[SulimanAbdulrazzaq]

Description

With --line-ranges, Black turns statements outside the selected ranges into a single STANDALONE_COMMENT using str(node). If such a statement has a # fmt: skip on the last line before a closing bracket, that comment has already become a nested STANDALONE_COMMENT whose value has no trailing newline (the line generator normally adds it). str(node) then glues the closing bracket onto the comment's line, so the bracket becomes part of the comment and Black fails on its own output:

$ printf 'x = 1\nv = (\n    a  # fmt: skip\n)\n' > t.py
$ black --line-ranges=1-1 t.py
error: cannot format t.py: INTERNAL ERROR: Black ... produced invalid code:
cannot parse: t.py:3:19
    a  # fmt: skip)
                  ^

Without --line-ranges the same file formats fine, and the lines outside the range should be left alone, so this is an internal error on code the user did not ask to touch.

The fix: when converting an unselected node (_convert_node_to_standalone_comment and _convert_nodes_to_standalone_comment in ranges.py) stringify the node with a helper that ends a nested STANDALONE_COMMENT with a newline when it ends in a # fmt: skip comment and the next leaf does not already start with a newline. Nothing else changes: a standalone comment that ends in code, such as a header followed by : (#5405), is joined as before.

Tests:

• tests/data/cases/line_ranges_fmt_skip_in_brackets.py: statements with # fmt: skip before a closing bracket (call, tuple, dict, if header, for iterable) outside the range stay untouched while the selected line is formatted.
• tests/data/line_ranges_formatted/fmt_skip_in_brackets.py: the existing line-by-line test runs --line-ranges over every line of a formatted file.

Both fail on main with INTERNAL ERROR ... produced invalid code and pass with the change.

Sibling code paths checked: the other str(node) consumer in this feature, _convert_nodes_to_standalone_comment (match/suite headers), is changed the same way; comments.py already adds newlines for nested standalone comments in its own stringification; visit_suite already skips suites that contain a standalone comment. Not in scope: some other --line-ranges + # fmt: off/# fmt: skip combinations still fail (for example an async for ... in ( # fmt: skip header outside the range, and tests/data/cases/fmtonoff5.py); they have different causes.

Performance: the helper walks the leaves of each converted node once, as str(node) did; I saw no consistent difference in a noisy local benchmark (--line-ranges=1-1 on a 6000-line file).

Checklist - did you ...

• Implement any code style changes under the --preview style, following the stability policy? (not a style change)
• Add an entry in CHANGES.md if necessary?
• Add / update tests if necessary?
• Add new / update outdated documentation? (no documentation change needed)

[github-actions]

diff-shades results comparing this PR (918a806) to main (6e94088):

--preview style: no changes

--stable style: no changes

---

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

[cobaltt7]

Thanks!