datamaestro 1.4.0__tar.gz → 1.4.2__tar.gz

{datamaestro-1.4.0 → datamaestro-1.4.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datamaestro
-Version: 1.4.0
+Version: 1.4.2
 Summary: "Dataset management command line and API"
 Home-page: https://github.com/experimaestro/datamaestro
 Author: Benjamin Piwowarski

{datamaestro-1.4.0 → datamaestro-1.4.2}/docs/source/api/download.rst

@@ -67,5 +67,5 @@ File hashes can be checked with the following checker
 Custom
 ======
 
-.. autofunction:: datamaestro.download.custom.Downloader
-.. autoclass:: datamaestro.download.custom.custom_download
+.. autoclass:: datamaestro.download.custom.Downloader
+.. autofunction:: datamaestro.download.custom.custom_download

{datamaestro-1.4.0 → datamaestro-1.4.2}/src/datamaestro/definitions.py

@@ -102,7 +102,7 @@ class DataDefinition(AbstractData):
             if components[0] == "datamaestro":
                 longest_ix = 0
 
-        return repository, components[(longest_ix + 1) :]
+        return repository, [s.lower() for s in components[(longest_ix + 1) :]]
 
     def ancestors(self):
         ancestors = []
@@ -293,7 +293,11 @@ class DatasetWrapper(AbstractDataset):
 
         # Builds the ID:
         # Removes module_name.config prefix
-        if annotation.id is None or annotation.id == "":
+        if (
+            (annotation.id is None)
+            or (annotation.id == "")
+            or ("." not in annotation.id)
+        ):
             # Computes an ID
             assert (
                 # id is empty string = use the module id
@@ -303,7 +307,15 @@ class DatasetWrapper(AbstractDataset):
                 "A @dataset without `id` should be in the "
                 f".config module (not {t.__module__})"
             )
-            path = ".".join(components[1:-1])
+
+            if annotation.id is None:
+                # There is nothing, use the full path
+                path = ".".join(components[1:])
+            else:
+                # Replace
+                path = ".".join(components[1:-1])
+                if annotation.id != "":
+                    path = f"{path}.{annotation.id}"
 
             self.id = path
         else:
@@ -369,33 +381,36 @@ class DatasetWrapper(AbstractDataset):
         if self.base is self.t:
             self.config = self.base.__create_dataset__(self)
 
-        # Construct the object
-        resources = {key: value.prepare() for key, value in self.resources.items()}
-
-        result = self.t(**resources)
+        else:
+            # Construct the object
+            resources = {key: value.prepare() for key, value in self.resources.items()}
 
-        # Download resources
-        logging.debug("Building with data type %s and dataset %s", self.base, self.t)
-        for hook in self.hooks["pre-use"]:
-            hook(self)
+            result = self.t(**resources)
 
-        if result is None:
-            name = self.t.__name__
-            filename = inspect.getfile(self.t)
-            raise Exception(
-                f"The dataset method {name} defined in "
-                f"{filename} returned a null object"
-            )
-
-        if isinstance(result, dict):
-            self.config = self.base(**result)
-        elif isinstance(result, self.base):
-            self.config = result
-        else:
-            raise RuntimeError(
-                f"The dataset method {name} defined in "
-                f"{filename} returned an object of type {type(dict)}"
+            # Download resources
+            logging.debug(
+                "Building with data type %s and dataset %s", self.base, self.t
             )
+            for hook in self.hooks["pre-use"]:
+                hook(self)
+
+            if result is None:
+                name = self.t.__name__
+                filename = inspect.getfile(self.t)
+                raise Exception(
+                    f"The dataset method {name} defined in "
+                    f"{filename} returned a null object"
+                )
+
+            if isinstance(result, dict):
+                self.config = self.base(**result)
+            elif isinstance(result, self.base):
+                self.config = result
+            else:
+                raise RuntimeError(
+                    f"The dataset method {name} defined in "
+                    f"{filename} returned an object of type {type(dict)}"
+                )
 
         # Setup ourself
         self.config.__datamaestro_dataset__ = self
@@ -557,13 +572,15 @@ class dataset:
             timestamp {bool} -- If the dataset evolves, specify its timestamp
             (default: None)
 
-            id {[type]} -- [description] (default: {None})
+            id {[type]} -- [description] (default: {None}) Gives the full ID of
+            the dataset if it contains a ., or just the last component otherwise
 
             url {[type]} -- [description] (default: {None})
 
             size {str} -- The size (should be a parsable format)
 
             doi {str} -- The DOI of the corresponding paper
+
         """
         if hasattr(base, "__datamaestro__") and isinstance(
             base.__datamaestro__, metadataset

{datamaestro-1.4.0 → datamaestro-1.4.2}/src/datamaestro/record.py

@@ -160,11 +160,20 @@ class Record:
     def __getitem__(self, key: Type[T]) -> T:
         """Get an item given its type"""
         base = key.__get_base__()
-        entry = self.items[base]
+        try:
+            entry = self.items[base]
+        except KeyError:
+            raise KeyError(
+                f"""No entry with type {key}: """
+                f"""{",".join(str(s) for s in self.items.keys())}"""
+            )
 
         # Check if this matches the expected class
         if not isinstance(entry, key):
-            raise KeyError(f"No entry with type {key}")
+            raise KeyError(
+                f"""No entry with type {key}: """
+                f"""{",".join(str(s) for s in self.items.keys())}"""
+            )
         return entry
 
     def update(self, *items: T, target: RecordType = None) -> "Record":

{datamaestro-1.4.0 → datamaestro-1.4.2}/src/datamaestro/version.py

@@ -17,5 +17,5 @@ __version__: str
 __version_tuple__: VERSION_TUPLE
 version_tuple: VERSION_TUPLE
 
-__version__ = version = '1.4.0'
-__version_tuple__ = version_tuple = (1, 4, 0)
+__version__ = version = '1.4.2'
+__version_tuple__ = version_tuple = (1, 4, 2)

{datamaestro-1.4.0 → datamaestro-1.4.2}/src/datamaestro.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datamaestro
-Version: 1.4.0
+Version: 1.4.2
 Summary: "Dataset management command line and API"
 Home-page: https://github.com/experimaestro/datamaestro
 Author: Benjamin Piwowarski