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")]
24 use cmp::{PartialEq, Eq};
25 use collections::Collection;
27 use iter::{Map, Iterator};
28 use iter::{DoubleEndedIterator, ExactSize};
30 use num::{CheckedMul, Saturating};
31 use option::{Option, None, Some};
33 use slice::{ImmutableSlice, MutableSlice};
38 Section: Creating a string
41 /// Converts a vector to a string slice without performing any allocations.
43 /// Once the slice has been validated as utf-8, it is transmuted in-place and
44 /// returned as a '&str' instead of a '&[u8]'
46 /// Returns None if the slice is not utf-8.
47 pub fn from_utf8<'a>(v: &'a [u8]) -> Option<&'a str> {
49 Some(unsafe { raw::from_utf8(v) })
53 /// Something that can be used to compare against a character
55 /// Determine if the splitter should split at the given character
56 fn matches(&mut self, char) -> bool;
57 /// Indicate if this is only concerned about ASCII characters,
58 /// which can allow for a faster implementation.
59 fn only_ascii(&self) -> bool;
62 impl CharEq for char {
64 fn matches(&mut self, c: char) -> bool { *self == c }
67 fn only_ascii(&self) -> bool { (*self as uint) < 128 }
70 impl<'a> CharEq for |char|: 'a -> bool {
72 fn matches(&mut self, c: char) -> bool { (*self)(c) }
75 fn only_ascii(&self) -> bool { false }
78 impl CharEq for extern "Rust" fn(char) -> bool {
80 fn matches(&mut self, c: char) -> bool { (*self)(c) }
83 fn only_ascii(&self) -> bool { false }
86 impl<'a> CharEq for &'a [char] {
88 fn matches(&mut self, c: char) -> bool {
89 self.iter().any(|&mut m| m.matches(c))
93 fn only_ascii(&self) -> bool {
94 self.iter().all(|m| m.only_ascii())
102 /// Iterator for the char (representing *Unicode Scalar Values*) of a string
104 /// Created with the method `.chars()`.
106 pub struct Chars<'a> {
107 iter: slice::Items<'a, u8>
110 // Return the initial codepoint accumulator for the first byte.
111 // The first byte is special, only want bottom 5 bits for width 2, 4 bits
112 // for width 3, and 3 bits for width 4
113 macro_rules! utf8_first_byte(
114 ($byte:expr, $width:expr) => (($byte & (0x7F >> $width)) as u32)
117 // return the value of $ch updated with continuation byte $byte
118 macro_rules! utf8_acc_cont_byte(
119 ($ch:expr, $byte:expr) => (($ch << 6) | ($byte & CONT_MASK) as u32)
122 macro_rules! utf8_is_cont_byte(
123 ($byte:expr) => (($byte & !CONT_MASK) == TAG_CONT_U8)
127 fn unwrap_or_0(opt: Option<&u8>) -> u8 {
134 impl<'a> Iterator<char> for Chars<'a> {
136 fn next(&mut self) -> Option<char> {
137 // Decode UTF-8, using the valid UTF-8 invariant
138 let x = match self.iter.next() {
140 Some(&next_byte) if next_byte < 128 => return Some(next_byte as char),
141 Some(&next_byte) => next_byte,
144 // Multibyte case follows
145 // Decode from a byte combination out of: [[[x y] z] w]
146 // NOTE: Performance is sensitive to the exact formulation here
147 let init = utf8_first_byte!(x, 2);
148 let y = unwrap_or_0(self.iter.next());
149 let mut ch = utf8_acc_cont_byte!(init, y);
152 // 5th bit in 0xE0 .. 0xEF is always clear, so `init` is still valid
153 let z = unwrap_or_0(self.iter.next());
154 let y_z = utf8_acc_cont_byte!((y & CONT_MASK) as u32, z);
155 ch = init << 12 | y_z;
158 // use only the lower 3 bits of `init`
159 let w = unwrap_or_0(self.iter.next());
160 ch = (init & 7) << 18 | utf8_acc_cont_byte!(y_z, w);
164 // str invariant says `ch` is a valid Unicode Scalar Value
166 Some(mem::transmute(ch))
171 fn size_hint(&self) -> (uint, Option<uint>) {
172 let (len, _) = self.iter.size_hint();
173 (len.saturating_add(3) / 4, Some(len))
177 impl<'a> DoubleEndedIterator<char> for Chars<'a> {
179 fn next_back(&mut self) -> Option<char> {
180 let w = match self.iter.next_back() {
182 Some(&back_byte) if back_byte < 128 => return Some(back_byte as char),
183 Some(&back_byte) => back_byte,
186 // Multibyte case follows
187 // Decode from a byte combination out of: [x [y [z w]]]
189 let z = unwrap_or_0(self.iter.next_back());
190 ch = utf8_first_byte!(z, 2);
191 if utf8_is_cont_byte!(z) {
192 let y = unwrap_or_0(self.iter.next_back());
193 ch = utf8_first_byte!(y, 3);
194 if utf8_is_cont_byte!(y) {
195 let x = unwrap_or_0(self.iter.next_back());
196 ch = utf8_first_byte!(x, 4);
197 ch = utf8_acc_cont_byte!(ch, y);
199 ch = utf8_acc_cont_byte!(ch, z);
201 ch = utf8_acc_cont_byte!(ch, w);
203 // str invariant says `ch` is a valid Unicode Scalar Value
205 Some(mem::transmute(ch))
210 /// External iterator for a string's characters and their byte offsets.
211 /// Use with the `std::iter` module.
213 pub struct CharOffsets<'a> {
218 impl<'a> Iterator<(uint, char)> for CharOffsets<'a> {
220 fn next(&mut self) -> Option<(uint, char)> {
221 let (pre_len, _) = self.iter.iter.size_hint();
222 match self.iter.next() {
225 let index = self.front_offset;
226 let (len, _) = self.iter.iter.size_hint();
227 self.front_offset += pre_len - len;
234 fn size_hint(&self) -> (uint, Option<uint>) {
235 self.iter.size_hint()
239 impl<'a> DoubleEndedIterator<(uint, char)> for CharOffsets<'a> {
241 fn next_back(&mut self) -> Option<(uint, char)> {
242 match self.iter.next_back() {
245 let (len, _) = self.iter.iter.size_hint();
246 let index = self.front_offset + len;
253 /// External iterator for a string's bytes.
254 /// Use with the `std::iter` module.
256 Map<'a, &'a u8, u8, slice::Items<'a, u8>>;
258 /// An iterator over the substrings of a string, separated by `sep`.
260 pub struct CharSplits<'a, Sep> {
261 /// The slice remaining to be iterated
264 /// Whether an empty string at the end is allowed
265 allow_trailing_empty: bool,
270 /// An iterator over the substrings of a string, separated by `sep`,
271 /// splitting at most `count` times.
273 pub struct CharSplitsN<'a, Sep> {
274 iter: CharSplits<'a, Sep>,
275 /// The number of splits remaining
280 /// An iterator over the lines of a string, separated by either `\n` or (`\r\n`).
281 pub type AnyLines<'a> =
282 Map<'a, &'a str, &'a str, CharSplits<'a, char>>;
284 impl<'a, Sep> CharSplits<'a, Sep> {
286 fn get_end(&mut self) -> Option<&'a str> {
287 if !self.finished && (self.allow_trailing_empty || self.string.len() > 0) {
288 self.finished = true;
296 impl<'a, Sep: CharEq> Iterator<&'a str> for CharSplits<'a, Sep> {
298 fn next(&mut self) -> Option<&'a str> {
299 if self.finished { return None }
301 let mut next_split = None;
303 for (idx, byte) in self.string.bytes().enumerate() {
304 if self.sep.matches(byte as char) && byte < 128u8 {
305 next_split = Some((idx, idx + 1));
310 for (idx, ch) in self.string.char_indices() {
311 if self.sep.matches(ch) {
312 next_split = Some((idx, self.string.char_range_at(idx).next));
318 Some((a, b)) => unsafe {
319 let elt = raw::slice_unchecked(self.string, 0, a);
320 self.string = raw::slice_unchecked(self.string, b, self.string.len());
323 None => self.get_end(),
328 impl<'a, Sep: CharEq> DoubleEndedIterator<&'a str>
329 for CharSplits<'a, Sep> {
331 fn next_back(&mut self) -> Option<&'a str> {
332 if self.finished { return None }
334 if !self.allow_trailing_empty {
335 self.allow_trailing_empty = true;
336 match self.next_back() {
337 Some(elt) if !elt.is_empty() => return Some(elt),
338 _ => if self.finished { return None }
341 let len = self.string.len();
342 let mut next_split = None;
345 for (idx, byte) in self.string.bytes().enumerate().rev() {
346 if self.sep.matches(byte as char) && byte < 128u8 {
347 next_split = Some((idx, idx + 1));
352 for (idx, ch) in self.string.char_indices().rev() {
353 if self.sep.matches(ch) {
354 next_split = Some((idx, self.string.char_range_at(idx).next));
360 Some((a, b)) => unsafe {
361 let elt = raw::slice_unchecked(self.string, b, len);
362 self.string = raw::slice_unchecked(self.string, 0, a);
365 None => { self.finished = true; Some(self.string) }
370 impl<'a, Sep: CharEq> Iterator<&'a str> for CharSplitsN<'a, Sep> {
372 fn next(&mut self) -> Option<&'a str> {
375 if self.invert { self.iter.next_back() } else { self.iter.next() }
382 /// The internal state of an iterator that searches for matches of a substring
383 /// within a larger string using naive search
385 struct NaiveSearcher {
390 fn new() -> NaiveSearcher {
391 NaiveSearcher { position: 0 }
394 fn next(&mut self, haystack: &[u8], needle: &[u8]) -> Option<(uint, uint)> {
395 while self.position + needle.len() <= haystack.len() {
396 if haystack.slice(self.position, self.position + needle.len()) == needle {
397 let match_pos = self.position;
398 self.position += needle.len(); // add 1 for all matches
399 return Some((match_pos, match_pos + needle.len()));
408 /// The internal state of an iterator that searches for matches of a substring
409 /// within a larger string using two-way search
411 struct TwoWaySearcher {
422 // This is the Two-Way search algorithm, which was introduced in the paper:
423 // Crochemore, M., Perrin, D., 1991, Two-way string-matching, Journal of the ACM 38(3):651-675.
424 impl TwoWaySearcher {
425 fn new(needle: &[u8]) -> TwoWaySearcher {
426 let (crit_pos1, period1) = TwoWaySearcher::maximal_suffix(needle, false);
427 let (crit_pos2, period2) = TwoWaySearcher::maximal_suffix(needle, true);
431 if crit_pos1 > crit_pos2 {
432 crit_pos = crit_pos1;
435 crit_pos = crit_pos2;
439 let byteset = needle.iter()
440 .fold(0, |a, &b| (1 << ((b & 0x3f) as uint)) | a);
442 // The logic here (calculating crit_pos and period, the final if statement to see which
443 // period to use for the TwoWaySearcher) is essentially an implementation of the
444 // "small-period" function from the paper (p. 670)
446 // In the paper they check whether `needle.slice_to(crit_pos)` is a suffix of
447 // `needle.slice(crit_pos, crit_pos + period)`, which is precisely what this does
448 if needle.slice_to(crit_pos) == needle.slice(period, period + crit_pos) {
460 period: cmp::max(crit_pos, needle.len() - crit_pos) + 1,
464 memory: uint::MAX // Dummy value to signify that the period is long
470 fn next(&mut self, haystack: &[u8], needle: &[u8], long_period: bool) -> Option<(uint, uint)> {
472 // Check that we have room to search in
473 if self.position + needle.len() > haystack.len() {
477 // Quickly skip by large portions unrelated to our substring
479 ((haystack[self.position + needle.len() - 1] & 0x3f)
481 self.position += needle.len();
485 // See if the right part of the needle matches
486 let start = if long_period { self.crit_pos }
487 else { cmp::max(self.crit_pos, self.memory) };
488 for i in range(start, needle.len()) {
489 if needle[i] != haystack[self.position + i] {
490 self.position += i - self.crit_pos + 1;
498 // See if the left part of the needle matches
499 let start = if long_period { 0 } else { self.memory };
500 for i in range(start, self.crit_pos).rev() {
501 if needle[i] != haystack[self.position + i] {
502 self.position += self.period;
504 self.memory = needle.len() - self.period;
510 // We have found a match!
511 let match_pos = self.position;
512 self.position += needle.len(); // add self.period for all matches
514 self.memory = 0; // set to needle.len() - self.period for all matches
516 return Some((match_pos, match_pos + needle.len()));
520 // returns (i, p) where i is the "critical position", the starting index of
521 // of maximal suffix, and p is the period of the suffix
522 // see p. 668 of the paper
524 fn maximal_suffix(arr: &[u8], reversed: bool) -> (uint, uint) {
525 let mut left = -1; // Corresponds to i in the paper
526 let mut right = 0; // Corresponds to j in the paper
527 let mut offset = 1; // Corresponds to k in the paper
528 let mut period = 1; // Corresponds to p in the paper
530 while right + offset < arr.len() {
534 a = arr[left + offset];
535 b = arr[right + offset];
537 a = arr[right + offset];
538 b = arr[left + offset];
541 // Suffix is smaller, period is entire prefix so far.
544 period = right - left;
546 // Advance through repetition of the current period.
547 if offset == period {
554 // Suffix is larger, start over from current location.
565 /// The internal state of an iterator that searches for matches of a substring
566 /// within a larger string using a dynamically chosen search algorithm
569 Naive(NaiveSearcher),
570 TwoWay(TwoWaySearcher),
571 TwoWayLong(TwoWaySearcher)
575 fn new(haystack: &[u8], needle: &[u8]) -> Searcher {
577 if needle.len() + 20 > haystack.len() {
578 Naive(NaiveSearcher::new())
580 let searcher = TwoWaySearcher::new(needle);
581 if searcher.memory == uint::MAX { // If the period is long
590 /// An iterator over the start and end indices of the matches of a
591 /// substring within a larger string
593 pub struct MatchIndices<'a> {
600 /// An iterator over the substrings of a string separated by a given
603 pub struct StrSplits<'a> {
604 it: MatchIndices<'a>,
609 impl<'a> Iterator<(uint, uint)> for MatchIndices<'a> {
611 fn next(&mut self) -> Option<(uint, uint)> {
612 match self.searcher {
613 Naive(ref mut searcher)
614 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes()),
615 TwoWay(ref mut searcher)
616 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes(), false),
617 TwoWayLong(ref mut searcher)
618 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes(), true)
623 impl<'a> Iterator<&'a str> for StrSplits<'a> {
625 fn next(&mut self) -> Option<&'a str> {
626 if self.finished { return None; }
628 match self.it.next() {
629 Some((from, to)) => {
630 let ret = Some(self.it.haystack.slice(self.last_end, from));
635 self.finished = true;
636 Some(self.it.haystack.slice(self.last_end, self.it.haystack.len()))
642 /// External iterator for a string's UTF16 codeunits.
643 /// Use with the `std::iter` module.
645 pub struct Utf16CodeUnits<'a> {
650 impl<'a> Iterator<u16> for Utf16CodeUnits<'a> {
652 fn next(&mut self) -> Option<u16> {
654 let tmp = self.extra;
659 let mut buf = [0u16, ..2];
660 self.chars.next().map(|ch| {
661 let n = ch.encode_utf16(buf.as_mut_slice()).unwrap_or(0);
662 if n == 2 { self.extra = buf[1]; }
668 fn size_hint(&self) -> (uint, Option<uint>) {
669 let (low, high) = self.chars.size_hint();
670 // every char gets either one u16 or two u16,
671 // so this iterator is between 1 or 2 times as
672 // long as the underlying iterator.
673 (low, high.and_then(|n| n.checked_mul(&2)))
678 Section: Comparing strings
681 // share the implementation of the lang-item vs. non-lang-item
683 /// NOTE: This function is (ab)used in rustc::middle::trans::_match
684 /// to compare &[u8] byte slices that are not necessarily valid UTF-8.
686 fn eq_slice_(a: &str, b: &str) -> bool {
688 extern { fn memcmp(s1: *const i8, s2: *const i8, n: uint) -> i32; }
689 a.len() == b.len() && unsafe {
690 memcmp(a.as_ptr() as *const i8,
691 b.as_ptr() as *const i8,
696 /// Bytewise slice equality
697 /// NOTE: This function is (ab)used in rustc::middle::trans::_match
698 /// to compare &[u8] byte slices that are not necessarily valid UTF-8.
701 pub fn eq_slice(a: &str, b: &str) -> bool {
709 /// Walk through `iter` checking that it's a valid UTF-8 sequence,
710 /// returning `true` in that case, or, if it is invalid, `false` with
711 /// `iter` reset such that it is pointing at the first byte in the
712 /// invalid sequence.
714 fn run_utf8_validation_iterator(iter: &mut slice::Items<u8>) -> bool {
716 // save the current thing we're pointing at.
719 // restore the iterator we had at the start of this codepoint.
720 macro_rules! err ( () => { {*iter = old; return false} });
721 macro_rules! next ( () => {
724 // we needed data, but there was none: error!
729 let first = match iter.next() {
731 // we're at the end of the iterator and a codepoint
732 // boundary at the same time, so this string is valid.
736 // ASCII characters are always valid, so only large
737 // bytes need more examination.
739 let w = utf8_char_width(first);
740 let second = next!();
741 // 2-byte encoding is for codepoints \u0080 to \u07ff
742 // first C2 80 last DF BF
743 // 3-byte encoding is for codepoints \u0800 to \uffff
744 // first E0 A0 80 last EF BF BF
745 // excluding surrogates codepoints \ud800 to \udfff
746 // ED A0 80 to ED BF BF
747 // 4-byte encoding is for codepoints \u10000 to \u10ffff
748 // first F0 90 80 80 last F4 8F BF BF
750 // Use the UTF-8 syntax from the RFC
752 // https://tools.ietf.org/html/rfc3629
754 // UTF8-2 = %xC2-DF UTF8-tail
755 // UTF8-3 = %xE0 %xA0-BF UTF8-tail / %xE1-EC 2( UTF8-tail ) /
756 // %xED %x80-9F UTF8-tail / %xEE-EF 2( UTF8-tail )
757 // UTF8-4 = %xF0 %x90-BF 2( UTF8-tail ) / %xF1-F3 3( UTF8-tail ) /
758 // %xF4 %x80-8F 2( UTF8-tail )
760 2 => if second & !CONT_MASK != TAG_CONT_U8 {err!()},
762 match (first, second, next!() & !CONT_MASK) {
763 (0xE0 , 0xA0 .. 0xBF, TAG_CONT_U8) |
764 (0xE1 .. 0xEC, 0x80 .. 0xBF, TAG_CONT_U8) |
765 (0xED , 0x80 .. 0x9F, TAG_CONT_U8) |
766 (0xEE .. 0xEF, 0x80 .. 0xBF, TAG_CONT_U8) => {}
771 match (first, second, next!() & !CONT_MASK, next!() & !CONT_MASK) {
772 (0xF0 , 0x90 .. 0xBF, TAG_CONT_U8, TAG_CONT_U8) |
773 (0xF1 .. 0xF3, 0x80 .. 0xBF, TAG_CONT_U8, TAG_CONT_U8) |
774 (0xF4 , 0x80 .. 0x8F, TAG_CONT_U8, TAG_CONT_U8) => {}
784 /// Determines if a vector of bytes contains valid UTF-8.
785 pub fn is_utf8(v: &[u8]) -> bool {
786 run_utf8_validation_iterator(&mut v.iter())
789 /// Determines if a vector of `u16` contains valid UTF-16
790 pub fn is_utf16(v: &[u16]) -> bool {
791 let mut it = v.iter();
792 macro_rules! next ( ($ret:expr) => {
793 match it.next() { Some(u) => *u, None => return $ret }
799 match char::from_u32(u as u32) {
802 let u2 = next!(false);
803 if u < 0xD7FF || u > 0xDBFF ||
804 u2 < 0xDC00 || u2 > 0xDFFF { return false; }
810 /// An iterator that decodes UTF-16 encoded codepoints from a vector
813 pub struct Utf16Items<'a> {
814 iter: slice::Items<'a, u16>
816 /// The possibilities for values decoded from a `u16` stream.
817 #[deriving(PartialEq, Eq, Clone, Show)]
819 /// A valid codepoint.
821 /// An invalid surrogate without its pair.
826 /// Convert `self` to a `char`, taking `LoneSurrogate`s to the
827 /// replacement character (U+FFFD).
829 pub fn to_char_lossy(&self) -> char {
832 LoneSurrogate(_) => '\uFFFD'
837 impl<'a> Iterator<Utf16Item> for Utf16Items<'a> {
838 fn next(&mut self) -> Option<Utf16Item> {
839 let u = match self.iter.next() {
844 if u < 0xD800 || 0xDFFF < u {
846 Some(ScalarValue(unsafe {mem::transmute(u as u32)}))
847 } else if u >= 0xDC00 {
848 // a trailing surrogate
849 Some(LoneSurrogate(u))
851 // preserve state for rewinding.
854 let u2 = match self.iter.next() {
857 None => return Some(LoneSurrogate(u))
859 if u2 < 0xDC00 || u2 > 0xDFFF {
860 // not a trailing surrogate so we're not a valid
861 // surrogate pair, so rewind to redecode u2 next time.
863 return Some(LoneSurrogate(u))
866 // all ok, so lets decode it.
867 let c = ((u - 0xD800) as u32 << 10 | (u2 - 0xDC00) as u32) + 0x1_0000;
868 Some(ScalarValue(unsafe {mem::transmute(c)}))
873 fn size_hint(&self) -> (uint, Option<uint>) {
874 let (low, high) = self.iter.size_hint();
875 // we could be entirely valid surrogates (2 elements per
876 // char), or entirely non-surrogates (1 element per char)
881 /// Create an iterator over the UTF-16 encoded codepoints in `v`,
882 /// returning invalid surrogates as `LoneSurrogate`s.
888 /// use std::str::{ScalarValue, LoneSurrogate};
890 /// // 𝄞mus<invalid>ic<invalid>
891 /// let v = [0xD834, 0xDD1E, 0x006d, 0x0075,
892 /// 0x0073, 0xDD1E, 0x0069, 0x0063,
895 /// assert_eq!(str::utf16_items(v).collect::<Vec<_>>(),
896 /// vec![ScalarValue('𝄞'),
897 /// ScalarValue('m'), ScalarValue('u'), ScalarValue('s'),
898 /// LoneSurrogate(0xDD1E),
899 /// ScalarValue('i'), ScalarValue('c'),
900 /// LoneSurrogate(0xD834)]);
902 pub fn utf16_items<'a>(v: &'a [u16]) -> Utf16Items<'a> {
903 Utf16Items { iter : v.iter() }
906 /// Return a slice of `v` ending at (and not including) the first NUL
915 /// let mut v = ['a' as u16, 'b' as u16, 'c' as u16, 'd' as u16];
916 /// // no NULs so no change
917 /// assert_eq!(str::truncate_utf16_at_nul(v), v.as_slice());
921 /// let b: &[_] = &['a' as u16, 'b' as u16];
922 /// assert_eq!(str::truncate_utf16_at_nul(v), b);
924 pub fn truncate_utf16_at_nul<'a>(v: &'a [u16]) -> &'a [u16] {
925 match v.iter().position(|c| *c == 0) {
926 // don't include the 0
927 Some(i) => v.slice_to(i),
932 // https://tools.ietf.org/html/rfc3629
933 static UTF8_CHAR_WIDTH: [u8, ..256] = [
934 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
935 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x1F
936 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
937 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x3F
938 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
939 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x5F
940 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
941 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x7F
942 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
943 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0x9F
944 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
945 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0xBF
946 0,0,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
947 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, // 0xDF
948 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, // 0xEF
949 4,4,4,4,4,0,0,0,0,0,0,0,0,0,0,0, // 0xFF
952 /// Given a first byte, determine how many bytes are in this UTF-8 character
954 pub fn utf8_char_width(b: u8) -> uint {
955 return UTF8_CHAR_WIDTH[b as uint] as uint;
958 /// Struct that contains a `char` and the index of the first byte of
959 /// the next `char` in a string. This can be used as a data structure
960 /// for iterating over the UTF-8 bytes of a string.
961 pub struct CharRange {
964 /// Index of the first byte of the next `char`
968 /// Mask of the value bits of a continuation byte
969 static CONT_MASK: u8 = 0b0011_1111u8;
970 /// Value of the tag bits (tag mask is !CONT_MASK) of a continuation byte
971 static TAG_CONT_U8: u8 = 0b1000_0000u8;
973 /// Unsafe operations
976 use collections::Collection;
979 use slice::{ImmutableSlice};
980 use str::{is_utf8, StrSlice};
982 /// Converts a slice of bytes to a string slice without checking
983 /// that the string contains valid UTF-8.
984 pub unsafe fn from_utf8<'a>(v: &'a [u8]) -> &'a str {
988 /// Form a slice from a C string. Unsafe because the caller must ensure the
989 /// C string has the static lifetime, or else the return value may be
990 /// invalidated later.
991 pub unsafe fn c_str_to_static_slice(s: *const i8) -> &'static str {
992 let s = s as *const u8;
997 curr = s.offset(len as int);
999 let v = Slice { data: s, len: len };
1000 assert!(is_utf8(::mem::transmute(v)));
1004 /// Takes a bytewise (not UTF-8) slice from a string.
1006 /// Returns the substring from [`begin`..`end`).
1010 /// If begin is greater than end.
1011 /// If end is greater than the length of the string.
1013 pub unsafe fn slice_bytes<'a>(s: &'a str, begin: uint, end: uint) -> &'a str {
1014 assert!(begin <= end);
1015 assert!(end <= s.len());
1016 slice_unchecked(s, begin, end)
1019 /// Takes a bytewise (not UTF-8) slice from a string.
1021 /// Returns the substring from [`begin`..`end`).
1023 /// Caller must check slice boundaries!
1025 pub unsafe fn slice_unchecked<'a>(s: &'a str, begin: uint, end: uint) -> &'a str {
1026 mem::transmute(Slice {
1027 data: s.as_ptr().offset(begin as int),
1034 Section: Trait implementations
1037 #[allow(missing_doc)]
1039 use cmp::{Ord, Ordering, Less, Equal, Greater, PartialEq, PartialOrd, Equiv, Eq};
1040 use collections::Collection;
1042 use option::{Option, Some};
1043 use str::{Str, StrSlice, eq_slice};
1045 impl<'a> Ord for &'a str {
1047 fn cmp(&self, other: & &'a str) -> Ordering {
1048 for (s_b, o_b) in self.bytes().zip(other.bytes()) {
1049 match s_b.cmp(&o_b) {
1050 Greater => return Greater,
1051 Less => return Less,
1056 self.len().cmp(&other.len())
1060 impl<'a> PartialEq for &'a str {
1062 fn eq(&self, other: & &'a str) -> bool {
1063 eq_slice((*self), (*other))
1066 fn ne(&self, other: & &'a str) -> bool { !(*self).eq(other) }
1069 impl<'a> Eq for &'a str {}
1071 impl<'a> PartialOrd for &'a str {
1073 fn partial_cmp(&self, other: &&'a str) -> Option<Ordering> {
1074 Some(self.cmp(other))
1078 impl<'a, S: Str> Equiv<S> for &'a str {
1080 fn equiv(&self, other: &S) -> bool { eq_slice(*self, other.as_slice()) }
1084 /// Any string that can be represented as a slice
1086 /// Work with `self` as a slice.
1087 fn as_slice<'a>(&'a self) -> &'a str;
1090 impl<'a> Str for &'a str {
1092 fn as_slice<'a>(&'a self) -> &'a str { *self }
1095 impl<'a> Collection for &'a str {
1097 fn len(&self) -> uint {
1102 /// Methods for string slices
1103 pub trait StrSlice<'a> {
1104 /// Returns true if one string contains another
1108 /// - needle - The string to look for
1113 /// assert!("bananas".contains("nana"));
1115 fn contains<'a>(&self, needle: &'a str) -> bool;
1117 /// Returns true if a string contains a char.
1121 /// - needle - The char to look for
1126 /// assert!("hello".contains_char('e'));
1128 fn contains_char(&self, needle: char) -> bool;
1130 /// An iterator over the characters of `self`. Note, this iterates
1131 /// over Unicode code-points, not Unicode graphemes.
1136 /// let v: Vec<char> = "abc åäö".chars().collect();
1137 /// assert_eq!(v, vec!['a', 'b', 'c', ' ', 'å', 'ä', 'ö']);
1139 fn chars(&self) -> Chars<'a>;
1141 /// An iterator over the bytes of `self`
1146 /// let v: Vec<u8> = "bors".bytes().collect();
1147 /// assert_eq!(v, b"bors".to_vec());
1149 fn bytes(&self) -> Bytes<'a>;
1151 /// An iterator over the characters of `self` and their byte offsets.
1152 fn char_indices(&self) -> CharOffsets<'a>;
1154 /// An iterator over substrings of `self`, separated by characters
1155 /// matched by `sep`.
1160 /// let v: Vec<&str> = "Mary had a little lamb".split(' ').collect();
1161 /// assert_eq!(v, vec!["Mary", "had", "a", "little", "lamb"]);
1163 /// let v: Vec<&str> = "abc1def2ghi".split(|c: char| c.is_digit()).collect();
1164 /// assert_eq!(v, vec!["abc", "def", "ghi"]);
1166 /// let v: Vec<&str> = "lionXXtigerXleopard".split('X').collect();
1167 /// assert_eq!(v, vec!["lion", "", "tiger", "leopard"]);
1169 /// let v: Vec<&str> = "".split('X').collect();
1170 /// assert_eq!(v, vec![""]);
1172 fn split<Sep: CharEq>(&self, sep: Sep) -> CharSplits<'a, Sep>;
1174 /// An iterator over substrings of `self`, separated by characters
1175 /// matched by `sep`, restricted to splitting at most `count`
1181 /// let v: Vec<&str> = "Mary had a little lambda".splitn(2, ' ').collect();
1182 /// assert_eq!(v, vec!["Mary", "had", "a little lambda"]);
1184 /// let v: Vec<&str> = "abc1def2ghi".splitn(1, |c: char| c.is_digit()).collect();
1185 /// assert_eq!(v, vec!["abc", "def2ghi"]);
1187 /// let v: Vec<&str> = "lionXXtigerXleopard".splitn(2, 'X').collect();
1188 /// assert_eq!(v, vec!["lion", "", "tigerXleopard"]);
1190 /// let v: Vec<&str> = "abcXdef".splitn(0, 'X').collect();
1191 /// assert_eq!(v, vec!["abcXdef"]);
1193 /// let v: Vec<&str> = "".splitn(1, 'X').collect();
1194 /// assert_eq!(v, vec![""]);
1196 fn splitn<Sep: CharEq>(&self, count: uint, sep: Sep) -> CharSplitsN<'a, Sep>;
1198 /// An iterator over substrings of `self`, separated by characters
1199 /// matched by `sep`.
1201 /// Equivalent to `split`, except that the trailing substring
1202 /// is skipped if empty (terminator semantics).
1207 /// let v: Vec<&str> = "A.B.".split_terminator('.').collect();
1208 /// assert_eq!(v, vec!["A", "B"]);
1210 /// let v: Vec<&str> = "A..B..".split_terminator('.').collect();
1211 /// assert_eq!(v, vec!["A", "", "B", ""]);
1213 /// let v: Vec<&str> = "Mary had a little lamb".split(' ').rev().collect();
1214 /// assert_eq!(v, vec!["lamb", "little", "a", "had", "Mary"]);
1216 /// let v: Vec<&str> = "abc1def2ghi".split(|c: char| c.is_digit()).rev().collect();
1217 /// assert_eq!(v, vec!["ghi", "def", "abc"]);
1219 /// let v: Vec<&str> = "lionXXtigerXleopard".split('X').rev().collect();
1220 /// assert_eq!(v, vec!["leopard", "tiger", "", "lion"]);
1222 fn split_terminator<Sep: CharEq>(&self, sep: Sep) -> CharSplits<'a, Sep>;
1224 /// An iterator over substrings of `self`, separated by characters
1225 /// matched by `sep`, starting from the end of the string.
1226 /// Restricted to splitting at most `count` times.
1231 /// let v: Vec<&str> = "Mary had a little lamb".rsplitn(2, ' ').collect();
1232 /// assert_eq!(v, vec!["lamb", "little", "Mary had a"]);
1234 /// let v: Vec<&str> = "abc1def2ghi".rsplitn(1, |c: char| c.is_digit()).collect();
1235 /// assert_eq!(v, vec!["ghi", "abc1def"]);
1237 /// let v: Vec<&str> = "lionXXtigerXleopard".rsplitn(2, 'X').collect();
1238 /// assert_eq!(v, vec!["leopard", "tiger", "lionX"]);
1240 fn rsplitn<Sep: CharEq>(&self, count: uint, sep: Sep) -> CharSplitsN<'a, Sep>;
1242 /// An iterator over the start and end indices of the disjoint
1243 /// matches of `sep` within `self`.
1245 /// That is, each returned value `(start, end)` satisfies
1246 /// `self.slice(start, end) == sep`. For matches of `sep` within
1247 /// `self` that overlap, only the indices corresponding to the
1248 /// first match are returned.
1253 /// let v: Vec<(uint, uint)> = "abcXXXabcYYYabc".match_indices("abc").collect();
1254 /// assert_eq!(v, vec![(0,3), (6,9), (12,15)]);
1256 /// let v: Vec<(uint, uint)> = "1abcabc2".match_indices("abc").collect();
1257 /// assert_eq!(v, vec![(1,4), (4,7)]);
1259 /// let v: Vec<(uint, uint)> = "ababa".match_indices("aba").collect();
1260 /// assert_eq!(v, vec![(0, 3)]); // only the first `aba`
1262 fn match_indices(&self, sep: &'a str) -> MatchIndices<'a>;
1264 /// An iterator over the substrings of `self` separated by `sep`.
1269 /// let v: Vec<&str> = "abcXXXabcYYYabc".split_str("abc").collect();
1270 /// assert_eq!(v, vec!["", "XXX", "YYY", ""]);
1272 /// let v: Vec<&str> = "1abcabc2".split_str("abc").collect();
1273 /// assert_eq!(v, vec!["1", "", "2"]);
1275 fn split_str(&self, &'a str) -> StrSplits<'a>;
1277 /// An iterator over the lines of a string (subsequences separated
1278 /// by `\n`). This does not include the empty string after a
1284 /// let four_lines = "foo\nbar\n\nbaz\n";
1285 /// let v: Vec<&str> = four_lines.lines().collect();
1286 /// assert_eq!(v, vec!["foo", "bar", "", "baz"]);
1288 fn lines(&self) -> CharSplits<'a, char>;
1290 /// An iterator over the lines of a string, separated by either
1291 /// `\n` or `\r\n`. As with `.lines()`, this does not include an
1292 /// empty trailing line.
1297 /// let four_lines = "foo\r\nbar\n\r\nbaz\n";
1298 /// let v: Vec<&str> = four_lines.lines_any().collect();
1299 /// assert_eq!(v, vec!["foo", "bar", "", "baz"]);
1301 fn lines_any(&self) -> AnyLines<'a>;
1303 /// Returns the number of Unicode code points (`char`) that a
1306 /// This does not perform any normalization, and is `O(n)`, since
1307 /// UTF-8 is a variable width encoding of code points.
1309 /// *Warning*: The number of code points in a string does not directly
1310 /// correspond to the number of visible characters or width of the
1311 /// visible text due to composing characters, and double- and
1312 /// zero-width ones.
1314 /// See also `.len()` for the byte length.
1319 /// // composed forms of `ö` and `é`
1320 /// let c = "Löwe 老虎 Léopard"; // German, Simplified Chinese, French
1321 /// // decomposed forms of `ö` and `é`
1322 /// let d = "Lo\u0308we 老虎 Le\u0301opard";
1324 /// assert_eq!(c.char_len(), 15);
1325 /// assert_eq!(d.char_len(), 17);
1327 /// assert_eq!(c.len(), 21);
1328 /// assert_eq!(d.len(), 23);
1330 /// // the two strings *look* the same
1331 /// println!("{}", c);
1332 /// println!("{}", d);
1334 fn char_len(&self) -> uint;
1336 /// Returns a slice of the given string from the byte range
1337 /// [`begin`..`end`).
1339 /// This operation is `O(1)`.
1341 /// Fails when `begin` and `end` do not point to valid characters
1342 /// or point beyond the last character of the string.
1344 /// See also `slice_to` and `slice_from` for slicing prefixes and
1345 /// suffixes of strings, and `slice_chars` for slicing based on
1346 /// code point counts.
1351 /// let s = "Löwe 老虎 Léopard";
1352 /// assert_eq!(s.slice(0, 1), "L");
1354 /// assert_eq!(s.slice(1, 9), "öwe 老");
1356 /// // these will fail:
1357 /// // byte 2 lies within `ö`:
1358 /// // s.slice(2, 3);
1360 /// // byte 8 lies within `老`
1361 /// // s.slice(1, 8);
1363 /// // byte 100 is outside the string
1364 /// // s.slice(3, 100);
1366 fn slice(&self, begin: uint, end: uint) -> &'a str;
1368 /// Returns a slice of the string from `begin` to its end.
1370 /// Equivalent to `self.slice(begin, self.len())`.
1372 /// Fails when `begin` does not point to a valid character, or is
1375 /// See also `slice`, `slice_to` and `slice_chars`.
1376 fn slice_from(&self, begin: uint) -> &'a str;
1378 /// Returns a slice of the string from the beginning to byte
1381 /// Equivalent to `self.slice(0, end)`.
1383 /// Fails when `end` does not point to a valid character, or is
1386 /// See also `slice`, `slice_from` and `slice_chars`.
1387 fn slice_to(&self, end: uint) -> &'a str;
1389 /// Returns a slice of the string from the character range
1390 /// [`begin`..`end`).
1392 /// That is, start at the `begin`-th code point of the string and
1393 /// continue to the `end`-th code point. This does not detect or
1394 /// handle edge cases such as leaving a combining character as the
1395 /// first code point of the string.
1397 /// Due to the design of UTF-8, this operation is `O(end)`.
1398 /// See `slice`, `slice_to` and `slice_from` for `O(1)`
1399 /// variants that use byte indices rather than code point
1402 /// Fails if `begin` > `end` or the either `begin` or `end` are
1403 /// beyond the last character of the string.
1408 /// let s = "Löwe 老虎 Léopard";
1409 /// assert_eq!(s.slice_chars(0, 4), "Löwe");
1410 /// assert_eq!(s.slice_chars(5, 7), "老虎");
1412 fn slice_chars(&self, begin: uint, end: uint) -> &'a str;
1414 /// Returns true if `needle` is a prefix of the string.
1419 /// assert!("banana".starts_with("ba"));
1421 fn starts_with(&self, needle: &str) -> bool;
1423 /// Returns true if `needle` is a suffix of the string.
1428 /// assert!("banana".ends_with("nana"));
1430 fn ends_with(&self, needle: &str) -> bool;
1432 /// Returns a string with characters that match `to_trim` removed.
1436 /// * to_trim - a character matcher
1441 /// assert_eq!("11foo1bar11".trim_chars('1'), "foo1bar")
1442 /// let x: &[_] = &['1', '2'];
1443 /// assert_eq!("12foo1bar12".trim_chars(x), "foo1bar")
1444 /// assert_eq!("123foo1bar123".trim_chars(|c: char| c.is_digit()), "foo1bar")
1446 fn trim_chars<C: CharEq>(&self, to_trim: C) -> &'a str;
1448 /// Returns a string with leading `chars_to_trim` removed.
1452 /// * to_trim - a character matcher
1457 /// assert_eq!("11foo1bar11".trim_left_chars('1'), "foo1bar11")
1458 /// let x: &[_] = &['1', '2'];
1459 /// assert_eq!("12foo1bar12".trim_left_chars(x), "foo1bar12")
1460 /// assert_eq!("123foo1bar123".trim_left_chars(|c: char| c.is_digit()), "foo1bar123")
1462 fn trim_left_chars<C: CharEq>(&self, to_trim: C) -> &'a str;
1464 /// Returns a string with trailing `chars_to_trim` removed.
1468 /// * to_trim - a character matcher
1473 /// assert_eq!("11foo1bar11".trim_right_chars('1'), "11foo1bar")
1474 /// let x: &[_] = &['1', '2'];
1475 /// assert_eq!("12foo1bar12".trim_right_chars(x), "12foo1bar")
1476 /// assert_eq!("123foo1bar123".trim_right_chars(|c: char| c.is_digit()), "123foo1bar")
1478 fn trim_right_chars<C: CharEq>(&self, to_trim: C) -> &'a str;
1480 /// Check that `index`-th byte lies at the start and/or end of a
1481 /// UTF-8 code point sequence.
1483 /// The start and end of the string (when `index == self.len()`)
1484 /// are considered to be boundaries.
1486 /// Fails if `index` is greater than `self.len()`.
1491 /// let s = "Löwe 老虎 Léopard";
1492 /// assert!(s.is_char_boundary(0));
1494 /// assert!(s.is_char_boundary(6));
1495 /// assert!(s.is_char_boundary(s.len()));
1497 /// // second byte of `ö`
1498 /// assert!(!s.is_char_boundary(2));
1500 /// // third byte of `老`
1501 /// assert!(!s.is_char_boundary(8));
1503 fn is_char_boundary(&self, index: uint) -> bool;
1505 /// Pluck a character out of a string and return the index of the next
1508 /// This function can be used to iterate over the Unicode characters of a
1513 /// This example manually iterates through the characters of a
1514 /// string; this should normally be done by `.chars()` or
1515 /// `.char_indices`.
1518 /// use std::str::CharRange;
1520 /// let s = "中华Việt Nam";
1522 /// while i < s.len() {
1523 /// let CharRange {ch, next} = s.char_range_at(i);
1524 /// println!("{}: {}", i, ch);
1546 /// * s - The string
1547 /// * i - The byte offset of the char to extract
1551 /// A record {ch: char, next: uint} containing the char value and the byte
1552 /// index of the next Unicode character.
1556 /// If `i` is greater than or equal to the length of the string.
1557 /// If `i` is not the index of the beginning of a valid UTF-8 character.
1558 fn char_range_at(&self, start: uint) -> CharRange;
1560 /// Given a byte position and a str, return the previous char and its position.
1562 /// This function can be used to iterate over a Unicode string in reverse.
1564 /// Returns 0 for next index if called on start index 0.
1568 /// If `i` is greater than the length of the string.
1569 /// If `i` is not an index following a valid UTF-8 character.
1570 fn char_range_at_reverse(&self, start: uint) -> CharRange;
1572 /// Plucks the character starting at the `i`th byte of a string.
1578 /// assert_eq!(s.char_at(1), 'b');
1579 /// assert_eq!(s.char_at(2), 'π');
1580 /// assert_eq!(s.char_at(4), 'c');
1585 /// If `i` is greater than or equal to the length of the string.
1586 /// If `i` is not the index of the beginning of a valid UTF-8 character.
1587 fn char_at(&self, i: uint) -> char;
1589 /// Plucks the character ending at the `i`th byte of a string.
1593 /// If `i` is greater than the length of the string.
1594 /// If `i` is not an index following a valid UTF-8 character.
1595 fn char_at_reverse(&self, i: uint) -> char;
1597 /// Work with the byte buffer of a string as a byte slice.
1602 /// assert_eq!("bors".as_bytes(), b"bors");
1604 fn as_bytes(&self) -> &'a [u8];
1606 /// Returns the byte index of the first character of `self` that
1607 /// matches `search`.
1611 /// `Some` containing the byte index of the last matching character
1612 /// or `None` if there is no match
1617 /// let s = "Löwe 老虎 Léopard";
1619 /// assert_eq!(s.find('L'), Some(0));
1620 /// assert_eq!(s.find('é'), Some(14));
1622 /// // the first space
1623 /// assert_eq!(s.find(|c: char| c.is_whitespace()), Some(5));
1625 /// // neither are found
1626 /// let x: &[_] = &['1', '2'];
1627 /// assert_eq!(s.find(x), None);
1629 fn find<C: CharEq>(&self, search: C) -> Option<uint>;
1631 /// Returns the byte index of the last character of `self` that
1632 /// matches `search`.
1636 /// `Some` containing the byte index of the last matching character
1637 /// or `None` if there is no match.
1642 /// let s = "Löwe 老虎 Léopard";
1644 /// assert_eq!(s.rfind('L'), Some(13));
1645 /// assert_eq!(s.rfind('é'), Some(14));
1647 /// // the second space
1648 /// assert_eq!(s.rfind(|c: char| c.is_whitespace()), Some(12));
1650 /// // searches for an occurrence of either `1` or `2`, but neither are found
1651 /// let x: &[_] = &['1', '2'];
1652 /// assert_eq!(s.rfind(x), None);
1654 fn rfind<C: CharEq>(&self, search: C) -> Option<uint>;
1656 /// Returns the byte index of the first matching substring
1660 /// * `needle` - The string to search for
1664 /// `Some` containing the byte index of the first matching substring
1665 /// or `None` if there is no match.
1670 /// let s = "Löwe 老虎 Léopard";
1672 /// assert_eq!(s.find_str("老虎 L"), Some(6));
1673 /// assert_eq!(s.find_str("muffin man"), None);
1675 fn find_str(&self, &str) -> Option<uint>;
1677 /// Retrieves the first character from a string slice and returns
1678 /// it. This does not allocate a new string; instead, it returns a
1679 /// slice that point one character beyond the character that was
1680 /// shifted. If the string does not contain any characters,
1681 /// a tuple of None and an empty string is returned instead.
1686 /// let s = "Löwe 老虎 Léopard";
1687 /// let (c, s1) = s.slice_shift_char();
1688 /// assert_eq!(c, Some('L'));
1689 /// assert_eq!(s1, "öwe 老虎 Léopard");
1691 /// let (c, s2) = s1.slice_shift_char();
1692 /// assert_eq!(c, Some('ö'));
1693 /// assert_eq!(s2, "we 老虎 Léopard");
1695 fn slice_shift_char(&self) -> (Option<char>, &'a str);
1697 /// Returns the byte offset of an inner slice relative to an enclosing outer slice.
1699 /// Fails if `inner` is not a direct slice contained within self.
1704 /// let string = "a\nb\nc";
1705 /// let lines: Vec<&str> = string.lines().collect();
1706 /// let lines = lines.as_slice();
1708 /// assert!(string.subslice_offset(lines[0]) == 0); // &"a"
1709 /// assert!(string.subslice_offset(lines[1]) == 2); // &"b"
1710 /// assert!(string.subslice_offset(lines[2]) == 4); // &"c"
1712 fn subslice_offset(&self, inner: &str) -> uint;
1714 /// Return an unsafe pointer to the strings buffer.
1716 /// The caller must ensure that the string outlives this pointer,
1717 /// and that it is not reallocated (e.g. by pushing to the
1719 fn as_ptr(&self) -> *const u8;
1721 /// Return an iterator of `u16` over the string encoded as UTF-16.
1722 fn utf16_units(&self) -> Utf16CodeUnits<'a>;
1726 fn slice_error_fail(s: &str, begin: uint, end: uint) -> ! {
1727 assert!(begin <= end);
1728 fail!("index {} and/or {} in `{}` do not lie on character boundary",
1732 impl<'a> StrSlice<'a> for &'a str {
1734 fn contains<'a>(&self, needle: &'a str) -> bool {
1735 self.find_str(needle).is_some()
1739 fn contains_char(&self, needle: char) -> bool {
1740 self.find(needle).is_some()
1744 fn chars(&self) -> Chars<'a> {
1745 Chars{iter: self.as_bytes().iter()}
1749 fn bytes(&self) -> Bytes<'a> {
1750 self.as_bytes().iter().map(|&b| b)
1754 fn char_indices(&self) -> CharOffsets<'a> {
1755 CharOffsets{front_offset: 0, iter: self.chars()}
1759 fn split<Sep: CharEq>(&self, sep: Sep) -> CharSplits<'a, Sep> {
1762 only_ascii: sep.only_ascii(),
1764 allow_trailing_empty: true,
1770 fn splitn<Sep: CharEq>(&self, count: uint, sep: Sep)
1771 -> CharSplitsN<'a, Sep> {
1773 iter: self.split(sep),
1780 fn split_terminator<Sep: CharEq>(&self, sep: Sep)
1781 -> CharSplits<'a, Sep> {
1783 allow_trailing_empty: false,
1789 fn rsplitn<Sep: CharEq>(&self, count: uint, sep: Sep)
1790 -> CharSplitsN<'a, Sep> {
1792 iter: self.split(sep),
1799 fn match_indices(&self, sep: &'a str) -> MatchIndices<'a> {
1800 assert!(!sep.is_empty())
1804 searcher: Searcher::new(self.as_bytes(), sep.as_bytes())
1809 fn split_str(&self, sep: &'a str) -> StrSplits<'a> {
1811 it: self.match_indices(sep),
1818 fn lines(&self) -> CharSplits<'a, char> {
1819 self.split_terminator('\n')
1822 fn lines_any(&self) -> AnyLines<'a> {
1823 self.lines().map(|line| {
1825 if l > 0 && line.as_bytes()[l - 1] == b'\r' { line.slice(0, l - 1) }
1831 fn char_len(&self) -> uint { self.chars().count() }
1834 fn slice(&self, begin: uint, end: uint) -> &'a str {
1835 // is_char_boundary checks that the index is in [0, .len()]
1837 self.is_char_boundary(begin) &&
1838 self.is_char_boundary(end) {
1839 unsafe { raw::slice_unchecked(*self, begin, end) }
1841 slice_error_fail(*self, begin, end)
1846 fn slice_from(&self, begin: uint) -> &'a str {
1847 // is_char_boundary checks that the index is in [0, .len()]
1848 if self.is_char_boundary(begin) {
1849 unsafe { raw::slice_unchecked(*self, begin, self.len()) }
1851 slice_error_fail(*self, begin, self.len())
1856 fn slice_to(&self, end: uint) -> &'a str {
1857 // is_char_boundary checks that the index is in [0, .len()]
1858 if self.is_char_boundary(end) {
1859 unsafe { raw::slice_unchecked(*self, 0, end) }
1861 slice_error_fail(*self, 0, end)
1865 fn slice_chars(&self, begin: uint, end: uint) -> &'a str {
1866 assert!(begin <= end);
1868 let mut begin_byte = None;
1869 let mut end_byte = None;
1871 // This could be even more efficient by not decoding,
1872 // only finding the char boundaries
1873 for (idx, _) in self.char_indices() {
1874 if count == begin { begin_byte = Some(idx); }
1875 if count == end { end_byte = Some(idx); break; }
1878 if begin_byte.is_none() && count == begin { begin_byte = Some(self.len()) }
1879 if end_byte.is_none() && count == end { end_byte = Some(self.len()) }
1881 match (begin_byte, end_byte) {
1882 (None, _) => fail!("slice_chars: `begin` is beyond end of string"),
1883 (_, None) => fail!("slice_chars: `end` is beyond end of string"),
1884 (Some(a), Some(b)) => unsafe { raw::slice_bytes(*self, a, b) }
1889 fn starts_with<'a>(&self, needle: &'a str) -> bool {
1890 let n = needle.len();
1891 self.len() >= n && needle.as_bytes() == self.as_bytes().slice_to(n)
1895 fn ends_with(&self, needle: &str) -> bool {
1896 let (m, n) = (self.len(), needle.len());
1897 m >= n && needle.as_bytes() == self.as_bytes().slice_from(m - n)
1901 fn trim_chars<C: CharEq>(&self, mut to_trim: C) -> &'a str {
1902 let cur = match self.find(|c: char| !to_trim.matches(c)) {
1904 Some(i) => unsafe { raw::slice_bytes(*self, i, self.len()) }
1906 match cur.rfind(|c: char| !to_trim.matches(c)) {
1909 let right = cur.char_range_at(i).next;
1910 unsafe { raw::slice_bytes(cur, 0, right) }
1916 fn trim_left_chars<C: CharEq>(&self, mut to_trim: C) -> &'a str {
1917 match self.find(|c: char| !to_trim.matches(c)) {
1919 Some(first) => unsafe { raw::slice_bytes(*self, first, self.len()) }
1924 fn trim_right_chars<C: CharEq>(&self, mut to_trim: C) -> &'a str {
1925 match self.rfind(|c: char| !to_trim.matches(c)) {
1928 let next = self.char_range_at(last).next;
1929 unsafe { raw::slice_bytes(*self, 0u, next) }
1935 fn is_char_boundary(&self, index: uint) -> bool {
1936 if index == self.len() { return true; }
1937 match self.as_bytes().get(index) {
1939 Some(&b) => b < 128u8 || b >= 192u8,
1944 fn char_range_at(&self, i: uint) -> CharRange {
1945 if self.as_bytes()[i] < 128u8 {
1946 return CharRange {ch: self.as_bytes()[i] as char, next: i + 1 };
1949 // Multibyte case is a fn to allow char_range_at to inline cleanly
1950 fn multibyte_char_range_at(s: &str, i: uint) -> CharRange {
1951 let mut val = s.as_bytes()[i] as u32;
1952 let w = UTF8_CHAR_WIDTH[val as uint] as uint;
1955 val = utf8_first_byte!(val, w);
1956 val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 1]);
1957 if w > 2 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 2]); }
1958 if w > 3 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 3]); }
1960 return CharRange {ch: unsafe { mem::transmute(val) }, next: i + w};
1963 return multibyte_char_range_at(*self, i);
1967 fn char_range_at_reverse(&self, start: uint) -> CharRange {
1968 let mut prev = start;
1970 prev = prev.saturating_sub(1);
1971 if self.as_bytes()[prev] < 128 {
1972 return CharRange{ch: self.as_bytes()[prev] as char, next: prev}
1975 // Multibyte case is a fn to allow char_range_at_reverse to inline cleanly
1976 fn multibyte_char_range_at_reverse(s: &str, mut i: uint) -> CharRange {
1977 // while there is a previous byte == 10......
1978 while i > 0 && s.as_bytes()[i] & !CONT_MASK == TAG_CONT_U8 {
1982 let mut val = s.as_bytes()[i] as u32;
1983 let w = UTF8_CHAR_WIDTH[val as uint] as uint;
1986 val = utf8_first_byte!(val, w);
1987 val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 1]);
1988 if w > 2 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 2]); }
1989 if w > 3 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 3]); }
1991 return CharRange {ch: unsafe { mem::transmute(val) }, next: i};
1994 return multibyte_char_range_at_reverse(*self, prev);
1998 fn char_at(&self, i: uint) -> char {
1999 self.char_range_at(i).ch
2003 fn char_at_reverse(&self, i: uint) -> char {
2004 self.char_range_at_reverse(i).ch
2008 fn as_bytes(&self) -> &'a [u8] {
2009 unsafe { mem::transmute(*self) }
2012 fn find<C: CharEq>(&self, mut search: C) -> Option<uint> {
2013 if search.only_ascii() {
2014 self.bytes().position(|b| search.matches(b as char))
2016 for (index, c) in self.char_indices() {
2017 if search.matches(c) { return Some(index); }
2023 fn rfind<C: CharEq>(&self, mut search: C) -> Option<uint> {
2024 if search.only_ascii() {
2025 self.bytes().rposition(|b| search.matches(b as char))
2027 for (index, c) in self.char_indices().rev() {
2028 if search.matches(c) { return Some(index); }
2034 fn find_str(&self, needle: &str) -> Option<uint> {
2035 if needle.is_empty() {
2038 self.match_indices(needle)
2040 .map(|(start, _end)| start)
2045 fn slice_shift_char(&self) -> (Option<char>, &'a str) {
2046 if self.is_empty() {
2047 return (None, *self);
2049 let CharRange {ch, next} = self.char_range_at(0u);
2050 let next_s = unsafe { raw::slice_bytes(*self, next, self.len()) };
2051 return (Some(ch), next_s);
2055 fn subslice_offset(&self, inner: &str) -> uint {
2056 let a_start = self.as_ptr() as uint;
2057 let a_end = a_start + self.len();
2058 let b_start = inner.as_ptr() as uint;
2059 let b_end = b_start + inner.len();
2061 assert!(a_start <= b_start);
2062 assert!(b_end <= a_end);
2067 fn as_ptr(&self) -> *const u8 {
2072 fn utf16_units(&self) -> Utf16CodeUnits<'a> {
2073 Utf16CodeUnits{ chars: self.chars(), extra: 0}
2077 impl<'a> Default for &'a str {
2078 fn default() -> &'a str { "" }