BUG: Infinite Loop in numpy.base_repr

[Stiibu]

Describe the issue:

A call to numpy.base_repr with a minimal value of the underlying int-type results in an infinite loop, e.g. -128 for np.int8. The problem seems to be a call to abs.

Reproduce the code example:

import numpy as np

np.base_repr(np.int8(-128))
# np.base_repr(np.int16(-2 ** 15)) fails as well

Error message:

RuntimeWarning: overflow encountered in scalar absolute num = abs(number)

Python and NumPy Versions:

Python 3.12.1
numpy 1.26.4

Runtime Environment:

[{'numpy_version': '1.26.4',
'python': '3.12.1 (tags/v3.12.1:2305ca5, Dec 7 2023, 22:03:25) [MSC v.1937 '
'64 bit (AMD64)]',
'uname': uname_result(system='Windows', node='EVK-NB-649', release='11', version='10.0.22621', machine='AMD64')},
{'simd_extensions': {'baseline': ['SSE', 'SSE2', 'SSE3'],
'found': ['SSSE3',
'SSE41',
'POPCNT',
'SSE42',
'AVX',
'F16C',
'FMA3',
'AVX2'],
'not_found': ['AVX512F',
'AVX512CD',
'AVX512_SKX',
'AVX512_CLX',
'AVX512_CNL',
'AVX512_ICL']}},
{'architecture': 'Haswell',
'filepath': 'C:\Users\$username\AppData\Local\Programs\Python\Python312\Lib\site-packages\numpy.libs\libopenblas64__v0.3.23-293-gc2f4bdbb-gcc_10_3_0-2bde3a66a51006b2b53eb373ff767a3f.dll',
'internal_api': 'openblas',
'num_threads': 16,
'prefix': 'libopenblas',
'threading_layer': 'pthreads',
'user_api': 'blas',
'version': '0.3.23.dev'}]
None

Context for the issue:

The current implementation of numpy.base_repr is not safely usable.

[krisrosengreen]

There seems to be an issue with the implementation of the abs function for int8, int16 ... int64. All corresponding negative edge values raise an overflow error when the absolute value is taken.

I'm not at all familiar with the Numpy codebase, so I couldn't find where the abs function for these is implemented.

[mhvk]

Thanks for reporting! This still is an issue on current main... np.base_repr? in ipython tells this is implemented as

numpy/numpy/_core/numeric.py

Lines 2139 to 2154 in e1bf1d6

	digits = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ'
	if base > len(digits):
	raise ValueError("Bases greater than 36 not handled in base_repr.")
	elif base < 2:
	raise ValueError("Bases less than 2 not handled in base_repr.")
	num = abs(number)
	res = []
	while num:
	res.append(digits[num % base])
	num //= base
	if padding:
	res.append('0' * padding)
	if number < 0:
	res.append('-')
	return ''.join(reversed(res or '0'))

I think the fix is simple: write num = abs(int(number)) to ensure one converts to a python int - but obviously one also has to add a test case. If you are able to do that, that would be great!

p.s. I don't quite like that np.abs(np.int8(-128)) yields -128 but that's probably best dealt with as a separate issue - #26145