1 // Copyright 2012-2014 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 // ignore-lexer-test FIXME #15679
13 //! String manipulation
15 //! For more details, see std::str
17 #![doc(primitive = "str")]
19 use self::Searcher::{Naive, TwoWay, TwoWayLong};
24 use iter::ExactSizeIterator;
25 use iter::{Map, Iterator, IteratorExt, DoubleEndedIterator};
30 use option::Option::{self, None, Some};
32 use raw::{Repr, Slice};
33 use result::Result::{self, Ok, Err};
34 use slice::{self, SliceExt};
37 macro_rules! delegate_iter {
38 (exact $te:ty in $ti:ty) => {
39 delegate_iter!{$te in $ti}
40 impl<'a> ExactSizeIterator for $ti {
42 fn rposition<P>(&mut self, predicate: P) -> Option<uint> where P: FnMut($te) -> bool{
43 self.0.rposition(predicate)
46 fn len(&self) -> uint {
51 ($te:ty in $ti:ty) => {
52 impl<'a> Iterator for $ti {
56 fn next(&mut self) -> Option<$te> {
60 fn size_hint(&self) -> (uint, Option<uint>) {
64 impl<'a> DoubleEndedIterator for $ti {
66 fn next_back(&mut self) -> Option<$te> {
71 (pattern $te:ty in $ti:ty) => {
72 impl<'a, P: CharEq> Iterator for $ti {
76 fn next(&mut self) -> Option<$te> {
80 fn size_hint(&self) -> (uint, Option<uint>) {
84 impl<'a, P: CharEq> DoubleEndedIterator for $ti {
86 fn next_back(&mut self) -> Option<$te> {
91 (pattern forward $te:ty in $ti:ty) => {
92 impl<'a, P: CharEq> Iterator for $ti {
96 fn next(&mut self) -> Option<$te> {
100 fn size_hint(&self) -> (uint, Option<uint>) {
107 /// A trait to abstract the idea of creating a new instance of a type from a
109 // FIXME(#17307): there should be an `E` associated type for a `Result` return
110 #[unstable = "will return a Result once associated types are working"]
112 /// Parses a string `s` to return an optional value of this type. If the
113 /// string is ill-formatted, the None is returned.
114 fn from_str(s: &str) -> Option<Self>;
117 impl FromStr for bool {
118 /// Parse a `bool` from a string.
120 /// Yields an `Option<bool>`, because `s` may or may not actually be parseable.
125 /// assert_eq!("true".parse(), Some(true));
126 /// assert_eq!("false".parse(), Some(false));
127 /// assert_eq!("not even a boolean".parse::<bool>(), None);
130 fn from_str(s: &str) -> Option<bool> {
132 "true" => Some(true),
133 "false" => Some(false),
140 Section: Creating a string
143 /// Errors which can occur when attempting to interpret a byte slice as a `str`.
144 #[derive(Copy, Eq, PartialEq, Clone)]
146 /// An invalid byte was detected at the byte offset given.
148 /// The offset is guaranteed to be in bounds of the slice in question, and
149 /// the byte at the specified offset was the first invalid byte in the
150 /// sequence detected.
153 /// The byte slice was invalid because more bytes were needed but no more
154 /// bytes were available.
158 /// Converts a slice of bytes to a string slice without performing any
161 /// Once the slice has been validated as utf-8, it is transmuted in-place and
162 /// returned as a '&str' instead of a '&[u8]'
166 /// Returns `Err` if the slice is not utf-8 with a description as to why the
167 /// provided slice is not utf-8.
168 pub fn from_utf8(v: &[u8]) -> Result<&str, Utf8Error> {
169 try!(run_utf8_validation_iterator(&mut v.iter()));
170 Ok(unsafe { from_utf8_unchecked(v) })
173 /// Converts a slice of bytes to a string slice without checking
174 /// that the string contains valid UTF-8.
176 pub unsafe fn from_utf8_unchecked<'a>(v: &'a [u8]) -> &'a str {
180 /// Constructs a static string slice from a given raw pointer.
182 /// This function will read memory starting at `s` until it finds a 0, and then
183 /// transmute the memory up to that point as a string slice, returning the
184 /// corresponding `&'static str` value.
186 /// This function is unsafe because the caller must ensure the C string itself
187 /// has the static lifetime and that the memory `s` is valid up to and including
188 /// the first null byte.
192 /// This function will panic if the string pointed to by `s` is not valid UTF-8.
193 #[unstable = "may change location based on the outcome of the c_str module"]
194 pub unsafe fn from_c_str(s: *const i8) -> &'static str {
195 let s = s as *const u8;
197 while *s.offset(len as int) != 0 {
200 let v: &'static [u8] = ::mem::transmute(Slice { data: s, len: len });
201 from_utf8(v).ok().expect("from_c_str passed invalid utf-8 data")
204 /// Something that can be used to compare against a character
205 #[unstable = "definition may change as pattern-related methods are stabilized"]
207 /// Determine if the splitter should split at the given character
208 fn matches(&mut self, char) -> bool;
209 /// Indicate if this is only concerned about ASCII characters,
210 /// which can allow for a faster implementation.
211 fn only_ascii(&self) -> bool;
214 impl CharEq for char {
216 fn matches(&mut self, c: char) -> bool { *self == c }
219 fn only_ascii(&self) -> bool { (*self as uint) < 128 }
222 impl<F> CharEq for F where F: FnMut(char) -> bool {
224 fn matches(&mut self, c: char) -> bool { (*self)(c) }
227 fn only_ascii(&self) -> bool { false }
230 impl<'a> CharEq for &'a [char] {
232 fn matches(&mut self, c: char) -> bool {
233 self.iter().any(|&mut m| m.matches(c))
237 fn only_ascii(&self) -> bool {
238 self.iter().all(|m| m.only_ascii())
246 /// Iterator for the char (representing *Unicode Scalar Values*) of a string
248 /// Created with the method `.chars()`.
249 #[derive(Clone, Copy)]
250 pub struct Chars<'a> {
251 iter: slice::Iter<'a, u8>
254 // Return the initial codepoint accumulator for the first byte.
255 // The first byte is special, only want bottom 5 bits for width 2, 4 bits
256 // for width 3, and 3 bits for width 4
257 macro_rules! utf8_first_byte {
258 ($byte:expr, $width:expr) => (($byte & (0x7F >> $width)) as u32)
261 // return the value of $ch updated with continuation byte $byte
262 macro_rules! utf8_acc_cont_byte {
263 ($ch:expr, $byte:expr) => (($ch << 6) | ($byte & CONT_MASK) as u32)
266 macro_rules! utf8_is_cont_byte {
267 ($byte:expr) => (($byte & !CONT_MASK) == TAG_CONT_U8)
271 fn unwrap_or_0(opt: Option<&u8>) -> u8 {
278 impl<'a> Iterator for Chars<'a> {
282 fn next(&mut self) -> Option<char> {
283 // Decode UTF-8, using the valid UTF-8 invariant
284 let x = match self.iter.next() {
286 Some(&next_byte) if next_byte < 128 => return Some(next_byte as char),
287 Some(&next_byte) => next_byte,
290 // Multibyte case follows
291 // Decode from a byte combination out of: [[[x y] z] w]
292 // NOTE: Performance is sensitive to the exact formulation here
293 let init = utf8_first_byte!(x, 2);
294 let y = unwrap_or_0(self.iter.next());
295 let mut ch = utf8_acc_cont_byte!(init, y);
298 // 5th bit in 0xE0 .. 0xEF is always clear, so `init` is still valid
299 let z = unwrap_or_0(self.iter.next());
300 let y_z = utf8_acc_cont_byte!((y & CONT_MASK) as u32, z);
301 ch = init << 12 | y_z;
304 // use only the lower 3 bits of `init`
305 let w = unwrap_or_0(self.iter.next());
306 ch = (init & 7) << 18 | utf8_acc_cont_byte!(y_z, w);
310 // str invariant says `ch` is a valid Unicode Scalar Value
312 Some(mem::transmute(ch))
317 fn size_hint(&self) -> (uint, Option<uint>) {
318 let (len, _) = self.iter.size_hint();
319 (len.saturating_add(3) / 4, Some(len))
323 impl<'a> DoubleEndedIterator for Chars<'a> {
325 fn next_back(&mut self) -> Option<char> {
326 let w = match self.iter.next_back() {
328 Some(&back_byte) if back_byte < 128 => return Some(back_byte as char),
329 Some(&back_byte) => back_byte,
332 // Multibyte case follows
333 // Decode from a byte combination out of: [x [y [z w]]]
335 let z = unwrap_or_0(self.iter.next_back());
336 ch = utf8_first_byte!(z, 2);
337 if utf8_is_cont_byte!(z) {
338 let y = unwrap_or_0(self.iter.next_back());
339 ch = utf8_first_byte!(y, 3);
340 if utf8_is_cont_byte!(y) {
341 let x = unwrap_or_0(self.iter.next_back());
342 ch = utf8_first_byte!(x, 4);
343 ch = utf8_acc_cont_byte!(ch, y);
345 ch = utf8_acc_cont_byte!(ch, z);
347 ch = utf8_acc_cont_byte!(ch, w);
349 // str invariant says `ch` is a valid Unicode Scalar Value
351 Some(mem::transmute(ch))
356 /// External iterator for a string's characters and their byte offsets.
357 /// Use with the `std::iter` module.
359 pub struct CharIndices<'a> {
364 impl<'a> Iterator for CharIndices<'a> {
365 type Item = (uint, char);
368 fn next(&mut self) -> Option<(uint, char)> {
369 let (pre_len, _) = self.iter.iter.size_hint();
370 match self.iter.next() {
373 let index = self.front_offset;
374 let (len, _) = self.iter.iter.size_hint();
375 self.front_offset += pre_len - len;
382 fn size_hint(&self) -> (uint, Option<uint>) {
383 self.iter.size_hint()
387 impl<'a> DoubleEndedIterator for CharIndices<'a> {
389 fn next_back(&mut self) -> Option<(uint, char)> {
390 match self.iter.next_back() {
393 let (len, _) = self.iter.iter.size_hint();
394 let index = self.front_offset + len;
401 /// External iterator for a string's bytes.
402 /// Use with the `std::iter` module.
404 /// Created with `StrExt::bytes`
407 pub struct Bytes<'a>(Map<&'a u8, u8, slice::Iter<'a, u8>, BytesDeref>);
408 delegate_iter!{exact u8 in Bytes<'a>}
410 /// A temporary fn new type that ensures that the `Bytes` iterator
412 #[derive(Copy, Clone)]
415 impl<'a> Fn(&'a u8) -> u8 for BytesDeref {
417 extern "rust-call" fn call(&self, (ptr,): (&'a u8,)) -> u8 {
422 /// An iterator over the substrings of a string, separated by `sep`.
424 struct CharSplits<'a, Sep> {
425 /// The slice remaining to be iterated
428 /// Whether an empty string at the end is allowed
429 allow_trailing_empty: bool,
434 /// An iterator over the substrings of a string, separated by `sep`,
435 /// splitting at most `count` times.
437 struct CharSplitsN<'a, Sep> {
438 iter: CharSplits<'a, Sep>,
439 /// The number of splits remaining
444 /// An iterator over the lines of a string, separated by `\n`.
446 pub struct Lines<'a> {
447 inner: CharSplits<'a, char>,
450 /// An iterator over the lines of a string, separated by either `\n` or (`\r\n`).
452 pub struct LinesAny<'a> {
453 inner: Map<&'a str, &'a str, Lines<'a>, fn(&str) -> &str>,
456 impl<'a, Sep> CharSplits<'a, Sep> {
458 fn get_end(&mut self) -> Option<&'a str> {
459 if !self.finished && (self.allow_trailing_empty || self.string.len() > 0) {
460 self.finished = true;
468 impl<'a, Sep: CharEq> Iterator for CharSplits<'a, Sep> {
472 fn next(&mut self) -> Option<&'a str> {
473 if self.finished { return None }
475 let mut next_split = None;
477 for (idx, byte) in self.string.bytes().enumerate() {
478 if self.sep.matches(byte as char) && byte < 128u8 {
479 next_split = Some((idx, idx + 1));
484 for (idx, ch) in self.string.char_indices() {
485 if self.sep.matches(ch) {
486 next_split = Some((idx, self.string.char_range_at(idx).next));
492 Some((a, b)) => unsafe {
493 let elt = self.string.slice_unchecked(0, a);
494 self.string = self.string.slice_unchecked(b, self.string.len());
497 None => self.get_end(),
502 impl<'a, Sep: CharEq> DoubleEndedIterator for CharSplits<'a, Sep> {
504 fn next_back(&mut self) -> Option<&'a str> {
505 if self.finished { return None }
507 if !self.allow_trailing_empty {
508 self.allow_trailing_empty = true;
509 match self.next_back() {
510 Some(elt) if !elt.is_empty() => return Some(elt),
511 _ => if self.finished { return None }
514 let len = self.string.len();
515 let mut next_split = None;
518 for (idx, byte) in self.string.bytes().enumerate().rev() {
519 if self.sep.matches(byte as char) && byte < 128u8 {
520 next_split = Some((idx, idx + 1));
525 for (idx, ch) in self.string.char_indices().rev() {
526 if self.sep.matches(ch) {
527 next_split = Some((idx, self.string.char_range_at(idx).next));
533 Some((a, b)) => unsafe {
534 let elt = self.string.slice_unchecked(b, len);
535 self.string = self.string.slice_unchecked(0, a);
538 None => { self.finished = true; Some(self.string) }
543 impl<'a, Sep: CharEq> Iterator for CharSplitsN<'a, Sep> {
547 fn next(&mut self) -> Option<&'a str> {
550 if self.invert { self.iter.next_back() } else { self.iter.next() }
557 /// The internal state of an iterator that searches for matches of a substring
558 /// within a larger string using naive search
560 struct NaiveSearcher {
565 fn new() -> NaiveSearcher {
566 NaiveSearcher { position: 0 }
569 fn next(&mut self, haystack: &[u8], needle: &[u8]) -> Option<(uint, uint)> {
570 while self.position + needle.len() <= haystack.len() {
571 if haystack[self.position .. self.position + needle.len()] == needle {
572 let match_pos = self.position;
573 self.position += needle.len(); // add 1 for all matches
574 return Some((match_pos, match_pos + needle.len()));
583 /// The internal state of an iterator that searches for matches of a substring
584 /// within a larger string using two-way search
586 struct TwoWaySearcher {
598 This is the Two-Way search algorithm, which was introduced in the paper:
599 Crochemore, M., Perrin, D., 1991, Two-way string-matching, Journal of the ACM 38(3):651-675.
601 Here's some background information.
603 A *word* is a string of symbols. The *length* of a word should be a familiar
604 notion, and here we denote it for any word x by |x|.
605 (We also allow for the possibility of the *empty word*, a word of length zero).
607 If x is any non-empty word, then an integer p with 0 < p <= |x| is said to be a
608 *period* for x iff for all i with 0 <= i <= |x| - p - 1, we have x[i] == x[i+p].
609 For example, both 1 and 2 are periods for the string "aa". As another example,
610 the only period of the string "abcd" is 4.
612 We denote by period(x) the *smallest* period of x (provided that x is non-empty).
613 This is always well-defined since every non-empty word x has at least one period,
614 |x|. We sometimes call this *the period* of x.
616 If u, v and x are words such that x = uv, where uv is the concatenation of u and
617 v, then we say that (u, v) is a *factorization* of x.
619 Let (u, v) be a factorization for a word x. Then if w is a non-empty word such
620 that both of the following hold
622 - either w is a suffix of u or u is a suffix of w
623 - either w is a prefix of v or v is a prefix of w
625 then w is said to be a *repetition* for the factorization (u, v).
627 Just to unpack this, there are four possibilities here. Let w = "abc". Then we
630 - w is a suffix of u and w is a prefix of v. ex: ("lolabc", "abcde")
631 - w is a suffix of u and v is a prefix of w. ex: ("lolabc", "ab")
632 - u is a suffix of w and w is a prefix of v. ex: ("bc", "abchi")
633 - u is a suffix of w and v is a prefix of w. ex: ("bc", "a")
635 Note that the word vu is a repetition for any factorization (u,v) of x = uv,
636 so every factorization has at least one repetition.
638 If x is a string and (u, v) is a factorization for x, then a *local period* for
639 (u, v) is an integer r such that there is some word w such that |w| = r and w is
640 a repetition for (u, v).
642 We denote by local_period(u, v) the smallest local period of (u, v). We sometimes
643 call this *the local period* of (u, v). Provided that x = uv is non-empty, this
644 is well-defined (because each non-empty word has at least one factorization, as
647 It can be proven that the following is an equivalent definition of a local period
648 for a factorization (u, v): any positive integer r such that x[i] == x[i+r] for
649 all i such that |u| - r <= i <= |u| - 1 and such that both x[i] and x[i+r] are
650 defined. (i.e. i > 0 and i + r < |x|).
652 Using the above reformulation, it is easy to prove that
654 1 <= local_period(u, v) <= period(uv)
656 A factorization (u, v) of x such that local_period(u,v) = period(x) is called a
657 *critical factorization*.
659 The algorithm hinges on the following theorem, which is stated without proof:
661 **Critical Factorization Theorem** Any word x has at least one critical
662 factorization (u, v) such that |u| < period(x).
664 The purpose of maximal_suffix is to find such a critical factorization.
667 impl TwoWaySearcher {
668 fn new(needle: &[u8]) -> TwoWaySearcher {
669 let (crit_pos1, period1) = TwoWaySearcher::maximal_suffix(needle, false);
670 let (crit_pos2, period2) = TwoWaySearcher::maximal_suffix(needle, true);
674 if crit_pos1 > crit_pos2 {
675 crit_pos = crit_pos1;
678 crit_pos = crit_pos2;
682 // This isn't in the original algorithm, as far as I'm aware.
683 let byteset = needle.iter()
684 .fold(0, |a, &b| (1 << ((b & 0x3f) as uint)) | a);
686 // A particularly readable explanation of what's going on here can be found
687 // in Crochemore and Rytter's book "Text Algorithms", ch 13. Specifically
688 // see the code for "Algorithm CP" on p. 323.
690 // What's going on is we have some critical factorization (u, v) of the
691 // needle, and we want to determine whether u is a suffix of
692 // v[..period]. If it is, we use "Algorithm CP1". Otherwise we use
693 // "Algorithm CP2", which is optimized for when the period of the needle
695 if needle[..crit_pos] == needle[period.. period + crit_pos] {
707 period: cmp::max(crit_pos, needle.len() - crit_pos) + 1,
711 memory: uint::MAX // Dummy value to signify that the period is long
716 // One of the main ideas of Two-Way is that we factorize the needle into
717 // two halves, (u, v), and begin trying to find v in the haystack by scanning
718 // left to right. If v matches, we try to match u by scanning right to left.
719 // How far we can jump when we encounter a mismatch is all based on the fact
720 // that (u, v) is a critical factorization for the needle.
722 fn next(&mut self, haystack: &[u8], needle: &[u8], long_period: bool) -> Option<(uint, uint)> {
724 // Check that we have room to search in
725 if self.position + needle.len() > haystack.len() {
729 // Quickly skip by large portions unrelated to our substring
731 ((haystack[self.position + needle.len() - 1] & 0x3f)
733 self.position += needle.len();
740 // See if the right part of the needle matches
741 let start = if long_period { self.crit_pos }
742 else { cmp::max(self.crit_pos, self.memory) };
743 for i in range(start, needle.len()) {
744 if needle[i] != haystack[self.position + i] {
745 self.position += i - self.crit_pos + 1;
753 // See if the left part of the needle matches
754 let start = if long_period { 0 } else { self.memory };
755 for i in range(start, self.crit_pos).rev() {
756 if needle[i] != haystack[self.position + i] {
757 self.position += self.period;
759 self.memory = needle.len() - self.period;
765 // We have found a match!
766 let match_pos = self.position;
767 self.position += needle.len(); // add self.period for all matches
769 self.memory = 0; // set to needle.len() - self.period for all matches
771 return Some((match_pos, match_pos + needle.len()));
775 // Computes a critical factorization (u, v) of `arr`.
776 // Specifically, returns (i, p), where i is the starting index of v in some
777 // critical factorization (u, v) and p = period(v)
779 fn maximal_suffix(arr: &[u8], reversed: bool) -> (uint, uint) {
780 let mut left = -1; // Corresponds to i in the paper
781 let mut right = 0; // Corresponds to j in the paper
782 let mut offset = 1; // Corresponds to k in the paper
783 let mut period = 1; // Corresponds to p in the paper
785 while right + offset < arr.len() {
789 a = arr[left + offset];
790 b = arr[right + offset];
792 a = arr[right + offset];
793 b = arr[left + offset];
796 // Suffix is smaller, period is entire prefix so far.
799 period = right - left;
801 // Advance through repetition of the current period.
802 if offset == period {
809 // Suffix is larger, start over from current location.
820 /// The internal state of an iterator that searches for matches of a substring
821 /// within a larger string using a dynamically chosen search algorithm
824 Naive(NaiveSearcher),
825 TwoWay(TwoWaySearcher),
826 TwoWayLong(TwoWaySearcher)
830 fn new(haystack: &[u8], needle: &[u8]) -> Searcher {
832 // FIXME(#16715): This unsigned integer addition will probably not
833 // overflow because that would mean that the memory almost solely
834 // consists of the needle. Needs #16715 to be formally fixed.
835 if needle.len() + 20 > haystack.len() {
836 Naive(NaiveSearcher::new())
838 let searcher = TwoWaySearcher::new(needle);
839 if searcher.memory == uint::MAX { // If the period is long
848 /// An iterator over the start and end indices of the matches of a
849 /// substring within a larger string
851 pub struct MatchIndices<'a> {
858 /// An iterator over the substrings of a string separated by a given
861 #[unstable = "Type might get removed"]
862 pub struct SplitStr<'a> {
863 it: MatchIndices<'a>,
868 impl<'a> Iterator for MatchIndices<'a> {
869 type Item = (uint, uint);
872 fn next(&mut self) -> Option<(uint, uint)> {
873 match self.searcher {
874 Naive(ref mut searcher)
875 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes()),
876 TwoWay(ref mut searcher)
877 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes(), false),
878 TwoWayLong(ref mut searcher)
879 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes(), true)
884 impl<'a> Iterator for SplitStr<'a> {
888 fn next(&mut self) -> Option<&'a str> {
889 if self.finished { return None; }
891 match self.it.next() {
892 Some((from, to)) => {
893 let ret = Some(self.it.haystack.slice(self.last_end, from));
898 self.finished = true;
899 Some(self.it.haystack.slice(self.last_end, self.it.haystack.len()))
907 Section: Comparing strings
910 // share the implementation of the lang-item vs. non-lang-item
912 /// NOTE: This function is (ab)used in rustc::middle::trans::_match
913 /// to compare &[u8] byte slices that are not necessarily valid UTF-8.
915 fn eq_slice_(a: &str, b: &str) -> bool {
916 #[allow(improper_ctypes)]
917 extern { fn memcmp(s1: *const i8, s2: *const i8, n: uint) -> i32; }
918 a.len() == b.len() && unsafe {
919 memcmp(a.as_ptr() as *const i8,
920 b.as_ptr() as *const i8,
925 /// Bytewise slice equality
926 /// NOTE: This function is (ab)used in rustc::middle::trans::_match
927 /// to compare &[u8] byte slices that are not necessarily valid UTF-8.
930 fn eq_slice(a: &str, b: &str) -> bool {
938 /// Walk through `iter` checking that it's a valid UTF-8 sequence,
939 /// returning `true` in that case, or, if it is invalid, `false` with
940 /// `iter` reset such that it is pointing at the first byte in the
941 /// invalid sequence.
943 fn run_utf8_validation_iterator(iter: &mut slice::Iter<u8>)
944 -> Result<(), Utf8Error> {
945 let whole = iter.as_slice();
947 // save the current thing we're pointing at.
950 // restore the iterator we had at the start of this codepoint.
951 macro_rules! err (() => { {
953 return Err(Utf8Error::InvalidByte(whole.len() - iter.as_slice().len()))
955 macro_rules! next ( () => {
958 // we needed data, but there was none: error!
959 None => return Err(Utf8Error::TooShort),
963 let first = match iter.next() {
965 // we're at the end of the iterator and a codepoint
966 // boundary at the same time, so this string is valid.
967 None => return Ok(())
970 // ASCII characters are always valid, so only large
971 // bytes need more examination.
973 let w = UTF8_CHAR_WIDTH[first as uint] as uint;
974 let second = next!();
975 // 2-byte encoding is for codepoints \u{0080} to \u{07ff}
976 // first C2 80 last DF BF
977 // 3-byte encoding is for codepoints \u{0800} to \u{ffff}
978 // first E0 A0 80 last EF BF BF
979 // excluding surrogates codepoints \u{d800} to \u{dfff}
980 // ED A0 80 to ED BF BF
981 // 4-byte encoding is for codepoints \u{1000}0 to \u{10ff}ff
982 // first F0 90 80 80 last F4 8F BF BF
984 // Use the UTF-8 syntax from the RFC
986 // https://tools.ietf.org/html/rfc3629
988 // UTF8-2 = %xC2-DF UTF8-tail
989 // UTF8-3 = %xE0 %xA0-BF UTF8-tail / %xE1-EC 2( UTF8-tail ) /
990 // %xED %x80-9F UTF8-tail / %xEE-EF 2( UTF8-tail )
991 // UTF8-4 = %xF0 %x90-BF 2( UTF8-tail ) / %xF1-F3 3( UTF8-tail ) /
992 // %xF4 %x80-8F 2( UTF8-tail )
994 2 => if second & !CONT_MASK != TAG_CONT_U8 {err!()},
996 match (first, second, next!() & !CONT_MASK) {
997 (0xE0 , 0xA0 ... 0xBF, TAG_CONT_U8) |
998 (0xE1 ... 0xEC, 0x80 ... 0xBF, TAG_CONT_U8) |
999 (0xED , 0x80 ... 0x9F, TAG_CONT_U8) |
1000 (0xEE ... 0xEF, 0x80 ... 0xBF, TAG_CONT_U8) => {}
1005 match (first, second, next!() & !CONT_MASK, next!() & !CONT_MASK) {
1006 (0xF0 , 0x90 ... 0xBF, TAG_CONT_U8, TAG_CONT_U8) |
1007 (0xF1 ... 0xF3, 0x80 ... 0xBF, TAG_CONT_U8, TAG_CONT_U8) |
1008 (0xF4 , 0x80 ... 0x8F, TAG_CONT_U8, TAG_CONT_U8) => {}
1018 // https://tools.ietf.org/html/rfc3629
1019 static UTF8_CHAR_WIDTH: [u8; 256] = [
1020 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1021 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x1F
1022 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1023 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x3F
1024 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1025 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x5F
1026 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1027 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x7F
1028 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
1029 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0x9F
1030 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
1031 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0xBF
1032 0,0,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
1033 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, // 0xDF
1034 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, // 0xEF
1035 4,4,4,4,4,0,0,0,0,0,0,0,0,0,0,0, // 0xFF
1038 /// Struct that contains a `char` and the index of the first byte of
1039 /// the next `char` in a string. This can be used as a data structure
1040 /// for iterating over the UTF-8 bytes of a string.
1042 #[unstable = "naming is uncertain with container conventions"]
1043 pub struct CharRange {
1046 /// Index of the first byte of the next `char`
1050 /// Mask of the value bits of a continuation byte
1051 const CONT_MASK: u8 = 0b0011_1111u8;
1052 /// Value of the tag bits (tag mask is !CONT_MASK) of a continuation byte
1053 const TAG_CONT_U8: u8 = 0b1000_0000u8;
1056 Section: Trait implementations
1059 #[allow(missing_docs)]
1061 use cmp::{Ordering, Ord, PartialEq, PartialOrd, Eq};
1062 use cmp::Ordering::{Less, Equal, Greater};
1063 use iter::IteratorExt;
1065 use option::Option::Some;
1067 use str::{StrExt, eq_slice};
1072 fn cmp(&self, other: &str) -> Ordering {
1073 for (s_b, o_b) in self.bytes().zip(other.bytes()) {
1074 match s_b.cmp(&o_b) {
1075 Greater => return Greater,
1076 Less => return Less,
1081 self.len().cmp(&other.len())
1086 impl PartialEq for str {
1088 fn eq(&self, other: &str) -> bool {
1089 eq_slice(self, other)
1092 fn ne(&self, other: &str) -> bool { !(*self).eq(other) }
1099 impl PartialOrd for str {
1101 fn partial_cmp(&self, other: &str) -> Option<Ordering> {
1102 Some(self.cmp(other))
1106 impl ops::Slice<uint, str> for str {
1108 fn as_slice_<'a>(&'a self) -> &'a str {
1113 fn slice_from_or_fail<'a>(&'a self, from: &uint) -> &'a str {
1114 self.slice_from(*from)
1118 fn slice_to_or_fail<'a>(&'a self, to: &uint) -> &'a str {
1123 fn slice_or_fail<'a>(&'a self, from: &uint, to: &uint) -> &'a str {
1124 self.slice(*from, *to)
1129 /// Any string that can be represented as a slice
1130 #[unstable = "Instead of taking this bound generically, this trait will be \
1131 replaced with one of slicing syntax, deref coercions, or \
1132 a more generic conversion trait"]
1133 pub trait Str for Sized? {
1134 /// Work with `self` as a slice.
1135 fn as_slice<'a>(&'a self) -> &'a str;
1140 fn as_slice<'a>(&'a self) -> &'a str { self }
1143 impl<'a, Sized? S> Str for &'a S where S: Str {
1145 fn as_slice(&self) -> &str { Str::as_slice(*self) }
1148 /// Return type of `StrExt::split`
1151 pub struct Split<'a, P>(CharSplits<'a, P>);
1152 delegate_iter!{pattern &'a str in Split<'a, P>}
1154 /// Return type of `StrExt::split_terminator`
1156 #[unstable = "might get removed in favour of a constructor method on Split"]
1157 pub struct SplitTerminator<'a, P>(CharSplits<'a, P>);
1158 delegate_iter!{pattern &'a str in SplitTerminator<'a, P>}
1160 /// Return type of `StrExt::splitn`
1163 pub struct SplitN<'a, P>(CharSplitsN<'a, P>);
1164 delegate_iter!{pattern forward &'a str in SplitN<'a, P>}
1166 /// Return type of `StrExt::rsplitn`
1169 pub struct RSplitN<'a, P>(CharSplitsN<'a, P>);
1170 delegate_iter!{pattern forward &'a str in RSplitN<'a, P>}
1172 /// Methods for string slices
1173 #[allow(missing_docs)]
1174 pub trait StrExt for Sized? {
1175 // NB there are no docs here are they're all located on the StrExt trait in
1176 // libcollections, not here.
1178 fn contains(&self, pat: &str) -> bool;
1179 fn contains_char<P: CharEq>(&self, pat: P) -> bool;
1180 fn chars<'a>(&'a self) -> Chars<'a>;
1181 fn bytes<'a>(&'a self) -> Bytes<'a>;
1182 fn char_indices<'a>(&'a self) -> CharIndices<'a>;
1183 fn split<'a, P: CharEq>(&'a self, pat: P) -> Split<'a, P>;
1184 fn splitn<'a, P: CharEq>(&'a self, count: uint, pat: P) -> SplitN<'a, P>;
1185 fn split_terminator<'a, P: CharEq>(&'a self, pat: P) -> SplitTerminator<'a, P>;
1186 fn rsplitn<'a, P: CharEq>(&'a self, count: uint, pat: P) -> RSplitN<'a, P>;
1187 fn match_indices<'a>(&'a self, sep: &'a str) -> MatchIndices<'a>;
1188 fn split_str<'a>(&'a self, pat: &'a str) -> SplitStr<'a>;
1189 fn lines<'a>(&'a self) -> Lines<'a>;
1190 fn lines_any<'a>(&'a self) -> LinesAny<'a>;
1191 fn char_len(&self) -> uint;
1192 fn slice<'a>(&'a self, begin: uint, end: uint) -> &'a str;
1193 fn slice_from<'a>(&'a self, begin: uint) -> &'a str;
1194 fn slice_to<'a>(&'a self, end: uint) -> &'a str;
1195 fn slice_chars<'a>(&'a self, begin: uint, end: uint) -> &'a str;
1196 unsafe fn slice_unchecked<'a>(&'a self, begin: uint, end: uint) -> &'a str;
1197 fn starts_with(&self, pat: &str) -> bool;
1198 fn ends_with(&self, pat: &str) -> bool;
1199 fn trim_matches<'a, P: CharEq>(&'a self, pat: P) -> &'a str;
1200 fn trim_left_matches<'a, P: CharEq>(&'a self, pat: P) -> &'a str;
1201 fn trim_right_matches<'a, P: CharEq>(&'a self, pat: P) -> &'a str;
1202 fn is_char_boundary(&self, index: uint) -> bool;
1203 fn char_range_at(&self, start: uint) -> CharRange;
1204 fn char_range_at_reverse(&self, start: uint) -> CharRange;
1205 fn char_at(&self, i: uint) -> char;
1206 fn char_at_reverse(&self, i: uint) -> char;
1207 fn as_bytes<'a>(&'a self) -> &'a [u8];
1208 fn find<P: CharEq>(&self, pat: P) -> Option<uint>;
1209 fn rfind<P: CharEq>(&self, pat: P) -> Option<uint>;
1210 fn find_str(&self, pat: &str) -> Option<uint>;
1211 fn slice_shift_char<'a>(&'a self) -> Option<(char, &'a str)>;
1212 fn subslice_offset(&self, inner: &str) -> uint;
1213 fn as_ptr(&self) -> *const u8;
1214 fn len(&self) -> uint;
1215 fn is_empty(&self) -> bool;
1216 fn parse<T: FromStr>(&self) -> Option<T>;
1220 fn slice_error_fail(s: &str, begin: uint, end: uint) -> ! {
1221 assert!(begin <= end);
1222 panic!("index {} and/or {} in `{}` do not lie on character boundary",
1226 impl StrExt for str {
1228 fn contains(&self, needle: &str) -> bool {
1229 self.find_str(needle).is_some()
1233 fn contains_char<P: CharEq>(&self, pat: P) -> bool {
1234 self.find(pat).is_some()
1238 fn chars(&self) -> Chars {
1239 Chars{iter: self.as_bytes().iter()}
1243 fn bytes(&self) -> Bytes {
1244 Bytes(self.as_bytes().iter().map(BytesDeref))
1248 fn char_indices(&self) -> CharIndices {
1249 CharIndices { front_offset: 0, iter: self.chars() }
1253 fn split<P: CharEq>(&self, pat: P) -> Split<P> {
1256 only_ascii: pat.only_ascii(),
1258 allow_trailing_empty: true,
1264 fn splitn<P: CharEq>(&self, count: uint, pat: P) -> SplitN<P> {
1265 SplitN(CharSplitsN {
1266 iter: self.split(pat).0,
1273 fn split_terminator<P: CharEq>(&self, pat: P) -> SplitTerminator<P> {
1274 SplitTerminator(CharSplits {
1275 allow_trailing_empty: false,
1281 fn rsplitn<P: CharEq>(&self, count: uint, pat: P) -> RSplitN<P> {
1282 RSplitN(CharSplitsN {
1283 iter: self.split(pat).0,
1290 fn match_indices<'a>(&'a self, sep: &'a str) -> MatchIndices<'a> {
1291 assert!(!sep.is_empty());
1295 searcher: Searcher::new(self.as_bytes(), sep.as_bytes())
1300 fn split_str<'a>(&'a self, sep: &'a str) -> SplitStr<'a> {
1302 it: self.match_indices(sep),
1309 fn lines(&self) -> Lines {
1310 Lines { inner: self.split_terminator('\n').0 }
1313 fn lines_any(&self) -> LinesAny {
1314 fn f(line: &str) -> &str {
1316 if l > 0 && line.as_bytes()[l - 1] == b'\r' { line.slice(0, l - 1) }
1320 let f: fn(&str) -> &str = f; // coerce to fn pointer
1321 LinesAny { inner: self.lines().map(f) }
1325 fn char_len(&self) -> uint { self.chars().count() }
1328 fn slice(&self, begin: uint, end: uint) -> &str {
1329 // is_char_boundary checks that the index is in [0, .len()]
1331 self.is_char_boundary(begin) &&
1332 self.is_char_boundary(end) {
1333 unsafe { self.slice_unchecked(begin, end) }
1335 slice_error_fail(self, begin, end)
1340 fn slice_from(&self, begin: uint) -> &str {
1341 // is_char_boundary checks that the index is in [0, .len()]
1342 if self.is_char_boundary(begin) {
1343 unsafe { self.slice_unchecked(begin, self.len()) }
1345 slice_error_fail(self, begin, self.len())
1350 fn slice_to(&self, end: uint) -> &str {
1351 // is_char_boundary checks that the index is in [0, .len()]
1352 if self.is_char_boundary(end) {
1353 unsafe { self.slice_unchecked(0, end) }
1355 slice_error_fail(self, 0, end)
1359 fn slice_chars(&self, begin: uint, end: uint) -> &str {
1360 assert!(begin <= end);
1362 let mut begin_byte = None;
1363 let mut end_byte = None;
1365 // This could be even more efficient by not decoding,
1366 // only finding the char boundaries
1367 for (idx, _) in self.char_indices() {
1368 if count == begin { begin_byte = Some(idx); }
1369 if count == end { end_byte = Some(idx); break; }
1372 if begin_byte.is_none() && count == begin { begin_byte = Some(self.len()) }
1373 if end_byte.is_none() && count == end { end_byte = Some(self.len()) }
1375 match (begin_byte, end_byte) {
1376 (None, _) => panic!("slice_chars: `begin` is beyond end of string"),
1377 (_, None) => panic!("slice_chars: `end` is beyond end of string"),
1378 (Some(a), Some(b)) => unsafe { self.slice_unchecked(a, b) }
1383 unsafe fn slice_unchecked(&self, begin: uint, end: uint) -> &str {
1384 mem::transmute(Slice {
1385 data: self.as_ptr().offset(begin as int),
1391 fn starts_with(&self, needle: &str) -> bool {
1392 let n = needle.len();
1393 self.len() >= n && needle.as_bytes() == self.as_bytes()[..n]
1397 fn ends_with(&self, needle: &str) -> bool {
1398 let (m, n) = (self.len(), needle.len());
1399 m >= n && needle.as_bytes() == self.as_bytes()[m-n..]
1403 fn trim_matches<P: CharEq>(&self, mut pat: P) -> &str {
1404 let cur = match self.find(|&mut: c: char| !pat.matches(c)) {
1406 Some(i) => unsafe { self.slice_unchecked(i, self.len()) }
1408 match cur.rfind(|&mut: c: char| !pat.matches(c)) {
1411 let right = cur.char_range_at(i).next;
1412 unsafe { cur.slice_unchecked(0, right) }
1418 fn trim_left_matches<P: CharEq>(&self, mut pat: P) -> &str {
1419 match self.find(|&mut: c: char| !pat.matches(c)) {
1421 Some(first) => unsafe { self.slice_unchecked(first, self.len()) }
1426 fn trim_right_matches<P: CharEq>(&self, mut pat: P) -> &str {
1427 match self.rfind(|&mut: c: char| !pat.matches(c)) {
1430 let next = self.char_range_at(last).next;
1431 unsafe { self.slice_unchecked(0u, next) }
1437 fn is_char_boundary(&self, index: uint) -> bool {
1438 if index == self.len() { return true; }
1439 match self.as_bytes().get(index) {
1441 Some(&b) => b < 128u8 || b >= 192u8,
1446 fn char_range_at(&self, i: uint) -> CharRange {
1447 if self.as_bytes()[i] < 128u8 {
1448 return CharRange {ch: self.as_bytes()[i] as char, next: i + 1 };
1451 // Multibyte case is a fn to allow char_range_at to inline cleanly
1452 fn multibyte_char_range_at(s: &str, i: uint) -> CharRange {
1453 let mut val = s.as_bytes()[i] as u32;
1454 let w = UTF8_CHAR_WIDTH[val as uint] as uint;
1457 val = utf8_first_byte!(val, w);
1458 val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 1]);
1459 if w > 2 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 2]); }
1460 if w > 3 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 3]); }
1462 return CharRange {ch: unsafe { mem::transmute(val) }, next: i + w};
1465 return multibyte_char_range_at(self, i);
1469 fn char_range_at_reverse(&self, start: uint) -> CharRange {
1470 let mut prev = start;
1472 prev = prev.saturating_sub(1);
1473 if self.as_bytes()[prev] < 128 {
1474 return CharRange{ch: self.as_bytes()[prev] as char, next: prev}
1477 // Multibyte case is a fn to allow char_range_at_reverse to inline cleanly
1478 fn multibyte_char_range_at_reverse(s: &str, mut i: uint) -> CharRange {
1479 // while there is a previous byte == 10......
1480 while i > 0 && s.as_bytes()[i] & !CONT_MASK == TAG_CONT_U8 {
1484 let mut val = s.as_bytes()[i] as u32;
1485 let w = UTF8_CHAR_WIDTH[val as uint] as uint;
1488 val = utf8_first_byte!(val, w);
1489 val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 1]);
1490 if w > 2 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 2]); }
1491 if w > 3 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 3]); }
1493 return CharRange {ch: unsafe { mem::transmute(val) }, next: i};
1496 return multibyte_char_range_at_reverse(self, prev);
1500 fn char_at(&self, i: uint) -> char {
1501 self.char_range_at(i).ch
1505 fn char_at_reverse(&self, i: uint) -> char {
1506 self.char_range_at_reverse(i).ch
1510 fn as_bytes(&self) -> &[u8] {
1511 unsafe { mem::transmute(self) }
1514 fn find<P: CharEq>(&self, mut pat: P) -> Option<uint> {
1515 if pat.only_ascii() {
1516 self.bytes().position(|b| pat.matches(b as char))
1518 for (index, c) in self.char_indices() {
1519 if pat.matches(c) { return Some(index); }
1525 fn rfind<P: CharEq>(&self, mut pat: P) -> Option<uint> {
1526 if pat.only_ascii() {
1527 self.bytes().rposition(|b| pat.matches(b as char))
1529 for (index, c) in self.char_indices().rev() {
1530 if pat.matches(c) { return Some(index); }
1536 fn find_str(&self, needle: &str) -> Option<uint> {
1537 if needle.is_empty() {
1540 self.match_indices(needle)
1542 .map(|(start, _end)| start)
1547 fn slice_shift_char(&self) -> Option<(char, &str)> {
1548 if self.is_empty() {
1551 let CharRange {ch, next} = self.char_range_at(0u);
1552 let next_s = unsafe { self.slice_unchecked(next, self.len()) };
1557 fn subslice_offset(&self, inner: &str) -> uint {
1558 let a_start = self.as_ptr() as uint;
1559 let a_end = a_start + self.len();
1560 let b_start = inner.as_ptr() as uint;
1561 let b_end = b_start + inner.len();
1563 assert!(a_start <= b_start);
1564 assert!(b_end <= a_end);
1569 fn as_ptr(&self) -> *const u8 {
1574 fn len(&self) -> uint { self.repr().len }
1577 fn is_empty(&self) -> bool { self.len() == 0 }
1580 fn parse<T: FromStr>(&self) -> Option<T> { FromStr::from_str(self) }
1584 impl<'a> Default for &'a str {
1586 fn default() -> &'a str { "" }
1589 impl<'a> Iterator for Lines<'a> {
1590 type Item = &'a str;
1593 fn next(&mut self) -> Option<&'a str> { self.inner.next() }
1595 fn size_hint(&self) -> (uint, Option<uint>) { self.inner.size_hint() }
1597 impl<'a> DoubleEndedIterator for Lines<'a> {
1599 fn next_back(&mut self) -> Option<&'a str> { self.inner.next_back() }
1601 impl<'a> Iterator for LinesAny<'a> {
1602 type Item = &'a str;
1605 fn next(&mut self) -> Option<&'a str> { self.inner.next() }
1607 fn size_hint(&self) -> (uint, Option<uint>) { self.inner.size_hint() }
1609 impl<'a> DoubleEndedIterator for LinesAny<'a> {
1611 fn next_back(&mut self) -> Option<&'a str> { self.inner.next_back() }