Qa/logging tap tester (singer-io#190)

*using tap-tester logger

* remove singer import

* appease Dylan (PR feedback)

* fix attr window test - broken with previous change

* update field selection test

* fix invalid attr window and start date tests

Co-authored-by: kspeer <kspeer@stitchdata.com>

tests/base.py

@@ -7,8 +7,8 @@
 from datetime import timedelta
 from datetime import datetime as dt
 
-import singer
 from tap_tester import connections, menagerie, runner
+from tap_tester.logger import LOGGER
 
 
 class FacebookBaseTest(unittest.TestCase):
@@ -29,7 +29,6 @@ class FacebookBaseTest(unittest.TestCase):
     FULL_TABLE = "FULL_TABLE"
     START_DATE_FORMAT = "%Y-%m-%dT00:00:00Z"
     BOOKMARK_COMPARISON_FORMAT = "%Y-%m-%dT00:00:00+00:00"
-    LOGGER = singer.get_logger()
 
     start_date = ""
 
@@ -216,7 +215,7 @@ def run_and_verify_check_mode(self, conn_id):
         found_catalog_names = set(map(lambda c: c['stream_name'], found_catalogs))
 
         self.assertSetEqual(self.expected_streams(), found_catalog_names, msg="discovered schemas do not match")
-        print("discovered schemas are OK")
+        LOGGER.info("discovered schemas are OK")
 
         return found_catalogs
 
@@ -240,7 +239,7 @@ def run_and_verify_sync(self, conn_id):
             sum(sync_record_count.values()), 0,
             msg="failed to replicate any data: {}".format(sync_record_count)
         )
-        print("total replicated row count: {}".format(sum(sync_record_count.values())))
+        LOGGER.info("total replicated row count: %s", sum(sync_record_count.values()))
 
         return sync_record_count
 
@@ -270,7 +269,7 @@ def perform_and_verify_table_and_field_selection(self,
 
             # Verify all testable streams are selected
             selected = catalog_entry.get('annotated-schema').get('selected')
-            print("Validating selection on {}: {}".format(cat['stream_name'], selected))
+            LOGGER.info("Validating selection on %s: %s", cat['stream_name'], selected)
             if cat['stream_name'] not in expected_selected:
                 self.assertFalse(selected, msg="Stream selected, but not testable.")
                 continue # Skip remaining assertions if we aren't selecting this stream
@@ -280,8 +279,8 @@ def perform_and_verify_table_and_field_selection(self,
                 # Verify all fields within each selected stream are selected
                 for field, field_props in catalog_entry.get('annotated-schema').get('properties').items():
                     field_selected = field_props.get('selected')
-                    print("\tValidating selection on {}.{}: {}".format(
-                        cat['stream_name'], field, field_selected))
+                    LOGGER.info("\tValidating selection on %s.%s: %s",
+                                cat['stream_name'], field, field_selected)
                     self.assertTrue(field_selected, msg="Field not selected.")
             else:
                 # Verify only automatic fields are selected
@@ -363,3 +362,15 @@ def timedelta_formatted(self, dtime, days=0):
     @staticmethod
     def is_insight(stream):
         return stream.startswith('ads_insights')
+
+    def setUp(self):
+        LOGGER.info("-------------------------------------------- STARTING TEST ---------------------------------------------------")
+        LOGGER.info("test: %s", self.name())
+        LOGGER.info("streams covered: %s", self.streams_to_test())
+        LOGGER.info("--------------------------------------------------------------------------------------------------------------")
+
+    def tearDown(self):
+        LOGGER.info("--------------------------------------------- ENDING TEST ----------------------------------------------------")
+        LOGGER.info("test: %s", self.name())
+        LOGGER.info("streams covered: %s", self.streams_to_test())
+        LOGGER.info("--------------------------------------------------------------------------------------------------------------")

tests/test_facebook_archived_data.py

@@ -10,6 +10,10 @@ class FacebookArchivedData(FacebookBaseTest):
     def name():
         return "tap_tester_facebook_archived_data"
 
+    def streams_to_test(self):
+        """include_deleted is supported for below streams only"""
+        return ['ads', 'adsets', 'campaigns']
+
     def get_properties(self, original: bool = True):
         """Configuration properties required for the tap."""
         return_value = {
@@ -29,8 +33,7 @@ def test_run(self):
         '''
             Testing the archived data with 'include_deleted' parameter
         '''
-        # include_deleted is supported for below streams only
-        expected_streams = ['ads', 'adsets', 'campaigns']
+        expected_streams = self.streams_to_test()
 
         ##########################################################################
         ### First Sync with include_deleted = false

tests/test_facebook_attribution_window.py

@@ -4,12 +4,17 @@
 
 from base import FacebookBaseTest
 
+
 class FacebookAttributionWindow(FacebookBaseTest):
 
     @staticmethod
     def name():
         return "tap_tester_facebook_attribution_window"
 
+    def streams_to_test(self):
+        """ 'attribution window' is only supported for 'ads_insights' streams """
+        return [stream for stream in self.expected_streams() if self.is_insight(stream)]
+
     def get_properties(self, original: bool = True):
         """Configuration properties required for the tap."""
         return_value = {
@@ -24,8 +29,11 @@ def get_properties(self, original: bool = True):
         return_value["start_date"] = self.start_date
         return return_value
 
-    # For the test ad set up in facebook ads manager we see data on April 7th, start date is based on this data
     def test_run(self):
+        """
+        For the test ad set up in facebook ads manager we see data
+        on April 7th, start date is based on this data
+        """
         # attrribution window = 7
         self.ATTRIBUTION_WINDOW = 7
         self.start_date = '2021-04-14T00:00:00Z'
@@ -44,30 +52,23 @@ def test_run(self):
         self.end_date = '2021-04-09T00:00:00Z'
         self.run_test(self.ATTRIBUTION_WINDOW, self.start_date, self.end_date)
 
-
     def run_test(self, attr_window, start_date, end_date):
         """
             Test to check the attribution window
         """
 
+        expected_streams = self.streams_to_test()
+
         conn_id = connections.ensure_connection(self)
 
         # calculate start date with attribution window
         start_date_with_attribution_window = self.timedelta_formatted(start_date, days=-attr_window)
 
-        # 'attribution window' is only supported for 'ads_insights' streams
-        expected_insights_streams = []
-        for stream in self.expected_streams():
-            if self.is_insight(stream):
-                expected_insights_streams.append(stream)
-
-        # core streams does not respect start date so will be descoped from this test
-
-        # # Run in check mode
+        # Run in check mode
         found_catalogs = self.run_and_verify_check_mode(conn_id)
 
         # Select only the expected streams tables
-        catalog_entries = [ce for ce in found_catalogs if ce['tap_stream_id'] in expected_insights_streams]
+        catalog_entries = [ce for ce in found_catalogs if ce['tap_stream_id'] in expected_streams]
         self.perform_and_verify_table_and_field_selection(conn_id, catalog_entries, select_all_fields=True)
 
         # Run a sync job using orchestrator
@@ -76,7 +77,7 @@ def run_test(self, attr_window, start_date, end_date):
 
         expected_replication_keys = self.expected_replication_keys()
 
-        for stream in expected_insights_streams:
+        for stream in expected_streams:
             with self.subTest(stream=stream):
 
                 replication_key = next(iter(expected_replication_keys[stream]))

tests/test_facebook_automatic_fields.py

@@ -15,6 +15,9 @@ class FacebookAutomaticFields(FacebookBaseTest):
     def name():
         return "tap_tester_facebook_automatic_fields"
 
+    def streams_to_test(self):
+        return self.expected_streams()
+
     def get_properties(self, original: bool = True):
         """Configuration properties required for the tap."""
         return_value = {
@@ -41,7 +44,7 @@ def test_run(self):
         that 251 (or more) records have been posted for that stream.
         """
 
-        expected_streams = self.expected_streams()
+        expected_streams = self.streams_to_test()
 
         # instantiate connection
         conn_id = connections.ensure_connection(self)

tests/test_facebook_bookmarks.py

@@ -13,6 +13,9 @@ class FacebookBookmarks(FacebookBaseTest):
     def name():
         return "tap_tester_facebook_bookmarks"
 
+    def streams_to_test(self):
+        return self.expected_streams()
+
     def get_properties(self, original: bool = True):
         """Configuration properties required for the tap."""
         return_value = {
@@ -96,7 +99,7 @@ def is_expected_date_format(self, date):
         return True
 
     def test_run(self):
-        expected_streams =  self.expected_streams()
+        expected_streams = self.streams_to_test()
         non_insight_streams = {stream for stream in expected_streams if not self.is_insight(stream)}
         insight_streams = {stream for stream in expected_streams if self.is_insight(stream)}
 

tests/test_facebook_discovery.py

@@ -8,11 +8,13 @@
 
 class DiscoveryTest(FacebookBaseTest):
     """Test tap discovery mode and metadata conforms to standards."""
-
     @staticmethod
     def name():
         return "tap_tester_facebook_discovery_test"
 
+    def streams_to_test(self):
+        return self.expected_streams()
+
     def test_run(self):
         """
         Testing that discovery creates the appropriate catalog with valid metadata.
@@ -30,7 +32,7 @@ def test_run(self):
           are given the inclusion of automatic.
         • verify that all other fields have inclusion of available metadata.
         """
-        streams_to_test = self.expected_streams()
+        streams_to_test = self.streams_to_test()
 
         conn_id = connections.ensure_connection(self)
 

tests/test_facebook_field_selection.py

@@ -13,7 +13,7 @@ def name():
         return "tap_tester_facebook_field_selection"
 
     @staticmethod
-    def expected_check_streams():
+    def streams_to_test():
         return {
             'ads',
             'adcreative',
@@ -29,23 +29,6 @@ def expected_check_streams():
             #'leads',
         }
 
-    @staticmethod
-    def expected_sync_streams():
-        return {
-            "ads",
-            "adcreative",
-            "adsets",
-            "campaigns",
-            "ads_insights",
-            "ads_insights_age_and_gender",
-            "ads_insights_country",
-            "ads_insights_platform_and_device",
-            "ads_insights_region",
-            "ads_insights_dma",
-            "ads_insights_hourly_advertiser",
-            #"leads",
-        }
-
     @staticmethod
     def expected_pks():
         return {
@@ -76,6 +59,9 @@ def get_properties(self):  # pylint: disable=arguments-differ
         }
 
     def test_run(self):
+
+        expected_streams = self.streams_to_test()
+
         conn_id = connections.ensure_connection(self)
 
         # run in check mode
@@ -90,7 +76,7 @@ def test_run(self):
 
         found_catalog_names = set(map(lambda c: c['tap_stream_id'], found_catalogs))
 
-        diff = self.expected_check_streams().symmetric_difference( found_catalog_names )
+        diff = expected_streams.symmetric_difference( found_catalog_names )
         self.assertEqual(len(diff), 0, msg="discovered schemas do not match: {}".format(diff))
         print("discovered schemas are kosher")
 
@@ -118,7 +104,7 @@ def test_run(self):
         menagerie.verify_sync_exit_status(self, exit_status, sync_job_name)
 
         # This should be validating the the PKs are written in each record
-        record_count_by_stream = runner.examine_target_output_file(self, conn_id, self.expected_sync_streams(), self.expected_pks())
+        record_count_by_stream = runner.examine_target_output_file(self, conn_id, expected_streams, self.expected_pks())
         replicated_row_count =  reduce(lambda accum,c : accum + c, record_count_by_stream.values())
         self.assertGreater(replicated_row_count, 0, msg="failed to replicate any data: {}".format(record_count_by_stream))
         print("total replicated row count: {}".format(replicated_row_count))

tests/test_facebook_invalid_attribution_window.py

@@ -11,6 +11,10 @@ class FacebookInvalidAttributionWindowInt(FacebookBaseTest):
     def name():
         return "tt_facebook_invalid_window_int"
 
+    @staticmethod
+    def streams_to_test():
+        return []
+
     def get_properties(self, original: bool = True):
         """Configuration properties required for the tap."""
         return_value = {
@@ -31,7 +35,7 @@ def test_run(self):
 
     def run_test(self):
         """
-            Test to verify that the error is raise when passing attribution window other than 1, 7 or 28
+        Test to verify that the error is raise when passing attribution window other than 1, 7 or 28
         """
         # create connection
         conn_id = connections.ensure_connection(self)

tests/test_facebook_start_date.py

@@ -14,6 +14,9 @@ class FacebookStartDateTest(FacebookBaseTest):
     def name():
         return "tap_tester_facebook_start_date_test"
 
+    def streams_to_test(self):
+        return self.expected_streams()
+
     def get_properties(self, original: bool = True):
         """Configuration properties required for the tap."""
         return_value = {
@@ -36,7 +39,7 @@ def test_run(self):
 
         self.start_date = self.start_date_1
 
-        expected_streams = self.expected_streams()
+        expected_streams = self.streams_to_test()
 
         ##########################################################################
         ### First Sync