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};
25 use iter::ExactSizeIterator;
27 use iter::{Map, Iterator, IteratorExt, DoubleEndedIterator};
32 use option::Option::{self, None, Some};
34 use raw::{Repr, Slice};
35 use result::Result::{self, Ok, Err};
36 use slice::{self, SliceExt};
39 macro_rules! delegate_iter {
40 (exact $te:ty : $ti:ty) => {
41 delegate_iter!{$te : $ti}
42 impl<'a> ExactSizeIterator for $ti {
44 fn len(&self) -> uint {
49 ($te:ty : $ti:ty) => {
51 impl<'a> Iterator for $ti {
55 fn next(&mut self) -> Option<$te> {
59 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 : $ti:ty) => {
73 impl<'a, P: CharEq> Iterator for $ti {
77 fn next(&mut self) -> Option<$te> {
81 fn size_hint(&self) -> (uint, Option<uint>) {
86 impl<'a, P: CharEq> DoubleEndedIterator for $ti {
88 fn next_back(&mut self) -> Option<$te> {
93 (pattern forward $te:ty : $ti:ty) => {
95 impl<'a, P: CharEq> Iterator for $ti {
99 fn next(&mut self) -> Option<$te> {
103 fn size_hint(&self) -> (uint, Option<uint>) {
110 /// A trait to abstract the idea of creating a new instance of a type from a
112 // FIXME(#17307): there should be an `E` associated type for a `Result` return
113 #[unstable = "will return a Result once associated types are working"]
115 /// Parses a string `s` to return an optional value of this type. If the
116 /// string is ill-formatted, the None is returned.
117 fn from_str(s: &str) -> Option<Self>;
120 impl FromStr for bool {
121 /// Parse a `bool` from a string.
123 /// Yields an `Option<bool>`, because `s` may or may not actually be parseable.
128 /// assert_eq!("true".parse(), Some(true));
129 /// assert_eq!("false".parse(), Some(false));
130 /// assert_eq!("not even a boolean".parse::<bool>(), None);
133 fn from_str(s: &str) -> Option<bool> {
135 "true" => Some(true),
136 "false" => Some(false),
143 Section: Creating a string
146 /// Errors which can occur when attempting to interpret a byte slice as a `str`.
147 #[derive(Copy, Eq, PartialEq, Clone, Show)]
148 #[unstable = "error enumeration recently added and definitions may be refined"]
150 /// An invalid byte was detected at the byte offset given.
152 /// The offset is guaranteed to be in bounds of the slice in question, and
153 /// the byte at the specified offset was the first invalid byte in the
154 /// sequence detected.
157 /// The byte slice was invalid because more bytes were needed but no more
158 /// bytes were available.
162 /// Converts a slice of bytes to a string slice without performing any
165 /// Once the slice has been validated as utf-8, it is transmuted in-place and
166 /// returned as a '&str' instead of a '&[u8]'
170 /// Returns `Err` if the slice is not utf-8 with a description as to why the
171 /// provided slice is not utf-8.
173 pub fn from_utf8(v: &[u8]) -> Result<&str, Utf8Error> {
174 try!(run_utf8_validation_iterator(&mut v.iter()));
175 Ok(unsafe { from_utf8_unchecked(v) })
178 /// Converts a slice of bytes to a string slice without checking
179 /// that the string contains valid UTF-8.
181 pub unsafe fn from_utf8_unchecked<'a>(v: &'a [u8]) -> &'a str {
185 /// Constructs a static string slice from a given raw pointer.
187 /// This function will read memory starting at `s` until it finds a 0, and then
188 /// transmute the memory up to that point as a string slice, returning the
189 /// corresponding `&'static str` value.
191 /// This function is unsafe because the caller must ensure the C string itself
192 /// has the static lifetime and that the memory `s` is valid up to and including
193 /// the first null byte.
197 /// This function will panic if the string pointed to by `s` is not valid UTF-8.
198 #[deprecated = "use std::ffi::c_str_to_bytes + str::from_utf8"]
199 pub unsafe fn from_c_str(s: *const i8) -> &'static str {
200 let s = s as *const u8;
202 while *s.offset(len as int) != 0 {
205 let v: &'static [u8] = ::mem::transmute(Slice { data: s, len: len });
206 from_utf8(v).ok().expect("from_c_str passed invalid utf-8 data")
209 /// Something that can be used to compare against a character
210 #[unstable = "definition may change as pattern-related methods are stabilized"]
212 /// Determine if the splitter should split at the given character
213 fn matches(&mut self, char) -> bool;
214 /// Indicate if this is only concerned about ASCII characters,
215 /// which can allow for a faster implementation.
216 fn only_ascii(&self) -> bool;
219 impl CharEq for char {
221 fn matches(&mut self, c: char) -> bool { *self == c }
224 fn only_ascii(&self) -> bool { (*self as uint) < 128 }
227 impl<F> CharEq for F where F: FnMut(char) -> bool {
229 fn matches(&mut self, c: char) -> bool { (*self)(c) }
232 fn only_ascii(&self) -> bool { false }
235 impl<'a> CharEq for &'a [char] {
237 fn matches(&mut self, c: char) -> bool {
238 self.iter().any(|&m| { let mut m = m; m.matches(c) })
242 fn only_ascii(&self) -> bool {
243 self.iter().all(|m| m.only_ascii())
248 impl Error for Utf8Error {
249 fn description(&self) -> &str {
251 Utf8Error::TooShort => "invalid utf-8: not enough bytes",
252 Utf8Error::InvalidByte(..) => "invalid utf-8: corrupt contents",
258 impl fmt::Display for Utf8Error {
259 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
261 Utf8Error::InvalidByte(n) => {
262 write!(f, "invalid utf-8: invalid byte at index {}", n)
264 Utf8Error::TooShort => {
265 write!(f, "invalid utf-8: byte slice too short")
275 /// Iterator for the char (representing *Unicode Scalar Values*) of a string
277 /// Created with the method `.chars()`.
278 #[derive(Clone, Copy)]
280 pub struct Chars<'a> {
281 iter: slice::Iter<'a, u8>
284 // Return the initial codepoint accumulator for the first byte.
285 // The first byte is special, only want bottom 5 bits for width 2, 4 bits
286 // for width 3, and 3 bits for width 4
287 macro_rules! utf8_first_byte {
288 ($byte:expr, $width:expr) => (($byte & (0x7F >> $width)) as u32)
291 // return the value of $ch updated with continuation byte $byte
292 macro_rules! utf8_acc_cont_byte {
293 ($ch:expr, $byte:expr) => (($ch << 6) | ($byte & CONT_MASK) as u32)
296 macro_rules! utf8_is_cont_byte {
297 ($byte:expr) => (($byte & !CONT_MASK) == TAG_CONT_U8)
301 fn unwrap_or_0(opt: Option<&u8>) -> u8 {
309 impl<'a> Iterator for Chars<'a> {
313 fn next(&mut self) -> Option<char> {
314 // Decode UTF-8, using the valid UTF-8 invariant
315 let x = match self.iter.next() {
317 Some(&next_byte) if next_byte < 128 => return Some(next_byte as char),
318 Some(&next_byte) => next_byte,
321 // Multibyte case follows
322 // Decode from a byte combination out of: [[[x y] z] w]
323 // NOTE: Performance is sensitive to the exact formulation here
324 let init = utf8_first_byte!(x, 2);
325 let y = unwrap_or_0(self.iter.next());
326 let mut ch = utf8_acc_cont_byte!(init, y);
329 // 5th bit in 0xE0 .. 0xEF is always clear, so `init` is still valid
330 let z = unwrap_or_0(self.iter.next());
331 let y_z = utf8_acc_cont_byte!((y & CONT_MASK) as u32, z);
332 ch = init << 12 | y_z;
335 // use only the lower 3 bits of `init`
336 let w = unwrap_or_0(self.iter.next());
337 ch = (init & 7) << 18 | utf8_acc_cont_byte!(y_z, w);
341 // str invariant says `ch` is a valid Unicode Scalar Value
343 Some(mem::transmute(ch))
348 fn size_hint(&self) -> (uint, Option<uint>) {
349 let (len, _) = self.iter.size_hint();
350 (len.saturating_add(3) / 4, Some(len))
355 impl<'a> DoubleEndedIterator for Chars<'a> {
357 fn next_back(&mut self) -> Option<char> {
358 let w = match self.iter.next_back() {
360 Some(&back_byte) if back_byte < 128 => return Some(back_byte as char),
361 Some(&back_byte) => back_byte,
364 // Multibyte case follows
365 // Decode from a byte combination out of: [x [y [z w]]]
367 let z = unwrap_or_0(self.iter.next_back());
368 ch = utf8_first_byte!(z, 2);
369 if utf8_is_cont_byte!(z) {
370 let y = unwrap_or_0(self.iter.next_back());
371 ch = utf8_first_byte!(y, 3);
372 if utf8_is_cont_byte!(y) {
373 let x = unwrap_or_0(self.iter.next_back());
374 ch = utf8_first_byte!(x, 4);
375 ch = utf8_acc_cont_byte!(ch, y);
377 ch = utf8_acc_cont_byte!(ch, z);
379 ch = utf8_acc_cont_byte!(ch, w);
381 // str invariant says `ch` is a valid Unicode Scalar Value
383 Some(mem::transmute(ch))
388 /// External iterator for a string's characters and their byte offsets.
389 /// Use with the `std::iter` module.
392 pub struct CharIndices<'a> {
398 impl<'a> Iterator for CharIndices<'a> {
399 type Item = (uint, char);
402 fn next(&mut self) -> Option<(uint, char)> {
403 let (pre_len, _) = self.iter.iter.size_hint();
404 match self.iter.next() {
407 let index = self.front_offset;
408 let (len, _) = self.iter.iter.size_hint();
409 self.front_offset += pre_len - len;
416 fn size_hint(&self) -> (uint, Option<uint>) {
417 self.iter.size_hint()
422 impl<'a> DoubleEndedIterator for CharIndices<'a> {
424 fn next_back(&mut self) -> Option<(uint, char)> {
425 match self.iter.next_back() {
428 let (len, _) = self.iter.iter.size_hint();
429 let index = self.front_offset + len;
436 /// External iterator for a string's bytes.
437 /// Use with the `std::iter` module.
439 /// Created with `StrExt::bytes`
442 pub struct Bytes<'a>(Map<&'a u8, u8, slice::Iter<'a, u8>, BytesDeref>);
443 delegate_iter!{exact u8 : Bytes<'a>}
445 /// A temporary fn new type that ensures that the `Bytes` iterator
447 #[derive(Copy, Clone)]
450 impl<'a> Fn(&'a u8) -> u8 for BytesDeref {
452 extern "rust-call" fn call(&self, (ptr,): (&'a u8,)) -> u8 {
457 /// An iterator over the substrings of a string, separated by `sep`.
459 struct CharSplits<'a, Sep> {
460 /// The slice remaining to be iterated
463 /// Whether an empty string at the end is allowed
464 allow_trailing_empty: bool,
469 /// An iterator over the substrings of a string, separated by `sep`,
470 /// splitting at most `count` times.
472 struct CharSplitsN<'a, Sep> {
473 iter: CharSplits<'a, Sep>,
474 /// The number of splits remaining
479 /// An iterator over the lines of a string, separated by `\n`.
481 pub struct Lines<'a> {
482 inner: CharSplits<'a, char>,
485 /// An iterator over the lines of a string, separated by either `\n` or (`\r\n`).
487 pub struct LinesAny<'a> {
488 inner: Map<&'a str, &'a str, Lines<'a>, fn(&str) -> &str>,
491 impl<'a, Sep> CharSplits<'a, Sep> {
493 fn get_end(&mut self) -> Option<&'a str> {
494 if !self.finished && (self.allow_trailing_empty || self.string.len() > 0) {
495 self.finished = true;
504 impl<'a, Sep: CharEq> Iterator for CharSplits<'a, Sep> {
508 fn next(&mut self) -> Option<&'a str> {
509 if self.finished { return None }
511 let mut next_split = None;
513 for (idx, byte) in self.string.bytes().enumerate() {
514 if self.sep.matches(byte as char) && byte < 128u8 {
515 next_split = Some((idx, idx + 1));
520 for (idx, ch) in self.string.char_indices() {
521 if self.sep.matches(ch) {
522 next_split = Some((idx, self.string.char_range_at(idx).next));
528 Some((a, b)) => unsafe {
529 let elt = self.string.slice_unchecked(0, a);
530 self.string = self.string.slice_unchecked(b, self.string.len());
533 None => self.get_end(),
539 impl<'a, Sep: CharEq> DoubleEndedIterator for CharSplits<'a, Sep> {
541 fn next_back(&mut self) -> Option<&'a str> {
542 if self.finished { return None }
544 if !self.allow_trailing_empty {
545 self.allow_trailing_empty = true;
546 match self.next_back() {
547 Some(elt) if !elt.is_empty() => return Some(elt),
548 _ => if self.finished { return None }
551 let len = self.string.len();
552 let mut next_split = None;
555 for (idx, byte) in self.string.bytes().enumerate().rev() {
556 if self.sep.matches(byte as char) && byte < 128u8 {
557 next_split = Some((idx, idx + 1));
562 for (idx, ch) in self.string.char_indices().rev() {
563 if self.sep.matches(ch) {
564 next_split = Some((idx, self.string.char_range_at(idx).next));
570 Some((a, b)) => unsafe {
571 let elt = self.string.slice_unchecked(b, len);
572 self.string = self.string.slice_unchecked(0, a);
575 None => { self.finished = true; Some(self.string) }
581 impl<'a, Sep: CharEq> Iterator for CharSplitsN<'a, Sep> {
585 fn next(&mut self) -> Option<&'a str> {
588 if self.invert { self.iter.next_back() } else { self.iter.next() }
595 /// The internal state of an iterator that searches for matches of a substring
596 /// within a larger string using naive search
598 struct NaiveSearcher {
603 fn new() -> NaiveSearcher {
604 NaiveSearcher { position: 0 }
607 fn next(&mut self, haystack: &[u8], needle: &[u8]) -> Option<(uint, uint)> {
608 while self.position + needle.len() <= haystack.len() {
609 if &haystack[self.position .. self.position + needle.len()] == needle {
610 let match_pos = self.position;
611 self.position += needle.len(); // add 1 for all matches
612 return Some((match_pos, match_pos + needle.len()));
621 /// The internal state of an iterator that searches for matches of a substring
622 /// within a larger string using two-way search
624 struct TwoWaySearcher {
636 This is the Two-Way search algorithm, which was introduced in the paper:
637 Crochemore, M., Perrin, D., 1991, Two-way string-matching, Journal of the ACM 38(3):651-675.
639 Here's some background information.
641 A *word* is a string of symbols. The *length* of a word should be a familiar
642 notion, and here we denote it for any word x by |x|.
643 (We also allow for the possibility of the *empty word*, a word of length zero).
645 If x is any non-empty word, then an integer p with 0 < p <= |x| is said to be a
646 *period* for x iff for all i with 0 <= i <= |x| - p - 1, we have x[i] == x[i+p].
647 For example, both 1 and 2 are periods for the string "aa". As another example,
648 the only period of the string "abcd" is 4.
650 We denote by period(x) the *smallest* period of x (provided that x is non-empty).
651 This is always well-defined since every non-empty word x has at least one period,
652 |x|. We sometimes call this *the period* of x.
654 If u, v and x are words such that x = uv, where uv is the concatenation of u and
655 v, then we say that (u, v) is a *factorization* of x.
657 Let (u, v) be a factorization for a word x. Then if w is a non-empty word such
658 that both of the following hold
660 - either w is a suffix of u or u is a suffix of w
661 - either w is a prefix of v or v is a prefix of w
663 then w is said to be a *repetition* for the factorization (u, v).
665 Just to unpack this, there are four possibilities here. Let w = "abc". Then we
668 - w is a suffix of u and w is a prefix of v. ex: ("lolabc", "abcde")
669 - w is a suffix of u and v is a prefix of w. ex: ("lolabc", "ab")
670 - u is a suffix of w and w is a prefix of v. ex: ("bc", "abchi")
671 - u is a suffix of w and v is a prefix of w. ex: ("bc", "a")
673 Note that the word vu is a repetition for any factorization (u,v) of x = uv,
674 so every factorization has at least one repetition.
676 If x is a string and (u, v) is a factorization for x, then a *local period* for
677 (u, v) is an integer r such that there is some word w such that |w| = r and w is
678 a repetition for (u, v).
680 We denote by local_period(u, v) the smallest local period of (u, v). We sometimes
681 call this *the local period* of (u, v). Provided that x = uv is non-empty, this
682 is well-defined (because each non-empty word has at least one factorization, as
685 It can be proven that the following is an equivalent definition of a local period
686 for a factorization (u, v): any positive integer r such that x[i] == x[i+r] for
687 all i such that |u| - r <= i <= |u| - 1 and such that both x[i] and x[i+r] are
688 defined. (i.e. i > 0 and i + r < |x|).
690 Using the above reformulation, it is easy to prove that
692 1 <= local_period(u, v) <= period(uv)
694 A factorization (u, v) of x such that local_period(u,v) = period(x) is called a
695 *critical factorization*.
697 The algorithm hinges on the following theorem, which is stated without proof:
699 **Critical Factorization Theorem** Any word x has at least one critical
700 factorization (u, v) such that |u| < period(x).
702 The purpose of maximal_suffix is to find such a critical factorization.
705 impl TwoWaySearcher {
706 fn new(needle: &[u8]) -> TwoWaySearcher {
707 let (crit_pos_false, period_false) = TwoWaySearcher::maximal_suffix(needle, false);
708 let (crit_pos_true, period_true) = TwoWaySearcher::maximal_suffix(needle, true);
710 let (crit_pos, period) =
711 if crit_pos_false > crit_pos_true {
712 (crit_pos_false, period_false)
714 (crit_pos_true, period_true)
717 // This isn't in the original algorithm, as far as I'm aware.
718 let byteset = needle.iter()
719 .fold(0, |a, &b| (1 << ((b & 0x3f) as uint)) | a);
721 // A particularly readable explanation of what's going on here can be found
722 // in Crochemore and Rytter's book "Text Algorithms", ch 13. Specifically
723 // see the code for "Algorithm CP" on p. 323.
725 // What's going on is we have some critical factorization (u, v) of the
726 // needle, and we want to determine whether u is a suffix of
727 // &v[..period]. If it is, we use "Algorithm CP1". Otherwise we use
728 // "Algorithm CP2", which is optimized for when the period of the needle
730 if &needle[..crit_pos] == &needle[period.. period + crit_pos] {
742 period: cmp::max(crit_pos, needle.len() - crit_pos) + 1,
746 memory: uint::MAX // Dummy value to signify that the period is long
751 // One of the main ideas of Two-Way is that we factorize the needle into
752 // two halves, (u, v), and begin trying to find v in the haystack by scanning
753 // left to right. If v matches, we try to match u by scanning right to left.
754 // How far we can jump when we encounter a mismatch is all based on the fact
755 // that (u, v) is a critical factorization for the needle.
757 fn next(&mut self, haystack: &[u8], needle: &[u8], long_period: bool) -> Option<(uint, uint)> {
759 // Check that we have room to search in
760 if self.position + needle.len() > haystack.len() {
764 // Quickly skip by large portions unrelated to our substring
766 ((haystack[self.position + needle.len() - 1] & 0x3f)
768 self.position += needle.len();
775 // See if the right part of the needle matches
776 let start = if long_period { self.crit_pos }
777 else { cmp::max(self.crit_pos, self.memory) };
778 for i in range(start, needle.len()) {
779 if needle[i] != haystack[self.position + i] {
780 self.position += i - self.crit_pos + 1;
788 // See if the left part of the needle matches
789 let start = if long_period { 0 } else { self.memory };
790 for i in range(start, self.crit_pos).rev() {
791 if needle[i] != haystack[self.position + i] {
792 self.position += self.period;
794 self.memory = needle.len() - self.period;
800 // We have found a match!
801 let match_pos = self.position;
802 self.position += needle.len(); // add self.period for all matches
804 self.memory = 0; // set to needle.len() - self.period for all matches
806 return Some((match_pos, match_pos + needle.len()));
810 // Computes a critical factorization (u, v) of `arr`.
811 // Specifically, returns (i, p), where i is the starting index of v in some
812 // critical factorization (u, v) and p = period(v)
814 fn maximal_suffix(arr: &[u8], reversed: bool) -> (uint, uint) {
815 let mut left = -1; // Corresponds to i in the paper
816 let mut right = 0; // Corresponds to j in the paper
817 let mut offset = 1; // Corresponds to k in the paper
818 let mut period = 1; // Corresponds to p in the paper
820 while right + offset < arr.len() {
824 a = arr[left + offset];
825 b = arr[right + offset];
827 a = arr[right + offset];
828 b = arr[left + offset];
831 // Suffix is smaller, period is entire prefix so far.
834 period = right - left;
836 // Advance through repetition of the current period.
837 if offset == period {
844 // Suffix is larger, start over from current location.
855 /// The internal state of an iterator that searches for matches of a substring
856 /// within a larger string using a dynamically chosen search algorithm
859 Naive(NaiveSearcher),
860 TwoWay(TwoWaySearcher),
861 TwoWayLong(TwoWaySearcher)
865 fn new(haystack: &[u8], needle: &[u8]) -> Searcher {
867 // FIXME(#16715): This unsigned integer addition will probably not
868 // overflow because that would mean that the memory almost solely
869 // consists of the needle. Needs #16715 to be formally fixed.
870 if needle.len() + 20 > haystack.len() {
871 Naive(NaiveSearcher::new())
873 let searcher = TwoWaySearcher::new(needle);
874 if searcher.memory == uint::MAX { // If the period is long
883 /// An iterator over the start and end indices of the matches of a
884 /// substring within a larger string
886 #[unstable = "type may be removed"]
887 pub struct MatchIndices<'a> {
894 /// An iterator over the substrings of a string separated by a given
897 #[unstable = "type may be removed"]
898 pub struct SplitStr<'a> {
899 it: MatchIndices<'a>,
905 impl<'a> Iterator for MatchIndices<'a> {
906 type Item = (uint, uint);
909 fn next(&mut self) -> Option<(uint, uint)> {
910 match self.searcher {
911 Naive(ref mut searcher)
912 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes()),
913 TwoWay(ref mut searcher)
914 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes(), false),
915 TwoWayLong(ref mut searcher)
916 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes(), true)
922 impl<'a> Iterator for SplitStr<'a> {
926 fn next(&mut self) -> Option<&'a str> {
927 if self.finished { return None; }
929 match self.it.next() {
930 Some((from, to)) => {
931 let ret = Some(self.it.haystack.slice(self.last_end, from));
936 self.finished = true;
937 Some(self.it.haystack.slice(self.last_end, self.it.haystack.len()))
945 Section: Comparing strings
948 // share the implementation of the lang-item vs. non-lang-item
950 /// NOTE: This function is (ab)used in rustc::middle::trans::_match
951 /// to compare &[u8] byte slices that are not necessarily valid UTF-8.
953 fn eq_slice_(a: &str, b: &str) -> bool {
954 #[allow(improper_ctypes)]
955 extern { fn memcmp(s1: *const i8, s2: *const i8, n: uint) -> i32; }
956 a.len() == b.len() && unsafe {
957 memcmp(a.as_ptr() as *const i8,
958 b.as_ptr() as *const i8,
963 /// Bytewise slice equality
964 /// NOTE: This function is (ab)used in rustc::middle::trans::_match
965 /// to compare &[u8] byte slices that are not necessarily valid UTF-8.
968 fn eq_slice(a: &str, b: &str) -> bool {
976 /// Walk through `iter` checking that it's a valid UTF-8 sequence,
977 /// returning `true` in that case, or, if it is invalid, `false` with
978 /// `iter` reset such that it is pointing at the first byte in the
979 /// invalid sequence.
981 fn run_utf8_validation_iterator(iter: &mut slice::Iter<u8>)
982 -> Result<(), Utf8Error> {
983 let whole = iter.as_slice();
985 // save the current thing we're pointing at.
988 // restore the iterator we had at the start of this codepoint.
989 macro_rules! err { () => {{
991 return Err(Utf8Error::InvalidByte(whole.len() - iter.as_slice().len()))
994 macro_rules! next { () => {
997 // we needed data, but there was none: error!
998 None => return Err(Utf8Error::TooShort),
1002 let first = match iter.next() {
1004 // we're at the end of the iterator and a codepoint
1005 // boundary at the same time, so this string is valid.
1006 None => return Ok(())
1009 // ASCII characters are always valid, so only large
1010 // bytes need more examination.
1012 let w = UTF8_CHAR_WIDTH[first as uint] as uint;
1013 let second = next!();
1014 // 2-byte encoding is for codepoints \u{0080} to \u{07ff}
1015 // first C2 80 last DF BF
1016 // 3-byte encoding is for codepoints \u{0800} to \u{ffff}
1017 // first E0 A0 80 last EF BF BF
1018 // excluding surrogates codepoints \u{d800} to \u{dfff}
1019 // ED A0 80 to ED BF BF
1020 // 4-byte encoding is for codepoints \u{1000}0 to \u{10ff}ff
1021 // first F0 90 80 80 last F4 8F BF BF
1023 // Use the UTF-8 syntax from the RFC
1025 // https://tools.ietf.org/html/rfc3629
1027 // UTF8-2 = %xC2-DF UTF8-tail
1028 // UTF8-3 = %xE0 %xA0-BF UTF8-tail / %xE1-EC 2( UTF8-tail ) /
1029 // %xED %x80-9F UTF8-tail / %xEE-EF 2( UTF8-tail )
1030 // UTF8-4 = %xF0 %x90-BF 2( UTF8-tail ) / %xF1-F3 3( UTF8-tail ) /
1031 // %xF4 %x80-8F 2( UTF8-tail )
1033 2 => if second & !CONT_MASK != TAG_CONT_U8 {err!()},
1035 match (first, second, next!() & !CONT_MASK) {
1036 (0xE0 , 0xA0 ... 0xBF, TAG_CONT_U8) |
1037 (0xE1 ... 0xEC, 0x80 ... 0xBF, TAG_CONT_U8) |
1038 (0xED , 0x80 ... 0x9F, TAG_CONT_U8) |
1039 (0xEE ... 0xEF, 0x80 ... 0xBF, TAG_CONT_U8) => {}
1044 match (first, second, next!() & !CONT_MASK, next!() & !CONT_MASK) {
1045 (0xF0 , 0x90 ... 0xBF, TAG_CONT_U8, TAG_CONT_U8) |
1046 (0xF1 ... 0xF3, 0x80 ... 0xBF, TAG_CONT_U8, TAG_CONT_U8) |
1047 (0xF4 , 0x80 ... 0x8F, TAG_CONT_U8, TAG_CONT_U8) => {}
1057 // https://tools.ietf.org/html/rfc3629
1058 static UTF8_CHAR_WIDTH: [u8; 256] = [
1059 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1060 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x1F
1061 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1062 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x3F
1063 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1064 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x5F
1065 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
1066 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x7F
1067 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
1068 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0x9F
1069 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
1070 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0xBF
1071 0,0,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
1072 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, // 0xDF
1073 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, // 0xEF
1074 4,4,4,4,4,0,0,0,0,0,0,0,0,0,0,0, // 0xFF
1077 /// Struct that contains a `char` and the index of the first byte of
1078 /// the next `char` in a string. This can be used as a data structure
1079 /// for iterating over the UTF-8 bytes of a string.
1081 #[unstable = "naming is uncertain with container conventions"]
1082 pub struct CharRange {
1085 /// Index of the first byte of the next `char`
1089 /// Mask of the value bits of a continuation byte
1090 const CONT_MASK: u8 = 0b0011_1111u8;
1091 /// Value of the tag bits (tag mask is !CONT_MASK) of a continuation byte
1092 const TAG_CONT_U8: u8 = 0b1000_0000u8;
1095 Section: Trait implementations
1099 use cmp::{Ordering, Ord, PartialEq, PartialOrd, Eq};
1100 use cmp::Ordering::{Less, Equal, Greater};
1101 use iter::IteratorExt;
1103 use option::Option::Some;
1105 use str::{StrExt, eq_slice};
1110 fn cmp(&self, other: &str) -> Ordering {
1111 for (s_b, o_b) in self.bytes().zip(other.bytes()) {
1112 match s_b.cmp(&o_b) {
1113 Greater => return Greater,
1114 Less => return Less,
1119 self.len().cmp(&other.len())
1124 impl PartialEq for str {
1126 fn eq(&self, other: &str) -> bool {
1127 eq_slice(self, other)
1130 fn ne(&self, other: &str) -> bool { !(*self).eq(other) }
1137 impl PartialOrd for str {
1139 fn partial_cmp(&self, other: &str) -> Option<Ordering> {
1140 Some(self.cmp(other))
1144 impl ops::Index<ops::Range<uint>> for str {
1147 fn index(&self, index: &ops::Range<uint>) -> &str {
1148 self.slice(index.start, index.end)
1151 impl ops::Index<ops::RangeTo<uint>> for str {
1154 fn index(&self, index: &ops::RangeTo<uint>) -> &str {
1155 self.slice_to(index.end)
1158 impl ops::Index<ops::RangeFrom<uint>> for str {
1161 fn index(&self, index: &ops::RangeFrom<uint>) -> &str {
1162 self.slice_from(index.start)
1165 impl ops::Index<ops::FullRange> for str {
1168 fn index(&self, _index: &ops::FullRange) -> &str {
1174 /// Any string that can be represented as a slice
1175 #[unstable = "Instead of taking this bound generically, this trait will be \
1176 replaced with one of slicing syntax (&foo[]), deref coercions, or \
1177 a more generic conversion trait"]
1179 /// Work with `self` as a slice.
1180 fn as_slice<'a>(&'a self) -> &'a str;
1185 fn as_slice<'a>(&'a self) -> &'a str { self }
1188 impl<'a, S: ?Sized> Str for &'a S where S: Str {
1190 fn as_slice(&self) -> &str { Str::as_slice(*self) }
1193 /// Return type of `StrExt::split`
1196 pub struct Split<'a, P>(CharSplits<'a, P>);
1197 delegate_iter!{pattern &'a str : Split<'a, P>}
1199 /// Return type of `StrExt::split_terminator`
1201 #[unstable = "might get removed in favour of a constructor method on Split"]
1202 pub struct SplitTerminator<'a, P>(CharSplits<'a, P>);
1203 delegate_iter!{pattern &'a str : SplitTerminator<'a, P>}
1205 /// Return type of `StrExt::splitn`
1208 pub struct SplitN<'a, P>(CharSplitsN<'a, P>);
1209 delegate_iter!{pattern forward &'a str : SplitN<'a, P>}
1211 /// Return type of `StrExt::rsplitn`
1214 pub struct RSplitN<'a, P>(CharSplitsN<'a, P>);
1215 delegate_iter!{pattern forward &'a str : RSplitN<'a, P>}
1217 /// Methods for string slices
1218 #[allow(missing_docs)]
1220 // NB there are no docs here are they're all located on the StrExt trait in
1221 // libcollections, not here.
1223 fn contains(&self, pat: &str) -> bool;
1224 fn contains_char<P: CharEq>(&self, pat: P) -> bool;
1225 fn chars<'a>(&'a self) -> Chars<'a>;
1226 fn bytes<'a>(&'a self) -> Bytes<'a>;
1227 fn char_indices<'a>(&'a self) -> CharIndices<'a>;
1228 fn split<'a, P: CharEq>(&'a self, pat: P) -> Split<'a, P>;
1229 fn splitn<'a, P: CharEq>(&'a self, count: uint, pat: P) -> SplitN<'a, P>;
1230 fn split_terminator<'a, P: CharEq>(&'a self, pat: P) -> SplitTerminator<'a, P>;
1231 fn rsplitn<'a, P: CharEq>(&'a self, count: uint, pat: P) -> RSplitN<'a, P>;
1232 fn match_indices<'a>(&'a self, sep: &'a str) -> MatchIndices<'a>;
1233 fn split_str<'a>(&'a self, pat: &'a str) -> SplitStr<'a>;
1234 fn lines<'a>(&'a self) -> Lines<'a>;
1235 fn lines_any<'a>(&'a self) -> LinesAny<'a>;
1236 fn char_len(&self) -> uint;
1237 fn slice<'a>(&'a self, begin: uint, end: uint) -> &'a str;
1238 fn slice_from<'a>(&'a self, begin: uint) -> &'a str;
1239 fn slice_to<'a>(&'a self, end: uint) -> &'a str;
1240 fn slice_chars<'a>(&'a self, begin: uint, end: uint) -> &'a str;
1241 unsafe fn slice_unchecked<'a>(&'a self, begin: uint, end: uint) -> &'a str;
1242 fn starts_with(&self, pat: &str) -> bool;
1243 fn ends_with(&self, pat: &str) -> bool;
1244 fn trim_matches<'a, P: CharEq>(&'a self, pat: P) -> &'a str;
1245 fn trim_left_matches<'a, P: CharEq>(&'a self, pat: P) -> &'a str;
1246 fn trim_right_matches<'a, P: CharEq>(&'a self, pat: P) -> &'a str;
1247 fn is_char_boundary(&self, index: uint) -> bool;
1248 fn char_range_at(&self, start: uint) -> CharRange;
1249 fn char_range_at_reverse(&self, start: uint) -> CharRange;
1250 fn char_at(&self, i: uint) -> char;
1251 fn char_at_reverse(&self, i: uint) -> char;
1252 fn as_bytes<'a>(&'a self) -> &'a [u8];
1253 fn find<P: CharEq>(&self, pat: P) -> Option<uint>;
1254 fn rfind<P: CharEq>(&self, pat: P) -> Option<uint>;
1255 fn find_str(&self, pat: &str) -> Option<uint>;
1256 fn slice_shift_char<'a>(&'a self) -> Option<(char, &'a str)>;
1257 fn subslice_offset(&self, inner: &str) -> uint;
1258 fn as_ptr(&self) -> *const u8;
1259 fn len(&self) -> uint;
1260 fn is_empty(&self) -> bool;
1261 fn parse<T: FromStr>(&self) -> Option<T>;
1265 fn slice_error_fail(s: &str, begin: uint, end: uint) -> ! {
1266 assert!(begin <= end);
1267 panic!("index {} and/or {} in `{}` do not lie on character boundary",
1271 impl StrExt for str {
1273 fn contains(&self, needle: &str) -> bool {
1274 self.find_str(needle).is_some()
1278 fn contains_char<P: CharEq>(&self, pat: P) -> bool {
1279 self.find(pat).is_some()
1283 fn chars(&self) -> Chars {
1284 Chars{iter: self.as_bytes().iter()}
1288 fn bytes(&self) -> Bytes {
1289 Bytes(self.as_bytes().iter().map(BytesDeref))
1293 fn char_indices(&self) -> CharIndices {
1294 CharIndices { front_offset: 0, iter: self.chars() }
1298 fn split<P: CharEq>(&self, pat: P) -> Split<P> {
1301 only_ascii: pat.only_ascii(),
1303 allow_trailing_empty: true,
1309 fn splitn<P: CharEq>(&self, count: uint, pat: P) -> SplitN<P> {
1310 SplitN(CharSplitsN {
1311 iter: self.split(pat).0,
1318 fn split_terminator<P: CharEq>(&self, pat: P) -> SplitTerminator<P> {
1319 SplitTerminator(CharSplits {
1320 allow_trailing_empty: false,
1326 fn rsplitn<P: CharEq>(&self, count: uint, pat: P) -> RSplitN<P> {
1327 RSplitN(CharSplitsN {
1328 iter: self.split(pat).0,
1335 fn match_indices<'a>(&'a self, sep: &'a str) -> MatchIndices<'a> {
1336 assert!(!sep.is_empty());
1340 searcher: Searcher::new(self.as_bytes(), sep.as_bytes())
1345 fn split_str<'a>(&'a self, sep: &'a str) -> SplitStr<'a> {
1347 it: self.match_indices(sep),
1354 fn lines(&self) -> Lines {
1355 Lines { inner: self.split_terminator('\n').0 }
1358 fn lines_any(&self) -> LinesAny {
1359 fn f(line: &str) -> &str {
1361 if l > 0 && line.as_bytes()[l - 1] == b'\r' { line.slice(0, l - 1) }
1365 let f: fn(&str) -> &str = f; // coerce to fn pointer
1366 LinesAny { inner: self.lines().map(f) }
1370 fn char_len(&self) -> uint { self.chars().count() }
1373 fn slice(&self, begin: uint, end: uint) -> &str {
1374 // is_char_boundary checks that the index is in [0, .len()]
1376 self.is_char_boundary(begin) &&
1377 self.is_char_boundary(end) {
1378 unsafe { self.slice_unchecked(begin, end) }
1380 slice_error_fail(self, begin, end)
1385 fn slice_from(&self, begin: uint) -> &str {
1386 // is_char_boundary checks that the index is in [0, .len()]
1387 if self.is_char_boundary(begin) {
1388 unsafe { self.slice_unchecked(begin, self.len()) }
1390 slice_error_fail(self, begin, self.len())
1395 fn slice_to(&self, end: uint) -> &str {
1396 // is_char_boundary checks that the index is in [0, .len()]
1397 if self.is_char_boundary(end) {
1398 unsafe { self.slice_unchecked(0, end) }
1400 slice_error_fail(self, 0, end)
1404 fn slice_chars(&self, begin: uint, end: uint) -> &str {
1405 assert!(begin <= end);
1407 let mut begin_byte = None;
1408 let mut end_byte = None;
1410 // This could be even more efficient by not decoding,
1411 // only finding the char boundaries
1412 for (idx, _) in self.char_indices() {
1413 if count == begin { begin_byte = Some(idx); }
1414 if count == end { end_byte = Some(idx); break; }
1417 if begin_byte.is_none() && count == begin { begin_byte = Some(self.len()) }
1418 if end_byte.is_none() && count == end { end_byte = Some(self.len()) }
1420 match (begin_byte, end_byte) {
1421 (None, _) => panic!("slice_chars: `begin` is beyond end of string"),
1422 (_, None) => panic!("slice_chars: `end` is beyond end of string"),
1423 (Some(a), Some(b)) => unsafe { self.slice_unchecked(a, b) }
1428 unsafe fn slice_unchecked(&self, begin: uint, end: uint) -> &str {
1429 mem::transmute(Slice {
1430 data: self.as_ptr().offset(begin as int),
1436 fn starts_with(&self, needle: &str) -> bool {
1437 let n = needle.len();
1438 self.len() >= n && needle.as_bytes() == &self.as_bytes()[..n]
1442 fn ends_with(&self, needle: &str) -> bool {
1443 let (m, n) = (self.len(), needle.len());
1444 m >= n && needle.as_bytes() == &self.as_bytes()[m-n..]
1448 fn trim_matches<P: CharEq>(&self, mut pat: P) -> &str {
1449 let cur = match self.find(|&mut: c: char| !pat.matches(c)) {
1451 Some(i) => unsafe { self.slice_unchecked(i, self.len()) }
1453 match cur.rfind(|&mut: c: char| !pat.matches(c)) {
1456 let right = cur.char_range_at(i).next;
1457 unsafe { cur.slice_unchecked(0, right) }
1463 fn trim_left_matches<P: CharEq>(&self, mut pat: P) -> &str {
1464 match self.find(|&mut: c: char| !pat.matches(c)) {
1466 Some(first) => unsafe { self.slice_unchecked(first, self.len()) }
1471 fn trim_right_matches<P: CharEq>(&self, mut pat: P) -> &str {
1472 match self.rfind(|&mut: c: char| !pat.matches(c)) {
1475 let next = self.char_range_at(last).next;
1476 unsafe { self.slice_unchecked(0u, next) }
1482 fn is_char_boundary(&self, index: uint) -> bool {
1483 if index == self.len() { return true; }
1484 match self.as_bytes().get(index) {
1486 Some(&b) => b < 128u8 || b >= 192u8,
1491 fn char_range_at(&self, i: uint) -> CharRange {
1492 if self.as_bytes()[i] < 128u8 {
1493 return CharRange {ch: self.as_bytes()[i] as char, next: i + 1 };
1496 // Multibyte case is a fn to allow char_range_at to inline cleanly
1497 fn multibyte_char_range_at(s: &str, i: uint) -> CharRange {
1498 let mut val = s.as_bytes()[i] as u32;
1499 let w = UTF8_CHAR_WIDTH[val as uint] as uint;
1502 val = utf8_first_byte!(val, w);
1503 val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 1]);
1504 if w > 2 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 2]); }
1505 if w > 3 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 3]); }
1507 return CharRange {ch: unsafe { mem::transmute(val) }, next: i + w};
1510 return multibyte_char_range_at(self, i);
1514 fn char_range_at_reverse(&self, start: uint) -> CharRange {
1515 let mut prev = start;
1517 prev = prev.saturating_sub(1);
1518 if self.as_bytes()[prev] < 128 {
1519 return CharRange{ch: self.as_bytes()[prev] as char, next: prev}
1522 // Multibyte case is a fn to allow char_range_at_reverse to inline cleanly
1523 fn multibyte_char_range_at_reverse(s: &str, mut i: uint) -> CharRange {
1524 // while there is a previous byte == 10......
1525 while i > 0 && s.as_bytes()[i] & !CONT_MASK == TAG_CONT_U8 {
1529 let mut val = s.as_bytes()[i] as u32;
1530 let w = UTF8_CHAR_WIDTH[val as uint] as uint;
1533 val = utf8_first_byte!(val, w);
1534 val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 1]);
1535 if w > 2 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 2]); }
1536 if w > 3 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 3]); }
1538 return CharRange {ch: unsafe { mem::transmute(val) }, next: i};
1541 return multibyte_char_range_at_reverse(self, prev);
1545 fn char_at(&self, i: uint) -> char {
1546 self.char_range_at(i).ch
1550 fn char_at_reverse(&self, i: uint) -> char {
1551 self.char_range_at_reverse(i).ch
1555 fn as_bytes(&self) -> &[u8] {
1556 unsafe { mem::transmute(self) }
1559 fn find<P: CharEq>(&self, mut pat: P) -> Option<uint> {
1560 if pat.only_ascii() {
1561 self.bytes().position(|b| pat.matches(b as char))
1563 for (index, c) in self.char_indices() {
1564 if pat.matches(c) { return Some(index); }
1570 fn rfind<P: CharEq>(&self, mut pat: P) -> Option<uint> {
1571 if pat.only_ascii() {
1572 self.bytes().rposition(|b| pat.matches(b as char))
1574 for (index, c) in self.char_indices().rev() {
1575 if pat.matches(c) { return Some(index); }
1581 fn find_str(&self, needle: &str) -> Option<uint> {
1582 if needle.is_empty() {
1585 self.match_indices(needle)
1587 .map(|(start, _end)| start)
1592 fn slice_shift_char(&self) -> Option<(char, &str)> {
1593 if self.is_empty() {
1596 let CharRange {ch, next} = self.char_range_at(0u);
1597 let next_s = unsafe { self.slice_unchecked(next, self.len()) };
1602 fn subslice_offset(&self, inner: &str) -> uint {
1603 let a_start = self.as_ptr() as uint;
1604 let a_end = a_start + self.len();
1605 let b_start = inner.as_ptr() as uint;
1606 let b_end = b_start + inner.len();
1608 assert!(a_start <= b_start);
1609 assert!(b_end <= a_end);
1614 fn as_ptr(&self) -> *const u8 {
1619 fn len(&self) -> uint { self.repr().len }
1622 fn is_empty(&self) -> bool { self.len() == 0 }
1625 fn parse<T: FromStr>(&self) -> Option<T> { FromStr::from_str(self) }
1629 impl<'a> Default for &'a str {
1631 fn default() -> &'a str { "" }
1635 impl<'a> Iterator for Lines<'a> {
1636 type Item = &'a str;
1639 fn next(&mut self) -> Option<&'a str> { self.inner.next() }
1641 fn size_hint(&self) -> (uint, Option<uint>) { self.inner.size_hint() }
1645 impl<'a> DoubleEndedIterator for Lines<'a> {
1647 fn next_back(&mut self) -> Option<&'a str> { self.inner.next_back() }
1651 impl<'a> Iterator for LinesAny<'a> {
1652 type Item = &'a str;
1655 fn next(&mut self) -> Option<&'a str> { self.inner.next() }
1657 fn size_hint(&self) -> (uint, Option<uint>) { self.inner.size_hint() }
1661 impl<'a> DoubleEndedIterator for LinesAny<'a> {
1663 fn next_back(&mut self) -> Option<&'a str> { self.inner.next_back() }