def _run_tests(args):
logging.info(f"loading entity labels...")
entity_labels = EntityLabelFileParser.create_from_file(args.entity_labels)
logging.info(f"loading task categories...")
categories = TaskCategoryFileParser.create_categories_from_file(
args.test_set_config, entity_labels)
logging.info(f"loading task configurations...")
task_configurations = TaskConfigurationFileParser.create_configurations_from_file(
args.test_set_config)
logging.info(f"loading entity linkings...")
entity_linkings = EntityLinkingFileParser.create_from_file(
args.entity_mapping)
logging.info(f"loading embeddings...")
embeddings = EmbeddingFileParser.create_from_file(args.embeddings)
test_configuration = TestConfiguration(embeddings, entity_linkings,
entity_labels, categories,
task_configurations)
logging.info(f"starting test execution...")
test_executor = TestExecutor(test_configuration)
test_category_results = list(test_executor.run())
logging.info(f"storing results...")
logging.info(f"Result size: {len(test_category_results)}")
args.test_results.parent.mkdir(parents=True, exist_ok=True)
with bz2.BZ2File(str(args.test_results.absolute()), "w") as output_stream:
pickle.dump(test_category_results, output_stream)
def test_create_categories_from_file(self):
file = self._random_test_file()
with open(file, "w+") as file_output:
print(f"""
configuration:
categories:
- category:
name: category_name_1
enabled: true
entities: {self.empty_file.absolute()}
tasks:
- task:
name: task_name_1
type: analogy
test_set: {self.empty_file.absolute()}
- task:
name: task_name_2
type: cosine_similarity
test_set: {self.empty_file.absolute()}
categories:
- category:
name: sub_category_name_1
enabled: false
entities: {self.empty_file.absolute()}
tasks:
categories:""",
file=file_output)
entity_labels = EntityLabels()
categories = TaskCategoryFileParser.create_categories_from_file(
file, entity_labels)
assert len(categories) == 1
category = categories[0]
assert category.name == "category_name_1"
assert category.enabled
assert len(category.tasks) == 2
assert len(categories) == 1
task = category.tasks[0]
assert task.name == "analogy: task_name_1"
assert task.__class__ == AnalogyTask
assert task.test_set == self.empty_file
task = category.tasks[1]
assert task.name == "cosine_similarity: task_name_2"
assert task.__class__ == CosineSimilarityTask
assert task.test_set == self.empty_file
subcategory = category.categories[0]
assert subcategory.name == "sub_category_name_1"
assert not subcategory.enabled
assert len(subcategory.tasks) == 0
assert len(subcategory.categories) == 0
def test_create_from_file_empty_categories(self):
file = self._random_test_file()
with open(file, "w+") as file_output:
print(f"""
configuration:
categories:""",
file=file_output)
entity_labels = EntityLabels()
categories = TaskCategoryFileParser.create_categories_from_file(
file, entity_labels)
assert len(categories) == 0
def _create_test_configuration(self):
task_configs = TaskConfigurationFileParser.create_configurations_from_file(
Path(self.resource_directory, "configuration.yaml"))
categories = TaskCategoryFileParser.create_categories_from_file(
Path(self.resource_directory, "configuration.yaml"),
EntityLabels())
entity_linking = EntityLinkingFileParser.create_from_file(
Path(self.resource_directory, "linking.csv"))
embedding = EmbeddingFileParser.create_from_file(
Path(self.resource_directory, "embedding"))
return TestConfiguration(embedding, entity_linking, EntityLabels(),
categories, task_configs)
def test_enabled_is_missing(self):
file = self._random_test_file()
with open(file, "w+") as file_output:
print(f"""
configuration:
categories:
- category:
name: category_name_1
tasks:
categories:""",
file=file_output)
with pytest.raises(KeyError):
entity_labels = EntityLabels()
categories = TaskCategoryFileParser.create_categories_from_file(
file, entity_labels)