immutable_chunkmap/
set.rs

1use crate::avl::{Iter, Tree, WeakTree};
2pub use crate::chunk::DEFAULT_SIZE;
3use core::{
4    borrow::Borrow,
5    cmp::{Eq, Ord, Ordering, PartialEq, PartialOrd},
6    default::Default,
7    fmt::{self, Debug, Formatter},
8    hash::{Hash, Hasher},
9    iter::FromIterator,
10    ops::{RangeBounds, RangeFull},
11};
12
13#[cfg(feature = "serde")]
14use serde::{
15    de::{SeqAccess, Visitor},
16    ser::SerializeSeq,
17    Deserialize, Deserializer, Serialize, Serializer,
18};
19
20#[cfg(feature = "serde")]
21use core::marker::PhantomData;
22
23#[cfg(feature = "rayon")]
24use rayon::{
25    iter::{FromParallelIterator, IntoParallelIterator},
26    prelude::*,
27};
28
29/// This set uses a similar strategy to BTreeSet to ensure cache
30/// efficient performance on modern hardware while still providing
31/// log(N) get, insert, and remove operations.
32/// # Examples
33/// ```
34/// # extern crate alloc;
35/// use alloc::string::String;
36/// use self::immutable_chunkmap::set::SetM;
37///
38/// let m =
39///    SetM::new()
40///    .insert(String::from("1")).0
41///    .insert(String::from("2")).0
42///    .insert(String::from("3")).0;
43///
44/// assert_eq!(m.contains("1"), true);
45/// assert_eq!(m.contains("2"), true);
46/// assert_eq!(m.contains("3"), true);
47/// assert_eq!(m.contains("4"), false);
48///
49/// for k in &m { println!("{}", k) }
50/// ```
51#[derive(Clone)]
52#[repr(transparent)]
53pub struct Set<K: Ord + Clone, const SIZE: usize>(Tree<K, (), SIZE>);
54
55/// set with a smaller chunk size, faster to update, slower to search
56pub type SetS<K> = Set<K, { DEFAULT_SIZE / 2 }>;
57
58/// set with the default chunk size, a good balance of search and update performance
59pub type SetM<K> = Set<K, DEFAULT_SIZE>;
60
61/// set with a larger chunk size, faster to search, slower to update
62pub type SetL<K> = Set<K, { DEFAULT_SIZE * 2 }>;
63
64#[derive(Clone)]
65pub struct WeakSetRef<K: Ord + Clone, const SIZE: usize>(WeakTree<K, (), SIZE>);
66
67pub type WeakSetRefS<K> = WeakSetRef<K, 32>;
68pub type WeakSetRefM<K> = WeakSetRef<K, 128>;
69pub type WeakSetRefL<K> = WeakSetRef<K, 512>;
70
71impl<K, const SIZE: usize> WeakSetRef<K, SIZE>
72where
73    K: Ord + Clone,
74{
75    pub fn upgrade(&self) -> Option<Set<K, SIZE>> {
76        self.0.upgrade().map(Set)
77    }
78}
79
80impl<K, const SIZE: usize> Hash for Set<K, SIZE>
81where
82    K: Hash + Ord + Clone,
83{
84    fn hash<H: Hasher>(&self, state: &mut H) {
85        self.0.hash(state)
86    }
87}
88
89impl<K, const SIZE: usize> Default for Set<K, SIZE>
90where
91    K: Ord + Clone,
92{
93    fn default() -> Set<K, SIZE> {
94        Set::new()
95    }
96}
97
98impl<K, const SIZE: usize> PartialEq for Set<K, SIZE>
99where
100    K: Ord + Clone,
101{
102    fn eq(&self, other: &Set<K, SIZE>) -> bool {
103        self.0 == other.0
104    }
105}
106
107impl<K, const SIZE: usize> Eq for Set<K, SIZE> where K: Eq + Ord + Clone {}
108
109impl<K, const SIZE: usize> PartialOrd for Set<K, SIZE>
110where
111    K: Ord + Clone,
112{
113    fn partial_cmp(&self, other: &Set<K, SIZE>) -> Option<Ordering> {
114        self.0.partial_cmp(&other.0)
115    }
116}
117
118impl<K, const SIZE: usize> Ord for Set<K, SIZE>
119where
120    K: Ord + Clone,
121{
122    fn cmp(&self, other: &Set<K, SIZE>) -> Ordering {
123        self.0.cmp(&other.0)
124    }
125}
126
127impl<K, const SIZE: usize> Debug for Set<K, SIZE>
128where
129    K: Debug + Ord + Clone,
130{
131    fn fmt(&self, f: &mut Formatter) -> fmt::Result {
132        f.debug_set().entries(self.into_iter()).finish()
133    }
134}
135
136impl<K, const SIZE: usize> FromIterator<K> for Set<K, SIZE>
137where
138    K: Ord + Clone,
139{
140    fn from_iter<T: IntoIterator<Item = K>>(iter: T) -> Self {
141        Set::new().insert_many(iter)
142    }
143}
144
145pub struct SetIter<
146    'a,
147    R: RangeBounds<Q> + 'a,
148    Q: Ord + ?Sized,
149    K: 'a + Clone + Ord + Borrow<Q>,
150    const SIZE: usize,
151>(Iter<'a, R, Q, K, (), SIZE>);
152
153impl<'a, R, Q, K, const SIZE: usize> Iterator for SetIter<'a, R, Q, K, SIZE>
154where
155    Q: Ord + ?Sized,
156    R: RangeBounds<Q> + 'a,
157    K: 'a + Clone + Ord + Borrow<Q>,
158{
159    type Item = &'a K;
160    fn next(&mut self) -> Option<Self::Item> {
161        self.0.next().map(|(k, ())| k)
162    }
163}
164
165impl<'a, R, Q, K, const SIZE: usize> DoubleEndedIterator for SetIter<'a, R, Q, K, SIZE>
166where
167    Q: Ord + ?Sized,
168    R: RangeBounds<Q> + 'a,
169    K: 'a + Clone + Ord + Borrow<Q>,
170{
171    fn next_back(&mut self) -> Option<Self::Item> {
172        self.0.next_back().map(|(k, ())| k)
173    }
174}
175
176impl<'a, K, const SIZE: usize> IntoIterator for &'a Set<K, SIZE>
177where
178    K: 'a + Ord + Clone,
179{
180    type Item = &'a K;
181    type IntoIter = SetIter<'a, RangeFull, K, K, SIZE>;
182    fn into_iter(self) -> Self::IntoIter {
183        SetIter(self.0.into_iter())
184    }
185}
186
187#[cfg(feature = "serde")]
188impl<V, const SIZE: usize> Serialize for Set<V, SIZE>
189where
190    V: Serialize + Clone + Ord,
191{
192    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
193    where
194        S: Serializer,
195    {
196        let mut seq = serializer.serialize_seq(Some(self.len()))?;
197        for v in self {
198            seq.serialize_element(v)?
199        }
200        seq.end()
201    }
202}
203
204#[cfg(feature = "serde")]
205struct SetVisitor<V: Clone + Ord, const SIZE: usize> {
206    marker: PhantomData<fn() -> Set<V, SIZE>>,
207}
208
209#[cfg(feature = "serde")]
210impl<'a, V, const SIZE: usize> Visitor<'a> for SetVisitor<V, SIZE>
211where
212    V: Deserialize<'a> + Clone + Ord,
213{
214    type Value = Set<V, SIZE>;
215
216    fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
217        formatter.write_str("expecting an immutable_chunkmap::Set")
218    }
219
220    fn visit_seq<A>(self, mut seq: A) -> Result<Self::Value, A::Error>
221    where
222        A: SeqAccess<'a>,
223    {
224        let mut t = Set::<V, SIZE>::new();
225        while let Some(v) = seq.next_element()? {
226            t.insert_cow(v);
227        }
228        Ok(t)
229    }
230}
231
232#[cfg(feature = "serde")]
233impl<'a, V, const SIZE: usize> Deserialize<'a> for Set<V, SIZE>
234where
235    V: Deserialize<'a> + Clone + Ord,
236{
237    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
238    where
239        D: Deserializer<'a>,
240    {
241        deserializer.deserialize_seq(SetVisitor {
242            marker: PhantomData,
243        })
244    }
245}
246
247#[cfg(feature = "rayon")]
248impl<'a, V, const SIZE: usize> IntoParallelIterator for &'a Set<V, SIZE>
249where
250    V: 'a + Ord + Clone + Send + Sync,
251{
252    type Item = &'a V;
253    type Iter = rayon::vec::IntoIter<&'a V>;
254
255    fn into_par_iter(self) -> Self::Iter {
256        self.into_iter().collect::<Vec<_>>().into_par_iter()
257    }
258}
259
260#[cfg(feature = "rayon")]
261impl<V, const SIZE: usize> FromParallelIterator<V> for Set<V, SIZE>
262where
263    V: Ord + Clone + Send + Sync,
264{
265    fn from_par_iter<I>(i: I) -> Self
266    where
267        I: IntoParallelIterator<Item = V>,
268    {
269        i.into_par_iter()
270            .fold_with(Set::new(), |mut m, v| {
271                m.insert_cow(v);
272                m
273            })
274            .reduce_with(|m0, m1| m0.union(&m1))
275            .unwrap_or_else(Set::new)
276    }
277}
278
279impl<K, const SIZE: usize> Set<K, SIZE>
280where
281    K: Ord + Clone,
282{
283    /// Create a new empty set
284    pub fn new() -> Self {
285        Set(Tree::new())
286    }
287
288    /// Create a weak reference to this set
289    pub fn downgrade(&self) -> WeakSetRef<K, SIZE> {
290        WeakSetRef(self.0.downgrade())
291    }
292
293    /// Return the number of strong references to this set (see Arc)
294    pub fn strong_count(&self) -> usize {
295        self.0.strong_count()
296    }
297
298    /// Return the number of weak references to this set (see Arc)
299    pub fn weak_count(&self) -> usize {
300        self.0.weak_count()
301    }
302
303    /// This will insert many elements at once, and is
304    /// potentially a lot faster than inserting one by one,
305    /// especially if the data is sorted.
306    ///
307    /// #Examples
308    ///```
309    /// use self::immutable_chunkmap::set::SetM;
310    ///
311    /// let mut v = vec![1, 10, -12, 44, 50];
312    /// v.sort_unstable();
313    ///
314    /// let m = SetM::new().insert_many(v.iter().map(|k| *k));
315    ///
316    /// for k in &v {
317    ///   assert_eq!(m.contains(k), true)
318    /// }
319    /// ```
320    pub fn insert_many<E: IntoIterator<Item = K>>(&self, elts: E) -> Self {
321        let root = self.0.insert_many(elts.into_iter().map(|k| (k, ())));
322        Set(root)
323    }
324
325    /// Remove multiple elements in a single pass. Similar performance
326    /// to insert_many.
327    pub fn remove_many<Q, E>(&self, elts: E) -> Self
328    where
329        Q: Ord,
330        K: Borrow<Q>,
331        E: IntoIterator<Item = Q>,
332    {
333        let root = self
334            .0
335            .update_many(elts.into_iter().map(|k| (k, ())), &mut |_, _, _| None);
336        Set(root)
337    }
338
339    /// This is just slightly wierd, however if you have a bunch of
340    /// borrowed forms of members of the set, and you want to look at
341    /// the real entries and possibly add/update/remove them, then
342    /// this method is for you.
343    pub fn update_many<Q, E, F>(&self, elts: E, mut f: F) -> Self
344    where
345        Q: Ord,
346        K: Borrow<Q>,
347        E: IntoIterator<Item = Q>,
348        F: FnMut(Q, Option<&K>) -> Option<K>,
349    {
350        let root =
351            self.0
352                .update_many(elts.into_iter().map(|k| (k, ())), &mut |q, (), cur| {
353                    let cur = cur.map(|(k, ())| k);
354                    f(q, cur).map(|k| (k, ()))
355                });
356        Set(root)
357    }
358
359    /// return a new set with k inserted into it. If k already
360    /// exists in the old set return true, else false. If the
361    /// element already exists in the set memory will not be
362    /// allocated.
363    pub fn insert(&self, k: K) -> (Self, bool) {
364        if self.contains(&k) {
365            (self.clone(), true)
366        } else {
367            (Set(self.0.insert(k, ()).0), false)
368        }
369    }
370
371    /// insert `k` with copy on write semantics. if `self` is a unique
372    /// reference to the set, then k will be inserted in
373    /// place. Otherwise, only the parts of the set necessary to
374    /// insert `k` will be copied, and then the copies will be
375    /// mutated. self will share all the parts that weren't modfied
376    /// with any previous clones.
377    pub fn insert_cow(&mut self, k: K) -> bool {
378        self.0.insert_cow(k, ()).is_some()
379    }
380
381    /// return true if the set contains k, else false. Runs in
382    /// log(N) time and constant space. where N is the size of
383    /// the set.
384    pub fn contains<'a, Q>(&'a self, k: &Q) -> bool
385    where
386        Q: ?Sized + Ord,
387        K: Borrow<Q>,
388    {
389        self.0.get(k).is_some()
390    }
391
392    /// return a reference to the item in the set that is equal to the
393    /// given value, or None if no such value exists.
394    pub fn get<'a, Q>(&'a self, k: &Q) -> Option<&'a K>
395    where
396        Q: ?Sized + Ord,
397        K: Borrow<Q>,
398    {
399        self.0.get_key(k)
400    }
401
402    /// return a new set with k removed. Runs in log(N) time
403    /// and log(N) space, where N is the size of the set
404    pub fn remove<Q: Sized + Ord>(&self, k: &Q) -> (Self, bool)
405    where
406        K: Borrow<Q>,
407    {
408        let (t, prev) = self.0.remove(k);
409        (Set(t), prev.is_some())
410    }
411
412    /// remove `k` from the set in place with copy on write semantics
413    /// (see `insert_cow`). return true if `k` was in the set.
414    pub fn remove_cow<Q: Sized + Ord>(&mut self, k: &Q) -> bool
415    where
416        K: Borrow<Q>,
417    {
418        self.0.remove_cow(k).is_some()
419    }
420
421    /// return the union of 2 sets. Runs in O(log(N) + M) time and
422    /// space, where N is the largest of the two sets, and M is the
423    /// number of chunks that intersect, which is roughly proportional
424    /// to the size of the intersection.
425    ///
426    /// # Examples
427    /// ```
428    /// use core::iter::FromIterator;
429    /// use self::immutable_chunkmap::set::SetM;
430    ///
431    /// let s0 = SetM::from_iter(0..10);
432    /// let s1 = SetM::from_iter(5..15);
433    /// let s2 = s0.union(&s1);
434    /// for i in 0..15 {
435    ///     assert!(s2.contains(&i));
436    /// }
437    /// ```
438    pub fn union(&self, other: &Set<K, SIZE>) -> Self {
439        Set(Tree::union(&self.0, &other.0, &mut |_, (), ()| Some(())))
440    }
441
442    /// return the intersection of 2 sets. Runs in O(log(N) + M) time
443    /// and space, where N is the smallest of the two sets, and M is
444    /// the number of intersecting chunks.
445    ///
446    /// # Examples
447    /// use core::iter::FromIterator;
448    /// use self::immutable_chunkmap::set::SetM;
449    ///
450    /// let s0 = SetM::from_iter(0..100);
451    /// let s1 = SetM::from_iter(20..50);
452    /// let s2 = s0.intersect(&s1);
453    ///
454    /// assert!(s2.len() == 30);
455    /// for i in 0..100 {
456    ///     if i < 20 || i >= 50 {
457    ///         assert!(!s2.contains(&i));
458    ///     } else {
459    ///         assert!(s2.contains(&i));
460    ///     }
461    /// }
462    pub fn intersect(&self, other: &Set<K, SIZE>) -> Self {
463        Set(Tree::intersect(
464            &self.0,
465            &other.0,
466            &mut |_, (), ()| Some(()),
467        ))
468    }
469
470    /// Return the difference of two sets. Runs in O(log(N) + M) time
471    /// and space, where N is the smallest of the two sets, and M is
472    /// the number of intersecting chunks.
473    ///
474    /// # Examples
475    /// ```
476    /// use core::iter::FromIterator;
477    /// use self::immutable_chunkmap::set::SetM;
478    ///
479    /// let s0 = SetM::from_iter(0..100);
480    /// let s1 = SetM::from_iter(0..50);
481    /// let s2 = s0.diff(&s1);
482    ///
483    /// assert!(s2.len() == 50);
484    /// for i in 0..50 {
485    ///     assert!(!s2.contains(&i));
486    /// }
487    /// for i in 50..100 {
488    ///     assert!(s2.contains(&i));
489    /// }
490    /// ```
491    pub fn diff(&self, other: &Set<K, SIZE>) -> Self
492    where
493        K: Debug,
494    {
495        Set(Tree::diff(&self.0, &other.0, &mut |_, (), ()| None))
496    }
497
498    /// get the number of elements in the map O(1) time and space
499    pub fn len(&self) -> usize {
500        self.0.len()
501    }
502
503    /// return an iterator over the subset of elements in the
504    /// set that are within the specified range.
505    ///
506    /// The returned iterator runs in O(log(N) + M) time, and
507    /// constant space. N is the number of elements in the
508    /// tree, and M is the number of elements you examine.
509    ///
510    /// if lbound >= ubound the returned iterator will be empty
511    pub fn range<'a, Q, R>(&'a self, r: R) -> SetIter<'a, R, Q, K, SIZE>
512    where
513        Q: Ord + ?Sized + 'a,
514        K: 'a + Clone + Ord + Borrow<Q>,
515        R: RangeBounds<Q> + 'a,
516    {
517        SetIter(self.0.range(r))
518    }
519}
520
521impl<K, const SIZE: usize> Set<K, SIZE>
522where
523    K: Ord + Clone + Debug,
524{
525    #[allow(dead_code)]
526    pub(crate) fn invariant(&self) -> () {
527        self.0.invariant()
528    }
529}
immutable_chunkmap/set.rs

immutable_chunkmap/
set.rs
