fix: merging

Author: agoose77

src/awkward/_do/content.py

@@ -9,14 +9,16 @@
 import awkward as ak
 from awkward._backends.backend import Backend
 from awkward._nplikes.numpy_like import NumpyMetadata
-from awkward._typing import Any, AxisMaybeNone, Literal
-from awkward.contents.content import ActionType, Content
+from awkward._typing import TYPE_CHECKING, Any, AxisMaybeNone, Literal
 from awkward.errors import AxisError
 from awkward.forms import form
-from awkward.record import Record
 
 np = NumpyMetadata.instance()
 
+if TYPE_CHECKING:
+    from awkward.contents.content import ActionType, Content
+    from awkward.record import Record
+
 
 def recursively_apply(
     layout: Content | Record,
@@ -32,6 +34,9 @@ def recursively_apply(
     function_name: str | None = None,
     regular_to_jagged=False,
 ) -> Content | Record | None:
+    from awkward.contents.content import Content
+    from awkward.record import Record
+
     if isinstance(layout, Content):
         return layout._recursively_apply(
             action,
@@ -201,6 +206,8 @@ def remove_structure(
     allow_records: bool = False,
     list_to_regular: bool = False,
 ):
+    from awkward.record import Record
+
     if isinstance(layout, Record):
         return remove_structure(
             layout._array[layout._at : layout._at + 1],

src/awkward/_do/meta.py

@@ -27,15 +27,11 @@
 
 def is_option(
     meta: Meta
-) -> TypeGuard[
-    IndexedOptionMeta | BitMaskedMeta | ByteMaskedMeta | UnmaskedMeta
-]:
+) -> TypeGuard[IndexedOptionMeta | BitMaskedMeta | ByteMaskedMeta | UnmaskedMeta]:
     return meta.is_option
 
 
-def is_list(
-    meta: Meta
-) -> TypeGuard[RegularMeta | ListOffsetMeta | ListMeta]:
+def is_list(meta: Meta) -> TypeGuard[RegularMeta | ListOffsetMeta | ListMeta]:
     return meta.is_list
 
 
@@ -59,19 +55,13 @@ def is_indexed(meta: Meta) -> TypeGuard[IndexedOptionMeta, IndexedMeta]:
     return meta.is_indexed
 
 
-class ImplementsTuple(RecordMeta):  # Intersection
-    _fields: None
-
-
-def is_record_tuple(meta: Meta) -> TypeGuard[ImplementsTuple]:
+# FIXME: narrow this to have `is_tuple` be a const True
+def is_record_tuple(meta: Meta) -> TypeGuard[RecordMeta]:
     return meta.is_record and meta.is_tuple
 
 
-class ImplementsRecord(RecordMeta):
-    _fields: list[str]
-
-
-def is_record_record(meta: Meta) -> TypeGuard[ImplementsRecord]:
+# FIXME: narrow this to have `is_tuple` be a const False
+def is_record_record(meta: Meta) -> TypeGuard[RecordMeta]:
     return meta.is_record and not meta.is_tuple
 
 

src/awkward/_meta/numpymeta.py

@@ -4,10 +4,12 @@
 
 from awkward._do.meta import is_indexed, is_numpy, is_option
 from awkward._meta.meta import Meta
+from awkward._nplikes.numpy_like import NumpyMetadata
 from awkward._nplikes.shape import ShapeItem
 from awkward._parameters import type_parameters_equal
-from awkward._typing import TYPE_CHECKING, JSONSerializable
+from awkward._typing import TYPE_CHECKING, DType, JSONSerializable
 
+np = NumpyMetadata.instance()
 if TYPE_CHECKING:
     from awkward._meta.regularmeta import RegularMeta
 
@@ -17,6 +19,10 @@ class NumpyMeta(Meta):
     is_leaf = True
     inner_shape: tuple[ShapeItem, ...]
 
+    @property
+    def dtype(self) -> DType:
+        raise NotImplementedError
+
     def purelist_parameters(self, *keys: str) -> JSONSerializable:
         if self._parameters is not None:
             for key in keys:
@@ -104,9 +110,9 @@ def _mergeable_next(self, other: Meta, mergebool: bool) -> bool:
 
             # Default merging (can we cast one to the other)
             else:
-                return self.backend.nplike.can_cast(
-                    self.dtype, other.dtype
-                ) or self.backend.nplike.can_cast(other.dtype, self.dtype)
+                return np.can_cast(
+                    self.dtype, other.dtype, casting="same_kind"
+                ) or np.can_cast(other.dtype, self.dtype, casting="same_kind")
 
         else:
             return False

src/awkward/_meta/recordmeta.py

@@ -170,10 +170,10 @@ def _mergeable_next(self, other: Meta, mergebool: bool) -> bool:
                     return False
 
             elif is_record_record(self) and is_record_record(other):
-                if set(self._fields) != set(other._fields):
+                if set(self._fields) != set(other._fields):  # type: ignore[arg-type]
                     return False
 
-                for i, field in enumerate(self._fields):
+                for i, field in enumerate(self._fields):  # type: ignore[arg-type]
                     x = self._contents[i]
                     y = other.contents[other.field_to_index(field)]
                     if not x._mergeable_next(y, mergebool):

src/awkward/_nplikes/array_module.py

@@ -653,11 +653,6 @@ def astype(
         assert not isinstance(x, PlaceholderArray)
         return x.astype(dtype, copy=copy)  # type: ignore[attr-defined]
 
-    def can_cast(
-        self, from_: DTypeLike | ArrayLikeT, to: DTypeLike | ArrayLikeT
-    ) -> bool:
-        return self._module.can_cast(from_, to, casting="same_kind")
-
     @classmethod
     def is_own_array(cls, obj) -> bool:
         return cls.is_own_array_type(type(obj))

src/awkward/_nplikes/numpy_like.py

@@ -83,7 +83,8 @@ class NumpyMetadata(PublicSingleton):
     datetime_data = staticmethod(numpy.datetime_data)
     issubdtype = staticmethod(numpy.issubdtype)
 
-    AxisError = numpy.AxisError
+    AxisError = staticmethod(numpy.AxisError)
+    can_cast = staticmethod(numpy.can_cast)
 
 
 if hasattr(numpy, "float16"):
@@ -537,10 +538,6 @@ def astype(
     ) -> ArrayLikeT:
         ...
 
-    @abstractmethod
-    def can_cast(self, from_: DType | ArrayLikeT, to: DType | ArrayLikeT) -> bool:
-        ...
-
     @abstractmethod
     def is_c_contiguous(self, x: ArrayLikeT | PlaceholderArray) -> bool:
         ...

src/awkward/_nplikes/typetracer.py

@@ -1512,11 +1512,6 @@ def astype(
         x.touch_data()
         return TypeTracerArray._new(np.dtype(dtype), x.shape)
 
-    def can_cast(
-        self, from_: DTypeLike | TypeTracerArray, to: DTypeLike | TypeTracerArray
-    ) -> bool:
-        return numpy.can_cast(from_, to, casting="same_kind")
-
     @classmethod
     def is_own_array_type(cls, type_: type) -> bool:
         return issubclass(type_, TypeTracerArray)

src/awkward/forms/numpyform.py

@@ -273,3 +273,9 @@ def _expected_from_buffers(
 
     def _to_regular_primitive(self) -> RegularForm | NumpyForm:
         return self.to_RegularForm()
+
+    @property
+    def dtype(self) -> DType:
+        from awkward.types.numpytype import primitive_to_dtype
+
+        return primitive_to_dtype(self.primitive)

src/awkward/operations/ak_strings_astype.py

@@ -4,7 +4,7 @@
 
 import awkward as ak
 from awkward._dispatch import high_level_function
-from awkward._do.content import recursively_apply
+from awkward._do.content import pad_none, recursively_apply
 from awkward._layout import HighLevelContext
 from awkward._nplikes.numpy import Numpy
 from awkward._nplikes.numpy_like import NumpyMetadata
@@ -68,7 +68,7 @@ def action(layout, **kwargs):
                 layout, highlevel=False, behavior=behavior
             )
             max_length = ak.operations.max(ak.operations.num(layout, behavior=behavior))
-            regulararray = ak._do.pad_none(layout, max_length, 1)
+            regulararray = pad_none(layout, max_length, 1)
             maskedarray = ak.operations.to_numpy(regulararray, allow_missing=True)
             npstrings = maskedarray.data
             if maskedarray.mask is not False:

src/awkward/operations/ak_values_astype.py

@@ -4,6 +4,7 @@
 
 import awkward as ak
 from awkward._dispatch import high_level_function
+from awkward._do.content import numbers_to_type
 from awkward._layout import HighLevelContext
 from awkward._nplikes.numpy_like import NumpyMetadata
 
@@ -73,5 +74,5 @@ def _impl(array, to, including_unknown, highlevel, behavior, attrs):
     with HighLevelContext(behavior=behavior, attrs=attrs) as ctx:
         layout = ctx.unwrap(array, allow_record=False, primitive_policy="error")
     to_str = ak.types.numpytype.dtype_to_primitive(np.dtype(to))
-    out = ak._do.numbers_to_type(layout, to_str, including_unknown)
+    out = numbers_to_type(layout, to_str, including_unknown)
     return ctx.wrap(out, highlevel=highlevel)