Added force_reload option to Dataset and InMemoryDataset [1/n]

CHANGELOG.md

@@ -7,6 +7,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 ### Added
 
+- Added `force_reload` option to `Dataset` and `InMemoryDataset` to reload datasets ([#8352](https://github.com/pyg-team/pytorch_geometric/pull/8352))
 - Added support for `torch.compile` in `MultiAggregation` ([#8345](https://github.com/pyg-team/pytorch_geometric/pull/8345))
 - Added support for `torch.compile` in `HeteroConv` ([#8344](https://github.com/pyg-team/pytorch_geometric/pull/8344))
 - Added support for weighted `sparse_cross_entropy` ([#8340](https://github.com/pyg-team/pytorch_geometric/pull/8340))

torch_geometric/data/dataset.py

@@ -44,6 +44,8 @@ class Dataset(torch.utils.data.Dataset, ABC):
             included in the final dataset. (default: :obj:`None`)
         log (bool, optional): Whether to print any console output while
             downloading and processing the dataset. (default: :obj:`True`)
+        force_reload (bool, optional): Whether to re-process the dataset.
+            (default: :obj:`False`)
     """
     @property
     def raw_file_names(self) -> Union[str, List[str], Tuple]:
@@ -84,6 +86,7 @@ def __init__(
         pre_transform: Optional[Callable] = None,
         pre_filter: Optional[Callable] = None,
         log: bool = True,
+        force_reload: bool = False,
     ):
         super().__init__()
 
@@ -96,6 +99,7 @@ def __init__(
         self.pre_filter = pre_filter
         self.log = log
         self._indices: Optional[Sequence] = None
+        self.force_reload = force_reload
 
         if self.has_download:
             self._download()
@@ -217,20 +221,20 @@ def _process(self):
         f = osp.join(self.processed_dir, 'pre_transform.pt')
         if osp.exists(f) and torch.load(f) != _repr(self.pre_transform):
             warnings.warn(
-                f"The `pre_transform` argument differs from the one used in "
-                f"the pre-processed version of this dataset. If you want to "
-                f"make use of another pre-processing technique, make sure to "
-                f"delete '{self.processed_dir}' first")
+                "The `pre_transform` argument differs from the one used in "
+                "the pre-processed version of this dataset. If you want to "
+                "make use of another pre-processing technique, pass "
+                "`force_reload=True` explicitly to reload the dataset.")
 
         f = osp.join(self.processed_dir, 'pre_filter.pt')
         if osp.exists(f) and torch.load(f) != _repr(self.pre_filter):
             warnings.warn(
                 "The `pre_filter` argument differs from the one used in "
                 "the pre-processed version of this dataset. If you want to "
-                "make use of another pre-fitering technique, make sure to "
-                "delete '{self.processed_dir}' first")
+                "make use of another pre-fitering technique, pass "
+                "`force_reload=True` explicitly to reload the dataset.")
 
-        if files_exist(self.processed_paths):  # pragma: no cover
+        if not self.force_reload and files_exist(self.processed_paths):
             return
 
         if self.log and 'pytest' not in sys.modules:

torch_geometric/data/in_memory_dataset.py

@@ -56,6 +56,8 @@ class InMemoryDataset(Dataset, ABC):
             included in the final dataset. (default: :obj:`None`)
         log (bool, optional): Whether to print any console output while
             downloading and processing the dataset. (default: :obj:`True`)
+        force_reload (bool, optional): Whether to re-process the dataset.
+            (default: :obj:`False`)
     """
     @property
     def raw_file_names(self) -> Union[str, List[str], Tuple]:
@@ -72,8 +74,10 @@ def __init__(
         pre_transform: Optional[Callable] = None,
         pre_filter: Optional[Callable] = None,
         log: bool = True,
+        force_reload: bool = False,
     ):
-        super().__init__(root, transform, pre_transform, pre_filter, log)
+        super().__init__(root, transform, pre_transform, pre_filter, log,
+                         force_reload)
         self._data = None
         self.slices = None
         self._data_list: Optional[List[BaseData]] = None

torch_geometric/datasets/actor.py

@@ -28,6 +28,8 @@ class Actor(InMemoryDataset):
             an :obj:`torch_geometric.data.Data` object and returns a
             transformed version. The data object will be transformed before
             being saved to disk. (default: :obj:`None`)
+        force_reload (bool, optional): Whether to re-process the dataset.
+            (default: :obj:`False`)
 
     **STATS:**
 
@@ -47,9 +49,15 @@ class Actor(InMemoryDataset):
 
     url = 'https://raw.githubusercontent.com/graphdml-uiuc-jlu/geom-gcn/master'
 
-    def __init__(self, root: str, transform: Optional[Callable] = None,
-                 pre_transform: Optional[Callable] = None):
-        super().__init__(root, transform, pre_transform)
+    def __init__(
+        self,
+        root: str,
+        transform: Optional[Callable] = None,
+        pre_transform: Optional[Callable] = None,
+        force_reload: bool = False,
+    ):
+        super().__init__(root, transform, pre_transform,
+                         force_reload=force_reload)
         self.load(self.processed_paths[0])
 
     @property