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::ImmutableVector;
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 matchPos = self.position;
398 self.position += needle.len(); // add 1 for all matches
399 return Some((matchPos, matchPos + 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 impl TwoWaySearcher {
423 fn new(needle: &[u8]) -> TwoWaySearcher {
424 let (critPos1, period1) = TwoWaySearcher::maximal_suffix(needle, false);
425 let (critPos2, period2) = TwoWaySearcher::maximal_suffix(needle, true);
429 if critPos1 > critPos2 {
437 let byteset = needle.iter()
438 .fold(0, |a, &b| (1 << ((b & 0x3f) as uint)) | a);
440 if needle.slice_to(critPos) == needle.slice_from(needle.len() - critPos) {
452 period: cmp::max(critPos, needle.len() - critPos) + 1,
456 memory: uint::MAX // Dummy value to signify that the period is long
462 fn next(&mut self, haystack: &[u8], needle: &[u8], longPeriod: bool) -> Option<(uint, uint)> {
464 // Check that we have room to search in
465 if self.position + needle.len() > haystack.len() {
469 // Quickly skip by large portions unrelated to our substring
471 ((haystack[self.position + needle.len() - 1] & 0x3f)
473 self.position += needle.len();
477 // See if the right part of the needle matches
478 let start = if longPeriod { self.critPos } else { cmp::max(self.critPos, self.memory) };
479 for i in range(start, needle.len()) {
480 if needle[i] != haystack[self.position + i] {
481 self.position += i - self.critPos + 1;
489 // See if the left part of the needle matches
490 let start = if longPeriod { 0 } else { self.memory };
491 for i in range(start, self.critPos).rev() {
492 if needle[i] != haystack[self.position + i] {
493 self.position += self.period;
495 self.memory = needle.len() - self.period;
501 // We have found a match!
502 let matchPos = self.position;
503 self.position += needle.len(); // add self.period for all matches
505 self.memory = 0; // set to needle.len() - self.period for all matches
507 return Some((matchPos, matchPos + needle.len()));
512 fn maximal_suffix(arr: &[u8], reversed: bool) -> (uint, uint) {
513 let mut left = -1; // Corresponds to i in the paper
514 let mut right = 0; // Corresponds to j in the paper
515 let mut offset = 1; // Corresponds to k in the paper
516 let mut period = 1; // Corresponds to p in the paper
518 while right + offset < arr.len() {
522 a = arr[left + offset];
523 b = arr[right + offset];
525 a = arr[right + offset];
526 b = arr[left + offset];
529 // Suffix is smaller, period is entire prefix so far.
532 period = right - left;
534 // Advance through repetition of the current period.
535 if offset == period {
542 // Suffix is larger, start over from current location.
553 /// The internal state of an iterator that searches for matches of a substring
554 /// within a larger string using a dynamically chosen search algorithm
557 Naive(NaiveSearcher),
558 TwoWay(TwoWaySearcher),
559 TwoWayLong(TwoWaySearcher)
563 fn new(haystack: &[u8], needle: &[u8]) -> Searcher {
565 if needle.len() > haystack.len() - 20 {
566 Naive(NaiveSearcher::new())
568 let searcher = TwoWaySearcher::new(needle);
569 if searcher.memory == uint::MAX { // If the period is long
578 /// An iterator over the start and end indices of the matches of a
579 /// substring within a larger string
581 pub struct MatchIndices<'a> {
588 /// An iterator over the substrings of a string separated by a given
591 pub struct StrSplits<'a> {
592 it: MatchIndices<'a>,
597 impl<'a> Iterator<(uint, uint)> for MatchIndices<'a> {
599 fn next(&mut self) -> Option<(uint, uint)> {
600 match self.searcher {
601 Naive(ref mut searcher)
602 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes()),
603 TwoWay(ref mut searcher)
604 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes(), false),
605 TwoWayLong(ref mut searcher)
606 => searcher.next(self.haystack.as_bytes(), self.needle.as_bytes(), true)
611 impl<'a> Iterator<&'a str> for StrSplits<'a> {
613 fn next(&mut self) -> Option<&'a str> {
614 if self.finished { return None; }
616 match self.it.next() {
617 Some((from, to)) => {
618 let ret = Some(self.it.haystack.slice(self.last_end, from));
623 self.finished = true;
624 Some(self.it.haystack.slice(self.last_end, self.it.haystack.len()))
630 /// External iterator for a string's UTF16 codeunits.
631 /// Use with the `std::iter` module.
633 pub struct Utf16CodeUnits<'a> {
638 impl<'a> Iterator<u16> for Utf16CodeUnits<'a> {
640 fn next(&mut self) -> Option<u16> {
642 let tmp = self.extra;
647 let mut buf = [0u16, ..2];
648 self.chars.next().map(|ch| {
649 let n = ch.encode_utf16(buf /* as mut slice! */);
650 if n == 2 { self.extra = buf[1]; }
656 fn size_hint(&self) -> (uint, Option<uint>) {
657 let (low, high) = self.chars.size_hint();
658 // every char gets either one u16 or two u16,
659 // so this iterator is between 1 or 2 times as
660 // long as the underlying iterator.
661 (low, high.and_then(|n| n.checked_mul(&2)))
666 Section: Comparing strings
669 // share the implementation of the lang-item vs. non-lang-item
671 /// NOTE: This function is (ab)used in rustc::middle::trans::_match
672 /// to compare &[u8] byte slices that are not necessarily valid UTF-8.
674 fn eq_slice_(a: &str, b: &str) -> bool {
676 extern { fn memcmp(s1: *const i8, s2: *const i8, n: uint) -> i32; }
677 a.len() == b.len() && unsafe {
678 memcmp(a.as_ptr() as *const i8,
679 b.as_ptr() as *const i8,
684 /// Bytewise slice equality
685 /// NOTE: This function is (ab)used in rustc::middle::trans::_match
686 /// to compare &[u8] byte slices that are not necessarily valid UTF-8.
689 pub fn eq_slice(a: &str, b: &str) -> bool {
697 /// Walk through `iter` checking that it's a valid UTF-8 sequence,
698 /// returning `true` in that case, or, if it is invalid, `false` with
699 /// `iter` reset such that it is pointing at the first byte in the
700 /// invalid sequence.
702 fn run_utf8_validation_iterator(iter: &mut slice::Items<u8>) -> bool {
704 // save the current thing we're pointing at.
707 // restore the iterator we had at the start of this codepoint.
708 macro_rules! err ( () => { {*iter = old; return false} });
709 macro_rules! next ( () => {
712 // we needed data, but there was none: error!
717 let first = match iter.next() {
719 // we're at the end of the iterator and a codepoint
720 // boundary at the same time, so this string is valid.
724 // ASCII characters are always valid, so only large
725 // bytes need more examination.
727 let w = utf8_char_width(first);
728 let second = next!();
729 // 2-byte encoding is for codepoints \u0080 to \u07ff
730 // first C2 80 last DF BF
731 // 3-byte encoding is for codepoints \u0800 to \uffff
732 // first E0 A0 80 last EF BF BF
733 // excluding surrogates codepoints \ud800 to \udfff
734 // ED A0 80 to ED BF BF
735 // 4-byte encoding is for codepoints \u10000 to \u10ffff
736 // first F0 90 80 80 last F4 8F BF BF
738 // Use the UTF-8 syntax from the RFC
740 // https://tools.ietf.org/html/rfc3629
742 // UTF8-2 = %xC2-DF UTF8-tail
743 // UTF8-3 = %xE0 %xA0-BF UTF8-tail / %xE1-EC 2( UTF8-tail ) /
744 // %xED %x80-9F UTF8-tail / %xEE-EF 2( UTF8-tail )
745 // UTF8-4 = %xF0 %x90-BF 2( UTF8-tail ) / %xF1-F3 3( UTF8-tail ) /
746 // %xF4 %x80-8F 2( UTF8-tail )
748 2 => if second & !CONT_MASK != TAG_CONT_U8 {err!()},
750 match (first, second, next!() & !CONT_MASK) {
751 (0xE0 , 0xA0 .. 0xBF, TAG_CONT_U8) |
752 (0xE1 .. 0xEC, 0x80 .. 0xBF, TAG_CONT_U8) |
753 (0xED , 0x80 .. 0x9F, TAG_CONT_U8) |
754 (0xEE .. 0xEF, 0x80 .. 0xBF, TAG_CONT_U8) => {}
759 match (first, second, next!() & !CONT_MASK, next!() & !CONT_MASK) {
760 (0xF0 , 0x90 .. 0xBF, TAG_CONT_U8, TAG_CONT_U8) |
761 (0xF1 .. 0xF3, 0x80 .. 0xBF, TAG_CONT_U8, TAG_CONT_U8) |
762 (0xF4 , 0x80 .. 0x8F, TAG_CONT_U8, TAG_CONT_U8) => {}
772 /// Determines if a vector of bytes contains valid UTF-8.
773 pub fn is_utf8(v: &[u8]) -> bool {
774 run_utf8_validation_iterator(&mut v.iter())
777 /// Determines if a vector of `u16` contains valid UTF-16
778 pub fn is_utf16(v: &[u16]) -> bool {
779 let mut it = v.iter();
780 macro_rules! next ( ($ret:expr) => {
781 match it.next() { Some(u) => *u, None => return $ret }
787 match char::from_u32(u as u32) {
790 let u2 = next!(false);
791 if u < 0xD7FF || u > 0xDBFF ||
792 u2 < 0xDC00 || u2 > 0xDFFF { return false; }
798 /// An iterator that decodes UTF-16 encoded codepoints from a vector
801 pub struct Utf16Items<'a> {
802 iter: slice::Items<'a, u16>
804 /// The possibilities for values decoded from a `u16` stream.
805 #[deriving(PartialEq, Eq, Clone, Show)]
807 /// A valid codepoint.
809 /// An invalid surrogate without its pair.
814 /// Convert `self` to a `char`, taking `LoneSurrogate`s to the
815 /// replacement character (U+FFFD).
817 pub fn to_char_lossy(&self) -> char {
820 LoneSurrogate(_) => '\uFFFD'
825 impl<'a> Iterator<Utf16Item> for Utf16Items<'a> {
826 fn next(&mut self) -> Option<Utf16Item> {
827 let u = match self.iter.next() {
832 if u < 0xD800 || 0xDFFF < u {
834 Some(ScalarValue(unsafe {mem::transmute(u as u32)}))
835 } else if u >= 0xDC00 {
836 // a trailing surrogate
837 Some(LoneSurrogate(u))
839 // preserve state for rewinding.
842 let u2 = match self.iter.next() {
845 None => return Some(LoneSurrogate(u))
847 if u2 < 0xDC00 || u2 > 0xDFFF {
848 // not a trailing surrogate so we're not a valid
849 // surrogate pair, so rewind to redecode u2 next time.
851 return Some(LoneSurrogate(u))
854 // all ok, so lets decode it.
855 let c = ((u - 0xD800) as u32 << 10 | (u2 - 0xDC00) as u32) + 0x1_0000;
856 Some(ScalarValue(unsafe {mem::transmute(c)}))
861 fn size_hint(&self) -> (uint, Option<uint>) {
862 let (low, high) = self.iter.size_hint();
863 // we could be entirely valid surrogates (2 elements per
864 // char), or entirely non-surrogates (1 element per char)
869 /// Create an iterator over the UTF-16 encoded codepoints in `v`,
870 /// returning invalid surrogates as `LoneSurrogate`s.
876 /// use std::str::{ScalarValue, LoneSurrogate};
878 /// // 𝄞mus<invalid>ic<invalid>
879 /// let v = [0xD834, 0xDD1E, 0x006d, 0x0075,
880 /// 0x0073, 0xDD1E, 0x0069, 0x0063,
883 /// assert_eq!(str::utf16_items(v).collect::<Vec<_>>(),
884 /// vec![ScalarValue('𝄞'),
885 /// ScalarValue('m'), ScalarValue('u'), ScalarValue('s'),
886 /// LoneSurrogate(0xDD1E),
887 /// ScalarValue('i'), ScalarValue('c'),
888 /// LoneSurrogate(0xD834)]);
890 pub fn utf16_items<'a>(v: &'a [u16]) -> Utf16Items<'a> {
891 Utf16Items { iter : v.iter() }
894 /// Return a slice of `v` ending at (and not including) the first NUL
903 /// let mut v = ['a' as u16, 'b' as u16, 'c' as u16, 'd' as u16];
904 /// // no NULs so no change
905 /// assert_eq!(str::truncate_utf16_at_nul(v), v.as_slice());
909 /// assert_eq!(str::truncate_utf16_at_nul(v),
910 /// &['a' as u16, 'b' as u16]);
912 pub fn truncate_utf16_at_nul<'a>(v: &'a [u16]) -> &'a [u16] {
913 match v.iter().position(|c| *c == 0) {
914 // don't include the 0
915 Some(i) => v.slice_to(i),
920 // https://tools.ietf.org/html/rfc3629
921 static UTF8_CHAR_WIDTH: [u8, ..256] = [
922 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
923 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x1F
924 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
925 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x3F
926 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
927 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x5F
928 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
929 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, // 0x7F
930 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
931 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0x9F
932 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
933 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, // 0xBF
934 0,0,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
935 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, // 0xDF
936 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, // 0xEF
937 4,4,4,4,4,0,0,0,0,0,0,0,0,0,0,0, // 0xFF
940 /// Given a first byte, determine how many bytes are in this UTF-8 character
942 pub fn utf8_char_width(b: u8) -> uint {
943 return UTF8_CHAR_WIDTH[b as uint] as uint;
946 /// Struct that contains a `char` and the index of the first byte of
947 /// the next `char` in a string. This can be used as a data structure
948 /// for iterating over the UTF-8 bytes of a string.
949 pub struct CharRange {
952 /// Index of the first byte of the next `char`
956 /// Mask of the value bits of a continuation byte
957 static CONT_MASK: u8 = 0b0011_1111u8;
958 /// Value of the tag bits (tag mask is !CONT_MASK) of a continuation byte
959 static TAG_CONT_U8: u8 = 0b1000_0000u8;
961 /// Unsafe operations
964 use collections::Collection;
967 use slice::{ImmutableVector};
968 use str::{is_utf8, StrSlice};
970 /// Converts a slice of bytes to a string slice without checking
971 /// that the string contains valid UTF-8.
972 pub unsafe fn from_utf8<'a>(v: &'a [u8]) -> &'a str {
976 /// Form a slice from a C string. Unsafe because the caller must ensure the
977 /// C string has the static lifetime, or else the return value may be
978 /// invalidated later.
979 pub unsafe fn c_str_to_static_slice(s: *const i8) -> &'static str {
980 let s = s as *const u8;
985 curr = s.offset(len as int);
987 let v = Slice { data: s, len: len };
988 assert!(is_utf8(::mem::transmute(v)));
992 /// Takes a bytewise (not UTF-8) slice from a string.
994 /// Returns the substring from [`begin`..`end`).
998 /// If begin is greater than end.
999 /// If end is greater than the length of the string.
1001 pub unsafe fn slice_bytes<'a>(s: &'a str, begin: uint, end: uint) -> &'a str {
1002 assert!(begin <= end);
1003 assert!(end <= s.len());
1004 slice_unchecked(s, begin, end)
1007 /// Takes a bytewise (not UTF-8) slice from a string.
1009 /// Returns the substring from [`begin`..`end`).
1011 /// Caller must check slice boundaries!
1013 pub unsafe fn slice_unchecked<'a>(s: &'a str, begin: uint, end: uint) -> &'a str {
1014 mem::transmute(Slice {
1015 data: s.as_ptr().offset(begin as int),
1022 Section: Trait implementations
1025 #[allow(missing_doc)]
1027 use cmp::{Ord, Ordering, Less, Equal, Greater, PartialEq, PartialOrd, Equiv, Eq};
1028 use collections::Collection;
1030 use option::{Option, Some};
1031 use str::{Str, StrSlice, eq_slice};
1033 impl<'a> Ord for &'a str {
1035 fn cmp(&self, other: & &'a str) -> Ordering {
1036 for (s_b, o_b) in self.bytes().zip(other.bytes()) {
1037 match s_b.cmp(&o_b) {
1038 Greater => return Greater,
1039 Less => return Less,
1044 self.len().cmp(&other.len())
1048 impl<'a> PartialEq for &'a str {
1050 fn eq(&self, other: & &'a str) -> bool {
1051 eq_slice((*self), (*other))
1054 fn ne(&self, other: & &'a str) -> bool { !(*self).eq(other) }
1057 impl<'a> Eq for &'a str {}
1059 impl<'a> PartialOrd for &'a str {
1061 fn partial_cmp(&self, other: &&'a str) -> Option<Ordering> {
1062 Some(self.cmp(other))
1066 impl<'a, S: Str> Equiv<S> for &'a str {
1068 fn equiv(&self, other: &S) -> bool { eq_slice(*self, other.as_slice()) }
1072 /// Any string that can be represented as a slice
1074 /// Work with `self` as a slice.
1075 fn as_slice<'a>(&'a self) -> &'a str;
1078 impl<'a> Str for &'a str {
1080 fn as_slice<'a>(&'a self) -> &'a str { *self }
1083 impl<'a> Collection for &'a str {
1085 fn len(&self) -> uint {
1090 /// Methods for string slices
1091 pub trait StrSlice<'a> {
1092 /// Returns true if one string contains another
1096 /// - needle - The string to look for
1097 fn contains<'a>(&self, needle: &'a str) -> bool;
1099 /// Returns true if a string contains a char.
1103 /// - needle - The char to look for
1104 fn contains_char(&self, needle: char) -> bool;
1106 /// An iterator over the characters of `self`. Note, this iterates
1107 /// over unicode code-points, not unicode graphemes.
1112 /// let v: Vec<char> = "abc åäö".chars().collect();
1113 /// assert_eq!(v, vec!['a', 'b', 'c', ' ', 'å', 'ä', 'ö']);
1115 fn chars(&self) -> Chars<'a>;
1117 /// An iterator over the bytes of `self`
1118 fn bytes(&self) -> Bytes<'a>;
1120 /// An iterator over the characters of `self` and their byte offsets.
1121 fn char_indices(&self) -> CharOffsets<'a>;
1123 /// An iterator over substrings of `self`, separated by characters
1124 /// matched by `sep`.
1129 /// let v: Vec<&str> = "Mary had a little lamb".split(' ').collect();
1130 /// assert_eq!(v, vec!["Mary", "had", "a", "little", "lamb"]);
1132 /// let v: Vec<&str> = "abc1def2ghi".split(|c: char| c.is_digit()).collect();
1133 /// assert_eq!(v, vec!["abc", "def", "ghi"]);
1135 /// let v: Vec<&str> = "lionXXtigerXleopard".split('X').collect();
1136 /// assert_eq!(v, vec!["lion", "", "tiger", "leopard"]);
1138 /// let v: Vec<&str> = "".split('X').collect();
1139 /// assert_eq!(v, vec![""]);
1141 fn split<Sep: CharEq>(&self, sep: Sep) -> CharSplits<'a, Sep>;
1143 /// An iterator over substrings of `self`, separated by characters
1144 /// matched by `sep`, restricted to splitting at most `count`
1150 /// let v: Vec<&str> = "Mary had a little lambda".splitn(' ', 2).collect();
1151 /// assert_eq!(v, vec!["Mary", "had", "a little lambda"]);
1153 /// let v: Vec<&str> = "abc1def2ghi".splitn(|c: char| c.is_digit(), 1).collect();
1154 /// assert_eq!(v, vec!["abc", "def2ghi"]);
1156 /// let v: Vec<&str> = "lionXXtigerXleopard".splitn('X', 2).collect();
1157 /// assert_eq!(v, vec!["lion", "", "tigerXleopard"]);
1159 /// let v: Vec<&str> = "abcXdef".splitn('X', 0).collect();
1160 /// assert_eq!(v, vec!["abcXdef"]);
1162 /// let v: Vec<&str> = "".splitn('X', 1).collect();
1163 /// assert_eq!(v, vec![""]);
1165 fn splitn<Sep: CharEq>(&self, sep: Sep, count: uint) -> CharSplitsN<'a, Sep>;
1167 /// An iterator over substrings of `self`, separated by characters
1168 /// matched by `sep`.
1170 /// Equivalent to `split`, except that the trailing substring
1171 /// is skipped if empty (terminator semantics).
1176 /// let v: Vec<&str> = "A.B.".split_terminator('.').collect();
1177 /// assert_eq!(v, vec!["A", "B"]);
1179 /// let v: Vec<&str> = "A..B..".split_terminator('.').collect();
1180 /// assert_eq!(v, vec!["A", "", "B", ""]);
1182 /// let v: Vec<&str> = "Mary had a little lamb".split(' ').rev().collect();
1183 /// assert_eq!(v, vec!["lamb", "little", "a", "had", "Mary"]);
1185 /// let v: Vec<&str> = "abc1def2ghi".split(|c: char| c.is_digit()).rev().collect();
1186 /// assert_eq!(v, vec!["ghi", "def", "abc"]);
1188 /// let v: Vec<&str> = "lionXXtigerXleopard".split('X').rev().collect();
1189 /// assert_eq!(v, vec!["leopard", "tiger", "", "lion"]);
1191 fn split_terminator<Sep: CharEq>(&self, sep: Sep) -> CharSplits<'a, Sep>;
1193 /// An iterator over substrings of `self`, separated by characters
1194 /// matched by `sep`, starting from the end of the string.
1195 /// Restricted to splitting at most `count` times.
1200 /// let v: Vec<&str> = "Mary had a little lamb".rsplitn(' ', 2).collect();
1201 /// assert_eq!(v, vec!["lamb", "little", "Mary had a"]);
1203 /// let v: Vec<&str> = "abc1def2ghi".rsplitn(|c: char| c.is_digit(), 1).collect();
1204 /// assert_eq!(v, vec!["ghi", "abc1def"]);
1206 /// let v: Vec<&str> = "lionXXtigerXleopard".rsplitn('X', 2).collect();
1207 /// assert_eq!(v, vec!["leopard", "tiger", "lionX"]);
1209 fn rsplitn<Sep: CharEq>(&self, sep: Sep, count: uint) -> CharSplitsN<'a, Sep>;
1211 /// An iterator over the start and end indices of the disjoint
1212 /// matches of `sep` within `self`.
1214 /// That is, each returned value `(start, end)` satisfies
1215 /// `self.slice(start, end) == sep`. For matches of `sep` within
1216 /// `self` that overlap, only the indices corresponding to the
1217 /// first match are returned.
1222 /// let v: Vec<(uint, uint)> = "abcXXXabcYYYabc".match_indices("abc").collect();
1223 /// assert_eq!(v, vec![(0,3), (6,9), (12,15)]);
1225 /// let v: Vec<(uint, uint)> = "1abcabc2".match_indices("abc").collect();
1226 /// assert_eq!(v, vec![(1,4), (4,7)]);
1228 /// let v: Vec<(uint, uint)> = "ababa".match_indices("aba").collect();
1229 /// assert_eq!(v, vec![(0, 3)]); // only the first `aba`
1231 fn match_indices(&self, sep: &'a str) -> MatchIndices<'a>;
1233 /// An iterator over the substrings of `self` separated by `sep`.
1238 /// let v: Vec<&str> = "abcXXXabcYYYabc".split_str("abc").collect();
1239 /// assert_eq!(v, vec!["", "XXX", "YYY", ""]);
1241 /// let v: Vec<&str> = "1abcabc2".split_str("abc").collect();
1242 /// assert_eq!(v, vec!["1", "", "2"]);
1244 fn split_str(&self, &'a str) -> StrSplits<'a>;
1246 /// An iterator over the lines of a string (subsequences separated
1247 /// by `\n`). This does not include the empty string after a
1253 /// let four_lines = "foo\nbar\n\nbaz\n";
1254 /// let v: Vec<&str> = four_lines.lines().collect();
1255 /// assert_eq!(v, vec!["foo", "bar", "", "baz"]);
1257 fn lines(&self) -> CharSplits<'a, char>;
1259 /// An iterator over the lines of a string, separated by either
1260 /// `\n` or `\r\n`. As with `.lines()`, this does not include an
1261 /// empty trailing line.
1266 /// let four_lines = "foo\r\nbar\n\r\nbaz\n";
1267 /// let v: Vec<&str> = four_lines.lines_any().collect();
1268 /// assert_eq!(v, vec!["foo", "bar", "", "baz"]);
1270 fn lines_any(&self) -> AnyLines<'a>;
1272 /// Returns the number of Unicode code points (`char`) that a
1275 /// This does not perform any normalization, and is `O(n)`, since
1276 /// UTF-8 is a variable width encoding of code points.
1278 /// *Warning*: The number of code points in a string does not directly
1279 /// correspond to the number of visible characters or width of the
1280 /// visible text due to composing characters, and double- and
1281 /// zero-width ones.
1283 /// See also `.len()` for the byte length.
1288 /// // composed forms of `ö` and `é`
1289 /// let c = "Löwe 老虎 Léopard"; // German, Simplified Chinese, French
1290 /// // decomposed forms of `ö` and `é`
1291 /// let d = "Lo\u0308we 老虎 Le\u0301opard";
1293 /// assert_eq!(c.char_len(), 15);
1294 /// assert_eq!(d.char_len(), 17);
1296 /// assert_eq!(c.len(), 21);
1297 /// assert_eq!(d.len(), 23);
1299 /// // the two strings *look* the same
1300 /// println!("{}", c);
1301 /// println!("{}", d);
1303 fn char_len(&self) -> uint;
1305 /// Returns a slice of the given string from the byte range
1306 /// [`begin`..`end`).
1308 /// This operation is `O(1)`.
1310 /// Fails when `begin` and `end` do not point to valid characters
1311 /// or point beyond the last character of the string.
1313 /// See also `slice_to` and `slice_from` for slicing prefixes and
1314 /// suffixes of strings, and `slice_chars` for slicing based on
1315 /// code point counts.
1320 /// let s = "Löwe 老虎 Léopard";
1321 /// assert_eq!(s.slice(0, 1), "L");
1323 /// assert_eq!(s.slice(1, 9), "öwe 老");
1325 /// // these will fail:
1326 /// // byte 2 lies within `ö`:
1327 /// // s.slice(2, 3);
1329 /// // byte 8 lies within `老`
1330 /// // s.slice(1, 8);
1332 /// // byte 100 is outside the string
1333 /// // s.slice(3, 100);
1335 fn slice(&self, begin: uint, end: uint) -> &'a str;
1337 /// Returns a slice of the string from `begin` to its end.
1339 /// Equivalent to `self.slice(begin, self.len())`.
1341 /// Fails when `begin` does not point to a valid character, or is
1344 /// See also `slice`, `slice_to` and `slice_chars`.
1345 fn slice_from(&self, begin: uint) -> &'a str;
1347 /// Returns a slice of the string from the beginning to byte
1350 /// Equivalent to `self.slice(0, end)`.
1352 /// Fails when `end` does not point to a valid character, or is
1355 /// See also `slice`, `slice_from` and `slice_chars`.
1356 fn slice_to(&self, end: uint) -> &'a str;
1358 /// Returns a slice of the string from the character range
1359 /// [`begin`..`end`).
1361 /// That is, start at the `begin`-th code point of the string and
1362 /// continue to the `end`-th code point. This does not detect or
1363 /// handle edge cases such as leaving a combining character as the
1364 /// first code point of the string.
1366 /// Due to the design of UTF-8, this operation is `O(end)`.
1367 /// See `slice`, `slice_to` and `slice_from` for `O(1)`
1368 /// variants that use byte indices rather than code point
1371 /// Fails if `begin` > `end` or the either `begin` or `end` are
1372 /// beyond the last character of the string.
1377 /// let s = "Löwe 老虎 Léopard";
1378 /// assert_eq!(s.slice_chars(0, 4), "Löwe");
1379 /// assert_eq!(s.slice_chars(5, 7), "老虎");
1381 fn slice_chars(&self, begin: uint, end: uint) -> &'a str;
1383 /// Returns true if `needle` is a prefix of the string.
1384 fn starts_with(&self, needle: &str) -> bool;
1386 /// Returns true if `needle` is a suffix of the string.
1387 fn ends_with(&self, needle: &str) -> bool;
1389 /// Returns a string with characters that match `to_trim` removed.
1393 /// * to_trim - a character matcher
1398 /// assert_eq!("11foo1bar11".trim_chars('1'), "foo1bar")
1399 /// assert_eq!("12foo1bar12".trim_chars(&['1', '2']), "foo1bar")
1400 /// assert_eq!("123foo1bar123".trim_chars(|c: char| c.is_digit()), "foo1bar")
1402 fn trim_chars<C: CharEq>(&self, to_trim: C) -> &'a str;
1404 /// Returns a string with leading `chars_to_trim` removed.
1408 /// * to_trim - a character matcher
1413 /// assert_eq!("11foo1bar11".trim_left_chars('1'), "foo1bar11")
1414 /// assert_eq!("12foo1bar12".trim_left_chars(&['1', '2']), "foo1bar12")
1415 /// assert_eq!("123foo1bar123".trim_left_chars(|c: char| c.is_digit()), "foo1bar123")
1417 fn trim_left_chars<C: CharEq>(&self, to_trim: C) -> &'a str;
1419 /// Returns a string with trailing `chars_to_trim` removed.
1423 /// * to_trim - a character matcher
1428 /// assert_eq!("11foo1bar11".trim_right_chars('1'), "11foo1bar")
1429 /// assert_eq!("12foo1bar12".trim_right_chars(&['1', '2']), "12foo1bar")
1430 /// assert_eq!("123foo1bar123".trim_right_chars(|c: char| c.is_digit()), "123foo1bar")
1432 fn trim_right_chars<C: CharEq>(&self, to_trim: C) -> &'a str;
1434 /// Check that `index`-th byte lies at the start and/or end of a
1435 /// UTF-8 code point sequence.
1437 /// The start and end of the string (when `index == self.len()`)
1438 /// are considered to be boundaries.
1440 /// Fails if `index` is greater than `self.len()`.
1445 /// let s = "Löwe 老虎 Léopard";
1446 /// assert!(s.is_char_boundary(0));
1448 /// assert!(s.is_char_boundary(6));
1449 /// assert!(s.is_char_boundary(s.len()));
1451 /// // second byte of `ö`
1452 /// assert!(!s.is_char_boundary(2));
1454 /// // third byte of `老`
1455 /// assert!(!s.is_char_boundary(8));
1457 fn is_char_boundary(&self, index: uint) -> bool;
1459 /// Pluck a character out of a string and return the index of the next
1462 /// This function can be used to iterate over the unicode characters of a
1467 /// This example manually iterate through the characters of a
1468 /// string; this should normally by done by `.chars()` or
1469 /// `.char_indices`.
1472 /// use std::str::CharRange;
1474 /// let s = "中华Việt Nam";
1476 /// while i < s.len() {
1477 /// let CharRange {ch, next} = s.char_range_at(i);
1478 /// println!("{}: {}", i, ch);
1500 /// * s - The string
1501 /// * i - The byte offset of the char to extract
1505 /// A record {ch: char, next: uint} containing the char value and the byte
1506 /// index of the next unicode character.
1510 /// If `i` is greater than or equal to the length of the string.
1511 /// If `i` is not the index of the beginning of a valid UTF-8 character.
1512 fn char_range_at(&self, start: uint) -> CharRange;
1514 /// Given a byte position and a str, return the previous char and its position.
1516 /// This function can be used to iterate over a unicode string in reverse.
1518 /// Returns 0 for next index if called on start index 0.
1522 /// If `i` is greater than the length of the string.
1523 /// If `i` is not an index following a valid UTF-8 character.
1524 fn char_range_at_reverse(&self, start: uint) -> CharRange;
1526 /// Plucks the character starting at the `i`th byte of a string.
1530 /// If `i` is greater than or equal to the length of the string.
1531 /// If `i` is not the index of the beginning of a valid UTF-8 character.
1532 fn char_at(&self, i: uint) -> char;
1534 /// Plucks the character ending at the `i`th byte of a string.
1538 /// If `i` is greater than the length of the string.
1539 /// If `i` is not an index following a valid UTF-8 character.
1540 fn char_at_reverse(&self, i: uint) -> char;
1542 /// Work with the byte buffer of a string as a byte slice.
1543 fn as_bytes(&self) -> &'a [u8];
1545 /// Returns the byte index of the first character of `self` that
1546 /// matches `search`.
1550 /// `Some` containing the byte index of the last matching character
1551 /// or `None` if there is no match
1556 /// let s = "Löwe 老虎 Léopard";
1558 /// assert_eq!(s.find('L'), Some(0));
1559 /// assert_eq!(s.find('é'), Some(14));
1561 /// // the first space
1562 /// assert_eq!(s.find(|c: char| c.is_whitespace()), Some(5));
1564 /// // neither are found
1565 /// assert_eq!(s.find(&['1', '2']), None);
1567 fn find<C: CharEq>(&self, search: C) -> Option<uint>;
1569 /// Returns the byte index of the last character of `self` that
1570 /// matches `search`.
1574 /// `Some` containing the byte index of the last matching character
1575 /// or `None` if there is no match.
1580 /// let s = "Löwe 老虎 Léopard";
1582 /// assert_eq!(s.rfind('L'), Some(13));
1583 /// assert_eq!(s.rfind('é'), Some(14));
1585 /// // the second space
1586 /// assert_eq!(s.rfind(|c: char| c.is_whitespace()), Some(12));
1588 /// // searches for an occurrence of either `1` or `2`, but neither are found
1589 /// assert_eq!(s.rfind(&['1', '2']), None);
1591 fn rfind<C: CharEq>(&self, search: C) -> Option<uint>;
1593 /// Returns the byte index of the first matching substring
1597 /// * `needle` - The string to search for
1601 /// `Some` containing the byte index of the first matching substring
1602 /// or `None` if there is no match.
1607 /// let s = "Löwe 老虎 Léopard";
1609 /// assert_eq!(s.find_str("老虎 L"), Some(6));
1610 /// assert_eq!(s.find_str("muffin man"), None);
1612 fn find_str(&self, &str) -> Option<uint>;
1614 /// Retrieves the first character from a string slice and returns
1615 /// it. This does not allocate a new string; instead, it returns a
1616 /// slice that point one character beyond the character that was
1617 /// shifted. If the string does not contain any characters,
1618 /// a tuple of None and an empty string is returned instead.
1623 /// let s = "Löwe 老虎 Léopard";
1624 /// let (c, s1) = s.slice_shift_char();
1625 /// assert_eq!(c, Some('L'));
1626 /// assert_eq!(s1, "öwe 老虎 Léopard");
1628 /// let (c, s2) = s1.slice_shift_char();
1629 /// assert_eq!(c, Some('ö'));
1630 /// assert_eq!(s2, "we 老虎 Léopard");
1632 fn slice_shift_char(&self) -> (Option<char>, &'a str);
1634 /// Returns the byte offset of an inner slice relative to an enclosing outer slice.
1636 /// Fails if `inner` is not a direct slice contained within self.
1641 /// let string = "a\nb\nc";
1642 /// let lines: Vec<&str> = string.lines().collect();
1643 /// let lines = lines.as_slice();
1645 /// assert!(string.subslice_offset(lines[0]) == 0); // &"a"
1646 /// assert!(string.subslice_offset(lines[1]) == 2); // &"b"
1647 /// assert!(string.subslice_offset(lines[2]) == 4); // &"c"
1649 fn subslice_offset(&self, inner: &str) -> uint;
1651 /// Return an unsafe pointer to the strings buffer.
1653 /// The caller must ensure that the string outlives this pointer,
1654 /// and that it is not reallocated (e.g. by pushing to the
1656 fn as_ptr(&self) -> *const u8;
1658 /// Return an iterator of `u16` over the string encoded as UTF-16.
1659 fn utf16_units(&self) -> Utf16CodeUnits<'a>;
1662 impl<'a> StrSlice<'a> for &'a str {
1664 fn contains<'a>(&self, needle: &'a str) -> bool {
1665 self.find_str(needle).is_some()
1669 fn contains_char(&self, needle: char) -> bool {
1670 self.find(needle).is_some()
1674 fn chars(&self) -> Chars<'a> {
1675 Chars{iter: self.as_bytes().iter()}
1679 fn bytes(&self) -> Bytes<'a> {
1680 self.as_bytes().iter().map(|&b| b)
1684 fn char_indices(&self) -> CharOffsets<'a> {
1685 CharOffsets{front_offset: 0, iter: self.chars()}
1689 fn split<Sep: CharEq>(&self, sep: Sep) -> CharSplits<'a, Sep> {
1692 only_ascii: sep.only_ascii(),
1694 allow_trailing_empty: true,
1700 fn splitn<Sep: CharEq>(&self, sep: Sep, count: uint)
1701 -> CharSplitsN<'a, Sep> {
1703 iter: self.split(sep),
1710 fn split_terminator<Sep: CharEq>(&self, sep: Sep)
1711 -> CharSplits<'a, Sep> {
1713 allow_trailing_empty: false,
1719 fn rsplitn<Sep: CharEq>(&self, sep: Sep, count: uint)
1720 -> CharSplitsN<'a, Sep> {
1722 iter: self.split(sep),
1729 fn match_indices(&self, sep: &'a str) -> MatchIndices<'a> {
1730 assert!(!sep.is_empty())
1734 searcher: Searcher::new(self.as_bytes(), sep.as_bytes())
1739 fn split_str(&self, sep: &'a str) -> StrSplits<'a> {
1741 it: self.match_indices(sep),
1748 fn lines(&self) -> CharSplits<'a, char> {
1749 self.split_terminator('\n')
1752 fn lines_any(&self) -> AnyLines<'a> {
1753 self.lines().map(|line| {
1755 if l > 0 && line.as_bytes()[l - 1] == b'\r' { line.slice(0, l - 1) }
1761 fn char_len(&self) -> uint { self.chars().count() }
1764 fn slice(&self, begin: uint, end: uint) -> &'a str {
1765 assert!(self.is_char_boundary(begin) && self.is_char_boundary(end),
1766 "index {} and/or {} in `{}` do not lie on character boundary", begin,
1768 unsafe { raw::slice_bytes(*self, begin, end) }
1772 fn slice_from(&self, begin: uint) -> &'a str {
1773 self.slice(begin, self.len())
1777 fn slice_to(&self, end: uint) -> &'a str {
1778 assert!(self.is_char_boundary(end), "index {} in `{}` does not lie on \
1779 a character boundary", end, *self);
1780 unsafe { raw::slice_bytes(*self, 0, end) }
1783 fn slice_chars(&self, begin: uint, end: uint) -> &'a str {
1784 assert!(begin <= end);
1786 let mut begin_byte = None;
1787 let mut end_byte = None;
1789 // This could be even more efficient by not decoding,
1790 // only finding the char boundaries
1791 for (idx, _) in self.char_indices() {
1792 if count == begin { begin_byte = Some(idx); }
1793 if count == end { end_byte = Some(idx); break; }
1796 if begin_byte.is_none() && count == begin { begin_byte = Some(self.len()) }
1797 if end_byte.is_none() && count == end { end_byte = Some(self.len()) }
1799 match (begin_byte, end_byte) {
1800 (None, _) => fail!("slice_chars: `begin` is beyond end of string"),
1801 (_, None) => fail!("slice_chars: `end` is beyond end of string"),
1802 (Some(a), Some(b)) => unsafe { raw::slice_bytes(*self, a, b) }
1807 fn starts_with<'a>(&self, needle: &'a str) -> bool {
1808 let n = needle.len();
1809 self.len() >= n && needle.as_bytes() == self.as_bytes().slice_to(n)
1813 fn ends_with(&self, needle: &str) -> bool {
1814 let (m, n) = (self.len(), needle.len());
1815 m >= n && needle.as_bytes() == self.as_bytes().slice_from(m - n)
1819 fn trim_chars<C: CharEq>(&self, mut to_trim: C) -> &'a str {
1820 let cur = match self.find(|c: char| !to_trim.matches(c)) {
1822 Some(i) => unsafe { raw::slice_bytes(*self, i, self.len()) }
1824 match cur.rfind(|c: char| !to_trim.matches(c)) {
1827 let right = cur.char_range_at(i).next;
1828 unsafe { raw::slice_bytes(cur, 0, right) }
1834 fn trim_left_chars<C: CharEq>(&self, mut to_trim: C) -> &'a str {
1835 match self.find(|c: char| !to_trim.matches(c)) {
1837 Some(first) => unsafe { raw::slice_bytes(*self, first, self.len()) }
1842 fn trim_right_chars<C: CharEq>(&self, mut to_trim: C) -> &'a str {
1843 match self.rfind(|c: char| !to_trim.matches(c)) {
1846 let next = self.char_range_at(last).next;
1847 unsafe { raw::slice_bytes(*self, 0u, next) }
1853 fn is_char_boundary(&self, index: uint) -> bool {
1854 if index == self.len() { return true; }
1855 if index > self.len() { return false; }
1856 let b = self.as_bytes()[index];
1857 return b < 128u8 || b >= 192u8;
1861 fn char_range_at(&self, i: uint) -> CharRange {
1862 if self.as_bytes()[i] < 128u8 {
1863 return CharRange {ch: self.as_bytes()[i] as char, next: i + 1 };
1866 // Multibyte case is a fn to allow char_range_at to inline cleanly
1867 fn multibyte_char_range_at(s: &str, i: uint) -> CharRange {
1868 let mut val = s.as_bytes()[i] as u32;
1869 let w = UTF8_CHAR_WIDTH[val as uint] as uint;
1872 val = utf8_first_byte!(val, w);
1873 val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 1]);
1874 if w > 2 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 2]); }
1875 if w > 3 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 3]); }
1877 return CharRange {ch: unsafe { mem::transmute(val) }, next: i + w};
1880 return multibyte_char_range_at(*self, i);
1884 fn char_range_at_reverse(&self, start: uint) -> CharRange {
1885 let mut prev = start;
1887 prev = prev.saturating_sub(1);
1888 if self.as_bytes()[prev] < 128 {
1889 return CharRange{ch: self.as_bytes()[prev] as char, next: prev}
1892 // Multibyte case is a fn to allow char_range_at_reverse to inline cleanly
1893 fn multibyte_char_range_at_reverse(s: &str, mut i: uint) -> CharRange {
1894 // while there is a previous byte == 10......
1895 while i > 0 && s.as_bytes()[i] & !CONT_MASK == TAG_CONT_U8 {
1899 let mut val = s.as_bytes()[i] as u32;
1900 let w = UTF8_CHAR_WIDTH[val as uint] as uint;
1903 val = utf8_first_byte!(val, w);
1904 val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 1]);
1905 if w > 2 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 2]); }
1906 if w > 3 { val = utf8_acc_cont_byte!(val, s.as_bytes()[i + 3]); }
1908 return CharRange {ch: unsafe { mem::transmute(val) }, next: i};
1911 return multibyte_char_range_at_reverse(*self, prev);
1915 fn char_at(&self, i: uint) -> char {
1916 self.char_range_at(i).ch
1920 fn char_at_reverse(&self, i: uint) -> char {
1921 self.char_range_at_reverse(i).ch
1925 fn as_bytes(&self) -> &'a [u8] {
1926 unsafe { mem::transmute(*self) }
1929 fn find<C: CharEq>(&self, mut search: C) -> Option<uint> {
1930 if search.only_ascii() {
1931 self.bytes().position(|b| search.matches(b as char))
1933 for (index, c) in self.char_indices() {
1934 if search.matches(c) { return Some(index); }
1940 fn rfind<C: CharEq>(&self, mut search: C) -> Option<uint> {
1941 if search.only_ascii() {
1942 self.bytes().rposition(|b| search.matches(b as char))
1944 for (index, c) in self.char_indices().rev() {
1945 if search.matches(c) { return Some(index); }
1951 fn find_str(&self, needle: &str) -> Option<uint> {
1952 if needle.is_empty() {
1955 self.match_indices(needle)
1957 .map(|(start, _end)| start)
1962 fn slice_shift_char(&self) -> (Option<char>, &'a str) {
1963 if self.is_empty() {
1964 return (None, *self);
1966 let CharRange {ch, next} = self.char_range_at(0u);
1967 let next_s = unsafe { raw::slice_bytes(*self, next, self.len()) };
1968 return (Some(ch), next_s);
1972 fn subslice_offset(&self, inner: &str) -> uint {
1973 let a_start = self.as_ptr() as uint;
1974 let a_end = a_start + self.len();
1975 let b_start = inner.as_ptr() as uint;
1976 let b_end = b_start + inner.len();
1978 assert!(a_start <= b_start);
1979 assert!(b_end <= a_end);
1984 fn as_ptr(&self) -> *const u8 {
1989 fn utf16_units(&self) -> Utf16CodeUnits<'a> {
1990 Utf16CodeUnits{ chars: self.chars(), extra: 0}
1994 impl<'a> Default for &'a str {
1995 fn default() -> &'a str { "" }