alloc/string.rs
1//! A UTF-8โencoded, growable string.
2//!
3//! This module contains the [`String`] type, the [`ToString`] trait for
4//! converting to strings, and several error types that may result from
5//! working with [`String`]s.
6//!
7//! # Examples
8//!
9//! There are multiple ways to create a new [`String`] from a string literal:
10//!
11//! ```
12//! let s = "Hello".to_string();
13//!
14//! let s = String::from("world");
15//! let s: String = "also this".into();
16//! ```
17//!
18//! You can create a new [`String`] from an existing one by concatenating with
19//! `+`:
20//!
21//! ```
22//! let s = "Hello".to_string();
23//!
24//! let message = s + " world!";
25//! ```
26//!
27//! If you have a vector of valid UTF-8 bytes, you can make a [`String`] out of
28//! it. You can do the reverse too.
29//!
30//! ```
31//! let sparkle_heart = vec![240, 159, 146, 150];
32//!
33//! // We know these bytes are valid, so we'll use `unwrap()`.
34//! let sparkle_heart = String::from_utf8(sparkle_heart).unwrap();
35//!
36//! assert_eq!("๐", sparkle_heart);
37//!
38//! let bytes = sparkle_heart.into_bytes();
39//!
40//! assert_eq!(bytes, [240, 159, 146, 150]);
41//! ```
42
43#![stable(feature = "rust1", since = "1.0.0")]
44
45use core::error::Error;
46use core::iter::FusedIterator;
47#[cfg(not(no_global_oom_handling))]
48use core::iter::from_fn;
49#[cfg(not(no_global_oom_handling))]
50use core::ops::Add;
51#[cfg(not(no_global_oom_handling))]
52use core::ops::AddAssign;
53#[cfg(not(no_global_oom_handling))]
54use core::ops::Bound::{Excluded, Included, Unbounded};
55use core::ops::{self, Range, RangeBounds};
56use core::str::pattern::{Pattern, Utf8Pattern};
57use core::{fmt, hash, ptr, slice};
58
59#[cfg(not(no_global_oom_handling))]
60use crate::alloc::Allocator;
61#[cfg(not(no_global_oom_handling))]
62use crate::borrow::{Cow, ToOwned};
63use crate::boxed::Box;
64use crate::collections::TryReserveError;
65use crate::str::{self, CharIndices, Chars, Utf8Error, from_utf8_unchecked_mut};
66#[cfg(not(no_global_oom_handling))]
67use crate::str::{FromStr, from_boxed_utf8_unchecked};
68use crate::vec::{self, Vec};
69
70/// A UTF-8โencoded, growable string.
71///
72/// `String` is the most common string type. It has ownership over the contents
73/// of the string, stored in a heap-allocated buffer (see [Representation](#representation)).
74/// It is closely related to its borrowed counterpart, the primitive [`str`].
75///
76/// # Examples
77///
78/// You can create a `String` from [a literal string][`&str`] with [`String::from`]:
79///
80/// [`String::from`]: From::from
81///
82/// ```
83/// let hello = String::from("Hello, world!");
84/// ```
85///
86/// You can append a [`char`] to a `String` with the [`push`] method, and
87/// append a [`&str`] with the [`push_str`] method:
88///
89/// ```
90/// let mut hello = String::from("Hello, ");
91///
92/// hello.push('w');
93/// hello.push_str("orld!");
94/// ```
95///
96/// [`push`]: String::push
97/// [`push_str`]: String::push_str
98///
99/// If you have a vector of UTF-8 bytes, you can create a `String` from it with
100/// the [`from_utf8`] method:
101///
102/// ```
103/// // some bytes, in a vector
104/// let sparkle_heart = vec![240, 159, 146, 150];
105///
106/// // We know these bytes are valid, so we'll use `unwrap()`.
107/// let sparkle_heart = String::from_utf8(sparkle_heart).unwrap();
108///
109/// assert_eq!("๐", sparkle_heart);
110/// ```
111///
112/// [`from_utf8`]: String::from_utf8
113///
114/// # UTF-8
115///
116/// `String`s are always valid UTF-8. If you need a non-UTF-8 string, consider
117/// [`OsString`]. It is similar, but without the UTF-8 constraint. Because UTF-8
118/// is a variable width encoding, `String`s are typically smaller than an array of
119/// the same `char`s:
120///
121/// ```
122/// use std::mem;
123///
124/// // `s` is ASCII which represents each `char` as one byte
125/// let s = "hello";
126/// assert_eq!(s.len(), 5);
127///
128/// // A `char` array with the same contents would be longer because
129/// // every `char` is four bytes
130/// let s = ['h', 'e', 'l', 'l', 'o'];
131/// let size: usize = s.into_iter().map(|c| mem::size_of_val(&c)).sum();
132/// assert_eq!(size, 20);
133///
134/// // However, for non-ASCII strings, the difference will be smaller
135/// // and sometimes they are the same
136/// let s = "๐๐๐๐๐";
137/// assert_eq!(s.len(), 20);
138///
139/// let s = ['๐', '๐', '๐', '๐', '๐'];
140/// let size: usize = s.into_iter().map(|c| mem::size_of_val(&c)).sum();
141/// assert_eq!(size, 20);
142/// ```
143///
144/// This raises interesting questions as to how `s[i]` should work.
145/// What should `i` be here? Several options include byte indices and
146/// `char` indices but, because of UTF-8 encoding, only byte indices
147/// would provide constant time indexing. Getting the `i`th `char`, for
148/// example, is available using [`chars`]:
149///
150/// ```
151/// let s = "hello";
152/// let third_character = s.chars().nth(2);
153/// assert_eq!(third_character, Some('l'));
154///
155/// let s = "๐๐๐๐๐";
156/// let third_character = s.chars().nth(2);
157/// assert_eq!(third_character, Some('๐'));
158/// ```
159///
160/// Next, what should `s[i]` return? Because indexing returns a reference
161/// to underlying data it could be `&u8`, `&[u8]`, or something else similar.
162/// Since we're only providing one index, `&u8` makes the most sense but that
163/// might not be what the user expects and can be explicitly achieved with
164/// [`as_bytes()`]:
165///
166/// ```
167/// // The first byte is 104 - the byte value of `'h'`
168/// let s = "hello";
169/// assert_eq!(s.as_bytes()[0], 104);
170/// // or
171/// assert_eq!(s.as_bytes()[0], b'h');
172///
173/// // The first byte is 240 which isn't obviously useful
174/// let s = "๐๐๐๐๐";
175/// assert_eq!(s.as_bytes()[0], 240);
176/// ```
177///
178/// Due to these ambiguities/restrictions, indexing with a `usize` is simply
179/// forbidden:
180///
181/// ```compile_fail,E0277
182/// let s = "hello";
183///
184/// // The following will not compile!
185/// println!("The first letter of s is {}", s[0]);
186/// ```
187///
188/// It is more clear, however, how `&s[i..j]` should work (that is,
189/// indexing with a range). It should accept byte indices (to be constant-time)
190/// and return a `&str` which is UTF-8 encoded. This is also called "string slicing".
191/// Note this will panic if the byte indices provided are not character
192/// boundaries - see [`is_char_boundary`] for more details. See the implementations
193/// for [`SliceIndex<str>`] for more details on string slicing. For a non-panicking
194/// version of string slicing, see [`get`].
195///
196/// [`OsString`]: ../../std/ffi/struct.OsString.html "ffi::OsString"
197/// [`SliceIndex<str>`]: core::slice::SliceIndex
198/// [`as_bytes()`]: str::as_bytes
199/// [`get`]: str::get
200/// [`is_char_boundary`]: str::is_char_boundary
201///
202/// The [`bytes`] and [`chars`] methods return iterators over the bytes and
203/// codepoints of the string, respectively. To iterate over codepoints along
204/// with byte indices, use [`char_indices`].
205///
206/// [`bytes`]: str::bytes
207/// [`chars`]: str::chars
208/// [`char_indices`]: str::char_indices
209///
210/// # Deref
211///
212/// `String` implements <code>[Deref]<Target = [str]></code>, and so inherits all of [`str`]'s
213/// methods. In addition, this means that you can pass a `String` to a
214/// function which takes a [`&str`] by using an ampersand (`&`):
215///
216/// ```
217/// fn takes_str(s: &str) { }
218///
219/// let s = String::from("Hello");
220///
221/// takes_str(&s);
222/// ```
223///
224/// This will create a [`&str`] from the `String` and pass it in. This
225/// conversion is very inexpensive, and so generally, functions will accept
226/// [`&str`]s as arguments unless they need a `String` for some specific
227/// reason.
228///
229/// In certain cases Rust doesn't have enough information to make this
230/// conversion, known as [`Deref`] coercion. In the following example a string
231/// slice [`&'a str`][`&str`] implements the trait `TraitExample`, and the function
232/// `example_func` takes anything that implements the trait. In this case Rust
233/// would need to make two implicit conversions, which Rust doesn't have the
234/// means to do. For that reason, the following example will not compile.
235///
236/// ```compile_fail,E0277
237/// trait TraitExample {}
238///
239/// impl<'a> TraitExample for &'a str {}
240///
241/// fn example_func<A: TraitExample>(example_arg: A) {}
242///
243/// let example_string = String::from("example_string");
244/// example_func(&example_string);
245/// ```
246///
247/// There are two options that would work instead. The first would be to
248/// change the line `example_func(&example_string);` to
249/// `example_func(example_string.as_str());`, using the method [`as_str()`]
250/// to explicitly extract the string slice containing the string. The second
251/// way changes `example_func(&example_string);` to
252/// `example_func(&*example_string);`. In this case we are dereferencing a
253/// `String` to a [`str`], then referencing the [`str`] back to
254/// [`&str`]. The second way is more idiomatic, however both work to do the
255/// conversion explicitly rather than relying on the implicit conversion.
256///
257/// # Representation
258///
259/// A `String` is made up of three components: a pointer to some bytes, a
260/// length, and a capacity. The pointer points to the internal buffer which `String`
261/// uses to store its data. The length is the number of bytes currently stored
262/// in the buffer, and the capacity is the size of the buffer in bytes. As such,
263/// the length will always be less than or equal to the capacity.
264///
265/// This buffer is always stored on the heap.
266///
267/// You can look at these with the [`as_ptr`], [`len`], and [`capacity`]
268/// methods:
269///
270/// ```
271/// use std::mem;
272///
273/// let story = String::from("Once upon a time...");
274///
275// FIXME Update this when vec_into_raw_parts is stabilized
276/// // Prevent automatically dropping the String's data
277/// let mut story = mem::ManuallyDrop::new(story);
278///
279/// let ptr = story.as_mut_ptr();
280/// let len = story.len();
281/// let capacity = story.capacity();
282///
283/// // story has nineteen bytes
284/// assert_eq!(19, len);
285///
286/// // We can re-build a String out of ptr, len, and capacity. This is all
287/// // unsafe because we are responsible for making sure the components are
288/// // valid:
289/// let s = unsafe { String::from_raw_parts(ptr, len, capacity) } ;
290///
291/// assert_eq!(String::from("Once upon a time..."), s);
292/// ```
293///
294/// [`as_ptr`]: str::as_ptr
295/// [`len`]: String::len
296/// [`capacity`]: String::capacity
297///
298/// If a `String` has enough capacity, adding elements to it will not
299/// re-allocate. For example, consider this program:
300///
301/// ```
302/// let mut s = String::new();
303///
304/// println!("{}", s.capacity());
305///
306/// for _ in 0..5 {
307/// s.push_str("hello");
308/// println!("{}", s.capacity());
309/// }
310/// ```
311///
312/// This will output the following:
313///
314/// ```text
315/// 0
316/// 8
317/// 16
318/// 16
319/// 32
320/// 32
321/// ```
322///
323/// At first, we have no memory allocated at all, but as we append to the
324/// string, it increases its capacity appropriately. If we instead use the
325/// [`with_capacity`] method to allocate the correct capacity initially:
326///
327/// ```
328/// let mut s = String::with_capacity(25);
329///
330/// println!("{}", s.capacity());
331///
332/// for _ in 0..5 {
333/// s.push_str("hello");
334/// println!("{}", s.capacity());
335/// }
336/// ```
337///
338/// [`with_capacity`]: String::with_capacity
339///
340/// We end up with a different output:
341///
342/// ```text
343/// 25
344/// 25
345/// 25
346/// 25
347/// 25
348/// 25
349/// ```
350///
351/// Here, there's no need to allocate more memory inside the loop.
352///
353/// [str]: prim@str "str"
354/// [`str`]: prim@str "str"
355/// [`&str`]: prim@str "&str"
356/// [Deref]: core::ops::Deref "ops::Deref"
357/// [`Deref`]: core::ops::Deref "ops::Deref"
358/// [`as_str()`]: String::as_str
359#[derive(PartialEq, PartialOrd, Eq, Ord)]
360#[stable(feature = "rust1", since = "1.0.0")]
361#[cfg_attr(not(test), lang = "String")]
362pub struct String {
363 vec: Vec<u8>,
364}
365
366/// A possible error value when converting a `String` from a UTF-8 byte vector.
367///
368/// This type is the error type for the [`from_utf8`] method on [`String`]. It
369/// is designed in such a way to carefully avoid reallocations: the
370/// [`into_bytes`] method will give back the byte vector that was used in the
371/// conversion attempt.
372///
373/// [`from_utf8`]: String::from_utf8
374/// [`into_bytes`]: FromUtf8Error::into_bytes
375///
376/// The [`Utf8Error`] type provided by [`std::str`] represents an error that may
377/// occur when converting a slice of [`u8`]s to a [`&str`]. In this sense, it's
378/// an analogue to `FromUtf8Error`, and you can get one from a `FromUtf8Error`
379/// through the [`utf8_error`] method.
380///
381/// [`Utf8Error`]: str::Utf8Error "std::str::Utf8Error"
382/// [`std::str`]: core::str "std::str"
383/// [`&str`]: prim@str "&str"
384/// [`utf8_error`]: FromUtf8Error::utf8_error
385///
386/// # Examples
387///
388/// ```
389/// // some invalid bytes, in a vector
390/// let bytes = vec![0, 159];
391///
392/// let value = String::from_utf8(bytes);
393///
394/// assert!(value.is_err());
395/// assert_eq!(vec![0, 159], value.unwrap_err().into_bytes());
396/// ```
397#[stable(feature = "rust1", since = "1.0.0")]
398#[cfg_attr(not(no_global_oom_handling), derive(Clone))]
399#[derive(Debug, PartialEq, Eq)]
400pub struct FromUtf8Error {
401 bytes: Vec<u8>,
402 error: Utf8Error,
403}
404
405/// A possible error value when converting a `String` from a UTF-16 byte slice.
406///
407/// This type is the error type for the [`from_utf16`] method on [`String`].
408///
409/// [`from_utf16`]: String::from_utf16
410///
411/// # Examples
412///
413/// ```
414/// // ๐mu<invalid>ic
415/// let v = &[0xD834, 0xDD1E, 0x006d, 0x0075,
416/// 0xD800, 0x0069, 0x0063];
417///
418/// assert!(String::from_utf16(v).is_err());
419/// ```
420#[stable(feature = "rust1", since = "1.0.0")]
421#[derive(Debug)]
422pub struct FromUtf16Error(());
423
424impl String {
425 /// Creates a new empty `String`.
426 ///
427 /// Given that the `String` is empty, this will not allocate any initial
428 /// buffer. While that means that this initial operation is very
429 /// inexpensive, it may cause excessive allocation later when you add
430 /// data. If you have an idea of how much data the `String` will hold,
431 /// consider the [`with_capacity`] method to prevent excessive
432 /// re-allocation.
433 ///
434 /// [`with_capacity`]: String::with_capacity
435 ///
436 /// # Examples
437 ///
438 /// ```
439 /// let s = String::new();
440 /// ```
441 #[inline]
442 #[rustc_const_stable(feature = "const_string_new", since = "1.39.0")]
443 #[cfg_attr(not(test), rustc_diagnostic_item = "string_new")]
444 #[stable(feature = "rust1", since = "1.0.0")]
445 #[must_use]
446 pub const fn new() -> String {
447 String { vec: Vec::new() }
448 }
449
450 /// Creates a new empty `String` with at least the specified capacity.
451 ///
452 /// `String`s have an internal buffer to hold their data. The capacity is
453 /// the length of that buffer, and can be queried with the [`capacity`]
454 /// method. This method creates an empty `String`, but one with an initial
455 /// buffer that can hold at least `capacity` bytes. This is useful when you
456 /// may be appending a bunch of data to the `String`, reducing the number of
457 /// reallocations it needs to do.
458 ///
459 /// [`capacity`]: String::capacity
460 ///
461 /// If the given capacity is `0`, no allocation will occur, and this method
462 /// is identical to the [`new`] method.
463 ///
464 /// [`new`]: String::new
465 ///
466 /// # Examples
467 ///
468 /// ```
469 /// let mut s = String::with_capacity(10);
470 ///
471 /// // The String contains no chars, even though it has capacity for more
472 /// assert_eq!(s.len(), 0);
473 ///
474 /// // These are all done without reallocating...
475 /// let cap = s.capacity();
476 /// for _ in 0..10 {
477 /// s.push('a');
478 /// }
479 ///
480 /// assert_eq!(s.capacity(), cap);
481 ///
482 /// // ...but this may make the string reallocate
483 /// s.push('a');
484 /// ```
485 #[cfg(not(no_global_oom_handling))]
486 #[inline]
487 #[stable(feature = "rust1", since = "1.0.0")]
488 #[must_use]
489 pub fn with_capacity(capacity: usize) -> String {
490 String { vec: Vec::with_capacity(capacity) }
491 }
492
493 /// Creates a new empty `String` with at least the specified capacity.
494 ///
495 /// # Errors
496 ///
497 /// Returns [`Err`] if the capacity exceeds `isize::MAX` bytes,
498 /// or if the memory allocator reports failure.
499 ///
500 #[inline]
501 #[unstable(feature = "try_with_capacity", issue = "91913")]
502 pub fn try_with_capacity(capacity: usize) -> Result<String, TryReserveError> {
503 Ok(String { vec: Vec::try_with_capacity(capacity)? })
504 }
505
506 // HACK(japaric): with cfg(test) the inherent `[T]::to_vec` method, which is
507 // required for this method definition, is not available. Since we don't
508 // require this method for testing purposes, I'll just stub it
509 // NB see the slice::hack module in slice.rs for more information
510 #[inline]
511 #[cfg(test)]
512 #[allow(missing_docs)]
513 pub fn from_str(_: &str) -> String {
514 panic!("not available with cfg(test)");
515 }
516
517 /// Converts a vector of bytes to a `String`.
518 ///
519 /// A string ([`String`]) is made of bytes ([`u8`]), and a vector of bytes
520 /// ([`Vec<u8>`]) is made of bytes, so this function converts between the
521 /// two. Not all byte slices are valid `String`s, however: `String`
522 /// requires that it is valid UTF-8. `from_utf8()` checks to ensure that
523 /// the bytes are valid UTF-8, and then does the conversion.
524 ///
525 /// If you are sure that the byte slice is valid UTF-8, and you don't want
526 /// to incur the overhead of the validity check, there is an unsafe version
527 /// of this function, [`from_utf8_unchecked`], which has the same behavior
528 /// but skips the check.
529 ///
530 /// This method will take care to not copy the vector, for efficiency's
531 /// sake.
532 ///
533 /// If you need a [`&str`] instead of a `String`, consider
534 /// [`str::from_utf8`].
535 ///
536 /// The inverse of this method is [`into_bytes`].
537 ///
538 /// # Errors
539 ///
540 /// Returns [`Err`] if the slice is not UTF-8 with a description as to why the
541 /// provided bytes are not UTF-8. The vector you moved in is also included.
542 ///
543 /// # Examples
544 ///
545 /// Basic usage:
546 ///
547 /// ```
548 /// // some bytes, in a vector
549 /// let sparkle_heart = vec![240, 159, 146, 150];
550 ///
551 /// // We know these bytes are valid, so we'll use `unwrap()`.
552 /// let sparkle_heart = String::from_utf8(sparkle_heart).unwrap();
553 ///
554 /// assert_eq!("๐", sparkle_heart);
555 /// ```
556 ///
557 /// Incorrect bytes:
558 ///
559 /// ```
560 /// // some invalid bytes, in a vector
561 /// let sparkle_heart = vec![0, 159, 146, 150];
562 ///
563 /// assert!(String::from_utf8(sparkle_heart).is_err());
564 /// ```
565 ///
566 /// See the docs for [`FromUtf8Error`] for more details on what you can do
567 /// with this error.
568 ///
569 /// [`from_utf8_unchecked`]: String::from_utf8_unchecked
570 /// [`Vec<u8>`]: crate::vec::Vec "Vec"
571 /// [`&str`]: prim@str "&str"
572 /// [`into_bytes`]: String::into_bytes
573 #[inline]
574 #[stable(feature = "rust1", since = "1.0.0")]
575 #[cfg_attr(not(test), rustc_diagnostic_item = "string_from_utf8")]
576 pub fn from_utf8(vec: Vec<u8>) -> Result<String, FromUtf8Error> {
577 match str::from_utf8(&vec) {
578 Ok(..) => Ok(String { vec }),
579 Err(e) => Err(FromUtf8Error { bytes: vec, error: e }),
580 }
581 }
582
583 /// Converts a slice of bytes to a string, including invalid characters.
584 ///
585 /// Strings are made of bytes ([`u8`]), and a slice of bytes
586 /// ([`&[u8]`][byteslice]) is made of bytes, so this function converts
587 /// between the two. Not all byte slices are valid strings, however: strings
588 /// are required to be valid UTF-8. During this conversion,
589 /// `from_utf8_lossy()` will replace any invalid UTF-8 sequences with
590 /// [`U+FFFD REPLACEMENT CHARACTER`][U+FFFD], which looks like this: ๏ฟฝ
591 ///
592 /// [byteslice]: prim@slice
593 /// [U+FFFD]: core::char::REPLACEMENT_CHARACTER
594 ///
595 /// If you are sure that the byte slice is valid UTF-8, and you don't want
596 /// to incur the overhead of the conversion, there is an unsafe version
597 /// of this function, [`from_utf8_unchecked`], which has the same behavior
598 /// but skips the checks.
599 ///
600 /// [`from_utf8_unchecked`]: String::from_utf8_unchecked
601 ///
602 /// This function returns a [`Cow<'a, str>`]. If our byte slice is invalid
603 /// UTF-8, then we need to insert the replacement characters, which will
604 /// change the size of the string, and hence, require a `String`. But if
605 /// it's already valid UTF-8, we don't need a new allocation. This return
606 /// type allows us to handle both cases.
607 ///
608 /// [`Cow<'a, str>`]: crate::borrow::Cow "borrow::Cow"
609 ///
610 /// # Examples
611 ///
612 /// Basic usage:
613 ///
614 /// ```
615 /// // some bytes, in a vector
616 /// let sparkle_heart = vec![240, 159, 146, 150];
617 ///
618 /// let sparkle_heart = String::from_utf8_lossy(&sparkle_heart);
619 ///
620 /// assert_eq!("๐", sparkle_heart);
621 /// ```
622 ///
623 /// Incorrect bytes:
624 ///
625 /// ```
626 /// // some invalid bytes
627 /// let input = b"Hello \xF0\x90\x80World";
628 /// let output = String::from_utf8_lossy(input);
629 ///
630 /// assert_eq!("Hello ๏ฟฝWorld", output);
631 /// ```
632 #[must_use]
633 #[cfg(not(no_global_oom_handling))]
634 #[stable(feature = "rust1", since = "1.0.0")]
635 pub fn from_utf8_lossy(v: &[u8]) -> Cow<'_, str> {
636 let mut iter = v.utf8_chunks();
637
638 let first_valid = if let Some(chunk) = iter.next() {
639 let valid = chunk.valid();
640 if chunk.invalid().is_empty() {
641 debug_assert_eq!(valid.len(), v.len());
642 return Cow::Borrowed(valid);
643 }
644 valid
645 } else {
646 return Cow::Borrowed("");
647 };
648
649 const REPLACEMENT: &str = "\u{FFFD}";
650
651 let mut res = String::with_capacity(v.len());
652 res.push_str(first_valid);
653 res.push_str(REPLACEMENT);
654
655 for chunk in iter {
656 res.push_str(chunk.valid());
657 if !chunk.invalid().is_empty() {
658 res.push_str(REPLACEMENT);
659 }
660 }
661
662 Cow::Owned(res)
663 }
664
665 /// Converts a [`Vec<u8>`] to a `String`, substituting invalid UTF-8
666 /// sequences with replacement characters.
667 ///
668 /// See [`from_utf8_lossy`] for more details.
669 ///
670 /// [`from_utf8_lossy`]: String::from_utf8_lossy
671 ///
672 /// Note that this function does not guarantee reuse of the original `Vec`
673 /// allocation.
674 ///
675 /// # Examples
676 ///
677 /// Basic usage:
678 ///
679 /// ```
680 /// #![feature(string_from_utf8_lossy_owned)]
681 /// // some bytes, in a vector
682 /// let sparkle_heart = vec![240, 159, 146, 150];
683 ///
684 /// let sparkle_heart = String::from_utf8_lossy_owned(sparkle_heart);
685 ///
686 /// assert_eq!(String::from("๐"), sparkle_heart);
687 /// ```
688 ///
689 /// Incorrect bytes:
690 ///
691 /// ```
692 /// #![feature(string_from_utf8_lossy_owned)]
693 /// // some invalid bytes
694 /// let input: Vec<u8> = b"Hello \xF0\x90\x80World".into();
695 /// let output = String::from_utf8_lossy_owned(input);
696 ///
697 /// assert_eq!(String::from("Hello ๏ฟฝWorld"), output);
698 /// ```
699 #[must_use]
700 #[cfg(not(no_global_oom_handling))]
701 #[unstable(feature = "string_from_utf8_lossy_owned", issue = "129436")]
702 pub fn from_utf8_lossy_owned(v: Vec<u8>) -> String {
703 if let Cow::Owned(string) = String::from_utf8_lossy(&v) {
704 string
705 } else {
706 // SAFETY: `String::from_utf8_lossy`'s contract ensures that if
707 // it returns a `Cow::Borrowed`, it is a valid UTF-8 string.
708 // Otherwise, it returns a new allocation of an owned `String`, with
709 // replacement characters for invalid sequences, which is returned
710 // above.
711 unsafe { String::from_utf8_unchecked(v) }
712 }
713 }
714
715 /// Decode a native endian UTF-16โencoded vector `v` into a `String`,
716 /// returning [`Err`] if `v` contains any invalid data.
717 ///
718 /// # Examples
719 ///
720 /// ```
721 /// // ๐music
722 /// let v = &[0xD834, 0xDD1E, 0x006d, 0x0075,
723 /// 0x0073, 0x0069, 0x0063];
724 /// assert_eq!(String::from("๐music"),
725 /// String::from_utf16(v).unwrap());
726 ///
727 /// // ๐mu<invalid>ic
728 /// let v = &[0xD834, 0xDD1E, 0x006d, 0x0075,
729 /// 0xD800, 0x0069, 0x0063];
730 /// assert!(String::from_utf16(v).is_err());
731 /// ```
732 #[cfg(not(no_global_oom_handling))]
733 #[stable(feature = "rust1", since = "1.0.0")]
734 pub fn from_utf16(v: &[u16]) -> Result<String, FromUtf16Error> {
735 // This isn't done via collect::<Result<_, _>>() for performance reasons.
736 // FIXME: the function can be simplified again when #48994 is closed.
737 let mut ret = String::with_capacity(v.len());
738 for c in char::decode_utf16(v.iter().cloned()) {
739 if let Ok(c) = c {
740 ret.push(c);
741 } else {
742 return Err(FromUtf16Error(()));
743 }
744 }
745 Ok(ret)
746 }
747
748 /// Decode a native endian UTF-16โencoded slice `v` into a `String`,
749 /// replacing invalid data with [the replacement character (`U+FFFD`)][U+FFFD].
750 ///
751 /// Unlike [`from_utf8_lossy`] which returns a [`Cow<'a, str>`],
752 /// `from_utf16_lossy` returns a `String` since the UTF-16 to UTF-8
753 /// conversion requires a memory allocation.
754 ///
755 /// [`from_utf8_lossy`]: String::from_utf8_lossy
756 /// [`Cow<'a, str>`]: crate::borrow::Cow "borrow::Cow"
757 /// [U+FFFD]: core::char::REPLACEMENT_CHARACTER
758 ///
759 /// # Examples
760 ///
761 /// ```
762 /// // ๐mus<invalid>ic<invalid>
763 /// let v = &[0xD834, 0xDD1E, 0x006d, 0x0075,
764 /// 0x0073, 0xDD1E, 0x0069, 0x0063,
765 /// 0xD834];
766 ///
767 /// assert_eq!(String::from("๐mus\u{FFFD}ic\u{FFFD}"),
768 /// String::from_utf16_lossy(v));
769 /// ```
770 #[cfg(not(no_global_oom_handling))]
771 #[must_use]
772 #[inline]
773 #[stable(feature = "rust1", since = "1.0.0")]
774 pub fn from_utf16_lossy(v: &[u16]) -> String {
775 char::decode_utf16(v.iter().cloned())
776 .map(|r| r.unwrap_or(char::REPLACEMENT_CHARACTER))
777 .collect()
778 }
779
780 /// Decode a UTF-16LEโencoded vector `v` into a `String`,
781 /// returning [`Err`] if `v` contains any invalid data.
782 ///
783 /// # Examples
784 ///
785 /// Basic usage:
786 ///
787 /// ```
788 /// #![feature(str_from_utf16_endian)]
789 /// // ๐music
790 /// let v = &[0x34, 0xD8, 0x1E, 0xDD, 0x6d, 0x00, 0x75, 0x00,
791 /// 0x73, 0x00, 0x69, 0x00, 0x63, 0x00];
792 /// assert_eq!(String::from("๐music"),
793 /// String::from_utf16le(v).unwrap());
794 ///
795 /// // ๐mu<invalid>ic
796 /// let v = &[0x34, 0xD8, 0x1E, 0xDD, 0x6d, 0x00, 0x75, 0x00,
797 /// 0x00, 0xD8, 0x69, 0x00, 0x63, 0x00];
798 /// assert!(String::from_utf16le(v).is_err());
799 /// ```
800 #[cfg(not(no_global_oom_handling))]
801 #[unstable(feature = "str_from_utf16_endian", issue = "116258")]
802 pub fn from_utf16le(v: &[u8]) -> Result<String, FromUtf16Error> {
803 if v.len() % 2 != 0 {
804 return Err(FromUtf16Error(()));
805 }
806 match (cfg!(target_endian = "little"), unsafe { v.align_to::<u16>() }) {
807 (true, ([], v, [])) => Self::from_utf16(v),
808 _ => char::decode_utf16(v.array_chunks::<2>().copied().map(u16::from_le_bytes))
809 .collect::<Result<_, _>>()
810 .map_err(|_| FromUtf16Error(())),
811 }
812 }
813
814 /// Decode a UTF-16LEโencoded slice `v` into a `String`, replacing
815 /// invalid data with [the replacement character (`U+FFFD`)][U+FFFD].
816 ///
817 /// Unlike [`from_utf8_lossy`] which returns a [`Cow<'a, str>`],
818 /// `from_utf16le_lossy` returns a `String` since the UTF-16 to UTF-8
819 /// conversion requires a memory allocation.
820 ///
821 /// [`from_utf8_lossy`]: String::from_utf8_lossy
822 /// [`Cow<'a, str>`]: crate::borrow::Cow "borrow::Cow"
823 /// [U+FFFD]: core::char::REPLACEMENT_CHARACTER
824 ///
825 /// # Examples
826 ///
827 /// Basic usage:
828 ///
829 /// ```
830 /// #![feature(str_from_utf16_endian)]
831 /// // ๐mus<invalid>ic<invalid>
832 /// let v = &[0x34, 0xD8, 0x1E, 0xDD, 0x6d, 0x00, 0x75, 0x00,
833 /// 0x73, 0x00, 0x1E, 0xDD, 0x69, 0x00, 0x63, 0x00,
834 /// 0x34, 0xD8];
835 ///
836 /// assert_eq!(String::from("๐mus\u{FFFD}ic\u{FFFD}"),
837 /// String::from_utf16le_lossy(v));
838 /// ```
839 #[cfg(not(no_global_oom_handling))]
840 #[unstable(feature = "str_from_utf16_endian", issue = "116258")]
841 pub fn from_utf16le_lossy(v: &[u8]) -> String {
842 match (cfg!(target_endian = "little"), unsafe { v.align_to::<u16>() }) {
843 (true, ([], v, [])) => Self::from_utf16_lossy(v),
844 (true, ([], v, [_remainder])) => Self::from_utf16_lossy(v) + "\u{FFFD}",
845 _ => {
846 let mut iter = v.array_chunks::<2>();
847 let string = char::decode_utf16(iter.by_ref().copied().map(u16::from_le_bytes))
848 .map(|r| r.unwrap_or(char::REPLACEMENT_CHARACTER))
849 .collect();
850 if iter.remainder().is_empty() { string } else { string + "\u{FFFD}" }
851 }
852 }
853 }
854
855 /// Decode a UTF-16BEโencoded vector `v` into a `String`,
856 /// returning [`Err`] if `v` contains any invalid data.
857 ///
858 /// # Examples
859 ///
860 /// Basic usage:
861 ///
862 /// ```
863 /// #![feature(str_from_utf16_endian)]
864 /// // ๐music
865 /// let v = &[0xD8, 0x34, 0xDD, 0x1E, 0x00, 0x6d, 0x00, 0x75,
866 /// 0x00, 0x73, 0x00, 0x69, 0x00, 0x63];
867 /// assert_eq!(String::from("๐music"),
868 /// String::from_utf16be(v).unwrap());
869 ///
870 /// // ๐mu<invalid>ic
871 /// let v = &[0xD8, 0x34, 0xDD, 0x1E, 0x00, 0x6d, 0x00, 0x75,
872 /// 0xD8, 0x00, 0x00, 0x69, 0x00, 0x63];
873 /// assert!(String::from_utf16be(v).is_err());
874 /// ```
875 #[cfg(not(no_global_oom_handling))]
876 #[unstable(feature = "str_from_utf16_endian", issue = "116258")]
877 pub fn from_utf16be(v: &[u8]) -> Result<String, FromUtf16Error> {
878 if v.len() % 2 != 0 {
879 return Err(FromUtf16Error(()));
880 }
881 match (cfg!(target_endian = "big"), unsafe { v.align_to::<u16>() }) {
882 (true, ([], v, [])) => Self::from_utf16(v),
883 _ => char::decode_utf16(v.array_chunks::<2>().copied().map(u16::from_be_bytes))
884 .collect::<Result<_, _>>()
885 .map_err(|_| FromUtf16Error(())),
886 }
887 }
888
889 /// Decode a UTF-16BEโencoded slice `v` into a `String`, replacing
890 /// invalid data with [the replacement character (`U+FFFD`)][U+FFFD].
891 ///
892 /// Unlike [`from_utf8_lossy`] which returns a [`Cow<'a, str>`],
893 /// `from_utf16le_lossy` returns a `String` since the UTF-16 to UTF-8
894 /// conversion requires a memory allocation.
895 ///
896 /// [`from_utf8_lossy`]: String::from_utf8_lossy
897 /// [`Cow<'a, str>`]: crate::borrow::Cow "borrow::Cow"
898 /// [U+FFFD]: core::char::REPLACEMENT_CHARACTER
899 ///
900 /// # Examples
901 ///
902 /// Basic usage:
903 ///
904 /// ```
905 /// #![feature(str_from_utf16_endian)]
906 /// // ๐mus<invalid>ic<invalid>
907 /// let v = &[0xD8, 0x34, 0xDD, 0x1E, 0x00, 0x6d, 0x00, 0x75,
908 /// 0x00, 0x73, 0xDD, 0x1E, 0x00, 0x69, 0x00, 0x63,
909 /// 0xD8, 0x34];
910 ///
911 /// assert_eq!(String::from("๐mus\u{FFFD}ic\u{FFFD}"),
912 /// String::from_utf16be_lossy(v));
913 /// ```
914 #[cfg(not(no_global_oom_handling))]
915 #[unstable(feature = "str_from_utf16_endian", issue = "116258")]
916 pub fn from_utf16be_lossy(v: &[u8]) -> String {
917 match (cfg!(target_endian = "big"), unsafe { v.align_to::<u16>() }) {
918 (true, ([], v, [])) => Self::from_utf16_lossy(v),
919 (true, ([], v, [_remainder])) => Self::from_utf16_lossy(v) + "\u{FFFD}",
920 _ => {
921 let mut iter = v.array_chunks::<2>();
922 let string = char::decode_utf16(iter.by_ref().copied().map(u16::from_be_bytes))
923 .map(|r| r.unwrap_or(char::REPLACEMENT_CHARACTER))
924 .collect();
925 if iter.remainder().is_empty() { string } else { string + "\u{FFFD}" }
926 }
927 }
928 }
929
930 /// Decomposes a `String` into its raw components: `(pointer, length, capacity)`.
931 ///
932 /// Returns the raw pointer to the underlying data, the length of
933 /// the string (in bytes), and the allocated capacity of the data
934 /// (in bytes). These are the same arguments in the same order as
935 /// the arguments to [`from_raw_parts`].
936 ///
937 /// After calling this function, the caller is responsible for the
938 /// memory previously managed by the `String`. The only way to do
939 /// this is to convert the raw pointer, length, and capacity back
940 /// into a `String` with the [`from_raw_parts`] function, allowing
941 /// the destructor to perform the cleanup.
942 ///
943 /// [`from_raw_parts`]: String::from_raw_parts
944 ///
945 /// # Examples
946 ///
947 /// ```
948 /// #![feature(vec_into_raw_parts)]
949 /// let s = String::from("hello");
950 ///
951 /// let (ptr, len, cap) = s.into_raw_parts();
952 ///
953 /// let rebuilt = unsafe { String::from_raw_parts(ptr, len, cap) };
954 /// assert_eq!(rebuilt, "hello");
955 /// ```
956 #[must_use = "losing the pointer will leak memory"]
957 #[unstable(feature = "vec_into_raw_parts", reason = "new API", issue = "65816")]
958 pub fn into_raw_parts(self) -> (*mut u8, usize, usize) {
959 self.vec.into_raw_parts()
960 }
961
962 /// Creates a new `String` from a pointer, a length and a capacity.
963 ///
964 /// # Safety
965 ///
966 /// This is highly unsafe, due to the number of invariants that aren't
967 /// checked:
968 ///
969 /// * all safety requirements for [`Vec::<u8>::from_raw_parts`].
970 /// * all safety requirements for [`String::from_utf8_unchecked`].
971 ///
972 /// Violating these may cause problems like corrupting the allocator's
973 /// internal data structures. For example, it is normally **not** safe to
974 /// build a `String` from a pointer to a C `char` array containing UTF-8
975 /// _unless_ you are certain that array was originally allocated by the
976 /// Rust standard library's allocator.
977 ///
978 /// The ownership of `buf` is effectively transferred to the
979 /// `String` which may then deallocate, reallocate or change the
980 /// contents of memory pointed to by the pointer at will. Ensure
981 /// that nothing else uses the pointer after calling this
982 /// function.
983 ///
984 /// # Examples
985 ///
986 /// ```
987 /// use std::mem;
988 ///
989 /// unsafe {
990 /// let s = String::from("hello");
991 ///
992 // FIXME Update this when vec_into_raw_parts is stabilized
993 /// // Prevent automatically dropping the String's data
994 /// let mut s = mem::ManuallyDrop::new(s);
995 ///
996 /// let ptr = s.as_mut_ptr();
997 /// let len = s.len();
998 /// let capacity = s.capacity();
999 ///
1000 /// let s = String::from_raw_parts(ptr, len, capacity);
1001 ///
1002 /// assert_eq!(String::from("hello"), s);
1003 /// }
1004 /// ```
1005 #[inline]
1006 #[stable(feature = "rust1", since = "1.0.0")]
1007 pub unsafe fn from_raw_parts(buf: *mut u8, length: usize, capacity: usize) -> String {
1008 unsafe { String { vec: Vec::from_raw_parts(buf, length, capacity) } }
1009 }
1010
1011 /// Converts a vector of bytes to a `String` without checking that the
1012 /// string contains valid UTF-8.
1013 ///
1014 /// See the safe version, [`from_utf8`], for more details.
1015 ///
1016 /// [`from_utf8`]: String::from_utf8
1017 ///
1018 /// # Safety
1019 ///
1020 /// This function is unsafe because it does not check that the bytes passed
1021 /// to it are valid UTF-8. If this constraint is violated, it may cause
1022 /// memory unsafety issues with future users of the `String`, as the rest of
1023 /// the standard library assumes that `String`s are valid UTF-8.
1024 ///
1025 /// # Examples
1026 ///
1027 /// ```
1028 /// // some bytes, in a vector
1029 /// let sparkle_heart = vec![240, 159, 146, 150];
1030 ///
1031 /// let sparkle_heart = unsafe {
1032 /// String::from_utf8_unchecked(sparkle_heart)
1033 /// };
1034 ///
1035 /// assert_eq!("๐", sparkle_heart);
1036 /// ```
1037 #[inline]
1038 #[must_use]
1039 #[stable(feature = "rust1", since = "1.0.0")]
1040 pub unsafe fn from_utf8_unchecked(bytes: Vec<u8>) -> String {
1041 String { vec: bytes }
1042 }
1043
1044 /// Converts a `String` into a byte vector.
1045 ///
1046 /// This consumes the `String`, so we do not need to copy its contents.
1047 ///
1048 /// # Examples
1049 ///
1050 /// ```
1051 /// let s = String::from("hello");
1052 /// let bytes = s.into_bytes();
1053 ///
1054 /// assert_eq!(&[104, 101, 108, 108, 111][..], &bytes[..]);
1055 /// ```
1056 #[inline]
1057 #[must_use = "`self` will be dropped if the result is not used"]
1058 #[stable(feature = "rust1", since = "1.0.0")]
1059 #[rustc_const_unstable(feature = "const_vec_string_slice", issue = "129041")]
1060 pub const fn into_bytes(self) -> Vec<u8> {
1061 self.vec
1062 }
1063
1064 /// Extracts a string slice containing the entire `String`.
1065 ///
1066 /// # Examples
1067 ///
1068 /// ```
1069 /// let s = String::from("foo");
1070 ///
1071 /// assert_eq!("foo", s.as_str());
1072 /// ```
1073 #[inline]
1074 #[must_use]
1075 #[stable(feature = "string_as_str", since = "1.7.0")]
1076 #[cfg_attr(not(test), rustc_diagnostic_item = "string_as_str")]
1077 #[rustc_const_unstable(feature = "const_vec_string_slice", issue = "129041")]
1078 pub const fn as_str(&self) -> &str {
1079 // SAFETY: String contents are stipulated to be valid UTF-8, invalid contents are an error
1080 // at construction.
1081 unsafe { str::from_utf8_unchecked(self.vec.as_slice()) }
1082 }
1083
1084 /// Converts a `String` into a mutable string slice.
1085 ///
1086 /// # Examples
1087 ///
1088 /// ```
1089 /// let mut s = String::from("foobar");
1090 /// let s_mut_str = s.as_mut_str();
1091 ///
1092 /// s_mut_str.make_ascii_uppercase();
1093 ///
1094 /// assert_eq!("FOOBAR", s_mut_str);
1095 /// ```
1096 #[inline]
1097 #[must_use]
1098 #[stable(feature = "string_as_str", since = "1.7.0")]
1099 #[cfg_attr(not(test), rustc_diagnostic_item = "string_as_mut_str")]
1100 #[rustc_const_unstable(feature = "const_vec_string_slice", issue = "129041")]
1101 pub const fn as_mut_str(&mut self) -> &mut str {
1102 // SAFETY: String contents are stipulated to be valid UTF-8, invalid contents are an error
1103 // at construction.
1104 unsafe { str::from_utf8_unchecked_mut(self.vec.as_mut_slice()) }
1105 }
1106
1107 /// Appends a given string slice onto the end of this `String`.
1108 ///
1109 /// # Examples
1110 ///
1111 /// ```
1112 /// let mut s = String::from("foo");
1113 ///
1114 /// s.push_str("bar");
1115 ///
1116 /// assert_eq!("foobar", s);
1117 /// ```
1118 #[cfg(not(no_global_oom_handling))]
1119 #[inline]
1120 #[stable(feature = "rust1", since = "1.0.0")]
1121 #[rustc_confusables("append", "push")]
1122 #[cfg_attr(not(test), rustc_diagnostic_item = "string_push_str")]
1123 pub fn push_str(&mut self, string: &str) {
1124 self.vec.extend_from_slice(string.as_bytes())
1125 }
1126
1127 /// Copies elements from `src` range to the end of the string.
1128 ///
1129 /// # Panics
1130 ///
1131 /// Panics if the starting point or end point do not lie on a [`char`]
1132 /// boundary, or if they're out of bounds.
1133 ///
1134 /// # Examples
1135 ///
1136 /// ```
1137 /// #![feature(string_extend_from_within)]
1138 /// let mut string = String::from("abcde");
1139 ///
1140 /// string.extend_from_within(2..);
1141 /// assert_eq!(string, "abcdecde");
1142 ///
1143 /// string.extend_from_within(..2);
1144 /// assert_eq!(string, "abcdecdeab");
1145 ///
1146 /// string.extend_from_within(4..8);
1147 /// assert_eq!(string, "abcdecdeabecde");
1148 /// ```
1149 #[cfg(not(no_global_oom_handling))]
1150 #[unstable(feature = "string_extend_from_within", issue = "103806")]
1151 pub fn extend_from_within<R>(&mut self, src: R)
1152 where
1153 R: RangeBounds<usize>,
1154 {
1155 let src @ Range { start, end } = slice::range(src, ..self.len());
1156
1157 assert!(self.is_char_boundary(start));
1158 assert!(self.is_char_boundary(end));
1159
1160 self.vec.extend_from_within(src);
1161 }
1162
1163 /// Returns this `String`'s capacity, in bytes.
1164 ///
1165 /// # Examples
1166 ///
1167 /// ```
1168 /// let s = String::with_capacity(10);
1169 ///
1170 /// assert!(s.capacity() >= 10);
1171 /// ```
1172 #[inline]
1173 #[must_use]
1174 #[stable(feature = "rust1", since = "1.0.0")]
1175 #[rustc_const_unstable(feature = "const_vec_string_slice", issue = "129041")]
1176 pub const fn capacity(&self) -> usize {
1177 self.vec.capacity()
1178 }
1179
1180 /// Reserves capacity for at least `additional` bytes more than the
1181 /// current length. The allocator may reserve more space to speculatively
1182 /// avoid frequent allocations. After calling `reserve`,
1183 /// capacity will be greater than or equal to `self.len() + additional`.
1184 /// Does nothing if capacity is already sufficient.
1185 ///
1186 /// # Panics
1187 ///
1188 /// Panics if the new capacity overflows [`usize`].
1189 ///
1190 /// # Examples
1191 ///
1192 /// Basic usage:
1193 ///
1194 /// ```
1195 /// let mut s = String::new();
1196 ///
1197 /// s.reserve(10);
1198 ///
1199 /// assert!(s.capacity() >= 10);
1200 /// ```
1201 ///
1202 /// This might not actually increase the capacity:
1203 ///
1204 /// ```
1205 /// let mut s = String::with_capacity(10);
1206 /// s.push('a');
1207 /// s.push('b');
1208 ///
1209 /// // s now has a length of 2 and a capacity of at least 10
1210 /// let capacity = s.capacity();
1211 /// assert_eq!(2, s.len());
1212 /// assert!(capacity >= 10);
1213 ///
1214 /// // Since we already have at least an extra 8 capacity, calling this...
1215 /// s.reserve(8);
1216 ///
1217 /// // ... doesn't actually increase.
1218 /// assert_eq!(capacity, s.capacity());
1219 /// ```
1220 #[cfg(not(no_global_oom_handling))]
1221 #[inline]
1222 #[stable(feature = "rust1", since = "1.0.0")]
1223 pub fn reserve(&mut self, additional: usize) {
1224 self.vec.reserve(additional)
1225 }
1226
1227 /// Reserves the minimum capacity for at least `additional` bytes more than
1228 /// the current length. Unlike [`reserve`], this will not
1229 /// deliberately over-allocate to speculatively avoid frequent allocations.
1230 /// After calling `reserve_exact`, capacity will be greater than or equal to
1231 /// `self.len() + additional`. Does nothing if the capacity is already
1232 /// sufficient.
1233 ///
1234 /// [`reserve`]: String::reserve
1235 ///
1236 /// # Panics
1237 ///
1238 /// Panics if the new capacity overflows [`usize`].
1239 ///
1240 /// # Examples
1241 ///
1242 /// Basic usage:
1243 ///
1244 /// ```
1245 /// let mut s = String::new();
1246 ///
1247 /// s.reserve_exact(10);
1248 ///
1249 /// assert!(s.capacity() >= 10);
1250 /// ```
1251 ///
1252 /// This might not actually increase the capacity:
1253 ///
1254 /// ```
1255 /// let mut s = String::with_capacity(10);
1256 /// s.push('a');
1257 /// s.push('b');
1258 ///
1259 /// // s now has a length of 2 and a capacity of at least 10
1260 /// let capacity = s.capacity();
1261 /// assert_eq!(2, s.len());
1262 /// assert!(capacity >= 10);
1263 ///
1264 /// // Since we already have at least an extra 8 capacity, calling this...
1265 /// s.reserve_exact(8);
1266 ///
1267 /// // ... doesn't actually increase.
1268 /// assert_eq!(capacity, s.capacity());
1269 /// ```
1270 #[cfg(not(no_global_oom_handling))]
1271 #[inline]
1272 #[stable(feature = "rust1", since = "1.0.0")]
1273 pub fn reserve_exact(&mut self, additional: usize) {
1274 self.vec.reserve_exact(additional)
1275 }
1276
1277 /// Tries to reserve capacity for at least `additional` bytes more than the
1278 /// current length. The allocator may reserve more space to speculatively
1279 /// avoid frequent allocations. After calling `try_reserve`, capacity will be
1280 /// greater than or equal to `self.len() + additional` if it returns
1281 /// `Ok(())`. Does nothing if capacity is already sufficient. This method
1282 /// preserves the contents even if an error occurs.
1283 ///
1284 /// # Errors
1285 ///
1286 /// If the capacity overflows, or the allocator reports a failure, then an error
1287 /// is returned.
1288 ///
1289 /// # Examples
1290 ///
1291 /// ```
1292 /// use std::collections::TryReserveError;
1293 ///
1294 /// fn process_data(data: &str) -> Result<String, TryReserveError> {
1295 /// let mut output = String::new();
1296 ///
1297 /// // Pre-reserve the memory, exiting if we can't
1298 /// output.try_reserve(data.len())?;
1299 ///
1300 /// // Now we know this can't OOM in the middle of our complex work
1301 /// output.push_str(data);
1302 ///
1303 /// Ok(output)
1304 /// }
1305 /// # process_data("rust").expect("why is the test harness OOMing on 4 bytes?");
1306 /// ```
1307 #[stable(feature = "try_reserve", since = "1.57.0")]
1308 pub fn try_reserve(&mut self, additional: usize) -> Result<(), TryReserveError> {
1309 self.vec.try_reserve(additional)
1310 }
1311
1312 /// Tries to reserve the minimum capacity for at least `additional` bytes
1313 /// more than the current length. Unlike [`try_reserve`], this will not
1314 /// deliberately over-allocate to speculatively avoid frequent allocations.
1315 /// After calling `try_reserve_exact`, capacity will be greater than or
1316 /// equal to `self.len() + additional` if it returns `Ok(())`.
1317 /// Does nothing if the capacity is already sufficient.
1318 ///
1319 /// Note that the allocator may give the collection more space than it
1320 /// requests. Therefore, capacity can not be relied upon to be precisely
1321 /// minimal. Prefer [`try_reserve`] if future insertions are expected.
1322 ///
1323 /// [`try_reserve`]: String::try_reserve
1324 ///
1325 /// # Errors
1326 ///
1327 /// If the capacity overflows, or the allocator reports a failure, then an error
1328 /// is returned.
1329 ///
1330 /// # Examples
1331 ///
1332 /// ```
1333 /// use std::collections::TryReserveError;
1334 ///
1335 /// fn process_data(data: &str) -> Result<String, TryReserveError> {
1336 /// let mut output = String::new();
1337 ///
1338 /// // Pre-reserve the memory, exiting if we can't
1339 /// output.try_reserve_exact(data.len())?;
1340 ///
1341 /// // Now we know this can't OOM in the middle of our complex work
1342 /// output.push_str(data);
1343 ///
1344 /// Ok(output)
1345 /// }
1346 /// # process_data("rust").expect("why is the test harness OOMing on 4 bytes?");
1347 /// ```
1348 #[stable(feature = "try_reserve", since = "1.57.0")]
1349 pub fn try_reserve_exact(&mut self, additional: usize) -> Result<(), TryReserveError> {
1350 self.vec.try_reserve_exact(additional)
1351 }
1352
1353 /// Shrinks the capacity of this `String` to match its length.
1354 ///
1355 /// # Examples
1356 ///
1357 /// ```
1358 /// let mut s = String::from("foo");
1359 ///
1360 /// s.reserve(100);
1361 /// assert!(s.capacity() >= 100);
1362 ///
1363 /// s.shrink_to_fit();
1364 /// assert_eq!(3, s.capacity());
1365 /// ```
1366 #[cfg(not(no_global_oom_handling))]
1367 #[inline]
1368 #[stable(feature = "rust1", since = "1.0.0")]
1369 pub fn shrink_to_fit(&mut self) {
1370 self.vec.shrink_to_fit()
1371 }
1372
1373 /// Shrinks the capacity of this `String` with a lower bound.
1374 ///
1375 /// The capacity will remain at least as large as both the length
1376 /// and the supplied value.
1377 ///
1378 /// If the current capacity is less than the lower limit, this is a no-op.
1379 ///
1380 /// # Examples
1381 ///
1382 /// ```
1383 /// let mut s = String::from("foo");
1384 ///
1385 /// s.reserve(100);
1386 /// assert!(s.capacity() >= 100);
1387 ///
1388 /// s.shrink_to(10);
1389 /// assert!(s.capacity() >= 10);
1390 /// s.shrink_to(0);
1391 /// assert!(s.capacity() >= 3);
1392 /// ```
1393 #[cfg(not(no_global_oom_handling))]
1394 #[inline]
1395 #[stable(feature = "shrink_to", since = "1.56.0")]
1396 pub fn shrink_to(&mut self, min_capacity: usize) {
1397 self.vec.shrink_to(min_capacity)
1398 }
1399
1400 /// Appends the given [`char`] to the end of this `String`.
1401 ///
1402 /// # Examples
1403 ///
1404 /// ```
1405 /// let mut s = String::from("abc");
1406 ///
1407 /// s.push('1');
1408 /// s.push('2');
1409 /// s.push('3');
1410 ///
1411 /// assert_eq!("abc123", s);
1412 /// ```
1413 #[cfg(not(no_global_oom_handling))]
1414 #[inline]
1415 #[stable(feature = "rust1", since = "1.0.0")]
1416 pub fn push(&mut self, ch: char) {
1417 match ch.len_utf8() {
1418 1 => self.vec.push(ch as u8),
1419 _ => {
1420 self.vec.extend_from_slice(ch.encode_utf8(&mut [0; char::MAX_LEN_UTF8]).as_bytes())
1421 }
1422 }
1423 }
1424
1425 /// Returns a byte slice of this `String`'s contents.
1426 ///
1427 /// The inverse of this method is [`from_utf8`].
1428 ///
1429 /// [`from_utf8`]: String::from_utf8
1430 ///
1431 /// # Examples
1432 ///
1433 /// ```
1434 /// let s = String::from("hello");
1435 ///
1436 /// assert_eq!(&[104, 101, 108, 108, 111], s.as_bytes());
1437 /// ```
1438 #[inline]
1439 #[must_use]
1440 #[stable(feature = "rust1", since = "1.0.0")]
1441 #[rustc_const_unstable(feature = "const_vec_string_slice", issue = "129041")]
1442 pub const fn as_bytes(&self) -> &[u8] {
1443 self.vec.as_slice()
1444 }
1445
1446 /// Shortens this `String` to the specified length.
1447 ///
1448 /// If `new_len` is greater than or equal to the string's current length, this has no
1449 /// effect.
1450 ///
1451 /// Note that this method has no effect on the allocated capacity
1452 /// of the string
1453 ///
1454 /// # Panics
1455 ///
1456 /// Panics if `new_len` does not lie on a [`char`] boundary.
1457 ///
1458 /// # Examples
1459 ///
1460 /// ```
1461 /// let mut s = String::from("hello");
1462 ///
1463 /// s.truncate(2);
1464 ///
1465 /// assert_eq!("he", s);
1466 /// ```
1467 #[inline]
1468 #[stable(feature = "rust1", since = "1.0.0")]
1469 pub fn truncate(&mut self, new_len: usize) {
1470 if new_len <= self.len() {
1471 assert!(self.is_char_boundary(new_len));
1472 self.vec.truncate(new_len)
1473 }
1474 }
1475
1476 /// Removes the last character from the string buffer and returns it.
1477 ///
1478 /// Returns [`None`] if this `String` is empty.
1479 ///
1480 /// # Examples
1481 ///
1482 /// ```
1483 /// let mut s = String::from("abฤ");
1484 ///
1485 /// assert_eq!(s.pop(), Some('ฤ'));
1486 /// assert_eq!(s.pop(), Some('b'));
1487 /// assert_eq!(s.pop(), Some('a'));
1488 ///
1489 /// assert_eq!(s.pop(), None);
1490 /// ```
1491 #[inline]
1492 #[stable(feature = "rust1", since = "1.0.0")]
1493 pub fn pop(&mut self) -> Option<char> {
1494 let ch = self.chars().rev().next()?;
1495 let newlen = self.len() - ch.len_utf8();
1496 unsafe {
1497 self.vec.set_len(newlen);
1498 }
1499 Some(ch)
1500 }
1501
1502 /// Removes a [`char`] from this `String` at a byte position and returns it.
1503 ///
1504 /// This is an *O*(*n*) operation, as it requires copying every element in the
1505 /// buffer.
1506 ///
1507 /// # Panics
1508 ///
1509 /// Panics if `idx` is larger than or equal to the `String`'s length,
1510 /// or if it does not lie on a [`char`] boundary.
1511 ///
1512 /// # Examples
1513 ///
1514 /// ```
1515 /// let mut s = String::from("abรง");
1516 ///
1517 /// assert_eq!(s.remove(0), 'a');
1518 /// assert_eq!(s.remove(1), 'รง');
1519 /// assert_eq!(s.remove(0), 'b');
1520 /// ```
1521 #[inline]
1522 #[stable(feature = "rust1", since = "1.0.0")]
1523 #[rustc_confusables("delete", "take")]
1524 pub fn remove(&mut self, idx: usize) -> char {
1525 let ch = match self[idx..].chars().next() {
1526 Some(ch) => ch,
1527 None => panic!("cannot remove a char from the end of a string"),
1528 };
1529
1530 let next = idx + ch.len_utf8();
1531 let len = self.len();
1532 unsafe {
1533 ptr::copy(self.vec.as_ptr().add(next), self.vec.as_mut_ptr().add(idx), len - next);
1534 self.vec.set_len(len - (next - idx));
1535 }
1536 ch
1537 }
1538
1539 /// Remove all matches of pattern `pat` in the `String`.
1540 ///
1541 /// # Examples
1542 ///
1543 /// ```
1544 /// #![feature(string_remove_matches)]
1545 /// let mut s = String::from("Trees are not green, the sky is not blue.");
1546 /// s.remove_matches("not ");
1547 /// assert_eq!("Trees are green, the sky is blue.", s);
1548 /// ```
1549 ///
1550 /// Matches will be detected and removed iteratively, so in cases where
1551 /// patterns overlap, only the first pattern will be removed:
1552 ///
1553 /// ```
1554 /// #![feature(string_remove_matches)]
1555 /// let mut s = String::from("banana");
1556 /// s.remove_matches("ana");
1557 /// assert_eq!("bna", s);
1558 /// ```
1559 #[cfg(not(no_global_oom_handling))]
1560 #[unstable(feature = "string_remove_matches", reason = "new API", issue = "72826")]
1561 pub fn remove_matches<P: Pattern>(&mut self, pat: P) {
1562 use core::str::pattern::Searcher;
1563
1564 let rejections = {
1565 let mut searcher = pat.into_searcher(self);
1566 // Per Searcher::next:
1567 //
1568 // A Match result needs to contain the whole matched pattern,
1569 // however Reject results may be split up into arbitrary many
1570 // adjacent fragments. Both ranges may have zero length.
1571 //
1572 // In practice the implementation of Searcher::next_match tends to
1573 // be more efficient, so we use it here and do some work to invert
1574 // matches into rejections since that's what we want to copy below.
1575 let mut front = 0;
1576 let rejections: Vec<_> = from_fn(|| {
1577 let (start, end) = searcher.next_match()?;
1578 let prev_front = front;
1579 front = end;
1580 Some((prev_front, start))
1581 })
1582 .collect();
1583 rejections.into_iter().chain(core::iter::once((front, self.len())))
1584 };
1585
1586 let mut len = 0;
1587 let ptr = self.vec.as_mut_ptr();
1588
1589 for (start, end) in rejections {
1590 let count = end - start;
1591 if start != len {
1592 // SAFETY: per Searcher::next:
1593 //
1594 // The stream of Match and Reject values up to a Done will
1595 // contain index ranges that are adjacent, non-overlapping,
1596 // covering the whole haystack, and laying on utf8
1597 // boundaries.
1598 unsafe {
1599 ptr::copy(ptr.add(start), ptr.add(len), count);
1600 }
1601 }
1602 len += count;
1603 }
1604
1605 unsafe {
1606 self.vec.set_len(len);
1607 }
1608 }
1609
1610 /// Retains only the characters specified by the predicate.
1611 ///
1612 /// In other words, remove all characters `c` such that `f(c)` returns `false`.
1613 /// This method operates in place, visiting each character exactly once in the
1614 /// original order, and preserves the order of the retained characters.
1615 ///
1616 /// # Examples
1617 ///
1618 /// ```
1619 /// let mut s = String::from("f_o_ob_ar");
1620 ///
1621 /// s.retain(|c| c != '_');
1622 ///
1623 /// assert_eq!(s, "foobar");
1624 /// ```
1625 ///
1626 /// Because the elements are visited exactly once in the original order,
1627 /// external state may be used to decide which elements to keep.
1628 ///
1629 /// ```
1630 /// let mut s = String::from("abcde");
1631 /// let keep = [false, true, true, false, true];
1632 /// let mut iter = keep.iter();
1633 /// s.retain(|_| *iter.next().unwrap());
1634 /// assert_eq!(s, "bce");
1635 /// ```
1636 #[inline]
1637 #[stable(feature = "string_retain", since = "1.26.0")]
1638 pub fn retain<F>(&mut self, mut f: F)
1639 where
1640 F: FnMut(char) -> bool,
1641 {
1642 struct SetLenOnDrop<'a> {
1643 s: &'a mut String,
1644 idx: usize,
1645 del_bytes: usize,
1646 }
1647
1648 impl<'a> Drop for SetLenOnDrop<'a> {
1649 fn drop(&mut self) {
1650 let new_len = self.idx - self.del_bytes;
1651 debug_assert!(new_len <= self.s.len());
1652 unsafe { self.s.vec.set_len(new_len) };
1653 }
1654 }
1655
1656 let len = self.len();
1657 let mut guard = SetLenOnDrop { s: self, idx: 0, del_bytes: 0 };
1658
1659 while guard.idx < len {
1660 let ch =
1661 // SAFETY: `guard.idx` is positive-or-zero and less that len so the `get_unchecked`
1662 // is in bound. `self` is valid UTF-8 like string and the returned slice starts at
1663 // a unicode code point so the `Chars` always return one character.
1664 unsafe { guard.s.get_unchecked(guard.idx..len).chars().next().unwrap_unchecked() };
1665 let ch_len = ch.len_utf8();
1666
1667 if !f(ch) {
1668 guard.del_bytes += ch_len;
1669 } else if guard.del_bytes > 0 {
1670 // SAFETY: `guard.idx` is in bound and `guard.del_bytes` represent the number of
1671 // bytes that are erased from the string so the resulting `guard.idx -
1672 // guard.del_bytes` always represent a valid unicode code point.
1673 //
1674 // `guard.del_bytes` >= `ch.len_utf8()`, so taking a slice with `ch.len_utf8()` len
1675 // is safe.
1676 ch.encode_utf8(unsafe {
1677 crate::slice::from_raw_parts_mut(
1678 guard.s.as_mut_ptr().add(guard.idx - guard.del_bytes),
1679 ch.len_utf8(),
1680 )
1681 });
1682 }
1683
1684 // Point idx to the next char
1685 guard.idx += ch_len;
1686 }
1687
1688 drop(guard);
1689 }
1690
1691 /// Inserts a character into this `String` at a byte position.
1692 ///
1693 /// This is an *O*(*n*) operation as it requires copying every element in the
1694 /// buffer.
1695 ///
1696 /// # Panics
1697 ///
1698 /// Panics if `idx` is larger than the `String`'s length, or if it does not
1699 /// lie on a [`char`] boundary.
1700 ///
1701 /// # Examples
1702 ///
1703 /// ```
1704 /// let mut s = String::with_capacity(3);
1705 ///
1706 /// s.insert(0, 'f');
1707 /// s.insert(1, 'o');
1708 /// s.insert(2, 'o');
1709 ///
1710 /// assert_eq!("foo", s);
1711 /// ```
1712 #[cfg(not(no_global_oom_handling))]
1713 #[inline]
1714 #[stable(feature = "rust1", since = "1.0.0")]
1715 #[rustc_confusables("set")]
1716 pub fn insert(&mut self, idx: usize, ch: char) {
1717 assert!(self.is_char_boundary(idx));
1718 let mut bits = [0; char::MAX_LEN_UTF8];
1719 let bits = ch.encode_utf8(&mut bits).as_bytes();
1720
1721 unsafe {
1722 self.insert_bytes(idx, bits);
1723 }
1724 }
1725
1726 #[cfg(not(no_global_oom_handling))]
1727 unsafe fn insert_bytes(&mut self, idx: usize, bytes: &[u8]) {
1728 let len = self.len();
1729 let amt = bytes.len();
1730 self.vec.reserve(amt);
1731
1732 unsafe {
1733 ptr::copy(self.vec.as_ptr().add(idx), self.vec.as_mut_ptr().add(idx + amt), len - idx);
1734 ptr::copy_nonoverlapping(bytes.as_ptr(), self.vec.as_mut_ptr().add(idx), amt);
1735 self.vec.set_len(len + amt);
1736 }
1737 }
1738
1739 /// Inserts a string slice into this `String` at a byte position.
1740 ///
1741 /// This is an *O*(*n*) operation as it requires copying every element in the
1742 /// buffer.
1743 ///
1744 /// # Panics
1745 ///
1746 /// Panics if `idx` is larger than the `String`'s length, or if it does not
1747 /// lie on a [`char`] boundary.
1748 ///
1749 /// # Examples
1750 ///
1751 /// ```
1752 /// let mut s = String::from("bar");
1753 ///
1754 /// s.insert_str(0, "foo");
1755 ///
1756 /// assert_eq!("foobar", s);
1757 /// ```
1758 #[cfg(not(no_global_oom_handling))]
1759 #[inline]
1760 #[stable(feature = "insert_str", since = "1.16.0")]
1761 #[cfg_attr(not(test), rustc_diagnostic_item = "string_insert_str")]
1762 pub fn insert_str(&mut self, idx: usize, string: &str) {
1763 assert!(self.is_char_boundary(idx));
1764
1765 unsafe {
1766 self.insert_bytes(idx, string.as_bytes());
1767 }
1768 }
1769
1770 /// Returns a mutable reference to the contents of this `String`.
1771 ///
1772 /// # Safety
1773 ///
1774 /// This function is unsafe because the returned `&mut Vec` allows writing
1775 /// bytes which are not valid UTF-8. If this constraint is violated, using
1776 /// the original `String` after dropping the `&mut Vec` may violate memory
1777 /// safety, as the rest of the standard library assumes that `String`s are
1778 /// valid UTF-8.
1779 ///
1780 /// # Examples
1781 ///
1782 /// ```
1783 /// let mut s = String::from("hello");
1784 ///
1785 /// unsafe {
1786 /// let vec = s.as_mut_vec();
1787 /// assert_eq!(&[104, 101, 108, 108, 111][..], &vec[..]);
1788 ///
1789 /// vec.reverse();
1790 /// }
1791 /// assert_eq!(s, "olleh");
1792 /// ```
1793 #[inline]
1794 #[stable(feature = "rust1", since = "1.0.0")]
1795 #[rustc_const_unstable(feature = "const_vec_string_slice", issue = "129041")]
1796 pub const unsafe fn as_mut_vec(&mut self) -> &mut Vec<u8> {
1797 &mut self.vec
1798 }
1799
1800 /// Returns the length of this `String`, in bytes, not [`char`]s or
1801 /// graphemes. In other words, it might not be what a human considers the
1802 /// length of the string.
1803 ///
1804 /// # Examples
1805 ///
1806 /// ```
1807 /// let a = String::from("foo");
1808 /// assert_eq!(a.len(), 3);
1809 ///
1810 /// let fancy_f = String::from("ฦoo");
1811 /// assert_eq!(fancy_f.len(), 4);
1812 /// assert_eq!(fancy_f.chars().count(), 3);
1813 /// ```
1814 #[inline]
1815 #[must_use]
1816 #[stable(feature = "rust1", since = "1.0.0")]
1817 #[rustc_const_unstable(feature = "const_vec_string_slice", issue = "129041")]
1818 #[rustc_confusables("length", "size")]
1819 pub const fn len(&self) -> usize {
1820 self.vec.len()
1821 }
1822
1823 /// Returns `true` if this `String` has a length of zero, and `false` otherwise.
1824 ///
1825 /// # Examples
1826 ///
1827 /// ```
1828 /// let mut v = String::new();
1829 /// assert!(v.is_empty());
1830 ///
1831 /// v.push('a');
1832 /// assert!(!v.is_empty());
1833 /// ```
1834 #[inline]
1835 #[must_use]
1836 #[stable(feature = "rust1", since = "1.0.0")]
1837 #[rustc_const_unstable(feature = "const_vec_string_slice", issue = "129041")]
1838 pub const fn is_empty(&self) -> bool {
1839 self.len() == 0
1840 }
1841
1842 /// Splits the string into two at the given byte index.
1843 ///
1844 /// Returns a newly allocated `String`. `self` contains bytes `[0, at)`, and
1845 /// the returned `String` contains bytes `[at, len)`. `at` must be on the
1846 /// boundary of a UTF-8 code point.
1847 ///
1848 /// Note that the capacity of `self` does not change.
1849 ///
1850 /// # Panics
1851 ///
1852 /// Panics if `at` is not on a `UTF-8` code point boundary, or if it is beyond the last
1853 /// code point of the string.
1854 ///
1855 /// # Examples
1856 ///
1857 /// ```
1858 /// # fn main() {
1859 /// let mut hello = String::from("Hello, World!");
1860 /// let world = hello.split_off(7);
1861 /// assert_eq!(hello, "Hello, ");
1862 /// assert_eq!(world, "World!");
1863 /// # }
1864 /// ```
1865 #[cfg(not(no_global_oom_handling))]
1866 #[inline]
1867 #[stable(feature = "string_split_off", since = "1.16.0")]
1868 #[must_use = "use `.truncate()` if you don't need the other half"]
1869 pub fn split_off(&mut self, at: usize) -> String {
1870 assert!(self.is_char_boundary(at));
1871 let other = self.vec.split_off(at);
1872 unsafe { String::from_utf8_unchecked(other) }
1873 }
1874
1875 /// Truncates this `String`, removing all contents.
1876 ///
1877 /// While this means the `String` will have a length of zero, it does not
1878 /// touch its capacity.
1879 ///
1880 /// # Examples
1881 ///
1882 /// ```
1883 /// let mut s = String::from("foo");
1884 ///
1885 /// s.clear();
1886 ///
1887 /// assert!(s.is_empty());
1888 /// assert_eq!(0, s.len());
1889 /// assert_eq!(3, s.capacity());
1890 /// ```
1891 #[inline]
1892 #[stable(feature = "rust1", since = "1.0.0")]
1893 pub fn clear(&mut self) {
1894 self.vec.clear()
1895 }
1896
1897 /// Removes the specified range from the string in bulk, returning all
1898 /// removed characters as an iterator.
1899 ///
1900 /// The returned iterator keeps a mutable borrow on the string to optimize
1901 /// its implementation.
1902 ///
1903 /// # Panics
1904 ///
1905 /// Panics if the starting point or end point do not lie on a [`char`]
1906 /// boundary, or if they're out of bounds.
1907 ///
1908 /// # Leaking
1909 ///
1910 /// If the returned iterator goes out of scope without being dropped (due to
1911 /// [`core::mem::forget`], for example), the string may still contain a copy
1912 /// of any drained characters, or may have lost characters arbitrarily,
1913 /// including characters outside the range.
1914 ///
1915 /// # Examples
1916 ///
1917 /// ```
1918 /// let mut s = String::from("ฮฑ is alpha, ฮฒ is beta");
1919 /// let beta_offset = s.find('ฮฒ').unwrap_or(s.len());
1920 ///
1921 /// // Remove the range up until the ฮฒ from the string
1922 /// let t: String = s.drain(..beta_offset).collect();
1923 /// assert_eq!(t, "ฮฑ is alpha, ");
1924 /// assert_eq!(s, "ฮฒ is beta");
1925 ///
1926 /// // A full range clears the string, like `clear()` does
1927 /// s.drain(..);
1928 /// assert_eq!(s, "");
1929 /// ```
1930 #[stable(feature = "drain", since = "1.6.0")]
1931 pub fn drain<R>(&mut self, range: R) -> Drain<'_>
1932 where
1933 R: RangeBounds<usize>,
1934 {
1935 // Memory safety
1936 //
1937 // The String version of Drain does not have the memory safety issues
1938 // of the vector version. The data is just plain bytes.
1939 // Because the range removal happens in Drop, if the Drain iterator is leaked,
1940 // the removal will not happen.
1941 let Range { start, end } = slice::range(range, ..self.len());
1942 assert!(self.is_char_boundary(start));
1943 assert!(self.is_char_boundary(end));
1944
1945 // Take out two simultaneous borrows. The &mut String won't be accessed
1946 // until iteration is over, in Drop.
1947 let self_ptr = self as *mut _;
1948 // SAFETY: `slice::range` and `is_char_boundary` do the appropriate bounds checks.
1949 let chars_iter = unsafe { self.get_unchecked(start..end) }.chars();
1950
1951 Drain { start, end, iter: chars_iter, string: self_ptr }
1952 }
1953
1954 /// Converts a `String` into an iterator over the [`char`]s of the string.
1955 ///
1956 /// As a string consists of valid UTF-8, we can iterate through a string
1957 /// by [`char`]. This method returns such an iterator.
1958 ///
1959 /// It's important to remember that [`char`] represents a Unicode Scalar
1960 /// Value, and might not match your idea of what a 'character' is. Iteration
1961 /// over grapheme clusters may be what you actually want. That functionality
1962 /// is not provided by Rust's standard library, check crates.io instead.
1963 ///
1964 /// # Examples
1965 ///
1966 /// Basic usage:
1967 ///
1968 /// ```
1969 /// #![feature(string_into_chars)]
1970 ///
1971 /// let word = String::from("goodbye");
1972 ///
1973 /// let mut chars = word.into_chars();
1974 ///
1975 /// assert_eq!(Some('g'), chars.next());
1976 /// assert_eq!(Some('o'), chars.next());
1977 /// assert_eq!(Some('o'), chars.next());
1978 /// assert_eq!(Some('d'), chars.next());
1979 /// assert_eq!(Some('b'), chars.next());
1980 /// assert_eq!(Some('y'), chars.next());
1981 /// assert_eq!(Some('e'), chars.next());
1982 ///
1983 /// assert_eq!(None, chars.next());
1984 /// ```
1985 ///
1986 /// Remember, [`char`]s might not match your intuition about characters:
1987 ///
1988 /// ```
1989 /// #![feature(string_into_chars)]
1990 ///
1991 /// let y = String::from("yฬ");
1992 ///
1993 /// let mut chars = y.into_chars();
1994 ///
1995 /// assert_eq!(Some('y'), chars.next()); // not 'yฬ'
1996 /// assert_eq!(Some('\u{0306}'), chars.next());
1997 ///
1998 /// assert_eq!(None, chars.next());
1999 /// ```
2000 ///
2001 /// [`char`]: prim@char
2002 #[inline]
2003 #[must_use = "`self` will be dropped if the result is not used"]
2004 #[unstable(feature = "string_into_chars", issue = "133125")]
2005 pub fn into_chars(self) -> IntoChars {
2006 IntoChars { bytes: self.into_bytes().into_iter() }
2007 }
2008
2009 /// Removes the specified range in the string,
2010 /// and replaces it with the given string.
2011 /// The given string doesn't need to be the same length as the range.
2012 ///
2013 /// # Panics
2014 ///
2015 /// Panics if the starting point or end point do not lie on a [`char`]
2016 /// boundary, or if they're out of bounds.
2017 ///
2018 /// # Examples
2019 ///
2020 /// ```
2021 /// let mut s = String::from("ฮฑ is alpha, ฮฒ is beta");
2022 /// let beta_offset = s.find('ฮฒ').unwrap_or(s.len());
2023 ///
2024 /// // Replace the range up until the ฮฒ from the string
2025 /// s.replace_range(..beta_offset, "ฮ is capital alpha; ");
2026 /// assert_eq!(s, "ฮ is capital alpha; ฮฒ is beta");
2027 /// ```
2028 #[cfg(not(no_global_oom_handling))]
2029 #[stable(feature = "splice", since = "1.27.0")]
2030 pub fn replace_range<R>(&mut self, range: R, replace_with: &str)
2031 where
2032 R: RangeBounds<usize>,
2033 {
2034 // Memory safety
2035 //
2036 // Replace_range does not have the memory safety issues of a vector Splice.
2037 // of the vector version. The data is just plain bytes.
2038
2039 // WARNING: Inlining this variable would be unsound (#81138)
2040 let start = range.start_bound();
2041 match start {
2042 Included(&n) => assert!(self.is_char_boundary(n)),
2043 Excluded(&n) => assert!(self.is_char_boundary(n + 1)),
2044 Unbounded => {}
2045 };
2046 // WARNING: Inlining this variable would be unsound (#81138)
2047 let end = range.end_bound();
2048 match end {
2049 Included(&n) => assert!(self.is_char_boundary(n + 1)),
2050 Excluded(&n) => assert!(self.is_char_boundary(n)),
2051 Unbounded => {}
2052 };
2053
2054 // Using `range` again would be unsound (#81138)
2055 // We assume the bounds reported by `range` remain the same, but
2056 // an adversarial implementation could change between calls
2057 unsafe { self.as_mut_vec() }.splice((start, end), replace_with.bytes());
2058 }
2059
2060 /// Converts this `String` into a <code>[Box]<[str]></code>.
2061 ///
2062 /// Before doing the conversion, this method discards excess capacity like [`shrink_to_fit`].
2063 /// Note that this call may reallocate and copy the bytes of the string.
2064 ///
2065 /// [`shrink_to_fit`]: String::shrink_to_fit
2066 /// [str]: prim@str "str"
2067 ///
2068 /// # Examples
2069 ///
2070 /// ```
2071 /// let s = String::from("hello");
2072 ///
2073 /// let b = s.into_boxed_str();
2074 /// ```
2075 #[cfg(not(no_global_oom_handling))]
2076 #[stable(feature = "box_str", since = "1.4.0")]
2077 #[must_use = "`self` will be dropped if the result is not used"]
2078 #[inline]
2079 pub fn into_boxed_str(self) -> Box<str> {
2080 let slice = self.vec.into_boxed_slice();
2081 unsafe { from_boxed_utf8_unchecked(slice) }
2082 }
2083
2084 /// Consumes and leaks the `String`, returning a mutable reference to the contents,
2085 /// `&'a mut str`.
2086 ///
2087 /// The caller has free choice over the returned lifetime, including `'static`. Indeed,
2088 /// this function is ideally used for data that lives for the remainder of the program's life,
2089 /// as dropping the returned reference will cause a memory leak.
2090 ///
2091 /// It does not reallocate or shrink the `String`, so the leaked allocation may include unused
2092 /// capacity that is not part of the returned slice. If you want to discard excess capacity,
2093 /// call [`into_boxed_str`], and then [`Box::leak`] instead. However, keep in mind that
2094 /// trimming the capacity may result in a reallocation and copy.
2095 ///
2096 /// [`into_boxed_str`]: Self::into_boxed_str
2097 ///
2098 /// # Examples
2099 ///
2100 /// ```
2101 /// let x = String::from("bucket");
2102 /// let static_ref: &'static mut str = x.leak();
2103 /// assert_eq!(static_ref, "bucket");
2104 /// # // FIXME(https://github.com/rust-lang/miri/issues/3670):
2105 /// # // use -Zmiri-disable-leak-check instead of unleaking in tests meant to leak.
2106 /// # drop(unsafe { Box::from_raw(static_ref) });
2107 /// ```
2108 #[stable(feature = "string_leak", since = "1.72.0")]
2109 #[inline]
2110 pub fn leak<'a>(self) -> &'a mut str {
2111 let slice = self.vec.leak();
2112 unsafe { from_utf8_unchecked_mut(slice) }
2113 }
2114}
2115
2116impl FromUtf8Error {
2117 /// Returns a slice of [`u8`]s bytes that were attempted to convert to a `String`.
2118 ///
2119 /// # Examples
2120 ///
2121 /// ```
2122 /// // some invalid bytes, in a vector
2123 /// let bytes = vec![0, 159];
2124 ///
2125 /// let value = String::from_utf8(bytes);
2126 ///
2127 /// assert_eq!(&[0, 159], value.unwrap_err().as_bytes());
2128 /// ```
2129 #[must_use]
2130 #[stable(feature = "from_utf8_error_as_bytes", since = "1.26.0")]
2131 pub fn as_bytes(&self) -> &[u8] {
2132 &self.bytes[..]
2133 }
2134
2135 /// Converts the bytes into a `String` lossily, substituting invalid UTF-8
2136 /// sequences with replacement characters.
2137 ///
2138 /// See [`String::from_utf8_lossy`] for more details on replacement of
2139 /// invalid sequences, and [`String::from_utf8_lossy_owned`] for the
2140 /// `String` function which corresponds to this function.
2141 ///
2142 /// # Examples
2143 ///
2144 /// ```
2145 /// #![feature(string_from_utf8_lossy_owned)]
2146 /// // some invalid bytes
2147 /// let input: Vec<u8> = b"Hello \xF0\x90\x80World".into();
2148 /// let output = String::from_utf8(input).unwrap_or_else(|e| e.into_utf8_lossy());
2149 ///
2150 /// assert_eq!(String::from("Hello ๏ฟฝWorld"), output);
2151 /// ```
2152 #[must_use]
2153 #[cfg(not(no_global_oom_handling))]
2154 #[unstable(feature = "string_from_utf8_lossy_owned", issue = "129436")]
2155 pub fn into_utf8_lossy(self) -> String {
2156 const REPLACEMENT: &str = "\u{FFFD}";
2157
2158 let mut res = {
2159 let mut v = Vec::with_capacity(self.bytes.len());
2160
2161 // `Utf8Error::valid_up_to` returns the maximum index of validated
2162 // UTF-8 bytes. Copy the valid bytes into the output buffer.
2163 v.extend_from_slice(&self.bytes[..self.error.valid_up_to()]);
2164
2165 // SAFETY: This is safe because the only bytes present in the buffer
2166 // were validated as UTF-8 by the call to `String::from_utf8` which
2167 // produced this `FromUtf8Error`.
2168 unsafe { String::from_utf8_unchecked(v) }
2169 };
2170
2171 let iter = self.bytes[self.error.valid_up_to()..].utf8_chunks();
2172
2173 for chunk in iter {
2174 res.push_str(chunk.valid());
2175 if !chunk.invalid().is_empty() {
2176 res.push_str(REPLACEMENT);
2177 }
2178 }
2179
2180 res
2181 }
2182
2183 /// Returns the bytes that were attempted to convert to a `String`.
2184 ///
2185 /// This method is carefully constructed to avoid allocation. It will
2186 /// consume the error, moving out the bytes, so that a copy of the bytes
2187 /// does not need to be made.
2188 ///
2189 /// # Examples
2190 ///
2191 /// ```
2192 /// // some invalid bytes, in a vector
2193 /// let bytes = vec![0, 159];
2194 ///
2195 /// let value = String::from_utf8(bytes);
2196 ///
2197 /// assert_eq!(vec![0, 159], value.unwrap_err().into_bytes());
2198 /// ```
2199 #[must_use = "`self` will be dropped if the result is not used"]
2200 #[stable(feature = "rust1", since = "1.0.0")]
2201 pub fn into_bytes(self) -> Vec<u8> {
2202 self.bytes
2203 }
2204
2205 /// Fetch a `Utf8Error` to get more details about the conversion failure.
2206 ///
2207 /// The [`Utf8Error`] type provided by [`std::str`] represents an error that may
2208 /// occur when converting a slice of [`u8`]s to a [`&str`]. In this sense, it's
2209 /// an analogue to `FromUtf8Error`. See its documentation for more details
2210 /// on using it.
2211 ///
2212 /// [`std::str`]: core::str "std::str"
2213 /// [`&str`]: prim@str "&str"
2214 ///
2215 /// # Examples
2216 ///
2217 /// ```
2218 /// // some invalid bytes, in a vector
2219 /// let bytes = vec![0, 159];
2220 ///
2221 /// let error = String::from_utf8(bytes).unwrap_err().utf8_error();
2222 ///
2223 /// // the first byte is invalid here
2224 /// assert_eq!(1, error.valid_up_to());
2225 /// ```
2226 #[must_use]
2227 #[stable(feature = "rust1", since = "1.0.0")]
2228 pub fn utf8_error(&self) -> Utf8Error {
2229 self.error
2230 }
2231}
2232
2233#[stable(feature = "rust1", since = "1.0.0")]
2234impl fmt::Display for FromUtf8Error {
2235 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2236 fmt::Display::fmt(&self.error, f)
2237 }
2238}
2239
2240#[stable(feature = "rust1", since = "1.0.0")]
2241impl fmt::Display for FromUtf16Error {
2242 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2243 fmt::Display::fmt("invalid utf-16: lone surrogate found", f)
2244 }
2245}
2246
2247#[stable(feature = "rust1", since = "1.0.0")]
2248impl Error for FromUtf8Error {
2249 #[allow(deprecated)]
2250 fn description(&self) -> &str {
2251 "invalid utf-8"
2252 }
2253}
2254
2255#[stable(feature = "rust1", since = "1.0.0")]
2256impl Error for FromUtf16Error {
2257 #[allow(deprecated)]
2258 fn description(&self) -> &str {
2259 "invalid utf-16"
2260 }
2261}
2262
2263#[cfg(not(no_global_oom_handling))]
2264#[stable(feature = "rust1", since = "1.0.0")]
2265impl Clone for String {
2266 fn clone(&self) -> Self {
2267 String { vec: self.vec.clone() }
2268 }
2269
2270 /// Clones the contents of `source` into `self`.
2271 ///
2272 /// This method is preferred over simply assigning `source.clone()` to `self`,
2273 /// as it avoids reallocation if possible.
2274 fn clone_from(&mut self, source: &Self) {
2275 self.vec.clone_from(&source.vec);
2276 }
2277}
2278
2279#[cfg(not(no_global_oom_handling))]
2280#[stable(feature = "rust1", since = "1.0.0")]
2281impl FromIterator<char> for String {
2282 fn from_iter<I: IntoIterator<Item = char>>(iter: I) -> String {
2283 let mut buf = String::new();
2284 buf.extend(iter);
2285 buf
2286 }
2287}
2288
2289#[cfg(not(no_global_oom_handling))]
2290#[stable(feature = "string_from_iter_by_ref", since = "1.17.0")]
2291impl<'a> FromIterator<&'a char> for String {
2292 fn from_iter<I: IntoIterator<Item = &'a char>>(iter: I) -> String {
2293 let mut buf = String::new();
2294 buf.extend(iter);
2295 buf
2296 }
2297}
2298
2299#[cfg(not(no_global_oom_handling))]
2300#[stable(feature = "rust1", since = "1.0.0")]
2301impl<'a> FromIterator<&'a str> for String {
2302 fn from_iter<I: IntoIterator<Item = &'a str>>(iter: I) -> String {
2303 let mut buf = String::new();
2304 buf.extend(iter);
2305 buf
2306 }
2307}
2308
2309#[cfg(not(no_global_oom_handling))]
2310#[stable(feature = "extend_string", since = "1.4.0")]
2311impl FromIterator<String> for String {
2312 fn from_iter<I: IntoIterator<Item = String>>(iter: I) -> String {
2313 let mut iterator = iter.into_iter();
2314
2315 // Because we're iterating over `String`s, we can avoid at least
2316 // one allocation by getting the first string from the iterator
2317 // and appending to it all the subsequent strings.
2318 match iterator.next() {
2319 None => String::new(),
2320 Some(mut buf) => {
2321 buf.extend(iterator);
2322 buf
2323 }
2324 }
2325 }
2326}
2327
2328#[cfg(not(no_global_oom_handling))]
2329#[stable(feature = "box_str2", since = "1.45.0")]
2330impl<A: Allocator> FromIterator<Box<str, A>> for String {
2331 fn from_iter<I: IntoIterator<Item = Box<str, A>>>(iter: I) -> String {
2332 let mut buf = String::new();
2333 buf.extend(iter);
2334 buf
2335 }
2336}
2337
2338#[cfg(not(no_global_oom_handling))]
2339#[stable(feature = "herd_cows", since = "1.19.0")]
2340impl<'a> FromIterator<Cow<'a, str>> for String {
2341 fn from_iter<I: IntoIterator<Item = Cow<'a, str>>>(iter: I) -> String {
2342 let mut iterator = iter.into_iter();
2343
2344 // Because we're iterating over CoWs, we can (potentially) avoid at least
2345 // one allocation by getting the first item and appending to it all the
2346 // subsequent items.
2347 match iterator.next() {
2348 None => String::new(),
2349 Some(cow) => {
2350 let mut buf = cow.into_owned();
2351 buf.extend(iterator);
2352 buf
2353 }
2354 }
2355 }
2356}
2357
2358#[cfg(not(no_global_oom_handling))]
2359#[stable(feature = "rust1", since = "1.0.0")]
2360impl Extend<char> for String {
2361 fn extend<I: IntoIterator<Item = char>>(&mut self, iter: I) {
2362 let iterator = iter.into_iter();
2363 let (lower_bound, _) = iterator.size_hint();
2364 self.reserve(lower_bound);
2365 iterator.for_each(move |c| self.push(c));
2366 }
2367
2368 #[inline]
2369 fn extend_one(&mut self, c: char) {
2370 self.push(c);
2371 }
2372
2373 #[inline]
2374 fn extend_reserve(&mut self, additional: usize) {
2375 self.reserve(additional);
2376 }
2377}
2378
2379#[cfg(not(no_global_oom_handling))]
2380#[stable(feature = "extend_ref", since = "1.2.0")]
2381impl<'a> Extend<&'a char> for String {
2382 fn extend<I: IntoIterator<Item = &'a char>>(&mut self, iter: I) {
2383 self.extend(iter.into_iter().cloned());
2384 }
2385
2386 #[inline]
2387 fn extend_one(&mut self, &c: &'a char) {
2388 self.push(c);
2389 }
2390
2391 #[inline]
2392 fn extend_reserve(&mut self, additional: usize) {
2393 self.reserve(additional);
2394 }
2395}
2396
2397#[cfg(not(no_global_oom_handling))]
2398#[stable(feature = "rust1", since = "1.0.0")]
2399impl<'a> Extend<&'a str> for String {
2400 fn extend<I: IntoIterator<Item = &'a str>>(&mut self, iter: I) {
2401 iter.into_iter().for_each(move |s| self.push_str(s));
2402 }
2403
2404 #[inline]
2405 fn extend_one(&mut self, s: &'a str) {
2406 self.push_str(s);
2407 }
2408}
2409
2410#[cfg(not(no_global_oom_handling))]
2411#[stable(feature = "box_str2", since = "1.45.0")]
2412impl<A: Allocator> Extend<Box<str, A>> for String {
2413 fn extend<I: IntoIterator<Item = Box<str, A>>>(&mut self, iter: I) {
2414 iter.into_iter().for_each(move |s| self.push_str(&s));
2415 }
2416}
2417
2418#[cfg(not(no_global_oom_handling))]
2419#[stable(feature = "extend_string", since = "1.4.0")]
2420impl Extend<String> for String {
2421 fn extend<I: IntoIterator<Item = String>>(&mut self, iter: I) {
2422 iter.into_iter().for_each(move |s| self.push_str(&s));
2423 }
2424
2425 #[inline]
2426 fn extend_one(&mut self, s: String) {
2427 self.push_str(&s);
2428 }
2429}
2430
2431#[cfg(not(no_global_oom_handling))]
2432#[stable(feature = "herd_cows", since = "1.19.0")]
2433impl<'a> Extend<Cow<'a, str>> for String {
2434 fn extend<I: IntoIterator<Item = Cow<'a, str>>>(&mut self, iter: I) {
2435 iter.into_iter().for_each(move |s| self.push_str(&s));
2436 }
2437
2438 #[inline]
2439 fn extend_one(&mut self, s: Cow<'a, str>) {
2440 self.push_str(&s);
2441 }
2442}
2443
2444#[cfg(not(no_global_oom_handling))]
2445#[unstable(feature = "ascii_char", issue = "110998")]
2446impl Extend<core::ascii::Char> for String {
2447 fn extend<I: IntoIterator<Item = core::ascii::Char>>(&mut self, iter: I) {
2448 self.vec.extend(iter.into_iter().map(|c| c.to_u8()));
2449 }
2450
2451 #[inline]
2452 fn extend_one(&mut self, c: core::ascii::Char) {
2453 self.vec.push(c.to_u8());
2454 }
2455}
2456
2457#[cfg(not(no_global_oom_handling))]
2458#[unstable(feature = "ascii_char", issue = "110998")]
2459impl<'a> Extend<&'a core::ascii::Char> for String {
2460 fn extend<I: IntoIterator<Item = &'a core::ascii::Char>>(&mut self, iter: I) {
2461 self.extend(iter.into_iter().cloned());
2462 }
2463
2464 #[inline]
2465 fn extend_one(&mut self, c: &'a core::ascii::Char) {
2466 self.vec.push(c.to_u8());
2467 }
2468}
2469
2470/// A convenience impl that delegates to the impl for `&str`.
2471///
2472/// # Examples
2473///
2474/// ```
2475/// assert_eq!(String::from("Hello world").find("world"), Some(6));
2476/// ```
2477#[unstable(
2478 feature = "pattern",
2479 reason = "API not fully fleshed out and ready to be stabilized",
2480 issue = "27721"
2481)]
2482impl<'b> Pattern for &'b String {
2483 type Searcher<'a> = <&'b str as Pattern>::Searcher<'a>;
2484
2485 fn into_searcher(self, haystack: &str) -> <&'b str as Pattern>::Searcher<'_> {
2486 self[..].into_searcher(haystack)
2487 }
2488
2489 #[inline]
2490 fn is_contained_in(self, haystack: &str) -> bool {
2491 self[..].is_contained_in(haystack)
2492 }
2493
2494 #[inline]
2495 fn is_prefix_of(self, haystack: &str) -> bool {
2496 self[..].is_prefix_of(haystack)
2497 }
2498
2499 #[inline]
2500 fn strip_prefix_of(self, haystack: &str) -> Option<&str> {
2501 self[..].strip_prefix_of(haystack)
2502 }
2503
2504 #[inline]
2505 fn is_suffix_of<'a>(self, haystack: &'a str) -> bool
2506 where
2507 Self::Searcher<'a>: core::str::pattern::ReverseSearcher<'a>,
2508 {
2509 self[..].is_suffix_of(haystack)
2510 }
2511
2512 #[inline]
2513 fn strip_suffix_of<'a>(self, haystack: &'a str) -> Option<&'a str>
2514 where
2515 Self::Searcher<'a>: core::str::pattern::ReverseSearcher<'a>,
2516 {
2517 self[..].strip_suffix_of(haystack)
2518 }
2519
2520 #[inline]
2521 fn as_utf8_pattern(&self) -> Option<Utf8Pattern<'_>> {
2522 Some(Utf8Pattern::StringPattern(self.as_bytes()))
2523 }
2524}
2525
2526macro_rules! impl_eq {
2527 ($lhs:ty, $rhs: ty) => {
2528 #[stable(feature = "rust1", since = "1.0.0")]
2529 #[allow(unused_lifetimes)]
2530 impl<'a, 'b> PartialEq<$rhs> for $lhs {
2531 #[inline]
2532 fn eq(&self, other: &$rhs) -> bool {
2533 PartialEq::eq(&self[..], &other[..])
2534 }
2535 #[inline]
2536 fn ne(&self, other: &$rhs) -> bool {
2537 PartialEq::ne(&self[..], &other[..])
2538 }
2539 }
2540
2541 #[stable(feature = "rust1", since = "1.0.0")]
2542 #[allow(unused_lifetimes)]
2543 impl<'a, 'b> PartialEq<$lhs> for $rhs {
2544 #[inline]
2545 fn eq(&self, other: &$lhs) -> bool {
2546 PartialEq::eq(&self[..], &other[..])
2547 }
2548 #[inline]
2549 fn ne(&self, other: &$lhs) -> bool {
2550 PartialEq::ne(&self[..], &other[..])
2551 }
2552 }
2553 };
2554}
2555
2556impl_eq! { String, str }
2557impl_eq! { String, &'a str }
2558#[cfg(not(no_global_oom_handling))]
2559impl_eq! { Cow<'a, str>, str }
2560#[cfg(not(no_global_oom_handling))]
2561impl_eq! { Cow<'a, str>, &'b str }
2562#[cfg(not(no_global_oom_handling))]
2563impl_eq! { Cow<'a, str>, String }
2564
2565#[stable(feature = "rust1", since = "1.0.0")]
2566impl Default for String {
2567 /// Creates an empty `String`.
2568 #[inline]
2569 fn default() -> String {
2570 String::new()
2571 }
2572}
2573
2574#[stable(feature = "rust1", since = "1.0.0")]
2575impl fmt::Display for String {
2576 #[inline]
2577 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2578 fmt::Display::fmt(&**self, f)
2579 }
2580}
2581
2582#[stable(feature = "rust1", since = "1.0.0")]
2583impl fmt::Debug for String {
2584 #[inline]
2585 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2586 fmt::Debug::fmt(&**self, f)
2587 }
2588}
2589
2590#[stable(feature = "rust1", since = "1.0.0")]
2591impl hash::Hash for String {
2592 #[inline]
2593 fn hash<H: hash::Hasher>(&self, hasher: &mut H) {
2594 (**self).hash(hasher)
2595 }
2596}
2597
2598/// Implements the `+` operator for concatenating two strings.
2599///
2600/// This consumes the `String` on the left-hand side and re-uses its buffer (growing it if
2601/// necessary). This is done to avoid allocating a new `String` and copying the entire contents on
2602/// every operation, which would lead to *O*(*n*^2) running time when building an *n*-byte string by
2603/// repeated concatenation.
2604///
2605/// The string on the right-hand side is only borrowed; its contents are copied into the returned
2606/// `String`.
2607///
2608/// # Examples
2609///
2610/// Concatenating two `String`s takes the first by value and borrows the second:
2611///
2612/// ```
2613/// let a = String::from("hello");
2614/// let b = String::from(" world");
2615/// let c = a + &b;
2616/// // `a` is moved and can no longer be used here.
2617/// ```
2618///
2619/// If you want to keep using the first `String`, you can clone it and append to the clone instead:
2620///
2621/// ```
2622/// let a = String::from("hello");
2623/// let b = String::from(" world");
2624/// let c = a.clone() + &b;
2625/// // `a` is still valid here.
2626/// ```
2627///
2628/// Concatenating `&str` slices can be done by converting the first to a `String`:
2629///
2630/// ```
2631/// let a = "hello";
2632/// let b = " world";
2633/// let c = a.to_string() + b;
2634/// ```
2635#[cfg(not(no_global_oom_handling))]
2636#[stable(feature = "rust1", since = "1.0.0")]
2637impl Add<&str> for String {
2638 type Output = String;
2639
2640 #[inline]
2641 fn add(mut self, other: &str) -> String {
2642 self.push_str(other);
2643 self
2644 }
2645}
2646
2647/// Implements the `+=` operator for appending to a `String`.
2648///
2649/// This has the same behavior as the [`push_str`][String::push_str] method.
2650#[cfg(not(no_global_oom_handling))]
2651#[stable(feature = "stringaddassign", since = "1.12.0")]
2652impl AddAssign<&str> for String {
2653 #[inline]
2654 fn add_assign(&mut self, other: &str) {
2655 self.push_str(other);
2656 }
2657}
2658
2659#[stable(feature = "rust1", since = "1.0.0")]
2660impl<I> ops::Index<I> for String
2661where
2662 I: slice::SliceIndex<str>,
2663{
2664 type Output = I::Output;
2665
2666 #[inline]
2667 fn index(&self, index: I) -> &I::Output {
2668 index.index(self.as_str())
2669 }
2670}
2671
2672#[stable(feature = "rust1", since = "1.0.0")]
2673impl<I> ops::IndexMut<I> for String
2674where
2675 I: slice::SliceIndex<str>,
2676{
2677 #[inline]
2678 fn index_mut(&mut self, index: I) -> &mut I::Output {
2679 index.index_mut(self.as_mut_str())
2680 }
2681}
2682
2683#[stable(feature = "rust1", since = "1.0.0")]
2684impl ops::Deref for String {
2685 type Target = str;
2686
2687 #[inline]
2688 fn deref(&self) -> &str {
2689 self.as_str()
2690 }
2691}
2692
2693#[unstable(feature = "deref_pure_trait", issue = "87121")]
2694unsafe impl ops::DerefPure for String {}
2695
2696#[stable(feature = "derefmut_for_string", since = "1.3.0")]
2697impl ops::DerefMut for String {
2698 #[inline]
2699 fn deref_mut(&mut self) -> &mut str {
2700 self.as_mut_str()
2701 }
2702}
2703
2704/// A type alias for [`Infallible`].
2705///
2706/// This alias exists for backwards compatibility, and may be eventually deprecated.
2707///
2708/// [`Infallible`]: core::convert::Infallible "convert::Infallible"
2709#[stable(feature = "str_parse_error", since = "1.5.0")]
2710pub type ParseError = core::convert::Infallible;
2711
2712#[cfg(not(no_global_oom_handling))]
2713#[stable(feature = "rust1", since = "1.0.0")]
2714impl FromStr for String {
2715 type Err = core::convert::Infallible;
2716 #[inline]
2717 fn from_str(s: &str) -> Result<String, Self::Err> {
2718 Ok(String::from(s))
2719 }
2720}
2721
2722/// A trait for converting a value to a `String`.
2723///
2724/// This trait is automatically implemented for any type which implements the
2725/// [`Display`] trait. As such, `ToString` shouldn't be implemented directly:
2726/// [`Display`] should be implemented instead, and you get the `ToString`
2727/// implementation for free.
2728///
2729/// [`Display`]: fmt::Display
2730#[cfg_attr(not(test), rustc_diagnostic_item = "ToString")]
2731#[stable(feature = "rust1", since = "1.0.0")]
2732pub trait ToString {
2733 /// Converts the given value to a `String`.
2734 ///
2735 /// # Examples
2736 ///
2737 /// ```
2738 /// let i = 5;
2739 /// let five = String::from("5");
2740 ///
2741 /// assert_eq!(five, i.to_string());
2742 /// ```
2743 #[rustc_conversion_suggestion]
2744 #[stable(feature = "rust1", since = "1.0.0")]
2745 #[cfg_attr(not(test), rustc_diagnostic_item = "to_string_method")]
2746 fn to_string(&self) -> String;
2747}
2748
2749/// # Panics
2750///
2751/// In this implementation, the `to_string` method panics
2752/// if the `Display` implementation returns an error.
2753/// This indicates an incorrect `Display` implementation
2754/// since `fmt::Write for String` never returns an error itself.
2755#[cfg(not(no_global_oom_handling))]
2756#[stable(feature = "rust1", since = "1.0.0")]
2757impl<T: fmt::Display + ?Sized> ToString for T {
2758 #[inline]
2759 fn to_string(&self) -> String {
2760 <Self as SpecToString>::spec_to_string(self)
2761 }
2762}
2763
2764#[cfg(not(no_global_oom_handling))]
2765trait SpecToString {
2766 fn spec_to_string(&self) -> String;
2767}
2768
2769#[cfg(not(no_global_oom_handling))]
2770impl<T: fmt::Display + ?Sized> SpecToString for T {
2771 // A common guideline is to not inline generic functions. However,
2772 // removing `#[inline]` from this method causes non-negligible regressions.
2773 // See <https://github.com/rust-lang/rust/pull/74852>, the last attempt
2774 // to try to remove it.
2775 #[inline]
2776 default fn spec_to_string(&self) -> String {
2777 let mut buf = String::new();
2778 let mut formatter =
2779 core::fmt::Formatter::new(&mut buf, core::fmt::FormattingOptions::new());
2780 // Bypass format_args!() to avoid write_str with zero-length strs
2781 fmt::Display::fmt(self, &mut formatter)
2782 .expect("a Display implementation returned an error unexpectedly");
2783 buf
2784 }
2785}
2786
2787#[cfg(not(no_global_oom_handling))]
2788impl SpecToString for core::ascii::Char {
2789 #[inline]
2790 fn spec_to_string(&self) -> String {
2791 self.as_str().to_owned()
2792 }
2793}
2794
2795#[cfg(not(no_global_oom_handling))]
2796impl SpecToString for char {
2797 #[inline]
2798 fn spec_to_string(&self) -> String {
2799 String::from(self.encode_utf8(&mut [0; char::MAX_LEN_UTF8]))
2800 }
2801}
2802
2803#[cfg(not(no_global_oom_handling))]
2804impl SpecToString for bool {
2805 #[inline]
2806 fn spec_to_string(&self) -> String {
2807 String::from(if *self { "true" } else { "false" })
2808 }
2809}
2810
2811#[cfg(not(no_global_oom_handling))]
2812impl SpecToString for u8 {
2813 #[inline]
2814 fn spec_to_string(&self) -> String {
2815 let mut buf = String::with_capacity(3);
2816 let mut n = *self;
2817 if n >= 10 {
2818 if n >= 100 {
2819 buf.push((b'0' + n / 100) as char);
2820 n %= 100;
2821 }
2822 buf.push((b'0' + n / 10) as char);
2823 n %= 10;
2824 }
2825 buf.push((b'0' + n) as char);
2826 buf
2827 }
2828}
2829
2830#[cfg(not(no_global_oom_handling))]
2831impl SpecToString for i8 {
2832 #[inline]
2833 fn spec_to_string(&self) -> String {
2834 let mut buf = String::with_capacity(4);
2835 if self.is_negative() {
2836 buf.push('-');
2837 }
2838 let mut n = self.unsigned_abs();
2839 if n >= 10 {
2840 if n >= 100 {
2841 buf.push('1');
2842 n -= 100;
2843 }
2844 buf.push((b'0' + n / 10) as char);
2845 n %= 10;
2846 }
2847 buf.push((b'0' + n) as char);
2848 buf
2849 }
2850}
2851
2852// Generic/generated code can sometimes have multiple, nested references
2853// for strings, including `&&&str`s that would never be written
2854// by hand. This macro generates twelve layers of nested `&`-impl
2855// for primitive strings.
2856#[cfg(not(no_global_oom_handling))]
2857macro_rules! to_string_str_wrap_in_ref {
2858 {x $($x:ident)*} => {
2859 &to_string_str_wrap_in_ref! { $($x)* }
2860 };
2861 {} => { str };
2862}
2863#[cfg(not(no_global_oom_handling))]
2864macro_rules! to_string_expr_wrap_in_deref {
2865 {$self:expr ; x $($x:ident)*} => {
2866 *(to_string_expr_wrap_in_deref! { $self ; $($x)* })
2867 };
2868 {$self:expr ;} => { $self };
2869}
2870#[cfg(not(no_global_oom_handling))]
2871macro_rules! to_string_str {
2872 {$($($x:ident)*),+} => {
2873 $(
2874 impl SpecToString for to_string_str_wrap_in_ref!($($x)*) {
2875 #[inline]
2876 fn spec_to_string(&self) -> String {
2877 String::from(to_string_expr_wrap_in_deref!(self ; $($x)*))
2878 }
2879 }
2880 )+
2881 };
2882}
2883
2884#[cfg(not(no_global_oom_handling))]
2885to_string_str! {
2886 x x x x x x x x x x x x,
2887 x x x x x x x x x x x,
2888 x x x x x x x x x x,
2889 x x x x x x x x x,
2890 x x x x x x x x,
2891 x x x x x x x,
2892 x x x x x x,
2893 x x x x x,
2894 x x x x,
2895 x x x,
2896 x x,
2897 x,
2898}
2899
2900#[cfg(not(no_global_oom_handling))]
2901impl SpecToString for Cow<'_, str> {
2902 #[inline]
2903 fn spec_to_string(&self) -> String {
2904 self[..].to_owned()
2905 }
2906}
2907
2908#[cfg(not(no_global_oom_handling))]
2909impl SpecToString for String {
2910 #[inline]
2911 fn spec_to_string(&self) -> String {
2912 self.to_owned()
2913 }
2914}
2915
2916#[cfg(not(no_global_oom_handling))]
2917impl SpecToString for fmt::Arguments<'_> {
2918 #[inline]
2919 fn spec_to_string(&self) -> String {
2920 crate::fmt::format(*self)
2921 }
2922}
2923
2924#[stable(feature = "rust1", since = "1.0.0")]
2925impl AsRef<str> for String {
2926 #[inline]
2927 fn as_ref(&self) -> &str {
2928 self
2929 }
2930}
2931
2932#[stable(feature = "string_as_mut", since = "1.43.0")]
2933impl AsMut<str> for String {
2934 #[inline]
2935 fn as_mut(&mut self) -> &mut str {
2936 self
2937 }
2938}
2939
2940#[stable(feature = "rust1", since = "1.0.0")]
2941impl AsRef<[u8]> for String {
2942 #[inline]
2943 fn as_ref(&self) -> &[u8] {
2944 self.as_bytes()
2945 }
2946}
2947
2948#[cfg(not(no_global_oom_handling))]
2949#[stable(feature = "rust1", since = "1.0.0")]
2950impl From<&str> for String {
2951 /// Converts a `&str` into a [`String`].
2952 ///
2953 /// The result is allocated on the heap.
2954 #[inline]
2955 fn from(s: &str) -> String {
2956 s.to_owned()
2957 }
2958}
2959
2960#[cfg(not(no_global_oom_handling))]
2961#[stable(feature = "from_mut_str_for_string", since = "1.44.0")]
2962impl From<&mut str> for String {
2963 /// Converts a `&mut str` into a [`String`].
2964 ///
2965 /// The result is allocated on the heap.
2966 #[inline]
2967 fn from(s: &mut str) -> String {
2968 s.to_owned()
2969 }
2970}
2971
2972#[cfg(not(no_global_oom_handling))]
2973#[stable(feature = "from_ref_string", since = "1.35.0")]
2974impl From<&String> for String {
2975 /// Converts a `&String` into a [`String`].
2976 ///
2977 /// This clones `s` and returns the clone.
2978 #[inline]
2979 fn from(s: &String) -> String {
2980 s.clone()
2981 }
2982}
2983
2984// note: test pulls in std, which causes errors here
2985#[cfg(not(test))]
2986#[stable(feature = "string_from_box", since = "1.18.0")]
2987impl From<Box<str>> for String {
2988 /// Converts the given boxed `str` slice to a [`String`].
2989 /// It is notable that the `str` slice is owned.
2990 ///
2991 /// # Examples
2992 ///
2993 /// ```
2994 /// let s1: String = String::from("hello world");
2995 /// let s2: Box<str> = s1.into_boxed_str();
2996 /// let s3: String = String::from(s2);
2997 ///
2998 /// assert_eq!("hello world", s3)
2999 /// ```
3000 fn from(s: Box<str>) -> String {
3001 s.into_string()
3002 }
3003}
3004
3005#[cfg(not(no_global_oom_handling))]
3006#[stable(feature = "box_from_str", since = "1.20.0")]
3007impl From<String> for Box<str> {
3008 /// Converts the given [`String`] to a boxed `str` slice that is owned.
3009 ///
3010 /// # Examples
3011 ///
3012 /// ```
3013 /// let s1: String = String::from("hello world");
3014 /// let s2: Box<str> = Box::from(s1);
3015 /// let s3: String = String::from(s2);
3016 ///
3017 /// assert_eq!("hello world", s3)
3018 /// ```
3019 fn from(s: String) -> Box<str> {
3020 s.into_boxed_str()
3021 }
3022}
3023
3024#[cfg(not(no_global_oom_handling))]
3025#[stable(feature = "string_from_cow_str", since = "1.14.0")]
3026impl<'a> From<Cow<'a, str>> for String {
3027 /// Converts a clone-on-write string to an owned
3028 /// instance of [`String`].
3029 ///
3030 /// This extracts the owned string,
3031 /// clones the string if it is not already owned.
3032 ///
3033 /// # Example
3034 ///
3035 /// ```
3036 /// # use std::borrow::Cow;
3037 /// // If the string is not owned...
3038 /// let cow: Cow<'_, str> = Cow::Borrowed("eggplant");
3039 /// // It will allocate on the heap and copy the string.
3040 /// let owned: String = String::from(cow);
3041 /// assert_eq!(&owned[..], "eggplant");
3042 /// ```
3043 fn from(s: Cow<'a, str>) -> String {
3044 s.into_owned()
3045 }
3046}
3047
3048#[cfg(not(no_global_oom_handling))]
3049#[stable(feature = "rust1", since = "1.0.0")]
3050impl<'a> From<&'a str> for Cow<'a, str> {
3051 /// Converts a string slice into a [`Borrowed`] variant.
3052 /// No heap allocation is performed, and the string
3053 /// is not copied.
3054 ///
3055 /// # Example
3056 ///
3057 /// ```
3058 /// # use std::borrow::Cow;
3059 /// assert_eq!(Cow::from("eggplant"), Cow::Borrowed("eggplant"));
3060 /// ```
3061 ///
3062 /// [`Borrowed`]: crate::borrow::Cow::Borrowed "borrow::Cow::Borrowed"
3063 #[inline]
3064 fn from(s: &'a str) -> Cow<'a, str> {
3065 Cow::Borrowed(s)
3066 }
3067}
3068
3069#[cfg(not(no_global_oom_handling))]
3070#[stable(feature = "rust1", since = "1.0.0")]
3071impl<'a> From<String> for Cow<'a, str> {
3072 /// Converts a [`String`] into an [`Owned`] variant.
3073 /// No heap allocation is performed, and the string
3074 /// is not copied.
3075 ///
3076 /// # Example
3077 ///
3078 /// ```
3079 /// # use std::borrow::Cow;
3080 /// let s = "eggplant".to_string();
3081 /// let s2 = "eggplant".to_string();
3082 /// assert_eq!(Cow::from(s), Cow::<'static, str>::Owned(s2));
3083 /// ```
3084 ///
3085 /// [`Owned`]: crate::borrow::Cow::Owned "borrow::Cow::Owned"
3086 #[inline]
3087 fn from(s: String) -> Cow<'a, str> {
3088 Cow::Owned(s)
3089 }
3090}
3091
3092#[cfg(not(no_global_oom_handling))]
3093#[stable(feature = "cow_from_string_ref", since = "1.28.0")]
3094impl<'a> From<&'a String> for Cow<'a, str> {
3095 /// Converts a [`String`] reference into a [`Borrowed`] variant.
3096 /// No heap allocation is performed, and the string
3097 /// is not copied.
3098 ///
3099 /// # Example
3100 ///
3101 /// ```
3102 /// # use std::borrow::Cow;
3103 /// let s = "eggplant".to_string();
3104 /// assert_eq!(Cow::from(&s), Cow::Borrowed("eggplant"));
3105 /// ```
3106 ///
3107 /// [`Borrowed`]: crate::borrow::Cow::Borrowed "borrow::Cow::Borrowed"
3108 #[inline]
3109 fn from(s: &'a String) -> Cow<'a, str> {
3110 Cow::Borrowed(s.as_str())
3111 }
3112}
3113
3114#[cfg(not(no_global_oom_handling))]
3115#[stable(feature = "cow_str_from_iter", since = "1.12.0")]
3116impl<'a> FromIterator<char> for Cow<'a, str> {
3117 fn from_iter<I: IntoIterator<Item = char>>(it: I) -> Cow<'a, str> {
3118 Cow::Owned(FromIterator::from_iter(it))
3119 }
3120}
3121
3122#[cfg(not(no_global_oom_handling))]
3123#[stable(feature = "cow_str_from_iter", since = "1.12.0")]
3124impl<'a, 'b> FromIterator<&'b str> for Cow<'a, str> {
3125 fn from_iter<I: IntoIterator<Item = &'b str>>(it: I) -> Cow<'a, str> {
3126 Cow::Owned(FromIterator::from_iter(it))
3127 }
3128}
3129
3130#[cfg(not(no_global_oom_handling))]
3131#[stable(feature = "cow_str_from_iter", since = "1.12.0")]
3132impl<'a> FromIterator<String> for Cow<'a, str> {
3133 fn from_iter<I: IntoIterator<Item = String>>(it: I) -> Cow<'a, str> {
3134 Cow::Owned(FromIterator::from_iter(it))
3135 }
3136}
3137
3138#[stable(feature = "from_string_for_vec_u8", since = "1.14.0")]
3139impl From<String> for Vec<u8> {
3140 /// Converts the given [`String`] to a vector [`Vec`] that holds values of type [`u8`].
3141 ///
3142 /// # Examples
3143 ///
3144 /// ```
3145 /// let s1 = String::from("hello world");
3146 /// let v1 = Vec::from(s1);
3147 ///
3148 /// for b in v1 {
3149 /// println!("{b}");
3150 /// }
3151 /// ```
3152 fn from(string: String) -> Vec<u8> {
3153 string.into_bytes()
3154 }
3155}
3156
3157#[stable(feature = "try_from_vec_u8_for_string", since = "CURRENT_RUSTC_VERSION")]
3158impl TryFrom<Vec<u8>> for String {
3159 type Error = FromUtf8Error;
3160 /// Converts the given [`Vec<u8>`] into a [`String`] if it contains valid UTF-8 data.
3161 ///
3162 /// # Examples
3163 ///
3164 /// ```
3165 /// let s1 = b"hello world".to_vec();
3166 /// let v1 = String::try_from(s1).unwrap();
3167 /// assert_eq!(v1, "hello world");
3168 ///
3169 /// ```
3170 fn try_from(bytes: Vec<u8>) -> Result<Self, Self::Error> {
3171 Self::from_utf8(bytes)
3172 }
3173}
3174
3175#[cfg(not(no_global_oom_handling))]
3176#[stable(feature = "rust1", since = "1.0.0")]
3177impl fmt::Write for String {
3178 #[inline]
3179 fn write_str(&mut self, s: &str) -> fmt::Result {
3180 self.push_str(s);
3181 Ok(())
3182 }
3183
3184 #[inline]
3185 fn write_char(&mut self, c: char) -> fmt::Result {
3186 self.push(c);
3187 Ok(())
3188 }
3189}
3190
3191/// An iterator over the [`char`]s of a string.
3192///
3193/// This struct is created by the [`into_chars`] method on [`String`].
3194/// See its documentation for more.
3195///
3196/// [`char`]: prim@char
3197/// [`into_chars`]: String::into_chars
3198#[cfg_attr(not(no_global_oom_handling), derive(Clone))]
3199#[must_use = "iterators are lazy and do nothing unless consumed"]
3200#[unstable(feature = "string_into_chars", issue = "133125")]
3201pub struct IntoChars {
3202 bytes: vec::IntoIter<u8>,
3203}
3204
3205#[unstable(feature = "string_into_chars", issue = "133125")]
3206impl fmt::Debug for IntoChars {
3207 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
3208 f.debug_tuple("IntoChars").field(&self.as_str()).finish()
3209 }
3210}
3211
3212impl IntoChars {
3213 /// Views the underlying data as a subslice of the original data.
3214 ///
3215 /// # Examples
3216 ///
3217 /// ```
3218 /// #![feature(string_into_chars)]
3219 ///
3220 /// let mut chars = String::from("abc").into_chars();
3221 ///
3222 /// assert_eq!(chars.as_str(), "abc");
3223 /// chars.next();
3224 /// assert_eq!(chars.as_str(), "bc");
3225 /// chars.next();
3226 /// chars.next();
3227 /// assert_eq!(chars.as_str(), "");
3228 /// ```
3229 #[unstable(feature = "string_into_chars", issue = "133125")]
3230 #[must_use]
3231 #[inline]
3232 pub fn as_str(&self) -> &str {
3233 // SAFETY: `bytes` is a valid UTF-8 string.
3234 unsafe { str::from_utf8_unchecked(self.bytes.as_slice()) }
3235 }
3236
3237 /// Consumes the `IntoChars`, returning the remaining string.
3238 ///
3239 /// # Examples
3240 ///
3241 /// ```
3242 /// #![feature(string_into_chars)]
3243 ///
3244 /// let chars = String::from("abc").into_chars();
3245 /// assert_eq!(chars.into_string(), "abc");
3246 ///
3247 /// let mut chars = String::from("def").into_chars();
3248 /// chars.next();
3249 /// assert_eq!(chars.into_string(), "ef");
3250 /// ```
3251 #[cfg(not(no_global_oom_handling))]
3252 #[unstable(feature = "string_into_chars", issue = "133125")]
3253 #[inline]
3254 pub fn into_string(self) -> String {
3255 // Safety: `bytes` are kept in UTF-8 form, only removing whole `char`s at a time.
3256 unsafe { String::from_utf8_unchecked(self.bytes.collect()) }
3257 }
3258
3259 #[inline]
3260 fn iter(&self) -> CharIndices<'_> {
3261 self.as_str().char_indices()
3262 }
3263}
3264
3265#[unstable(feature = "string_into_chars", issue = "133125")]
3266impl Iterator for IntoChars {
3267 type Item = char;
3268
3269 #[inline]
3270 fn next(&mut self) -> Option<char> {
3271 let mut iter = self.iter();
3272 match iter.next() {
3273 None => None,
3274 Some((_, ch)) => {
3275 let offset = iter.offset();
3276 // `offset` is a valid index.
3277 let _ = self.bytes.advance_by(offset);
3278 Some(ch)
3279 }
3280 }
3281 }
3282
3283 #[inline]
3284 fn count(self) -> usize {
3285 self.iter().count()
3286 }
3287
3288 #[inline]
3289 fn size_hint(&self) -> (usize, Option<usize>) {
3290 self.iter().size_hint()
3291 }
3292
3293 #[inline]
3294 fn last(mut self) -> Option<char> {
3295 self.next_back()
3296 }
3297}
3298
3299#[unstable(feature = "string_into_chars", issue = "133125")]
3300impl DoubleEndedIterator for IntoChars {
3301 #[inline]
3302 fn next_back(&mut self) -> Option<char> {
3303 let len = self.as_str().len();
3304 let mut iter = self.iter();
3305 match iter.next_back() {
3306 None => None,
3307 Some((idx, ch)) => {
3308 // `idx` is a valid index.
3309 let _ = self.bytes.advance_back_by(len - idx);
3310 Some(ch)
3311 }
3312 }
3313 }
3314}
3315
3316#[unstable(feature = "string_into_chars", issue = "133125")]
3317impl FusedIterator for IntoChars {}
3318
3319/// A draining iterator for `String`.
3320///
3321/// This struct is created by the [`drain`] method on [`String`]. See its
3322/// documentation for more.
3323///
3324/// [`drain`]: String::drain
3325#[stable(feature = "drain", since = "1.6.0")]
3326pub struct Drain<'a> {
3327 /// Will be used as &'a mut String in the destructor
3328 string: *mut String,
3329 /// Start of part to remove
3330 start: usize,
3331 /// End of part to remove
3332 end: usize,
3333 /// Current remaining range to remove
3334 iter: Chars<'a>,
3335}
3336
3337#[stable(feature = "collection_debug", since = "1.17.0")]
3338impl fmt::Debug for Drain<'_> {
3339 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
3340 f.debug_tuple("Drain").field(&self.as_str()).finish()
3341 }
3342}
3343
3344#[stable(feature = "drain", since = "1.6.0")]
3345unsafe impl Sync for Drain<'_> {}
3346#[stable(feature = "drain", since = "1.6.0")]
3347unsafe impl Send for Drain<'_> {}
3348
3349#[stable(feature = "drain", since = "1.6.0")]
3350impl Drop for Drain<'_> {
3351 fn drop(&mut self) {
3352 unsafe {
3353 // Use Vec::drain. "Reaffirm" the bounds checks to avoid
3354 // panic code being inserted again.
3355 let self_vec = (*self.string).as_mut_vec();
3356 if self.start <= self.end && self.end <= self_vec.len() {
3357 self_vec.drain(self.start..self.end);
3358 }
3359 }
3360 }
3361}
3362
3363impl<'a> Drain<'a> {
3364 /// Returns the remaining (sub)string of this iterator as a slice.
3365 ///
3366 /// # Examples
3367 ///
3368 /// ```
3369 /// let mut s = String::from("abc");
3370 /// let mut drain = s.drain(..);
3371 /// assert_eq!(drain.as_str(), "abc");
3372 /// let _ = drain.next().unwrap();
3373 /// assert_eq!(drain.as_str(), "bc");
3374 /// ```
3375 #[must_use]
3376 #[stable(feature = "string_drain_as_str", since = "1.55.0")]
3377 pub fn as_str(&self) -> &str {
3378 self.iter.as_str()
3379 }
3380}
3381
3382#[stable(feature = "string_drain_as_str", since = "1.55.0")]
3383impl<'a> AsRef<str> for Drain<'a> {
3384 fn as_ref(&self) -> &str {
3385 self.as_str()
3386 }
3387}
3388
3389#[stable(feature = "string_drain_as_str", since = "1.55.0")]
3390impl<'a> AsRef<[u8]> for Drain<'a> {
3391 fn as_ref(&self) -> &[u8] {
3392 self.as_str().as_bytes()
3393 }
3394}
3395
3396#[stable(feature = "drain", since = "1.6.0")]
3397impl Iterator for Drain<'_> {
3398 type Item = char;
3399
3400 #[inline]
3401 fn next(&mut self) -> Option<char> {
3402 self.iter.next()
3403 }
3404
3405 fn size_hint(&self) -> (usize, Option<usize>) {
3406 self.iter.size_hint()
3407 }
3408
3409 #[inline]
3410 fn last(mut self) -> Option<char> {
3411 self.next_back()
3412 }
3413}
3414
3415#[stable(feature = "drain", since = "1.6.0")]
3416impl DoubleEndedIterator for Drain<'_> {
3417 #[inline]
3418 fn next_back(&mut self) -> Option<char> {
3419 self.iter.next_back()
3420 }
3421}
3422
3423#[stable(feature = "fused", since = "1.26.0")]
3424impl FusedIterator for Drain<'_> {}
3425
3426#[cfg(not(no_global_oom_handling))]
3427#[stable(feature = "from_char_for_string", since = "1.46.0")]
3428impl From<char> for String {
3429 /// Allocates an owned [`String`] from a single character.
3430 ///
3431 /// # Example
3432 /// ```rust
3433 /// let c: char = 'a';
3434 /// let s: String = String::from(c);
3435 /// assert_eq!("a", &s[..]);
3436 /// ```
3437 #[inline]
3438 fn from(c: char) -> Self {
3439 c.to_string()
3440 }
3441}