def test_missing_in_row_dict_raise_valueerror(self):
database = Database("whoah")
database.register()
dtype = [
(numpy_string('a'), np.uint32),
(numpy_string('b'), np.uint32),
(numpy_string('row'), np.uint32),
(numpy_string('col'), np.uint32),
(numpy_string('values'), np.float32),
]
array = np.array([
(1, 2, MAX_INT_32, MAX_INT_32, 99),
(1, 4, MAX_INT_32, MAX_INT_32, 99),
],
dtype=dtype)
row_dict = {1: 0}
col_dict = {2: 0}
np.save(database.filepath_processed(), array, allow_pickle=False)
with self.assertRaises(ValueError):
MatrixBuilder.build([database.filepath_processed()],
"values",
"a",
"row",
"b",
"col",
row_dict,
col_dict,
drop_missing=False)
def test_base_class(self):
database = DatabaseChooser("a database")
self.assertEqual(database._metadata, databases)
self.assertEqual(
[x[0] for x in database.dtype_fields],
[numpy_string(x) for x in ('input', 'output', 'row', 'col', 'type')]
)
def test_multiple_values_same_exchange(self):
"""Values for same (row, col) should add together"""
a = np.zeros((2, ),
dtype=[(numpy_string('values'), np.float64),
(numpy_string('rows'), np.uint32),
(numpy_string('cols'), np.uint32)])
a[0] = (9, 1, 1)
a[1] = (33, 1, 1)
r = [0, 0, 0] # Just need right length
matrix = MatrixBuilder.build_matrix(array=a,
row_dict=r,
col_dict=r,
row_index_label='rows',
col_index_label='cols',
data_label='values')
answer = np.array(((0, 0, 0), (0, 42, 0), (0, 0, 0)))
self.assertTrue(np.allclose(answer, matrix.toarray()))
class PDS(ProcessedDataStore):
_metadata = metadata
dtype_fields = [
(numpy_string('input'), np.uint32),
]
def process_data(self, row):
return (row['input'], ), row
def test_base_class(self):
method = Method(("a", "method"))
self.assertEqual(method.validator, ia_validator)
self.assertEqual(method._metadata, methods)
method.register()
self.assertTrue(isinstance(method.metadata, dict))
self.assertEqual(
[x[0] for x in method.dtype_fields],
[numpy_string(x) for x in ('flow', 'geo', 'row', 'col')])
def test_build_matrix(self):
a = np.zeros((4, ),
dtype=[(numpy_string('values'), np.float64),
(numpy_string('rows'), np.uint32),
(numpy_string('cols'), np.uint32)])
a[0] = (4.2, 0, 2)
a[1] = (6.6, 1, 1)
a[2] = (1.3, 2, 1)
a[3] = (10, 2, 2)
r = [0, 0, 0] # Just need right length
matrix = MatrixBuilder.build_matrix(array=a,
row_dict=r,
col_dict=r,
row_index_label='rows',
col_index_label='cols',
data_label='values')
answer = np.array(((0, 0, 4.2), (0, 6.6, 0), (0, 1.3, 10)))
self.assertTrue(np.allclose(answer, matrix.toarray()))
class Loading(ProcessedDataStore):
""""""
_metadata = loadings
validator = loading_validator
dtype_fields = [(numpy_string("geo"), np.uint32),
(numpy_string("row"), np.uint32)]
def add_mappings(self, data):
"""In theory, this shouldn't do anything, as all spatial units should be in defined by the method."""
geomapping.add({obj[1] for obj in data})
def process_data(self, row):
return (geomapping[row[1]], MAX_INT_32), row[0]
@property
def filename(self):
return super(Loading, self).filename + ".loading"
def write_to_map(self, method, flow, geocollection=None):
map_obj = get_pandarus_map(method, geocollection)
data = {x[1][1]: x[0] for x in self.load()}
if map_obj.raster:
self._write_raster_loadings_to_map(map_obj, data)
else:
self._write_vector_loadings_to_map(map_obj, data)
def _write_raster_loadings_to_map(self, map_obj, data):
NODATA = -9999.0
filepath = os.path.join(projects.output_dir, self.filename + ".tiff")
array = np.zeros(map_obj.file.array().shape) + NODATA
for obj in map_obj:
if obj["label"] in data:
array[obj["row"], obj["col"]] = data[obj["label"]]
map_obj.file.write_modified_array(filepath, array, nodata=NODATA)
return filepath
def _write_vector_loadings_to_map(self, map_obj, data):
raise NotImplementedError
def test_build_one_d_drop_missing(self):
database = Database("ghost")
database.register()
dtype = [
(numpy_string('a'), np.uint32),
(numpy_string('row'), np.uint32),
(numpy_string('values'), np.float32),
]
array = np.array([
(1, MAX_INT_32, 99),
(2, MAX_INT_32, 99),
(3, MAX_INT_32, 99),
],
dtype=dtype)
row_dict = {1: 0, 2: 1}
np.save(database.filepath_processed(), array, allow_pickle=False)
values = MatrixBuilder.build([database.filepath_processed()],
"values",
"a",
"row",
row_dict=row_dict,
one_d=True)[0]
self.assertEqual(values.shape, (2, ))
def test_build_one_d(self):
database = Database("sour")
database.register()
dtype = [
(numpy_string('a'), np.uint32),
(numpy_string('row'), np.uint32),
(numpy_string('values'), np.float32),
]
array = np.array([
(1, MAX_INT_32, 99),
(2, MAX_INT_32, 100),
],
dtype=dtype)
row_dict = {1: 0, 2: 1}
np.save(database.filepath_processed(), array, allow_pickle=False)
matrix = MatrixBuilder.build([database.filepath_processed()],
"values",
"a",
"row",
row_dict=row_dict,
one_d=True)[3]
self.assertTrue(
np.allclose(matrix.toarray(), np.array(((99, 0), (0, 100)))))
def test_build_drop_missing(self):
database = Database("boo")
database.register()
dtype = [
(numpy_string('a'), np.uint32),
(numpy_string('b'), np.uint32),
(numpy_string('row'), np.uint32),
(numpy_string('col'), np.uint32),
(numpy_string('values'), np.float32),
]
array = np.array([
(1, 2, MAX_INT_32, MAX_INT_32, 99),
(3, 4, MAX_INT_32, MAX_INT_32, 99),
(3, 2, MAX_INT_32, MAX_INT_32, 99),
(5, 6, MAX_INT_32, MAX_INT_32, 99),
],
dtype=dtype)
row_dict = {1: 0, 3: 1}
col_dict = {2: 0, 6: 1}
np.save(database.filepath_processed(), array, allow_pickle=False)
values = MatrixBuilder.build([database.filepath_processed()], "values",
"a", "row", "b", "col", row_dict,
col_dict)[0]
self.assertEqual(values.shape, (2, ))
def test_base_class(self):
norm = Normalization(("foo", ))
self.assertEqual(norm.validator, normalization_validator)
self.assertEqual(norm._metadata, normalizations)
self.assertEqual([x[0] for x in norm.dtype_fields],
[numpy_string(x) for x in ('flow', 'index')])
