implement DatasetDict (#296)

* add utils.hydra.resolve_target()

* implement DatasetDict

* add documentation to methods

* rename from_hf_dataset to from_hf and allow HF Dataset or HF IterableDataset as input

* improve documentation

* fix tests

src/pytorch_ie/data/__init__.py

@@ -1,13 +1,8 @@
-from typing import Dict, Union
-
-from datasets import Split
-
 from .builder import GeneratorBasedBuilder
 from .dataset import Dataset, IterableDataset
+from .dataset_dict import DatasetDict
 from .dataset_formatter import DocumentFormatter
 
-DatasetDict = Dict[Union[str, Split], Dataset]
-
 __all__ = [
     "GeneratorBasedBuilder",
     "Dataset",

src/pytorch_ie/data/common.py

@@ -0,0 +1,40 @@
+from abc import ABC, abstractmethod
+from typing import Optional, Union
+
+from .dataset import Dataset, IterableDataset
+
+
+class EnterDatasetMixin(ABC):
+    """Mixin for processors that enter a dataset context."""
+
+    @abstractmethod
+    def enter_dataset(
+        self, dataset: Union[Dataset, IterableDataset], name: Optional[str] = None
+    ) -> None:
+        """Enter dataset context."""
+
+
+class ExitDatasetMixin(ABC):
+    """Mixin for processors that exit a dataset context."""
+
+    @abstractmethod
+    def exit_dataset(
+        self, dataset: Union[Dataset, IterableDataset], name: Optional[str] = None
+    ) -> None:
+        """Exit dataset context."""
+
+
+class EnterDatasetDictMixin(ABC):
+    """Mixin for processors that enter a dataset dict context."""
+
+    @abstractmethod
+    def enter_dataset_dict(self, dataset_dict) -> None:
+        """Enter dataset dict context."""
+
+
+class ExitDatasetDictMixin(ABC):
+    """Mixin for processors that exit a dataset dict context."""
+
+    @abstractmethod
+    def exit_dataset_dict(self, dataset_dict) -> None:
+        """Exit dataset dict context."""