fix: Updated schema validation logic to column as 'validation_status' (…

…#455) * fix: Updated schema validation logic to column as validation_status * fix: fixed the col 'schema' referring in sqlserver and bigquery tests files, while running integration tests
GoogleCloudPlatform · Apr 25, 2022 · e30c337 · e30c337
1 parent 4093625
commit e30c337
Show file tree

Hide file tree

Showing 5 changed files with 6 additions and 6 deletions.
diff --git a/data_validation/schema_validation.py b/data_validation/schema_validation.py
@@ -58,7 +58,7 @@ def execute(self):
                 "target_column_name",
                 "source_agg_value",
                 "target_agg_value",
-                "status",
+                "validation_status",
                 "error_result.details",
             ],
         )

diff --git a/tests/system/data_sources/test_bigquery.py b/tests/system/data_sources/test_bigquery.py
@@ -392,7 +392,7 @@ def test_schema_validation():
     df = validator.execute()
 
     for validation in df.to_dict(orient="records"):
-        assert validation["status"] == consts.VALIDATION_STATUS_SUCCESS
+        assert validation["validation_status"] == consts.VALIDATION_STATUS_SUCCESS
 
 
 def test_cli_store_yaml_then_run_gcs():

diff --git a/tests/system/data_sources/test_sql_server.py b/tests/system/data_sources/test_sql_server.py
@@ -115,4 +115,4 @@ def test_schema_validation():
     df = validator.execute()
 
     for validation in df.to_dict(orient="records"):
-        assert validation["status"] == consts.VALIDATION_STATUS_SUCCESS
+        assert validation["validation_status"] == consts.VALIDATION_STATUS_SUCCESS
diff --git a/tests/system/result_handlers/test_bigquery.py b/tests/system/result_handlers/test_bigquery.py
@@ -61,7 +61,7 @@ def get_dataframe(bigquery_client, table_id):
             "source_column_name, target_table_name, target_column_name, "
             "validation_type, aggregation_type, validation_name, "
             "source_agg_value, target_agg_value, group_by_columns, "
-            "difference, pct_difference, pct_threshold, status, labels "
+            "difference, pct_difference, pct_threshold, validation_status, labels "
             f" FROM `{table_id}` ORDER BY target_agg_value ASC"
         ).to_dataframe()
 
@@ -136,7 +136,7 @@ def test_execute_with_nan(bigquery_client, bigquery_dataset_id):
             "difference": [-1.0, -1.0, _NAN, _NAN, _NAN, _NAN],
             "pct_difference": [-50.0, -25.0, _NAN, _NAN, _NAN, _NAN],
             "pct_threshold": [25.0, 25.0, _NAN, _NAN, _NAN, _NAN],
-            "status": [
+            "validation_status": [
                 consts.VALIDATION_STATUS_FAIL,
                 consts.VALIDATION_STATUS_SUCCESS,
                 _NAN,

diff --git a/tests/unit/test_schema_validation.py b/tests/unit/test_schema_validation.py
@@ -200,7 +200,7 @@ def test_execute(module_under_test, fs):
     dv_client = data_validation.DataValidation(SAMPLE_SCHEMA_CONFIG, verbose=True)
     result_df = dv_client.schema_validator.execute()
     failures = result_df[
-        result_df["status"].str.contains(consts.VALIDATION_STATUS_FAIL)
+        result_df["validation_status"].str.contains(consts.VALIDATION_STATUS_FAIL)
     ]
 
     assert len(result_df) == len(source_data[0]) + 1