Fix freshness if no data, hanlde corner cases better

soda-core/src/soda_core/contracts/impl/check_types/freshness_check.py

@@ -91,7 +91,13 @@ def evaluate(self, measurement_values: MeasurementValues, contract: Contract) ->
         threshold_metric_name: str = f"freshness_in_{self.unit}s"
 
         threshold_value: Optional[float] = None
-        if data_timestamp_utc and max_timestamp_utc:
+        if max_timestamp_utc is None:
+            outcome = CheckOutcome.FAILED
+
+        elif data_timestamp_utc is not None:
+            logger.debug(
+                f"Calculating freshness using '{max_timestamp}' as 'max' and '{data_timestamp}' as 'now' values"
+            )
             freshness = data_timestamp_utc - max_timestamp_utc
             freshness_in_seconds = freshness.total_seconds()
 
@@ -111,6 +117,8 @@ def evaluate(self, measurement_values: MeasurementValues, contract: Contract) ->
                 else:
                     outcome = CheckOutcome.FAILED
 
+        freshness_str: Optional[str] = str(freshness) if freshness is not None else None
+
         return FreshnessCheckResult(
             contract=contract,
             check=self._build_check_info(),
@@ -121,27 +129,32 @@ def evaluate(self, measurement_values: MeasurementValues, contract: Contract) ->
             max_timestamp_utc=max_timestamp_utc,
             data_timestamp=data_timestamp,
             data_timestamp_utc=data_timestamp_utc,
-            freshness=str(freshness),
+            freshness=freshness_str,
             freshness_in_seconds=freshness_in_seconds,
             unit=self.unit,
         )
 
     def _get_max_timestamp(self, measurement_values: MeasurementValues) -> Optional[datetime]:
         max_timestamp: Optional[datetime] = measurement_values.get_value(self.max_timestamp_metric)
-        if not isinstance(max_timestamp, datetime):
+        if max_timestamp is None:
+            logger.warning(
+                f"Freshness metric '{self.max_timestamp_metric.type}' for column '{self.column}' "
+                f"returned no value. Does the table or partition have rows?"
+            )
+            return None
+        elif not isinstance(max_timestamp, datetime):
+            logger.debug(
+                f"Attempting to convert freshness value '{max_timestamp}' of data type '{type(max_timestamp).__name__}' to datetime"
+            )
             if isinstance(max_timestamp, date):
                 max_timestamp = datetime.combine(max_timestamp, datetime.min.time())
             elif isinstance(max_timestamp, str):
                 max_timestamp = convert_str_to_datetime(max_timestamp)
-            else:
-                logger.error(
-                    f"Freshness metric '{self.max_timestamp_metric.type}' for column '{self.column}' "
-                    f"has an invalid data type '({type(max_timestamp).__name__})'. "
-                    f"Is the column a timestamp or a timestamp-compatible type?"
-                )
 
         if not isinstance(max_timestamp, datetime):
-            logger.error(f"Freshness column '{self.column}' does not have timestamp values: '{max_timestamp}'")
+            logger.error(
+                f"Freshness column '{self.column}' returned value '{max_timestamp}' of data type '{type(max_timestamp).__name__}' which is not a datetime or datetime-compatible type."
+            )
             max_timestamp = None
 
         return max_timestamp

soda-tests/tests/features/test_freshness_check.py

@@ -115,3 +115,35 @@ def test_freshness_now_variable(data_source_test_helper: DataSourceTestHelper):
     assert str(check_result.freshness_in_seconds) == "3600.0"
     assert str(check_result.unit) == "hour"
     assert 0.99 < get_diagnostic_value(check_result, "freshness_in_hours") < 1.01
+
+
+def test_freshness_no_rows(data_source_test_helper: DataSourceTestHelper):
+    test_table = data_source_test_helper.ensure_test_table(test_table_specification)
+
+    id_quoted = data_source_test_helper.quote_column("id")
+
+    contract_yaml_str: str = f"""
+        filter: |
+          {id_quoted} > 10
+        checks:
+          - freshness:
+              column: created_at
+              threshold:
+                must_be_less_than: 2
+    """
+
+    with freeze_time(datetime(year=2025, month=1, day=4, hour=10, minute=0, second=0)):
+        contract_verification_result_t1: ContractVerificationResult = data_source_test_helper.assert_contract_fail(
+            test_table=test_table, contract_yaml_str=contract_yaml_str
+        )
+        check_result: FreshnessCheckResult = contract_verification_result_t1.check_results[0]
+        assert check_result.max_timestamp is None
+        assert check_result.max_timestamp_utc is None
+        assert str(check_result.data_timestamp) == "2025-01-04 10:00:00+00:00"
+        assert str(check_result.data_timestamp_utc) == "2025-01-04 10:00:00+00:00"
+        assert check_result.freshness is None
+        assert check_result.freshness_in_seconds is None
+        assert str(check_result.unit) == "hour"
+        assert len(check_result.diagnostic_metric_values) == 0
+
+        assert not contract_verification_result_t1.has_errors()