regex/re_unicode.rs
1use std::borrow::Cow;
2use std::collections::HashMap;
3use std::fmt;
4use std::ops::Index;
5use std::str::FromStr;
6use std::sync::Arc;
7
8use find_byte::find_byte;
9use syntax;
10
11use error::Error;
12use exec::{Exec, ExecNoSyncStr};
13use expand::expand_str;
14use re_builder::unicode::RegexBuilder;
15use re_trait::{self, RegularExpression, SubCapturesPosIter};
16
17/// Escapes all regular expression meta characters in `text`.
18///
19/// The string returned may be safely used as a literal in a regular
20/// expression.
21pub fn escape(text: &str) -> String {
22 syntax::escape(text)
23}
24
25/// Match represents a single match of a regex in a haystack.
26///
27/// The lifetime parameter `'t` refers to the lifetime of the matched text.
28#[derive(Copy, Clone, Debug, Eq, PartialEq)]
29pub struct Match<'t> {
30 text: &'t str,
31 start: usize,
32 end: usize,
33}
34
35impl<'t> Match<'t> {
36 /// Returns the starting byte offset of the match in the haystack.
37 #[inline]
38 pub fn start(&self) -> usize {
39 self.start
40 }
41
42 /// Returns the ending byte offset of the match in the haystack.
43 #[inline]
44 pub fn end(&self) -> usize {
45 self.end
46 }
47
48 /// Returns the matched text.
49 #[inline]
50 pub fn as_str(&self) -> &'t str {
51 &self.text[self.start..self.end]
52 }
53
54 /// Creates a new match from the given haystack and byte offsets.
55 #[inline]
56 fn new(haystack: &'t str, start: usize, end: usize) -> Match<'t> {
57 Match { text: haystack, start: start, end: end }
58 }
59}
60
61impl<'t> From<Match<'t>> for &'t str {
62 fn from(m: Match<'t>) -> &'t str {
63 m.as_str()
64 }
65}
66
67/// A compiled regular expression for matching Unicode strings.
68///
69/// It is represented as either a sequence of bytecode instructions (dynamic)
70/// or as a specialized Rust function (native). It can be used to search, split
71/// or replace text. All searching is done with an implicit `.*?` at the
72/// beginning and end of an expression. To force an expression to match the
73/// whole string (or a prefix or a suffix), you must use an anchor like `^` or
74/// `$` (or `\A` and `\z`).
75///
76/// While this crate will handle Unicode strings (whether in the regular
77/// expression or in the search text), all positions returned are **byte
78/// indices**. Every byte index is guaranteed to be at a Unicode code point
79/// boundary.
80///
81/// The lifetimes `'r` and `'t` in this crate correspond to the lifetime of a
82/// compiled regular expression and text to search, respectively.
83///
84/// The only methods that allocate new strings are the string replacement
85/// methods. All other methods (searching and splitting) return borrowed
86/// pointers into the string given.
87///
88/// # Examples
89///
90/// Find the location of a US phone number:
91///
92/// ```rust
93/// # use regex::Regex;
94/// let re = Regex::new("[0-9]{3}-[0-9]{3}-[0-9]{4}").unwrap();
95/// let mat = re.find("phone: 111-222-3333").unwrap();
96/// assert_eq!((mat.start(), mat.end()), (7, 19));
97/// ```
98///
99/// # Using the `std::str::pattern` methods with `Regex`
100///
101/// > **Note**: This section requires that this crate is compiled with the
102/// > `pattern` Cargo feature enabled, which **requires nightly Rust**.
103///
104/// Since `Regex` implements `Pattern`, you can use regexes with methods
105/// defined on `&str`. For example, `is_match`, `find`, `find_iter`
106/// and `split` can be replaced with `str::contains`, `str::find`,
107/// `str::match_indices` and `str::split`.
108///
109/// Here are some examples:
110///
111/// ```rust,ignore
112/// # use regex::Regex;
113/// let re = Regex::new(r"\d+").unwrap();
114/// let haystack = "a111b222c";
115///
116/// assert!(haystack.contains(&re));
117/// assert_eq!(haystack.find(&re), Some(1));
118/// assert_eq!(haystack.match_indices(&re).collect::<Vec<_>>(),
119/// vec![(1, 4), (5, 8)]);
120/// assert_eq!(haystack.split(&re).collect::<Vec<_>>(), vec!["a", "b", "c"]);
121/// ```
122#[derive(Clone)]
123pub struct Regex(Exec);
124
125impl fmt::Display for Regex {
126 /// Shows the original regular expression.
127 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
128 write!(f, "{}", self.as_str())
129 }
130}
131
132impl fmt::Debug for Regex {
133 /// Shows the original regular expression.
134 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
135 fmt::Display::fmt(self, f)
136 }
137}
138
139#[doc(hidden)]
140impl From<Exec> for Regex {
141 fn from(exec: Exec) -> Regex {
142 Regex(exec)
143 }
144}
145
146impl FromStr for Regex {
147 type Err = Error;
148
149 /// Attempts to parse a string into a regular expression
150 fn from_str(s: &str) -> Result<Regex, Error> {
151 Regex::new(s)
152 }
153}
154
155/// Core regular expression methods.
156impl Regex {
157 /// Compiles a regular expression. Once compiled, it can be used repeatedly
158 /// to search, split or replace text in a string.
159 ///
160 /// If an invalid expression is given, then an error is returned.
161 pub fn new(re: &str) -> Result<Regex, Error> {
162 RegexBuilder::new(re).build()
163 }
164
165 /// Returns true if and only if the regex matches the string given.
166 ///
167 /// It is recommended to use this method if all you need to do is test
168 /// a match, since the underlying matching engine may be able to do less
169 /// work.
170 ///
171 /// # Example
172 ///
173 /// Test if some text contains at least one word with exactly 13
174 /// Unicode word characters:
175 ///
176 /// ```rust
177 /// # extern crate regex; use regex::Regex;
178 /// # fn main() {
179 /// let text = "I categorically deny having triskaidekaphobia.";
180 /// assert!(Regex::new(r"\b\w{13}\b").unwrap().is_match(text));
181 /// # }
182 /// ```
183 pub fn is_match(&self, text: &str) -> bool {
184 self.is_match_at(text, 0)
185 }
186
187 /// Returns the start and end byte range of the leftmost-first match in
188 /// `text`. If no match exists, then `None` is returned.
189 ///
190 /// Note that this should only be used if you want to discover the position
191 /// of the match. Testing the existence of a match is faster if you use
192 /// `is_match`.
193 ///
194 /// # Example
195 ///
196 /// Find the start and end location of the first word with exactly 13
197 /// Unicode word characters:
198 ///
199 /// ```rust
200 /// # extern crate regex; use regex::Regex;
201 /// # fn main() {
202 /// let text = "I categorically deny having triskaidekaphobia.";
203 /// let mat = Regex::new(r"\b\w{13}\b").unwrap().find(text).unwrap();
204 /// assert_eq!(mat.start(), 2);
205 /// assert_eq!(mat.end(), 15);
206 /// # }
207 /// ```
208 pub fn find<'t>(&self, text: &'t str) -> Option<Match<'t>> {
209 self.find_at(text, 0)
210 }
211
212 /// Returns an iterator for each successive non-overlapping match in
213 /// `text`, returning the start and end byte indices with respect to
214 /// `text`.
215 ///
216 /// # Example
217 ///
218 /// Find the start and end location of every word with exactly 13 Unicode
219 /// word characters:
220 ///
221 /// ```rust
222 /// # extern crate regex; use regex::Regex;
223 /// # fn main() {
224 /// let text = "Retroactively relinquishing remunerations is reprehensible.";
225 /// for mat in Regex::new(r"\b\w{13}\b").unwrap().find_iter(text) {
226 /// println!("{:?}", mat);
227 /// }
228 /// # }
229 /// ```
230 pub fn find_iter<'r, 't>(&'r self, text: &'t str) -> Matches<'r, 't> {
231 Matches(self.0.searcher_str().find_iter(text))
232 }
233
234 /// Returns the capture groups corresponding to the leftmost-first
235 /// match in `text`. Capture group `0` always corresponds to the entire
236 /// match. If no match is found, then `None` is returned.
237 ///
238 /// You should only use `captures` if you need access to the location of
239 /// capturing group matches. Otherwise, `find` is faster for discovering
240 /// the location of the overall match.
241 ///
242 /// # Examples
243 ///
244 /// Say you have some text with movie names and their release years,
245 /// like "'Citizen Kane' (1941)". It'd be nice if we could search for text
246 /// looking like that, while also extracting the movie name and its release
247 /// year separately.
248 ///
249 /// ```rust
250 /// # extern crate regex; use regex::Regex;
251 /// # fn main() {
252 /// let re = Regex::new(r"'([^']+)'\s+\((\d{4})\)").unwrap();
253 /// let text = "Not my favorite movie: 'Citizen Kane' (1941).";
254 /// let caps = re.captures(text).unwrap();
255 /// assert_eq!(caps.get(1).unwrap().as_str(), "Citizen Kane");
256 /// assert_eq!(caps.get(2).unwrap().as_str(), "1941");
257 /// assert_eq!(caps.get(0).unwrap().as_str(), "'Citizen Kane' (1941)");
258 /// // You can also access the groups by index using the Index notation.
259 /// // Note that this will panic on an invalid index.
260 /// assert_eq!(&caps[1], "Citizen Kane");
261 /// assert_eq!(&caps[2], "1941");
262 /// assert_eq!(&caps[0], "'Citizen Kane' (1941)");
263 /// # }
264 /// ```
265 ///
266 /// Note that the full match is at capture group `0`. Each subsequent
267 /// capture group is indexed by the order of its opening `(`.
268 ///
269 /// We can make this example a bit clearer by using *named* capture groups:
270 ///
271 /// ```rust
272 /// # extern crate regex; use regex::Regex;
273 /// # fn main() {
274 /// let re = Regex::new(r"'(?P<title>[^']+)'\s+\((?P<year>\d{4})\)")
275 /// .unwrap();
276 /// let text = "Not my favorite movie: 'Citizen Kane' (1941).";
277 /// let caps = re.captures(text).unwrap();
278 /// assert_eq!(caps.name("title").unwrap().as_str(), "Citizen Kane");
279 /// assert_eq!(caps.name("year").unwrap().as_str(), "1941");
280 /// assert_eq!(caps.get(0).unwrap().as_str(), "'Citizen Kane' (1941)");
281 /// // You can also access the groups by name using the Index notation.
282 /// // Note that this will panic on an invalid group name.
283 /// assert_eq!(&caps["title"], "Citizen Kane");
284 /// assert_eq!(&caps["year"], "1941");
285 /// assert_eq!(&caps[0], "'Citizen Kane' (1941)");
286 ///
287 /// # }
288 /// ```
289 ///
290 /// Here we name the capture groups, which we can access with the `name`
291 /// method or the `Index` notation with a `&str`. Note that the named
292 /// capture groups are still accessible with `get` or the `Index` notation
293 /// with a `usize`.
294 ///
295 /// The `0`th capture group is always unnamed, so it must always be
296 /// accessed with `get(0)` or `[0]`.
297 pub fn captures<'t>(&self, text: &'t str) -> Option<Captures<'t>> {
298 let mut locs = self.capture_locations();
299 self.captures_read_at(&mut locs, text, 0).map(move |_| Captures {
300 text: text,
301 locs: locs.0,
302 named_groups: self.0.capture_name_idx().clone(),
303 })
304 }
305
306 /// Returns an iterator over all the non-overlapping capture groups matched
307 /// in `text`. This is operationally the same as `find_iter`, except it
308 /// yields information about capturing group matches.
309 ///
310 /// # Example
311 ///
312 /// We can use this to find all movie titles and their release years in
313 /// some text, where the movie is formatted like "'Title' (xxxx)":
314 ///
315 /// ```rust
316 /// # extern crate regex; use regex::Regex;
317 /// # fn main() {
318 /// let re = Regex::new(r"'(?P<title>[^']+)'\s+\((?P<year>\d{4})\)")
319 /// .unwrap();
320 /// let text = "'Citizen Kane' (1941), 'The Wizard of Oz' (1939), 'M' (1931).";
321 /// for caps in re.captures_iter(text) {
322 /// println!("Movie: {:?}, Released: {:?}",
323 /// &caps["title"], &caps["year"]);
324 /// }
325 /// // Output:
326 /// // Movie: Citizen Kane, Released: 1941
327 /// // Movie: The Wizard of Oz, Released: 1939
328 /// // Movie: M, Released: 1931
329 /// # }
330 /// ```
331 pub fn captures_iter<'r, 't>(
332 &'r self,
333 text: &'t str,
334 ) -> CaptureMatches<'r, 't> {
335 CaptureMatches(self.0.searcher_str().captures_iter(text))
336 }
337
338 /// Returns an iterator of substrings of `text` delimited by a match of the
339 /// regular expression. Namely, each element of the iterator corresponds to
340 /// text that *isn't* matched by the regular expression.
341 ///
342 /// This method will *not* copy the text given.
343 ///
344 /// # Example
345 ///
346 /// To split a string delimited by arbitrary amounts of spaces or tabs:
347 ///
348 /// ```rust
349 /// # extern crate regex; use regex::Regex;
350 /// # fn main() {
351 /// let re = Regex::new(r"[ \t]+").unwrap();
352 /// let fields: Vec<&str> = re.split("a b \t c\td e").collect();
353 /// assert_eq!(fields, vec!["a", "b", "c", "d", "e"]);
354 /// # }
355 /// ```
356 pub fn split<'r, 't>(&'r self, text: &'t str) -> Split<'r, 't> {
357 Split { finder: self.find_iter(text), last: 0 }
358 }
359
360 /// Returns an iterator of at most `limit` substrings of `text` delimited
361 /// by a match of the regular expression. (A `limit` of `0` will return no
362 /// substrings.) Namely, each element of the iterator corresponds to text
363 /// that *isn't* matched by the regular expression. The remainder of the
364 /// string that is not split will be the last element in the iterator.
365 ///
366 /// This method will *not* copy the text given.
367 ///
368 /// # Example
369 ///
370 /// Get the first two words in some text:
371 ///
372 /// ```rust
373 /// # extern crate regex; use regex::Regex;
374 /// # fn main() {
375 /// let re = Regex::new(r"\W+").unwrap();
376 /// let fields: Vec<&str> = re.splitn("Hey! How are you?", 3).collect();
377 /// assert_eq!(fields, vec!("Hey", "How", "are you?"));
378 /// # }
379 /// ```
380 pub fn splitn<'r, 't>(
381 &'r self,
382 text: &'t str,
383 limit: usize,
384 ) -> SplitN<'r, 't> {
385 SplitN { splits: self.split(text), n: limit }
386 }
387
388 /// Replaces the leftmost-first match with the replacement provided.
389 /// The replacement can be a regular string (where `$N` and `$name` are
390 /// expanded to match capture groups) or a function that takes the matches'
391 /// `Captures` and returns the replaced string.
392 ///
393 /// If no match is found, then a copy of the string is returned unchanged.
394 ///
395 /// # Replacement string syntax
396 ///
397 /// All instances of `$name` in the replacement text is replaced with the
398 /// corresponding capture group `name`.
399 ///
400 /// `name` may be an integer corresponding to the index of the
401 /// capture group (counted by order of opening parenthesis where `0` is the
402 /// entire match) or it can be a name (consisting of letters, digits or
403 /// underscores) corresponding to a named capture group.
404 ///
405 /// If `name` isn't a valid capture group (whether the name doesn't exist
406 /// or isn't a valid index), then it is replaced with the empty string.
407 ///
408 /// The longest possible name is used. e.g., `$1a` looks up the capture
409 /// group named `1a` and not the capture group at index `1`. To exert more
410 /// precise control over the name, use braces, e.g., `${1}a`.
411 ///
412 /// To write a literal `$` use `$$`.
413 ///
414 /// # Examples
415 ///
416 /// Note that this function is polymorphic with respect to the replacement.
417 /// In typical usage, this can just be a normal string:
418 ///
419 /// ```rust
420 /// # extern crate regex; use regex::Regex;
421 /// # fn main() {
422 /// let re = Regex::new("[^01]+").unwrap();
423 /// assert_eq!(re.replace("1078910", ""), "1010");
424 /// # }
425 /// ```
426 ///
427 /// But anything satisfying the `Replacer` trait will work. For example,
428 /// a closure of type `|&Captures| -> String` provides direct access to the
429 /// captures corresponding to a match. This allows one to access
430 /// capturing group matches easily:
431 ///
432 /// ```rust
433 /// # extern crate regex; use regex::Regex;
434 /// # use regex::Captures; fn main() {
435 /// let re = Regex::new(r"([^,\s]+),\s+(\S+)").unwrap();
436 /// let result = re.replace("Springsteen, Bruce", |caps: &Captures| {
437 /// format!("{} {}", &caps[2], &caps[1])
438 /// });
439 /// assert_eq!(result, "Bruce Springsteen");
440 /// # }
441 /// ```
442 ///
443 /// But this is a bit cumbersome to use all the time. Instead, a simple
444 /// syntax is supported that expands `$name` into the corresponding capture
445 /// group. Here's the last example, but using this expansion technique
446 /// with named capture groups:
447 ///
448 /// ```rust
449 /// # extern crate regex; use regex::Regex;
450 /// # fn main() {
451 /// let re = Regex::new(r"(?P<last>[^,\s]+),\s+(?P<first>\S+)").unwrap();
452 /// let result = re.replace("Springsteen, Bruce", "$first $last");
453 /// assert_eq!(result, "Bruce Springsteen");
454 /// # }
455 /// ```
456 ///
457 /// Note that using `$2` instead of `$first` or `$1` instead of `$last`
458 /// would produce the same result. To write a literal `$` use `$$`.
459 ///
460 /// Sometimes the replacement string requires use of curly braces to
461 /// delineate a capture group replacement and surrounding literal text.
462 /// For example, if we wanted to join two words together with an
463 /// underscore:
464 ///
465 /// ```rust
466 /// # extern crate regex; use regex::Regex;
467 /// # fn main() {
468 /// let re = Regex::new(r"(?P<first>\w+)\s+(?P<second>\w+)").unwrap();
469 /// let result = re.replace("deep fried", "${first}_$second");
470 /// assert_eq!(result, "deep_fried");
471 /// # }
472 /// ```
473 ///
474 /// Without the curly braces, the capture group name `first_` would be
475 /// used, and since it doesn't exist, it would be replaced with the empty
476 /// string.
477 ///
478 /// Finally, sometimes you just want to replace a literal string with no
479 /// regard for capturing group expansion. This can be done by wrapping a
480 /// byte string with `NoExpand`:
481 ///
482 /// ```rust
483 /// # extern crate regex; use regex::Regex;
484 /// # fn main() {
485 /// use regex::NoExpand;
486 ///
487 /// let re = Regex::new(r"(?P<last>[^,\s]+),\s+(\S+)").unwrap();
488 /// let result = re.replace("Springsteen, Bruce", NoExpand("$2 $last"));
489 /// assert_eq!(result, "$2 $last");
490 /// # }
491 /// ```
492 pub fn replace<'t, R: Replacer>(
493 &self,
494 text: &'t str,
495 rep: R,
496 ) -> Cow<'t, str> {
497 self.replacen(text, 1, rep)
498 }
499
500 /// Replaces all non-overlapping matches in `text` with the replacement
501 /// provided. This is the same as calling `replacen` with `limit` set to
502 /// `0`.
503 ///
504 /// See the documentation for `replace` for details on how to access
505 /// capturing group matches in the replacement string.
506 pub fn replace_all<'t, R: Replacer>(
507 &self,
508 text: &'t str,
509 rep: R,
510 ) -> Cow<'t, str> {
511 self.replacen(text, 0, rep)
512 }
513
514 /// Replaces at most `limit` non-overlapping matches in `text` with the
515 /// replacement provided. If `limit` is 0, then all non-overlapping matches
516 /// are replaced.
517 ///
518 /// See the documentation for `replace` for details on how to access
519 /// capturing group matches in the replacement string.
520 pub fn replacen<'t, R: Replacer>(
521 &self,
522 text: &'t str,
523 limit: usize,
524 mut rep: R,
525 ) -> Cow<'t, str> {
526 // If we know that the replacement doesn't have any capture expansions,
527 // then we can fast path. The fast path can make a tremendous
528 // difference:
529 //
530 // 1) We use `find_iter` instead of `captures_iter`. Not asking for
531 // captures generally makes the regex engines faster.
532 // 2) We don't need to look up all of the capture groups and do
533 // replacements inside the replacement string. We just push it
534 // at each match and be done with it.
535 if let Some(rep) = rep.no_expansion() {
536 let mut it = self.find_iter(text).enumerate().peekable();
537 if it.peek().is_none() {
538 return Cow::Borrowed(text);
539 }
540 let mut new = String::with_capacity(text.len());
541 let mut last_match = 0;
542 for (i, m) in it {
543 if limit > 0 && i >= limit {
544 break;
545 }
546 new.push_str(&text[last_match..m.start()]);
547 new.push_str(&rep);
548 last_match = m.end();
549 }
550 new.push_str(&text[last_match..]);
551 return Cow::Owned(new);
552 }
553
554 // The slower path, which we use if the replacement needs access to
555 // capture groups.
556 let mut it = self.captures_iter(text).enumerate().peekable();
557 if it.peek().is_none() {
558 return Cow::Borrowed(text);
559 }
560 let mut new = String::with_capacity(text.len());
561 let mut last_match = 0;
562 for (i, cap) in it {
563 if limit > 0 && i >= limit {
564 break;
565 }
566 // unwrap on 0 is OK because captures only reports matches
567 let m = cap.get(0).unwrap();
568 new.push_str(&text[last_match..m.start()]);
569 rep.replace_append(&cap, &mut new);
570 last_match = m.end();
571 }
572 new.push_str(&text[last_match..]);
573 Cow::Owned(new)
574 }
575}
576
577/// Advanced or "lower level" search methods.
578impl Regex {
579 /// Returns the end location of a match in the text given.
580 ///
581 /// This method may have the same performance characteristics as
582 /// `is_match`, except it provides an end location for a match. In
583 /// particular, the location returned *may be shorter* than the proper end
584 /// of the leftmost-first match.
585 ///
586 /// # Example
587 ///
588 /// Typically, `a+` would match the entire first sequence of `a` in some
589 /// text, but `shortest_match` can give up as soon as it sees the first
590 /// `a`.
591 ///
592 /// ```rust
593 /// # extern crate regex; use regex::Regex;
594 /// # fn main() {
595 /// let text = "aaaaa";
596 /// let pos = Regex::new(r"a+").unwrap().shortest_match(text);
597 /// assert_eq!(pos, Some(1));
598 /// # }
599 /// ```
600 pub fn shortest_match(&self, text: &str) -> Option<usize> {
601 self.shortest_match_at(text, 0)
602 }
603
604 /// Returns the same as shortest_match, but starts the search at the given
605 /// offset.
606 ///
607 /// The significance of the starting point is that it takes the surrounding
608 /// context into consideration. For example, the `\A` anchor can only
609 /// match when `start == 0`.
610 pub fn shortest_match_at(
611 &self,
612 text: &str,
613 start: usize,
614 ) -> Option<usize> {
615 self.0.searcher_str().shortest_match_at(text, start)
616 }
617
618 /// Returns the same as is_match, but starts the search at the given
619 /// offset.
620 ///
621 /// The significance of the starting point is that it takes the surrounding
622 /// context into consideration. For example, the `\A` anchor can only
623 /// match when `start == 0`.
624 pub fn is_match_at(&self, text: &str, start: usize) -> bool {
625 self.shortest_match_at(text, start).is_some()
626 }
627
628 /// Returns the same as find, but starts the search at the given
629 /// offset.
630 ///
631 /// The significance of the starting point is that it takes the surrounding
632 /// context into consideration. For example, the `\A` anchor can only
633 /// match when `start == 0`.
634 pub fn find_at<'t>(
635 &self,
636 text: &'t str,
637 start: usize,
638 ) -> Option<Match<'t>> {
639 self.0
640 .searcher_str()
641 .find_at(text, start)
642 .map(|(s, e)| Match::new(text, s, e))
643 }
644
645 /// This is like `captures`, but uses
646 /// [`CaptureLocations`](struct.CaptureLocations.html)
647 /// instead of
648 /// [`Captures`](struct.Captures.html) in order to amortize allocations.
649 ///
650 /// To create a `CaptureLocations` value, use the
651 /// `Regex::capture_locations` method.
652 ///
653 /// This returns the overall match if this was successful, which is always
654 /// equivalence to the `0`th capture group.
655 pub fn captures_read<'t>(
656 &self,
657 locs: &mut CaptureLocations,
658 text: &'t str,
659 ) -> Option<Match<'t>> {
660 self.captures_read_at(locs, text, 0)
661 }
662
663 /// Returns the same as captures, but starts the search at the given
664 /// offset and populates the capture locations given.
665 ///
666 /// The significance of the starting point is that it takes the surrounding
667 /// context into consideration. For example, the `\A` anchor can only
668 /// match when `start == 0`.
669 pub fn captures_read_at<'t>(
670 &self,
671 locs: &mut CaptureLocations,
672 text: &'t str,
673 start: usize,
674 ) -> Option<Match<'t>> {
675 self.0
676 .searcher_str()
677 .captures_read_at(&mut locs.0, text, start)
678 .map(|(s, e)| Match::new(text, s, e))
679 }
680
681 /// An undocumented alias for `captures_read_at`.
682 ///
683 /// The `regex-capi` crate previously used this routine, so to avoid
684 /// breaking that crate, we continue to provide the name as an undocumented
685 /// alias.
686 #[doc(hidden)]
687 pub fn read_captures_at<'t>(
688 &self,
689 locs: &mut CaptureLocations,
690 text: &'t str,
691 start: usize,
692 ) -> Option<Match<'t>> {
693 self.captures_read_at(locs, text, start)
694 }
695}
696
697/// Auxiliary methods.
698impl Regex {
699 /// Returns the original string of this regex.
700 pub fn as_str(&self) -> &str {
701 &self.0.regex_strings()[0]
702 }
703
704 /// Returns an iterator over the capture names.
705 pub fn capture_names(&self) -> CaptureNames {
706 CaptureNames(self.0.capture_names().iter())
707 }
708
709 /// Returns the number of captures.
710 pub fn captures_len(&self) -> usize {
711 self.0.capture_names().len()
712 }
713
714 /// Returns an empty set of capture locations that can be reused in
715 /// multiple calls to `captures_read` or `captures_read_at`.
716 pub fn capture_locations(&self) -> CaptureLocations {
717 CaptureLocations(self.0.searcher_str().locations())
718 }
719
720 /// An alias for `capture_locations` to preserve backward compatibility.
721 ///
722 /// The `regex-capi` crate uses this method, so to avoid breaking that
723 /// crate, we continue to export it as an undocumented API.
724 #[doc(hidden)]
725 pub fn locations(&self) -> CaptureLocations {
726 CaptureLocations(self.0.searcher_str().locations())
727 }
728}
729
730/// An iterator over the names of all possible captures.
731///
732/// `None` indicates an unnamed capture; the first element (capture 0, the
733/// whole matched region) is always unnamed.
734///
735/// `'r` is the lifetime of the compiled regular expression.
736pub struct CaptureNames<'r>(::std::slice::Iter<'r, Option<String>>);
737
738impl<'r> Iterator for CaptureNames<'r> {
739 type Item = Option<&'r str>;
740
741 fn next(&mut self) -> Option<Option<&'r str>> {
742 self.0
743 .next()
744 .as_ref()
745 .map(|slot| slot.as_ref().map(|name| name.as_ref()))
746 }
747
748 fn size_hint(&self) -> (usize, Option<usize>) {
749 self.0.size_hint()
750 }
751}
752
753/// Yields all substrings delimited by a regular expression match.
754///
755/// `'r` is the lifetime of the compiled regular expression and `'t` is the
756/// lifetime of the string being split.
757pub struct Split<'r, 't> {
758 finder: Matches<'r, 't>,
759 last: usize,
760}
761
762impl<'r, 't> Iterator for Split<'r, 't> {
763 type Item = &'t str;
764
765 fn next(&mut self) -> Option<&'t str> {
766 let text = self.finder.0.text();
767 match self.finder.next() {
768 None => {
769 if self.last >= text.len() {
770 None
771 } else {
772 let s = &text[self.last..];
773 self.last = text.len();
774 Some(s)
775 }
776 }
777 Some(m) => {
778 let matched = &text[self.last..m.start()];
779 self.last = m.end();
780 Some(matched)
781 }
782 }
783 }
784}
785
786/// Yields at most `N` substrings delimited by a regular expression match.
787///
788/// The last substring will be whatever remains after splitting.
789///
790/// `'r` is the lifetime of the compiled regular expression and `'t` is the
791/// lifetime of the string being split.
792pub struct SplitN<'r, 't> {
793 splits: Split<'r, 't>,
794 n: usize,
795}
796
797impl<'r, 't> Iterator for SplitN<'r, 't> {
798 type Item = &'t str;
799
800 fn next(&mut self) -> Option<&'t str> {
801 if self.n == 0 {
802 return None;
803 }
804 self.n -= 1;
805 if self.n == 0 {
806 let text = self.splits.finder.0.text();
807 Some(&text[self.splits.last..])
808 } else {
809 self.splits.next()
810 }
811 }
812}
813
814/// CaptureLocations is a low level representation of the raw offsets of each
815/// submatch.
816///
817/// You can think of this as a lower level
818/// [`Captures`](struct.Captures.html), where this type does not support
819/// named capturing groups directly and it does not borrow the text that these
820/// offsets were matched on.
821///
822/// Primarily, this type is useful when using the lower level `Regex` APIs
823/// such as `read_captures`, which permits amortizing the allocation in which
824/// capture match locations are stored.
825///
826/// In order to build a value of this type, you'll need to call the
827/// `capture_locations` method on the `Regex` being used to execute the search.
828/// The value returned can then be reused in subsequent searches.
829#[derive(Clone, Debug)]
830pub struct CaptureLocations(re_trait::Locations);
831
832/// A type alias for `CaptureLocations` for backwards compatibility.
833///
834/// Previously, we exported `CaptureLocations` as `Locations` in an
835/// undocumented API. To prevent breaking that code (e.g., in `regex-capi`),
836/// we continue re-exporting the same undocumented API.
837#[doc(hidden)]
838pub type Locations = CaptureLocations;
839
840impl CaptureLocations {
841 /// Returns the start and end positions of the Nth capture group. Returns
842 /// `None` if `i` is not a valid capture group or if the capture group did
843 /// not match anything. The positions returned are *always* byte indices
844 /// with respect to the original string matched.
845 #[inline]
846 pub fn get(&self, i: usize) -> Option<(usize, usize)> {
847 self.0.pos(i)
848 }
849
850 /// Returns the total number of capturing groups.
851 ///
852 /// This is always at least `1` since every regex has at least `1`
853 /// capturing group that corresponds to the entire match.
854 #[inline]
855 pub fn len(&self) -> usize {
856 self.0.len()
857 }
858
859 /// An alias for the `get` method for backwards compatibility.
860 ///
861 /// Previously, we exported `get` as `pos` in an undocumented API. To
862 /// prevent breaking that code (e.g., in `regex-capi`), we continue
863 /// re-exporting the same undocumented API.
864 #[doc(hidden)]
865 #[inline]
866 pub fn pos(&self, i: usize) -> Option<(usize, usize)> {
867 self.get(i)
868 }
869}
870
871/// Captures represents a group of captured strings for a single match.
872///
873/// The 0th capture always corresponds to the entire match. Each subsequent
874/// index corresponds to the next capture group in the regex. If a capture
875/// group is named, then the matched string is *also* available via the `name`
876/// method. (Note that the 0th capture is always unnamed and so must be
877/// accessed with the `get` method.)
878///
879/// Positions returned from a capture group are always byte indices.
880///
881/// `'t` is the lifetime of the matched text.
882pub struct Captures<'t> {
883 text: &'t str,
884 locs: re_trait::Locations,
885 named_groups: Arc<HashMap<String, usize>>,
886}
887
888impl<'t> Captures<'t> {
889 /// Returns the match associated with the capture group at index `i`. If
890 /// `i` does not correspond to a capture group, or if the capture group
891 /// did not participate in the match, then `None` is returned.
892 ///
893 /// # Examples
894 ///
895 /// Get the text of the match with a default of an empty string if this
896 /// group didn't participate in the match:
897 ///
898 /// ```rust
899 /// # use regex::Regex;
900 /// let re = Regex::new(r"[a-z]+(?:([0-9]+)|([A-Z]+))").unwrap();
901 /// let caps = re.captures("abc123").unwrap();
902 ///
903 /// let text1 = caps.get(1).map_or("", |m| m.as_str());
904 /// let text2 = caps.get(2).map_or("", |m| m.as_str());
905 /// assert_eq!(text1, "123");
906 /// assert_eq!(text2, "");
907 /// ```
908 pub fn get(&self, i: usize) -> Option<Match<'t>> {
909 self.locs.pos(i).map(|(s, e)| Match::new(self.text, s, e))
910 }
911
912 /// Returns the match for the capture group named `name`. If `name` isn't a
913 /// valid capture group or didn't match anything, then `None` is returned.
914 pub fn name(&self, name: &str) -> Option<Match<'t>> {
915 self.named_groups.get(name).and_then(|&i| self.get(i))
916 }
917
918 /// An iterator that yields all capturing matches in the order in which
919 /// they appear in the regex. If a particular capture group didn't
920 /// participate in the match, then `None` is yielded for that capture.
921 ///
922 /// The first match always corresponds to the overall match of the regex.
923 pub fn iter<'c>(&'c self) -> SubCaptureMatches<'c, 't> {
924 SubCaptureMatches { caps: self, it: self.locs.iter() }
925 }
926
927 /// Expands all instances of `$name` in `replacement` to the corresponding
928 /// capture group `name`, and writes them to the `dst` buffer given.
929 ///
930 /// `name` may be an integer corresponding to the index of the
931 /// capture group (counted by order of opening parenthesis where `0` is the
932 /// entire match) or it can be a name (consisting of letters, digits or
933 /// underscores) corresponding to a named capture group.
934 ///
935 /// If `name` isn't a valid capture group (whether the name doesn't exist
936 /// or isn't a valid index), then it is replaced with the empty string.
937 ///
938 /// The longest possible name is used. e.g., `$1a` looks up the capture
939 /// group named `1a` and not the capture group at index `1`. To exert more
940 /// precise control over the name, use braces, e.g., `${1}a`.
941 ///
942 /// To write a literal `$` use `$$`.
943 pub fn expand(&self, replacement: &str, dst: &mut String) {
944 expand_str(self, replacement, dst)
945 }
946
947 /// Returns the number of captured groups.
948 ///
949 /// This is always at least `1`, since every regex has at least one capture
950 /// group that corresponds to the full match.
951 #[inline]
952 pub fn len(&self) -> usize {
953 self.locs.len()
954 }
955}
956
957impl<'t> fmt::Debug for Captures<'t> {
958 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
959 f.debug_tuple("Captures").field(&CapturesDebug(self)).finish()
960 }
961}
962
963struct CapturesDebug<'c, 't: 'c>(&'c Captures<'t>);
964
965impl<'c, 't> fmt::Debug for CapturesDebug<'c, 't> {
966 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
967 // We'd like to show something nice here, even if it means an
968 // allocation to build a reverse index.
969 let slot_to_name: HashMap<&usize, &String> =
970 self.0.named_groups.iter().map(|(a, b)| (b, a)).collect();
971 let mut map = f.debug_map();
972 for (slot, m) in self.0.locs.iter().enumerate() {
973 let m = m.map(|(s, e)| &self.0.text[s..e]);
974 if let Some(name) = slot_to_name.get(&slot) {
975 map.entry(&name, &m);
976 } else {
977 map.entry(&slot, &m);
978 }
979 }
980 map.finish()
981 }
982}
983
984/// Get a group by index.
985///
986/// `'t` is the lifetime of the matched text.
987///
988/// The text can't outlive the `Captures` object if this method is
989/// used, because of how `Index` is defined (normally `a[i]` is part
990/// of `a` and can't outlive it); to do that, use `get()` instead.
991///
992/// # Panics
993///
994/// If there is no group at the given index.
995impl<'t> Index<usize> for Captures<'t> {
996 type Output = str;
997
998 fn index(&self, i: usize) -> &str {
999 self.get(i)
1000 .map(|m| m.as_str())
1001 .unwrap_or_else(|| panic!("no group at index '{}'", i))
1002 }
1003}
1004
1005/// Get a group by name.
1006///
1007/// `'t` is the lifetime of the matched text and `'i` is the lifetime
1008/// of the group name (the index).
1009///
1010/// The text can't outlive the `Captures` object if this method is
1011/// used, because of how `Index` is defined (normally `a[i]` is part
1012/// of `a` and can't outlive it); to do that, use `name` instead.
1013///
1014/// # Panics
1015///
1016/// If there is no group named by the given value.
1017impl<'t, 'i> Index<&'i str> for Captures<'t> {
1018 type Output = str;
1019
1020 fn index<'a>(&'a self, name: &'i str) -> &'a str {
1021 self.name(name)
1022 .map(|m| m.as_str())
1023 .unwrap_or_else(|| panic!("no group named '{}'", name))
1024 }
1025}
1026
1027/// An iterator that yields all capturing matches in the order in which they
1028/// appear in the regex.
1029///
1030/// If a particular capture group didn't participate in the match, then `None`
1031/// is yielded for that capture. The first match always corresponds to the
1032/// overall match of the regex.
1033///
1034/// The lifetime `'c` corresponds to the lifetime of the `Captures` value, and
1035/// the lifetime `'t` corresponds to the originally matched text.
1036pub struct SubCaptureMatches<'c, 't: 'c> {
1037 caps: &'c Captures<'t>,
1038 it: SubCapturesPosIter<'c>,
1039}
1040
1041impl<'c, 't> Iterator for SubCaptureMatches<'c, 't> {
1042 type Item = Option<Match<'t>>;
1043
1044 fn next(&mut self) -> Option<Option<Match<'t>>> {
1045 self.it
1046 .next()
1047 .map(|cap| cap.map(|(s, e)| Match::new(self.caps.text, s, e)))
1048 }
1049}
1050
1051/// An iterator that yields all non-overlapping capture groups matching a
1052/// particular regular expression.
1053///
1054/// The iterator stops when no more matches can be found.
1055///
1056/// `'r` is the lifetime of the compiled regular expression and `'t` is the
1057/// lifetime of the matched string.
1058pub struct CaptureMatches<'r, 't>(
1059 re_trait::CaptureMatches<'t, ExecNoSyncStr<'r>>,
1060);
1061
1062impl<'r, 't> Iterator for CaptureMatches<'r, 't> {
1063 type Item = Captures<'t>;
1064
1065 fn next(&mut self) -> Option<Captures<'t>> {
1066 self.0.next().map(|locs| Captures {
1067 text: self.0.text(),
1068 locs: locs,
1069 named_groups: self.0.regex().capture_name_idx().clone(),
1070 })
1071 }
1072}
1073
1074/// An iterator over all non-overlapping matches for a particular string.
1075///
1076/// The iterator yields a `Match` value. The iterator stops when no more
1077/// matches can be found.
1078///
1079/// `'r` is the lifetime of the compiled regular expression and `'t` is the
1080/// lifetime of the matched string.
1081pub struct Matches<'r, 't>(re_trait::Matches<'t, ExecNoSyncStr<'r>>);
1082
1083impl<'r, 't> Iterator for Matches<'r, 't> {
1084 type Item = Match<'t>;
1085
1086 fn next(&mut self) -> Option<Match<'t>> {
1087 let text = self.0.text();
1088 self.0.next().map(|(s, e)| Match::new(text, s, e))
1089 }
1090}
1091
1092/// Replacer describes types that can be used to replace matches in a string.
1093///
1094/// In general, users of this crate shouldn't need to implement this trait,
1095/// since implementations are already provided for `&str` and
1096/// `FnMut(&Captures) -> String` (or any `FnMut(&Captures) -> T`
1097/// where `T: AsRef<str>`), which covers most use cases.
1098pub trait Replacer {
1099 /// Appends text to `dst` to replace the current match.
1100 ///
1101 /// The current match is represented by `caps`, which is guaranteed to
1102 /// have a match at capture group `0`.
1103 ///
1104 /// For example, a no-op replacement would be
1105 /// `dst.extend(caps.get(0).unwrap().as_str())`.
1106 fn replace_append(&mut self, caps: &Captures, dst: &mut String);
1107
1108 /// Return a fixed unchanging replacement string.
1109 ///
1110 /// When doing replacements, if access to `Captures` is not needed (e.g.,
1111 /// the replacement byte string does not need `$` expansion), then it can
1112 /// be beneficial to avoid finding sub-captures.
1113 ///
1114 /// In general, this is called once for every call to `replacen`.
1115 fn no_expansion<'r>(&'r mut self) -> Option<Cow<'r, str>> {
1116 None
1117 }
1118
1119 /// Return a `Replacer` that borrows and wraps this `Replacer`.
1120 ///
1121 /// This is useful when you want to take a generic `Replacer` (which might
1122 /// not be cloneable) and use it without consuming it, so it can be used
1123 /// more than once.
1124 ///
1125 /// # Example
1126 ///
1127 /// ```
1128 /// use regex::{Regex, Replacer};
1129 ///
1130 /// fn replace_all_twice<R: Replacer>(
1131 /// re: Regex,
1132 /// src: &str,
1133 /// mut rep: R,
1134 /// ) -> String {
1135 /// let dst = re.replace_all(src, rep.by_ref());
1136 /// let dst = re.replace_all(&dst, rep.by_ref());
1137 /// dst.into_owned()
1138 /// }
1139 /// ```
1140 fn by_ref<'r>(&'r mut self) -> ReplacerRef<'r, Self> {
1141 ReplacerRef(self)
1142 }
1143}
1144
1145/// By-reference adaptor for a `Replacer`
1146///
1147/// Returned by [`Replacer::by_ref`](trait.Replacer.html#method.by_ref).
1148#[derive(Debug)]
1149pub struct ReplacerRef<'a, R: ?Sized + 'a>(&'a mut R);
1150
1151impl<'a, R: Replacer + ?Sized + 'a> Replacer for ReplacerRef<'a, R> {
1152 fn replace_append(&mut self, caps: &Captures, dst: &mut String) {
1153 self.0.replace_append(caps, dst)
1154 }
1155 fn no_expansion(&mut self) -> Option<Cow<str>> {
1156 self.0.no_expansion()
1157 }
1158}
1159
1160impl<'a> Replacer for &'a str {
1161 fn replace_append(&mut self, caps: &Captures, dst: &mut String) {
1162 caps.expand(*self, dst);
1163 }
1164
1165 fn no_expansion(&mut self) -> Option<Cow<str>> {
1166 match find_byte(b'$', self.as_bytes()) {
1167 Some(_) => None,
1168 None => Some(Cow::Borrowed(*self)),
1169 }
1170 }
1171}
1172
1173impl<F, T> Replacer for F
1174where
1175 F: FnMut(&Captures) -> T,
1176 T: AsRef<str>,
1177{
1178 fn replace_append(&mut self, caps: &Captures, dst: &mut String) {
1179 dst.push_str((*self)(caps).as_ref());
1180 }
1181}
1182
1183/// `NoExpand` indicates literal string replacement.
1184///
1185/// It can be used with `replace` and `replace_all` to do a literal string
1186/// replacement without expanding `$name` to their corresponding capture
1187/// groups. This can be both convenient (to avoid escaping `$`, for example)
1188/// and performant (since capture groups don't need to be found).
1189///
1190/// `'t` is the lifetime of the literal text.
1191pub struct NoExpand<'t>(pub &'t str);
1192
1193impl<'t> Replacer for NoExpand<'t> {
1194 fn replace_append(&mut self, _: &Captures, dst: &mut String) {
1195 dst.push_str(self.0);
1196 }
1197
1198 fn no_expansion(&mut self) -> Option<Cow<str>> {
1199 Some(Cow::Borrowed(self.0))
1200 }
1201}