1 //! Unicode string slices.
3 //! *[See also the `str` primitive type](../../std/primitive.str.html).*
5 //! The `&str` type is one of the two main string types, the other being `String`.
6 //! Unlike its `String` counterpart, its contents are borrowed.
10 //! A basic string declaration of `&str` type:
13 //! let hello_world = "Hello, World!";
16 //! Here we have declared a string literal, also known as a string slice.
17 //! String literals have a static lifetime, which means the string `hello_world`
18 //! is guaranteed to be valid for the duration of the entire program.
19 //! We can explicitly specify `hello_world`'s lifetime as well:
22 //! let hello_world: &'static str = "Hello, world!";
25 #![stable(feature = "rust1", since = "1.0.0")]
27 // Many of the usings in this module are only used in the test configuration.
28 // It's cleaner to just turn off the unused_imports warning than to fix them.
29 #![allow(unused_imports)]
31 use core::borrow::Borrow;
33 use core::str as core_str;
34 use core::str::pattern::{Pattern, Searcher, ReverseSearcher, DoubleEndedSearcher};
37 use core::iter::FusedIterator;
38 use core::unicode::conversions;
40 use crate::borrow::ToOwned;
41 use crate::boxed::Box;
42 use crate::slice::{SliceConcatExt, SliceIndex};
43 use crate::string::String;
46 #[stable(feature = "rust1", since = "1.0.0")]
47 pub use core::str::{FromStr, Utf8Error};
49 #[stable(feature = "rust1", since = "1.0.0")]
50 pub use core::str::{Lines, LinesAny};
51 #[stable(feature = "rust1", since = "1.0.0")]
52 pub use core::str::{Split, RSplit};
53 #[stable(feature = "rust1", since = "1.0.0")]
54 pub use core::str::{SplitN, RSplitN};
55 #[stable(feature = "rust1", since = "1.0.0")]
56 pub use core::str::{SplitTerminator, RSplitTerminator};
57 #[stable(feature = "rust1", since = "1.0.0")]
58 pub use core::str::{Matches, RMatches};
59 #[stable(feature = "rust1", since = "1.0.0")]
60 pub use core::str::{MatchIndices, RMatchIndices};
61 #[stable(feature = "rust1", since = "1.0.0")]
62 pub use core::str::{from_utf8, from_utf8_mut, Chars, CharIndices, Bytes};
63 #[stable(feature = "rust1", since = "1.0.0")]
64 pub use core::str::{from_utf8_unchecked, from_utf8_unchecked_mut, ParseBoolError};
65 #[stable(feature = "rust1", since = "1.0.0")]
66 pub use core::str::SplitWhitespace;
67 #[stable(feature = "rust1", since = "1.0.0")]
68 pub use core::str::pattern;
69 #[stable(feature = "encode_utf16", since = "1.8.0")]
70 pub use core::str::EncodeUtf16;
71 #[stable(feature = "split_ascii_whitespace", since = "1.34.0")]
72 pub use core::str::SplitAsciiWhitespace;
74 #[unstable(feature = "slice_concat_ext",
75 reason = "trait should not have to exist",
77 impl<S: Borrow<str>> SliceConcatExt<str> for [S] {
80 fn concat(&self) -> String {
84 fn join(&self, sep: &str) -> String {
86 String::from_utf8_unchecked( join_generic_copy(self, sep.as_bytes()) )
90 fn connect(&self, sep: &str) -> String {
95 macro_rules! spezialize_for_lengths {
96 ($separator:expr, $target:expr, $iter:expr; $($num:expr),*) => {
97 let mut target = $target;
99 let sep_bytes = $separator;
100 match $separator.len() {
102 // loops with hardcoded sizes run much faster
103 // specialize the cases with small separator lengths
106 copy_slice_and_advance!(target, sep_bytes);
107 copy_slice_and_advance!(target, s.borrow().as_ref());
112 // arbitrary non-zero size fallback
114 copy_slice_and_advance!(target, sep_bytes);
115 copy_slice_and_advance!(target, s.borrow().as_ref());
122 macro_rules! copy_slice_and_advance {
123 ($target:expr, $bytes:expr) => {
124 let len = $bytes.len();
125 let (head, tail) = {$target}.split_at_mut(len);
126 head.copy_from_slice($bytes);
131 // Optimized join implementation that works for both Vec<T> (T: Copy) and String's inner vec
132 // Currently (2018-05-13) there is a bug with type inference and specialization (see issue #36262)
133 // For this reason SliceConcatExt<T> is not specialized for T: Copy and SliceConcatExt<str> is the
134 // only user of this function. It is left in place for the time when that is fixed.
136 // the bounds for String-join are S: Borrow<str> and for Vec-join Borrow<[T]>
137 // [T] and str both impl AsRef<[T]> for some T
138 // => s.borrow().as_ref() and we always have slices
139 fn join_generic_copy<B, T, S>(slice: &[S], sep: &[T]) -> Vec<T>
142 B: AsRef<[T]> + ?Sized,
145 let sep_len = sep.len();
146 let mut iter = slice.iter();
148 // the first slice is the only one without a separator preceding it
149 let first = match iter.next() {
150 Some(first) => first,
151 None => return vec![],
154 // compute the exact total length of the joined Vec
155 // if the `len` calculation overflows, we'll panic
156 // we would have run out of memory anyway and the rest of the function requires
157 // the entire Vec pre-allocated for safety
158 let len = sep_len.checked_mul(iter.len()).and_then(|n| {
160 .map(|s| s.borrow().as_ref().len())
161 .try_fold(n, usize::checked_add)
162 }).expect("attempt to join into collection with len > usize::MAX");
164 // crucial for safety
165 let mut result = Vec::with_capacity(len);
166 assert!(result.capacity() >= len);
168 result.extend_from_slice(first.borrow().as_ref());
172 let pos = result.len();
173 let target = result.get_unchecked_mut(pos..len);
175 // copy separator and slices over without bounds checks
176 // generate loops with hardcoded offsets for small separators
177 // massive improvements possible (~ x2)
178 spezialize_for_lengths!(sep, target, iter; 0, 1, 2, 3, 4);
185 #[stable(feature = "rust1", since = "1.0.0")]
186 impl Borrow<str> for String {
188 fn borrow(&self) -> &str {
193 #[stable(feature = "rust1", since = "1.0.0")]
194 impl ToOwned for str {
197 fn to_owned(&self) -> String {
198 unsafe { String::from_utf8_unchecked(self.as_bytes().to_owned()) }
201 fn clone_into(&self, target: &mut String) {
202 let mut b = mem::replace(target, String::new()).into_bytes();
203 self.as_bytes().clone_into(&mut b);
204 *target = unsafe { String::from_utf8_unchecked(b) }
208 /// Methods for string slices.
209 #[lang = "str_alloc"]
212 /// Converts a `Box<str>` into a `Box<[u8]>` without copying or allocating.
219 /// let s = "this is a string";
220 /// let boxed_str = s.to_owned().into_boxed_str();
221 /// let boxed_bytes = boxed_str.into_boxed_bytes();
222 /// assert_eq!(*boxed_bytes, *s.as_bytes());
224 #[stable(feature = "str_box_extras", since = "1.20.0")]
226 pub fn into_boxed_bytes(self: Box<str>) -> Box<[u8]> {
230 /// Replaces all matches of a pattern with another string.
232 /// `replace` creates a new [`String`], and copies the data from this string slice into it.
233 /// While doing so, it attempts to find matches of a pattern. If it finds any, it
234 /// replaces them with the replacement string slice.
236 /// [`String`]: string/struct.String.html
243 /// let s = "this is old";
245 /// assert_eq!("this is new", s.replace("old", "new"));
248 /// When the pattern doesn't match:
251 /// let s = "this is old";
252 /// assert_eq!(s, s.replace("cookie monster", "little lamb"));
254 #[must_use = "this returns the replaced string as a new allocation, \
255 without modifying the original"]
256 #[stable(feature = "rust1", since = "1.0.0")]
258 pub fn replace<'a, P: Pattern<'a>>(&'a self, from: P, to: &str) -> String {
259 let mut result = String::new();
260 let mut last_end = 0;
261 for (start, part) in self.match_indices(from) {
262 result.push_str(unsafe { self.get_unchecked(last_end..start) });
264 last_end = start + part.len();
266 result.push_str(unsafe { self.get_unchecked(last_end..self.len()) });
270 /// Replaces first N matches of a pattern with another string.
272 /// `replacen` creates a new [`String`], and copies the data from this string slice into it.
273 /// While doing so, it attempts to find matches of a pattern. If it finds any, it
274 /// replaces them with the replacement string slice at most `count` times.
276 /// [`String`]: string/struct.String.html
283 /// let s = "foo foo 123 foo";
284 /// assert_eq!("new new 123 foo", s.replacen("foo", "new", 2));
285 /// assert_eq!("faa fao 123 foo", s.replacen('o', "a", 3));
286 /// assert_eq!("foo foo new23 foo", s.replacen(char::is_numeric, "new", 1));
289 /// When the pattern doesn't match:
292 /// let s = "this is old";
293 /// assert_eq!(s, s.replacen("cookie monster", "little lamb", 10));
295 #[must_use = "this returns the replaced string as a new allocation, \
296 without modifying the original"]
297 #[stable(feature = "str_replacen", since = "1.16.0")]
298 pub fn replacen<'a, P: Pattern<'a>>(&'a self, pat: P, to: &str, count: usize) -> String {
299 // Hope to reduce the times of re-allocation
300 let mut result = String::with_capacity(32);
301 let mut last_end = 0;
302 for (start, part) in self.match_indices(pat).take(count) {
303 result.push_str(unsafe { self.get_unchecked(last_end..start) });
305 last_end = start + part.len();
307 result.push_str(unsafe { self.get_unchecked(last_end..self.len()) });
311 /// Returns the lowercase equivalent of this string slice, as a new [`String`].
313 /// 'Lowercase' is defined according to the terms of the Unicode Derived Core Property
316 /// Since some characters can expand into multiple characters when changing
317 /// the case, this function returns a [`String`] instead of modifying the
318 /// parameter in-place.
320 /// [`String`]: string/struct.String.html
329 /// assert_eq!("hello", s.to_lowercase());
332 /// A tricky example, with sigma:
337 /// assert_eq!("σ", sigma.to_lowercase());
339 /// // but at the end of a word, it's ς, not σ:
340 /// let odysseus = "ὈΔΥΣΣΕΎΣ";
342 /// assert_eq!("ὀδυσσεύς", odysseus.to_lowercase());
345 /// Languages without case are not changed:
348 /// let new_year = "农历新年";
350 /// assert_eq!(new_year, new_year.to_lowercase());
352 #[stable(feature = "unicode_case_mapping", since = "1.2.0")]
353 pub fn to_lowercase(&self) -> String {
354 let mut s = String::with_capacity(self.len());
355 for (i, c) in self[..].char_indices() {
357 // Σ maps to σ, except at the end of a word where it maps to ς.
358 // This is the only conditional (contextual) but language-independent mapping
359 // in `SpecialCasing.txt`,
360 // so hard-code it rather than have a generic "condition" mechanism.
361 // See https://github.com/rust-lang/rust/issues/26035
362 map_uppercase_sigma(self, i, &mut s)
364 match conversions::to_lower(c) {
365 [a, '\0', _] => s.push(a),
380 fn map_uppercase_sigma(from: &str, i: usize, to: &mut String) {
381 // See http://www.unicode.org/versions/Unicode7.0.0/ch03.pdf#G33992
382 // for the definition of `Final_Sigma`.
383 debug_assert!('Σ'.len_utf8() == 2);
384 let is_word_final = case_ignoreable_then_cased(from[..i].chars().rev()) &&
385 !case_ignoreable_then_cased(from[i + 2..].chars());
386 to.push_str(if is_word_final { "ς" } else { "σ" });
389 fn case_ignoreable_then_cased<I: Iterator<Item = char>>(iter: I) -> bool {
390 use core::unicode::derived_property::{Cased, Case_Ignorable};
391 match iter.skip_while(|&c| Case_Ignorable(c)).next() {
398 /// Returns the uppercase equivalent of this string slice, as a new [`String`].
400 /// 'Uppercase' is defined according to the terms of the Unicode Derived Core Property
403 /// Since some characters can expand into multiple characters when changing
404 /// the case, this function returns a [`String`] instead of modifying the
405 /// parameter in-place.
407 /// [`String`]: string/struct.String.html
416 /// assert_eq!("HELLO", s.to_uppercase());
419 /// Scripts without case are not changed:
422 /// let new_year = "农历新年";
424 /// assert_eq!(new_year, new_year.to_uppercase());
426 #[stable(feature = "unicode_case_mapping", since = "1.2.0")]
427 pub fn to_uppercase(&self) -> String {
428 let mut s = String::with_capacity(self.len());
429 for c in self[..].chars() {
430 match conversions::to_upper(c) {
431 [a, '\0', _] => s.push(a),
446 /// Escapes each char in `s` with [`char::escape_debug`].
448 /// Note: only extended grapheme codepoints that begin the string will be
451 /// [`char::escape_debug`]: primitive.char.html#method.escape_debug
452 #[unstable(feature = "str_escape",
453 reason = "return type may change to be an iterator",
455 pub fn escape_debug(&self) -> String {
456 let mut string = String::with_capacity(self.len());
457 let mut chars = self.chars();
458 if let Some(first) = chars.next() {
459 string.extend(first.escape_debug_ext(true))
461 string.extend(chars.flat_map(|c| c.escape_debug_ext(false)));
465 /// Escapes each char in `s` with [`char::escape_default`].
467 /// [`char::escape_default`]: primitive.char.html#method.escape_default
468 #[unstable(feature = "str_escape",
469 reason = "return type may change to be an iterator",
471 pub fn escape_default(&self) -> String {
472 self.chars().flat_map(|c| c.escape_default()).collect()
475 /// Escapes each char in `s` with [`char::escape_unicode`].
477 /// [`char::escape_unicode`]: primitive.char.html#method.escape_unicode
478 #[unstable(feature = "str_escape",
479 reason = "return type may change to be an iterator",
481 pub fn escape_unicode(&self) -> String {
482 self.chars().flat_map(|c| c.escape_unicode()).collect()
485 /// Converts a [`Box<str>`] into a [`String`] without copying or allocating.
487 /// [`String`]: string/struct.String.html
488 /// [`Box<str>`]: boxed/struct.Box.html
495 /// let string = String::from("birthday gift");
496 /// let boxed_str = string.clone().into_boxed_str();
498 /// assert_eq!(boxed_str.into_string(), string);
500 #[stable(feature = "box_str", since = "1.4.0")]
502 pub fn into_string(self: Box<str>) -> String {
503 let slice = Box::<[u8]>::from(self);
504 unsafe { String::from_utf8_unchecked(slice.into_vec()) }
507 /// Creates a new [`String`] by repeating a string `n` times.
511 /// This function will panic if the capacity would overflow.
513 /// [`String`]: string/struct.String.html
520 /// assert_eq!("abc".repeat(4), String::from("abcabcabcabc"));
523 /// A panic upon overflow:
527 /// // this will panic at runtime
528 /// "0123456789abcdef".repeat(usize::max_value());
531 #[stable(feature = "repeat_str", since = "1.16.0")]
532 pub fn repeat(&self, n: usize) -> String {
533 unsafe { String::from_utf8_unchecked(self.as_bytes().repeat(n)) }
536 /// Returns a copy of this string where each character is mapped to its
537 /// ASCII upper case equivalent.
539 /// ASCII letters 'a' to 'z' are mapped to 'A' to 'Z',
540 /// but non-ASCII letters are unchanged.
542 /// To uppercase the value in-place, use [`make_ascii_uppercase`].
544 /// To uppercase ASCII characters in addition to non-ASCII characters, use
545 /// [`to_uppercase`].
550 /// let s = "Grüße, Jürgen ❤";
552 /// assert_eq!("GRüßE, JüRGEN ❤", s.to_ascii_uppercase());
555 /// [`make_ascii_uppercase`]: #method.make_ascii_uppercase
556 /// [`to_uppercase`]: #method.to_uppercase
557 #[stable(feature = "ascii_methods_on_intrinsics", since = "1.23.0")]
559 pub fn to_ascii_uppercase(&self) -> String {
560 let mut bytes = self.as_bytes().to_vec();
561 bytes.make_ascii_uppercase();
562 // make_ascii_uppercase() preserves the UTF-8 invariant.
563 unsafe { String::from_utf8_unchecked(bytes) }
566 /// Returns a copy of this string where each character is mapped to its
567 /// ASCII lower case equivalent.
569 /// ASCII letters 'A' to 'Z' are mapped to 'a' to 'z',
570 /// but non-ASCII letters are unchanged.
572 /// To lowercase the value in-place, use [`make_ascii_lowercase`].
574 /// To lowercase ASCII characters in addition to non-ASCII characters, use
575 /// [`to_lowercase`].
580 /// let s = "Grüße, Jürgen ❤";
582 /// assert_eq!("grüße, jürgen ❤", s.to_ascii_lowercase());
585 /// [`make_ascii_lowercase`]: #method.make_ascii_lowercase
586 /// [`to_lowercase`]: #method.to_lowercase
587 #[stable(feature = "ascii_methods_on_intrinsics", since = "1.23.0")]
589 pub fn to_ascii_lowercase(&self) -> String {
590 let mut bytes = self.as_bytes().to_vec();
591 bytes.make_ascii_lowercase();
592 // make_ascii_lowercase() preserves the UTF-8 invariant.
593 unsafe { String::from_utf8_unchecked(bytes) }
597 /// Converts a boxed slice of bytes to a boxed string slice without checking
598 /// that the string contains valid UTF-8.
605 /// let smile_utf8 = Box::new([226, 152, 186]);
606 /// let smile = unsafe { std::str::from_boxed_utf8_unchecked(smile_utf8) };
608 /// assert_eq!("☺", &*smile);
610 #[stable(feature = "str_box_extras", since = "1.20.0")]
612 pub unsafe fn from_boxed_utf8_unchecked(v: Box<[u8]>) -> Box<str> {
613 Box::from_raw(Box::into_raw(v) as *mut str)