Pre-commit autoupdate (#148)

.pre-commit-config.yaml

@@ -1,22 +1,22 @@
 repos:
   - repo: https://github.com/Quantco/pre-commit-mirrors-ruff
-    rev: 0.2.2
+    rev: 0.3.4
     hooks:
       - id: ruff-conda
       - id: ruff-format-conda
   - repo: https://github.com/Quantco/pre-commit-mirrors-mypy
-    rev: "1.8.0"
+    rev: "1.9.0"
     hooks:
       - id: mypy-conda
         additional_dependencies: [-c, conda-forge, types-setuptools]
   - repo: https://github.com/Quantco/pre-commit-mirrors-pyupgrade
-    rev: 3.15.0
+    rev: 3.15.2
     hooks:
       - id: pyupgrade-conda
         args:
           - --py38
   - repo: https://github.com/Quantco/pre-commit-mirrors-prettier
-    rev: 3.2.4
+    rev: 3.2.5
     hooks:
       - id: prettier-conda
         files: "\\.md$"

src/spox/_schemas.py

@@ -1,4 +1,5 @@
 """Exposes information related to reference ONNX operator schemas, used by StandardOpNode."""
+
 import itertools
 from typing import (
     Callable,
@@ -16,11 +17,9 @@
 
 
 class _Comparable(Protocol):
-    def __lt__(self, other):
-        ...
+    def __lt__(self, other): ...
 
-    def __gt__(self, other):
-        ...
+    def __gt__(self, other): ...
 
 
 S = TypeVar("S")

src/spox/_scope.py

@@ -59,12 +59,10 @@ def __contains__(self, item: Union[str, H]) -> bool:
         )
 
     @overload
-    def __getitem__(self, item: H) -> str:
-        ...
+    def __getitem__(self, item: H) -> str: ...
 
     @overload
-    def __getitem__(self, item: str) -> H:
-        ...
+    def __getitem__(self, item: str) -> H: ...
 
     def __getitem__(self, item: Union[str, H]):
         """Access the name of an object or an object with a given name in this (or outer) namespace."""
@@ -76,12 +74,10 @@ def __getitem__(self, item: Union[str, H]):
             return self.name_of[item]
 
     @overload
-    def __setitem__(self, key: str, value: H):
-        ...
+    def __setitem__(self, key: str, value: H): ...
 
     @overload
-    def __setitem__(self, key: H, value: str):
-        ...
+    def __setitem__(self, key: H, value: str): ...
 
     def __setitem__(self, _key, _value):
         """Set the name of an object in exactly this namespace. Both ``[name] = obj`` and ``[obj] = name`` work."""

src/spox/_standard.py

@@ -1,4 +1,5 @@
 """Module implementing a base for standard ONNX operators, which use the functionality of ONNX node-level inference."""
+
 from typing import TYPE_CHECKING, Callable, Dict, Tuple
 
 import numpy

src/spox/opset/ai/onnx/ml/v3.py

@@ -1836,12 +1836,14 @@ def tree_ensemble_regressor(
         'SUM,' 'MIN,' 'MAX.'
     base_values
         Attribute.
-        Base values for classification, added to final class score; the size
-        must be the same as the classes or can be left unassigned (assumed 0)
+        Base values for regression, added to final prediction after applying
+        aggregate_function; the size must be the same as the classes or can be
+        left unassigned (assumed 0)
     base_values_as_tensor
         Attribute.
-        Base values for classification, added to final class score; the size
-        must be the same as the classes or can be left unassigned (assumed 0)
+        Base values for regression, added to final prediction after applying
+        aggregate_function; the size must be the same as the classes or can be
+        left unassigned (assumed 0)
     n_targets
         Attribute.
         The total number of targets.

src/spox/opset/ai/onnx/ml/v4.py

@@ -64,7 +64,7 @@ class Attributes(BaseAttributes):
         default_float: AttrFloat32
         default_int64: AttrInt64
         default_string: AttrString
-        default_tensor: AttrTensor
+        default_tensor: Optional[AttrTensor]
         keys_floats: Optional[AttrFloat32s]
         keys_int64s: Optional[AttrInt64s]
         keys_strings: Optional[AttrStrings]
@@ -95,7 +95,7 @@ def label_encoder(
     default_float: float = -0.0,
     default_int64: int = -1,
     default_string: str = "_Unused",
-    default_tensor: np.ndarray,
+    default_tensor: Optional[np.ndarray] = None,
     keys_floats: Optional[Iterable[float]] = None,
     keys_int64s: Optional[Iterable[int]] = None,
     keys_strings: Optional[Iterable[str]] = None,
@@ -144,7 +144,8 @@ def label_encoder(
         A string.
     default_tensor
         Attribute.
-        A default tensor.
+        A default tensor. {"*Unused"} if values*\ \* has string type, {-1} if
+        values\_\* has integral type, and {-0.f} if values\_\* has float type.
     keys_floats
         Attribute.
         A list of floats.
@@ -192,7 +193,7 @@ def label_encoder(
             default_float=AttrFloat32(default_float, name="default_float"),
             default_int64=AttrInt64(default_int64, name="default_int64"),
             default_string=AttrString(default_string, name="default_string"),
-            default_tensor=AttrTensor(default_tensor, name="default_tensor"),
+            default_tensor=AttrTensor.maybe(default_tensor, name="default_tensor"),
             keys_floats=AttrFloat32s.maybe(keys_floats, name="keys_floats"),
             keys_int64s=AttrInt64s.maybe(keys_int64s, name="keys_int64s"),
             keys_strings=AttrStrings.maybe(keys_strings, name="keys_strings"),

src/spox/opset/ai/onnx/v17.py

@@ -4615,7 +4615,7 @@ def batch_normalization(
     training_mode
         Attribute.
         If set to true, it indicates BatchNormalization is being used for
-        training, and outputs 1, 2, 3, and 4 would be populated.
+        training, and outputs 1 and 2 are to be computed.
 
     Returns
     =======
@@ -8532,7 +8532,11 @@ def layer_normalization(
     indicate the i-th dimension of ``X``. If ``X``'s shape is
     ``[d[0], ..., d[axis-1], d[axis], ..., d[rank-1]]``, the shape of
     ``Mean`` and ``InvStdDev`` is ``[d[0], ..., d[axis-1], 1, ..., 1]``.
-    ``Y`` and ``X`` have the same shape.
+    ``Y`` and ``X`` have the same shape. This operator supports
+    unidirectional broadcasting (tensors ``Scale`` and ``B`` should be
+    unidirectional broadcastable to tensor ``X``); for more details please
+    check `the
+    doc <https://github.com/onnx/onnx/blob/main/docs/Broadcasting.md>`__.
 
     Parameters
     ==========
@@ -9302,7 +9306,8 @@ def max_pool(
        output_spatial_shape[i] = ceil((input_spatial_shape[i] + pad_shape[i] - dilation[i] * (kernel_shape[i] - 1) - 1) / strides_spatial_shape[i] + 1)
 
     if ceil_mode is enabled. ``pad_shape[i]`` is the sum of pads along axis
-    ``i``.
+    ``i``. Sliding windows that would start in the right padded region are
+    ignored.
 
     ``auto_pad`` is a DEPRECATED attribute. If you are using them currently,
     the output spatial shape will be following when ceil_mode is enabled:
@@ -14978,16 +14983,16 @@ def top_k(
 ) -> Tuple[Var, Var]:
     r"""
     Retrieve the top-K largest or smallest elements along a specified axis.
-    Given an input tensor of shape [a_1, a_2, ..., a_n, r] and integer
+    Given an input tensor of shape [a_0, a_1, ..., a\_{n-1}] and integer
     argument k, return two outputs:
 
-    -  Value tensor of shape [a_1, a_2, ..., a\_{axis-1}, k, a\_{axis+1},
-       ... a_n] which contains the values of the top k elements along the
-       specified axis
+    -  Value tensor of shape [a_0, a_1, ..., a\_{axis-1}, k, a\_{axis+1},
+       ... a\_{n-1}] which contains the values of the top k elements along
+       the specified axis
 
-    -  Index tensor of shape [a_1, a_2, ..., a\_{axis-1}, k, a\_{axis+1},
-       ... a_n] which contains the indices of the top k elements (original
-       indices from the input tensor).
+    -  Index tensor of shape [a_0, a_1, ..., a\_{axis-1}, k, a\_{axis+1},
+       ... a\_{n-1}] which contains the indices of the top k elements
+       (original indices from the input tensor).
 
     -  If "largest" is 1 (the default value) then the k largest elements are
        returned.
@@ -15006,7 +15011,7 @@ def top_k(
     ==========
     X
         Type T.
-        Tensor of shape [a_1, a_2, ..., a_n, r]
+        Tensor of shape [a_0, a_1, ..., a\_{n-1}]
     K
         Type tensor(int64).
         A 1-D tensor containing a single positive value corresponding to the
@@ -15027,12 +15032,13 @@ def top_k(
     =======
     Values : Var
         Type T.
-        Tensor of shape [a_1, a_2, ..., a\_{axis-1}, k, a\_{axis+1}, ... a_n]
-        containing top K values from the input tensor
+        Tensor of shape [a_0, a_1, ..., a\_{axis-1}, k, a\_{axis+1}, ...
+        a\_{n-1}] containing top K values from the input tensor
     Indices : Var
         Type I.
-        Tensor of shape [a_1, a_2, ..., a\_{axis-1}, k, a\_{axis+1}, ... a_n]
-        containing the corresponding input tensor indices for the top K values.
+        Tensor of shape [a_0, a_1, ..., a\_{axis-1}, k, a\_{axis+1}, ...
+        a\_{n-1}] containing the corresponding input tensor indices for the top
+        K values.
 
     Notes
     =====

src/spox/opset/ai/onnx/v19.py

@@ -821,7 +821,8 @@ def average_pool(
        output_spatial_shape[i] = ceil((input_spatial_shape[i] + pad_shape[i] - dilation[i] * (kernel_shape[i] - 1) - 1) / strides_spatial_shape[i] + 1)
 
     if ceil_mode is enabled. ``pad_shape[i]`` is the sum of pads along axis
-    ``i``.
+    ``i``. Sliding windows that would start in the right padded region are
+    ignored.
 
     ``auto_pad`` is a DEPRECATED attribute. If you are using them currently,
     the output spatial shape will be following when ceil_mode is enabled:

tests/test_custom_operator.py

@@ -5,6 +5,7 @@
 for the respective fields ``attrs/inputs/outputs`` and ``infer_output_types`` will be useful as well.
 Of these, ``propagate_values`` is probably least common.
 """
+
 from dataclasses import dataclass
 from typing import Dict