def test_profile_excludes_citations(
alice_columnar_table_single_batch_context,
alice_columnar_table_single_batch,
):
# Load data context
data_context: DataContext = alice_columnar_table_single_batch_context
# Load profiler configs & loop (run tests for each one)
yaml_config: str = alice_columnar_table_single_batch["profiler_config"]
# Instantiate Profiler
profiler_config: dict = yaml.load(yaml_config)
# Roundtrip through schema validation to remove any illegal fields add/or restore any missing fields.
deserialized_config: dict = ruleBasedProfilerConfigSchema.load(profiler_config)
serialized_config: dict = ruleBasedProfilerConfigSchema.dump(deserialized_config)
# `class_name`/`module_name` are generally consumed through `instantiate_class_from_config`
# so we need to manually remove those values if we wish to use the **kwargs instantiation pattern
serialized_config.pop("class_name")
serialized_config.pop("module_name")
profiler: RuleBasedProfiler = RuleBasedProfiler(
**serialized_config,
data_context=data_context,
)
expectation_suite: ExpectationSuite = profiler.run(
expectation_suite_name=alice_columnar_table_single_batch[
"expected_expectation_suite_name"
],
include_citation=False,
)
assert expectation_suite.meta.get("citations") is None
def test_resolve_config_using_acceptable_arguments(
profiler_with_placeholder_args: RuleBasedProfiler, ) -> None:
old_config: RuleBasedProfilerConfig = profiler_with_placeholder_args.config
# Roundtrip through schema validation to add/or restore any missing fields.
old_deserialized_config: dict = ruleBasedProfilerConfigSchema.load(
old_config.to_json_dict())
old_deserialized_config.pop("class_name")
old_deserialized_config.pop("module_name")
old_config = RuleBasedProfilerConfig(**old_deserialized_config)
# Brand new config is created but existing attributes are unchanged
new_config: RuleBasedProfilerConfig = (
RuleBasedProfilerConfig.resolve_config_using_acceptable_arguments(
profiler=profiler_with_placeholder_args, ))
# Roundtrip through schema validation to add/or restore any missing fields.
# new_deserialized_config: dict = ruleBasedProfilerConfigSchema.load(new_config.to_json_dict())
new_deserialized_config: dict = new_config.to_json_dict()
new_deserialized_config.pop("class_name")
new_deserialized_config.pop("module_name")
new_config = RuleBasedProfilerConfig(**new_deserialized_config)
assert id(old_config) != id(new_config)
assert all(old_config[attr] == new_config[attr]
for attr in ("config_version", "name"))
def test_bobster_profiler_user_workflow_multi_batch_row_count_range_rule_bootstrap_sampling_method(
bobster_columnar_table_multi_batch_normal_mean_5000_stdev_1000_data_context,
bobster_columnar_table_multi_batch_normal_mean_5000_stdev_1000,
):
# Load data context
data_context: DataContext = (
bobster_columnar_table_multi_batch_normal_mean_5000_stdev_1000_data_context
)
# Load profiler configs & loop (run tests for each one)
yaml_config: str = bobster_columnar_table_multi_batch_normal_mean_5000_stdev_1000[
"profiler_config"]
# Instantiate Profiler
profiler_config: CommentedMap = yaml.load(yaml_config)
# Roundtrip through schema validation to remove any illegal fields add/or restore any missing fields.
deserialized_config: dict = ruleBasedProfilerConfigSchema.load(
profiler_config)
serialized_config: dict = ruleBasedProfilerConfigSchema.dump(
deserialized_config)
# `class_name`/`module_name` are generally consumed through `instantiate_class_from_config`
# so we need to manually remove those values if we wish to use the **kwargs instantiation pattern
serialized_config.pop("class_name")
serialized_config.pop("module_name")
profiler: RuleBasedProfiler = RuleBasedProfiler(
**serialized_config,
data_context=data_context,
)
expectation_suite: ExpectationSuite = profiler.run(
expectation_suite_name=
bobster_columnar_table_multi_batch_normal_mean_5000_stdev_1000[
"test_configuration_bootstrap_sampling_method"]
["expectation_suite_name"], )
expect_table_row_count_to_be_between_expectation_configuration_kwargs: dict = (
expectation_suite.to_json_dict()["expectations"][0]["kwargs"])
min_value: int = (
expect_table_row_count_to_be_between_expectation_configuration_kwargs[
"min_value"])
max_value: int = (
expect_table_row_count_to_be_between_expectation_configuration_kwargs[
"max_value"])
assert (bobster_columnar_table_multi_batch_normal_mean_5000_stdev_1000[
"test_configuration_bootstrap_sampling_method"]
["expect_table_row_count_to_be_between_min_value_mean_value"] <
min_value <
bobster_columnar_table_multi_batch_normal_mean_5000_stdev_1000[
"test_configuration_bootstrap_sampling_method"]
["expect_table_row_count_to_be_between_mean_value"])
assert (bobster_columnar_table_multi_batch_normal_mean_5000_stdev_1000[
"test_configuration_bootstrap_sampling_method"]
["expect_table_row_count_to_be_between_mean_value"] < max_value <
bobster_columnar_table_multi_batch_normal_mean_5000_stdev_1000[
"test_configuration_bootstrap_sampling_method"]
["expect_table_row_count_to_be_between_max_value_mean_value"])
def test_bobby_profiler_user_workflow_multi_batch_row_count_range_rule_and_column_ranges_rule_oneshot_sampling_method(
bobby_columnar_table_multi_batch_deterministic_data_context,
bobby_columnar_table_multi_batch,
):
# Load data context
data_context: DataContext = (
bobby_columnar_table_multi_batch_deterministic_data_context)
# Load profiler configs & loop (run tests for each one)
yaml_config: str = bobby_columnar_table_multi_batch["profiler_config"]
# Instantiate Profiler
profiler_config: dict = yaml.load(yaml_config)
# Roundtrip through schema validation to remove any illegal fields add/or restore any missing fields.
deserialized_config: dict = ruleBasedProfilerConfigSchema.load(
profiler_config)
serialized_config: dict = ruleBasedProfilerConfigSchema.dump(
deserialized_config)
# `class_name`/`module_name` are generally consumed through `instantiate_class_from_config`
# so we need to manually remove those values if we wish to use the **kwargs instantiation pattern
serialized_config.pop("class_name")
serialized_config.pop("module_name")
profiler: RuleBasedProfiler = RuleBasedProfiler(
**serialized_config,
data_context=data_context,
)
expectation_suite: ExpectationSuite = profiler.run(
expectation_suite_name=bobby_columnar_table_multi_batch[
"test_configuration_oneshot_sampling_method"]
["expectation_suite_name"],
include_citation=True,
)
assert (expectation_suite == bobby_columnar_table_multi_batch[
"test_configuration_oneshot_sampling_method"]
["expected_expectation_suite"])
def test_notebook_execution_rule_based_profiler_with_pandas_backend(
titanic_v013_multi_datasource_pandas_data_context_with_checkpoints_v1_with_empty_store_stats_enabled,
bobby_columnar_table_multi_batch,
):
"""
To set this test up we:
- create a suite using Rule-Based Profiler
- verify that no validations have happened
- create the suite edit notebook by hijacking the private cli method
We then:
- execute that notebook (Note this will raise various errors like
CellExecutionError if any cell in the notebook fails
- create a new context from disk
- verify that a validation has been run with our expectation suite
"""
context: DataContext = titanic_v013_multi_datasource_pandas_data_context_with_checkpoints_v1_with_empty_store_stats_enabled
root_dir: str = context.root_directory
uncommitted_dir: str = os.path.join(root_dir, "uncommitted")
expectation_suite_name: str = "warning"
context.create_expectation_suite(
expectation_suite_name=expectation_suite_name)
batch_request: dict = {
"datasource_name": "my_datasource",
"data_connector_name": "my_basic_data_connector",
"data_asset_name": "Titanic_1912",
}
# Sanity check test setup
original_suite: ExpectationSuite = context.get_expectation_suite(
expectation_suite_name=expectation_suite_name)
assert len(original_suite.expectations) == 0
assert context.list_expectation_suite_names() == [expectation_suite_name]
assert context.list_datasources() == [
{
"name": "my_datasource",
"class_name": "Datasource",
"module_name": "great_expectations.datasource",
"execution_engine": {
"class_name": "PandasExecutionEngine",
"module_name": "great_expectations.execution_engine",
},
"data_connectors": {
"my_basic_data_connector": {
"module_name":
"great_expectations.datasource.data_connector",
"base_directory": f"{root_dir}/../data/titanic",
"default_regex": {
"pattern": "(.*)\\.csv",
"group_names": ["data_asset_name"],
},
"class_name": "InferredAssetFilesystemDataConnector",
},
"my_special_data_connector": {
"glob_directive": "*.csv",
"assets": {
"users": {
"pattern":
"(.+)_(\\d+)_(\\d+)\\.csv",
"group_names": ["name", "timestamp", "size"],
"class_name":
"Asset",
"base_directory":
f"{root_dir}/../data/titanic",
"module_name":
"great_expectations.datasource.data_connector.asset",
}
},
"module_name":
"great_expectations.datasource.data_connector",
"base_directory": f"{root_dir}/../data/titanic",
"default_regex": {
"pattern": "(.+)\\.csv",
"group_names": ["name"]
},
"class_name": "ConfiguredAssetFilesystemDataConnector",
},
"my_other_data_connector": {
"glob_directive": "*.csv",
"assets": {
"users": {
"class_name":
"Asset",
"module_name":
"great_expectations.datasource.data_connector.asset",
}
},
"module_name":
"great_expectations.datasource.data_connector",
"base_directory": f"{root_dir}/../data/titanic",
"default_regex": {
"pattern": "(.+)\\.csv",
"group_names": ["name"]
},
"class_name": "ConfiguredAssetFilesystemDataConnector",
},
"my_runtime_data_connector": {
"module_name":
"great_expectations.datasource.data_connector",
"batch_identifiers":
["pipeline_stage_name", "airflow_run_id"],
"class_name": "RuntimeDataConnector",
},
},
},
{
"name": "my_additional_datasource",
"class_name": "Datasource",
"module_name": "great_expectations.datasource",
"execution_engine": {
"module_name": "great_expectations.execution_engine",
"class_name": "PandasExecutionEngine",
},
"data_connectors": {
"my_additional_data_connector": {
"module_name":
"great_expectations.datasource.data_connector",
"default_regex": {
"pattern": "(.*)\\.csv",
"group_names": ["data_asset_name"],
},
"base_directory": f"{root_dir}/../data/titanic",
"class_name": "InferredAssetFilesystemDataConnector",
}
},
},
]
assert context.get_validation_result(
expectation_suite_name="warning") == {}
# Load profiler configs & loop (run tests for each one)
yaml_config: str = bobby_columnar_table_multi_batch["profiler_config"]
# Instantiate Profiler
profiler_config: dict = yaml.load(yaml_config)
# Roundtrip through schema validation to remove any illegal fields add/or restore any missing fields.
deserialized_config: dict = ruleBasedProfilerConfigSchema.load(
profiler_config)
serialized_config: dict = ruleBasedProfilerConfigSchema.dump(
deserialized_config)
# `class_name`/`module_name` are generally consumed through `instantiate_class_from_config`
# so we need to manually remove those values if we wish to use the **kwargs instantiation pattern
serialized_config.pop("class_name")
serialized_config.pop("module_name")
profiler: RuleBasedProfiler = RuleBasedProfiler(
**serialized_config,
data_context=context,
)
profiler_name: str = "bobby_user_workflow"
context.save_profiler(
profiler=profiler,
name=profiler_name,
)
# Create notebook
# do not want to actually send usage_message, since the function call is not the result of actual usage
_suite_edit_workflow(
context=context,
expectation_suite_name=expectation_suite_name,
profile=True,
profiler_name=profiler_name,
usage_event="test_notebook_execution",
interactive_mode=CLISuiteInteractiveFlagCombinations.
UNPROMPTED_INTERACTIVE_FALSE_MANUAL_TRUE,
no_jupyter=True,
create_if_not_exist=False,
datasource_name=None,
batch_request=batch_request,
additional_batch_request_args=None,
suppress_usage_message=True,
assume_yes=True,
)
edit_notebook_path: str = os.path.join(uncommitted_dir,
"edit_warning.ipynb")
assert os.path.isfile(edit_notebook_path)
run_notebook(
notebook_path=edit_notebook_path,
notebook_dir=uncommitted_dir,
string_to_be_replaced=
"context.open_data_docs(resource_identifier=validation_result_identifier)",
replacement_string="",
)
# Assertions about output
context = DataContext(context_root_dir=root_dir)
obs_validation_result: ExpectationSuiteValidationResult = (
context.get_validation_result(expectation_suite_name="warning"))
assert obs_validation_result.statistics == {
"evaluated_expectations": 13,
"successful_expectations": 13,
"unsuccessful_expectations": 0,
"success_percent": 100.0,
}
expected_expectation_configurations: List[ExpectationConfiguration] = [
ExpectationConfiguration(
**{
"meta": {
"profiler_details": {
"metric_configuration": {
"domain_kwargs": {},
"metric_dependencies": None,
"metric_name": "table.row_count",
"metric_value_kwargs": None,
},
"num_batches": 1,
}
},
"kwargs": {
"max_value": 1313,
"min_value": 1313
},
"expectation_type": "expect_table_row_count_to_be_between",
}),
ExpectationConfiguration(
**{
"meta": {
"profiler_details": {
"metric_configuration": {
"domain_kwargs": {
"column": "Unnamed: 0"
},
"metric_dependencies": None,
"metric_name": "column.min",
"metric_value_kwargs": None,
},
"num_batches": 1,
}
},
"kwargs": {
"column": "Unnamed: 0",
"max_value": 1,
"min_value": 1,
"mostly": 1.0,
},
"expectation_type": "expect_column_min_to_be_between",
}),
ExpectationConfiguration(
**{
"meta": {
"profiler_details": {
"metric_configuration": {
"domain_kwargs": {
"column": "Unnamed: 0"
},
"metric_dependencies": None,
"metric_name": "column.max",
"metric_value_kwargs": None,
},
"num_batches": 1,
}
},
"kwargs": {
"column": "Unnamed: 0",
"max_value": 1313,
"min_value": 1313,
"mostly": 1.0,
},
"expectation_type": "expect_column_max_to_be_between",
}),
ExpectationConfiguration(
**{
"meta": {
"profiler_details": {
"metric_configuration": {
"domain_kwargs": {
"column": "Age"
},
"metric_dependencies": None,
"metric_name": "column.min",
"metric_value_kwargs": None,
},
"num_batches": 1,
}
},
"kwargs": {
"column": "Age",
"max_value": 0.17,
"min_value": 0.17,
"mostly": 1.0,
},
"expectation_type": "expect_column_min_to_be_between",
}),
ExpectationConfiguration(
**{
"meta": {
"profiler_details": {
"metric_configuration": {
"domain_kwargs": {
"column": "Age"
},
"metric_dependencies": None,
"metric_name": "column.max",
"metric_value_kwargs": None,
},
"num_batches": 1,
}
},
"kwargs": {
"column": "Age",
"max_value": 71.0,
"min_value": 71.0,
"mostly": 1.0,
},
"expectation_type": "expect_column_max_to_be_between",
}),
ExpectationConfiguration(
**{
"meta": {
"profiler_details": {
"metric_configuration": {
"domain_kwargs": {
"column": "Survived"
},
"metric_dependencies": None,
"metric_name": "column.min",
"metric_value_kwargs": None,
},
"num_batches": 1,
}
},
"kwargs": {
"column": "Survived",
"max_value": 0,
"min_value": 0,
"mostly": 1.0,
},
"expectation_type": "expect_column_min_to_be_between",
}),
ExpectationConfiguration(
**{
"meta": {
"profiler_details": {
"metric_configuration": {
"domain_kwargs": {
"column": "Survived"
},
"metric_dependencies": None,
"metric_name": "column.max",
"metric_value_kwargs": None,
},
"num_batches": 1,
}
},
"kwargs": {
"column": "Survived",
"max_value": 1,
"min_value": 1,
"mostly": 1.0,
},
"expectation_type": "expect_column_max_to_be_between",
}),
ExpectationConfiguration(
**{
"meta": {
"profiler_details": {
"metric_configuration": {
"domain_kwargs": {
"column": "SexCode"
},
"metric_dependencies": None,
"metric_name": "column.min",
"metric_value_kwargs": None,
},
"num_batches": 1,
}
},
"kwargs": {
"column": "SexCode",
"max_value": 0,
"min_value": 0,
"mostly": 1.0,
},
"expectation_type": "expect_column_min_to_be_between",
}),
ExpectationConfiguration(
**{
"meta": {
"profiler_details": {
"metric_configuration": {
"domain_kwargs": {
"column": "SexCode"
},
"metric_dependencies": None,
"metric_name": "column.max",
"metric_value_kwargs": None,
},
"num_batches": 1,
}
},
"kwargs": {
"column": "SexCode",
"max_value": 1,
"min_value": 1,
"mostly": 1.0,
},
"expectation_type": "expect_column_max_to_be_between",
}),
ExpectationConfiguration(
**{
"meta": {},
"kwargs": {
"column": "PClass",
"value_set": [
"*",
"1st",
"2nd",
"3rd",
],
},
"expectation_type": "expect_column_values_to_be_in_set",
}),
ExpectationConfiguration(
**{
"meta": {},
"kwargs": {
"column": "Sex",
"value_set": ["female", "male"]
},
"expectation_type": "expect_column_values_to_be_in_set",
}),
ExpectationConfiguration(
**{
"meta": {},
"kwargs": {
"column": "Survived",
"value_set": [0, 1]
},
"expectation_type": "expect_column_values_to_be_in_set",
}),
ExpectationConfiguration(
**{
"meta": {},
"kwargs": {
"column": "SexCode",
"value_set": [0, 1]
},
"expectation_type": "expect_column_values_to_be_in_set",
}),
]
suite: ExpectationSuite = context.get_expectation_suite(
expectation_suite_name=expectation_suite_name)
expectation_configurations: List[ExpectationConfiguration] = []
expectation_configuration: ExpectationConfiguration
for expectation_configuration in suite.expectations:
kwargs: dict = expectation_configuration.kwargs
key: str
value: Any
kwargs = {
key: sorted(value) if isinstance(value,
(list, set, tuple)) else value
for key, value in kwargs.items()
}
expectation_configuration.kwargs = kwargs
expectation_configurations.append(expectation_configuration)
assert expectation_configurations == expected_expectation_configurations
def test_quentin_profiler_user_workflow_multi_batch_quantiles_value_ranges_rule(
quentin_columnar_table_multi_batch_data_context,
quentin_columnar_table_multi_batch,
):
# Load data context
data_context: DataContext = quentin_columnar_table_multi_batch_data_context
# Load profiler configs & loop (run tests for each one)
yaml_config: str = quentin_columnar_table_multi_batch["profiler_config"]
# Instantiate Profiler
profiler_config: CommentedMap = yaml.load(yaml_config)
# Roundtrip through schema validation to remove any illegal fields add/or restore any missing fields.
deserialized_config: dict = ruleBasedProfilerConfigSchema.load(
profiler_config)
serialized_config: dict = ruleBasedProfilerConfigSchema.dump(
deserialized_config)
# `class_name`/`module_name` are generally consumed through `instantiate_class_from_config`
# so we need to manually remove those values if we wish to use the **kwargs instantiation pattern
serialized_config.pop("class_name")
serialized_config.pop("module_name")
profiler: RuleBasedProfiler = RuleBasedProfiler(
**serialized_config,
data_context=data_context,
)
expectation_suite: ExpectationSuite = profiler.run(
expectation_suite_name=quentin_columnar_table_multi_batch[
"test_configuration"]["expectation_suite_name"], )
expectation_configuration_dict: dict
column_name: str
expectation_kwargs: dict
expect_column_quantile_values_to_be_between_expectation_configurations_kwargs_dict: Dict[
str, dict] = {
expectation_configuration_dict["kwargs"]["column"]:
expectation_configuration_dict["kwargs"]
for expectation_configuration_dict in
expectation_suite.to_json_dict()["expectations"]
}
expect_column_quantile_values_to_be_between_expectation_configurations_value_ranges_by_column: Dict[
str, List[List[Number]]] = {
column_name: expectation_kwargs["value_ranges"]
for column_name, expectation_kwargs in
expect_column_quantile_values_to_be_between_expectation_configurations_kwargs_dict
.items()
}
assert (
expect_column_quantile_values_to_be_between_expectation_configurations_value_ranges_by_column[
"tolls_amount"] ==
quentin_columnar_table_multi_batch["test_configuration"]
["expect_column_quantile_values_to_be_between_quantile_ranges_by_column"]
["tolls_amount"])
# Measure of "closeness" between "actual" and "desired" is computed as: atol + rtol * abs(desired)
# (see "https://numpy.org/doc/stable/reference/generated/numpy.testing.assert_allclose.html" for details).
rtol: float = 1.0e-7
atol: float = 5.0e-2
value_range: List[Number]
paired_quantiles: zip
column_quantiles: List[List[Number]]
idx: int
for (
column_name,
column_quantiles,
) in (expect_column_quantile_values_to_be_between_expectation_configurations_value_ranges_by_column
.items()):
paired_quantiles = zip(
column_quantiles,
quentin_columnar_table_multi_batch["test_configuration"]
["expect_column_quantile_values_to_be_between_quantile_ranges_by_column"]
[column_name],
)
for value_range in list(paired_quantiles):
for idx in range(2):
np.testing.assert_allclose(
actual=value_range[0][idx],
desired=value_range[1][idx],
rtol=rtol,
atol=atol,
err_msg=
f"Actual value of {value_range[0][idx]} differs from expected value of {value_range[1][idx]} by more than {atol + rtol * abs(value_range[1][idx])} tolerance.",
)