Interactive Mode suddenly prevents indexing "[]"

[alex77g2]

Bug report

Bug description:

import numpy as np
m = np.arange(16).reshape(4,4)
s = "m[1]" # see text below
m[1]  # array([4, 5, 6, 7])

The code above works as expected in Python 3.12 (file.py + interactive mode) and Python 3.13 (from file.py).
But it fails in Interactive Mode (console), at least under Windows 10.
Running same code from file.py works ( "print(m[1])" ).
Python 3.13.3 (tags/v3.13.3:6280bb5, Apr 8 2025, 14:47:33) [MSC v.1943 64 bit (AMD64)] on win32
It is nearly impossible to enter "m[1]". If testing Console of 3.12 it is fine. Console 3.13 is broken.
Reopen and retry does not fix it. After some time trying, I got it only once via entering strings "[1]" and removing string limiters '"'. If code is somehow entered, it works, but usually it is impossible to enter now.

CPython versions tested on:

3.13

Operating systems tested on:

Windows

[alex77g2]

I use python console mode every week and I like it. Yesterday, I started using Python 3.13 after 1 year daily using Python 3.12 under Windwos and Ubuntu/Linux. I would see me as an educated power user.

[chris-eibl]

Maybe a duplicate of #132439?

[alex77g2]

Maybe a duplicate of #132439?

I can confirm, that I also have german keyboard layout (as in the suggested duplicate).
I can also confirm, that it affects {} and [] but not () . This is in line with duplicate suggestion.
I found, that entering "[" many times helps sometimes, but then the closing one "]" is impossible.

[alex77g2]

Another note: we do not need numpy, list indexing is enough.

a = (1,2,3,4)
s = "a[1]" # sometimes possible after hard trying
a[1] # 2

Pressing "[" many time allows for a single "[" char. "[[[[" e.g. gives "[", if you open multiple visible ones, closing a single one is often possible, but only one closing "]". Is there some counting inside, that is intended to avoid unmatched [] or {} ???
I tried on another german Windows computer and it still fails.

[StanFromIreland]

I can confirm

You confirmed this is a duplicate. The fix was not backported, are there plans to do so? Discussion on weather it should be should continue on the pr, not here.

[chris-eibl]

Fixed in 3.14 beta 1 and also backported to 3.13. Will be in the upcoming 3.13.4. I think we can close here?

[alex77g2]

Fixed in 3.14 beta 1 and also backported to 3.13. Will be in the upcoming 3.13.4. I think we can close here?

OK, closing. A final question: are there nightly builds with the fix available?

[chris-eibl]

3.14 b1 can be downloaded from python.org

[chris-eibl]

AFAIR, 3.13.2 did not have the bug, so this might be another option from python.org

[chris-eibl]

Or just disable the new REPL using the environment variable PYTHON_BASIC_REPL=1.

[alex77g2]

Or just disable the new REPL using the environment variable PYTHON_BASIC_REPL=1.

rhis helps.