error('room_type is empty %s', self._message_id)
name_dict = {
'Guests': 'number_of_guests',
'Rooms': 'number_of_rooms',
'Check In': 'check_in_raw',
'Check Out': 'check_out_raw'
}
for i in reservation_info:
title = strip_str(take_first(i, './td[1]/strong/text()'))
value = remove_space(take_first(i, './td[2]/text()'))
if title in name_dict.keys():
result[name_dict[title]] = value
else:
self.logger.warning('%s is %s %s', title, value,
self._message_id)
else:
self.logger.error('reservation_information is empty %s',
self._message_id)
charge_names = strip_list(self._etree.xpath(hilton_xp.CHARGE_NAME))
charge_values = strip_list(self._etree.xpath(hilton_xp.CHARGE_VALUE))
charge_values = [remove_space(i) for i in charge_values]
charge_names = [
remove_space(i) for i in charge_names if 'Stay per Room' not in i
]
if charge_names and charge_values:
name_dict = {
'Taxes': 'taxes_fee_per_room',
'Total': 'total_per_room',
'Total for Stay': 'total_cost',
'Service Charge': 'service_charge_per_room',
'Rate': 'price_per_room',
'Rate per night': 'price',
}
for i, j in zip(charge_names, charge_values):
if i in name_dict.keys():
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('price_details is empty %s', self._message_id)
cancellation = strip_list(self._etree.xpath(hilton_xp.CANCELLATION))
if cancellation:
result['cancellation_policies'] = cancellation
else:
self.logger.error('cancellation is empty %s', self._message_id)
rate_rule_cancellation = \
strip_list(self._etree.xpath(hilton_xp.RATE_RULE_CANCELLATION))
if rate_rule_cancellation:
result['rate_rules'] = rate_rule_cancellation
else:
self.logger.error('rate_rule_cancellation is empty %s',
self._message_id)
additional_info_names = strip_list(
self._etree.xpath(hilton_xp.ADDITIONAL_INFO_NAME))
additional_info_values = \
strip_list(self._etree.xpath(hilton_xp.ADDITIONAL_INFO_VALUE))
name_dict = {
'Service Charges': 'service_charge',
'Self parking': 'guest_parking'
}
if additional_info_names and additional_info_values:
for i, j in zip(additional_info_names, additional_info_values):
if i in name_dict.keys():
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('additional_information is empty %s',
self._message_id)
notice = strip_list(self._etree.xpath(hilton_xp.NOTICE))
if notice:
result['notice'] = notice
return result
def _parse(self):
result = self._order
result['confirm_code'] = strip_str(
take_first(self._etree, ba_xp.CONFIRM_CODE))
if not result['confirm_code']:
raise MissingColumnError('confirm_code', self._message_id)
try:
result['flight_number'] = self._etree.xpath(ba_xp.FLIGHT_NUM)[0]
except IndexError as e:
raise MissingColumnError('flight_number', self._message_id) from e
itinerary = strip_list(self._etree.xpath(ba_xp.ITINERARY))
if len(itinerary) != 8:
raise MissingColumnError('itinerary', self._message_id)
result['depart_date'], result['arrive_date'] = itinerary[::4]
result['depart_time'], result['arrive_time'] = itinerary[1::4]
result['depart_city'], result['arrive_city'] = itinerary[2::4]
result['depart_terminal'] = itinerary[3]
result['arrive_terminal'] = itinerary[-1]
tz_dp, tz_ar = to_timezone(itinerary[2]), to_timezone(itinerary[6])
dp_dt, ar_dt = ' '.join(itinerary[:2]), ' '.join(itinerary[4:6])
# TODO(LensHo): 小时的格式可能为H
dt_format = 'd MMM YYYY HH:mm'
result['depart_datetime_formatted'] = DateTime(dp_dt, dt_format) \
.tz_to_datetime(tz_dp)
if not result['depart_datetime_formatted']:
result.pop('depart_datetime_formatted')
self.logger.error('depart_datetime_formatted is empty %s',
self._message_id)
result['arrive_datetime_formatted'] = DateTime(ar_dt, dt_format) \
.tz_to_datetime(tz_ar)
if not result['arrive_datetime_formatted']:
result.pop('arrive_datetime_formatted')
self.logger.error('arrive_datetime_formatted is empty %s',
self._message_id)
guest = strip_list(self._etree.xpath(ba_xp.GUEST))
if guest:
guest = [name for name in guest if name != 'Passenger']
result['guest_name'] = ', '.join(guest).strip(', ')
result['guest_names_list'] = guest
else:
self.logger.error('guest_name is empty %s', self._message_id)
baggage_names = strip_list(self._etree.xpath(ba_xp.BAGGAGE_NAMES))
baggage_values = strip_list(self._etree.xpath(ba_xp.BAGGAGE_VALUES))
if baggage_names and baggage_values:
for name, value in zip(baggage_names, baggage_values):
if name in self.name_dict:
result[self.name_dict[name]] = value
else:
self.logger.warning('baggage information %s is %s', name,
value)
else:
self.logger.error('baggage_allowance is empty %s',
self._message_id)
airline_link = self._etree.xpath(ba_xp.AIRLINE_LINK)
airline_text = self._etree.xpath(ba_xp.AIRLINE_TEXT)
if airline_link:
result['carrier_link'] = airline_link[0]
result['carrier_name'] = airline_text[0]
if len(airline_link) >= 2:
result['trans_airline_link'] = airline_link[1]
result['trans_airline_name'] = airline_text[1]
else:
self.logger.warning('airline name is empty %s', self._message_id)
result['disability_assistance'] = strip_str(
take_first(self._etree, ba_xp.DISABILITY_ASSISTANCE))
if not result['disability_assistance']:
result.pop('disability_assistance')
self.logger.warning('disability assistance is empty %s',
self._message_id)
payment = self._etree.xpath(ba_xp.PAYMENT)
if payment:
for item in payment:
item = strip_list(item.xpath('./td/text()'))
if item:
if item[0] in self.name_dict:
result[self.name_dict[item[0]]] = remove_space(item[1])
elif len(item) == 1:
result['ticket_number'] = ', '.join([
result.get('ticket_number', ''),
remove_space(item[0])
])
else:
self.logger.warning('%s is %s %s', item[0], item[1],
self._message_id)
else:
self.logger.error('payment is empty %s', self._message_id)
result['modify_link'] = take_first(self._etree, ba_xp.CHANGE_LINKS)
if not result['modify_link']:
result.pop('modify_link')
self.logger.error('modify link is empty %s', self._message_id)
links = self._etree.xpath(ba_xp.LINKS)
if links:
result['related_links'] = []
for item in links:
name = take_first(item, './text()')
link = take_first(item, './@href')
if name and link:
result['related_links'].append({
'name': name,
'link': link
})
if not result:
result.pop('related_links')
self.logger.error('related_links is empty %s',
self._message_id)
else:
self.logger.error('related_links is empty %s', self._message_id)
return result
def _parse(self): # pylint: disable=R0912,R0915
result = self._order
hotel_name = strip_str(take_first(self._etree, marriott_xp.HOTEL_NAME))
if hotel_name:
result['hotel_name'] = hotel_name
else:
self.logger.error('hotel_name is empty %s', self._message_id)
address = strip_str(take_first(self._etree, marriott_xp.ADDRESS))
if address:
result['address'] = address
else:
self.logger.error('address is empty %s', self._message_id)
tel = take_first(self._etree, marriott_xp.PHONE)
if tel:
result['telephone'] = tel
else:
self.logger.error('telephone is empty %s', self._message_id)
confirm_num = take_first(self._etree, marriott_xp.CONFIRM_NUM)
if confirm_num:
confirm_num = confirm_num.split(': ')[-1]
result['confirm_code'] = confirm_num
else:
self.logger.error('confirm_number is empty %s', self._message_id)
guest = take_first(self._etree, marriott_xp.GUEST)
if guest:
guest = guest.split('For ')[-1]
result['guest_name'] = guest
else:
self.logger.error('guest_name is empty %s', self._message_id)
check_in_out_time = strip_list(
self._etree.xpath(marriott_xp.CHECK_IN_OUT_TIME))
if len(check_in_out_time) == 2:
result['check_in_time'] = check_in_out_time[0]
result['check_out_time'] = check_in_out_time[1]
else:
self.logger.error('check_in_date and check_out_time is empty %s',
self._message_id)
check_in_out_date = strip_list(
self._etree.xpath(marriott_xp.CHECK_IN_OUT_DATE))
check_in_date, check_out_date = unpack(check_in_out_date)
if check_in_date and check_out_date:
result['check_in_date'] = check_in_date
result['check_out_date'] = check_out_date
tz = to_timezone(address)
check_in_date_formatted = \
DateTime(check_in_date, 'MMMM DD, YYYY').tz_to_datetime(tz)
if check_in_date_formatted:
result['check_in_date_formatted'] = check_in_date_formatted
else:
self.logger.error('check_in_date_formatted is empty %s',
self._message_id)
check_out_date_formatted = \
DateTime(check_out_date, 'MMMM DD, YYYY').tz_to_datetime(tz)
if check_out_date_formatted:
result['check_out_date_formatted'] = check_out_date_formatted
else:
self.logger.error('check_out_date_formatted is empty %s',
self._message_id)
else:
self.logger.error('check_in_date and check_out_date is empty %s',
self._message_id)
related_links = self._etree.xpath(marriott_xp.RELATED_LINK)
related_text = self._etree.xpath(marriott_xp.RELATED_TEXT)
related_links = to_dict(related_text, related_links)
if related_links:
related = []
for i in related_links:
if 'Hotel Website' in i.get('name'):
result['hotel_link'] = i.get('value')
related_links.remove(i)
elif 'Map & Directions' in i.get('name'):
result['map_link'] = i.get('value')
related_links.remove(i)
elif 'Cancel' in i.get('name'):
result['cancellation_link'] = i.get('value')
else:
related.append(i)
if related:
result['related_links'] = related_links
else:
self.logger.error('related_links is empty %s', self._message_id)
room_type = take_first(self._etree, marriott_xp.ROOM_TYPE)
room_type_value = take_first(self._etree, marriott_xp.ROOM_TYPE_VALUE)
if room_type and room_type_value:
result['room_type'] = room_type_value
else:
self.logger.error('room_type is empty %s', self._message_id)
room_num_guest = strip_list(
self._etree.xpath(marriott_xp.ROOM_NUM_GUEST))
room_num_guest_name, room_num_guest_value = group(room_num_guest)
if room_num_guest_name and room_num_guest_value:
for i, j in zip(room_num_guest_name, room_num_guest_value):
if 'NUMBER OF ROOMS' in i:
result['number_of_rooms'] = j
elif 'GUESTS PER ROOM' in i:
continue
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('room number is empty %s', self._message_id)
guarantee = strip_list(self._etree.xpath(
marriott_xp.GUARANTEED_METHOD))
if guarantee:
result['guarantee_policies'] = [guarantee[-1]]
else:
self.logger.error('guarantee is empty %s', self._message_id)
price_des = strip_str(
take_first(self._etree, marriott_xp.CHARGE_DESCRIPTION))
if price_des:
result['price_tips'] = [price_des]
else:
self.logger.warning('price_description is empty %s',
self._message_id)
notice = strip_list(self._etree.xpath(marriott_xp.HOTEL_ALERT))
if notice:
result['notice'] = notice
else:
self.logger.error('notice is empty %s', self._message_id)
rates = strip_list(self._etree.xpath(marriott_xp.RATES))
if rates:
rates_type = rates.pop()
if 'Best Available rate' in rates_type:
nights = sum([int(i.split(' ')[0]) for i in rates[1::3]])
price = sum([float(i.split(' ')[0]) for i in rates[2::3]])
currency = ' ' + rates[2].split(' ')[-1]
result['price'] = str(round(price / nights, 2)) + currency
else:
result['price'] = rates[-1]
self.logger.error('price is empty %s', self._message_id)
else:
self.logger.error('rates is empty %s', self._message_id)
taxes = strip_list(self._etree.xpath(marriott_xp.TAXES))
name, value = unpack(taxes)
if name and value:
if 'TAXES & FEES' in name:
result['taxes_fee'] = taxes[-1]
else:
self.logger.error('%s is %s %s', name, value, self._message_id)
else:
self.logger.error('taxes is empty %s', self._message_id)
total = strip_list(self._etree.xpath(marriott_xp.TOTAL))
name, value = unpack(total)
if total:
if 'Total' in name:
result['total_cost'] = total[-1]
else:
self.logger.error('%s is %s %s', name, value, self._message_id)
else:
self.logger.error('total_price is empty %s', self._message_id)
other_charge = strip_list(self._etree.xpath(marriott_xp.OTHER_CHARGE))
other_charge = [
i for i in other_charge if i != '\u2022' and i != 'Other Charges'
]
if other_charge:
result['other_charges'] = other_charge
else:
self.logger.warning('other_charge is empty %s', self._message_id)
cancellation = take_first(self._etree,
marriott_xp.RATE_CANCELLATION_DETAILS)
if cancellation is not None:
cancellation = replace_tags(etree.tostring(cancellation))
cancellation = cancellation.split('• \n')
result['cancellation_policies'] = strip_list(cancellation)
else:
self.logger.error('cancellation_policy is empty %s',
self._message_id)
rate_guarantee_title = take_first(self._etree,
marriott_xp.RATE_GUARANTEE_TITLE)
rate_guarantee = strip_list(
self._etree.xpath(marriott_xp.RATE_GUARANTEE))
rate_guarantee = [i for i in rate_guarantee if i != '\u2022']
if rate_guarantee and 'GUARANTEE' in rate_guarantee_title:
guarantee = result.get('guarantee_policies')
if guarantee:
result['guarantee_policies'].extend(rate_guarantee)
else:
result['guarantee_policies'] = rate_guarantee
else:
self.logger.error('rate guarantee is empty %s', self._message_id)
addition_title = take_first(self._etree,
marriott_xp.ADDITION_INFO_TITLE)
addition_link = strip_list(
self._etree.xpath(marriott_xp.ADDITION_INFO_LINK))
addition_text = strip_list(
self._etree.xpath(marriott_xp.ADDITION_INFO_TEXT))
if addition_text and 'ADDITIONAL' in addition_title:
result['additional_information'] = to_dict(addition_text,
addition_link)
else:
self.logger.error('additional information is empty %s',
self._message_id)
contact_links = self._etree.xpath(marriott_xp.CONTACT_LINK)
contact_texts = strip_list(self._etree.xpath(marriott_xp.CONTACT_TEXT))
contact = strip_list(self._etree.xpath(marriott_xp.CONTACT_1))
contact_1 = [{'name': i, 'value': i} for i in contact]
contact = chain(to_dict(contact_texts, contact_links), contact_1)
contact = filter_dict_value(contact)
if contact:
result['contact_information'] = contact
return result
def _parse(self): # pylint: disable=R0912, R0915
result = self._order
address = strip_str(take_first(self._etree, hotels_xp.ADDRESS))
if address:
result['address'] = address
else:
self.logger.error('address is empty %s', self._message_id)
confirm_num = take_first(self._etree, hotels_xp.CONFIRM_NUMBER)
if confirm_num:
result['confirm_code'] = confirm_num
else:
self.logger.error('conform_code is empty %s', self._message_id)
tz = to_timezone(address)
check_in_date = take_first(self._etree, hotels_xp.CHECK_IN_DATE)
if check_in_date:
check_in_date_formatted = DateTime(
check_in_date,
'MMMM D, YYYY'
) \
.tz_to_datetime(tz) #
if check_in_date_formatted:
result['check_in_date_formatted'] = check_in_date_formatted
result['check_in_date'] = check_in_date
else:
self.logger.error('check_in_date is empty %s', self._message_id)
check_in_time = take_first(self._etree, hotels_xp.CHECK_IN_TIME)
if check_in_time:
result['check_in_time'] = check_in_time.strip('(').strip(')')
else:
self.logger.error('check_in_time is empty %s', self._message_id)
check_out_date = take_first(self._etree, hotels_xp.CHECK_OUT_DATE)
if check_out_date:
check_out_date_formatted = DateTime(
check_out_date,
'MMMM D, YYYY'
) \
.tz_to_datetime(tz)
if check_out_date_formatted:
result['check_out_date_formatted'] = check_out_date_formatted
result['check_out_date'] = check_out_date
else:
self.logger.error('check_out_date is empty %s', self._message_id)
check_out_time = take_first(self._etree, hotels_xp.CHECK_OUT_TIME)
if check_out_time:
result['check_out_time'] = check_out_time.strip('(').strip(')')
else:
self.logger.error('check_out_time is empty %s', self._message_id)
stay = take_first(self._etree, hotels_xp.STAY)
if stay:
result['number_of_nights'] = stay.split(', ')[0]
result['number_of_rooms'] = stay.split(', ')[-1]
else:
self.logger.error('stay is empty %s', self._message_id)
cancellation_details = strip_list([
strip_str(take_first(self._etree, hotels_xp.CANCELLATION)),
strip_str(
take_first(self._etree, hotels_xp.CANCELLATION_POLICY_1)
),
strip_str(
take_first(self._etree, hotels_xp.CANCELLATION_POLICY_2)
)
])
if cancellation_details:
result['cancellation_policies'] = cancellation_details
else:
self.logger.error('policies is empty %s', self._message_id)
hotel_name = strip_str(take_first(self._etree, hotels_xp.NAME))
if hotel_name:
result['hotel_name'] = hotel_name
else:
self.logger.error('hotel_name is empty %s', self._message_id)
tel = take_first(self._etree, hotels_xp.TELEPHONE)
if tel:
result['telephone'] = tel.split(':')[-1].strip()
else:
self.logger.error('telephone is empty %s', self._message_id)
notice = self._etree.xpath(hotels_xp.IMPORTANT_NOTICE)
if not notice:
self.logger.warning('notice is empty %s', self._message_id)
required = self._etree.xpath(hotels_xp.REQUIRED)
if not required:
self.logger.warning('required is empty %s', self._message_id)
if notice or required:
result['notice'] = list(chain(notice or [], required or []))
price = strip_str(take_first(self._etree, hotels_xp.PRICE))
if price:
result['price'] = price
else:
self.logger.error('price is empty %s', self._message_id)
total = strip_str(take_first(self._etree, hotels_xp.COST))
if total:
result['total_cost'] = total
else:
self.logger.error('total_cost is empty %s', self._message_id)
guest_name = ' '.join(strip_list(self._etree.xpath(hotels_xp.ROOM_2)))
if guest_name:
result['guest_name'] = guest_name
else:
self.logger.error('guest_name is empty %s', self._message_id)
icon_explanation = take_first(self._etree, hotels_xp.ICON_EXPLANATION)
if icon_explanation:
result['icon_explanation'] = icon_explanation
else:
self.logger.warning(
'icon_explanation is empty %s', self._message_id
)
room_all = self._etree.xpath(hotels_xp.ROOM_ALL)
bed_type = room_all.pop(0)
if bed_type:
result['bed_type'] = bed_type
else:
self.logger.error('bed_type is empty %s', self._message_id)
values = []
room_tip = []
for i in room_all:
if len(i.strip()) < 5:
continue
if '-sq-' in i:
result['room_area'] = i.strip()
elif i.startswith(' - '):
values.append(i.split(' - ')[-1].strip())
else:
room_tip.append(i.strip())
if room_tip:
result['room_tips'] = room_tip
else:
self.logger.error('room_tips is empty %s', self._message_id)
room_type = strip_str(take_first(self._etree, hotels_xp.ROOM_1))
if room_type:
result['room_type'] = room_type
else:
self.logger.error('room_type is empty %s', self._message_id)
preference = strip_str(take_first(self._etree, hotels_xp.PREFERENCE))
if preference:
result['preference'] = preference
else:
self.logger.error('preference is empty %s', self._message_id)
note = strip_str(take_first(self._etree, hotels_xp.NOTE))
if note:
result['tip'] = note
else:
self.logger.error('tip is empty %s', self._message_id)
facilities_keys = self._etree.xpath(hotels_xp.FACILITIES_KEYS)
name_dict = {
'Internet': 'internet',
'Entertainment': 'entertainment',
'Food & Drink': 'food_drink',
'Sleep': 'sleep',
'Bathroom': 'bathroom',
'Practical': 'practical',
'Comfort': 'comfort'
}
if facilities_keys and values:
for i, j in zip(facilities_keys, values):
if i in name_dict.keys():
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('room_details is empty %s', self._message_id)
return result
def _parse(self) -> Dict: # pylint: disable=R0912,R0915,R0916
result = self._order
itinerary = strip_str(
take_first(self._etree, expedia_flight_xp.ITINERARY))
if itinerary:
result['itinerary_code'] = itinerary
else:
self.logger.error('itinerary is empty %s', self._message_id)
confirm_num = strip_str(
take_first(self._etree, expedia_flight_xp.CONFIRM_NUM))
if confirm_num:
result['confirm_code'] = confirm_num
else:
self.logger.error('confirm_code is empty %s', self._message_id)
booking_id = strip_str(
take_first(self._etree, expedia_flight_xp.BOOKING_ID))
if booking_id:
result['booking_id'] = booking_id
else:
self.logger.error('booking_id is empty %s', self._message_id)
ticket = strip_str(take_first(self._etree, expedia_flight_xp.TICKET))
if ticket:
result['ticket_number'] = ticket.split(' ')[0] \
if ' ' in ticket else ticket
else:
self.logger.error('ticket_number is empty %s', self._message_id)
depart_date, arrive_date = \
unpack(strip_list(take_first(self._etree,
expedia_flight_xp.DATES).split(' - ')))
if depart_date and arrive_date:
result['depart_date'] = depart_date
result['arrive_date'] = arrive_date
else:
self.logger.error('depart_date and arrive_date is empty %s',
self._message_id)
related_links_1 = self._etree.xpath(expedia_flight_xp.RELATED_LINK_1)
if related_links_1:
name_dict = {
'Change': 'change_link',
'cancel': 'cancellation_link',
'Customer Support': 'customer_support_link'
}
for i in related_links_1:
link = take_first(i, './@href')
text = take_first(i, './text()')
if text in name_dict.keys():
result[name_dict[text]] = link
else:
self.logger.warning('%s is %s %s', text, link,
self._message_id)
else:
self.logger.error('part 1 of related_links is empty %s',
self._message_id)
related_links_2 = self._etree.xpath(expedia_flight_xp.RELATED_LINK_2)
related_texts_2 = self._etree.xpath(expedia_flight_xp.RELATED_TEXT_2)
if related_links_2 and related_texts_2:
result['related_links'] = to_dict(related_texts_2, related_links_2)
else:
self.logger.warning('part 2 of related_links is empty %s',
self._message_id)
flight_name = strip_str(
take_first(self._etree, expedia_flight_xp.FLIGHT_NAME))
if flight_name:
result['flight_name'] = flight_name
else:
self.logger.error('flight_name is empty %s', self._message_id)
policies = strip_str(
take_first(self._etree, expedia_flight_xp.CANCELLATION))
if policies:
result['policies'] = [policies]
else:
self.logger.error('policies is empty %s', self._message_id)
depart_station = strip_str(
take_first(self._etree, expedia_flight_xp.DEPART_STATION))
if depart_station:
result['depart_station'] = depart_station
else:
self.logger.error('depart_station is empty %s', self._message_id)
depart_time = strip_str(
take_first(self._etree, expedia_flight_xp.DEPART_TIME))
if depart_time:
result['depart_time'] = depart_time
else:
self.logger.error('depart_time is empty %s', self._message_id)
depart_terminal = strip_str(
take_first(self._etree, expedia_flight_xp.DEPART_TERMINAL))
if depart_terminal:
result['depart_terminal'] = depart_terminal
else:
self.logger.error('depart_terminal is empty %s', self._message_id)
arrive_station = strip_str(
take_first(self._etree, expedia_flight_xp.ARRIVE_STATION))
if arrive_station:
result['arrive_station'] = arrive_station
else:
self.logger.error('arrive_station is empty %s', self._message_id)
arrive_time = strip_str(
take_first(self._etree, expedia_flight_xp.ARRIVE_TIME))
if arrive_time:
result['arrive_time'] = arrive_time
else:
self.logger.error('arrive_time is empty %s', self._message_id)
arrive_terminal = strip_str(
take_first(self._etree, expedia_flight_xp.ARRIVE_TERMINAL))
if arrive_terminal:
result['arrive_terminal'] = arrive_terminal
else:
self.logger.error('arrive_terminal is empty %s', self._message_id)
if arrive_station \
and depart_station \
and arrive_time \
and arrive_date \
and depart_time \
and depart_date:
tz_depart = to_timezone(depart_station)
tz_arrive = to_timezone(arrive_station)
depart_datetime = ' '.join([depart_date, depart_time])
arrive_datetime = ' '.join([arrive_date, arrive_time])
depart_datetime = \
DateTime(depart_datetime, # TODO: (LensHo) datetime formatted
'MMM DD, YYYY h:mma').tz_to_datetime(tz_depart)
if depart_datetime:
result['depart_datetime_formatted'] = depart_datetime
else:
self.logger.error('depart_datetime_formatted is empty %s',
self._message_id)
arrive_datetime = \
DateTime(arrive_datetime,
'MMM DD, YYYY h:mma').tz_to_datetime(tz_arrive)
if arrive_datetime:
result['arrive_datetime_formatted'] = arrive_datetime
else:
self.logger.error('arrive_datetime_formatted is empty %s',
self._message_id)
cabin = strip_str(take_first(self._etree, expedia_flight_xp.CABIN))
if cabin:
result['cabin'] = cabin
else:
self.logger.error('cabin is empty %s', self._message_id)
flight_info_divs = self._etree.xpath(
expedia_flight_xp.FLIGHT_INFO_DIVS)
if flight_info_divs:
name_dict = {
'Included': 'included',
'Fee applies': 'fee_applies',
'Not included': 'not_included'
}
for i in flight_info_divs:
name = take_first(i, expedia_flight_xp.FLIGHT_INFO_DIV_NAME)
value = i.xpath(expedia_flight_xp.FLIGHT_INFO_DIV_VALUE)
if name in name_dict.keys():
result[name_dict[name]] = value
else:
self.logger.warning('%s is %s %s', name, value,
self._message_id)
flight_duration = take_first(self._etree,
expedia_flight_xp.FLIGHT_DURATION)
flight_duration = strip_str(flight_duration)
if flight_duration:
result['flight_duration'] = flight_duration
else:
self.logger.error('flight_duration is empty %s', self._message_id)
guest_name = strip_str(take_first(self._etree,
expedia_flight_xp.GUEST))
if guest_name:
result['guest_name'] = guest_name
else:
self.logger.error('guest_name is empty %s', self._message_id)
price = strip_list(self._etree.xpath(expedia_flight_xp.PRICE))
if price:
name_dict = {'Flight': 'price', 'Taxes & Fees': 'taxes_fee'}
for i, j in zip(price[::2], price[1::2]):
if 'Traveler' in i:
continue
if i in name_dict.keys():
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('price is empty %s', self._message_id)
total = strip_str(take_first(self._etree, expedia_flight_xp.TOTAL))
if total:
result['total_cost'] = total
else:
self.logger.error('total_cost is empty %s', self._message_id)
united_restrictions = self._etree.xpath(
expedia_flight_xp.UNITED_RESTRICTIONS)
united_restrictions = strip_list(
[remove_tags(etree.tostring(i)) for i in united_restrictions])
res_des = strip_list(
self._etree.xpath(expedia_flight_xp.UNITED_RESTRICTIONS_DES))
if united_restrictions:
if res_des:
des_num = range(len(res_des))
name_dict = {str(k + 1): v for k, v in zip(des_num, res_des)}
restrictions = []
for i in united_restrictions:
if i[-1].isdigit() and res_des:
restrictions.append({
'text': i,
'explanation': name_dict[i[-1]]
})
else:
restrictions.append({'text': i})
if restrictions:
result['united_restrictions'] = restrictions
else:
self.logger.error('united_res is empty %s', self._message_id)
else:
self.logger.error('united_restrictions is empty %s',
self._message_id)
notice = strip_list(self._etree.xpath(expedia_flight_xp.NOTICE))
if notice:
result['notice'] = notice
else:
self.logger.error('notice is empty %s', self._message_id)
airline_rule = strip_str(
take_first(self._etree, expedia_flight_xp.AIRLINE_RULE))
if airline_rule:
result['airline_rule'] = airline_rule
else:
self.logger.error('airline_rule is empty %s', self._message_id)
tel = strip_str(take_first(self._etree, expedia_flight_xp.PHONE))
if tel:
result['telephone'] = tel.split(' ')[-1].strip('.')
else:
self.logger.error('telephone is empty %s', self._message_id)
return result
def _parse(self): # pylint: disable=R0912, R0915
result = self._order
confirm_number = strip_str(
take_first(self._etree, united_xp.CONFIRM_NUM))
if confirm_number:
result['confirm_code'] = confirm_number
else:
self.logger.error('conform_number is empty %s', self._message_id)
traveler_info_values = self._etree.xpath(
united_xp.TRAVELER_INFO_VALUES)
if traveler_info_values:
result['guest_name'] = traveler_info_values[0]
result['ticket_number'] = traveler_info_values[1]
if len(traveler_info_values) == 3:
result['seats'] = traveler_info_values[2]
elif len(traveler_info_values) == 4:
result['frequent_flyer'] = traveler_info_values[2]
result['seats'] = traveler_info_values[3]
else:
self.logger.error('traveler info increase %s',
self._message_id)
else:
self.logger.error('traveler info is empty %s', self._message_id)
names = [
'flight_date', 'flight_number', 'class', 'depart_city',
'arrive_city', 'plane_model'
]
flight_info_divs = self._etree.xpath(united_xp.FLIGHT_INFO_VALUES)
if flight_info_divs:
for i, j in zip(names, flight_info_divs):
val = j.xpath('./p/span/text() | ./span/text()')
if val:
if i == 'flight_date' and len(val) == 2:
result['depart_date'] = val[0]
result['arrive_date'] = val[1]
elif i == 'flight_date' and len(val) == 1:
result['arrive_date'] = result['depart_date'] = val[0]
else:
result[i] = ' '.join(val)
else:
self.logger.error('flight info is empty %s', self._message_id)
flight_time = self._etree.xpath(united_xp.FLIGHT_TIME)
if len(flight_time) == 2:
result['depart_time'] = flight_time[0]
result['arrive_time'] = flight_time[1]
else:
self.logger.error('flight_time is empty %s', self._message_id)
if (result.get('depart_date') and result.get('depart_time')
and result.get('depart_city')):
dt_depart = ' '.join(
[result['depart_date'], result['depart_time']])
tz = to_timezone(result['depart_city'])
depart_formatted = DateTime(dt_depart,
'DDMMMYY h:mm A').tz_to_datetime(tz)
if depart_formatted:
result['depart_datetime_formatted'] = depart_formatted
else:
self.logger.error('depart_datetime_formatted is empty %s',
self._message_id)
else:
self.logger.error('depart_datetime_formatted is empty %s',
self._message_id)
if (result.get('arrive_date') and result.get('arrive_time')
and result.get('arrive_city')):
dt_arrive = ' '.join(
[result['arrive_date'], result['arrive_time']])
tz = to_timezone(result['arrive_city'])
arrive_formatted = DateTime(dt_arrive,
'DDMMMYY h:mm A').tz_to_datetime(tz)
if arrive_formatted:
result['arrive_datetime_formatted'] = arrive_formatted
else:
self.logger.error('arrive_datetime_formatted is empty %s',
self._message_id)
else:
self.logger.error('arrive_datetime_formatted is empty %s',
self._message_id)
currency = strip_str(take_first(self._etree, united_xp.CURRENCY))
airfare = strip_str(take_first(self._etree, united_xp.AIRFARE_PRICE))
if airfare:
result['price'] = ' '.join([airfare, currency])
else:
self.logger.error('airfare is empty %s', self._message_id)
tax_fee_price = strip_str(
take_first(self._etree, united_xp.TAX_FEE_PRICE))
if tax_fee_price:
result['taxes_fee'] = tax_fee_price
else:
self.logger.error('taxes fee is empty %s', self._message_id)
per_person = strip_str(
take_first(self._etree, united_xp.TOTAL_PER_PERSON))
if per_person:
result['subtotal'] = ' '.join([per_person, currency])
else:
self.logger.error('per person is empty %s', self._message_id)
fare_total = strip_str(take_first(self._etree, united_xp.TOTAL_PRICE))
if fare_total:
result['total_cost'] = ' '.join([fare_total, currency])
else:
self.logger.error('total fare is empty %s', self._message_id)
operator = strip_str(take_first(self._etree, united_xp.OPERATED))
if operator:
result['operator'] = operator
else:
self.logger.error('operator is empty %s', self._message_id)
payment_form = strip_str(
take_first(self._etree, united_xp.PAYMENT_FORM))
if payment_form:
result['payment_form'] = payment_form
else:
self.logger.error('payment_form is empty %s', self._message_id)
tax_summary = strip_str(take_first(self._etree, united_xp.TAX_SUMMARY))
if tax_summary:
result['tax_summary'] = tax_summary
else:
self.logger.error('taxes_summary is empty %s', self._message_id)
fare_summary = strip_str(
take_first(self._etree, united_xp.FARE_SUMMARY))
if fare_summary:
result['fare_summary'] = fare_summary
else:
self.logger.error('fare_summary is empty %s', self._message_id)
fare_rule = strip_str(take_first(self._etree, united_xp.FARE_RULE))
if fare_rule:
result['fare_rule'] = fare_rule
else:
self.logger.error('fare_rule is empty %s', self._message_id)
bag_fee_1 = take_first(self._etree, united_xp.BAG_FEE_1)
bag_fee_2 = take_first(self._etree, united_xp.BAG_FEE_2)
bag_fee_3 = take_first(self._etree, united_xp.BAG_FEE_3)
bag_fee = strip_list([bag_fee_1, bag_fee_2, bag_fee_3])
if bag_fee:
result['bag_fee_summary'] = bag_fee
else:
self.logger.error('bag_fee_summary is empty %s', self._message_id)
bag_names = self._etree.xpath(united_xp.BAG_FEE_TABLE_KEYS)
bag_table_names = []
if bag_names:
bag_table_values = self._etree.xpath(
united_xp.BAG_FEE_TABLE_VALUES)
for i, j in enumerate(bag_names):
s = remove_tags(etree.tostring(j))
if i == 3:
s = strip_str(take_first(self._etree, united_xp.MAX_WT))
bag_table_names.append(s)
table = filter_dict_value(
to_dict(bag_table_names, bag_table_values))
else:
table = ''
self.logger.error('bag_fee_table is empty %s', self._message_id)
if table:
result['bag_fee_table'] = table
marketing_info_trs = self._etree.xpath(united_xp.MARKETING_INFO)
if marketing_info_trs:
titles = marketing_info_trs[1::3]
contents = marketing_info_trs[2::3]
titles_list = [
strip_str(take_first(i, united_xp.MARKETING_INFO_TITLES))
for i in titles
]
name_dict = {
'Important Information about MileagePlus Earning':
'mileage_plus_earning',
'International eTicket Reminders': 'eticket_reminders',
'Customer Care Contact Information': 'customer_care',
'Refunds Within 24 Hours': 'refund',
'Hazardous materials': 'hazardous_materials',
'Proud Member of Star Alliance': 'star_alliance'
}
for i, j in zip(titles_list, contents):
s = j.xpath(united_xp.MARKETING_INFO_CONTENTS)
if s:
a = []
for k in s:
links = k.xpath('.//a/@href')
link_texts = k.xpath('.//a/text()')
texts = strip_str(remove_tags(etree.tostring(k)))
if links:
a.append({
'text': texts,
'links': {
'name': link_texts,
'value': links
}
})
else:
a.append({'text': texts})
result[name_dict[i]] = a
else:
self.logger.error('marketing_info is empty %s', self._message_id)
notice_titles = strip_list(self._etree.xpath(united_xp.NOTICES_TITLE))
notice_contents = strip_list(
self._etree.xpath(united_xp.NOTICE_CONTENTS))
last_advice = strip_str(take_first(self._etree, united_xp.LAST_ADVICE))
name_dict = {
'Notice of Baggage Liability Limitations':
'baggage_liability_limit',
'Notice of Incorporated Terms': 'incorporated_terms',
'Notice of Certain Terms': 'certain_terms',
'Notice of Boarding Times': 'boarding_times',
'Advice to International Passengers on Carrier Liability':
'carrier_liability',
'Notice - Overbooking of Flights': 'overbooking'
}
if notice_contents and notice_titles:
notice_contents[-1] += last_advice
for i, j in zip(notice_titles, notice_contents):
if i in name_dict.keys():
result[name_dict[i]] = j.strip('- ')
else:
self.logger.warning('notice %s is %s %s', i, j,
self._message_id)
else:
self.logger.error('notice is empty %s', self._message_id)
return result
def _parse(self): # pylint: disable=R0912,R0915
result = self._order
hotel_name = strip_str(take_first(self._etree, ihg_xp.HOTEL_NAME))
if hotel_name:
result['hotel_name'] = hotel_name
else:
self.logger.error('hotel_name is empty %s', self._message_id)
confirm_num = strip_str(take_first(self._etree, ihg_xp.CONFIRM_NUM))
if confirm_num:
result['confirm_code'] = confirm_num
else:
self.logger.error('confirm_number is empty %s', self._message_id)
address = ', '.join(strip_list(self._etree.xpath(ihg_xp.ADDRESS)))
if address:
result['address'] = address
else:
self.logger.error('address is empty %s', self._message_id)
telephone = strip_str(take_first(self._etree, ihg_xp.PHONE))
if telephone:
result['telephone'] = telephone.split(': ')[-1]
else:
self.logger.error('telephone is empty %s', self._message_id)
tz = to_timezone(address)
check_in_date = strip_str(take_first(self._etree,
ihg_xp.CHECK_IN_DATE))
if check_in_date:
result['check_in_date'] = check_in_date
else:
self.logger.error('check_in_date is empty %s', self._message_id)
check_out_date = strip_str(
take_first(self._etree, ihg_xp.CHECK_OUT_DATE))
if check_out_date:
result['check_out_date'] = check_out_date
else:
self.logger.error('check_out_date is empty %s', self._message_id)
check_in_time = strip_str(take_first(self._etree,
ihg_xp.CHECK_IN_TIME))
if check_in_time:
result['check_in_time'] = check_in_time
else:
self.logger.error('check_in_time is empty %s', self._message_id)
check_out_time = strip_str(
take_first(self._etree, ihg_xp.CHECK_OUT_TIME))
if check_out_time:
result['check_out_time'] = check_out_time
else:
self.logger.error('check_out_time is empty %s', self._message_id)
check_in_date_formatted = DateTime(check_in_date,
'MM/DD/YY').tz_to_datetime(tz) #
if check_in_date_formatted:
result['check_in_date_formatted'] = check_in_date_formatted
check_out_date_formatted = DateTime(check_out_date,
'MM/DD/YY').tz_to_datetime(tz)
if check_out_date_formatted:
result['check_out_date_formatted'] = check_out_date_formatted
guest_name = strip_str(take_first(self._etree, ihg_xp.GUEST_NAME))
if guest_name:
result['guest_name'] = guest_name
else:
self.logger.error('guest_name is empty %s', self._message_id)
room_num = strip_str(take_first(self._etree, ihg_xp.ROOMS))
guest_num = strip_str(take_first(self._etree, ihg_xp.ADULTS))
room_type = strip_str(take_first(self._etree, ihg_xp.ROOM_TYPE))
if room_num:
result['number_of_rooms'] = room_num
else:
self.logger.error('number of rooms is empty %s', self._message_id)
if guest_num:
result['number_of_guests'] = guest_num
else:
self.logger.error('number of guests is empty %s', self._message_id)
if room_type:
result['room_type'] = room_type
else:
self.logger.error('room_type is empty %s', self._message_id)
price_des = strip_str(take_first(self._etree, ihg_xp.RATE_NOTICE))
if price_des:
result['price_tips'] = [price_des]
else:
self.logger.error('price_description is empty %s',
self._message_id)
rate_names = strip_list(self._etree.xpath(ihg_xp.RATE_NAME))
rate_values = strip_list(self._etree.xpath(ihg_xp.RATE_VALUE))
if rate_values and rate_names:
_, price = rate_names.pop(0), rate_values.pop(0) # noqa
result['price'] = price
name_dict = {
'Service Charge': 'service_charge',
'Total Taxes': 'taxes_fee',
'Estimated Total Price': 'total_cost'
}
for i, j in zip(rate_names, rate_values):
if i in name_dict:
result[name_dict[i]] = j
else:
self.logger.warning(' %s is %s %s', i, j, self._message_id)
cancellation_name = strip_str(
take_first(self._etree, ihg_xp.CANCELLATION_NAME))
cancellation_value = strip_str(
take_first(self._etree, ihg_xp.CANCELLATION_VALUE))
if 'Cancellation' in cancellation_name and cancellation_value:
result['cancellation_policy'] = cancellation_value
else:
self.logger.error('cancellation_policy is empty %s',
self._message_id)
rate_des_name = strip_str(take_first(self._etree,
ihg_xp.RATE_DES_NAME))
rate_des_value = strip_str(
take_first(self._etree, ihg_xp.RATE_DES_VALUE))
if 'Rate' in rate_des_name and rate_des_value:
result['rate_tip'] = rate_des_value
else:
self.logger.error('rate_tip is empty %s', self._message_id)
hotel_link = take_first(self._etree, ihg_xp.HOTEL_LINK)
if hotel_link:
result['hotel_link'] = hotel_link
else:
self.logger.error('hotel_link is empty %s', self._message_id)
preference_link = take_first(self._etree, ihg_xp.PREFERENCE_LINK)
if preference_link:
result['preference_link'] = preference_link
else:
self.logger.error('preference_link is empty %s', self._message_id)
related_links = self._etree.xpath(ihg_xp.RELATED_LINK)
if related_links:
related = []
for i in related_links:
link = i.xpath('./a/@href')
text = i.xpath('./a/text()')
if link and text:
if 'CANCEL' in text[0]:
result['cancellation_link'] = link
elif 'MODIFY' in text[0]:
result['modify_link'] = link
elif 'DOWNLOAD' in text[0]:
pass
else:
related.append({'name': text[0], 'value': link})
if related:
result['related_links'] = related
else:
self.logger.error('related_links is empty %s',
self._message_id)
notice = self._etree.xpath(ihg_xp.NOTICE)
if notice:
notice = notice[0].xpath('./text() | ./*')
a = []
name_dict = {
'Early Departure Fee:': 'early_departure_fee',
'Daily Valet Parking Fee:': 'daily_valet_parking_fee',
'Pet Policy:': 'pet_policy',
'Payment Card Authorization Form:': 'payment_card_auth_form'
}
for i in notice:
if not isinstance(i, str) and i.tag == 'br':
continue
elif not isinstance(i, str) and i.tag == 'span':
if a and not a[-1]['links'] == 0:
del a[-1]['links']
span = i.xpath('./span/text()')
if span:
try:
name = name_dict[span[0]]
b = {'text': '', 'links': []}
for j in i.xpath('./text() | ./*'):
if isinstance(j, str):
b['text'] = ''.join([b['text'], j])
elif not isinstance(j, str) and j.tag == 'a':
b['text'] = ''.join([b['text'], j.text])
c = {
'name': j.text,
'value': j.get('href')
}
b['links'].append(c)
else:
pass
result[name] = b
a.append(b)
except KeyError as e:
self.logger.exception(e)
self.logger.warning('unknown key %s %s', span[0],
self._message_id)
else:
try:
name = name_dict[i.text]
b = {'text': '', 'links': []}
result[name] = b
a.append(b)
except KeyError as exc:
self.logger.exception(exc)
self.logger.warning('unknown key %s %s',
name_dict[i.text],
self._message_id)
elif isinstance(i, str):
if a:
a[-1]['text'] = ''.join([a[-1]['text'], i])
else:
self.logger.error('string before span %s %s', i,
self._message_id)
else:
self.logger.warning('unknown tag %s %s', i.tag,
self._message_id)
if a and not a[-1]['links']:
del a[-1]['links']
else:
self.logger.error('notice is empty %s', self._message_id)
tip = take_first(self._etree, ihg_xp.ADDITION_FEE)
if tip:
result['notice'] = tip
else:
self.logger.warning('hotel_information_notice is empty %s',
self._message_id)
return result
def parse(self):
result = self._order
hotel_name = strip_str(take_first(self._etree, hyatt_xp.HOTEL_NAME))
if hotel_name:
result['hotel_name'] = hotel_name
else:
self.logger.error('hotel_name is empty %s', self._message_id)
confirm_num = strip_str(take_first(self._etree, hyatt_xp.CONFIRM_NUM))
if confirm_num:
confirm_num = confirm_num.split(': ')[-1]
result['confirm_code'] = confirm_num
else:
self.logger.error('conform_number is empty %s', self._message_id)
address = ', '.join(strip_list(self._etree.xpath(hyatt_xp.ADDRESS)))
if address:
result['address'] = address
else:
self.logger.error('address is empty %s', self._message_id)
telephone = strip_str(take_first(self._etree, hyatt_xp.PHONE))
if telephone:
result['telephone'] = telephone
else:
self.logger.error('telephone is empty %s', self._message_id)
check_in_out_datetime = strip_list(
self._etree.xpath(hyatt_xp.CHECK_IN_OUT_DATETIME))
check_in_date, check_in_time, check_out_date, check_out_time = \
unpack(check_in_out_datetime, 4)
if check_out_date and check_in_date:
tz = to_timezone(address)
check_in_date = remove_space(check_in_date)
check_out_date = remove_space(check_out_date)
result['check_in_date'] = check_in_date
result['check_out_date'] = check_out_date
check_in_date_formatted = DateTime(
check_in_date,
'MMMM DD, YYYY'
) \
.tz_to_datetime(tz)
if check_in_date_formatted:
result['check_in_date_formatted'] = check_in_date_formatted
else:
self.logger.error('check_in_date_formatted is empty %s',
self._message_id)
check_out_date_formatted = DateTime(
check_out_date,
'MMMM DD, YYYY'
) \
.tz_to_datetime(tz)
if check_out_date_formatted:
result['check_out_date_formatted'] = check_out_date_formatted
else:
self.logger.error('check_out_date_formatted is empty %s',
self._message_id)
else:
self.logger.error('check_in_date and check_out_date is empty %s',
self._message_id)
if check_in_time and check_out_time:
result['check_in_time'] = check_in_time
result['check_out_time'] = check_out_time
else:
self.logger.error('check_in_time and check_out_time is empty %s',
self._message_id)
related_links = self._etree.xpath(hyatt_xp.RELATED_LINKS)
related_text = strip_list(
self._etree.xpath(hyatt_xp.RELATED_LINKS_TEXT))
plan_link = self._etree.xpath(hyatt_xp.PLAN_LINK)
plan_text = strip_list(self._etree.xpath(hyatt_xp.PLAN_TEXT))
if not related_links and related_text:
self.logger.error('related_links is empty %s', self._message_id)
if not plan_text and plan_link:
self.logger.error('plan_links is empty %s', self._message_id)
related_links = chain(related_links or [], plan_link or [])
related_texts = chain(related_text, plan_text)
if related_links and related_texts:
name_dict = {
'Cancel Reservation': 'cancellation_link',
'Modify Reservation': 'modify_link',
'Get to know our hotel': 'hotel_link',
'Add Reservation': 'add_reservation_link',
'Customer Service': 'customer_service_link'
}
for i, j in zip(related_texts, related_links):
if i in name_dict:
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('related_links is empty %s', self._message_id)
reservation_info = self._etree.xpath(hyatt_xp.RESERVATION_INFO)
if reservation_info:
note_1 = reservation_info.pop(0)
note_1 = strip_list(
note_1.xpath('./table/tr/td/table/tr[2]/td/text()'))
note_2 = reservation_info.pop()
note_2 = strip_list(note_2.xpath('.//td/text()'))
note = list(chain([note_1, note_2]))
if note:
result['notice'] = note
else:
self.logger.error('notice is empty %s', self._message_id)
cancellation = reservation_info.pop()
cancellation = cancellation.xpath(
'./table/tr/td/table/tr[2]/td/text()')
cancellation = strip_list(cancellation)
if cancellation:
result['cancellation_policies'] = cancellation
else:
self.logger.error('policies is empty %s', self._message_id)
name_dict = {
'Guest Name': 'guest_name',
'Number of Adults': 'number_of_guests',
'Room(s) Booked': 'number_of_rooms',
'Room Type': 'room_type',
'Type of Rate': 'rate_type',
'SERVICE CHARGE': 'service_charge',
'SALES TAX': 'taxes_fee',
}
for i in reservation_info:
title = strip_str(take_first(i, './b/text()'))
value = '\n'.join(strip_list(i.xpath('./text()')))
if title and value:
if 'Room Description' in title:
result['room_tips'] = [value.strip('- ')]
elif 'Rate Information' in title:
result['rate_tips'] = remove_space(value)
elif 'Policies' in title:
result['policies'] = strip_list(value.split('\n'))
elif title in name_dict:
result[name_dict[title]] = value
else:
self.logger.warning('%s is %s %s', title, value,
self._message_id)
else:
self.logger.error('reservation_information is empty %s',
self._message_id)
return result
def _parse(self): # pylint: disable=R0912,R0915
result = self._order
related_link = self._etree.xpath(spg_xp.RELATED_LINKS)
related_text = strip_list(self._etree.xpath(spg_xp.RELATED_TEXT))
if related_link:
result['related_links'] = to_dict(related_text, related_link)
else:
self.logger.error('related_links is empty %s', self._message_id)
hotel_name = strip_str(take_first(self._etree, spg_xp.HOTEL_NAME))
if hotel_name:
result['hotel_name'] = hotel_name
else:
self.logger.error('hotel_name is empty %s', self._message_id)
address = self._etree.xpath(spg_xp.ADDRESS)
if address and len(address) >= 3:
address = remove_space(' '.join(address[:2]).replace('\t', ''))
result['address'] = address
else:
address = ''
self.logger.error('address is empty %s', self._message_id)
tel = strip_str(take_first(self._etree, spg_xp.PHONE))
if tel:
result['telephone'] = tel
else:
self.logger.error('telephone is empty %s', self._message_id)
fax = strip_str(take_first(self._etree, spg_xp.FAX))
if fax:
result['fax'] = fax.split(': ')[-1]
else:
self.logger.warning('fax is empty %s', self._message_id)
confirm_num = strip_list(self._etree.xpath(spg_xp.ADDRESS))
if confirm_num and len(confirm_num) >= 4:
result['confirm_code'] = confirm_num[3]
else:
self.logger.error('confirm_number is empty %s', self._message_id)
reservation_explanation = strip_str(
take_first(self._etree, spg_xp.EXPLANATION)
)
if reservation_explanation:
result['tip'] = reservation_explanation
else:
self.logger.warning(
'reservation_explanation is empty %s', self._message_id
)
reservation_keys = strip_list(
self._etree.xpath(spg_xp.RESERVATION_INFO_KEYS)
)
reservation_values = strip_list(
self._etree.xpath(spg_xp.RESERVATION_INFO_VALUES)
)
if reservation_keys and reservation_values:
tz = to_timezone(address)
for i, j in zip(reservation_keys, reservation_values):
if 'Check In' in i:
check_in_date = j.split('\n')[0]
check_in_time = j.split(' - ')[-1].strip(' *')
if check_in_time:
result['check_in_time'] = check_in_time
else:
self.logger.error(
'check_in_time is empty %s', self._message_id
)
if check_in_date:
result['check_in_date'] = check_in_date
formatted = DateTime(check_in_date,
'DD-MMM-YYYY').tz_to_datetime(tz)
if formatted:
result['check_in_date_formatted'] = formatted
else:
self.logger.error(
'check_in_date_formatted is empty %s',
self._message_id
)
else:
self.logger.error(
'check_in_date is empty %s', self._message_id
)
elif 'Check Out' in i:
check_out_date = j.split('\n')[0]
check_out_time = j.split(' - ')[-1].strip(' *')
if check_out_time:
result['check_out_time'] = check_out_time
else:
self.logger.error(
'check_out_time is empty %s', self._message_id
)
if check_out_date:
result['check_out_date'] = check_out_date
formatted = DateTime(check_out_date,
'DD-MMM-YYYY').tz_to_datetime(tz)
if formatted:
result['check_out_date_formatted'] = formatted
else:
self.logger.error(
'check_out_date_formatted is empty %s',
self._message_id
)
else:
self.logger.error(
'check_in_date is empty %s', self._message_id
)
elif 'Number of Guests' in i:
result['number_of_guests'] = j
elif 'Number of Rooms' in i:
result['number_of_rooms'] = j
else:
self.logger.warning(' %s is %s %s', i, j, self._message_id)
else:
self.logger.error(
'reservation_information is empty %s', self._message_id
)
accommodation_keys = strip_list(
self._etree.xpath(spg_xp.ACCOMMODATION_INFO_KEYS)
)
accommodation_values = \
strip_list(self._etree.xpath(spg_xp.ACCOMMODATION_INFO_VALUES))
if accommodation_keys and accommodation_values:
for i, j in zip(accommodation_keys, accommodation_values):
if 'Guest Name' in i:
result['guest_name'] = j # TODO: (LensHo) 多个名字
elif 'Number of Adults' in i:
result['number_of_adults'] = j
elif 'Number of Children' in i:
result['number_of_children'] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('accommodation is empty %s', self._message_id)
accommodation_details = strip_list(
self._etree.xpath(spg_xp.ACCOMMODATION_DETAILS)
)
if accommodation_details:
result['room_tips'] = accommodation_details
else:
self.logger.warning(
'accommodation_details is empty %s', self._message_id
)
room_description = strip_str(
take_last(self._etree, spg_xp.ROOM_DESCRIPTION)
)
if room_description:
bed_type = room_description.split(': ')[-1]
result['bed_type'] = bed_type
room_tips = result.get('room_tips')
if room_tips:
result['room_tips'].append(room_description)
else:
result['room_tips'] = [room_description]
else:
self.logger.warning('room_tip is empty %s', self._message_id)
currency = strip_str(take_first(self._etree, spg_xp.CURRENCY))
if not currency:
self.logger.error('currency is empty %s', self._message_id)
room_rate_avg = strip_str(
take_first(self._etree, spg_xp.PER_ROOM_RATE)
)
if room_rate_avg:
result['price'] = ' '.join([room_rate_avg, currency])
else:
self.logger.error('room_rate_avg is empty %s', self._message_id)
estimated_avg = strip_str(
take_first(self._etree, spg_xp.ESTIMATED_TOTAL_1)
)
if estimated_avg:
result['total_cost'] = ' '.join([estimated_avg, currency])
else:
self.logger.error('total is empty %s', self._message_id)
other_fees_name = strip_list(self._etree.xpath(spg_xp.OTHER_COST_KEYS))
other_fees_avg = strip_list(
self._etree.xpath(spg_xp.OTHER_COST_VALUES_1)
)
if other_fees_name and other_fees_avg:
other_fees_name = other_fees_name[::2]
other_fees_value = map(remove_space, other_fees_name[1::2])
other_fees_avg = other_fees_avg[1::2]
other_fees = zip(other_fees_name, other_fees_value, other_fees_avg)
for name, fee, avg in other_fees:
if 'Value Added Tax' in name:
result['taxes_fee'] = ' '.join([fee, avg, currency])
elif 'Service Charge' in name:
result['service_charge'] = ' '.join([fee, avg, currency])
else:
self.logger.warning(
'%s is %s %s', name, fee + avg + currency,
self._message_id
)
else:
self.logger.error('other_fee is empty %s', self._message_id)
price_explanation = strip_str(
take_first(self._etree, spg_xp.EXPLANATION_PRICE)
)
if price_explanation:
result['price_tips'] = [price_explanation]
else:
self.logger.warning(
'price_explanation is empty %s', self._message_id
)
policies = strip_list(self._etree.xpath(spg_xp.CANCELLATION))
if policies:
result['policies'] = policies
else:
self.logger.error('policies is empty %s', self._message_id)
privacy_links = self._etree.xpath(spg_xp.PRIVACY_LINK)
privacy_texts = strip_list(self._etree.xpath(spg_xp.PRIVACY_LINK_TEXT))
privacy = take_first(self._etree, spg_xp.PRIVACY)
if privacy is not None:
result['privacies'] = [{
'text': strip_str(remove_tags(etree.tostring(privacy))),
'links': to_dict(privacy_texts, privacy_links)
}]
else:
self.logger.error('privacy is empty %s', self._message_id)
disclosure = self._etree.xpath(spg_xp.DISCLOSURE)
if disclosure:
disclosure = disclosure[0].xpath('./text() | ./*')
a = []
for i in disclosure:
if i == '\n' or not isinstance(i, str) and i.tag == 'br':
continue
elif not isinstance(i, str) and i.tag == 'strong':
b = {'name': i.text, 'text': '', 'links': []}
if a and not a[-1]['links']:
del a[-1]['links']
a.append(b)
elif isinstance(i, str):
if a:
a[-1]['text'] = ''.join([a[-1]['text'], i])
else:
self.logger.error(
'string before strong %s %s', i, self._message_id
)
elif i.tag == 'a':
if a:
a[-1]['text'] = ''.join([a[-1]['text'], i.text])
b = {'name': i.text, 'value': i.get('href')}
a[-1]['links'].append(b)
else:
self.logger.error(
'tag before strong %s %s', i.tag, self._message_id
)
else:
self.logger.error(
'unknown tag %s %s', i.tag, self._message_id
)
if a and not a[-1]['links']:
del a[-1]['links']
result['disclosures'] = a
else:
self.logger.error('disclosure is empty %s', self._message_id)
else:
self.logger.error('disclosure_raw is empty %s', self._message_id)
return result
def _parse(self):
self._order.set('guest_name', take_first(self._etree,
wdm_xp.GUEST_NAME))
self._order['confirm_code'] = take_first(self._etree,
wdm_xp.CONFIRM_NUM)
if not self._order['confirm_code']:
raise MissingColumnError('confirm_code', self._message_id)
self._order['hotel_name'] = take_first(self._etree, wdm_xp.HOTEL_NAME)
if not self._order['hotel_name']:
raise MissingColumnError('hotel_name', self._message_id)
address = strip_list(self._etree.xpath(wdm_xp.ADDRESS))
if not address:
raise MissingColumnError('address and telephone', self._message_id)
self._order['telephone'] = address.pop().strip('Phone: ')
self._order['address'] = remove_space(', '.join(address))
if not (self._order['telephone'] and self._order['address']):
raise MissingColumnError('address or telephone', self._message_id)
self._order.set('hotel_link', take_first(self._etree,
wdm_xp.HOTEL_LINK))
self._order.set('email', take_first(self._etree, wdm_xp.EMAIL))
self._order.set('modify_link',
take_first(self._etree, wdm_xp.MODIFY_LINK))
self._order.set('cancellation_link',
take_first(self._etree, wdm_xp.CANCEL_LINK))
reservation = strip_list(self._etree.xpath(wdm_xp.RESERVATION))
if len(reservation) < 5:
raise MissingColumnError('reservation', self._message_id)
self._order.set('room_tips', strip_list(reservation[0].split('\n')))
self._order.set('number_of_rooms', reservation[1].split(';')[0])
self._order.set('number_of_nights', reservation[1].split(';')[1])
self._order.set('number_of_guests', reservation[2])
check_in = reservation[3].split(' ')
self._order['check_in_date'] = check_in.pop(0)
self._order['check_in_time'] = ' '.join(check_in)
if not (self._order['check_in_date'] and self._order['check_in_time']):
raise MissingColumnError('check_in', self._message_id)
checkout = reservation[4].split(' ')
self._order['check_out_date'] = checkout.pop(0)
self._order['check_out_time'] = ' '.join(checkout)
if not (self._order['check_out_date']
and self._order['check_out_time']):
raise MissingColumnError('checkout', self._message_id)
tz = to_timezone(self._order['address'])
self._order['check_in_date_formatted'] = DateTime(
self._order['check_in_date'], 'MM/DD/YYYY').tz_to_datetime(tz)
self._order['check_out_date_formatted'] = DateTime(
self._order['check_out_date'], 'MM/DD/YYYY').tz_to_datetime(tz)
if not (self._order['check_in_date_formatted']
and self._order['check_out_date_formatted']):
raise MissingColumnError('check_in_out_date_formatted',
self._message_id)
price = self._etree.xpath(wdm_xp.PRICE)
if price:
for item in price:
name = strip_str(take_first(item, './td[1]/text()'))
value = strip_str(take_first(item, './td[2]/text()'))
currency = strip_str(take_first(item, './td[2]/span/text()'))
if name in self.name_dict:
self._order[self.name_dict[name]] = value + currency
else:
self.logger.warning('price %s is %s %s', name, value,
self._message_id)
other_info = strip_list(self._etree.xpath(wdm_xp.OTHER_INFO))
for info in other_info:
if info.startswith('Cancellation Policy'):
self._order.set('cancellation_policies',
info.strip('Cancellation Policy: ').split(';'))
elif info.startswith('Payment Method: '):
self._order.set('payment_card', info.strip('Payment Method: '))
else:
self.logger.warning('other information %s %s', info,
self._message_id)
return self._order
def _parse(self): # pylint: disable=R0912,R0915
result = self._order
guest_name = strip_str(take_first(self._etree, accor_xp.GUEST))
if guest_name:
result['guest_name'] = guest_name
else:
self.logger.error('guest_name is empty %s', self._message_id)
confirm_num = strip_str(take_first(self._etree, accor_xp.CONFIRM_NUM))
if confirm_num:
result['confirm_code'] = confirm_num
else:
self.logger.error('confirm_number is empty %s', self._message_id)
address = ', '.join(strip_list(self._etree.xpath(accor_xp.ADDRESS)))
if address:
result['address'] = address
else:
self.logger.error('address is empty %s', self._message_id)
check_in_out_date = strip_str(
take_first(
self._etree,
accor_xp.CHECK_IN_OUT_DATE
)
) \
.split(' ')
if not check_in_out_date:
self.logger.error('check_in_out_date is empty %s',
self._message_id)
if len(check_in_out_date) == 4:
check_in_date = check_in_out_date[1]
check_out_date = check_in_out_date[-1]
result['check_in_date'] = check_in_date
result['check_out_date'] = check_out_date
tz = to_timezone(address)
check_in_date_formatted = DateTime(check_in_date,
'MM/DD/YYYY').tz_to_datetime(tz)
if check_in_date_formatted:
result['check_in_date_formatted'] = check_in_date_formatted
check_out_date_formatted = DateTime(
check_out_date,
'MM/DD/YYYY'
) \
.tz_to_datetime(tz)
if check_out_date_formatted:
result['check_out_date_formatted'] = check_out_date_formatted
hotel_link = take_first(self._etree, accor_xp.HOTEL_LINK)
if hotel_link:
result['hotel_link'] = hotel_link
else:
self.logger.error('hotel_link is empty %s', self._message_id)
map_link = take_first(self._etree, accor_xp.MAP_LINK)
if map_link:
result['map_link'] = map_link
else:
self.logger.error('map_link is empty %s', self._message_id)
related_links = self._etree.xpath(accor_xp.RELATED_LINKS)
if not related_links:
self.logger.error('related_links is empty %s', self._message_id)
related_texts = strip_list(self._etree.xpath(accor_xp.RELATED_TEXTS))
related_links = to_dict(related_texts, related_links)
if related_links:
result['related_links'] = related_links
for i in related_links:
if i.get('name') in 'Cancel':
related_links.remove(i)
result['cancellation_link'] = i.get('value')
break
hotel_name = strip_str(take_first(self._etree, accor_xp.HOTEL_NAME))
if hotel_name:
result['hotel_name'] = hotel_name
else:
self.logger.error('hotel_name is empty %s', self._message_id)
phone_email = strip_list(self._etree.xpath(accor_xp.PHONE_EMAIL))
phone = phone_email.pop(0) if phone_email else ''
email = phone_email.pop(0) if phone_email else ''
if phone:
result['telephone'] = phone.split(': ')[-1]
else:
self.logger.error('telephone is empty %s', self._message_id)
if email:
result['email'] = email
else:
self.logger.warning('email is empty %s', self._message_id)
bed_type = strip_str(take_first(self._etree, accor_xp.BED_TYPE))
if bed_type:
result['bed_type'] = bed_type
else:
self.logger.error('bed_type is empty %s', self._message_id)
room_detail = strip_str(take_first(self._etree, accor_xp.DETAILS))
if room_detail:
result['room_tips'] = [room_detail]
else:
self.logger.error('room_details is empty %s', self._message_id)
stay = strip_str(take_first(self._etree, accor_xp.STAY))
if stay:
days = stay.split(', ')[-1].split(' ')[0]
result['number_of_rooms'] = stay.split(' ')[0]
result['number_of_nights'] = stay.split(', ')[-1]
else:
days = ''
self.logger.error('stay is empty %s', self._message_id)
room_price_values = strip_list(
self._etree.xpath(accor_xp.ROOM_PRICE_VALUE))
if room_price_values:
price_sum = 0
currency = ''
for i in room_price_values:
price_sum += float(i.split(' ')[-1])
currency = i.split(' ')[0] + ' '
if days:
result['price'] = currency + \
str(round(price_sum / int(days), 2))
else:
self.logger.error('price is empty %s', self._message_id)
total_price_names = strip_list(
self._etree.xpath(accor_xp.TOTAL_PRICE_NAME))
total_price_values = strip_list(
self._etree.xpath(accor_xp.TOTAL_PRICE_VALUE))
if len(total_price_values) > 1:
# last two --> total
unused_total_name, total_value = total_price_values[-2:]
result['total_cost'] = total_value
total_price_values.pop(0) # Total price of stay
else:
self.logger.error('total_price is empty %s', self._message_id)
if total_price_names and total_price_values:
for i, j in zip(total_price_names, total_price_values):
if i in 'Total amount including VAT':
result['total_amount_including_vat'] = j
elif i in 'Other taxes excluded':
result['other_taxes_excluded'] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
info_names = strip_list(self._etree.xpath(accor_xp.INFO_NAME))
info_values = self._etree.xpath(accor_xp.INFO_VALUE)
if info_values and info_names:
name_dict = {
'Practical information': 'practical_information',
'Special requirements': 'your_special_requirements',
'Sales conditions': 'sales_conditions',
'Taxes': 'taxes-fee'
}
info_values = map(remove_tags, map(etree.tostring, info_values))
for i, j in zip(info_names, strip_list(info_values)):
if i in name_dict:
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('notice is empty %s', self._message_id)
require = result.get('your_special_requirements')
if require:
result['your_special_requirements'] = [require]
policy_names = strip_list(self._etree.xpath(accor_xp.POLICIES_NAMES))
policy_values = strip_list(self._etree.xpath(accor_xp.POLICIES_VALUES))
if policy_names and policy_values:
policy_values = policy_values[2:]
for i, j in zip(policy_names, policy_values):
if 'Check in Policy' in i:
result['check_in_policies'] = [j]
check_in_time = str(DateTime(j, 'HH:mm').to_time()) # H
if check_in_time:
result['check_in_time'] = check_in_time
elif 'Check out Policy' in i:
result['check_out_policies'] = [j]
check_out_time = str(DateTime(j, 'HH:mm').to_time())
if check_out_time:
result['check_out_time'] = check_out_time
elif 'Cancellation policy' in i:
result['cancellation_policies'] = [j]
elif 'Guarantee Policy' in i:
result['guarantee_policies'] = [j]
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
return result
def _parse(self): # pylint: disable=R0912,R0915
result = self._order
if dt.now().timestamp() / int(self.received_time) < 1:
self.received_time = int(int(self.received_time) / 1000)
record_locator = strip_str(
take_first(self._etree, aa_xp.RECORD_LOCATOR))
if record_locator:
result['record_locator'] = record_locator
else:
self.logger.error('recoder_locator is empty %s', self._message_id)
itinerary_info_guest = strip_list(self._etree.xpath(aa_xp.INFO_2))
if itinerary_info_guest:
result['guest_name'] = itinerary_info_guest[0]
# TODO (LensHo): to be fix
if len(itinerary_info_guest) > 1:
result['meal'] = itinerary_info_guest[-1]
else:
self.logger.warning('afford is empty %s', self._message_id)
else:
self.logger.error('guest_name is empty %s', self._message_id)
itinerary_info_1 = strip_list(self._etree.xpath(aa_xp.INFO_1))
if len(itinerary_info_1) == 9:
result['carrier'] = itinerary_info_1[0]
result['flight_number'] = itinerary_info_1[1]
result['depart_city'] = itinerary_info_1[2]
result['depart_date'] = itinerary_info_1[3]
result['depart_time'] = itinerary_info_1[4]
result['arrive_city'] = itinerary_info_1[5]
result['arrive_date'] = itinerary_info_1[6]
result['arrive_time'] = itinerary_info_1[7]
result['booking_code'] = itinerary_info_1[8]
depart_date_time = ' '.join(
[itinerary_info_1[3], itinerary_info_1[4]])
arrive_date_time = ' '.join(
[itinerary_info_1[6], itinerary_info_1[7]])
tz_depart = to_timezone(result['depart_city'])
tz_arrive = to_timezone(result['arrive_city'])
depart_dt = DateTime(depart_date_time, 'DDMMM h:mm A')
depart_year = depart_dt.received_time_to_year(self.received_time)
depart_datetime_formatted = depart_dt.year_to_datetime(
depart_year, tz_depart)
if depart_datetime_formatted:
result['depart_datetime_formatted'] = depart_datetime_formatted
else:
self.logger.error('depart_datetime_formatted is empty %s',
self._message_id)
arrive_dt = DateTime(arrive_date_time, 'DDMMM h:mm A')
arrive_year = arrive_dt.received_time_to_year(self.received_time)
arrive_datetime_formatted = arrive_dt.year_to_datetime(
arrive_year, tz_arrive)
if arrive_datetime_formatted:
result['arrive_datetime_formatted'] = arrive_datetime_formatted
else:
self.logger.error('arrive_datetime_formatted is empty %s',
self._message_id)
else:
self.logger.error('itinerary_details is empty %s',
self._message_id)
# TODO: (LensHo) might not exist
itinerary_info_2 = strip_list(self._etree.xpath(aa_xp.INFO_3))
itinerary_info_trans = strip_list(self._etree.xpath(aa_xp.INFO_4))
if itinerary_info_trans:
result['transfer_information'] = itinerary_info_trans[0]
else:
self.logger.info('transfer_information is empty %s',
self._message_id)
itinerary_info_afford = strip_list(self._etree.xpath(aa_xp.INFO_5))
if itinerary_info_afford:
result['trans_meal'] = itinerary_info_afford[-1]
if len(itinerary_info_2) == 9:
depart_date_time = ' '.join(
[itinerary_info_2[3], itinerary_info_2[4]])
arrive_date_time = ' '.join(
[itinerary_info_2[6], itinerary_info_2[7]])
result['trans_carrier'] = itinerary_info_2[0]
result['trans_flight_number'] = itinerary_info_2[1]
result['trans_depart_city'] = itinerary_info_2[2]
result['trans_depart_date'] = itinerary_info_2[3]
result['trans_depart_time'] = itinerary_info_2[4]
result['trans_arrive_city'] = itinerary_info_2[5]
result['trans_arrive_date'] = itinerary_info_2[6]
result['trans_arrive_time'] = itinerary_info_2[7]
result['trans_booking_code'] = itinerary_info_2[8]
tz_depart = to_timezone(result['trans_depart_city'])
tz_arrive = to_timezone(result['trans_arrive_city'])
depart_dt = DateTime(depart_date_time, 'DDMMM h:mm A')
depart_year = depart_dt.received_time_to_year(self.received_time)
depart_datetime_formatted_2 = depart_dt.year_to_datetime(
depart_year, tz_depart)
if depart_datetime_formatted_2:
result['trans_depart_datetime_formatted'] = \
depart_datetime_formatted_2
else:
self.logger.error(
'trans_depart_datetime_formatted is empty %s',
self._message_id)
arrive_dt = DateTime(arrive_date_time, 'DDMMM h:mm A')
arrive_year = arrive_dt.received_time_to_year(self.received_time)
arrive_datetime_formatted_2 = arrive_dt.year_to_datetime(
arrive_year, tz_arrive)
if arrive_datetime_formatted_2:
result['trans_arrive_datetime_formatted'] = \
arrive_datetime_formatted_2
else:
self.logger.error(
'trans_arrive_datetime_formatted is empty %s',
self._message_id)
else:
self.logger.info('transfer information is empty %s',
self._message_id)
receipt_info = self._etree.xpath(aa_xp.INFO_6)
if receipt_info:
name_dict = {
'TICKET NUMBER': 'ticket_number',
'FARE-SGD': 'fare_sgd',
'EQUIV FARE-EUR': 'fare_equal_to_eur',
'TAXES AND CARRIER-IMPOSED FEES': 'taxes_fee'
}
for i in receipt_info[0]:
name = take_first(i, './/strong/text()')
value = take_first(i, './/td/text()')
if value and name in name_dict.keys():
result[name_dict[name]] = value
elif name in ['TICKET TOTAL', 'PASSENGER']:
pass
else:
self.logger.warning(
'%s: passengers is empty',
self._message_id,
)
else:
self.logger.error('receipt_info is empty %s', self._message_id)
total_price = strip_str(take_first(self._etree, aa_xp.TOTAL))
if total_price:
result['total_cost'] = total_price
else:
self.logger.error('price_details is empty %s', self._message_id)
notice_1 = take_first(self._etree, aa_xp.NOTICE_1)
notice_3 = take_first(self._etree, aa_xp.NOTICE_3)
notice_1 = notice_1 and strip_str(remove_tags(
etree.tostring(notice_1)))
if not notice_1:
self.logger.warning('part1 of notice is empty %s',
self._message_id)
notice_2 = strip_str(take_first(self._etree, aa_xp.NOTICE_2))
if not notice_2:
self.logger.warning('part2 of notice is empty %s',
self._message_id)
notice_3 = strip_str(remove_tags(etree.tostring(notice_3))) \
if notice_3 else ''
if not notice_3:
self.logger.warning('part3 of notice is empty %s',
self._message_id)
if notice_1 or notice_2 or notice_3:
result['notice'] = strip_list([notice_1, notice_2, notice_3])
related_links_1 = self._etree.xpath(aa_xp.RELATED_LINK_1)
if not related_links_1:
self.logger.warning('part1 of related_links is empty %s',
self._message_id)
related_links_3 = self._etree.xpath(aa_xp.RELATED_LINK_3)
if not related_links_3:
self.logger.warning('part2 of related_links is empty %s',
self._message_id)
related_texts_1 = strip_list(self._etree.xpath(aa_xp.RELATED_TEXT_1))
related_texts_3 = strip_list(self._etree.xpath(aa_xp.RELATED_TEXT_3))
links = chain(related_links_1 or [], related_links_3 or [])
texts = chain(related_texts_1 or [], related_texts_3 or [])
name_dict = {
'Check-In Options': 'check_in_options_link',
'Baggage Information.': 'baggage_information_link',
'U.S. Entry Requirements.': 'us_entry_requirements_link',
'Contact American.': 'contact_american_link',
'Worldwide Phone Numbers': 'worldwide_phone_numbers_link',
'Conditions of Carriage': 'conditions_of_carriage_link'
}
for i, j in zip(texts, links):
if i in name_dict:
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
return result
def _parse(self):
result = self._order
restaurant_name = take_first(self._etree, opentable_xp.NAME)
if restaurant_name:
result['restaurant_name'] = restaurant_name
else:
self.logger.error('restaurant_name is empty %s', self._message_id)
address = ', '.join(strip_list(self._etree.xpath(
opentable_xp.ADDRESS)))
if restaurant_name:
result['address'] = address
else:
self.logger.error('address is empty %s', self._message_id)
guest_name = take_first(self._etree, opentable_xp.USER_NAME)
if guest_name:
result['guest_name'] = guest_name
else:
self.logger.error('guest_name is empty %s', self._message_id)
confirm_number = take_first(self._etree,
opentable_xp.CONFIRMATION_NUMBER)
if confirm_number:
result['confirm_code'] = confirm_number
else:
self.logger.error('confirm_number is empty %s', self._message_id)
check_in_datetime = take_first(self._etree, opentable_xp.DATETIME)
if check_in_datetime:
result['check_in_datetime'] = check_in_datetime
tz = to_timezone(address)
dt = DateTime(check_in_datetime, 'MMMM D, YYYY .. h:mm A')
check_in_datetime_formatted = dt.tz_to_datetime(tz)
if check_in_datetime_formatted:
result['check_in_datetime_formatted'] = \
check_in_datetime_formatted
else:
self.logger.error('check_in_datetime_formatted is empty %s',
self._message_id)
else:
self.logger.error('check_in_datetime is empty %s',
self._message_id)
tel = take_first(self._etree, opentable_xp.PHONE)
if tel:
result['telephone'] = tel
else:
self.logger.error('telephone is empty %s', self._message_id)
title = take_first(self._etree, opentable_xp.PROMOTED_OFFER_TITLE)
if not title:
self.logger.warning('promote_offer_title is empty %s',
self._message_id)
details = take_first(self._etree, opentable_xp.PROMOTED_OFFER_DETAILS)
if not details:
self.logger.warning('promote_offer_details is empty %s',
self._message_id)
condition = take_first(self._etree,
opentable_xp.PROMOTED_OFFER_CONDITIONS)
if not condition:
self.logger.warning('promote_offer_condition is empty %s',
self._message_id)
promote_offer_values = [title, details, condition]
promote_offer_names = ['title', 'details', 'conditions']
promote_offer = to_dict(promote_offer_names, promote_offer_values)
promote_offer = filter_dict_value(promote_offer)
if promote_offer:
result['promote_offer'] = promote_offer
menu = take_first(self._etree, opentable_xp.MENU)
if menu:
result['menu_link'] = menu
else:
self.logger.warning('menu_link is empty %s', self._message_id)
direction = take_first(self._etree, opentable_xp.DIRECTION)
if direction:
result['map_link'] = direction
else:
self.logger.warning('direction_link is empty %s', self._message_id)
restaurant_link = take_first(self._etree, opentable_xp.NAME_LINK)
if restaurant_link:
result['restaurant_link'] = restaurant_link
else:
self.logger.warning('restaurant_link is empty %s',
self._message_id)
links = self._etree.xpath(opentable_xp.LINK)
links_name = strip_list(self._etree.xpath(opentable_xp.LINK_NAME))
name_dict = {
'Cancel': 'cancellation_link',
'Modify': 'modify_link',
'Calendar': 'calendar_link',
'Share': 'share_link'
}
if links and links_name:
for i, j in zip(links_name, links):
if i in name_dict.keys():
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('related_links is empty %s', self._message_id)
confirmation = '\n'.join(
self._etree.xpath(opentable_xp.RESTAURANT_CONFIRMATION))
if confirmation:
result['restaurant_confirmation'] = confirmation
else:
self.logger.warning('restaurant_confirmation is empty %s',
self._message_id)
restaurant_details = strip_str(
take_first(self._etree, opentable_xp.RESTAURANT_DETAILS))
if restaurant_details:
result['restaurant_details'] = restaurant_details
else:
self.logger.warning('restaurant_details is empty %s',
self._message_id)
notice = strip_list(self._etree.xpath(opentable_xp.NOTE))
if notice:
result['notice'] = [i.split(': ')[-1] for i in notice]
else:
self.logger.warning('notice is empty %s', self._message_id)
your_special_requirement = strip_list(
self._etree.xpath(opentable_xp.SPECIAL_NOTE))
if your_special_requirement:
result['your_special_requirement'] = your_special_requirement
else:
self.logger.info('your_special_requirement is empty %s',
self._message_id)
return result
def _parse(self):
result = self._order
links = self._etree.xpath(expedia_xp.LINK)
if links and len(links) == 2:
result['hotel_link'] = links[0]
result['map_link'] = links[1]
else:
self.logger.warning('hotel and map links is empty %s',
self._message_id)
tel_fax = take_first(self._etree, expedia_xp.HELP)
if tel_fax:
if 'Fax' in tel_fax:
result['telephone'] = tel_fax.split(', Fax: ')[0].split(
': ')[-1]
result['fax'] = tel_fax.split(':')[-1].strip()
else:
result['telephone'] = tel_fax.split(': ')[-1]
else:
self.logger.error('telephone is empty %s', self._message_id)
hotel_name = strip_str(take_first(self._etree, expedia_xp.NAME))
if hotel_name:
result['hotel_name'] = hotel_name
else:
self.logger.error('hotel_name is empty %s', self._message_id)
address = strip_str(take_first(self._etree, expedia_xp.ADDRESS))
if address:
result['address'] = address
else:
self.logger.error('address is empty %s', self._message_id)
reservation_date = strip_str(
take_first(self._etree, expedia_xp.RESERVATION_DATE))
if reservation_date:
check_in_date, check_out_date = unpack(
reservation_date.split(' - '))
else:
check_in_date = check_out_date = ''
self.logger.error('reservation_date_raw is empty %s',
self._message_id)
if check_out_date and check_in_date:
result['check_in_date'] = check_in_date
result['check_out_date'] = check_out_date
tz = to_timezone(address)
# date format is unsure
check_in_date_formatted = DateTime(
check_in_date,
'MMM D, YYYY'
) \
.tz_to_datetime(tz)
if check_in_date_formatted:
result['check_in_date_formatted'] = check_in_date_formatted
check_out_date_formatted = DateTime(
check_out_date,
'MMM D, YYYY'
) \
.tz_to_datetime(tz)
if check_out_date_formatted:
result['check_out_date_formatted'] = check_out_date_formatted
confirm_number = strip_str(
take_first(self._etree, expedia_xp.COMFIRM_NUMBER))
if confirm_number:
result['confirm_code'] = confirm_number
else:
self.logger.error('confirm_number is empty %s', self._message_id)
reservation_time = strip_list(
self._etree.xpath(expedia_xp.CHECK_IN_OUT_TIME))
if reservation_time:
check_in_time, check_out_time = unpack(reservation_time)
else:
check_out_time = check_in_time = ''
self.logger.error('reservation_time is empty %s', self._message_id)
if check_in_time and check_out_time:
result['check_in_time'] = check_in_time
result['check_out_time'] = check_out_time
guest = strip_list(self._etree.xpath(expedia_xp.GUEST))
guest_name = guest[1].split(
'for ')[-1] if guest and len(guest) > 1 else ''
if guest_name:
result['guest_name'] = guest_name
else:
self.logger.error('guest_name is empty %s', self._message_id)
room = strip_list(self._etree.xpath(expedia_xp.ROOM))
room_type = room.pop(0) if room and len(room) >= 1 else ''
if room_type:
result['room_type'] = room_type
else:
self.logger.error('room_type is empty %s', self._message_id)
if room:
result['included_amenities'] = [remove_space(i) for i in room]
else:
self.logger.error('included_amenities is empty %s',
self._message_id)
room_request = strip_list(self._etree.xpath(expedia_xp.ROOM_REQUEST))
if len(room_request) > 1:
result['bed_type'] = room_request.pop(1)
else:
self.logger.error('bed_type is empty %s', self._message_id)
if len(room_request) > 1:
result['room_tips'] = room_request[1:]
else:
self.logger.error('room_tip is empty %s', self._message_id)
price = strip_str(take_last(self._etree, expedia_xp.PRICE))
if price:
result['price'] = price
else:
self.logger.error('price is empty %s', self._message_id)
total = strip_str(take_first(self._etree, expedia_xp.TOTAL))
if total:
result['total_cost'] = ' '.join(remove_space(total).split(' ')[1:])
else:
self.logger.error('price total is empty %s', self._message_id)
price_des = strip_str(take_last(self._etree, expedia_xp.PRICE_DETAILS))
if price_des:
result['price_tips'] = [price_des]
else:
self.logger.error('price describe is empty %s', self._message_id)
taxes = strip_str(take_last(self._etree, expedia_xp.TAXES))
if taxes:
result['taxes_fee'] = strip_str(taxes.split(':')[-1])
else:
self.logger.warning('taxes is empty %s', self._message_id)
rules = strip_list(self._etree.xpath(expedia_xp.RULES))
if rules:
result['notice'] = [i for i in rules if '.' in i]
else:
self.logger.error('notice is empty %s', self._message_id)
additional_hotel_fee = take_first(self._etree,
expedia_xp.ADDITIONAL_HOTEL_FEE)
if additional_hotel_fee is not None:
additional_hotel_fee = strip_list(
remove_tags(etree.tostring(additional_hotel_fee)).replace(
' ', '').split('\n'))
if len(additional_hotel_fee) > 1:
result['additional_hotel_fee'] = additional_hotel_fee[1:]
else:
self.logger.warning('additional_hotel_fee is empty %s',
self._message_id)
check_in_policy = strip_list(
self._etree.xpath(expedia_xp.CHECK_IN_POLICY))
if check_in_policy:
result['check_in_policies'] = check_in_policy[1:]
else:
self.logger.warning('policies is empty %s', self._message_id)
return result
def _parse(self): # pylint: disable=R0912,R0915
result = self._order
name = remove_space(take_first(self._etree, priceline_xp.NAME))
if name:
result['hotel_name'] = name
else:
self.logger.error('hotel_name is empty %s', self._message_id)
address = ', '.join(self._etree.xpath(priceline_xp.ADDRESS)).strip()
if address:
result['address'] = address
else:
self.logger.error('address is empty %s', self._message_id)
check_in_date = strip_str(
take_last(self._etree, priceline_xp.CHECK_IN_DATE))
tz = to_timezone(address)
if check_in_date:
result['check_in_date'] = check_in_date
check_in_date_formatted = \
DateTime(check_in_date, 'MMMM DD, YYYY').tz_to_datetime(tz)
if check_in_date_formatted:
result['check_in_date_formatted'] = check_in_date_formatted
else:
self.logger.error('check_in_date_formatted is empty %s',
self._message_id)
else:
self.logger.error('check_in_date is empty %s', self._message_id)
check_in_time = strip_str(
take_last(self._etree, priceline_xp.CHECK_IN_TIME))
if check_in_time:
result['check_in_time'] = check_in_time.strip('(').strip(')')
else:
self.logger.error('check_in_time is empty %s', self._message_id)
check_out_date = strip_str(
take_last(self._etree, priceline_xp.CHECK_OUT_DATE))
if check_out_date:
result['check_out_date'] = check_out_date
check_out_date_formatted = \
DateTime(check_out_date, 'MMMM DD, YYYY').tz_to_datetime(tz)
if check_out_date_formatted:
result['check_out_date_formatted'] = check_out_date_formatted
else:
self.logger.error('check_out_date_formatted is empty %s',
self._message_id)
else:
self.logger.error('check_out_date is empty %s', self._message_id)
check_out_time = strip_str(
take_last(self._etree, priceline_xp.CHECK_OUT_TIME))
if check_out_time:
result['check_out_time'] = check_out_time.strip('(').strip(')')
else:
self.logger.error('check_out_time is empty %s', self._message_id)
tel = strip_str(take_last(self._etree, priceline_xp.PHONE))
if tel:
result['telephone'] = tel
else:
self.logger.error('telephone is empty %s', self._message_id)
confirm_num = self._etree.xpath(priceline_xp.CONFIRM_NUM)
if confirm_num:
confirm_num = ''.join(confirm_num).strip()
pin = confirm_num.split(': ')[-1].strip(')')
confirm_num = confirm_num.split(' (')[0]
result['confirm_code'] = confirm_num
result['pin_code'] = pin
else:
self.logger.error('confirm_number is empty %s', self._message_id)
room_keys = strip_list(self._etree.xpath(priceline_xp.ROOM_KEYS))
room_values = strip_list(self._etree.xpath(priceline_xp.ROOM_VALUES))
if len(room_keys) == len(room_values) and room_keys:
name_dict = {
'Deal Type': 'deal_type',
'Room Price': 'price',
'Number of rooms': 'number_of_rooms',
'Number of nights': 'number_of_nights'
}
for i, j in zip(room_keys, room_values):
if i in name_dict:
result[name_dict[i]] = j
else:
self.logger.warning('%s is%s %s', i, j, self._message_id)
else:
self.logger.error('price is empty %s', self._message_id)
room_info_keys = strip_list(
self._etree.xpath(priceline_xp.ROOM_INFO_KEYS))
room_info_values = strip_list(
self._etree.xpath(priceline_xp.ROOM_INFO_VALUES))
room_info_values = list(map(remove_space, room_info_values))
if room_info_keys and room_info_values:
name_dict = {
'Internet': 'internet',
'Guest Parking': 'guest_parking',
'Prepayment': 'prepayment',
'Meal Plan': 'meal_plan'
}
for k, v in zip(room_info_keys[::-1], room_info_values[::-1]):
if k in name_dict:
result[name_dict[k]] = v
elif 'Room Type' in k:
room_type = strip_list(
self._etree.xpath(priceline_xp.ROOM_TYPE))
if room_type:
result['room_type'] = ' '.join(room_type)
else:
self.logger.error('room_type is empty %s',
self._message_id)
else:
self.logger.warning('%s is %s %s', k, v, self._message_id)
else:
self.logger.error('room_info is empty %s', self._message_id)
reservation_name = strip_list(
self._etree.xpath(priceline_xp.RESERVATION_NAME))
if reservation_name:
result['guest_name'] = reservation_name[0].split(': ')[-1]
else:
self.logger.error('guest_name is empty %s', self._message_id)
total = strip_str(
take_first(self._etree, priceline_xp.ROOM_PRICE_TOTAL))
if total:
result['total_cost'] = total
else:
self.logger.error('total_price is empty %s', self._message_id)
room_price_keys = strip_list(
self._etree.xpath(priceline_xp.ROOM_PRICE_KEYS))
room_price_keys = strip_list(room_price_keys)
room_price_values = strip_list(
self._etree.xpath(priceline_xp.ROOM_PRICE_VALUES))
if room_price_keys and room_price_values:
name_dict = {
'Room Subtotal': 'subtotal',
'Hotel Fee': 'service_charge',
'Taxes & Fees': 'taxes_fee'
}
for i, j in zip(room_price_keys, room_price_values):
if i in name_dict:
result[name_dict[i]] = j
else:
self.logger.warning('%s is %s %s', i, j, self._message_id)
else:
self.logger.error('price_details is empty %s', self._message_id)
price_extra = strip_list(self._etree.xpath(priceline_xp.PRICE_EXTRA))
price_extra = [remove_space(j) for j in price_extra]
if price_extra:
result['price_tips'] = price_extra
else:
self.logger.error('price_extra is empty %s', self._message_id)
notice = strip_list(self._etree.xpath(priceline_xp.NOTICE))
notice = [remove_space(j) for j in notice]
if notice:
result['notice'] = notice
else:
self.logger.error('notice is empty %s', self._message_id)
payment_type = strip_str(
take_first(self._etree, priceline_xp.PAYMENT_TYPE))
if payment_type:
result['payment_forms'] = payment_type
else:
self.logger.error('payment_form is empty %s', self._message_id)
return result