def test_diff():
diff: AddressDiff = create_diff()
assert len(diff.matches) == 3
assert len(diff.missing) == 2
assert len(diff.surplus) == 2
assert diff.num_matches == 3
assert diff.num_missing_not_ignored == 1
assert diff.num_surplus_not_ignored == 1
assert round(abs(diff.coverage_in_percent - 75.0), 1) == 0
assert diff.matches[Address('Identical Street', '1')] == MatchInformation(
OsmPrimitive('way', 1), Coordinate(1.1, 1.2))
assert diff.matches[Address('Identical Street', '7')] == MatchInformation(
OsmPrimitive('way', 7), Coordinate(7.1, 7.2))
assert diff.matches[Address('Identical Street', '13')] == MatchInformation(
OsmPrimitive('way', 13), Coordinate(13.1, 13.2))
assert diff.missing[Address('Missing Street',
'2')] == MissingAddressInformation(
Coordinate(2.1, 2.2), None)
assert diff.missing[Address('Missing Street', '3')] == \
MissingAddressInformation(Coordinate(3.1, 3.2), IgnoredAddressInformation(False, 'building does not exist', ignore_date))
assert diff.surplus[0] == (Address('Surplus Street', '4'),
SurplusInformation(
OsmPrimitive('node', 31),
IgnoredAddressInformation(
True, 'building does exist',
ignore_date)))
assert diff.surplus[1] == (Address('Surplus Street', '3'),
SurplusInformation(OsmPrimitive('way', 32),
None))
return diff
def create_diff():
address_diff = AddressDiffComputation()
reference = {
Address('Identical Street', '1'): Coordinate(1.1, 1.2),
Address('Identical Street', '7'): Coordinate(7.1, 7.2),
Address('Identical Street', '13'): Coordinate(13.1, 13.2),
Address('Missing Street', '2'): Coordinate(2.1, 2.2),
Address('Missing Street', '2'):
Coordinate(2.1, 2.2), # duplicate in reference set
Address('Missing Street', '3'): Coordinate(3.1, 3.2)
}
actual = [(Address('Identical Street', '1'), OsmPrimitive('way', 1)),
(Address('Identical Street', '7'), OsmPrimitive('way', 7)),
(Address('Identical Street', '13'), OsmPrimitive('way', 13)),
(Address('Surplus Street', '4'), OsmPrimitive('node', 31)),
(Address('Surplus Street', '3'), OsmPrimitive('way', 32))
] # way and node with same address, to be reported twice
ignored = {
Address('Missing Street', '3'):
IgnoredAddressInformation(False, 'building does not exist',
ignore_date),
Address('Surplus Street', '4'):
IgnoredAddressInformation(True, 'building does exist', ignore_date)
}
diff: AddressDiff = address_diff.compute(reference, actual, ignored)
return diff
def test_add():
file_path, keeper = create_keeper('keeper_addresses_to_be_ignored.csv')
keeper.add_ignored(
Address('Ignored Street', '1'),
IgnoredAddressInformation(False, "building does not exist",
ignore_date))
ignored = CsvAddressListImport.read_to_be_ignored(file_path)
assert len(ignored) == 1
assert ignored[Address('Ignored Street',
'1')] == IgnoredAddressInformation(
False, "building does not exist", ignore_date)
def test_remove():
file_path, keeper = create_keeper(
'keeper_addresses_to_be_ignored_not_anymore.csv')
address = Address('Ignored Street', '1')
keeper.add_ignored(
address,
IgnoredAddressInformation(False, "building does not exist",
ignore_date))
keeper.remove_ignored(address)
ignored = CsvAddressListImport.read_to_be_ignored(file_path)
assert len(ignored) == 0
def read_to_be_ignored(
file_path_or_buffer) -> Dict[Address, IgnoredAddressInformation]:
csv_file = pandas.read_csv(file_path_or_buffer)
addresses = {}
for index, row in csv_file.iterrows():
ignore_date_string: str = row.get('ignore_date')
try:
ignore_date = IgnoredAddressInformation.parse_ignore_date(
ignore_date_string)
except (ValueError, TypeError):
print(
f'To be ignored address has invalid ignore date "{ignore_date_string}", so we assume 2020-05-01'
)
ignore_date = IgnoredAddressInformation.parse_ignore_date(
'2020-05-01')
try:
addresses[CsvAddressListImport.get_address(row)] = \
IgnoredAddressInformation(bool(row.get('exists')), row.get('reason'), ignore_date)
except InvalidHousenumber as i:
print(f'Invalid housenumber "{i.input_string}": {i.message}"')
# skip this address, continue with others
return addresses
from missing_addresses import Address, MissingAddressInformation, SurplusInformation, Coordinate, IgnoredAddressInformation, MatchInformation, \
OsmPrimitive, AddressDiffComputation, AddressDiff
ignore_date = IgnoredAddressInformation.parse_ignore_date('2020-05-13')
def create_diff():
address_diff = AddressDiffComputation()
reference = {
Address('Identical Street', '1'): Coordinate(1.1, 1.2),
Address('Identical Street', '7'): Coordinate(7.1, 7.2),
Address('Identical Street', '13'): Coordinate(13.1, 13.2),
Address('Missing Street', '2'): Coordinate(2.1, 2.2),
Address('Missing Street', '2'):
Coordinate(2.1, 2.2), # duplicate in reference set
Address('Missing Street', '3'): Coordinate(3.1, 3.2)
}
actual = [(Address('Identical Street', '1'), OsmPrimitive('way', 1)),
(Address('Identical Street', '7'), OsmPrimitive('way', 7)),
(Address('Identical Street', '13'), OsmPrimitive('way', 13)),
(Address('Surplus Street', '4'), OsmPrimitive('node', 31)),
(Address('Surplus Street', '3'), OsmPrimitive('way', 32))
] # way and node with same address, to be reported twice
ignored = {
Address('Missing Street', '3'):
IgnoredAddressInformation(False, 'building does not exist',
ignore_date),
Address('Surplus Street', '4'):
IgnoredAddressInformation(True, 'building does exist', ignore_date)
}
def save_ignored(self, addresses: Dict[Address, IgnoredAddressInformation]):
with open(self.file_path, 'w', newline='', encoding='UTF-8') as csvfile:
w = writer(csvfile)
w.writerow(('addr:street', 'addr:housenumber', 'exists', 'reason', 'ignore_date'))
for (a, i) in addresses.items():
w.writerow([a.street, str(a.housenumber), i.exists, i.reason, i.ignore_date.strftime(IgnoredAddressInformation.get_date_format())])