paths.rs

  1use std::{
  2    ffi::OsStr,
  3    path::{Path, PathBuf},
  4};
  5
  6use globset::{Glob, GlobMatcher};
  7use serde::{Deserialize, Serialize};
  8
  9lazy_static::lazy_static! {
 10    pub static ref HOME: PathBuf = dirs::home_dir().expect("failed to determine home directory");
 11}
 12
 13pub trait PathExt {
 14    fn compact(&self) -> PathBuf;
 15    fn icon_stem_or_suffix(&self) -> Option<&str>;
 16    fn extension_or_hidden_file_name(&self) -> Option<&str>;
 17    fn try_from_bytes<'a>(bytes: &'a [u8]) -> anyhow::Result<Self>
 18    where
 19        Self: From<&'a Path>,
 20    {
 21        #[cfg(unix)]
 22        {
 23            use std::os::unix::prelude::OsStrExt;
 24            Ok(Self::from(Path::new(OsStr::from_bytes(bytes))))
 25        }
 26        #[cfg(windows)]
 27        {
 28            use anyhow::anyhow;
 29            use tendril::fmt::{Format, WTF8};
 30            WTF8::validate(bytes)
 31                .then(|| {
 32                    // Safety: bytes are valid WTF-8 sequence.
 33                    Self::from(Path::new(unsafe {
 34                        OsStr::from_encoded_bytes_unchecked(bytes)
 35                    }))
 36                })
 37                .ok_or_else(|| anyhow!("Invalid WTF-8 sequence: {bytes:?}"))
 38        }
 39    }
 40}
 41
 42impl<T: AsRef<Path>> PathExt for T {
 43    /// Compacts a given file path by replacing the user's home directory
 44    /// prefix with a tilde (`~`).
 45    ///
 46    /// # Returns
 47    ///
 48    /// * A `PathBuf` containing the compacted file path. If the input path
 49    ///   does not have the user's home directory prefix, or if we are not on
 50    ///   Linux or macOS, the original path is returned unchanged.
 51    fn compact(&self) -> PathBuf {
 52        if cfg!(target_os = "linux") || cfg!(target_os = "macos") {
 53            match self.as_ref().strip_prefix(HOME.as_path()) {
 54                Ok(relative_path) => {
 55                    let mut shortened_path = PathBuf::new();
 56                    shortened_path.push("~");
 57                    shortened_path.push(relative_path);
 58                    shortened_path
 59                }
 60                Err(_) => self.as_ref().to_path_buf(),
 61            }
 62        } else {
 63            self.as_ref().to_path_buf()
 64        }
 65    }
 66
 67    /// Returns either the suffix if available, or the file stem otherwise to determine which file icon to use
 68    fn icon_stem_or_suffix(&self) -> Option<&str> {
 69        let path = self.as_ref();
 70        let file_name = path.file_name()?.to_str()?;
 71        if file_name.starts_with('.') {
 72            return file_name.strip_prefix('.');
 73        }
 74
 75        path.extension()
 76            .and_then(|e| e.to_str())
 77            .or_else(|| path.file_stem()?.to_str())
 78    }
 79
 80    /// Returns a file's extension or, if the file is hidden, its name without the leading dot
 81    fn extension_or_hidden_file_name(&self) -> Option<&str> {
 82        if let Some(extension) = self.as_ref().extension() {
 83            return extension.to_str();
 84        }
 85
 86        self.as_ref().file_name()?.to_str()?.split('.').last()
 87    }
 88}
 89
 90/// A delimiter to use in `path_query:row_number:column_number` strings parsing.
 91pub const FILE_ROW_COLUMN_DELIMITER: char = ':';
 92
 93/// A representation of a path-like string with optional row and column numbers.
 94/// Matching values example: `te`, `test.rs:22`, `te:22:5`, etc.
 95#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Hash)]
 96pub struct PathLikeWithPosition<P> {
 97    pub path_like: P,
 98    pub row: Option<u32>,
 99    // Absent if row is absent.
100    pub column: Option<u32>,
101}
102
103impl<P> PathLikeWithPosition<P> {
104    /// Parses a string that possibly has `:row:column` suffix.
105    /// Ignores trailing `:`s, so `test.rs:22:` is parsed as `test.rs:22`.
106    /// If any of the row/column component parsing fails, the whole string is then parsed as a path like.
107    pub fn parse_str<E>(
108        s: &str,
109        parse_path_like_str: impl Fn(&str) -> Result<P, E>,
110    ) -> Result<Self, E> {
111        let fallback = |fallback_str| {
112            Ok(Self {
113                path_like: parse_path_like_str(fallback_str)?,
114                row: None,
115                column: None,
116            })
117        };
118
119        let trimmed = s.trim();
120
121        #[cfg(target_os = "windows")]
122        {
123            let is_absolute = trimmed.starts_with(r"\\?\");
124            if is_absolute {
125                return Self::parse_absolute_path(trimmed, parse_path_like_str);
126            }
127        }
128
129        match trimmed.split_once(FILE_ROW_COLUMN_DELIMITER) {
130            Some((path_like_str, maybe_row_and_col_str)) => {
131                let path_like_str = path_like_str.trim();
132                let maybe_row_and_col_str = maybe_row_and_col_str.trim();
133                if path_like_str.is_empty() {
134                    fallback(s)
135                } else if maybe_row_and_col_str.is_empty() {
136                    fallback(path_like_str)
137                } else {
138                    let (row_parse_result, maybe_col_str) =
139                        match maybe_row_and_col_str.split_once(FILE_ROW_COLUMN_DELIMITER) {
140                            Some((maybe_row_str, maybe_col_str)) => {
141                                (maybe_row_str.parse::<u32>(), maybe_col_str.trim())
142                            }
143                            None => (maybe_row_and_col_str.parse::<u32>(), ""),
144                        };
145
146                    match row_parse_result {
147                        Ok(row) => {
148                            if maybe_col_str.is_empty() {
149                                Ok(Self {
150                                    path_like: parse_path_like_str(path_like_str)?,
151                                    row: Some(row),
152                                    column: None,
153                                })
154                            } else {
155                                let (maybe_col_str, _) =
156                                    maybe_col_str.split_once(':').unwrap_or((maybe_col_str, ""));
157                                match maybe_col_str.parse::<u32>() {
158                                    Ok(col) => Ok(Self {
159                                        path_like: parse_path_like_str(path_like_str)?,
160                                        row: Some(row),
161                                        column: Some(col),
162                                    }),
163                                    Err(_) => Ok(Self {
164                                        path_like: parse_path_like_str(path_like_str)?,
165                                        row: Some(row),
166                                        column: None,
167                                    }),
168                                }
169                            }
170                        }
171                        Err(_) => Ok(Self {
172                            path_like: parse_path_like_str(path_like_str)?,
173                            row: None,
174                            column: None,
175                        }),
176                    }
177                }
178            }
179            None => fallback(s),
180        }
181    }
182
183    /// This helper function is used for parsing absolute paths on Windows. It exists because absolute paths on Windows are quite different from other platforms. See [this page](https://learn.microsoft.com/en-us/dotnet/standard/io/file-path-formats#dos-device-paths) for more information.
184    #[cfg(target_os = "windows")]
185    fn parse_absolute_path<E>(
186        s: &str,
187        parse_path_like_str: impl Fn(&str) -> Result<P, E>,
188    ) -> Result<Self, E> {
189        let fallback = |fallback_str| {
190            Ok(Self {
191                path_like: parse_path_like_str(fallback_str)?,
192                row: None,
193                column: None,
194            })
195        };
196
197        let mut iterator = s.split(FILE_ROW_COLUMN_DELIMITER);
198
199        let drive_prefix = iterator.next().unwrap_or_default();
200        let file_path = iterator.next().unwrap_or_default();
201
202        // TODO: How to handle drives without a letter? UNC paths?
203        let complete_path = drive_prefix.replace("\\\\?\\", "") + ":" + &file_path;
204
205        if let Some(row_str) = iterator.next() {
206            if let Some(column_str) = iterator.next() {
207                match row_str.parse::<u32>() {
208                    Ok(row) => match column_str.parse::<u32>() {
209                        Ok(col) => {
210                            return Ok(Self {
211                                path_like: parse_path_like_str(&complete_path)?,
212                                row: Some(row),
213                                column: Some(col),
214                            });
215                        }
216
217                        Err(_) => {
218                            return Ok(Self {
219                                path_like: parse_path_like_str(&complete_path)?,
220                                row: Some(row),
221                                column: None,
222                            });
223                        }
224                    },
225
226                    Err(_) => {
227                        return fallback(&complete_path);
228                    }
229                }
230            }
231        }
232        return fallback(&complete_path);
233    }
234
235    pub fn map_path_like<P2, E>(
236        self,
237        mapping: impl FnOnce(P) -> Result<P2, E>,
238    ) -> Result<PathLikeWithPosition<P2>, E> {
239        Ok(PathLikeWithPosition {
240            path_like: mapping(self.path_like)?,
241            row: self.row,
242            column: self.column,
243        })
244    }
245
246    pub fn to_string(&self, path_like_to_string: impl Fn(&P) -> String) -> String {
247        let path_like_string = path_like_to_string(&self.path_like);
248        if let Some(row) = self.row {
249            if let Some(column) = self.column {
250                format!("{path_like_string}:{row}:{column}")
251            } else {
252                format!("{path_like_string}:{row}")
253            }
254        } else {
255            path_like_string
256        }
257    }
258}
259
260#[derive(Clone, Debug)]
261pub struct PathMatcher {
262    source: String,
263    glob: GlobMatcher,
264}
265
266impl std::fmt::Display for PathMatcher {
267    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
268        self.source.fmt(f)
269    }
270}
271
272impl PartialEq for PathMatcher {
273    fn eq(&self, other: &Self) -> bool {
274        self.source.eq(&other.source)
275    }
276}
277
278impl Eq for PathMatcher {}
279
280impl PathMatcher {
281    pub fn new(source: &str) -> Result<Self, globset::Error> {
282        Ok(PathMatcher {
283            glob: Glob::new(source)?.compile_matcher(),
284            source: String::from(source),
285        })
286    }
287
288    pub fn source(&self) -> &str {
289        &self.source
290    }
291
292    pub fn is_match<P: AsRef<Path>>(&self, other: P) -> bool {
293        let other_path = other.as_ref();
294        other_path.starts_with(Path::new(&self.source))
295            || other_path.ends_with(Path::new(&self.source))
296            || self.glob.is_match(other_path)
297            || self.check_with_end_separator(other_path)
298    }
299
300    fn check_with_end_separator(&self, path: &Path) -> bool {
301        let path_str = path.to_string_lossy();
302        let separator = std::path::MAIN_SEPARATOR_STR;
303        if path_str.ends_with(separator) {
304            self.glob.is_match(path)
305        } else {
306            self.glob.is_match(path_str.to_string() + separator)
307        }
308    }
309}
310
311#[cfg(test)]
312mod tests {
313    use super::*;
314
315    type TestPath = PathLikeWithPosition<String>;
316
317    fn parse_str(s: &str) -> TestPath {
318        TestPath::parse_str(s, |s| Ok::<_, std::convert::Infallible>(s.to_string()))
319            .expect("infallible")
320    }
321
322    #[test]
323    fn path_with_position_parsing_positive() {
324        let input_and_expected = [
325            (
326                "test_file.rs",
327                PathLikeWithPosition {
328                    path_like: "test_file.rs".to_string(),
329                    row: None,
330                    column: None,
331                },
332            ),
333            (
334                "test_file.rs:1",
335                PathLikeWithPosition {
336                    path_like: "test_file.rs".to_string(),
337                    row: Some(1),
338                    column: None,
339                },
340            ),
341            (
342                "test_file.rs:1:2",
343                PathLikeWithPosition {
344                    path_like: "test_file.rs".to_string(),
345                    row: Some(1),
346                    column: Some(2),
347                },
348            ),
349        ];
350
351        for (input, expected) in input_and_expected {
352            let actual = parse_str(input);
353            assert_eq!(
354                actual, expected,
355                "For positive case input str '{input}', got a parse mismatch"
356            );
357        }
358    }
359
360    #[test]
361    fn path_with_position_parsing_negative() {
362        for (input, row, column) in [
363            ("test_file.rs:a", None, None),
364            ("test_file.rs:a:b", None, None),
365            ("test_file.rs::", None, None),
366            ("test_file.rs::1", None, None),
367            ("test_file.rs:1::", Some(1), None),
368            ("test_file.rs::1:2", None, None),
369            ("test_file.rs:1::2", Some(1), None),
370            ("test_file.rs:1:2:3", Some(1), Some(2)),
371        ] {
372            let actual = parse_str(input);
373            assert_eq!(
374                actual,
375                PathLikeWithPosition {
376                    path_like: "test_file.rs".to_string(),
377                    row,
378                    column,
379                },
380                "For negative case input str '{input}', got a parse mismatch"
381            );
382        }
383    }
384
385    // Trim off trailing `:`s for otherwise valid input.
386    #[test]
387    fn path_with_position_parsing_special() {
388        #[cfg(not(target_os = "windows"))]
389        let input_and_expected = [
390            (
391                "test_file.rs:",
392                PathLikeWithPosition {
393                    path_like: "test_file.rs".to_string(),
394                    row: None,
395                    column: None,
396                },
397            ),
398            (
399                "test_file.rs:1:",
400                PathLikeWithPosition {
401                    path_like: "test_file.rs".to_string(),
402                    row: Some(1),
403                    column: None,
404                },
405            ),
406            (
407                "crates/file_finder/src/file_finder.rs:1902:13:",
408                PathLikeWithPosition {
409                    path_like: "crates/file_finder/src/file_finder.rs".to_string(),
410                    row: Some(1902),
411                    column: Some(13),
412                },
413            ),
414        ];
415
416        #[cfg(target_os = "windows")]
417        let input_and_expected = [
418            (
419                "test_file.rs:",
420                PathLikeWithPosition {
421                    path_like: "test_file.rs".to_string(),
422                    row: None,
423                    column: None,
424                },
425            ),
426            (
427                "test_file.rs:1:",
428                PathLikeWithPosition {
429                    path_like: "test_file.rs".to_string(),
430                    row: Some(1),
431                    column: None,
432                },
433            ),
434            (
435                "\\\\?\\C:\\Users\\someone\\test_file.rs:1902:13:",
436                PathLikeWithPosition {
437                    path_like: "C:\\Users\\someone\\test_file.rs".to_string(),
438                    row: Some(1902),
439                    column: Some(13),
440                },
441            ),
442            (
443                "\\\\?\\C:\\Users\\someone\\test_file.rs:1902:13:15:",
444                PathLikeWithPosition {
445                    path_like: "C:\\Users\\someone\\test_file.rs".to_string(),
446                    row: Some(1902),
447                    column: Some(13),
448                },
449            ),
450            (
451                "\\\\?\\C:\\Users\\someone\\test_file.rs:1902:::15:",
452                PathLikeWithPosition {
453                    path_like: "C:\\Users\\someone\\test_file.rs".to_string(),
454                    row: Some(1902),
455                    column: None,
456                },
457            ),
458        ];
459
460        for (input, expected) in input_and_expected {
461            let actual = parse_str(input);
462            assert_eq!(
463                actual, expected,
464                "For special case input str '{input}', got a parse mismatch"
465            );
466        }
467    }
468
469    #[test]
470    fn test_path_compact() {
471        let path: PathBuf = [
472            HOME.to_string_lossy().to_string(),
473            "some_file.txt".to_string(),
474        ]
475        .iter()
476        .collect();
477        if cfg!(target_os = "linux") || cfg!(target_os = "macos") {
478            assert_eq!(path.compact().to_str(), Some("~/some_file.txt"));
479        } else {
480            assert_eq!(path.compact().to_str(), path.to_str());
481        }
482    }
483
484    #[test]
485    fn test_icon_stem_or_suffix() {
486        // No dots in name
487        let path = Path::new("/a/b/c/file_name.rs");
488        assert_eq!(path.icon_stem_or_suffix(), Some("rs"));
489
490        // Single dot in name
491        let path = Path::new("/a/b/c/file.name.rs");
492        assert_eq!(path.icon_stem_or_suffix(), Some("rs"));
493
494        // No suffix
495        let path = Path::new("/a/b/c/file");
496        assert_eq!(path.icon_stem_or_suffix(), Some("file"));
497
498        // Multiple dots in name
499        let path = Path::new("/a/b/c/long.file.name.rs");
500        assert_eq!(path.icon_stem_or_suffix(), Some("rs"));
501
502        // Hidden file, no extension
503        let path = Path::new("/a/b/c/.gitignore");
504        assert_eq!(path.icon_stem_or_suffix(), Some("gitignore"));
505
506        // Hidden file, with extension
507        let path = Path::new("/a/b/c/.eslintrc.js");
508        assert_eq!(path.icon_stem_or_suffix(), Some("eslintrc.js"));
509    }
510
511    #[test]
512    fn test_extension_or_hidden_file_name() {
513        // No dots in name
514        let path = Path::new("/a/b/c/file_name.rs");
515        assert_eq!(path.extension_or_hidden_file_name(), Some("rs"));
516
517        // Single dot in name
518        let path = Path::new("/a/b/c/file.name.rs");
519        assert_eq!(path.extension_or_hidden_file_name(), Some("rs"));
520
521        // Multiple dots in name
522        let path = Path::new("/a/b/c/long.file.name.rs");
523        assert_eq!(path.extension_or_hidden_file_name(), Some("rs"));
524
525        // Hidden file, no extension
526        let path = Path::new("/a/b/c/.gitignore");
527        assert_eq!(path.extension_or_hidden_file_name(), Some("gitignore"));
528
529        // Hidden file, with extension
530        let path = Path::new("/a/b/c/.eslintrc.js");
531        assert_eq!(path.extension_or_hidden_file_name(), Some("js"));
532    }
533
534    #[test]
535    fn edge_of_glob() {
536        let path = Path::new("/work/node_modules");
537        let path_matcher = PathMatcher::new("**/node_modules/**").unwrap();
538        assert!(
539            path_matcher.is_match(path),
540            "Path matcher {path_matcher} should match {path:?}"
541        );
542    }
543
544    #[test]
545    fn project_search() {
546        let path = Path::new("/Users/someonetoignore/work/zed/zed.dev/node_modules");
547        let path_matcher = PathMatcher::new("**/node_modules/**").unwrap();
548        assert!(
549            path_matcher.is_match(path),
550            "Path matcher {path_matcher} should match {path:?}"
551        );
552    }
553}