def test_address_extractor_find_city_names():
c1 = City("abc", "12345", None)
c2 = City("def g", "12345", None)
ae = AddressExtractor([c1, c2])
assert ae.find_city_names("without city") == []
assert ae.find_city_names("with def and g") == []
assert ae.find_city_names("with the abc city") == [(c1, 9, 12)]
assert ae.find_city_names("with the def g city") == [(c2, 9, 14)]
assert ae.find_city_names("with def g and abc cities") == [
(c2, 5, 10),
(c1, 15, 18),
]
def test_address_extractor_find_nearby_postal_code(mocker):
c = City("abc", "12345", None)
ae = AddressExtractor([c], postal_code_search_distance=8)
assert ae.find_nearby_postal_code("blah abc 12345", c, 5,
8) == ("12345", 9, 14)
assert ae.find_nearby_postal_code("abc fr12345-", c, 0,
3) == ("12345", 6, 11)
assert ae.find_nearby_postal_code("12345- abc fr", c, 7,
10) == ("12345", 0, 5)
assert ae.find_nearby_postal_code("blah abc 123456", c, 5, 8) is None
assert ae.find_nearby_postal_code("blah abc foo 12345", c, 5, 8) is None
assert ae.find_nearby_postal_code("12345 blah abc foo", c, 11, 14) is None
# Search substring matching with postal code start
assert ae.find_nearby_postal_code("foo 12345fr abc", c, 12,
15) == ("12345", 4, 9)
# Invalid postal code (not 5 digits)
m_get_logger = mocker.patch(f"{module}.get_logger")
c2 = City("abc", "12A42", None)
assert ae.find_nearby_postal_code("foo 12A42 abc", c2, 12, 15) is None
m_get_logger.assert_called_once_with(f"{module}.AddressExtractor")
m_get_logger.return_value.error.assert_called_once_with(
"postal code contains non-digit characters: %s", c2)
def test_load_cities_fr(mocker):
m_gzip_open = mocker.patch(f"{module}.gzip.open")
m_json_load = mocker.patch(
f"{module}.json.load",
return_value=[
{
"fields": {
"nom_de_la_commune": "PARIS",
"code_postal": "75000",
"coordonnees_gps": [48.866667, 2.333333],
},
},
{
"fields": {
"nom_de_la_commune": "POYA",
"code_postal": "98827"
}
},
],
)
res = load_cities_fr()
m_gzip_open.assert_called_once_with(settings.OCR_CITIES_FR_PATH, "rb")
m_json_load.assert_called_once_with(
m_gzip_open.return_value.__enter__.return_value)
assert res == {
City("paris", "75000", (48.866667, 2.333333)),
City("poya", "98827", None),
}
# Error with postal code with bad length
mocker.resetall()
m_json_load.return_value = [
{
"fields": {
"nom_de_la_commune": "YOLO",
"code_postal": "123"
}
},
]
with pytest.raises(ValueError,
match="'123', invalid FR postal code for city 'yolo'"):
load_cities_fr()
m_gzip_open.assert_called_once_with(settings.OCR_CITIES_FR_PATH, "rb")
m_json_load.assert_called_once_with(
m_gzip_open.return_value.__enter__.return_value)
# Error with non-digit postal code
mocker.resetall()
m_json_load.return_value = [
{
"fields": {
"nom_de_la_commune": "YOLO",
"code_postal": "12A42"
}
},
]
with pytest.raises(
ValueError,
match="'12A42', invalid FR postal code for city 'yolo'"):
load_cities_fr()
m_gzip_open.assert_called_once_with(settings.OCR_CITIES_FR_PATH, "rb")
m_json_load.assert_called_once_with(
m_gzip_open.return_value.__enter__.return_value)
def cities():
return [
City("paris", "75000", (48.866667, 2.333333)),
City("poya", "98827", None)
]