Product.aggregate([
{
$match: {
_id: "5cc9441feed4c258881c99cd"
}
},
{
$lookup: {
from: "comments",
localField: "_id",
foreignField: "product",
as: "comments"
}
},
{
$unwind: "$comments"
},
{
$lookup: {
from: "users",
localField: "comments.user",
foreignField: "_id",
as: "comments.user"
}
},
{
$unwind: "$comments.user"
},
{
$group: {
_id: "$_id",
// add other fields you want to include
comments: {
$addToSet: "$comments"
}
}
},
])
我不再需要非相交的值。
如何从集合use std::collections::HashSet;
let mut a: HashSet<T> = HashSet::new();
let mut b: HashSet<T> = HashSet::new();
let mut c: HashSet<T> = a.intersection(&b).collect();
// Error: a collection of type `std::collections::HashSet<T>` cannot be built from an iterator over elements of type `&T`
和a
中窃取/移动数据到b
中而不进行复制或克隆?理想情况下,这将具有理论上最佳的时间复杂度:O(min(a,b))。
答案 0 :(得分:4)
由编译器强加的别名规则要求您前后移动值。尽管可以无条件地将值从集合中消耗掉。但是,如果我们跟踪应移动的值和应保留在新集中的值,则可以将某些值发回。然后,retain
允许我们从第二组中删除公用值。
use std::collections::HashSet;
use std::hash::Hash;
/// Extracts the common values in `a` and `b` into a new set.
fn inplace_intersection<T>(a: &mut HashSet<T>, b: &mut HashSet<T>) -> HashSet<T>
where
T: Hash,
T: Eq,
{
let x: HashSet<(T, bool)> = a
.drain()
.map(|v| {
let intersects = b.contains(&v);
(v, intersects)
})
.collect();
let mut c = HashSet::new();
for (v, is_inter) in x {
if is_inter {
c.insert(v);
} else {
a.insert(v);
}
}
b.retain(|v| !c.contains(&v));
c
}
使用:
use itertools::Itertools; // for .sorted()
let mut a: HashSet<_> = [1, 2, 3].iter().cloned().collect();
let mut b: HashSet<_> = [4, 2, 3].iter().cloned().collect();
let c = inplace_intersection(&mut a, &mut b);
let a: Vec<_> = a.into_iter().sorted().collect();
let b: Vec<_> = b.into_iter().sorted().collect();
let c: Vec<_> = c.into_iter().sorted().collect();
assert_eq!(&a, &[1]);
assert_eq!(&b, &[4]);
assert_eq!(&c, &[2, 3]);
答案 1 :(得分:3)
或者,如果您可以对这些集合本身拥有所有权,并且不关心在其他集合中保留非相交的值,则可以执行以下操作:
use std::hash::Hash;
use std::collections::HashSet;
fn intersection<T: Eq + Hash>(a: HashSet<T>, b: &HashSet<T>) -> HashSet<T> {
a.into_iter().filter(|e| b.contains(e)).collect()
}
这将b中包含的a中的元素收集到一个新的HashSet中。
答案 2 :(得分:2)
另一种解决方案,类似于E_net4的解决方案,但是该解决方案不涉及先清空然后重新填充第一组。恕我直言,它也更容易阅读。
fn inplace_intersection<T>(a: &mut HashSet<T>, b: &mut HashSet<T>) -> HashSet<T>
where
T: Hash,
T: Eq,
{
let mut c = HashSet::new();
for v in a.iter() {
if let Some(found) = b.take(v) {
c.insert(found);
}
}
a.retain(|v| !c.contains(&v));
c
}
写完这个之后,我意识到它可以变得更简单:
fn inplace_intersection<T>(a: &mut HashSet<T>, b: &mut HashSet<T>) -> HashSet<T>
where
T: Hash,
T: Eq,
{
let c: HashSet<T> = a.iter().filter_map(|v| b.take(v)).collect();
a.retain(|v| !c.contains(&v));
c
}