add lookup of tiger data

This commit is contained in:
Sarah Hoffmann 2023-02-02 15:31:42 +01:00
parent f1ceefe9a6
commit 70f6f9a711
4 changed files with 188 additions and 42 deletions

View File

@ -93,6 +93,27 @@ async def find_in_osmline(conn: SearchConnection, place: ntyp.PlaceRef,
return (await conn.execute(sql)).one_or_none()
async def find_in_tiger(conn: SearchConnection, place: ntyp.PlaceRef,
details: ntyp.LookupDetails) -> Optional[SaRow]:
""" Search for the given place in table of Tiger addresses and return the
base information.
"""
t = conn.t.tiger
sql = sa.select(t.c.place_id, t.c.parent_place_id,
t.c.startnumber, t.c.endnumber, t.c.step,
t.c.postcode,
sa.func.ST_X(sa.func.ST_Centroid(t.c.linegeo)).label('x'),
sa.func.ST_Y(sa.func.ST_Centroid(t.c.linegeo)).label('y'),
_select_column_geometry(t.c.linegeo, details.geometry_output))
if isinstance(place, ntyp.PlaceID):
sql = sql.where(t.c.place_id == place.place_id)
else:
return None
return (await conn.execute(sql)).one_or_none()
async def get_place_by_id(conn: SearchConnection, place: ntyp.PlaceRef,
details: ntyp.LookupDetails) -> Optional[nres.SearchResult]:
""" Retrieve a place with additional details from the database.
@ -102,13 +123,19 @@ async def get_place_by_id(conn: SearchConnection, place: ntyp.PlaceRef,
row = await find_in_placex(conn, place, details)
if row is not None:
result = nres.create_from_placex_row(row=row)
result = nres.create_from_placex_row(row)
await nres.add_result_details(conn, result, details)
return result
row = await find_in_osmline(conn, place, details)
if row is not None:
result = nres.create_from_osmline_row(row=row)
result = nres.create_from_osmline_row(row)
await nres.add_result_details(conn, result, details)
return result
row = await find_in_tiger(conn, place, details)
if row is not None:
result = nres.create_from_tiger_row(row)
await nres.add_result_details(conn, result, details)
return result

View File

@ -144,50 +144,63 @@ def create_from_placex_row(row: SaRow) -> SearchResult:
""" Construct a new SearchResult and add the data from the result row
from the placex table.
"""
result = SearchResult(source_table=SourceTable.PLACEX,
place_id=row.place_id,
parent_place_id=row.parent_place_id,
linked_place_id=row.linked_place_id,
osm_object=(row.osm_type, row.osm_id),
category=(row.class_, row.type),
admin_level=row.admin_level,
names=row.name,
address=row.address,
extratags=row.extratags,
housenumber=row.housenumber,
postcode=row.postcode,
wikipedia=row.wikipedia,
rank_address=row.rank_address,
rank_search=row.rank_search,
importance=row.importance,
country_code=row.country_code,
indexed_date=getattr(row, 'indexed_date'),
centroid=Point(row.x, row.y),
geometry = _filter_geometries(row))
return result
return SearchResult(source_table=SourceTable.PLACEX,
place_id=row.place_id,
parent_place_id=row.parent_place_id,
linked_place_id=row.linked_place_id,
osm_object=(row.osm_type, row.osm_id),
category=(row.class_, row.type),
admin_level=row.admin_level,
names=row.name,
address=row.address,
extratags=row.extratags,
housenumber=row.housenumber,
postcode=row.postcode,
wikipedia=row.wikipedia,
rank_address=row.rank_address,
rank_search=row.rank_search,
importance=row.importance,
country_code=row.country_code,
indexed_date=getattr(row, 'indexed_date'),
centroid=Point(row.x, row.y),
geometry=_filter_geometries(row))
def create_from_osmline_row(row: SaRow) -> SearchResult:
""" Construct a new SearchResult and add the data from the result row
from the osmline table.
"""
result = SearchResult(source_table=SourceTable.OSMLINE,
place_id=row.place_id,
parent_place_id=row.parent_place_id,
osm_object=('W', row.osm_id),
category=('place', 'houses'),
address=row.address,
postcode=row.postcode,
extratags={'startnumber': str(row.startnumber),
'endnumber': str(row.endnumber),
'step': str(row.step)},
country_code=row.country_code,
indexed_date=getattr(row, 'indexed_date'),
centroid=Point(row.x, row.y),
geometry = _filter_geometries(row))
return SearchResult(source_table=SourceTable.OSMLINE,
place_id=row.place_id,
parent_place_id=row.parent_place_id,
osm_object=('W', row.osm_id),
category=('place', 'houses'),
address=row.address,
postcode=row.postcode,
extratags={'startnumber': str(row.startnumber),
'endnumber': str(row.endnumber),
'step': str(row.step)},
country_code=row.country_code,
indexed_date=getattr(row, 'indexed_date'),
centroid=Point(row.x, row.y),
geometry=_filter_geometries(row))
return result
def create_from_tiger_row(row: SaRow) -> SearchResult:
""" Construct a new SearchResult and add the data from the result row
from the Tiger table.
"""
return SearchResult(source_table=SourceTable.TIGER,
place_id=row.place_id,
parent_place_id=row.parent_place_id,
category=('place', 'houses'),
postcode=row.postcode,
extratags={'startnumber': str(row.startnumber),
'endnumber': str(row.endnumber),
'step': str(row.step)},
country_code='us',
centroid=Point(row.x, row.y),
geometry=_filter_geometries(row))
async def add_result_details(conn: SearchConnection, result: SearchResult,

View File

@ -18,7 +18,7 @@ from nominatim.db.sql_preprocessor import SQLPreprocessor
class APITester:
def __init__(self):
self.api = napi.NominatimAPI(Path('/invalid'), {})
self.api = napi.NominatimAPI(Path('/invalid'))
self.async_to_sync(self.api._async_api.setup_database())
@ -93,6 +93,16 @@ class APITester:
'linegeo': 'SRID=4326;' + kw.get('geometry', 'LINESTRING(1.1 -0.2, 1.09 -0.22)')})
def add_tiger(self, **kw):
self.add_data('tiger',
{'place_id': kw.get('place_id', 30000),
'parent_place_id': kw.get('parent_place_id'),
'startnumber': kw.get('startnumber', 2),
'endnumber': kw.get('endnumber', 6),
'step': kw.get('step', 2),
'postcode': kw.get('postcode'),
'linegeo': 'SRID=4326;' + kw.get('geometry', 'LINESTRING(1.1 -0.2, 1.09 -0.22)')})
async def exec_async(self, sql, *args, **kwargs):
async with self.api._async_api.begin() as conn:
return await conn.execute(sql, *args, **kwargs)
@ -104,9 +114,10 @@ class APITester:
@pytest.fixture
def apiobj(temp_db_with_extensions, temp_db_conn):
def apiobj(temp_db_with_extensions, temp_db_conn, monkeypatch):
""" Create an asynchronous SQLAlchemy engine for the test DB.
"""
monkeypatch.setenv('NOMINATIM_USE_US_TIGER_DATA', 'yes')
testapi = APITester()
testapi.async_to_sync(testapi.create_tables())

View File

@ -168,7 +168,6 @@ def test_lookup_placex_with_address_details(apiobj):
names={'ref': 'pl'}, extratags={},
admin_level=None, fromarea=True, isaddress=False,
rank_address=4, distance=0.0)
]
@ -370,7 +369,103 @@ def test_lookup_osmline_with_address_details(apiobj):
names={'ref': 'pl'}, extratags={},
admin_level=None, fromarea=True, isaddress=False,
rank_address=4, distance=0.0)
]
def test_lookup_in_tiger(apiobj):
apiobj.add_tiger(place_id=4924,
parent_place_id=12,
startnumber=1, endnumber=4, step=1,
postcode='34425',
geometry='LINESTRING(23 34, 23 35)')
result = apiobj.api.lookup(napi.PlaceID(4924), napi.LookupDetails())
assert result is not None
assert result.source_table.name == 'TIGER'
assert result.category == ('place', 'houses')
assert result.centroid == (pytest.approx(23.0), pytest.approx(34.5))
assert result.place_id == 4924
assert result.parent_place_id == 12
assert result.linked_place_id is None
assert result.osm_object is None
assert result.admin_level == 15
assert result.names is None
assert result.address is None
assert result.extratags == {'startnumber': '1', 'endnumber': '4', 'step': '1'}
assert result.housenumber is None
assert result.postcode == '34425'
assert result.wikipedia is None
assert result.rank_search == 30
assert result.rank_address == 30
assert result.importance is None
assert result.country_code == 'us'
assert result.indexed_date is None
assert result.address_rows is None
assert result.linked_rows is None
assert result.parented_rows is None
assert result.name_keywords is None
assert result.address_keywords is None
assert result.geometry == {'type': 'ST_LineString'}
def test_lookup_tiger_with_address_details(apiobj):
apiobj.add_tiger(place_id=9000,
startnumber=2, endnumber=4, step=1,
parent_place_id=332)
apiobj.add_placex(place_id=332, osm_type='W', osm_id=4,
class_='highway', type='residential', name='Street',
country_code='us',
rank_search=27, rank_address=26)
apiobj.add_address_placex(332, fromarea=False, isaddress=False,
distance=0.0034,
place_id=1000, osm_type='N', osm_id=3333,
class_='place', type='suburb', name='Smallplace',
country_code='us', admin_level=13,
rank_search=24, rank_address=23)
apiobj.add_address_placex(332, fromarea=True, isaddress=True,
place_id=1001, osm_type='N', osm_id=3334,
class_='place', type='city', name='Bigplace',
country_code='us',
rank_search=17, rank_address=16)
result = apiobj.api.lookup(napi.PlaceID(9000),
napi.LookupDetails(address_details=True))
assert result.address_rows == [
napi.AddressLine(place_id=None, osm_object=None,
category=('place', 'house_number'),
names={'ref': '2'}, extratags={},
admin_level=None, fromarea=True, isaddress=True,
rank_address=28, distance=0.0),
napi.AddressLine(place_id=332, osm_object=('W', 4),
category=('highway', 'residential'),
names={'name': 'Street'}, extratags={},
admin_level=15, fromarea=True, isaddress=True,
rank_address=26, distance=0.0),
napi.AddressLine(place_id=1000, osm_object=('N', 3333),
category=('place', 'suburb'),
names={'name': 'Smallplace'}, extratags={},
admin_level=13, fromarea=False, isaddress=True,
rank_address=23, distance=0.0034),
napi.AddressLine(place_id=1001, osm_object=('N', 3334),
category=('place', 'city'),
names={'name': 'Bigplace'}, extratags={},
admin_level=15, fromarea=True, isaddress=True,
rank_address=16, distance=0.0),
napi.AddressLine(place_id=None, osm_object=None,
category=('place', 'country_code'),
names={'ref': 'us'}, extratags={},
admin_level=None, fromarea=True, isaddress=False,
rank_address=4, distance=0.0)
]