fix nwis_client bug causing KeyError when station does not return data

python/nwis_client/setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = hydrotools.nwis_client
-version = 3.0.4
+version = 3.0.5
 author = Austin Raney
 author_email = [email protected]
 description = A convenient interface to the USGS NWIS Instantaneous Values (IV) REST Service API.

python/nwis_client/src/hydrotools/nwis_client/iv.py

@@ -247,17 +247,25 @@ def list_to_df_helper(item: dict):
 
             return df
 
+        def empty_df_warning_helper():
+            warning_message = "No data was returned by the request."
+            warnings.warn(warning_message)
+
         list_of_frames = list(map(list_to_df_helper, raw_data))
 
         # Empty list. No data was returned in the request
         if not list_of_frames:
-            warning_message = "No data was returned by the request."
-            warnings.warn(warning_message)
-            return pd.DataFrame(None)
+            empty_df_warning_helper()
+            return _create_empty_canonical_df()
 
         # Concatenate list in single pd.DataFrame
         dfs = pd.concat(list_of_frames, ignore_index=True)
 
+        # skip data processing steps if no data was retrieved and return empty canonical df
+        if dfs.empty:
+            empty_df_warning_helper()
+            return _create_empty_canonical_df()
+
         # Convert values to numbers
         dfs.loc[:, "value"] = pd.to_numeric(dfs["value"], downcast="float")
 
@@ -809,3 +817,17 @@ def flatten_and_stringify(v):
     value_groups = np.array_split(values, n_groups)
 
     return list(map(lambda i: ",".join(i), value_groups))
+
+
+def _create_empty_canonical_df() -> pd.DataFrame:
+    """Returns an empty hydrotools canonical dataframe with correct field datatypes."""
+    cols = {
+        "value_time": pd.Series(dtype="datetime64[ns]"),
+        "variable_name": pd.Series(dtype="category"),
+        "usgs_site_code": pd.Series(dtype="category"),
+        "measurement_unit": pd.Series(dtype="category"),
+        "value": pd.Series(dtype="float32"),
+        "qualifiers": pd.Series(dtype="category"),
+        "series": pd.Series(dtype="category"),
+    }
+    return pd.DataFrame(cols, index=[])

python/nwis_client/tests/test_nwis.py

@@ -423,3 +423,17 @@ def test_get_raw_countyCd(setup_iv):
 def test_splitting_bbox(test, validation):
     v = iv._bbox_split(test)
     assert v == validation
+
+
+def test_get_returns_empty_canonical_dataframe(setup_iv, monkeypatch):
+    """Verify that `get` can returns an empty canonical dataframe."""
+
+    def get_raw_mock(*args, **kwargs):
+        return [{"values": []}]
+
+    monkeypatch.setattr(iv.IVDataService, "get_raw", get_raw_mock)
+    df = setup_iv.get(
+        sites=["01189000"], startDt="2015-12-01T00:00", endDt="2015-12-31T23:45"
+    )
+    canonical_df = iv._create_empty_canonical_df()
+    assert df.equals(canonical_df)