From c3507cccec0fea4c46677919df06f38919d7c01e Mon Sep 17 00:00:00 2001 From: Pablo Martin Date: Wed, 29 Mar 2023 17:17:44 +0200 Subject: [PATCH 1/7] Refactor shared elements to follow DRY --- tests/test_integration/test_data_testing.py | 85 ++++++++++----------- 1 file changed, 40 insertions(+), 45 deletions(-) diff --git a/tests/test_integration/test_data_testing.py b/tests/test_integration/test_data_testing.py index 8b456f2..5894dd1 100644 --- a/tests/test_integration/test_data_testing.py +++ b/tests/test_integration/test_data_testing.py @@ -19,29 +19,44 @@ from lolafect.connections import open_ssh_tunnel_with_s3_pkey, close_ssh_tunnel TEST_LOLACONFIG = build_lolaconfig(flow_name="testing-suite") +TEST_QUERY = """ +SELECT 1 AS a_one, + "lol" AS a_string, + NULL AS a_null +""" + +TEST_EXPECTATIONS_THAT_FIT_DATA = [ + ExpectationConfiguration( + expectation_type="expect_column_values_to_be_between", + kwargs={"column": "a_one", "min_value": 1, "max_value": 1}, + ), + ExpectationConfiguration( + expectation_type="expect_column_values_to_match_like_pattern", + kwargs={"column": "a_string", "like_pattern": "%lol%"}, + ), + ExpectationConfiguration( + expectation_type="expect_column_values_to_be_null", + kwargs={"column": "a_null"}, + ), +] + +TEST_EXPECTATIONS_THAT_DONT_FIT_DATA = [ + ExpectationConfiguration( + expectation_type="expect_column_values_to_be_between", + kwargs={"column": "a_one", "min_value": 2, "max_value": 2}, + ), + ExpectationConfiguration( + expectation_type="expect_column_values_to_match_like_pattern", + kwargs={"column": "a_string", "like_pattern": "%xD%"}, + ), + ExpectationConfiguration( + expectation_type="expect_column_values_to_not_be_null", + kwargs={"column": "a_null"}, + ), +] + def test_validation_on_mysql_succeeds(): - - test_query = """ - SELECT 1 AS a_one, - "lol" AS a_string, - NULL AS a_null - """ - test_expectations = [ - ExpectationConfiguration( - expectation_type="expect_column_values_to_be_between", - kwargs={"column": "a_one", "min_value": 1, "max_value": 1}, - ), - ExpectationConfiguration( - expectation_type="expect_column_values_to_match_like_pattern", - kwargs={"column": "a_string", "like_pattern": "%lol%"}, - ), - ExpectationConfiguration( - expectation_type="expect_column_values_to_be_null", - kwargs={"column": "a_null"}, - ), - ] - ssh_tunnel = open_ssh_tunnel_with_s3_pkey.run( s3_bucket_name=TEST_LOLACONFIG.S3_BUCKET_NAME, ssh_tunnel_credentials=TEST_LOLACONFIG.SSH_TUNNEL_CREDENTIALS, @@ -58,8 +73,8 @@ def test_validation_on_mysql_succeeds(): "password": TEST_LOLACONFIG.DW_CREDENTIALS["password"], "db": TEST_LOLACONFIG.DW_CREDENTIALS["default_db"], }, - query=test_query, - expectation_configurations=test_expectations, + query=TEST_QUERY, + expectation_configurations=TEST_EXPECTATIONS_THAT_FIT_DATA, ) closed_tunnel = close_ssh_tunnel.run(ssh_tunnel) @@ -70,26 +85,6 @@ def test_validation_on_mysql_succeeds(): def test_validation_on_mysql_fails(): - test_query = """ - SELECT 1 AS a_one, - "lol" AS a_string, - NULL AS a_null - """ - test_expectations = [ - ExpectationConfiguration( - expectation_type="expect_column_values_to_be_between", - kwargs={"column": "a_one", "min_value": 2, "max_value": 2}, - ), - ExpectationConfiguration( - expectation_type="expect_column_values_to_match_like_pattern", - kwargs={"column": "a_string", "like_pattern": "%xD%"}, - ), - ExpectationConfiguration( - expectation_type="expect_column_values_to_not_be_null", - kwargs={"column": "a_null"}, - ), - ] - ssh_tunnel = open_ssh_tunnel_with_s3_pkey.run( s3_bucket_name=TEST_LOLACONFIG.S3_BUCKET_NAME, ssh_tunnel_credentials=TEST_LOLACONFIG.SSH_TUNNEL_CREDENTIALS, @@ -106,8 +101,8 @@ def test_validation_on_mysql_fails(): "password": TEST_LOLACONFIG.DW_CREDENTIALS["password"], "db": TEST_LOLACONFIG.DW_CREDENTIALS["default_db"], }, - query=test_query, - expectation_configurations=test_expectations, + query=TEST_QUERY, + expectation_configurations=TEST_EXPECTATIONS_THAT_DONT_FIT_DATA, ) closed_tunnel = close_ssh_tunnel.run(ssh_tunnel) From b51c810a394909e74f4ec6aac6d6b9214cf255ab Mon Sep 17 00:00:00 2001 From: Pablo Martin Date: Wed, 29 Mar 2023 17:18:00 +0200 Subject: [PATCH 2/7] Add tests for trino feature --- tests/test_integration/test_data_testing.py | 40 ++++++++++++++++++++- 1 file changed, 39 insertions(+), 1 deletion(-) diff --git a/tests/test_integration/test_data_testing.py b/tests/test_integration/test_data_testing.py index 5894dd1..919903d 100644 --- a/tests/test_integration/test_data_testing.py +++ b/tests/test_integration/test_data_testing.py @@ -1,7 +1,7 @@ from great_expectations.core.expectation_configuration import ExpectationConfiguration from lolafect.lolaconfig import build_lolaconfig -from lolafect.data_testing import run_data_test_on_mysql +from lolafect.data_testing import run_data_test_on_mysql, run_data_test_on_trino from lolafect.connections import open_ssh_tunnel_with_s3_pkey, close_ssh_tunnel # __ __ _____ _ _ _____ _ _ _____ _ @@ -110,3 +110,41 @@ def test_validation_on_mysql_fails(): data_test_failed = validation_result["success"] == False assert data_test_failed + + +def test_validation_on_trino_succeeds(): + validation_result = run_data_test_on_trino.run( + name="lolafect-testing-test_validation_on_mysql_fails", + trino_credentials={ + "host": TEST_LOLACONFIG.TRINO_CREDENTIALS["host"], + "port": TEST_LOLACONFIG.TRINO_CREDENTIALS["port"], + "user": TEST_LOLACONFIG.TRINO_CREDENTIALS["user"], + "password": TEST_LOLACONFIG.TRINO_CREDENTIALS["password"], + "db": TEST_LOLACONFIG.TRINO_CREDENTIALS["default_db"], + }, + query=TEST_QUERY, + expectation_configurations=TEST_EXPECTATIONS_THAT_FIT_DATA, + ) + + data_test_passed = validation_result["success"] == True + + assert data_test_passed + + +def test_validation_on_trino_fails(): + validation_result = run_data_test_on_trino.run( + name="lolafect-testing-test_validation_on_mysql_fails", + trino_credentials={ + "host": TEST_LOLACONFIG.TRINO_CREDENTIALS["host"], + "port": TEST_LOLACONFIG.TRINO_CREDENTIALS["port"], + "user": TEST_LOLACONFIG.TRINO_CREDENTIALS["user"], + "password": TEST_LOLACONFIG.TRINO_CREDENTIALS["password"], + "db": TEST_LOLACONFIG.TRINO_CREDENTIALS["default_db"], + }, + query=TEST_QUERY, + expectation_configurations=TEST_EXPECTATIONS_THAT_DONT_FIT_DATA, + ) + + data_test_failed = validation_result["success"] == False + + assert data_test_failed From 814a376e02818f08671075dd3532fd272bc1a709 Mon Sep 17 00:00:00 2001 From: Pablo Martin Date: Wed, 29 Mar 2023 17:18:14 +0200 Subject: [PATCH 3/7] Add entry in readme showcasing new trino feature --- README.md | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) diff --git a/README.md b/README.md index d9cbbf1..134b28d 100644 --- a/README.md +++ b/README.md @@ -143,6 +143,28 @@ with Flow(...) as flow: print("The data is bad!!!") ``` +**Run a Great Expectations validation on a Trino query** + +```python +from lolafect.data_testing import run_data_test_on_trino + +with Flow(...) as flow: + + my_query = """SELECT something FROM somewhere""" + my_expectations = {...} # A bunch of things you want to validate on the result of the query + + validation_results = run_data_test_on_trino( + name="my-cool-validation", + trino_credentials={...}, + query=my_query, + expectations=my_expectations + ) + + if not validation_results["success"]: + print("The data is bad!!!") +``` + + ### Slack **Send a warning message to slack if your tasks fails** From 5b97864e8d3d458c39375bb3f08ed7d4e1ca088a Mon Sep 17 00:00:00 2001 From: Pablo Martin Date: Wed, 29 Mar 2023 17:18:41 +0200 Subject: [PATCH 4/7] Signature for new task --- lolafect/data_testing.py | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/lolafect/data_testing.py b/lolafect/data_testing.py index 4216d4b..863d6fe 100644 --- a/lolafect/data_testing.py +++ b/lolafect/data_testing.py @@ -60,6 +60,15 @@ def run_data_test_on_mysql( return results +@task() +def run_data_test_on_trino( + name: str, + trino_credentials: dict, + query: str, + expectation_configurations: List[ExpectationConfiguration], + great_expectations_s3_bucket: str = DEFAULT_GREAT_EXPECTATIONS_S3_BUCKET, +) -> dict: + raise NotImplementedError("WIP") def _create_in_memory_data_context_for_mysql( mysql_credentials: dict, From 19211eee735575678dff3df6b336272c4bf0e0ed Mon Sep 17 00:00:00 2001 From: Pablo Martin Date: Wed, 29 Mar 2023 17:28:16 +0200 Subject: [PATCH 5/7] Naive duplication of MySQL elements but for Trino. Not working yet. --- lolafect/data_testing.py | 88 +++++++++++++++++++++++++++++++++++++++- 1 file changed, 87 insertions(+), 1 deletion(-) diff --git a/lolafect/data_testing.py b/lolafect/data_testing.py index 863d6fe..b1fe584 100644 --- a/lolafect/data_testing.py +++ b/lolafect/data_testing.py @@ -68,7 +68,43 @@ def run_data_test_on_trino( expectation_configurations: List[ExpectationConfiguration], great_expectations_s3_bucket: str = DEFAULT_GREAT_EXPECTATIONS_S3_BUCKET, ) -> dict: - raise NotImplementedError("WIP") + """ + Validate a query and an expectation suite against a given Trino server. + + :param name: a unique name for the data test. + :param trino_credentials: credentials for the Trino cluster. + :param query: the query to test against. + :param expectation_configurations: the expectations on the dataset. + :param great_expectations_s3_bucket: the bucket where Great Expectations + files live. + :return: the result of the data test. + """ + logger = prefect.context.get("logger") + + logger.info("Creating data context.") + data_context = _create_in_memory_data_context_for_trino( + trino_credentials, great_expectations_s3_bucket + ) + logger.info("Data context created.") + logger.info("Creating expectation suite.") + data_context = _create_expectation_suite( + data_context, name, expectation_configurations + ) + logger.info("Expectation suite created.") + logger.info("Creating checkpoint.") + data_context = _create_checkpoint( + data_context, + f"{trino_credentials['host']}:{trino_credentials['port']}", + query, + name, + ) + logger.info("Checkpoint created.") + logger.info("Running checkpoint.") + results = data_context.run_checkpoint(f"{name}_checkpoint") + logger.info("Checkpoint finished.") + logger.info(f"Validation result: {results['success']}") + + return results def _create_in_memory_data_context_for_mysql( mysql_credentials: dict, @@ -120,6 +156,56 @@ def _create_in_memory_data_context_for_mysql( return data_context +def _create_in_memory_data_context_for_trino( + trino_credentials: dict, + great_expectations_s3_bucket: str, +) -> AbstractDataContext: + """ + Create a DataContext without a YAML config file and specify a Trino + datasource. + + :param trino_credentials: the creds to the mysql where the query will be + executed. + :param great_expectations_s3_bucket: the name of the bucket where Great + Exepctations files while be stored. + :return: the data context. + """ + + data_context = BaseDataContext( + project_config=DataContextConfig( + datasources={ + f"{trino_credentials['host']}:{trino_credentials['port']}": DatasourceConfig( + class_name="Datasource", + execution_engine={ + "class_name": "SqlAlchemyExecutionEngine", + "connection_string": f"trino://%s:%s@%s:%s/{trino_credentials['db']}/{'SCHEMA_GOES_HERE'}" + % ( + trino_credentials["user"], + urlquote(trino_credentials["password"]), + trino_credentials["host"], + trino_credentials["port"], + ), + }, + data_connectors={ + "default_runtime_data_connector_name": { + "class_name": "RuntimeDataConnector", + "batch_identifiers": ["default_identifier_name"], + }, + "default_inferred_data_connector_name": { + "class_name": "InferredAssetSqlDataConnector", + "name": "whole_table", + }, + }, + ) + }, + store_backend_defaults=S3StoreBackendDefaults( + default_bucket_name=great_expectations_s3_bucket + ), + ) + ) + + return data_context + def _create_expectation_suite( data_context: AbstractDataContext, From 5066fe438221d38442f1e422b2fa45da6a0e7568 Mon Sep 17 00:00:00 2001 From: Pablo Martin Date: Thu, 30 Mar 2023 13:58:15 +0200 Subject: [PATCH 6/7] Use single quotes so that query works both in MySQL and Trino. --- tests/test_integration/test_data_testing.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/tests/test_integration/test_data_testing.py b/tests/test_integration/test_data_testing.py index 919903d..bcd8d28 100644 --- a/tests/test_integration/test_data_testing.py +++ b/tests/test_integration/test_data_testing.py @@ -21,7 +21,7 @@ TEST_LOLACONFIG = build_lolaconfig(flow_name="testing-suite") TEST_QUERY = """ SELECT 1 AS a_one, - "lol" AS a_string, + 'lol' AS a_string, NULL AS a_null """ From d9c131dcc6266d5022dfca7cdabaffbed468fac5 Mon Sep 17 00:00:00 2001 From: Pablo Martin Date: Thu, 20 Apr 2023 15:36:16 +0200 Subject: [PATCH 7/7] Unfinished work --- lolafect/data_testing.py | 4 +- tests/test_integration/test_data_testing.py | 130 ++++++++++---------- 2 files changed, 71 insertions(+), 63 deletions(-) diff --git a/lolafect/data_testing.py b/lolafect/data_testing.py index b1fe584..9e3e777 100644 --- a/lolafect/data_testing.py +++ b/lolafect/data_testing.py @@ -178,12 +178,14 @@ def _create_in_memory_data_context_for_trino( class_name="Datasource", execution_engine={ "class_name": "SqlAlchemyExecutionEngine", - "connection_string": f"trino://%s:%s@%s:%s/{trino_credentials['db']}/{'SCHEMA_GOES_HERE'}" + "connection_string": f"trino://%s:%s@%s:%s/%s/%s" % ( trino_credentials["user"], urlquote(trino_credentials["password"]), trino_credentials["host"], trino_credentials["port"], + trino_credentials["catalog"], + trino_credentials["schema"], ), }, data_connectors={ diff --git a/tests/test_integration/test_data_testing.py b/tests/test_integration/test_data_testing.py index bcd8d28..a6c564a 100644 --- a/tests/test_integration/test_data_testing.py +++ b/tests/test_integration/test_data_testing.py @@ -18,11 +18,13 @@ from lolafect.connections import open_ssh_tunnel_with_s3_pkey, close_ssh_tunnel # are working properly. TEST_LOLACONFIG = build_lolaconfig(flow_name="testing-suite") - +#1 AS a_one, +# 'lol' AS a_string, +# NULL AS a_null TEST_QUERY = """ -SELECT 1 AS a_one, - 'lol' AS a_string, - NULL AS a_null +SELECT * +from app_lm_mysql_pl.comprea.market +where id = 1 """ TEST_EXPECTATIONS_THAT_FIT_DATA = [ @@ -55,62 +57,62 @@ TEST_EXPECTATIONS_THAT_DONT_FIT_DATA = [ ), ] - -def test_validation_on_mysql_succeeds(): - ssh_tunnel = open_ssh_tunnel_with_s3_pkey.run( - s3_bucket_name=TEST_LOLACONFIG.S3_BUCKET_NAME, - ssh_tunnel_credentials=TEST_LOLACONFIG.SSH_TUNNEL_CREDENTIALS, - remote_target_host=TEST_LOLACONFIG.DW_CREDENTIALS["host"], - remote_target_port=TEST_LOLACONFIG.DW_CREDENTIALS["port"], - ) - - validation_result = run_data_test_on_mysql.run( - name="lolafect-testing-test_validation_on_mysql_succeeds", - mysql_credentials={ - "host": ssh_tunnel.local_bind_address[0], - "port": ssh_tunnel.local_bind_address[1], - "user": TEST_LOLACONFIG.DW_CREDENTIALS["user"], - "password": TEST_LOLACONFIG.DW_CREDENTIALS["password"], - "db": TEST_LOLACONFIG.DW_CREDENTIALS["default_db"], - }, - query=TEST_QUERY, - expectation_configurations=TEST_EXPECTATIONS_THAT_FIT_DATA, - ) - - closed_tunnel = close_ssh_tunnel.run(ssh_tunnel) - - data_test_passed = validation_result["success"] == True - - assert data_test_passed - - -def test_validation_on_mysql_fails(): - ssh_tunnel = open_ssh_tunnel_with_s3_pkey.run( - s3_bucket_name=TEST_LOLACONFIG.S3_BUCKET_NAME, - ssh_tunnel_credentials=TEST_LOLACONFIG.SSH_TUNNEL_CREDENTIALS, - remote_target_host=TEST_LOLACONFIG.DW_CREDENTIALS["host"], - remote_target_port=TEST_LOLACONFIG.DW_CREDENTIALS["port"], - ) - - validation_result = run_data_test_on_mysql.run( - name="lolafect-testing-test_validation_on_mysql_fails", - mysql_credentials={ - "host": ssh_tunnel.local_bind_address[0], - "port": ssh_tunnel.local_bind_address[1], - "user": TEST_LOLACONFIG.DW_CREDENTIALS["user"], - "password": TEST_LOLACONFIG.DW_CREDENTIALS["password"], - "db": TEST_LOLACONFIG.DW_CREDENTIALS["default_db"], - }, - query=TEST_QUERY, - expectation_configurations=TEST_EXPECTATIONS_THAT_DONT_FIT_DATA, - ) - - closed_tunnel = close_ssh_tunnel.run(ssh_tunnel) - - data_test_failed = validation_result["success"] == False - - assert data_test_failed - +# +# def test_validation_on_mysql_succeeds(): +# ssh_tunnel = open_ssh_tunnel_with_s3_pkey.run( +# s3_bucket_name=TEST_LOLACONFIG.S3_BUCKET_NAME, +# ssh_tunnel_credentials=TEST_LOLACONFIG.SSH_TUNNEL_CREDENTIALS, +# remote_target_host=TEST_LOLACONFIG.DW_CREDENTIALS["host"], +# remote_target_port=TEST_LOLACONFIG.DW_CREDENTIALS["port"], +# ) +# +# validation_result = run_data_test_on_mysql.run( +# name="lolafect-testing-test_validation_on_mysql_succeeds", +# mysql_credentials={ +# "host": ssh_tunnel.local_bind_address[0], +# "port": ssh_tunnel.local_bind_address[1], +# "user": TEST_LOLACONFIG.DW_CREDENTIALS["user"], +# "password": TEST_LOLACONFIG.DW_CREDENTIALS["password"], +# "db": TEST_LOLACONFIG.DW_CREDENTIALS["default_db"], +# }, +# query=TEST_QUERY, +# expectation_configurations=TEST_EXPECTATIONS_THAT_FIT_DATA, +# ) +# +# closed_tunnel = close_ssh_tunnel.run(ssh_tunnel) +# +# data_test_passed = validation_result["success"] == True +# +# assert data_test_passed +# +# +# def test_validation_on_mysql_fails(): +# ssh_tunnel = open_ssh_tunnel_with_s3_pkey.run( +# s3_bucket_name=TEST_LOLACONFIG.S3_BUCKET_NAME, +# ssh_tunnel_credentials=TEST_LOLACONFIG.SSH_TUNNEL_CREDENTIALS, +# remote_target_host=TEST_LOLACONFIG.DW_CREDENTIALS["host"], +# remote_target_port=TEST_LOLACONFIG.DW_CREDENTIALS["port"], +# ) +# +# validation_result = run_data_test_on_mysql.run( +# name="lolafect-testing-test_validation_on_mysql_fails", +# mysql_credentials={ +# "host": ssh_tunnel.local_bind_address[0], +# "port": ssh_tunnel.local_bind_address[1], +# "user": TEST_LOLACONFIG.DW_CREDENTIALS["user"], +# "password": TEST_LOLACONFIG.DW_CREDENTIALS["password"], +# "db": TEST_LOLACONFIG.DW_CREDENTIALS["default_db"], +# }, +# query=TEST_QUERY, +# expectation_configurations=TEST_EXPECTATIONS_THAT_DONT_FIT_DATA, +# ) +# +# closed_tunnel = close_ssh_tunnel.run(ssh_tunnel) +# +# data_test_failed = validation_result["success"] == False +# +# assert data_test_failed +# def test_validation_on_trino_succeeds(): validation_result = run_data_test_on_trino.run( @@ -120,11 +122,14 @@ def test_validation_on_trino_succeeds(): "port": TEST_LOLACONFIG.TRINO_CREDENTIALS["port"], "user": TEST_LOLACONFIG.TRINO_CREDENTIALS["user"], "password": TEST_LOLACONFIG.TRINO_CREDENTIALS["password"], - "db": TEST_LOLACONFIG.TRINO_CREDENTIALS["default_db"], + "catalog": "data_dw", + "schema": "sandbox" }, query=TEST_QUERY, expectation_configurations=TEST_EXPECTATIONS_THAT_FIT_DATA, ) + print("###############\n" * 20) + print(validation_result) data_test_passed = validation_result["success"] == True @@ -139,7 +144,8 @@ def test_validation_on_trino_fails(): "port": TEST_LOLACONFIG.TRINO_CREDENTIALS["port"], "user": TEST_LOLACONFIG.TRINO_CREDENTIALS["user"], "password": TEST_LOLACONFIG.TRINO_CREDENTIALS["password"], - "db": TEST_LOLACONFIG.TRINO_CREDENTIALS["default_db"], + "catalog": "data_dw", + "schema": "sandbox" }, query=TEST_QUERY, expectation_configurations=TEST_EXPECTATIONS_THAT_DONT_FIT_DATA,