fix: cursor hooks on windows #11616
fix: cursor hooks on windows #11616
Conversation
|
@Rettend is attempting to deploy a commit to the GitButler Team on Vercel. A member of the Team first needs to authorize it. |
There was a problem hiding this comment.
Pull request overview
This PR fixes a Windows-specific issue where Cursor hooks fail because Cursor sends workspace paths in the format /c:/Users/... instead of the standard C:/Users/.... The leading slash causes GitButler CLI commands to fail with "Failed to access a directory" errors.
Key Changes:
- Added
cursor_path_to_pathbuf()function that detects and strips the leading slash from Windows drive paths in the/c:/format - Updated
handle_after_edit()andhandle_stop()to use the new path conversion function - Added basic unit tests to verify the path transformation behavior
| fn test_cursor_path_to_pathbuf_windows_drive_paths() { | ||
| #[cfg(target_os = "windows")] | ||
| { | ||
| assert_eq!( | ||
| cursor_path_to_pathbuf("/C:/repo").to_string_lossy(), | ||
| "C:/repo" | ||
| ); | ||
| assert_eq!( | ||
| cursor_path_to_pathbuf("/M:/work/repo").to_string_lossy(), | ||
| "M:/work/repo" | ||
| ); | ||
| } |
There was a problem hiding this comment.
The function should also handle edge cases where paths are malformed or have lowercase drive letters. Consider adding test cases for:
- Lowercase drive letters (e.g.,
/c:/repoinstead of/C:/repo) - Edge case where the path is exactly
/C:(length < 3 after stripping prefix) - Paths with backslashes (e.g.,
/C:\repo)
The current byte checking at position 1 would be correct for lowercase drive letters since ASCII 'c' and 'C' both have ':' at position 1 when checking without_leading_slash.as_bytes()[1], but the tests should verify this behavior explicitly.
| fn test_cursor_path_to_pathbuf_windows_drive_paths() { | ||
| #[cfg(target_os = "windows")] | ||
| { | ||
| assert_eq!( | ||
| cursor_path_to_pathbuf("/C:/repo").to_string_lossy(), | ||
| "C:/repo" | ||
| ); | ||
| assert_eq!( | ||
| cursor_path_to_pathbuf("/M:/work/repo").to_string_lossy(), | ||
| "M:/work/repo" | ||
| ); | ||
| } |
There was a problem hiding this comment.
The test should also verify that paths without the problematic format are left unchanged on Windows. For example, regular Windows paths like C:/repo (without the leading slash) should remain as C:/repo, and Unix-style paths like /home/user should remain unchanged. This ensures the function doesn't modify paths unnecessarily.
| { | ||
| if let Some(without_leading_slash) = input.strip_prefix('/') { | ||
| let b = without_leading_slash.as_bytes(); | ||
| if b.len() >= 3 && b[1] == b':' && (b[2] == b'/' || b[2] == b'\\') { |
There was a problem hiding this comment.
The byte index check uses hardcoded positions (b[1] and b[2]) but doesn't explicitly validate that b[0] is an ASCII alphabetic character representing a drive letter. While the current logic may work in practice since Cursor sends valid drive letters, it would be more robust to check that b[0] is in the range A-Z or a-z. This would prevent false positives for paths like /1:/foo being treated as Windows drive paths.
| if b.len() >= 3 && b[1] == b':' && (b[2] == b'/' || b[2] == b'\\') { | |
| if b.len() >= 3 | |
| && ((b[0] >= b'a' && b[0] <= b'z') || (b[0] >= b'A' && b[0] <= b'Z')) | |
| && b[1] == b':' | |
| && (b[2] == b'/' || b[2] == b'\\') | |
| { |
fix #10665
Cursor sends paths in this format:
/c:/Users, I'm not sure why, but to my knowledge this prevents GitButler from working with Cursor Hooks on Windows.The fix just removes the slash.
I verified that after the fix
but cursor after-editandbut cursor stopare triggered and finish.