I have a Vec<(String, i64)> and need to iterate over the Strings and move them and then iterate over the i64s.
However, if I move the Strings I have to store the i64 again into another Vec:
let l: Vec<_> = l
.into_iter()
.map(|(string, int)| {
drop(string);
int
})
.collect();
for i in l {
process(i);
}
How can I iterate over the Strings and i64s separately without incurring any additional performance overhead.
The only solution I can think of at the moment that will not cause additional operations is to store the Strings and i64s separately.
You can use std::mem::take() while iterating over the Vec in a first pass to take ownership of the String element while putting a non-allocating Default in its place. This allows you to keep the Vec in its original form, so no extra container is required.
fn foo(mut inp: Vec<(String, i64)>) {
// First pass over the Vec "extracts" the owned Strings, replacing the content
// in the Vec by a non-allocating empty String, which is close to zero cost;
// this leaves the Vec as is, so no intermediate representation is needed.
for s in inp.iter_mut().map(|(s, _)| std::mem::take(s)) {
// Process String
}
// Something happens
// Second pass ignores the empty strings, processes the integers
for i in inp.into_iter().map(|(_, i)| i) {
// Process the integers
}
}
If the type of the list can be changed to Vec<Option<String>, i64> from Vec<String, i64>, then you can try the following way.
fn main() {
let mut l = Vec::new();
l.push((Some("a".to_string()), 1i64));
l.push((Some("b".to_string()), 2));
l.push((Some("c".to_string()), 3));
l.push((Some("d".to_string()), 4));
l.iter_mut().for_each(|(s, _)| {
if let Some(x) = s.take() {
println!("Processing string: {}", x);
}
});
l.iter().for_each(|(_, i)| {
println!("Processing int: {}", i);
});
}
Playground
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With