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 : $ti:ty) => {
39 delegate_iter!{$te : $ti}
40 impl<'a> ExactSizeIterator for $ti {
42 fn len(&self) -> uint {
47 ($te:ty : $ti:ty) => {
49 impl<'a> Iterator for $ti {
53 fn next(&mut self) -> Option<$te> {
57 fn size_hint(&self) -> (uint, Option<uint>) {
62 impl<'a> DoubleEndedIterator for $ti {
64 fn next_back(&mut self) -> Option<$te> {
69 (pattern $te:ty : $ti:ty) => {
71 impl<'a, P: CharEq> Iterator for $ti {
75 fn next(&mut self) -> Option<$te> {
79 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 : $ti:ty) => {
93 impl<'a, P: CharEq> Iterator for $ti {
97 fn next(&mut self) -> Option<$te> {
101 fn size_hint(&self) -> (uint, Option<uint>) {
108 /// A trait to abstract the idea of creating a new instance of a type from a
110 // FIXME(#17307): there should be an `E` associated type for a `Result` return
111 #[unstable = "will return a Result once associated types are working"]
113 /// Parses a string `s` to return an optional value of this type. If the
114 /// string is ill-formatted, the None is returned.
115 fn from_str(s: &str) -> Option<Self>;
118 impl FromStr for bool {
119 /// Parse a `bool` from a string.
121 /// Yields an `Option<bool>`, because `s` may or may not actually be parseable.
126 /// assert_eq!("true".parse(), Some(true));
127 /// assert_eq!("false".parse(), Some(false));
128 /// assert_eq!("not even a boolean".parse::<bool>(), None);
131 fn from_str(s: &str) -> Option<bool> {
133 "true" => Some(true),
134 "false" => Some(false),
141 Section: Creating a string
144 /// Errors which can occur when attempting to interpret a byte slice as a `str`.
145 #[derive(Copy, Eq, PartialEq, Clone, Show)]
146 #[unstable = "error enumeration recently added and definitions may be refined"]
148 /// An invalid byte was detected at the byte offset given.
150 /// The offset is guaranteed to be in bounds of the slice in question, and
151 /// the byte at the specified offset was the first invalid byte in the
152 /// sequence detected.
155 /// The byte slice was invalid because more bytes were needed but no more
156 /// bytes were available.
160 /// Converts a slice of bytes to a string slice without performing any
163 /// Once the slice has been validated as utf-8, it is transmuted in-place and
164 /// returned as a '&str' instead of a '&[u8]'
168 /// Returns `Err` if the slice is not utf-8 with a description as to why the
169 /// provided slice is not utf-8.
171 pub fn from_utf8(v: &[u8]) -> Result<&str, Utf8Error> {
172 try!(run_utf8_validation_iterator(&mut v.iter()));
173 Ok(unsafe { from_utf8_unchecked(v) })
176 /// Converts a slice of bytes to a string slice without checking
177 /// that the string contains valid UTF-8.
179 pub unsafe fn from_utf8_unchecked<'a>(v: &'a [u8]) -> &'a str {
183 /// Constructs a static string slice from a given raw pointer.
185 /// This function will read memory starting at `s` until it finds a 0, and then
186 /// transmute the memory up to that point as a string slice, returning the
187 /// corresponding `&'static str` value.
189 /// This function is unsafe because the caller must ensure the C string itself
190 /// has the static lifetime and that the memory `s` is valid up to and including
191 /// the first null byte.
195 /// This function will panic if the string pointed to by `s` is not valid UTF-8.
196 #[deprecated = "use std::ffi::c_str_to_bytes + str::from_utf8"]
197 pub unsafe fn from_c_str(s: *const i8) -> &'static str {
198 let s = s as *const u8;
200 while *s.offset(len as int) != 0 {
203 let v: &'static [u8] = ::mem::transmute(Slice { data: s, len: len });
204 from_utf8(v).ok().expect("from_c_str passed invalid utf-8 data")
207 /// Something that can be used to compare against a character
208 #[unstable = "definition may change as pattern-related methods are stabilized"]
210 /// Determine if the splitter should split at the given character
211 fn matches(&mut self, char) -> bool;
212 /// Indicate if this is only concerned about ASCII characters,
213 /// which can allow for a faster implementation.
214 fn only_ascii(&self) -> bool;
217 impl CharEq for char {
219 fn matches(&mut self, c: char) -> bool { *self == c }
222 fn only_ascii(&self) -> bool { (*self as uint) < 128 }
225 impl<F> CharEq for F where F: FnMut(char) -> bool {
227 fn matches(&mut self, c: char) -> bool { (*self)(c) }
230 fn only_ascii(&self) -> bool { false }
233 impl<'a> CharEq for &'a [char] {
235 fn matches(&mut self, c: char) -> bool {
236 self.iter().any(|&m| { let mut m = m; m.matches(c) })
240 fn only_ascii(&self) -> bool {
241 self.iter().all(|m| m.only_ascii())
249 /// Iterator for the char (representing *Unicode Scalar Values*) of a string
251 /// Created with the method `.chars()`.
252 #[derive(Clone, Copy)]
254 pub struct Chars<'a> {
255 iter: slice::Iter<'a, u8>
258 // Return the initial codepoint accumulator for the first byte.
259 // The first byte is special, only want bottom 5 bits for width 2, 4 bits
260 // for width 3, and 3 bits for width 4
261 macro_rules! utf8_first_byte {
262 ($byte:expr, $width:expr) => (($byte & (0x7F >> $width)) as u32)
265 // return the value of $ch updated with continuation byte $byte
266 macro_rules! utf8_acc_cont_byte {
267 ($ch:expr, $byte:expr) => (($ch << 6) | ($byte & CONT_MASK) as u32)
270 macro_rules! utf8_is_cont_byte {
271 ($byte:expr) => (($byte & !CONT_MASK) == TAG_CONT_U8)
275 fn unwrap_or_0(opt: Option<&u8>) -> u8 {
283 impl<'a> Iterator for Chars<'a> {
287 fn next(&mut self) -> Option<char> {
288 // Decode UTF-8, using the valid UTF-8 invariant
289 let x = match self.iter.next() {
291 Some(&next_byte) if next_byte < 128 => return Some(next_byte as char),
292 Some(&next_byte) => next_byte,
295 // Multibyte case follows
296 // Decode from a byte combination out of: [[[x y] z] w]
297 // NOTE: Performance is sensitive to the exact formulation here
298 let init = utf8_first_byte!(x, 2);
299 let y = unwrap_or_0(self.iter.next());
300 let mut ch = utf8_acc_cont_byte!(init, y);
303 // 5th bit in 0xE0 .. 0xEF is always clear, so `init` is still valid
304 let z = unwrap_or_0(self.iter.next());
305 let y_z = utf8_acc_cont_byte!((y & CONT_MASK) as u32, z);
306 ch = init << 12 | y_z;
309 // use only the lower 3 bits of `init`
310 let w = unwrap_or_0(self.iter.next());
311 ch = (init & 7) << 18 | utf8_acc_cont_byte!(y_z, w);
315 // str invariant says `ch` is a valid Unicode Scalar Value
317 Some(mem::transmute(ch))
322 fn size_hint(&self) -> (uint, Option<uint>) {
323 let (len, _) = self.iter.size_hint();
324 (len.saturating_add(3) / 4, Some(len))
329 impl<'a> DoubleEndedIterator for Chars<'a> {
331 fn next_back(&mut self) -> Option<char> {
332 let w = match self.iter.next_back() {
334 Some(&back_byte) if back_byte < 128 => return Some(back_byte as char),
335 Some(&back_byte) => back_byte,
338 // Multibyte case follows
339 // Decode from a byte combination out of: [x [y [z w]]]
341 let z = unwrap_or_0(self.iter.next_back());
342 ch = utf8_first_byte!(z, 2);
343 if utf8_is_cont_byte!(z) {
344 let y = unwrap_or_0(self.iter.next_back());
345 ch = utf8_first_byte!(y, 3);
346 if utf8_is_cont_byte!(y) {
347 let x = unwrap_or_0(self.iter.next_back());
348 ch = utf8_first_byte!(x, 4);
349 ch = utf8_acc_cont_byte!(ch, y);
351 ch = utf8_acc_cont_byte!(ch, z);
353 ch = utf8_acc_cont_byte!(ch, w);
355 // str invariant says `ch` is a valid Unicode Scalar Value
357 Some(mem::transmute(ch))
362 /// External iterator for a string's characters and their byte offsets.
363 /// Use with the `std::iter` module.
366 pub struct CharIndices<'a> {
372 impl<'a> Iterator for CharIndices<'a> {
373 type Item = (uint, char);
376 fn next(&mut self) -> Option<(uint, char)> {
377 let (pre_len, _) = self.iter.iter.size_hint();
378 match self.iter.next() {
381 let index = self.front_offset;
382 let (len, _) = self.iter.iter.size_hint();
383 self.front_offset += pre_len - len;
390 fn size_hint(&self) -> (uint, Option<uint>) {
391 self.iter.size_hint()
396 impl<'a> DoubleEndedIterator for CharIndices<'a> {
398 fn next_back(&mut self) -> Option<(uint, char)> {
399 match self.iter.next_back() {
402 let (len, _) = self.iter.iter.size_hint();
403 let index = self.front_offset + len;
410 /// External iterator for a string's bytes.
411 /// Use with the `std::iter` module.
413 /// Created with `StrExt::bytes`
416 pub struct Bytes<'a>(Map<&'a u8, u8, slice::Iter<'a, u8>, BytesDeref>);
417 delegate_iter!{exact u8 : Bytes<'a>}
419 /// A temporary fn new type that ensures that the `Bytes` iterator
421 #[derive(Copy, Clone)]
424 impl<'a> Fn(&'a u8) -> u8 for BytesDeref {
426 extern "rust-call" fn call(&self, (ptr,): (&'a u8,)) -> u8 {
431 /// An iterator over the substrings of a string, separated by `sep`.
433 struct CharSplits<'a, Sep> {
434 /// The slice remaining to be iterated
437 /// Whether an empty string at the end is allowed
438 allow_trailing_empty: bool,
443 /// An iterator over the substrings of a string, separated by `sep`,
444 /// splitting at most `count` times.
446 struct CharSplitsN<'a, Sep> {
447 iter: CharSplits<'a, Sep>,
448 /// The number of splits remaining
453 /// An iterator over the lines of a string, separated by `\n`.
455 pub struct Lines<'a> {
456 inner: CharSplits<'a, char>,
459 /// An iterator over the lines of a string, separated by either `\n` or (`\r\n`).
461 pub struct LinesAny<'a> {
462 inner: Map<&'a str, &'a str, Lines<'a>, fn(&str) -> &str>,
465 impl<'a, Sep> CharSplits<'a, Sep> {
467 fn get_end(&mut self) -> Option<&'a str> {
468 if !self.finished && (self.allow_trailing_empty || self.string.len() > 0) {
469 self.finished = true;
478 impl<'a, Sep: CharEq> Iterator for CharSplits<'a, Sep> {
482 fn next(&mut self) -> Option<&'a str> {
483 if self.finished { return None }
485 let mut next_split = None;
487 for (idx, byte) in self.string.bytes().enumerate() {
488 if self.sep.matches(byte as char) && byte < 128u8 {
489 next_split = Some((idx, idx + 1));
494 for (idx, ch) in self.string.char_indices() {
495 if self.sep.matches(ch) {
496 next_split = Some((idx, self.string.char_range_at(idx).next));
502 Some((a, b)) => unsafe {
503 let elt = self.string.slice_unchecked(0, a);
504 self.string = self.string.slice_unchecked(b, self.string.len());
507 None => self.get_end(),
513 impl<'a, Sep: CharEq> DoubleEndedIterator for CharSplits<'a, Sep> {
515 fn next_back(&mut self) -> Option<&'a str> {
516 if self.finished { return None }
518 if !self.allow_trailing_empty {
519 self.allow_trailing_empty = true;
520 match self.next_back() {
521 Some(elt) if !elt.is_empty() => return Some(elt),
522 _ => if self.finished { return None }
525 let len = self.string.len();
526 let mut next_split = None;
529 for (idx, byte) in self.string.bytes().enumerate().rev() {
530 if self.sep.matches(byte as char) && byte < 128u8 {
531 next_split = Some((idx, idx + 1));
536 for (idx, ch) in self.string.char_indices().rev() {
537 if self.sep.matches(ch) {
538 next_split = Some((idx, self.string.char_range_at(idx).next));
544 Some((a, b)) => unsafe {
545 let elt = self.string.slice_unchecked(b, len);
546 self.string = self.string.slice_unchecked(0, a);
549 None => { self.finished = true; Some(self.string) }
555 impl<'a, Sep: CharEq> Iterator for CharSplitsN<'a, Sep> {
559 fn next(&mut self) -> Option<&'a str> {
562 if self.invert { self.iter.next_back() } else { self.iter.next() }
569 /// The internal state of an iterator that searches for matches of a substring
570 /// within a larger string using naive search
572 struct NaiveSearcher {
577 fn new() -> NaiveSearcher {
578 NaiveSearcher { position: 0 }
581 fn next(&mut self, haystack: &[u8], needle: &[u8]) -> Option<(uint, uint)> {
582 while self.position + needle.len() <= haystack.len() {
583 if &haystack[self.position .. self.position + needle.len()] == needle {
584 let match_pos = self.position;
585 self.position += needle.len(); // add 1 for all matches
586 return Some((match_pos, match_pos + needle.len()));
595 /// The internal state of an iterator that searches for matches of a substring
596 /// within a larger string using two-way search
598 struct TwoWaySearcher {
610 This is the Two-Way search algorithm, which was introduced in the paper:
611 Crochemore, M., Perrin, D., 1991, Two-way string-matching, Journal of the ACM 38(3):651-675.
613 Here's some background information.
615 A *word* is a string of symbols. The *length* of a word should be a familiar
616 notion, and here we denote it for any word x by |x|.
617 (We also allow for the possibility of the *empty word*, a word of length zero).
619 If x is any non-empty word, then an integer p with 0 < p <= |x| is said to be a
620 *period* for x iff for all i with 0 <= i <= |x| - p - 1, we have x[i] == x[i+p].
621 For example, both 1 and 2 are periods for the string "aa". As another example,
622 the only period of the string "abcd" is 4.
624 We denote by period(x) the *smallest* period of x (provided that x is non-empty).
625 This is always well-defined since every non-empty word x has at least one period,
626 |x|. We sometimes call this *the period* of x.
628 If u, v and x are words such that x = uv, where uv is the concatenation of u and
629 v, then we say that (u, v) is a *factorization* of x.
631 Let (u, v) be a factorization for a word x. Then if w is a non-empty word such
632 that both of the following hold
634 - either w is a suffix of u or u is a suffix of w
635 - either w is a prefix of v or v is a prefix of w
637 then w is said to be a *repetition* for the factorization (u, v).
639 Just to unpack this, there are four possibilities here. Let w = "abc". Then we
642 - w is a suffix of u and w is a prefix of v. ex: ("lolabc", "abcde")
643 - w is a suffix of u and v is a prefix of w. ex: ("lolabc", "ab")
644 - u is a suffix of w and w is a prefix of v. ex: ("bc", "abchi")
645 - u is a suffix of w and v is a prefix of w. ex: ("bc", "a")
647 Note that the word vu is a repetition for any factorization (u,v) of x = uv,
648 so every factorization has at least one repetition.
650 If x is a string and (u, v) is a factorization for x, then a *local period* for
651 (u, v) is an integer r such that there is some word w such that |w| = r and w is
652 a repetition for (u, v).
654 We denote by local_period(u, v) the smallest local period of (u, v). We sometimes
655 call this *the local period* of (u, v). Provided that x = uv is non-empty, this
656 is well-defined (because each non-empty word has at least one factorization, as
659 It can be proven that the following is an equivalent definition of a local period
660 for a factorization (u, v): any positive integer r such that x[i] == x[i+r] for
661 all i such that |u| - r <= i <= |u| - 1 and such that both x[i] and x[i+r] are
662 defined. (i.e. i > 0 and i + r < |x|).
664 Using the above reformulation, it is easy to prove that
666 1 <= local_period(u, v) <= period(uv)
668 A factorization (u, v) of x such that local_period(u,v) = period(x) is called a
669 *critical factorization*.
671 The algorithm hinges on the following theorem, which is stated without proof:
673 **Critical Factorization Theorem** Any word x has at least one critical
674 factorization (u, v) such that |u| < period(x).
676 The purpose of maximal_suffix is to find such a critical factorization.
679 impl TwoWaySearcher {
680 fn new(needle: &[u8]) -> TwoWaySearcher {
681 let (crit_pos1, period1) = TwoWaySearcher::maximal_suffix(needle, false);
682 let (crit_pos2, period2) = TwoWaySearcher::maximal_suffix(needle, true);
686 if crit_pos1 > crit_pos2 {
687 crit_pos = crit_pos1;
690 crit_pos = crit_pos2;
694 // This isn't in the original algorithm, as far as I'm aware.
695 let byteset = needle.iter()
696 .fold(0, |a, &b| (1 << ((b & 0x3f) as uint)) | a);
698 // A particularly readable explanation of what's going on here can be found
699 // in Crochemore and Rytter's book "Text Algorithms", ch 13. Specifically
700 // see the code for "Algorithm CP" on p. 323.
702 // What's going on is we have some critical factorization (u, v) of the
703 // needle, and we want to determine whether u is a suffix of
704 // &v[..period]. If it is, we use "Algorithm CP1". Otherwise we use
705 // "Algorithm CP2", which is optimized for when the period of the needle
707 if &needle[..crit_pos] == &needle[period.. period + crit_pos] {
719 period: cmp::max(crit_pos, needle.len() - crit_pos) + 1,
723 memory: uint::MAX // Dummy value to signify that the period is long
728 // One of the main ideas of Two-Way is that we factorize the needle into
729 // two halves, (u, v), and begin trying to find v in the haystack by scanning
730 // left to right. If v matches, we try to match u by scanning right to left.
731 // How far we can jump when we encounter a mismatch is all based on the fact
732 // that (u, v) is a critical factorization for the needle.
734 fn next(&mut self, haystack: &[u8], needle: &[u8], long_period: bool) -> Option<(uint, uint)> {
736 // Check that we have room to search in
737 if self.position + needle.len() > haystack.len() {
741 // Quickly skip by large portions unrelated to our substring
743 ((haystack[self.position + needle.len() - 1] & 0x3f)
745 self.position += needle.len();
752 // See if the right part of the needle matches
753 let start = if long_period { self.crit_pos }
754 else { cmp::max(self.crit_pos, self.memory) };
755 for i in range(start, needle.len()) {
756 if needle[i] != haystack[self.position + i] {
757 self.position += i - self.crit_pos + 1;
765 // See if the left part of the needle matches
766 let start = if long_period { 0 } else { self.memory };
767 for i in range(start, self.crit_pos).rev() {
768 if needle[i] != haystack[self.position + i] {
769 self.position += self.period;
771 self.memory = needle.len() - self.period;
777 // We have found a match!
778 let match_pos = self.position;
779 self.position += needle.len(); // add self.period for all matches
781 self.memory = 0; // set to needle.len() - self.period for all matches
783 return Some((match_pos, match_pos + needle.len()));
787 // Computes a critical factorization (u, v) of `arr`.
788 // Specifically, returns (i, p), where i is the starting index of v in some
789 // critical factorization (u, v) and p = period(v)
791 fn maximal_suffix(arr: &[u8], reversed: bool) -> (uint, uint) {
792 let mut left = -1; // Corresponds to i in the paper
793 let mut right = 0; // Corresponds to j in the paper
794 let mut offset = 1; // Corresponds to k in the paper
795 let mut period = 1; // Corresponds to p in the paper
797 while right + offset < arr.len() {
801 a = arr[left + offset];
802 b = arr[right + offset];
804 a = arr[right + offset];
805 b = arr[left + offset];
808 // Suffix is smaller, period is entire prefix so far.
811 period = right - left;
813 // Advance through repetition of the current period.
814 if offset == period {
821 // Suffix is larger, start over from current location.
832 /// The internal state of an iterator that searches for matches of a substring
833 /// within a larger string using a dynamically chosen search algorithm
836 Naive(NaiveSearcher),
837 TwoWay(TwoWaySearcher),
838 TwoWayLong(TwoWaySearcher)
842 fn new(haystack: &[u8], needle: &[u8]) -> Searcher {
844 // FIXME(#16715): This unsigned integer addition will probably not
845 // overflow because that would mean that the memory almost solely
846 // consists of the needle. Needs #16715 to be formally fixed.
847 if needle.len() + 20 > haystack.len() {
848 Naive(NaiveSearcher::new())
850 let searcher = TwoWaySearcher::new(needle);
851 if searcher.memory == uint::MAX { // If the period is long
860 /// An iterator over the start and end indices of the matches of a
861 /// substring within a larger string
863 #[unstable = "type may be removed"]
864 pub struct MatchIndices<'a> {
871 /// An iterator over the substrings of a string separated by a given
874 #[unstable = "type may be removed"]
875 pub struct SplitStr<'a> {
876 it: MatchIndices<'a>,
882 impl<'a> Iterator for MatchIndices<'a> {
883 type Item = (uint, uint);
886 fn next(&mut self) -> Option<(uint, uint)> {
887 match self.searcher {
888 Naive(ref mut searcher)
889 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes()),
890 TwoWay(ref mut searcher)
891 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes(), false),
892 TwoWayLong(ref mut searcher)
893 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes(), true)
899 impl<'a> Iterator for SplitStr<'a> {
903 fn next(&mut self) -> Option<&'a str> {
904 if self.finished { return None; }
906 match self.it.next() {
907 Some((from, to)) => {
908 let ret = Some(self.it.haystack.slice(self.last_end, from));
913 self.finished = true;
914 Some(self.it.haystack.slice(self.last_end, self.it.haystack.len()))
922 Section: Comparing strings
925 // share the implementation of the lang-item vs. non-lang-item
927 /// NOTE: This function is (ab)used in rustc::middle::trans::_match
928 /// to compare &[u8] byte slices that are not necessarily valid UTF-8.
930 fn eq_slice_(a: &str, b: &str) -> bool {
931 #[allow(improper_ctypes)]
932 extern { fn memcmp(s1: *const i8, s2: *const i8, n: uint) -> i32; }
933 a.len() == b.len() && unsafe {
934 memcmp(a.as_ptr() as *const i8,
935 b.as_ptr() as *const i8,
940 /// Bytewise slice equality
941 /// NOTE: This function is (ab)used in rustc::middle::trans::_match
942 /// to compare &[u8] byte slices that are not necessarily valid UTF-8.
945 fn eq_slice(a: &str, b: &str) -> bool {
953 /// Walk through `iter` checking that it's a valid UTF-8 sequence,
954 /// returning `true` in that case, or, if it is invalid, `false` with
955 /// `iter` reset such that it is pointing at the first byte in the
956 /// invalid sequence.
958 fn run_utf8_validation_iterator(iter: &mut slice::Iter<u8>)
959 -> Result<(), Utf8Error> {
960 let whole = iter.as_slice();
962 // save the current thing we're pointing at.
965 // restore the iterator we had at the start of this codepoint.
966 macro_rules! err { () => {{
968 return Err(Utf8Error::InvalidByte(whole.len() - iter.as_slice().len()))
971 macro_rules! next { () => {
974 // we needed data, but there was none: error!
975 None => return Err(Utf8Error::TooShort),
979 let first = match iter.next() {
981 // we're at the end of the iterator and a codepoint
982 // boundary at the same time, so this string is valid.
983 None => return Ok(())
986 // ASCII characters are always valid, so only large
987 // bytes need more examination.
989 let w = UTF8_CHAR_WIDTH[first as uint] as uint;
990 let second = next!();
991 // 2-byte encoding is for codepoints \u{0080} to \u{07ff}
992 // first C2 80 last DF BF
993 // 3-byte encoding is for codepoints \u{0800} to \u{ffff}
994 // first E0 A0 80 last EF BF BF
995 // excluding surrogates codepoints \u{d800} to \u{dfff}
996 // ED A0 80 to ED BF BF
997 // 4-byte encoding is for codepoints \u{1000}0 to \u{10ff}ff
998 // first F0 90 80 80 last F4 8F BF BF
1000 // Use the UTF-8 syntax from the RFC
1002 // https://tools.ietf.org/html/rfc3629
1004 // UTF8-2 = %xC2-DF UTF8-tail
1005 // UTF8-3 = %xE0 %xA0-BF UTF8-tail / %xE1-EC 2( UTF8-tail ) /
1006 // %xED %x80-9F UTF8-tail / %xEE-EF 2( UTF8-tail )
1007 // UTF8-4 = %xF0 %x90-BF 2( UTF8-tail ) / %xF1-F3 3( UTF8-tail ) /
1008 // %xF4 %x80-8F 2( UTF8-tail )
1010 2 => if second & !CONT_MASK != TAG_CONT_U8 {err!()},
1012 match (first, second, next!() & !CONT_MASK) {
1013 (0xE0 , 0xA0 ... 0xBF, TAG_CONT_U8) |
1014 (0xE1 ... 0xEC, 0x80 ... 0xBF, TAG_CONT_U8) |
1015 (0xED , 0x80 ... 0x9F, TAG_CONT_U8) |
1016 (0xEE ... 0xEF, 0x80 ... 0xBF, TAG_CONT_U8) => {}
1021 match (first, second, next!() & !CONT_MASK, next!() & !CONT_MASK) {
1022 (0xF0 , 0x90 ... 0xBF, TAG_CONT_U8, TAG_CONT_U8) |
1023 (0xF1 ... 0xF3, 0x80 ... 0xBF, TAG_CONT_U8, TAG_CONT_U8) |
1024 (0xF4 , 0x80 ... 0x8F, TAG_CONT_U8, TAG_CONT_U8) => {}
1034 // https://tools.ietf.org/html/rfc3629
1035 static UTF8_CHAR_WIDTH: [u8; 256] = [
1036 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1037 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x1F
1038 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1039 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x3F
1040 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1041 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x5F
1042 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1043 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x7F
1044 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
1045 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0x9F
1046 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
1047 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0xBF
1048 0,0,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
1049 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, // 0xDF
1050 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, // 0xEF
1051 4,4,4,4,4,0,0,0,0,0,0,0,0,0,0,0, // 0xFF
1054 /// Struct that contains a `char` and the index of the first byte of
1055 /// the next `char` in a string. This can be used as a data structure
1056 /// for iterating over the UTF-8 bytes of a string.
1058 #[unstable = "naming is uncertain with container conventions"]
1059 pub struct CharRange {
1062 /// Index of the first byte of the next `char`
1066 /// Mask of the value bits of a continuation byte
1067 const CONT_MASK: u8 = 0b0011_1111u8;
1068 /// Value of the tag bits (tag mask is !CONT_MASK) of a continuation byte
1069 const TAG_CONT_U8: u8 = 0b1000_0000u8;
1072 Section: Trait implementations
1076 use cmp::{Ordering, Ord, PartialEq, PartialOrd, Eq};
1077 use cmp::Ordering::{Less, Equal, Greater};
1078 use iter::IteratorExt;
1080 use option::Option::Some;
1082 use str::{StrExt, eq_slice};
1087 fn cmp(&self, other: &str) -> Ordering {
1088 for (s_b, o_b) in self.bytes().zip(other.bytes()) {
1089 match s_b.cmp(&o_b) {
1090 Greater => return Greater,
1091 Less => return Less,
1096 self.len().cmp(&other.len())
1101 impl PartialEq for str {
1103 fn eq(&self, other: &str) -> bool {
1104 eq_slice(self, other)
1107 fn ne(&self, other: &str) -> bool { !(*self).eq(other) }
1114 impl PartialOrd for str {
1116 fn partial_cmp(&self, other: &str) -> Option<Ordering> {
1117 Some(self.cmp(other))
1121 impl ops::Index<ops::Range<uint>> for str {
1124 fn index(&self, index: &ops::Range<uint>) -> &str {
1125 self.slice(index.start, index.end)
1128 impl ops::Index<ops::RangeTo<uint>> for str {
1131 fn index(&self, index: &ops::RangeTo<uint>) -> &str {
1132 self.slice_to(index.end)
1135 impl ops::Index<ops::RangeFrom<uint>> for str {
1138 fn index(&self, index: &ops::RangeFrom<uint>) -> &str {
1139 self.slice_from(index.start)
1142 impl ops::Index<ops::FullRange> for str {
1145 fn index(&self, _index: &ops::FullRange) -> &str {
1151 /// Any string that can be represented as a slice
1152 #[unstable = "Instead of taking this bound generically, this trait will be \
1153 replaced with one of slicing syntax, deref coercions, or \
1154 a more generic conversion trait"]
1156 /// Work with `self` as a slice.
1157 fn as_slice<'a>(&'a self) -> &'a str;
1162 fn as_slice<'a>(&'a self) -> &'a str { self }
1165 impl<'a, S: ?Sized> Str for &'a S where S: Str {
1167 fn as_slice(&self) -> &str { Str::as_slice(*self) }
1170 /// Return type of `StrExt::split`
1173 pub struct Split<'a, P>(CharSplits<'a, P>);
1174 delegate_iter!{pattern &'a str : Split<'a, P>}
1176 /// Return type of `StrExt::split_terminator`
1178 #[unstable = "might get removed in favour of a constructor method on Split"]
1179 pub struct SplitTerminator<'a, P>(CharSplits<'a, P>);
1180 delegate_iter!{pattern &'a str : SplitTerminator<'a, P>}
1182 /// Return type of `StrExt::splitn`
1185 pub struct SplitN<'a, P>(CharSplitsN<'a, P>);
1186 delegate_iter!{pattern forward &'a str : SplitN<'a, P>}
1188 /// Return type of `StrExt::rsplitn`
1191 pub struct RSplitN<'a, P>(CharSplitsN<'a, P>);
1192 delegate_iter!{pattern forward &'a str : RSplitN<'a, P>}
1194 /// Methods for string slices
1195 #[allow(missing_docs)]
1197 // NB there are no docs here are they're all located on the StrExt trait in
1198 // libcollections, not here.
1200 fn contains(&self, pat: &str) -> bool;
1201 fn contains_char<P: CharEq>(&self, pat: P) -> bool;
1202 fn chars<'a>(&'a self) -> Chars<'a>;
1203 fn bytes<'a>(&'a self) -> Bytes<'a>;
1204 fn char_indices<'a>(&'a self) -> CharIndices<'a>;
1205 fn split<'a, P: CharEq>(&'a self, pat: P) -> Split<'a, P>;
1206 fn splitn<'a, P: CharEq>(&'a self, count: uint, pat: P) -> SplitN<'a, P>;
1207 fn split_terminator<'a, P: CharEq>(&'a self, pat: P) -> SplitTerminator<'a, P>;
1208 fn rsplitn<'a, P: CharEq>(&'a self, count: uint, pat: P) -> RSplitN<'a, P>;
1209 fn match_indices<'a>(&'a self, sep: &'a str) -> MatchIndices<'a>;
1210 fn split_str<'a>(&'a self, pat: &'a str) -> SplitStr<'a>;
1211 fn lines<'a>(&'a self) -> Lines<'a>;
1212 fn lines_any<'a>(&'a self) -> LinesAny<'a>;
1213 fn char_len(&self) -> uint;
1214 fn slice<'a>(&'a self, begin: uint, end: uint) -> &'a str;
1215 fn slice_from<'a>(&'a self, begin: uint) -> &'a str;
1216 fn slice_to<'a>(&'a self, end: uint) -> &'a str;
1217 fn slice_chars<'a>(&'a self, begin: uint, end: uint) -> &'a str;
1218 unsafe fn slice_unchecked<'a>(&'a self, begin: uint, end: uint) -> &'a str;
1219 fn starts_with(&self, pat: &str) -> bool;
1220 fn ends_with(&self, pat: &str) -> bool;
1221 fn trim_matches<'a, P: CharEq>(&'a self, pat: P) -> &'a str;
1222 fn trim_left_matches<'a, P: CharEq>(&'a self, pat: P) -> &'a str;
1223 fn trim_right_matches<'a, P: CharEq>(&'a self, pat: P) -> &'a str;
1224 fn is_char_boundary(&self, index: uint) -> bool;
1225 fn char_range_at(&self, start: uint) -> CharRange;
1226 fn char_range_at_reverse(&self, start: uint) -> CharRange;
1227 fn char_at(&self, i: uint) -> char;
1228 fn char_at_reverse(&self, i: uint) -> char;
1229 fn as_bytes<'a>(&'a self) -> &'a [u8];
1230 fn find<P: CharEq>(&self, pat: P) -> Option<uint>;
1231 fn rfind<P: CharEq>(&self, pat: P) -> Option<uint>;
1232 fn find_str(&self, pat: &str) -> Option<uint>;
1233 fn slice_shift_char<'a>(&'a self) -> Option<(char, &'a str)>;
1234 fn subslice_offset(&self, inner: &str) -> uint;
1235 fn as_ptr(&self) -> *const u8;
1236 fn len(&self) -> uint;
1237 fn is_empty(&self) -> bool;
1238 fn parse<T: FromStr>(&self) -> Option<T>;
1242 fn slice_error_fail(s: &str, begin: uint, end: uint) -> ! {
1243 assert!(begin <= end);
1244 panic!("index {} and/or {} in `{}` do not lie on character boundary",
1248 impl StrExt for str {
1250 fn contains(&self, needle: &str) -> bool {
1251 self.find_str(needle).is_some()
1255 fn contains_char<P: CharEq>(&self, pat: P) -> bool {
1256 self.find(pat).is_some()
1260 fn chars(&self) -> Chars {
1261 Chars{iter: self.as_bytes().iter()}
1265 fn bytes(&self) -> Bytes {
1266 Bytes(self.as_bytes().iter().map(BytesDeref))
1270 fn char_indices(&self) -> CharIndices {
1271 CharIndices { front_offset: 0, iter: self.chars() }
1275 fn split<P: CharEq>(&self, pat: P) -> Split<P> {
1278 only_ascii: pat.only_ascii(),
1280 allow_trailing_empty: true,
1286 fn splitn<P: CharEq>(&self, count: uint, pat: P) -> SplitN<P> {
1287 SplitN(CharSplitsN {
1288 iter: self.split(pat).0,
1295 fn split_terminator<P: CharEq>(&self, pat: P) -> SplitTerminator<P> {
1296 SplitTerminator(CharSplits {
1297 allow_trailing_empty: false,
1303 fn rsplitn<P: CharEq>(&self, count: uint, pat: P) -> RSplitN<P> {
1304 RSplitN(CharSplitsN {
1305 iter: self.split(pat).0,
1312 fn match_indices<'a>(&'a self, sep: &'a str) -> MatchIndices<'a> {
1313 assert!(!sep.is_empty());
1317 searcher: Searcher::new(self.as_bytes(), sep.as_bytes())
1322 fn split_str<'a>(&'a self, sep: &'a str) -> SplitStr<'a> {
1324 it: self.match_indices(sep),
1331 fn lines(&self) -> Lines {
1332 Lines { inner: self.split_terminator('\n').0 }
1335 fn lines_any(&self) -> LinesAny {
1336 fn f(line: &str) -> &str {
1338 if l > 0 && line.as_bytes()[l - 1] == b'\r' { line.slice(0, l - 1) }
1342 let f: fn(&str) -> &str = f; // coerce to fn pointer
1343 LinesAny { inner: self.lines().map(f) }
1347 fn char_len(&self) -> uint { self.chars().count() }
1350 fn slice(&self, begin: uint, end: uint) -> &str {
1351 // is_char_boundary checks that the index is in [0, .len()]
1353 self.is_char_boundary(begin) &&
1354 self.is_char_boundary(end) {
1355 unsafe { self.slice_unchecked(begin, end) }
1357 slice_error_fail(self, begin, end)
1362 fn slice_from(&self, begin: uint) -> &str {
1363 // is_char_boundary checks that the index is in [0, .len()]
1364 if self.is_char_boundary(begin) {
1365 unsafe { self.slice_unchecked(begin, self.len()) }
1367 slice_error_fail(self, begin, self.len())
1372 fn slice_to(&self, end: uint) -> &str {
1373 // is_char_boundary checks that the index is in [0, .len()]
1374 if self.is_char_boundary(end) {
1375 unsafe { self.slice_unchecked(0, end) }
1377 slice_error_fail(self, 0, end)
1381 fn slice_chars(&self, begin: uint, end: uint) -> &str {
1382 assert!(begin <= end);
1384 let mut begin_byte = None;
1385 let mut end_byte = None;
1387 // This could be even more efficient by not decoding,
1388 // only finding the char boundaries
1389 for (idx, _) in self.char_indices() {
1390 if count == begin { begin_byte = Some(idx); }
1391 if count == end { end_byte = Some(idx); break; }
1394 if begin_byte.is_none() && count == begin { begin_byte = Some(self.len()) }
1395 if end_byte.is_none() && count == end { end_byte = Some(self.len()) }
1397 match (begin_byte, end_byte) {
1398 (None, _) => panic!("slice_chars: `begin` is beyond end of string"),
1399 (_, None) => panic!("slice_chars: `end` is beyond end of string"),
1400 (Some(a), Some(b)) => unsafe { self.slice_unchecked(a, b) }
1405 unsafe fn slice_unchecked(&self, begin: uint, end: uint) -> &str {
1406 mem::transmute(Slice {
1407 data: self.as_ptr().offset(begin as int),
1413 fn starts_with(&self, needle: &str) -> bool {
1414 let n = needle.len();
1415 self.len() >= n && needle.as_bytes() == &self.as_bytes()[..n]
1419 fn ends_with(&self, needle: &str) -> bool {
1420 let (m, n) = (self.len(), needle.len());
1421 m >= n && needle.as_bytes() == &self.as_bytes()[(m-n)..]
1425 fn trim_matches<P: CharEq>(&self, mut pat: P) -> &str {
1426 let cur = match self.find(|&mut: c: char| !pat.matches(c)) {
1428 Some(i) => unsafe { self.slice_unchecked(i, self.len()) }
1430 match cur.rfind(|&mut: c: char| !pat.matches(c)) {
1433 let right = cur.char_range_at(i).next;
1434 unsafe { cur.slice_unchecked(0, right) }
1440 fn trim_left_matches<P: CharEq>(&self, mut pat: P) -> &str {
1441 match self.find(|&mut: c: char| !pat.matches(c)) {
1443 Some(first) => unsafe { self.slice_unchecked(first, self.len()) }
1448 fn trim_right_matches<P: CharEq>(&self, mut pat: P) -> &str {
1449 match self.rfind(|&mut: c: char| !pat.matches(c)) {
1452 let next = self.char_range_at(last).next;
1453 unsafe { self.slice_unchecked(0u, next) }
1459 fn is_char_boundary(&self, index: uint) -> bool {
1460 if index == self.len() { return true; }
1461 match self.as_bytes().get(index) {
1463 Some(&b) => b < 128u8 || b >= 192u8,
1468 fn char_range_at(&self, i: uint) -> CharRange {
1469 if self.as_bytes()[i] < 128u8 {
1470 return CharRange {ch: self.as_bytes()[i] as char, next: i + 1 };
1473 // Multibyte case is a fn to allow char_range_at to inline cleanly
1474 fn multibyte_char_range_at(s: &str, i: uint) -> CharRange {
1475 let mut val = s.as_bytes()[i] as u32;
1476 let w = UTF8_CHAR_WIDTH[val as uint] as uint;
1479 val = utf8_first_byte!(val, w);
1480 val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 1]);
1481 if w > 2 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 2]); }
1482 if w > 3 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 3]); }
1484 return CharRange {ch: unsafe { mem::transmute(val) }, next: i + w};
1487 return multibyte_char_range_at(self, i);
1491 fn char_range_at_reverse(&self, start: uint) -> CharRange {
1492 let mut prev = start;
1494 prev = prev.saturating_sub(1);
1495 if self.as_bytes()[prev] < 128 {
1496 return CharRange{ch: self.as_bytes()[prev] as char, next: prev}
1499 // Multibyte case is a fn to allow char_range_at_reverse to inline cleanly
1500 fn multibyte_char_range_at_reverse(s: &str, mut i: uint) -> CharRange {
1501 // while there is a previous byte == 10......
1502 while i > 0 && s.as_bytes()[i] & !CONT_MASK == TAG_CONT_U8 {
1506 let mut val = s.as_bytes()[i] as u32;
1507 let w = UTF8_CHAR_WIDTH[val as uint] as uint;
1510 val = utf8_first_byte!(val, w);
1511 val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 1]);
1512 if w > 2 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 2]); }
1513 if w > 3 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 3]); }
1515 return CharRange {ch: unsafe { mem::transmute(val) }, next: i};
1518 return multibyte_char_range_at_reverse(self, prev);
1522 fn char_at(&self, i: uint) -> char {
1523 self.char_range_at(i).ch
1527 fn char_at_reverse(&self, i: uint) -> char {
1528 self.char_range_at_reverse(i).ch
1532 fn as_bytes(&self) -> &[u8] {
1533 unsafe { mem::transmute(self) }
1536 fn find<P: CharEq>(&self, mut pat: P) -> Option<uint> {
1537 if pat.only_ascii() {
1538 self.bytes().position(|b| pat.matches(b as char))
1540 for (index, c) in self.char_indices() {
1541 if pat.matches(c) { return Some(index); }
1547 fn rfind<P: CharEq>(&self, mut pat: P) -> Option<uint> {
1548 if pat.only_ascii() {
1549 self.bytes().rposition(|b| pat.matches(b as char))
1551 for (index, c) in self.char_indices().rev() {
1552 if pat.matches(c) { return Some(index); }
1558 fn find_str(&self, needle: &str) -> Option<uint> {
1559 if needle.is_empty() {
1562 self.match_indices(needle)
1564 .map(|(start, _end)| start)
1569 fn slice_shift_char(&self) -> Option<(char, &str)> {
1570 if self.is_empty() {
1573 let CharRange {ch, next} = self.char_range_at(0u);
1574 let next_s = unsafe { self.slice_unchecked(next, self.len()) };
1579 fn subslice_offset(&self, inner: &str) -> uint {
1580 let a_start = self.as_ptr() as uint;
1581 let a_end = a_start + self.len();
1582 let b_start = inner.as_ptr() as uint;
1583 let b_end = b_start + inner.len();
1585 assert!(a_start <= b_start);
1586 assert!(b_end <= a_end);
1591 fn as_ptr(&self) -> *const u8 {
1596 fn len(&self) -> uint { self.repr().len }
1599 fn is_empty(&self) -> bool { self.len() == 0 }
1602 fn parse<T: FromStr>(&self) -> Option<T> { FromStr::from_str(self) }
1606 impl<'a> Default for &'a str {
1608 fn default() -> &'a str { "" }
1612 impl<'a> Iterator for Lines<'a> {
1613 type Item = &'a str;
1616 fn next(&mut self) -> Option<&'a str> { self.inner.next() }
1618 fn size_hint(&self) -> (uint, Option<uint>) { self.inner.size_hint() }
1622 impl<'a> DoubleEndedIterator for Lines<'a> {
1624 fn next_back(&mut self) -> Option<&'a str> { self.inner.next_back() }
1628 impl<'a> Iterator for LinesAny<'a> {
1629 type Item = &'a str;
1632 fn next(&mut self) -> Option<&'a str> { self.inner.next() }
1634 fn size_hint(&self) -> (uint, Option<uint>) { self.inner.size_hint() }
1638 impl<'a> DoubleEndedIterator for LinesAny<'a> {
1640 fn next_back(&mut self) -> Option<&'a str> { self.inner.next_back() }