From 1d78c33a5f5bed1a3698d493dbdff7893e3231a7 Mon Sep 17 00:00:00 2001
From: Ben Murray <ben.murray@gmail.com>
Date: Fri, 13 Oct 2023 17:03:26 +0100
Subject: [PATCH] Initial commit of a solution for issue #7130

Signed-off-by: Ben Murray <ben.murray@gmail.com>
---
 monai/data/dataset.py         | 10 +++++-----
 monai/transforms/compose.py   | 21 +++++++++++++++++++++
 monai/transforms/transform.py | 12 ++++++++++++
 3 files changed, 38 insertions(+), 5 deletions(-)

diff --git a/monai/data/dataset.py b/monai/data/dataset.py
index 4f2061426e..629a30257c 100644
--- a/monai/data/dataset.py
+++ b/monai/data/dataset.py
@@ -288,7 +288,7 @@ def set_transform_hash(self, hash_xform_func: Callable[..., bytes]):
         inherit from MONAI's `Transform` class."""
         hashable_transforms = []
         for _tr in self.transform.flatten().transforms:
-            if isinstance(_tr, RandomizableTrait) or not isinstance(_tr, Transform):
+            if (isinstance(_tr, RandomizableTrait) and _tr.is_random() is True) or not isinstance(_tr, Transform):
                 break
             hashable_transforms.append(_tr)
         # Try to hash. Fall back to a hash of their names
@@ -327,7 +327,7 @@ def _pre_transform(self, item_transformed):
             raise ValueError("transform must be an instance of monai.transforms.Compose.")
 
         first_random = self.transform.get_index_of_first(
-            lambda t: isinstance(t, RandomizableTrait) or not isinstance(t, Transform)
+            lambda t: (isinstance(t, RandomizableTrait) and t.is_random() is True) or not isinstance(t, Transform)
         )
         item_transformed = self.transform(item_transformed, end=first_random, threading=True)
 
@@ -350,7 +350,7 @@ def _post_transform(self, item_transformed):
             raise ValueError("transform must be an instance of monai.transforms.Compose.")
 
         first_random = self.transform.get_index_of_first(
-            lambda t: isinstance(t, RandomizableTrait) or not isinstance(t, Transform)
+            lambda t: isinstance(t, RandomizableTrait) and t.is_random() is True or not isinstance(t, Transform)
         )
         if first_random is not None:
             item_transformed = self.transform(item_transformed, start=first_random)
@@ -887,7 +887,7 @@ def _load_cache_item(self, idx: int):
         item = self.data[idx]
 
         first_random = self.transform.get_index_of_first(
-            lambda t: isinstance(t, RandomizableTrait) or not isinstance(t, Transform)
+            lambda t: (isinstance(t, RandomizableTrait) and t.is_random() is True) or not isinstance(t, Transform)
         )
         item = self.transform(item, end=first_random, threading=True)
 
@@ -921,7 +921,7 @@ def _transform(self, index: int):
             raise ValueError("transform must be an instance of monai.transforms.Compose.")
 
         first_random = self.transform.get_index_of_first(
-            lambda t: isinstance(t, RandomizableTrait) or not isinstance(t, Transform)
+            lambda t: (isinstance(t, RandomizableTrait) and t.is_random() is True) or not isinstance(t, Transform)
         )
         if first_random is not None:
             data = deepcopy(data) if self.copy_cache is True else data
diff --git a/monai/transforms/compose.py b/monai/transforms/compose.py
index 1614913f5e..816d94537d 100644
--- a/monai/transforms/compose.py
+++ b/monai/transforms/compose.py
@@ -334,6 +334,18 @@ def __call__(self, input_, start=0, end=None, threading=False, lazy: bool | None
 
         return result
 
+    def is_random(self):
+        def _recursive_check(tx):
+            if isinstance(Randomizable) and tx.is_random():
+                return True
+            return False
+        
+        for t in self.transforms:
+            if _recursive_check(t) is True:
+                return True
+
+        return False
+
     def inverse(self, data):
         self._raise_if_not_invertible(data)
 
@@ -481,6 +493,9 @@ def __call__(self, data, start=0, end=None, threading=False, lazy: str | bool |
                     self.push_transform(data[key], extra_info={"index": index})
         return data
 
+    def is_random(self):
+        return True
+
     def inverse(self, data):
         if len(self.transforms) == 0:
             return data
@@ -575,6 +590,9 @@ def __call__(self, input_, start=0, end=None, threading=False, lazy: bool | None
                     self.push_transform(input_[key], extra_info={"applied_order": applied_order})
         return input_
 
+    def is_random(self):
+        return True
+
     def inverse(self, data):
         if len(self.transforms) == 0:
             return data
@@ -740,6 +758,9 @@ def __call__(self, data, start=0, end=None, threading=False, lazy: bool | None =
 
         return data
 
+    def is_random(self):
+        return True
+
     # From RandomOrder
     def inverse(self, data):
         if len(self.transforms) == 0:
diff --git a/monai/transforms/transform.py b/monai/transforms/transform.py
index e35335ba0e..a77b6098b7 100644
--- a/monai/transforms/transform.py
+++ b/monai/transforms/transform.py
@@ -230,6 +230,18 @@ def randomize(self, data: Any) -> None:
             NotImplementedError: When the subclass does not override this method.
         """
         raise NotImplementedError(f"Subclass {self.__class__.__name__} must implement this method.")
+    
+    def is_random(self):
+        """
+        This method indicates whether this particular instance of a Randomizable is operating randomly or not.
+        This allows objects that have the capacity to act in a random fashion to indicate whether they are acting
+        randomly or not, given their current state.
+        This method should be overridden by objects that have this capacity. ``Compose`` is an example of this.
+
+        Returns:
+            True if the object is acting randomly, False otherwise
+        """
+        return True
 
 
 class Transform(ABC):