GeezerVison

The World Wide Web, restored to 800 pixels of glory.

GeezerVison is showing github.com — Pasting zero-width characters into PyREPL results in incorrect cursor position · Issue #139246 · python/cpython

Pasting zero-width characters into PyREPL results in incorrect cursor position #139246

New issue

Copy link

New issue

Copy link

Open

Pasting zero-width characters into PyREPL results in incorrect cursor position#139246

Copy link

Labels

stdlibStandard Library Python modules in the Lib/ directoryStandard Library Python modules in the Lib/ directorytopic-replRelated to the interactive shellRelated to the interactive shelltype-bugAn unexpected behavior, bug, or errorAn unexpected behavior, bug, or error

Description

grayjk

opened

on Sep 22, 2025

Issue body actions

Bug report

Bug description:

To reproduce

Tested with Python 3.14 rc3 on Linux

Open the new REPL
Paste: ééééé

The paste-able string above is five letter 'e's each followed by \u0301. To create it use:

print('e\u0301' * 5)

The \u0301 is a combining mark with zero-width display length

Current behavior

The cursor position is five spaces after the pasted string

Expected behavior

The cursor position is immediately after the pasted string

CPython versions tested on:

3.14

Operating systems tested on:

Linux

Linked PRs

Activity

grayjk
added
type-bugAn unexpected behavior, bug, or errorAn unexpected behavior, bug, or error
on Sep 22, 2025
grayjk commented on Sep 22, 2025
grayjk
on Sep 22, 2025
ContributorAuthor
More actions
The root cause seems to be related to pyrepl wlen miscalculating the length of the zero-width combining mark

wlen('e\u0301') is 2 instead of 1
StanFromIreland
added
stdlibStandard Library Python modules in the Lib/ directoryStandard Library Python modules in the Lib/ directory

topic-replRelated to the interactive shellRelated to the interactive shell
on Sep 22, 2025

StanFromIreland commented on Sep 22, 2025

StanFromIreland

on Sep 22, 2025

Member

More actions

Possibly related: #130273

>>> ééééé/0
Traceback (most recent call last):
  File "<python-input-6>", line 1, in <module>
    ééééé/0
    ~~~~~~~~~~^~
ZeroDivisionError: division by zero

Reacted by grayjk

grayjk commented on Sep 22, 2025
grayjk
on Sep 22, 2025
ContributorAuthor
More actions
@StanFromIreland It does look like traceback has a similar mishandling of zero-width characters in _display_width
yihong0618 commented on Sep 23, 2025
yihong0618
on Sep 23, 2025
Contributor
More actions
something like
seems width is wrong here

é ȩ
yihong0618 commented on Sep 23, 2025
yihong0618
on Sep 23, 2025
Contributor
More actions
I create a draft patch maybe can fix this can you help test?
cc @grayjk
grayjk commented on Sep 24, 2025
grayjk
on Sep 24, 2025
ContributorAuthor
More actions
Also affects Python 3.13
encukou
added a commit that references this issue on Oct 30, 2025
gh-139246: zero-width word paste can be wrong in default repl (GH-139254
4e6dba0
miss-islington
added a commit that references this issue on Oct 30, 2025
pythongh-139246: zero-width word paste can be wrong in default repl (p…
1d1a971
StanFromIreland
added a commit that references this issue on Nov 6, 2025
[3.13] pythongh-139246: zero-width word paste can be wrong in default…
1a8b6b5
encukou
added 2 commits that reference this issue on Nov 7, 2025
[3.14] gh-139246: zero-width word paste can be wrong in default repl (G…
d729087
[3.13] gh-139246: zero-width word paste can be wrong in default repl (G…
0923704
encukou commented on Nov 7, 2025
encukou
on Nov 7, 2025
Member
More actions
Thank you for the fix!
Should we keep this open for consolidating this and the calculations for tracebacks?
yihong0618 commented on Nov 7, 2025
yihong0618
on Nov 7, 2025
Contributor
More actions

Thank you for the fix! Should we keep this open for consolidating this and the calculations for tracebacks?

yes I think keep it open is better since there's a lot of case for the wcwidth
grayjk commented on Dec 1, 2025
grayjk
on Dec 1, 2025
ContributorAuthor
More actions
Hangul Jamo (https://www.unicode.org/faq/korean#3) also has zero width conjoining characters: https://wcwidth.readthedocs.io/en/latest/specs.html#width-of-0

An example is:

print("\u1100\u1161")

which displays:

가

The pyrepl wlen function in 3.15.0a2 returns 3 for this character. The length should be 2

Since the wlen return is incorrect, pasting the character results in an incorrect cursor position
yihong0618 commented on Dec 3, 2025
yihong0618
on Dec 3, 2025
Contributor
More actions

가

This is the same behavior in basic repl and other repl I think its fine here?
typenoob commented on Dec 3, 2025
typenoob
on Dec 3, 2025
More actions

가

It somehow looks ugly in my 3.14.0 REPL. (I think it's the OS problem, it looks fine on Linux)
And yes, we should calc its wlen as 2.
typenoob commented on Dec 3, 2025
typenoob
on Dec 3, 2025
More actions
Do you think we should improve wlen and str_width based on wcwidth, or we should only fix when there's incorrect behavior occurs? @grayjk
StanFromIreland
added a commit that references this issue on Dec 6, 2025
pythongh-139246: zero-width word paste can be wrong in default repl (p…
7281913
StanFromIreland commented on May 9, 2026
StanFromIreland
on May 9, 2026 · Hidden as outdated
show commentMore actions
StanFromIreland
closed this as a duplicateof Traceback colors are shifted when the line contains wide unicode characters #130273 on May 9, 2026
StanFromIreland commented on May 9, 2026
StanFromIreland
on May 9, 2026 · Hidden as resolved
show commentMore actions
StanFromIreland
reopened this on May 9, 2026
StanFromIreland commented on May 9, 2026
StanFromIreland
on May 9, 2026
Member
More actions
We get an incorrect number of carets for 가, and when trying delete it, on the second backspace it takes too much:

>>> 가/0 Traceback (most recent call last): File "<python-input-4>", line 1, in <module> 가/0 ^^^ NameError: name '가' is not defined

And, when trying delete it, on the second backspace it takes too much:

>>
grayjk commented on May 11, 2026
grayjk
on May 11, 2026
ContributorAuthor
More actions
@StanFromIreland what Python version?

I tested in 3.15.0 beta 1 with Putty and it worked (i.e. displayed 2 carets).

The backspace problem is tracked in issue #142162

Metadata

Assignees

No one assigned

Labels

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions

You are visitor 01402503 — GeezerVison does not store the pages you look up.
Best viewed with Netscape Navigator 4.7 or Internet Explorer 6 at 800×600. No cookies required.

GeezerVison

Pasting zero-width characters into PyREPL results in incorrect cursor position #139246

Description

Bug report

Bug description:

To reproduce

Current behavior

Expected behavior

CPython versions tested on:

Operating systems tested on:

Linked PRs

Activity

grayjk commented on Sep 22, 2025

StanFromIreland commented on Sep 22, 2025

grayjk commented on Sep 22, 2025

yihong0618 commented on Sep 23, 2025

yihong0618 commented on Sep 23, 2025

grayjk commented on Sep 24, 2025

encukou commented on Nov 7, 2025

yihong0618 commented on Nov 7, 2025

grayjk commented on Dec 1, 2025

yihong0618 commented on Dec 3, 2025

typenoob commented on Dec 3, 2025

typenoob commented on Dec 3, 2025

StanFromIreland commented on May 9, 2026

StanFromIreland commented on May 9, 2026

StanFromIreland commented on May 9, 2026

grayjk commented on May 11, 2026

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions