def test_reorder():
assert match.name_match('Renaissance Center Tower 300',
'Renaissance Center 300 Tower',
endings=['tower'])
assert match.name_match('Renaissance Center Tower 300',
'Renaissance Center 300 Tower')
def test_match_name_parish_church():
osm = 'Church of St Peter & St Paul'
wd = 'St Peter and St Paul, Bromley'
assert match.name_match(osm, wd, ['church of'])
osm = 'Bromley Parish Church of St Peter & St Paul'
wd = 'St Peter and St Paul, Bromley'
assert match.name_match(osm, wd, ['Parish Church of'])
def test_name_match_trim_both():
m = match.name_match('Oxmoor Mall', 'Oxmoor Center',
endings=['mall', 'center'])
assert m.match_type.name == 'both_trimmed'
m = match.name_match('Castle House', 'The Castle Inn',
endings=['house', 'inn'])
assert m.match_type.name == 'both_trimmed'
def test_name_with_dashes():
wikidata = "Hôpital Saint-François d'Assise"
osm = "Hôpital Saint-François-d'Assise"
assert match.name_match(osm, wikidata)
wikidata = 'Walton-on-the-Hill'
osm = 'Walton on the Hill'
assert match.name_match(osm, wikidata)
def test_strip_words():
osm = 'Rio de la Tetta'
wd = 'Rio Tetta'
assert match.name_match(osm, wd)
osm = 'Holy Trinity Church'
wd = 'Church Of The Holy Trinity'
assert match.name_match(osm, wd, endings=['church'])
def test_number_in_name():
# https://www.wikidata.org/wiki/Q88276810
n1 = 'Lake Number Ten'
n2 = 'Lake No. 10'
assert bool(match.name_match(n1, n2))
assert bool(match.name_match(n2, n1))
n1 = 'Test forty two'
n2 = 'Test 42'
assert bool(match.name_match(n1, n2))
assert bool(match.name_match(n2, n1))
def test_match_name_abbreviation():
wikidata_names = [
'Bishop Justus Church of England School',
'Bishop Justus CE School',
]
for wd in wikidata_names:
assert match.name_match('Bishop Justus CofE School ', wd)
osm = 'Mullard Radio Astronomy Observatory (MRAO)'
wikidata = 'Mullard Radio Astronomy Observatory'
assert match.name_match(osm, wikidata)
def test_match_name_abbreviation():
wikidata_names = [
'Bishop Justus Church of England School',
'Bishop Justus CE School',
]
for wd in wikidata_names:
assert match.name_match('Bishop Justus CofE School ', wd)
assert match.name_match('St Peter', 'Saint Peter')
assert match.name_match('Test Roman Catholic church', 'Test RC church')
osm = 'Mullard Radio Astronomy Observatory (MRAO)'
wikidata = 'Mullard Radio Astronomy Observatory'
assert match.name_match(osm, wikidata)
def test_match_with_place_names():
osm = 'Hungarian house'
wd = 'Hungarian House of New York'
place_names = ['Manhattan', 'New York City', 'New York',
'United States of America']
assert match.name_match(osm, wd, place_names=place_names)
def test_name_match_trim_to_empty():
osm = 'Hall'
wd = 'Post Office'
endings = ['hall', 'post office']
assert not match.match_with_words_removed(osm.lower(), wd.lower(), endings)
assert not match.name_match(osm, wd, endings=endings)
def test_name_match_numbers():
assert match.name_match('Manhattan Community Board 1',
'Manhattan Community Board 1')
assert not match.name_match('Manhattan Community Board 11',
'Manhattan Community Board 1')
assert not match.name_match('Manhattan Community Board 1',
'Manhattan Community Board 11')
assert not match.name_containing_initials('Manhattan Community Board 1',
'Manhattan Community Board 11')
osm_tags = {
'name': 'Manhattan Community Board 11',
}
wikidata_names = {
'1-й общественный совет': [('label', 'ru')],
'Manhattan Community Board 1': [('label', 'en'),
('sitelink', 'enwiki'),
('extract', 'enwiki')],
}
assert not match.check_for_match(osm_tags, wikidata_names)
def test_ignore_apostrophe_s_in_match():
osm = 'Augustine Steward House'
wikidata = "Augustine Steward's House"
cur = match.name_match(osm, wikidata)
assert cur.match_type == match.MatchType.good
def test_name_match_roman_catholic():
assert match.name_match("St. Paul's Roman Catholic Church",
"St. Paul's Catholic Church")
def test_name_match_initials_then_name():
osm = 'RGC – Rainbow Grocery Coop'
wd = 'Rainbow Grocery Coop'
assert match.name_match(osm, wd)
def test_strict_stable_name_match():
assert not match.name_match('Nazeing Park',
'Stable At Nazeing Park',
endings=['stable'])
def test_name_starts_with_the_old():
assert match.name_match('The Old Rectory', 'The Rectory')
def test_name_match_dash_and_both_trim():
n1 = 'Sint Pieters Museum'
n2 = 'Museum Sint-Pieters'
assert match.name_match(n1, n2, endings=['museum'])
def test_no_trim_s_on_single_term_name():
osm = 'Boots'
wd = 'The Boot Inn'
assert not match.name_match(osm, wd, endings=['inn'])
def test_name_match():
assert not match.name_match('', '')
assert match.name_match('test', 'test')
assert match.name_match('the old shop', 'old shop')
assert not match.name_match('test', '')
assert not match.name_match('', 'test')
assert match.name_match('test', 'test.')
assert match.name_match('test.', 'test')
assert not match.name_match('test', '.')
assert not match.name_match('.', 'test')
assert not match.name_match('aaa', 'bbb')
assert not match.name_match('aaa', 'the ')
assert match.name_match('aaa-bbb', 'aaa bbb')
assert match.name_match('the old shop', 'old shop')
assert match.name_match('the bull', 'bull public house',
['public house'])
assert match.name_match('TIAT', 'This Is A Test')
assert match.name_match('John Smith', 'Statue of John Smith')
assert match.name_match('John Smith', 'Tomb of John Smith')
assert match.name_match('Lake Test', 'Test', ['lake'])
assert match.name_match('Test', 'Lake Test', ['lake'])
assert match.name_match('Test', 'Test, Washington, DC')
assert match.name_match('aaa bbb', 'bbb aaa')
assert match.name_match('Vif', 'gare de Vif', endings=['gare'])
assert match.name_match('Vif', 'station Vif', endings=['station'])
assert match.name_match('Sliabh Liag', 'Sliabh a Liag')
assert match.name_match('Beulah', 'Beulah, Powys')
assert match.name_match('Four Crosses', 'Four Crosses, Powys') is not None
assert match.name_match('The Ship', "'The Ship', Derriford")
assert match.name_match('Place Bellecour', ' La Place Bellecour')
assert match.name_match('Lamott', 'La Mott, Pennsylvania')
assert match.name_match('Ті-Ді гарден', 'Թի Դի Գարդեն')
assert match.name_match('Maria-Hilf-Kirche', 'Mariahilfkirche, Munich')
assert match.name_match('Kunkelspass', 'Кункелспас') is not None
assert match.name_match('Bethanien-Kirche', 'Bethanienkirche, Berlin')
assert match.name_match('Tricketts Cross', "Trickett's Cross, Dorset")
assert match.name_match('Кастелец', 'Кастелець')
assert match.name_match('Church and 18th Street', 'Church Street & 18th Street')
assert match.name_match('Church and 18th Street',
'Church Street & 18th Street', endings={'test'})
assert match.name_match('Boulder Railroad Depot (Historic)',
'Boulder Railroad Depot')
assert match.name_match('Times Furnishing', 'Times Furnishing Company Building')
assert match.name_match('Bailiffscourt Hotel And Country Club',
'Bailiffscourt Hotel') is not None
osm = 'St Peter & St Paul'
wd = 'St Peter and St Paul, Bromley'
assert match.name_match(osm, wd)
osm = 'New York Skyports Incorporated Seaplane Base'
wikidata = 'New York Skyports Inc. Seaplane Base'
assert match.name_match(osm, wikidata)
osm = 'Disneyland Pacific Hotel; Pacific Hotel'
wikidata = 'Disneyland Pacific Hotel'
assert match.name_match(osm, wikidata)
osm = 'Leeds Bradford International'
wikidata = 'Leeds Bradford International Airport'
trim = ['airport', 'international airport']
assert match.name_match(osm, wikidata, endings=trim)
osm = 'Bresso'
wikidata = 'Aeroporto di Bresso'
trim = ['aeroporto']
assert match.name_match(osm, wikidata, endings=trim)
assert match.name_match('Rainbow Grocery Coop',
'Rainbow Grocery Cooperative')
assert match.name_match('Kirkwood Inn', "Kirkwood's", endings=['inn'])
osm = 'ESCOLA DE NAUTICA DE BARCELONA'
wikidata = 'Escola de Nàutica de Barcelona'
cur = match.name_match(osm, wikidata)
assert cur.match_type == match.MatchType.good
osm = 'Lombard Buildings'
wikidata = 'Lombard Building'
cur = match.name_match(osm, wikidata, endings=['building'])
assert cur.match_type == match.MatchType.good
assert match.name_match('Boxers', 'The Boxers')
osm = 'The Landers'
wikidata = 'Landers Theatre'
assert match.name_match('The Landers', 'Landers Theatre', endings=['theatre'])
osm = 'Main Street Station'
wikidata = 'Richmond Main Street Station'
place_names = {'Richmond City', 'Virginia'}
assert match.name_match(osm, wikidata, place_names=place_names)
osm = 'Manor Buildings'
wikidata = 'Manor House Buildings'
assert match.name_match(osm, wikidata, endings={'house'})
assert match.name_match('site of Pegwell Lodge', 'Pegwell Lodge')
n1 = 'City of Birmingham Symphony Orchestra'
n2 = 'CBSO Centre'
assert match.name_match(n1, n2)
osm = 'Wabasca Indian Reserve #166'
wikidata = 'Wabasca 166'
assert match.name_match(osm, wikidata, endings={'Indian reserve'})
def test_church_names():
name = "St John's Church"
assert match.name_match(name, name + ' And Attached Railings')
assert match.name_match('Church building', 'Church')
assert match.name_match('Church', 'Church building')
osm = "St. Michael's Church"
wikidata = 'Church Of St Michael'
trim = ['church', 'church of']
assert match.name_match(osm, wikidata, endings=trim)
osm = 'Saint Vitus Catholic Church'
wikidata = "St. Vitus's Church, Cleveland"
trim = ['church', 'church of', 'catholic church', 'rc church']
place_names = {'Cleveland', 'Cuyahoga County', 'Ohio'}
assert match.name_match(osm, wikidata, endings=trim, place_names=place_names)
assert match.name_match("St. Paul's Roman Catholic Church",
"St. Paul's Catholic Church")
assert match.name_match('St Peter', 'Saint Peter')
assert match.name_match('Test Roman Catholic church', 'Test RC church')
assert match.name_match('Church of Ss Peter and Paul',
"St Peter and St Paul's Church",
endings=['church', 'church of'])
assert match.name_match('Church of Ss Peter and Paul',
"St Peter and St Paul's Church",
endings=['church', 'church of'])
assert match.name_match('Michaelstow Church',
'Michaelstow DSC 8243',
endings=['church', 'church of'])
assert match.name_match('Saint Edmund and Saint George',
'Church of St Edmund and George',
endings=['church', 'church of', 'st'])
assert match.name_match('Parish Church of St Mary',
'St Mary the Virgin',
endings=['parish church', 'church'])
assert match.name_match('St John the Evangelist parish church',
'Parish Church of St John, the Evangelist',
endings=['parish church', 'parish', 'church'])
assert match.name_match('St Mary and St Edmund',
'Church of St Edmund and St Mary',
endings=['parish church', 'parish', 'church', 'church of'])
def test_match_with_missing_house_number():
assert match.name_match('1-3 Rectory Cottages', 'Rectory Cottages')
def test_at_symbol_match():
a = 'HEB Center @ Cedar Park'
b = 'H-E-B Center at Cedar Park'
assert match.name_match(a, b)
assert match.name_match(b, a)
def test_strip_place_name():
osm = 'Danmarks ambassade'
wd = 'Danmarks ambassade i Oslo'
assert match.name_match(osm, wd, place_names=['Oslo'])
def test_street_address():
assert not match.name_match('17 Mill Lane', '1, Mill Lane')
assert match.name_match('110 Livingston', '110 Livingston Street')
assert not match.name_match('Livingston', 'Livingston Street')
def test_name_match_church():
n1 = "St Andrew"
n2 = "St Andrew's Church"
assert match.name_match(n1, n2, endings=['church'])
def test_name_match():
assert not match.name_match('', '')
assert match.name_match('test', 'test')
assert match.name_match('the old shop', 'old shop')
assert not match.name_match('test', '')
assert not match.name_match('', 'test')
assert match.name_match('test', 'test.')
assert match.name_match('test.', 'test')
assert not match.name_match('test', '.')
assert not match.name_match('.', 'test')
assert not match.name_match('aaa', 'bbb')
assert not match.name_match('aaa', 'the ')
assert match.name_match('aaa-bbb', 'aaa bbb')
assert match.name_match('the old shop', 'old shop')
assert match.name_match('the bull', 'bull public house',
['public house'])
assert match.name_match('TIAT', 'This Is A Test')
assert match.name_match('John Smith', 'Statue of John Smith')
assert match.name_match('John Smith', 'Tomb of John Smith')
name = "St John's Church"
assert match.name_match(name, name + ' And Attached Railings')
assert match.name_match('Church building', 'Church')
assert match.name_match('Church', 'Church building')
assert match.name_match('Lake Test', 'Test', ['lake'])
assert match.name_match('Test', 'Lake Test', ['lake'])
assert match.name_match('Test', 'Test, Washington, DC')
assert match.name_match('aaa bbb', 'bbb aaa')
assert match.name_match('Vif', 'gare de Vif', endings=['gare'])
assert match.name_match('Vif', 'station Vif', endings=['station'])
osm = 'St Peter & St Paul'
wd = 'St Peter and St Paul, Bromley'
assert match.name_match(osm, wd)
osm = 'New York Skyports Incorporated Seaplane Base'
wikidata = 'New York Skyports Inc. Seaplane Base'
assert match.name_match(osm, wikidata)
osm = 'Disneyland Pacific Hotel; Pacific Hotel'
wikidata = 'Disneyland Pacific Hotel'
assert match.name_match(osm, wikidata)
osm = 'Leeds Bradford International'
wikidata = 'Leeds Bradford International Airport'
trim = ['airport', 'international airport']
assert match.name_match(osm, wikidata, endings=trim)
osm = 'Bresso'
wikidata = 'Aeroporto di Bresso'
trim = ['aeroporto']
assert match.name_match(osm, wikidata, endings=trim)
assert match.name_match('Rainbow Grocery Coop',
'Rainbow Grocery Cooperative')
assert match.name_match('Kirkwood Inn', "Kirkwood's", endings=['inn'])
osm = 'ESCOLA DE NAUTICA DE BARCELONA'
wikidata = 'Escola de Nàutica de Barcelona'
cur = match.name_match(osm, wikidata)
assert cur.match_type == match.MatchType.good
osm = 'Lombard Buildings'
wikidata = 'Lombard Building'
cur = match.name_match(osm, wikidata, endings=['building'])
assert cur.match_type == match.MatchType.good
assert match.name_match('Boxers', 'The Boxers')
osm = 'The Landers'
wikidata = 'Landers Theatre'
assert match.name_match('The Landers', 'Landers Theatre', endings=['theatre'])
osm = "St. Michael's Church"
wikidata = 'Church Of St Michael'
trim = ['church', 'church of']
assert match.name_match(osm, wikidata, endings=trim)
osm = 'Saint Vitus Catholic Church'
wikidata = "St. Vitus's Church, Cleveland"
trim = ['church', 'church of', 'catholic church', 'rc church']
place_names = {'Cleveland', 'Cuyahoga County', 'Ohio'}
assert match.name_match(osm, wikidata, endings=trim, place_names=place_names)
osm = 'Main Street Station'
wikidata = 'Richmond Main Street Station'
place_names = {'Richmond City', 'Virginia'}
assert match.name_match(osm, wikidata, place_names=place_names)
osm = 'Manor Buildings'
wikidata = 'Manor House Buildings'
assert match.name_match(osm, wikidata)
assert match.name_match('site of Pegwell Lodge', 'Pegwell Lodge')
