Bug#1117200: pandas: Compatibility with numexpr v2.13

[Antonio Valentino]

Source: pandas
Version: 2.3.2+dfsg-2
Severity: normal
Tags: patch
X-Debbugs-Cc: antonio.valentino at tiscali.it

Dear Maintainer,
currently pandas is not compatible with numexpr >= v2.13 (already in 
unstable).

The incompatibility is caused buy a change in the error type raised by 
some invalid expressions in numexpr v2.13.
See https://github.com/PyTables/PyTables/pull/1256.

The issue has been fixed upstream.
See: https://github.com/pandas-dev/pandas/pull/62452.

I kindly ask to apply the attached patch (I can also open a merge 
request on salsa if it is considered a better option).

-- System Information:
Debian Release: forky/sid
   APT prefers unstable
   APT policy: (500, 'unstable'), (1, 'experimental')
Architecture: amd64 (x86_64)

Kernel: Linux 6.14.0-33-generic (SMP w/8 CPU threads; PREEMPT)
Locale: LANG=C, LC_CTYPE=C.UTF-8 (charmap=UTF-8), LANGUAGE not set
Shell: /bin/sh linked to /usr/bin/dash
Init: unable to detect
-------------- next part --------------
commit 179258fb8048999ba4cad675a7608f09fff88579
Author: Matthew Roeschke <10647082+mroeschke at users.noreply.github.com>
Date:   Thu Sep 25 11:43:11 2025 -0700

    TST: Adjust tests for numexpr 2.13 (#62452)

Index: pandas/pandas/tests/computation/test_eval.py
===================================================================
--- pandas.orig/pandas/tests/computation/test_eval.py
+++ pandas/pandas/tests/computation/test_eval.py
@@ -8,6 +8,7 @@ import numpy as np
 import pytest
 
 from pandas.compat import PY312
+from pandas.compat._optional import import_optional_dependency
 from pandas.errors import (
     NumExprClobberingError,
     PerformanceWarning,
@@ -54,6 +55,9 @@ from pandas.core.computation.ops import
     _unary_math_ops,
 )
 from pandas.core.computation.scope import DEFAULT_GLOBALS
+from pandas.util.version import Version
+
+numexpr = import_optional_dependency("numexpr", errors="ignore")
 
 
 @pytest.fixture(
@@ -322,7 +326,9 @@ class TestEval:
     def test_floor_division(self, lhs, rhs, engine, parser):
         ex = "lhs // rhs"
 
-        if engine == "python":
+        if engine == "python" or (
+            engine == "numexpr" and Version(numexpr.__version__) >= Version("2.13.0")
+        ):
             res = pd.eval(ex, engine=engine, parser=parser)
             expected = lhs // rhs
             tm.assert_equal(res, expected)
@@ -393,7 +399,7 @@ class TestEval:
 
         # int raises on numexpr
         lhs = DataFrame(np.random.default_rng(2).integers(5, size=(5, 2)))
-        if engine == "numexpr":
+        if engine == "numexpr" and Version(numexpr.__version__) < Version("2.13.0"):
             msg = "couldn't find matching opcode for 'invert"
             with pytest.raises(NotImplementedError, match=msg):
                 pd.eval(expr, engine=engine, parser=parser)
@@ -438,7 +444,7 @@ class TestEval:
 
         # int raises on numexpr
         lhs = Series(np.random.default_rng(2).integers(5, size=5))
-        if engine == "numexpr":
+        if engine == "numexpr" and Version(numexpr.__version__) < Version("2.13.0"):
             msg = "couldn't find matching opcode for 'invert"
             with pytest.raises(NotImplementedError, match=msg):
                 pd.eval(expr, engine=engine, parser=parser)