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 let len = self.len();
1405 let ch_len = ch.len_utf8();
1406 self.reserve(ch_len);
1407
1408 // SAFETY: Just reserved capacity for at least the length needed to encode `ch`.
1409 unsafe {
1410 core::char::encode_utf8_raw_unchecked(ch as u32, self.vec.as_mut_ptr().add(self.len()));
1411 self.vec.set_len(len + ch_len);
1412 }
1413 }
1414
1415 /// Returns a byte slice of this `String`'s contents.
1416 ///
1417 /// The inverse of this method is [`from_utf8`].
1418 ///
1419 /// [`from_utf8`]: String::from_utf8
1420 ///
1421 /// # Examples
1422 ///
1423 /// ```
1424 /// let s = String::from("hello");
1425 ///
1426 /// assert_eq!(&[104, 101, 108, 108, 111], s.as_bytes());
1427 /// ```
1428 #[inline]
1429 #[must_use]
1430 #[stable(feature = "rust1", since = "1.0.0")]
1431 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1432 pub const fn as_bytes(&self) -> &[u8] {
1433 self.vec.as_slice()
1434 }
1435
1436 /// Shortens this `String` to the specified length.
1437 ///
1438 /// If `new_len` is greater than or equal to the string's current length, this has no
1439 /// effect.
1440 ///
1441 /// Note that this method has no effect on the allocated capacity
1442 /// of the string
1443 ///
1444 /// # Panics
1445 ///
1446 /// Panics if `new_len` does not lie on a [`char`] boundary.
1447 ///
1448 /// # Examples
1449 ///
1450 /// ```
1451 /// let mut s = String::from("hello");
1452 ///
1453 /// s.truncate(2);
1454 ///
1455 /// assert_eq!("he", s);
1456 /// ```
1457 #[inline]
1458 #[stable(feature = "rust1", since = "1.0.0")]
1459 pub fn truncate(&mut self, new_len: usize) {
1460 if new_len <= self.len() {
1461 assert!(self.is_char_boundary(new_len));
1462 self.vec.truncate(new_len)
1463 }
1464 }
1465
1466 /// Removes the last character from the string buffer and returns it.
1467 ///
1468 /// Returns [`None`] if this `String` is empty.
1469 ///
1470 /// # Examples
1471 ///
1472 /// ```
1473 /// let mut s = String::from("abč");
1474 ///
1475 /// assert_eq!(s.pop(), Some('č'));
1476 /// assert_eq!(s.pop(), Some('b'));
1477 /// assert_eq!(s.pop(), Some('a'));
1478 ///
1479 /// assert_eq!(s.pop(), None);
1480 /// ```
1481 #[inline]
1482 #[stable(feature = "rust1", since = "1.0.0")]
1483 pub fn pop(&mut self) -> Option<char> {
1484 let ch = self.chars().rev().next()?;
1485 let newlen = self.len() - ch.len_utf8();
1486 unsafe {
1487 self.vec.set_len(newlen);
1488 }
1489 Some(ch)
1490 }
1491
1492 /// Removes a [`char`] from this `String` at a byte position and returns it.
1493 ///
1494 /// This is an *O*(*n*) operation, as it requires copying every element in the
1495 /// buffer.
1496 ///
1497 /// # Panics
1498 ///
1499 /// Panics if `idx` is larger than or equal to the `String`'s length,
1500 /// or if it does not lie on a [`char`] boundary.
1501 ///
1502 /// # Examples
1503 ///
1504 /// ```
1505 /// let mut s = String::from("abç");
1506 ///
1507 /// assert_eq!(s.remove(0), 'a');
1508 /// assert_eq!(s.remove(1), 'Γ§');
1509 /// assert_eq!(s.remove(0), 'b');
1510 /// ```
1511 #[inline]
1512 #[stable(feature = "rust1", since = "1.0.0")]
1513 #[rustc_confusables("delete", "take")]
1514 pub fn remove(&mut self, idx: usize) -> char {
1515 let ch = match self[idx..].chars().next() {
1516 Some(ch) => ch,
1517 None => panic!("cannot remove a char from the end of a string"),
1518 };
1519
1520 let next = idx + ch.len_utf8();
1521 let len = self.len();
1522 unsafe {
1523 ptr::copy(self.vec.as_ptr().add(next), self.vec.as_mut_ptr().add(idx), len - next);
1524 self.vec.set_len(len - (next - idx));
1525 }
1526 ch
1527 }
1528
1529 /// Remove all matches of pattern `pat` in the `String`.
1530 ///
1531 /// # Examples
1532 ///
1533 /// ```
1534 /// #![feature(string_remove_matches)]
1535 /// let mut s = String::from("Trees are not green, the sky is not blue.");
1536 /// s.remove_matches("not ");
1537 /// assert_eq!("Trees are green, the sky is blue.", s);
1538 /// ```
1539 ///
1540 /// Matches will be detected and removed iteratively, so in cases where
1541 /// patterns overlap, only the first pattern will be removed:
1542 ///
1543 /// ```
1544 /// #![feature(string_remove_matches)]
1545 /// let mut s = String::from("banana");
1546 /// s.remove_matches("ana");
1547 /// assert_eq!("bna", s);
1548 /// ```
1549 #[cfg(not(no_global_oom_handling))]
1550 #[unstable(feature = "string_remove_matches", reason = "new API", issue = "72826")]
1551 pub fn remove_matches<P: Pattern>(&mut self, pat: P) {
1552 use core::str::pattern::Searcher;
1553
1554 let rejections = {
1555 let mut searcher = pat.into_searcher(self);
1556 // Per Searcher::next:
1557 //
1558 // A Match result needs to contain the whole matched pattern,
1559 // however Reject results may be split up into arbitrary many
1560 // adjacent fragments. Both ranges may have zero length.
1561 //
1562 // In practice the implementation of Searcher::next_match tends to
1563 // be more efficient, so we use it here and do some work to invert
1564 // matches into rejections since that's what we want to copy below.
1565 let mut front = 0;
1566 let rejections: Vec<_> = from_fn(|| {
1567 let (start, end) = searcher.next_match()?;
1568 let prev_front = front;
1569 front = end;
1570 Some((prev_front, start))
1571 })
1572 .collect();
1573 rejections.into_iter().chain(core::iter::once((front, self.len())))
1574 };
1575
1576 let mut len = 0;
1577 let ptr = self.vec.as_mut_ptr();
1578
1579 for (start, end) in rejections {
1580 let count = end - start;
1581 if start != len {
1582 // SAFETY: per Searcher::next:
1583 //
1584 // The stream of Match and Reject values up to a Done will
1585 // contain index ranges that are adjacent, non-overlapping,
1586 // covering the whole haystack, and laying on utf8
1587 // boundaries.
1588 unsafe {
1589 ptr::copy(ptr.add(start), ptr.add(len), count);
1590 }
1591 }
1592 len += count;
1593 }
1594
1595 unsafe {
1596 self.vec.set_len(len);
1597 }
1598 }
1599
1600 /// Retains only the characters specified by the predicate.
1601 ///
1602 /// In other words, remove all characters `c` such that `f(c)` returns `false`.
1603 /// This method operates in place, visiting each character exactly once in the
1604 /// original order, and preserves the order of the retained characters.
1605 ///
1606 /// # Examples
1607 ///
1608 /// ```
1609 /// let mut s = String::from("f_o_ob_ar");
1610 ///
1611 /// s.retain(|c| c != '_');
1612 ///
1613 /// assert_eq!(s, "foobar");
1614 /// ```
1615 ///
1616 /// Because the elements are visited exactly once in the original order,
1617 /// external state may be used to decide which elements to keep.
1618 ///
1619 /// ```
1620 /// let mut s = String::from("abcde");
1621 /// let keep = [false, true, true, false, true];
1622 /// let mut iter = keep.iter();
1623 /// s.retain(|_| *iter.next().unwrap());
1624 /// assert_eq!(s, "bce");
1625 /// ```
1626 #[inline]
1627 #[stable(feature = "string_retain", since = "1.26.0")]
1628 pub fn retain<F>(&mut self, mut f: F)
1629 where
1630 F: FnMut(char) -> bool,
1631 {
1632 struct SetLenOnDrop<'a> {
1633 s: &'a mut String,
1634 idx: usize,
1635 del_bytes: usize,
1636 }
1637
1638 impl<'a> Drop for SetLenOnDrop<'a> {
1639 fn drop(&mut self) {
1640 let new_len = self.idx - self.del_bytes;
1641 debug_assert!(new_len <= self.s.len());
1642 unsafe { self.s.vec.set_len(new_len) };
1643 }
1644 }
1645
1646 let len = self.len();
1647 let mut guard = SetLenOnDrop { s: self, idx: 0, del_bytes: 0 };
1648
1649 while guard.idx < len {
1650 let ch =
1651 // SAFETY: `guard.idx` is positive-or-zero and less that len so the `get_unchecked`
1652 // is in bound. `self` is valid UTF-8 like string and the returned slice starts at
1653 // a unicode code point so the `Chars` always return one character.
1654 unsafe { guard.s.get_unchecked(guard.idx..len).chars().next().unwrap_unchecked() };
1655 let ch_len = ch.len_utf8();
1656
1657 if !f(ch) {
1658 guard.del_bytes += ch_len;
1659 } else if guard.del_bytes > 0 {
1660 // SAFETY: `guard.idx` is in bound and `guard.del_bytes` represent the number of
1661 // bytes that are erased from the string so the resulting `guard.idx -
1662 // guard.del_bytes` always represent a valid unicode code point.
1663 //
1664 // `guard.del_bytes` >= `ch.len_utf8()`, so taking a slice with `ch.len_utf8()` len
1665 // is safe.
1666 ch.encode_utf8(unsafe {
1667 crate::slice::from_raw_parts_mut(
1668 guard.s.as_mut_ptr().add(guard.idx - guard.del_bytes),
1669 ch.len_utf8(),
1670 )
1671 });
1672 }
1673
1674 // Point idx to the next char
1675 guard.idx += ch_len;
1676 }
1677
1678 drop(guard);
1679 }
1680
1681 /// Inserts a character into this `String` at a byte position.
1682 ///
1683 /// This is an *O*(*n*) operation as it requires copying every element in the
1684 /// buffer.
1685 ///
1686 /// # Panics
1687 ///
1688 /// Panics if `idx` is larger than the `String`'s length, or if it does not
1689 /// lie on a [`char`] boundary.
1690 ///
1691 /// # Examples
1692 ///
1693 /// ```
1694 /// let mut s = String::with_capacity(3);
1695 ///
1696 /// s.insert(0, 'f');
1697 /// s.insert(1, 'o');
1698 /// s.insert(2, 'o');
1699 ///
1700 /// assert_eq!("foo", s);
1701 /// ```
1702 #[cfg(not(no_global_oom_handling))]
1703 #[inline]
1704 #[stable(feature = "rust1", since = "1.0.0")]
1705 #[rustc_confusables("set")]
1706 pub fn insert(&mut self, idx: usize, ch: char) {
1707 assert!(self.is_char_boundary(idx));
1708
1709 let len = self.len();
1710 let ch_len = ch.len_utf8();
1711 self.reserve(ch_len);
1712
1713 // SAFETY: Move the bytes starting from `idx` to their new location `ch_len`
1714 // bytes ahead. This is safe because sufficient capacity was reserved, and `idx`
1715 // is a char boundary.
1716 unsafe {
1717 ptr::copy(
1718 self.vec.as_ptr().add(idx),
1719 self.vec.as_mut_ptr().add(idx + ch_len),
1720 len - idx,
1721 );
1722 }
1723
1724 // SAFETY: Encode the character into the vacated region if `idx != len`,
1725 // or into the uninitialized spare capacity otherwise.
1726 unsafe {
1727 core::char::encode_utf8_raw_unchecked(ch as u32, self.vec.as_mut_ptr().add(idx));
1728 }
1729
1730 // SAFETY: Update the length to include the newly added bytes.
1731 unsafe {
1732 self.vec.set_len(len + ch_len);
1733 }
1734 }
1735
1736 /// Inserts a string slice into this `String` at a byte position.
1737 ///
1738 /// This is an *O*(*n*) operation as it requires copying every element in the
1739 /// buffer.
1740 ///
1741 /// # Panics
1742 ///
1743 /// Panics if `idx` is larger than the `String`'s length, or if it does not
1744 /// lie on a [`char`] boundary.
1745 ///
1746 /// # Examples
1747 ///
1748 /// ```
1749 /// let mut s = String::from("bar");
1750 ///
1751 /// s.insert_str(0, "foo");
1752 ///
1753 /// assert_eq!("foobar", s);
1754 /// ```
1755 #[cfg(not(no_global_oom_handling))]
1756 #[inline]
1757 #[stable(feature = "insert_str", since = "1.16.0")]
1758 #[rustc_diagnostic_item = "string_insert_str"]
1759 pub fn insert_str(&mut self, idx: usize, string: &str) {
1760 assert!(self.is_char_boundary(idx));
1761
1762 let len = self.len();
1763 let amt = string.len();
1764 self.reserve(amt);
1765
1766 // SAFETY: Move the bytes starting from `idx` to their new location `amt` bytes
1767 // ahead. This is safe because sufficient capacity was just reserved, and `idx`
1768 // is a char boundary.
1769 unsafe {
1770 ptr::copy(self.vec.as_ptr().add(idx), self.vec.as_mut_ptr().add(idx + amt), len - idx);
1771 }
1772
1773 // SAFETY: Copy the new string slice into the vacated region if `idx != len`,
1774 // or into the uninitialized spare capacity otherwise. The borrow checker
1775 // ensures that the source and destination do not overlap.
1776 unsafe {
1777 ptr::copy_nonoverlapping(string.as_ptr(), self.vec.as_mut_ptr().add(idx), amt);
1778 }
1779
1780 // SAFETY: Update the length to include the newly added bytes.
1781 unsafe {
1782 self.vec.set_len(len + amt);
1783 }
1784 }
1785
1786 /// Returns a mutable reference to the contents of this `String`.
1787 ///
1788 /// # Safety
1789 ///
1790 /// This function is unsafe because the returned `&mut Vec` allows writing
1791 /// bytes which are not valid UTF-8. If this constraint is violated, using
1792 /// the original `String` after dropping the `&mut Vec` may violate memory
1793 /// safety, as the rest of the standard library assumes that `String`s are
1794 /// valid UTF-8.
1795 ///
1796 /// # Examples
1797 ///
1798 /// ```
1799 /// let mut s = String::from("hello");
1800 ///
1801 /// unsafe {
1802 /// let vec = s.as_mut_vec();
1803 /// assert_eq!(&[104, 101, 108, 108, 111][..], &vec[..]);
1804 ///
1805 /// vec.reverse();
1806 /// }
1807 /// assert_eq!(s, "olleh");
1808 /// ```
1809 #[inline]
1810 #[stable(feature = "rust1", since = "1.0.0")]
1811 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1812 pub const unsafe fn as_mut_vec(&mut self) -> &mut Vec<u8> {
1813 &mut self.vec
1814 }
1815
1816 /// Returns the length of this `String`, in bytes, not [`char`]s or
1817 /// graphemes. In other words, it might not be what a human considers the
1818 /// length of the string.
1819 ///
1820 /// # Examples
1821 ///
1822 /// ```
1823 /// let a = String::from("foo");
1824 /// assert_eq!(a.len(), 3);
1825 ///
1826 /// let fancy_f = String::from("Ζ’oo");
1827 /// assert_eq!(fancy_f.len(), 4);
1828 /// assert_eq!(fancy_f.chars().count(), 3);
1829 /// ```
1830 #[inline]
1831 #[must_use]
1832 #[stable(feature = "rust1", since = "1.0.0")]
1833 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1834 #[rustc_confusables("length", "size")]
1835 #[rustc_no_implicit_autorefs]
1836 pub const fn len(&self) -> usize {
1837 self.vec.len()
1838 }
1839
1840 /// Returns `true` if this `String` has a length of zero, and `false` otherwise.
1841 ///
1842 /// # Examples
1843 ///
1844 /// ```
1845 /// let mut v = String::new();
1846 /// assert!(v.is_empty());
1847 ///
1848 /// v.push('a');
1849 /// assert!(!v.is_empty());
1850 /// ```
1851 #[inline]
1852 #[must_use]
1853 #[stable(feature = "rust1", since = "1.0.0")]
1854 #[rustc_const_stable(feature = "const_vec_string_slice", since = "1.87.0")]
1855 #[rustc_no_implicit_autorefs]
1856 pub const fn is_empty(&self) -> bool {
1857 self.len() == 0
1858 }
1859
1860 /// Splits the string into two at the given byte index.
1861 ///
1862 /// Returns a newly allocated `String`. `self` contains bytes `[0, at)`, and
1863 /// the returned `String` contains bytes `[at, len)`. `at` must be on the
1864 /// boundary of a UTF-8 code point.
1865 ///
1866 /// Note that the capacity of `self` does not change.
1867 ///
1868 /// # Panics
1869 ///
1870 /// Panics if `at` is not on a `UTF-8` code point boundary, or if it is beyond the last
1871 /// code point of the string.
1872 ///
1873 /// # Examples
1874 ///
1875 /// ```
1876 /// # fn main() {
1877 /// let mut hello = String::from("Hello, World!");
1878 /// let world = hello.split_off(7);
1879 /// assert_eq!(hello, "Hello, ");
1880 /// assert_eq!(world, "World!");
1881 /// # }
1882 /// ```
1883 #[cfg(not(no_global_oom_handling))]
1884 #[inline]
1885 #[stable(feature = "string_split_off", since = "1.16.0")]
1886 #[must_use = "use `.truncate()` if you don't need the other half"]
1887 pub fn split_off(&mut self, at: usize) -> String {
1888 assert!(self.is_char_boundary(at));
1889 let other = self.vec.split_off(at);
1890 unsafe { String::from_utf8_unchecked(other) }
1891 }
1892
1893 /// Truncates this `String`, removing all contents.
1894 ///
1895 /// While this means the `String` will have a length of zero, it does not
1896 /// touch its capacity.
1897 ///
1898 /// # Examples
1899 ///
1900 /// ```
1901 /// let mut s = String::from("foo");
1902 ///
1903 /// s.clear();
1904 ///
1905 /// assert!(s.is_empty());
1906 /// assert_eq!(0, s.len());
1907 /// assert_eq!(3, s.capacity());
1908 /// ```
1909 #[inline]
1910 #[stable(feature = "rust1", since = "1.0.0")]
1911 pub fn clear(&mut self) {
1912 self.vec.clear()
1913 }
1914
1915 /// Removes the specified range from the string in bulk, returning all
1916 /// removed characters as an iterator.
1917 ///
1918 /// The returned iterator keeps a mutable borrow on the string to optimize
1919 /// its implementation.
1920 ///
1921 /// # Panics
1922 ///
1923 /// Panics if the starting point or end point do not lie on a [`char`]
1924 /// boundary, or if they're out of bounds.
1925 ///
1926 /// # Leaking
1927 ///
1928 /// If the returned iterator goes out of scope without being dropped (due to
1929 /// [`core::mem::forget`], for example), the string may still contain a copy
1930 /// of any drained characters, or may have lost characters arbitrarily,
1931 /// including characters outside the range.
1932 ///
1933 /// # Examples
1934 ///
1935 /// ```
1936 /// let mut s = String::from("Ξ± is alpha, Ξ² is beta");
1937 /// let beta_offset = s.find('Ξ²').unwrap_or(s.len());
1938 ///
1939 /// // Remove the range up until the Ξ² from the string
1940 /// let t: String = s.drain(..beta_offset).collect();
1941 /// assert_eq!(t, "Ξ± is alpha, ");
1942 /// assert_eq!(s, "Ξ² is beta");
1943 ///
1944 /// // A full range clears the string, like `clear()` does
1945 /// s.drain(..);
1946 /// assert_eq!(s, "");
1947 /// ```
1948 #[stable(feature = "drain", since = "1.6.0")]
1949 pub fn drain<R>(&mut self, range: R) -> Drain<'_>
1950 where
1951 R: RangeBounds<usize>,
1952 {
1953 // Memory safety
1954 //
1955 // The String version of Drain does not have the memory safety issues
1956 // of the vector version. The data is just plain bytes.
1957 // Because the range removal happens in Drop, if the Drain iterator is leaked,
1958 // the removal will not happen.
1959 let Range { start, end } = slice::range(range, ..self.len());
1960 assert!(self.is_char_boundary(start));
1961 assert!(self.is_char_boundary(end));
1962
1963 // Take out two simultaneous borrows. The &mut String won't be accessed
1964 // until iteration is over, in Drop.
1965 let self_ptr = self as *mut _;
1966 // SAFETY: `slice::range` and `is_char_boundary` do the appropriate bounds checks.
1967 let chars_iter = unsafe { self.get_unchecked(start..end) }.chars();
1968
1969 Drain { start, end, iter: chars_iter, string: self_ptr }
1970 }
1971
1972 /// Converts a `String` into an iterator over the [`char`]s of the string.
1973 ///
1974 /// As a string consists of valid UTF-8, we can iterate through a string
1975 /// by [`char`]. This method returns such an iterator.
1976 ///
1977 /// It's important to remember that [`char`] represents a Unicode Scalar
1978 /// Value, and might not match your idea of what a 'character' is. Iteration
1979 /// over grapheme clusters may be what you actually want. That functionality
1980 /// is not provided by Rust's standard library, check crates.io instead.
1981 ///
1982 /// # Examples
1983 ///
1984 /// Basic usage:
1985 ///
1986 /// ```
1987 /// #![feature(string_into_chars)]
1988 ///
1989 /// let word = String::from("goodbye");
1990 ///
1991 /// let mut chars = word.into_chars();
1992 ///
1993 /// assert_eq!(Some('g'), chars.next());
1994 /// assert_eq!(Some('o'), chars.next());
1995 /// assert_eq!(Some('o'), chars.next());
1996 /// assert_eq!(Some('d'), chars.next());
1997 /// assert_eq!(Some('b'), chars.next());
1998 /// assert_eq!(Some('y'), chars.next());
1999 /// assert_eq!(Some('e'), chars.next());
2000 ///
2001 /// assert_eq!(None, chars.next());
2002 /// ```
2003 ///
2004 /// Remember, [`char`]s might not match your intuition about characters:
2005 ///
2006 /// ```
2007 /// #![feature(string_into_chars)]
2008 ///
2009 /// let y = String::from("y̆");
2010 ///
2011 /// let mut chars = y.into_chars();
2012 ///
2013 /// assert_eq!(Some('y'), chars.next()); // not 'y̆'
2014 /// assert_eq!(Some('\u{0306}'), chars.next());
2015 ///
2016 /// assert_eq!(None, chars.next());
2017 /// ```
2018 ///
2019 /// [`char`]: prim@char
2020 #[inline]
2021 #[must_use = "`self` will be dropped if the result is not used"]
2022 #[unstable(feature = "string_into_chars", issue = "133125")]
2023 pub fn into_chars(self) -> IntoChars {
2024 IntoChars { bytes: self.into_bytes().into_iter() }
2025 }
2026
2027 /// Removes the specified range in the string,
2028 /// and replaces it with the given string.
2029 /// The given string doesn't need to be the same length as the range.
2030 ///
2031 /// # Panics
2032 ///
2033 /// Panics if the starting point or end point do not lie on a [`char`]
2034 /// boundary, or if they're out of bounds.
2035 ///
2036 /// # Examples
2037 ///
2038 /// ```
2039 /// let mut s = String::from("Ξ± is alpha, Ξ² is beta");
2040 /// let beta_offset = s.find('Ξ²').unwrap_or(s.len());
2041 ///
2042 /// // Replace the range up until the Ξ² from the string
2043 /// s.replace_range(..beta_offset, "Ξ‘ is capital alpha; ");
2044 /// assert_eq!(s, "Ξ‘ is capital alpha; Ξ² is beta");
2045 /// ```
2046 #[cfg(not(no_global_oom_handling))]
2047 #[stable(feature = "splice", since = "1.27.0")]
2048 pub fn replace_range<R>(&mut self, range: R, replace_with: &str)
2049 where
2050 R: RangeBounds<usize>,
2051 {
2052 // Memory safety
2053 //
2054 // Replace_range does not have the memory safety issues of a vector Splice.
2055 // of the vector version. The data is just plain bytes.
2056
2057 // WARNING: Inlining this variable would be unsound (#81138)
2058 let start = range.start_bound();
2059 match start {
2060 Included(&n) => assert!(self.is_char_boundary(n)),
2061 Excluded(&n) => assert!(self.is_char_boundary(n + 1)),
2062 Unbounded => {}
2063 };
2064 // WARNING: Inlining this variable would be unsound (#81138)
2065 let end = range.end_bound();
2066 match end {
2067 Included(&n) => assert!(self.is_char_boundary(n + 1)),
2068 Excluded(&n) => assert!(self.is_char_boundary(n)),
2069 Unbounded => {}
2070 };
2071
2072 // Using `range` again would be unsound (#81138)
2073 // We assume the bounds reported by `range` remain the same, but
2074 // an adversarial implementation could change between calls
2075 unsafe { self.as_mut_vec() }.splice((start, end), replace_with.bytes());
2076 }
2077
2078 /// Converts this `String` into a <code>[Box]<[str]></code>.
2079 ///
2080 /// Before doing the conversion, this method discards excess capacity like [`shrink_to_fit`].
2081 /// Note that this call may reallocate and copy the bytes of the string.
2082 ///
2083 /// [`shrink_to_fit`]: String::shrink_to_fit
2084 /// [str]: prim@str "str"
2085 ///
2086 /// # Examples
2087 ///
2088 /// ```
2089 /// let s = String::from("hello");
2090 ///
2091 /// let b = s.into_boxed_str();
2092 /// ```
2093 #[cfg(not(no_global_oom_handling))]
2094 #[stable(feature = "box_str", since = "1.4.0")]
2095 #[must_use = "`self` will be dropped if the result is not used"]
2096 #[inline]
2097 pub fn into_boxed_str(self) -> Box<str> {
2098 let slice = self.vec.into_boxed_slice();
2099 unsafe { from_boxed_utf8_unchecked(slice) }
2100 }
2101
2102 /// Consumes and leaks the `String`, returning a mutable reference to the contents,
2103 /// `&'a mut str`.
2104 ///
2105 /// The caller has free choice over the returned lifetime, including `'static`. Indeed,
2106 /// this function is ideally used for data that lives for the remainder of the program's life,
2107 /// as dropping the returned reference will cause a memory leak.
2108 ///
2109 /// It does not reallocate or shrink the `String`, so the leaked allocation may include unused
2110 /// capacity that is not part of the returned slice. If you want to discard excess capacity,
2111 /// call [`into_boxed_str`], and then [`Box::leak`] instead. However, keep in mind that
2112 /// trimming the capacity may result in a reallocation and copy.
2113 ///
2114 /// [`into_boxed_str`]: Self::into_boxed_str
2115 ///
2116 /// # Examples
2117 ///
2118 /// ```
2119 /// let x = String::from("bucket");
2120 /// let static_ref: &'static mut str = x.leak();
2121 /// assert_eq!(static_ref, "bucket");
2122 /// # // FIXME(https://github.com/rust-lang/miri/issues/3670):
2123 /// # // use -Zmiri-disable-leak-check instead of unleaking in tests meant to leak.
2124 /// # drop(unsafe { Box::from_raw(static_ref) });
2125 /// ```
2126 #[stable(feature = "string_leak", since = "1.72.0")]
2127 #[inline]
2128 pub fn leak<'a>(self) -> &'a mut str {
2129 let slice = self.vec.leak();
2130 unsafe { from_utf8_unchecked_mut(slice) }
2131 }
2132}
2133
2134impl FromUtf8Error {
2135 /// Returns a slice of [`u8`]s bytes that were attempted to convert to a `String`.
2136 ///
2137 /// # Examples
2138 ///
2139 /// ```
2140 /// // some invalid bytes, in a vector
2141 /// let bytes = vec![0, 159];
2142 ///
2143 /// let value = String::from_utf8(bytes);
2144 ///
2145 /// assert_eq!(&[0, 159], value.unwrap_err().as_bytes());
2146 /// ```
2147 #[must_use]
2148 #[stable(feature = "from_utf8_error_as_bytes", since = "1.26.0")]
2149 pub fn as_bytes(&self) -> &[u8] {
2150 &self.bytes[..]
2151 }
2152
2153 /// Converts the bytes into a `String` lossily, substituting invalid UTF-8
2154 /// sequences with replacement characters.
2155 ///
2156 /// See [`String::from_utf8_lossy`] for more details on replacement of
2157 /// invalid sequences, and [`String::from_utf8_lossy_owned`] for the
2158 /// `String` function which corresponds to this function.
2159 ///
2160 /// # Examples
2161 ///
2162 /// ```
2163 /// #![feature(string_from_utf8_lossy_owned)]
2164 /// // some invalid bytes
2165 /// let input: Vec<u8> = b"Hello \xF0\x90\x80World".into();
2166 /// let output = String::from_utf8(input).unwrap_or_else(|e| e.into_utf8_lossy());
2167 ///
2168 /// assert_eq!(String::from("Hello οΏ½World"), output);
2169 /// ```
2170 #[must_use]
2171 #[cfg(not(no_global_oom_handling))]
2172 #[unstable(feature = "string_from_utf8_lossy_owned", issue = "129436")]
2173 pub fn into_utf8_lossy(self) -> String {
2174 const REPLACEMENT: &str = "\u{FFFD}";
2175
2176 let mut res = {
2177 let mut v = Vec::with_capacity(self.bytes.len());
2178
2179 // `Utf8Error::valid_up_to` returns the maximum index of validated
2180 // UTF-8 bytes. Copy the valid bytes into the output buffer.
2181 v.extend_from_slice(&self.bytes[..self.error.valid_up_to()]);
2182
2183 // SAFETY: This is safe because the only bytes present in the buffer
2184 // were validated as UTF-8 by the call to `String::from_utf8` which
2185 // produced this `FromUtf8Error`.
2186 unsafe { String::from_utf8_unchecked(v) }
2187 };
2188
2189 let iter = self.bytes[self.error.valid_up_to()..].utf8_chunks();
2190
2191 for chunk in iter {
2192 res.push_str(chunk.valid());
2193 if !chunk.invalid().is_empty() {
2194 res.push_str(REPLACEMENT);
2195 }
2196 }
2197
2198 res
2199 }
2200
2201 /// Returns the bytes that were attempted to convert to a `String`.
2202 ///
2203 /// This method is carefully constructed to avoid allocation. It will
2204 /// consume the error, moving out the bytes, so that a copy of the bytes
2205 /// does not need to be made.
2206 ///
2207 /// # Examples
2208 ///
2209 /// ```
2210 /// // some invalid bytes, in a vector
2211 /// let bytes = vec![0, 159];
2212 ///
2213 /// let value = String::from_utf8(bytes);
2214 ///
2215 /// assert_eq!(vec![0, 159], value.unwrap_err().into_bytes());
2216 /// ```
2217 #[must_use = "`self` will be dropped if the result is not used"]
2218 #[stable(feature = "rust1", since = "1.0.0")]
2219 pub fn into_bytes(self) -> Vec<u8> {
2220 self.bytes
2221 }
2222
2223 /// Fetch a `Utf8Error` to get more details about the conversion failure.
2224 ///
2225 /// The [`Utf8Error`] type provided by [`std::str`] represents an error that may
2226 /// occur when converting a slice of [`u8`]s to a [`&str`]. In this sense, it's
2227 /// an analogue to `FromUtf8Error`. See its documentation for more details
2228 /// on using it.
2229 ///
2230 /// [`std::str`]: core::str "std::str"
2231 /// [`&str`]: prim@str "&str"
2232 ///
2233 /// # Examples
2234 ///
2235 /// ```
2236 /// // some invalid bytes, in a vector
2237 /// let bytes = vec![0, 159];
2238 ///
2239 /// let error = String::from_utf8(bytes).unwrap_err().utf8_error();
2240 ///
2241 /// // the first byte is invalid here
2242 /// assert_eq!(1, error.valid_up_to());
2243 /// ```
2244 #[must_use]
2245 #[stable(feature = "rust1", since = "1.0.0")]
2246 pub fn utf8_error(&self) -> Utf8Error {
2247 self.error
2248 }
2249}
2250
2251#[stable(feature = "rust1", since = "1.0.0")]
2252impl fmt::Display for FromUtf8Error {
2253 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2254 fmt::Display::fmt(&self.error, f)
2255 }
2256}
2257
2258#[stable(feature = "rust1", since = "1.0.0")]
2259impl fmt::Display for FromUtf16Error {
2260 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2261 fmt::Display::fmt(self:"invalid utf-16: lone surrogate found", f)
2262 }
2263}
2264
2265#[stable(feature = "rust1", since = "1.0.0")]
2266impl Error for FromUtf8Error {
2267 #[allow(deprecated)]
2268 fn description(&self) -> &str {
2269 "invalid utf-8"
2270 }
2271}
2272
2273#[stable(feature = "rust1", since = "1.0.0")]
2274impl Error for FromUtf16Error {
2275 #[allow(deprecated)]
2276 fn description(&self) -> &str {
2277 "invalid utf-16"
2278 }
2279}
2280
2281#[cfg(not(no_global_oom_handling))]
2282#[stable(feature = "rust1", since = "1.0.0")]
2283impl Clone for String {
2284 fn clone(&self) -> Self {
2285 String { vec: self.vec.clone() }
2286 }
2287
2288 /// Clones the contents of `source` into `self`.
2289 ///
2290 /// This method is preferred over simply assigning `source.clone()` to `self`,
2291 /// as it avoids reallocation if possible.
2292 fn clone_from(&mut self, source: &Self) {
2293 self.vec.clone_from(&source.vec);
2294 }
2295}
2296
2297#[cfg(not(no_global_oom_handling))]
2298#[stable(feature = "rust1", since = "1.0.0")]
2299impl FromIterator<char> for String {
2300 fn from_iter<I: IntoIterator<Item = char>>(iter: I) -> String {
2301 let mut buf: String = String::new();
2302 buf.extend(iter);
2303 buf
2304 }
2305}
2306
2307#[cfg(not(no_global_oom_handling))]
2308#[stable(feature = "string_from_iter_by_ref", since = "1.17.0")]
2309impl<'a> FromIterator<&'a char> for String {
2310 fn from_iter<I: IntoIterator<Item = &'a char>>(iter: I) -> String {
2311 let mut buf: String = String::new();
2312 buf.extend(iter);
2313 buf
2314 }
2315}
2316
2317#[cfg(not(no_global_oom_handling))]
2318#[stable(feature = "rust1", since = "1.0.0")]
2319impl<'a> FromIterator<&'a str> for String {
2320 fn from_iter<I: IntoIterator<Item = &'a str>>(iter: I) -> String {
2321 let mut buf: String = String::new();
2322 buf.extend(iter);
2323 buf
2324 }
2325}
2326
2327#[cfg(not(no_global_oom_handling))]
2328#[stable(feature = "extend_string", since = "1.4.0")]
2329impl FromIterator<String> for String {
2330 fn from_iter<I: IntoIterator<Item = String>>(iter: I) -> String {
2331 let mut iterator: ::IntoIter = iter.into_iter();
2332
2333 // Because we're iterating over `String`s, we can avoid at least
2334 // one allocation by getting the first string from the iterator
2335 // and appending to it all the subsequent strings.
2336 match iterator.next() {
2337 None => String::new(),
2338 Some(mut buf: String) => {
2339 buf.extend(iter:iterator);
2340 buf
2341 }
2342 }
2343 }
2344}
2345
2346#[cfg(not(no_global_oom_handling))]
2347#[stable(feature = "box_str2", since = "1.45.0")]
2348impl<A: Allocator> FromIterator<Box<str, A>> for String {
2349 fn from_iter<I: IntoIterator<Item = Box<str, A>>>(iter: I) -> String {
2350 let mut buf: String = String::new();
2351 buf.extend(iter);
2352 buf
2353 }
2354}
2355
2356#[cfg(not(no_global_oom_handling))]
2357#[stable(feature = "herd_cows", since = "1.19.0")]
2358impl<'a> FromIterator<Cow<'a, str>> for String {
2359 fn from_iter<I: IntoIterator<Item = Cow<'a, str>>>(iter: I) -> String {
2360 let mut iterator: ::IntoIter = iter.into_iter();
2361
2362 // Because we're iterating over CoWs, we can (potentially) avoid at least
2363 // one allocation by getting the first item and appending to it all the
2364 // subsequent items.
2365 match iterator.next() {
2366 None => String::new(),
2367 Some(cow: Cow<'a, str>) => {
2368 let mut buf: String = cow.into_owned();
2369 buf.extend(iter:iterator);
2370 buf
2371 }
2372 }
2373 }
2374}
2375
2376#[cfg(not(no_global_oom_handling))]
2377#[stable(feature = "rust1", since = "1.0.0")]
2378impl Extend<char> for String {
2379 fn extend<I: IntoIterator<Item = char>>(&mut self, iter: I) {
2380 let iterator: ::IntoIter = iter.into_iter();
2381 let (lower_bound: usize, _) = iterator.size_hint();
2382 self.reserve(additional:lower_bound);
2383 iterator.for_each(move |c: char| self.push(ch:c));
2384 }
2385
2386 #[inline]
2387 fn extend_one(&mut self, c: char) {
2388 self.push(ch:c);
2389 }
2390
2391 #[inline]
2392 fn extend_reserve(&mut self, additional: usize) {
2393 self.reserve(additional);
2394 }
2395}
2396
2397#[cfg(not(no_global_oom_handling))]
2398#[stable(feature = "extend_ref", since = "1.2.0")]
2399impl<'a> Extend<&'a char> for String {
2400 fn extend<I: IntoIterator<Item = &'a char>>(&mut self, iter: I) {
2401 self.extend(iter.into_iter().cloned());
2402 }
2403
2404 #[inline]
2405 fn extend_one(&mut self, &c: char: &'a char) {
2406 self.push(ch:c);
2407 }
2408
2409 #[inline]
2410 fn extend_reserve(&mut self, additional: usize) {
2411 self.reserve(additional);
2412 }
2413}
2414
2415#[cfg(not(no_global_oom_handling))]
2416#[stable(feature = "rust1", since = "1.0.0")]
2417impl<'a> Extend<&'a str> for String {
2418 fn extend<I: IntoIterator<Item = &'a str>>(&mut self, iter: I) {
2419 iter.into_iter().for_each(move |s: &'a str| self.push_str(string:s));
2420 }
2421
2422 #[inline]
2423 fn extend_one(&mut self, s: &'a str) {
2424 self.push_str(string:s);
2425 }
2426}
2427
2428#[cfg(not(no_global_oom_handling))]
2429#[stable(feature = "box_str2", since = "1.45.0")]
2430impl<A: Allocator> Extend<Box<str, A>> for String {
2431 fn extend<I: IntoIterator<Item = Box<str, A>>>(&mut self, iter: I) {
2432 iter.into_iter().for_each(move |s: Box| self.push_str(&s));
2433 }
2434}
2435
2436#[cfg(not(no_global_oom_handling))]
2437#[stable(feature = "extend_string", since = "1.4.0")]
2438impl Extend<String> for String {
2439 fn extend<I: IntoIterator<Item = String>>(&mut self, iter: I) {
2440 iter.into_iter().for_each(move |s: String| self.push_str(&s));
2441 }
2442
2443 #[inline]
2444 fn extend_one(&mut self, s: String) {
2445 self.push_str(&s);
2446 }
2447}
2448
2449#[cfg(not(no_global_oom_handling))]
2450#[stable(feature = "herd_cows", since = "1.19.0")]
2451impl<'a> Extend<Cow<'a, str>> for String {
2452 fn extend<I: IntoIterator<Item = Cow<'a, str>>>(&mut self, iter: I) {
2453 iter.into_iter().for_each(move |s: Cow<'a, str>| self.push_str(&s));
2454 }
2455
2456 #[inline]
2457 fn extend_one(&mut self, s: Cow<'a, str>) {
2458 self.push_str(&s);
2459 }
2460}
2461
2462#[cfg(not(no_global_oom_handling))]
2463#[unstable(feature = "ascii_char", issue = "110998")]
2464impl Extend<core::ascii::Char> for String {
2465 fn extend<I: IntoIterator<Item = core::ascii::Char>>(&mut self, iter: I) {
2466 self.vec.extend(iter.into_iter().map(|c: AsciiChar| c.to_u8()));
2467 }
2468
2469 #[inline]
2470 fn extend_one(&mut self, c: core::ascii::Char) {
2471 self.vec.push(c.to_u8());
2472 }
2473}
2474
2475#[cfg(not(no_global_oom_handling))]
2476#[unstable(feature = "ascii_char", issue = "110998")]
2477impl<'a> Extend<&'a core::ascii::Char> for String {
2478 fn extend<I: IntoIterator<Item = &'a core::ascii::Char>>(&mut self, iter: I) {
2479 self.extend(iter.into_iter().cloned());
2480 }
2481
2482 #[inline]
2483 fn extend_one(&mut self, c: &'a core::ascii::Char) {
2484 self.vec.push(c.to_u8());
2485 }
2486}
2487
2488/// A convenience impl that delegates to the impl for `&str`.
2489///
2490/// # Examples
2491///
2492/// ```
2493/// assert_eq!(String::from("Hello world").find("world"), Some(6));
2494/// ```
2495#[unstable(
2496 feature = "pattern",
2497 reason = "API not fully fleshed out and ready to be stabilized",
2498 issue = "27721"
2499)]
2500impl<'b> Pattern for &'b String {
2501 type Searcher<'a> = <&'b str as Pattern>::Searcher<'a>;
2502
2503 fn into_searcher(self, haystack: &str) -> <&'b str as Pattern>::Searcher<'_> {
2504 self[..].into_searcher(haystack)
2505 }
2506
2507 #[inline]
2508 fn is_contained_in(self, haystack: &str) -> bool {
2509 self[..].is_contained_in(haystack)
2510 }
2511
2512 #[inline]
2513 fn is_prefix_of(self, haystack: &str) -> bool {
2514 self[..].is_prefix_of(haystack)
2515 }
2516
2517 #[inline]
2518 fn strip_prefix_of(self, haystack: &str) -> Option<&str> {
2519 self[..].strip_prefix_of(haystack)
2520 }
2521
2522 #[inline]
2523 fn is_suffix_of<'a>(self, haystack: &'a str) -> bool
2524 where
2525 Self::Searcher<'a>: core::str::pattern::ReverseSearcher<'a>,
2526 {
2527 self[..].is_suffix_of(haystack)
2528 }
2529
2530 #[inline]
2531 fn strip_suffix_of<'a>(self, haystack: &'a str) -> Option<&'a str>
2532 where
2533 Self::Searcher<'a>: core::str::pattern::ReverseSearcher<'a>,
2534 {
2535 self[..].strip_suffix_of(haystack)
2536 }
2537
2538 #[inline]
2539 fn as_utf8_pattern(&self) -> Option<Utf8Pattern<'_>> {
2540 Some(Utf8Pattern::StringPattern(self.as_bytes()))
2541 }
2542}
2543
2544macro_rules! impl_eq {
2545 ($lhs:ty, $rhs: ty) => {
2546 #[stable(feature = "rust1", since = "1.0.0")]
2547 #[allow(unused_lifetimes)]
2548 impl<'a, 'b> PartialEq<$rhs> for $lhs {
2549 #[inline]
2550 fn eq(&self, other: &$rhs) -> bool {
2551 PartialEq::eq(&self[..], &other[..])
2552 }
2553 #[inline]
2554 fn ne(&self, other: &$rhs) -> bool {
2555 PartialEq::ne(&self[..], &other[..])
2556 }
2557 }
2558
2559 #[stable(feature = "rust1", since = "1.0.0")]
2560 #[allow(unused_lifetimes)]
2561 impl<'a, 'b> PartialEq<$lhs> for $rhs {
2562 #[inline]
2563 fn eq(&self, other: &$lhs) -> bool {
2564 PartialEq::eq(&self[..], &other[..])
2565 }
2566 #[inline]
2567 fn ne(&self, other: &$lhs) -> bool {
2568 PartialEq::ne(&self[..], &other[..])
2569 }
2570 }
2571 };
2572}
2573
2574impl_eq! { String, str }
2575impl_eq! { String, &'a str }
2576#[cfg(not(no_global_oom_handling))]
2577impl_eq! { Cow<'a, str>, str }
2578#[cfg(not(no_global_oom_handling))]
2579impl_eq! { Cow<'a, str>, &'b str }
2580#[cfg(not(no_global_oom_handling))]
2581impl_eq! { Cow<'a, str>, String }
2582
2583#[stable(feature = "rust1", since = "1.0.0")]
2584impl Default for String {
2585 /// Creates an empty `String`.
2586 #[inline]
2587 fn default() -> String {
2588 String::new()
2589 }
2590}
2591
2592#[stable(feature = "rust1", since = "1.0.0")]
2593impl fmt::Display for String {
2594 #[inline]
2595 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2596 fmt::Display::fmt(&**self, f)
2597 }
2598}
2599
2600#[stable(feature = "rust1", since = "1.0.0")]
2601impl fmt::Debug for String {
2602 #[inline]
2603 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
2604 fmt::Debug::fmt(&**self, f)
2605 }
2606}
2607
2608#[stable(feature = "rust1", since = "1.0.0")]
2609impl hash::Hash for String {
2610 #[inline]
2611 fn hash<H: hash::Hasher>(&self, hasher: &mut H) {
2612 (**self).hash(state:hasher)
2613 }
2614}
2615
2616/// Implements the `+` operator for concatenating two strings.
2617///
2618/// This consumes the `String` on the left-hand side and re-uses its buffer (growing it if
2619/// necessary). This is done to avoid allocating a new `String` and copying the entire contents on
2620/// every operation, which would lead to *O*(*n*^2) running time when building an *n*-byte string by
2621/// repeated concatenation.
2622///
2623/// The string on the right-hand side is only borrowed; its contents are copied into the returned
2624/// `String`.
2625///
2626/// # Examples
2627///
2628/// Concatenating two `String`s takes the first by value and borrows the second:
2629///
2630/// ```
2631/// let a = String::from("hello");
2632/// let b = String::from(" world");
2633/// let c = a + &b;
2634/// // `a` is moved and can no longer be used here.
2635/// ```
2636///
2637/// If you want to keep using the first `String`, you can clone it and append to the clone instead:
2638///
2639/// ```
2640/// let a = String::from("hello");
2641/// let b = String::from(" world");
2642/// let c = a.clone() + &b;
2643/// // `a` is still valid here.
2644/// ```
2645///
2646/// Concatenating `&str` slices can be done by converting the first to a `String`:
2647///
2648/// ```
2649/// let a = "hello";
2650/// let b = " world";
2651/// let c = a.to_string() + b;
2652/// ```
2653#[cfg(not(no_global_oom_handling))]
2654#[stable(feature = "rust1", since = "1.0.0")]
2655impl Add<&str> for String {
2656 type Output = String;
2657
2658 #[inline]
2659 fn add(mut self, other: &str) -> String {
2660 self.push_str(string:other);
2661 self
2662 }
2663}
2664
2665/// Implements the `+=` operator for appending to a `String`.
2666///
2667/// This has the same behavior as the [`push_str`][String::push_str] method.
2668#[cfg(not(no_global_oom_handling))]
2669#[stable(feature = "stringaddassign", since = "1.12.0")]
2670impl AddAssign<&str> for String {
2671 #[inline]
2672 fn add_assign(&mut self, other: &str) {
2673 self.push_str(string:other);
2674 }
2675}
2676
2677#[stable(feature = "rust1", since = "1.0.0")]
2678impl<I> ops::Index<I> for String
2679where
2680 I: slice::SliceIndex<str>,
2681{
2682 type Output = I::Output;
2683
2684 #[inline]
2685 fn index(&self, index: I) -> &I::Output {
2686 index.index(self.as_str())
2687 }
2688}
2689
2690#[stable(feature = "rust1", since = "1.0.0")]
2691impl<I> ops::IndexMut<I> for String
2692where
2693 I: slice::SliceIndex<str>,
2694{
2695 #[inline]
2696 fn index_mut(&mut self, index: I) -> &mut I::Output {
2697 index.index_mut(self.as_mut_str())
2698 }
2699}
2700
2701#[stable(feature = "rust1", since = "1.0.0")]
2702impl ops::Deref for String {
2703 type Target = str;
2704
2705 #[inline]
2706 fn deref(&self) -> &str {
2707 self.as_str()
2708 }
2709}
2710
2711#[unstable(feature = "deref_pure_trait", issue = "87121")]
2712unsafe impl ops::DerefPure for String {}
2713
2714#[stable(feature = "derefmut_for_string", since = "1.3.0")]
2715impl ops::DerefMut for String {
2716 #[inline]
2717 fn deref_mut(&mut self) -> &mut str {
2718 self.as_mut_str()
2719 }
2720}
2721
2722/// A type alias for [`Infallible`].
2723///
2724/// This alias exists for backwards compatibility, and may be eventually deprecated.
2725///
2726/// [`Infallible`]: core::convert::Infallible "convert::Infallible"
2727#[stable(feature = "str_parse_error", since = "1.5.0")]
2728pub type ParseError = core::convert::Infallible;
2729
2730#[cfg(not(no_global_oom_handling))]
2731#[stable(feature = "rust1", since = "1.0.0")]
2732impl FromStr for String {
2733 type Err = core::convert::Infallible;
2734 #[inline]
2735 fn from_str(s: &str) -> Result<String, Self::Err> {
2736 Ok(String::from(s))
2737 }
2738}
2739
2740/// A trait for converting a value to a `String`.
2741///
2742/// This trait is automatically implemented for any type which implements the
2743/// [`Display`] trait. As such, `ToString` shouldn't be implemented directly:
2744/// [`Display`] should be implemented instead, and you get the `ToString`
2745/// implementation for free.
2746///
2747/// [`Display`]: fmt::Display
2748#[rustc_diagnostic_item = "ToString"]
2749#[stable(feature = "rust1", since = "1.0.0")]
2750pub trait ToString {
2751 /// Converts the given value to a `String`.
2752 ///
2753 /// # Examples
2754 ///
2755 /// ```
2756 /// let i = 5;
2757 /// let five = String::from("5");
2758 ///
2759 /// assert_eq!(five, i.to_string());
2760 /// ```
2761 #[rustc_conversion_suggestion]
2762 #[stable(feature = "rust1", since = "1.0.0")]
2763 #[rustc_diagnostic_item = "to_string_method"]
2764 fn to_string(&self) -> String;
2765}
2766
2767/// # Panics
2768///
2769/// In this implementation, the `to_string` method panics
2770/// if the `Display` implementation returns an error.
2771/// This indicates an incorrect `Display` implementation
2772/// since `fmt::Write for String` never returns an error itself.
2773#[cfg(not(no_global_oom_handling))]
2774#[stable(feature = "rust1", since = "1.0.0")]
2775impl<T: fmt::Display + ?Sized> ToString for T {
2776 #[inline]
2777 fn to_string(&self) -> String {
2778 <Self as SpecToString>::spec_to_string(self)
2779 }
2780}
2781
2782#[cfg(not(no_global_oom_handling))]
2783trait SpecToString {
2784 fn spec_to_string(&self) -> String;
2785}
2786
2787#[cfg(not(no_global_oom_handling))]
2788impl<T: fmt::Display + ?Sized> SpecToString for T {
2789 // A common guideline is to not inline generic functions. However,
2790 // removing `#[inline]` from this method causes non-negligible regressions.
2791 // See <https://github.com/rust-lang/rust/pull/74852>, the last attempt
2792 // to try to remove it.
2793 #[inline]
2794 default fn spec_to_string(&self) -> String {
2795 let mut buf: String = String::new();
2796 let mut formatter: Formatter<'_> =
2797 core::fmt::Formatter::new(&mut buf, options:core::fmt::FormattingOptions::new());
2798 // Bypass format_args!() to avoid write_str with zero-length strs
2799 fmt::Display::fmt(self, &mut formatter)
2800 .expect(msg:"a Display implementation returned an error unexpectedly");
2801 buf
2802 }
2803}
2804
2805#[cfg(not(no_global_oom_handling))]
2806impl SpecToString for core::ascii::Char {
2807 #[inline]
2808 fn spec_to_string(&self) -> String {
2809 self.as_str().to_owned()
2810 }
2811}
2812
2813#[cfg(not(no_global_oom_handling))]
2814impl SpecToString for char {
2815 #[inline]
2816 fn spec_to_string(&self) -> String {
2817 String::from(self.encode_utf8(&mut [0; char::MAX_LEN_UTF8]))
2818 }
2819}
2820
2821#[cfg(not(no_global_oom_handling))]
2822impl SpecToString for bool {
2823 #[inline]
2824 fn spec_to_string(&self) -> String {
2825 String::from(if *self { "true" } else { "false" })
2826 }
2827}
2828
2829macro_rules! impl_to_string {
2830 ($($signed:ident, $unsigned:ident,)*) => {
2831 $(
2832 #[cfg(not(no_global_oom_handling))]
2833 #[cfg(not(feature = "optimize_for_size"))]
2834 impl SpecToString for $signed {
2835 #[inline]
2836 fn spec_to_string(&self) -> String {
2837 const SIZE: usize = $signed::MAX.ilog(10) as usize + 1;
2838 let mut buf = [core::mem::MaybeUninit::<u8>::uninit(); SIZE];
2839 // Only difference between signed and unsigned are these 8 lines.
2840 let mut out;
2841 if *self < 0 {
2842 out = String::with_capacity(SIZE + 1);
2843 out.push('-');
2844 } else {
2845 out = String::with_capacity(SIZE);
2846 }
2847
2848 out.push_str(self.unsigned_abs()._fmt(&mut buf));
2849 out
2850 }
2851 }
2852 #[cfg(not(no_global_oom_handling))]
2853 #[cfg(not(feature = "optimize_for_size"))]
2854 impl SpecToString for $unsigned {
2855 #[inline]
2856 fn spec_to_string(&self) -> String {
2857 const SIZE: usize = $unsigned::MAX.ilog(10) as usize + 1;
2858 let mut buf = [core::mem::MaybeUninit::<u8>::uninit(); SIZE];
2859
2860 self._fmt(&mut buf).to_string()
2861 }
2862 }
2863 )*
2864 }
2865}
2866
2867impl_to_string! {
2868 i8, u8,
2869 i16, u16,
2870 i32, u32,
2871 i64, u64,
2872 isize, usize,
2873}
2874
2875#[cfg(not(no_global_oom_handling))]
2876#[cfg(feature = "optimize_for_size")]
2877impl SpecToString for u8 {
2878 #[inline]
2879 fn spec_to_string(&self) -> String {
2880 let mut buf = String::with_capacity(3);
2881 let mut n = *self;
2882 if n >= 10 {
2883 if n >= 100 {
2884 buf.push((b'0' + n / 100) as char);
2885 n %= 100;
2886 }
2887 buf.push((b'0' + n / 10) as char);
2888 n %= 10;
2889 }
2890 buf.push((b'0' + n) as char);
2891 buf
2892 }
2893}
2894
2895#[cfg(not(no_global_oom_handling))]
2896#[cfg(feature = "optimize_for_size")]
2897impl SpecToString for i8 {
2898 #[inline]
2899 fn spec_to_string(&self) -> String {
2900 let mut buf = String::with_capacity(4);
2901 if self.is_negative() {
2902 buf.push('-');
2903 }
2904 let mut n = self.unsigned_abs();
2905 if n >= 10 {
2906 if n >= 100 {
2907 buf.push('1');
2908 n -= 100;
2909 }
2910 buf.push((b'0' + n / 10) as char);
2911 n %= 10;
2912 }
2913 buf.push((b'0' + n) as char);
2914 buf
2915 }
2916}
2917
2918// Generic/generated code can sometimes have multiple, nested references
2919// for strings, including `&&&str`s that would never be written
2920// by hand. This macro generates twelve layers of nested `&`-impl
2921// for primitive strings.
2922#[cfg(not(no_global_oom_handling))]
2923macro_rules! to_string_str_wrap_in_ref {
2924 {x $($x:ident)*} => {
2925 &to_string_str_wrap_in_ref! { $($x)* }
2926 };
2927 {} => { str };
2928}
2929#[cfg(not(no_global_oom_handling))]
2930macro_rules! to_string_expr_wrap_in_deref {
2931 {$self:expr ; x $($x:ident)*} => {
2932 *(to_string_expr_wrap_in_deref! { $self ; $($x)* })
2933 };
2934 {$self:expr ;} => { $self };
2935}
2936#[cfg(not(no_global_oom_handling))]
2937macro_rules! to_string_str {
2938 {$($($x:ident)*),+} => {
2939 $(
2940 impl SpecToString for to_string_str_wrap_in_ref!($($x)*) {
2941 #[inline]
2942 fn spec_to_string(&self) -> String {
2943 String::from(to_string_expr_wrap_in_deref!(self ; $($x)*))
2944 }
2945 }
2946 )+
2947 };
2948}
2949
2950#[cfg(not(no_global_oom_handling))]
2951to_string_str! {
2952 x x x x x x x x x x x x,
2953 x x x x x x x x x x x,
2954 x x x x x x x x x x,
2955 x x x x x x x x x,
2956 x x x x x x x x,
2957 x x x x x x x,
2958 x x x x x x,
2959 x x x x x,
2960 x x x x,
2961 x x x,
2962 x x,
2963 x,
2964}
2965
2966#[cfg(not(no_global_oom_handling))]
2967impl SpecToString for Cow<'_, str> {
2968 #[inline]
2969 fn spec_to_string(&self) -> String {
2970 self[..].to_owned()
2971 }
2972}
2973
2974#[cfg(not(no_global_oom_handling))]
2975impl SpecToString for String {
2976 #[inline]
2977 fn spec_to_string(&self) -> String {
2978 self.to_owned()
2979 }
2980}
2981
2982#[cfg(not(no_global_oom_handling))]
2983impl SpecToString for fmt::Arguments<'_> {
2984 #[inline]
2985 fn spec_to_string(&self) -> String {
2986 crate::fmt::format(*self)
2987 }
2988}
2989
2990#[stable(feature = "rust1", since = "1.0.0")]
2991impl AsRef<str> for String {
2992 #[inline]
2993 fn as_ref(&self) -> &str {
2994 self
2995 }
2996}
2997
2998#[stable(feature = "string_as_mut", since = "1.43.0")]
2999impl AsMut<str> for String {
3000 #[inline]
3001 fn as_mut(&mut self) -> &mut str {
3002 self
3003 }
3004}
3005
3006#[stable(feature = "rust1", since = "1.0.0")]
3007impl AsRef<[u8]> for String {
3008 #[inline]
3009 fn as_ref(&self) -> &[u8] {
3010 self.as_bytes()
3011 }
3012}
3013
3014#[cfg(not(no_global_oom_handling))]
3015#[stable(feature = "rust1", since = "1.0.0")]
3016impl From<&str> for String {
3017 /// Converts a `&str` into a [`String`].
3018 ///
3019 /// The result is allocated on the heap.
3020 #[inline]
3021 fn from(s: &str) -> String {
3022 s.to_owned()
3023 }
3024}
3025
3026#[cfg(not(no_global_oom_handling))]
3027#[stable(feature = "from_mut_str_for_string", since = "1.44.0")]
3028impl From<&mut str> for String {
3029 /// Converts a `&mut str` into a [`String`].
3030 ///
3031 /// The result is allocated on the heap.
3032 #[inline]
3033 fn from(s: &mut str) -> String {
3034 s.to_owned()
3035 }
3036}
3037
3038#[cfg(not(no_global_oom_handling))]
3039#[stable(feature = "from_ref_string", since = "1.35.0")]
3040impl From<&String> for String {
3041 /// Converts a `&String` into a [`String`].
3042 ///
3043 /// This clones `s` and returns the clone.
3044 #[inline]
3045 fn from(s: &String) -> String {
3046 s.clone()
3047 }
3048}
3049
3050// note: test pulls in std, which causes errors here
3051#[stable(feature = "string_from_box", since = "1.18.0")]
3052impl From<Box<str>> for String {
3053 /// Converts the given boxed `str` slice to a [`String`].
3054 /// It is notable that the `str` slice is owned.
3055 ///
3056 /// # Examples
3057 ///
3058 /// ```
3059 /// let s1: String = String::from("hello world");
3060 /// let s2: Box<str> = s1.into_boxed_str();
3061 /// let s3: String = String::from(s2);
3062 ///
3063 /// assert_eq!("hello world", s3)
3064 /// ```
3065 fn from(s: Box<str>) -> String {
3066 s.into_string()
3067 }
3068}
3069
3070#[cfg(not(no_global_oom_handling))]
3071#[stable(feature = "box_from_str", since = "1.20.0")]
3072impl From<String> for Box<str> {
3073 /// Converts the given [`String`] to a boxed `str` slice that is owned.
3074 ///
3075 /// # Examples
3076 ///
3077 /// ```
3078 /// let s1: String = String::from("hello world");
3079 /// let s2: Box<str> = Box::from(s1);
3080 /// let s3: String = String::from(s2);
3081 ///
3082 /// assert_eq!("hello world", s3)
3083 /// ```
3084 fn from(s: String) -> Box<str> {
3085 s.into_boxed_str()
3086 }
3087}
3088
3089#[cfg(not(no_global_oom_handling))]
3090#[stable(feature = "string_from_cow_str", since = "1.14.0")]
3091impl<'a> From<Cow<'a, str>> for String {
3092 /// Converts a clone-on-write string to an owned
3093 /// instance of [`String`].
3094 ///
3095 /// This extracts the owned string,
3096 /// clones the string if it is not already owned.
3097 ///
3098 /// # Example
3099 ///
3100 /// ```
3101 /// # use std::borrow::Cow;
3102 /// // If the string is not owned...
3103 /// let cow: Cow<'_, str> = Cow::Borrowed("eggplant");
3104 /// // It will allocate on the heap and copy the string.
3105 /// let owned: String = String::from(cow);
3106 /// assert_eq!(&owned[..], "eggplant");
3107 /// ```
3108 fn from(s: Cow<'a, str>) -> String {
3109 s.into_owned()
3110 }
3111}
3112
3113#[cfg(not(no_global_oom_handling))]
3114#[stable(feature = "rust1", since = "1.0.0")]
3115impl<'a> From<&'a str> for Cow<'a, str> {
3116 /// Converts a string slice into a [`Borrowed`] variant.
3117 /// No heap allocation is performed, and the string
3118 /// is not copied.
3119 ///
3120 /// # Example
3121 ///
3122 /// ```
3123 /// # use std::borrow::Cow;
3124 /// assert_eq!(Cow::from("eggplant"), Cow::Borrowed("eggplant"));
3125 /// ```
3126 ///
3127 /// [`Borrowed`]: crate::borrow::Cow::Borrowed "borrow::Cow::Borrowed"
3128 #[inline]
3129 fn from(s: &'a str) -> Cow<'a, str> {
3130 Cow::Borrowed(s)
3131 }
3132}
3133
3134#[cfg(not(no_global_oom_handling))]
3135#[stable(feature = "rust1", since = "1.0.0")]
3136impl<'a> From<String> for Cow<'a, str> {
3137 /// Converts a [`String`] into an [`Owned`] variant.
3138 /// No heap allocation is performed, and the string
3139 /// is not copied.
3140 ///
3141 /// # Example
3142 ///
3143 /// ```
3144 /// # use std::borrow::Cow;
3145 /// let s = "eggplant".to_string();
3146 /// let s2 = "eggplant".to_string();
3147 /// assert_eq!(Cow::from(s), Cow::<'static, str>::Owned(s2));
3148 /// ```
3149 ///
3150 /// [`Owned`]: crate::borrow::Cow::Owned "borrow::Cow::Owned"
3151 #[inline]
3152 fn from(s: String) -> Cow<'a, str> {
3153 Cow::Owned(s)
3154 }
3155}
3156
3157#[cfg(not(no_global_oom_handling))]
3158#[stable(feature = "cow_from_string_ref", since = "1.28.0")]
3159impl<'a> From<&'a String> for Cow<'a, str> {
3160 /// Converts a [`String`] reference into a [`Borrowed`] variant.
3161 /// No heap allocation is performed, and the string
3162 /// is not copied.
3163 ///
3164 /// # Example
3165 ///
3166 /// ```
3167 /// # use std::borrow::Cow;
3168 /// let s = "eggplant".to_string();
3169 /// assert_eq!(Cow::from(&s), Cow::Borrowed("eggplant"));
3170 /// ```
3171 ///
3172 /// [`Borrowed`]: crate::borrow::Cow::Borrowed "borrow::Cow::Borrowed"
3173 #[inline]
3174 fn from(s: &'a String) -> Cow<'a, str> {
3175 Cow::Borrowed(s.as_str())
3176 }
3177}
3178
3179#[cfg(not(no_global_oom_handling))]
3180#[stable(feature = "cow_str_from_iter", since = "1.12.0")]
3181impl<'a> FromIterator<char> for Cow<'a, str> {
3182 fn from_iter<I: IntoIterator<Item = char>>(it: I) -> Cow<'a, str> {
3183 Cow::Owned(FromIterator::from_iter(it))
3184 }
3185}
3186
3187#[cfg(not(no_global_oom_handling))]
3188#[stable(feature = "cow_str_from_iter", since = "1.12.0")]
3189impl<'a, 'b> FromIterator<&'b str> for Cow<'a, str> {
3190 fn from_iter<I: IntoIterator<Item = &'b str>>(it: I) -> Cow<'a, str> {
3191 Cow::Owned(FromIterator::from_iter(it))
3192 }
3193}
3194
3195#[cfg(not(no_global_oom_handling))]
3196#[stable(feature = "cow_str_from_iter", since = "1.12.0")]
3197impl<'a> FromIterator<String> for Cow<'a, str> {
3198 fn from_iter<I: IntoIterator<Item = String>>(it: I) -> Cow<'a, str> {
3199 Cow::Owned(FromIterator::from_iter(it))
3200 }
3201}
3202
3203#[stable(feature = "from_string_for_vec_u8", since = "1.14.0")]
3204impl From<String> for Vec<u8> {
3205 /// Converts the given [`String`] to a vector [`Vec`] that holds values of type [`u8`].
3206 ///
3207 /// # Examples
3208 ///
3209 /// ```
3210 /// let s1 = String::from("hello world");
3211 /// let v1 = Vec::from(s1);
3212 ///
3213 /// for b in v1 {
3214 /// println!("{b}");
3215 /// }
3216 /// ```
3217 fn from(string: String) -> Vec<u8> {
3218 string.into_bytes()
3219 }
3220}
3221
3222#[stable(feature = "try_from_vec_u8_for_string", since = "1.87.0")]
3223impl TryFrom<Vec<u8>> for String {
3224 type Error = FromUtf8Error;
3225 /// Converts the given [`Vec<u8>`] into a [`String`] if it contains valid UTF-8 data.
3226 ///
3227 /// # Examples
3228 ///
3229 /// ```
3230 /// let s1 = b"hello world".to_vec();
3231 /// let v1 = String::try_from(s1).unwrap();
3232 /// assert_eq!(v1, "hello world");
3233 ///
3234 /// ```
3235 fn try_from(bytes: Vec<u8>) -> Result<Self, Self::Error> {
3236 Self::from_utf8(vec:bytes)
3237 }
3238}
3239
3240#[cfg(not(no_global_oom_handling))]
3241#[stable(feature = "rust1", since = "1.0.0")]
3242impl fmt::Write for String {
3243 #[inline]
3244 fn write_str(&mut self, s: &str) -> fmt::Result {
3245 self.push_str(string:s);
3246 Ok(())
3247 }
3248
3249 #[inline]
3250 fn write_char(&mut self, c: char) -> fmt::Result {
3251 self.push(ch:c);
3252 Ok(())
3253 }
3254}
3255
3256/// An iterator over the [`char`]s of a string.
3257///
3258/// This struct is created by the [`into_chars`] method on [`String`].
3259/// See its documentation for more.
3260///
3261/// [`char`]: prim@char
3262/// [`into_chars`]: String::into_chars
3263#[cfg_attr(not(no_global_oom_handling), derive(Clone))]
3264#[must_use = "iterators are lazy and do nothing unless consumed"]
3265#[unstable(feature = "string_into_chars", issue = "133125")]
3266pub struct IntoChars {
3267 bytes: vec::IntoIter<u8>,
3268}
3269
3270#[unstable(feature = "string_into_chars", issue = "133125")]
3271impl fmt::Debug for IntoChars {
3272 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
3273 f.debug_tuple(name:"IntoChars").field(&self.as_str()).finish()
3274 }
3275}
3276
3277impl IntoChars {
3278 /// Views the underlying data as a subslice of the original data.
3279 ///
3280 /// # Examples
3281 ///
3282 /// ```
3283 /// #![feature(string_into_chars)]
3284 ///
3285 /// let mut chars = String::from("abc").into_chars();
3286 ///
3287 /// assert_eq!(chars.as_str(), "abc");
3288 /// chars.next();
3289 /// assert_eq!(chars.as_str(), "bc");
3290 /// chars.next();
3291 /// chars.next();
3292 /// assert_eq!(chars.as_str(), "");
3293 /// ```
3294 #[unstable(feature = "string_into_chars", issue = "133125")]
3295 #[must_use]
3296 #[inline]
3297 pub fn as_str(&self) -> &str {
3298 // SAFETY: `bytes` is a valid UTF-8 string.
3299 unsafe { str::from_utf8_unchecked(self.bytes.as_slice()) }
3300 }
3301
3302 /// Consumes the `IntoChars`, returning the remaining string.
3303 ///
3304 /// # Examples
3305 ///
3306 /// ```
3307 /// #![feature(string_into_chars)]
3308 ///
3309 /// let chars = String::from("abc").into_chars();
3310 /// assert_eq!(chars.into_string(), "abc");
3311 ///
3312 /// let mut chars = String::from("def").into_chars();
3313 /// chars.next();
3314 /// assert_eq!(chars.into_string(), "ef");
3315 /// ```
3316 #[cfg(not(no_global_oom_handling))]
3317 #[unstable(feature = "string_into_chars", issue = "133125")]
3318 #[inline]
3319 pub fn into_string(self) -> String {
3320 // Safety: `bytes` are kept in UTF-8 form, only removing whole `char`s at a time.
3321 unsafe { String::from_utf8_unchecked(self.bytes.collect()) }
3322 }
3323
3324 #[inline]
3325 fn iter(&self) -> CharIndices<'_> {
3326 self.as_str().char_indices()
3327 }
3328}
3329
3330#[unstable(feature = "string_into_chars", issue = "133125")]
3331impl Iterator for IntoChars {
3332 type Item = char;
3333
3334 #[inline]
3335 fn next(&mut self) -> Option<char> {
3336 let mut iter = self.iter();
3337 match iter.next() {
3338 None => None,
3339 Some((_, ch)) => {
3340 let offset = iter.offset();
3341 // `offset` is a valid index.
3342 let _ = self.bytes.advance_by(offset);
3343 Some(ch)
3344 }
3345 }
3346 }
3347
3348 #[inline]
3349 fn count(self) -> usize {
3350 self.iter().count()
3351 }
3352
3353 #[inline]
3354 fn size_hint(&self) -> (usize, Option<usize>) {
3355 self.iter().size_hint()
3356 }
3357
3358 #[inline]
3359 fn last(mut self) -> Option<char> {
3360 self.next_back()
3361 }
3362}
3363
3364#[unstable(feature = "string_into_chars", issue = "133125")]
3365impl DoubleEndedIterator for IntoChars {
3366 #[inline]
3367 fn next_back(&mut self) -> Option<char> {
3368 let len: usize = self.as_str().len();
3369 let mut iter: CharIndices<'_> = self.iter();
3370 match iter.next_back() {
3371 None => None,
3372 Some((idx: usize, ch: char)) => {
3373 // `idx` is a valid index.
3374 let _ = self.bytes.advance_back_by(len - idx);
3375 Some(ch)
3376 }
3377 }
3378 }
3379}
3380
3381#[unstable(feature = "string_into_chars", issue = "133125")]
3382impl FusedIterator for IntoChars {}
3383
3384/// A draining iterator for `String`.
3385///
3386/// This struct is created by the [`drain`] method on [`String`]. See its
3387/// documentation for more.
3388///
3389/// [`drain`]: String::drain
3390#[stable(feature = "drain", since = "1.6.0")]
3391pub struct Drain<'a> {
3392 /// Will be used as &'a mut String in the destructor
3393 string: *mut String,
3394 /// Start of part to remove
3395 start: usize,
3396 /// End of part to remove
3397 end: usize,
3398 /// Current remaining range to remove
3399 iter: Chars<'a>,
3400}
3401
3402#[stable(feature = "collection_debug", since = "1.17.0")]
3403impl fmt::Debug for Drain<'_> {
3404 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
3405 f.debug_tuple(name:"Drain").field(&self.as_str()).finish()
3406 }
3407}
3408
3409#[stable(feature = "drain", since = "1.6.0")]
3410unsafe impl Sync for Drain<'_> {}
3411#[stable(feature = "drain", since = "1.6.0")]
3412unsafe impl Send for Drain<'_> {}
3413
3414#[stable(feature = "drain", since = "1.6.0")]
3415impl Drop for Drain<'_> {
3416 fn drop(&mut self) {
3417 unsafe {
3418 // Use Vec::drain. "Reaffirm" the bounds checks to avoid
3419 // panic code being inserted again.
3420 let self_vec: &mut Vec = (*self.string).as_mut_vec();
3421 if self.start <= self.end && self.end <= self_vec.len() {
3422 self_vec.drain(self.start..self.end);
3423 }
3424 }
3425 }
3426}
3427
3428impl<'a> Drain<'a> {
3429 /// Returns the remaining (sub)string of this iterator as a slice.
3430 ///
3431 /// # Examples
3432 ///
3433 /// ```
3434 /// let mut s = String::from("abc");
3435 /// let mut drain = s.drain(..);
3436 /// assert_eq!(drain.as_str(), "abc");
3437 /// let _ = drain.next().unwrap();
3438 /// assert_eq!(drain.as_str(), "bc");
3439 /// ```
3440 #[must_use]
3441 #[stable(feature = "string_drain_as_str", since = "1.55.0")]
3442 pub fn as_str(&self) -> &str {
3443 self.iter.as_str()
3444 }
3445}
3446
3447#[stable(feature = "string_drain_as_str", since = "1.55.0")]
3448impl<'a> AsRef<str> for Drain<'a> {
3449 fn as_ref(&self) -> &str {
3450 self.as_str()
3451 }
3452}
3453
3454#[stable(feature = "string_drain_as_str", since = "1.55.0")]
3455impl<'a> AsRef<[u8]> for Drain<'a> {
3456 fn as_ref(&self) -> &[u8] {
3457 self.as_str().as_bytes()
3458 }
3459}
3460
3461#[stable(feature = "drain", since = "1.6.0")]
3462impl Iterator for Drain<'_> {
3463 type Item = char;
3464
3465 #[inline]
3466 fn next(&mut self) -> Option<char> {
3467 self.iter.next()
3468 }
3469
3470 fn size_hint(&self) -> (usize, Option<usize>) {
3471 self.iter.size_hint()
3472 }
3473
3474 #[inline]
3475 fn last(mut self) -> Option<char> {
3476 self.next_back()
3477 }
3478}
3479
3480#[stable(feature = "drain", since = "1.6.0")]
3481impl DoubleEndedIterator for Drain<'_> {
3482 #[inline]
3483 fn next_back(&mut self) -> Option<char> {
3484 self.iter.next_back()
3485 }
3486}
3487
3488#[stable(feature = "fused", since = "1.26.0")]
3489impl FusedIterator for Drain<'_> {}
3490
3491#[cfg(not(no_global_oom_handling))]
3492#[stable(feature = "from_char_for_string", since = "1.46.0")]
3493impl From<char> for String {
3494 /// Allocates an owned [`String`] from a single character.
3495 ///
3496 /// # Example
3497 /// ```rust
3498 /// let c: char = 'a';
3499 /// let s: String = String::from(c);
3500 /// assert_eq!("a", &s[..]);
3501 /// ```
3502 #[inline]
3503 fn from(c: char) -> Self {
3504 c.to_string()
3505 }
3506}
3507