_bulk_geocode logic extraction

This commit is contained in:
Juan Ignacio Sánchez Lara
2018-07-10 15:17:14 +02:00
parent a6c5c21131
commit 286a75fa8e
8 changed files with 115 additions and 123 deletions

View File

@@ -8,6 +8,67 @@ from ..helpers.integration_test_helper import IntegrationTestHelper
from ..helpers.integration_test_helper import assert_close_enough
class TestStreetFunctionsSetUp(TestCase):
provider = None
fixture_points = None
GOOGLE_POINTS = {
'Plaza Mayor, Valladolid': [-4.728252, 41.6517025],
'Paseo Zorrilla, Valladolid': [-4.7404453, 41.6314339],
'1900 amphitheatre parkway': [-122.0875324, 37.4227968],
'1901 amphitheatre parkway': [-122.0885504, 37.4238657],
'1902 amphitheatre parkway': [-122.0876674, 37.4235729],
'Valladolid': [-4.7245321, 41.652251],
'Valladolid, Spain': [-4.7245321, 41.652251],
'Valladolid, Mexico': [-88.2022488, 20.68964],
'Madrid': [-3.7037902, 40.4167754],
'Logroño, Spain': [-2.4449852, 42.4627195],
'Logroño, Argentina': [-61.6961807, -29.5031057],
'Plaza España 1, Barcelona': [2.1482563, 41.375485]
}
HERE_POINTS = {
'Plaza Mayor, Valladolid': [-4.72979, 41.65258],
'Paseo Zorrilla, Valladolid': [-4.73869, 41.63817],
'1900 amphitheatre parkway': [-122.0879468, 37.4234763],
'1901 amphitheatre parkway': [-122.0879253, 37.4238725],
'1902 amphitheatre parkway': [-122.0879531, 37.4234775],
'Valladolid': [-4.73214, 41.6542],
'Valladolid, Spain': [-4.73214, 41.6542],
'Valladolid, Mexico': [-88.20117, 20.69021],
'Madrid': [-3.70578, 40.42028],
'Logroño, Spain': [-2.45194, 42.46592],
'Logroño, Argentina': [-61.69604, -29.50425],
'Plaza España 1, Barcelona': [2.1735699, 41.3823] # TODO: not ideal
}
TOMTOM_POINTS = HERE_POINTS.copy()
TOMTOM_POINTS.update({
'Plaza Mayor, Valladolid': [-4.72183, 41.5826],
'Paseo Zorrilla, Valladolid': [-4.74031, 41.63181],
'Valladolid': [-4.72838, 41.6542],
'Valladolid, Spain': [-4.72838, 41.6542],
'Madrid': [-3.70035, 40.42028],
'Logroño, Spain': [-2.44998, 42.46592],
'Plaza España 1, Barcelona': [2.07479, 41.36818] # TODO: not ideal
})
MAPBOX_POINTS = GOOGLE_POINTS.copy()
MAPBOX_POINTS.update({
'Logroño, Spain': [-2.44556, 42.47],
'Logroño, Argentina': [-70.687195, -33.470901], # TODO: huge mismatch
'Valladolid': [-4.72856, 41.652251],
'Valladolid, Spain': [-4.72856, 41.652251],
'1902 amphitheatre parkway': [-118.03, 34.06], # TODO: huge mismatch
'Madrid': [-3.69194, 40.4167754],
'Plaza España 1, Barcelona': [2.245969, 41.452483] # TODO: not ideal
})
FIXTURE_POINTS = {
'google': GOOGLE_POINTS,
'heremaps': HERE_POINTS,
'tomtom': TOMTOM_POINTS,
'mapbox': MAPBOX_POINTS
}
def setUp(self):
self.env_variables = IntegrationTestHelper.get_environment_variables()
@@ -18,6 +79,15 @@ class TestStreetFunctionsSetUp(TestCase):
self.env_variables['api_key']
)
if not self.fixture_points:
query = "select provider from " \
"cdb_dataservices_client.cdb_service_quota_info() " \
"where service = 'hires_geocoder'"
response = self._run_authenticated(query)
provider = response['rows'][0]['provider']
self.fixture_points = self.FIXTURE_POINTS[provider]
def _run_authenticated(self, query):
authenticated_query = "{}&api_key={}".format(query,
self.env_variables[
@@ -54,77 +124,9 @@ class TestStreetFunctions(TestStreetFunctionsSetUp):
row = response['rows'][0]
x_y = [row['st_x'], row['st_y']]
# Wrong coordinates (Plaza España, Madrid): [-3.7138975, 40.4256762]
assert_close_enough(x_y, [2.1482563, 41.375485])
assert_close_enough(x_y, self.fixture_points['Plaza España 1, Barcelona'])
class TestBulkStreetFunctions(TestStreetFunctionsSetUp):
provider = None
fixture_points = None
GOOGLE_POINTS = {
'Plaza Mayor, Valladolid': [-4.728252, 41.6517025],
'Paseo Zorrilla, Valladolid': [-4.7404453, 41.6314339],
'1900 amphitheatre parkway': [-122.0875324, 37.4227968],
'1901 amphitheatre parkway': [-122.0885504, 37.4238657],
'1902 amphitheatre parkway': [-122.0876674, 37.4235729],
'Valladolid': [-4.7245321, 41.652251],
'Valladolid, Spain': [-4.7245321, 41.652251],
'Valladolid, Mexico': [-88.2022488, 20.68964],
'Madrid': [-3.7037902, 40.4167754],
'Logroño, Spain': [-2.4449852, 42.4627195],
'Logroño, Argentina': [-61.6961807, -29.5031057]
}
HERE_POINTS = {
'Plaza Mayor, Valladolid': [-4.72979, 41.65258],
'Paseo Zorrilla, Valladolid': [-4.73869, 41.63817],
'1900 amphitheatre parkway': [-122.0879468, 37.4234763],
'1901 amphitheatre parkway': [-122.0879253, 37.4238725],
'1902 amphitheatre parkway': [-122.0879531, 37.4234775],
'Valladolid': [-4.73214, 41.6542],
'Valladolid, Spain': [-4.73214, 41.6542],
'Valladolid, Mexico': [-88.20117, 20.69021],
'Madrid': [-3.70578, 40.42028],
'Logroño, Spain': [-2.45194, 42.46592],
'Logroño, Argentina': [-61.69604, -29.50425]
}
TOMTOM_POINTS = HERE_POINTS.copy()
TOMTOM_POINTS.update({
'Plaza Mayor, Valladolid': [-4.72183, 41.5826],
'Paseo Zorrilla, Valladolid': [-4.74031, 41.63181],
'Valladolid': [-4.72838, 41.6542],
'Valladolid, Spain': [-4.72838, 41.6542],
'Madrid': [-3.70035, 40.42028],
'Logroño, Spain': [-2.44998, 42.46592],
})
MAPBOX_POINTS = GOOGLE_POINTS.copy()
MAPBOX_POINTS.update({
'Logroño, Spain': [-2.44556, 42.47],
'Logroño, Argentina': [-70.687195, -33.470901], # TODO: huge mismatch
'Valladolid': [-4.72856, 41.652251],
'Valladolid, Spain': [-4.72856, 41.652251],
'1902 amphitheatre parkway': [-118.03, 34.06], # TODO: huge mismatch
'Madrid': [-3.69194, 40.4167754],
})
FIXTURE_POINTS = {
'google': GOOGLE_POINTS,
'heremaps': HERE_POINTS,
'tomtom': TOMTOM_POINTS,
'mapbox': MAPBOX_POINTS
}
def setUp(self):
TestStreetFunctionsSetUp.setUp(self)
if not self.fixture_points:
query = "select provider from " \
"cdb_dataservices_client.cdb_service_quota_info() " \
"where service = 'hires_geocoder'"
response = self._run_authenticated(query)
provider = response['rows'][0]['provider']
self.fixture_points = self.FIXTURE_POINTS[provider]
def test_full_spec(self):
query = "select cartodb_id, st_x(the_geom), st_y(the_geom) " \
@@ -257,7 +259,7 @@ class TestBulkStreetFunctions(TestStreetFunctionsSetUp):
"""
Useful just to test a good batch size
"""
n = 10
n = 50
streets = []
for i in range(0, n):
streets.append('{{"cartodb_id": {}, "address": "{} Yonge Street, ' \
@@ -292,6 +294,19 @@ class TestBulkStreetFunctions(TestStreetFunctionsSetUp):
x_y_by_cartodb_id = self._x_y_by_cartodb_id(response)
assert_equal(x_y_by_cartodb_id[1], x_y_by_cartodb_id[2])
# "'Plaza España 1', 'Barcelona', null, 'Spain') as the_geom) _x"
def test_component_aggregation(self):
query = "select cartodb_id, st_x(the_geom), st_y(the_geom) " \
"FROM cdb_dataservices_client.cdb_bulk_geocode_street_point(" \
"'select 1 as cartodb_id, ''Spain'' as country, " \
"''Barcelona'' as city, " \
"''Plaza España 1'' as street' " \
", 'street', 'city', NULL, 'country')"
response = self._run_authenticated(query)
assert_close_enough(self._x_y_by_cartodb_id(response)[1],
self.fixture_points['Plaza España 1, Barcelona'])
def _run_authenticated(self, query):
authenticated_query = "{}&api_key={}".format(query,
self.env_variables[