1//! A UTF-8–encoded, 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!("πŸ’–", 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-8–encoded, 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!("πŸ’–", 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/// // `s` is ASCII which represents each `char` as one byte
123/// let s = "hello";
124/// assert_eq!(s.len(), 5);
125///
126/// // A `char` array with the same contents would be longer because
127/// // every `char` is four bytes
128/// let s = ['h', 'e', 'l', 'l', 'o'];
129/// let size: usize = s.into_iter().map(|c| size_of_val(&c)).sum();
130/// assert_eq!(size, 20);
131///
132/// // However, for non-ASCII strings, the difference will be smaller
133/// // and sometimes they are the same
134/// let s = "πŸ’–πŸ’–πŸ’–πŸ’–πŸ’–";
135/// assert_eq!(s.len(), 20);
136///
137/// let s = ['πŸ’–', 'πŸ’–', 'πŸ’–', 'πŸ’–', 'πŸ’–'];
138/// let size: usize = s.into_iter().map(|c| size_of_val(&c)).sum();
139/// assert_eq!(size, 20);
140/// ```
141///
142/// This raises interesting questions as to how `s[i]` should work.
143/// What should `i` be here? Several options include byte indices and
144/// `char` indices but, because of UTF-8 encoding, only byte indices
145/// would provide constant time indexing. Getting the `i`th `char`, for
146/// example, is available using [`chars`]:
147///
148/// ```
149/// let s = "hello";
150/// let third_character = s.chars().nth(2);
151/// assert_eq!(third_character, Some('l'));
152///
153/// let s = "πŸ’–πŸ’–πŸ’–πŸ’–πŸ’–";
154/// let third_character = s.chars().nth(2);
155/// assert_eq!(third_character, Some('πŸ’–'));
156/// ```
157///
158/// Next, what should `s[i]` return? Because indexing returns a reference
159/// to underlying data it could be `&u8`, `&[u8]`, or something else similar.
160/// Since we're only providing one index, `&u8` makes the most sense but that
161/// might not be what the user expects and can be explicitly achieved with
162/// [`as_bytes()`]:
163///
164/// ```
165/// // The first byte is 104 - the byte value of `'h'`
166/// let s = "hello";
167/// assert_eq!(s.as_bytes()[0], 104);
168/// // or
169/// assert_eq!(s.as_bytes()[0], b'h');
170///
171/// // The first byte is 240 which isn't obviously useful
172/// let s = "πŸ’–πŸ’–πŸ’–πŸ’–πŸ’–";
173/// assert_eq!(s.as_bytes()[0], 240);
174/// ```
175///
176/// Due to these ambiguities/restrictions, indexing with a `usize` is simply
177/// forbidden:
178///
179/// ```compile_fail,E0277
180/// let s = "hello";
181///
182/// // The following will not compile!
183/// println!("The first letter of s is {}", s[0]);
184/// ```
185///
186/// It is more clear, however, how `&s[i..j]` should work (that is,
187/// indexing with a range). It should accept byte indices (to be constant-time)
188/// and return a `&str` which is UTF-8 encoded. This is also called "string slicing".
189/// Note this will panic if the byte indices provided are not character
190/// boundaries - see [`is_char_boundary`] for more details. See the implementations
191/// for [`SliceIndex<str>`] for more details on string slicing. For a non-panicking
192/// version of string slicing, see [`get`].
193///
194/// [`OsString`]: ../../std/ffi/struct.OsString.html "ffi::OsString"
195/// [`SliceIndex<str>`]: core::slice::SliceIndex
196/// [`as_bytes()`]: str::as_bytes
197/// [`get`]: str::get
198/// [`is_char_boundary`]: str::is_char_boundary
199///
200/// The [`bytes`] and [`chars`] methods return iterators over the bytes and
201/// codepoints of the string, respectively. To iterate over codepoints along
202/// with byte indices, use [`char_indices`].
203///
204/// [`bytes`]: str::bytes
205/// [`chars`]: str::chars
206/// [`char_indices`]: str::char_indices
207///
208/// # Deref
209///
210/// `String` implements <code>[Deref]<Target = [str]></code>, and so inherits all of [`str`]'s
211/// methods. In addition, this means that you can pass a `String` to a
212/// function which takes a [`&str`] by using an ampersand (`&`):
213///
214/// ```
215/// fn takes_str(s: &str) { }
216///
217/// let s = String::from("Hello");
218///
219/// takes_str(&s);
220/// ```
221///
222/// This will create a [`&str`] from the `String` and pass it in. This
223/// conversion is very inexpensive, and so generally, functions will accept
224/// [`&str`]s as arguments unless they need a `String` for some specific
225/// reason.
226///
227/// In certain cases Rust doesn't have enough information to make this
228/// conversion, known as [`Deref`] coercion. In the following example a string
229/// slice [`&'a str`][`&str`] implements the trait `TraitExample`, and the function
230/// `example_func` takes anything that implements the trait. In this case Rust
231/// would need to make two implicit conversions, which Rust doesn't have the
232/// means to do. For that reason, the following example will not compile.
233///
234/// ```compile_fail,E0277
235/// trait TraitExample {}
236///
237/// impl<'a> TraitExample for &'a str {}
238///
239/// fn example_func<A: TraitExample>(example_arg: A) {}
240///
241/// let example_string = String::from("example_string");
242/// example_func(&example_string);
243/// ```
244///
245/// There are two options that would work instead. The first would be to
246/// change the line `example_func(&example_string);` to
247/// `example_func(example_string.as_str());`, using the method [`as_str()`]
248/// to explicitly extract the string slice containing the string. The second
249/// way changes `example_func(&example_string);` to
250/// `example_func(&*example_string);`. In this case we are dereferencing a
251/// `String` to a [`str`], then referencing the [`str`] back to
252/// [`&str`]. The second way is more idiomatic, however both work to do the
253/// conversion explicitly rather than relying on the implicit conversion.
254///
255/// # Representation
256///
257/// A `String` is made up of three components: a pointer to some bytes, a
258/// length, and a capacity. The pointer points to the internal buffer which `String`
259/// uses to store its data. The length is the number of bytes currently stored
260/// in the buffer, and the capacity is the size of the buffer in bytes. As such,
261/// the length will always be less than or equal to the capacity.
262///
263/// This buffer is always stored on the heap.
264///
265/// You can look at these with the [`as_ptr`], [`len`], and [`capacity`]
266/// methods:
267///
268/// ```
269/// use std::mem;
270///
271/// let story = String::from("Once upon a time...");
272///
273// FIXME Update this when vec_into_raw_parts is stabilized
274/// // Prevent automatically dropping the String's data
275/// let mut story = mem::ManuallyDrop::new(story);
276///
277/// let ptr = story.as_mut_ptr();
278/// let len = story.len();
279/// let capacity = story.capacity();
280///
281/// // story has nineteen bytes
282/// assert_eq!(19, len);
283///
284/// // We can re-build a String out of ptr, len, and capacity. This is all
285/// // unsafe because we are responsible for making sure the components are
286/// // valid:
287/// let s = unsafe { String::from_raw_parts(ptr, len, capacity) } ;
288///
289/// assert_eq!(String::from("Once upon a time..."), s);
290/// ```
291///
292/// [`as_ptr`]: str::as_ptr
293/// [`len`]: String::len
294/// [`capacity`]: String::capacity
295///
296/// If a `String` has enough capacity, adding elements to it will not
297/// re-allocate. For example, consider this program:
298///
299/// ```
300/// let mut s = String::new();
301///
302/// println!("{}", s.capacity());
303///
304/// for _ in 0..5 {
305/// s.push_str("hello");
306/// println!("{}", s.capacity());
307/// }
308/// ```
309///
310/// This will output the following:
311///
312/// ```text
313/// 0
314/// 8
315/// 16
316/// 16
317/// 32
318/// 32
319/// ```
320///
321/// At first, we have no memory allocated at all, but as we append to the
322/// string, it increases its capacity appropriately. If we instead use the
323/// [`with_capacity`] method to allocate the correct capacity initially:
324///
325/// ```
326/// let mut s = String::with_capacity(25);
327///
328/// println!("{}", s.capacity());
329///
330/// for _ in 0..5 {
331/// s.push_str("hello");
332/// println!("{}", s.capacity());
333/// }
334/// ```
335///
336/// [`with_capacity`]: String::with_capacity
337///
338/// We end up with a different output:
339///
340/// ```text
341/// 25
342/// 25
343/// 25
344/// 25
345/// 25
346/// 25
347/// ```
348///
349/// Here, there's no need to allocate more memory inside the loop.
350///
351/// [str]: prim@str "str"
352/// [`str`]: prim@str "str"
353/// [`&str`]: prim@str "&str"
354/// [Deref]: core::ops::Deref "ops::Deref"
355/// [`Deref`]: core::ops::Deref "ops::Deref"
356/// [`as_str()`]: String::as_str
357#[derive(PartialEq, PartialOrd, Eq, Ord)]
358#[stable(feature = "rust1", since = "1.0.0")]
359#[lang = "String"]
360pub struct String {
361 vec: Vec<u8>,
362}
363
364/// A possible error value when converting a `String` from a UTF-8 byte vector.
365///
366/// This type is the error type for the [`from_utf8`] method on [`String`]. It
367/// is designed in such a way to carefully avoid reallocations: the
368/// [`into_bytes`] method will give back the byte vector that was used in the
369/// conversion attempt.
370///
371/// [`from_utf8`]: String::from_utf8
372/// [`into_bytes`]: FromUtf8Error::into_bytes
373///
374/// The [`Utf8Error`] type provided by [`std::str`] represents an error that may
375/// occur when converting a slice of [`u8`]s to a [`&str`]. In this sense, it's
376/// an analogue to `FromUtf8Error`, and you can get one from a `FromUtf8Error`
377/// through the [`utf8_error`] method.
378///
379/// [`Utf8Error`]: str::Utf8Error "std::str::Utf8Error"
380/// [`std::str`]: core::str "std::str"
381/// [`&str`]: prim@str "&str"
382/// [`utf8_error`]: FromUtf8Error::utf8_error
383///
384/// # Examples
385///
386/// ```
387/// // some invalid bytes, in a vector
388/// let bytes = vec![0, 159];
389///
390/// let value = String::from_utf8(bytes);
391///
392/// assert!(value.is_err());
393/// assert_eq!(vec![0, 159], value.unwrap_err().into_bytes());
394/// ```
395#[stable(feature = "rust1", since = "1.0.0")]
396#[cfg_attr(not(no_global_oom_handling), derive(Clone))]
397#[derive(Debug, PartialEq, Eq)]
398pub struct FromUtf8Error {
399 bytes: Vec<u8>,
400 error: Utf8Error,
401}
402
403/// A possible error value when converting a `String` from a UTF-16 byte slice.
404///
405/// This type is the error type for the [`from_utf16`] method on [`String`].
406///
407/// [`from_utf16`]: String::from_utf16
408///
409/// # Examples
410///
411/// ```
412/// // π„žmu<invalid>ic
413/// let v = &[0xD834, 0xDD1E, 0x006d, 0x0075,
414/// 0xD800, 0x0069, 0x0063];
415///
416/// assert!(String::from_utf16(v).is_err());
417/// ```
418#[stable(feature = "rust1", since = "1.0.0")]
419#[derive(Debug)]
420pub struct FromUtf16Error(());
421
422impl String {
423 /// Creates a new empty `String`.
424 ///
425 /// Given that the `String` is empty, this will not allocate any initial
426 /// buffer. While that means that this initial operation is very
427 /// inexpensive, it may cause excessive allocation later when you add
428 /// data. If you have an idea of how much data the `String` will hold,
429 /// consider the [`with_capacity`] method to prevent excessive
430 /// re-allocation.
431 ///
432 /// [`with_capacity`]: String::with_capacity
433 ///
434 /// # Examples
435 ///
436 /// ```
437 /// let s = String::new();
438 /// ```
439 #[inline]
440 #[rustc_const_stable(feature = "const_string_new", since = "1.39.0")]
441 #[rustc_diagnostic_item = "string_new"]
442 #[stable(feature = "rust1", since = "1.0.0")]
443 #[must_use]
444 pub const fn new() -> String {
445 String { vec: Vec::new() }
446 }
447
448 /// Creates a new empty `String` with at least the specified capacity.
449 ///
450 /// `String`s have an internal buffer to hold their data. The capacity is
451 /// the length of that buffer, and can be queried with the [`capacity`]
452 /// method. This method creates an empty `String`, but one with an initial
453 /// buffer that can hold at least `capacity` bytes. This is useful when you
454 /// may be appending a bunch of data to the `String`, reducing the number of
455 /// reallocations it needs to do.
456 ///
457 /// [`capacity`]: String::capacity
458 ///
459 /// If the given capacity is `0`, no allocation will occur, and this method
460 /// is identical to the [`new`] method.
461 ///
462 /// [`new`]: String::new
463 ///
464 /// # Examples
465 ///
466 /// ```
467 /// let mut s = String::with_capacity(10);
468 ///
469 /// // The String contains no chars, even though it has capacity for more
470 /// assert_eq!(s.len(), 0);
471 ///
472 /// // These are all done without reallocating...
473 /// let cap = s.capacity();
474 /// for _ in 0..10 {
475 /// s.push('a');
476 /// }
477 ///
478 /// assert_eq!(s.capacity(), cap);
479 ///
480 /// // ...but this may make the string reallocate
481 /// s.push('a');
482 /// ```
483 #[cfg(not(no_global_oom_handling))]
484 #[inline]
485 #[stable(feature = "rust1", since = "1.0.0")]
486 #[must_use]
487 pub fn with_capacity(capacity: usize) -> String {
488 String { vec: Vec::with_capacity(capacity) }
489 }
490
491 /// Creates a new empty `String` with at least the specified capacity.
492 ///
493 /// # Errors
494 ///
495 /// Returns [`Err`] if the capacity exceeds `isize::MAX` bytes,
496 /// or if the memory allocator reports failure.
497 ///
498 #[inline]
499 #[unstable(feature = "try_with_capacity", issue = "91913")]
500 pub fn try_with_capacity(capacity: usize) -> Result<String, TryReserveError> {
501 Ok(String { vec: Vec::try_with_capacity(capacity)? })
502 }
503
504 /// Converts a vector of bytes to a `String`.
505 ///
506 /// A string ([`String`]) is made of bytes ([`u8`]), and a vector of bytes
507 /// ([`Vec<u8>`]) is made of bytes, so this function converts between the
508 /// two. Not all byte slices are valid `String`s, however: `String`
509 /// requires that it is valid UTF-8. `from_utf8()` checks to ensure that
510 /// the bytes are valid UTF-8, and then does the conversion.
511 ///
512 /// If you are sure that the byte slice is valid UTF-8, and you don't want
513 /// to incur the overhead of the validity check, there is an unsafe version
514 /// of this function, [`from_utf8_unchecked`], which has the same behavior
515 /// but skips the check.
516 ///
517 /// This method will take care to not copy the vector, for efficiency's
518 /// sake.
519 ///
520 /// If you need a [`&str`] instead of a `String`, consider
521 /// [`str::from_utf8`].
522 ///
523 /// The inverse of this method is [`into_bytes`].
524 ///
525 /// # Errors
526 ///
527 /// Returns [`Err`] if the slice is not UTF-8 with a description as to why the
528 /// provided bytes are not UTF-8. The vector you moved in is also included.
529 ///
530 /// # Examples
531 ///
532 /// Basic usage:
533 ///
534 /// ```
535 /// // some bytes, in a vector
536 /// let sparkle_heart = vec![240, 159, 146, 150];
537 ///
538 /// // We know these bytes are valid, so we'll use `unwrap()`.
539 /// let sparkle_heart = String::from_utf8(sparkle_heart).unwrap();
540 ///
541 /// assert_eq!("πŸ’–", sparkle_heart);
542 /// ```
543 ///
544 /// Incorrect bytes:
545 ///
546 /// ```
547 /// // some invalid bytes, in a vector
548 /// let sparkle_heart = vec![0, 159, 146, 150];
549 ///
550 /// assert!(String::from_utf8(sparkle_heart).is_err());
551 /// ```
552 ///
553 /// See the docs for [`FromUtf8Error`] for more details on what you can do
554 /// with this error.
555 ///
556 /// [`from_utf8_unchecked`]: String::from_utf8_unchecked
557 /// [`Vec<u8>`]: crate::vec::Vec "Vec"
558 /// [`&str`]: prim@str "&str"
559 /// [`into_bytes`]: String::into_bytes
560 #[inline]
561 #[stable(feature = "rust1", since = "1.0.0")]
562 #[rustc_diagnostic_item = "string_from_utf8"]
563 pub fn from_utf8(vec: Vec<u8>) -> Result<String, FromUtf8Error> {
564 match str::from_utf8(&vec) {
565 Ok(..) => Ok(String { vec }),
566 Err(e) => Err(FromUtf8Error { bytes: vec, error: e }),
567 }
568 }
569
570 /// Converts a slice of bytes to a string, including invalid characters.
571 ///
572 /// Strings are made of bytes ([`u8`]), and a slice of bytes
573 /// ([`&[u8]`][byteslice]) is made of bytes, so this function converts
574 /// between the two. Not all byte slices are valid strings, however: strings
575 /// are required to be valid UTF-8. During this conversion,
576 /// `from_utf8_lossy()` will replace any invalid UTF-8 sequences with
577 /// [`U+FFFD REPLACEMENT CHARACTER`][U+FFFD], which looks like this: οΏ½
578 ///
579 /// [byteslice]: prim@slice
580 /// [U+FFFD]: core::char::REPLACEMENT_CHARACTER
581 ///
582 /// If you are sure that the byte slice is valid UTF-8, and you don't want
583 /// to incur the overhead of the conversion, there is an unsafe version
584 /// of this function, [`from_utf8_unchecked`], which has the same behavior
585 /// but skips the checks.
586 ///
587 /// [`from_utf8_unchecked`]: String::from_utf8_unchecked
588 ///
589 /// This function returns a [`Cow<'a, str>`]. If our byte slice is invalid
590 /// UTF-8, then we need to insert the replacement characters, which will
591 /// change the size of the string, and hence, require a `String`. But if
592 /// it's already valid UTF-8, we don't need a new allocation. This return
593 /// type allows us to handle both cases.
594 ///
595 /// [`Cow<'a, str>`]: crate::borrow::Cow "borrow::Cow"
596 ///
597 /// # Examples
598 ///
599 /// Basic usage:
600 ///
601 /// ```
602 /// // some bytes, in a vector
603 /// let sparkle_heart = vec![240, 159, 146, 150];
604 ///
605 /// let sparkle_heart = String::from_utf8_lossy(&sparkle_heart);
606 ///
607 /// assert_eq!("πŸ’–", sparkle_heart);
608 /// ```
609 ///
610 /// Incorrect bytes:
611 ///
612 /// ```
613 /// // some invalid bytes
614 /// let input = b"Hello \xF0\x90\x80World";
615 /// let output = String::from_utf8_lossy(input);
616 ///
617 /// assert_eq!("Hello οΏ½World", output);
618 /// ```
619 #[must_use]
620 #[cfg(not(no_global_oom_handling))]
621 #[stable(feature = "rust1", since = "1.0.0")]
622 pub fn from_utf8_lossy(v: &[u8]) -> Cow<'_, str> {
623 let mut iter = v.utf8_chunks();
624
625 let first_valid = if let Some(chunk) = iter.next() {
626 let valid = chunk.valid();
627 if chunk.invalid().is_empty() {
628 debug_assert_eq!(valid.len(), v.len());
629 return Cow::Borrowed(valid);
630 }
631 valid
632 } else {
633 return Cow::Borrowed("");
634 };
635
636 const REPLACEMENT: &str = "\u{FFFD}";
637
638 let mut res = String::with_capacity(v.len());
639 res.push_str(first_valid);
640 res.push_str(REPLACEMENT);
641
642 for chunk in iter {
643 res.push_str(chunk.valid());
644 if !chunk.invalid().is_empty() {
645 res.push_str(REPLACEMENT);
646 }
647 }
648
649 Cow::Owned(res)
650 }
651
652 /// Converts a [`Vec<u8>`] to a `String`, substituting invalid UTF-8
653 /// sequences with replacement characters.
654 ///
655 /// See [`from_utf8_lossy`] for more details.
656 ///
657 /// [`from_utf8_lossy`]: String::from_utf8_lossy
658 ///
659 /// Note that this function does not guarantee reuse of the original `Vec`
660 /// allocation.
661 ///
662 /// # Examples
663 ///
664 /// Basic usage:
665 ///
666 /// ```
667 /// #![feature(string_from_utf8_lossy_owned)]
668 /// // some bytes, in a vector
669 /// let sparkle_heart = vec![240, 159, 146, 150];
670 ///
671 /// let sparkle_heart = String::from_utf8_lossy_owned(sparkle_heart);
672 ///
673 /// assert_eq!(String::from("πŸ’–"), sparkle_heart);
674 /// ```
675 ///
676 /// Incorrect bytes:
677 ///
678 /// ```
679 /// #![feature(string_from_utf8_lossy_owned)]
680 /// // some invalid bytes
681 /// let input: Vec<u8> = b"Hello \xF0\x90\x80World".into();
682 /// let output = String::from_utf8_lossy_owned(input);
683 ///
684 /// assert_eq!(String::from("Hello οΏ½World"), output);
685 /// ```
686 #[must_use]
687 #[cfg(not(no_global_oom_handling))]
688 #[unstable(feature = "string_from_utf8_lossy_owned", issue = "129436")]
689 pub fn from_utf8_lossy_owned(v: Vec<u8>) -> String {
690 if let Cow::Owned(string) = String::from_utf8_lossy(&v) {
691 string
692 } else {
693 // SAFETY: `String::from_utf8_lossy`'s contract ensures that if
694 // it returns a `Cow::Borrowed`, it is a valid UTF-8 string.
695 // Otherwise, it returns a new allocation of an owned `String`, with
696 // replacement characters for invalid sequences, which is returned
697 // above.
698 unsafe { String::from_utf8_unchecked(v) }
699 }
700 }
701
702 /// Decode a native endian UTF-16–encoded vector `v` into a `String`,
703 /// returning [`Err`] if `v` contains any invalid data.
704 ///
705 /// # Examples
706 ///
707 /// ```
708 /// // π„žmusic
709 /// let v = &[0xD834, 0xDD1E, 0x006d, 0x0075,
710 /// 0x0073, 0x0069, 0x0063];
711 /// assert_eq!(String::from("π„žmusic"),
712 /// String::from_utf16(v).unwrap());
713 ///
714 /// // π„žmu<invalid>ic
715 /// let v = &[0xD834, 0xDD1E, 0x006d, 0x0075,
716 /// 0xD800, 0x0069, 0x0063];
717 /// assert!(String::from_utf16(v).is_err());
718 /// ```
719 #[cfg(not(no_global_oom_handling))]
720 #[stable(feature = "rust1", since = "1.0.0")]
721 pub fn from_utf16(v: &[u16]) -> Result<String, FromUtf16Error> {
722 // This isn't done via collect::<Result<_, _>>() for performance reasons.
723 // FIXME: the function can be simplified again when #48994 is closed.
724 let mut ret = String::with_capacity(v.len());
725 for c in char::decode_utf16(v.iter().cloned()) {
726 if let Ok(c) = c {
727 ret.push(c);
728 } else {
729 return Err(FromUtf16Error(()));
730 }
731 }
732 Ok(ret)
733 }
734
735 /// Decode a native endian UTF-16–encoded slice `v` into a `String`,
736 /// replacing invalid data with [the replacement character (`U+FFFD`)][U+FFFD].
737 ///
738 /// Unlike [`from_utf8_lossy`] which returns a [`Cow<'a, str>`],
739 /// `from_utf16_lossy` returns a `String` since the UTF-16 to UTF-8
740 /// conversion requires a memory allocation.
741 ///
742 /// [`from_utf8_lossy`]: String::from_utf8_lossy
743 /// [`Cow<'a, str>`]: crate::borrow::Cow "borrow::Cow"
744 /// [U+FFFD]: core::char::REPLACEMENT_CHARACTER
745 ///
746 /// # Examples
747 ///
748 /// ```
749 /// // π„žmus<invalid>ic<invalid>
750 /// let v = &[0xD834, 0xDD1E, 0x006d, 0x0075,
751 /// 0x0073, 0xDD1E, 0x0069, 0x0063,
752 /// 0xD834];
753 ///
754 /// assert_eq!(String::from("π„žmus\u{FFFD}ic\u{FFFD}"),
755 /// String::from_utf16_lossy(v));
756 /// ```
757 #[cfg(not(no_global_oom_handling))]
758 #[must_use]
759 #[inline]
760 #[stable(feature = "rust1", since = "1.0.0")]
761 pub fn from_utf16_lossy(v: &[u16]) -> String {
762 char::decode_utf16(v.iter().cloned())
763 .map(|r| r.unwrap_or(char::REPLACEMENT_CHARACTER))
764 .collect()
765 }
766
767 /// Decode a UTF-16LE–encoded vector `v` into a `String`,
768 /// returning [`Err`] if `v` contains any invalid data.
769 ///
770 /// # Examples
771 ///
772 /// Basic usage:
773 ///
774 /// ```
775 /// #![feature(str_from_utf16_endian)]
776 /// // π„žmusic
777 /// let v = &[0x34, 0xD8, 0x1E, 0xDD, 0x6d, 0x00, 0x75, 0x00,
778 /// 0x73, 0x00, 0x69, 0x00, 0x63, 0x00];
779 /// assert_eq!(String::from("π„žmusic"),
780 /// String::from_utf16le(v).unwrap());
781 ///
782 /// // π„žmu<invalid>ic
783 /// let v = &[0x34, 0xD8, 0x1E, 0xDD, 0x6d, 0x00, 0x75, 0x00,
784 /// 0x00, 0xD8, 0x69, 0x00, 0x63, 0x00];
785 /// assert!(String::from_utf16le(v).is_err());
786 /// ```
787 #[cfg(not(no_global_oom_handling))]
788 #[unstable(feature = "str_from_utf16_endian", issue = "116258")]
789 pub fn from_utf16le(v: &[u8]) -> Result<String, FromUtf16Error> {
790 if v.len() % 2 != 0 {
791 return Err(FromUtf16Error(()));
792 }
793 match (cfg!(target_endian = "little"), unsafe { v.align_to::<u16>() }) {
794 (true, ([], v, [])) => Self::from_utf16(v),
795 _ => char::decode_utf16(v.array_chunks::<2>().copied().map(u16::from_le_bytes))
796 .collect::<Result<_, _>>()
797 .map_err(|_| FromUtf16Error(())),
798 }
799 }
800
801 /// Decode a UTF-16LE–encoded slice `v` into a `String`, replacing
802 /// invalid data with [the replacement character (`U+FFFD`)][U+FFFD].
803 ///
804 /// Unlike [`from_utf8_lossy`] which returns a [`Cow<'a, str>`],
805 /// `from_utf16le_lossy` returns a `String` since the UTF-16 to UTF-8
806 /// conversion requires a memory allocation.
807 ///
808 /// [`from_utf8_lossy`]: String::from_utf8_lossy
809 /// [`Cow<'a, str>`]: crate::borrow::Cow "borrow::Cow"
810 /// [U+FFFD]: core::char::REPLACEMENT_CHARACTER
811 ///
812 /// # Examples
813 ///
814 /// Basic usage:
815 ///
816 /// ```
817 /// #![feature(str_from_utf16_endian)]
818 /// // π„žmus<invalid>ic<invalid>
819 /// let v = &[0x34, 0xD8, 0x1E, 0xDD, 0x6d, 0x00, 0x75, 0x00,
820 /// 0x73, 0x00, 0x1E, 0xDD, 0x69, 0x00, 0x63, 0x00,
821 /// 0x34, 0xD8];
822 ///
823 /// assert_eq!(String::from("π„žmus\u{FFFD}ic\u{FFFD}"),
824 /// String::from_utf16le_lossy(v));
825 /// ```
826 #[cfg(not(no_global_oom_handling))]
827 #[unstable(feature = "str_from_utf16_endian", issue = "116258")]
828 pub fn from_utf16le_lossy(v: &[u8]) -> String {
829 match (cfg!(target_endian = "little"), unsafe { v.align_to::<u16>() }) {
830 (true, ([], v, [])) => Self::from_utf16_lossy(v),
831 (true, ([], v, [_remainder])) => Self::from_utf16_lossy(v) + "\u{FFFD}",
832 _ => {
833 let mut iter = v.array_chunks::<2>();
834 let string = char::decode_utf16(iter.by_ref().copied().map(u16::from_le_bytes))
835 .map(|r| r.unwrap_or(char::REPLACEMENT_CHARACTER))
836 .collect();
837 if iter.remainder().is_empty() { string } else { string + "\u{FFFD}" }
838 }
839 }
840 }
841
842 /// Decode a UTF-16BE–encoded vector `v` into a `String`,
843 /// returning [`Err`] if `v` contains any invalid data.
844 ///
845 /// # Examples
846 ///
847 /// Basic usage:
848 ///
849 /// ```
850 /// #![feature(str_from_utf16_endian)]
851 /// // π„žmusic
852 /// let v = &[0xD8, 0x34, 0xDD, 0x1E, 0x00, 0x6d, 0x00, 0x75,
853 /// 0x00, 0x73, 0x00, 0x69, 0x00, 0x63];
854 /// assert_eq!(String::from("π„žmusic"),
855 /// String::from_utf16be(v).unwrap());
856 ///
857 /// // π„žmu<invalid>ic
858 /// let v = &[0xD8, 0x34, 0xDD, 0x1E, 0x00, 0x6d, 0x00, 0x75,
859 /// 0xD8, 0x00, 0x00, 0x69, 0x00, 0x63];
860 /// assert!(String::from_utf16be(v).is_err());
861 /// ```
862 #[cfg(not(no_global_oom_handling))]
863 #[unstable(feature = "str_from_utf16_endian", issue = "116258")]
864 pub fn from_utf16be(v: &[u8]) -> Result<String, FromUtf16Error> {
865 if v.len() % 2 != 0 {
866 return Err(FromUtf16Error(()));
867 }
868 match (cfg!(target_endian = "big"), unsafe { v.align_to::<u16>() }) {
869 (true, ([], v, [])) => Self::from_utf16(v),
870 _ => char::decode_utf16(v.array_chunks::<2>().copied().map(u16::from_be_bytes))
871 .collect::<Result<_, _>>()
872 .map_err(|_| FromUtf16Error(())),
873 }
874 }
875
876 /// Decode a UTF-16BE–encoded slice `v` into a `String`, replacing
877 /// invalid data with [the replacement character (`U+FFFD`)][U+FFFD].
878 ///
879 /// Unlike [`from_utf8_lossy`] which returns a [`Cow<'a, str>`],
880 /// `from_utf16le_lossy` returns a `String` since the UTF-16 to UTF-8
881 /// conversion requires a memory allocation.
882 ///
883 /// [`from_utf8_lossy`]: String::from_utf8_lossy
884 /// [`Cow<'a, str>`]: crate::borrow::Cow "borrow::Cow"
885 /// [U+FFFD]: core::char::REPLACEMENT_CHARACTER
886 ///
887 /// # Examples
888 ///
889 /// Basic usage:
890 ///
891 /// ```
892 /// #![feature(str_from_utf16_endian)]
893 /// // π„žmus<invalid>ic<invalid>
894 /// let v = &[0xD8, 0x34, 0xDD, 0x1E, 0x00, 0x6d, 0x00, 0x75,
895 /// 0x00, 0x73, 0xDD, 0x1E, 0x00, 0x69, 0x00, 0x63,
896 /// 0xD8, 0x34];
897 ///
898 /// assert_eq!(String::from("π„žmus\u{FFFD}ic\u{FFFD}"),
899 /// String::from_utf16be_lossy(v));
900 /// ```
901 #[cfg(not(no_global_oom_handling))]
902 #[unstable(feature = "str_from_utf16_endian", issue = "116258")]
903 pub fn from_utf16be_lossy(v: &[u8]) -> String {
904 match (cfg!(target_endian = "big"), unsafe { v.align_to::<u16>() }) {
905 (true, ([], v, [])) => Self::from_utf16_lossy(v),
906 (true, ([], v, [_remainder])) => Self::from_utf16_lossy(v) + "\u{FFFD}",
907 _ => {
908 let mut iter = v.array_chunks::<2>();
909 let string = char::decode_utf16(iter.by_ref().copied().map(u16::from_be_bytes))
910 .map(|r| r.unwrap_or(char::REPLACEMENT_CHARACTER))
911 .collect();
912 if iter.remainder().is_empty() { string } else { string + "\u{FFFD}" }
913 }
914 }
915 }
916
917 /// Decomposes a `String` into its raw components: `(pointer, length, capacity)`.
918 ///
919 /// Returns the raw pointer to the underlying data, the length of
920 /// the string (in bytes), and the allocated capacity of the data
921 /// (in bytes). These are the same arguments in the same order as
922 /// the arguments to [`from_raw_parts`].
923 ///
924 /// After calling this function, the caller is responsible for the
925 /// memory previously managed by the `String`. The only way to do
926 /// this is to convert the raw pointer, length, and capacity back
927 /// into a `String` with the [`from_raw_parts`] function, allowing
928 /// the destructor to perform the cleanup.
929 ///
930 /// [`from_raw_parts`]: String::from_raw_parts
931 ///
932 /// # Examples
933 ///
934 /// ```
935 /// #![feature(vec_into_raw_parts)]
936 /// let s = String::from("hello");
937 ///
938 /// let (ptr, len, cap) = s.into_raw_parts();
939 ///
940 /// let rebuilt = unsafe { String::from_raw_parts(ptr, len, cap) };
941 /// assert_eq!(rebuilt, "hello");
942 /// ```
943 #[must_use = "losing the pointer will leak memory"]
944 #[unstable(feature = "vec_into_raw_parts", reason = "new API", issue = "65816")]
945 pub fn into_raw_parts(self) -> (*mut u8, usize, usize) {
946 self.vec.into_raw_parts()
947 }
948
949 /// Creates a new `String` from a pointer, a length and a capacity.
950 ///
951 /// # Safety
952 ///
953 /// This is highly unsafe, due to the number of invariants that aren't
954 /// checked:
955 ///
956 /// * all safety requirements for [`Vec::<u8>::from_raw_parts`].
957 /// * all safety requirements for [`String::from_utf8_unchecked`].
958 ///
959 /// Violating these may cause problems like corrupting the allocator's
960 /// internal data structures. For example, it is normally **not** safe to
961 /// build a `String` from a pointer to a C `char` array containing UTF-8
962 /// _unless_ you are certain that array was originally allocated by the
963 /// Rust standard library's allocator.
964 ///
965 /// The ownership of `buf` is effectively transferred to the
966 /// `String` which may then deallocate, reallocate or change the
967 /// contents of memory pointed to by the pointer at will. Ensure
968 /// that nothing else uses the pointer after calling this
969 /// function.
970 ///
971 /// # Examples
972 ///
973 /// ```
974 /// use std::mem;
975 ///
976 /// unsafe {
977 /// let s = String::from("hello");
978 ///
979 // FIXME Update this when vec_into_raw_parts is stabilized
980 /// // Prevent automatically dropping the String's data
981 /// let mut s = mem::ManuallyDrop::new(s);
982 ///
983 /// let ptr = s.as_mut_ptr();
984 /// let len = s.len();
985 /// let capacity = s.capacity();
986 ///
987 /// let s = String::from_raw_parts(ptr, len, capacity);
988 ///
989 /// assert_eq!(String::from("hello"), s);
990 /// }
991 /// ```
992 #[inline]
993 #[stable(feature = "rust1", since = "1.0.0")]
994 pub unsafe fn from_raw_parts(buf: *mut u8, length: usize, capacity: usize) -> String {
995 unsafe { String { vec: Vec::from_raw_parts(buf, length, capacity) } }
996 }
997
998 /// Converts a vector of bytes to a `String` without checking that the
999 /// string contains valid UTF-8.
1000 ///
1001 /// See the safe version, [`from_utf8`], for more details.
1002 ///
1003 /// [`from_utf8`]: String::from_utf8
1004 ///
1005 /// # Safety
1006 ///
1007 /// This function is unsafe because it does not check that the bytes passed
1008 /// to it are valid UTF-8. If this constraint is violated, it may cause
1009 /// memory unsafety issues with future users of the `String`, as the rest of
1010 /// the standard library assumes that `String`s are valid UTF-8.
1011 ///
1012 /// # Examples
1013 ///
1014 /// ```
1015 /// // some bytes, in a vector
1016 /// let sparkle_heart = vec![240, 159, 146, 150];
1017 ///
1018 /// let sparkle_heart = unsafe {
1019 /// String::from_utf8_unchecked(sparkle_heart)
1020 /// };
1021 ///
1022 /// assert_eq!("πŸ’–", sparkle_heart);
1023 /// ```
1024 #[inline]
1025 #[must_use]
1026 #[stable(feature = "rust1", since = "1.0.0")]
1027 pub unsafe fn from_utf8_unchecked(bytes: Vec<u8>) -> String {
1028 String { vec: bytes }
1029 }
1030
1031 /// Converts a `String` into a byte vector.
1032 ///
1033 /// This consumes the `String`, so we do not need to copy its contents.
1034 ///
1035 /// # Examples
1036 ///
1037 /// ```
1038 /// let s = String::from("hello");
1039 /// let bytes = s.into_bytes();
1040 ///
1041 /// assert_eq!(&[104, 101, 108, 108, 111][..], &bytes[..]);
1042 /// ```
1043 #[inline]
1044 #[must_use = "`self` will be dropped if the result is not used"]
1045 #[stable(feature = "rust1", since = "1.0.0")]
1046 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1047 #[rustc_allow_const_fn_unstable(const_precise_live_drops)]
1048 pub const fn into_bytes(self) -> Vec<u8> {
1049 self.vec
1050 }
1051
1052 /// Extracts a string slice containing the entire `String`.
1053 ///
1054 /// # Examples
1055 ///
1056 /// ```
1057 /// let s = String::from("foo");
1058 ///
1059 /// assert_eq!("foo", s.as_str());
1060 /// ```
1061 #[inline]
1062 #[must_use]
1063 #[stable(feature = "string_as_str", since = "1.7.0")]
1064 #[rustc_diagnostic_item = "string_as_str"]
1065 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1066 pub const fn as_str(&self) -> &str {
1067 // SAFETY: String contents are stipulated to be valid UTF-8, invalid contents are an error
1068 // at construction.
1069 unsafe { str::from_utf8_unchecked(self.vec.as_slice()) }
1070 }
1071
1072 /// Converts a `String` into a mutable string slice.
1073 ///
1074 /// # Examples
1075 ///
1076 /// ```
1077 /// let mut s = String::from("foobar");
1078 /// let s_mut_str = s.as_mut_str();
1079 ///
1080 /// s_mut_str.make_ascii_uppercase();
1081 ///
1082 /// assert_eq!("FOOBAR", s_mut_str);
1083 /// ```
1084 #[inline]
1085 #[must_use]
1086 #[stable(feature = "string_as_str", since = "1.7.0")]
1087 #[rustc_diagnostic_item = "string_as_mut_str"]
1088 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1089 pub const fn as_mut_str(&mut self) -> &mut str {
1090 // SAFETY: String contents are stipulated to be valid UTF-8, invalid contents are an error
1091 // at construction.
1092 unsafe { str::from_utf8_unchecked_mut(self.vec.as_mut_slice()) }
1093 }
1094
1095 /// Appends a given string slice onto the end of this `String`.
1096 ///
1097 /// # Examples
1098 ///
1099 /// ```
1100 /// let mut s = String::from("foo");
1101 ///
1102 /// s.push_str("bar");
1103 ///
1104 /// assert_eq!("foobar", s);
1105 /// ```
1106 #[cfg(not(no_global_oom_handling))]
1107 #[inline]
1108 #[stable(feature = "rust1", since = "1.0.0")]
1109 #[rustc_confusables("append", "push")]
1110 #[rustc_diagnostic_item = "string_push_str"]
1111 pub fn push_str(&mut self, string: &str) {
1112 self.vec.extend_from_slice(string.as_bytes())
1113 }
1114
1115 /// Copies elements from `src` range to the end of the string.
1116 ///
1117 /// # Panics
1118 ///
1119 /// Panics if the starting point or end point do not lie on a [`char`]
1120 /// boundary, or if they're out of bounds.
1121 ///
1122 /// # Examples
1123 ///
1124 /// ```
1125 /// let mut string = String::from("abcde");
1126 ///
1127 /// string.extend_from_within(2..);
1128 /// assert_eq!(string, "abcdecde");
1129 ///
1130 /// string.extend_from_within(..2);
1131 /// assert_eq!(string, "abcdecdeab");
1132 ///
1133 /// string.extend_from_within(4..8);
1134 /// assert_eq!(string, "abcdecdeabecde");
1135 /// ```
1136 #[cfg(not(no_global_oom_handling))]
1137 #[stable(feature = "string_extend_from_within", since = "1.87.0")]
1138 pub fn extend_from_within<R>(&mut self, src: R)
1139 where
1140 R: RangeBounds<usize>,
1141 {
1142 let src @ Range { start, end } = slice::range(src, ..self.len());
1143
1144 assert!(self.is_char_boundary(start));
1145 assert!(self.is_char_boundary(end));
1146
1147 self.vec.extend_from_within(src);
1148 }
1149
1150 /// Returns this `String`'s capacity, in bytes.
1151 ///
1152 /// # Examples
1153 ///
1154 /// ```
1155 /// let s = String::with_capacity(10);
1156 ///
1157 /// assert!(s.capacity() >= 10);
1158 /// ```
1159 #[inline]
1160 #[must_use]
1161 #[stable(feature = "rust1", since = "1.0.0")]
1162 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1163 pub const fn capacity(&self) -> usize {
1164 self.vec.capacity()
1165 }
1166
1167 /// Reserves capacity for at least `additional` bytes more than the
1168 /// current length. The allocator may reserve more space to speculatively
1169 /// avoid frequent allocations. After calling `reserve`,
1170 /// capacity will be greater than or equal to `self.len() + additional`.
1171 /// Does nothing if capacity is already sufficient.
1172 ///
1173 /// # Panics
1174 ///
1175 /// Panics if the new capacity overflows [`usize`].
1176 ///
1177 /// # Examples
1178 ///
1179 /// Basic usage:
1180 ///
1181 /// ```
1182 /// let mut s = String::new();
1183 ///
1184 /// s.reserve(10);
1185 ///
1186 /// assert!(s.capacity() >= 10);
1187 /// ```
1188 ///
1189 /// This might not actually increase the capacity:
1190 ///
1191 /// ```
1192 /// let mut s = String::with_capacity(10);
1193 /// s.push('a');
1194 /// s.push('b');
1195 ///
1196 /// // s now has a length of 2 and a capacity of at least 10
1197 /// let capacity = s.capacity();
1198 /// assert_eq!(2, s.len());
1199 /// assert!(capacity >= 10);
1200 ///
1201 /// // Since we already have at least an extra 8 capacity, calling this...
1202 /// s.reserve(8);
1203 ///
1204 /// // ... doesn't actually increase.
1205 /// assert_eq!(capacity, s.capacity());
1206 /// ```
1207 #[cfg(not(no_global_oom_handling))]
1208 #[inline]
1209 #[stable(feature = "rust1", since = "1.0.0")]
1210 pub fn reserve(&mut self, additional: usize) {
1211 self.vec.reserve(additional)
1212 }
1213
1214 /// Reserves the minimum capacity for at least `additional` bytes more than
1215 /// the current length. Unlike [`reserve`], this will not
1216 /// deliberately over-allocate to speculatively avoid frequent allocations.
1217 /// After calling `reserve_exact`, capacity will be greater than or equal to
1218 /// `self.len() + additional`. Does nothing if the capacity is already
1219 /// sufficient.
1220 ///
1221 /// [`reserve`]: String::reserve
1222 ///
1223 /// # Panics
1224 ///
1225 /// Panics if the new capacity overflows [`usize`].
1226 ///
1227 /// # Examples
1228 ///
1229 /// Basic usage:
1230 ///
1231 /// ```
1232 /// let mut s = String::new();
1233 ///
1234 /// s.reserve_exact(10);
1235 ///
1236 /// assert!(s.capacity() >= 10);
1237 /// ```
1238 ///
1239 /// This might not actually increase the capacity:
1240 ///
1241 /// ```
1242 /// let mut s = String::with_capacity(10);
1243 /// s.push('a');
1244 /// s.push('b');
1245 ///
1246 /// // s now has a length of 2 and a capacity of at least 10
1247 /// let capacity = s.capacity();
1248 /// assert_eq!(2, s.len());
1249 /// assert!(capacity >= 10);
1250 ///
1251 /// // Since we already have at least an extra 8 capacity, calling this...
1252 /// s.reserve_exact(8);
1253 ///
1254 /// // ... doesn't actually increase.
1255 /// assert_eq!(capacity, s.capacity());
1256 /// ```
1257 #[cfg(not(no_global_oom_handling))]
1258 #[inline]
1259 #[stable(feature = "rust1", since = "1.0.0")]
1260 pub fn reserve_exact(&mut self, additional: usize) {
1261 self.vec.reserve_exact(additional)
1262 }
1263
1264 /// Tries to reserve capacity for at least `additional` bytes more than the
1265 /// current length. The allocator may reserve more space to speculatively
1266 /// avoid frequent allocations. After calling `try_reserve`, capacity will be
1267 /// greater than or equal to `self.len() + additional` if it returns
1268 /// `Ok(())`. Does nothing if capacity is already sufficient. This method
1269 /// preserves the contents even if an error occurs.
1270 ///
1271 /// # Errors
1272 ///
1273 /// If the capacity overflows, or the allocator reports a failure, then an error
1274 /// is returned.
1275 ///
1276 /// # Examples
1277 ///
1278 /// ```
1279 /// use std::collections::TryReserveError;
1280 ///
1281 /// fn process_data(data: &str) -> Result<String, TryReserveError> {
1282 /// let mut output = String::new();
1283 ///
1284 /// // Pre-reserve the memory, exiting if we can't
1285 /// output.try_reserve(data.len())?;
1286 ///
1287 /// // Now we know this can't OOM in the middle of our complex work
1288 /// output.push_str(data);
1289 ///
1290 /// Ok(output)
1291 /// }
1292 /// # process_data("rust").expect("why is the test harness OOMing on 4 bytes?");
1293 /// ```
1294 #[stable(feature = "try_reserve", since = "1.57.0")]
1295 pub fn try_reserve(&mut self, additional: usize) -> Result<(), TryReserveError> {
1296 self.vec.try_reserve(additional)
1297 }
1298
1299 /// Tries to reserve the minimum capacity for at least `additional` bytes
1300 /// more than the current length. Unlike [`try_reserve`], this will not
1301 /// deliberately over-allocate to speculatively avoid frequent allocations.
1302 /// After calling `try_reserve_exact`, capacity will be greater than or
1303 /// equal to `self.len() + additional` if it returns `Ok(())`.
1304 /// Does nothing if the capacity is already sufficient.
1305 ///
1306 /// Note that the allocator may give the collection more space than it
1307 /// requests. Therefore, capacity can not be relied upon to be precisely
1308 /// minimal. Prefer [`try_reserve`] if future insertions are expected.
1309 ///
1310 /// [`try_reserve`]: String::try_reserve
1311 ///
1312 /// # Errors
1313 ///
1314 /// If the capacity overflows, or the allocator reports a failure, then an error
1315 /// is returned.
1316 ///
1317 /// # Examples
1318 ///
1319 /// ```
1320 /// use std::collections::TryReserveError;
1321 ///
1322 /// fn process_data(data: &str) -> Result<String, TryReserveError> {
1323 /// let mut output = String::new();
1324 ///
1325 /// // Pre-reserve the memory, exiting if we can't
1326 /// output.try_reserve_exact(data.len())?;
1327 ///
1328 /// // Now we know this can't OOM in the middle of our complex work
1329 /// output.push_str(data);
1330 ///
1331 /// Ok(output)
1332 /// }
1333 /// # process_data("rust").expect("why is the test harness OOMing on 4 bytes?");
1334 /// ```
1335 #[stable(feature = "try_reserve", since = "1.57.0")]
1336 pub fn try_reserve_exact(&mut self, additional: usize) -> Result<(), TryReserveError> {
1337 self.vec.try_reserve_exact(additional)
1338 }
1339
1340 /// Shrinks the capacity of this `String` to match its length.
1341 ///
1342 /// # Examples
1343 ///
1344 /// ```
1345 /// let mut s = String::from("foo");
1346 ///
1347 /// s.reserve(100);
1348 /// assert!(s.capacity() >= 100);
1349 ///
1350 /// s.shrink_to_fit();
1351 /// assert_eq!(3, s.capacity());
1352 /// ```
1353 #[cfg(not(no_global_oom_handling))]
1354 #[inline]
1355 #[stable(feature = "rust1", since = "1.0.0")]
1356 pub fn shrink_to_fit(&mut self) {
1357 self.vec.shrink_to_fit()
1358 }
1359
1360 /// Shrinks the capacity of this `String` with a lower bound.
1361 ///
1362 /// The capacity will remain at least as large as both the length
1363 /// and the supplied value.
1364 ///
1365 /// If the current capacity is less than the lower limit, this is a no-op.
1366 ///
1367 /// # Examples
1368 ///
1369 /// ```
1370 /// let mut s = String::from("foo");
1371 ///
1372 /// s.reserve(100);
1373 /// assert!(s.capacity() >= 100);
1374 ///
1375 /// s.shrink_to(10);
1376 /// assert!(s.capacity() >= 10);
1377 /// s.shrink_to(0);
1378 /// assert!(s.capacity() >= 3);
1379 /// ```
1380 #[cfg(not(no_global_oom_handling))]
1381 #[inline]
1382 #[stable(feature = "shrink_to", since = "1.56.0")]
1383 pub fn shrink_to(&mut self, min_capacity: usize) {
1384 self.vec.shrink_to(min_capacity)
1385 }
1386
1387 /// Appends the given [`char`] to the end of this `String`.
1388 ///
1389 /// # Examples
1390 ///
1391 /// ```
1392 /// let mut s = String::from("abc");
1393 ///
1394 /// s.push('1');
1395 /// s.push('2');
1396 /// s.push('3');
1397 ///
1398 /// assert_eq!("abc123", s);
1399 /// ```
1400 #[cfg(not(no_global_oom_handling))]
1401 #[inline]
1402 #[stable(feature = "rust1", since = "1.0.0")]
1403 pub fn push(&mut self, ch: char) {
1404 match ch.len_utf8() {
1405 1 => self.vec.push(ch as u8),
1406 _ => {
1407 self.vec.extend_from_slice(ch.encode_utf8(&mut [0; char::MAX_LEN_UTF8]).as_bytes())
1408 }
1409 }
1410 }
1411
1412 /// Returns a byte slice of this `String`'s contents.
1413 ///
1414 /// The inverse of this method is [`from_utf8`].
1415 ///
1416 /// [`from_utf8`]: String::from_utf8
1417 ///
1418 /// # Examples
1419 ///
1420 /// ```
1421 /// let s = String::from("hello");
1422 ///
1423 /// assert_eq!(&[104, 101, 108, 108, 111], s.as_bytes());
1424 /// ```
1425 #[inline]
1426 #[must_use]
1427 #[stable(feature = "rust1", since = "1.0.0")]
1428 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1429 pub const fn as_bytes(&self) -> &[u8] {
1430 self.vec.as_slice()
1431 }
1432
1433 /// Shortens this `String` to the specified length.
1434 ///
1435 /// If `new_len` is greater than or equal to the string's current length, this has no
1436 /// effect.
1437 ///
1438 /// Note that this method has no effect on the allocated capacity
1439 /// of the string
1440 ///
1441 /// # Panics
1442 ///
1443 /// Panics if `new_len` does not lie on a [`char`] boundary.
1444 ///
1445 /// # Examples
1446 ///
1447 /// ```
1448 /// let mut s = String::from("hello");
1449 ///
1450 /// s.truncate(2);
1451 ///
1452 /// assert_eq!("he", s);
1453 /// ```
1454 #[inline]
1455 #[stable(feature = "rust1", since = "1.0.0")]
1456 pub fn truncate(&mut self, new_len: usize) {
1457 if new_len <= self.len() {
1458 assert!(self.is_char_boundary(new_len));
1459 self.vec.truncate(new_len)
1460 }
1461 }
1462
1463 /// Removes the last character from the string buffer and returns it.
1464 ///
1465 /// Returns [`None`] if this `String` is empty.
1466 ///
1467 /// # Examples
1468 ///
1469 /// ```
1470 /// let mut s = String::from("abč");
1471 ///
1472 /// assert_eq!(s.pop(), Some('č'));
1473 /// assert_eq!(s.pop(), Some('b'));
1474 /// assert_eq!(s.pop(), Some('a'));
1475 ///
1476 /// assert_eq!(s.pop(), None);
1477 /// ```
1478 #[inline]
1479 #[stable(feature = "rust1", since = "1.0.0")]
1480 pub fn pop(&mut self) -> Option<char> {
1481 let ch = self.chars().rev().next()?;
1482 let newlen = self.len() - ch.len_utf8();
1483 unsafe {
1484 self.vec.set_len(newlen);
1485 }
1486 Some(ch)
1487 }
1488
1489 /// Removes a [`char`] from this `String` at a byte position and returns it.
1490 ///
1491 /// This is an *O*(*n*) operation, as it requires copying every element in the
1492 /// buffer.
1493 ///
1494 /// # Panics
1495 ///
1496 /// Panics if `idx` is larger than or equal to the `String`'s length,
1497 /// or if it does not lie on a [`char`] boundary.
1498 ///
1499 /// # Examples
1500 ///
1501 /// ```
1502 /// let mut s = String::from("abç");
1503 ///
1504 /// assert_eq!(s.remove(0), 'a');
1505 /// assert_eq!(s.remove(1), 'Γ§');
1506 /// assert_eq!(s.remove(0), 'b');
1507 /// ```
1508 #[inline]
1509 #[stable(feature = "rust1", since = "1.0.0")]
1510 #[rustc_confusables("delete", "take")]
1511 pub fn remove(&mut self, idx: usize) -> char {
1512 let ch = match self[idx..].chars().next() {
1513 Some(ch) => ch,
1514 None => panic!("cannot remove a char from the end of a string"),
1515 };
1516
1517 let next = idx + ch.len_utf8();
1518 let len = self.len();
1519 unsafe {
1520 ptr::copy(self.vec.as_ptr().add(next), self.vec.as_mut_ptr().add(idx), len - next);
1521 self.vec.set_len(len - (next - idx));
1522 }
1523 ch
1524 }
1525
1526 /// Remove all matches of pattern `pat` in the `String`.
1527 ///
1528 /// # Examples
1529 ///
1530 /// ```
1531 /// #![feature(string_remove_matches)]
1532 /// let mut s = String::from("Trees are not green, the sky is not blue.");
1533 /// s.remove_matches("not ");
1534 /// assert_eq!("Trees are green, the sky is blue.", s);
1535 /// ```
1536 ///
1537 /// Matches will be detected and removed iteratively, so in cases where
1538 /// patterns overlap, only the first pattern will be removed:
1539 ///
1540 /// ```
1541 /// #![feature(string_remove_matches)]
1542 /// let mut s = String::from("banana");
1543 /// s.remove_matches("ana");
1544 /// assert_eq!("bna", s);
1545 /// ```
1546 #[cfg(not(no_global_oom_handling))]
1547 #[unstable(feature = "string_remove_matches", reason = "new API", issue = "72826")]
1548 pub fn remove_matches<P: Pattern>(&mut self, pat: P) {
1549 use core::str::pattern::Searcher;
1550
1551 let rejections = {
1552 let mut searcher = pat.into_searcher(self);
1553 // Per Searcher::next:
1554 //
1555 // A Match result needs to contain the whole matched pattern,
1556 // however Reject results may be split up into arbitrary many
1557 // adjacent fragments. Both ranges may have zero length.
1558 //
1559 // In practice the implementation of Searcher::next_match tends to
1560 // be more efficient, so we use it here and do some work to invert
1561 // matches into rejections since that's what we want to copy below.
1562 let mut front = 0;
1563 let rejections: Vec<_> = from_fn(|| {
1564 let (start, end) = searcher.next_match()?;
1565 let prev_front = front;
1566 front = end;
1567 Some((prev_front, start))
1568 })
1569 .collect();
1570 rejections.into_iter().chain(core::iter::once((front, self.len())))
1571 };
1572
1573 let mut len = 0;
1574 let ptr = self.vec.as_mut_ptr();
1575
1576 for (start, end) in rejections {
1577 let count = end - start;
1578 if start != len {
1579 // SAFETY: per Searcher::next:
1580 //
1581 // The stream of Match and Reject values up to a Done will
1582 // contain index ranges that are adjacent, non-overlapping,
1583 // covering the whole haystack, and laying on utf8
1584 // boundaries.
1585 unsafe {
1586 ptr::copy(ptr.add(start), ptr.add(len), count);
1587 }
1588 }
1589 len += count;
1590 }
1591
1592 unsafe {
1593 self.vec.set_len(len);
1594 }
1595 }
1596
1597 /// Retains only the characters specified by the predicate.
1598 ///
1599 /// In other words, remove all characters `c` such that `f(c)` returns `false`.
1600 /// This method operates in place, visiting each character exactly once in the
1601 /// original order, and preserves the order of the retained characters.
1602 ///
1603 /// # Examples
1604 ///
1605 /// ```
1606 /// let mut s = String::from("f_o_ob_ar");
1607 ///
1608 /// s.retain(|c| c != '_');
1609 ///
1610 /// assert_eq!(s, "foobar");
1611 /// ```
1612 ///
1613 /// Because the elements are visited exactly once in the original order,
1614 /// external state may be used to decide which elements to keep.
1615 ///
1616 /// ```
1617 /// let mut s = String::from("abcde");
1618 /// let keep = [false, true, true, false, true];
1619 /// let mut iter = keep.iter();
1620 /// s.retain(|_| *iter.next().unwrap());
1621 /// assert_eq!(s, "bce");
1622 /// ```
1623 #[inline]
1624 #[stable(feature = "string_retain", since = "1.26.0")]
1625 pub fn retain<F>(&mut self, mut f: F)
1626 where
1627 F: FnMut(char) -> bool,
1628 {
1629 struct SetLenOnDrop<'a> {
1630 s: &'a mut String,
1631 idx: usize,
1632 del_bytes: usize,
1633 }
1634
1635 impl<'a> Drop for SetLenOnDrop<'a> {
1636 fn drop(&mut self) {
1637 let new_len = self.idx - self.del_bytes;
1638 debug_assert!(new_len <= self.s.len());
1639 unsafe { self.s.vec.set_len(new_len) };
1640 }
1641 }
1642
1643 let len = self.len();
1644 let mut guard = SetLenOnDrop { s: self, idx: 0, del_bytes: 0 };
1645
1646 while guard.idx < len {
1647 let ch =
1648 // SAFETY: `guard.idx` is positive-or-zero and less that len so the `get_unchecked`
1649 // is in bound. `self` is valid UTF-8 like string and the returned slice starts at
1650 // a unicode code point so the `Chars` always return one character.
1651 unsafe { guard.s.get_unchecked(guard.idx..len).chars().next().unwrap_unchecked() };
1652 let ch_len = ch.len_utf8();
1653
1654 if !f(ch) {
1655 guard.del_bytes += ch_len;
1656 } else if guard.del_bytes > 0 {
1657 // SAFETY: `guard.idx` is in bound and `guard.del_bytes` represent the number of
1658 // bytes that are erased from the string so the resulting `guard.idx -
1659 // guard.del_bytes` always represent a valid unicode code point.
1660 //
1661 // `guard.del_bytes` >= `ch.len_utf8()`, so taking a slice with `ch.len_utf8()` len
1662 // is safe.
1663 ch.encode_utf8(unsafe {
1664 crate::slice::from_raw_parts_mut(
1665 guard.s.as_mut_ptr().add(guard.idx - guard.del_bytes),
1666 ch.len_utf8(),
1667 )
1668 });
1669 }
1670
1671 // Point idx to the next char
1672 guard.idx += ch_len;
1673 }
1674
1675 drop(guard);
1676 }
1677
1678 /// Inserts a character into this `String` at a byte position.
1679 ///
1680 /// This is an *O*(*n*) operation as it requires copying every element in the
1681 /// buffer.
1682 ///
1683 /// # Panics
1684 ///
1685 /// Panics if `idx` is larger than the `String`'s length, or if it does not
1686 /// lie on a [`char`] boundary.
1687 ///
1688 /// # Examples
1689 ///
1690 /// ```
1691 /// let mut s = String::with_capacity(3);
1692 ///
1693 /// s.insert(0, 'f');
1694 /// s.insert(1, 'o');
1695 /// s.insert(2, 'o');
1696 ///
1697 /// assert_eq!("foo", s);
1698 /// ```
1699 #[cfg(not(no_global_oom_handling))]
1700 #[inline]
1701 #[stable(feature = "rust1", since = "1.0.0")]
1702 #[rustc_confusables("set")]
1703 pub fn insert(&mut self, idx: usize, ch: char) {
1704 assert!(self.is_char_boundary(idx));
1705 let mut bits = [0; char::MAX_LEN_UTF8];
1706 let bits = ch.encode_utf8(&mut bits).as_bytes();
1707
1708 unsafe {
1709 self.insert_bytes(idx, bits);
1710 }
1711 }
1712
1713 #[cfg(not(no_global_oom_handling))]
1714 unsafe fn insert_bytes(&mut self, idx: usize, bytes: &[u8]) {
1715 let len = self.len();
1716 let amt = bytes.len();
1717 self.vec.reserve(amt);
1718
1719 unsafe {
1720 ptr::copy(self.vec.as_ptr().add(idx), self.vec.as_mut_ptr().add(idx + amt), len - idx);
1721 ptr::copy_nonoverlapping(bytes.as_ptr(), self.vec.as_mut_ptr().add(idx), amt);
1722 self.vec.set_len(len + amt);
1723 }
1724 }
1725
1726 /// Inserts a string slice into this `String` at a byte position.
1727 ///
1728 /// This is an *O*(*n*) operation as it requires copying every element in the
1729 /// buffer.
1730 ///
1731 /// # Panics
1732 ///
1733 /// Panics if `idx` is larger than the `String`'s length, or if it does not
1734 /// lie on a [`char`] boundary.
1735 ///
1736 /// # Examples
1737 ///
1738 /// ```
1739 /// let mut s = String::from("bar");
1740 ///
1741 /// s.insert_str(0, "foo");
1742 ///
1743 /// assert_eq!("foobar", s);
1744 /// ```
1745 #[cfg(not(no_global_oom_handling))]
1746 #[inline]
1747 #[stable(feature = "insert_str", since = "1.16.0")]
1748 #[rustc_diagnostic_item = "string_insert_str"]
1749 pub fn insert_str(&mut self, idx: usize, string: &str) {
1750 assert!(self.is_char_boundary(idx));
1751
1752 unsafe {
1753 self.insert_bytes(idx, string.as_bytes());
1754 }
1755 }
1756
1757 /// Returns a mutable reference to the contents of this `String`.
1758 ///
1759 /// # Safety
1760 ///
1761 /// This function is unsafe because the returned `&mut Vec` allows writing
1762 /// bytes which are not valid UTF-8. If this constraint is violated, using
1763 /// the original `String` after dropping the `&mut Vec` may violate memory
1764 /// safety, as the rest of the standard library assumes that `String`s are
1765 /// valid UTF-8.
1766 ///
1767 /// # Examples
1768 ///
1769 /// ```
1770 /// let mut s = String::from("hello");
1771 ///
1772 /// unsafe {
1773 /// let vec = s.as_mut_vec();
1774 /// assert_eq!(&[104, 101, 108, 108, 111][..], &vec[..]);
1775 ///
1776 /// vec.reverse();
1777 /// }
1778 /// assert_eq!(s, "olleh");
1779 /// ```
1780 #[inline]
1781 #[stable(feature = "rust1", since = "1.0.0")]
1782 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1783 pub const unsafe fn as_mut_vec(&mut self) -> &mut Vec<u8> {
1784 &mut self.vec
1785 }
1786
1787 /// Returns the length of this `String`, in bytes, not [`char`]s or
1788 /// graphemes. In other words, it might not be what a human considers the
1789 /// length of the string.
1790 ///
1791 /// # Examples
1792 ///
1793 /// ```
1794 /// let a = String::from("foo");
1795 /// assert_eq!(a.len(), 3);
1796 ///
1797 /// let fancy_f = String::from("Ζ’oo");
1798 /// assert_eq!(fancy_f.len(), 4);
1799 /// assert_eq!(fancy_f.chars().count(), 3);
1800 /// ```
1801 #[inline]
1802 #[must_use]
1803 #[stable(feature = "rust1", since = "1.0.0")]
1804 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1805 #[rustc_confusables("length", "size")]
1806 pub const fn len(&self) -> usize {
1807 self.vec.len()
1808 }
1809
1810 /// Returns `true` if this `String` has a length of zero, and `false` otherwise.
1811 ///
1812 /// # Examples
1813 ///
1814 /// ```
1815 /// let mut v = String::new();
1816 /// assert!(v.is_empty());
1817 ///
1818 /// v.push('a');
1819 /// assert!(!v.is_empty());
1820 /// ```
1821 #[inline]
1822 #[must_use]
1823 #[stable(feature = "rust1", since = "1.0.0")]
1824 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1825 pub const fn is_empty(&self) -> bool {
1826 self.len() == 0
1827 }
1828
1829 /// Splits the string into two at the given byte index.
1830 ///
1831 /// Returns a newly allocated `String`. `self` contains bytes `[0, at)`, and
1832 /// the returned `String` contains bytes `[at, len)`. `at` must be on the
1833 /// boundary of a UTF-8 code point.
1834 ///
1835 /// Note that the capacity of `self` does not change.
1836 ///
1837 /// # Panics
1838 ///
1839 /// Panics if `at` is not on a `UTF-8` code point boundary, or if it is beyond the last
1840 /// code point of the string.
1841 ///
1842 /// # Examples
1843 ///
1844 /// ```
1845 /// # fn main() {
1846 /// let mut hello = String::from("Hello, World!");
1847 /// let world = hello.split_off(7);
1848 /// assert_eq!(hello, "Hello, ");
1849 /// assert_eq!(world, "World!");
1850 /// # }
1851 /// ```
1852 #[cfg(not(no_global_oom_handling))]
1853 #[inline]
1854 #[stable(feature = "string_split_off", since = "1.16.0")]
1855 #[must_use = "use `.truncate()` if you don't need the other half"]
1856 pub fn split_off(&mut self, at: usize) -> String {
1857 assert!(self.is_char_boundary(at));
1858 let other = self.vec.split_off(at);
1859 unsafe { String::from_utf8_unchecked(other) }
1860 }
1861
1862 /// Truncates this `String`, removing all contents.
1863 ///
1864 /// While this means the `String` will have a length of zero, it does not
1865 /// touch its capacity.
1866 ///
1867 /// # Examples
1868 ///
1869 /// ```
1870 /// let mut s = String::from("foo");
1871 ///
1872 /// s.clear();
1873 ///
1874 /// assert!(s.is_empty());
1875 /// assert_eq!(0, s.len());
1876 /// assert_eq!(3, s.capacity());
1877 /// ```
1878 #[inline]
1879 #[stable(feature = "rust1", since = "1.0.0")]
1880 pub fn clear(&mut self) {
1881 self.vec.clear()
1882 }
1883
1884 /// Removes the specified range from the string in bulk, returning all
1885 /// removed characters as an iterator.
1886 ///
1887 /// The returned iterator keeps a mutable borrow on the string to optimize
1888 /// its implementation.
1889 ///
1890 /// # Panics
1891 ///
1892 /// Panics if the starting point or end point do not lie on a [`char`]
1893 /// boundary, or if they're out of bounds.
1894 ///
1895 /// # Leaking
1896 ///
1897 /// If the returned iterator goes out of scope without being dropped (due to
1898 /// [`core::mem::forget`], for example), the string may still contain a copy
1899 /// of any drained characters, or may have lost characters arbitrarily,
1900 /// including characters outside the range.
1901 ///
1902 /// # Examples
1903 ///
1904 /// ```
1905 /// let mut s = String::from("Ξ± is alpha, Ξ² is beta");
1906 /// let beta_offset = s.find('Ξ²').unwrap_or(s.len());
1907 ///
1908 /// // Remove the range up until the Ξ² from the string
1909 /// let t: String = s.drain(..beta_offset).collect();
1910 /// assert_eq!(t, "Ξ± is alpha, ");
1911 /// assert_eq!(s, "Ξ² is beta");
1912 ///
1913 /// // A full range clears the string, like `clear()` does
1914 /// s.drain(..);
1915 /// assert_eq!(s, "");
1916 /// ```
1917 #[stable(feature = "drain", since = "1.6.0")]
1918 pub fn drain<R>(&mut self, range: R) -> Drain<'_>
1919 where
1920 R: RangeBounds<usize>,
1921 {
1922 // Memory safety
1923 //
1924 // The String version of Drain does not have the memory safety issues
1925 // of the vector version. The data is just plain bytes.
1926 // Because the range removal happens in Drop, if the Drain iterator is leaked,
1927 // the removal will not happen.
1928 let Range { start, end } = slice::range(range, ..self.len());
1929 assert!(self.is_char_boundary(start));
1930 assert!(self.is_char_boundary(end));
1931
1932 // Take out two simultaneous borrows. The &mut String won't be accessed
1933 // until iteration is over, in Drop.
1934 let self_ptr = self as *mut _;
1935 // SAFETY: `slice::range` and `is_char_boundary` do the appropriate bounds checks.
1936 let chars_iter = unsafe { self.get_unchecked(start..end) }.chars();
1937
1938 Drain { start, end, iter: chars_iter, string: self_ptr }
1939 }
1940
1941 /// Converts a `String` into an iterator over the [`char`]s of the string.
1942 ///
1943 /// As a string consists of valid UTF-8, we can iterate through a string
1944 /// by [`char`]. This method returns such an iterator.
1945 ///
1946 /// It's important to remember that [`char`] represents a Unicode Scalar
1947 /// Value, and might not match your idea of what a 'character' is. Iteration
1948 /// over grapheme clusters may be what you actually want. That functionality
1949 /// is not provided by Rust's standard library, check crates.io instead.
1950 ///
1951 /// # Examples
1952 ///
1953 /// Basic usage:
1954 ///
1955 /// ```
1956 /// #![feature(string_into_chars)]
1957 ///
1958 /// let word = String::from("goodbye");
1959 ///
1960 /// let mut chars = word.into_chars();
1961 ///
1962 /// assert_eq!(Some('g'), chars.next());
1963 /// assert_eq!(Some('o'), chars.next());
1964 /// assert_eq!(Some('o'), chars.next());
1965 /// assert_eq!(Some('d'), chars.next());
1966 /// assert_eq!(Some('b'), chars.next());
1967 /// assert_eq!(Some('y'), chars.next());
1968 /// assert_eq!(Some('e'), chars.next());
1969 ///
1970 /// assert_eq!(None, chars.next());
1971 /// ```
1972 ///
1973 /// Remember, [`char`]s might not match your intuition about characters:
1974 ///
1975 /// ```
1976 /// #![feature(string_into_chars)]
1977 ///
1978 /// let y = String::from("y̆");
1979 ///
1980 /// let mut chars = y.into_chars();
1981 ///
1982 /// assert_eq!(Some('y'), chars.next()); // not 'y̆'
1983 /// assert_eq!(Some('\u{0306}'), chars.next());
1984 ///
1985 /// assert_eq!(None, chars.next());
1986 /// ```
1987 ///
1988 /// [`char`]: prim@char
1989 #[inline]
1990 #[must_use = "`self` will be dropped if the result is not used"]
1991 #[unstable(feature = "string_into_chars", issue = "133125")]
1992 pub fn into_chars(self) -> IntoChars {
1993 IntoChars { bytes: self.into_bytes().into_iter() }
1994 }
1995
1996 /// Removes the specified range in the string,
1997 /// and replaces it with the given string.
1998 /// The given string doesn't need to be the same length as the range.
1999 ///
2000 /// # Panics
2001 ///
2002 /// Panics if the starting point or end point do not lie on a [`char`]
2003 /// boundary, or if they're out of bounds.
2004 ///
2005 /// # Examples
2006 ///
2007 /// ```
2008 /// let mut s = String::from("Ξ± is alpha, Ξ² is beta");
2009 /// let beta_offset = s.find('Ξ²').unwrap_or(s.len());
2010 ///
2011 /// // Replace the range up until the Ξ² from the string
2012 /// s.replace_range(..beta_offset, "Ξ‘ is capital alpha; ");
2013 /// assert_eq!(s, "Ξ‘ is capital alpha; Ξ² is beta");
2014 /// ```
2015 #[cfg(not(no_global_oom_handling))]
2016 #[stable(feature = "splice", since = "1.27.0")]
2017 pub fn replace_range<R>(&mut self, range: R, replace_with: &str)
2018 where
2019 R: RangeBounds<usize>,
2020 {
2021 // Memory safety
2022 //
2023 // Replace_range does not have the memory safety issues of a vector Splice.
2024 // of the vector version. The data is just plain bytes.
2025
2026 // WARNING: Inlining this variable would be unsound (#81138)
2027 let start = range.start_bound();
2028 match start {
2029 Included(&n) => assert!(self.is_char_boundary(n)),
2030 Excluded(&n) => assert!(self.is_char_boundary(n + 1)),
2031 Unbounded => {}
2032 };
2033 // WARNING: Inlining this variable would be unsound (#81138)
2034 let end = range.end_bound();
2035 match end {
2036 Included(&n) => assert!(self.is_char_boundary(n + 1)),
2037 Excluded(&n) => assert!(self.is_char_boundary(n)),
2038 Unbounded => {}
2039 };
2040
2041 // Using `range` again would be unsound (#81138)
2042 // We assume the bounds reported by `range` remain the same, but
2043 // an adversarial implementation could change between calls
2044 unsafe { self.as_mut_vec() }.splice((start, end), replace_with.bytes());
2045 }
2046
2047 /// Converts this `String` into a <code>[Box]<[str]></code>.
2048 ///
2049 /// Before doing the conversion, this method discards excess capacity like [`shrink_to_fit`].
2050 /// Note that this call may reallocate and copy the bytes of the string.
2051 ///
2052 /// [`shrink_to_fit`]: String::shrink_to_fit
2053 /// [str]: prim@str "str"
2054 ///
2055 /// # Examples
2056 ///
2057 /// ```
2058 /// let s = String::from("hello");
2059 ///
2060 /// let b = s.into_boxed_str();
2061 /// ```
2062 #[cfg(not(no_global_oom_handling))]
2063 #[stable(feature = "box_str", since = "1.4.0")]
2064 #[must_use = "`self` will be dropped if the result is not used"]
2065 #[inline]
2066 pub fn into_boxed_str(self) -> Box<str> {
2067 let slice = self.vec.into_boxed_slice();
2068 unsafe { from_boxed_utf8_unchecked(slice) }
2069 }
2070
2071 /// Consumes and leaks the `String`, returning a mutable reference to the contents,
2072 /// `&'a mut str`.
2073 ///
2074 /// The caller has free choice over the returned lifetime, including `'static`. Indeed,
2075 /// this function is ideally used for data that lives for the remainder of the program's life,
2076 /// as dropping the returned reference will cause a memory leak.
2077 ///
2078 /// It does not reallocate or shrink the `String`, so the leaked allocation may include unused
2079 /// capacity that is not part of the returned slice. If you want to discard excess capacity,
2080 /// call [`into_boxed_str`], and then [`Box::leak`] instead. However, keep in mind that
2081 /// trimming the capacity may result in a reallocation and copy.
2082 ///
2083 /// [`into_boxed_str`]: Self::into_boxed_str
2084 ///
2085 /// # Examples
2086 ///
2087 /// ```
2088 /// let x = String::from("bucket");
2089 /// let static_ref: &'static mut str = x.leak();
2090 /// assert_eq!(static_ref, "bucket");
2091 /// # // FIXME(https://github.com/rust-lang/miri/issues/3670):
2092 /// # // use -Zmiri-disable-leak-check instead of unleaking in tests meant to leak.
2093 /// # drop(unsafe { Box::from_raw(static_ref) });
2094 /// ```
2095 #[stable(feature = "string_leak", since = "1.72.0")]
2096 #[inline]
2097 pub fn leak<'a>(self) -> &'a mut str {
2098 let slice = self.vec.leak();
2099 unsafe { from_utf8_unchecked_mut(slice) }
2100 }
2101}
2102
2103impl FromUtf8Error {
2104 /// Returns a slice of [`u8`]s bytes that were attempted to convert to a `String`.
2105 ///
2106 /// # Examples
2107 ///
2108 /// ```
2109 /// // some invalid bytes, in a vector
2110 /// let bytes = vec![0, 159];
2111 ///
2112 /// let value = String::from_utf8(bytes);
2113 ///
2114 /// assert_eq!(&[0, 159], value.unwrap_err().as_bytes());
2115 /// ```
2116 #[must_use]
2117 #[stable(feature = "from_utf8_error_as_bytes", since = "1.26.0")]
2118 pub fn as_bytes(&self) -> &[u8] {
2119 &self.bytes[..]
2120 }
2121
2122 /// Converts the bytes into a `String` lossily, substituting invalid UTF-8
2123 /// sequences with replacement characters.
2124 ///
2125 /// See [`String::from_utf8_lossy`] for more details on replacement of
2126 /// invalid sequences, and [`String::from_utf8_lossy_owned`] for the
2127 /// `String` function which corresponds to this function.
2128 ///
2129 /// # Examples
2130 ///
2131 /// ```
2132 /// #![feature(string_from_utf8_lossy_owned)]
2133 /// // some invalid bytes
2134 /// let input: Vec<u8> = b"Hello \xF0\x90\x80World".into();
2135 /// let output = String::from_utf8(input).unwrap_or_else(|e| e.into_utf8_lossy());
2136 ///
2137 /// assert_eq!(String::from("Hello οΏ½World"), output);
2138 /// ```
2139 #[must_use]
2140 #[cfg(not(no_global_oom_handling))]
2141 #[unstable(feature = "string_from_utf8_lossy_owned", issue = "129436")]
2142 pub fn into_utf8_lossy(self) -> String {
2143 const REPLACEMENT: &str = "\u{FFFD}";
2144
2145 let mut res = {
2146 let mut v = Vec::with_capacity(self.bytes.len());
2147
2148 // `Utf8Error::valid_up_to` returns the maximum index of validated
2149 // UTF-8 bytes. Copy the valid bytes into the output buffer.
2150 v.extend_from_slice(&self.bytes[..self.error.valid_up_to()]);
2151
2152 // SAFETY: This is safe because the only bytes present in the buffer
2153 // were validated as UTF-8 by the call to `String::from_utf8` which
2154 // produced this `FromUtf8Error`.
2155 unsafe { String::from_utf8_unchecked(v) }
2156 };
2157
2158 let iter = self.bytes[self.error.valid_up_to()..].utf8_chunks();
2159
2160 for chunk in iter {
2161 res.push_str(chunk.valid());
2162 if !chunk.invalid().is_empty() {
2163 res.push_str(REPLACEMENT);
2164 }
2165 }
2166
2167 res
2168 }
2169
2170 /// Returns the bytes that were attempted to convert to a `String`.
2171 ///
2172 /// This method is carefully constructed to avoid allocation. It will
2173 /// consume the error, moving out the bytes, so that a copy of the bytes
2174 /// does not need to be made.
2175 ///
2176 /// # Examples
2177 ///
2178 /// ```
2179 /// // some invalid bytes, in a vector
2180 /// let bytes = vec![0, 159];
2181 ///
2182 /// let value = String::from_utf8(bytes);
2183 ///
2184 /// assert_eq!(vec![0, 159], value.unwrap_err().into_bytes());
2185 /// ```
2186 #[must_use = "`self` will be dropped if the result is not used"]
2187 #[stable(feature = "rust1", since = "1.0.0")]
2188 pub fn into_bytes(self) -> Vec<u8> {
2189 self.bytes
2190 }
2191
2192 /// Fetch a `Utf8Error` to get more details about the conversion failure.
2193 ///
2194 /// The [`Utf8Error`] type provided by [`std::str`] represents an error that may
2195 /// occur when converting a slice of [`u8`]s to a [`&str`]. In this sense, it's
2196 /// an analogue to `FromUtf8Error`. See its documentation for more details
2197 /// on using it.
2198 ///
2199 /// [`std::str`]: core::str "std::str"
2200 /// [`&str`]: prim@str "&str"
2201 ///
2202 /// # Examples
2203 ///
2204 /// ```
2205 /// // some invalid bytes, in a vector
2206 /// let bytes = vec![0, 159];
2207 ///
2208 /// let error = String::from_utf8(bytes).unwrap_err().utf8_error();
2209 ///
2210 /// // the first byte is invalid here
2211 /// assert_eq!(1, error.valid_up_to());
2212 /// ```
2213 #[must_use]
2214 #[stable(feature = "rust1", since = "1.0.0")]
2215 pub fn utf8_error(&self) -> Utf8Error {
2216 self.error
2217 }
2218}
2219
2220#[stable(feature = "rust1", since = "1.0.0")]
2221impl fmt::Display for FromUtf8Error {
2222 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2223 fmt::Display::fmt(&self.error, f)
2224 }
2225}
2226
2227#[stable(feature = "rust1", since = "1.0.0")]
2228impl fmt::Display for FromUtf16Error {
2229 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2230 fmt::Display::fmt(self:"invalid utf-16: lone surrogate found", f)
2231 }
2232}
2233
2234#[stable(feature = "rust1", since = "1.0.0")]
2235impl Error for FromUtf8Error {
2236 #[allow(deprecated)]
2237 fn description(&self) -> &str {
2238 "invalid utf-8"
2239 }
2240}
2241
2242#[stable(feature = "rust1", since = "1.0.0")]
2243impl Error for FromUtf16Error {
2244 #[allow(deprecated)]
2245 fn description(&self) -> &str {
2246 "invalid utf-16"
2247 }
2248}
2249
2250#[cfg(not(no_global_oom_handling))]
2251#[stable(feature = "rust1", since = "1.0.0")]
2252impl Clone for String {
2253 fn clone(&self) -> Self {
2254 String { vec: self.vec.clone() }
2255 }
2256
2257 /// Clones the contents of `source` into `self`.
2258 ///
2259 /// This method is preferred over simply assigning `source.clone()` to `self`,
2260 /// as it avoids reallocation if possible.
2261 fn clone_from(&mut self, source: &Self) {
2262 self.vec.clone_from(&source.vec);
2263 }
2264}
2265
2266#[cfg(not(no_global_oom_handling))]
2267#[stable(feature = "rust1", since = "1.0.0")]
2268impl FromIterator<char> for String {
2269 fn from_iter<I: IntoIterator<Item = char>>(iter: I) -> String {
2270 let mut buf: String = String::new();
2271 buf.extend(iter);
2272 buf
2273 }
2274}
2275
2276#[cfg(not(no_global_oom_handling))]
2277#[stable(feature = "string_from_iter_by_ref", since = "1.17.0")]
2278impl<'a> FromIterator<&'a char> for String {
2279 fn from_iter<I: IntoIterator<Item = &'a char>>(iter: I) -> String {
2280 let mut buf: String = String::new();
2281 buf.extend(iter);
2282 buf
2283 }
2284}
2285
2286#[cfg(not(no_global_oom_handling))]
2287#[stable(feature = "rust1", since = "1.0.0")]
2288impl<'a> FromIterator<&'a str> for String {
2289 fn from_iter<I: IntoIterator<Item = &'a str>>(iter: I) -> String {
2290 let mut buf: String = String::new();
2291 buf.extend(iter);
2292 buf
2293 }
2294}
2295
2296#[cfg(not(no_global_oom_handling))]
2297#[stable(feature = "extend_string", since = "1.4.0")]
2298impl FromIterator<String> for String {
2299 fn from_iter<I: IntoIterator<Item = String>>(iter: I) -> String {
2300 let mut iterator: ::IntoIter = iter.into_iter();
2301
2302 // Because we're iterating over `String`s, we can avoid at least
2303 // one allocation by getting the first string from the iterator
2304 // and appending to it all the subsequent strings.
2305 match iterator.next() {
2306 None => String::new(),
2307 Some(mut buf: String) => {
2308 buf.extend(iter:iterator);
2309 buf
2310 }
2311 }
2312 }
2313}
2314
2315#[cfg(not(no_global_oom_handling))]
2316#[stable(feature = "box_str2", since = "1.45.0")]
2317impl<A: Allocator> FromIterator<Box<str, A>> for String {
2318 fn from_iter<I: IntoIterator<Item = Box<str, A>>>(iter: I) -> String {
2319 let mut buf: String = String::new();
2320 buf.extend(iter);
2321 buf
2322 }
2323}
2324
2325#[cfg(not(no_global_oom_handling))]
2326#[stable(feature = "herd_cows", since = "1.19.0")]
2327impl<'a> FromIterator<Cow<'a, str>> for String {
2328 fn from_iter<I: IntoIterator<Item = Cow<'a, str>>>(iter: I) -> String {
2329 let mut iterator: ::IntoIter = iter.into_iter();
2330
2331 // Because we're iterating over CoWs, we can (potentially) avoid at least
2332 // one allocation by getting the first item and appending to it all the
2333 // subsequent items.
2334 match iterator.next() {
2335 None => String::new(),
2336 Some(cow: Cow<'a, str>) => {
2337 let mut buf: String = cow.into_owned();
2338 buf.extend(iter:iterator);
2339 buf
2340 }
2341 }
2342 }
2343}
2344
2345#[cfg(not(no_global_oom_handling))]
2346#[stable(feature = "rust1", since = "1.0.0")]
2347impl Extend<char> for String {
2348 fn extend<I: IntoIterator<Item = char>>(&mut self, iter: I) {
2349 let iterator: ::IntoIter = iter.into_iter();
2350 let (lower_bound: usize, _) = iterator.size_hint();
2351 self.reserve(additional:lower_bound);
2352 iterator.for_each(move |c: char| self.push(ch:c));
2353 }
2354
2355 #[inline]
2356 fn extend_one(&mut self, c: char) {
2357 self.push(ch:c);
2358 }
2359
2360 #[inline]
2361 fn extend_reserve(&mut self, additional: usize) {
2362 self.reserve(additional);
2363 }
2364}
2365
2366#[cfg(not(no_global_oom_handling))]
2367#[stable(feature = "extend_ref", since = "1.2.0")]
2368impl<'a> Extend<&'a char> for String {
2369 fn extend<I: IntoIterator<Item = &'a char>>(&mut self, iter: I) {
2370 self.extend(iter.into_iter().cloned());
2371 }
2372
2373 #[inline]
2374 fn extend_one(&mut self, &c: char: &'a char) {
2375 self.push(ch:c);
2376 }
2377
2378 #[inline]
2379 fn extend_reserve(&mut self, additional: usize) {
2380 self.reserve(additional);
2381 }
2382}
2383
2384#[cfg(not(no_global_oom_handling))]
2385#[stable(feature = "rust1", since = "1.0.0")]
2386impl<'a> Extend<&'a str> for String {
2387 fn extend<I: IntoIterator<Item = &'a str>>(&mut self, iter: I) {
2388 iter.into_iter().for_each(move |s: &'a str| self.push_str(string:s));
2389 }
2390
2391 #[inline]
2392 fn extend_one(&mut self, s: &'a str) {
2393 self.push_str(string:s);
2394 }
2395}
2396
2397#[cfg(not(no_global_oom_handling))]
2398#[stable(feature = "box_str2", since = "1.45.0")]
2399impl<A: Allocator> Extend<Box<str, A>> for String {
2400 fn extend<I: IntoIterator<Item = Box<str, A>>>(&mut self, iter: I) {
2401 iter.into_iter().for_each(move |s: Box| self.push_str(&s));
2402 }
2403}
2404
2405#[cfg(not(no_global_oom_handling))]
2406#[stable(feature = "extend_string", since = "1.4.0")]
2407impl Extend<String> for String {
2408 fn extend<I: IntoIterator<Item = String>>(&mut self, iter: I) {
2409 iter.into_iter().for_each(move |s: String| self.push_str(&s));
2410 }
2411
2412 #[inline]
2413 fn extend_one(&mut self, s: String) {
2414 self.push_str(&s);
2415 }
2416}
2417
2418#[cfg(not(no_global_oom_handling))]
2419#[stable(feature = "herd_cows", since = "1.19.0")]
2420impl<'a> Extend<Cow<'a, str>> for String {
2421 fn extend<I: IntoIterator<Item = Cow<'a, str>>>(&mut self, iter: I) {
2422 iter.into_iter().for_each(move |s: Cow<'a, str>| self.push_str(&s));
2423 }
2424
2425 #[inline]
2426 fn extend_one(&mut self, s: Cow<'a, str>) {
2427 self.push_str(&s);
2428 }
2429}
2430
2431#[cfg(not(no_global_oom_handling))]
2432#[unstable(feature = "ascii_char", issue = "110998")]
2433impl Extend<core::ascii::Char> for String {
2434 fn extend<I: IntoIterator<Item = core::ascii::Char>>(&mut self, iter: I) {
2435 self.vec.extend(iter.into_iter().map(|c: AsciiChar| c.to_u8()));
2436 }
2437
2438 #[inline]
2439 fn extend_one(&mut self, c: core::ascii::Char) {
2440 self.vec.push(c.to_u8());
2441 }
2442}
2443
2444#[cfg(not(no_global_oom_handling))]
2445#[unstable(feature = "ascii_char", issue = "110998")]
2446impl<'a> Extend<&'a core::ascii::Char> for String {
2447 fn extend<I: IntoIterator<Item = &'a core::ascii::Char>>(&mut self, iter: I) {
2448 self.extend(iter.into_iter().cloned());
2449 }
2450
2451 #[inline]
2452 fn extend_one(&mut self, c: &'a core::ascii::Char) {
2453 self.vec.push(c.to_u8());
2454 }
2455}
2456
2457/// A convenience impl that delegates to the impl for `&str`.
2458///
2459/// # Examples
2460///
2461/// ```
2462/// assert_eq!(String::from("Hello world").find("world"), Some(6));
2463/// ```
2464#[unstable(
2465 feature = "pattern",
2466 reason = "API not fully fleshed out and ready to be stabilized",
2467 issue = "27721"
2468)]
2469impl<'b> Pattern for &'b String {
2470 type Searcher<'a> = <&'b str as Pattern>::Searcher<'a>;
2471
2472 fn into_searcher(self, haystack: &str) -> <&'b str as Pattern>::Searcher<'_> {
2473 self[..].into_searcher(haystack)
2474 }
2475
2476 #[inline]
2477 fn is_contained_in(self, haystack: &str) -> bool {
2478 self[..].is_contained_in(haystack)
2479 }
2480
2481 #[inline]
2482 fn is_prefix_of(self, haystack: &str) -> bool {
2483 self[..].is_prefix_of(haystack)
2484 }
2485
2486 #[inline]
2487 fn strip_prefix_of(self, haystack: &str) -> Option<&str> {
2488 self[..].strip_prefix_of(haystack)
2489 }
2490
2491 #[inline]
2492 fn is_suffix_of<'a>(self, haystack: &'a str) -> bool
2493 where
2494 Self::Searcher<'a>: core::str::pattern::ReverseSearcher<'a>,
2495 {
2496 self[..].is_suffix_of(haystack)
2497 }
2498
2499 #[inline]
2500 fn strip_suffix_of<'a>(self, haystack: &'a str) -> Option<&'a str>
2501 where
2502 Self::Searcher<'a>: core::str::pattern::ReverseSearcher<'a>,
2503 {
2504 self[..].strip_suffix_of(haystack)
2505 }
2506
2507 #[inline]
2508 fn as_utf8_pattern(&self) -> Option<Utf8Pattern<'_>> {
2509 Some(Utf8Pattern::StringPattern(self.as_bytes()))
2510 }
2511}
2512
2513macro_rules! impl_eq {
2514 ($lhs:ty, $rhs: ty) => {
2515 #[stable(feature = "rust1", since = "1.0.0")]
2516 #[allow(unused_lifetimes)]
2517 impl<'a, 'b> PartialEq<$rhs> for $lhs {
2518 #[inline]
2519 fn eq(&self, other: &$rhs) -> bool {
2520 PartialEq::eq(&self[..], &other[..])
2521 }
2522 #[inline]
2523 fn ne(&self, other: &$rhs) -> bool {
2524 PartialEq::ne(&self[..], &other[..])
2525 }
2526 }
2527
2528 #[stable(feature = "rust1", since = "1.0.0")]
2529 #[allow(unused_lifetimes)]
2530 impl<'a, 'b> PartialEq<$lhs> for $rhs {
2531 #[inline]
2532 fn eq(&self, other: &$lhs) -> bool {
2533 PartialEq::eq(&self[..], &other[..])
2534 }
2535 #[inline]
2536 fn ne(&self, other: &$lhs) -> bool {
2537 PartialEq::ne(&self[..], &other[..])
2538 }
2539 }
2540 };
2541}
2542
2543impl_eq! { String, str }
2544impl_eq! { String, &'a str }
2545#[cfg(not(no_global_oom_handling))]
2546impl_eq! { Cow<'a, str>, str }
2547#[cfg(not(no_global_oom_handling))]
2548impl_eq! { Cow<'a, str>, &'b str }
2549#[cfg(not(no_global_oom_handling))]
2550impl_eq! { Cow<'a, str>, String }
2551
2552#[stable(feature = "rust1", since = "1.0.0")]
2553impl Default for String {
2554 /// Creates an empty `String`.
2555 #[inline]
2556 fn default() -> String {
2557 String::new()
2558 }
2559}
2560
2561#[stable(feature = "rust1", since = "1.0.0")]
2562impl fmt::Display for String {
2563 #[inline]
2564 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2565 fmt::Display::fmt(&**self, f)
2566 }
2567}
2568
2569#[stable(feature = "rust1", since = "1.0.0")]
2570impl fmt::Debug for String {
2571 #[inline]
2572 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2573 fmt::Debug::fmt(&**self, f)
2574 }
2575}
2576
2577#[stable(feature = "rust1", since = "1.0.0")]
2578impl hash::Hash for String {
2579 #[inline]
2580 fn hash<H: hash::Hasher>(&self, hasher: &mut H) {
2581 (**self).hash(state:hasher)
2582 }
2583}
2584
2585/// Implements the `+` operator for concatenating two strings.
2586///
2587/// This consumes the `String` on the left-hand side and re-uses its buffer (growing it if
2588/// necessary). This is done to avoid allocating a new `String` and copying the entire contents on
2589/// every operation, which would lead to *O*(*n*^2) running time when building an *n*-byte string by
2590/// repeated concatenation.
2591///
2592/// The string on the right-hand side is only borrowed; its contents are copied into the returned
2593/// `String`.
2594///
2595/// # Examples
2596///
2597/// Concatenating two `String`s takes the first by value and borrows the second:
2598///
2599/// ```
2600/// let a = String::from("hello");
2601/// let b = String::from(" world");
2602/// let c = a + &b;
2603/// // `a` is moved and can no longer be used here.
2604/// ```
2605///
2606/// If you want to keep using the first `String`, you can clone it and append to the clone instead:
2607///
2608/// ```
2609/// let a = String::from("hello");
2610/// let b = String::from(" world");
2611/// let c = a.clone() + &b;
2612/// // `a` is still valid here.
2613/// ```
2614///
2615/// Concatenating `&str` slices can be done by converting the first to a `String`:
2616///
2617/// ```
2618/// let a = "hello";
2619/// let b = " world";
2620/// let c = a.to_string() + b;
2621/// ```
2622#[cfg(not(no_global_oom_handling))]
2623#[stable(feature = "rust1", since = "1.0.0")]
2624impl Add<&str> for String {
2625 type Output = String;
2626
2627 #[inline]
2628 fn add(mut self, other: &str) -> String {
2629 self.push_str(string:other);
2630 self
2631 }
2632}
2633
2634/// Implements the `+=` operator for appending to a `String`.
2635///
2636/// This has the same behavior as the [`push_str`][String::push_str] method.
2637#[cfg(not(no_global_oom_handling))]
2638#[stable(feature = "stringaddassign", since = "1.12.0")]
2639impl AddAssign<&str> for String {
2640 #[inline]
2641 fn add_assign(&mut self, other: &str) {
2642 self.push_str(string:other);
2643 }
2644}
2645
2646#[stable(feature = "rust1", since = "1.0.0")]
2647impl<I> ops::Index<I> for String
2648where
2649 I: slice::SliceIndex<str>,
2650{
2651 type Output = I::Output;
2652
2653 #[inline]
2654 fn index(&self, index: I) -> &I::Output {
2655 index.index(self.as_str())
2656 }
2657}
2658
2659#[stable(feature = "rust1", since = "1.0.0")]
2660impl<I> ops::IndexMut<I> for String
2661where
2662 I: slice::SliceIndex<str>,
2663{
2664 #[inline]
2665 fn index_mut(&mut self, index: I) -> &mut I::Output {
2666 index.index_mut(self.as_mut_str())
2667 }
2668}
2669
2670#[stable(feature = "rust1", since = "1.0.0")]
2671impl ops::Deref for String {
2672 type Target = str;
2673
2674 #[inline]
2675 fn deref(&self) -> &str {
2676 self.as_str()
2677 }
2678}
2679
2680#[unstable(feature = "deref_pure_trait", issue = "87121")]
2681unsafe impl ops::DerefPure for String {}
2682
2683#[stable(feature = "derefmut_for_string", since = "1.3.0")]
2684impl ops::DerefMut for String {
2685 #[inline]
2686 fn deref_mut(&mut self) -> &mut str {
2687 self.as_mut_str()
2688 }
2689}
2690
2691/// A type alias for [`Infallible`].
2692///
2693/// This alias exists for backwards compatibility, and may be eventually deprecated.
2694///
2695/// [`Infallible`]: core::convert::Infallible "convert::Infallible"
2696#[stable(feature = "str_parse_error", since = "1.5.0")]
2697pub type ParseError = core::convert::Infallible;
2698
2699#[cfg(not(no_global_oom_handling))]
2700#[stable(feature = "rust1", since = "1.0.0")]
2701impl FromStr for String {
2702 type Err = core::convert::Infallible;
2703 #[inline]
2704 fn from_str(s: &str) -> Result<String, Self::Err> {
2705 Ok(String::from(s))
2706 }
2707}
2708
2709/// A trait for converting a value to a `String`.
2710///
2711/// This trait is automatically implemented for any type which implements the
2712/// [`Display`] trait. As such, `ToString` shouldn't be implemented directly:
2713/// [`Display`] should be implemented instead, and you get the `ToString`
2714/// implementation for free.
2715///
2716/// [`Display`]: fmt::Display
2717#[rustc_diagnostic_item = "ToString"]
2718#[stable(feature = "rust1", since = "1.0.0")]
2719pub trait ToString {
2720 /// Converts the given value to a `String`.
2721 ///
2722 /// # Examples
2723 ///
2724 /// ```
2725 /// let i = 5;
2726 /// let five = String::from("5");
2727 ///
2728 /// assert_eq!(five, i.to_string());
2729 /// ```
2730 #[rustc_conversion_suggestion]
2731 #[stable(feature = "rust1", since = "1.0.0")]
2732 #[rustc_diagnostic_item = "to_string_method"]
2733 fn to_string(&self) -> String;
2734}
2735
2736/// # Panics
2737///
2738/// In this implementation, the `to_string` method panics
2739/// if the `Display` implementation returns an error.
2740/// This indicates an incorrect `Display` implementation
2741/// since `fmt::Write for String` never returns an error itself.
2742#[cfg(not(no_global_oom_handling))]
2743#[stable(feature = "rust1", since = "1.0.0")]
2744impl<T: fmt::Display + ?Sized> ToString for T {
2745 #[inline]
2746 fn to_string(&self) -> String {
2747 <Self as SpecToString>::spec_to_string(self)
2748 }
2749}
2750
2751#[cfg(not(no_global_oom_handling))]
2752trait SpecToString {
2753 fn spec_to_string(&self) -> String;
2754}
2755
2756#[cfg(not(no_global_oom_handling))]
2757impl<T: fmt::Display + ?Sized> SpecToString for T {
2758 // A common guideline is to not inline generic functions. However,
2759 // removing `#[inline]` from this method causes non-negligible regressions.
2760 // See <https://github.com/rust-lang/rust/pull/74852>, the last attempt
2761 // to try to remove it.
2762 #[inline]
2763 default fn spec_to_string(&self) -> String {
2764 let mut buf: String = String::new();
2765 let mut formatter: Formatter<'_> =
2766 core::fmt::Formatter::new(&mut buf, options:core::fmt::FormattingOptions::new());
2767 // Bypass format_args!() to avoid write_str with zero-length strs
2768 fmt::Display::fmt(self, &mut formatter)
2769 .expect(msg:"a Display implementation returned an error unexpectedly");
2770 buf
2771 }
2772}
2773
2774#[cfg(not(no_global_oom_handling))]
2775impl SpecToString for core::ascii::Char {
2776 #[inline]
2777 fn spec_to_string(&self) -> String {
2778 self.as_str().to_owned()
2779 }
2780}
2781
2782#[cfg(not(no_global_oom_handling))]
2783impl SpecToString for char {
2784 #[inline]
2785 fn spec_to_string(&self) -> String {
2786 String::from(self.encode_utf8(&mut [0; char::MAX_LEN_UTF8]))
2787 }
2788}
2789
2790#[cfg(not(no_global_oom_handling))]
2791impl SpecToString for bool {
2792 #[inline]
2793 fn spec_to_string(&self) -> String {
2794 String::from(if *self { "true" } else { "false" })
2795 }
2796}
2797
2798#[cfg(not(no_global_oom_handling))]
2799impl SpecToString for u8 {
2800 #[inline]
2801 fn spec_to_string(&self) -> String {
2802 let mut buf: String = String::with_capacity(3);
2803 let mut n: u8 = *self;
2804 if n >= 10 {
2805 if n >= 100 {
2806 buf.push((b'0' + n / 100) as char);
2807 n %= 100;
2808 }
2809 buf.push((b'0' + n / 10) as char);
2810 n %= 10;
2811 }
2812 buf.push((b'0' + n) as char);
2813 buf
2814 }
2815}
2816
2817#[cfg(not(no_global_oom_handling))]
2818impl SpecToString for i8 {
2819 #[inline]
2820 fn spec_to_string(&self) -> String {
2821 let mut buf: String = String::with_capacity(4);
2822 if self.is_negative() {
2823 buf.push(ch:'-');
2824 }
2825 let mut n: u8 = self.unsigned_abs();
2826 if n >= 10 {
2827 if n >= 100 {
2828 buf.push(ch:'1');
2829 n -= 100;
2830 }
2831 buf.push((b'0' + n / 10) as char);
2832 n %= 10;
2833 }
2834 buf.push((b'0' + n) as char);
2835 buf
2836 }
2837}
2838
2839// Generic/generated code can sometimes have multiple, nested references
2840// for strings, including `&&&str`s that would never be written
2841// by hand. This macro generates twelve layers of nested `&`-impl
2842// for primitive strings.
2843#[cfg(not(no_global_oom_handling))]
2844macro_rules! to_string_str_wrap_in_ref {
2845 {x $($x:ident)*} => {
2846 &to_string_str_wrap_in_ref! { $($x)* }
2847 };
2848 {} => { str };
2849}
2850#[cfg(not(no_global_oom_handling))]
2851macro_rules! to_string_expr_wrap_in_deref {
2852 {$self:expr ; x $($x:ident)*} => {
2853 *(to_string_expr_wrap_in_deref! { $self ; $($x)* })
2854 };
2855 {$self:expr ;} => { $self };
2856}
2857#[cfg(not(no_global_oom_handling))]
2858macro_rules! to_string_str {
2859 {$($($x:ident)*),+} => {
2860 $(
2861 impl SpecToString for to_string_str_wrap_in_ref!($($x)*) {
2862 #[inline]
2863 fn spec_to_string(&self) -> String {
2864 String::from(to_string_expr_wrap_in_deref!(self ; $($x)*))
2865 }
2866 }
2867 )+
2868 };
2869}
2870
2871#[cfg(not(no_global_oom_handling))]
2872to_string_str! {
2873 x x x x x x x x x x x x,
2874 x x x x x x x x x x x,
2875 x x x x x x x x x x,
2876 x x x x x x x x x,
2877 x x x x x x x x,
2878 x x x x x x x,
2879 x x x x x x,
2880 x x x x x,
2881 x x x x,
2882 x x x,
2883 x x,
2884 x,
2885}
2886
2887#[cfg(not(no_global_oom_handling))]
2888impl SpecToString for Cow<'_, str> {
2889 #[inline]
2890 fn spec_to_string(&self) -> String {
2891 self[..].to_owned()
2892 }
2893}
2894
2895#[cfg(not(no_global_oom_handling))]
2896impl SpecToString for String {
2897 #[inline]
2898 fn spec_to_string(&self) -> String {
2899 self.to_owned()
2900 }
2901}
2902
2903#[cfg(not(no_global_oom_handling))]
2904impl SpecToString for fmt::Arguments<'_> {
2905 #[inline]
2906 fn spec_to_string(&self) -> String {
2907 crate::fmt::format(*self)
2908 }
2909}
2910
2911#[stable(feature = "rust1", since = "1.0.0")]
2912impl AsRef<str> for String {
2913 #[inline]
2914 fn as_ref(&self) -> &str {
2915 self
2916 }
2917}
2918
2919#[stable(feature = "string_as_mut", since = "1.43.0")]
2920impl AsMut<str> for String {
2921 #[inline]
2922 fn as_mut(&mut self) -> &mut str {
2923 self
2924 }
2925}
2926
2927#[stable(feature = "rust1", since = "1.0.0")]
2928impl AsRef<[u8]> for String {
2929 #[inline]
2930 fn as_ref(&self) -> &[u8] {
2931 self.as_bytes()
2932 }
2933}
2934
2935#[cfg(not(no_global_oom_handling))]
2936#[stable(feature = "rust1", since = "1.0.0")]
2937impl From<&str> for String {
2938 /// Converts a `&str` into a [`String`].
2939 ///
2940 /// The result is allocated on the heap.
2941 #[inline]
2942 fn from(s: &str) -> String {
2943 s.to_owned()
2944 }
2945}
2946
2947#[cfg(not(no_global_oom_handling))]
2948#[stable(feature = "from_mut_str_for_string", since = "1.44.0")]
2949impl From<&mut str> for String {
2950 /// Converts a `&mut str` into a [`String`].
2951 ///
2952 /// The result is allocated on the heap.
2953 #[inline]
2954 fn from(s: &mut str) -> String {
2955 s.to_owned()
2956 }
2957}
2958
2959#[cfg(not(no_global_oom_handling))]
2960#[stable(feature = "from_ref_string", since = "1.35.0")]
2961impl From<&String> for String {
2962 /// Converts a `&String` into a [`String`].
2963 ///
2964 /// This clones `s` and returns the clone.
2965 #[inline]
2966 fn from(s: &String) -> String {
2967 s.clone()
2968 }
2969}
2970
2971// note: test pulls in std, which causes errors here
2972#[stable(feature = "string_from_box", since = "1.18.0")]
2973impl From<Box<str>> for String {
2974 /// Converts the given boxed `str` slice to a [`String`].
2975 /// It is notable that the `str` slice is owned.
2976 ///
2977 /// # Examples
2978 ///
2979 /// ```
2980 /// let s1: String = String::from("hello world");
2981 /// let s2: Box<str> = s1.into_boxed_str();
2982 /// let s3: String = String::from(s2);
2983 ///
2984 /// assert_eq!("hello world", s3)
2985 /// ```
2986 fn from(s: Box<str>) -> String {
2987 s.into_string()
2988 }
2989}
2990
2991#[cfg(not(no_global_oom_handling))]
2992#[stable(feature = "box_from_str", since = "1.20.0")]
2993impl From<String> for Box<str> {
2994 /// Converts the given [`String`] to a boxed `str` slice that is owned.
2995 ///
2996 /// # Examples
2997 ///
2998 /// ```
2999 /// let s1: String = String::from("hello world");
3000 /// let s2: Box<str> = Box::from(s1);
3001 /// let s3: String = String::from(s2);
3002 ///
3003 /// assert_eq!("hello world", s3)
3004 /// ```
3005 fn from(s: String) -> Box<str> {
3006 s.into_boxed_str()
3007 }
3008}
3009
3010#[cfg(not(no_global_oom_handling))]
3011#[stable(feature = "string_from_cow_str", since = "1.14.0")]
3012impl<'a> From<Cow<'a, str>> for String {
3013 /// Converts a clone-on-write string to an owned
3014 /// instance of [`String`].
3015 ///
3016 /// This extracts the owned string,
3017 /// clones the string if it is not already owned.
3018 ///
3019 /// # Example
3020 ///
3021 /// ```
3022 /// # use std::borrow::Cow;
3023 /// // If the string is not owned...
3024 /// let cow: Cow<'_, str> = Cow::Borrowed("eggplant");
3025 /// // It will allocate on the heap and copy the string.
3026 /// let owned: String = String::from(cow);
3027 /// assert_eq!(&owned[..], "eggplant");
3028 /// ```
3029 fn from(s: Cow<'a, str>) -> String {
3030 s.into_owned()
3031 }
3032}
3033
3034#[cfg(not(no_global_oom_handling))]
3035#[stable(feature = "rust1", since = "1.0.0")]
3036impl<'a> From<&'a str> for Cow<'a, str> {
3037 /// Converts a string slice into a [`Borrowed`] variant.
3038 /// No heap allocation is performed, and the string
3039 /// is not copied.
3040 ///
3041 /// # Example
3042 ///
3043 /// ```
3044 /// # use std::borrow::Cow;
3045 /// assert_eq!(Cow::from("eggplant"), Cow::Borrowed("eggplant"));
3046 /// ```
3047 ///
3048 /// [`Borrowed`]: crate::borrow::Cow::Borrowed "borrow::Cow::Borrowed"
3049 #[inline]
3050 fn from(s: &'a str) -> Cow<'a, str> {
3051 Cow::Borrowed(s)
3052 }
3053}
3054
3055#[cfg(not(no_global_oom_handling))]
3056#[stable(feature = "rust1", since = "1.0.0")]
3057impl<'a> From<String> for Cow<'a, str> {
3058 /// Converts a [`String`] into an [`Owned`] variant.
3059 /// No heap allocation is performed, and the string
3060 /// is not copied.
3061 ///
3062 /// # Example
3063 ///
3064 /// ```
3065 /// # use std::borrow::Cow;
3066 /// let s = "eggplant".to_string();
3067 /// let s2 = "eggplant".to_string();
3068 /// assert_eq!(Cow::from(s), Cow::<'static, str>::Owned(s2));
3069 /// ```
3070 ///
3071 /// [`Owned`]: crate::borrow::Cow::Owned "borrow::Cow::Owned"
3072 #[inline]
3073 fn from(s: String) -> Cow<'a, str> {
3074 Cow::Owned(s)
3075 }
3076}
3077
3078#[cfg(not(no_global_oom_handling))]
3079#[stable(feature = "cow_from_string_ref", since = "1.28.0")]
3080impl<'a> From<&'a String> for Cow<'a, str> {
3081 /// Converts a [`String`] reference into a [`Borrowed`] variant.
3082 /// No heap allocation is performed, and the string
3083 /// is not copied.
3084 ///
3085 /// # Example
3086 ///
3087 /// ```
3088 /// # use std::borrow::Cow;
3089 /// let s = "eggplant".to_string();
3090 /// assert_eq!(Cow::from(&s), Cow::Borrowed("eggplant"));
3091 /// ```
3092 ///
3093 /// [`Borrowed`]: crate::borrow::Cow::Borrowed "borrow::Cow::Borrowed"
3094 #[inline]
3095 fn from(s: &'a String) -> Cow<'a, str> {
3096 Cow::Borrowed(s.as_str())
3097 }
3098}
3099
3100#[cfg(not(no_global_oom_handling))]
3101#[stable(feature = "cow_str_from_iter", since = "1.12.0")]
3102impl<'a> FromIterator<char> for Cow<'a, str> {
3103 fn from_iter<I: IntoIterator<Item = char>>(it: I) -> Cow<'a, str> {
3104 Cow::Owned(FromIterator::from_iter(it))
3105 }
3106}
3107
3108#[cfg(not(no_global_oom_handling))]
3109#[stable(feature = "cow_str_from_iter", since = "1.12.0")]
3110impl<'a, 'b> FromIterator<&'b str> for Cow<'a, str> {
3111 fn from_iter<I: IntoIterator<Item = &'b str>>(it: I) -> Cow<'a, str> {
3112 Cow::Owned(FromIterator::from_iter(it))
3113 }
3114}
3115
3116#[cfg(not(no_global_oom_handling))]
3117#[stable(feature = "cow_str_from_iter", since = "1.12.0")]
3118impl<'a> FromIterator<String> for Cow<'a, str> {
3119 fn from_iter<I: IntoIterator<Item = String>>(it: I) -> Cow<'a, str> {
3120 Cow::Owned(FromIterator::from_iter(it))
3121 }
3122}
3123
3124#[stable(feature = "from_string_for_vec_u8", since = "1.14.0")]
3125impl From<String> for Vec<u8> {
3126 /// Converts the given [`String`] to a vector [`Vec`] that holds values of type [`u8`].
3127 ///
3128 /// # Examples
3129 ///
3130 /// ```
3131 /// let s1 = String::from("hello world");
3132 /// let v1 = Vec::from(s1);
3133 ///
3134 /// for b in v1 {
3135 /// println!("{b}");
3136 /// }
3137 /// ```
3138 fn from(string: String) -> Vec<u8> {
3139 string.into_bytes()
3140 }
3141}
3142
3143#[stable(feature = "try_from_vec_u8_for_string", since = "1.87.0")]
3144impl TryFrom<Vec<u8>> for String {
3145 type Error = FromUtf8Error;
3146 /// Converts the given [`Vec<u8>`] into a [`String`] if it contains valid UTF-8 data.
3147 ///
3148 /// # Examples
3149 ///
3150 /// ```
3151 /// let s1 = b"hello world".to_vec();
3152 /// let v1 = String::try_from(s1).unwrap();
3153 /// assert_eq!(v1, "hello world");
3154 ///
3155 /// ```
3156 fn try_from(bytes: Vec<u8>) -> Result<Self, Self::Error> {
3157 Self::from_utf8(vec:bytes)
3158 }
3159}
3160
3161#[cfg(not(no_global_oom_handling))]
3162#[stable(feature = "rust1", since = "1.0.0")]
3163impl fmt::Write for String {
3164 #[inline]
3165 fn write_str(&mut self, s: &str) -> fmt::Result {
3166 self.push_str(string:s);
3167 Ok(())
3168 }
3169
3170 #[inline]
3171 fn write_char(&mut self, c: char) -> fmt::Result {
3172 self.push(ch:c);
3173 Ok(())
3174 }
3175}
3176
3177/// An iterator over the [`char`]s of a string.
3178///
3179/// This struct is created by the [`into_chars`] method on [`String`].
3180/// See its documentation for more.
3181///
3182/// [`char`]: prim@char
3183/// [`into_chars`]: String::into_chars
3184#[cfg_attr(not(no_global_oom_handling), derive(Clone))]
3185#[must_use = "iterators are lazy and do nothing unless consumed"]
3186#[unstable(feature = "string_into_chars", issue = "133125")]
3187pub struct IntoChars {
3188 bytes: vec::IntoIter<u8>,
3189}
3190
3191#[unstable(feature = "string_into_chars", issue = "133125")]
3192impl fmt::Debug for IntoChars {
3193 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
3194 f.debug_tuple(name:"IntoChars").field(&self.as_str()).finish()
3195 }
3196}
3197
3198impl IntoChars {
3199 /// Views the underlying data as a subslice of the original data.
3200 ///
3201 /// # Examples
3202 ///
3203 /// ```
3204 /// #![feature(string_into_chars)]
3205 ///
3206 /// let mut chars = String::from("abc").into_chars();
3207 ///
3208 /// assert_eq!(chars.as_str(), "abc");
3209 /// chars.next();
3210 /// assert_eq!(chars.as_str(), "bc");
3211 /// chars.next();
3212 /// chars.next();
3213 /// assert_eq!(chars.as_str(), "");
3214 /// ```
3215 #[unstable(feature = "string_into_chars", issue = "133125")]
3216 #[must_use]
3217 #[inline]
3218 pub fn as_str(&self) -> &str {
3219 // SAFETY: `bytes` is a valid UTF-8 string.
3220 unsafe { str::from_utf8_unchecked(self.bytes.as_slice()) }
3221 }
3222
3223 /// Consumes the `IntoChars`, returning the remaining string.
3224 ///
3225 /// # Examples
3226 ///
3227 /// ```
3228 /// #![feature(string_into_chars)]
3229 ///
3230 /// let chars = String::from("abc").into_chars();
3231 /// assert_eq!(chars.into_string(), "abc");
3232 ///
3233 /// let mut chars = String::from("def").into_chars();
3234 /// chars.next();
3235 /// assert_eq!(chars.into_string(), "ef");
3236 /// ```
3237 #[cfg(not(no_global_oom_handling))]
3238 #[unstable(feature = "string_into_chars", issue = "133125")]
3239 #[inline]
3240 pub fn into_string(self) -> String {
3241 // Safety: `bytes` are kept in UTF-8 form, only removing whole `char`s at a time.
3242 unsafe { String::from_utf8_unchecked(self.bytes.collect()) }
3243 }
3244
3245 #[inline]
3246 fn iter(&self) -> CharIndices<'_> {
3247 self.as_str().char_indices()
3248 }
3249}
3250
3251#[unstable(feature = "string_into_chars", issue = "133125")]
3252impl Iterator for IntoChars {
3253 type Item = char;
3254
3255 #[inline]
3256 fn next(&mut self) -> Option<char> {
3257 let mut iter = self.iter();
3258 match iter.next() {
3259 None => None,
3260 Some((_, ch)) => {
3261 let offset = iter.offset();
3262 // `offset` is a valid index.
3263 let _ = self.bytes.advance_by(offset);
3264 Some(ch)
3265 }
3266 }
3267 }
3268
3269 #[inline]
3270 fn count(self) -> usize {
3271 self.iter().count()
3272 }
3273
3274 #[inline]
3275 fn size_hint(&self) -> (usize, Option<usize>) {
3276 self.iter().size_hint()
3277 }
3278
3279 #[inline]
3280 fn last(mut self) -> Option<char> {
3281 self.next_back()
3282 }
3283}
3284
3285#[unstable(feature = "string_into_chars", issue = "133125")]
3286impl DoubleEndedIterator for IntoChars {
3287 #[inline]
3288 fn next_back(&mut self) -> Option<char> {
3289 let len: usize = self.as_str().len();
3290 let mut iter: CharIndices<'_> = self.iter();
3291 match iter.next_back() {
3292 None => None,
3293 Some((idx: usize, ch: char)) => {
3294 // `idx` is a valid index.
3295 let _ = self.bytes.advance_back_by(len - idx);
3296 Some(ch)
3297 }
3298 }
3299 }
3300}
3301
3302#[unstable(feature = "string_into_chars", issue = "133125")]
3303impl FusedIterator for IntoChars {}
3304
3305/// A draining iterator for `String`.
3306///
3307/// This struct is created by the [`drain`] method on [`String`]. See its
3308/// documentation for more.
3309///
3310/// [`drain`]: String::drain
3311#[stable(feature = "drain", since = "1.6.0")]
3312pub struct Drain<'a> {
3313 /// Will be used as &'a mut String in the destructor
3314 string: *mut String,
3315 /// Start of part to remove
3316 start: usize,
3317 /// End of part to remove
3318 end: usize,
3319 /// Current remaining range to remove
3320 iter: Chars<'a>,
3321}
3322
3323#[stable(feature = "collection_debug", since = "1.17.0")]
3324impl fmt::Debug for Drain<'_> {
3325 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
3326 f.debug_tuple(name:"Drain").field(&self.as_str()).finish()
3327 }
3328}
3329
3330#[stable(feature = "drain", since = "1.6.0")]
3331unsafe impl Sync for Drain<'_> {}
3332#[stable(feature = "drain", since = "1.6.0")]
3333unsafe impl Send for Drain<'_> {}
3334
3335#[stable(feature = "drain", since = "1.6.0")]
3336impl Drop for Drain<'_> {
3337 fn drop(&mut self) {
3338 unsafe {
3339 // Use Vec::drain. "Reaffirm" the bounds checks to avoid
3340 // panic code being inserted again.
3341 let self_vec: &mut Vec = (*self.string).as_mut_vec();
3342 if self.start <= self.end && self.end <= self_vec.len() {
3343 self_vec.drain(self.start..self.end);
3344 }
3345 }
3346 }
3347}
3348
3349impl<'a> Drain<'a> {
3350 /// Returns the remaining (sub)string of this iterator as a slice.
3351 ///
3352 /// # Examples
3353 ///
3354 /// ```
3355 /// let mut s = String::from("abc");
3356 /// let mut drain = s.drain(..);
3357 /// assert_eq!(drain.as_str(), "abc");
3358 /// let _ = drain.next().unwrap();
3359 /// assert_eq!(drain.as_str(), "bc");
3360 /// ```
3361 #[must_use]
3362 #[stable(feature = "string_drain_as_str", since = "1.55.0")]
3363 pub fn as_str(&self) -> &str {
3364 self.iter.as_str()
3365 }
3366}
3367
3368#[stable(feature = "string_drain_as_str", since = "1.55.0")]
3369impl<'a> AsRef<str> for Drain<'a> {
3370 fn as_ref(&self) -> &str {
3371 self.as_str()
3372 }
3373}
3374
3375#[stable(feature = "string_drain_as_str", since = "1.55.0")]
3376impl<'a> AsRef<[u8]> for Drain<'a> {
3377 fn as_ref(&self) -> &[u8] {
3378 self.as_str().as_bytes()
3379 }
3380}
3381
3382#[stable(feature = "drain", since = "1.6.0")]
3383impl Iterator for Drain<'_> {
3384 type Item = char;
3385
3386 #[inline]
3387 fn next(&mut self) -> Option<char> {
3388 self.iter.next()
3389 }
3390
3391 fn size_hint(&self) -> (usize, Option<usize>) {
3392 self.iter.size_hint()
3393 }
3394
3395 #[inline]
3396 fn last(mut self) -> Option<char> {
3397 self.next_back()
3398 }
3399}
3400
3401#[stable(feature = "drain", since = "1.6.0")]
3402impl DoubleEndedIterator for Drain<'_> {
3403 #[inline]
3404 fn next_back(&mut self) -> Option<char> {
3405 self.iter.next_back()
3406 }
3407}
3408
3409#[stable(feature = "fused", since = "1.26.0")]
3410impl FusedIterator for Drain<'_> {}
3411
3412#[cfg(not(no_global_oom_handling))]
3413#[stable(feature = "from_char_for_string", since = "1.46.0")]
3414impl From<char> for String {
3415 /// Allocates an owned [`String`] from a single character.
3416 ///
3417 /// # Example
3418 /// ```rust
3419 /// let c: char = 'a';
3420 /// let s: String = String::from(c);
3421 /// assert_eq!("a", &s[..]);
3422 /// ```
3423 #[inline]
3424 fn from(c: char) -> Self {
3425 c.to_string()
3426 }
3427}
3428