comments/grammar.rs:105–125: the common leading-whitespace amount is computed as a byte offset via str::find, min'd across lines, then applied to every line with replace_range(..common, "").
If lines have different-byte-width leading whitespace (an ASCII space on one line, a multi-byte Unicode space like NBSP U+00A0 on another — a common copy-paste artifact), the byte offset lands mid-codepoint → panic end of range should be a character boundary. Parsed at construction time, before validation:
Reproduction
printf '/// a\n///\xc2\xa0b\nstruct S {}\n' > nbsp.slice && ./slicec nbsp.slice --dry-run
It's also a latent correctness bug (mixed indentation stripped by the wrong amount).
Fix
Measure and strip in chars, not bytes: char_indices().nth(common).
comments/grammar.rs:105–125: the common leading-whitespace amount is computed as a byte offset viastr::find, min'd across lines, then applied to every line withreplace_range(..common, "").If lines have different-byte-width leading whitespace (an ASCII space on one line, a multi-byte Unicode space like NBSP U+00A0 on another — a common copy-paste artifact), the byte offset lands mid-codepoint → panic
end of range should be a character boundary. Parsed at construction time, before validation:Reproduction
It's also a latent correctness bug (mixed indentation stripped by the wrong amount).
Fix
Measure and strip in chars, not bytes:
char_indices().nth(common).