fix import issues with the latest onnx

CHANGELOGS.rst

@@ -4,6 +4,7 @@ Change Logs
 0.3.2
 +++++
 
+* :pr:`103`: fix import issue with the latest onnx version
 * :pr:`101`: fix as_tensor in onnx_text_plot_tree
 
 0.3.1

_unittests/ut_npx/test_npx.py

@@ -1268,7 +1268,7 @@ def test_numpy_op_bin_reduce(self):
             "xor", lambda x, y: (x.sum() == y.sum()) ^ (((-x).sum()) == y.sum())
         )
 
-    def common_test_inline(self, fonx, fnp, tcst=0):
+    def common_test_inline(self, fonx, fnp, tcst=0, atol=1e-10):
         f = fonx(Input("A"))
         self.assertIsInstance(f, Var)
         onx = f.to_onnx(constraints={0: Float64[None], (0, False): Float64[None]})
@@ -1277,7 +1277,7 @@ def common_test_inline(self, fonx, fnp, tcst=0):
         y = fnp(x)
         ref = ReferenceEvaluator(onx)
         got = ref.run(None, {"A": x})
-        self.assertEqualArray(y, got[0], atol=1e-10)
+        self.assertEqualArray(y, got[0], atol=atol)
 
     def common_test_inline_bin(self, fonx, fnp, tcst=0):
         f = fonx(Input("A"), Input("B"))
@@ -1470,7 +1470,7 @@ def test_equal(self):
 
     @unittest.skipIf(scipy is None, reason="scipy is not installed.")
     def test_erf(self):
-        self.common_test_inline(erf_inline, scipy.special.erf)
+        self.common_test_inline(erf_inline, scipy.special.erf, atol=1e-7)
 
     def test_exp(self):
         self.common_test_inline(exp_inline, np.exp)

_unittests/ut_translate_api/test_translate_classic.py

@@ -406,7 +406,7 @@ def _run(cls, code):
         import onnx.helper
         import onnx.numpy_helper
         import onnx_array_api.translate_api.make_helper
-        import onnx.reference.custom_element_types
+        import ml_dtypes
 
         def from_array_extended(tensor, name=None):
             dt = tensor.dtype
@@ -433,7 +433,7 @@ def from_array_extended(tensor, name=None):
         globs.update(onnx.helper.__dict__)
         globs.update(onnx.numpy_helper.__dict__)
         globs.update(onnx_array_api.translate_api.make_helper.__dict__)
-        globs.update(onnx.reference.custom_element_types.__dict__)
+        globs.update(ml_dtypes.__dict__)
         globs["from_array_extended"] = from_array_extended
         locs = {}
         try:

_unittests/ut_xrun_doc/test_documentation_examples.py

@@ -40,7 +40,7 @@ def run_test(self, fold: str, name: str, verbose=0) -> int:
             cmds = [sys.executable, "-u", os.path.join(fold, name)]
             p = subprocess.Popen(cmds, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
             res = p.communicate()
-            out, err = res
+            _out, err = res
             st = err.decode("ascii", errors="ignore")
             if st and "Traceback" in st:
                 if '"dot" not found in path.' in st:

onnx_array_api/profiling.py

@@ -281,7 +281,12 @@ def align_text(text, size):
             return text[:h] + "..." + text[-h + 1 :]
 
         dicts = self.as_dict(filter_node=filter_node, sort_key=sort_key)
-        max_nc = max(max(_["nc1"] for _ in dicts), max(_["nc2"] for _ in dicts))
+        set1 = [_["nc1"] for _ in dicts]
+        set2 = [_["nc1"] for _ in dicts]
+        if set1 or set2:
+            max_nc = max([*set1, *set2])
+        else:
+            max_nc = 1
         dg = int(math.log(max_nc) / math.log(10) + 1.5)
         line_format = (
             "{indent}{fct} -- {nc1: %dd} {nc2: %dd} -- {tin:1.5f} {tall:1.5f}"

onnx_array_api/translate_api/base_emitter.py

@@ -129,8 +129,9 @@ def render_attribute_value(self, value: Any) -> Tuple[List[str], str]:
         if value[0].type == AttributeProto.TENSOR:
             repl = {"bool": "bool_", "object": "object_", "str": "str_"}
             sdtype = repl.get(str(v.dtype), str(str(v.dtype)))
+            package = "np" if hasattr(np, sdtype) else "ml_dtypes"
             return [], (
-                f"from_array(np.array({v.tolist()}, dtype=np.{sdtype}), "
+                f"from_array(np.array({v.tolist()}, dtype={package}.{sdtype}), "
                 f"name={value[0].name!r})"
             )
         if isinstance(v, (int, float, list)):

onnx_array_api/translate_api/builder_emitter.py

@@ -1,4 +1,5 @@
 from typing import Any, Dict, List
+import numpy as np
 from onnx import TensorProto
 from onnx.numpy_helper import to_array
 from .base_emitter import BaseEmitter
@@ -135,7 +136,10 @@ def _emit_end_signature(self, **kwargs: Dict[str, Any]) -> List[str]:
             val = to_array(init)
             stype = str(val.dtype).split(".")[-1]
             name = self._clean_result_name(init.name)
-            rows.append(f"    {name} = np.array({val.tolist()}, dtype=np.{stype})")
+            package = "np" if hasattr(np, stype) else "ml_dtypes"
+            rows.append(
+                f"    {name} = np.array({val.tolist()}, dtype={package}.{stype})"
+            )
         return rows
 
     def _emit_begin_return(self, **kwargs: Dict[str, Any]) -> List[str]:

onnx_array_api/translate_api/inner_emitter.py

@@ -1,4 +1,5 @@
 from typing import Any, Dict, List, Optional, Tuple
+import numpy as np
 from onnx import AttributeProto
 from ..annotations import ELEMENT_TYPE_NAME
 from .base_emitter import BaseEmitter
@@ -105,7 +106,7 @@ def _emit_initializer(self, **kwargs: Dict[str, Any]) -> List[str]:
             else:
                 raise NotImplementedError(f"Unexpected dtype={sdtype}.")
         else:
-            sdtype = f"np.{sdtype}"
+            sdtype = f"np.{sdtype}" if hasattr(np, sdtype) else f"ml_dtypes.{sdtype}"
 
         return [
             "initializers.append(",
@@ -233,7 +234,7 @@ def _emit_initializer(self, **kwargs: Dict[str, Any]) -> List[str]:
             else:
                 raise NotImplementedError(f"Unexpected dtype={sdtype}.")
         else:
-            sdtype = f"np.{sdtype}"
+            sdtype = f"np.{sdtype}" if hasattr(np, sdtype) else f"ml_dtypes.{sdtype}"
         if value.size <= 16:
             return [
                 "initializers.append(",

onnx_array_api/translate_api/light_emitter.py

@@ -1,4 +1,5 @@
 from typing import Any, Dict, List
+import numpy as np
 from ..annotations import ELEMENT_TYPE_NAME
 from .base_emitter import BaseEmitter
 
@@ -43,8 +44,9 @@ def _emit_initializer(self, **kwargs: Dict[str, Any]) -> List[str]:
         value = kwargs["value"]
         repl = {"bool": "bool_", "object": "object_", "str": "str_"}
         sdtype = repl.get(str(value.dtype), str(str(value.dtype)))
+        package = "np" if hasattr(np, sdtype) else "ml_dtypes"
         return [
-            f"cst(np.array({value.tolist()}, dtype=np.{sdtype}))",
+            f"cst(np.array({value.tolist()}, dtype={package}.{sdtype}))",
             f"rename({name!r})",
         ]
 

pyproject.toml

@@ -39,7 +39,12 @@ select = [
 ]
 
 [tool.ruff.lint.per-file-ignores]
-"**" = ["B905", "C401", "C408", "C413", "PYI041", "RUF012", "RUF100", "RUF010", "SIM108", "SIM910", "SIM110", "SIM102", "SIM114", "SIM103", "UP015", "UP027", "UP031", "UP034", "UP032", "UP006", "UP035", "UP007", "UP038"]
+"**" = [
+    "B905", "C401", "C408", "C413", "PYI041",
+    "RUF012", "RUF100", "RUF010",
+    "SIM108", "SIM910", "SIM110", "SIM102", "SIM114", "SIM103",
+    "UP015", "UP027", "UP031", "UP034", "UP032", "UP006", "UP035", "UP007", "UP038", "UP045"
+]
 "**/plot*.py" = ["B018"]
 "_doc/examples/plot_first_example.py" = ["E402", "F811"]
 "_doc/examples/plot_onnxruntime.py" = ["E402", "F811"]