better handling of load cache vs. dont load cache

NREL · Nov 27, 2023 · b32d7d6 · b32d7d6
1 parent 357440f
commit b32d7d6
Showing 3 changed files with 13 additions and 10 deletions.
diff --git a/sup3r/preprocessing/data_handling/dual_data_handling.py b/sup3r/preprocessing/data_handling/dual_data_handling.py
@@ -185,7 +185,8 @@ def means(self):
 
         if self.hr_dh.data is None:
             msg = ('High-res DataHandler object has DataHandler.data=None! '
-                   'Try initializing DualDataHandler with load_cached=True')
+                   'Try initializing the high-res handler with '
+                   'load_cached=True')
             logger.error(msg)
             raise RuntimeError(msg)
 
@@ -206,7 +207,8 @@ def stds(self):
 
         if self.hr_dh.data is None:
             msg = ('High-res DataHandler object has DataHandler.data=None! '
-                   'Try initializing DualDataHandler with load_cached=True')
+                   'Try initializing the high-res handler with '
+                   'load_cached=True')
             logger.error(msg)
             raise RuntimeError(msg)
 
@@ -234,10 +236,10 @@ def normalize(self, means=None, stds=None, max_workers=None):
             Has no effect. Used to match MixIn class signature.
         """
 
-        if self.lr_dh.data is None or self.hr_dh.data is None:
-            msg = ('Low-res or High-res DataHandler object has '
-                   'DataHandler.data=None! Try initializing DualDataHandler '
-                   'with load_cached=True')
+        if self.hr_dh.data is None:
+            msg = ('High-res DataHandler object has DataHandler.data=None! '
+                   'Try initializing the high-res handler with '
+                   'load_cached=True')
             logger.error(msg)
             raise RuntimeError(msg)
 
@@ -254,6 +256,7 @@ def normalize(self, means=None, stds=None, max_workers=None):
         self.lr_dh.normalize(means=means, stds=stds,
                              features=self.lr_dh.features,
                              max_workers=self.lr_dh.norm_workers)
+
         logger.info('Normalizing high resolution data features='
                     f'{self.hr_dh.features}')
         self.hr_dh.normalize(means=means, stds=stds,

diff --git a/sup3r/preprocessing/data_handling/mixin.py b/sup3r/preprocessing/data_handling/mixin.py
@@ -1156,7 +1156,7 @@ def normalize(self, means=None, stds=None, features=None,
 
         if self._is_normalized:
             logger.info('Skipping DataHandler, already normalized')
-        else:
+        elif self.data is not None:
             self._normalize(self.data,
                             self.val_data,
                             features=features,

diff --git a/tests/data_handling/test_dual_data_handling.py b/tests/data_handling/test_dual_data_handling.py
@@ -615,13 +615,13 @@ def test_bad_cache_load():
                                        t_enhance=t_enhance,
                                        cache_pattern=dual_cache,
                                        load_cached=False,
-                                       val_split=0.1)
+                                       val_split=0.0)
 
         # because load_cached is False
         assert hr_handler.data is None
         assert lr_handler.data is not None
 
-        good_err = "Try initializing DualDataHandler with load_cached=True"
+        good_err = "DataHandler.data=None!"
         with pytest.raises(RuntimeError) as ex:
             _ = copy.deepcopy(dual_handler.means)
         assert good_err in str(ex.value)
@@ -640,7 +640,7 @@ def test_bad_cache_load():
                                        t_enhance=t_enhance,
                                        cache_pattern=dual_cache,
                                        load_cached=True,
-                                       val_split=0.1)
+                                       val_split=0.0)
 
         # because load_cached is True
         assert hr_handler.data is not None