1use std::collections::hash_map::Entry;
2use std::collections::HashMap;
3use std::fmt;
4use std::hash::Hash;
5use std::iter::FusedIterator;
6
7/// An iterator adapter to filter out duplicate elements.
8///
9/// See [`.unique_by()`](crate::Itertools::unique) for more information.
10#[derive(Clone)]
11#[must_use = "iterator adaptors are lazy and do nothing unless consumed"]
12pub struct UniqueBy<I: Iterator, V, F> {
13 iter: I,
14 // Use a Hashmap for the Entry API in order to prevent hashing twice.
15 // This can maybe be replaced with a HashSet once `get_or_insert_with`
16 // or a proper Entry API for Hashset is stable and meets this msrv
17 used: HashMap<V, ()>,
18 f: F,
19}
20
21impl<I, V, F> fmt::Debug for UniqueBy<I, V, F>
22where
23 I: Iterator + fmt::Debug,
24 V: fmt::Debug + Hash + Eq,
25{
26 debug_fmt_fields!(UniqueBy, iter, used);
27}
28
29/// Create a new `UniqueBy` iterator.
30pub fn unique_by<I, V, F>(iter: I, f: F) -> UniqueBy<I, V, F>
31where
32 V: Eq + Hash,
33 F: FnMut(&I::Item) -> V,
34 I: Iterator,
35{
36 UniqueBy {
37 iter,
38 used: HashMap::new(),
39 f,
40 }
41}
42
43// count the number of new unique keys in iterable (`used` is the set already seen)
44fn count_new_keys<I, K>(mut used: HashMap<K, ()>, iterable: I) -> usize
45where
46 I: IntoIterator<Item = K>,
47 K: Hash + Eq,
48{
49 let iter: ::IntoIter = iterable.into_iter();
50 let current_used: usize = used.len();
51 used.extend(iter.map(|key: K| (key, ())));
52 used.len() - current_used
53}
54
55impl<I, V, F> Iterator for UniqueBy<I, V, F>
56where
57 I: Iterator,
58 V: Eq + Hash,
59 F: FnMut(&I::Item) -> V,
60{
61 type Item = I::Item;
62
63 fn next(&mut self) -> Option<Self::Item> {
64 let Self { iter: &mut I, used: &mut HashMap, f: &mut F } = self;
65 iter.find(|v: &::Item| used.insert(k:f(v), ()).is_none())
66 }
67
68 #[inline]
69 fn size_hint(&self) -> (usize, Option<usize>) {
70 let (low: usize, hi: Option) = self.iter.size_hint();
71 ((low > 0 && self.used.is_empty()) as usize, hi)
72 }
73
74 fn count(self) -> usize {
75 let mut key_f: F = self.f;
76 count_new_keys(self.used, self.iter.map(move |elt: ::Item| key_f(&elt)))
77 }
78}
79
80impl<I, V, F> DoubleEndedIterator for UniqueBy<I, V, F>
81where
82 I: DoubleEndedIterator,
83 V: Eq + Hash,
84 F: FnMut(&I::Item) -> V,
85{
86 fn next_back(&mut self) -> Option<Self::Item> {
87 let Self { iter: &mut I, used: &mut HashMap, f: &mut F } = self;
88 iter.rfind(|v: &::Item| used.insert(k:f(v), ()).is_none())
89 }
90}
91
92impl<I, V, F> FusedIterator for UniqueBy<I, V, F>
93where
94 I: FusedIterator,
95 V: Eq + Hash,
96 F: FnMut(&I::Item) -> V,
97{
98}
99
100impl<I> Iterator for Unique<I>
101where
102 I: Iterator,
103 I::Item: Eq + Hash + Clone,
104{
105 type Item = I::Item;
106
107 fn next(&mut self) -> Option<Self::Item> {
108 let UniqueBy { iter, used, .. } = &mut self.iter;
109 iter.find_map(|v| {
110 if let Entry::Vacant(entry) = used.entry(v) {
111 let elt = entry.key().clone();
112 entry.insert(());
113 return Some(elt);
114 }
115 None
116 })
117 }
118
119 #[inline]
120 fn size_hint(&self) -> (usize, Option<usize>) {
121 let (low, hi) = self.iter.iter.size_hint();
122 ((low > 0 && self.iter.used.is_empty()) as usize, hi)
123 }
124
125 fn count(self) -> usize {
126 count_new_keys(self.iter.used, self.iter.iter)
127 }
128}
129
130impl<I> DoubleEndedIterator for Unique<I>
131where
132 I: DoubleEndedIterator,
133 I::Item: Eq + Hash + Clone,
134{
135 fn next_back(&mut self) -> Option<Self::Item> {
136 let UniqueBy { iter: &mut I, used: &mut HashMap, .. } = &mut self.iter;
137 iter.rev().find_map(|v: impl Eq + Hash + Clone| {
138 if let Entry::Vacant(entry: VacantEntry<'_, impl Eq + Hash + Clone, …>) = used.entry(key:v) {
139 let elt: impl Eq + Hash + Clone = entry.key().clone();
140 entry.insert(());
141 return Some(elt);
142 }
143 None
144 })
145 }
146}
147
148impl<I> FusedIterator for Unique<I>
149where
150 I: FusedIterator,
151 I::Item: Eq + Hash + Clone,
152{
153}
154
155/// An iterator adapter to filter out duplicate elements.
156///
157/// See [`.unique()`](crate::Itertools::unique) for more information.
158#[derive(Clone)]
159#[must_use = "iterator adaptors are lazy and do nothing unless consumed"]
160pub struct Unique<I>
161where
162 I: Iterator,
163 I::Item: Eq + Hash + Clone,
164{
165 iter: UniqueBy<I, I::Item, ()>,
166}
167
168impl<I> fmt::Debug for Unique<I>
169where
170 I: Iterator + fmt::Debug,
171 I::Item: Hash + Eq + fmt::Debug + Clone,
172{
173 debug_fmt_fields!(Unique, iter);
174}
175
176pub fn unique<I>(iter: I) -> Unique<I>
177where
178 I: Iterator,
179 I::Item: Eq + Hash + Clone,
180{
181 Unique {
182 iter: UniqueBy {
183 iter,
184 used: HashMap::new(),
185 f: (),
186 },
187 }
188}
189