1//! A hash set implemented using [`IndexMap`]
2
3mod iter;
4mod mutable;
5mod slice;
6
7#[cfg(test)]
8mod tests;
9
10pub use self::iter::{
11 Difference, Drain, Intersection, IntoIter, Iter, Splice, SymmetricDifference, Union,
12};
13pub use self::mutable::MutableValues;
14pub use self::slice::Slice;
15
16#[cfg(feature = "rayon")]
17pub use crate::rayon::set as rayon;
18use crate::TryReserveError;
19
20#[cfg(feature = "std")]
21use std::collections::hash_map::RandomState;
22
23use crate::util::try_simplify_range;
24use alloc::boxed::Box;
25use alloc::vec::Vec;
26use core::cmp::Ordering;
27use core::fmt;
28use core::hash::{BuildHasher, Hash};
29use core::ops::{BitAnd, BitOr, BitXor, Index, RangeBounds, Sub};
30
31use super::{Entries, Equivalent, IndexMap};
32
33type Bucket<T> = super::Bucket<T, ()>;
34
35/// A hash set where the iteration order of the values is independent of their
36/// hash values.
37///
38/// The interface is closely compatible with the standard
39/// [`HashSet`][std::collections::HashSet],
40/// but also has additional features.
41///
42/// # Order
43///
44/// The values have a consistent order that is determined by the sequence of
45/// insertion and removal calls on the set. The order does not depend on the
46/// values or the hash function at all. Note that insertion order and value
47/// are not affected if a re-insertion is attempted once an element is
48/// already present.
49///
50/// All iterators traverse the set *in order*. Set operation iterators like
51/// [`IndexSet::union`] produce a concatenated order, as do their matching "bitwise"
52/// operators. See their documentation for specifics.
53///
54/// The insertion order is preserved, with **notable exceptions** like the
55/// [`.remove()`][Self::remove] or [`.swap_remove()`][Self::swap_remove] methods.
56/// Methods such as [`.sort_by()`][Self::sort_by] of
57/// course result in a new order, depending on the sorting order.
58///
59/// # Indices
60///
61/// The values are indexed in a compact range without holes in the range
62/// `0..self.len()`. For example, the method `.get_full` looks up the index for
63/// a value, and the method `.get_index` looks up the value by index.
64///
65/// # Complexity
66///
67/// Internally, `IndexSet<T, S>` just holds an [`IndexMap<T, (), S>`](IndexMap). Thus the complexity
68/// of the two are the same for most methods.
69///
70/// # Examples
71///
72/// ```
73/// use indexmap::IndexSet;
74///
75/// // Collects which letters appear in a sentence.
76/// let letters: IndexSet<_> = "a short treatise on fungi".chars().collect();
77///
78/// assert!(letters.contains(&'s'));
79/// assert!(letters.contains(&'t'));
80/// assert!(letters.contains(&'u'));
81/// assert!(!letters.contains(&'y'));
82/// ```
83#[cfg(feature = "std")]
84pub struct IndexSet<T, S = RandomState> {
85 pub(crate) map: IndexMap<T, (), S>,
86}
87#[cfg(not(feature = "std"))]
88pub struct IndexSet<T, S> {
89 pub(crate) map: IndexMap<T, (), S>,
90}
91
92impl<T, S> Clone for IndexSet<T, S>
93where
94 T: Clone,
95 S: Clone,
96{
97 fn clone(&self) -> Self {
98 IndexSet {
99 map: self.map.clone(),
100 }
101 }
102
103 fn clone_from(&mut self, other: &Self) {
104 self.map.clone_from(&other.map);
105 }
106}
107
108impl<T, S> Entries for IndexSet<T, S> {
109 type Entry = Bucket<T>;
110
111 #[inline]
112 fn into_entries(self) -> Vec<Self::Entry> {
113 self.map.into_entries()
114 }
115
116 #[inline]
117 fn as_entries(&self) -> &[Self::Entry] {
118 self.map.as_entries()
119 }
120
121 #[inline]
122 fn as_entries_mut(&mut self) -> &mut [Self::Entry] {
123 self.map.as_entries_mut()
124 }
125
126 fn with_entries<F>(&mut self, f: F)
127 where
128 F: FnOnce(&mut [Self::Entry]),
129 {
130 self.map.with_entries(f);
131 }
132}
133
134impl<T, S> fmt::Debug for IndexSet<T, S>
135where
136 T: fmt::Debug,
137{
138 #[cfg(not(feature = "test_debug"))]
139 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
140 f.debug_set().entries(self.iter()).finish()
141 }
142
143 #[cfg(feature = "test_debug")]
144 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
145 // Let the inner `IndexMap` print all of its details
146 f.debug_struct("IndexSet").field("map", &self.map).finish()
147 }
148}
149
150#[cfg(feature = "std")]
151#[cfg_attr(docsrs, doc(cfg(feature = "std")))]
152impl<T> IndexSet<T> {
153 /// Create a new set. (Does not allocate.)
154 pub fn new() -> Self {
155 IndexSet {
156 map: IndexMap::new(),
157 }
158 }
159
160 /// Create a new set with capacity for `n` elements.
161 /// (Does not allocate if `n` is zero.)
162 ///
163 /// Computes in **O(n)** time.
164 pub fn with_capacity(n: usize) -> Self {
165 IndexSet {
166 map: IndexMap::with_capacity(n),
167 }
168 }
169}
170
171impl<T, S> IndexSet<T, S> {
172 /// Create a new set with capacity for `n` elements.
173 /// (Does not allocate if `n` is zero.)
174 ///
175 /// Computes in **O(n)** time.
176 pub fn with_capacity_and_hasher(n: usize, hash_builder: S) -> Self {
177 IndexSet {
178 map: IndexMap::with_capacity_and_hasher(n, hash_builder),
179 }
180 }
181
182 /// Create a new set with `hash_builder`.
183 ///
184 /// This function is `const`, so it
185 /// can be called in `static` contexts.
186 pub const fn with_hasher(hash_builder: S) -> Self {
187 IndexSet {
188 map: IndexMap::with_hasher(hash_builder),
189 }
190 }
191
192 /// Return the number of elements the set can hold without reallocating.
193 ///
194 /// This number is a lower bound; the set might be able to hold more,
195 /// but is guaranteed to be able to hold at least this many.
196 ///
197 /// Computes in **O(1)** time.
198 pub fn capacity(&self) -> usize {
199 self.map.capacity()
200 }
201
202 /// Return a reference to the set's `BuildHasher`.
203 pub fn hasher(&self) -> &S {
204 self.map.hasher()
205 }
206
207 /// Return the number of elements in the set.
208 ///
209 /// Computes in **O(1)** time.
210 pub fn len(&self) -> usize {
211 self.map.len()
212 }
213
214 /// Returns true if the set contains no elements.
215 ///
216 /// Computes in **O(1)** time.
217 pub fn is_empty(&self) -> bool {
218 self.map.is_empty()
219 }
220
221 /// Return an iterator over the values of the set, in their order
222 pub fn iter(&self) -> Iter<'_, T> {
223 Iter::new(self.as_entries())
224 }
225
226 /// Remove all elements in the set, while preserving its capacity.
227 ///
228 /// Computes in **O(n)** time.
229 pub fn clear(&mut self) {
230 self.map.clear();
231 }
232
233 /// Shortens the set, keeping the first `len` elements and dropping the rest.
234 ///
235 /// If `len` is greater than the set's current length, this has no effect.
236 pub fn truncate(&mut self, len: usize) {
237 self.map.truncate(len);
238 }
239
240 /// Clears the `IndexSet` in the given index range, returning those values
241 /// as a drain iterator.
242 ///
243 /// The range may be any type that implements [`RangeBounds<usize>`],
244 /// including all of the `std::ops::Range*` types, or even a tuple pair of
245 /// `Bound` start and end values. To drain the set entirely, use `RangeFull`
246 /// like `set.drain(..)`.
247 ///
248 /// This shifts down all entries following the drained range to fill the
249 /// gap, and keeps the allocated memory for reuse.
250 ///
251 /// ***Panics*** if the starting point is greater than the end point or if
252 /// the end point is greater than the length of the set.
253 #[track_caller]
254 pub fn drain<R>(&mut self, range: R) -> Drain<'_, T>
255 where
256 R: RangeBounds<usize>,
257 {
258 Drain::new(self.map.core.drain(range))
259 }
260
261 /// Splits the collection into two at the given index.
262 ///
263 /// Returns a newly allocated set containing the elements in the range
264 /// `[at, len)`. After the call, the original set will be left containing
265 /// the elements `[0, at)` with its previous capacity unchanged.
266 ///
267 /// ***Panics*** if `at > len`.
268 #[track_caller]
269 pub fn split_off(&mut self, at: usize) -> Self
270 where
271 S: Clone,
272 {
273 Self {
274 map: self.map.split_off(at),
275 }
276 }
277
278 /// Reserve capacity for `additional` more values.
279 ///
280 /// Computes in **O(n)** time.
281 pub fn reserve(&mut self, additional: usize) {
282 self.map.reserve(additional);
283 }
284
285 /// Reserve capacity for `additional` more values, without over-allocating.
286 ///
287 /// Unlike `reserve`, this does not deliberately over-allocate the entry capacity to avoid
288 /// frequent re-allocations. However, the underlying data structures may still have internal
289 /// capacity requirements, and the allocator itself may give more space than requested, so this
290 /// cannot be relied upon to be precisely minimal.
291 ///
292 /// Computes in **O(n)** time.
293 pub fn reserve_exact(&mut self, additional: usize) {
294 self.map.reserve_exact(additional);
295 }
296
297 /// Try to reserve capacity for `additional` more values.
298 ///
299 /// Computes in **O(n)** time.
300 pub fn try_reserve(&mut self, additional: usize) -> Result<(), TryReserveError> {
301 self.map.try_reserve(additional)
302 }
303
304 /// Try to reserve capacity for `additional` more values, without over-allocating.
305 ///
306 /// Unlike `try_reserve`, this does not deliberately over-allocate the entry capacity to avoid
307 /// frequent re-allocations. However, the underlying data structures may still have internal
308 /// capacity requirements, and the allocator itself may give more space than requested, so this
309 /// cannot be relied upon to be precisely minimal.
310 ///
311 /// Computes in **O(n)** time.
312 pub fn try_reserve_exact(&mut self, additional: usize) -> Result<(), TryReserveError> {
313 self.map.try_reserve_exact(additional)
314 }
315
316 /// Shrink the capacity of the set as much as possible.
317 ///
318 /// Computes in **O(n)** time.
319 pub fn shrink_to_fit(&mut self) {
320 self.map.shrink_to_fit();
321 }
322
323 /// Shrink the capacity of the set with a lower limit.
324 ///
325 /// Computes in **O(n)** time.
326 pub fn shrink_to(&mut self, min_capacity: usize) {
327 self.map.shrink_to(min_capacity);
328 }
329}
330
331impl<T, S> IndexSet<T, S>
332where
333 T: Hash + Eq,
334 S: BuildHasher,
335{
336 /// Insert the value into the set.
337 ///
338 /// If an equivalent item already exists in the set, it returns
339 /// `false` leaving the original value in the set and without
340 /// altering its insertion order. Otherwise, it inserts the new
341 /// item and returns `true`.
342 ///
343 /// Computes in **O(1)** time (amortized average).
344 pub fn insert(&mut self, value: T) -> bool {
345 self.map.insert(value, ()).is_none()
346 }
347
348 /// Insert the value into the set, and get its index.
349 ///
350 /// If an equivalent item already exists in the set, it returns
351 /// the index of the existing item and `false`, leaving the
352 /// original value in the set and without altering its insertion
353 /// order. Otherwise, it inserts the new item and returns the index
354 /// of the inserted item and `true`.
355 ///
356 /// Computes in **O(1)** time (amortized average).
357 pub fn insert_full(&mut self, value: T) -> (usize, bool) {
358 let (index, existing) = self.map.insert_full(value, ());
359 (index, existing.is_none())
360 }
361
362 /// Insert the value into the set at its ordered position among sorted values.
363 ///
364 /// This is equivalent to finding the position with
365 /// [`binary_search`][Self::binary_search], and if needed calling
366 /// [`insert_before`][Self::insert_before] for a new value.
367 ///
368 /// If the sorted item is found in the set, it returns the index of that
369 /// existing item and `false`, without any change. Otherwise, it inserts the
370 /// new item and returns its sorted index and `true`.
371 ///
372 /// If the existing items are **not** already sorted, then the insertion
373 /// index is unspecified (like [`slice::binary_search`]), but the value
374 /// is moved to or inserted at that position regardless.
375 ///
376 /// Computes in **O(n)** time (average). Instead of repeating calls to
377 /// `insert_sorted`, it may be faster to call batched [`insert`][Self::insert]
378 /// or [`extend`][Self::extend] and only call [`sort`][Self::sort] or
379 /// [`sort_unstable`][Self::sort_unstable] once.
380 pub fn insert_sorted(&mut self, value: T) -> (usize, bool)
381 where
382 T: Ord,
383 {
384 let (index, existing) = self.map.insert_sorted(value, ());
385 (index, existing.is_none())
386 }
387
388 /// Insert the value into the set before the value at the given index, or at the end.
389 ///
390 /// If an equivalent item already exists in the set, it returns `false` leaving the
391 /// original value in the set, but moved to the new position. The returned index
392 /// will either be the given index or one less, depending on how the value moved.
393 /// (See [`shift_insert`](Self::shift_insert) for different behavior here.)
394 ///
395 /// Otherwise, it inserts the new value exactly at the given index and returns `true`.
396 ///
397 /// ***Panics*** if `index` is out of bounds.
398 /// Valid indices are `0..=set.len()` (inclusive).
399 ///
400 /// Computes in **O(n)** time (average).
401 ///
402 /// # Examples
403 ///
404 /// ```
405 /// use indexmap::IndexSet;
406 /// let mut set: IndexSet<char> = ('a'..='z').collect();
407 ///
408 /// // The new value '*' goes exactly at the given index.
409 /// assert_eq!(set.get_index_of(&'*'), None);
410 /// assert_eq!(set.insert_before(10, '*'), (10, true));
411 /// assert_eq!(set.get_index_of(&'*'), Some(10));
412 ///
413 /// // Moving the value 'a' up will shift others down, so this moves *before* 10 to index 9.
414 /// assert_eq!(set.insert_before(10, 'a'), (9, false));
415 /// assert_eq!(set.get_index_of(&'a'), Some(9));
416 /// assert_eq!(set.get_index_of(&'*'), Some(10));
417 ///
418 /// // Moving the value 'z' down will shift others up, so this moves to exactly 10.
419 /// assert_eq!(set.insert_before(10, 'z'), (10, false));
420 /// assert_eq!(set.get_index_of(&'z'), Some(10));
421 /// assert_eq!(set.get_index_of(&'*'), Some(11));
422 ///
423 /// // Moving or inserting before the endpoint is also valid.
424 /// assert_eq!(set.len(), 27);
425 /// assert_eq!(set.insert_before(set.len(), '*'), (26, false));
426 /// assert_eq!(set.get_index_of(&'*'), Some(26));
427 /// assert_eq!(set.insert_before(set.len(), '+'), (27, true));
428 /// assert_eq!(set.get_index_of(&'+'), Some(27));
429 /// assert_eq!(set.len(), 28);
430 /// ```
431 #[track_caller]
432 pub fn insert_before(&mut self, index: usize, value: T) -> (usize, bool) {
433 let (index, existing) = self.map.insert_before(index, value, ());
434 (index, existing.is_none())
435 }
436
437 /// Insert the value into the set at the given index.
438 ///
439 /// If an equivalent item already exists in the set, it returns `false` leaving
440 /// the original value in the set, but moved to the given index.
441 /// Note that existing values **cannot** be moved to `index == set.len()`!
442 /// (See [`insert_before`](Self::insert_before) for different behavior here.)
443 ///
444 /// Otherwise, it inserts the new value at the given index and returns `true`.
445 ///
446 /// ***Panics*** if `index` is out of bounds.
447 /// Valid indices are `0..set.len()` (exclusive) when moving an existing value, or
448 /// `0..=set.len()` (inclusive) when inserting a new value.
449 ///
450 /// Computes in **O(n)** time (average).
451 ///
452 /// # Examples
453 ///
454 /// ```
455 /// use indexmap::IndexSet;
456 /// let mut set: IndexSet<char> = ('a'..='z').collect();
457 ///
458 /// // The new value '*' goes exactly at the given index.
459 /// assert_eq!(set.get_index_of(&'*'), None);
460 /// assert_eq!(set.shift_insert(10, '*'), true);
461 /// assert_eq!(set.get_index_of(&'*'), Some(10));
462 ///
463 /// // Moving the value 'a' up to 10 will shift others down, including the '*' that was at 10.
464 /// assert_eq!(set.shift_insert(10, 'a'), false);
465 /// assert_eq!(set.get_index_of(&'a'), Some(10));
466 /// assert_eq!(set.get_index_of(&'*'), Some(9));
467 ///
468 /// // Moving the value 'z' down to 9 will shift others up, including the '*' that was at 9.
469 /// assert_eq!(set.shift_insert(9, 'z'), false);
470 /// assert_eq!(set.get_index_of(&'z'), Some(9));
471 /// assert_eq!(set.get_index_of(&'*'), Some(10));
472 ///
473 /// // Existing values can move to len-1 at most, but new values can insert at the endpoint.
474 /// assert_eq!(set.len(), 27);
475 /// assert_eq!(set.shift_insert(set.len() - 1, '*'), false);
476 /// assert_eq!(set.get_index_of(&'*'), Some(26));
477 /// assert_eq!(set.shift_insert(set.len(), '+'), true);
478 /// assert_eq!(set.get_index_of(&'+'), Some(27));
479 /// assert_eq!(set.len(), 28);
480 /// ```
481 ///
482 /// ```should_panic
483 /// use indexmap::IndexSet;
484 /// let mut set: IndexSet<char> = ('a'..='z').collect();
485 ///
486 /// // This is an invalid index for moving an existing value!
487 /// set.shift_insert(set.len(), 'a');
488 /// ```
489 #[track_caller]
490 pub fn shift_insert(&mut self, index: usize, value: T) -> bool {
491 self.map.shift_insert(index, value, ()).is_none()
492 }
493
494 /// Adds a value to the set, replacing the existing value, if any, that is
495 /// equal to the given one, without altering its insertion order. Returns
496 /// the replaced value.
497 ///
498 /// Computes in **O(1)** time (average).
499 pub fn replace(&mut self, value: T) -> Option<T> {
500 self.replace_full(value).1
501 }
502
503 /// Adds a value to the set, replacing the existing value, if any, that is
504 /// equal to the given one, without altering its insertion order. Returns
505 /// the index of the item and its replaced value.
506 ///
507 /// Computes in **O(1)** time (average).
508 pub fn replace_full(&mut self, value: T) -> (usize, Option<T>) {
509 let hash = self.map.hash(&value);
510 match self.map.core.replace_full(hash, value, ()) {
511 (i, Some((replaced, ()))) => (i, Some(replaced)),
512 (i, None) => (i, None),
513 }
514 }
515
516 /// Return an iterator over the values that are in `self` but not `other`.
517 ///
518 /// Values are produced in the same order that they appear in `self`.
519 pub fn difference<'a, S2>(&'a self, other: &'a IndexSet<T, S2>) -> Difference<'a, T, S2>
520 where
521 S2: BuildHasher,
522 {
523 Difference::new(self, other)
524 }
525
526 /// Return an iterator over the values that are in `self` or `other`,
527 /// but not in both.
528 ///
529 /// Values from `self` are produced in their original order, followed by
530 /// values from `other` in their original order.
531 pub fn symmetric_difference<'a, S2>(
532 &'a self,
533 other: &'a IndexSet<T, S2>,
534 ) -> SymmetricDifference<'a, T, S, S2>
535 where
536 S2: BuildHasher,
537 {
538 SymmetricDifference::new(self, other)
539 }
540
541 /// Return an iterator over the values that are in both `self` and `other`.
542 ///
543 /// Values are produced in the same order that they appear in `self`.
544 pub fn intersection<'a, S2>(&'a self, other: &'a IndexSet<T, S2>) -> Intersection<'a, T, S2>
545 where
546 S2: BuildHasher,
547 {
548 Intersection::new(self, other)
549 }
550
551 /// Return an iterator over all values that are in `self` or `other`.
552 ///
553 /// Values from `self` are produced in their original order, followed by
554 /// values that are unique to `other` in their original order.
555 pub fn union<'a, S2>(&'a self, other: &'a IndexSet<T, S2>) -> Union<'a, T, S>
556 where
557 S2: BuildHasher,
558 {
559 Union::new(self, other)
560 }
561
562 /// Creates a splicing iterator that replaces the specified range in the set
563 /// with the given `replace_with` iterator and yields the removed items.
564 /// `replace_with` does not need to be the same length as `range`.
565 ///
566 /// The `range` is removed even if the iterator is not consumed until the
567 /// end. It is unspecified how many elements are removed from the set if the
568 /// `Splice` value is leaked.
569 ///
570 /// The input iterator `replace_with` is only consumed when the `Splice`
571 /// value is dropped. If a value from the iterator matches an existing entry
572 /// in the set (outside of `range`), then the original will be unchanged.
573 /// Otherwise, the new value will be inserted in the replaced `range`.
574 ///
575 /// ***Panics*** if the starting point is greater than the end point or if
576 /// the end point is greater than the length of the set.
577 ///
578 /// # Examples
579 ///
580 /// ```
581 /// use indexmap::IndexSet;
582 ///
583 /// let mut set = IndexSet::from([0, 1, 2, 3, 4]);
584 /// let new = [5, 4, 3, 2, 1];
585 /// let removed: Vec<_> = set.splice(2..4, new).collect();
586 ///
587 /// // 1 and 4 kept their positions, while 5, 3, and 2 were newly inserted.
588 /// assert!(set.into_iter().eq([0, 1, 5, 3, 2, 4]));
589 /// assert_eq!(removed, &[2, 3]);
590 /// ```
591 #[track_caller]
592 pub fn splice<R, I>(&mut self, range: R, replace_with: I) -> Splice<'_, I::IntoIter, T, S>
593 where
594 R: RangeBounds<usize>,
595 I: IntoIterator<Item = T>,
596 {
597 Splice::new(self, range, replace_with.into_iter())
598 }
599
600 /// Moves all values from `other` into `self`, leaving `other` empty.
601 ///
602 /// This is equivalent to calling [`insert`][Self::insert] for each value
603 /// from `other` in order, which means that values that already exist
604 /// in `self` are unchanged in their current position.
605 ///
606 /// See also [`union`][Self::union] to iterate the combined values by
607 /// reference, without modifying `self` or `other`.
608 ///
609 /// # Examples
610 ///
611 /// ```
612 /// use indexmap::IndexSet;
613 ///
614 /// let mut a = IndexSet::from([3, 2, 1]);
615 /// let mut b = IndexSet::from([3, 4, 5]);
616 /// let old_capacity = b.capacity();
617 ///
618 /// a.append(&mut b);
619 ///
620 /// assert_eq!(a.len(), 5);
621 /// assert_eq!(b.len(), 0);
622 /// assert_eq!(b.capacity(), old_capacity);
623 ///
624 /// assert!(a.iter().eq(&[3, 2, 1, 4, 5]));
625 /// ```
626 pub fn append<S2>(&mut self, other: &mut IndexSet<T, S2>) {
627 self.map.append(&mut other.map);
628 }
629}
630
631impl<T, S> IndexSet<T, S>
632where
633 S: BuildHasher,
634{
635 /// Return `true` if an equivalent to `value` exists in the set.
636 ///
637 /// Computes in **O(1)** time (average).
638 pub fn contains<Q>(&self, value: &Q) -> bool
639 where
640 Q: ?Sized + Hash + Equivalent<T>,
641 {
642 self.map.contains_key(value)
643 }
644
645 /// Return a reference to the value stored in the set, if it is present,
646 /// else `None`.
647 ///
648 /// Computes in **O(1)** time (average).
649 pub fn get<Q>(&self, value: &Q) -> Option<&T>
650 where
651 Q: ?Sized + Hash + Equivalent<T>,
652 {
653 self.map.get_key_value(value).map(|(x, &())| x)
654 }
655
656 /// Return item index and value
657 pub fn get_full<Q>(&self, value: &Q) -> Option<(usize, &T)>
658 where
659 Q: ?Sized + Hash + Equivalent<T>,
660 {
661 self.map.get_full(value).map(|(i, x, &())| (i, x))
662 }
663
664 /// Return item index, if it exists in the set
665 ///
666 /// Computes in **O(1)** time (average).
667 pub fn get_index_of<Q>(&self, value: &Q) -> Option<usize>
668 where
669 Q: ?Sized + Hash + Equivalent<T>,
670 {
671 self.map.get_index_of(value)
672 }
673
674 /// Remove the value from the set, and return `true` if it was present.
675 ///
676 /// **NOTE:** This is equivalent to [`.swap_remove(value)`][Self::swap_remove], replacing this
677 /// value's position with the last element, and it is deprecated in favor of calling that
678 /// explicitly. If you need to preserve the relative order of the values in the set, use
679 /// [`.shift_remove(value)`][Self::shift_remove] instead.
680 #[deprecated(note = "`remove` disrupts the set order -- \
681 use `swap_remove` or `shift_remove` for explicit behavior.")]
682 pub fn remove<Q>(&mut self, value: &Q) -> bool
683 where
684 Q: ?Sized + Hash + Equivalent<T>,
685 {
686 self.swap_remove(value)
687 }
688
689 /// Remove the value from the set, and return `true` if it was present.
690 ///
691 /// Like [`Vec::swap_remove`], the value is removed by swapping it with the
692 /// last element of the set and popping it off. **This perturbs
693 /// the position of what used to be the last element!**
694 ///
695 /// Return `false` if `value` was not in the set.
696 ///
697 /// Computes in **O(1)** time (average).
698 pub fn swap_remove<Q>(&mut self, value: &Q) -> bool
699 where
700 Q: ?Sized + Hash + Equivalent<T>,
701 {
702 self.map.swap_remove(value).is_some()
703 }
704
705 /// Remove the value from the set, and return `true` if it was present.
706 ///
707 /// Like [`Vec::remove`], the value is removed by shifting all of the
708 /// elements that follow it, preserving their relative order.
709 /// **This perturbs the index of all of those elements!**
710 ///
711 /// Return `false` if `value` was not in the set.
712 ///
713 /// Computes in **O(n)** time (average).
714 pub fn shift_remove<Q>(&mut self, value: &Q) -> bool
715 where
716 Q: ?Sized + Hash + Equivalent<T>,
717 {
718 self.map.shift_remove(value).is_some()
719 }
720
721 /// Removes and returns the value in the set, if any, that is equal to the
722 /// given one.
723 ///
724 /// **NOTE:** This is equivalent to [`.swap_take(value)`][Self::swap_take], replacing this
725 /// value's position with the last element, and it is deprecated in favor of calling that
726 /// explicitly. If you need to preserve the relative order of the values in the set, use
727 /// [`.shift_take(value)`][Self::shift_take] instead.
728 #[deprecated(note = "`take` disrupts the set order -- \
729 use `swap_take` or `shift_take` for explicit behavior.")]
730 pub fn take<Q>(&mut self, value: &Q) -> Option<T>
731 where
732 Q: ?Sized + Hash + Equivalent<T>,
733 {
734 self.swap_take(value)
735 }
736
737 /// Removes and returns the value in the set, if any, that is equal to the
738 /// given one.
739 ///
740 /// Like [`Vec::swap_remove`], the value is removed by swapping it with the
741 /// last element of the set and popping it off. **This perturbs
742 /// the position of what used to be the last element!**
743 ///
744 /// Return `None` if `value` was not in the set.
745 ///
746 /// Computes in **O(1)** time (average).
747 pub fn swap_take<Q>(&mut self, value: &Q) -> Option<T>
748 where
749 Q: ?Sized + Hash + Equivalent<T>,
750 {
751 self.map.swap_remove_entry(value).map(|(x, ())| x)
752 }
753
754 /// Removes and returns the value in the set, if any, that is equal to the
755 /// given one.
756 ///
757 /// Like [`Vec::remove`], the value is removed by shifting all of the
758 /// elements that follow it, preserving their relative order.
759 /// **This perturbs the index of all of those elements!**
760 ///
761 /// Return `None` if `value` was not in the set.
762 ///
763 /// Computes in **O(n)** time (average).
764 pub fn shift_take<Q>(&mut self, value: &Q) -> Option<T>
765 where
766 Q: ?Sized + Hash + Equivalent<T>,
767 {
768 self.map.shift_remove_entry(value).map(|(x, ())| x)
769 }
770
771 /// Remove the value from the set return it and the index it had.
772 ///
773 /// Like [`Vec::swap_remove`], the value is removed by swapping it with the
774 /// last element of the set and popping it off. **This perturbs
775 /// the position of what used to be the last element!**
776 ///
777 /// Return `None` if `value` was not in the set.
778 pub fn swap_remove_full<Q>(&mut self, value: &Q) -> Option<(usize, T)>
779 where
780 Q: ?Sized + Hash + Equivalent<T>,
781 {
782 self.map.swap_remove_full(value).map(|(i, x, ())| (i, x))
783 }
784
785 /// Remove the value from the set return it and the index it had.
786 ///
787 /// Like [`Vec::remove`], the value is removed by shifting all of the
788 /// elements that follow it, preserving their relative order.
789 /// **This perturbs the index of all of those elements!**
790 ///
791 /// Return `None` if `value` was not in the set.
792 pub fn shift_remove_full<Q>(&mut self, value: &Q) -> Option<(usize, T)>
793 where
794 Q: ?Sized + Hash + Equivalent<T>,
795 {
796 self.map.shift_remove_full(value).map(|(i, x, ())| (i, x))
797 }
798}
799
800impl<T, S> IndexSet<T, S> {
801 /// Remove the last value
802 ///
803 /// This preserves the order of the remaining elements.
804 ///
805 /// Computes in **O(1)** time (average).
806 #[doc(alias = "pop_last")] // like `BTreeSet`
807 pub fn pop(&mut self) -> Option<T> {
808 self.map.pop().map(|(x, ())| x)
809 }
810
811 /// Scan through each value in the set and keep those where the
812 /// closure `keep` returns `true`.
813 ///
814 /// The elements are visited in order, and remaining elements keep their
815 /// order.
816 ///
817 /// Computes in **O(n)** time (average).
818 pub fn retain<F>(&mut self, mut keep: F)
819 where
820 F: FnMut(&T) -> bool,
821 {
822 self.map.retain(move |x, &mut ()| keep(x))
823 }
824
825 /// Sort the set’s values by their default ordering.
826 ///
827 /// This is a stable sort -- but equivalent values should not normally coexist in
828 /// a set at all, so [`sort_unstable`][Self::sort_unstable] is preferred
829 /// because it is generally faster and doesn't allocate auxiliary memory.
830 ///
831 /// See [`sort_by`](Self::sort_by) for details.
832 pub fn sort(&mut self)
833 where
834 T: Ord,
835 {
836 self.map.sort_keys()
837 }
838
839 /// Sort the set’s values in place using the comparison function `cmp`.
840 ///
841 /// Computes in **O(n log n)** time and **O(n)** space. The sort is stable.
842 pub fn sort_by<F>(&mut self, mut cmp: F)
843 where
844 F: FnMut(&T, &T) -> Ordering,
845 {
846 self.map.sort_by(move |a, _, b, _| cmp(a, b));
847 }
848
849 /// Sort the values of the set and return a by-value iterator of
850 /// the values with the result.
851 ///
852 /// The sort is stable.
853 pub fn sorted_by<F>(self, mut cmp: F) -> IntoIter<T>
854 where
855 F: FnMut(&T, &T) -> Ordering,
856 {
857 let mut entries = self.into_entries();
858 entries.sort_by(move |a, b| cmp(&a.key, &b.key));
859 IntoIter::new(entries)
860 }
861
862 /// Sort the set's values by their default ordering.
863 ///
864 /// See [`sort_unstable_by`](Self::sort_unstable_by) for details.
865 pub fn sort_unstable(&mut self)
866 where
867 T: Ord,
868 {
869 self.map.sort_unstable_keys()
870 }
871
872 /// Sort the set's values in place using the comparison function `cmp`.
873 ///
874 /// Computes in **O(n log n)** time. The sort is unstable.
875 pub fn sort_unstable_by<F>(&mut self, mut cmp: F)
876 where
877 F: FnMut(&T, &T) -> Ordering,
878 {
879 self.map.sort_unstable_by(move |a, _, b, _| cmp(a, b))
880 }
881
882 /// Sort the values of the set and return a by-value iterator of
883 /// the values with the result.
884 pub fn sorted_unstable_by<F>(self, mut cmp: F) -> IntoIter<T>
885 where
886 F: FnMut(&T, &T) -> Ordering,
887 {
888 let mut entries = self.into_entries();
889 entries.sort_unstable_by(move |a, b| cmp(&a.key, &b.key));
890 IntoIter::new(entries)
891 }
892
893 /// Sort the set’s values in place using a key extraction function.
894 ///
895 /// During sorting, the function is called at most once per entry, by using temporary storage
896 /// to remember the results of its evaluation. The order of calls to the function is
897 /// unspecified and may change between versions of `indexmap` or the standard library.
898 ///
899 /// Computes in **O(m n + n log n + c)** time () and **O(n)** space, where the function is
900 /// **O(m)**, *n* is the length of the map, and *c* the capacity. The sort is stable.
901 pub fn sort_by_cached_key<K, F>(&mut self, mut sort_key: F)
902 where
903 K: Ord,
904 F: FnMut(&T) -> K,
905 {
906 self.with_entries(move |entries| {
907 entries.sort_by_cached_key(move |a| sort_key(&a.key));
908 });
909 }
910
911 /// Search over a sorted set for a value.
912 ///
913 /// Returns the position where that value is present, or the position where it can be inserted
914 /// to maintain the sort. See [`slice::binary_search`] for more details.
915 ///
916 /// Computes in **O(log(n))** time, which is notably less scalable than looking the value up
917 /// using [`get_index_of`][IndexSet::get_index_of], but this can also position missing values.
918 pub fn binary_search(&self, x: &T) -> Result<usize, usize>
919 where
920 T: Ord,
921 {
922 self.as_slice().binary_search(x)
923 }
924
925 /// Search over a sorted set with a comparator function.
926 ///
927 /// Returns the position where that value is present, or the position where it can be inserted
928 /// to maintain the sort. See [`slice::binary_search_by`] for more details.
929 ///
930 /// Computes in **O(log(n))** time.
931 #[inline]
932 pub fn binary_search_by<'a, F>(&'a self, f: F) -> Result<usize, usize>
933 where
934 F: FnMut(&'a T) -> Ordering,
935 {
936 self.as_slice().binary_search_by(f)
937 }
938
939 /// Search over a sorted set with an extraction function.
940 ///
941 /// Returns the position where that value is present, or the position where it can be inserted
942 /// to maintain the sort. See [`slice::binary_search_by_key`] for more details.
943 ///
944 /// Computes in **O(log(n))** time.
945 #[inline]
946 pub fn binary_search_by_key<'a, B, F>(&'a self, b: &B, f: F) -> Result<usize, usize>
947 where
948 F: FnMut(&'a T) -> B,
949 B: Ord,
950 {
951 self.as_slice().binary_search_by_key(b, f)
952 }
953
954 /// Returns the index of the partition point of a sorted set according to the given predicate
955 /// (the index of the first element of the second partition).
956 ///
957 /// See [`slice::partition_point`] for more details.
958 ///
959 /// Computes in **O(log(n))** time.
960 #[must_use]
961 pub fn partition_point<P>(&self, pred: P) -> usize
962 where
963 P: FnMut(&T) -> bool,
964 {
965 self.as_slice().partition_point(pred)
966 }
967
968 /// Reverses the order of the set’s values in place.
969 ///
970 /// Computes in **O(n)** time and **O(1)** space.
971 pub fn reverse(&mut self) {
972 self.map.reverse()
973 }
974
975 /// Returns a slice of all the values in the set.
976 ///
977 /// Computes in **O(1)** time.
978 pub fn as_slice(&self) -> &Slice<T> {
979 Slice::from_slice(self.as_entries())
980 }
981
982 /// Converts into a boxed slice of all the values in the set.
983 ///
984 /// Note that this will drop the inner hash table and any excess capacity.
985 pub fn into_boxed_slice(self) -> Box<Slice<T>> {
986 Slice::from_boxed(self.into_entries().into_boxed_slice())
987 }
988
989 /// Get a value by index
990 ///
991 /// Valid indices are `0 <= index < self.len()`.
992 ///
993 /// Computes in **O(1)** time.
994 pub fn get_index(&self, index: usize) -> Option<&T> {
995 self.as_entries().get(index).map(Bucket::key_ref)
996 }
997
998 /// Returns a slice of values in the given range of indices.
999 ///
1000 /// Valid indices are `0 <= index < self.len()`.
1001 ///
1002 /// Computes in **O(1)** time.
1003 pub fn get_range<R: RangeBounds<usize>>(&self, range: R) -> Option<&Slice<T>> {
1004 let entries = self.as_entries();
1005 let range = try_simplify_range(range, entries.len())?;
1006 entries.get(range).map(Slice::from_slice)
1007 }
1008
1009 /// Get the first value
1010 ///
1011 /// Computes in **O(1)** time.
1012 pub fn first(&self) -> Option<&T> {
1013 self.as_entries().first().map(Bucket::key_ref)
1014 }
1015
1016 /// Get the last value
1017 ///
1018 /// Computes in **O(1)** time.
1019 pub fn last(&self) -> Option<&T> {
1020 self.as_entries().last().map(Bucket::key_ref)
1021 }
1022
1023 /// Remove the value by index
1024 ///
1025 /// Valid indices are `0 <= index < self.len()`.
1026 ///
1027 /// Like [`Vec::swap_remove`], the value is removed by swapping it with the
1028 /// last element of the set and popping it off. **This perturbs
1029 /// the position of what used to be the last element!**
1030 ///
1031 /// Computes in **O(1)** time (average).
1032 pub fn swap_remove_index(&mut self, index: usize) -> Option<T> {
1033 self.map.swap_remove_index(index).map(|(x, ())| x)
1034 }
1035
1036 /// Remove the value by index
1037 ///
1038 /// Valid indices are `0 <= index < self.len()`.
1039 ///
1040 /// Like [`Vec::remove`], the value is removed by shifting all of the
1041 /// elements that follow it, preserving their relative order.
1042 /// **This perturbs the index of all of those elements!**
1043 ///
1044 /// Computes in **O(n)** time (average).
1045 pub fn shift_remove_index(&mut self, index: usize) -> Option<T> {
1046 self.map.shift_remove_index(index).map(|(x, ())| x)
1047 }
1048
1049 /// Moves the position of a value from one index to another
1050 /// by shifting all other values in-between.
1051 ///
1052 /// * If `from < to`, the other values will shift down while the targeted value moves up.
1053 /// * If `from > to`, the other values will shift up while the targeted value moves down.
1054 ///
1055 /// ***Panics*** if `from` or `to` are out of bounds.
1056 ///
1057 /// Computes in **O(n)** time (average).
1058 #[track_caller]
1059 pub fn move_index(&mut self, from: usize, to: usize) {
1060 self.map.move_index(from, to)
1061 }
1062
1063 /// Swaps the position of two values in the set.
1064 ///
1065 /// ***Panics*** if `a` or `b` are out of bounds.
1066 ///
1067 /// Computes in **O(1)** time (average).
1068 #[track_caller]
1069 pub fn swap_indices(&mut self, a: usize, b: usize) {
1070 self.map.swap_indices(a, b)
1071 }
1072}
1073
1074/// Access [`IndexSet`] values at indexed positions.
1075///
1076/// # Examples
1077///
1078/// ```
1079/// use indexmap::IndexSet;
1080///
1081/// let mut set = IndexSet::new();
1082/// for word in "Lorem ipsum dolor sit amet".split_whitespace() {
1083/// set.insert(word.to_string());
1084/// }
1085/// assert_eq!(set[0], "Lorem");
1086/// assert_eq!(set[1], "ipsum");
1087/// set.reverse();
1088/// assert_eq!(set[0], "amet");
1089/// assert_eq!(set[1], "sit");
1090/// set.sort();
1091/// assert_eq!(set[0], "Lorem");
1092/// assert_eq!(set[1], "amet");
1093/// ```
1094///
1095/// ```should_panic
1096/// use indexmap::IndexSet;
1097///
1098/// let mut set = IndexSet::new();
1099/// set.insert("foo");
1100/// println!("{:?}", set[10]); // panics!
1101/// ```
1102impl<T, S> Index<usize> for IndexSet<T, S> {
1103 type Output = T;
1104
1105 /// Returns a reference to the value at the supplied `index`.
1106 ///
1107 /// ***Panics*** if `index` is out of bounds.
1108 fn index(&self, index: usize) -> &T {
1109 self.get_index(index).unwrap_or_else(|| {
1110 panic!(
1111 "index out of bounds: the len is {len} but the index is {index}",
1112 len = self.len()
1113 );
1114 })
1115 }
1116}
1117
1118impl<T, S> FromIterator<T> for IndexSet<T, S>
1119where
1120 T: Hash + Eq,
1121 S: BuildHasher + Default,
1122{
1123 fn from_iter<I: IntoIterator<Item = T>>(iterable: I) -> Self {
1124 let iter: impl Iterator = iterable.into_iter().map(|x: T| (x, ()));
1125 IndexSet {
1126 map: IndexMap::from_iter(iter),
1127 }
1128 }
1129}
1130
1131#[cfg(feature = "std")]
1132#[cfg_attr(docsrs, doc(cfg(feature = "std")))]
1133impl<T, const N: usize> From<[T; N]> for IndexSet<T, RandomState>
1134where
1135 T: Eq + Hash,
1136{
1137 /// # Examples
1138 ///
1139 /// ```
1140 /// use indexmap::IndexSet;
1141 ///
1142 /// let set1 = IndexSet::from([1, 2, 3, 4]);
1143 /// let set2: IndexSet<_> = [1, 2, 3, 4].into();
1144 /// assert_eq!(set1, set2);
1145 /// ```
1146 fn from(arr: [T; N]) -> Self {
1147 Self::from_iter(arr)
1148 }
1149}
1150
1151impl<T, S> Extend<T> for IndexSet<T, S>
1152where
1153 T: Hash + Eq,
1154 S: BuildHasher,
1155{
1156 fn extend<I: IntoIterator<Item = T>>(&mut self, iterable: I) {
1157 let iter: impl Iterator = iterable.into_iter().map(|x: T| (x, ()));
1158 self.map.extend(iter);
1159 }
1160}
1161
1162impl<'a, T, S> Extend<&'a T> for IndexSet<T, S>
1163where
1164 T: Hash + Eq + Copy + 'a,
1165 S: BuildHasher,
1166{
1167 fn extend<I: IntoIterator<Item = &'a T>>(&mut self, iterable: I) {
1168 let iter: impl Iterator = iterable.into_iter().copied();
1169 self.extend(iter);
1170 }
1171}
1172
1173impl<T, S> Default for IndexSet<T, S>
1174where
1175 S: Default,
1176{
1177 /// Return an empty [`IndexSet`]
1178 fn default() -> Self {
1179 IndexSet {
1180 map: IndexMap::default(),
1181 }
1182 }
1183}
1184
1185impl<T, S1, S2> PartialEq<IndexSet<T, S2>> for IndexSet<T, S1>
1186where
1187 T: Hash + Eq,
1188 S1: BuildHasher,
1189 S2: BuildHasher,
1190{
1191 fn eq(&self, other: &IndexSet<T, S2>) -> bool {
1192 self.len() == other.len() && self.is_subset(other)
1193 }
1194}
1195
1196impl<T, S> Eq for IndexSet<T, S>
1197where
1198 T: Eq + Hash,
1199 S: BuildHasher,
1200{
1201}
1202
1203impl<T, S> IndexSet<T, S>
1204where
1205 T: Eq + Hash,
1206 S: BuildHasher,
1207{
1208 /// Returns `true` if `self` has no elements in common with `other`.
1209 pub fn is_disjoint<S2>(&self, other: &IndexSet<T, S2>) -> bool
1210 where
1211 S2: BuildHasher,
1212 {
1213 if self.len() <= other.len() {
1214 self.iter().all(move |value| !other.contains(value))
1215 } else {
1216 other.iter().all(move |value| !self.contains(value))
1217 }
1218 }
1219
1220 /// Returns `true` if all elements of `self` are contained in `other`.
1221 pub fn is_subset<S2>(&self, other: &IndexSet<T, S2>) -> bool
1222 where
1223 S2: BuildHasher,
1224 {
1225 self.len() <= other.len() && self.iter().all(move |value| other.contains(value))
1226 }
1227
1228 /// Returns `true` if all elements of `other` are contained in `self`.
1229 pub fn is_superset<S2>(&self, other: &IndexSet<T, S2>) -> bool
1230 where
1231 S2: BuildHasher,
1232 {
1233 other.is_subset(self)
1234 }
1235}
1236
1237impl<T, S1, S2> BitAnd<&IndexSet<T, S2>> for &IndexSet<T, S1>
1238where
1239 T: Eq + Hash + Clone,
1240 S1: BuildHasher + Default,
1241 S2: BuildHasher,
1242{
1243 type Output = IndexSet<T, S1>;
1244
1245 /// Returns the set intersection, cloned into a new set.
1246 ///
1247 /// Values are collected in the same order that they appear in `self`.
1248 fn bitand(self, other: &IndexSet<T, S2>) -> Self::Output {
1249 self.intersection(other).cloned().collect()
1250 }
1251}
1252
1253impl<T, S1, S2> BitOr<&IndexSet<T, S2>> for &IndexSet<T, S1>
1254where
1255 T: Eq + Hash + Clone,
1256 S1: BuildHasher + Default,
1257 S2: BuildHasher,
1258{
1259 type Output = IndexSet<T, S1>;
1260
1261 /// Returns the set union, cloned into a new set.
1262 ///
1263 /// Values from `self` are collected in their original order, followed by
1264 /// values that are unique to `other` in their original order.
1265 fn bitor(self, other: &IndexSet<T, S2>) -> Self::Output {
1266 self.union(other).cloned().collect()
1267 }
1268}
1269
1270impl<T, S1, S2> BitXor<&IndexSet<T, S2>> for &IndexSet<T, S1>
1271where
1272 T: Eq + Hash + Clone,
1273 S1: BuildHasher + Default,
1274 S2: BuildHasher,
1275{
1276 type Output = IndexSet<T, S1>;
1277
1278 /// Returns the set symmetric-difference, cloned into a new set.
1279 ///
1280 /// Values from `self` are collected in their original order, followed by
1281 /// values from `other` in their original order.
1282 fn bitxor(self, other: &IndexSet<T, S2>) -> Self::Output {
1283 self.symmetric_difference(other).cloned().collect()
1284 }
1285}
1286
1287impl<T, S1, S2> Sub<&IndexSet<T, S2>> for &IndexSet<T, S1>
1288where
1289 T: Eq + Hash + Clone,
1290 S1: BuildHasher + Default,
1291 S2: BuildHasher,
1292{
1293 type Output = IndexSet<T, S1>;
1294
1295 /// Returns the set difference, cloned into a new set.
1296 ///
1297 /// Values are collected in the same order that they appear in `self`.
1298 fn sub(self, other: &IndexSet<T, S2>) -> Self::Output {
1299 self.difference(other).cloned().collect()
1300 }
1301}
1302