def __parse_phones(self, item):
if len(item('td')) > 1:
more_info_html = replace_br(item('td:last').html(), ';;;')
more_info_items = map(strip, PQ(more_info_html).text().split(';;;'))
if more_info_items and more_info_items[0].startswith('('):
return normalize_phones(more_info_items[0].split(','))
return []
def __parse_base(self, item, city, point_type):
point = Point()
point.prov = self.uid
point.type = point_type
point.name = normalize_text(item('.b-map-side>h5').text())
point.address, point.place = split_address_place(u'г. %s, %s' % (city, item('.b-map-side>p span:eq(0)').text()))
coordinates = item('.b-map-side>p span:eq(1)').text()
if coordinates:
point.lat, point.lng = map(strip, coordinates.split(','))
text_html = replace_br(item('.b-map-side-more').html(), ';;;')
time_items = []
for sub_item in map(normalize_text, PQ(text_html).text().split(';;;')):
if not sub_item:
continue
if sub_item.startswith(u'Телефон:'):
point.phones = normalize_phones(sub_item[len(u'Телефон:')].split(','))
continue
time_items.append(sub_item)
point.time = normalize_time(', '.join(time_items))
point.check_information = CHECK_OFFICIAL
if point.lat and point.lng:
point.check_coordinates = CHECK_OFFICIAL
else:
warning_not_official_coordinates(point)
return point
def get_offices(self):
points = []
items_tree = ET.fromstring(get_url(self.__offices_xml_url))
for item in items_tree.iter('item'):
point = self.__parse_office(item)
if point:
points.append(point)
page = PQ(get_url(self.__regional_offices_page_url))
point = None
for item in map(PQ, page('#content_internal span:eq(0)').children()):
if item[0].tag not in self.__regional_offices_tags:
continue
if item[0].tag == 'h2':
point = Point()
point.prov = self.uid
point.type = TYPE_OFFICE
point.name = trim_spaces_and_commas(normalize_text(item.text()))
point.check_information = CHECK_OFFICIAL
continue
if not point:
continue
item_html = replace_br(item.html(), ';;;')
sub_items = PQ(item_html).text().split(';;;')
point.address, point.place = split_address_place(sub_items[0])
for sub_item in map(normalize_text, sub_items[1:]):
if sub_item.startswith(u'т.ф.:'):
point.phone = normalize_phones(sub_item[len(u'т.ф.:'):].split(','))
warning_not_official_coordinates(point)
points.append(point)
point = None
return points
def __parse_base_office_exchange(self, item, map_points, point_type, start_names):
point = Point()
point.prov = self.uid
point.type = point_type
point.name = normalize_text(item('.name').text())
if not point.name.startswith(start_names):
return None
point.address, point.place = split_address_place(item('.addres strong').text())
sub_item = item('.item_block tr:last')
point.phones = normalize_phones(sub_item('td:eq(0)').text().split(','))
mon_thu = u'пн-чт: ' + sub_item('td:eq(2)').text()
fri = u'пт: ' + sub_item('td:eq(3)').text()
sat = u'сб: ' + sub_item('td:eq(4)').text()
sun = u'вс: ' + sub_item('td:eq(5)').text()
point.time = normalize_time(', '.join([mon_thu, fri, sat, sun]))
point.check_information = CHECK_OFFICIAL
for lng, lat, name, address, place in map_points:
if (point.name in name if point.name and name else True) and\
(point.address and address and point.address in address) and\
(point.place in place if point.place and place else True):
point.lat = lat
point.lng = lng
point.check_coordinates = CHECK_OFFICIAL
break
else:
warning_not_official_coordinates(point)
return point
def __parse_base_office_exchange(self, item):
point = Point()
point.prov = self.uid
point.name = normalize_text(item('td:eq(1)').text())
point.address, point.place = split_address_place(item('td:eq(2)').text())
point.time = normalize_time(item('td:eq(3)').text())
point.phones = normalize_phones(item('td:eq(4)').text().split(','))
point.check_information = CHECK_OFFICIAL
return point
def __parse_office(self, item):
point = Point()
point.prov = self.uid
point.type = TYPE_OFFICE
point.name = normalize_text(item('td:eq(0)').text())
address_items = item('td:eq(1)').text().split(u'тел.')
address = address_items[0]
point.address, point.place = split_address_place(address)
if len(address_items) > 1:
phone = address_items[1]
phones_items = phone.split(u'доб')
point.phones = normalize_phones(phones_items[0].split(','))
point.time = self.__parse_time(item)
point.check_information = CHECK_OFFICIAL
warning_not_official_coordinates(point)
return point
def __parse_office(self, item):
point = Point()
point.prov = self.uid
point.type = TYPE_OFFICE
point.name = normalize_text(item('h1').text())
point.address, point.place = split_address_place(item('tr:eq(2) td:eq(1)').text())
phones = []
phone_html = replace_br(item('tr:eq(5) td:eq(1)').html(), ';;;')
if phone_html:
phones += map(strip, PQ(phone_html).text().split(';;;'))
phone_html = replace_br(item('tr:eq(6) td:eq(1)').html(), ';;;')
if phone_html:
phones += map(strip, PQ(phone_html).text().split(';;;'))
point.phones = normalize_phones(filter(lambda phone: phone.startswith((u'+', u'тел')), phones))
point.time = normalize_time(item('tr:eq(8) td:eq(1)').text())
point.check_information = CHECK_OFFICIAL
warning_not_official_coordinates(point)
return point
def __parse_office(self, item, city):
point = Point()
point.prov = self.uid
point.type = TYPE_OFFICE
point.name = normalize_text(item('th .pointShowMaps span:eq(0)').text())
address = item('th .pointShowMaps span:eq(1)').text()
point.address, point.place = split_address_place(u'г. %s, %s' % (city, address))
time_html = replace_br(item('td:eq(0)').html(), ', ')
point.time = normalize_time(PQ(time_html).text())
phones_html = replace_br(item('td:eq(1)').html(), ', ')
point.phones = normalize_phones(PQ(phones_html).text().split(','))
point.lat = normalize_text(item('th .item_coords .coord1').text())
point.lng = normalize_text(item('th .item_coords .coord2').text())
point.check_information = CHECK_OFFICIAL
if point.lat and point.lng:
point.check_coordinates = CHECK_OFFICIAL
else:
warning_not_official_coordinates(point)
return point
def __parse_office_exchange(self, item, city, coordinates, point_type, point_keywords):
point = Point()
point.prov = self.uid
point.type = point_type
point.name = normalize_text(item('th:eq(0)').text())
if not point.name.startswith(point_keywords):
return None
address_html = replace_br(item('td:eq(0)').html(), ';;;')
address_items = PQ(address_html).text().split(';;;', 1)
point.address = normalize_address(u'%s, %s' % (city, address_items[0]))
if len(address_items) > 1:
point.place = normalize_text(address_items[1])
item('td:eq(1) ul, td:eq(1) li').remove()
point.time = normalize_time(item('td:eq(1)').text())
point.phones = normalize_phones(map(lambda phone_item: PQ(phone_item).text(), item('td:eq(2) p') or item('td:eq(2)')))
point.check_information = CHECK_OFFICIAL
point.lat, point.lng = self.__get_point_coordinate(point, coordinates)
if point.lat and point.lng:
point.check_coordinates = CHECK_OFFICIAL
else:
warning_not_official_coordinates(point)
return point
def __parse_exchange(self, item):
point = Point()
point.prov = self.uid
point.type = TYPE_EXCHANGE
point.name = normalize_text(item.find('name').text)
city = item.find('region').text if item.find('region') else u'Минск'
address = item.find('address').text
point.address = normalize_address(u'г. %s, %s' % (city.title(), address))
point.place = normalize_text(item.find('location').text)
point.lat = item.find('lattitude').text
point.lng = item.find('longitude').text
point.time = normalize_time(item.find('time').text)
if item.find('phones').text:
point.phones = normalize_phones(item.find('phones').text.split(','))
point.check_information = CHECK_OFFICIAL
if point.lat and point.lng:
point.check_coordinates = CHECK_OFFICIAL
else:
warning_not_official_coordinates(point)
return point
def __parse_office(self, item, city_name):
point = self.__parse_base(item, city_name, TYPE_OFFICE)
point.name = item('td:eq(1) a').text()
point.phones = normalize_phones(self.__phone_splitter.split(item('td:eq(1) i').text() or ''))
point.time = self.__parse_time(item('td:eq(2)'))
return point
