[IR] Refactor TensorBase to simplify implementation

[justinchuby]

Move name, doc_string, meta and metadata fields to the base class

PR Overview

This PR refactors the TensorBase class to simplify its implementation by consolidating common fields (name, doc_string, metadata, and metadata_props) to the base class. Key changes include:

• Adding an init method and slots in TensorBase to manage common attributes.
• Removing duplicate metadata properties from Tensor, ExternalTensor, and StringTensor.
• Updating TensorProtoTensor in the serde module to call the new TensorBase constructor.

Reviewed Changes

File	Description
onnxscript/ir/_core.py	Refactors TensorBase by adding common initialization and removing metadata duplicates in subclasses.
onnxscript/ir/serde.py	Updates TensorProtoTensor by removing inlined metadata_props deserialization and adjusting imports.

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (1)

onnxscript/ir/serde.py:246

• The metadata_props deserialization from proto (previously done via deserialize_metadata_props) has been removed in TensorProtoTensor.init. If metadata is expected to be propagated from the proto, consider reintroducing this assignment.

super().__init__()

[codecov]

❌ 42 Tests Failed:

Tests completed	Failed	Passed	Skipped
9670	42	9628	1942

View the top 3 failed test(s) by shortest run time

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0753_test_optional_has_element_empty_no_input_name_optional_input

Stack Traces | 0.004s run time

onnxscript\backend\onnx_export_test.py:137: in extract_functions
    mod = importlib.import_module(import_name)
C:\hostedtoolcache\windows\Python\3.11.9\x64\Lib\importlib\__init__.py:126: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
E   ModuleNotFoundError: No module named 'tests.onnx_backend_test_code.test_optional_has_element_empty_no_input_name_optional_input'

The above exception was the direct cause of the following exception:
.nox\test_onnx_weekly\Lib\site-packages\parameterized\parameterized.py:620: in standalone_func
    return func(*(a + p.args), **p.kwargs, **kw)
onnxscript\backend\onnx_export_test.py:271: in test_export2python_produces_correct_onnx_script_model
    functions = extract_functions(backend_test.name, code, self.test_folder)
onnxscript\backend\onnx_export_test.py:139: in extract_functions
    raise AssertionError(
E   AssertionError: Unable to import 'tests.onnx_backend_test_code.test_optional_has_element_empty_no_input_name_optional_input' (e=No module named 'tests.onnx_backend_test_code.test_optional_has_element_empty_no_input_name_optional_input') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_optional_has_element_empty_no_input_name_optional_input.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_optional_has_element_empty_no_input_name_optional_input.py', current folder: D:\a\onnxscript\onnxscript
E   ---- CONTENT --
E   import numpy
E   from onnx import TensorProto
E   from onnx.helper import make_tensor
E   from onnxscript import script, external_tensor
E   from onnxscript.values import Opset
E   from onnxscript.onnx_types import BOOL
E   from onnxscript.onnx_opset import opset18
E   
E   @script()
E   def bck_test_optional_has_element_empty_no_input_name_optional_input() -> (BOOL):
E       output = opset18.OptionalHasElement(None)
E       return output

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0945_test_reshape_reordered_last_dims

Stack Traces | 0.004s run time

onnxscript\backend\onnx_export_test.py:137: in extract_functions
    mod = importlib.import_module(import_name)
C:\hostedtoolcache\windows\Python\3.12.9\x64\Lib\importlib\__init__.py:90: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
E   ModuleNotFoundError: No module named 'tests.onnx_backend_test_code.test_reshape_reordered_last_dims'

The above exception was the direct cause of the following exception:
.nox\test\Lib\site-packages\parameterized\parameterized.py:620: in standalone_func
    return func(*(a + p.args), **p.kwargs, **kw)
onnxscript\backend\onnx_export_test.py:271: in test_export2python_produces_correct_onnx_script_model
    functions = extract_functions(backend_test.name, code, self.test_folder)
onnxscript\backend\onnx_export_test.py:139: in extract_functions
    raise AssertionError(
E   AssertionError: Unable to import 'tests.onnx_backend_test_code.test_reshape_reordered_last_dims' (e=No module named 'tests.onnx_backend_test_code.test_reshape_reordered_last_dims') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_reshape_reordered_last_dims.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_reshape_reordered_last_dims.py', current folder: D:\a\onnxscript\onnxscript
E   ---- CONTENT --
E   import numpy
E   from onnx import TensorProto
E   from onnx.helper import make_tensor
E   from onnxscript import script, external_tensor
E   from onnxscript.values import Opset
E   from onnxscript.onnx_types import FLOAT, INT64
E   from onnxscript.onnx_opset import opset21
E   
E   @script()
E   def bck_test_reshape_reordered_last_dims(data: FLOAT[2,3,4], shape: INT64[3]) -> (FLOAT[2,4,3]):
E       reshaped = opset21.Reshape(data, shape)
E       return reshaped

onnxscript.backend.onnx_export_test.TestOnnxBackEnd::test_export2python_produces_correct_onnx_script_model_0975_test_resize_upsample_scales_nearest

Stack Traces | 0.004s run time

onnxscript\backend\onnx_export_test.py:137: in extract_functions
    mod = importlib.import_module(import_name)
C:\hostedtoolcache\windows\Python\3.10.11\x64\lib\importlib\__init__.py:126: in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
E   ModuleNotFoundError: No module named 'tests.onnx_backend_test_code.test_resize_upsample_scales_nearest'

The above exception was the direct cause of the following exception:
.nox\test\lib\site-packages\parameterized\parameterized.py:620: in standalone_func
    return func(*(a + p.args), **p.kwargs, **kw)
onnxscript\backend\onnx_export_test.py:271: in test_export2python_produces_correct_onnx_script_model
    functions = extract_functions(backend_test.name, code, self.test_folder)
onnxscript\backend\onnx_export_test.py:139: in extract_functions
    raise AssertionError(
E   AssertionError: Unable to import 'tests.onnx_backend_test_code.test_resize_upsample_scales_nearest' (e=No module named 'tests.onnx_backend_test_code.test_resize_upsample_scales_nearest') (file: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_resize_upsample_scales_nearest.py', absolute path: 'D:\\a\\onnxscript\\onnxscript\\tests\\onnx_backend_test_code\\test_resize_upsample_scales_nearest.py', current folder: D:\a\onnxscript\onnxscript
E   ---- CONTENT --
E   import numpy
E   from onnx import TensorProto
E   from onnx.helper import make_tensor
E   from onnxscript import script, external_tensor
E   from onnxscript.values import Opset
E   from onnxscript.onnx_types import FLOAT
E   from onnxscript.onnx_opset import opset19
E   
E   @script()
E   def bck_test_resize_upsample_scales_nearest(X: FLOAT[1,1,2,2], scales: FLOAT[4]) -> (FLOAT[1,1,4,6]):
E       Y = opset19.Resize(X, None, scales, mode='nearest')
E       return Y

To view more test analytics, go to the Test Analytics Dashboard
_{📋 Got 3 mins? Take this short survey to help us improve Test Analytics.}

PR Overview

This PR refactors the TensorBase implementation by moving metadata, documentation, and name fields to the base class, thereby simplifying the initialization logic across tensor-related classes.

• Consolidated common field initialization (name, doc_string, metadata_props) into TensorBase.
• Removed duplicate property definitions and assignments from Tensor subclasses.
• Updated import statements to reflect the removal of now-unnecessary dependencies.

Reviewed Changes

File	Description
onnxscript/ir/serde.py	Updated TensorProtoTensor init to delegate metadata_props to the base class.
onnxscript/ir/_core.py	Added common fields to TensorBase and updated subclass constructors accordingly.

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

Comments suppressed due to low confidence (3)

onnxscript/ir/serde.py:247

• The refactor moves metadata properties initialization to the base class. Consider adding tests to ensure that metadata_props deserialization from proto remains consistent.

def __init__(self, proto: onnx.TensorProto) -> None:

onnxscript/ir/_core.py:107

• The new init in TensorBase consolidates common fields; please ensure that all subclasses correctly pass these parameters to maintain consistent behavior. Adding unit tests for these changes is recommended.

def __init__(self, name: str | None = None, doc_string: str | None = None, metadata_props: dict[str, str] | None = None) -> None:

onnxscript/ir/_core.py:397

• The removal of redundant assignments (e.g. name, doc_string, metadata_props) in the Tensor subclass appears appropriate; please verify through tests that these fields are correctly initialized via the base class.

super().__init__(name=name, doc_string=doc_string, metadata_props=metadata_props)