def test_ValidationsStore_with_InMemoryStoreBackend():
my_store = ValidationsStore(
store_backend={
"module_name": "great_expectations.data_context.store",
"class_name": "InMemoryStoreBackend",
})
with pytest.raises(TypeError):
my_store.get("not_a_ValidationResultIdentifier")
ns_1 = ValidationResultIdentifier.from_tuple(
("a", "b", "c", "quarantine", "prod-100"))
my_store.set(ns_1, ExpectationSuiteValidationResult(success=True))
assert my_store.get(ns_1) == ExpectationSuiteValidationResult(
success=True, statistics={}, results=[])
ns_2 = ValidationResultIdentifier.from_tuple(
("a", "b", "c", "quarantine", "prod-200"))
my_store.set(ns_2, ExpectationSuiteValidationResult(success=False))
assert my_store.get(ns_2) == ExpectationSuiteValidationResult(
success=False, statistics={}, results=[])
assert set(my_store.list_keys()) == {
ns_1,
ns_2,
}
def test_ValidationsStore_with_InMemoryStoreBackend():
my_store = ValidationsStore(
store_backend={
"module_name": "great_expectations.data_context.store",
"class_name": "InMemoryStoreBackend",
}
)
with pytest.raises(TypeError):
my_store.get("not_a_ValidationResultIdentifier")
ns_1 = ValidationResultIdentifier.from_tuple(
(
"a",
"b",
"c",
"quarantine",
datetime.datetime.now(datetime.timezone.utc),
"prod-100",
)
)
my_store.set(ns_1, ExpectationSuiteValidationResult(success=True))
assert my_store.get(ns_1) == ExpectationSuiteValidationResult(
success=True, statistics={}, results=[]
)
ns_2 = ValidationResultIdentifier.from_tuple(
(
"a",
"b",
"c",
"quarantine",
datetime.datetime.now(datetime.timezone.utc),
"prod-200",
)
)
my_store.set(ns_2, ExpectationSuiteValidationResult(success=False))
assert my_store.get(ns_2) == ExpectationSuiteValidationResult(
success=False, statistics={}, results=[]
)
assert set(my_store.list_keys()) == {
ns_1,
ns_2,
}
"""
What does this test and why?
A Store should be able to report it's store_backend_id
which is set when the StoreBackend is instantiated.
"""
# Check that store_backend_id exists can be read
assert my_store.store_backend_id is not None
# Check that store_backend_id is a valid UUID
assert test_utils.validate_uuid4(my_store.store_backend_id)
def _build_validation_and_profiling_result_site_keys(
self,
skip_and_clean_missing: bool) -> List[ValidationResultIdentifier]:
validation_and_profiling_result_site_keys = []
validations = self.site_section_builders_config.get(
"validations", "None")
profiling = self.site_section_builders_config.get("profiling", "None")
if (validations and validations not in FALSEY_YAML_STRINGS) or (
profiling and profiling not in FALSEY_YAML_STRINGS):
source_store = ("validations" if
(validations
and validations not in FALSEY_YAML_STRINGS) else
"profiling")
validation_and_profiling_result_source_keys = set(
self.data_context.stores[self.site_section_builders_config[
source_store].get("source_store_name")].list_keys())
validation_and_profiling_result_site_keys = [
ValidationResultIdentifier.from_tuple(validation_result_tuple)
for validation_result_tuple in self.target_store.
store_backends[ValidationResultIdentifier].list_keys()
]
if skip_and_clean_missing:
cleaned_keys = []
for (validation_result_site_key
) in validation_and_profiling_result_site_keys:
if (validation_result_site_key not in
validation_and_profiling_result_source_keys):
self.target_store.store_backends[
ValidationResultIdentifier].remove_key(
validation_result_site_key)
else:
cleaned_keys.append(validation_result_site_key)
validation_and_profiling_result_site_keys = cleaned_keys
return validation_and_profiling_result_site_keys
def test_ValidationsStore_with_TupleFileSystemStoreBackend(tmp_path_factory):
path = str(
tmp_path_factory.mktemp(
"test_ValidationResultStore_with_TupleFileSystemStoreBackend__dir")
)
project_path = str(tmp_path_factory.mktemp("my_dir"))
my_store = ValidationsStore(
store_backend={
"module_name": "great_expectations.data_context.store",
"class_name": "TupleFilesystemStoreBackend",
"base_directory": "my_store/",
},
runtime_environment={"root_directory": path},
)
with pytest.raises(TypeError):
my_store.get("not_a_ValidationResultIdentifier")
ns_1 = ValidationResultIdentifier(
expectation_suite_identifier=ExpectationSuiteIdentifier(
"asset.quarantine"),
run_id="prod-100",
batch_identifier="batch_id",
)
my_store.set(ns_1, ExpectationSuiteValidationResult(success=True))
assert my_store.get(ns_1) == ExpectationSuiteValidationResult(
success=True, statistics={}, results=[])
ns_2 = ValidationResultIdentifier.from_tuple((
"asset",
"quarantine",
"prod-20",
datetime.datetime.now(datetime.timezone.utc),
"batch_id",
))
my_store.set(ns_2, ExpectationSuiteValidationResult(success=False))
assert my_store.get(ns_2) == ExpectationSuiteValidationResult(
success=False, statistics={}, results=[])
print(my_store.list_keys())
assert set(my_store.list_keys()) == {
ns_1,
ns_2,
}
print(gen_directory_tree_str(path))
assert (gen_directory_tree_str(path) == """\
test_ValidationResultStore_with_TupleFileSystemStoreBackend__dir0/
my_store/
asset/
quarantine/
prod-100/
20190926T134241.000000Z/
batch_id.json
prod-20/
20190926T134241.000000Z/
batch_id.json
""")
def test_HtmlSiteStore_filesystem_backend(tmp_path_factory):
path = str(tmp_path_factory.mktemp('test_HtmlSiteStore_with_TupleFileSystemStoreBackend__dir'))
my_store = HtmlSiteStore(
store_backend={
"class_name": "TupleFilesystemStoreBackend",
"base_directory": "my_store"
},
runtime_environment={
"root_directory": path
}
)
with pytest.raises(TypeError):
my_store.get("not_a_ValidationResultIdentifier")
with pytest.raises(ValidationError):
my_store.get(validationResultIdentifierSchema.load({}).data)
ns_1 = SiteSectionIdentifier(
site_section_name="validations",
resource_identifier=ValidationResultIdentifier.from_tuple(('a', 'b', 'c', 'quarantine', 'prod-100'))
)
my_store.set(ns_1, "aaa")
# assert my_store.get(ns_1) == "aaa"
ns_2 = SiteSectionIdentifier(
site_section_name="validations",
resource_identifier=ValidationResultIdentifier.from_tuple(('a', 'b', 'c', 'quarantine', 'prod-20'))
)
my_store.set(ns_2, "bbb")
# assert my_store.get(ns_2) == {"B": "bbb"}
print(my_store.list_keys())
# WARNING: OBSERVE THAT SITE_SECTION_NAME IS LOST IN THE CALL TO LIST_KEYS
assert set(my_store.list_keys()) == {
ns_1.resource_identifier,
ns_2.resource_identifier,
}
print(gen_directory_tree_str(path))
assert gen_directory_tree_str(path) == """\
def build(self):
# Loop over sections in the HtmlStore
logger.debug("DefaultSiteIndexBuilder.build")
expectation_suite_keys = [
ExpectationSuiteIdentifier.from_tuple(expectation_suite_tuple)
for expectation_suite_tuple in self.target_store.store_backends[
ExpectationSuiteIdentifier
].list_keys()
]
validation_and_profiling_result_keys = [
ValidationResultIdentifier.from_tuple(validation_result_tuple)
for validation_result_tuple in self.target_store.store_backends[
ValidationResultIdentifier
].list_keys()
]
profiling_result_keys = [
validation_result_key
for validation_result_key in validation_and_profiling_result_keys
if validation_result_key.run_id.run_name == "profiling"
]
validation_result_keys = [
validation_result_key
for validation_result_key in validation_and_profiling_result_keys
if validation_result_key.run_id.run_name != "profiling"
]
validation_result_keys = sorted(
validation_result_keys, key=lambda x: x.run_id.run_time, reverse=True
)
if self.validation_results_limit:
validation_result_keys = validation_result_keys[
: self.validation_results_limit
]
index_links_dict = OrderedDict()
index_links_dict["site_name"] = self.site_name
if self.show_how_to_buttons:
index_links_dict["cta_object"] = self.get_calls_to_action()
for expectation_suite_key in expectation_suite_keys:
self.add_resource_info_to_index_links_dict(
index_links_dict=index_links_dict,
expectation_suite_name=expectation_suite_key.expectation_suite_name,
section_name="expectations",
)
for profiling_result_key in profiling_result_keys:
try:
validation = self.data_context.get_validation_result(
batch_identifier=profiling_result_key.batch_identifier,
expectation_suite_name=profiling_result_key.expectation_suite_identifier.expectation_suite_name,
run_id=profiling_result_key.run_id,
validations_store_name=self.source_stores.get("profiling"),
)
batch_kwargs = validation.meta.get("batch_kwargs", {})
self.add_resource_info_to_index_links_dict(
index_links_dict=index_links_dict,
expectation_suite_name=profiling_result_key.expectation_suite_identifier.expectation_suite_name,
section_name="profiling",
batch_identifier=profiling_result_key.batch_identifier,
run_id=profiling_result_key.run_id,
run_time=profiling_result_key.run_id.run_time,
run_name=profiling_result_key.run_id.run_name,
asset_name=batch_kwargs.get("data_asset_name"),
batch_kwargs=batch_kwargs,
)
except Exception:
error_msg = "Profiling result not found: {0:s} - skipping".format(
str(profiling_result_key.to_tuple())
)
logger.warning(error_msg)
for validation_result_key in validation_result_keys:
try:
validation = self.data_context.get_validation_result(
batch_identifier=validation_result_key.batch_identifier,
expectation_suite_name=validation_result_key.expectation_suite_identifier.expectation_suite_name,
run_id=validation_result_key.run_id,
validations_store_name=self.source_stores.get("validations"),
)
validation_success = validation.success
batch_kwargs = validation.meta.get("batch_kwargs", {})
self.add_resource_info_to_index_links_dict(
index_links_dict=index_links_dict,
expectation_suite_name=validation_result_key.expectation_suite_identifier.expectation_suite_name,
section_name="validations",
batch_identifier=validation_result_key.batch_identifier,
run_id=validation_result_key.run_id,
validation_success=validation_success,
run_time=validation_result_key.run_id.run_time,
run_name=validation_result_key.run_id.run_name,
asset_name=batch_kwargs.get("data_asset_name"),
batch_kwargs=batch_kwargs,
)
except Exception:
error_msg = "Validation result not found: {0:s} - skipping".format(
str(validation_result_key.to_tuple())
)
logger.warning(error_msg)
try:
rendered_content = self.renderer_class.render(index_links_dict)
viewable_content = self.view_class.render(
rendered_content,
data_context_id=self.data_context_id,
show_how_to_buttons=self.show_how_to_buttons,
)
except Exception as e:
exception_message = f"""\
An unexpected Exception occurred during data docs rendering. Because of this error, certain parts of data docs will \
not be rendered properly and/or may not appear altogether. Please use the trace, included in this message, to \
diagnose and repair the underlying issue. Detailed information follows:
"""
exception_traceback = traceback.format_exc()
exception_message += (
f'{type(e).__name__}: "{str(e)}". Traceback: "{exception_traceback}".'
)
logger.error(exception_message, e, exc_info=True)
return (self.target_store.write_index_page(viewable_content), index_links_dict)
def build(self, skip_and_clean_missing=True, build_index: bool = True):
"""
:param skip_and_clean_missing: if True, target html store keys without corresponding source store keys will
be skipped and removed from the target store
:param build_index: a flag if False, skips building the index page
:return: tuple(index_page_url, index_links_dict)
"""
# Loop over sections in the HtmlStore
logger.debug("DefaultSiteIndexBuilder.build")
if not build_index:
logger.debug("Skipping index rendering")
return None, None
index_links_dict = OrderedDict()
index_links_dict["site_name"] = self.site_name
if self.show_how_to_buttons:
index_links_dict["cta_object"] = self.get_calls_to_action()
if (
# TODO why is this duplicated?
self.site_section_builders_config.get("expectations", "None")
and self.site_section_builders_config.get(
"expectations", "None") not in FALSEY_YAML_STRINGS):
expectation_suite_source_keys = self.data_context.stores[
self.site_section_builders_config["expectations"].get(
"source_store_name")].list_keys()
expectation_suite_site_keys = [
ExpectationSuiteIdentifier.from_tuple(expectation_suite_tuple)
for expectation_suite_tuple in self.target_store.
store_backends[ExpectationSuiteIdentifier].list_keys()
]
if skip_and_clean_missing:
cleaned_keys = []
for expectation_suite_site_key in expectation_suite_site_keys:
if expectation_suite_site_key not in expectation_suite_source_keys:
self.target_store.store_backends[
ExpectationSuiteIdentifier].remove_key(
expectation_suite_site_key)
else:
cleaned_keys.append(expectation_suite_site_key)
expectation_suite_site_keys = cleaned_keys
for expectation_suite_key in expectation_suite_site_keys:
self.add_resource_info_to_index_links_dict(
index_links_dict=index_links_dict,
expectation_suite_name=expectation_suite_key.
expectation_suite_name,
section_name="expectations",
)
validation_and_profiling_result_site_keys = []
if (
# TODO why is this duplicated?
self.site_section_builders_config.get("validations", "None")
and self.site_section_builders_config.get(
"validations", "None") not in FALSEY_YAML_STRINGS
or self.site_section_builders_config.get("profiling", "None")
and self.site_section_builders_config.get(
"profiling", "None") not in FALSEY_YAML_STRINGS):
source_store = (
"validations"
# TODO why is this duplicated?
if
self.site_section_builders_config.get("validations", "None")
and self.site_section_builders_config.get(
"validations", "None") not in FALSEY_YAML_STRINGS else
"profiling")
validation_and_profiling_result_source_keys = self.data_context.stores[
self.site_section_builders_config[source_store].get(
"source_store_name")].list_keys()
validation_and_profiling_result_site_keys = [
ValidationResultIdentifier.from_tuple(validation_result_tuple)
for validation_result_tuple in self.target_store.
store_backends[ValidationResultIdentifier].list_keys()
]
if skip_and_clean_missing:
cleaned_keys = []
for (validation_result_site_key
) in validation_and_profiling_result_site_keys:
if (validation_result_site_key not in
validation_and_profiling_result_source_keys):
self.target_store.store_backends[
ValidationResultIdentifier].remove_key(
validation_result_site_key)
else:
cleaned_keys.append(validation_result_site_key)
validation_and_profiling_result_site_keys = cleaned_keys
if (
# TODO why is this duplicated?
self.site_section_builders_config.get("profiling", "None")
and self.site_section_builders_config.get(
"profiling", "None") not in FALSEY_YAML_STRINGS):
profiling_run_name_filter = self.site_section_builders_config[
"profiling"]["run_name_filter"]
profiling_result_site_keys = [
validation_result_key for validation_result_key in
validation_and_profiling_result_site_keys
if resource_key_passes_run_name_filter(
validation_result_key, profiling_run_name_filter)
]
for profiling_result_key in profiling_result_site_keys:
try:
validation = self.data_context.get_validation_result(
batch_identifier=profiling_result_key.batch_identifier,
expectation_suite_name=profiling_result_key.
expectation_suite_identifier.expectation_suite_name,
run_id=profiling_result_key.run_id,
validations_store_name=self.source_stores.get(
"profiling"),
)
batch_kwargs = validation.meta.get("batch_kwargs", {})
batch_spec = validation.meta.get("batch_spec", {})
self.add_resource_info_to_index_links_dict(
index_links_dict=index_links_dict,
expectation_suite_name=profiling_result_key.
expectation_suite_identifier.expectation_suite_name,
section_name="profiling",
batch_identifier=profiling_result_key.batch_identifier,
run_id=profiling_result_key.run_id,
run_time=profiling_result_key.run_id.run_time,
run_name=profiling_result_key.run_id.run_name,
asset_name=batch_kwargs.get("data_asset_name")
or batch_spec.get("data_asset_name"),
batch_kwargs=batch_kwargs,
batch_spec=batch_spec,
)
except Exception:
error_msg = "Profiling result not found: {:s} - skipping".format(
str(profiling_result_key.to_tuple()))
logger.warning(error_msg)
if (
# TODO why is this duplicated?
self.site_section_builders_config.get("validations", "None")
and self.site_section_builders_config.get(
"validations", "None") not in FALSEY_YAML_STRINGS):
validations_run_name_filter = self.site_section_builders_config[
"validations"]["run_name_filter"]
validation_result_site_keys = [
validation_result_key for validation_result_key in
validation_and_profiling_result_site_keys
if resource_key_passes_run_name_filter(
validation_result_key, validations_run_name_filter)
]
validation_result_site_keys = sorted(
validation_result_site_keys,
key=lambda x: x.run_id.run_time,
reverse=True,
)
if self.validation_results_limit:
validation_result_site_keys = validation_result_site_keys[:self
.
validation_results_limit]
for validation_result_key in validation_result_site_keys:
try:
validation = self.data_context.get_validation_result(
batch_identifier=validation_result_key.
batch_identifier,
expectation_suite_name=validation_result_key.
expectation_suite_identifier.expectation_suite_name,
run_id=validation_result_key.run_id,
validations_store_name=self.source_stores.get(
"validations"),
)
validation_success = validation.success
batch_kwargs = validation.meta.get("batch_kwargs", {})
batch_spec = validation.meta.get("batch_spec", {})
self.add_resource_info_to_index_links_dict(
index_links_dict=index_links_dict,
expectation_suite_name=validation_result_key.
expectation_suite_identifier.expectation_suite_name,
section_name="validations",
batch_identifier=validation_result_key.
batch_identifier,
run_id=validation_result_key.run_id,
validation_success=validation_success,
run_time=validation_result_key.run_id.run_time,
run_name=validation_result_key.run_id.run_name,
asset_name=batch_kwargs.get("data_asset_name")
or batch_spec.get("data_asset_name"),
batch_kwargs=batch_kwargs,
batch_spec=batch_spec,
)
except Exception:
error_msg = "Validation result not found: {:s} - skipping".format(
str(validation_result_key.to_tuple()))
logger.warning(error_msg)
try:
rendered_content = self.renderer_class.render(index_links_dict)
viewable_content = self.view_class.render(
rendered_content,
data_context_id=self.data_context_id,
show_how_to_buttons=self.show_how_to_buttons,
)
except Exception as e:
exception_message = f"""\
An unexpected Exception occurred during data docs rendering. Because of this error, certain parts of data docs will \
not be rendered properly and/or may not appear altogether. Please use the trace, included in this message, to \
diagnose and repair the underlying issue. Detailed information follows:
"""
exception_traceback = traceback.format_exc()
exception_message += (
f'{type(e).__name__}: "{str(e)}". Traceback: "{exception_traceback}".'
)
logger.error(exception_message)
return (self.target_store.write_index_page(viewable_content),
index_links_dict)
def test_HtmlSiteStore_filesystem_backend(tmp_path_factory):
path = str(
tmp_path_factory.mktemp(
"test_HtmlSiteStore_with_TupleFileSystemStoreBackend__dir"))
my_store = HtmlSiteStore(
store_backend={
"class_name": "TupleFilesystemStoreBackend",
"base_directory": "my_store",
},
runtime_environment={"root_directory": path},
)
with pytest.raises(TypeError):
my_store.get("not_a_ValidationResultIdentifier")
with pytest.raises(ValidationError):
my_store.get(validationResultIdentifierSchema.load({}))
ns_1 = SiteSectionIdentifier(
site_section_name="validations",
resource_identifier=ValidationResultIdentifier.from_tuple((
"a",
"b",
"c",
"quarantine",
datetime.datetime.now(datetime.timezone.utc),
"prod-100",
)),
)
my_store.set(ns_1, "aaa")
# assert my_store.get(ns_1) == "aaa"
ns_2 = SiteSectionIdentifier(
site_section_name="validations",
resource_identifier=ValidationResultIdentifier.from_tuple((
"a",
"b",
"c",
"quarantine",
datetime.datetime.now(datetime.timezone.utc),
"prod-20",
)),
)
my_store.set(ns_2, "bbb")
# assert my_store.get(ns_2) == {"B": "bbb"}
print(my_store.list_keys())
# WARNING: OBSERVE THAT SITE_SECTION_NAME IS LOST IN THE CALL TO LIST_KEYS
assert set(my_store.list_keys()) == {
ns_1.resource_identifier,
ns_2.resource_identifier,
}
print(gen_directory_tree_str(path))
assert (gen_directory_tree_str(path) == """\
test_HtmlSiteStore_with_TupleFileSystemStoreBackend__dir0/
my_store/
validations/
a/
b/
c/
quarantine/
20190926T134241.000000Z/
prod-100.html
prod-20.html
""")
def test_configuration_driven_site_builder_skip_and_clean_missing(
site_builder_data_context_with_html_store_titanic_random,
):
# tests auto-cleaning functionality of DefaultSiteIndexBuilder
# when index page is built, if an HTML page is present without corresponding suite or validation result,
# the HTML page should be removed and not appear on index page
context = site_builder_data_context_with_html_store_titanic_random
context.add_validation_operator(
"validate_and_store",
{
"class_name": "ActionListValidationOperator",
"action_list": [
{
"name": "store_validation_result",
"action": {
"class_name": "StoreValidationResultAction",
"target_store_name": "validations_store",
},
},
{
"name": "extract_and_store_eval_parameters",
"action": {
"class_name": "StoreEvaluationParametersAction",
"target_store_name": "evaluation_parameter_store",
},
},
],
},
)
# profiling the Titanic datasource will generate one expectation suite and one validation
# that is a profiling result
datasource_name = "titanic"
data_asset_name = "Titanic"
profiler_name = "BasicDatasetProfiler"
generator_name = "subdir_reader"
context.profile_datasource(datasource_name)
# creating another validation result using the profiler's suite (no need to use a new expectation suite
# for this test). having two validation results - one with run id "profiling" - allows us to test
# the logic of run_name_filter that helps filtering validation results to be included in
# the profiling and the validation sections.
batch_kwargs = context.build_batch_kwargs(
datasource=datasource_name,
batch_kwargs_generator=generator_name,
data_asset_name=data_asset_name,
)
expectation_suite_name = "{}.{}.{}.{}".format(
datasource_name, generator_name, data_asset_name, profiler_name
)
batch = context.get_batch(
batch_kwargs=batch_kwargs,
expectation_suite_name=expectation_suite_name,
)
run_id = RunIdentifier(run_name="test_run_id_12345")
context.run_validation_operator(
assets_to_validate=[batch],
run_id=run_id,
validation_operator_name="validate_and_store",
)
data_docs_config = context._project_config.data_docs_sites
local_site_config = data_docs_config["local_site"]
validations_set = set(context.stores["validations_store"].list_keys())
assert len(validations_set) == 6
expectation_suite_set = set(context.stores["expectations_store"].list_keys())
assert len(expectation_suite_set) == 5
site_builder = SiteBuilder(
data_context=context,
runtime_environment={"root_directory": context.root_directory},
**local_site_config
)
site_builder.build()
# test expectation suite pages
expectation_suite_html_pages = {
ExpectationSuiteIdentifier.from_tuple(suite_tuple)
for suite_tuple in site_builder.target_store.store_backends[
ExpectationSuiteIdentifier
].list_keys()
}
# suites in expectations store should match html pages
assert expectation_suite_set == expectation_suite_html_pages
# remove suites from expectations store
for i in range(2):
context.stores["expectations_store"].remove_key(list(expectation_suite_set)[i])
# re-build data docs, which should remove suite HTML pages that no longer have corresponding suite in
# expectations store
site_builder.build()
expectation_suite_set = set(context.stores["expectations_store"].list_keys())
expectation_suite_html_pages = {
ExpectationSuiteIdentifier.from_tuple(suite_tuple)
for suite_tuple in site_builder.target_store.store_backends[
ExpectationSuiteIdentifier
].list_keys()
}
assert expectation_suite_set == expectation_suite_html_pages
# test validation result pages
validation_html_pages = {
ValidationResultIdentifier.from_tuple(result_tuple)
for result_tuple in site_builder.target_store.store_backends[
ValidationResultIdentifier
].list_keys()
}
# validations in store should match html pages
assert validations_set == validation_html_pages
# remove validations from store
for i in range(2):
context.stores["validations_store"].store_backend.remove_key(
list(validations_set)[i]
)
# re-build data docs, which should remove validation HTML pages that no longer have corresponding validation in
# validations store
site_builder.build()
validations_set = set(context.stores["validations_store"].list_keys())
validation_html_pages = {
ValidationResultIdentifier.from_tuple(result_tuple)
for result_tuple in site_builder.target_store.store_backends[
ValidationResultIdentifier
].list_keys()
}
assert validations_set == validation_html_pages
def test_ValidationsStore_with_TupleFileSystemStoreBackend(tmp_path_factory):
path = str(
tmp_path_factory.mktemp(
"test_ValidationResultStore_with_TupleFileSystemStoreBackend__dir"
)
)
project_path = str(tmp_path_factory.mktemp("my_dir"))
my_store = ValidationsStore(
store_backend={
"module_name": "great_expectations.data_context.store",
"class_name": "TupleFilesystemStoreBackend",
"base_directory": "my_store/",
},
runtime_environment={"root_directory": path},
)
with pytest.raises(TypeError):
my_store.get("not_a_ValidationResultIdentifier")
ns_1 = ValidationResultIdentifier(
expectation_suite_identifier=ExpectationSuiteIdentifier("asset.quarantine"),
run_id="prod-100",
batch_identifier="batch_id",
)
my_store.set(ns_1, ExpectationSuiteValidationResult(success=True))
assert my_store.get(ns_1) == ExpectationSuiteValidationResult(
success=True, statistics={}, results=[]
)
ns_2 = ValidationResultIdentifier.from_tuple(
(
"asset",
"quarantine",
"prod-20",
datetime.datetime.now(datetime.timezone.utc),
"batch_id",
)
)
my_store.set(ns_2, ExpectationSuiteValidationResult(success=False))
assert my_store.get(ns_2) == ExpectationSuiteValidationResult(
success=False, statistics={}, results=[]
)
print(my_store.list_keys())
assert set(my_store.list_keys()) == {
ns_1,
ns_2,
}
print(gen_directory_tree_str(path))
assert (
gen_directory_tree_str(path)
== """\
test_ValidationResultStore_with_TupleFileSystemStoreBackend__dir0/
my_store/
.ge_store_backend_id
asset/
quarantine/
prod-100/
20190926T134241.000000Z/
batch_id.json
prod-20/
20190926T134241.000000Z/
batch_id.json
"""
)
"""
What does this test and why?
A Store should be able to report it's store_backend_id
which is set when the StoreBackend is instantiated.
"""
# Check that store_backend_id exists can be read
assert my_store.store_backend_id is not None
# Check that store_backend_id is a valid UUID
assert test_utils.validate_uuid4(my_store.store_backend_id)
# Check that another store with the same configuration shares the same store_backend_id
my_store_duplicate = ValidationsStore(
store_backend={
"module_name": "great_expectations.data_context.store",
"class_name": "TupleFilesystemStoreBackend",
"base_directory": "my_store/",
},
runtime_environment={"root_directory": path},
)
assert my_store.store_backend_id == my_store_duplicate.store_backend_id
def build(self):
# Loop over sections in the HtmlStore
logger.debug("DefaultSiteIndexBuilder.build")
expectation_suite_keys = [
ExpectationSuiteIdentifier.from_tuple(expectation_suite_tuple) for expectation_suite_tuple in
self.target_store.store_backends[ExpectationSuiteIdentifier].list_keys()
]
validation_and_profiling_result_keys = [
ValidationResultIdentifier.from_tuple(validation_result_tuple) for validation_result_tuple in
self.target_store.store_backends[ValidationResultIdentifier].list_keys()
]
profiling_result_keys = [
validation_result_key for validation_result_key in validation_and_profiling_result_keys
if validation_result_key.run_id == "profiling"
]
validation_result_keys = [
validation_result_key for validation_result_key in validation_and_profiling_result_keys
if validation_result_key.run_id != "profiling"
]
validation_result_keys = sorted(validation_result_keys, key=lambda x: x.run_id, reverse=True)
if self.validation_results_limit:
validation_result_keys = validation_result_keys[:self.validation_results_limit]
index_links_dict = OrderedDict()
index_links_dict["site_name"] = self.site_name
if self.show_cta_footer:
index_links_dict["cta_object"] = self.get_calls_to_action()
for expectation_suite_key in expectation_suite_keys:
self.add_resource_info_to_index_links_dict(
index_links_dict=index_links_dict,
expectation_suite_name=expectation_suite_key.expectation_suite_name,
section_name="expectations"
)
for profiling_result_key in profiling_result_keys:
try:
validation = self.data_context.get_validation_result(
batch_identifier=profiling_result_key.batch_identifier,
expectation_suite_name=profiling_result_key.expectation_suite_identifier.expectation_suite_name,
run_id=profiling_result_key.run_id
)
validation_success = validation.success
self.add_resource_info_to_index_links_dict(
index_links_dict=index_links_dict,
expectation_suite_name=profiling_result_key.expectation_suite_identifier.expectation_suite_name,
section_name="profiling",
batch_identifier=profiling_result_key.batch_identifier,
run_id=profiling_result_key.run_id,
validation_success=validation_success
)
except Exception as e:
error_msg = "Profiling result not found: {0:s} - skipping".format(str(profiling_result_key.to_tuple()))
logger.warning(error_msg)
for validation_result_key in validation_result_keys:
try:
validation = self.data_context.get_validation_result(
batch_identifier=validation_result_key.batch_identifier,
expectation_suite_name=validation_result_key.expectation_suite_identifier.expectation_suite_name,
run_id=validation_result_key.run_id
)
validation_success = validation.success
self.add_resource_info_to_index_links_dict(
index_links_dict=index_links_dict,
expectation_suite_name=validation_result_key.expectation_suite_identifier.expectation_suite_name,
section_name="validations",
batch_identifier=validation_result_key.batch_identifier,
run_id=validation_result_key.run_id,
validation_success=validation_success
)
except Exception as e:
error_msg = "Validation result not found: {0:s} - skipping".format(str(validation_result_key.to_tuple()))
logger.warning(error_msg)
try:
rendered_content = self.renderer_class.render(index_links_dict)
viewable_content = self.view_class.render(rendered_content)
except Exception as e:
logger.error("Exception occurred during data docs rendering: ", e, exc_info=True)
return None
return (
self.target_store.write_index_page(viewable_content),
index_links_dict
)