Set Default Include Anonymous Visitors, handle None values, add unit tests

tap_pendo/streams.py

@@ -28,6 +28,7 @@
 session = requests.Session()
 # timeout request after 300 seconds
 REQUEST_TIMEOUT = 300
+DEFAULT_INCLUDE_ANONYMOUS_VISITORS = 'false'
 
 def get_abs_path(path):
     return os.path.join(os.path.dirname(os.path.realpath(__file__)), path)
@@ -1000,7 +1001,8 @@ class Visitors(LazyAggregationStream):
     key_properties = ["visitor_id"]
 
     def get_body(self):
-        include_anonymous_visitors = bool(self.config.get('include_anonymous_visitors', 'false').lower() == 'true')
+        include_anonymous_visitors = self.config.get('include_anonymous_visitors') or DEFAULT_INCLUDE_ANONYMOUS_VISITORS
+        anons = str(include_anonymous_visitors).lower() == 'true'
         return {
             "response": {
                 "mimeType": "application/json"
@@ -1011,7 +1013,7 @@ def get_body(self):
                 "pipeline": [{
                     "source": {
                         "visitors": {
-                            "identified": not include_anonymous_visitors
+                            "identified": not anons
                         }
                     }
                 }],

tests/tap_tester/test_start_date.py

@@ -17,7 +17,8 @@ def name(self):
         return "pendo_start_date_test"
 
     def test_run(self):
-        self.run_test("2021-09-09T00:00:00Z", "2022-05-01T00:00:00Z", {"accounts", "visitors", "metadata_visitors", "metadata_accounts"})
+        self.run_test("2022-06-10T00:00:00Z", "2022-07-22T00:00:00Z", {"accounts"})
+        self.run_test("2021-09-09T00:00:00Z", "2022-05-01T00:00:00Z", {"visitors", "metadata_visitors", "metadata_accounts"})
         self.run_test("2020-09-01T00:00:00Z", "2021-03-01T00:00:00Z", {"features", "feature_events", "pages", "page_events", "events"})
         self.run_test("2021-09-09T00:00:00Z", "2021-09-16T00:00:00Z", {"guides", "guide_events"})
         self.run_test("2021-09-13T00:00:00Z", "2021-09-15T00:00:00Z", {"track_types", "track_events"})

tests/unittests/test_lazy_aggregation_sync.py

@@ -63,7 +63,7 @@ def test_lazzy_aggregation_with_sub_stream(self, mocked_substream, mocked_select
     def test_lazzy_aggregation_without_sub_stream(self, mocked_substream, mocked_selected, mocked_request):
         '''
             Verify that if sub stream is not selected then also all data should be return for super stream
-            and sync_substream should not be called 
+            and sync_substream should not be called
         '''
         expected_data = [{"id":1}, {"id":2}, {"id":3}]
         records = '{"results": [{"id":1}, {"id":2}, {"id":3}]}'
@@ -77,3 +77,111 @@ def test_lazzy_aggregation_without_sub_stream(self, mocked_substream, mocked_sel
 
         self.assertEqual(list(stream_response), expected_data)
         self.assertEqual(mocked_substream.call_count, 0)
+
+class TestConfigParsing(unittest.TestCase):
+
+    def test_reading_include_anonymous_visitors_provided_valid_input_true(self):
+        config = {
+            'include_anonymous_visitors': 'true'
+        }
+
+        my_visitor = Visitors(config)
+
+
+        return_value = my_visitor.get_body()
+        expected_value = True
+
+        # This is the value that matches the config
+        actual_value = not return_value['request']['pipeline'][0]['source']['visitors']['identified']
+
+        self.assertEqual(expected_value, actual_value)
+
+    def test_reading_include_anonymous_visitors_provided_valid_input_TRUE(self):
+        config = {
+            'include_anonymous_visitors': 'TRUE'
+        }
+
+        my_visitor = Visitors(config)
+
+
+        return_value = my_visitor.get_body()
+        expected_value = True
+
+        # This is the value that matches the config
+        actual_value = not return_value['request']['pipeline'][0]['source']['visitors']['identified']
+
+        self.assertEqual(expected_value, actual_value)
+
+    def test_reading_include_anonymous_visitors_provided_valid_input_false(self):
+        config = {
+            'include_anonymous_visitors': 'false'
+        }
+
+        my_visitor = Visitors(config)
+
+
+        return_value = my_visitor.get_body()
+        expected_value = False
+
+        # This is the value that matches the config
+        actual_value = not return_value['request']['pipeline'][0]['source']['visitors']['identified']
+
+        self.assertEqual(expected_value, actual_value)
+
+    def test_reading_include_anonymous_visitors_provided_invalid_input(self):
+        config = {
+            'include_anonymous_visitors': 'bad input'
+        }
+
+        my_visitor = Visitors(config)
+
+        return_value = my_visitor.get_body()
+        expected_value = False
+        actual_value = not return_value['request']['pipeline'][0]['source']['visitors']['identified']
+        self.assertEqual(expected_value, actual_value)
+
+    def test_reading_include_anonymous_visitors_provided_none(self):
+        config = {
+            'include_anonymous_visitors': None
+        }
+
+        my_visitor = Visitors(config)
+
+        return_value = my_visitor.get_body()
+        expected_value = False
+        actual_value = not return_value['request']['pipeline'][0]['source']['visitors']['identified']
+        self.assertEqual(expected_value, actual_value)
+
+    def test_reading_include_anonymous_visitors_no_provided_input(self):
+        config = {}
+
+        my_visitor = Visitors(config)
+
+        return_value = my_visitor.get_body()
+        expected_value = False
+        actual_value = not return_value['request']['pipeline'][0]['source']['visitors']['identified']
+        self.assertEqual(expected_value, actual_value)
+
+    def test_reading_include_anonymous_visitors_provided_boolean_true(self):
+        config = {
+            'include_anonymous_visitors': True
+        }
+
+        my_visitor = Visitors(config)
+
+        return_value = my_visitor.get_body()
+        expected_value = True
+        actual_value = not return_value['request']['pipeline'][0]['source']['visitors']['identified']
+        self.assertEqual(expected_value, actual_value)
+
+    def test_reading_include_anonymous_visitors_provided_boolean_false(self):
+        config = {
+            'include_anonymous_visitors': False
+        }
+
+        my_visitor = Visitors(config)
+
+        return_value = my_visitor.get_body()
+        expected_value = False
+        actual_value = not return_value['request']['pipeline'][0]['source']['visitors']['identified']
+        self.assertEqual(expected_value, actual_value)