Add observed grouper/grouping as cached method

Author: Kei

pandas/core/groupby/generic.py

@@ -2040,7 +2040,6 @@ def _gotitem(self, key, ndim: int, subset=None):
                 group_keys=self.group_keys,
                 observed=self.observed,
                 dropna=self.dropna,
-                observed_grouper=self.observed_grouper,
             )
         elif ndim == 1:
             if subset is None:
@@ -2058,7 +2057,6 @@ def _gotitem(self, key, ndim: int, subset=None):
                 group_keys=self.group_keys,
                 observed=self.observed,
                 dropna=self.dropna,
-                observed_grouper=self.observed_grouper,
             )
 
         raise AssertionError("invalid ndim for _gotitem")

pandas/core/groupby/groupby.py

@@ -588,7 +588,6 @@ class BaseGroupBy(PandasObject, SelectionMixin[NDFrameT], GroupByIndexingMixin):
         "obj",
         "observed",
         "sort",
-        "observed_grouper",
     }
 
     _grouper: ops.BaseGrouper
@@ -1107,7 +1106,6 @@ def __init__(
         group_keys: bool = True,
         observed: bool = False,
         dropna: bool = True,
-        observed_grouper: ops.BaseGrouper | None = None,
     ) -> None:
         self._selection = selection
 
@@ -1120,18 +1118,6 @@ def __init__(
         self.group_keys = group_keys
         self.dropna = dropna
 
-        if not observed and grouper is None:
-            observed_grouper, _, _ = get_grouper(
-                obj,
-                self.keys,
-                level=self.level,
-                sort=self.sort,
-                observed=True,
-                dropna=self.dropna,
-            )
-
-        self.observed_grouper = observed_grouper
-
         if grouper is None:
             grouper, exclusions, obj = get_grouper(
                 obj,
@@ -1899,7 +1885,7 @@ def _transform(self, func, *args, engine=None, engine_kwargs=None, **kwargs):
 
             with (
                 com.temp_setattr(self, "observed", True),
-                com.temp_setattr(self, "_grouper", self.observed_grouper),
+                com.temp_setattr(self, "_grouper", self._grouper.observed_grouper),
             ):
                 return self._reduction_kernel_transform(
                     func, *args, engine=engine, engine_kwargs=engine_kwargs, **kwargs

pandas/core/groupby/grouper.py

@@ -668,6 +668,24 @@ def groups(self) -> dict[Hashable, Index]:
         cats = Categorical.from_codes(codes, uniques, validate=False)
         return self._index.groupby(cats)
 
+    @cache_readonly
+    def observed_grouping(self) -> Grouper:
+        if self._observed:
+            return self
+
+        grouping = Grouping(
+            self._index,
+            self._orig_grouper,
+            obj=self.obj,
+            level=self.level,
+            sort=self._sort,
+            observed=True,
+            in_axis=self.in_axis,
+            dropna=self._dropna,
+            uniques=self._uniques,
+        )
+        return grouping
+
 
 def get_grouper(
     obj: NDFrameT,

pandas/core/groupby/ops.py

@@ -823,6 +823,15 @@ def result_index_and_ids(self) -> tuple[Index, npt.NDArray[np.intp]]:
 
         return result_index, ids
 
+    @cache_readonly
+    def observed_grouper(self) -> BaseGrouper:
+        if all(ping._observed for ping in self.groupings):
+            return self
+
+        groupings = [ping.observed_grouping for ping in self.groupings]
+        grouper = BaseGrouper(self.axis, groupings, sort=self._sort, dropna=self.dropna)
+        return grouper
+
     def _ob_index_and_ids(
         self,
         levels: list[Index],
@@ -1154,6 +1163,10 @@ def groupings(self) -> list[grouper.Grouping]:
         )
         return [ping]
 
+    @cache_readonly
+    def observed_grouper(self) -> BinGrouper:
+        return self
+
 
 def _is_indexed_like(obj, axes) -> bool:
     if isinstance(obj, Series):