Merge branch '7dav_new_geo_ids' of https://github.com/cmu-delphi/covidcast-indicators into 7dav_new_geo_ids

Author: qx-teo

_delphi_utils_python/delphi_utils/validator/dynamic.py

@@ -183,22 +183,25 @@ def check_na_vals(self, geo_sig_df, geo_type, signal_type, report):
         Returns:
             - None
         """
-        def replace_first_six(df):
+        def replace_first_six(df, start_date):
             x = df.val.isnull()
             # First 6 days have to be null
             x.iloc[:6] = False
-            return df.time_value[x]
+            df = df[x]
+            return df.time_value[df.time_value >= start_date]
 
         grouped_df = geo_sig_df.groupby('geo_id')
-        error_df = grouped_df.apply(replace_first_six)
+        error_df = grouped_df.apply(replace_first_six,
+            start_date = self.params.time_window.start_date)
 
         if not error_df.empty:
             for index, value in error_df.iteritems():
                 report.add_raised_error(
-                    ValidationFailure(f"check_val_missing (geo_id {index[0]})",
+                    ValidationFailure("check_val_missing",
                                       geo_type=geo_type,
                                       signal=signal_type,
-                                      date=value
+                                      date=value,
+                                      message=f"geo_id {index[0]}"
                                       )
                 )
 

_delphi_utils_python/tests/validator/test_dynamic.py

@@ -105,25 +105,29 @@ def test_0_vs_many(self):
 
         assert len(report.raised_errors) == 1
         assert report.raised_errors[0].check_name == "check_rapid_change_num_rows"
+
 class TestCheckNaVals:
     params = {
         "common": {
             "data_source": "",
-            "span_length": 1,
+            "span_length": 14,
             "end_date": "2020-09-02"
         }
     }
     def test_missing(self):
         validator = DynamicValidator(self.params)
         report = ValidationReport([])
-        data = {"val": [np.nan] * 14, "geo_id": [0,1] * 7, "time_value": ["2021-09-01"] * 14 }
+        data = {"val": [np.nan] * 15, "geo_id": [0,1] * 7 + [2], 
+            "time_value": ["2021-08-30"] * 14 + ["2021-05-01"]}
         df = pd.DataFrame(data)
-        #df.set_index(range(7), inplace=True)
+        df.time_value = (pd.to_datetime(df.time_value)).dt.date
         validator.check_na_vals(df, "geo", "signal", report)
 
         assert len(report.raised_errors) == 2
-        assert report.raised_errors[0].check_name == "check_val_missing (geo_id 0)"
-        assert report.raised_errors[1].check_name == "check_val_missing (geo_id 1)"
+        assert report.raised_errors[0].check_name == "check_val_missing"
+        assert report.raised_errors[0].message == "geo_id 0"
+        assert report.raised_errors[1].check_name == "check_val_missing"
+        assert report.raised_errors[1].message == "geo_id 1"
 
 class TestCheckAvgValDiffs:
     params = {