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.
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.
11 pub use self::NamesIter::*;
12 pub use self::Regex::*;
14 use std::collections::HashMap;
16 use std::str::{MaybeOwned, Owned, Slice};
21 use vm::{CaptureLocs, MatchKind, Exists, Location, Submatches};
23 /// Escapes all regular expression meta characters in `text`.
25 /// The string returned may be safely used as a literal in a regular
27 pub fn quote(text: &str) -> String {
28 let mut quoted = String::with_capacity(text.len());
29 for c in text.chars() {
30 if parse::is_punct(c) {
38 /// Tests if the given regular expression matches somewhere in the text given.
40 /// If there was a problem compiling the regular expression, an error is
43 /// To find submatches, split or replace text, you'll need to compile an
46 /// Note that you should prefer the `regex!` macro when possible. For example,
47 /// `regex!("...").is_match("...")`.
48 pub fn is_match(regex: &str, text: &str) -> Result<bool, parse::Error> {
49 Regex::new(regex).map(|r| r.is_match(text))
52 /// A compiled regular expression
54 /// It is represented as either a sequence of bytecode instructions (dynamic)
55 /// or as a specialized Rust function (native). It can be used to search, split
56 /// or replace text. All searching is done with an implicit `.*?` at the
57 /// beginning and end of an expression. To force an expression to match the
58 /// whole string (or a prefix or a suffix), you must use an anchor like `^` or
59 /// `$` (or `\A` and `\z`).
61 /// While this crate will handle Unicode strings (whether in the regular
62 /// expression or in the search text), all positions returned are **byte
63 /// indices**. Every byte index is guaranteed to be at a Unicode code point
66 /// The lifetimes `'r` and `'t` in this crate correspond to the lifetime of a
67 /// compiled regular expression and text to search, respectively.
69 /// The only methods that allocate new strings are the string replacement
70 /// methods. All other methods (searching and splitting) return borrowed
71 /// pointers into the string given.
75 /// Find the location of a US phone number:
78 /// # use regex::Regex;
79 /// let re = match Regex::new("[0-9]{3}-[0-9]{3}-[0-9]{4}") {
81 /// Err(err) => panic!("{}", err),
83 /// assert_eq!(re.find("phone: 111-222-3333"), Some((7, 19)));
86 /// You can also use the `regex!` macro to compile a regular expression when
87 /// you compile your program:
90 /// #![feature(phase)]
91 /// extern crate regex;
92 /// #[phase(plugin)] extern crate regex_macros;
95 /// let re = regex!(r"\d+");
96 /// assert_eq!(re.find("123 abc"), Some((0, 3)));
100 /// Given an incorrect regular expression, `regex!` will cause the Rust
101 /// compiler to produce a compile time error.
102 /// Note that `regex!` will compile the expression to native Rust code, which
103 /// makes it much faster when searching text.
104 /// More details about the `regex!` macro can be found in the `regex` crate
108 // The representation of `Regex` is exported to support the `regex!`
109 // syntax extension. Do not rely on it.
111 // See the comments for the `program` module in `lib.rs` for a more
112 // detailed explanation for what `regex!` requires.
121 pub struct ExDynamic {
123 names: Vec<Option<String>>,
129 pub struct ExNative {
131 pub original: &'static str,
133 pub names: &'static &'static [Option<&'static str>],
135 pub prog: fn(MatchKind, &str, uint, uint) -> Vec<Option<uint>>
138 impl Clone for ExNative {
139 fn clone(&self) -> ExNative { *self }
142 impl fmt::Show for Regex {
143 /// Shows the original regular expression.
144 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
145 write!(f, "{}", self.as_str())
150 /// Compiles a dynamic regular expression. Once compiled, it can be
151 /// used repeatedly to search, split or replace text in a string.
153 /// When possible, you should prefer the `regex!` macro since it is
154 /// safer and always faster.
156 /// If an invalid expression is given, then an error is returned.
157 pub fn new(re: &str) -> Result<Regex, parse::Error> {
158 let ast = try!(parse::parse(re));
159 let (prog, names) = Program::new(ast);
160 Ok(Dynamic(ExDynamic {
161 original: re.to_string(),
167 /// Returns true if and only if the regex matches the string given.
171 /// Test if some text contains at least one word with exactly 13
175 /// # #![feature(phase)]
176 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
178 /// let text = "I categorically deny having triskaidekaphobia.";
179 /// let matched = regex!(r"\b\w{13}\b").is_match(text);
180 /// assert!(matched);
183 pub fn is_match(&self, text: &str) -> bool {
184 has_match(&exec(self, Exists, text))
187 /// Returns the start and end byte range of the leftmost-first match in
188 /// `text`. If no match exists, then `None` is returned.
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
196 /// Find the start and end location of the first word with exactly 13
200 /// # #![feature(phase)]
201 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
203 /// let text = "I categorically deny having triskaidekaphobia.";
204 /// let pos = regex!(r"\b\w{13}\b").find(text);
205 /// assert_eq!(pos, Some((2, 15)));
208 pub fn find(&self, text: &str) -> Option<(uint, uint)> {
209 let caps = exec(self, Location, text);
210 if has_match(&caps) {
211 Some((caps[0].unwrap(), caps[1].unwrap()))
217 /// Returns an iterator for each successive non-overlapping match in
218 /// `text`, returning the start and end byte indices with respect to
223 /// Find the start and end location of every word with exactly 13
227 /// # #![feature(phase)]
228 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
230 /// let text = "Retroactively relinquishing remunerations is reprehensible.";
231 /// for pos in regex!(r"\b\w{13}\b").find_iter(text) {
232 /// println!("{}", pos);
241 pub fn find_iter<'r, 't>(&'r self, text: &'t str) -> FindMatches<'r, 't> {
250 /// Returns the capture groups corresponding to the leftmost-first
251 /// match in `text`. Capture group `0` always corresponds to the entire
252 /// match. If no match is found, then `None` is returned.
254 /// You should only use `captures` if you need access to submatches.
255 /// Otherwise, `find` is faster for discovering the location of the overall
260 /// Say you have some text with movie names and their release years,
261 /// like "'Citizen Kane' (1941)". It'd be nice if we could search for text
262 /// looking like that, while also extracting the movie name and its release
266 /// # #![feature(phase)]
267 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
269 /// let re = regex!(r"'([^']+)'\s+\((\d{4})\)");
270 /// let text = "Not my favorite movie: 'Citizen Kane' (1941).";
271 /// let caps = re.captures(text).unwrap();
272 /// assert_eq!(caps.at(1), "Citizen Kane");
273 /// assert_eq!(caps.at(2), "1941");
274 /// assert_eq!(caps.at(0), "'Citizen Kane' (1941)");
278 /// Note that the full match is at capture group `0`. Each subsequent
279 /// capture group is indexed by the order of its opening `(`.
281 /// We can make this example a bit clearer by using *named* capture groups:
284 /// # #![feature(phase)]
285 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
287 /// let re = regex!(r"'(?P<title>[^']+)'\s+\((?P<year>\d{4})\)");
288 /// let text = "Not my favorite movie: 'Citizen Kane' (1941).";
289 /// let caps = re.captures(text).unwrap();
290 /// assert_eq!(caps.name("title"), "Citizen Kane");
291 /// assert_eq!(caps.name("year"), "1941");
292 /// assert_eq!(caps.at(0), "'Citizen Kane' (1941)");
296 /// Here we name the capture groups, which we can access with the `name`
297 /// method. Note that the named capture groups are still accessible with
300 /// The `0`th capture group is always unnamed, so it must always be
301 /// accessed with `at(0)`.
302 pub fn captures<'t>(&self, text: &'t str) -> Option<Captures<'t>> {
303 let caps = exec(self, Submatches, text);
304 Captures::new(self, text, caps)
307 /// Returns an iterator over all the non-overlapping capture groups matched
308 /// in `text`. This is operationally the same as `find_iter` (except it
309 /// yields information about submatches).
313 /// We can use this to find all movie titles and their release years in
314 /// some text, where the movie is formatted like "'Title' (xxxx)":
317 /// # #![feature(phase)]
318 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
320 /// let re = regex!(r"'(?P<title>[^']+)'\s+\((?P<year>\d{4})\)");
321 /// let text = "'Citizen Kane' (1941), 'The Wizard of Oz' (1939), 'M' (1931).";
322 /// for caps in re.captures_iter(text) {
323 /// println!("Movie: {}, Released: {}", caps.name("title"), caps.name("year"));
326 /// // Movie: Citizen Kane, Released: 1941
327 /// // Movie: The Wizard of Oz, Released: 1939
328 /// // Movie: M, Released: 1931
331 pub fn captures_iter<'r, 't>(&'r self, text: &'t str)
332 -> FindCaptures<'r, 't> {
341 /// Returns an iterator of substrings of `text` delimited by a match
342 /// of the regular expression.
343 /// Namely, each element of the iterator corresponds to text that *isn't*
344 /// matched by the regular expression.
346 /// This method will *not* copy the text given.
350 /// To split a string delimited by arbitrary amounts of spaces or tabs:
353 /// # #![feature(phase)]
354 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
356 /// let re = regex!(r"[ \t]+");
357 /// let fields: Vec<&str> = re.split("a b \t c\td e").collect();
358 /// assert_eq!(fields, vec!("a", "b", "c", "d", "e"));
361 pub fn split<'r, 't>(&'r self, text: &'t str) -> RegexSplits<'r, 't> {
363 finder: self.find_iter(text),
368 /// Returns an iterator of at most `limit` substrings of `text` delimited
369 /// by a match of the regular expression. (A `limit` of `0` will return no
371 /// Namely, each element of the iterator corresponds to text that *isn't*
372 /// matched by the regular expression.
373 /// The remainder of the string that is not split will be the last element
376 /// This method will *not* copy the text given.
380 /// Get the first two words in some text:
383 /// # #![feature(phase)]
384 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
386 /// let re = regex!(r"\W+");
387 /// let fields: Vec<&str> = re.splitn("Hey! How are you?", 3).collect();
388 /// assert_eq!(fields, vec!("Hey", "How", "are you?"));
391 pub fn splitn<'r, 't>(&'r self, text: &'t str, limit: uint)
392 -> RegexSplitsN<'r, 't> {
394 splits: self.split(text),
400 /// Replaces the leftmost-first match with the replacement provided.
401 /// The replacement can be a regular string (where `$N` and `$name` are
402 /// expanded to match capture groups) or a function that takes the matches'
403 /// `Captures` and returns the replaced string.
405 /// If no match is found, then a copy of the string is returned unchanged.
409 /// Note that this function is polymorphic with respect to the replacement.
410 /// In typical usage, this can just be a normal string:
413 /// # #![feature(phase)]
414 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
416 /// let re = regex!("[^01]+");
417 /// assert_eq!(re.replace("1078910", "").as_slice(), "1010");
421 /// But anything satisfying the `Replacer` trait will work. For example,
422 /// a closure of type `|&Captures| -> String` provides direct access to the
423 /// captures corresponding to a match. This allows one to access
424 /// submatches easily:
427 /// # #![feature(phase)]
428 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
429 /// # use regex::Captures; fn main() {
430 /// let re = regex!(r"([^,\s]+),\s+(\S+)");
431 /// let result = re.replace("Springsteen, Bruce", |caps: &Captures| {
432 /// format!("{} {}", caps.at(2), caps.at(1))
434 /// assert_eq!(result.as_slice(), "Bruce Springsteen");
438 /// But this is a bit cumbersome to use all the time. Instead, a simple
439 /// syntax is supported that expands `$name` into the corresponding capture
440 /// group. Here's the last example, but using this expansion technique
441 /// with named capture groups:
444 /// # #![feature(phase)]
445 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
447 /// let re = regex!(r"(?P<last>[^,\s]+),\s+(?P<first>\S+)");
448 /// let result = re.replace("Springsteen, Bruce", "$first $last");
449 /// assert_eq!(result.as_slice(), "Bruce Springsteen");
453 /// Note that using `$2` instead of `$first` or `$1` instead of `$last`
454 /// would produce the same result. To write a literal `$` use `$$`.
456 /// Finally, sometimes you just want to replace a literal string with no
457 /// submatch expansion. This can be done by wrapping a string with
461 /// # #![feature(phase)]
462 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
464 /// use regex::NoExpand;
466 /// let re = regex!(r"(?P<last>[^,\s]+),\s+(\S+)");
467 /// let result = re.replace("Springsteen, Bruce", NoExpand("$2 $last"));
468 /// assert_eq!(result.as_slice(), "$2 $last");
471 pub fn replace<R: Replacer>(&self, text: &str, rep: R) -> String {
472 self.replacen(text, 1, rep)
475 /// Replaces all non-overlapping matches in `text` with the
476 /// replacement provided. This is the same as calling `replacen` with
477 /// `limit` set to `0`.
479 /// See the documentation for `replace` for details on how to access
480 /// submatches in the replacement string.
481 pub fn replace_all<R: Replacer>(&self, text: &str, rep: R) -> String {
482 self.replacen(text, 0, rep)
485 /// Replaces at most `limit` non-overlapping matches in `text` with the
486 /// replacement provided. If `limit` is 0, then all non-overlapping matches
489 /// See the documentation for `replace` for details on how to access
490 /// submatches in the replacement string.
491 pub fn replacen<R: Replacer>
492 (&self, text: &str, limit: uint, mut rep: R) -> String {
493 let mut new = String::with_capacity(text.len());
494 let mut last_match = 0u;
496 for (i, cap) in self.captures_iter(text).enumerate() {
497 // It'd be nicer to use the 'take' iterator instead, but it seemed
498 // awkward given that '0' => no limit.
499 if limit > 0 && i >= limit {
503 let (s, e) = cap.pos(0).unwrap(); // captures only reports matches
504 new.push_str(text.slice(last_match, s));
505 new.push_str(rep.reg_replace(&cap).as_slice());
508 new.push_str(text.slice(last_match, text.len()));
512 /// Returns the original string of this regex.
513 pub fn as_str<'a>(&'a self) -> &'a str {
515 Dynamic(ExDynamic { ref original, .. }) => original.as_slice(),
516 Native(ExNative { ref original, .. }) => original.as_slice(),
522 pub fn names_iter<'a>(&'a self) -> NamesIter<'a> {
524 Native(ref n) => NamesIterNative(n.names.iter()),
525 Dynamic(ref d) => NamesIterDynamic(d.names.iter())
529 fn names_len(&self) -> uint {
531 Native(ref n) => n.names.len(),
532 Dynamic(ref d) => d.names.len()
538 pub enum NamesIter<'a> {
539 NamesIterNative(::std::slice::Items<'a, Option<&'static str>>),
540 NamesIterDynamic(::std::slice::Items<'a, Option<String>>)
543 impl<'a> Iterator<Option<String>> for NamesIter<'a> {
544 fn next(&mut self) -> Option<Option<String>> {
546 NamesIterNative(ref mut i) => i.next().map(|x| x.map(|s| s.to_string())),
547 NamesIterDynamic(ref mut i) => i.next().map(|x| x.as_ref().map(|s| s.to_string())),
552 /// NoExpand indicates literal string replacement.
554 /// It can be used with `replace` and `replace_all` to do a literal
555 /// string replacement without expanding `$name` to their corresponding
558 /// `'r` is the lifetime of the literal text.
559 pub struct NoExpand<'t>(pub &'t str);
561 /// Replacer describes types that can be used to replace matches in a string.
563 /// Returns a possibly owned string that is used to replace the match
564 /// corresponding the the `caps` capture group.
566 /// The `'a` lifetime refers to the lifetime of a borrowed string when
567 /// a new owned string isn't needed (e.g., for `NoExpand`).
568 fn reg_replace<'a>(&'a mut self, caps: &Captures) -> MaybeOwned<'a>;
571 impl<'t> Replacer for NoExpand<'t> {
572 fn reg_replace<'a>(&'a mut self, _: &Captures) -> MaybeOwned<'a> {
573 let NoExpand(s) = *self;
578 impl<'t> Replacer for &'t str {
579 fn reg_replace<'a>(&'a mut self, caps: &Captures) -> MaybeOwned<'a> {
580 Owned(caps.expand(*self))
584 impl<'t> Replacer for |&Captures|: 't -> String {
585 fn reg_replace<'a>(&'a mut self, caps: &Captures) -> MaybeOwned<'a> {
590 /// Yields all substrings delimited by a regular expression match.
592 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
593 /// of the string being split.
594 pub struct RegexSplits<'r, 't> {
595 finder: FindMatches<'r, 't>,
599 impl<'r, 't> Iterator<&'t str> for RegexSplits<'r, 't> {
600 fn next(&mut self) -> Option<&'t str> {
601 let text = self.finder.search;
602 match self.finder.next() {
604 if self.last >= text.len() {
607 let s = text.slice(self.last, text.len());
608 self.last = text.len();
613 let matched = text.slice(self.last, s);
621 /// Yields at most `N` substrings delimited by a regular expression match.
623 /// The last substring will be whatever remains after splitting.
625 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
626 /// of the string being split.
627 pub struct RegexSplitsN<'r, 't> {
628 splits: RegexSplits<'r, 't>,
633 impl<'r, 't> Iterator<&'t str> for RegexSplitsN<'r, 't> {
634 fn next(&mut self) -> Option<&'t str> {
635 let text = self.splits.finder.search;
636 if self.cur >= self.limit {
640 if self.cur >= self.limit {
641 Some(text.slice(self.splits.last, text.len()))
649 /// Captures represents a group of captured strings for a single match.
651 /// The 0th capture always corresponds to the entire match. Each subsequent
652 /// index corresponds to the next capture group in the regex.
653 /// If a capture group is named, then the matched string is *also* available
654 /// via the `name` method. (Note that the 0th capture is always unnamed and so
655 /// must be accessed with the `at` method.)
657 /// Positions returned from a capture group are always byte indices.
659 /// `'t` is the lifetime of the matched text.
660 pub struct Captures<'t> {
663 named: Option<HashMap<String, uint>>,
666 impl<'t> Captures<'t> {
667 #[allow(experimental)]
668 fn new(re: &Regex, search: &'t str, locs: CaptureLocs)
669 -> Option<Captures<'t>> {
670 if !has_match(&locs) {
675 if re.names_len() == 0 {
678 let mut named = HashMap::new();
679 for (i, name) in re.names_iter().enumerate() {
683 named.insert(name, i);
696 /// Returns the start and end positions of the Nth capture group.
697 /// Returns `None` if `i` is not a valid capture group or if the capture
698 /// group did not match anything.
699 /// The positions returned are *always* byte indices with respect to the
700 /// original string matched.
701 pub fn pos(&self, i: uint) -> Option<(uint, uint)> {
702 let (s, e) = (i * 2, i * 2 + 1);
703 if e >= self.locs.len() || self.locs[s].is_none() {
704 // VM guarantees that each pair of locations are both Some or None.
707 Some((self.locs[s].unwrap(), self.locs[e].unwrap()))
710 /// Returns the matched string for the capture group `i`.
711 /// If `i` isn't a valid capture group or didn't match anything, then the
712 /// empty string is returned.
713 pub fn at(&self, i: uint) -> &'t str {
717 self.text.slice(s, e)
722 /// Returns the matched string for the capture group named `name`.
723 /// If `name` isn't a valid capture group or didn't match anything, then
724 /// the empty string is returned.
725 pub fn name(&self, name: &str) -> &'t str {
729 match h.find_equiv(name) {
731 Some(i) => self.at(*i),
737 /// Creates an iterator of all the capture groups in order of appearance
738 /// in the regular expression.
739 pub fn iter(&'t self) -> SubCaptures<'t> {
740 SubCaptures { idx: 0, caps: self, }
743 /// Creates an iterator of all the capture group positions in order of
744 /// appearance in the regular expression. Positions are byte indices
745 /// in terms of the original string matched.
746 pub fn iter_pos(&'t self) -> SubCapturesPos<'t> {
747 SubCapturesPos { idx: 0, caps: self, }
750 /// Expands all instances of `$name` in `text` to the corresponding capture
753 /// `name` may be an integer corresponding to the index of the
754 /// capture group (counted by order of opening parenthesis where `0` is the
755 /// entire match) or it can be a name (consisting of letters, digits or
756 /// underscores) corresponding to a named capture group.
758 /// If `name` isn't a valid capture group (whether the name doesn't exist or
759 /// isn't a valid index), then it is replaced with the empty string.
761 /// To write a literal `$` use `$$`.
762 pub fn expand(&self, text: &str) -> String {
763 // How evil can you get?
764 // FIXME: Don't use regexes for this. It's completely unnecessary.
765 let re = Regex::new(r"(^|[^$]|\b)\$(\w+)").unwrap();
766 let text = re.replace_all(text, |refs: &Captures| -> String {
767 let (pre, name) = (refs.at(1), refs.at(2));
769 match from_str::<uint>(name.as_slice()) {
770 None => self.name(name).to_string(),
771 Some(i) => self.at(i).to_string(),
774 let re = Regex::new(r"\$\$").unwrap();
775 re.replace_all(text.as_slice(), NoExpand("$"))
778 /// Returns the number of captured groups.
780 pub fn len(&self) -> uint { self.locs.len() / 2 }
782 /// Returns if there are no captured groups.
784 pub fn is_empty(&self) -> bool { self.len() == 0 }
787 /// An iterator over capture groups for a particular match of a regular
790 /// `'t` is the lifetime of the matched text.
791 pub struct SubCaptures<'t> {
793 caps: &'t Captures<'t>,
796 impl<'t> Iterator<&'t str> for SubCaptures<'t> {
797 fn next(&mut self) -> Option<&'t str> {
798 if self.idx < self.caps.len() {
800 Some(self.caps.at(self.idx - 1))
807 /// An iterator over capture group positions for a particular match of a
808 /// regular expression.
810 /// Positions are byte indices in terms of the original string matched.
812 /// `'t` is the lifetime of the matched text.
813 pub struct SubCapturesPos<'t> {
815 caps: &'t Captures<'t>,
818 impl<'t> Iterator<Option<(uint, uint)>> for SubCapturesPos<'t> {
819 fn next(&mut self) -> Option<Option<(uint, uint)>> {
820 if self.idx < self.caps.len() {
822 Some(self.caps.pos(self.idx - 1))
829 /// An iterator that yields all non-overlapping capture groups matching a
830 /// particular regular expression.
832 /// The iterator stops when no more matches can be found.
834 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
835 /// of the matched string.
836 pub struct FindCaptures<'r, 't> {
839 last_match: Option<uint>,
843 impl<'r, 't> Iterator<Captures<'t>> for FindCaptures<'r, 't> {
844 fn next(&mut self) -> Option<Captures<'t>> {
845 if self.last_end > self.search.len() {
849 let caps = exec_slice(self.re, Submatches, self.search,
850 self.last_end, self.search.len());
852 if !has_match(&caps) {
855 (caps[0].unwrap(), caps[1].unwrap())
858 // Don't accept empty matches immediately following a match.
859 // i.e., no infinite loops please.
860 if e == s && Some(self.last_end) == self.last_match {
865 self.last_match = Some(self.last_end);
866 Captures::new(self.re, self.search, caps)
870 /// An iterator over all non-overlapping matches for a particular string.
872 /// The iterator yields a tuple of integers corresponding to the start and end
873 /// of the match. The indices are byte offsets. The iterator stops when no more
874 /// matches can be found.
876 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
877 /// of the matched string.
878 pub struct FindMatches<'r, 't> {
881 last_match: Option<uint>,
885 impl<'r, 't> Iterator<(uint, uint)> for FindMatches<'r, 't> {
886 fn next(&mut self) -> Option<(uint, uint)> {
887 if self.last_end > self.search.len() {
891 let caps = exec_slice(self.re, Location, self.search,
892 self.last_end, self.search.len());
894 if !has_match(&caps) {
897 (caps[0].unwrap(), caps[1].unwrap())
900 // Don't accept empty matches immediately following a match.
901 // i.e., no infinite loops please.
902 if e == s && Some(self.last_end) == self.last_match {
907 self.last_match = Some(self.last_end);
912 fn exec(re: &Regex, which: MatchKind, input: &str) -> CaptureLocs {
913 exec_slice(re, which, input, 0, input.len())
916 fn exec_slice(re: &Regex, which: MatchKind,
917 input: &str, s: uint, e: uint) -> CaptureLocs {
919 Dynamic(ExDynamic { ref prog, .. }) => vm::run(which, prog, input, s, e),
920 Native(ExNative { prog, .. }) => prog(which, input, s, e),
925 fn has_match(caps: &CaptureLocs) -> bool {
926 caps.len() >= 2 && caps[0].is_some() && caps[1].is_some()