7.1.0 (#188)

* fix: don't panic when jitter is 0
* fix: support percent encoding in urls
* feat: add from tuple for RedisValue / MultipleKeys
* feat: make CLIENT ID checks optional

---------

Co-authored-by: 3Ti <[email protected]>
Co-authored-by: dupu <[email protected]>

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "fred"
-version = "7.0.0"
+version = "7.1.0"
 authors = ["Alec Embke <[email protected]>"]
 edition = "2021"
 description = "An async Redis client built on Tokio."
@@ -38,23 +38,24 @@ features = [
 rustdoc-args = ["--cfg", "docsrs"]
 
 [dependencies]
-arc-swap = "1.5"
-tokio = { version = "1.19.0", features = ["net", "sync", "rt", "rt-multi-thread", "macros"] }
-tokio-util = { version = "0.7.1", features = ["codec"] }
-bytes = "1.1"
+arc-swap = "1.6"
+tokio = { version = "1.34", features = ["net", "sync", "rt", "rt-multi-thread", "macros"] }
+tokio-util = { version = "0.7", features = ["codec"] }
+bytes = "1.5"
 bytes-utils = "0.1"
-futures = "0.3"
+futures = { version = "0.3", features = ["std"] }
 parking_lot = "0.12"
 lazy_static = "1.4"
 redis-protocol = { version = "4.1", features = ["decode-mut"] }
 log = "0.4"
 float-cmp = "0.9"
-url = "2.3"
-tokio-stream = "0.1.1"
+url = "2.4"
+tokio-stream = "0.1"
 sha-1 = { version = "0.10", optional = true }
 rand = "0.8"
 semver = "1.0"
 socket2 = "0.5"
+urlencoding = "2.1"
 async-trait = { version = "0.1" }
 rustls = { version = "0.21", optional = true }
 native-tls = { version = "0.2", optional = true }
@@ -72,8 +73,9 @@ trust-dns-resolver = { version = "0.23", optional = true }
 base64 = "0.21"
 subprocess = "0.2"
 pretty_env_logger = "0.5"
-bollard = "0.14"
+bollard = "0.15"
 serde = "1.0"
+tokio-stream = { version = "0.1", features = ["sync"] }
 
 [lib]
 doc = true
@@ -108,6 +110,10 @@ required-features = ["client-tracking"]
 name = "lua"
 required-features = ["sha-1"]
 
+[[example]]
+name = "events"
+required-features = ["tokio-stream/sync"]
+
 [features]
 default = ["ignore-auth-error", "pool-prefer-active"]
 serde-json = ["serde_json"]

examples/README.md

@@ -18,6 +18,8 @@ Examples
 * [Custom](./custom.rs) - Send custom commands or operate on RESP frames.
 * [DNS](./dns.rs) - Customize the DNS resolution logic. 
 * [Client Tracking](./client_tracking.rs) - Implement [client side caching](https://redis.io/docs/manual/client-side-caching/). 
+* [Events](./events.rs) - Respond to connection events with the `EventsInterface`.
+* [Keyspace Notifications](./keyspace.rs) - Use the [keyspace notifications](https://redis.io/docs/manual/keyspace-notifications/) interface.
 * [Misc](./misc.rs) - Miscellaneous or advanced features. 
 
 Or see the [tests](../tests/integration) for more examples.

examples/events.rs

@@ -0,0 +1,106 @@
+#![allow(clippy::disallowed_names)]
+#![allow(clippy::let_underscore_future)]
+
+use fred::prelude::*;
+use futures::StreamExt;
+
+// requires tokio_stream 0.1.3 or later
+use tokio_stream::wrappers::BroadcastStream;
+
+/// There are two interfaces for interacting with connection events on the `EventInterface`.
+///
+/// * The `on_*` functions are generally easier to use but require spawning a new tokio task. They also currently only
+///   support synchronous functions.
+/// * The `*_rx` functions are somewhat more complicated to use but allow the caller to control the underlying channel
+///   receiver directly. Additionally, these functions do not require spawning a new tokio task.
+///
+/// See the source for `on_any` for an example of how one might handle multiple receivers in one task.
+///
+/// The best approach depends on how many tasks the caller is willing to create. The `setup_pool` function shows
+/// how one might combine multiple receiver streams in a `RedisPool` to minimize the overhead of new tokio tasks for
+/// each underlying client.
+#[tokio::main]
+async fn main() -> Result<(), RedisError> {
+  let client = Builder::default_centralized().build()?;
+
+  // use the on_* functions
+  let reconnect_task = client.on_reconnect(|server| {
+    println!("Reconnected to {}", server);
+    Ok(())
+  });
+  let error_task = client.on_error(|error| {
+    println!("Connection error: {:?}", error);
+    Ok(())
+  });
+
+  // use the *_rx functions to do the same thing shown above. although not shown here, callers have more freedom to
+  // reduce the number of spawned tokio tasks with this interface.
+  let mut reconnect_rx = client.reconnect_rx();
+  let reconnect_task_2 = tokio::spawn(async move {
+    while let Ok(server) = reconnect_rx.recv().await {
+      println!("Reconnected to {}", server);
+    }
+  });
+
+  let mut error_rx = client.error_rx();
+  let error_task_2 = tokio::spawn(async move {
+    while let Ok(error) = error_rx.recv().await {
+      println!("Connection error: {:?}", error);
+    }
+  });
+
+  client.connect();
+  client.wait_for_connect().await?;
+
+  // ...
+
+  client.quit().await?;
+  reconnect_task.await??;
+  error_task.await??;
+  reconnect_task_2.await?;
+  error_task_2.await?;
+  Ok(())
+}
+
+/// Shows how to combine multiple event streams from multiple clients into one tokio task.
+#[allow(dead_code)]
+async fn setup_pool() -> Result<(), RedisError> {
+  let pool = Builder::default_centralized().build_pool(5)?;
+
+  // `select_all` does most of the work here but requires that the channel receivers implement `Stream`. unfortunately
+  // `tokio::sync::broadcast::Receiver` does not do this, so we use `tokio_stream::wrappers::BroadcastStream`.
+  let error_rxs: Vec<_> = pool
+    .clients()
+    .iter()
+    .map(|client| BroadcastStream::new(client.error_rx()))
+    .collect();
+  let reconnect_rxs: Vec<_> = pool
+    .clients()
+    .iter()
+    .map(|client| BroadcastStream::new(client.reconnect_rx()))
+    .collect();
+  let mut error_rx = futures::stream::select_all(error_rxs);
+  let mut reconnect_rx = futures::stream::select_all(reconnect_rxs);
+
+  let all_events_task = tokio::spawn(async move {
+    loop {
+      tokio::select! {
+        Some(Ok(error)) = error_rx.next() => {
+          println!("Error: {:?}", error);
+        }
+        Some(Ok(server)) = reconnect_rx.next() => {
+          println!("Reconnected to {}", server);
+        }
+      }
+    }
+  });
+
+  pool.connect();
+  pool.wait_for_connect().await?;
+
+  // ...
+
+  pool.quit().await?;
+  all_events_task.abort();
+  Ok(())
+}

examples/keyspace.rs

@@ -0,0 +1,136 @@
+#![allow(clippy::disallowed_names)]
+#![allow(clippy::let_underscore_future)]
+
+use fred::prelude::*;
+use std::time::Duration;
+use tokio::time::sleep;
+
+async fn fake_traffic(client: &RedisClient, amount: usize) -> Result<(), RedisError> {
+  // use a new client since the provided client is subscribed to keyspace events
+  let client = client.clone_new();
+  client.connect();
+  client.wait_for_connect().await?;
+
+  for idx in 0 .. amount {
+    let key: RedisKey = format!("foo-{}", idx).into();
+
+    client.set(&key, 1, None, None, false).await?;
+    client.incr(&key).await?;
+    client.del(&key).await?;
+  }
+
+  client.quit().await?;
+  Ok(())
+}
+
+/// Examples showing how to set up keyspace notifications with clustered or centralized/sentinel deployments.
+///
+/// The most complicated part of this process involves safely handling reconnections. Keyspace events rely on the
+/// pubsub interface, and clients are required to subscribe or resubscribe whenever a new connection is created. These
+/// examples show how to manually handle reconnections, but the caller can also use the `SubscriberClient` interface
+/// to remove some of the boilerplate.
+///
+/// If callers do not need the keyspace subscriptions to survive reconnects then the process is more
+/// straightforward.
+///
+/// Both examples assume that the server has been configured to emit keyspace events (via `notify-keyspace-events`).
+#[tokio::main]
+async fn main() -> Result<(), RedisError> {
+  clustered_keyspace_events().await?;
+  centralized_keyspace_events().await?;
+  Ok(())
+}
+
+async fn centralized_keyspace_events() -> Result<(), RedisError> {
+  let subscriber = Builder::default_centralized().build()?;
+
+  let reconnect_subscriber = subscriber.clone();
+  // resubscribe to the foo- prefix whenever we reconnect to a server
+  let reconnect_task = tokio::spawn(async move {
+    let mut reconnect_rx = reconnect_subscriber.reconnect_rx();
+
+    while let Ok(server) = reconnect_rx.recv().await {
+      println!("Reconnected to {}. Subscribing to keyspace events...", server);
+      reconnect_subscriber.psubscribe("__key__*:foo*").await?;
+    }
+
+    Ok::<_, RedisError>(())
+  });
+
+  // connect after setting up the reconnection logic
+  subscriber.connect();
+  subscriber.wait_for_connect().await?;
+
+  let mut keyspace_rx = subscriber.on_keyspace_event();
+  // set up a task that listens for keyspace events
+  let keyspace_task = tokio::spawn(async move {
+    while let Ok(event) = keyspace_rx.recv().await {
+      println!(
+        "Recv: {} on {} in db {}",
+        event.operation,
+        event.key.as_str_lossy(),
+        event.db
+      );
+    }
+
+    Ok::<_, RedisError>(())
+  });
+
+  // generate fake traffic and wait a second
+  fake_traffic(&subscriber, 1_000).await?;
+  sleep(Duration::from_secs(1)).await;
+  subscriber.quit().await?;
+  keyspace_task.await??;
+  reconnect_task.await??;
+
+  Ok(())
+}
+
+async fn clustered_keyspace_events() -> Result<(), RedisError> {
+  let subscriber = Builder::default_clustered().build()?;
+
+  let reconnect_subscriber = subscriber.clone();
+  // resubscribe to the foo- prefix whenever we reconnect to a server
+  let reconnect_task = tokio::spawn(async move {
+    let mut reconnect_rx = reconnect_subscriber.reconnect_rx();
+
+    // in 7.x the reconnection interface added a `Server` struct to reconnect events to make this easier.
+    while let Ok(server) = reconnect_rx.recv().await {
+      println!("Reconnected to {}. Subscribing to keyspace events...", server);
+      reconnect_subscriber
+        .with_cluster_node(server)
+        .psubscribe("__key__*:foo*")
+        .await?;
+    }
+
+    Ok::<_, RedisError>(())
+  });
+
+  // connect after setting up the reconnection logic
+  subscriber.connect();
+  subscriber.wait_for_connect().await?;
+
+  let mut keyspace_rx = subscriber.on_keyspace_event();
+  // set up a task that listens for keyspace events
+  let keyspace_task = tokio::spawn(async move {
+    while let Ok(event) = keyspace_rx.recv().await {
+      println!(
+        "Recv: {} on {} in db {}",
+        event.operation,
+        event.key.as_str_lossy(),
+        event.db
+      );
+    }
+
+    Ok::<_, RedisError>(())
+  });
+
+  // generate fake traffic and wait a second
+  fake_traffic(&subscriber, 1_000).await?;
+  sleep(Duration::from_secs(1)).await;
+  subscriber.quit().await?;
+  keyspace_task.await??;
+  reconnect_task.await??;
+
+  Ok(())
+}

examples/pool.rs

@@ -6,9 +6,14 @@ use fred::prelude::*;
 #[tokio::main]
 async fn main() -> Result<(), RedisError> {
   let pool = Builder::default_centralized().build_pool(5)?;
-  let _ = pool.connect();
+  pool.connect();
   pool.wait_for_connect().await?;
 
+  // use the pool like other clients
+  pool.get("foo").await?;
+  pool.set("foo", "bar", None, None, false).await?;
+  pool.get("foo").await?;
+
   // interact with specific clients via next(), last(), or clients()
   let pipeline = pool.next().pipeline();
   pipeline.incr("foo").await?;
@@ -17,19 +22,8 @@ async fn main() -> Result<(), RedisError> {
 
   for client in pool.clients() {
     println!("{} connected to {:?}", client.id(), client.active_connections().await?);
-
-    // set up event listeners on each client
-    client.on_error(|error| {
-      println!("Connection error: {:?}", error);
-      Ok(())
-    });
   }
 
-  // or use the pool like any other RedisClient
-  pool.get("foo").await?;
-  pool.set("foo", "bar", None, None, false).await?;
-  pool.get("foo").await?;
-
-  let _ = pool.quit().await;
+  pool.quit().await?;
   Ok(())
 }