Merge branch 'refactor_24' into 88-make-readers-return-sasdata-objects

sasdata/metadata.py

@@ -338,10 +338,10 @@ def summary(self) -> str:
 
 class Instrument:
     def __init__(self, target: AccessorTarget):
-        self.aperture = Aperture(target.with_path_prefix("sasaperture"))
-        self.collimation = Collimation(target.with_path_prefix("sascollimation"))
-        self.detector = Detector(target.with_path_prefix("sasdetector"))
-        self.source = Source(target.with_path_prefix("sassource"))
+        self.aperture = Aperture(target.with_path_prefix("sasaperture|aperture"))
+        self.collimation = Collimation(target.with_path_prefix("sascollimation|collimation"))
+        self.detector = Detector(target.with_path_prefix("sasdetector|detector"))
+        self.source = Source(target.with_path_prefix("sassource|source"))
 
     def summary(self):
         return (
@@ -350,27 +350,51 @@ def summary(self):
             self.detector.summary() +
             self.source.summary())
 
+def decode_string(data):
+    """ This is some crazy stuff"""
+
+    if isinstance(data, str):
+        return data
+
+    elif isinstance(data, np.ndarray):
+
+        if data.dtype == object:
+
+            data = data.reshape(-1)
+            data = data[0]
+
+            if isinstance(data, bytes):
+                return data.decode("utf-8")
+
+            return str(data)
+
+        else:
+            return data.tobytes().decode("utf-8")
+
+    else:
+        return str(data)
 
 class Metadata:
     def __init__(self, target: AccessorTarget):
         self._target = target
 
-        self.instrument = Instrument(target.with_path_prefix("sasinstrument"))
-        self.process = Process(target.with_path_prefix("sasprocess"))
-        self.sample = Sample(target.with_path_prefix("sassample"))
-        self.transmission_spectrum = TransmissionSpectrum(target.with_path_prefix("sastransmission_spectrum"))
+        self.instrument = Instrument(target.with_path_prefix("sasinstrument|instrument"))
+        self.process = Process(target.with_path_prefix("sasprocess|process"))
+        self.sample = Sample(target.with_path_prefix("sassample|sample"))
+        self.transmission_spectrum = TransmissionSpectrum(target.with_path_prefix("sastransmission_spectrum|transmission_spectrum"))
 
         self._title = StringAccessor(target, "title")
         self._run = StringAccessor(target, "run")
-        self._definitiion = StringAccessor(target, "definition")
+        self._definition = StringAccessor(target, "definition")
 
-        self.title: str = self._title.value
-        self.run: str = self._run.value
-        self.definitiion: str = self._definitiion.value
+        self.title: str = decode_string(self._title.value)
+        self.run: str = decode_string(self._run.value)
+        self.definition: str = decode_string(self._definition.value)
 
     def summary(self):
         return (
-            f"  {self.title}, Run: {self.run}\n" + "  " + "="*len(self.title) +
+            f"  {self.title}, Run: {self.run}\n" +
+            "  " + "="*len(self.title) +
                            "=======" +
             "="*len(self.run) + "\n\n" +
             f"Definition: {self.title}\n" +

sasdata/quantities/_accessor_base.py

@@ -47,21 +47,40 @@ def get_value(self, path: str):
         current_tree_position: Group | Dataset = self._data
 
         for token in tokens:
+
+            options = token.split("|")
+
             if isinstance(current_tree_position, Group):
-                if token in current_tree_position.children:
-                    current_tree_position = current_tree_position.children[token]
-                else:
+
+                found = False
+                for option in options:
+                    if option in current_tree_position.children:
+                        current_tree_position = current_tree_position.children[option]
+                        found = True
+
+                        if self.verbose:
+                            logger.info(f"Found option: {option}")
+
+                if not found:
                     if self.verbose:
-                        logger.info(f"Failed at token {token} on group {current_tree_position.name}. Options: " +
+                        logger.info(f"Failed to find any of {options} on group {current_tree_position.name}. Options: " +
                                     ",".join([key for key in current_tree_position.children]))
                     return None
 
             elif isinstance(current_tree_position, Dataset):
-                if token in current_tree_position.attributes:
-                    current_tree_position = current_tree_position.attributes[token]
-                else:
+
+                found = False
+                for option in options:
+                    if option in current_tree_position.attributes:
+                        current_tree_position = current_tree_position.attributes[option]
+                        found = True
+
+                        if self.verbose:
+                            logger.info(f"Found option: {option}")
+
+                if not found:
                     if self.verbose:
-                        logger.info(f"Failed at token {token} on attribute {current_tree_position.name}. Options: " +
+                        logger.info(f"Failed to find any of {options} on attribute {current_tree_position.name}. Options: " +
                                     ",".join([key for key in current_tree_position.attributes]))
                     return None
 

sasdata/quantities/accessors.py

@@ -127,21 +127,40 @@ def get_value(self, path: str):
         current_tree_position: Group | Dataset = self._data
 
         for token in tokens:
+
+            options = token.split("|")
+
             if isinstance(current_tree_position, Group):
-                if token in current_tree_position.children:
-                    current_tree_position = current_tree_position.children[token]
-                else:
+
+                found = False
+                for option in options:
+                    if option in current_tree_position.children:
+                        current_tree_position = current_tree_position.children[option]
+                        found = True
+
+                        if self.verbose:
+                            logger.info(f"Found option: {option}")
+
+                if not found:
                     if self.verbose:
-                        logger.info(f"Failed at token {token} on group {current_tree_position.name}. Options: " +
+                        logger.info(f"Failed to find any of {options} on group {current_tree_position.name}. Options: " +
                                     ",".join([key for key in current_tree_position.children]))
                     return None
 
             elif isinstance(current_tree_position, Dataset):
-                if token in current_tree_position.attributes:
-                    current_tree_position = current_tree_position.attributes[token]
-                else:
+
+                found = False
+                for option in options:
+                    if option in current_tree_position.attributes:
+                        current_tree_position = current_tree_position.attributes[option]
+                        found = True
+
+                        if self.verbose:
+                            logger.info(f"Found option: {option}")
+
+                if not found:
                     if self.verbose:
-                        logger.info(f"Failed at token {token} on attribute {current_tree_position.name}. Options: " +
+                        logger.info(f"Failed to find any of {options} on attribute {current_tree_position.name}. Options: " +
                                     ",".join([key for key in current_tree_position.attributes]))
                     return None
 

sasdata/temp_hdf5_reader.py

@@ -19,7 +19,10 @@
 from sasdata.quantities.unit_parser import parse
 
 # test_file = "./example_data/1d_data/33837rear_1D_1.75_16.5_NXcanSAS_v3.h5"
-test_file = "./example_data/1d_data/33837rear_1D_1.75_16.5_NXcanSAS.h5"
+# test_file = "./example_data/1d_data/33837rear_1D_1.75_16.5_NXcanSAS.h5"
+# test_file = "./example_data/2d_data/BAM_2D.h5"
+test_file = "./example_data/2d_data/14250_2D_NoDetInfo_NXcanSAS_v3.h5"
+# test_file = "./example_data/2d_data/33837rear_2D_1.75_16.5_NXcanSAS_v3.h5"
 
 logger = logging.getLogger(__name__)
 
@@ -79,8 +82,12 @@ def connected_data(node: SASDataGroup, name_prefix="") -> list[NamedQuantity]:
                                  value=child.data,
                                  units=units)
 
-        if "uncertainty" in child.attributes:
-            uncertainty_name = child.attributes["uncertainty"]
+        # Turns out people can't be trusted to use the same keys here
+        if "uncertainty" in child.attributes or "uncertainties" in child.attributes:
+            try:
+                uncertainty_name = child.attributes["uncertainty"]
+            except:
+                uncertainty_name = child.attributes["uncertainties"]
             uncertainty_map[name] = uncertainty_name
             uncertainties.add(uncertainty_name)
 
@@ -114,12 +121,13 @@ def load_data(filename) -> list[SasData]:
 
             entry_keys = [key for key in entry.keys()]
 
-            if "sasdata" not in entry_keys:
-                logger.warning("No sasdata key")
+            if "sasdata" not in entry_keys and "data" not in entry_keys:
+                logger.warning("No sasdata or data key")
 
             for key in entry_keys:
                 component = entry[key]
-                if key.lower() == "sasdata":
+                lower_key = key.lower()
+                if lower_key == "sasdata" or lower_key == "data":
                     datum = recurse_hdf5(component)
                     # TODO: Use named identifier
                     data_contents = connected_data(datum, "FILE_ID_HERE")
@@ -143,4 +151,4 @@ def load_data(filename) -> list[SasData]:
 data = load_data(test_file)
 
 for dataset in data:
-    print(dataset.summary(include_raw=True))
+    print(dataset.summary(include_raw=False))