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::CowString;
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>>,
130 pub struct ExNative {
132 pub original: &'static str,
134 pub names: &'static &'static [Option<&'static str>],
136 pub prog: fn(MatchKind, &str, uint, uint) -> Vec<Option<uint>>
139 impl Clone for ExNative {
140 fn clone(&self) -> ExNative {
145 impl fmt::Show for Regex {
146 /// Shows the original regular expression.
147 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
148 write!(f, "{}", self.as_str())
153 /// Compiles a dynamic regular expression. Once compiled, it can be
154 /// used repeatedly to search, split or replace text in a string.
156 /// When possible, you should prefer the `regex!` macro since it is
157 /// safer and always faster.
159 /// If an invalid expression is given, then an error is returned.
160 pub fn new(re: &str) -> Result<Regex, parse::Error> {
161 let ast = try!(parse::parse(re));
162 let (prog, names) = Program::new(ast);
163 Ok(Dynamic(ExDynamic {
164 original: re.to_string(),
170 /// Returns true if and only if the regex matches the string given.
174 /// Test if some text contains at least one word with exactly 13
178 /// # #![feature(phase)]
179 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
181 /// let text = "I categorically deny having triskaidekaphobia.";
182 /// let matched = regex!(r"\b\w{13}\b").is_match(text);
183 /// assert!(matched);
186 pub fn is_match(&self, text: &str) -> bool {
187 has_match(&exec(self, Exists, text))
190 /// Returns the start and end byte range of the leftmost-first match in
191 /// `text`. If no match exists, then `None` is returned.
193 /// Note that this should only be used if you want to discover the position
194 /// of the match. Testing the existence of a match is faster if you use
199 /// Find the start and end location of the first word with exactly 13
203 /// # #![feature(phase)]
204 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
206 /// let text = "I categorically deny having triskaidekaphobia.";
207 /// let pos = regex!(r"\b\w{13}\b").find(text);
208 /// assert_eq!(pos, Some((2, 15)));
211 pub fn find(&self, text: &str) -> Option<(uint, uint)> {
212 let caps = exec(self, Location, text);
213 if has_match(&caps) {
214 Some((caps[0].unwrap(), caps[1].unwrap()))
220 /// Returns an iterator for each successive non-overlapping match in
221 /// `text`, returning the start and end byte indices with respect to
226 /// Find the start and end location of every word with exactly 13
230 /// # #![feature(phase)]
231 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
233 /// let text = "Retroactively relinquishing remunerations is reprehensible.";
234 /// for pos in regex!(r"\b\w{13}\b").find_iter(text) {
235 /// println!("{}", pos);
244 pub fn find_iter<'r, 't>(&'r self, text: &'t str) -> FindMatches<'r, 't> {
253 /// Returns the capture groups corresponding to the leftmost-first
254 /// match in `text`. Capture group `0` always corresponds to the entire
255 /// match. If no match is found, then `None` is returned.
257 /// You should only use `captures` if you need access to submatches.
258 /// Otherwise, `find` is faster for discovering the location of the overall
263 /// Say you have some text with movie names and their release years,
264 /// like "'Citizen Kane' (1941)". It'd be nice if we could search for text
265 /// looking like that, while also extracting the movie name and its release
269 /// # #![feature(phase)]
270 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
272 /// let re = regex!(r"'([^']+)'\s+\((\d{4})\)");
273 /// let text = "Not my favorite movie: 'Citizen Kane' (1941).";
274 /// let caps = re.captures(text).unwrap();
275 /// assert_eq!(caps.at(1), Some("Citizen Kane"));
276 /// assert_eq!(caps.at(2), Some("1941"));
277 /// assert_eq!(caps.at(0), Some("'Citizen Kane' (1941)"));
281 /// Note that the full match is at capture group `0`. Each subsequent
282 /// capture group is indexed by the order of its opening `(`.
284 /// We can make this example a bit clearer by using *named* capture groups:
287 /// # #![feature(phase)]
288 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
290 /// let re = regex!(r"'(?P<title>[^']+)'\s+\((?P<year>\d{4})\)");
291 /// let text = "Not my favorite movie: 'Citizen Kane' (1941).";
292 /// let caps = re.captures(text).unwrap();
293 /// assert_eq!(caps.name("title"), Some("Citizen Kane"));
294 /// assert_eq!(caps.name("year"), Some("1941"));
295 /// assert_eq!(caps.at(0), Some("'Citizen Kane' (1941)"));
299 /// Here we name the capture groups, which we can access with the `name`
300 /// method. Note that the named capture groups are still accessible with
303 /// The `0`th capture group is always unnamed, so it must always be
304 /// accessed with `at(0)`.
305 pub fn captures<'t>(&self, text: &'t str) -> Option<Captures<'t>> {
306 let caps = exec(self, Submatches, text);
307 Captures::new(self, text, caps)
310 /// Returns an iterator over all the non-overlapping capture groups matched
311 /// in `text`. This is operationally the same as `find_iter` (except it
312 /// yields information about submatches).
316 /// We can use this to find all movie titles and their release years in
317 /// some text, where the movie is formatted like "'Title' (xxxx)":
320 /// # #![feature(phase)]
321 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
323 /// let re = regex!(r"'(?P<title>[^']+)'\s+\((?P<year>\d{4})\)");
324 /// let text = "'Citizen Kane' (1941), 'The Wizard of Oz' (1939), 'M' (1931).";
325 /// for caps in re.captures_iter(text) {
326 /// println!("Movie: {}, Released: {}", caps.name("title"), caps.name("year"));
329 /// // Movie: Citizen Kane, Released: 1941
330 /// // Movie: The Wizard of Oz, Released: 1939
331 /// // Movie: M, Released: 1931
334 pub fn captures_iter<'r, 't>(&'r self, text: &'t str)
335 -> FindCaptures<'r, 't> {
344 /// Returns an iterator of substrings of `text` delimited by a match
345 /// of the regular expression.
346 /// Namely, each element of the iterator corresponds to text that *isn't*
347 /// matched by the regular expression.
349 /// This method will *not* copy the text given.
353 /// To split a string delimited by arbitrary amounts of spaces or tabs:
356 /// # #![feature(phase)]
357 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
359 /// let re = regex!(r"[ \t]+");
360 /// let fields: Vec<&str> = re.split("a b \t c\td e").collect();
361 /// assert_eq!(fields, vec!("a", "b", "c", "d", "e"));
364 pub fn split<'r, 't>(&'r self, text: &'t str) -> RegexSplits<'r, 't> {
366 finder: self.find_iter(text),
371 /// Returns an iterator of at most `limit` substrings of `text` delimited
372 /// by a match of the regular expression. (A `limit` of `0` will return no
374 /// Namely, each element of the iterator corresponds to text that *isn't*
375 /// matched by the regular expression.
376 /// The remainder of the string that is not split will be the last element
379 /// This method will *not* copy the text given.
383 /// Get the first two words in some text:
386 /// # #![feature(phase)]
387 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
389 /// let re = regex!(r"\W+");
390 /// let fields: Vec<&str> = re.splitn("Hey! How are you?", 3).collect();
391 /// assert_eq!(fields, vec!("Hey", "How", "are you?"));
394 pub fn splitn<'r, 't>(&'r self, text: &'t str, limit: uint)
395 -> RegexSplitsN<'r, 't> {
397 splits: self.split(text),
403 /// Replaces the leftmost-first match with the replacement provided.
404 /// The replacement can be a regular string (where `$N` and `$name` are
405 /// expanded to match capture groups) or a function that takes the matches'
406 /// `Captures` and returns the replaced string.
408 /// If no match is found, then a copy of the string is returned unchanged.
412 /// Note that this function is polymorphic with respect to the replacement.
413 /// In typical usage, this can just be a normal string:
416 /// # #![feature(phase)]
417 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
419 /// let re = regex!("[^01]+");
420 /// assert_eq!(re.replace("1078910", ""), "1010");
424 /// But anything satisfying the `Replacer` trait will work. For example,
425 /// a closure of type `|&Captures| -> String` provides direct access to the
426 /// captures corresponding to a match. This allows one to access
427 /// submatches easily:
430 /// # #![feature(phase)]
431 /// # #![feature(unboxed_closures)]
432 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
433 /// # use regex::Captures; fn main() {
434 /// let re = regex!(r"([^,\s]+),\s+(\S+)");
435 /// let result = re.replace("Springsteen, Bruce", |&: caps: &Captures| {
436 /// format!("{} {}", caps.at(2).unwrap_or(""), caps.at(1).unwrap_or(""))
438 /// assert_eq!(result, "Bruce Springsteen");
442 /// But this is a bit cumbersome to use all the time. Instead, a simple
443 /// syntax is supported that expands `$name` into the corresponding capture
444 /// group. Here's the last example, but using this expansion technique
445 /// with named capture groups:
448 /// # #![feature(phase)]
449 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
451 /// let re = regex!(r"(?P<last>[^,\s]+),\s+(?P<first>\S+)");
452 /// let result = re.replace("Springsteen, Bruce", "$first $last");
453 /// assert_eq!(result, "Bruce Springsteen");
457 /// Note that using `$2` instead of `$first` or `$1` instead of `$last`
458 /// would produce the same result. To write a literal `$` use `$$`.
460 /// Finally, sometimes you just want to replace a literal string with no
461 /// submatch expansion. This can be done by wrapping a string with
465 /// # #![feature(phase)]
466 /// # extern crate regex; #[phase(plugin)] extern crate regex_macros;
468 /// use regex::NoExpand;
470 /// let re = regex!(r"(?P<last>[^,\s]+),\s+(\S+)");
471 /// let result = re.replace("Springsteen, Bruce", NoExpand("$2 $last"));
472 /// assert_eq!(result, "$2 $last");
475 pub fn replace<R: Replacer>(&self, text: &str, rep: R) -> String {
476 self.replacen(text, 1, rep)
479 /// Replaces all non-overlapping matches in `text` with the
480 /// replacement provided. This is the same as calling `replacen` with
481 /// `limit` set to `0`.
483 /// See the documentation for `replace` for details on how to access
484 /// submatches in the replacement string.
485 pub fn replace_all<R: Replacer>(&self, text: &str, rep: R) -> String {
486 self.replacen(text, 0, rep)
489 /// Replaces at most `limit` non-overlapping matches in `text` with the
490 /// replacement provided. If `limit` is 0, then all non-overlapping matches
493 /// See the documentation for `replace` for details on how to access
494 /// submatches in the replacement string.
495 pub fn replacen<R: Replacer>
496 (&self, text: &str, limit: uint, mut rep: R) -> String {
497 let mut new = String::with_capacity(text.len());
498 let mut last_match = 0u;
500 for (i, cap) in self.captures_iter(text).enumerate() {
501 // It'd be nicer to use the 'take' iterator instead, but it seemed
502 // awkward given that '0' => no limit.
503 if limit > 0 && i >= limit {
507 let (s, e) = cap.pos(0).unwrap(); // captures only reports matches
508 new.push_str(text[last_match..s]);
509 new.push_str(rep.reg_replace(&cap)[]);
512 new.push_str(text[last_match..text.len()]);
516 /// Returns the original string of this regex.
517 pub fn as_str<'a>(&'a self) -> &'a str {
519 Dynamic(ExDynamic { ref original, .. }) => original[],
520 Native(ExNative { ref original, .. }) => original[],
526 pub fn names_iter<'a>(&'a self) -> NamesIter<'a> {
528 Native(ref n) => NamesIterNative(n.names.iter()),
529 Dynamic(ref d) => NamesIterDynamic(d.names.iter())
533 fn names_len(&self) -> uint {
535 Native(ref n) => n.names.len(),
536 Dynamic(ref d) => d.names.len()
542 pub enum NamesIter<'a> {
543 NamesIterNative(::std::slice::Iter<'a, Option<&'static str>>),
544 NamesIterDynamic(::std::slice::Iter<'a, Option<String>>)
547 impl<'a> Iterator<Option<String>> for NamesIter<'a> {
548 fn next(&mut self) -> Option<Option<String>> {
550 NamesIterNative(ref mut i) => i.next().map(|x| x.map(|s| s.to_string())),
551 NamesIterDynamic(ref mut i) => i.next().map(|x| x.as_ref().map(|s| s.to_string())),
556 /// NoExpand indicates literal string replacement.
558 /// It can be used with `replace` and `replace_all` to do a literal
559 /// string replacement without expanding `$name` to their corresponding
562 /// `'r` is the lifetime of the literal text.
563 pub struct NoExpand<'t>(pub &'t str);
565 /// Replacer describes types that can be used to replace matches in a string.
567 /// Returns a possibly owned string that is used to replace the match
568 /// corresponding to the `caps` capture group.
570 /// The `'a` lifetime refers to the lifetime of a borrowed string when
571 /// a new owned string isn't needed (e.g., for `NoExpand`).
572 fn reg_replace<'a>(&'a mut self, caps: &Captures) -> CowString<'a>;
575 impl<'t> Replacer for NoExpand<'t> {
576 fn reg_replace<'a>(&'a mut self, _: &Captures) -> CowString<'a> {
577 let NoExpand(s) = *self;
582 impl<'t> Replacer for &'t str {
583 fn reg_replace<'a>(&'a mut self, caps: &Captures) -> CowString<'a> {
584 caps.expand(*self).into_cow()
588 impl<F> Replacer for F where F: FnMut(&Captures) -> String {
589 fn reg_replace<'a>(&'a mut self, caps: &Captures) -> CowString<'a> {
590 (*self)(caps).into_cow()
594 /// Yields all substrings delimited by a regular expression match.
596 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
597 /// of the string being split.
598 pub struct RegexSplits<'r, 't> {
599 finder: FindMatches<'r, 't>,
603 impl<'r, 't> Iterator<&'t str> for RegexSplits<'r, 't> {
604 fn next(&mut self) -> Option<&'t str> {
605 let text = self.finder.search;
606 match self.finder.next() {
608 if self.last >= text.len() {
611 let s = text[self.last..text.len()];
612 self.last = text.len();
617 let matched = text[self.last..s];
625 /// Yields at most `N` substrings delimited by a regular expression match.
627 /// The last substring will be whatever remains after splitting.
629 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
630 /// of the string being split.
631 pub struct RegexSplitsN<'r, 't> {
632 splits: RegexSplits<'r, 't>,
637 impl<'r, 't> Iterator<&'t str> for RegexSplitsN<'r, 't> {
638 fn next(&mut self) -> Option<&'t str> {
639 let text = self.splits.finder.search;
640 if self.cur >= self.limit {
644 if self.cur >= self.limit {
645 Some(text[self.splits.last..text.len()])
653 /// Captures represents a group of captured strings for a single match.
655 /// The 0th capture always corresponds to the entire match. Each subsequent
656 /// index corresponds to the next capture group in the regex.
657 /// If a capture group is named, then the matched string is *also* available
658 /// via the `name` method. (Note that the 0th capture is always unnamed and so
659 /// must be accessed with the `at` method.)
661 /// Positions returned from a capture group are always byte indices.
663 /// `'t` is the lifetime of the matched text.
664 pub struct Captures<'t> {
667 named: Option<HashMap<String, uint>>,
670 impl<'t> Captures<'t> {
671 #[allow(experimental)]
672 fn new(re: &Regex, search: &'t str, locs: CaptureLocs)
673 -> Option<Captures<'t>> {
674 if !has_match(&locs) {
679 if re.names_len() == 0 {
682 let mut named = HashMap::new();
683 for (i, name) in re.names_iter().enumerate() {
687 named.insert(name, i);
700 /// Returns the start and end positions of the Nth capture group.
701 /// Returns `None` if `i` is not a valid capture group or if the capture
702 /// group did not match anything.
703 /// The positions returned are *always* byte indices with respect to the
704 /// original string matched.
705 pub fn pos(&self, i: uint) -> Option<(uint, uint)> {
706 let (s, e) = (i * 2, i * 2 + 1);
707 if e >= self.locs.len() || self.locs[s].is_none() {
708 // VM guarantees that each pair of locations are both Some or None.
711 Some((self.locs[s].unwrap(), self.locs[e].unwrap()))
714 /// Returns the matched string for the capture group `i`. If `i` isn't
715 /// a valid capture group or didn't match anything, then `None` is
717 pub fn at(&self, i: uint) -> Option<&'t str> {
720 Some((s, e)) => Some(self.text.slice(s, e))
724 /// Returns the matched string for the capture group named `name`. If
725 /// `name` isn't a valid capture group or didn't match anything, then
726 /// `None` is returned.
727 pub fn name(&self, name: &str) -> Option<&'t str> {
733 Some(i) => self.at(*i),
739 /// Creates an iterator of all the capture groups in order of appearance
740 /// in the regular expression.
741 pub fn iter(&'t self) -> SubCaptures<'t> {
742 SubCaptures { idx: 0, caps: self, }
745 /// Creates an iterator of all the capture group positions in order of
746 /// appearance in the regular expression. Positions are byte indices
747 /// in terms of the original string matched.
748 pub fn iter_pos(&'t self) -> SubCapturesPos<'t> {
749 SubCapturesPos { idx: 0, caps: self, }
752 /// Expands all instances of `$name` in `text` to the corresponding capture
755 /// `name` may be an integer corresponding to the index of the
756 /// capture group (counted by order of opening parenthesis where `0` is the
757 /// entire match) or it can be a name (consisting of letters, digits or
758 /// underscores) corresponding to a named capture group.
760 /// If `name` isn't a valid capture group (whether the name doesn't exist or
761 /// isn't a valid index), then it is replaced with the empty string.
763 /// To write a literal `$` use `$$`.
764 pub fn expand(&self, text: &str) -> String {
765 // How evil can you get?
766 // FIXME: Don't use regexes for this. It's completely unnecessary.
767 let re = Regex::new(r"(^|[^$]|\b)\$(\w+)").unwrap();
768 let text = re.replace_all(text, |&mut: refs: &Captures| -> String {
769 let pre = refs.at(1).unwrap_or("");
770 let name = refs.at(2).unwrap_or("");
772 match name.parse::<uint>() {
773 None => self.name(name).unwrap_or("").to_string(),
774 Some(i) => self.at(i).unwrap_or("").to_string(),
777 let re = Regex::new(r"\$\$").unwrap();
778 re.replace_all(text[], NoExpand("$"))
781 /// Returns the number of captured groups.
783 pub fn len(&self) -> uint { self.locs.len() / 2 }
785 /// Returns if there are no captured groups.
787 pub fn is_empty(&self) -> bool { self.len() == 0 }
790 /// An iterator over capture groups for a particular match of a regular
793 /// `'t` is the lifetime of the matched text.
794 pub struct SubCaptures<'t> {
796 caps: &'t Captures<'t>,
799 impl<'t> Iterator<&'t str> for SubCaptures<'t> {
800 fn next(&mut self) -> Option<&'t str> {
801 if self.idx < self.caps.len() {
803 Some(self.caps.at(self.idx - 1).unwrap_or(""))
810 /// An iterator over capture group positions for a particular match of a
811 /// regular expression.
813 /// Positions are byte indices in terms of the original string matched.
815 /// `'t` is the lifetime of the matched text.
816 pub struct SubCapturesPos<'t> {
818 caps: &'t Captures<'t>,
821 impl<'t> Iterator<Option<(uint, uint)>> for SubCapturesPos<'t> {
822 fn next(&mut self) -> Option<Option<(uint, uint)>> {
823 if self.idx < self.caps.len() {
825 Some(self.caps.pos(self.idx - 1))
832 /// An iterator that yields all non-overlapping capture groups matching a
833 /// particular regular expression.
835 /// The iterator stops when no more matches can be found.
837 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
838 /// of the matched string.
839 pub struct FindCaptures<'r, 't> {
842 last_match: Option<uint>,
846 impl<'r, 't> Iterator<Captures<'t>> for FindCaptures<'r, 't> {
847 fn next(&mut self) -> Option<Captures<'t>> {
848 if self.last_end > self.search.len() {
852 let caps = exec_slice(self.re, Submatches, self.search,
853 self.last_end, self.search.len());
855 if !has_match(&caps) {
858 (caps[0].unwrap(), caps[1].unwrap())
861 // Don't accept empty matches immediately following a match.
862 // i.e., no infinite loops please.
863 if e == s && Some(self.last_end) == self.last_match {
868 self.last_match = Some(self.last_end);
869 Captures::new(self.re, self.search, caps)
873 /// An iterator over all non-overlapping matches for a particular string.
875 /// The iterator yields a tuple of integers corresponding to the start and end
876 /// of the match. The indices are byte offsets. The iterator stops when no more
877 /// matches can be found.
879 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
880 /// of the matched string.
881 pub struct FindMatches<'r, 't> {
884 last_match: Option<uint>,
888 impl<'r, 't> Iterator<(uint, uint)> for FindMatches<'r, 't> {
889 fn next(&mut self) -> Option<(uint, uint)> {
890 if self.last_end > self.search.len() {
894 let caps = exec_slice(self.re, Location, self.search,
895 self.last_end, self.search.len());
897 if !has_match(&caps) {
900 (caps[0].unwrap(), caps[1].unwrap())
903 // Don't accept empty matches immediately following a match.
904 // i.e., no infinite loops please.
905 if e == s && Some(self.last_end) == self.last_match {
910 self.last_match = Some(self.last_end);
915 fn exec(re: &Regex, which: MatchKind, input: &str) -> CaptureLocs {
916 exec_slice(re, which, input, 0, input.len())
919 fn exec_slice(re: &Regex, which: MatchKind,
920 input: &str, s: uint, e: uint) -> CaptureLocs {
922 Dynamic(ExDynamic { ref prog, .. }) => vm::run(which, prog, input, s, e),
923 Native(ExNative { ref prog, .. }) => (*prog)(which, input, s, e),
928 fn has_match(caps: &CaptureLocs) -> bool {
929 caps.len() >= 2 && caps[0].is_some() && caps[1].is_some()