Preserve prefixes outside --line-ranges

[cyphercodes]

Description

Preserves comments and blank lines that are outside the requested --line-ranges instead of normalizing them while unchanged code is converted to standalone-comment leaves for range formatting.

This keeps the issue repro constrained to the requested lines: comments and separation before unchanged statements are preserved when the selected range ends immediately before them.

Fixes #4430.

Checklist - did you ...

• Implement any code style changes under the --preview style, following the stability policy?
• Add an entry in CHANGES.md if necessary?
• Add / update tests if necessary?
• Add new / update outdated documentation?

Verification

• python -m black --check src/black/comments.py src/black/linegen.py src/black/lines.py src/black/ranges.py tests/test_ranges.py
• python -m pytest tests/test_ranges.py tests/test_format.py -q
• git diff --check

[github-actions]

diff-shades results comparing this PR (fb7be7d) to main (c7801d9):

--preview style: no changes

--stable style: no changes

---

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

[cobaltt7]

Thanks - this fixes issues where whitespace at the beginning of lines before comments is stripped when using --line-ranges. However, this doesn't address trailing whitespace on prefix lines, which I think was the main issue #4430 was implying.

As an example:

   #  format whitespace   
print( "format me" )   
      

   #  don't format whitespace   
print("don't format me"  )     
      

Output with --line-ranges 1-3:

#  format whitespace
print("format me")


   #  don't format whitespace
print("don't format me"  )     
[empty line at EOF]

Expected output:

#  format whitespace
print( "format me" )


   #  don't format whitespace   
print("don't format me"  )     
      

[cobaltt7]

I also found an edge case where this fix doesn't seem to be applied. Example with --line-ranges 1-2:

print( "format me" )
   #  format whitespace

print("don't format me"  )

The spaces before the comment are retained, but they should be removed here. They are removed when running Black on the whole file.

[cobaltt7]

@cyphercodes are you still updating this?

[cyphercodes]

Updated — merged current main and resolved the src/black/ranges.py conflict by preserving the upstream batched replacement logic plus the PR's line-range conversion marker.

Verification:

• python -m pytest tests/test_black.py -k 'line_ranges or preserve_outside_prefix' -q
• git diff --check

[cobaltt7]

@cyphercodes Please resolve my two review comments above.

[cyphercodes]

Updated — addressed both line-ranges prefix cases from the review:

• unselected prefix comment lines now keep their original leading/trailing whitespace
• prefix comments that fall inside the selected --line-ranges are still normalized
• trailing whitespace-only blank lines outside the selected ranges are preserved

Verification:

• .venv/bin/python -m black src/black/__init__.py src/black/comments.py src/black/linegen.py src/black/ranges.py src/blib2to3/pytree.py tests/test_black.py
• .venv/bin/python -m pytest tests/test_black.py -k 'line_ranges or preserve_outside_prefix' -q
• .venv/bin/python -m pytest tests/test_format.py -k line_ranges -q
• git diff --check

[cobaltt7]

Is there a reason these aren't their own test case files? They should be if possible.

[cobaltt7]

Wouldn't line_ranges_first_lineno != 0 or line_ranges_selected is not None always be equal to line_ranges_converted? Could we replace line_ranges_converted with one of those?

[cobaltt7]

Also, when I first read through, I assumed that min(line_ranges_selected) would be the same as line_ranges_first_lineno, but I don't think that's always the case after reading through the code. It may be worth adding comments describing the use of both and the differences.

[cobaltt7]

Could we use comparison operators (<=/>=) instead of creating a set and checking inclusion?