Testing data normalisation

.gitignore

@@ -172,3 +172,6 @@ src/mypy.txt
 
 # Ignore .DS_Store files
 .DS_Store
+
+.ruff_cache/
+temp_dir/

packages/data/src/pyearthtools/data/transforms/normalisation/__init__.py

@@ -14,7 +14,7 @@
 
 
 """
-A sophisticated [Transform][pyearthtools.data.transforms.Transform] to normalise and unnormalise data.
+A sophisticated [Transform][pyearthtools.data.transforms.Transform] to normalise and denormalise data.
 
 ## Methods
 | Name        | Description |
@@ -30,10 +30,10 @@
 ## Transforms
 [Normalise][pyearthtools.data.transforms.normalisation.normalise] provides the Transforms to normalise incoming data
 
-[UnNormalise][pyearthtools.data.transforms.normalisation.unnormalise] provides the Transforms to unnormalise incoming data
+[Denormalise][pyearthtools.data.transforms.normalisation.denormalise] provides the Transforms to denormalise incoming data
 
 """
 
 from pyearthtools.data.transforms.normalisation import _utils
 from pyearthtools.data.transforms.normalisation.normalise import Normalise
-from pyearthtools.data.transforms.normalisation.unnormalise import Unnormalise
+from pyearthtools.data.transforms.normalisation.denormalise import Denormalise

packages/data/src/pyearthtools/data/transforms/normalisation/default.py

@@ -78,9 +78,9 @@ def __init__(
         """
         Base Normalise Class
 
-        Setup Transformer Class to normalise and unnormalise data
+        Setup Transformer Class to normalise and denormalise data
 
-        Can't be used directly, see `Normalise`, & `Unnormalise`.
+        Can't be used directly, see `Normalise`, & `Denormalise`.
 
         Anomaly, Range, Deviation all require `start`, `end`, and `interval` to be given or `file`.
 
@@ -122,14 +122,14 @@ def __init__(
         kwargs.update(skip_invalid=True)
         self.retrieval_arguments = dict(start=start, end=end, interval=interval, verbose=verbose, **kwargs)
 
+        # TODO - Consider wrapping in a try/except block to catch if TemporaryDirectory() fails
         if cache == "temp":
             temp_dir = tempfile.TemporaryDirectory()
             cache = temp_dir.name
             self.temp_dir = temp_dir
 
         self.cache_dir = cache
         self.index = index
-
         self._function = function
         self.override = override
         self.verbose = verbose
@@ -156,7 +156,7 @@ def check_init_args(self):
         for arg in ["start", "end", "interval"]:
             if not self.retrieval_arguments[arg]:
                 raise RuntimeError(
-                    f"`override`, or (`start`, `end` and `interval`) was not given in `__init__`."
+                    "`override`, or (`start`, `end` and `interval`) was not given in `__init__`."
                     "These must be given in order to find the normalisation values."
                 )
         return True
@@ -356,7 +356,7 @@ def as_float(**kwargs):
         return {variable_name: range[variable_name]}
 
     def _find_user_normaliser(self, key: str):
-        if not "Normaliser" in key:
+        if "Normaliser" not in key:
             raise AttributeError(f"{key!r} does not contain 'Normaliser', so is being ignored")
         try:
             return dynamic_import(key)(self)
@@ -480,7 +480,7 @@ def apply(self, dataset: xr.Dataset):
 
     def __call__(self, method: str | dict | tuple, default: str | None = None) -> Transform:
         """
-        Get Transform to Normaliser or UnNormalise based on provided method
+        Get Transform to Normaliser or Denormalise based on provided method
 
         Args:
             method (dict): Dictionary assigning variable names to normalisation methods

packages/data/src/pyearthtools/data/transforms/normalisation/unnormalise.py → packages/data/src/pyearthtools/data/transforms/normalisation/denormalise.py

@@ -27,53 +27,53 @@
 xr.set_options(keep_attrs=True)
 
 
-class Unnormalise(Normaliser):
-    """Unnormalise Incoming Data"""
+class Denormalise(Normaliser):
+    """Denormalise Incoming Data"""
 
     @functools.wraps(Normaliser)
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 
     def __getattr__(self, key: str):
-        function = self._find_user_normaliser(key).unnormalise
+        function = self._find_user_normaliser(key).denormalise
         if isinstance(function, Transform):
             return function
         return FunctionTransform(function)
 
     @property
     def anomaly(normalise_self):
         """
-        UnNormalise using anomalies
+        Denormalise using anomalies
         """
 
-        class AnomalyUnNormaliser(Transform):
+        class AnomalyDenormaliser(Transform):
             @property
             def _info_(self):
                 return normalise_self._info_
 
             def apply(self, dataset: xr.Dataset):
-                """UnNormalise Dataset by creating anomalies"""
+                """Denormalise Dataset by creating anomalies"""
                 dataset = xr.Dataset(dataset)
                 for variable_name in dataset:
                     average = normalise_self.get_anomaly(variable_name)
                     dataset[variable_name] = dataset[variable_name] + average[variable_name]
                 return dataset
 
-        return AnomalyUnNormaliser()
+        return AnomalyDenormaliser()
 
     @property
     def range(normalise_self):
         """
-        UnNormalise using range
+        Denormalise using range
         """
 
-        class RangeUnNormaliser(Transform):
+        class RangeDenormaliser(Transform):
             @property
             def _info_(self):
                 return normalise_self._info_
 
             def apply(self, dataset: xr.Dataset):
-                """UnNormalise between 0-1 with ranges"""
+                """Denormalise between 0-1 with ranges"""
                 dataset = xr.Dataset(dataset)
                 for variable_name in dataset:
                     range = normalise_self.get_range(variable_name)
@@ -85,7 +85,7 @@ def apply(self, dataset: xr.Dataset):
 
                 return dataset
 
-        return RangeUnNormaliser()
+        return RangeDenormaliser()
 
     def manual_range(normalise_self, min: float, max: float):
         warnings.warn(
@@ -116,16 +116,16 @@ def apply(self, dataset: xr.Dataset):
     @property
     def deviation(normalise_self):
         """
-        UnNormalise using mean & standard deviation
+        Denormalise using mean & standard deviation
         """
 
-        class DeviationUnNormaliser(Transform):
+        class DeviationDenormaliser(Transform):
             @property
             def _info_(self):
                 return normalise_self._info_
 
             def apply(self, dataset: xr.Dataset):
-                """UnNormalise Dataset using mean & standard deviation"""
+                """Denormalise Dataset using mean & standard deviation"""
                 dataset = xr.Dataset(dataset)
                 for variable_name in dataset:
                     average, deviation = normalise_self.get_deviation(variable_name)
@@ -135,15 +135,15 @@ def apply(self, dataset: xr.Dataset):
                     ]
                 return dataset
 
-        return DeviationUnNormaliser()
+        return DeviationDenormaliser()
 
     @property
     def temporal_difference(normalise_self):
         """
         Normalise datasets using mean & standard deviation
         """
 
-        class TemporalDifferenceUnNormaliser(Transform):
+        class TemporalDifferenceDenormaliser(Transform):
             """Normalise Dataset using mean & standard deviation"""
 
             @property
@@ -159,21 +159,21 @@ def apply(self, dataset: xr.Dataset):
 
                 return dataset
 
-        return TemporalDifferenceUnNormaliser()
+        return TemporalDifferenceDenormaliser()
 
     @property
     def deviation_spatial(normalise_self):
         """
-        UnNormalise using mean & standard deviation
+        Denormalise using mean & standard deviation
         """
 
-        class DeviationUnNormaliser(Transform):
+        class DeviationDenormaliser(Transform):
             @property
             def _info_(self):
                 return normalise_self._info_
 
             def apply(self, dataset: xr.Dataset):
-                """UnNormalise Dataset using mean & standard deviation spatially"""
+                """Denormalise Dataset using mean & standard deviation spatially"""
                 dataset = xr.Dataset(dataset)
                 for variable_name in dataset:
                     average, deviation = normalise_self.get_deviation(variable_name, spatial=True)
@@ -183,27 +183,27 @@ def apply(self, dataset: xr.Dataset):
                     ]
                 return dataset
 
-        return DeviationUnNormaliser()
+        return DeviationDenormaliser()
 
     @property
     def log(normalise_self):
         """
-        UnNormalise using exp
+        Denormalise using exp
         """
 
-        class LogUnNormaliser(Transform):
+        class LogDenormaliser(Transform):
             @property
             def _info_(self):
                 return normalise_self._info_
 
             def apply(self, dataset: xr.Dataset):
-                """UnNormalise Dataset with exp"""
+                """Denormalise Dataset with exp"""
                 dataset = xr.Dataset(dataset)
                 for variable_name in dataset:
                     dataset[variable_name] = np.exp(dataset[variable_name])
                 return dataset
 
-        return LogUnNormaliser()
+        return LogDenormaliser()
 
     @property
     def function(self) -> FunctionTransform:
@@ -215,4 +215,4 @@ def function(self) -> FunctionTransform:
         """
         if self._function is None:
             raise ValueError("Cannot use function transform without a given function.\nTry giving `function` to init")
-        return FunctionTransform(self._function.unnormalise)
+        return FunctionTransform(self._function.denormalise)

packages/data/tests/download/test_arcoera5.py

@@ -2,6 +2,7 @@
 
 from pyearthtools.data.download import arcoera5
 
+# FIXME: Skip slow downloads unless chosen specifically
 
 def _load_sample(variables, levels, sample_time):
     arco = arcoera5.ARCOERA5(variables, levels=levels)