1 // Copyright 2014-2015 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 use std::borrow::Cow; 12 use std::collections::HashMap; 13 use std::fmt; 14 use std::ops::Index; 15 use std::str::FromStr; 16 use std::sync::Arc; 17 18 use memchr::memchr; 19 20 use exec::{Exec, ExecNoSync}; 21 use expand::expand_bytes; 22 use error::Error; 23 use re_builder::bytes::RegexBuilder; 24 use re_trait::{self, RegularExpression, Locations, SubCapturesPosIter}; 25 26 /// Match represents a single match of a regex in a haystack. 27 /// 28 /// The lifetime parameter `'t` refers to the lifetime of the matched text. 29 #[derive(Copy, Clone, Debug, Eq, PartialEq)] 30 pub struct Match<'t> { 31 text: &'t [u8], 32 start: usize, 33 end: usize, 34 } 35 36 impl<'t> Match<'t> { 37 /// Returns the starting byte offset of the match in the haystack. 38 #[inline] start(&self) -> usize39 pub fn start(&self) -> usize { 40 self.start 41 } 42 43 /// Returns the ending byte offset of the match in the haystack. 44 #[inline] end(&self) -> usize45 pub fn end(&self) -> usize { 46 self.end 47 } 48 49 /// Returns the matched text. 50 #[inline] as_bytes(&self) -> &'t [u8]51 pub fn as_bytes(&self) -> &'t [u8] { 52 &self.text[self.start..self.end] 53 } 54 55 /// Creates a new match from the given haystack and byte offsets. 56 #[inline] new(haystack: &'t [u8], start: usize, end: usize) -> Match<'t>57 fn new(haystack: &'t [u8], start: usize, end: usize) -> Match<'t> { 58 Match { 59 text: haystack, 60 start: start, 61 end: end, 62 } 63 } 64 } 65 66 /// A compiled regular expression for matching arbitrary bytes. 67 /// 68 /// It can be used to search, split or replace text. All searching is done with 69 /// an implicit `.*?` at the beginning and end of an expression. To force an 70 /// expression to match the whole string (or a prefix or a suffix), you must 71 /// use an anchor like `^` or `$` (or `\A` and `\z`). 72 /// 73 /// Like the `Regex` type in the parent module, matches with this regex return 74 /// byte offsets into the search text. **Unlike** the parent `Regex` type, 75 /// these byte offsets may not correspond to UTF-8 sequence boundaries since 76 /// the regexes in this module can match arbitrary bytes. 77 #[derive(Clone)] 78 pub struct Regex(Exec); 79 80 impl fmt::Display for Regex { 81 /// Shows the original regular expression. fmt(&self, f: &mut fmt::Formatter) -> fmt::Result82 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { 83 write!(f, "{}", self.as_str()) 84 } 85 } 86 87 impl fmt::Debug for Regex { 88 /// Shows the original regular expression. fmt(&self, f: &mut fmt::Formatter) -> fmt::Result89 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { 90 fmt::Display::fmt(self, f) 91 } 92 } 93 94 /// A constructor for Regex from an Exec. 95 /// 96 /// This is hidden because Exec isn't actually part of the public API. 97 #[doc(hidden)] 98 impl From<Exec> for Regex { from(exec: Exec) -> Regex99 fn from(exec: Exec) -> Regex { 100 Regex(exec) 101 } 102 } 103 104 impl FromStr for Regex { 105 type Err = Error; 106 107 /// Attempts to parse a string into a regular expression from_str(s: &str) -> Result<Regex, Error>108 fn from_str(s: &str) -> Result<Regex, Error> { 109 Regex::new(s) 110 } 111 } 112 113 /// Core regular expression methods. 114 impl Regex { 115 /// Compiles a regular expression. Once compiled, it can be used repeatedly 116 /// to search, split or replace text in a string. 117 /// 118 /// If an invalid expression is given, then an error is returned. new(re: &str) -> Result<Regex, Error>119 pub fn new(re: &str) -> Result<Regex, Error> { 120 RegexBuilder::new(re).build() 121 } 122 123 /// Returns true if and only if the regex matches the string given. 124 /// 125 /// It is recommended to use this method if all you need to do is test 126 /// a match, since the underlying matching engine may be able to do less 127 /// work. 128 /// 129 /// # Example 130 /// 131 /// Test if some text contains at least one word with exactly 13 ASCII word 132 /// bytes: 133 /// 134 /// ```rust 135 /// # extern crate regex; use regex::bytes::Regex; 136 /// # fn main() { 137 /// let text = b"I categorically deny having triskaidekaphobia."; 138 /// assert!(Regex::new(r"\b\w{13}\b").unwrap().is_match(text)); 139 /// # } 140 /// ``` is_match(&self, text: &[u8]) -> bool141 pub fn is_match(&self, text: &[u8]) -> bool { 142 self.is_match_at(text, 0) 143 } 144 145 /// Returns the start and end byte range of the leftmost-first match in 146 /// `text`. If no match exists, then `None` is returned. 147 /// 148 /// Note that this should only be used if you want to discover the position 149 /// of the match. Testing the existence of a match is faster if you use 150 /// `is_match`. 151 /// 152 /// # Example 153 /// 154 /// Find the start and end location of the first word with exactly 13 155 /// ASCII word bytes: 156 /// 157 /// ```rust 158 /// # extern crate regex; use regex::bytes::Regex; 159 /// # fn main() { 160 /// let text = b"I categorically deny having triskaidekaphobia."; 161 /// let mat = Regex::new(r"\b\w{13}\b").unwrap().find(text).unwrap(); 162 /// assert_eq!((mat.start(), mat.end()), (2, 15)); 163 /// # } 164 /// ``` find<'t>(&self, text: &'t [u8]) -> Option<Match<'t>>165 pub fn find<'t>(&self, text: &'t [u8]) -> Option<Match<'t>> { 166 self.find_at(text, 0) 167 } 168 169 /// Returns an iterator for each successive non-overlapping match in 170 /// `text`, returning the start and end byte indices with respect to 171 /// `text`. 172 /// 173 /// # Example 174 /// 175 /// Find the start and end location of every word with exactly 13 ASCII 176 /// word bytes: 177 /// 178 /// ```rust 179 /// # extern crate regex; use regex::bytes::Regex; 180 /// # fn main() { 181 /// let text = b"Retroactively relinquishing remunerations is reprehensible."; 182 /// for mat in Regex::new(r"\b\w{13}\b").unwrap().find_iter(text) { 183 /// println!("{:?}", mat); 184 /// } 185 /// # } 186 /// ``` find_iter<'r, 't>(&'r self, text: &'t [u8]) -> Matches<'r, 't>187 pub fn find_iter<'r, 't>(&'r self, text: &'t [u8]) -> Matches<'r, 't> { 188 Matches(self.0.searcher().find_iter(text)) 189 } 190 191 /// Returns the capture groups corresponding to the leftmost-first 192 /// match in `text`. Capture group `0` always corresponds to the entire 193 /// match. If no match is found, then `None` is returned. 194 /// 195 /// You should only use `captures` if you need access to the location of 196 /// capturing group matches. Otherwise, `find` is faster for discovering 197 /// the location of the overall match. 198 /// 199 /// # Examples 200 /// 201 /// Say you have some text with movie names and their release years, 202 /// like "'Citizen Kane' (1941)". It'd be nice if we could search for text 203 /// looking like that, while also extracting the movie name and its release 204 /// year separately. 205 /// 206 /// ```rust 207 /// # extern crate regex; use regex::bytes::Regex; 208 /// # fn main() { 209 /// let re = Regex::new(r"'([^']+)'\s+\((\d{4})\)").unwrap(); 210 /// let text = b"Not my favorite movie: 'Citizen Kane' (1941)."; 211 /// let caps = re.captures(text).unwrap(); 212 /// assert_eq!(&caps[1], &b"Citizen Kane"[..]); 213 /// assert_eq!(&caps[2], &b"1941"[..]); 214 /// assert_eq!(&caps[0], &b"'Citizen Kane' (1941)"[..]); 215 /// // You can also access the groups by index using the Index notation. 216 /// // Note that this will panic on an invalid index. 217 /// assert_eq!(&caps[1], b"Citizen Kane"); 218 /// assert_eq!(&caps[2], b"1941"); 219 /// assert_eq!(&caps[0], b"'Citizen Kane' (1941)"); 220 /// # } 221 /// ``` 222 /// 223 /// Note that the full match is at capture group `0`. Each subsequent 224 /// capture group is indexed by the order of its opening `(`. 225 /// 226 /// We can make this example a bit clearer by using *named* capture groups: 227 /// 228 /// ```rust 229 /// # extern crate regex; use regex::bytes::Regex; 230 /// # fn main() { 231 /// let re = Regex::new(r"'(?P<title>[^']+)'\s+\((?P<year>\d{4})\)") 232 /// .unwrap(); 233 /// let text = b"Not my favorite movie: 'Citizen Kane' (1941)."; 234 /// let caps = re.captures(text).unwrap(); 235 /// assert_eq!(&caps["title"], &b"Citizen Kane"[..]); 236 /// assert_eq!(&caps["year"], &b"1941"[..]); 237 /// assert_eq!(&caps[0], &b"'Citizen Kane' (1941)"[..]); 238 /// // You can also access the groups by name using the Index notation. 239 /// // Note that this will panic on an invalid group name. 240 /// assert_eq!(&caps["title"], b"Citizen Kane"); 241 /// assert_eq!(&caps["year"], b"1941"); 242 /// assert_eq!(&caps[0], b"'Citizen Kane' (1941)"); 243 /// 244 /// # } 245 /// ``` 246 /// 247 /// Here we name the capture groups, which we can access with the `name` 248 /// method or the `Index` notation with a `&str`. Note that the named 249 /// capture groups are still accessible with `get` or the `Index` notation 250 /// with a `usize`. 251 /// 252 /// The `0`th capture group is always unnamed, so it must always be 253 /// accessed with `get(0)` or `[0]`. captures<'t>(&self, text: &'t [u8]) -> Option<Captures<'t>>254 pub fn captures<'t>(&self, text: &'t [u8]) -> Option<Captures<'t>> { 255 let mut locs = self.locations(); 256 self.read_captures_at(&mut locs, text, 0).map(|_| Captures { 257 text: text, 258 locs: locs, 259 named_groups: self.0.capture_name_idx().clone(), 260 }) 261 } 262 263 /// Returns an iterator over all the non-overlapping capture groups matched 264 /// in `text`. This is operationally the same as `find_iter`, except it 265 /// yields information about capturing group matches. 266 /// 267 /// # Example 268 /// 269 /// We can use this to find all movie titles and their release years in 270 /// some text, where the movie is formatted like "'Title' (xxxx)": 271 /// 272 /// ```rust 273 /// # extern crate regex; use std::str; use regex::bytes::Regex; 274 /// # fn main() { 275 /// let re = Regex::new(r"'(?P<title>[^']+)'\s+\((?P<year>\d{4})\)") 276 /// .unwrap(); 277 /// let text = b"'Citizen Kane' (1941), 'The Wizard of Oz' (1939), 'M' (1931)."; 278 /// for caps in re.captures_iter(text) { 279 /// let title = str::from_utf8(&caps["title"]).unwrap(); 280 /// let year = str::from_utf8(&caps["year"]).unwrap(); 281 /// println!("Movie: {:?}, Released: {:?}", title, year); 282 /// } 283 /// // Output: 284 /// // Movie: Citizen Kane, Released: 1941 285 /// // Movie: The Wizard of Oz, Released: 1939 286 /// // Movie: M, Released: 1931 287 /// # } 288 /// ``` captures_iter<'r, 't>( &'r self, text: &'t [u8], ) -> CaptureMatches<'r, 't>289 pub fn captures_iter<'r, 't>( 290 &'r self, 291 text: &'t [u8], 292 ) -> CaptureMatches<'r, 't> { 293 CaptureMatches(self.0.searcher().captures_iter(text)) 294 } 295 296 /// Returns an iterator of substrings of `text` delimited by a match of the 297 /// regular expression. Namely, each element of the iterator corresponds to 298 /// text that *isn't* matched by the regular expression. 299 /// 300 /// This method will *not* copy the text given. 301 /// 302 /// # Example 303 /// 304 /// To split a string delimited by arbitrary amounts of spaces or tabs: 305 /// 306 /// ```rust 307 /// # extern crate regex; use regex::bytes::Regex; 308 /// # fn main() { 309 /// let re = Regex::new(r"[ \t]+").unwrap(); 310 /// let fields: Vec<&[u8]> = re.split(b"a b \t c\td e").collect(); 311 /// assert_eq!(fields, vec![ 312 /// &b"a"[..], &b"b"[..], &b"c"[..], &b"d"[..], &b"e"[..], 313 /// ]); 314 /// # } 315 /// ``` split<'r, 't>(&'r self, text: &'t [u8]) -> Split<'r, 't>316 pub fn split<'r, 't>(&'r self, text: &'t [u8]) -> Split<'r, 't> { 317 Split { 318 finder: self.find_iter(text), 319 last: 0, 320 } 321 } 322 323 /// Returns an iterator of at most `limit` substrings of `text` delimited 324 /// by a match of the regular expression. (A `limit` of `0` will return no 325 /// substrings.) Namely, each element of the iterator corresponds to text 326 /// that *isn't* matched by the regular expression. The remainder of the 327 /// string that is not split will be the last element in the iterator. 328 /// 329 /// This method will *not* copy the text given. 330 /// 331 /// # Example 332 /// 333 /// Get the first two words in some text: 334 /// 335 /// ```rust 336 /// # extern crate regex; use regex::bytes::Regex; 337 /// # fn main() { 338 /// let re = Regex::new(r"\W+").unwrap(); 339 /// let fields: Vec<&[u8]> = re.splitn(b"Hey! How are you?", 3).collect(); 340 /// assert_eq!(fields, vec![&b"Hey"[..], &b"How"[..], &b"are you?"[..]]); 341 /// # } 342 /// ``` splitn<'r, 't>( &'r self, text: &'t [u8], limit: usize, ) -> SplitN<'r, 't>343 pub fn splitn<'r, 't>( 344 &'r self, 345 text: &'t [u8], 346 limit: usize, 347 ) -> SplitN<'r, 't> { 348 SplitN { 349 splits: self.split(text), 350 n: limit, 351 } 352 } 353 354 /// Replaces the leftmost-first match with the replacement provided. The 355 /// replacement can be a regular byte string (where `$N` and `$name` are 356 /// expanded to match capture groups) or a function that takes the matches' 357 /// `Captures` and returns the replaced byte string. 358 /// 359 /// If no match is found, then a copy of the byte string is returned 360 /// unchanged. 361 /// 362 /// # Replacement string syntax 363 /// 364 /// All instances of `$name` in the replacement text is replaced with the 365 /// corresponding capture group `name`. 366 /// 367 /// `name` may be an integer corresponding to the index of the 368 /// capture group (counted by order of opening parenthesis where `0` is the 369 /// entire match) or it can be a name (consisting of letters, digits or 370 /// underscores) corresponding to a named capture group. 371 /// 372 /// If `name` isn't a valid capture group (whether the name doesn't exist 373 /// or isn't a valid index), then it is replaced with the empty string. 374 /// 375 /// The longest possible name is used. e.g., `$1a` looks up the capture 376 /// group named `1a` and not the capture group at index `1`. To exert more 377 /// precise control over the name, use braces, e.g., `${1}a`. 378 /// 379 /// To write a literal `$` use `$$`. 380 /// 381 /// # Examples 382 /// 383 /// Note that this function is polymorphic with respect to the replacement. 384 /// In typical usage, this can just be a normal byte string: 385 /// 386 /// ```rust 387 /// # extern crate regex; use regex::bytes::Regex; 388 /// # fn main() { 389 /// let re = Regex::new("[^01]+").unwrap(); 390 /// assert_eq!(re.replace(b"1078910", &b""[..]), &b"1010"[..]); 391 /// # } 392 /// ``` 393 /// 394 /// But anything satisfying the `Replacer` trait will work. For example, a 395 /// closure of type `|&Captures| -> Vec<u8>` provides direct access to the 396 /// captures corresponding to a match. This allows one to access capturing 397 /// group matches easily: 398 /// 399 /// ```rust 400 /// # extern crate regex; use regex::bytes::Regex; 401 /// # use regex::bytes::Captures; fn main() { 402 /// let re = Regex::new(r"([^,\s]+),\s+(\S+)").unwrap(); 403 /// let result = re.replace(b"Springsteen, Bruce", |caps: &Captures| { 404 /// let mut replacement = caps[2].to_owned(); 405 /// replacement.push(b' '); 406 /// replacement.extend(&caps[1]); 407 /// replacement 408 /// }); 409 /// assert_eq!(result, &b"Bruce Springsteen"[..]); 410 /// # } 411 /// ``` 412 /// 413 /// But this is a bit cumbersome to use all the time. Instead, a simple 414 /// syntax is supported that expands `$name` into the corresponding capture 415 /// group. Here's the last example, but using this expansion technique 416 /// with named capture groups: 417 /// 418 /// ```rust 419 /// # extern crate regex; use regex::bytes::Regex; 420 /// # fn main() { 421 /// let re = Regex::new(r"(?P<last>[^,\s]+),\s+(?P<first>\S+)").unwrap(); 422 /// let result = re.replace(b"Springsteen, Bruce", &b"$first $last"[..]); 423 /// assert_eq!(result, &b"Bruce Springsteen"[..]); 424 /// # } 425 /// ``` 426 /// 427 /// Note that using `$2` instead of `$first` or `$1` instead of `$last` 428 /// would produce the same result. To write a literal `$` use `$$`. 429 /// 430 /// Sometimes the replacement string requires use of curly braces to 431 /// delineate a capture group replacement and surrounding literal text. 432 /// For example, if we wanted to join two words together with an 433 /// underscore: 434 /// 435 /// ```rust 436 /// # extern crate regex; use regex::bytes::Regex; 437 /// # fn main() { 438 /// let re = Regex::new(r"(?P<first>\w+)\s+(?P<second>\w+)").unwrap(); 439 /// let result = re.replace(b"deep fried", &b"${first}_$second"[..]); 440 /// assert_eq!(result, &b"deep_fried"[..]); 441 /// # } 442 /// ``` 443 /// 444 /// Without the curly braces, the capture group name `first_` would be 445 /// used, and since it doesn't exist, it would be replaced with the empty 446 /// string. 447 /// 448 /// Finally, sometimes you just want to replace a literal string with no 449 /// regard for capturing group expansion. This can be done by wrapping a 450 /// byte string with `NoExpand`: 451 /// 452 /// ```rust 453 /// # extern crate regex; use regex::bytes::Regex; 454 /// # fn main() { 455 /// use regex::bytes::NoExpand; 456 /// 457 /// let re = Regex::new(r"(?P<last>[^,\s]+),\s+(\S+)").unwrap(); 458 /// let result = re.replace(b"Springsteen, Bruce", NoExpand(b"$2 $last")); 459 /// assert_eq!(result, &b"$2 $last"[..]); 460 /// # } 461 /// ``` replace<'t, R: Replacer>( &self, text: &'t [u8], rep: R, ) -> Cow<'t, [u8]>462 pub fn replace<'t, R: Replacer>( 463 &self, 464 text: &'t [u8], 465 rep: R, 466 ) -> Cow<'t, [u8]> { 467 self.replacen(text, 1, rep) 468 } 469 470 /// Replaces all non-overlapping matches in `text` with the replacement 471 /// provided. This is the same as calling `replacen` with `limit` set to 472 /// `0`. 473 /// 474 /// See the documentation for `replace` for details on how to access 475 /// capturing group matches in the replacement text. replace_all<'t, R: Replacer>( &self, text: &'t [u8], rep: R, ) -> Cow<'t, [u8]>476 pub fn replace_all<'t, R: Replacer>( 477 &self, 478 text: &'t [u8], 479 rep: R, 480 ) -> Cow<'t, [u8]> { 481 self.replacen(text, 0, rep) 482 } 483 484 /// Replaces at most `limit` non-overlapping matches in `text` with the 485 /// replacement provided. If `limit` is 0, then all non-overlapping matches 486 /// are replaced. 487 /// 488 /// See the documentation for `replace` for details on how to access 489 /// capturing group matches in the replacement text. replacen<'t, R: Replacer>( &self, text: &'t [u8], limit: usize, mut rep: R, ) -> Cow<'t, [u8]>490 pub fn replacen<'t, R: Replacer>( 491 &self, 492 text: &'t [u8], 493 limit: usize, 494 mut rep: R, 495 ) -> Cow<'t, [u8]> { 496 if let Some(rep) = rep.no_expansion() { 497 let mut it = self.find_iter(text).enumerate().peekable(); 498 if it.peek().is_none() { 499 return Cow::Borrowed(text); 500 } 501 let mut new = Vec::with_capacity(text.len()); 502 let mut last_match = 0; 503 for (i, m) in it { 504 if limit > 0 && i >= limit { 505 break 506 } 507 new.extend_from_slice(&text[last_match..m.start()]); 508 new.extend_from_slice(&rep); 509 last_match = m.end(); 510 } 511 new.extend_from_slice(&text[last_match..]); 512 return Cow::Owned(new); 513 } 514 515 // The slower path, which we use if the replacement needs access to 516 // capture groups. 517 let mut it = self.captures_iter(text).enumerate().peekable(); 518 if it.peek().is_none() { 519 return Cow::Borrowed(text); 520 } 521 let mut new = Vec::with_capacity(text.len()); 522 let mut last_match = 0; 523 for (i, cap) in it { 524 if limit > 0 && i >= limit { 525 break 526 } 527 // unwrap on 0 is OK because captures only reports matches 528 let m = cap.get(0).unwrap(); 529 new.extend_from_slice(&text[last_match..m.start()]); 530 rep.replace_append(&cap, &mut new); 531 last_match = m.end(); 532 } 533 new.extend_from_slice(&text[last_match..]); 534 Cow::Owned(new) 535 } 536 } 537 538 /// Advanced or "lower level" search methods. 539 impl Regex { 540 /// Returns the end location of a match in the text given. 541 /// 542 /// This method may have the same performance characteristics as 543 /// `is_match`, except it provides an end location for a match. In 544 /// particular, the location returned *may be shorter* than the proper end 545 /// of the leftmost-first match. 546 /// 547 /// # Example 548 /// 549 /// Typically, `a+` would match the entire first sequence of `a` in some 550 /// text, but `shortest_match` can give up as soon as it sees the first 551 /// `a`. 552 /// 553 /// ```rust 554 /// # extern crate regex; use regex::bytes::Regex; 555 /// # fn main() { 556 /// let text = b"aaaaa"; 557 /// let pos = Regex::new(r"a+").unwrap().shortest_match(text); 558 /// assert_eq!(pos, Some(1)); 559 /// # } 560 /// ``` shortest_match(&self, text: &[u8]) -> Option<usize>561 pub fn shortest_match(&self, text: &[u8]) -> Option<usize> { 562 self.shortest_match_at(text, 0) 563 } 564 565 /// Returns the same as shortest_match, but starts the search at the given 566 /// offset. 567 /// 568 /// The significance of the starting point is that it takes the surrounding 569 /// context into consideration. For example, the `\A` anchor can only 570 /// match when `start == 0`. 571 #[doc(hidden)] shortest_match_at( &self, text: &[u8], start: usize, ) -> Option<usize>572 pub fn shortest_match_at( 573 &self, 574 text: &[u8], 575 start: usize, 576 ) -> Option<usize> { 577 self.0.searcher().shortest_match_at(text, start) 578 } 579 580 /// Returns the same as is_match, but starts the search at the given 581 /// offset. 582 /// 583 /// The significance of the starting point is that it takes the surrounding 584 /// context into consideration. For example, the `\A` anchor can only 585 /// match when `start == 0`. 586 #[doc(hidden)] is_match_at(&self, text: &[u8], start: usize) -> bool587 pub fn is_match_at(&self, text: &[u8], start: usize) -> bool { 588 self.shortest_match_at(text, start).is_some() 589 } 590 591 /// Returns the same as find, but starts the search at the given 592 /// offset. 593 /// 594 /// The significance of the starting point is that it takes the surrounding 595 /// context into consideration. For example, the `\A` anchor can only 596 /// match when `start == 0`. 597 #[doc(hidden)] find_at<'t>( &self, text: &'t [u8], start: usize, ) -> Option<Match<'t>>598 pub fn find_at<'t>( 599 &self, 600 text: &'t [u8], 601 start: usize, 602 ) -> Option<Match<'t>> { 603 self.0.searcher().find_at(text, start) 604 .map(|(s, e)| Match::new(text, s, e)) 605 } 606 607 /// Returns the same as captures, but starts the search at the given 608 /// offset and populates the capture locations given. 609 /// 610 /// The significance of the starting point is that it takes the surrounding 611 /// context into consideration. For example, the `\A` anchor can only 612 /// match when `start == 0`. 613 #[doc(hidden)] read_captures_at<'t>( &self, locs: &mut Locations, text: &'t [u8], start: usize, ) -> Option<Match<'t>>614 pub fn read_captures_at<'t>( 615 &self, 616 locs: &mut Locations, 617 text: &'t [u8], 618 start: usize, 619 ) -> Option<Match<'t>> { 620 self.0.searcher().read_captures_at(locs, text, start) 621 .map(|(s, e)| Match::new(text, s, e)) 622 } 623 } 624 625 /// Auxiliary methods. 626 impl Regex { 627 /// Returns the original string of this regex. as_str(&self) -> &str628 pub fn as_str(&self) -> &str { 629 &self.0.regex_strings()[0] 630 } 631 632 /// Returns an iterator over the capture names. capture_names(&self) -> CaptureNames633 pub fn capture_names(&self) -> CaptureNames { 634 CaptureNames(self.0.capture_names().iter()) 635 } 636 637 /// Returns the number of captures. captures_len(&self) -> usize638 pub fn captures_len(&self) -> usize { 639 self.0.capture_names().len() 640 } 641 642 /// Returns an empty set of locations that can be reused in multiple calls 643 /// to `read_captures`. 644 #[doc(hidden)] locations(&self) -> Locations645 pub fn locations(&self) -> Locations { 646 self.0.searcher().locations() 647 } 648 } 649 650 /// An iterator over all non-overlapping matches for a particular string. 651 /// 652 /// The iterator yields a tuple of integers corresponding to the start and end 653 /// of the match. The indices are byte offsets. The iterator stops when no more 654 /// matches can be found. 655 /// 656 /// `'r` is the lifetime of the compiled regular expression and `'t` is the 657 /// lifetime of the matched byte string. 658 pub struct Matches<'r, 't>(re_trait::Matches<'t, ExecNoSync<'r>>); 659 660 impl<'r, 't> Iterator for Matches<'r, 't> { 661 type Item = Match<'t>; 662 next(&mut self) -> Option<Match<'t>>663 fn next(&mut self) -> Option<Match<'t>> { 664 let text = self.0.text(); 665 self.0.next().map(|(s, e)| Match::new(text, s, e)) 666 } 667 } 668 669 /// An iterator that yields all non-overlapping capture groups matching a 670 /// particular regular expression. 671 /// 672 /// The iterator stops when no more matches can be found. 673 /// 674 /// `'r` is the lifetime of the compiled regular expression and `'t` is the 675 /// lifetime of the matched byte string. 676 pub struct CaptureMatches<'r, 't>(re_trait::CaptureMatches<'t, ExecNoSync<'r>>); 677 678 impl<'r, 't> Iterator for CaptureMatches<'r, 't> { 679 type Item = Captures<'t>; 680 next(&mut self) -> Option<Captures<'t>>681 fn next(&mut self) -> Option<Captures<'t>> { 682 self.0.next().map(|locs| Captures { 683 text: self.0.text(), 684 locs: locs, 685 named_groups: self.0.regex().capture_name_idx().clone(), 686 }) 687 } 688 } 689 690 /// Yields all substrings delimited by a regular expression match. 691 /// 692 /// `'r` is the lifetime of the compiled regular expression and `'t` is the 693 /// lifetime of the byte string being split. 694 pub struct Split<'r, 't> { 695 finder: Matches<'r, 't>, 696 last: usize, 697 } 698 699 impl<'r, 't> Iterator for Split<'r, 't> { 700 type Item = &'t [u8]; 701 next(&mut self) -> Option<&'t [u8]>702 fn next(&mut self) -> Option<&'t [u8]> { 703 let text = self.finder.0.text(); 704 match self.finder.next() { 705 None => { 706 if self.last >= text.len() { 707 None 708 } else { 709 let s = &text[self.last..]; 710 self.last = text.len(); 711 Some(s) 712 } 713 } 714 Some(m) => { 715 let matched = &text[self.last..m.start()]; 716 self.last = m.end(); 717 Some(matched) 718 } 719 } 720 } 721 } 722 723 /// Yields at most `N` substrings delimited by a regular expression match. 724 /// 725 /// The last substring will be whatever remains after splitting. 726 /// 727 /// `'r` is the lifetime of the compiled regular expression and `'t` is the 728 /// lifetime of the byte string being split. 729 pub struct SplitN<'r, 't> { 730 splits: Split<'r, 't>, 731 n: usize, 732 } 733 734 impl<'r, 't> Iterator for SplitN<'r, 't> { 735 type Item = &'t [u8]; 736 next(&mut self) -> Option<&'t [u8]>737 fn next(&mut self) -> Option<&'t [u8]> { 738 if self.n == 0 { 739 return None 740 } 741 self.n -= 1; 742 if self.n == 0 { 743 let text = self.splits.finder.0.text(); 744 Some(&text[self.splits.last..]) 745 } else { 746 self.splits.next() 747 } 748 } 749 } 750 751 /// An iterator over the names of all possible captures. 752 /// 753 /// `None` indicates an unnamed capture; the first element (capture 0, the 754 /// whole matched region) is always unnamed. 755 /// 756 /// `'r` is the lifetime of the compiled regular expression. 757 pub struct CaptureNames<'r>(::std::slice::Iter<'r, Option<String>>); 758 759 impl<'r> Iterator for CaptureNames<'r> { 760 type Item = Option<&'r str>; 761 next(&mut self) -> Option<Option<&'r str>>762 fn next(&mut self) -> Option<Option<&'r str>> { 763 self.0.next().as_ref() 764 .map(|slot| slot.as_ref().map(|name| name.as_ref())) 765 } 766 size_hint(&self) -> (usize, Option<usize>)767 fn size_hint(&self) -> (usize, Option<usize>) { 768 self.0.size_hint() 769 } 770 } 771 772 /// Captures represents a group of captured byte strings for a single match. 773 /// 774 /// The 0th capture always corresponds to the entire match. Each subsequent 775 /// index corresponds to the next capture group in the regex. If a capture 776 /// group is named, then the matched byte string is *also* available via the 777 /// `name` method. (Note that the 0th capture is always unnamed and so must be 778 /// accessed with the `get` method.) 779 /// 780 /// Positions returned from a capture group are always byte indices. 781 /// 782 /// `'t` is the lifetime of the matched text. 783 pub struct Captures<'t> { 784 text: &'t [u8], 785 locs: Locations, 786 named_groups: Arc<HashMap<String, usize>>, 787 } 788 789 impl<'t> Captures<'t> { 790 /// Returns the match associated with the capture group at index `i`. If 791 /// `i` does not correspond to a capture group, or if the capture group 792 /// did not participate in the match, then `None` is returned. 793 /// 794 /// # Examples 795 /// 796 /// Get the text of the match with a default of an empty string if this 797 /// group didn't participate in the match: 798 /// 799 /// ```rust 800 /// # use regex::bytes::Regex; 801 /// let re = Regex::new(r"[a-z]+(?:([0-9]+)|([A-Z]+))").unwrap(); 802 /// let caps = re.captures(b"abc123").unwrap(); 803 /// 804 /// let text1 = caps.get(1).map_or(&b""[..], |m| m.as_bytes()); 805 /// let text2 = caps.get(2).map_or(&b""[..], |m| m.as_bytes()); 806 /// assert_eq!(text1, &b"123"[..]); 807 /// assert_eq!(text2, &b""[..]); 808 /// ``` get(&self, i: usize) -> Option<Match<'t>>809 pub fn get(&self, i: usize) -> Option<Match<'t>> { 810 self.locs.pos(i).map(|(s, e)| Match::new(self.text, s, e)) 811 } 812 813 /// Returns the match for the capture group named `name`. If `name` isn't a 814 /// valid capture group or didn't match anything, then `None` is returned. name(&self, name: &str) -> Option<Match<'t>>815 pub fn name(&self, name: &str) -> Option<Match<'t>> { 816 self.named_groups.get(name).and_then(|&i| self.get(i)) 817 } 818 819 /// An iterator that yields all capturing matches in the order in which 820 /// they appear in the regex. If a particular capture group didn't 821 /// participate in the match, then `None` is yielded for that capture. 822 /// 823 /// The first match always corresponds to the overall match of the regex. iter<'c>(&'c self) -> SubCaptureMatches<'c, 't>824 pub fn iter<'c>(&'c self) -> SubCaptureMatches<'c, 't> { 825 SubCaptureMatches { 826 caps: self, 827 it: self.locs.iter(), 828 } 829 } 830 831 /// Expands all instances of `$name` in `replacement` to the corresponding 832 /// capture group `name`, and writes them to the `dst` buffer given. 833 /// 834 /// `name` may be an integer corresponding to the index of the 835 /// capture group (counted by order of opening parenthesis where `0` is the 836 /// entire match) or it can be a name (consisting of letters, digits or 837 /// underscores) corresponding to a named capture group. 838 /// 839 /// If `name` isn't a valid capture group (whether the name doesn't exist 840 /// or isn't a valid index), then it is replaced with the empty string. 841 /// 842 /// The longest possible name is used. e.g., `$1a` looks up the capture 843 /// group named `1a` and not the capture group at index `1`. To exert more 844 /// precise control over the name, use braces, e.g., `${1}a`. 845 /// 846 /// To write a literal `$` use `$$`. expand(&self, replacement: &[u8], dst: &mut Vec<u8>)847 pub fn expand(&self, replacement: &[u8], dst: &mut Vec<u8>) { 848 expand_bytes(self, replacement, dst) 849 } 850 851 /// Returns the number of captured groups. 852 /// 853 /// This is always at least `1`, since every regex has at least one capture 854 /// group that corresponds to the full match. 855 #[inline] len(&self) -> usize856 pub fn len(&self) -> usize { 857 self.locs.len() 858 } 859 } 860 861 impl<'t> fmt::Debug for Captures<'t> { fmt(&self, f: &mut fmt::Formatter) -> fmt::Result862 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { 863 f.debug_tuple("Captures").field(&CapturesDebug(self)).finish() 864 } 865 } 866 867 struct CapturesDebug<'c, 't: 'c>(&'c Captures<'t>); 868 869 impl<'c, 't> fmt::Debug for CapturesDebug<'c, 't> { fmt(&self, f: &mut fmt::Formatter) -> fmt::Result870 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { 871 fn escape_bytes(bytes: &[u8]) -> String { 872 let mut s = String::new(); 873 for &b in bytes { 874 s.push_str(&escape_byte(b)); 875 } 876 s 877 } 878 879 fn escape_byte(byte: u8) -> String { 880 use std::ascii::escape_default; 881 882 let escaped: Vec<u8> = escape_default(byte).collect(); 883 String::from_utf8_lossy(&escaped).into_owned() 884 } 885 886 // We'd like to show something nice here, even if it means an 887 // allocation to build a reverse index. 888 let slot_to_name: HashMap<&usize, &String> = 889 self.0.named_groups.iter().map(|(a, b)| (b, a)).collect(); 890 let mut map = f.debug_map(); 891 for (slot, m) in self.0.locs.iter().enumerate() { 892 let m = m.map(|(s, e)| escape_bytes(&self.0.text[s..e])); 893 if let Some(name) = slot_to_name.get(&slot) { 894 map.entry(&name, &m); 895 } else { 896 map.entry(&slot, &m); 897 } 898 } 899 map.finish() 900 } 901 } 902 903 /// Get a group by index. 904 /// 905 /// `'t` is the lifetime of the matched text. 906 /// 907 /// The text can't outlive the `Captures` object if this method is 908 /// used, because of how `Index` is defined (normally `a[i]` is part 909 /// of `a` and can't outlive it); to do that, use `get()` instead. 910 /// 911 /// # Panics 912 /// 913 /// If there is no group at the given index. 914 impl<'t> Index<usize> for Captures<'t> { 915 type Output = [u8]; 916 index(&self, i: usize) -> &[u8]917 fn index(&self, i: usize) -> &[u8] { 918 self.get(i).map(|m| m.as_bytes()) 919 .unwrap_or_else(|| panic!("no group at index '{}'", i)) 920 } 921 } 922 923 /// Get a group by name. 924 /// 925 /// `'t` is the lifetime of the matched text and `'i` is the lifetime 926 /// of the group name (the index). 927 /// 928 /// The text can't outlive the `Captures` object if this method is 929 /// used, because of how `Index` is defined (normally `a[i]` is part 930 /// of `a` and can't outlive it); to do that, use `name` instead. 931 /// 932 /// # Panics 933 /// 934 /// If there is no group named by the given value. 935 impl<'t, 'i> Index<&'i str> for Captures<'t> { 936 type Output = [u8]; 937 index<'a>(&'a self, name: &'i str) -> &'a [u8]938 fn index<'a>(&'a self, name: &'i str) -> &'a [u8] { 939 self.name(name).map(|m| m.as_bytes()) 940 .unwrap_or_else(|| panic!("no group named '{}'", name)) 941 } 942 } 943 944 /// An iterator that yields all capturing matches in the order in which they 945 /// appear in the regex. 946 /// 947 /// If a particular capture group didn't participate in the match, then `None` 948 /// is yielded for that capture. The first match always corresponds to the 949 /// overall match of the regex. 950 /// 951 /// The lifetime `'c` corresponds to the lifetime of the `Captures` value, and 952 /// the lifetime `'t` corresponds to the originally matched text. 953 pub struct SubCaptureMatches<'c, 't: 'c> { 954 caps: &'c Captures<'t>, 955 it: SubCapturesPosIter<'c>, 956 } 957 958 impl<'c, 't> Iterator for SubCaptureMatches<'c, 't> { 959 type Item = Option<Match<'t>>; 960 next(&mut self) -> Option<Option<Match<'t>>>961 fn next(&mut self) -> Option<Option<Match<'t>>> { 962 self.it.next() 963 .map(|cap| cap.map(|(s, e)| Match::new(self.caps.text, s, e))) 964 } 965 } 966 967 /// Replacer describes types that can be used to replace matches in a byte 968 /// string. 969 /// 970 /// In general, users of this crate shouldn't need to implement this trait, 971 /// since implementations are already provided for `&[u8]` and 972 /// `FnMut(&Captures) -> Vec<u8>`, which covers most use cases. 973 pub trait Replacer { 974 /// Appends text to `dst` to replace the current match. 975 /// 976 /// The current match is represented by `caps`, which is guaranteed to 977 /// have a match at capture group `0`. 978 /// 979 /// For example, a no-op replacement would be 980 /// `dst.extend(&caps[0])`. replace_append(&mut self, caps: &Captures, dst: &mut Vec<u8>)981 fn replace_append(&mut self, caps: &Captures, dst: &mut Vec<u8>); 982 983 /// Return a fixed unchanging replacement byte string. 984 /// 985 /// When doing replacements, if access to `Captures` is not needed (e.g., 986 /// the replacement byte string does not need `$` expansion), then it can 987 /// be beneficial to avoid finding sub-captures. 988 /// 989 /// In general, this is called once for every call to `replacen`. no_expansion<'r>(&'r mut self) -> Option<Cow<'r, [u8]>>990 fn no_expansion<'r>(&'r mut self) -> Option<Cow<'r, [u8]>> { 991 None 992 } 993 994 /// Return a `Replacer` that borrows and wraps this `Replacer`. 995 /// 996 /// This is useful when you want to take a generic `Replacer` (which might 997 /// not be cloneable) and use it without consuming it, so it can be used 998 /// more than once. 999 /// 1000 /// # Example 1001 /// 1002 /// ``` 1003 /// use regex::bytes::{Regex, Replacer}; 1004 /// 1005 /// fn replace_all_twice<R: Replacer>( 1006 /// re: Regex, 1007 /// src: &[u8], 1008 /// mut rep: R, 1009 /// ) -> Vec<u8> { 1010 /// let dst = re.replace_all(src, rep.by_ref()); 1011 /// let dst = re.replace_all(&dst, rep.by_ref()); 1012 /// dst.into_owned() 1013 /// } 1014 /// ``` by_ref<'r>(&'r mut self) -> ReplacerRef<'r, Self>1015 fn by_ref<'r>(&'r mut self) -> ReplacerRef<'r, Self> { 1016 ReplacerRef(self) 1017 } 1018 } 1019 1020 /// By-reference adaptor for a `Replacer` 1021 /// 1022 /// Returned by [`Replacer::by_ref`](trait.Replacer.html#method.by_ref). 1023 #[derive(Debug)] 1024 pub struct ReplacerRef<'a, R: ?Sized + 'a>(&'a mut R); 1025 1026 impl<'a, R: Replacer + ?Sized + 'a> Replacer for ReplacerRef<'a, R> { replace_append(&mut self, caps: &Captures, dst: &mut Vec<u8>)1027 fn replace_append(&mut self, caps: &Captures, dst: &mut Vec<u8>) { 1028 self.0.replace_append(caps, dst) 1029 } no_expansion<'r>(&'r mut self) -> Option<Cow<'r, [u8]>>1030 fn no_expansion<'r>(&'r mut self) -> Option<Cow<'r, [u8]>> { 1031 self.0.no_expansion() 1032 } 1033 } 1034 1035 impl<'a> Replacer for &'a [u8] { replace_append(&mut self, caps: &Captures, dst: &mut Vec<u8>)1036 fn replace_append(&mut self, caps: &Captures, dst: &mut Vec<u8>) { 1037 caps.expand(*self, dst); 1038 } 1039 no_expansion(&mut self) -> Option<Cow<[u8]>>1040 fn no_expansion(&mut self) -> Option<Cow<[u8]>> { 1041 match memchr(b'$', *self) { 1042 Some(_) => None, 1043 None => Some(Cow::Borrowed(*self)), 1044 } 1045 } 1046 } 1047 1048 impl<F> Replacer for F where F: FnMut(&Captures) -> Vec<u8> { replace_append(&mut self, caps: &Captures, dst: &mut Vec<u8>)1049 fn replace_append(&mut self, caps: &Captures, dst: &mut Vec<u8>) { 1050 dst.extend_from_slice(&(*self)(caps)); 1051 } 1052 } 1053 1054 /// `NoExpand` indicates literal byte string replacement. 1055 /// 1056 /// It can be used with `replace` and `replace_all` to do a literal byte string 1057 /// replacement without expanding `$name` to their corresponding capture 1058 /// groups. This can be both convenient (to avoid escaping `$`, for example) 1059 /// and performant (since capture groups don't need to be found). 1060 /// 1061 /// `'t` is the lifetime of the literal text. 1062 pub struct NoExpand<'t>(pub &'t [u8]); 1063 1064 impl<'t> Replacer for NoExpand<'t> { replace_append(&mut self, _: &Captures, dst: &mut Vec<u8>)1065 fn replace_append(&mut self, _: &Captures, dst: &mut Vec<u8>) { 1066 dst.extend_from_slice(self.0); 1067 } 1068 no_expansion(&mut self) -> Option<Cow<[u8]>>1069 fn no_expansion(&mut self) -> Option<Cow<[u8]>> { 1070 Some(Cow::Borrowed(self.0)) 1071 } 1072 } 1073