Cloud diffs in progress

data_diff/cloud/datafold_api.py

@@ -6,6 +6,9 @@
 import pydantic
 import requests
 
+from ..utils import getLogger
+
+logger = getLogger(__name__)
 
 Self = TypeVar("Self", bound=pydantic.BaseModel)
 
@@ -216,11 +219,12 @@ def poll_data_diff_results(self, diff_id: int) -> TCloudApiDataDiffSummaryResult
         summary_results = None
         start_time = time.monotonic()
         sleep_interval = 5  # starts at 5 sec
-        max_sleep_interval = 60
+        max_sleep_interval = 30
         max_wait_time = 300
 
         diff_url = f"{self.host}/datadiffs/{diff_id}/overview"
         while not summary_results:
+            logger.debug(f"Polling: {diff_url}")
             response = self.make_get_request(url=f"api/v1/datadiffs/{diff_id}/summary_results")
             response_json = response.json()
             if response_json["status"] == "success":

data_diff/dbt.py

@@ -118,6 +118,7 @@ def dbt_diff(
                     "Datasource ID not found, include it as a dbt variable in the dbt_project.yml. "
                     "\nvars:\n data_diff:\n   datasource_id: 1234"
                 )
+        rich.print("[green][bold]\nDiffs in progress...[/][/]\n")
 
     else:
         dbt_parser.set_connection()
@@ -288,14 +289,14 @@ def _cloud_diff(diff_vars: DiffVars, datasource_id: int, api: DatafoldAPI) -> No
     diff_url = None
     try:
         diff_id = api.create_data_diff(payload=payload)
+        diff_url = f"{api.host}/datadiffs/{diff_id}/overview"
+        rich.print(f"{diff_vars.dev_path[2]}: {diff_url}")
 
         if diff_id is None:
             raise Exception(f"Api response did not contain a diff_id")
 
         diff_results = api.poll_data_diff_results(diff_id)
 
-        diff_url = f"{api.host}/datadiffs/{diff_id}/overview"
-
         rows_added_count = diff_results.pks.exclusives[1]
         rows_removed_count = diff_results.pks.exclusives[0]
 
@@ -352,4 +353,4 @@ def _cloud_diff(diff_vars: DiffVars, datasource_id: int, api: DatafoldAPI) -> No
 
 
 def _diff_output_base(dev_path: str, prod_path: str) -> str:
-    return f"[green]{prod_path} <> {dev_path}[/] \n"
+    return f"\n[green]{prod_path} <> {dev_path}[/] \n"

tests/test_dbt.py

@@ -499,7 +499,7 @@ def test_cloud_diff(self, mock_api, mock_os_environ, mock_print):
         _cloud_diff(diff_vars, expected_datasource_id, api=mock_api)
 
         mock_api.create_data_diff.assert_called_once()
-        mock_print.assert_called_once()
+        self.assertEqual(mock_print.call_count, 2)
 
         payload = mock_api.create_data_diff.call_args[1]["payload"]
         self.assertEqual(payload.data_source1_id, expected_datasource_id)
@@ -541,7 +541,7 @@ def test_diff_is_cloud(
         mock_initialize_api.assert_called_once()
         mock_cloud_diff.assert_called_once_with(expected_diff_vars, 1, api)
         mock_local_diff.assert_not_called()
-        mock_print.assert_not_called()
+        mock_print.assert_called_once()
 
     @patch("data_diff.dbt._initialize_api")
     @patch("data_diff.dbt._get_diff_vars")
@@ -722,7 +722,7 @@ def test_diff_is_cloud_no_pks(
         mock_dbt_parser_inst.set_connection.assert_not_called()
         mock_cloud_diff.assert_not_called()
         mock_local_diff.assert_not_called()
-        self.assertEqual(mock_print.call_count, 1)
+        self.assertEqual(mock_print.call_count, 2)
 
     @patch("data_diff.dbt._get_diff_vars")
     @patch("data_diff.dbt._local_diff")