Module std::iter

1.0.0 · source ·
Expand description

Composable external iteration.

If you’ve found yourself with a collection of some kind, and needed to perform an operation on the elements of said collection, you’ll quickly run into ‘iterators’. Iterators are heavily used in idiomatic Rust code, so it’s worth becoming familiar with them.

Before explaining more, let’s talk about how this module is structured:

Organization

This module is largely organized by type:

  • Traits are the core portion: these traits define what kind of iterators exist and what you can do with them. The methods of these traits are worth putting some extra study time into.
  • Functions provide some helpful ways to create some basic iterators.
  • Structs are often the return types of the various methods on this module’s traits. You’ll usually want to look at the method that creates the struct, rather than the struct itself. For more detail about why, see ‘Implementing Iterator’.

That’s it! Let’s dig into iterators.

Iterator

The heart and soul of this module is the Iterator trait. The core of Iterator looks like this:

trait Iterator {
    type Item;
    fn next(&mut self) -> Option<Self::Item>;
}
Run

An iterator has a method, next, which when called, returns an Option<Item>. Calling next will return Some(Item) as long as there are elements, and once they’ve all been exhausted, will return None to indicate that iteration is finished. Individual iterators may choose to resume iteration, and so calling next again may or may not eventually start returning Some(Item) again at some point (for example, see TryIter).

Iterator’s full definition includes a number of other methods as well, but they are default methods, built on top of next, and so you get them for free.

Iterators are also composable, and it’s common to chain them together to do more complex forms of processing. See the Adapters section below for more details.

The three forms of iteration

There are three common methods which can create iterators from a collection:

  • iter(), which iterates over &T.
  • iter_mut(), which iterates over &mut T.
  • into_iter(), which iterates over T.

Various things in the standard library may implement one or more of the three, where appropriate.

Implementing Iterator

Creating an iterator of your own involves two steps: creating a struct to hold the iterator’s state, and then implementing Iterator for that struct. This is why there are so many structs in this module: there is one for each iterator and iterator adapter.

Let’s make an iterator named Counter which counts from 1 to 5:

// First, the struct:

/// An iterator which counts from one to five
struct Counter {
    count: usize,
}

// we want our count to start at one, so let's add a new() method to help.
// This isn't strictly necessary, but is convenient. Note that we start
// `count` at zero, we'll see why in `next()`'s implementation below.
impl Counter {
    fn new() -> Counter {
        Counter { count: 0 }
    }
}

// Then, we implement `Iterator` for our `Counter`:

impl Iterator for Counter {
    // we will be counting with usize
    type Item = usize;

    // next() is the only required method
    fn next(&mut self) -> Option<Self::Item> {
        // Increment our count. This is why we started at zero.
        self.count += 1;

        // Check to see if we've finished counting or not.
        if self.count < 6 {
            Some(self.count)
        } else {
            None
        }
    }
}

// And now we can use it!

let mut counter = Counter::new();

assert_eq!(counter.next(), Some(1));
assert_eq!(counter.next(), Some(2));
assert_eq!(counter.next(), Some(3));
assert_eq!(counter.next(), Some(4));
assert_eq!(counter.next(), Some(5));
assert_eq!(counter.next(), None);
Run

Calling next this way gets repetitive. Rust has a construct which can call next on your iterator, until it reaches None. Let’s go over that next.

Also note that Iterator provides a default implementation of methods such as nth and fold which call next internally. However, it is also possible to write a custom implementation of methods like nth and fold if an iterator can compute them more efficiently without calling next.

for loops and IntoIterator

Rust’s for loop syntax is actually sugar for iterators. Here’s a basic example of for:

let values = vec![1, 2, 3, 4, 5];

for x in values {
    println!("{x}");
}
Run

This will print the numbers one through five, each on their own line. But you’ll notice something here: we never called anything on our vector to produce an iterator. What gives?

There’s a trait in the standard library for converting something into an iterator: IntoIterator. This trait has one method, into_iter, which converts the thing implementing IntoIterator into an iterator. Let’s take a look at that for loop again, and what the compiler converts it into:

let values = vec![1, 2, 3, 4, 5];

for x in values {
    println!("{x}");
}
Run

Rust de-sugars this into:

let values = vec![1, 2, 3, 4, 5];
{
    let result = match IntoIterator::into_iter(values) {
        mut iter => loop {
            let next;
            match iter.next() {
                Some(val) => next = val,
                None => break,
            };
            let x = next;
            let () = { println!("{x}"); };
        },
    };
    result
}
Run

First, we call into_iter() on the value. Then, we match on the iterator that returns, calling next over and over until we see a None. At that point, we break out of the loop, and we’re done iterating.

There’s one more subtle bit here: the standard library contains an interesting implementation of IntoIterator:

impl<I: Iterator> IntoIterator for I
Run

In other words, all Iterators implement IntoIterator, by just returning themselves. This means two things:

  1. If you’re writing an Iterator, you can use it with a for loop.
  2. If you’re creating a collection, implementing IntoIterator for it will allow your collection to be used with the for loop.

Iterating by reference

Since into_iter() takes self by value, using a for loop to iterate over a collection consumes that collection. Often, you may want to iterate over a collection without consuming it. Many collections offer methods that provide iterators over references, conventionally called iter() and iter_mut() respectively:

let mut values = vec![41];
for x in values.iter_mut() {
    *x += 1;
}
for x in values.iter() {
    assert_eq!(*x, 42);
}
assert_eq!(values.len(), 1); // `values` is still owned by this function.
Run

If a collection type C provides iter(), it usually also implements IntoIterator for &C, with an implementation that just calls iter(). Likewise, a collection C that provides iter_mut() generally implements IntoIterator for &mut C by delegating to iter_mut(). This enables a convenient shorthand:

let mut values = vec![41];
for x in &mut values { // same as `values.iter_mut()`
    *x += 1;
}
for x in &values { // same as `values.iter()`
    assert_eq!(*x, 42);
}
assert_eq!(values.len(), 1);
Run

While many collections offer iter(), not all offer iter_mut(). For example, mutating the keys of a HashSet<T> could put the collection into an inconsistent state if the key hashes change, so this collection only offers iter().

Adapters

Functions which take an Iterator and return another Iterator are often called ‘iterator adapters’, as they’re a form of the ‘adapter pattern’.

Common iterator adapters include map, take, and filter. For more, see their documentation.

If an iterator adapter panics, the iterator will be in an unspecified (but memory safe) state. This state is also not guaranteed to stay the same across versions of Rust, so you should avoid relying on the exact values returned by an iterator which panicked.

Laziness

Iterators (and iterator adapters) are lazy. This means that just creating an iterator doesn’t do a whole lot. Nothing really happens until you call next. This is sometimes a source of confusion when creating an iterator solely for its side effects. For example, the map method calls a closure on each element it iterates over:

let v = vec![1, 2, 3, 4, 5];
v.iter().map(|x| println!("{x}"));
Run

This will not print any values, as we only created an iterator, rather than using it. The compiler will warn us about this kind of behavior:

warning: unused result that must be used: iterators are lazy and
do nothing unless consumed

The idiomatic way to write a map for its side effects is to use a for loop or call the for_each method:

let v = vec![1, 2, 3, 4, 5];

v.iter().for_each(|x| println!("{x}"));
// or
for x in &v {
    println!("{x}");
}
Run

Another common way to evaluate an iterator is to use the collect method to produce a new collection.

Infinity

Iterators do not have to be finite. As an example, an open-ended range is an infinite iterator:

let numbers = 0..;
Run

It is common to use the take iterator adapter to turn an infinite iterator into a finite one:

let numbers = 0..;
let five_numbers = numbers.take(5);

for number in five_numbers {
    println!("{number}");
}
Run

This will print the numbers 0 through 4, each on their own line.

Bear in mind that methods on infinite iterators, even those for which a result can be determined mathematically in finite time, might not terminate. Specifically, methods such as min, which in the general case require traversing every element in the iterator, are likely not to return successfully for any infinite iterators.

let ones = std::iter::repeat(1);
let least = ones.min().unwrap(); // Oh no! An infinite loop!
// `ones.min()` causes an infinite loop, so we won't reach this point!
println!("The smallest number one is {least}.");
Run

Structs

  • ArrayChunksExperimental
    An iterator over N elements of the iterator at a time.
  • ByRefSizedExperimental
    Like Iterator::by_ref, but requiring Sized so it can forward generics.
  • IntersperseExperimental
    An iterator adapter that places a separator between all elements.
  • IntersperseWithExperimental
    An iterator adapter that places a separator between all elements.
  • MapWindowsExperimental
    An iterator over the mapped windows of another iterator.
  • An iterator that links two iterators together, in a chain.
  • An iterator that clones the elements of an underlying iterator.
  • An iterator that copies the elements of an underlying iterator.
  • An iterator that repeats endlessly.
  • An iterator that yields nothing.
  • An iterator that yields the current count and the element during iteration.
  • An iterator that filters the elements of iter with predicate.
  • An iterator that uses f to both filter and map elements from iter.
  • An iterator that maps each element to an iterator, and yields the elements of the produced iterators.
  • An iterator that flattens one level of nesting in an iterator of things that can be turned into iterators.
  • An iterator where each iteration calls the provided closure F: FnMut() -> Option<T>.
  • An iterator that yields None forever after the underlying iterator yields None once.
  • An iterator that calls a function with a reference to each element before yielding it.
  • An iterator that maps the values of iter with f.
  • An iterator that only accepts elements while predicate returns Some(_).
  • An iterator that yields an element exactly once.
  • An iterator that yields a single element of type A by applying the provided closure F: FnOnce() -> A.
  • An iterator with a peek() that returns an optional reference to the next element.
  • An iterator that repeats an element endlessly.
  • An iterator that repeats elements of type A endlessly by applying the provided closure F: FnMut() -> A.
  • A double-ended iterator with the direction inverted.
  • An iterator to maintain state while iterating another iterator.
  • An iterator that skips over n elements of iter.
  • An iterator that rejects elements while predicate returns true.
  • An iterator for stepping iterators by a custom amount.
  • A new iterator where each successive item is computed based on the preceding one.
  • An iterator that only iterates over the first n iterations of iter.
  • An iterator that only accepts elements while predicate returns true.
  • An iterator that iterates two other iterators simultaneously.

Traits

  • StepExperimental
    Objects that have a notion of successor and predecessor operations.
  • TrustedLenExperimental
    An iterator that reports an accurate length using size_hint.
  • TrustedStepExperimental
    A type that upholds all invariants of Step.
  • An iterator able to yield elements from both ends.
  • An iterator that knows its exact length.
  • Extend a collection with the contents of an iterator.
  • Conversion from an Iterator.
  • An iterator that always continues to yield None when exhausted.
  • Conversion into an Iterator.
  • A trait for dealing with iterators.
  • Trait to represent types that can be created by multiplying elements of an iterator.
  • Trait to represent types that can be created by summing up an iterator.

Functions

  • from_coroutineExperimental
    Creates a new iterator where each iteration calls the provided coroutine.
  • Creates an iterator that yields nothing.
  • Creates a new iterator where each iteration calls the provided closure F: FnMut() -> Option<T>.
  • Creates an iterator that yields an element exactly once.
  • Creates an iterator that lazily generates a value exactly once by invoking the provided closure.
  • Creates a new iterator that endlessly repeats a single element.
  • Creates a new iterator that repeats elements of type A endlessly by applying the provided closure, the repeater, F: FnMut() -> A.
  • Creates a new iterator where each successive item is computed based on the preceding one.
  • Converts the arguments to iterators and zips them.