C++ refactoring: nbytes

src/awkward/_v2/contents/bitmaskedarray.py

@@ -474,6 +474,12 @@ def _validityerror(self, path):
         else:
             return self._content.validityerror(path + ".content")
 
+    def _nbytes_part(self, largest):
+        self.mask._nbytes_part(largest)
+        self.content._nbytes_part(largest)
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def _rpad(self, target, axis, depth, clip):
         return self.toByteMaskedArray()._rpad(target, axis, depth, clip)
 

src/awkward/_v2/contents/bytemaskedarray.py

@@ -704,6 +704,12 @@ def _validityerror(self, path):
         else:
             return self._content.validityerror(path + ".content")
 
+    def _nbytes_part(self, largest):
+        self.mask._nbytes_part(largest)
+        self.content._nbytes_part(largest)
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def bytemask(self):
         if not self._valid_when:
             return self._mask

src/awkward/_v2/contents/content.py

@@ -1013,6 +1013,12 @@ def validityerror(self, path="layout"):
             return paramcheck
         return self._validityerror(path)
 
+    @property
+    def nbytes(self):
+        largest = {0: 0}
+        self._nbytes_part(largest)
+        return sum(largest.values())
+
     def purelist_parameter(self, key):
         return self.Form.purelist_parameter(self, key)
 

src/awkward/_v2/contents/emptyarray.py

@@ -213,6 +213,10 @@ def _reduce_next(
     def _validityerror(self, path):
         return ""
 
+    def _nbytes_part(self, largest):
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def _rpad(self, target, axis, depth, clip):
         posaxis = self.axis_wrap_if_negative(axis)
         if posaxis != depth:

src/awkward/_v2/contents/indexedarray.py

@@ -927,6 +927,12 @@ def _validityerror(self, path):
         else:
             return self._content.validityerror(path + ".content")
 
+    def _nbytes_part(self, largest):
+        self.index._nbytes_part(largest)
+        self.content._nbytes_part(largest)
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def _rpad(self, target, axis, depth, clip):
         posaxis = self.axis_wrap_if_negative(axis)
         if posaxis == depth:

src/awkward/_v2/contents/indexedoptionarray.py

@@ -1376,6 +1376,12 @@ def _validityerror(self, path):
         else:
             return self._content.validityerror(path + ".content")
 
+    def _nbytes_part(self, largest):
+        self.index._nbytes_part(largest)
+        self.content._nbytes_part(largest)
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def bytemask(self):
         out = ak._v2.index.Index8.empty(len(self.index), self.nplike)
         self._handle_error(

src/awkward/_v2/contents/listarray.py

@@ -985,6 +985,13 @@ def _validityerror(self, path):
             else:
                 return self._content.validityerror(path + ".content")
 
+    def _nbytes_part(self, largest):
+        self.starts._nbytes_part(largest)
+        self.stops._nbytes_part(largest)
+        self.content._nbytes_part(largest)
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def _rpad(self, target, axis, depth, clip):
         if not clip:
             posaxis = self.axis_wrap_if_negative(axis)

src/awkward/_v2/contents/listoffsetarray.py

@@ -1556,6 +1556,12 @@ def _validityerror(self, path):
             else:
                 return self._content.validityerror(path + ".content")
 
+    def _nbytes_part(self, largest):
+        self.offsets._nbytes_part(largest)
+        self.content._nbytes_part(largest)
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def _rpad(self, target, axis, depth, clip):
         posaxis = self.axis_wrap_if_negative(axis)
         if posaxis == depth:

src/awkward/_v2/contents/numpyarray.py

@@ -1003,6 +1003,13 @@ def _rpad(self, target, axis, depth, clip):
         else:
             return self.rpad_axis0(target, clip=True)
 
+    def _nbytes_part(self, largest):
+        it = id(self.ptr)
+        if it not in largest or largest[it] < self.data.nbytes:
+            largest[it] = self.data.nbytes
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def _to_arrow(self, pyarrow, mask_node, validbytes, length, options):
         if self._data.ndim != 1:
             return self.toRegularArray()._to_arrow(

src/awkward/_v2/contents/recordarray.py

@@ -651,6 +651,12 @@ def _validityerror(self, path):
                 return sub
         return ""
 
+    def _nbytes_part(self, largest):
+        for content in self.contents:
+            content._nbytes_part(largest)
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def _rpad(self, target, axis, depth, clip):
         posaxis = self.axis_wrap_if_negative(axis)
         if posaxis == depth:

src/awkward/_v2/contents/regulararray.py

@@ -854,6 +854,11 @@ def _validityerror(self, path):
         else:
             return self._content.validityerror(path + ".content")
 
+    def _nbytes_part(self, largest):
+        self.content._nbytes_part(largest)
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def _rpad(self, target, axis, depth, clip):
         posaxis = self.axis_wrap_if_negative(axis)
         if posaxis == depth:

src/awkward/_v2/contents/unionarray.py

@@ -810,6 +810,12 @@ def _validityerror(self, path):
                     return sub
             return ""
 
+    def _nbytes_part(self, largest):
+        for content in self.contents:
+            content._nbytes_part(largest)
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def _rpad(self, target, axis, depth, clip):
         posaxis = self.axis_wrap_if_negative(axis)
         if posaxis == depth:

src/awkward/_v2/contents/unmaskedarray.py

@@ -391,6 +391,12 @@ def _validityerror(self, path):
         else:
             return self._content.validityerror(path + ".content")
 
+    def _nbytes_part(self, largest):
+        self.content._nbytes_part(largest)
+        # FIXME: identifier is not counted in C++ code?
+        if self.identifier is not None:
+            self.identifier._nbytes_part(largest)
+
     def _rpad(self, target, axis, depth, clip):
         posaxis = self.axis_wrap_if_negative(axis)
         if posaxis == depth:

src/awkward/_v2/highlevel.py

@@ -455,25 +455,25 @@ def to_numpy(self, allow_missing=True):
         """
         return ak._v2.operations.convert.to_numpy(self, allow_missing=allow_missing)
 
-    #     @property
-    #     def nbytes(self):
-    #         """
-    #         The total number of bytes in all the #ak.layout.Index,
-    #         #ak.layout.Identities, and #ak.layout.NumpyArray buffers in this
-    #         array tree.
+    @property
+    def nbytes(self):
+        """
+        The total number of bytes in all the #ak.layout.Index,
+        #ak.layout.Identities, and #ak.layout.NumpyArray buffers in this
+        array tree.
 
-    #         Note: this calculation takes overlapping buffers into account, to the
-    #         extent that overlaps are not double-counted, but overlaps are currently
-    #         assumed to be complete subsets of one another, and so it is
-    #         theoretically possible (though unlikely) that this number is an
-    #         underestimate of the true usage.
+        Note: this calculation takes overlapping buffers into account, to the
+        extent that overlaps are not double-counted, but overlaps are currently
+        assumed to be complete subsets of one another, and so it is
+        theoretically possible (though unlikely) that this number is an
+        underestimate of the true usage.
 
-    #         It also does not count buffers that must be kept in memory because
-    #         of ownership, but are not directly used in the array. Nor does it count
-    #         the (small) C++ nodes or Python objects that reference the (large)
-    #         array buffers.
-    #         """
-    #         return self._layout.nbytes
+        It also does not count buffers that must be kept in memory because
+        of ownership, but are not directly used in the array. Nor does it count
+        the (small) C++ nodes or Python objects that reference the (large)
+        array buffers.
+        """
+        return self._layout.nbytes
 
     @property
     def ndim(self):

src/awkward/_v2/identifier.py

@@ -115,3 +115,8 @@ def referentially_equal(self, other):
             and self._data.strides == other._data.strides
             and self._data.dtype == other._data.dtype
         )
+
+    def _nbytes_part(self, largest):
+        it = id(self.ref)
+        if it not in largest or largest[it] < self.data.nbytes:
+            largest[it] = self.data.nbytes

src/awkward/_v2/index.py

@@ -150,6 +150,11 @@ def to64(self):
     def __copy__(self):
         return Index(self._data.copy())
 
+    def _nbytes_part(self, largest):
+        it = id(self.data)
+        if it not in largest or largest[it] < self.data.nbytes:
+            largest[it] = self.data.nbytes
+
 
 class Index8(Index):
     _expected_dtype = np.dtype(np.int8)