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 __get_point_coordinate(self, point_address, coordinates):
lower_point_address = point_address.lower()
for lat, lng, address, place in coordinates:
for token in address.split():
if trim_spaces_and_commas(token.lower()) not in lower_point_address:
break
else:
return lat, lng
return None, None
def __parse_base_atm_terminals(self, item, map_points, point_type, start_names):
point = Point()
point.prov = self.uid
point.type = point_type
if not item('.name').text().split()[0].startswith(start_names):
return None
point.address, point.place = split_address_place(' '.join(item('.name').text().strip().split()[1:]))
point.place = trim_spaces_and_commas(normalize_text(item('.addres strong').text()))
point.check_information = CHECK_OFFICIAL
for lat, lng, name, address, place in map_points:
if (name in start_names) 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
