Struct moka::future::Cache

pub struct Cache<K, V, S = RandomState> { /* fields omitted */ }

A thread-safe, futures-aware concurrent in-memory cache.

Cache supports full concurrency of retrievals and a high expected concurrency for updates. It can be accessed inside and outside of asynchronous contexts.

Cache utilizes a lock-free concurrent hash table SegmentedHashMap from the moka-cht crate for the central key-value storage. Cache performs a best-effort bounding of the map using an entry replacement algorithm to determine which entries to evict when the capacity is exceeded.

To use this cache, enable a crate feature called “future”.

Examples

Cache entries are manually added using an insert method, and are stored in the cache until either evicted or manually invalidated:

• Inside an async context (async fn or async block), use insert or invalidate method for updating the cache and await them.
• Outside any async context, use blocking_insert or blocking_invalidate methods. They will block for a short time under heavy updates.

Here’s an example of reading and updating a cache by using multiple asynchronous tasks with Tokio runtime:

 // Cargo.toml
 //
 // [dependencies]
 // moka = { version = "0.6", features = ["future"] }
 // tokio = { version = "1", features = ["rt-multi-thread", "macros" ] }
 // futures = "0.3"

 use moka::future::Cache;

 #[tokio::main]
 async fn main() {
     const NUM_TASKS: usize = 16;
     const NUM_KEYS_PER_TASK: usize = 64;

     fn value(n: usize) -> String {
         format!("value {}", n)
     }

     // Create a cache that can store up to 10,000 entries.
     let cache = Cache::new(10_000);

     // Spawn async tasks and write to and read from the cache.
     let tasks: Vec<_> = (0..NUM_TASKS)
         .map(|i| {
             // To share the same cache across the async tasks, clone it.
             // This is a cheap operation.
             let my_cache = cache.clone();
             let start = i * NUM_KEYS_PER_TASK;
             let end = (i + 1) * NUM_KEYS_PER_TASK;

             tokio::spawn(async move {
                 // Insert 64 entries. (NUM_KEYS_PER_TASK = 64)
                 for key in start..end {
                     // insert() is an async method, so await it.
                     my_cache.insert(key, value(key)).await;
                     // get() returns Option<String>, a clone of the stored value.
                     assert_eq!(my_cache.get(&key), Some(value(key)));
                 }

                 // Invalidate every 4 element of the inserted entries.
                 for key in (start..end).step_by(4) {
                     // invalidate() is an async method, so await it.
                     my_cache.invalidate(&key).await;
                 }
             })
         })
         .collect();

     // Wait for all tasks to complete.
     futures_util::future::join_all(tasks).await;

     // Verify the result.
     for key in 0..(NUM_TASKS * NUM_KEYS_PER_TASK) {
         if key % 4 == 0 {
             assert_eq!(cache.get(&key), None);
         } else {
             assert_eq!(cache.get(&key), Some(value(key)));
         }
     }
 }

Thread Safety

All methods provided by the Cache are considered thread-safe, and can be safely accessed by multiple concurrent threads.

• Cache<K, V, S> requires trait bounds Send, Sync and 'static for K (key), V (value) and S (hasher state).
• Cache<K, V, S> will implement Send and Sync.

Sharing a cache across asynchronous tasks

To share a cache across async tasks (or OS threads), do one of the followings:

• Create a clone of the cache by calling its clone method and pass it to other task.
• Wrap the cache by a sync::OnceCell or sync::Lazy from once_cell create, and set it to a static variable.

Cloning is a cheap operation for Cache as it only creates thread-safe reference-counted pointers to the internal data structures.

Avoiding to clone the value at get

The return type of get method is Option<V> instead of Option<&V>. Every time get is called for an existing key, it creates a clone of the stored value V and returns it. This is because the Cache allows concurrent updates from threads so a value stored in the cache can be dropped or replaced at any time by any other thread. get cannot return a reference &V as it is impossible to guarantee the value outlives the reference.

If you want to store values that will be expensive to clone, wrap them by std::sync::Arc before storing in a cache. Arc is a thread-safe reference-counted pointer and its clone() method is cheap.

Expiration Policies

Cache supports the following expiration policies:

• Time to live: A cached entry will be expired after the specified duration past from insert.
• Time to idle: A cached entry will be expired after the specified duration past from get or insert.

See the CacheBuilder’s doc for how to configure a cache with them.

Hashing Algorithm

By default, Cache uses a hashing algorithm selected to provide resistance against HashDoS attacks. It will be the same one used by std::collections::HashMap, which is currently SipHash 1-3.

While SipHash’s performance is very competitive for medium sized keys, other hashing algorithms will outperform it for small keys such as integers as well as large keys such as long strings. However those algorithms will typically not protect against attacks such as HashDoS.

The hashing algorithm can be replaced on a per-Cache basis using the build_with_hasher method of the CacheBuilder. Many alternative algorithms are available on crates.io, such as the aHash crate.

Implementations

impl<K, V> Cache<K, V, RandomState> where
    K: Hash + Eq + Send + Sync + 'static,
    V: Clone + Send + Sync + 'static, 

pub fn new(max_capacity: usize) -> Self

Constructs a new Cache<K, V> that will store up to the max_capacity entries.

To adjust various configuration knobs such as initial_capacity or time_to_live, use the CacheBuilder.

impl<K, V, S> Cache<K, V, S> where
    K: Hash + Eq + Send + Sync + 'static,
    V: Clone + Send + Sync + 'static,
    S: BuildHasher + Clone + Send + Sync + 'static, 

pub fn get<Q>(&self, key: &Q) -> Option<V> where
    Arc<K>: Borrow<Q>,
    Q: Hash + Eq + ?Sized, 

Returns a clone of the value corresponding to the key.

If you want to store values that will be expensive to clone, wrap them by std::sync::Arc before storing in a cache. Arc is a thread-safe reference-counted pointer and its clone() method is cheap.

The key may be any borrowed form of the cache’s key type, but Hash and Eq on the borrowed form must match those for the key type.

pub async fn get_or_insert_with<F>(&self, key: K, init: F) -> V where
    F: Future<Output = V>, 

Ensures the value of the key exists by inserting the output of the init future if not exist, and returns a clone of the value.

This method prevents to resolve the init future multiple times on the same key even if the method is concurrently called by many async tasks; only one of the calls resolves its future, and other calls wait for that future to complete.

Example

// Cargo.toml
//
// [dependencies]
// moka = { version = "0.6", features = ["future"] }
// futures = "0.3"
// tokio = { version = "1", features = ["rt-multi-thread", "macros" ] }
use moka::future::Cache;
use std::sync::Arc;

#[tokio::main]
async fn main() {
    const TEN_MIB: usize = 10 * 1024 * 1024; // 10MiB
    let cache = Cache::new(100);

    // Spawn four async tasks.
    let tasks: Vec<_> = (0..4_u8)
        .map(|task_id| {
            let my_cache = cache.clone();
            tokio::spawn(async move {
                println!("Task {} started.", task_id);

                // Insert and get the value for key1. Although all four async tasks
                // will call `get_or_insert_with` at the same time, the `init` async
                // block must be resolved only once.
                let value = my_cache
                    .get_or_insert_with("key1", async move {
                        println!("Task {} inserting a value.", task_id);
                        Arc::new(vec![0u8; TEN_MIB])
                    })
                    .await;

                // Ensure the value exists now.
                assert_eq!(value.len(), TEN_MIB);
                assert!(my_cache.get(&"key1").is_some());

                println!("Task {} got the value. (len: {})", task_id, value.len());
            })
        })
        .collect();

    // Run all tasks concurrently and wait for them to complete.
    futures_util::future::join_all(tasks).await;
}

A Sample Result

• The init future (async black) was resolved exactly once by task 3.
• Other tasks were blocked until task 3 inserted the value.

Task 0 started.
Task 3 started.
Task 1 started.
Task 2 started.
Task 3 inserting a value.
Task 3 got the value. (len: 10485760)
Task 0 got the value. (len: 10485760)
Task 1 got the value. (len: 10485760)
Task 2 got the value. (len: 10485760)

Panics

This method panics when the init future has been panicked. When it happens, only the caller whose init future panicked will get the panic (e.g. only task 3 in the above sample). If there are other calls in progress (e.g. task 0, 1 and 2 above), this method will restart and resolve one of the remaining init futures.

pub async fn get_or_try_insert_with<F, E>(
    &self,
    key: K,
    init: F
) -> Result<V, Arc<E>> where
    F: Future<Output = Result<V, E>>,
    E: Send + Sync + 'static, 

Try to ensure the value of the key exists by inserting an Ok output of the init future if not exist, and returns a clone of the value or the Err produced by the future.

This method prevents to resolve the init future multiple times on the same key even if the method is concurrently called by many async tasks; only one of the calls resolves its future (as long as these futures return the same error type), and other calls wait for that future to complete.

Example

// Cargo.toml
//
// [dependencies]
// moka = { version = "0.6", features = ["future"] }
// futures = "0.3"
// reqwest = "0.11"
// tokio = { version = "1", features = ["rt-multi-thread", "macros" ] }
use moka::future::Cache;

// This async function tries to get HTML from the given URI.
async fn get_html(task_id: u8, uri: &str) -> Result<String, reqwest::Error> {
    println!("get_html() called by task {}.", task_id);
    Ok(reqwest::get(uri).await?.text().await?)
}

#[tokio::main]
async fn main() {
    let cache = Cache::new(100);

    // Spawn four async tasks.
    let tasks: Vec<_> = (0..4_u8)
        .map(|task_id| {
            let my_cache = cache.clone();
            tokio::spawn(async move {
                println!("Task {} started.", task_id);

                // Try to insert and get the value for key1. Although
                // all four async tasks will call `get_or_try_insert_with`
                // at the same time, get_html() must be called only once.
                let value = my_cache
                    .get_or_try_insert_with(
                        "key1",
                        get_html(task_id, "https://www.rust-lang.org"),
                    ).await;

                // Ensure the value exists now.
                assert!(value.is_ok());
                assert!(my_cache.get(&"key1").is_some());

                println!(
                    "Task {} got the value. (len: {})",
                    task_id,
                    value.unwrap().len()
                );
            })
        })
        .collect();

    // Run all tasks concurrently and wait for them to complete.
    futures_util::future::join_all(tasks).await;
}

A Sample Result

• get_html() was called exactly once by task 2.
• Other tasks were blocked until task 2 inserted the value.

Task 1 started.
Task 0 started.
Task 2 started.
Task 3 started.
get_html() called by task 2.
Task 2 got the value. (len: 19419)
Task 1 got the value. (len: 19419)
Task 0 got the value. (len: 19419)
Task 3 got the value. (len: 19419)

Panics

This method panics when the init future has been panicked. When it happens, only the caller whose init future panicked will get the panic (e.g. only task 2 in the above sample). If there are other calls in progress (e.g. task 0, 1 and 3 above), this method will restart and resolve one of the remaining init futures.

pub async fn insert(&self, key: K, value: V)

Inserts a key-value pair into the cache.

If the cache has this key present, the value is updated.

pub fn blocking_insert(&self, key: K, value: V)

Blocking insert to call outside of asynchronous contexts.

This method is intended for use cases where you are inserting from synchronous code.

pub async fn invalidate<Q>(&self, key: &Q) where
    Arc<K>: Borrow<Q>,
    Q: Hash + Eq + ?Sized, 

Discards any cached value for the key.

The key may be any borrowed form of the cache’s key type, but Hash and Eq on the borrowed form must match those for the key type.

pub fn blocking_invalidate<Q>(&self, key: &Q) where
    Arc<K>: Borrow<Q>,
    Q: Hash + Eq + ?Sized, 

Blocking invalidate to call outside of asynchronous contexts.

This method is intended for use cases where you are invalidating from synchronous code.

pub fn invalidate_all(&self)

Discards all cached values.

This method returns immediately and a background thread will evict all the cached values inserted before the time when this method was called. It is guaranteed that the get method must not return these invalidated values even if they have not been evicted.

Like the invalidate method, this method does not clear the historic popularity estimator of keys so that it retains the client activities of trying to retrieve an item.

pub fn invalidate_entries_if<F>(
    &self,
    predicate: F
) -> Result<PredicateId, PredicateError> where
    F: Fn(&K, &V) -> bool + Send + Sync + 'static, 

Discards cached values that satisfy a predicate.

invalidate_entries_if takes a closure that returns true or false. This method returns immediately and a background thread will apply the closure to each cached value inserted before the time when invalidate_entries_if was called. If the closure returns true on a value, that value will be evicted from the cache.

Also the get method will apply the closure to a value to determine if it should have been invalidated. Therefore, it is guaranteed that the get method must not return invalidated values.

Note that you must call CacheBuilder::support_invalidation_closures at the cache creation time as the cache needs to maintain additional internal data structures to support this method. Otherwise, calling this method will fail with a PredicateError::InvalidationClosuresDisabled.

Like the invalidate method, this method does not clear the historic popularity estimator of keys so that it retains the client activities of trying to retrieve an item.

pub fn max_capacity(&self) -> usize

Returns the max_capacity of this cache.

pub fn time_to_live(&self) -> Option<Duration>

Returns the time_to_live of this cache.

pub fn time_to_idle(&self) -> Option<Duration>

Returns the time_to_idle of this cache.

pub fn num_segments(&self) -> usize

Returns the number of internal segments of this cache.

Cache always returns 1.

Trait Implementations

impl<K: Clone, V: Clone, S: Clone> Clone for Cache<K, V, S>

fn clone(&self) -> Cache<K, V, S>

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<K, V, S> ConcurrentCacheExt<K, V> for Cache<K, V, S> where
    K: Hash + Eq + Send + Sync + 'static,
    V: Send + Sync + 'static,
    S: BuildHasher + Clone + Send + Sync + 'static, 

fn sync(&self)

Performs any pending maintenance operations needed by the cache.

impl<K, V, S> Send for Cache<K, V, S> where
    K: Send + Sync,
    V: Send + Sync,
    S: Send, 

impl<K, V, S> Sync for Cache<K, V, S> where
    K: Send + Sync,
    V: Send + Sync,
    S: Sync,