Add support for non-numeric keys in KeyData.as_single_value via uniqueness check

Author: philsmt

extra_data/keydata.py

@@ -327,7 +327,7 @@ def data_counts(self, labelled=True):
 
             return res
 
-    def as_single_value(self, rtol=1e-5, atol=0.0, reduce_by='median'):
+    def as_single_value(self, rtol=1e-5, atol=0.0, reduce_by=None):
         """Retrieve a single reduced value if within tolerances.
 
         The relative and absolute tolerances *rtol* and *atol* work the
@@ -340,13 +340,33 @@ def as_single_value(self, rtol=1e-5, atol=0.0, reduce_by='median'):
         the first value encountered. By default, 'median' is used.
 
         If within tolerances, the reduced value is returned.
+
+        For non-numerical keys like strings, the method instead always
+        checks for uniqueness and returns such a value, if present.
         """
 
         data = self.ndarray()
 
         if len(data) == 0:
             raise NoDataError(self.source, self.key)
 
+        if not np.issubdtype(self.dtype, np.number):
+            # Handle non-numeric types first.
+
+            if reduce_by is not None:
+                raise TypeError('custom reduce method not supported for '
+                                'non-numeric type')
+
+            unique_values = np.unique(data, axis=None)
+
+            if len(unique_values) > 1:
+                raise ValueError(f'str values are not unique: {unique_values}')
+
+            return unique_values[0]
+
+        elif reduce_by is None:
+            reduce_by = 'median'
+
         if callable(reduce_by):
             value = reduce_by(data)
         elif isinstance(reduce_by, str) and hasattr(np, reduce_by):

extra_data/tests/test_keydata.py

@@ -11,6 +11,7 @@
 from .mockdata import write_file
 from .mockdata.xgm import XGM
 
+
 def test_get_keydata(mock_spb_raw_run):
     run = RunDirectory(mock_spb_raw_run)
     print(run.instrument_sources)
@@ -43,6 +44,7 @@ def test_get_keydata(mock_spb_raw_run):
     with pytest.raises(TypeError):
         iter(xgm_beam_x)
 
+
 def test_select_trains(mock_spb_raw_run):
     run = RunDirectory(mock_spb_raw_run)
     xgm_beam_x = run['SPB_XTD9_XGM/DOOCS/MAIN', 'beamPosition.ixPos.value']
@@ -277,6 +279,7 @@ def test_single_value(mock_sa3_control_data, monkeypatch):
 
     imager = f['SA3_XTD10_IMGFEL/CAM/BEAMVIEW:daqOutput', 'data.image.pixels']
     flux = f['SA3_XTD10_XGM/XGM/DOOCS', 'pulseEnergy.photonFlux']
+    state = f['SA3_XTD10_XGM/XGM/DOOCS', 'state']
 
     # Try without data for a source and key.
     with pytest.raises(NoDataError):
@@ -309,6 +312,17 @@ def test_single_value(mock_sa3_control_data, monkeypatch):
     assert flux.as_single_value(rtol=1, reduce_by=np.mean) == np.mean(data)
     assert flux.as_single_value(atol=len(data)-1, reduce_by='first') == 0
 
+    # Try strings.
+    assert state[5:].as_single_value() == 'ON'
+
+    with pytest.raises(ValueError):
+        # Contains two unique values.
+        state.as_single_value()
+
+    with pytest.raises(TypeError):
+        # Does not accept reduce_by
+        state.as_single_value(reduce_by='mean')
+
     # Try vector data.
     intensity = f['SA3_XTD10_XGM/XGM/DOOCS:output', 'data.intensityTD']
     data = np.repeat(data, intensity.shape[1]).reshape(-1, intensity.shape[-1])
@@ -405,6 +419,6 @@ def test_units(mock_sa3_control_data):
 
     # Check that it still works after selecting 0 trains
     assert xgm_intensity.select_trains(np.s_[:0]).units == 'μJ'
-    
+
     # units are added to xarray's attributes
     assert xgm_intensity.xarray().attrs['units'] == 'μJ'