[sjh9714]

Description

Fixes #5187.

Comment-only # fmt: off / # fmt: on blocks before a with statement could be emitted on the same logical line as the following with statement after a preceding standalone comment. When the line was split, the separator before with became a line prefix, so Black produced invalid code with a leading space at top level.

This normalizes the prefix copied onto comment-only fmt blocks and flushes those blocks when they follow Black's invisible context-manager parenthesis, while leaving fmt blocks inside real bracketed expressions on the existing path.

Checklist - did you ...

• Implement any code style changes under the --preview style, following the stability policy? (N/A: this fixes invalid output, not a style change.)
• Add an entry in CHANGES.md if necessary?
• Add / update tests if necessary?
• Add new / update outdated documentation? (N/A.)

Testing

• /tmp/black-pr-5187-venv/bin/python -m pytest tests/test_format.py -k 'fmtonoff_comment_only_with or fmtonoff6 or fmtonoff3 or fmtonoff2 or fmtskip11'
• /tmp/black-pr-5187-venv/bin/python -m pytest tests/test_format.py
• /tmp/black-pr-5187-venv/bin/python -m black --check src/black/linegen.py src/black/comments.py
• env -u NO_COLOR TERM=xterm-256color /tmp/black-pr-5187-venv/bin/python -m pytest tests --run-optional no_jupyter

Note: I used Codex while preparing this change, reviewed the final diff, and ran the listed checks locally.

[maflAT]

This issue can be reduced to a much simpler case.

if False is True or False:  #fmt: skip
    pass#this comment should begin with a space

2026-06-22 08:30:07.037 [info] error: cannot format c:_WIP_\black.py: Cannot parse: 2:26: False is True or False: #fmt: skip

Oh no! 💥 💔 💥
1 file failed to reformat.

Black seems to crash on any slightly complex condition with a fmt: skip comment.
The following 2 cases work perfectly fine:

if False is True :  #fmt: skip
    pass  # this comment should begin with a space

if False is True or False:
    pass  # this comment should begin with a space

[github-actions]

diff-shades results comparing this PR (558d44c) to main (b74b230):

--preview style: no changes

--stable style: no changes

---

What is this? | Workflow run | diff-shades documentation

[cobaltt7]

@maflAT I could not reproduce a crash with either of those snippets on main nor this PR. Please open a new issue with a reproduction if you're experiencing further issues.

@sjh9714 Thanks for the PR!