| 1 | // Copyright 2014 The Rust Project Developers. See the COPYRIGHT |
| 2 | // file at the top-level directory of this distribution and at |
| 3 | // http://rust-lang.org/COPYRIGHT. |
| 4 | // |
| 5 | // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or |
| 6 | // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license |
| 7 | // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your |
| 8 | // option. This file may not be copied, modified, or distributed |
| 9 | // except according to those terms. |
| 10 | |
| 11 | //! Support for matching file paths against Unix shell style patterns. |
| 12 | //! |
| 13 | //! The `glob` and `glob_with` functions allow querying the filesystem for all |
| 14 | //! files that match a particular pattern (similar to the libc `glob` function). |
| 15 | //! The methods on the `Pattern` type provide functionality for checking if |
| 16 | //! individual paths match a particular pattern (similar to the libc `fnmatch` |
| 17 | //! function). |
| 18 | //! |
| 19 | //! For consistency across platforms, and for Windows support, this module |
| 20 | //! is implemented entirely in Rust rather than deferring to the libc |
| 21 | //! `glob`/`fnmatch` functions. |
| 22 | //! |
| 23 | //! # Examples |
| 24 | //! |
| 25 | //! To print all jpg files in `/media/` and all of its subdirectories. |
| 26 | //! |
| 27 | //! ```rust,no_run |
| 28 | //! use glob::glob; |
| 29 | //! |
| 30 | //! for entry in glob("/media/**/*.jpg" ).expect("Failed to read glob pattern" ) { |
| 31 | //! match entry { |
| 32 | //! Ok(path) => println!("{:?}" , path.display()), |
| 33 | //! Err(e) => println!("{:?}" , e), |
| 34 | //! } |
| 35 | //! } |
| 36 | //! ``` |
| 37 | //! |
| 38 | //! To print all files containing the letter "a", case insensitive, in a `local` |
| 39 | //! directory relative to the current working directory. This ignores errors |
| 40 | //! instead of printing them. |
| 41 | //! |
| 42 | //! ```rust,no_run |
| 43 | //! use glob::glob_with; |
| 44 | //! use glob::MatchOptions; |
| 45 | //! |
| 46 | //! let options = MatchOptions { |
| 47 | //! case_sensitive: false, |
| 48 | //! require_literal_separator: false, |
| 49 | //! require_literal_leading_dot: false, |
| 50 | //! }; |
| 51 | //! for entry in glob_with("local/*a*" , options).unwrap() { |
| 52 | //! if let Ok(path) = entry { |
| 53 | //! println!("{:?}" , path.display()) |
| 54 | //! } |
| 55 | //! } |
| 56 | //! ``` |
| 57 | |
| 58 | #![doc ( |
| 59 | html_logo_url = "https://www.rust-lang.org/logos/rust-logo-128x128-blk-v2.png" , |
| 60 | html_favicon_url = "https://www.rust-lang.org/favicon.ico" , |
| 61 | html_root_url = "https://docs.rs/glob/0.3.1" |
| 62 | )] |
| 63 | #![deny (missing_docs)] |
| 64 | |
| 65 | #[cfg (test)] |
| 66 | #[macro_use ] |
| 67 | extern crate doc_comment; |
| 68 | |
| 69 | #[cfg (test)] |
| 70 | doctest!("../README.md" ); |
| 71 | |
| 72 | use std::cmp; |
| 73 | use std::error::Error; |
| 74 | use std::fmt; |
| 75 | use std::fs; |
| 76 | use std::fs::DirEntry; |
| 77 | use std::io; |
| 78 | use std::ops::Deref; |
| 79 | use std::path::{self, Component, Path, PathBuf}; |
| 80 | use std::str::FromStr; |
| 81 | |
| 82 | use CharSpecifier::{CharRange, SingleChar}; |
| 83 | use MatchResult::{EntirePatternDoesntMatch, Match, SubPatternDoesntMatch}; |
| 84 | use PatternToken::AnyExcept; |
| 85 | use PatternToken::{AnyChar, AnyRecursiveSequence, AnySequence, AnyWithin, Char}; |
| 86 | |
| 87 | /// An iterator that yields `Path`s from the filesystem that match a particular |
| 88 | /// pattern. |
| 89 | /// |
| 90 | /// Note that it yields `GlobResult` in order to report any `IoErrors` that may |
| 91 | /// arise during iteration. If a directory matches but is unreadable, |
| 92 | /// thereby preventing its contents from being checked for matches, a |
| 93 | /// `GlobError` is returned to express this. |
| 94 | /// |
| 95 | /// See the `glob` function for more details. |
| 96 | #[derive (Debug)] |
| 97 | pub struct Paths { |
| 98 | dir_patterns: Vec<Pattern>, |
| 99 | require_dir: bool, |
| 100 | options: MatchOptions, |
| 101 | todo: Vec<Result<(PathWrapper, usize), GlobError>>, |
| 102 | scope: Option<PathWrapper>, |
| 103 | } |
| 104 | |
| 105 | /// Return an iterator that produces all the `Path`s that match the given |
| 106 | /// pattern using default match options, which may be absolute or relative to |
| 107 | /// the current working directory. |
| 108 | /// |
| 109 | /// This may return an error if the pattern is invalid. |
| 110 | /// |
| 111 | /// This method uses the default match options and is equivalent to calling |
| 112 | /// `glob_with(pattern, MatchOptions::new())`. Use `glob_with` directly if you |
| 113 | /// want to use non-default match options. |
| 114 | /// |
| 115 | /// When iterating, each result is a `GlobResult` which expresses the |
| 116 | /// possibility that there was an `IoError` when attempting to read the contents |
| 117 | /// of the matched path. In other words, each item returned by the iterator |
| 118 | /// will either be an `Ok(Path)` if the path matched, or an `Err(GlobError)` if |
| 119 | /// the path (partially) matched _but_ its contents could not be read in order |
| 120 | /// to determine if its contents matched. |
| 121 | /// |
| 122 | /// See the `Paths` documentation for more information. |
| 123 | /// |
| 124 | /// # Examples |
| 125 | /// |
| 126 | /// Consider a directory `/media/pictures` containing only the files |
| 127 | /// `kittens.jpg`, `puppies.jpg` and `hamsters.gif`: |
| 128 | /// |
| 129 | /// ```rust,no_run |
| 130 | /// use glob::glob; |
| 131 | /// |
| 132 | /// for entry in glob("/media/pictures/*.jpg" ).unwrap() { |
| 133 | /// match entry { |
| 134 | /// Ok(path) => println!("{:?}" , path.display()), |
| 135 | /// |
| 136 | /// // if the path matched but was unreadable, |
| 137 | /// // thereby preventing its contents from matching |
| 138 | /// Err(e) => println!("{:?}" , e), |
| 139 | /// } |
| 140 | /// } |
| 141 | /// ``` |
| 142 | /// |
| 143 | /// The above code will print: |
| 144 | /// |
| 145 | /// ```ignore |
| 146 | /// /media/pictures/kittens.jpg |
| 147 | /// /media/pictures/puppies.jpg |
| 148 | /// ``` |
| 149 | /// |
| 150 | /// If you want to ignore unreadable paths, you can use something like |
| 151 | /// `filter_map`: |
| 152 | /// |
| 153 | /// ```rust |
| 154 | /// use glob::glob; |
| 155 | /// use std::result::Result; |
| 156 | /// |
| 157 | /// for path in glob("/media/pictures/*.jpg" ).unwrap().filter_map(Result::ok) { |
| 158 | /// println!("{}" , path.display()); |
| 159 | /// } |
| 160 | /// ``` |
| 161 | /// Paths are yielded in alphabetical order. |
| 162 | pub fn glob(pattern: &str) -> Result<Paths, PatternError> { |
| 163 | glob_with(pattern, options:MatchOptions::new()) |
| 164 | } |
| 165 | |
| 166 | /// Return an iterator that produces all the `Path`s that match the given |
| 167 | /// pattern using the specified match options, which may be absolute or relative |
| 168 | /// to the current working directory. |
| 169 | /// |
| 170 | /// This may return an error if the pattern is invalid. |
| 171 | /// |
| 172 | /// This function accepts Unix shell style patterns as described by |
| 173 | /// `Pattern::new(..)`. The options given are passed through unchanged to |
| 174 | /// `Pattern::matches_with(..)` with the exception that |
| 175 | /// `require_literal_separator` is always set to `true` regardless of the value |
| 176 | /// passed to this function. |
| 177 | /// |
| 178 | /// Paths are yielded in alphabetical order. |
| 179 | pub fn glob_with(pattern: &str, options: MatchOptions) -> Result<Paths, PatternError> { |
| 180 | #[cfg (windows)] |
| 181 | fn check_windows_verbatim(p: &Path) -> bool { |
| 182 | match p.components().next() { |
| 183 | Some(Component::Prefix(ref p)) => { |
| 184 | // Allow VerbatimDisk paths. std canonicalize() generates them, and they work fine |
| 185 | p.kind().is_verbatim() |
| 186 | && if let std::path::Prefix::VerbatimDisk(_) = p.kind() { |
| 187 | false |
| 188 | } else { |
| 189 | true |
| 190 | } |
| 191 | } |
| 192 | _ => false, |
| 193 | } |
| 194 | } |
| 195 | #[cfg (not(windows))] |
| 196 | fn check_windows_verbatim(_: &Path) -> bool { |
| 197 | false |
| 198 | } |
| 199 | |
| 200 | #[cfg (windows)] |
| 201 | fn to_scope(p: &Path) -> PathBuf { |
| 202 | // FIXME handle volume relative paths here |
| 203 | p.to_path_buf() |
| 204 | } |
| 205 | #[cfg (not(windows))] |
| 206 | fn to_scope(p: &Path) -> PathBuf { |
| 207 | p.to_path_buf() |
| 208 | } |
| 209 | |
| 210 | // make sure that the pattern is valid first, else early return with error |
| 211 | if let Err(err) = Pattern::new(pattern) { |
| 212 | return Err(err); |
| 213 | } |
| 214 | |
| 215 | let mut components = Path::new(pattern).components().peekable(); |
| 216 | loop { |
| 217 | match components.peek() { |
| 218 | Some(&Component::Prefix(..)) | Some(&Component::RootDir) => { |
| 219 | components.next(); |
| 220 | } |
| 221 | _ => break, |
| 222 | } |
| 223 | } |
| 224 | let rest = components.map(|s| s.as_os_str()).collect::<PathBuf>(); |
| 225 | let normalized_pattern = Path::new(pattern).iter().collect::<PathBuf>(); |
| 226 | let root_len = normalized_pattern.to_str().unwrap().len() - rest.to_str().unwrap().len(); |
| 227 | let root = if root_len > 0 { |
| 228 | Some(Path::new(&pattern[..root_len])) |
| 229 | } else { |
| 230 | None |
| 231 | }; |
| 232 | |
| 233 | if root_len > 0 && check_windows_verbatim(root.unwrap()) { |
| 234 | // FIXME: How do we want to handle verbatim paths? I'm inclined to |
| 235 | // return nothing, since we can't very well find all UNC shares with a |
| 236 | // 1-letter server name. |
| 237 | return Ok(Paths { |
| 238 | dir_patterns: Vec::new(), |
| 239 | require_dir: false, |
| 240 | options, |
| 241 | todo: Vec::new(), |
| 242 | scope: None, |
| 243 | }); |
| 244 | } |
| 245 | |
| 246 | let scope = root.map_or_else(|| PathBuf::from("." ), to_scope); |
| 247 | let scope = PathWrapper::from_path(scope); |
| 248 | |
| 249 | let mut dir_patterns = Vec::new(); |
| 250 | let components = |
| 251 | pattern[cmp::min(root_len, pattern.len())..].split_terminator(path::is_separator); |
| 252 | |
| 253 | for component in components { |
| 254 | dir_patterns.push(Pattern::new(component)?); |
| 255 | } |
| 256 | |
| 257 | if root_len == pattern.len() { |
| 258 | dir_patterns.push(Pattern { |
| 259 | original: "" .to_string(), |
| 260 | tokens: Vec::new(), |
| 261 | is_recursive: false, |
| 262 | }); |
| 263 | } |
| 264 | |
| 265 | let last_is_separator = pattern.chars().next_back().map(path::is_separator); |
| 266 | let require_dir = last_is_separator == Some(true); |
| 267 | let todo = Vec::new(); |
| 268 | |
| 269 | Ok(Paths { |
| 270 | dir_patterns, |
| 271 | require_dir, |
| 272 | options, |
| 273 | todo, |
| 274 | scope: Some(scope), |
| 275 | }) |
| 276 | } |
| 277 | |
| 278 | /// A glob iteration error. |
| 279 | /// |
| 280 | /// This is typically returned when a particular path cannot be read |
| 281 | /// to determine if its contents match the glob pattern. This is possible |
| 282 | /// if the program lacks the appropriate permissions, for example. |
| 283 | #[derive (Debug)] |
| 284 | pub struct GlobError { |
| 285 | path: PathBuf, |
| 286 | error: io::Error, |
| 287 | } |
| 288 | |
| 289 | impl GlobError { |
| 290 | /// The Path that the error corresponds to. |
| 291 | pub fn path(&self) -> &Path { |
| 292 | &self.path |
| 293 | } |
| 294 | |
| 295 | /// The error in question. |
| 296 | pub fn error(&self) -> &io::Error { |
| 297 | &self.error |
| 298 | } |
| 299 | |
| 300 | /// Consumes self, returning the _raw_ underlying `io::Error` |
| 301 | pub fn into_error(self) -> io::Error { |
| 302 | self.error |
| 303 | } |
| 304 | } |
| 305 | |
| 306 | impl Error for GlobError { |
| 307 | #[allow (deprecated)] |
| 308 | fn description(&self) -> &str { |
| 309 | self.error.description() |
| 310 | } |
| 311 | |
| 312 | #[allow (unknown_lints, bare_trait_objects)] |
| 313 | fn cause(&self) -> Option<&dynError> { |
| 314 | Some(&self.error) |
| 315 | } |
| 316 | } |
| 317 | |
| 318 | impl fmt::Display for GlobError { |
| 319 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
| 320 | write!( |
| 321 | f, |
| 322 | "attempting to read ` {}` resulted in an error: {}" , |
| 323 | self.path.display(), |
| 324 | self.error |
| 325 | ) |
| 326 | } |
| 327 | } |
| 328 | |
| 329 | #[derive (Debug)] |
| 330 | struct PathWrapper { |
| 331 | path: PathBuf, |
| 332 | is_directory: bool, |
| 333 | } |
| 334 | |
| 335 | impl PathWrapper { |
| 336 | fn from_dir_entry(path: PathBuf, e: DirEntry) -> Self { |
| 337 | let is_directory = e |
| 338 | .file_type() |
| 339 | .ok() |
| 340 | .and_then(|file_type| { |
| 341 | // We need to use fs::metadata to resolve the actual path |
| 342 | // if it's a symlink. |
| 343 | if file_type.is_symlink() { |
| 344 | None |
| 345 | } else { |
| 346 | Some(file_type.is_dir()) |
| 347 | } |
| 348 | }) |
| 349 | .or_else(|| fs::metadata(&path).map(|m| m.is_dir()).ok()) |
| 350 | .unwrap_or(false); |
| 351 | Self { path, is_directory } |
| 352 | } |
| 353 | fn from_path(path: PathBuf) -> Self { |
| 354 | let is_directory = fs::metadata(&path).map(|m| m.is_dir()).unwrap_or(false); |
| 355 | Self { path, is_directory } |
| 356 | } |
| 357 | |
| 358 | fn into_path(self) -> PathBuf { |
| 359 | self.path |
| 360 | } |
| 361 | } |
| 362 | |
| 363 | impl Deref for PathWrapper { |
| 364 | type Target = Path; |
| 365 | |
| 366 | fn deref(&self) -> &Self::Target { |
| 367 | self.path.deref() |
| 368 | } |
| 369 | } |
| 370 | |
| 371 | impl AsRef<Path> for PathWrapper { |
| 372 | fn as_ref(&self) -> &Path { |
| 373 | self.path.as_ref() |
| 374 | } |
| 375 | } |
| 376 | |
| 377 | /// An alias for a glob iteration result. |
| 378 | /// |
| 379 | /// This represents either a matched path or a glob iteration error, |
| 380 | /// such as failing to read a particular directory's contents. |
| 381 | pub type GlobResult = Result<PathBuf, GlobError>; |
| 382 | |
| 383 | impl Iterator for Paths { |
| 384 | type Item = GlobResult; |
| 385 | |
| 386 | fn next(&mut self) -> Option<GlobResult> { |
| 387 | // the todo buffer hasn't been initialized yet, so it's done at this |
| 388 | // point rather than in glob() so that the errors are unified that is, |
| 389 | // failing to fill the buffer is an iteration error construction of the |
| 390 | // iterator (i.e. glob()) only fails if it fails to compile the Pattern |
| 391 | if let Some(scope) = self.scope.take() { |
| 392 | if !self.dir_patterns.is_empty() { |
| 393 | // Shouldn't happen, but we're using -1 as a special index. |
| 394 | assert!(self.dir_patterns.len() < !0 as usize); |
| 395 | |
| 396 | fill_todo(&mut self.todo, &self.dir_patterns, 0, &scope, self.options); |
| 397 | } |
| 398 | } |
| 399 | |
| 400 | loop { |
| 401 | if self.dir_patterns.is_empty() || self.todo.is_empty() { |
| 402 | return None; |
| 403 | } |
| 404 | |
| 405 | let (path, mut idx) = match self.todo.pop().unwrap() { |
| 406 | Ok(pair) => pair, |
| 407 | Err(e) => return Some(Err(e)), |
| 408 | }; |
| 409 | |
| 410 | // idx -1: was already checked by fill_todo, maybe path was '.' or |
| 411 | // '..' that we can't match here because of normalization. |
| 412 | if idx == !0 as usize { |
| 413 | if self.require_dir && !path.is_directory { |
| 414 | continue; |
| 415 | } |
| 416 | return Some(Ok(path.into_path())); |
| 417 | } |
| 418 | |
| 419 | if self.dir_patterns[idx].is_recursive { |
| 420 | let mut next = idx; |
| 421 | |
| 422 | // collapse consecutive recursive patterns |
| 423 | while (next + 1) < self.dir_patterns.len() |
| 424 | && self.dir_patterns[next + 1].is_recursive |
| 425 | { |
| 426 | next += 1; |
| 427 | } |
| 428 | |
| 429 | if path.is_directory { |
| 430 | // the path is a directory, so it's a match |
| 431 | |
| 432 | // push this directory's contents |
| 433 | fill_todo( |
| 434 | &mut self.todo, |
| 435 | &self.dir_patterns, |
| 436 | next, |
| 437 | &path, |
| 438 | self.options, |
| 439 | ); |
| 440 | |
| 441 | if next == self.dir_patterns.len() - 1 { |
| 442 | // pattern ends in recursive pattern, so return this |
| 443 | // directory as a result |
| 444 | return Some(Ok(path.into_path())); |
| 445 | } else { |
| 446 | // advanced to the next pattern for this path |
| 447 | idx = next + 1; |
| 448 | } |
| 449 | } else if next == self.dir_patterns.len() - 1 { |
| 450 | // not a directory and it's the last pattern, meaning no |
| 451 | // match |
| 452 | continue; |
| 453 | } else { |
| 454 | // advanced to the next pattern for this path |
| 455 | idx = next + 1; |
| 456 | } |
| 457 | } |
| 458 | |
| 459 | // not recursive, so match normally |
| 460 | if self.dir_patterns[idx].matches_with( |
| 461 | { |
| 462 | match path.file_name().and_then(|s| s.to_str()) { |
| 463 | // FIXME (#9639): How do we handle non-utf8 filenames? |
| 464 | // Ignore them for now; ideally we'd still match them |
| 465 | // against a * |
| 466 | None => continue, |
| 467 | Some(x) => x, |
| 468 | } |
| 469 | }, |
| 470 | self.options, |
| 471 | ) { |
| 472 | if idx == self.dir_patterns.len() - 1 { |
| 473 | // it is not possible for a pattern to match a directory |
| 474 | // *AND* its children so we don't need to check the |
| 475 | // children |
| 476 | |
| 477 | if !self.require_dir || path.is_directory { |
| 478 | return Some(Ok(path.into_path())); |
| 479 | } |
| 480 | } else { |
| 481 | fill_todo( |
| 482 | &mut self.todo, |
| 483 | &self.dir_patterns, |
| 484 | idx + 1, |
| 485 | &path, |
| 486 | self.options, |
| 487 | ); |
| 488 | } |
| 489 | } |
| 490 | } |
| 491 | } |
| 492 | } |
| 493 | |
| 494 | /// A pattern parsing error. |
| 495 | #[derive (Debug)] |
| 496 | #[allow (missing_copy_implementations)] |
| 497 | pub struct PatternError { |
| 498 | /// The approximate character index of where the error occurred. |
| 499 | pub pos: usize, |
| 500 | |
| 501 | /// A message describing the error. |
| 502 | pub msg: &'static str, |
| 503 | } |
| 504 | |
| 505 | impl Error for PatternError { |
| 506 | fn description(&self) -> &str { |
| 507 | self.msg |
| 508 | } |
| 509 | } |
| 510 | |
| 511 | impl fmt::Display for PatternError { |
| 512 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
| 513 | write!( |
| 514 | f, |
| 515 | "Pattern syntax error near position {}: {}" , |
| 516 | self.pos, self.msg |
| 517 | ) |
| 518 | } |
| 519 | } |
| 520 | |
| 521 | /// A compiled Unix shell style pattern. |
| 522 | /// |
| 523 | /// - `?` matches any single character. |
| 524 | /// |
| 525 | /// - `*` matches any (possibly empty) sequence of characters. |
| 526 | /// |
| 527 | /// - `**` matches the current directory and arbitrary |
| 528 | /// subdirectories. To match files in arbitrary subdiretories, use |
| 529 | /// `**/*`. |
| 530 | /// |
| 531 | /// This sequence **must** form a single path component, so both |
| 532 | /// `**a` and `b**` are invalid and will result in an error. A |
| 533 | /// sequence of more than two consecutive `*` characters is also |
| 534 | /// invalid. |
| 535 | /// |
| 536 | /// - `[...]` matches any character inside the brackets. Character sequences |
| 537 | /// can also specify ranges of characters, as ordered by Unicode, so e.g. |
| 538 | /// `[0-9]` specifies any character between 0 and 9 inclusive. An unclosed |
| 539 | /// bracket is invalid. |
| 540 | /// |
| 541 | /// - `[!...]` is the negation of `[...]`, i.e. it matches any characters |
| 542 | /// **not** in the brackets. |
| 543 | /// |
| 544 | /// - The metacharacters `?`, `*`, `[`, `]` can be matched by using brackets |
| 545 | /// (e.g. `[?]`). When a `]` occurs immediately following `[` or `[!` then it |
| 546 | /// is interpreted as being part of, rather then ending, the character set, so |
| 547 | /// `]` and NOT `]` can be matched by `[]]` and `[!]]` respectively. The `-` |
| 548 | /// character can be specified inside a character sequence pattern by placing |
| 549 | /// it at the start or the end, e.g. `[abc-]`. |
| 550 | #[derive (Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Default, Debug)] |
| 551 | pub struct Pattern { |
| 552 | original: String, |
| 553 | tokens: Vec<PatternToken>, |
| 554 | is_recursive: bool, |
| 555 | } |
| 556 | |
| 557 | /// Show the original glob pattern. |
| 558 | impl fmt::Display for Pattern { |
| 559 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
| 560 | self.original.fmt(f) |
| 561 | } |
| 562 | } |
| 563 | |
| 564 | impl FromStr for Pattern { |
| 565 | type Err = PatternError; |
| 566 | |
| 567 | fn from_str(s: &str) -> Result<Self, PatternError> { |
| 568 | Self::new(pattern:s) |
| 569 | } |
| 570 | } |
| 571 | |
| 572 | #[derive (Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Debug)] |
| 573 | enum PatternToken { |
| 574 | Char(char), |
| 575 | AnyChar, |
| 576 | AnySequence, |
| 577 | AnyRecursiveSequence, |
| 578 | AnyWithin(Vec<CharSpecifier>), |
| 579 | AnyExcept(Vec<CharSpecifier>), |
| 580 | } |
| 581 | |
| 582 | #[derive (Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Debug)] |
| 583 | enum CharSpecifier { |
| 584 | SingleChar(char), |
| 585 | CharRange(char, char), |
| 586 | } |
| 587 | |
| 588 | #[derive (Copy, Clone, PartialEq)] |
| 589 | enum MatchResult { |
| 590 | Match, |
| 591 | SubPatternDoesntMatch, |
| 592 | EntirePatternDoesntMatch, |
| 593 | } |
| 594 | |
| 595 | const ERROR_WILDCARDS: &str = "wildcards are either regular `*` or recursive `**`" ; |
| 596 | const ERROR_RECURSIVE_WILDCARDS: &str = "recursive wildcards must form a single path \ |
| 597 | component" ; |
| 598 | const ERROR_INVALID_RANGE: &str = "invalid range pattern" ; |
| 599 | |
| 600 | impl Pattern { |
| 601 | /// This function compiles Unix shell style patterns. |
| 602 | /// |
| 603 | /// An invalid glob pattern will yield a `PatternError`. |
| 604 | pub fn new(pattern: &str) -> Result<Self, PatternError> { |
| 605 | let chars = pattern.chars().collect::<Vec<_>>(); |
| 606 | let mut tokens = Vec::new(); |
| 607 | let mut is_recursive = false; |
| 608 | let mut i = 0; |
| 609 | |
| 610 | while i < chars.len() { |
| 611 | match chars[i] { |
| 612 | '?' => { |
| 613 | tokens.push(AnyChar); |
| 614 | i += 1; |
| 615 | } |
| 616 | '*' => { |
| 617 | let old = i; |
| 618 | |
| 619 | while i < chars.len() && chars[i] == '*' { |
| 620 | i += 1; |
| 621 | } |
| 622 | |
| 623 | let count = i - old; |
| 624 | |
| 625 | if count > 2 { |
| 626 | return Err(PatternError { |
| 627 | pos: old + 2, |
| 628 | msg: ERROR_WILDCARDS, |
| 629 | }); |
| 630 | } else if count == 2 { |
| 631 | // ** can only be an entire path component |
| 632 | // i.e. a/**/b is valid, but a**/b or a/**b is not |
| 633 | // invalid matches are treated literally |
| 634 | let is_valid = if i == 2 || path::is_separator(chars[i - count - 1]) { |
| 635 | // it ends in a '/' |
| 636 | if i < chars.len() && path::is_separator(chars[i]) { |
| 637 | i += 1; |
| 638 | true |
| 639 | // or the pattern ends here |
| 640 | // this enables the existing globbing mechanism |
| 641 | } else if i == chars.len() { |
| 642 | true |
| 643 | // `**` ends in non-separator |
| 644 | } else { |
| 645 | return Err(PatternError { |
| 646 | pos: i, |
| 647 | msg: ERROR_RECURSIVE_WILDCARDS, |
| 648 | }); |
| 649 | } |
| 650 | // `**` begins with non-separator |
| 651 | } else { |
| 652 | return Err(PatternError { |
| 653 | pos: old - 1, |
| 654 | msg: ERROR_RECURSIVE_WILDCARDS, |
| 655 | }); |
| 656 | }; |
| 657 | |
| 658 | if is_valid { |
| 659 | // collapse consecutive AnyRecursiveSequence to a |
| 660 | // single one |
| 661 | |
| 662 | let tokens_len = tokens.len(); |
| 663 | |
| 664 | if !(tokens_len > 1 && tokens[tokens_len - 1] == AnyRecursiveSequence) { |
| 665 | is_recursive = true; |
| 666 | tokens.push(AnyRecursiveSequence); |
| 667 | } |
| 668 | } |
| 669 | } else { |
| 670 | tokens.push(AnySequence); |
| 671 | } |
| 672 | } |
| 673 | '[' => { |
| 674 | if i + 4 <= chars.len() && chars[i + 1] == '!' { |
| 675 | match chars[i + 3..].iter().position(|x| *x == ']' ) { |
| 676 | None => (), |
| 677 | Some(j) => { |
| 678 | let chars = &chars[i + 2..i + 3 + j]; |
| 679 | let cs = parse_char_specifiers(chars); |
| 680 | tokens.push(AnyExcept(cs)); |
| 681 | i += j + 4; |
| 682 | continue; |
| 683 | } |
| 684 | } |
| 685 | } else if i + 3 <= chars.len() && chars[i + 1] != '!' { |
| 686 | match chars[i + 2..].iter().position(|x| *x == ']' ) { |
| 687 | None => (), |
| 688 | Some(j) => { |
| 689 | let cs = parse_char_specifiers(&chars[i + 1..i + 2 + j]); |
| 690 | tokens.push(AnyWithin(cs)); |
| 691 | i += j + 3; |
| 692 | continue; |
| 693 | } |
| 694 | } |
| 695 | } |
| 696 | |
| 697 | // if we get here then this is not a valid range pattern |
| 698 | return Err(PatternError { |
| 699 | pos: i, |
| 700 | msg: ERROR_INVALID_RANGE, |
| 701 | }); |
| 702 | } |
| 703 | c => { |
| 704 | tokens.push(Char(c)); |
| 705 | i += 1; |
| 706 | } |
| 707 | } |
| 708 | } |
| 709 | |
| 710 | Ok(Self { |
| 711 | tokens, |
| 712 | original: pattern.to_string(), |
| 713 | is_recursive, |
| 714 | }) |
| 715 | } |
| 716 | |
| 717 | /// Escape metacharacters within the given string by surrounding them in |
| 718 | /// brackets. The resulting string will, when compiled into a `Pattern`, |
| 719 | /// match the input string and nothing else. |
| 720 | pub fn escape(s: &str) -> String { |
| 721 | let mut escaped = String::new(); |
| 722 | for c in s.chars() { |
| 723 | match c { |
| 724 | // note that ! does not need escaping because it is only special |
| 725 | // inside brackets |
| 726 | '?' | '*' | '[' | ']' => { |
| 727 | escaped.push('[' ); |
| 728 | escaped.push(c); |
| 729 | escaped.push(']' ); |
| 730 | } |
| 731 | c => { |
| 732 | escaped.push(c); |
| 733 | } |
| 734 | } |
| 735 | } |
| 736 | escaped |
| 737 | } |
| 738 | |
| 739 | /// Return if the given `str` matches this `Pattern` using the default |
| 740 | /// match options (i.e. `MatchOptions::new()`). |
| 741 | /// |
| 742 | /// # Examples |
| 743 | /// |
| 744 | /// ```rust |
| 745 | /// use glob::Pattern; |
| 746 | /// |
| 747 | /// assert!(Pattern::new("c?t" ).unwrap().matches("cat" )); |
| 748 | /// assert!(Pattern::new("k[!e]tteh" ).unwrap().matches("kitteh" )); |
| 749 | /// assert!(Pattern::new("d*g" ).unwrap().matches("doog" )); |
| 750 | /// ``` |
| 751 | pub fn matches(&self, str: &str) -> bool { |
| 752 | self.matches_with(str, MatchOptions::new()) |
| 753 | } |
| 754 | |
| 755 | /// Return if the given `Path`, when converted to a `str`, matches this |
| 756 | /// `Pattern` using the default match options (i.e. `MatchOptions::new()`). |
| 757 | pub fn matches_path(&self, path: &Path) -> bool { |
| 758 | // FIXME (#9639): This needs to handle non-utf8 paths |
| 759 | path.to_str().map_or(false, |s| self.matches(s)) |
| 760 | } |
| 761 | |
| 762 | /// Return if the given `str` matches this `Pattern` using the specified |
| 763 | /// match options. |
| 764 | pub fn matches_with(&self, str: &str, options: MatchOptions) -> bool { |
| 765 | self.matches_from(true, str.chars(), 0, options) == Match |
| 766 | } |
| 767 | |
| 768 | /// Return if the given `Path`, when converted to a `str`, matches this |
| 769 | /// `Pattern` using the specified match options. |
| 770 | pub fn matches_path_with(&self, path: &Path, options: MatchOptions) -> bool { |
| 771 | // FIXME (#9639): This needs to handle non-utf8 paths |
| 772 | path.to_str() |
| 773 | .map_or(false, |s| self.matches_with(s, options)) |
| 774 | } |
| 775 | |
| 776 | /// Access the original glob pattern. |
| 777 | pub fn as_str(&self) -> &str { |
| 778 | &self.original |
| 779 | } |
| 780 | |
| 781 | fn matches_from( |
| 782 | &self, |
| 783 | mut follows_separator: bool, |
| 784 | mut file: std::str::Chars, |
| 785 | i: usize, |
| 786 | options: MatchOptions, |
| 787 | ) -> MatchResult { |
| 788 | for (ti, token) in self.tokens[i..].iter().enumerate() { |
| 789 | match *token { |
| 790 | AnySequence | AnyRecursiveSequence => { |
| 791 | // ** must be at the start. |
| 792 | debug_assert!(match *token { |
| 793 | AnyRecursiveSequence => follows_separator, |
| 794 | _ => true, |
| 795 | }); |
| 796 | |
| 797 | // Empty match |
| 798 | match self.matches_from(follows_separator, file.clone(), i + ti + 1, options) { |
| 799 | SubPatternDoesntMatch => (), // keep trying |
| 800 | m => return m, |
| 801 | }; |
| 802 | |
| 803 | while let Some(c) = file.next() { |
| 804 | if follows_separator && options.require_literal_leading_dot && c == '.' { |
| 805 | return SubPatternDoesntMatch; |
| 806 | } |
| 807 | follows_separator = path::is_separator(c); |
| 808 | match *token { |
| 809 | AnyRecursiveSequence if !follows_separator => continue, |
| 810 | AnySequence |
| 811 | if options.require_literal_separator && follows_separator => |
| 812 | { |
| 813 | return SubPatternDoesntMatch |
| 814 | } |
| 815 | _ => (), |
| 816 | } |
| 817 | match self.matches_from( |
| 818 | follows_separator, |
| 819 | file.clone(), |
| 820 | i + ti + 1, |
| 821 | options, |
| 822 | ) { |
| 823 | SubPatternDoesntMatch => (), // keep trying |
| 824 | m => return m, |
| 825 | } |
| 826 | } |
| 827 | } |
| 828 | _ => { |
| 829 | let c = match file.next() { |
| 830 | Some(c) => c, |
| 831 | None => return EntirePatternDoesntMatch, |
| 832 | }; |
| 833 | |
| 834 | let is_sep = path::is_separator(c); |
| 835 | |
| 836 | if !match *token { |
| 837 | AnyChar | AnyWithin(..) | AnyExcept(..) |
| 838 | if (options.require_literal_separator && is_sep) |
| 839 | || (follows_separator |
| 840 | && options.require_literal_leading_dot |
| 841 | && c == '.' ) => |
| 842 | { |
| 843 | false |
| 844 | } |
| 845 | AnyChar => true, |
| 846 | AnyWithin(ref specifiers) => in_char_specifiers(&specifiers, c, options), |
| 847 | AnyExcept(ref specifiers) => !in_char_specifiers(&specifiers, c, options), |
| 848 | Char(c2) => chars_eq(c, c2, options.case_sensitive), |
| 849 | AnySequence | AnyRecursiveSequence => unreachable!(), |
| 850 | } { |
| 851 | return SubPatternDoesntMatch; |
| 852 | } |
| 853 | follows_separator = is_sep; |
| 854 | } |
| 855 | } |
| 856 | } |
| 857 | |
| 858 | // Iter is fused. |
| 859 | if file.next().is_none() { |
| 860 | Match |
| 861 | } else { |
| 862 | SubPatternDoesntMatch |
| 863 | } |
| 864 | } |
| 865 | } |
| 866 | |
| 867 | // Fills `todo` with paths under `path` to be matched by `patterns[idx]`, |
| 868 | // special-casing patterns to match `.` and `..`, and avoiding `readdir()` |
| 869 | // calls when there are no metacharacters in the pattern. |
| 870 | fn fill_todo( |
| 871 | todo: &mut Vec<Result<(PathWrapper, usize), GlobError>>, |
| 872 | patterns: &[Pattern], |
| 873 | idx: usize, |
| 874 | path: &PathWrapper, |
| 875 | options: MatchOptions, |
| 876 | ) { |
| 877 | // convert a pattern that's just many Char(_) to a string |
| 878 | fn pattern_as_str(pattern: &Pattern) -> Option<String> { |
| 879 | let mut s = String::new(); |
| 880 | for token in &pattern.tokens { |
| 881 | match *token { |
| 882 | Char(c) => s.push(c), |
| 883 | _ => return None, |
| 884 | } |
| 885 | } |
| 886 | |
| 887 | Some(s) |
| 888 | } |
| 889 | |
| 890 | let add = |todo: &mut Vec<_>, next_path: PathWrapper| { |
| 891 | if idx + 1 == patterns.len() { |
| 892 | // We know it's good, so don't make the iterator match this path |
| 893 | // against the pattern again. In particular, it can't match |
| 894 | // . or .. globs since these never show up as path components. |
| 895 | todo.push(Ok((next_path, !0 as usize))); |
| 896 | } else { |
| 897 | fill_todo(todo, patterns, idx + 1, &next_path, options); |
| 898 | } |
| 899 | }; |
| 900 | |
| 901 | let pattern = &patterns[idx]; |
| 902 | let is_dir = path.is_directory; |
| 903 | let curdir = path.as_ref() == Path::new("." ); |
| 904 | match pattern_as_str(pattern) { |
| 905 | Some(s) => { |
| 906 | // This pattern component doesn't have any metacharacters, so we |
| 907 | // don't need to read the current directory to know where to |
| 908 | // continue. So instead of passing control back to the iterator, |
| 909 | // we can just check for that one entry and potentially recurse |
| 910 | // right away. |
| 911 | let special = "." == s || ".." == s; |
| 912 | let next_path = if curdir { |
| 913 | PathBuf::from(s) |
| 914 | } else { |
| 915 | path.join(&s) |
| 916 | }; |
| 917 | let next_path = PathWrapper::from_path(next_path); |
| 918 | if (special && is_dir) |
| 919 | || (!special |
| 920 | && (fs::metadata(&next_path).is_ok() |
| 921 | || fs::symlink_metadata(&next_path).is_ok())) |
| 922 | { |
| 923 | add(todo, next_path); |
| 924 | } |
| 925 | } |
| 926 | None if is_dir => { |
| 927 | let dirs = fs::read_dir(path).and_then(|d| { |
| 928 | d.map(|e| { |
| 929 | e.map(|e| { |
| 930 | let path = if curdir { |
| 931 | PathBuf::from(e.path().file_name().unwrap()) |
| 932 | } else { |
| 933 | e.path() |
| 934 | }; |
| 935 | PathWrapper::from_dir_entry(path, e) |
| 936 | }) |
| 937 | }) |
| 938 | .collect::<Result<Vec<_>, _>>() |
| 939 | }); |
| 940 | match dirs { |
| 941 | Ok(mut children) => { |
| 942 | if options.require_literal_leading_dot { |
| 943 | children |
| 944 | .retain(|x| !x.file_name().unwrap().to_str().unwrap().starts_with("." )); |
| 945 | } |
| 946 | children.sort_by(|p1, p2| p2.file_name().cmp(&p1.file_name())); |
| 947 | todo.extend(children.into_iter().map(|x| Ok((x, idx)))); |
| 948 | |
| 949 | // Matching the special directory entries . and .. that |
| 950 | // refer to the current and parent directory respectively |
| 951 | // requires that the pattern has a leading dot, even if the |
| 952 | // `MatchOptions` field `require_literal_leading_dot` is not |
| 953 | // set. |
| 954 | if !pattern.tokens.is_empty() && pattern.tokens[0] == Char('.' ) { |
| 955 | for &special in &["." , ".." ] { |
| 956 | if pattern.matches_with(special, options) { |
| 957 | add(todo, PathWrapper::from_path(path.join(special))); |
| 958 | } |
| 959 | } |
| 960 | } |
| 961 | } |
| 962 | Err(e) => { |
| 963 | todo.push(Err(GlobError { |
| 964 | path: path.to_path_buf(), |
| 965 | error: e, |
| 966 | })); |
| 967 | } |
| 968 | } |
| 969 | } |
| 970 | None => { |
| 971 | // not a directory, nothing more to find |
| 972 | } |
| 973 | } |
| 974 | } |
| 975 | |
| 976 | fn parse_char_specifiers(s: &[char]) -> Vec<CharSpecifier> { |
| 977 | let mut cs: Vec = Vec::new(); |
| 978 | let mut i: usize = 0; |
| 979 | while i < s.len() { |
| 980 | if i + 3 <= s.len() && s[i + 1] == '-' { |
| 981 | cs.push(CharRange(s[i], s[i + 2])); |
| 982 | i += 3; |
| 983 | } else { |
| 984 | cs.push(SingleChar(s[i])); |
| 985 | i += 1; |
| 986 | } |
| 987 | } |
| 988 | cs |
| 989 | } |
| 990 | |
| 991 | fn in_char_specifiers(specifiers: &[CharSpecifier], c: char, options: MatchOptions) -> bool { |
| 992 | for &specifier in specifiers.iter() { |
| 993 | match specifier { |
| 994 | SingleChar(sc) => { |
| 995 | if chars_eq(c, sc, options.case_sensitive) { |
| 996 | return true; |
| 997 | } |
| 998 | } |
| 999 | CharRange(start, end) => { |
| 1000 | // FIXME: work with non-ascii chars properly (issue #1347) |
| 1001 | if !options.case_sensitive && c.is_ascii() && start.is_ascii() && end.is_ascii() { |
| 1002 | let start = start.to_ascii_lowercase(); |
| 1003 | let end = end.to_ascii_lowercase(); |
| 1004 | |
| 1005 | let start_up = start.to_uppercase().next().unwrap(); |
| 1006 | let end_up = end.to_uppercase().next().unwrap(); |
| 1007 | |
| 1008 | // only allow case insensitive matching when |
| 1009 | // both start and end are within a-z or A-Z |
| 1010 | if start != start_up && end != end_up { |
| 1011 | let c = c.to_ascii_lowercase(); |
| 1012 | if c >= start && c <= end { |
| 1013 | return true; |
| 1014 | } |
| 1015 | } |
| 1016 | } |
| 1017 | |
| 1018 | if c >= start && c <= end { |
| 1019 | return true; |
| 1020 | } |
| 1021 | } |
| 1022 | } |
| 1023 | } |
| 1024 | |
| 1025 | false |
| 1026 | } |
| 1027 | |
| 1028 | /// A helper function to determine if two chars are (possibly case-insensitively) equal. |
| 1029 | fn chars_eq(a: char, b: char, case_sensitive: bool) -> bool { |
| 1030 | if cfg!(windows) && path::is_separator(a) && path::is_separator(b) { |
| 1031 | true |
| 1032 | } else if !case_sensitive && a.is_ascii() && b.is_ascii() { |
| 1033 | // FIXME: work with non-ascii chars properly (issue #9084) |
| 1034 | a.to_ascii_lowercase() == b.to_ascii_lowercase() |
| 1035 | } else { |
| 1036 | a == b |
| 1037 | } |
| 1038 | } |
| 1039 | |
| 1040 | /// Configuration options to modify the behaviour of `Pattern::matches_with(..)`. |
| 1041 | #[allow (missing_copy_implementations)] |
| 1042 | #[derive (Debug, Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Default)] |
| 1043 | pub struct MatchOptions { |
| 1044 | /// Whether or not patterns should be matched in a case-sensitive manner. |
| 1045 | /// This currently only considers upper/lower case relationships between |
| 1046 | /// ASCII characters, but in future this might be extended to work with |
| 1047 | /// Unicode. |
| 1048 | pub case_sensitive: bool, |
| 1049 | |
| 1050 | /// Whether or not path-component separator characters (e.g. `/` on |
| 1051 | /// Posix) must be matched by a literal `/`, rather than by `*` or `?` or |
| 1052 | /// `[...]`. |
| 1053 | pub require_literal_separator: bool, |
| 1054 | |
| 1055 | /// Whether or not paths that contain components that start with a `.` |
| 1056 | /// will require that `.` appears literally in the pattern; `*`, `?`, `**`, |
| 1057 | /// or `[...]` will not match. This is useful because such files are |
| 1058 | /// conventionally considered hidden on Unix systems and it might be |
| 1059 | /// desirable to skip them when listing files. |
| 1060 | pub require_literal_leading_dot: bool, |
| 1061 | } |
| 1062 | |
| 1063 | impl MatchOptions { |
| 1064 | /// Constructs a new `MatchOptions` with default field values. This is used |
| 1065 | /// when calling functions that do not take an explicit `MatchOptions` |
| 1066 | /// parameter. |
| 1067 | /// |
| 1068 | /// This function always returns this value: |
| 1069 | /// |
| 1070 | /// ```rust,ignore |
| 1071 | /// MatchOptions { |
| 1072 | /// case_sensitive: true, |
| 1073 | /// require_literal_separator: false, |
| 1074 | /// require_literal_leading_dot: false |
| 1075 | /// } |
| 1076 | /// ``` |
| 1077 | /// |
| 1078 | /// # Note |
| 1079 | /// The behavior of this method doesn't match `default()`'s. This returns |
| 1080 | /// `case_sensitive` as `true` while `default()` does it as `false`. |
| 1081 | // FIXME: Consider unity the behavior with `default()` in a next major release. |
| 1082 | pub fn new() -> Self { |
| 1083 | Self { |
| 1084 | case_sensitive: true, |
| 1085 | require_literal_separator: false, |
| 1086 | require_literal_leading_dot: false, |
| 1087 | } |
| 1088 | } |
| 1089 | } |
| 1090 | |
| 1091 | #[cfg (test)] |
| 1092 | mod test { |
| 1093 | use super::{glob, MatchOptions, Pattern}; |
| 1094 | use std::path::Path; |
| 1095 | |
| 1096 | #[test ] |
| 1097 | fn test_pattern_from_str() { |
| 1098 | assert!("a*b" .parse::<Pattern>().unwrap().matches("a_b" )); |
| 1099 | assert!("a/**b" .parse::<Pattern>().unwrap_err().pos == 4); |
| 1100 | } |
| 1101 | |
| 1102 | #[test ] |
| 1103 | fn test_wildcard_errors() { |
| 1104 | assert!(Pattern::new("a/**b" ).unwrap_err().pos == 4); |
| 1105 | assert!(Pattern::new("a/bc**" ).unwrap_err().pos == 3); |
| 1106 | assert!(Pattern::new("a/*****" ).unwrap_err().pos == 4); |
| 1107 | assert!(Pattern::new("a/b**c**d" ).unwrap_err().pos == 2); |
| 1108 | assert!(Pattern::new("a**b" ).unwrap_err().pos == 0); |
| 1109 | } |
| 1110 | |
| 1111 | #[test ] |
| 1112 | fn test_unclosed_bracket_errors() { |
| 1113 | assert!(Pattern::new("abc[def" ).unwrap_err().pos == 3); |
| 1114 | assert!(Pattern::new("abc[!def" ).unwrap_err().pos == 3); |
| 1115 | assert!(Pattern::new("abc[" ).unwrap_err().pos == 3); |
| 1116 | assert!(Pattern::new("abc[!" ).unwrap_err().pos == 3); |
| 1117 | assert!(Pattern::new("abc[d" ).unwrap_err().pos == 3); |
| 1118 | assert!(Pattern::new("abc[!d" ).unwrap_err().pos == 3); |
| 1119 | assert!(Pattern::new("abc[]" ).unwrap_err().pos == 3); |
| 1120 | assert!(Pattern::new("abc[!]" ).unwrap_err().pos == 3); |
| 1121 | } |
| 1122 | |
| 1123 | #[test ] |
| 1124 | fn test_glob_errors() { |
| 1125 | assert!(glob("a/**b" ).err().unwrap().pos == 4); |
| 1126 | assert!(glob("abc[def" ).err().unwrap().pos == 3); |
| 1127 | } |
| 1128 | |
| 1129 | // this test assumes that there is a /root directory and that |
| 1130 | // the user running this test is not root or otherwise doesn't |
| 1131 | // have permission to read its contents |
| 1132 | #[cfg (all(unix, not(target_os = "macos" )))] |
| 1133 | #[test ] |
| 1134 | fn test_iteration_errors() { |
| 1135 | use std::io; |
| 1136 | let mut iter = glob("/root/*" ).unwrap(); |
| 1137 | |
| 1138 | // GlobErrors shouldn't halt iteration |
| 1139 | let next = iter.next(); |
| 1140 | assert!(next.is_some()); |
| 1141 | |
| 1142 | let err = next.unwrap(); |
| 1143 | assert!(err.is_err()); |
| 1144 | |
| 1145 | let err = err.err().unwrap(); |
| 1146 | assert!(err.path() == Path::new("/root" )); |
| 1147 | assert!(err.error().kind() == io::ErrorKind::PermissionDenied); |
| 1148 | } |
| 1149 | |
| 1150 | #[test ] |
| 1151 | fn test_absolute_pattern() { |
| 1152 | assert!(glob("/" ).unwrap().next().is_some()); |
| 1153 | assert!(glob("//" ).unwrap().next().is_some()); |
| 1154 | |
| 1155 | // assume that the filesystem is not empty! |
| 1156 | assert!(glob("/*" ).unwrap().next().is_some()); |
| 1157 | |
| 1158 | #[cfg (not(windows))] |
| 1159 | fn win() {} |
| 1160 | |
| 1161 | #[cfg (windows)] |
| 1162 | fn win() { |
| 1163 | use std::env::current_dir; |
| 1164 | use std::path::Component; |
| 1165 | |
| 1166 | // check windows absolute paths with host/device components |
| 1167 | let root_with_device = current_dir() |
| 1168 | .ok() |
| 1169 | .and_then(|p| match p.components().next().unwrap() { |
| 1170 | Component::Prefix(prefix_component) => { |
| 1171 | let path = Path::new(prefix_component.as_os_str()); |
| 1172 | path.join("*" ); |
| 1173 | Some(path.to_path_buf()) |
| 1174 | } |
| 1175 | _ => panic!("no prefix in this path" ), |
| 1176 | }) |
| 1177 | .unwrap(); |
| 1178 | // FIXME (#9639): This needs to handle non-utf8 paths |
| 1179 | assert!(glob(root_with_device.as_os_str().to_str().unwrap()) |
| 1180 | .unwrap() |
| 1181 | .next() |
| 1182 | .is_some()); |
| 1183 | } |
| 1184 | win() |
| 1185 | } |
| 1186 | |
| 1187 | #[test ] |
| 1188 | fn test_wildcards() { |
| 1189 | assert!(Pattern::new("a*b" ).unwrap().matches("a_b" )); |
| 1190 | assert!(Pattern::new("a*b*c" ).unwrap().matches("abc" )); |
| 1191 | assert!(!Pattern::new("a*b*c" ).unwrap().matches("abcd" )); |
| 1192 | assert!(Pattern::new("a*b*c" ).unwrap().matches("a_b_c" )); |
| 1193 | assert!(Pattern::new("a*b*c" ).unwrap().matches("a___b___c" )); |
| 1194 | assert!(Pattern::new("abc*abc*abc" ) |
| 1195 | .unwrap() |
| 1196 | .matches("abcabcabcabcabcabcabc" )); |
| 1197 | assert!(!Pattern::new("abc*abc*abc" ) |
| 1198 | .unwrap() |
| 1199 | .matches("abcabcabcabcabcabcabca" )); |
| 1200 | assert!(Pattern::new("a*a*a*a*a*a*a*a*a" ) |
| 1201 | .unwrap() |
| 1202 | .matches("aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" )); |
| 1203 | assert!(Pattern::new("a*b[xyz]c*d" ).unwrap().matches("abxcdbxcddd" )); |
| 1204 | } |
| 1205 | |
| 1206 | #[test ] |
| 1207 | fn test_recursive_wildcards() { |
| 1208 | let pat = Pattern::new("some/**/needle.txt" ).unwrap(); |
| 1209 | assert!(pat.matches("some/needle.txt" )); |
| 1210 | assert!(pat.matches("some/one/needle.txt" )); |
| 1211 | assert!(pat.matches("some/one/two/needle.txt" )); |
| 1212 | assert!(pat.matches("some/other/needle.txt" )); |
| 1213 | assert!(!pat.matches("some/other/notthis.txt" )); |
| 1214 | |
| 1215 | // a single ** should be valid, for globs |
| 1216 | // Should accept anything |
| 1217 | let pat = Pattern::new("**" ).unwrap(); |
| 1218 | assert!(pat.is_recursive); |
| 1219 | assert!(pat.matches("abcde" )); |
| 1220 | assert!(pat.matches("" )); |
| 1221 | assert!(pat.matches(".asdf" )); |
| 1222 | assert!(pat.matches("/x/.asdf" )); |
| 1223 | |
| 1224 | // collapse consecutive wildcards |
| 1225 | let pat = Pattern::new("some/**/**/needle.txt" ).unwrap(); |
| 1226 | assert!(pat.matches("some/needle.txt" )); |
| 1227 | assert!(pat.matches("some/one/needle.txt" )); |
| 1228 | assert!(pat.matches("some/one/two/needle.txt" )); |
| 1229 | assert!(pat.matches("some/other/needle.txt" )); |
| 1230 | assert!(!pat.matches("some/other/notthis.txt" )); |
| 1231 | |
| 1232 | // ** can begin the pattern |
| 1233 | let pat = Pattern::new("**/test" ).unwrap(); |
| 1234 | assert!(pat.matches("one/two/test" )); |
| 1235 | assert!(pat.matches("one/test" )); |
| 1236 | assert!(pat.matches("test" )); |
| 1237 | |
| 1238 | // /** can begin the pattern |
| 1239 | let pat = Pattern::new("/**/test" ).unwrap(); |
| 1240 | assert!(pat.matches("/one/two/test" )); |
| 1241 | assert!(pat.matches("/one/test" )); |
| 1242 | assert!(pat.matches("/test" )); |
| 1243 | assert!(!pat.matches("/one/notthis" )); |
| 1244 | assert!(!pat.matches("/notthis" )); |
| 1245 | |
| 1246 | // Only start sub-patterns on start of path segment. |
| 1247 | let pat = Pattern::new("**/.*" ).unwrap(); |
| 1248 | assert!(pat.matches(".abc" )); |
| 1249 | assert!(pat.matches("abc/.abc" )); |
| 1250 | assert!(!pat.matches("ab.c" )); |
| 1251 | assert!(!pat.matches("abc/ab.c" )); |
| 1252 | } |
| 1253 | |
| 1254 | #[test ] |
| 1255 | fn test_lots_of_files() { |
| 1256 | // this is a good test because it touches lots of differently named files |
| 1257 | glob("/*/*/*/*" ).unwrap().skip(10000).next(); |
| 1258 | } |
| 1259 | |
| 1260 | #[test ] |
| 1261 | fn test_range_pattern() { |
| 1262 | let pat = Pattern::new("a[0-9]b" ).unwrap(); |
| 1263 | for i in 0..10 { |
| 1264 | assert!(pat.matches(&format!("a{}b" , i))); |
| 1265 | } |
| 1266 | assert!(!pat.matches("a_b" )); |
| 1267 | |
| 1268 | let pat = Pattern::new("a[!0-9]b" ).unwrap(); |
| 1269 | for i in 0..10 { |
| 1270 | assert!(!pat.matches(&format!("a{}b" , i))); |
| 1271 | } |
| 1272 | assert!(pat.matches("a_b" )); |
| 1273 | |
| 1274 | let pats = ["[a-z123]" , "[1a-z23]" , "[123a-z]" ]; |
| 1275 | for &p in pats.iter() { |
| 1276 | let pat = Pattern::new(p).unwrap(); |
| 1277 | for c in "abcdefghijklmnopqrstuvwxyz" .chars() { |
| 1278 | assert!(pat.matches(&c.to_string())); |
| 1279 | } |
| 1280 | for c in "ABCDEFGHIJKLMNOPQRSTUVWXYZ" .chars() { |
| 1281 | let options = MatchOptions { |
| 1282 | case_sensitive: false, |
| 1283 | ..MatchOptions::new() |
| 1284 | }; |
| 1285 | assert!(pat.matches_with(&c.to_string(), options)); |
| 1286 | } |
| 1287 | assert!(pat.matches("1" )); |
| 1288 | assert!(pat.matches("2" )); |
| 1289 | assert!(pat.matches("3" )); |
| 1290 | } |
| 1291 | |
| 1292 | let pats = ["[abc-]" , "[-abc]" , "[a-c-]" ]; |
| 1293 | for &p in pats.iter() { |
| 1294 | let pat = Pattern::new(p).unwrap(); |
| 1295 | assert!(pat.matches("a" )); |
| 1296 | assert!(pat.matches("b" )); |
| 1297 | assert!(pat.matches("c" )); |
| 1298 | assert!(pat.matches("-" )); |
| 1299 | assert!(!pat.matches("d" )); |
| 1300 | } |
| 1301 | |
| 1302 | let pat = Pattern::new("[2-1]" ).unwrap(); |
| 1303 | assert!(!pat.matches("1" )); |
| 1304 | assert!(!pat.matches("2" )); |
| 1305 | |
| 1306 | assert!(Pattern::new("[-]" ).unwrap().matches("-" )); |
| 1307 | assert!(!Pattern::new("[!-]" ).unwrap().matches("-" )); |
| 1308 | } |
| 1309 | |
| 1310 | #[test ] |
| 1311 | fn test_pattern_matches() { |
| 1312 | let txt_pat = Pattern::new("*hello.txt" ).unwrap(); |
| 1313 | assert!(txt_pat.matches("hello.txt" )); |
| 1314 | assert!(txt_pat.matches("gareth_says_hello.txt" )); |
| 1315 | assert!(txt_pat.matches("some/path/to/hello.txt" )); |
| 1316 | assert!(txt_pat.matches("some \\path \\to \\hello.txt" )); |
| 1317 | assert!(txt_pat.matches("/an/absolute/path/to/hello.txt" )); |
| 1318 | assert!(!txt_pat.matches("hello.txt-and-then-some" )); |
| 1319 | assert!(!txt_pat.matches("goodbye.txt" )); |
| 1320 | |
| 1321 | let dir_pat = Pattern::new("*some/path/to/hello.txt" ).unwrap(); |
| 1322 | assert!(dir_pat.matches("some/path/to/hello.txt" )); |
| 1323 | assert!(dir_pat.matches("a/bigger/some/path/to/hello.txt" )); |
| 1324 | assert!(!dir_pat.matches("some/path/to/hello.txt-and-then-some" )); |
| 1325 | assert!(!dir_pat.matches("some/other/path/to/hello.txt" )); |
| 1326 | } |
| 1327 | |
| 1328 | #[test ] |
| 1329 | fn test_pattern_escape() { |
| 1330 | let s = "_[_]_?_*_!_" ; |
| 1331 | assert_eq!(Pattern::escape(s), "_[[]_[]]_[?]_[*]_!_" .to_string()); |
| 1332 | assert!(Pattern::new(&Pattern::escape(s)).unwrap().matches(s)); |
| 1333 | } |
| 1334 | |
| 1335 | #[test ] |
| 1336 | fn test_pattern_matches_case_insensitive() { |
| 1337 | let pat = Pattern::new("aBcDeFg" ).unwrap(); |
| 1338 | let options = MatchOptions { |
| 1339 | case_sensitive: false, |
| 1340 | require_literal_separator: false, |
| 1341 | require_literal_leading_dot: false, |
| 1342 | }; |
| 1343 | |
| 1344 | assert!(pat.matches_with("aBcDeFg" , options)); |
| 1345 | assert!(pat.matches_with("abcdefg" , options)); |
| 1346 | assert!(pat.matches_with("ABCDEFG" , options)); |
| 1347 | assert!(pat.matches_with("AbCdEfG" , options)); |
| 1348 | } |
| 1349 | |
| 1350 | #[test ] |
| 1351 | fn test_pattern_matches_case_insensitive_range() { |
| 1352 | let pat_within = Pattern::new("[a]" ).unwrap(); |
| 1353 | let pat_except = Pattern::new("[!a]" ).unwrap(); |
| 1354 | |
| 1355 | let options_case_insensitive = MatchOptions { |
| 1356 | case_sensitive: false, |
| 1357 | require_literal_separator: false, |
| 1358 | require_literal_leading_dot: false, |
| 1359 | }; |
| 1360 | let options_case_sensitive = MatchOptions { |
| 1361 | case_sensitive: true, |
| 1362 | require_literal_separator: false, |
| 1363 | require_literal_leading_dot: false, |
| 1364 | }; |
| 1365 | |
| 1366 | assert!(pat_within.matches_with("a" , options_case_insensitive)); |
| 1367 | assert!(pat_within.matches_with("A" , options_case_insensitive)); |
| 1368 | assert!(!pat_within.matches_with("A" , options_case_sensitive)); |
| 1369 | |
| 1370 | assert!(!pat_except.matches_with("a" , options_case_insensitive)); |
| 1371 | assert!(!pat_except.matches_with("A" , options_case_insensitive)); |
| 1372 | assert!(pat_except.matches_with("A" , options_case_sensitive)); |
| 1373 | } |
| 1374 | |
| 1375 | #[test ] |
| 1376 | fn test_pattern_matches_require_literal_separator() { |
| 1377 | let options_require_literal = MatchOptions { |
| 1378 | case_sensitive: true, |
| 1379 | require_literal_separator: true, |
| 1380 | require_literal_leading_dot: false, |
| 1381 | }; |
| 1382 | let options_not_require_literal = MatchOptions { |
| 1383 | case_sensitive: true, |
| 1384 | require_literal_separator: false, |
| 1385 | require_literal_leading_dot: false, |
| 1386 | }; |
| 1387 | |
| 1388 | assert!(Pattern::new("abc/def" ) |
| 1389 | .unwrap() |
| 1390 | .matches_with("abc/def" , options_require_literal)); |
| 1391 | assert!(!Pattern::new("abc?def" ) |
| 1392 | .unwrap() |
| 1393 | .matches_with("abc/def" , options_require_literal)); |
| 1394 | assert!(!Pattern::new("abc*def" ) |
| 1395 | .unwrap() |
| 1396 | .matches_with("abc/def" , options_require_literal)); |
| 1397 | assert!(!Pattern::new("abc[/]def" ) |
| 1398 | .unwrap() |
| 1399 | .matches_with("abc/def" , options_require_literal)); |
| 1400 | |
| 1401 | assert!(Pattern::new("abc/def" ) |
| 1402 | .unwrap() |
| 1403 | .matches_with("abc/def" , options_not_require_literal)); |
| 1404 | assert!(Pattern::new("abc?def" ) |
| 1405 | .unwrap() |
| 1406 | .matches_with("abc/def" , options_not_require_literal)); |
| 1407 | assert!(Pattern::new("abc*def" ) |
| 1408 | .unwrap() |
| 1409 | .matches_with("abc/def" , options_not_require_literal)); |
| 1410 | assert!(Pattern::new("abc[/]def" ) |
| 1411 | .unwrap() |
| 1412 | .matches_with("abc/def" , options_not_require_literal)); |
| 1413 | } |
| 1414 | |
| 1415 | #[test ] |
| 1416 | fn test_pattern_matches_require_literal_leading_dot() { |
| 1417 | let options_require_literal_leading_dot = MatchOptions { |
| 1418 | case_sensitive: true, |
| 1419 | require_literal_separator: false, |
| 1420 | require_literal_leading_dot: true, |
| 1421 | }; |
| 1422 | let options_not_require_literal_leading_dot = MatchOptions { |
| 1423 | case_sensitive: true, |
| 1424 | require_literal_separator: false, |
| 1425 | require_literal_leading_dot: false, |
| 1426 | }; |
| 1427 | |
| 1428 | let f = |options| { |
| 1429 | Pattern::new("*.txt" ) |
| 1430 | .unwrap() |
| 1431 | .matches_with(".hello.txt" , options) |
| 1432 | }; |
| 1433 | assert!(f(options_not_require_literal_leading_dot)); |
| 1434 | assert!(!f(options_require_literal_leading_dot)); |
| 1435 | |
| 1436 | let f = |options| { |
| 1437 | Pattern::new(".*.*" ) |
| 1438 | .unwrap() |
| 1439 | .matches_with(".hello.txt" , options) |
| 1440 | }; |
| 1441 | assert!(f(options_not_require_literal_leading_dot)); |
| 1442 | assert!(f(options_require_literal_leading_dot)); |
| 1443 | |
| 1444 | let f = |options| { |
| 1445 | Pattern::new("aaa/bbb/*" ) |
| 1446 | .unwrap() |
| 1447 | .matches_with("aaa/bbb/.ccc" , options) |
| 1448 | }; |
| 1449 | assert!(f(options_not_require_literal_leading_dot)); |
| 1450 | assert!(!f(options_require_literal_leading_dot)); |
| 1451 | |
| 1452 | let f = |options| { |
| 1453 | Pattern::new("aaa/bbb/*" ) |
| 1454 | .unwrap() |
| 1455 | .matches_with("aaa/bbb/c.c.c." , options) |
| 1456 | }; |
| 1457 | assert!(f(options_not_require_literal_leading_dot)); |
| 1458 | assert!(f(options_require_literal_leading_dot)); |
| 1459 | |
| 1460 | let f = |options| { |
| 1461 | Pattern::new("aaa/bbb/.*" ) |
| 1462 | .unwrap() |
| 1463 | .matches_with("aaa/bbb/.ccc" , options) |
| 1464 | }; |
| 1465 | assert!(f(options_not_require_literal_leading_dot)); |
| 1466 | assert!(f(options_require_literal_leading_dot)); |
| 1467 | |
| 1468 | let f = |options| { |
| 1469 | Pattern::new("aaa/?bbb" ) |
| 1470 | .unwrap() |
| 1471 | .matches_with("aaa/.bbb" , options) |
| 1472 | }; |
| 1473 | assert!(f(options_not_require_literal_leading_dot)); |
| 1474 | assert!(!f(options_require_literal_leading_dot)); |
| 1475 | |
| 1476 | let f = |options| { |
| 1477 | Pattern::new("aaa/[.]bbb" ) |
| 1478 | .unwrap() |
| 1479 | .matches_with("aaa/.bbb" , options) |
| 1480 | }; |
| 1481 | assert!(f(options_not_require_literal_leading_dot)); |
| 1482 | assert!(!f(options_require_literal_leading_dot)); |
| 1483 | |
| 1484 | let f = |options| Pattern::new("**/*" ).unwrap().matches_with(".bbb" , options); |
| 1485 | assert!(f(options_not_require_literal_leading_dot)); |
| 1486 | assert!(!f(options_require_literal_leading_dot)); |
| 1487 | } |
| 1488 | |
| 1489 | #[test ] |
| 1490 | fn test_matches_path() { |
| 1491 | // on windows, (Path::new("a/b").as_str().unwrap() == "a\\b"), so this |
| 1492 | // tests that / and \ are considered equivalent on windows |
| 1493 | assert!(Pattern::new("a/b" ).unwrap().matches_path(&Path::new("a/b" ))); |
| 1494 | } |
| 1495 | |
| 1496 | #[test ] |
| 1497 | fn test_path_join() { |
| 1498 | let pattern = Path::new("one" ).join(&Path::new("**/*.rs" )); |
| 1499 | assert!(Pattern::new(pattern.to_str().unwrap()).is_ok()); |
| 1500 | } |
| 1501 | } |
| 1502 | |