Rust Iterators

Iterator Patterns

Efficient data processing with lazy iterator chains.

Basic Iterator Chain

// Single pass, lazy evaluation, no intermediate allocations
let result: Vec<_> = data
    .iter()
    .filter(|&&x| x > 2)
    .map(|&x| x * 2)
    .collect();

// Sum without collecting
let sum: i32 = data.iter().filter(|&&x| x > 0).sum();

// Count matches
let count = data.iter().filter(|x| x.is_valid()).count();

Common Iterator Methods

// Transform each element
.map(|x| transform(x))

// Keep elements matching predicate
.filter(|x| predicate(x))

// Transform + flatten (Option/Result/Vec)
.filter_map(|x| x.ok())          // Filter None/Err, unwrap Some/Ok
.flat_map(|x| x.into_iter())     // Flatten nested iterables

// Take/skip elements
.take(10)                         // First 10 elements
.skip(5)                          // Skip first 5
.take_while(|x| x < &100)        // Take while condition true
.skip_while(|x| x < &10)         // Skip while condition true

// Find elements
.find(|x| predicate(x))          // First match (Option)
.position(|x| predicate(x))      // Index of first match
.any(|x| predicate(x))           // Any match exists
.all(|x| predicate(x))           // All match

// Reduce/fold
.fold(0, |acc, x| acc + x)       // Reduce with initial value
.reduce(|a, b| a + b)            // Reduce without initial
.sum::<i32>()                    // Sum numbers
.product::<i32>()                // Multiply numbers

// Collect to different types
.collect::<Vec<_>>()
.collect::<HashSet<_>>()
.collect::<String>()
.collect::<Result<Vec<_>, _>>()  // Collect Results, fail on first Err

Parallel Iteration with Rayon

use rayon::prelude::*;

// Parallel iteration (automatic work stealing)
let sum: i32 = data.par_iter().map(|x| expensive(x)).sum();

// Parallel sort
let mut data = data;
data.par_sort();

// Parallel filter + map
let results: Vec<_> = data
    .par_iter()
    .filter(|x| predicate(x))
    .map(|x| transform(x))
    .collect();

Enumerate and Zip

// Get index with element
for (index, item) in items.iter().enumerate() {
    println!("{}: {}", index, item);
}

// Pair two iterators
let pairs: Vec<_> = keys.iter().zip(values.iter()).collect();

// Unzip pairs
let (keys, values): (Vec<_>, Vec<_>) = pairs.into_iter().unzip();

Chain and Flatten

// Concatenate iterators
let combined: Vec<_> = first.iter()
    .chain(second.iter())
    .chain(third.iter())
    .collect();

// Flatten nested iterables
let flat: Vec<_> = nested_vecs.iter().flatten().collect();

// Flatten with transformation
let all_items: Vec<_> = containers
    .iter()
    .flat_map(|c| c.items())
    .collect();

Peekable Iterator

let mut iter = data.iter().peekable();

while let Some(&current) = iter.next() {
    // Look ahead without consuming
    if let Some(&&next) = iter.peek() {
        if next > current {
            // Do something
        }
    }
}

Windows and Chunks

// Overlapping windows of size 3
for window in data.windows(3) {
    let [a, b, c] = window else { continue };
    // Process window
}

// Non-overlapping chunks
for chunk in data.chunks(100) {
    process_batch(chunk);
}

// Exact chunks (panic if not divisible)
for chunk in data.chunks_exact(100) {
    process_batch(chunk);
}

Custom Iterator

struct Counter {
    current: usize,
    max: usize,
}

impl Iterator for Counter {
    type Item = usize;

    fn next(&mut self) -> Option<Self::Item> {
        if self.current < self.max {
            let result = self.current;
            self.current += 1;
            Some(result)
        } else {
            None
        }
    }

    // Optional: Provide size hint for better allocation
    fn size_hint(&self) -> (usize, Option<usize>) {
        let remaining = self.max - self.current;
        (remaining, Some(remaining))
    }
}

// ExactSizeIterator if size is known
impl ExactSizeIterator for Counter {}

Collect into Result

// Collect Vec<Result<T, E>> into Result<Vec<T>, E>
let results: Result<Vec<_>, _> = items
    .iter()
    .map(|item| fallible_operation(item))
    .collect();

// Short-circuits on first error
match results {
    Ok(values) => process(values),
    Err(e) => handle_error(e),
}

Performance Tips

// Avoid: Collecting intermediate results
let temp: Vec<_> = data.iter().filter(|x| x > &0).collect();
let result: i32 = temp.iter().sum();

// Better: Chain without collecting
let result: i32 = data.iter().filter(|&&x| x > 0).sum();

// Use iterator methods over manual loops
// Avoid:
let mut sum = 0;
for x in &data {
    if *x > 0 {
        sum += x;
    }
}

// Prefer:
let sum: i32 = data.iter().filter(|&&x| x > 0).sum();

Guidelines

• Prefer iterator chains over manual loops
• Avoid collecting intermediate results
• Use filter_map to combine filter + map for Options
• Use flat_map to flatten nested structures
• Use Rayon for CPU-bound parallel work
• Implement size_hint for custom iterators
• Use chunks for batch processing

Examples

See hercules-local-algo/src/pipeline/ for iterator patterns.