Add a cache for state_getKeysPaged requests (#361)

* Add a cache for state_getKeysPaged requests

* PR link

light-base/src/json_rpc_service/background.rs

@@ -215,6 +215,13 @@ struct Cache {
         >,
         fnv::FnvBuildHasher,
     >,
+
+    /// When `state_getKeysPaged` is called and the response is truncated, the response is
+    /// inserted in this cache. The API user is likely to call `state_getKeysPaged` again with
+    /// the same parameters, in which case we hit the cache and avoid the networking requests.
+    /// The keys are `(block_hash, prefix)` and values are list of keys.
+    state_get_keys_paged:
+        lru::LruCache<([u8; 32], Option<methods::HexString>), Vec<Vec<u8>>, fnv::FnvBuildHasher>,
 }
 
 pub(super) fn start<TPlat: Platform>(
@@ -249,6 +256,10 @@ pub(super) fn start<TPlat: Platform>(
                 NonZeroUsize::new(32).unwrap(),
                 Default::default(),
             ),
+            state_get_keys_paged: lru::LruCache::with_hasher(
+                NonZeroUsize::new(2).unwrap(),
+                Default::default(),
+            ),
         }),
         genesis_block_hash: config.genesis_block_hash,
         printed_legacy_json_rpc_warning: atomic::AtomicBool::new(false),

light-base/src/json_rpc_service/background/state_chain.rs

@@ -1052,6 +1052,33 @@ impl<TPlat: Platform> Background<TPlat> {
             ),
         };
 
+        // Because the user is likely to call this function multiple times in a row with the exact
+        // same parameters, we store the untruncated responses in a cache. Check if we hit the
+        // cache.
+        if let Some(keys) = self
+            .cache
+            .lock()
+            .await
+            .state_get_keys_paged
+            .get(&(hash, prefix.clone()))
+        {
+            let out = keys
+                .iter()
+                .filter(|k| start_key.as_ref().map_or(true, |start| *k >= &start.0)) // TODO: not sure if start should be in the set or not?
+                .cloned()
+                .map(methods::HexString)
+                .take(usize::try_from(count).unwrap_or(usize::max_value()))
+                .collect::<Vec<_>>();
+
+            self.requests_subscriptions
+                .respond(
+                    request_id.1,
+                    methods::Response::state_getKeysPaged(out).to_json_response(request_id.0),
+                )
+                .await;
+            return;
+        }
+
         // Obtain the state trie root and height of the requested block.
         // This is necessary to perform network storage queries.
         let (state_root, block_number) = match self.state_trie_root_hash(&hash).await {
@@ -1080,7 +1107,7 @@ impl<TPlat: Platform> Background<TPlat> {
             .storage_prefix_keys_query(
                 block_number,
                 &hash,
-                &prefix.unwrap().0, // TODO: don't unwrap! what is this Option?
+                &prefix.as_ref().unwrap().0, // TODO: don't unwrap! what is this Option?
                 &state_root,
                 3,
                 Duration::from_secs(12),
@@ -1092,11 +1119,24 @@ impl<TPlat: Platform> Background<TPlat> {
             Ok(keys) => {
                 // TODO: instead of requesting all keys with that prefix from the network, pass `start_key` to the network service
                 let out = keys
-                    .into_iter()
-                    .filter(|k| start_key.as_ref().map_or(true, |start| k >= &start.0)) // TODO: not sure if start should be in the set or not?
+                    .iter()
+                    .filter(|k| start_key.as_ref().map_or(true, |start| *k >= &start.0)) // TODO: not sure if start should be in the set or not?
+                    .cloned() // TODO: instead of cloning, make `Response::state_getKeysPaged` accept references
                     .map(methods::HexString)
                     .take(usize::try_from(count).unwrap_or(usize::max_value()))
                     .collect::<Vec<_>>();
+
+                // If the returned response is somehow truncated, it is very likely that the
+                // JSON-RPC client will call the function again with the exact same parameters.
+                // Thus, store the results in a cache.
+                if out.len() != keys.len() {
+                    self.cache
+                        .lock()
+                        .await
+                        .state_get_keys_paged
+                        .push((hash, prefix), keys);
+                }
+
                 methods::Response::state_getKeysPaged(out).to_json_response(request_id.0)
             }
             Err(error) => json_rpc::parse::build_error_response(

wasm-node/CHANGELOG.md

@@ -9,6 +9,7 @@
 - Rename `/webrtc` to `/webrtc-direct` in multiaddresses, in accordance with the rest of the libp2p ecosystem. ([#326](https://github.com/smol-dot/smoldot/pull/326))
 - Improved the ganularity of the tasks that handle JSON-RPC requests and libp2p connections. Smoldot now yields more often to the browser, reducing the chances and the severity of freezes during the rendering of the web page. ([#349](https://github.com/smol-dot/smoldot/pull/349))
 - Smoldot is now compiled with the `bulk-memory-operations` and `sign-extensions-ops` WebAssembly features enabled. This is expected to considerably speed up its execution. The minimum version required to run smoldot is now Chrome 75, Firefox 79, NodeJS v12.5, and Deno v0.4. ([#356](https://github.com/smol-dot/smoldot/pull/356))
+- When the `state_getKeysPaged` JSON-RPC function is called, and the list of keys returned in the response is truncated (due to the `count` and `startKey` parameters), the rest of the keys are now put in a cache with the expectation that `state_getKeysPaged` is called again in order to obtain the rest of the keys. The `state_getKeysPaged` JSON-RPC function is unfortunately very often used by PolkadotJS despite being completely unsuitable for light clients. ([#361](https://github.com/smol-dot/smoldot/pull/361))
 
 ### Fixed