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 use collections::HashMap;
13 use std::from_str::from_str;
14 use std::str::{MaybeOwned, Owned, Slice};
19 use vm::{CaptureLocs, MatchKind, Exists, Location, Submatches};
21 /// Escapes all regular expression meta characters in `text` so that it may be
22 /// safely used in a regular expression as a literal string.
23 pub fn quote(text: &str) -> String {
24 let mut quoted = String::with_capacity(text.len());
25 for c in text.chars() {
26 if parse::is_punct(c) {
27 quoted.push_char('\\')
34 /// Tests if the given regular expression matches somewhere in the text given.
36 /// If there was a problem compiling the regular expression, an error is
39 /// To find submatches, split or replace text, you'll need to compile an
42 /// Note that you should prefer the `regex!` macro when possible. For example,
43 /// `regex!("...").is_match("...")`.
44 pub fn is_match(regex: &str, text: &str) -> Result<bool, parse::Error> {
45 Regex::new(regex).map(|r| r.is_match(text))
48 /// Regex is a compiled regular expression, represented as either a sequence
49 /// of bytecode instructions (dynamic) or as a specialized Rust function
50 /// (native). It can be used to search, split
51 /// or replace text. All searching is done with an implicit `.*?` at the
52 /// beginning and end of an expression. To force an expression to match the
53 /// whole string (or a prefix or a suffix), you must use an anchor like `^` or
54 /// `$` (or `\A` and `\z`).
56 /// While this crate will handle Unicode strings (whether in the regular
57 /// expression or in the search text), all positions returned are **byte
58 /// indices**. Every byte index is guaranteed to be at a UTF8 codepoint
61 /// The lifetimes `'r` and `'t` in this crate correspond to the lifetime of a
62 /// compiled regular expression and text to search, respectively.
64 /// The only methods that allocate new strings are the string replacement
65 /// methods. All other methods (searching and splitting) return borrowed
66 /// pointers into the string given.
70 /// Find the location of a US phone number:
73 /// # use regex::Regex;
74 /// let re = match Regex::new("[0-9]{3}-[0-9]{3}-[0-9]{4}") {
76 /// Err(err) => fail!("{}", err),
78 /// assert_eq!(re.find("phone: 111-222-3333"), Some((7, 19)));
81 /// You can also use the `regex!` macro to compile a regular expression when
82 /// you compile your program:
85 /// #![feature(phase)]
86 /// extern crate regex;
87 /// #[phase(syntax)] extern crate regex_macros;
90 /// let re = regex!(r"\d+");
91 /// assert_eq!(re.find("123 abc"), Some((0, 3)));
95 /// Given an incorrect regular expression, `regex!` will cause the Rust
96 /// compiler to produce a compile time error.
97 /// Note that `regex!` will compile the expression to native Rust code, which
98 /// makes it much faster when searching text.
99 /// More details about the `regex!` macro can be found in the `regex` crate
102 #[allow(visible_private_types)]
104 // The representation of `Regex` is exported to support the `regex!`
105 // syntax extension. Do not rely on it.
107 // See the comments for the `program` module in `lib.rs` for a more
108 // detailed explanation for what `regex!` requires.
119 names: Vec<Option<String>>,
127 pub original: &'static str,
129 pub names: &'static [Option<&'static str>],
131 pub prog: fn(MatchKind, &str, uint, uint) -> Vec<Option<uint>>
134 impl Clone for Native {
135 fn clone(&self) -> Native { *self }
138 impl fmt::Show for Regex {
139 /// Shows the original regular expression.
140 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
141 write!(f, "{}", self.as_str())
146 /// Compiles a dynamic regular expression. Once compiled, it can be
147 /// used repeatedly to search, split or replace text in a string.
149 /// When possible, you should prefer the `regex!` macro since it is
150 /// safer and always faster.
152 /// If an invalid expression is given, then an error is returned.
153 pub fn new(re: &str) -> Result<Regex, parse::Error> {
154 let ast = try!(parse::parse(re));
155 let (prog, names) = Program::new(ast);
157 original: re.to_string(),
163 /// Returns true if and only if the regex matches the string given.
167 /// Test if some text contains at least one word with exactly 13
171 /// # #![feature(phase)]
172 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
174 /// let text = "I categorically deny having triskaidekaphobia.";
175 /// let matched = regex!(r"\b\w{13}\b").is_match(text);
176 /// assert!(matched);
179 pub fn is_match(&self, text: &str) -> bool {
180 has_match(&exec(self, Exists, text))
183 /// Returns the start and end byte range of the leftmost-first match in
184 /// `text`. If no match exists, then `None` is returned.
186 /// Note that this should only be used if you want to discover the position
187 /// of the match. Testing the existence of a match is faster if you use
192 /// Find the start and end location of every word with exactly 13
196 /// # #![feature(phase)]
197 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
199 /// let text = "I categorically deny having triskaidekaphobia.";
200 /// let pos = regex!(r"\b\w{13}\b").find(text);
201 /// assert_eq!(pos, Some((2, 15)));
204 pub fn find(&self, text: &str) -> Option<(uint, uint)> {
205 let caps = exec(self, Location, text);
206 if has_match(&caps) {
207 Some((caps.get(0).unwrap(), caps.get(1).unwrap()))
213 /// Returns an iterator for each successive non-overlapping match in
214 /// `text`, returning the start and end byte indices with respect to
219 /// Find the start and end location of the first word with exactly 13
223 /// # #![feature(phase)]
224 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
226 /// let text = "Retroactively relinquishing remunerations is reprehensible.";
227 /// for pos in regex!(r"\b\w{13}\b").find_iter(text) {
228 /// println!("{}", pos);
237 pub fn find_iter<'r, 't>(&'r self, text: &'t str) -> FindMatches<'r, 't> {
246 /// Returns the capture groups corresponding to the leftmost-first
247 /// match in `text`. Capture group `0` always corresponds to the entire
248 /// match. If no match is found, then `None` is returned.
250 /// You should only use `captures` if you need access to submatches.
251 /// Otherwise, `find` is faster for discovering the location of the overall
256 /// Say you have some text with movie names and their release years,
257 /// like "'Citizen Kane' (1941)". It'd be nice if we could search for text
258 /// looking like that, while also extracting the movie name and its release
262 /// # #![feature(phase)]
263 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
265 /// let re = regex!(r"'([^']+)'\s+\((\d{4})\)");
266 /// let text = "Not my favorite movie: 'Citizen Kane' (1941).";
267 /// let caps = re.captures(text).unwrap();
268 /// assert_eq!(caps.at(1), "Citizen Kane");
269 /// assert_eq!(caps.at(2), "1941");
270 /// assert_eq!(caps.at(0), "'Citizen Kane' (1941)");
274 /// Note that the full match is at capture group `0`. Each subsequent
275 /// capture group is indexed by the order of its opening `(`.
277 /// We can make this example a bit clearer by using *named* capture groups:
280 /// # #![feature(phase)]
281 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
283 /// let re = regex!(r"'(?P<title>[^']+)'\s+\((?P<year>\d{4})\)");
284 /// let text = "Not my favorite movie: 'Citizen Kane' (1941).";
285 /// let caps = re.captures(text).unwrap();
286 /// assert_eq!(caps.name("title"), "Citizen Kane");
287 /// assert_eq!(caps.name("year"), "1941");
288 /// assert_eq!(caps.at(0), "'Citizen Kane' (1941)");
292 /// Here we name the capture groups, which we can access with the `name`
293 /// method. Note that the named capture groups are still accessible with
296 /// The `0`th capture group is always unnamed, so it must always be
297 /// accessed with `at(0)`.
298 pub fn captures<'t>(&self, text: &'t str) -> Option<Captures<'t>> {
299 let caps = exec(self, Submatches, text);
300 Captures::new(self, text, caps)
303 /// Returns an iterator over all the non-overlapping capture groups matched
304 /// in `text`. This is operationally the same as `find_iter` (except it
305 /// yields information about submatches).
309 /// We can use this to find all movie titles and their release years in
310 /// some text, where the movie is formatted like "'Title' (xxxx)":
313 /// # #![feature(phase)]
314 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
316 /// let re = regex!(r"'(?P<title>[^']+)'\s+\((?P<year>\d{4})\)");
317 /// let text = "'Citizen Kane' (1941), 'The Wizard of Oz' (1939), 'M' (1931).";
318 /// for caps in re.captures_iter(text) {
319 /// println!("Movie: {}, Released: {}", caps.name("title"), caps.name("year"));
322 /// // Movie: Citizen Kane, Released: 1941
323 /// // Movie: The Wizard of Oz, Released: 1939
324 /// // Movie: M, Released: 1931
327 pub fn captures_iter<'r, 't>(&'r self, text: &'t str)
328 -> FindCaptures<'r, 't> {
337 /// Returns an iterator of substrings of `text` delimited by a match
338 /// of the regular expression.
339 /// Namely, each element of the iterator corresponds to text that *isn't*
340 /// matched by the regular expression.
342 /// This method will *not* copy the text given.
346 /// To split a string delimited by arbitrary amounts of spaces or tabs:
349 /// # #![feature(phase)]
350 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
352 /// let re = regex!(r"[ \t]+");
353 /// let fields: Vec<&str> = re.split("a b \t c\td e").collect();
354 /// assert_eq!(fields, vec!("a", "b", "c", "d", "e"));
357 pub fn split<'r, 't>(&'r self, text: &'t str) -> RegexSplits<'r, 't> {
359 finder: self.find_iter(text),
364 /// Returns an iterator of at most `limit` substrings of `text` delimited
365 /// by a match of the regular expression. (A `limit` of `0` will return no
367 /// Namely, each element of the iterator corresponds to text that *isn't*
368 /// matched by the regular expression.
369 /// The remainder of the string that is not split will be the last element
372 /// This method will *not* copy the text given.
376 /// Get the first two words in some text:
379 /// # #![feature(phase)]
380 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
382 /// let re = regex!(r"\W+");
383 /// let fields: Vec<&str> = re.splitn("Hey! How are you?", 3).collect();
384 /// assert_eq!(fields, vec!("Hey", "How", "are you?"));
387 pub fn splitn<'r, 't>(&'r self, text: &'t str, limit: uint)
388 -> RegexSplitsN<'r, 't> {
390 splits: self.split(text),
396 /// Replaces the leftmost-first match with the replacement provided.
397 /// The replacement can be a regular string (where `$N` and `$name` are
398 /// expanded to match capture groups) or a function that takes the matches'
399 /// `Captures` and returns the replaced string.
401 /// If no match is found, then a copy of the string is returned unchanged.
405 /// Note that this function is polymorphic with respect to the replacement.
406 /// In typical usage, this can just be a normal string:
409 /// # #![feature(phase)]
410 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
412 /// let re = regex!("[^01]+");
413 /// assert_eq!(re.replace("1078910", "").as_slice(), "1010");
417 /// But anything satisfying the `Replacer` trait will work. For example,
418 /// a closure of type `|&Captures| -> String` provides direct access to the
419 /// captures corresponding to a match. This allows one to access
420 /// submatches easily:
423 /// # #![feature(phase)]
424 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
425 /// # use regex::Captures; fn main() {
426 /// let re = regex!(r"([^,\s]+),\s+(\S+)");
427 /// let result = re.replace("Springsteen, Bruce", |caps: &Captures| {
428 /// format!("{} {}", caps.at(2), caps.at(1))
430 /// assert_eq!(result.as_slice(), "Bruce Springsteen");
434 /// But this is a bit cumbersome to use all the time. Instead, a simple
435 /// syntax is supported that expands `$name` into the corresponding capture
436 /// group. Here's the last example, but using this expansion technique
437 /// with named capture groups:
440 /// # #![feature(phase)]
441 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
443 /// let re = regex!(r"(?P<last>[^,\s]+),\s+(?P<first>\S+)");
444 /// let result = re.replace("Springsteen, Bruce", "$first $last");
445 /// assert_eq!(result.as_slice(), "Bruce Springsteen");
449 /// Note that using `$2` instead of `$first` or `$1` instead of `$last`
450 /// would produce the same result. To write a literal `$` use `$$`.
452 /// Finally, sometimes you just want to replace a literal string with no
453 /// submatch expansion. This can be done by wrapping a string with
457 /// # #![feature(phase)]
458 /// # extern crate regex; #[phase(syntax)] extern crate regex_macros;
460 /// use regex::NoExpand;
462 /// let re = regex!(r"(?P<last>[^,\s]+),\s+(\S+)");
463 /// let result = re.replace("Springsteen, Bruce", NoExpand("$2 $last"));
464 /// assert_eq!(result.as_slice(), "$2 $last");
467 pub fn replace<R: Replacer>(&self, text: &str, rep: R) -> String {
468 self.replacen(text, 1, rep)
471 /// Replaces all non-overlapping matches in `text` with the
472 /// replacement provided. This is the same as calling `replacen` with
473 /// `limit` set to `0`.
475 /// See the documentation for `replace` for details on how to access
476 /// submatches in the replacement string.
477 pub fn replace_all<R: Replacer>(&self, text: &str, rep: R) -> String {
478 self.replacen(text, 0, rep)
481 /// Replaces at most `limit` non-overlapping matches in `text` with the
482 /// replacement provided. If `limit` is 0, then all non-overlapping matches
485 /// See the documentation for `replace` for details on how to access
486 /// submatches in the replacement string.
487 pub fn replacen<R: Replacer>
488 (&self, text: &str, limit: uint, mut rep: R) -> String {
489 let mut new = String::with_capacity(text.len());
490 let mut last_match = 0u;
492 for (i, cap) in self.captures_iter(text).enumerate() {
493 // It'd be nicer to use the 'take' iterator instead, but it seemed
494 // awkward given that '0' => no limit.
495 if limit > 0 && i >= limit {
499 let (s, e) = cap.pos(0).unwrap(); // captures only reports matches
500 new.push_str(text.slice(last_match, s));
501 new.push_str(rep.reg_replace(&cap).as_slice());
504 new.append(text.slice(last_match, text.len()))
507 /// Returns the original string of this regex.
508 pub fn as_str<'a>(&'a self) -> &'a str {
510 Dynamic(Dynamic { ref original, .. }) => original.as_slice(),
511 Native(Native { ref original, .. }) => original.as_slice(),
516 #[allow(visible_private_types)]
518 pub fn names_iter<'a>(&'a self) -> NamesIter<'a> {
520 Native(ref n) => NamesIterNative(n.names.iter()),
521 Dynamic(ref d) => NamesIterDynamic(d.names.iter())
525 fn names_len(&self) -> uint {
527 Native(ref n) => n.names.len(),
528 Dynamic(ref d) => d.names.len()
535 NamesIterNative(::std::slice::Items<'a, Option<&'static str>>),
536 NamesIterDynamic(::std::slice::Items<'a, Option<String>>)
539 impl<'a> Iterator<Option<String>> for NamesIter<'a> {
540 fn next(&mut self) -> Option<Option<String>> {
542 NamesIterNative(ref mut i) => i.next().map(|x| x.map(|s| s.to_string())),
543 NamesIterDynamic(ref mut i) => i.next().map(|x| x.as_ref().map(|s| s.to_string())),
548 /// NoExpand indicates literal string replacement.
550 /// It can be used with `replace` and `replace_all` to do a literal
551 /// string replacement without expanding `$name` to their corresponding
554 /// `'r` is the lifetime of the literal text.
555 pub struct NoExpand<'t>(pub &'t str);
557 /// Replacer describes types that can be used to replace matches in a string.
559 /// Returns a possibly owned string that is used to replace the match
560 /// corresponding the the `caps` capture group.
562 /// The `'a` lifetime refers to the lifetime of a borrowed string when
563 /// a new owned string isn't needed (e.g., for `NoExpand`).
564 fn reg_replace<'a>(&'a mut self, caps: &Captures) -> MaybeOwned<'a>;
567 impl<'t> Replacer for NoExpand<'t> {
568 fn reg_replace<'a>(&'a mut self, _: &Captures) -> MaybeOwned<'a> {
569 let NoExpand(s) = *self;
574 impl<'t> Replacer for &'t str {
575 fn reg_replace<'a>(&'a mut self, caps: &Captures) -> MaybeOwned<'a> {
576 Owned(caps.expand(*self))
580 impl<'a> Replacer for |&Captures|: 'a -> String {
581 fn reg_replace<'r>(&'r mut self, caps: &Captures) -> MaybeOwned<'r> {
586 /// Yields all substrings delimited by a regular expression match.
588 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
589 /// of the string being split.
590 pub struct RegexSplits<'r, 't> {
591 finder: FindMatches<'r, 't>,
595 impl<'r, 't> Iterator<&'t str> for RegexSplits<'r, 't> {
596 fn next(&mut self) -> Option<&'t str> {
597 let text = self.finder.search;
598 match self.finder.next() {
600 if self.last >= text.len() {
603 let s = text.slice(self.last, text.len());
604 self.last = text.len();
609 let matched = text.slice(self.last, s);
617 /// Yields at most `N` substrings delimited by a regular expression match.
619 /// The last substring will be whatever remains after splitting.
621 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
622 /// of the string being split.
623 pub struct RegexSplitsN<'r, 't> {
624 splits: RegexSplits<'r, 't>,
629 impl<'r, 't> Iterator<&'t str> for RegexSplitsN<'r, 't> {
630 fn next(&mut self) -> Option<&'t str> {
631 let text = self.splits.finder.search;
632 if self.cur >= self.limit {
636 if self.cur >= self.limit {
637 Some(text.slice(self.splits.last, text.len()))
645 /// Captures represents a group of captured strings for a single match.
647 /// The 0th capture always corresponds to the entire match. Each subsequent
648 /// index corresponds to the next capture group in the regex.
649 /// If a capture group is named, then the matched string is *also* available
650 /// via the `name` method. (Note that the 0th capture is always unnamed and so
651 /// must be accessed with the `at` method.)
653 /// Positions returned from a capture group are always byte indices.
655 /// `'t` is the lifetime of the matched text.
656 pub struct Captures<'t> {
659 named: Option<HashMap<String, uint>>,
662 impl<'t> Captures<'t> {
663 #[allow(experimental)]
664 fn new(re: &Regex, search: &'t str, locs: CaptureLocs)
665 -> Option<Captures<'t>> {
666 if !has_match(&locs) {
671 if re.names_len() == 0 {
674 let mut named = HashMap::new();
675 for (i, name) in re.names_iter().enumerate() {
679 named.insert(name, i);
692 /// Returns the start and end positions of the Nth capture group.
693 /// Returns `None` if `i` is not a valid capture group or if the capture
694 /// group did not match anything.
695 /// The positions returned are *always* byte indices with respect to the
696 /// original string matched.
697 pub fn pos(&self, i: uint) -> Option<(uint, uint)> {
698 let (s, e) = (i * 2, i * 2 + 1);
699 if e >= self.locs.len() || self.locs.get(s).is_none() {
700 // VM guarantees that each pair of locations are both Some or None.
703 Some((self.locs.get(s).unwrap(), self.locs.get(e).unwrap()))
706 /// Returns the matched string for the capture group `i`.
707 /// If `i` isn't a valid capture group or didn't match anything, then the
708 /// empty string is returned.
709 pub fn at(&self, i: uint) -> &'t str {
713 self.text.slice(s, e)
718 /// Returns the matched string for the capture group named `name`.
719 /// If `name` isn't a valid capture group or didn't match anything, then
720 /// the empty string is returned.
721 pub fn name(&self, name: &str) -> &'t str {
725 match h.find_equiv(&name) {
727 Some(i) => self.at(*i),
733 /// Creates an iterator of all the capture groups in order of appearance
734 /// in the regular expression.
735 pub fn iter(&'t self) -> SubCaptures<'t> {
736 SubCaptures { idx: 0, caps: self, }
739 /// Creates an iterator of all the capture group positions in order of
740 /// appearance in the regular expression. Positions are byte indices
741 /// in terms of the original string matched.
742 pub fn iter_pos(&'t self) -> SubCapturesPos<'t> {
743 SubCapturesPos { idx: 0, caps: self, }
746 /// Expands all instances of `$name` in `text` to the corresponding capture
749 /// `name` may be an integer corresponding to the index of the
750 /// capture group (counted by order of opening parenthesis where `0` is the
751 /// entire match) or it can be a name (consisting of letters, digits or
752 /// underscores) corresponding to a named capture group.
754 /// If `name` isn't a valid capture group (whether the name doesn't exist or
755 /// isn't a valid index), then it is replaced with the empty string.
757 /// To write a literal `$` use `$$`.
758 pub fn expand(&self, text: &str) -> String {
759 // How evil can you get?
760 // FIXME: Don't use regexes for this. It's completely unnecessary.
761 let re = Regex::new(r"(^|[^$]|\b)\$(\w+)").unwrap();
762 let text = re.replace_all(text, |refs: &Captures| -> String {
763 let (pre, name) = (refs.at(1), refs.at(2));
765 match from_str::<uint>(name.as_slice()) {
766 None => self.name(name).to_string(),
767 Some(i) => self.at(i).to_string(),
770 let re = Regex::new(r"\$\$").unwrap();
771 re.replace_all(text.as_slice(), NoExpand("$"))
775 impl<'t> Container for Captures<'t> {
776 /// Returns the number of captured groups.
778 fn len(&self) -> uint {
783 /// An iterator over capture groups for a particular match of a regular
786 /// `'t` is the lifetime of the matched text.
787 pub struct SubCaptures<'t> {
789 caps: &'t Captures<'t>,
792 impl<'t> Iterator<&'t str> for SubCaptures<'t> {
793 fn next(&mut self) -> Option<&'t str> {
794 if self.idx < self.caps.len() {
796 Some(self.caps.at(self.idx - 1))
803 /// An iterator over capture group positions for a particular match of a
804 /// regular expression.
806 /// Positions are byte indices in terms of the original string matched.
808 /// `'t` is the lifetime of the matched text.
809 pub struct SubCapturesPos<'t> {
811 caps: &'t Captures<'t>,
814 impl<'t> Iterator<Option<(uint, uint)>> for SubCapturesPos<'t> {
815 fn next(&mut self) -> Option<Option<(uint, uint)>> {
816 if self.idx < self.caps.len() {
818 Some(self.caps.pos(self.idx - 1))
825 /// An iterator that yields all non-overlapping capture groups matching a
826 /// particular regular expression. The iterator stops when no more matches can
829 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
830 /// of the matched string.
831 pub struct FindCaptures<'r, 't> {
834 last_match: Option<uint>,
838 impl<'r, 't> Iterator<Captures<'t>> for FindCaptures<'r, 't> {
839 fn next(&mut self) -> Option<Captures<'t>> {
840 if self.last_end > self.search.len() {
844 let caps = exec_slice(self.re, Submatches, self.search,
845 self.last_end, self.search.len());
847 if !has_match(&caps) {
850 (caps.get(0).unwrap(), caps.get(1).unwrap())
853 // Don't accept empty matches immediately following a match.
854 // i.e., no infinite loops please.
855 if e == s && Some(self.last_end) == self.last_match {
860 self.last_match = Some(self.last_end);
861 Captures::new(self.re, self.search, caps)
865 /// An iterator over all non-overlapping matches for a particular string.
867 /// The iterator yields a tuple of integers corresponding to the start and end
868 /// of the match. The indices are byte offsets. The iterator stops when no more
869 /// matches can be found.
871 /// `'r` is the lifetime of the compiled expression and `'t` is the lifetime
872 /// of the matched string.
873 pub struct FindMatches<'r, 't> {
876 last_match: Option<uint>,
880 impl<'r, 't> Iterator<(uint, uint)> for FindMatches<'r, 't> {
881 fn next(&mut self) -> Option<(uint, uint)> {
882 if self.last_end > self.search.len() {
886 let caps = exec_slice(self.re, Location, self.search,
887 self.last_end, self.search.len());
889 if !has_match(&caps) {
892 (caps.get(0).unwrap(), caps.get(1).unwrap())
895 // Don't accept empty matches immediately following a match.
896 // i.e., no infinite loops please.
897 if e == s && Some(self.last_end) == self.last_match {
902 self.last_match = Some(self.last_end);
907 fn exec(re: &Regex, which: MatchKind, input: &str) -> CaptureLocs {
908 exec_slice(re, which, input, 0, input.len())
911 fn exec_slice(re: &Regex, which: MatchKind,
912 input: &str, s: uint, e: uint) -> CaptureLocs {
914 Dynamic(Dynamic { ref prog, .. }) => vm::run(which, prog, input, s, e),
915 Native(Native { prog, .. }) => prog(which, input, s, e),
920 fn has_match(caps: &CaptureLocs) -> bool {
921 caps.len() >= 2 && caps.get(0).is_some() && caps.get(1).is_some()