1use anyhow::Context;
2use globset::{GlobBuilder, GlobSet, GlobSetBuilder};
3use itertools::Itertools;
4use regex::Regex;
5use serde::{Deserialize, Serialize};
6use std::borrow::Cow;
7use std::cmp::Ordering;
8use std::error::Error;
9use std::fmt::{Display, Formatter};
10use std::mem;
11use std::path::StripPrefixError;
12use std::sync::{Arc, OnceLock};
13use std::{
14 ffi::OsStr,
15 path::{Path, PathBuf},
16 sync::LazyLock,
17};
18
19use crate::rel_path::RelPathBuf;
20use crate::{rel_path::RelPath, shell::ShellKind};
21
22static HOME_DIR: OnceLock<PathBuf> = OnceLock::new();
23
24/// Returns the path to the user's home directory.
25pub fn home_dir() -> &'static PathBuf {
26 HOME_DIR.get_or_init(|| {
27 if cfg!(any(test, feature = "test-support")) {
28 if cfg!(target_os = "macos") {
29 PathBuf::from("/Users/zed")
30 } else if cfg!(target_os = "windows") {
31 PathBuf::from("C:\\Users\\zed")
32 } else {
33 PathBuf::from("/home/zed")
34 }
35 } else {
36 dirs::home_dir().expect("failed to determine home directory")
37 }
38 })
39}
40
41pub trait PathExt {
42 /// Compacts a given file path by replacing the user's home directory
43 /// prefix with a tilde (`~`).
44 ///
45 /// # Returns
46 ///
47 /// * A `PathBuf` containing the compacted file path. If the input path
48 /// does not have the user's home directory prefix, or if we are not on
49 /// Linux or macOS, the original path is returned unchanged.
50 fn compact(&self) -> PathBuf;
51
52 /// Returns a file's extension or, if the file is hidden, its name without the leading dot
53 fn extension_or_hidden_file_name(&self) -> Option<&str>;
54
55 fn try_from_bytes<'a>(bytes: &'a [u8]) -> anyhow::Result<Self>
56 where
57 Self: From<&'a Path>,
58 {
59 #[cfg(unix)]
60 {
61 use std::os::unix::prelude::OsStrExt;
62 Ok(Self::from(Path::new(OsStr::from_bytes(bytes))))
63 }
64 #[cfg(windows)]
65 {
66 use tendril::fmt::{Format, WTF8};
67 WTF8::validate(bytes)
68 .then(|| {
69 // Safety: bytes are valid WTF-8 sequence.
70 Self::from(Path::new(unsafe {
71 OsStr::from_encoded_bytes_unchecked(bytes)
72 }))
73 })
74 .with_context(|| format!("Invalid WTF-8 sequence: {bytes:?}"))
75 }
76 }
77
78 /// Converts a local path to one that can be used inside of WSL.
79 /// Returns `None` if the path cannot be converted into a WSL one (network share).
80 fn local_to_wsl(&self) -> Option<PathBuf>;
81
82 /// Returns a file's "full" joined collection of extensions, in the case where a file does not
83 /// just have a singular extension but instead has multiple (e.g File.tar.gz, Component.stories.tsx)
84 ///
85 /// Will provide back the extensions joined together such as tar.gz or stories.tsx
86 fn multiple_extensions(&self) -> Option<String>;
87
88 /// Try to make a shell-safe representation of the path.
89 fn try_shell_safe(&self, shell_kind: ShellKind) -> anyhow::Result<String>;
90}
91
92impl<T: AsRef<Path>> PathExt for T {
93 fn compact(&self) -> PathBuf {
94 if cfg!(any(target_os = "linux", target_os = "freebsd")) || cfg!(target_os = "macos") {
95 match self.as_ref().strip_prefix(home_dir().as_path()) {
96 Ok(relative_path) => {
97 let mut shortened_path = PathBuf::new();
98 shortened_path.push("~");
99 shortened_path.push(relative_path);
100 shortened_path
101 }
102 Err(_) => self.as_ref().to_path_buf(),
103 }
104 } else {
105 self.as_ref().to_path_buf()
106 }
107 }
108
109 fn extension_or_hidden_file_name(&self) -> Option<&str> {
110 let path = self.as_ref();
111 let file_name = path.file_name()?.to_str()?;
112 if file_name.starts_with('.') {
113 return file_name.strip_prefix('.');
114 }
115
116 path.extension()
117 .and_then(|e| e.to_str())
118 .or_else(|| path.file_stem()?.to_str())
119 }
120
121 fn local_to_wsl(&self) -> Option<PathBuf> {
122 // quite sketchy to convert this back to path at the end, but a lot of functions only accept paths
123 // todo: ideally rework them..?
124 let mut new_path = std::ffi::OsString::new();
125 for component in self.as_ref().components() {
126 match component {
127 std::path::Component::Prefix(prefix) => {
128 let drive_letter = prefix.as_os_str().to_string_lossy().to_lowercase();
129 let drive_letter = drive_letter.strip_suffix(':')?;
130
131 new_path.push(format!("/mnt/{}", drive_letter));
132 }
133 std::path::Component::RootDir => {}
134 std::path::Component::CurDir => {
135 new_path.push("/.");
136 }
137 std::path::Component::ParentDir => {
138 new_path.push("/..");
139 }
140 std::path::Component::Normal(os_str) => {
141 new_path.push("/");
142 new_path.push(os_str);
143 }
144 }
145 }
146
147 Some(new_path.into())
148 }
149
150 fn multiple_extensions(&self) -> Option<String> {
151 let path = self.as_ref();
152 let file_name = path.file_name()?.to_str()?;
153
154 let parts: Vec<&str> = file_name
155 .split('.')
156 // Skip the part with the file name extension
157 .skip(1)
158 .collect();
159
160 if parts.len() < 2 {
161 return None;
162 }
163
164 Some(parts.into_iter().join("."))
165 }
166
167 fn try_shell_safe(&self, shell_kind: ShellKind) -> anyhow::Result<String> {
168 let path_str = self
169 .as_ref()
170 .to_str()
171 .with_context(|| "Path contains invalid UTF-8")?;
172 shell_kind
173 .try_quote(path_str)
174 .as_deref()
175 .map(ToOwned::to_owned)
176 .context("Failed to quote path")
177 }
178}
179
180pub fn path_ends_with(base: &Path, suffix: &Path) -> bool {
181 strip_path_suffix(base, suffix).is_some()
182}
183
184pub fn strip_path_suffix<'a>(base: &'a Path, suffix: &Path) -> Option<&'a Path> {
185 if let Some(remainder) = base
186 .as_os_str()
187 .as_encoded_bytes()
188 .strip_suffix(suffix.as_os_str().as_encoded_bytes())
189 {
190 if remainder
191 .last()
192 .is_none_or(|last_byte| std::path::is_separator(*last_byte as char))
193 {
194 let os_str = unsafe {
195 OsStr::from_encoded_bytes_unchecked(
196 &remainder[0..remainder.len().saturating_sub(1)],
197 )
198 };
199 return Some(Path::new(os_str));
200 }
201 }
202 None
203}
204
205/// In memory, this is identical to `Path`. On non-Windows conversions to this type are no-ops. On
206/// windows, these conversions sanitize UNC paths by removing the `\\\\?\\` prefix.
207#[derive(Eq, PartialEq, Hash, Ord, PartialOrd)]
208#[repr(transparent)]
209pub struct SanitizedPath(Path);
210
211impl SanitizedPath {
212 pub fn new<T: AsRef<Path> + ?Sized>(path: &T) -> &Self {
213 #[cfg(not(target_os = "windows"))]
214 return Self::unchecked_new(path.as_ref());
215
216 #[cfg(target_os = "windows")]
217 return Self::unchecked_new(dunce::simplified(path.as_ref()));
218 }
219
220 pub fn unchecked_new<T: AsRef<Path> + ?Sized>(path: &T) -> &Self {
221 // safe because `Path` and `SanitizedPath` have the same repr and Drop impl
222 unsafe { mem::transmute::<&Path, &Self>(path.as_ref()) }
223 }
224
225 pub fn from_arc(path: Arc<Path>) -> Arc<Self> {
226 // safe because `Path` and `SanitizedPath` have the same repr and Drop impl
227 #[cfg(not(target_os = "windows"))]
228 return unsafe { mem::transmute::<Arc<Path>, Arc<Self>>(path) };
229
230 #[cfg(target_os = "windows")]
231 {
232 let simplified = dunce::simplified(path.as_ref());
233 if simplified == path.as_ref() {
234 // safe because `Path` and `SanitizedPath` have the same repr and Drop impl
235 unsafe { mem::transmute::<Arc<Path>, Arc<Self>>(path) }
236 } else {
237 Self::unchecked_new(simplified).into()
238 }
239 }
240 }
241
242 pub fn new_arc<T: AsRef<Path> + ?Sized>(path: &T) -> Arc<Self> {
243 Self::new(path).into()
244 }
245
246 pub fn cast_arc(path: Arc<Self>) -> Arc<Path> {
247 // safe because `Path` and `SanitizedPath` have the same repr and Drop impl
248 unsafe { mem::transmute::<Arc<Self>, Arc<Path>>(path) }
249 }
250
251 pub fn cast_arc_ref(path: &Arc<Self>) -> &Arc<Path> {
252 // safe because `Path` and `SanitizedPath` have the same repr and Drop impl
253 unsafe { mem::transmute::<&Arc<Self>, &Arc<Path>>(path) }
254 }
255
256 pub fn starts_with(&self, prefix: &Self) -> bool {
257 self.0.starts_with(&prefix.0)
258 }
259
260 pub fn as_path(&self) -> &Path {
261 &self.0
262 }
263
264 pub fn file_name(&self) -> Option<&std::ffi::OsStr> {
265 self.0.file_name()
266 }
267
268 pub fn extension(&self) -> Option<&std::ffi::OsStr> {
269 self.0.extension()
270 }
271
272 pub fn join<P: AsRef<Path>>(&self, path: P) -> PathBuf {
273 self.0.join(path)
274 }
275
276 pub fn parent(&self) -> Option<&Self> {
277 self.0.parent().map(Self::unchecked_new)
278 }
279
280 pub fn strip_prefix(&self, base: &Self) -> Result<&Path, StripPrefixError> {
281 self.0.strip_prefix(base.as_path())
282 }
283
284 pub fn to_str(&self) -> Option<&str> {
285 self.0.to_str()
286 }
287
288 pub fn to_path_buf(&self) -> PathBuf {
289 self.0.to_path_buf()
290 }
291}
292
293impl std::fmt::Debug for SanitizedPath {
294 fn fmt(&self, formatter: &mut Formatter<'_>) -> std::fmt::Result {
295 std::fmt::Debug::fmt(&self.0, formatter)
296 }
297}
298
299impl Display for SanitizedPath {
300 fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
301 write!(f, "{}", self.0.display())
302 }
303}
304
305impl From<&SanitizedPath> for Arc<SanitizedPath> {
306 fn from(sanitized_path: &SanitizedPath) -> Self {
307 let path: Arc<Path> = sanitized_path.0.into();
308 // safe because `Path` and `SanitizedPath` have the same repr and Drop impl
309 unsafe { mem::transmute(path) }
310 }
311}
312
313impl From<&SanitizedPath> for PathBuf {
314 fn from(sanitized_path: &SanitizedPath) -> Self {
315 sanitized_path.as_path().into()
316 }
317}
318
319impl AsRef<Path> for SanitizedPath {
320 fn as_ref(&self) -> &Path {
321 &self.0
322 }
323}
324
325#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
326pub enum PathStyle {
327 Posix,
328 Windows,
329}
330
331impl PathStyle {
332 #[cfg(target_os = "windows")]
333 pub const fn local() -> Self {
334 PathStyle::Windows
335 }
336
337 #[cfg(not(target_os = "windows"))]
338 pub const fn local() -> Self {
339 PathStyle::Posix
340 }
341
342 #[inline]
343 pub fn primary_separator(&self) -> &'static str {
344 match self {
345 PathStyle::Posix => "/",
346 PathStyle::Windows => "\\",
347 }
348 }
349
350 pub fn separators(&self) -> &'static [&'static str] {
351 match self {
352 PathStyle::Posix => &["/"],
353 PathStyle::Windows => &["\\", "/"],
354 }
355 }
356
357 pub fn separators_ch(&self) -> &'static [char] {
358 match self {
359 PathStyle::Posix => &['/'],
360 PathStyle::Windows => &['\\', '/'],
361 }
362 }
363
364 pub fn is_windows(&self) -> bool {
365 *self == PathStyle::Windows
366 }
367
368 pub fn is_posix(&self) -> bool {
369 *self == PathStyle::Posix
370 }
371
372 pub fn join(self, left: impl AsRef<Path>, right: impl AsRef<Path>) -> Option<String> {
373 let right = right.as_ref().to_str()?;
374 if is_absolute(right, self) {
375 return None;
376 }
377 let left = left.as_ref().to_str()?;
378 if left.is_empty() {
379 Some(right.into())
380 } else {
381 Some(format!(
382 "{left}{}{right}",
383 if left.ends_with(self.primary_separator()) {
384 ""
385 } else {
386 self.primary_separator()
387 }
388 ))
389 }
390 }
391
392 pub fn split(self, path_like: &str) -> (Option<&str>, &str) {
393 let Some(pos) = path_like.rfind(self.primary_separator()) else {
394 return (None, path_like);
395 };
396 let filename_start = pos + self.primary_separator().len();
397 (
398 Some(&path_like[..filename_start]),
399 &path_like[filename_start..],
400 )
401 }
402
403 pub fn strip_prefix<'a>(
404 &self,
405 child: &'a Path,
406 parent: &'a Path,
407 ) -> Option<std::borrow::Cow<'a, RelPath>> {
408 let parent = parent.to_str()?;
409 if parent.is_empty() {
410 return RelPath::new(child, *self).ok();
411 }
412 let parent = self
413 .separators()
414 .iter()
415 .find_map(|sep| parent.strip_suffix(sep))
416 .unwrap_or(parent);
417 let child = child.to_str()?;
418 let stripped = child.strip_prefix(parent)?;
419 if let Some(relative) = self
420 .separators()
421 .iter()
422 .find_map(|sep| stripped.strip_prefix(sep))
423 {
424 RelPath::new(relative.as_ref(), *self).ok()
425 } else if stripped.is_empty() {
426 Some(Cow::Borrowed(RelPath::empty()))
427 } else {
428 None
429 }
430 }
431}
432
433#[derive(Debug, Clone)]
434pub struct RemotePathBuf {
435 style: PathStyle,
436 string: String,
437}
438
439impl RemotePathBuf {
440 pub fn new(string: String, style: PathStyle) -> Self {
441 Self { style, string }
442 }
443
444 pub fn from_str(path: &str, style: PathStyle) -> Self {
445 Self::new(path.to_string(), style)
446 }
447
448 pub fn path_style(&self) -> PathStyle {
449 self.style
450 }
451
452 pub fn to_proto(self) -> String {
453 self.string
454 }
455}
456
457impl Display for RemotePathBuf {
458 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
459 write!(f, "{}", self.string)
460 }
461}
462
463pub fn is_absolute(path_like: &str, path_style: PathStyle) -> bool {
464 path_like.starts_with('/')
465 || path_style == PathStyle::Windows
466 && (path_like.starts_with('\\')
467 || path_like
468 .chars()
469 .next()
470 .is_some_and(|c| c.is_ascii_alphabetic())
471 && path_like[1..]
472 .strip_prefix(':')
473 .is_some_and(|path| path.starts_with('/') || path.starts_with('\\')))
474}
475
476#[derive(Debug, PartialEq)]
477#[non_exhaustive]
478pub struct NormalizeError;
479
480impl Error for NormalizeError {}
481
482impl std::fmt::Display for NormalizeError {
483 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
484 f.write_str("parent reference `..` points outside of base directory")
485 }
486}
487
488/// Copied from stdlib where it's unstable.
489///
490/// Normalize a path, including `..` without traversing the filesystem.
491///
492/// Returns an error if normalization would leave leading `..` components.
493///
494/// <div class="warning">
495///
496/// This function always resolves `..` to the "lexical" parent.
497/// That is "a/b/../c" will always resolve to `a/c` which can change the meaning of the path.
498/// In particular, `a/c` and `a/b/../c` are distinct on many systems because `b` may be a symbolic link, so its parent isn't `a`.
499///
500/// </div>
501///
502/// [`path::absolute`](absolute) is an alternative that preserves `..`.
503/// Or [`Path::canonicalize`] can be used to resolve any `..` by querying the filesystem.
504pub fn normalize_lexically(path: &Path) -> Result<PathBuf, NormalizeError> {
505 use std::path::Component;
506
507 let mut lexical = PathBuf::new();
508 let mut iter = path.components().peekable();
509
510 // Find the root, if any, and add it to the lexical path.
511 // Here we treat the Windows path "C:\" as a single "root" even though
512 // `components` splits it into two: (Prefix, RootDir).
513 let root = match iter.peek() {
514 Some(Component::ParentDir) => return Err(NormalizeError),
515 Some(p @ Component::RootDir) | Some(p @ Component::CurDir) => {
516 lexical.push(p);
517 iter.next();
518 lexical.as_os_str().len()
519 }
520 Some(Component::Prefix(prefix)) => {
521 lexical.push(prefix.as_os_str());
522 iter.next();
523 if let Some(p @ Component::RootDir) = iter.peek() {
524 lexical.push(p);
525 iter.next();
526 }
527 lexical.as_os_str().len()
528 }
529 None => return Ok(PathBuf::new()),
530 Some(Component::Normal(_)) => 0,
531 };
532
533 for component in iter {
534 match component {
535 Component::RootDir => unreachable!(),
536 Component::Prefix(_) => return Err(NormalizeError),
537 Component::CurDir => continue,
538 Component::ParentDir => {
539 // It's an error if ParentDir causes us to go above the "root".
540 if lexical.as_os_str().len() == root {
541 return Err(NormalizeError);
542 } else {
543 lexical.pop();
544 }
545 }
546 Component::Normal(path) => lexical.push(path),
547 }
548 }
549 Ok(lexical)
550}
551
552/// A delimiter to use in `path_query:row_number:column_number` strings parsing.
553pub const FILE_ROW_COLUMN_DELIMITER: char = ':';
554
555const ROW_COL_CAPTURE_REGEX: &str = r"(?xs)
556 ([^\(]+)\:(?:
557 \((\d+)[,:](\d+)\) # filename:(row,column), filename:(row:column)
558 |
559 \((\d+)\)() # filename:(row)
560 )
561 |
562 ([^\(]+)(?:
563 \((\d+)[,:](\d+)\) # filename(row,column), filename(row:column)
564 |
565 \((\d+)\)() # filename(row)
566 )
567 |
568 (.+?)(?:
569 \:+(\d+)\:(\d+)\:*$ # filename:row:column
570 |
571 \:+(\d+)\:*()$ # filename:row
572 |
573 \:+()()$
574 )";
575
576/// A representation of a path-like string with optional row and column numbers.
577/// Matching values example: `te`, `test.rs:22`, `te:22:5`, `test.c(22)`, `test.c(22,5)`etc.
578#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Hash)]
579pub struct PathWithPosition {
580 pub path: PathBuf,
581 pub row: Option<u32>,
582 // Absent if row is absent.
583 pub column: Option<u32>,
584}
585
586impl PathWithPosition {
587 /// Returns a PathWithPosition from a path.
588 pub fn from_path(path: PathBuf) -> Self {
589 Self {
590 path,
591 row: None,
592 column: None,
593 }
594 }
595
596 /// Parses a string that possibly has `:row:column` or `(row, column)` suffix.
597 /// Parenthesis format is used by [MSBuild](https://learn.microsoft.com/en-us/visualstudio/msbuild/msbuild-diagnostic-format-for-tasks) compatible tools
598 /// Ignores trailing `:`s, so `test.rs:22:` is parsed as `test.rs:22`.
599 /// If the suffix parsing fails, the whole string is parsed as a path.
600 ///
601 /// Be mindful that `test_file:10:1:` is a valid posix filename.
602 /// `PathWithPosition` class assumes that the ending position-like suffix is **not** part of the filename.
603 ///
604 /// # Examples
605 ///
606 /// ```
607 /// # use util::paths::PathWithPosition;
608 /// # use std::path::PathBuf;
609 /// assert_eq!(PathWithPosition::parse_str("test_file"), PathWithPosition {
610 /// path: PathBuf::from("test_file"),
611 /// row: None,
612 /// column: None,
613 /// });
614 /// assert_eq!(PathWithPosition::parse_str("test_file:10"), PathWithPosition {
615 /// path: PathBuf::from("test_file"),
616 /// row: Some(10),
617 /// column: None,
618 /// });
619 /// assert_eq!(PathWithPosition::parse_str("test_file.rs"), PathWithPosition {
620 /// path: PathBuf::from("test_file.rs"),
621 /// row: None,
622 /// column: None,
623 /// });
624 /// assert_eq!(PathWithPosition::parse_str("test_file.rs:1"), PathWithPosition {
625 /// path: PathBuf::from("test_file.rs"),
626 /// row: Some(1),
627 /// column: None,
628 /// });
629 /// assert_eq!(PathWithPosition::parse_str("test_file.rs:1:2"), PathWithPosition {
630 /// path: PathBuf::from("test_file.rs"),
631 /// row: Some(1),
632 /// column: Some(2),
633 /// });
634 /// ```
635 ///
636 /// # Expected parsing results when encounter ill-formatted inputs.
637 /// ```
638 /// # use util::paths::PathWithPosition;
639 /// # use std::path::PathBuf;
640 /// assert_eq!(PathWithPosition::parse_str("test_file.rs:a"), PathWithPosition {
641 /// path: PathBuf::from("test_file.rs:a"),
642 /// row: None,
643 /// column: None,
644 /// });
645 /// assert_eq!(PathWithPosition::parse_str("test_file.rs:a:b"), PathWithPosition {
646 /// path: PathBuf::from("test_file.rs:a:b"),
647 /// row: None,
648 /// column: None,
649 /// });
650 /// assert_eq!(PathWithPosition::parse_str("test_file.rs"), PathWithPosition {
651 /// path: PathBuf::from("test_file.rs"),
652 /// row: None,
653 /// column: None,
654 /// });
655 /// assert_eq!(PathWithPosition::parse_str("test_file.rs::1"), PathWithPosition {
656 /// path: PathBuf::from("test_file.rs"),
657 /// row: Some(1),
658 /// column: None,
659 /// });
660 /// assert_eq!(PathWithPosition::parse_str("test_file.rs:1::"), PathWithPosition {
661 /// path: PathBuf::from("test_file.rs"),
662 /// row: Some(1),
663 /// column: None,
664 /// });
665 /// assert_eq!(PathWithPosition::parse_str("test_file.rs::1:2"), PathWithPosition {
666 /// path: PathBuf::from("test_file.rs"),
667 /// row: Some(1),
668 /// column: Some(2),
669 /// });
670 /// assert_eq!(PathWithPosition::parse_str("test_file.rs:1::2"), PathWithPosition {
671 /// path: PathBuf::from("test_file.rs:1"),
672 /// row: Some(2),
673 /// column: None,
674 /// });
675 /// assert_eq!(PathWithPosition::parse_str("test_file.rs:1:2:3"), PathWithPosition {
676 /// path: PathBuf::from("test_file.rs:1"),
677 /// row: Some(2),
678 /// column: Some(3),
679 /// });
680 /// ```
681 pub fn parse_str(s: &str) -> Self {
682 let trimmed = s.trim();
683 let path = Path::new(trimmed);
684 let Some(maybe_file_name_with_row_col) = path.file_name().unwrap_or_default().to_str()
685 else {
686 return Self {
687 path: Path::new(s).to_path_buf(),
688 row: None,
689 column: None,
690 };
691 };
692 if maybe_file_name_with_row_col.is_empty() {
693 return Self {
694 path: Path::new(s).to_path_buf(),
695 row: None,
696 column: None,
697 };
698 }
699
700 // Let's avoid repeated init cost on this. It is subject to thread contention, but
701 // so far this code isn't called from multiple hot paths. Getting contention here
702 // in the future seems unlikely.
703 static SUFFIX_RE: LazyLock<Regex> =
704 LazyLock::new(|| Regex::new(ROW_COL_CAPTURE_REGEX).unwrap());
705 match SUFFIX_RE
706 .captures(maybe_file_name_with_row_col)
707 .map(|caps| caps.extract())
708 {
709 Some((_, [file_name, maybe_row, maybe_column])) => {
710 let row = maybe_row.parse::<u32>().ok();
711 let column = maybe_column.parse::<u32>().ok();
712
713 let (_, suffix) = trimmed.split_once(file_name).unwrap();
714 let path_without_suffix = &trimmed[..trimmed.len() - suffix.len()];
715
716 Self {
717 path: Path::new(path_without_suffix).to_path_buf(),
718 row,
719 column,
720 }
721 }
722 None => {
723 // The `ROW_COL_CAPTURE_REGEX` deals with separated digits only,
724 // but in reality there could be `foo/bar.py:22:in` inputs which we want to match too.
725 // The regex mentioned is not very extendable with "digit or random string" checks, so do this here instead.
726 let delimiter = ':';
727 let mut path_parts = s
728 .rsplitn(3, delimiter)
729 .collect::<Vec<_>>()
730 .into_iter()
731 .rev()
732 .fuse();
733 let mut path_string = path_parts.next().expect("rsplitn should have the rest of the string as its last parameter that we reversed").to_owned();
734 let mut row = None;
735 let mut column = None;
736 if let Some(maybe_row) = path_parts.next() {
737 if let Ok(parsed_row) = maybe_row.parse::<u32>() {
738 row = Some(parsed_row);
739 if let Some(parsed_column) = path_parts
740 .next()
741 .and_then(|maybe_col| maybe_col.parse::<u32>().ok())
742 {
743 column = Some(parsed_column);
744 }
745 } else {
746 path_string.push(delimiter);
747 path_string.push_str(maybe_row);
748 }
749 }
750 for split in path_parts {
751 path_string.push(delimiter);
752 path_string.push_str(split);
753 }
754
755 Self {
756 path: PathBuf::from(path_string),
757 row,
758 column,
759 }
760 }
761 }
762 }
763
764 pub fn map_path<E>(
765 self,
766 mapping: impl FnOnce(PathBuf) -> Result<PathBuf, E>,
767 ) -> Result<PathWithPosition, E> {
768 Ok(PathWithPosition {
769 path: mapping(self.path)?,
770 row: self.row,
771 column: self.column,
772 })
773 }
774
775 pub fn to_string(&self, path_to_string: impl Fn(&PathBuf) -> String) -> String {
776 let path_string = path_to_string(&self.path);
777 if let Some(row) = self.row {
778 if let Some(column) = self.column {
779 format!("{path_string}:{row}:{column}")
780 } else {
781 format!("{path_string}:{row}")
782 }
783 } else {
784 path_string
785 }
786 }
787}
788
789#[derive(Clone, Debug)]
790pub struct PathMatcher {
791 sources: Vec<(String, RelPathBuf, /*trailing separator*/ bool)>,
792 glob: GlobSet,
793 path_style: PathStyle,
794}
795
796impl PartialEq for PathMatcher {
797 fn eq(&self, other: &Self) -> bool {
798 self.sources.eq(&other.sources)
799 }
800}
801
802impl Eq for PathMatcher {}
803
804impl PathMatcher {
805 pub fn new(
806 globs: impl IntoIterator<Item = impl AsRef<str>>,
807 path_style: PathStyle,
808 ) -> Result<Self, globset::Error> {
809 let globs = globs
810 .into_iter()
811 .map(|as_str| {
812 GlobBuilder::new(as_str.as_ref())
813 .backslash_escape(path_style.is_posix())
814 .build()
815 })
816 .collect::<Result<Vec<_>, _>>()?;
817 let sources = globs
818 .iter()
819 .filter_map(|glob| {
820 let glob = glob.glob();
821 Some((
822 glob.to_string(),
823 RelPath::new(&glob.as_ref(), path_style)
824 .ok()
825 .map(std::borrow::Cow::into_owned)?,
826 glob.ends_with(path_style.separators_ch()),
827 ))
828 })
829 .collect();
830 let mut glob_builder = GlobSetBuilder::new();
831 for single_glob in globs {
832 glob_builder.add(single_glob);
833 }
834 let glob = glob_builder.build()?;
835 Ok(PathMatcher {
836 glob,
837 sources,
838 path_style,
839 })
840 }
841
842 pub fn sources(&self) -> impl Iterator<Item = &str> + Clone {
843 self.sources.iter().map(|(source, ..)| source.as_str())
844 }
845
846 pub fn is_match<P: AsRef<RelPath>>(&self, other: P) -> bool {
847 if self.sources.iter().any(|(_, source, _)| {
848 other.as_ref().starts_with(source) || other.as_ref().ends_with(source)
849 }) {
850 return true;
851 }
852 let other_path = other.as_ref().display(self.path_style);
853
854 if self.glob.is_match(&*other_path) {
855 return true;
856 }
857
858 self.glob
859 .is_match(other_path.into_owned() + self.path_style.primary_separator())
860 }
861}
862
863impl Default for PathMatcher {
864 fn default() -> Self {
865 Self {
866 path_style: PathStyle::local(),
867 glob: GlobSet::empty(),
868 sources: vec![],
869 }
870 }
871}
872
873/// Compares two sequences of consecutive digits for natural sorting.
874///
875/// This function is a core component of natural sorting that handles numeric comparison
876/// in a way that feels natural to humans. It extracts and compares consecutive digit
877/// sequences from two iterators, handling various cases like leading zeros and very large numbers.
878///
879/// # Behavior
880///
881/// The function implements the following comparison rules:
882/// 1. Different numeric values: Compares by actual numeric value (e.g., "2" < "10")
883/// 2. Leading zeros: When values are equal, longer sequence wins (e.g., "002" > "2")
884/// 3. Large numbers: Falls back to string comparison for numbers that would overflow u128
885///
886/// # Examples
887///
888/// ```text
889/// "1" vs "2" -> Less (different values)
890/// "2" vs "10" -> Less (numeric comparison)
891/// "002" vs "2" -> Greater (leading zeros)
892/// "10" vs "010" -> Less (leading zeros)
893/// "999..." vs "1000..." -> Less (large number comparison)
894/// ```
895///
896/// # Implementation Details
897///
898/// 1. Extracts consecutive digits into strings
899/// 2. Compares sequence lengths for leading zero handling
900/// 3. For equal lengths, compares digit by digit
901/// 4. For different lengths:
902/// - Attempts numeric comparison first (for numbers up to 2^128 - 1)
903/// - Falls back to string comparison if numbers would overflow
904///
905/// The function advances both iterators past their respective numeric sequences,
906/// regardless of the comparison result.
907fn compare_numeric_segments<I>(
908 a_iter: &mut std::iter::Peekable<I>,
909 b_iter: &mut std::iter::Peekable<I>,
910) -> Ordering
911where
912 I: Iterator<Item = char>,
913{
914 // Collect all consecutive digits into strings
915 let mut a_num_str = String::new();
916 let mut b_num_str = String::new();
917
918 while let Some(&c) = a_iter.peek() {
919 if !c.is_ascii_digit() {
920 break;
921 }
922
923 a_num_str.push(c);
924 a_iter.next();
925 }
926
927 while let Some(&c) = b_iter.peek() {
928 if !c.is_ascii_digit() {
929 break;
930 }
931
932 b_num_str.push(c);
933 b_iter.next();
934 }
935
936 // First compare lengths (handle leading zeros)
937 match a_num_str.len().cmp(&b_num_str.len()) {
938 Ordering::Equal => {
939 // Same length, compare digit by digit
940 match a_num_str.cmp(&b_num_str) {
941 Ordering::Equal => Ordering::Equal,
942 ordering => ordering,
943 }
944 }
945
946 // Different lengths but same value means leading zeros
947 ordering => {
948 // Try parsing as numbers first
949 if let (Ok(a_val), Ok(b_val)) = (a_num_str.parse::<u128>(), b_num_str.parse::<u128>()) {
950 match a_val.cmp(&b_val) {
951 Ordering::Equal => ordering, // Same value, longer one is greater (leading zeros)
952 ord => ord,
953 }
954 } else {
955 // If parsing fails (overflow), compare as strings
956 a_num_str.cmp(&b_num_str)
957 }
958 }
959 }
960}
961
962/// Performs natural sorting comparison between two strings.
963///
964/// Natural sorting is an ordering that handles numeric sequences in a way that matches human expectations.
965/// For example, "file2" comes before "file10" (unlike standard lexicographic sorting).
966///
967/// # Characteristics
968///
969/// * Case-sensitive with lowercase priority: When comparing same letters, lowercase comes before uppercase
970/// * Numbers are compared by numeric value, not character by character
971/// * Leading zeros affect ordering when numeric values are equal
972/// * Can handle numbers larger than u128::MAX (falls back to string comparison)
973/// * When strings are equal case-insensitively, lowercase is prioritized (lowercase < uppercase)
974///
975/// # Algorithm
976///
977/// The function works by:
978/// 1. Processing strings character by character in a case-insensitive manner
979/// 2. When encountering digits, treating consecutive digits as a single number
980/// 3. Comparing numbers by their numeric value rather than lexicographically
981/// 4. For non-numeric characters, using case-insensitive comparison
982/// 5. If everything is equal case-insensitively, using case-sensitive comparison as final tie-breaker
983pub fn natural_sort(a: &str, b: &str) -> Ordering {
984 let mut a_iter = a.chars().peekable();
985 let mut b_iter = b.chars().peekable();
986
987 loop {
988 match (a_iter.peek(), b_iter.peek()) {
989 (None, None) => {
990 return b.cmp(a);
991 }
992 (None, _) => return Ordering::Less,
993 (_, None) => return Ordering::Greater,
994 (Some(&a_char), Some(&b_char)) => {
995 if a_char.is_ascii_digit() && b_char.is_ascii_digit() {
996 match compare_numeric_segments(&mut a_iter, &mut b_iter) {
997 Ordering::Equal => continue,
998 ordering => return ordering,
999 }
1000 } else {
1001 match a_char
1002 .to_ascii_lowercase()
1003 .cmp(&b_char.to_ascii_lowercase())
1004 {
1005 Ordering::Equal => {
1006 a_iter.next();
1007 b_iter.next();
1008 }
1009 ordering => return ordering,
1010 }
1011 }
1012 }
1013 }
1014 }
1015}
1016
1017/// Case-insensitive natural sort without applying the final lowercase/uppercase tie-breaker.
1018/// This is useful when comparing individual path components where we want to keep walking
1019/// deeper components before deciding on casing.
1020fn natural_sort_no_tiebreak(a: &str, b: &str) -> Ordering {
1021 if a.eq_ignore_ascii_case(b) {
1022 Ordering::Equal
1023 } else {
1024 natural_sort(a, b)
1025 }
1026}
1027
1028fn stem_and_extension(filename: &str) -> (Option<&str>, Option<&str>) {
1029 if filename.is_empty() {
1030 return (None, None);
1031 }
1032
1033 match filename.rsplit_once('.') {
1034 // Case 1: No dot was found. The entire name is the stem.
1035 None => (Some(filename), None),
1036
1037 // Case 2: A dot was found.
1038 Some((before, after)) => {
1039 // This is the crucial check for dotfiles like ".bashrc".
1040 // If `before` is empty, the dot was the first character.
1041 // In that case, we revert to the "whole name is the stem" logic.
1042 if before.is_empty() {
1043 (Some(filename), None)
1044 } else {
1045 // Otherwise, we have a standard stem and extension.
1046 (Some(before), Some(after))
1047 }
1048 }
1049 }
1050}
1051
1052pub fn compare_rel_paths(
1053 (path_a, a_is_file): (&RelPath, bool),
1054 (path_b, b_is_file): (&RelPath, bool),
1055) -> Ordering {
1056 let mut components_a = path_a.components();
1057 let mut components_b = path_b.components();
1058 loop {
1059 match (components_a.next(), components_b.next()) {
1060 (Some(component_a), Some(component_b)) => {
1061 let a_is_file = a_is_file && components_a.rest().is_empty();
1062 let b_is_file = b_is_file && components_b.rest().is_empty();
1063
1064 let ordering = a_is_file.cmp(&b_is_file).then_with(|| {
1065 let (a_stem, a_extension) = a_is_file
1066 .then(|| stem_and_extension(component_a))
1067 .unwrap_or_default();
1068 let path_string_a = if a_is_file { a_stem } else { Some(component_a) };
1069
1070 let (b_stem, b_extension) = b_is_file
1071 .then(|| stem_and_extension(component_b))
1072 .unwrap_or_default();
1073 let path_string_b = if b_is_file { b_stem } else { Some(component_b) };
1074
1075 let compare_components = match (path_string_a, path_string_b) {
1076 (Some(a), Some(b)) => natural_sort(&a, &b),
1077 (Some(_), None) => Ordering::Greater,
1078 (None, Some(_)) => Ordering::Less,
1079 (None, None) => Ordering::Equal,
1080 };
1081
1082 compare_components.then_with(|| {
1083 if a_is_file && b_is_file {
1084 let ext_a = a_extension.unwrap_or_default();
1085 let ext_b = b_extension.unwrap_or_default();
1086 ext_a.cmp(ext_b)
1087 } else {
1088 Ordering::Equal
1089 }
1090 })
1091 });
1092
1093 if !ordering.is_eq() {
1094 return ordering;
1095 }
1096 }
1097 (Some(_), None) => break Ordering::Greater,
1098 (None, Some(_)) => break Ordering::Less,
1099 (None, None) => break Ordering::Equal,
1100 }
1101 }
1102}
1103
1104/// Compare two relative paths with mixed files and directories using
1105/// case-insensitive natural sorting. For example, "Apple", "aardvark.txt",
1106/// and "Zebra" would be sorted as: aardvark.txt, Apple, Zebra
1107/// (case-insensitive alphabetical).
1108pub fn compare_rel_paths_mixed(
1109 (path_a, a_is_file): (&RelPath, bool),
1110 (path_b, b_is_file): (&RelPath, bool),
1111) -> Ordering {
1112 let original_paths_equal = std::ptr::eq(path_a, path_b) || path_a == path_b;
1113 let mut components_a = path_a.components();
1114 let mut components_b = path_b.components();
1115
1116 loop {
1117 match (components_a.next(), components_b.next()) {
1118 (Some(component_a), Some(component_b)) => {
1119 let a_leaf_file = a_is_file && components_a.rest().is_empty();
1120 let b_leaf_file = b_is_file && components_b.rest().is_empty();
1121
1122 let (a_stem, a_ext) = a_leaf_file
1123 .then(|| stem_and_extension(component_a))
1124 .unwrap_or_default();
1125 let (b_stem, b_ext) = b_leaf_file
1126 .then(|| stem_and_extension(component_b))
1127 .unwrap_or_default();
1128 let a_key = if a_leaf_file {
1129 a_stem
1130 } else {
1131 Some(component_a)
1132 };
1133 let b_key = if b_leaf_file {
1134 b_stem
1135 } else {
1136 Some(component_b)
1137 };
1138
1139 let ordering = match (a_key, b_key) {
1140 (Some(a), Some(b)) => natural_sort_no_tiebreak(a, b)
1141 .then_with(|| match (a_leaf_file, b_leaf_file) {
1142 (true, false) if a == b => Ordering::Greater,
1143 (false, true) if a == b => Ordering::Less,
1144 _ => Ordering::Equal,
1145 })
1146 .then_with(|| {
1147 if a_leaf_file && b_leaf_file {
1148 let a_ext_str = a_ext.unwrap_or_default().to_lowercase();
1149 let b_ext_str = b_ext.unwrap_or_default().to_lowercase();
1150 b_ext_str.cmp(&a_ext_str)
1151 } else {
1152 Ordering::Equal
1153 }
1154 }),
1155 (Some(_), None) => Ordering::Greater,
1156 (None, Some(_)) => Ordering::Less,
1157 (None, None) => Ordering::Equal,
1158 };
1159
1160 if !ordering.is_eq() {
1161 return ordering;
1162 }
1163 }
1164 (Some(_), None) => return Ordering::Greater,
1165 (None, Some(_)) => return Ordering::Less,
1166 (None, None) => {
1167 // Deterministic tie-break: use natural sort to prefer lowercase when paths
1168 // are otherwise equal but still differ in casing.
1169 if !original_paths_equal {
1170 return natural_sort(path_a.as_unix_str(), path_b.as_unix_str());
1171 }
1172 return Ordering::Equal;
1173 }
1174 }
1175 }
1176}
1177
1178/// Compare two relative paths with files before directories using
1179/// case-insensitive natural sorting. At each directory level, all files
1180/// are sorted before all directories, with case-insensitive alphabetical
1181/// ordering within each group.
1182pub fn compare_rel_paths_files_first(
1183 (path_a, a_is_file): (&RelPath, bool),
1184 (path_b, b_is_file): (&RelPath, bool),
1185) -> Ordering {
1186 let original_paths_equal = std::ptr::eq(path_a, path_b) || path_a == path_b;
1187 let mut components_a = path_a.components();
1188 let mut components_b = path_b.components();
1189
1190 loop {
1191 match (components_a.next(), components_b.next()) {
1192 (Some(component_a), Some(component_b)) => {
1193 let a_leaf_file = a_is_file && components_a.rest().is_empty();
1194 let b_leaf_file = b_is_file && components_b.rest().is_empty();
1195
1196 let (a_stem, a_ext) = a_leaf_file
1197 .then(|| stem_and_extension(component_a))
1198 .unwrap_or_default();
1199 let (b_stem, b_ext) = b_leaf_file
1200 .then(|| stem_and_extension(component_b))
1201 .unwrap_or_default();
1202 let a_key = if a_leaf_file {
1203 a_stem
1204 } else {
1205 Some(component_a)
1206 };
1207 let b_key = if b_leaf_file {
1208 b_stem
1209 } else {
1210 Some(component_b)
1211 };
1212
1213 let ordering = match (a_key, b_key) {
1214 (Some(a), Some(b)) => {
1215 if a_leaf_file && !b_leaf_file {
1216 Ordering::Less
1217 } else if !a_leaf_file && b_leaf_file {
1218 Ordering::Greater
1219 } else {
1220 natural_sort_no_tiebreak(a, b).then_with(|| {
1221 if a_leaf_file && b_leaf_file {
1222 let a_ext_str = a_ext.unwrap_or_default().to_lowercase();
1223 let b_ext_str = b_ext.unwrap_or_default().to_lowercase();
1224 a_ext_str.cmp(&b_ext_str)
1225 } else {
1226 Ordering::Equal
1227 }
1228 })
1229 }
1230 }
1231 (Some(_), None) => Ordering::Greater,
1232 (None, Some(_)) => Ordering::Less,
1233 (None, None) => Ordering::Equal,
1234 };
1235
1236 if !ordering.is_eq() {
1237 return ordering;
1238 }
1239 }
1240 (Some(_), None) => return Ordering::Greater,
1241 (None, Some(_)) => return Ordering::Less,
1242 (None, None) => {
1243 // Deterministic tie-break: use natural sort to prefer lowercase when paths
1244 // are otherwise equal but still differ in casing.
1245 if !original_paths_equal {
1246 return natural_sort(path_a.as_unix_str(), path_b.as_unix_str());
1247 }
1248 return Ordering::Equal;
1249 }
1250 }
1251 }
1252}
1253
1254pub fn compare_paths(
1255 (path_a, a_is_file): (&Path, bool),
1256 (path_b, b_is_file): (&Path, bool),
1257) -> Ordering {
1258 let mut components_a = path_a.components().peekable();
1259 let mut components_b = path_b.components().peekable();
1260
1261 loop {
1262 match (components_a.next(), components_b.next()) {
1263 (Some(component_a), Some(component_b)) => {
1264 let a_is_file = components_a.peek().is_none() && a_is_file;
1265 let b_is_file = components_b.peek().is_none() && b_is_file;
1266
1267 let ordering = a_is_file.cmp(&b_is_file).then_with(|| {
1268 let path_a = Path::new(component_a.as_os_str());
1269 let path_string_a = if a_is_file {
1270 path_a.file_stem()
1271 } else {
1272 path_a.file_name()
1273 }
1274 .map(|s| s.to_string_lossy());
1275
1276 let path_b = Path::new(component_b.as_os_str());
1277 let path_string_b = if b_is_file {
1278 path_b.file_stem()
1279 } else {
1280 path_b.file_name()
1281 }
1282 .map(|s| s.to_string_lossy());
1283
1284 let compare_components = match (path_string_a, path_string_b) {
1285 (Some(a), Some(b)) => natural_sort(&a, &b),
1286 (Some(_), None) => Ordering::Greater,
1287 (None, Some(_)) => Ordering::Less,
1288 (None, None) => Ordering::Equal,
1289 };
1290
1291 compare_components.then_with(|| {
1292 if a_is_file && b_is_file {
1293 let ext_a = path_a.extension().unwrap_or_default();
1294 let ext_b = path_b.extension().unwrap_or_default();
1295 ext_a.cmp(ext_b)
1296 } else {
1297 Ordering::Equal
1298 }
1299 })
1300 });
1301
1302 if !ordering.is_eq() {
1303 return ordering;
1304 }
1305 }
1306 (Some(_), None) => break Ordering::Greater,
1307 (None, Some(_)) => break Ordering::Less,
1308 (None, None) => break Ordering::Equal,
1309 }
1310 }
1311}
1312
1313#[derive(Debug, Clone, PartialEq, Eq)]
1314pub struct WslPath {
1315 pub distro: String,
1316
1317 // the reason this is an OsString and not any of the path types is that it needs to
1318 // represent a unix path (with '/' separators) on windows. `from_path` does this by
1319 // manually constructing it from the path components of a given windows path.
1320 pub path: std::ffi::OsString,
1321}
1322
1323impl WslPath {
1324 pub fn from_path<P: AsRef<Path>>(path: P) -> Option<WslPath> {
1325 if cfg!(not(target_os = "windows")) {
1326 return None;
1327 }
1328 use std::{
1329 ffi::OsString,
1330 path::{Component, Prefix},
1331 };
1332
1333 let mut components = path.as_ref().components();
1334 let Some(Component::Prefix(prefix)) = components.next() else {
1335 return None;
1336 };
1337 let (server, distro) = match prefix.kind() {
1338 Prefix::UNC(server, distro) => (server, distro),
1339 Prefix::VerbatimUNC(server, distro) => (server, distro),
1340 _ => return None,
1341 };
1342 let Some(Component::RootDir) = components.next() else {
1343 return None;
1344 };
1345
1346 let server_str = server.to_string_lossy();
1347 if server_str == "wsl.localhost" || server_str == "wsl$" {
1348 let mut result = OsString::from("");
1349 for c in components {
1350 use Component::*;
1351 match c {
1352 Prefix(p) => unreachable!("got {p:?}, but already stripped prefix"),
1353 RootDir => unreachable!("got root dir, but already stripped root"),
1354 CurDir => continue,
1355 ParentDir => result.push("/.."),
1356 Normal(s) => {
1357 result.push("/");
1358 result.push(s);
1359 }
1360 }
1361 }
1362 if result.is_empty() {
1363 result.push("/");
1364 }
1365 Some(WslPath {
1366 distro: distro.to_string_lossy().to_string(),
1367 path: result,
1368 })
1369 } else {
1370 None
1371 }
1372 }
1373}
1374
1375#[cfg(test)]
1376mod tests {
1377 use crate::rel_path::rel_path;
1378
1379 use super::*;
1380 use util_macros::perf;
1381
1382 #[perf]
1383 fn compare_paths_with_dots() {
1384 let mut paths = vec![
1385 (Path::new("test_dirs"), false),
1386 (Path::new("test_dirs/1.46"), false),
1387 (Path::new("test_dirs/1.46/bar_1"), true),
1388 (Path::new("test_dirs/1.46/bar_2"), true),
1389 (Path::new("test_dirs/1.45"), false),
1390 (Path::new("test_dirs/1.45/foo_2"), true),
1391 (Path::new("test_dirs/1.45/foo_1"), true),
1392 ];
1393 paths.sort_by(|&a, &b| compare_paths(a, b));
1394 assert_eq!(
1395 paths,
1396 vec![
1397 (Path::new("test_dirs"), false),
1398 (Path::new("test_dirs/1.45"), false),
1399 (Path::new("test_dirs/1.45/foo_1"), true),
1400 (Path::new("test_dirs/1.45/foo_2"), true),
1401 (Path::new("test_dirs/1.46"), false),
1402 (Path::new("test_dirs/1.46/bar_1"), true),
1403 (Path::new("test_dirs/1.46/bar_2"), true),
1404 ]
1405 );
1406 let mut paths = vec![
1407 (Path::new("root1/one.txt"), true),
1408 (Path::new("root1/one.two.txt"), true),
1409 ];
1410 paths.sort_by(|&a, &b| compare_paths(a, b));
1411 assert_eq!(
1412 paths,
1413 vec![
1414 (Path::new("root1/one.txt"), true),
1415 (Path::new("root1/one.two.txt"), true),
1416 ]
1417 );
1418 }
1419
1420 #[perf]
1421 fn compare_paths_with_same_name_different_extensions() {
1422 let mut paths = vec![
1423 (Path::new("test_dirs/file.rs"), true),
1424 (Path::new("test_dirs/file.txt"), true),
1425 (Path::new("test_dirs/file.md"), true),
1426 (Path::new("test_dirs/file"), true),
1427 (Path::new("test_dirs/file.a"), true),
1428 ];
1429 paths.sort_by(|&a, &b| compare_paths(a, b));
1430 assert_eq!(
1431 paths,
1432 vec![
1433 (Path::new("test_dirs/file"), true),
1434 (Path::new("test_dirs/file.a"), true),
1435 (Path::new("test_dirs/file.md"), true),
1436 (Path::new("test_dirs/file.rs"), true),
1437 (Path::new("test_dirs/file.txt"), true),
1438 ]
1439 );
1440 }
1441
1442 #[perf]
1443 fn compare_paths_case_semi_sensitive() {
1444 let mut paths = vec![
1445 (Path::new("test_DIRS"), false),
1446 (Path::new("test_DIRS/foo_1"), true),
1447 (Path::new("test_DIRS/foo_2"), true),
1448 (Path::new("test_DIRS/bar"), true),
1449 (Path::new("test_DIRS/BAR"), true),
1450 (Path::new("test_dirs"), false),
1451 (Path::new("test_dirs/foo_1"), true),
1452 (Path::new("test_dirs/foo_2"), true),
1453 (Path::new("test_dirs/bar"), true),
1454 (Path::new("test_dirs/BAR"), true),
1455 ];
1456 paths.sort_by(|&a, &b| compare_paths(a, b));
1457 assert_eq!(
1458 paths,
1459 vec![
1460 (Path::new("test_dirs"), false),
1461 (Path::new("test_dirs/bar"), true),
1462 (Path::new("test_dirs/BAR"), true),
1463 (Path::new("test_dirs/foo_1"), true),
1464 (Path::new("test_dirs/foo_2"), true),
1465 (Path::new("test_DIRS"), false),
1466 (Path::new("test_DIRS/bar"), true),
1467 (Path::new("test_DIRS/BAR"), true),
1468 (Path::new("test_DIRS/foo_1"), true),
1469 (Path::new("test_DIRS/foo_2"), true),
1470 ]
1471 );
1472 }
1473
1474 #[perf]
1475 fn compare_paths_mixed_case_numeric_ordering() {
1476 let mut entries = [
1477 (Path::new(".config"), false),
1478 (Path::new("Dir1"), false),
1479 (Path::new("dir01"), false),
1480 (Path::new("dir2"), false),
1481 (Path::new("Dir02"), false),
1482 (Path::new("dir10"), false),
1483 (Path::new("Dir10"), false),
1484 ];
1485
1486 entries.sort_by(|&a, &b| compare_paths(a, b));
1487
1488 let ordered: Vec<&str> = entries
1489 .iter()
1490 .map(|(path, _)| path.to_str().unwrap())
1491 .collect();
1492
1493 assert_eq!(
1494 ordered,
1495 vec![
1496 ".config", "Dir1", "dir01", "dir2", "Dir02", "dir10", "Dir10"
1497 ]
1498 );
1499 }
1500
1501 #[perf]
1502 fn compare_rel_paths_mixed_case_insensitive() {
1503 // Test that mixed mode is case-insensitive
1504 let mut paths = vec![
1505 (RelPath::unix("zebra.txt").unwrap(), true),
1506 (RelPath::unix("Apple").unwrap(), false),
1507 (RelPath::unix("banana.rs").unwrap(), true),
1508 (RelPath::unix("Carrot").unwrap(), false),
1509 (RelPath::unix("aardvark.txt").unwrap(), true),
1510 ];
1511 paths.sort_by(|&a, &b| compare_rel_paths_mixed(a, b));
1512 // Case-insensitive: aardvark < Apple < banana < Carrot < zebra
1513 assert_eq!(
1514 paths,
1515 vec![
1516 (RelPath::unix("aardvark.txt").unwrap(), true),
1517 (RelPath::unix("Apple").unwrap(), false),
1518 (RelPath::unix("banana.rs").unwrap(), true),
1519 (RelPath::unix("Carrot").unwrap(), false),
1520 (RelPath::unix("zebra.txt").unwrap(), true),
1521 ]
1522 );
1523 }
1524
1525 #[perf]
1526 fn compare_rel_paths_files_first_basic() {
1527 // Test that files come before directories
1528 let mut paths = vec![
1529 (RelPath::unix("zebra.txt").unwrap(), true),
1530 (RelPath::unix("Apple").unwrap(), false),
1531 (RelPath::unix("banana.rs").unwrap(), true),
1532 (RelPath::unix("Carrot").unwrap(), false),
1533 (RelPath::unix("aardvark.txt").unwrap(), true),
1534 ];
1535 paths.sort_by(|&a, &b| compare_rel_paths_files_first(a, b));
1536 // Files first (case-insensitive), then directories (case-insensitive)
1537 assert_eq!(
1538 paths,
1539 vec![
1540 (RelPath::unix("aardvark.txt").unwrap(), true),
1541 (RelPath::unix("banana.rs").unwrap(), true),
1542 (RelPath::unix("zebra.txt").unwrap(), true),
1543 (RelPath::unix("Apple").unwrap(), false),
1544 (RelPath::unix("Carrot").unwrap(), false),
1545 ]
1546 );
1547 }
1548
1549 #[perf]
1550 fn compare_rel_paths_files_first_case_insensitive() {
1551 // Test case-insensitive sorting within files and directories
1552 let mut paths = vec![
1553 (RelPath::unix("Zebra.txt").unwrap(), true),
1554 (RelPath::unix("apple").unwrap(), false),
1555 (RelPath::unix("Banana.rs").unwrap(), true),
1556 (RelPath::unix("carrot").unwrap(), false),
1557 (RelPath::unix("Aardvark.txt").unwrap(), true),
1558 ];
1559 paths.sort_by(|&a, &b| compare_rel_paths_files_first(a, b));
1560 assert_eq!(
1561 paths,
1562 vec![
1563 (RelPath::unix("Aardvark.txt").unwrap(), true),
1564 (RelPath::unix("Banana.rs").unwrap(), true),
1565 (RelPath::unix("Zebra.txt").unwrap(), true),
1566 (RelPath::unix("apple").unwrap(), false),
1567 (RelPath::unix("carrot").unwrap(), false),
1568 ]
1569 );
1570 }
1571
1572 #[perf]
1573 fn compare_rel_paths_files_first_numeric() {
1574 // Test natural number sorting with files first
1575 let mut paths = vec![
1576 (RelPath::unix("file10.txt").unwrap(), true),
1577 (RelPath::unix("dir2").unwrap(), false),
1578 (RelPath::unix("file2.txt").unwrap(), true),
1579 (RelPath::unix("dir10").unwrap(), false),
1580 (RelPath::unix("file1.txt").unwrap(), true),
1581 ];
1582 paths.sort_by(|&a, &b| compare_rel_paths_files_first(a, b));
1583 assert_eq!(
1584 paths,
1585 vec![
1586 (RelPath::unix("file1.txt").unwrap(), true),
1587 (RelPath::unix("file2.txt").unwrap(), true),
1588 (RelPath::unix("file10.txt").unwrap(), true),
1589 (RelPath::unix("dir2").unwrap(), false),
1590 (RelPath::unix("dir10").unwrap(), false),
1591 ]
1592 );
1593 }
1594
1595 #[perf]
1596 fn compare_rel_paths_mixed_case() {
1597 // Test case-insensitive sorting with varied capitalization
1598 let mut paths = vec![
1599 (RelPath::unix("README.md").unwrap(), true),
1600 (RelPath::unix("readme.txt").unwrap(), true),
1601 (RelPath::unix("ReadMe.rs").unwrap(), true),
1602 ];
1603 paths.sort_by(|&a, &b| compare_rel_paths_mixed(a, b));
1604 // All "readme" variants should group together, sorted by extension
1605 assert_eq!(
1606 paths,
1607 vec![
1608 (RelPath::unix("readme.txt").unwrap(), true),
1609 (RelPath::unix("ReadMe.rs").unwrap(), true),
1610 (RelPath::unix("README.md").unwrap(), true),
1611 ]
1612 );
1613 }
1614
1615 #[perf]
1616 fn compare_rel_paths_mixed_files_and_dirs() {
1617 // Verify directories and files are still mixed
1618 let mut paths = vec![
1619 (RelPath::unix("file2.txt").unwrap(), true),
1620 (RelPath::unix("Dir1").unwrap(), false),
1621 (RelPath::unix("file1.txt").unwrap(), true),
1622 (RelPath::unix("dir2").unwrap(), false),
1623 ];
1624 paths.sort_by(|&a, &b| compare_rel_paths_mixed(a, b));
1625 // Case-insensitive: dir1, dir2, file1, file2 (all mixed)
1626 assert_eq!(
1627 paths,
1628 vec![
1629 (RelPath::unix("Dir1").unwrap(), false),
1630 (RelPath::unix("dir2").unwrap(), false),
1631 (RelPath::unix("file1.txt").unwrap(), true),
1632 (RelPath::unix("file2.txt").unwrap(), true),
1633 ]
1634 );
1635 }
1636
1637 #[perf]
1638 fn compare_rel_paths_mixed_with_nested_paths() {
1639 // Test that nested paths still work correctly
1640 let mut paths = vec![
1641 (RelPath::unix("src/main.rs").unwrap(), true),
1642 (RelPath::unix("Cargo.toml").unwrap(), true),
1643 (RelPath::unix("src").unwrap(), false),
1644 (RelPath::unix("target").unwrap(), false),
1645 ];
1646 paths.sort_by(|&a, &b| compare_rel_paths_mixed(a, b));
1647 assert_eq!(
1648 paths,
1649 vec![
1650 (RelPath::unix("Cargo.toml").unwrap(), true),
1651 (RelPath::unix("src").unwrap(), false),
1652 (RelPath::unix("src/main.rs").unwrap(), true),
1653 (RelPath::unix("target").unwrap(), false),
1654 ]
1655 );
1656 }
1657
1658 #[perf]
1659 fn compare_rel_paths_files_first_with_nested() {
1660 // Files come before directories, even with nested paths
1661 let mut paths = vec![
1662 (RelPath::unix("src/lib.rs").unwrap(), true),
1663 (RelPath::unix("README.md").unwrap(), true),
1664 (RelPath::unix("src").unwrap(), false),
1665 (RelPath::unix("tests").unwrap(), false),
1666 ];
1667 paths.sort_by(|&a, &b| compare_rel_paths_files_first(a, b));
1668 assert_eq!(
1669 paths,
1670 vec![
1671 (RelPath::unix("README.md").unwrap(), true),
1672 (RelPath::unix("src").unwrap(), false),
1673 (RelPath::unix("src/lib.rs").unwrap(), true),
1674 (RelPath::unix("tests").unwrap(), false),
1675 ]
1676 );
1677 }
1678
1679 #[perf]
1680 fn compare_rel_paths_mixed_dotfiles() {
1681 // Test that dotfiles are handled correctly in mixed mode
1682 let mut paths = vec![
1683 (RelPath::unix(".gitignore").unwrap(), true),
1684 (RelPath::unix("README.md").unwrap(), true),
1685 (RelPath::unix(".github").unwrap(), false),
1686 (RelPath::unix("src").unwrap(), false),
1687 ];
1688 paths.sort_by(|&a, &b| compare_rel_paths_mixed(a, b));
1689 assert_eq!(
1690 paths,
1691 vec![
1692 (RelPath::unix(".github").unwrap(), false),
1693 (RelPath::unix(".gitignore").unwrap(), true),
1694 (RelPath::unix("README.md").unwrap(), true),
1695 (RelPath::unix("src").unwrap(), false),
1696 ]
1697 );
1698 }
1699
1700 #[perf]
1701 fn compare_rel_paths_files_first_dotfiles() {
1702 // Test that dotfiles come first when they're files
1703 let mut paths = vec![
1704 (RelPath::unix(".gitignore").unwrap(), true),
1705 (RelPath::unix("README.md").unwrap(), true),
1706 (RelPath::unix(".github").unwrap(), false),
1707 (RelPath::unix("src").unwrap(), false),
1708 ];
1709 paths.sort_by(|&a, &b| compare_rel_paths_files_first(a, b));
1710 assert_eq!(
1711 paths,
1712 vec![
1713 (RelPath::unix(".gitignore").unwrap(), true),
1714 (RelPath::unix("README.md").unwrap(), true),
1715 (RelPath::unix(".github").unwrap(), false),
1716 (RelPath::unix("src").unwrap(), false),
1717 ]
1718 );
1719 }
1720
1721 #[perf]
1722 fn compare_rel_paths_mixed_same_stem_different_extension() {
1723 // Files with same stem but different extensions should sort by extension
1724 let mut paths = vec![
1725 (RelPath::unix("file.rs").unwrap(), true),
1726 (RelPath::unix("file.md").unwrap(), true),
1727 (RelPath::unix("file.txt").unwrap(), true),
1728 ];
1729 paths.sort_by(|&a, &b| compare_rel_paths_mixed(a, b));
1730 assert_eq!(
1731 paths,
1732 vec![
1733 (RelPath::unix("file.txt").unwrap(), true),
1734 (RelPath::unix("file.rs").unwrap(), true),
1735 (RelPath::unix("file.md").unwrap(), true),
1736 ]
1737 );
1738 }
1739
1740 #[perf]
1741 fn compare_rel_paths_files_first_same_stem() {
1742 // Same stem files should still sort by extension with files_first
1743 let mut paths = vec![
1744 (RelPath::unix("main.rs").unwrap(), true),
1745 (RelPath::unix("main.c").unwrap(), true),
1746 (RelPath::unix("main").unwrap(), false),
1747 ];
1748 paths.sort_by(|&a, &b| compare_rel_paths_files_first(a, b));
1749 assert_eq!(
1750 paths,
1751 vec![
1752 (RelPath::unix("main.c").unwrap(), true),
1753 (RelPath::unix("main.rs").unwrap(), true),
1754 (RelPath::unix("main").unwrap(), false),
1755 ]
1756 );
1757 }
1758
1759 #[perf]
1760 fn compare_rel_paths_mixed_deep_nesting() {
1761 // Test sorting with deeply nested paths
1762 let mut paths = vec![
1763 (RelPath::unix("a/b/c.txt").unwrap(), true),
1764 (RelPath::unix("A/B.txt").unwrap(), true),
1765 (RelPath::unix("a.txt").unwrap(), true),
1766 (RelPath::unix("A.txt").unwrap(), true),
1767 ];
1768 paths.sort_by(|&a, &b| compare_rel_paths_mixed(a, b));
1769 assert_eq!(
1770 paths,
1771 vec![
1772 (RelPath::unix("A/B.txt").unwrap(), true),
1773 (RelPath::unix("a/b/c.txt").unwrap(), true),
1774 (RelPath::unix("a.txt").unwrap(), true),
1775 (RelPath::unix("A.txt").unwrap(), true),
1776 ]
1777 );
1778 }
1779
1780 #[perf]
1781 fn path_with_position_parse_posix_path() {
1782 // Test POSIX filename edge cases
1783 // Read more at https://en.wikipedia.org/wiki/Filename
1784 assert_eq!(
1785 PathWithPosition::parse_str("test_file"),
1786 PathWithPosition {
1787 path: PathBuf::from("test_file"),
1788 row: None,
1789 column: None
1790 }
1791 );
1792
1793 assert_eq!(
1794 PathWithPosition::parse_str("a:bc:.zip:1"),
1795 PathWithPosition {
1796 path: PathBuf::from("a:bc:.zip"),
1797 row: Some(1),
1798 column: None
1799 }
1800 );
1801
1802 assert_eq!(
1803 PathWithPosition::parse_str("one.second.zip:1"),
1804 PathWithPosition {
1805 path: PathBuf::from("one.second.zip"),
1806 row: Some(1),
1807 column: None
1808 }
1809 );
1810
1811 // Trim off trailing `:`s for otherwise valid input.
1812 assert_eq!(
1813 PathWithPosition::parse_str("test_file:10:1:"),
1814 PathWithPosition {
1815 path: PathBuf::from("test_file"),
1816 row: Some(10),
1817 column: Some(1)
1818 }
1819 );
1820
1821 assert_eq!(
1822 PathWithPosition::parse_str("test_file.rs:"),
1823 PathWithPosition {
1824 path: PathBuf::from("test_file.rs"),
1825 row: None,
1826 column: None
1827 }
1828 );
1829
1830 assert_eq!(
1831 PathWithPosition::parse_str("test_file.rs:1:"),
1832 PathWithPosition {
1833 path: PathBuf::from("test_file.rs"),
1834 row: Some(1),
1835 column: None
1836 }
1837 );
1838
1839 assert_eq!(
1840 PathWithPosition::parse_str("ab\ncd"),
1841 PathWithPosition {
1842 path: PathBuf::from("ab\ncd"),
1843 row: None,
1844 column: None
1845 }
1846 );
1847
1848 assert_eq!(
1849 PathWithPosition::parse_str("👋\nab"),
1850 PathWithPosition {
1851 path: PathBuf::from("👋\nab"),
1852 row: None,
1853 column: None
1854 }
1855 );
1856
1857 assert_eq!(
1858 PathWithPosition::parse_str("Types.hs:(617,9)-(670,28):"),
1859 PathWithPosition {
1860 path: PathBuf::from("Types.hs"),
1861 row: Some(617),
1862 column: Some(9),
1863 }
1864 );
1865 }
1866
1867 #[perf]
1868 #[cfg(not(target_os = "windows"))]
1869 fn path_with_position_parse_posix_path_with_suffix() {
1870 assert_eq!(
1871 PathWithPosition::parse_str("foo/bar:34:in"),
1872 PathWithPosition {
1873 path: PathBuf::from("foo/bar"),
1874 row: Some(34),
1875 column: None,
1876 }
1877 );
1878 assert_eq!(
1879 PathWithPosition::parse_str("foo/bar.rs:1902:::15:"),
1880 PathWithPosition {
1881 path: PathBuf::from("foo/bar.rs:1902"),
1882 row: Some(15),
1883 column: None
1884 }
1885 );
1886
1887 assert_eq!(
1888 PathWithPosition::parse_str("app-editors:zed-0.143.6:20240710-201212.log:34:"),
1889 PathWithPosition {
1890 path: PathBuf::from("app-editors:zed-0.143.6:20240710-201212.log"),
1891 row: Some(34),
1892 column: None,
1893 }
1894 );
1895
1896 assert_eq!(
1897 PathWithPosition::parse_str("crates/file_finder/src/file_finder.rs:1902:13:"),
1898 PathWithPosition {
1899 path: PathBuf::from("crates/file_finder/src/file_finder.rs"),
1900 row: Some(1902),
1901 column: Some(13),
1902 }
1903 );
1904
1905 assert_eq!(
1906 PathWithPosition::parse_str("crate/utils/src/test:today.log:34"),
1907 PathWithPosition {
1908 path: PathBuf::from("crate/utils/src/test:today.log"),
1909 row: Some(34),
1910 column: None,
1911 }
1912 );
1913 assert_eq!(
1914 PathWithPosition::parse_str("/testing/out/src/file_finder.odin(7:15)"),
1915 PathWithPosition {
1916 path: PathBuf::from("/testing/out/src/file_finder.odin"),
1917 row: Some(7),
1918 column: Some(15),
1919 }
1920 );
1921 }
1922
1923 #[perf]
1924 #[cfg(target_os = "windows")]
1925 fn path_with_position_parse_windows_path() {
1926 assert_eq!(
1927 PathWithPosition::parse_str("crates\\utils\\paths.rs"),
1928 PathWithPosition {
1929 path: PathBuf::from("crates\\utils\\paths.rs"),
1930 row: None,
1931 column: None
1932 }
1933 );
1934
1935 assert_eq!(
1936 PathWithPosition::parse_str("C:\\Users\\someone\\test_file.rs"),
1937 PathWithPosition {
1938 path: PathBuf::from("C:\\Users\\someone\\test_file.rs"),
1939 row: None,
1940 column: None
1941 }
1942 );
1943 }
1944
1945 #[perf]
1946 #[cfg(target_os = "windows")]
1947 fn path_with_position_parse_windows_path_with_suffix() {
1948 assert_eq!(
1949 PathWithPosition::parse_str("crates\\utils\\paths.rs:101"),
1950 PathWithPosition {
1951 path: PathBuf::from("crates\\utils\\paths.rs"),
1952 row: Some(101),
1953 column: None
1954 }
1955 );
1956
1957 assert_eq!(
1958 PathWithPosition::parse_str("\\\\?\\C:\\Users\\someone\\test_file.rs:1:20"),
1959 PathWithPosition {
1960 path: PathBuf::from("\\\\?\\C:\\Users\\someone\\test_file.rs"),
1961 row: Some(1),
1962 column: Some(20)
1963 }
1964 );
1965
1966 assert_eq!(
1967 PathWithPosition::parse_str("C:\\Users\\someone\\test_file.rs(1902,13)"),
1968 PathWithPosition {
1969 path: PathBuf::from("C:\\Users\\someone\\test_file.rs"),
1970 row: Some(1902),
1971 column: Some(13)
1972 }
1973 );
1974
1975 // Trim off trailing `:`s for otherwise valid input.
1976 assert_eq!(
1977 PathWithPosition::parse_str("\\\\?\\C:\\Users\\someone\\test_file.rs:1902:13:"),
1978 PathWithPosition {
1979 path: PathBuf::from("\\\\?\\C:\\Users\\someone\\test_file.rs"),
1980 row: Some(1902),
1981 column: Some(13)
1982 }
1983 );
1984
1985 assert_eq!(
1986 PathWithPosition::parse_str("\\\\?\\C:\\Users\\someone\\test_file.rs:1902:13:15:"),
1987 PathWithPosition {
1988 path: PathBuf::from("\\\\?\\C:\\Users\\someone\\test_file.rs:1902"),
1989 row: Some(13),
1990 column: Some(15)
1991 }
1992 );
1993
1994 assert_eq!(
1995 PathWithPosition::parse_str("\\\\?\\C:\\Users\\someone\\test_file.rs:1902:::15:"),
1996 PathWithPosition {
1997 path: PathBuf::from("\\\\?\\C:\\Users\\someone\\test_file.rs:1902"),
1998 row: Some(15),
1999 column: None
2000 }
2001 );
2002
2003 assert_eq!(
2004 PathWithPosition::parse_str("\\\\?\\C:\\Users\\someone\\test_file.rs(1902,13):"),
2005 PathWithPosition {
2006 path: PathBuf::from("\\\\?\\C:\\Users\\someone\\test_file.rs"),
2007 row: Some(1902),
2008 column: Some(13),
2009 }
2010 );
2011
2012 assert_eq!(
2013 PathWithPosition::parse_str("\\\\?\\C:\\Users\\someone\\test_file.rs(1902):"),
2014 PathWithPosition {
2015 path: PathBuf::from("\\\\?\\C:\\Users\\someone\\test_file.rs"),
2016 row: Some(1902),
2017 column: None,
2018 }
2019 );
2020
2021 assert_eq!(
2022 PathWithPosition::parse_str("C:\\Users\\someone\\test_file.rs:1902:13:"),
2023 PathWithPosition {
2024 path: PathBuf::from("C:\\Users\\someone\\test_file.rs"),
2025 row: Some(1902),
2026 column: Some(13),
2027 }
2028 );
2029
2030 assert_eq!(
2031 PathWithPosition::parse_str("C:\\Users\\someone\\test_file.rs(1902,13):"),
2032 PathWithPosition {
2033 path: PathBuf::from("C:\\Users\\someone\\test_file.rs"),
2034 row: Some(1902),
2035 column: Some(13),
2036 }
2037 );
2038
2039 assert_eq!(
2040 PathWithPosition::parse_str("C:\\Users\\someone\\test_file.rs(1902):"),
2041 PathWithPosition {
2042 path: PathBuf::from("C:\\Users\\someone\\test_file.rs"),
2043 row: Some(1902),
2044 column: None,
2045 }
2046 );
2047
2048 assert_eq!(
2049 PathWithPosition::parse_str("crates/utils/paths.rs:101"),
2050 PathWithPosition {
2051 path: PathBuf::from("crates\\utils\\paths.rs"),
2052 row: Some(101),
2053 column: None,
2054 }
2055 );
2056 }
2057
2058 #[perf]
2059 fn test_path_compact() {
2060 let path: PathBuf = [
2061 home_dir().to_string_lossy().into_owned(),
2062 "some_file.txt".to_string(),
2063 ]
2064 .iter()
2065 .collect();
2066 if cfg!(any(target_os = "linux", target_os = "freebsd")) || cfg!(target_os = "macos") {
2067 assert_eq!(path.compact().to_str(), Some("~/some_file.txt"));
2068 } else {
2069 assert_eq!(path.compact().to_str(), path.to_str());
2070 }
2071 }
2072
2073 #[perf]
2074 fn test_extension_or_hidden_file_name() {
2075 // No dots in name
2076 let path = Path::new("/a/b/c/file_name.rs");
2077 assert_eq!(path.extension_or_hidden_file_name(), Some("rs"));
2078
2079 // Single dot in name
2080 let path = Path::new("/a/b/c/file.name.rs");
2081 assert_eq!(path.extension_or_hidden_file_name(), Some("rs"));
2082
2083 // Multiple dots in name
2084 let path = Path::new("/a/b/c/long.file.name.rs");
2085 assert_eq!(path.extension_or_hidden_file_name(), Some("rs"));
2086
2087 // Hidden file, no extension
2088 let path = Path::new("/a/b/c/.gitignore");
2089 assert_eq!(path.extension_or_hidden_file_name(), Some("gitignore"));
2090
2091 // Hidden file, with extension
2092 let path = Path::new("/a/b/c/.eslintrc.js");
2093 assert_eq!(path.extension_or_hidden_file_name(), Some("eslintrc.js"));
2094 }
2095
2096 #[perf]
2097 // fn edge_of_glob() {
2098 // let path = Path::new("/work/node_modules");
2099 // let path_matcher =
2100 // PathMatcher::new(&["**/node_modules/**".to_owned()], PathStyle::Posix).unwrap();
2101 // assert!(
2102 // path_matcher.is_match(path),
2103 // "Path matcher should match {path:?}"
2104 // );
2105 // }
2106
2107 // #[perf]
2108 // fn file_in_dirs() {
2109 // let path = Path::new("/work/.env");
2110 // let path_matcher = PathMatcher::new(&["**/.env".to_owned()], PathStyle::Posix).unwrap();
2111 // assert!(
2112 // path_matcher.is_match(path),
2113 // "Path matcher should match {path:?}"
2114 // );
2115 // let path = Path::new("/work/package.json");
2116 // assert!(
2117 // !path_matcher.is_match(path),
2118 // "Path matcher should not match {path:?}"
2119 // );
2120 // }
2121
2122 // #[perf]
2123 // fn project_search() {
2124 // let path = Path::new("/Users/someonetoignore/work/zed/zed.dev/node_modules");
2125 // let path_matcher =
2126 // PathMatcher::new(&["**/node_modules/**".to_owned()], PathStyle::Posix).unwrap();
2127 // assert!(
2128 // path_matcher.is_match(path),
2129 // "Path matcher should match {path:?}"
2130 // );
2131 // }
2132 #[perf]
2133 #[cfg(target_os = "windows")]
2134 fn test_sanitized_path() {
2135 let path = Path::new("C:\\Users\\someone\\test_file.rs");
2136 let sanitized_path = SanitizedPath::new(path);
2137 assert_eq!(
2138 sanitized_path.to_string(),
2139 "C:\\Users\\someone\\test_file.rs"
2140 );
2141
2142 let path = Path::new("\\\\?\\C:\\Users\\someone\\test_file.rs");
2143 let sanitized_path = SanitizedPath::new(path);
2144 assert_eq!(
2145 sanitized_path.to_string(),
2146 "C:\\Users\\someone\\test_file.rs"
2147 );
2148 }
2149
2150 #[perf]
2151 fn test_compare_numeric_segments() {
2152 // Helper function to create peekable iterators and test
2153 fn compare(a: &str, b: &str) -> Ordering {
2154 let mut a_iter = a.chars().peekable();
2155 let mut b_iter = b.chars().peekable();
2156
2157 let result = compare_numeric_segments(&mut a_iter, &mut b_iter);
2158
2159 // Verify iterators advanced correctly
2160 assert!(
2161 !a_iter.next().is_some_and(|c| c.is_ascii_digit()),
2162 "Iterator a should have consumed all digits"
2163 );
2164 assert!(
2165 !b_iter.next().is_some_and(|c| c.is_ascii_digit()),
2166 "Iterator b should have consumed all digits"
2167 );
2168
2169 result
2170 }
2171
2172 // Basic numeric comparisons
2173 assert_eq!(compare("0", "0"), Ordering::Equal);
2174 assert_eq!(compare("1", "2"), Ordering::Less);
2175 assert_eq!(compare("9", "10"), Ordering::Less);
2176 assert_eq!(compare("10", "9"), Ordering::Greater);
2177 assert_eq!(compare("99", "100"), Ordering::Less);
2178
2179 // Leading zeros
2180 assert_eq!(compare("0", "00"), Ordering::Less);
2181 assert_eq!(compare("00", "0"), Ordering::Greater);
2182 assert_eq!(compare("01", "1"), Ordering::Greater);
2183 assert_eq!(compare("001", "1"), Ordering::Greater);
2184 assert_eq!(compare("001", "01"), Ordering::Greater);
2185
2186 // Same value different representation
2187 assert_eq!(compare("000100", "100"), Ordering::Greater);
2188 assert_eq!(compare("100", "0100"), Ordering::Less);
2189 assert_eq!(compare("0100", "00100"), Ordering::Less);
2190
2191 // Large numbers
2192 assert_eq!(compare("9999999999", "10000000000"), Ordering::Less);
2193 assert_eq!(
2194 compare(
2195 "340282366920938463463374607431768211455", // u128::MAX
2196 "340282366920938463463374607431768211456"
2197 ),
2198 Ordering::Less
2199 );
2200 assert_eq!(
2201 compare(
2202 "340282366920938463463374607431768211456", // > u128::MAX
2203 "340282366920938463463374607431768211455"
2204 ),
2205 Ordering::Greater
2206 );
2207
2208 // Iterator advancement verification
2209 let mut a_iter = "123abc".chars().peekable();
2210 let mut b_iter = "456def".chars().peekable();
2211
2212 compare_numeric_segments(&mut a_iter, &mut b_iter);
2213
2214 assert_eq!(a_iter.collect::<String>(), "abc");
2215 assert_eq!(b_iter.collect::<String>(), "def");
2216 }
2217
2218 #[perf]
2219 fn test_natural_sort() {
2220 // Basic alphanumeric
2221 assert_eq!(natural_sort("a", "b"), Ordering::Less);
2222 assert_eq!(natural_sort("b", "a"), Ordering::Greater);
2223 assert_eq!(natural_sort("a", "a"), Ordering::Equal);
2224
2225 // Case sensitivity
2226 assert_eq!(natural_sort("a", "A"), Ordering::Less);
2227 assert_eq!(natural_sort("A", "a"), Ordering::Greater);
2228 assert_eq!(natural_sort("aA", "aa"), Ordering::Greater);
2229 assert_eq!(natural_sort("aa", "aA"), Ordering::Less);
2230
2231 // Numbers
2232 assert_eq!(natural_sort("1", "2"), Ordering::Less);
2233 assert_eq!(natural_sort("2", "10"), Ordering::Less);
2234 assert_eq!(natural_sort("02", "10"), Ordering::Less);
2235 assert_eq!(natural_sort("02", "2"), Ordering::Greater);
2236
2237 // Mixed alphanumeric
2238 assert_eq!(natural_sort("a1", "a2"), Ordering::Less);
2239 assert_eq!(natural_sort("a2", "a10"), Ordering::Less);
2240 assert_eq!(natural_sort("a02", "a2"), Ordering::Greater);
2241 assert_eq!(natural_sort("a1b", "a1c"), Ordering::Less);
2242
2243 // Multiple numeric segments
2244 assert_eq!(natural_sort("1a2", "1a10"), Ordering::Less);
2245 assert_eq!(natural_sort("1a10", "1a2"), Ordering::Greater);
2246 assert_eq!(natural_sort("2a1", "10a1"), Ordering::Less);
2247
2248 // Special characters
2249 assert_eq!(natural_sort("a-1", "a-2"), Ordering::Less);
2250 assert_eq!(natural_sort("a_1", "a_2"), Ordering::Less);
2251 assert_eq!(natural_sort("a.1", "a.2"), Ordering::Less);
2252
2253 // Unicode
2254 assert_eq!(natural_sort("文1", "文2"), Ordering::Less);
2255 assert_eq!(natural_sort("文2", "文10"), Ordering::Less);
2256 assert_eq!(natural_sort("🔤1", "🔤2"), Ordering::Less);
2257
2258 // Empty and special cases
2259 assert_eq!(natural_sort("", ""), Ordering::Equal);
2260 assert_eq!(natural_sort("", "a"), Ordering::Less);
2261 assert_eq!(natural_sort("a", ""), Ordering::Greater);
2262 assert_eq!(natural_sort(" ", " "), Ordering::Less);
2263
2264 // Mixed everything
2265 assert_eq!(natural_sort("File-1.txt", "File-2.txt"), Ordering::Less);
2266 assert_eq!(natural_sort("File-02.txt", "File-2.txt"), Ordering::Greater);
2267 assert_eq!(natural_sort("File-2.txt", "File-10.txt"), Ordering::Less);
2268 assert_eq!(natural_sort("File_A1", "File_A2"), Ordering::Less);
2269 assert_eq!(natural_sort("File_a1", "File_A1"), Ordering::Less);
2270 }
2271
2272 #[perf]
2273 fn test_compare_paths() {
2274 // Helper function for cleaner tests
2275 fn compare(a: &str, is_a_file: bool, b: &str, is_b_file: bool) -> Ordering {
2276 compare_paths((Path::new(a), is_a_file), (Path::new(b), is_b_file))
2277 }
2278
2279 // Basic path comparison
2280 assert_eq!(compare("a", true, "b", true), Ordering::Less);
2281 assert_eq!(compare("b", true, "a", true), Ordering::Greater);
2282 assert_eq!(compare("a", true, "a", true), Ordering::Equal);
2283
2284 // Files vs Directories
2285 assert_eq!(compare("a", true, "a", false), Ordering::Greater);
2286 assert_eq!(compare("a", false, "a", true), Ordering::Less);
2287 assert_eq!(compare("b", false, "a", true), Ordering::Less);
2288
2289 // Extensions
2290 assert_eq!(compare("a.txt", true, "a.md", true), Ordering::Greater);
2291 assert_eq!(compare("a.md", true, "a.txt", true), Ordering::Less);
2292 assert_eq!(compare("a", true, "a.txt", true), Ordering::Less);
2293
2294 // Nested paths
2295 assert_eq!(compare("dir/a", true, "dir/b", true), Ordering::Less);
2296 assert_eq!(compare("dir1/a", true, "dir2/a", true), Ordering::Less);
2297 assert_eq!(compare("dir/sub/a", true, "dir/a", true), Ordering::Less);
2298
2299 // Case sensitivity in paths
2300 assert_eq!(
2301 compare("Dir/file", true, "dir/file", true),
2302 Ordering::Greater
2303 );
2304 assert_eq!(
2305 compare("dir/File", true, "dir/file", true),
2306 Ordering::Greater
2307 );
2308 assert_eq!(compare("dir/file", true, "Dir/File", true), Ordering::Less);
2309
2310 // Hidden files and special names
2311 assert_eq!(compare(".hidden", true, "visible", true), Ordering::Less);
2312 assert_eq!(compare("_special", true, "normal", true), Ordering::Less);
2313 assert_eq!(compare(".config", false, ".data", false), Ordering::Less);
2314
2315 // Mixed numeric paths
2316 assert_eq!(
2317 compare("dir1/file", true, "dir2/file", true),
2318 Ordering::Less
2319 );
2320 assert_eq!(
2321 compare("dir2/file", true, "dir10/file", true),
2322 Ordering::Less
2323 );
2324 assert_eq!(
2325 compare("dir02/file", true, "dir2/file", true),
2326 Ordering::Greater
2327 );
2328
2329 // Root paths
2330 assert_eq!(compare("/a", true, "/b", true), Ordering::Less);
2331 assert_eq!(compare("/", false, "/a", true), Ordering::Less);
2332
2333 // Complex real-world examples
2334 assert_eq!(
2335 compare("project/src/main.rs", true, "project/src/lib.rs", true),
2336 Ordering::Greater
2337 );
2338 assert_eq!(
2339 compare(
2340 "project/tests/test_1.rs",
2341 true,
2342 "project/tests/test_2.rs",
2343 true
2344 ),
2345 Ordering::Less
2346 );
2347 assert_eq!(
2348 compare(
2349 "project/v1.0.0/README.md",
2350 true,
2351 "project/v1.10.0/README.md",
2352 true
2353 ),
2354 Ordering::Less
2355 );
2356 }
2357
2358 #[perf]
2359 fn test_natural_sort_case_sensitivity() {
2360 std::thread::sleep(std::time::Duration::from_millis(100));
2361 // Same letter different case - lowercase should come first
2362 assert_eq!(natural_sort("a", "A"), Ordering::Less);
2363 assert_eq!(natural_sort("A", "a"), Ordering::Greater);
2364 assert_eq!(natural_sort("a", "a"), Ordering::Equal);
2365 assert_eq!(natural_sort("A", "A"), Ordering::Equal);
2366
2367 // Mixed case strings
2368 assert_eq!(natural_sort("aaa", "AAA"), Ordering::Less);
2369 assert_eq!(natural_sort("AAA", "aaa"), Ordering::Greater);
2370 assert_eq!(natural_sort("aAa", "AaA"), Ordering::Less);
2371
2372 // Different letters
2373 assert_eq!(natural_sort("a", "b"), Ordering::Less);
2374 assert_eq!(natural_sort("A", "b"), Ordering::Less);
2375 assert_eq!(natural_sort("a", "B"), Ordering::Less);
2376 }
2377
2378 #[perf]
2379 fn test_natural_sort_with_numbers() {
2380 // Basic number ordering
2381 assert_eq!(natural_sort("file1", "file2"), Ordering::Less);
2382 assert_eq!(natural_sort("file2", "file10"), Ordering::Less);
2383 assert_eq!(natural_sort("file10", "file2"), Ordering::Greater);
2384
2385 // Numbers in different positions
2386 assert_eq!(natural_sort("1file", "2file"), Ordering::Less);
2387 assert_eq!(natural_sort("file1text", "file2text"), Ordering::Less);
2388 assert_eq!(natural_sort("text1file", "text2file"), Ordering::Less);
2389
2390 // Multiple numbers in string
2391 assert_eq!(natural_sort("file1-2", "file1-10"), Ordering::Less);
2392 assert_eq!(natural_sort("2-1file", "10-1file"), Ordering::Less);
2393
2394 // Leading zeros
2395 assert_eq!(natural_sort("file002", "file2"), Ordering::Greater);
2396 assert_eq!(natural_sort("file002", "file10"), Ordering::Less);
2397
2398 // Very large numbers
2399 assert_eq!(
2400 natural_sort("file999999999999999999999", "file999999999999999999998"),
2401 Ordering::Greater
2402 );
2403
2404 // u128 edge cases
2405
2406 // Numbers near u128::MAX (340,282,366,920,938,463,463,374,607,431,768,211,455)
2407 assert_eq!(
2408 natural_sort(
2409 "file340282366920938463463374607431768211454",
2410 "file340282366920938463463374607431768211455"
2411 ),
2412 Ordering::Less
2413 );
2414
2415 // Equal length numbers that overflow u128
2416 assert_eq!(
2417 natural_sort(
2418 "file340282366920938463463374607431768211456",
2419 "file340282366920938463463374607431768211455"
2420 ),
2421 Ordering::Greater
2422 );
2423
2424 // Different length numbers that overflow u128
2425 assert_eq!(
2426 natural_sort(
2427 "file3402823669209384634633746074317682114560",
2428 "file340282366920938463463374607431768211455"
2429 ),
2430 Ordering::Greater
2431 );
2432
2433 // Leading zeros with numbers near u128::MAX
2434 assert_eq!(
2435 natural_sort(
2436 "file0340282366920938463463374607431768211455",
2437 "file340282366920938463463374607431768211455"
2438 ),
2439 Ordering::Greater
2440 );
2441
2442 // Very large numbers with different lengths (both overflow u128)
2443 assert_eq!(
2444 natural_sort(
2445 "file999999999999999999999999999999999999999999999999",
2446 "file9999999999999999999999999999999999999999999999999"
2447 ),
2448 Ordering::Less
2449 );
2450 }
2451
2452 #[perf]
2453 fn test_natural_sort_case_sensitive() {
2454 // Numerically smaller values come first.
2455 assert_eq!(natural_sort("File1", "file2"), Ordering::Less);
2456 assert_eq!(natural_sort("file1", "File2"), Ordering::Less);
2457
2458 // Numerically equal values: the case-insensitive comparison decides first.
2459 // Case-sensitive comparison only occurs when both are equal case-insensitively.
2460 assert_eq!(natural_sort("Dir1", "dir01"), Ordering::Less);
2461 assert_eq!(natural_sort("dir2", "Dir02"), Ordering::Less);
2462 assert_eq!(natural_sort("dir2", "dir02"), Ordering::Less);
2463
2464 // Numerically equal and case-insensitively equal:
2465 // the lexicographically smaller (case-sensitive) one wins.
2466 assert_eq!(natural_sort("dir1", "Dir1"), Ordering::Less);
2467 assert_eq!(natural_sort("dir02", "Dir02"), Ordering::Less);
2468 assert_eq!(natural_sort("dir10", "Dir10"), Ordering::Less);
2469 }
2470
2471 #[perf]
2472 fn test_natural_sort_edge_cases() {
2473 // Empty strings
2474 assert_eq!(natural_sort("", ""), Ordering::Equal);
2475 assert_eq!(natural_sort("", "a"), Ordering::Less);
2476 assert_eq!(natural_sort("a", ""), Ordering::Greater);
2477
2478 // Special characters
2479 assert_eq!(natural_sort("file-1", "file_1"), Ordering::Less);
2480 assert_eq!(natural_sort("file.1", "file_1"), Ordering::Less);
2481 assert_eq!(natural_sort("file 1", "file_1"), Ordering::Less);
2482
2483 // Unicode characters
2484 // 9312 vs 9313
2485 assert_eq!(natural_sort("file①", "file②"), Ordering::Less);
2486 // 9321 vs 9313
2487 assert_eq!(natural_sort("file⑩", "file②"), Ordering::Greater);
2488 // 28450 vs 23383
2489 assert_eq!(natural_sort("file漢", "file字"), Ordering::Greater);
2490
2491 // Mixed alphanumeric with special chars
2492 assert_eq!(natural_sort("file-1a", "file-1b"), Ordering::Less);
2493 assert_eq!(natural_sort("file-1.2", "file-1.10"), Ordering::Less);
2494 assert_eq!(natural_sort("file-1.10", "file-1.2"), Ordering::Greater);
2495 }
2496
2497 #[test]
2498 fn test_multiple_extensions() {
2499 // No extensions
2500 let path = Path::new("/a/b/c/file_name");
2501 assert_eq!(path.multiple_extensions(), None);
2502
2503 // Only one extension
2504 let path = Path::new("/a/b/c/file_name.tsx");
2505 assert_eq!(path.multiple_extensions(), None);
2506
2507 // Stories sample extension
2508 let path = Path::new("/a/b/c/file_name.stories.tsx");
2509 assert_eq!(path.multiple_extensions(), Some("stories.tsx".to_string()));
2510
2511 // Longer sample extension
2512 let path = Path::new("/a/b/c/long.app.tar.gz");
2513 assert_eq!(path.multiple_extensions(), Some("app.tar.gz".to_string()));
2514 }
2515
2516 #[test]
2517 fn test_strip_path_suffix() {
2518 let base = Path::new("/a/b/c/file_name");
2519 let suffix = Path::new("file_name");
2520 assert_eq!(strip_path_suffix(base, suffix), Some(Path::new("/a/b/c")));
2521
2522 let base = Path::new("/a/b/c/file_name.tsx");
2523 let suffix = Path::new("file_name.tsx");
2524 assert_eq!(strip_path_suffix(base, suffix), Some(Path::new("/a/b/c")));
2525
2526 let base = Path::new("/a/b/c/file_name.stories.tsx");
2527 let suffix = Path::new("c/file_name.stories.tsx");
2528 assert_eq!(strip_path_suffix(base, suffix), Some(Path::new("/a/b")));
2529
2530 let base = Path::new("/a/b/c/long.app.tar.gz");
2531 let suffix = Path::new("b/c/long.app.tar.gz");
2532 assert_eq!(strip_path_suffix(base, suffix), Some(Path::new("/a")));
2533
2534 let base = Path::new("/a/b/c/long.app.tar.gz");
2535 let suffix = Path::new("/a/b/c/long.app.tar.gz");
2536 assert_eq!(strip_path_suffix(base, suffix), Some(Path::new("")));
2537
2538 let base = Path::new("/a/b/c/long.app.tar.gz");
2539 let suffix = Path::new("/a/b/c/no_match.app.tar.gz");
2540 assert_eq!(strip_path_suffix(base, suffix), None);
2541
2542 let base = Path::new("/a/b/c/long.app.tar.gz");
2543 let suffix = Path::new("app.tar.gz");
2544 assert_eq!(strip_path_suffix(base, suffix), None);
2545 }
2546
2547 #[test]
2548 fn test_strip_prefix() {
2549 let expected = [
2550 (
2551 PathStyle::Posix,
2552 "/a/b/c",
2553 "/a/b",
2554 Some(rel_path("c").into_arc()),
2555 ),
2556 (
2557 PathStyle::Posix,
2558 "/a/b/c",
2559 "/a/b/",
2560 Some(rel_path("c").into_arc()),
2561 ),
2562 (
2563 PathStyle::Posix,
2564 "/a/b/c",
2565 "/",
2566 Some(rel_path("a/b/c").into_arc()),
2567 ),
2568 (PathStyle::Posix, "/a/b/c", "", None),
2569 (PathStyle::Posix, "/a/b//c", "/a/b/", None),
2570 (PathStyle::Posix, "/a/bc", "/a/b", None),
2571 (
2572 PathStyle::Posix,
2573 "/a/b/c",
2574 "/a/b/c",
2575 Some(rel_path("").into_arc()),
2576 ),
2577 (
2578 PathStyle::Windows,
2579 "C:\\a\\b\\c",
2580 "C:\\a\\b",
2581 Some(rel_path("c").into_arc()),
2582 ),
2583 (
2584 PathStyle::Windows,
2585 "C:\\a\\b\\c",
2586 "C:\\a\\b\\",
2587 Some(rel_path("c").into_arc()),
2588 ),
2589 (
2590 PathStyle::Windows,
2591 "C:\\a\\b\\c",
2592 "C:\\",
2593 Some(rel_path("a/b/c").into_arc()),
2594 ),
2595 (PathStyle::Windows, "C:\\a\\b\\c", "", None),
2596 (PathStyle::Windows, "C:\\a\\b\\\\c", "C:\\a\\b\\", None),
2597 (PathStyle::Windows, "C:\\a\\bc", "C:\\a\\b", None),
2598 (
2599 PathStyle::Windows,
2600 "C:\\a\\b/c",
2601 "C:\\a\\b",
2602 Some(rel_path("c").into_arc()),
2603 ),
2604 (
2605 PathStyle::Windows,
2606 "C:\\a\\b/c",
2607 "C:\\a\\b\\",
2608 Some(rel_path("c").into_arc()),
2609 ),
2610 (
2611 PathStyle::Windows,
2612 "C:\\a\\b/c",
2613 "C:\\a\\b/",
2614 Some(rel_path("c").into_arc()),
2615 ),
2616 ];
2617 let actual = expected.clone().map(|(style, child, parent, _)| {
2618 (
2619 style,
2620 child,
2621 parent,
2622 style
2623 .strip_prefix(child.as_ref(), parent.as_ref())
2624 .map(|rel_path| rel_path.into_arc()),
2625 )
2626 });
2627 pretty_assertions::assert_eq!(actual, expected);
2628 }
2629
2630 #[cfg(target_os = "windows")]
2631 #[test]
2632 fn test_wsl_path() {
2633 use super::WslPath;
2634 let path = "/a/b/c";
2635 assert_eq!(WslPath::from_path(&path), None);
2636
2637 let path = r"\\wsl.localhost";
2638 assert_eq!(WslPath::from_path(&path), None);
2639
2640 let path = r"\\wsl.localhost\Distro";
2641 assert_eq!(
2642 WslPath::from_path(&path),
2643 Some(WslPath {
2644 distro: "Distro".to_owned(),
2645 path: "/".into(),
2646 })
2647 );
2648
2649 let path = r"\\wsl.localhost\Distro\blue";
2650 assert_eq!(
2651 WslPath::from_path(&path),
2652 Some(WslPath {
2653 distro: "Distro".to_owned(),
2654 path: "/blue".into()
2655 })
2656 );
2657
2658 let path = r"\\wsl$\archlinux\tomato\.\paprika\..\aubergine.txt";
2659 assert_eq!(
2660 WslPath::from_path(&path),
2661 Some(WslPath {
2662 distro: "archlinux".to_owned(),
2663 path: "/tomato/paprika/../aubergine.txt".into()
2664 })
2665 );
2666
2667 let path = r"\\windows.localhost\Distro\foo";
2668 assert_eq!(WslPath::from_path(&path), None);
2669 }
2670}