def test_indirect_matrices_shape_with_missing_sources(self):
dataset_name = "SAMPLE_WARC"
dataset_builder = DatasetBuilder(dataset_name)
dataset_builder.build()
top = dataset_builder.artifacts[0]
middle = dataset_builder.artifacts[1]
bottom = dataset_builder.artifacts[2]
expected_top_middle_shape = (len(top), len(middle))
expected_middle_bottom_shape = (len(middle), len(bottom))
expected_top_bottom_shape = (len(top), len(bottom))
assert_shapes_equal(
dataset_builder.trace_matrix_map["0-1"].matrix, expected_top_middle_shape
)
assert_shapes_equal(
dataset_builder.trace_matrix_map["0-2"].matrix, expected_top_bottom_shape
)
assert_shapes_equal(
dataset_builder.trace_matrix_map["1-2"].matrix,
expected_middle_bottom_shape,
)
self.assertEqual(
dataset_builder.trace_matrix_map["1-2"].matrix.shape,
expected_middle_bottom_shape,
)
def test_creating_missing_matrices(self):
"""
For MockDataset, removes trace matrix 0-2 from structure definition and checks that builder is able to
construct one using transitive traces from the other matrices.
:return:
"""
dataset_builder = DatasetBuilder("MockDataset")
dataset_builder.structure_definition["traces"]["0-2"] = None
dataset_builder.build()
trace_matrix_builder = dataset_builder.trace_matrix_builder
graph = trace_matrix_builder.create_trace_matrix_dependency_graph()
self.assertEqual(
3, len(trace_matrix_builder.trace_matrix_map.get_trace_ids()))
self.assertIsNotNone(trace_matrix_builder.trace_matrix_map["0-2"])
self.assertIsNotNone(
trace_matrix_builder.trace_matrix_map["0-2"].matrix)
self.assertEqual(
1, trace_matrix_builder.trace_matrix_map["0-2"].matrix[0, 0])
self.assertEqual(
2, trace_matrix_builder.trace_matrix_map["0-2"].matrix.sum(
axis=1).sum())
self.assertEqual(3, len(graph.es))
def test_create_transitive_trace_matrices_with_warc(self):
dataset_builder = DatasetBuilder("SAMPLE_WARC")
dataset_builder.build()
trace_matrix_creator = dataset_builder.trace_matrix_builder
graph = (dataset_builder.trace_matrix_builder.
create_trace_matrix_dependency_graph())
self.assertIsNotNone(trace_matrix_creator.trace_matrix_map["1-2"])
self.assertEqual(3, len(graph.es))
def run(self) -> Table:
with create_loading_bar(
EXPERIMENT_NAME, DATASET_COLUMN_ORDER, len(DATASET_COLUMN_ORDER)
) as d_iterable:
for dataset_name in d_iterable:
builder = DatasetBuilder(dataset_name)
builder.build()
builder.export()
print(f"{dataset_name} exported.")
return Table()
def test_export_dataset(self):
"""
Tests that after MockDataset is exported all required folders have been updated.
:return:
"""
# Setup
dataset_name = "MockDataset"
dataset_builder = DatasetBuilder(dataset_name)
dataset_builder.build()
# Work
dataset_builder.export()
folders = ["Artifacts", "Oracles"]
for folder_rel_path in folders:
path_to_folder = os.path.join(
PATH_TO_SAMPLE_DATASETS, dataset_name, folder_rel_path
)
check_folder_has_updated(path_to_folder)
def test_normalize_original_matrices(self):
db = DatasetBuilder("MockDataset")
db.build()
self.assertEqual(1, db.trace_matrix_builder["0-2"].matrix[0][2])
from api.datasets.builder.dataset_builder import DatasetBuilder
from api.datasets.dataset import Dataset
from api.tracer import Tracer
DATASET_NAME = "IllustrativeExample"
TOP_TECHNIQUE_NAME = "(. (VSM NT) (0 1))"
BOTTOM_TECHNIQUE_NAME = "(. (VSM NT) (1 2))"
DIRECT_TECHNIQUE_NAME = "(. (VSM NT) (0 2))"
TECHNIQUE_NAME = "(x (MAX INDEPENDENT) ((. (VSM NT) (0 1)) (. (VSM NT) (1 2))))"
REBUILD = False
if __name__ == "__main__":
if REBUILD:
dataset_builder = DatasetBuilder(DATASET_NAME)
dataset_builder.build()
dataset_builder.export()
dataset = Dataset(DATASET_NAME)
tracer = Tracer()
top_technique_data = tracer.get_technique_data(DATASET_NAME,
TOP_TECHNIQUE_NAME)
bottom_technique_data = tracer.get_technique_data(DATASET_NAME,
BOTTOM_TECHNIQUE_NAME)
direct_technique_data = tracer.get_technique_data(DATASET_NAME,
DIRECT_TECHNIQUE_NAME)
top_score = top_technique_data.similarity_matrix[0][0]
bottom_score = bottom_technique_data.similarity_matrix[0][0]
transitive_score = top_score * bottom_score
direct_score = direct_technique_data.similarity_matrix[0][0]