fix: parse filenames w.o. prefix, quoted unicode. Improve parse errors #439
Conversation
There was a problem hiding this comment.
Pull Request Overview
This PR enhances diff and status parsing to correctly handle filenames without a/b prefixes and quoted/unicode paths, and improves parse error diagnostics. Key updates include a new FilePath type with quoting support, a revamped parser error model with richer context, and updates across ops and UI to use the new path formatting. Tests were added for non-ASCII filenames, quoted escapes, and ambiguous cases.
- Add FilePath with fmt() to unescape quoted paths (using smashquote) and switch DiffHeader to use FilePath.
- Overhaul parser errors to accumulate ThinParseError with contextual formatting and logging; support headers without prefixes and robust newline-or-EOF handling.
- Update ops, UI rendering, and status parsing to use Strings/Cow for paths; add tests for non-ASCII and quoted filenames.
Reviewed Changes
Copilot reviewed 13 out of 14 changed files in this pull request and generated 11 comments.
Show a summary per file
| File | Description |
|---|---|
| src/tests/snapshots/gitu__tests__non_ascii_filename.snap | Adds snapshot to validate UI with non-ASCII filenames. |
| src/tests/mod.rs | Adds test to exercise non-ASCII filename handling. |
| src/screen/status.rs | Switches untracked paths to String-based flow and PathBuf conversion. |
| src/ops/unstage.rs | Uses FilePath::fmt() for path resolution in unstage. |
| src/ops/stage.rs | Uses FilePath::fmt() for path resolution in stage. |
| src/ops/show.rs | Uses FilePath::fmt() for editor paths. |
| src/ops/discard.rs | Uses FilePath::fmt() for paths; adjusts discard behavior for Added/Renamed/Other. |
| src/items.rs | Formats displayed filenames using FilePath::fmt(). |
| src/highlight.rs | Uses FilePath::fmt() for syntax highlight path discrimination. |
| src/gitu_diff.rs | Major parser refactor: FilePath, improved errors, support for quoted/unprefixed paths. |
| src/git/status.rs | Changes StatusFile path types to String. |
| src/git/parse/status/mod.rs | Parses status paths as Strings and unescapes quoted paths (for path; not yet for new_path). |
| Cargo.toml | Adds tinyvec and smashquote dependencies. |
Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.
| ItemData::Delta { diff, file_i } => { | ||
| let file_path = &diff.file_diffs[*file_i].header.new_file; | ||
| let path: &str = &file_path.fmt(&diff.text); | ||
| editor(Path::new(path), None) | ||
| } |
There was a problem hiding this comment.
This takes a reference to a temporary Cow returned by fmt(), resulting in a dangling reference. Bind the Cow (or an owned String) to a local variable so it lives through the call.
| } => { | ||
| let file_path = &diff.file_diffs[*file_i].header.new_file; | ||
| let path: &str = &file_path.fmt(&diff.text); | ||
| editor( | ||
| Path::new(path), | ||
| Some(diff.file_line_of_first_diff(*file_i, *hunk_i) as u32), | ||
| ) | ||
| } |
There was a problem hiding this comment.
Same temporary reference issue as above; path borrows from a temporary Cow. Store the Cow or an owned String in a local variable before passing a reference to Path::new.
| let old_mask = diff.mask_old_hunk(file_index, hunk_index); | ||
| let old_file_range = file_diff.header.old_file.clone(); | ||
| let old_path = &diff.text[old_file_range]; | ||
| let old_path = &old_file_range.fmt(&diff.text); |
There was a problem hiding this comment.
Both lines take references to temporary Cow values returned by fmt(), which will drop immediately, leaving dangling references. Assign the Cow to a local variable and pass a reference to its contents.
| let new_mask = diff.mask_new_hunk(file_index, hunk_index); | ||
| let new_file_range = file_diff.header.new_file.clone(); | ||
| let new_path = &diff.text[new_file_range]; | ||
| let new_path = &new_file_range.fmt(&diff.text); |
There was a problem hiding this comment.
Both lines take references to temporary Cow values returned by fmt(), which will drop immediately, leaving dangling references. Assign the Cow to a local variable and pass a reference to its contents.
| // TODO Need to handle invalid unicode | ||
| let line_start = self.input[..cursor].rfind('\n').unwrap_or(0); | ||
| let line_end = self.input[line_start..] | ||
| .find('\n') |
There was a problem hiding this comment.
line_end is computed as an offset within the slice self.input[line_start..] but then used as an absolute index, which can cause slicing panics. Convert the relative offset to an absolute index.
| .find('\n') | |
| .find('\n') | |
| .map(|offset| line_start + offset) |
| let file_path = &diff.file_diffs[*file_i].header.new_file; | ||
| let path: &str = &file_path.fmt(&diff.text); | ||
| editor(Path::new(path), None) |
There was a problem hiding this comment.
Suggested fix: bind the Cow to a local variable to avoid borrowing a temporary. For example: let path = file_path.fmt(&diff.text); editor(Path::new(path.as_ref()), None).
| let file_path = &diff.file_diffs[*file_i].header.new_file; | ||
| let path: &str = &file_path.fmt(&diff.text); | ||
| editor( | ||
| Path::new(path), | ||
| Some(diff.file_line_of_first_diff(*file_i, *hunk_i) as u32), | ||
| ) |
There was a problem hiding this comment.
Suggested fix: let path = file_path.fmt(&diff.text); editor(Path::new(path.as_ref()), Some(...)). This ensures the referent outlives the call.
| let old_mask = diff.mask_old_hunk(file_index, hunk_index); | ||
| let old_file_range = file_diff.header.old_file.clone(); | ||
| let old_path = &diff.text[old_file_range]; | ||
| let old_path = &old_file_range.fmt(&diff.text); |
There was a problem hiding this comment.
Suggested fix: store the Cow in locals so their lifetimes cover the call: let old = old_file_range.fmt(&diff.text); let newp = new_file_range.fmt(&diff.text); then pass old.as_ref(), newp.as_ref() to iter_syntax_highlights.
| let new_mask = diff.mask_new_hunk(file_index, hunk_index); | ||
| let new_file_range = file_diff.header.new_file.clone(); | ||
| let new_path = &diff.text[new_file_range]; | ||
| let new_path = &new_file_range.fmt(&diff.text); |
There was a problem hiding this comment.
Suggested fix: store the Cow in locals so their lifetimes cover the call: let old = old_file_range.fmt(&diff.text); let newp = new_file_range.fmt(&diff.text); then pass old.as_ref(), newp.as_ref() to iter_syntax_highlights.
21610ed to
ba69ec4
Compare
No description provided.