introduce lightweight (5 mins) caching for crate metadata

Cargo.toml

@@ -10,6 +10,7 @@ futures = "0.1.18"
 hyper = "0.11.15"
 hyper-tls = "0.1.2"
 lazy_static = "1.0.0"
+lru-cache = "0.1.1"
 maud = "0.17.2"
 ordermap = { version = "0.4.0", features = ["serde-1"] }
 relative-path = { version = "0.3.7", features = ["serde"] }

src/engine/mod.rs

@@ -15,32 +15,40 @@ use tokio_service::Service;
 mod machines;
 mod futures;
 
-use ::utils::throttle::Throttle;
+use ::utils::cache::Cache;
 
 use ::models::repo::{Repository, RepoPath};
 use ::models::crates::{CrateName, CrateRelease, AnalyzedDependencies};
 
-use ::interactors::crates::query_crate;
+use ::interactors::crates::QueryCrate;
-use ::interactors::github::retrieve_file_at_path;
+use ::interactors::github::{GetPopularRepos, RetrieveFileAtPath};
-use ::interactors::github::GetPopularRepos;
 
 use self::futures::AnalyzeDependenciesFuture;
 use self::futures::CrawlManifestFuture;
 
+type HttpClient = Client<HttpsConnector<HttpConnector>>;
+
 #[derive(Clone, Debug)]
 pub struct Engine {
-    client: Client<HttpsConnector<HttpConnector>>,
+    client: HttpClient,
     logger: Logger,
 
-    get_popular_repos: Arc<Throttle<GetPopularRepos<Client<HttpsConnector<HttpConnector>>>>>
+    query_crate: Arc<Cache<QueryCrate<HttpClient>>>,
+    get_popular_repos: Arc<Cache<GetPopularRepos<HttpClient>>>,
+    retrieve_file_at_path: Arc<RetrieveFileAtPath<HttpClient>>
 }
 
 impl Engine {
     pub fn new(client: Client<HttpsConnector<HttpConnector>>, logger: Logger) -> Engine {
+        let query_crate = Cache::new(QueryCrate(client.clone()), Duration::from_secs(300), 500);
+        let get_popular_repos = Cache::new(GetPopularRepos(client.clone()), Duration::from_secs(10), 1);
+
         Engine {
             client: client.clone(), logger,
 
-            get_popular_repos: Arc::new(Throttle::new(GetPopularRepos(client), Duration::from_secs(10)))
+            query_crate: Arc::new(query_crate),
+            get_popular_repos: Arc::new(get_popular_repos),
+            retrieve_file_at_path: Arc::new(RetrieveFileAtPath(client))
         }
     }
 }
@@ -97,10 +105,11 @@ impl Engine {
     fn fetch_releases<I: IntoIterator<Item=CrateName>>(&self, names: I) ->
         impl Iterator<Item=impl Future<Item=Vec<CrateRelease>, Error=Error>>
     {
-        let client = self.client.clone();
+        let engine = self.clone();
         names.into_iter().map(move |name| {
-            query_crate(client.clone(), name)
+            engine.query_crate.call(name)
-                .map(|resp| resp.releases)
+                .from_err()
+                .map(|resp| resp.releases.clone())
         })
     }
 
@@ -108,7 +117,7 @@ impl Engine {
         impl Future<Item=String, Error=Error>
     {
         let manifest_path = path.join(RelativePath::new("Cargo.toml"));
-        retrieve_file_at_path(self.client.clone(), &repo_path, &manifest_path).from_err()
+        self.retrieve_file_at_path.call((repo_path.clone(), manifest_path))
     }
 }
 

src/interactors/crates.rs

@@ -36,10 +36,21 @@ pub struct QueryCrateResponse {
     pub releases: Vec<CrateRelease>
 }
 
-pub fn query_crate<S>(service: S, crate_name: CrateName) ->
+#[derive(Debug, Clone)]
-    impl Future<Item=QueryCrateResponse, Error=Error>
+pub struct QueryCrate<S>(pub S);
-    where S: Service<Request=Request, Response=Response, Error=HyperError>
+
+impl<S> Service for QueryCrate<S>
+    where S: Service<Request=Request, Response=Response, Error=HyperError> + Clone + 'static,
+          S::Future: 'static
 {
+    type Request = CrateName;
+    type Response = QueryCrateResponse;
+    type Error = Error;
+    type Future = Box<Future<Item=Self::Response, Error=Self::Error>>;
+
+    fn call(&self,  crate_name: CrateName) -> Self::Future {
+        let service = self.0.clone();
+
         let lower_name = crate_name.as_ref().to_lowercase();
 
         let path = match lower_name.len() {
@@ -52,7 +63,7 @@ pub fn query_crate<S>(service: S, crate_name: CrateName) ->
         let uri_future = format!("{}/master/{}", CRATES_INDEX_BASE_URI, path)
             .parse::<Uri>().into_future().from_err();
 
-    uri_future.and_then(move |uri| {
+        Box::new(uri_future.and_then(move |uri| {
             let request = Request::new(Method::Get, uri.clone());
 
             service.call(request).from_err().and_then(move |response| {
@@ -74,5 +85,6 @@ pub fn query_crate<S>(service: S, crate_name: CrateName) ->
                     future::Either::B(convert_future)
                 }
             })
-    })
+        }))
+    }
 }

src/interactors/github.rs

@@ -11,10 +11,22 @@ use ::models::repo::{Repository, RepoPath};
 const GITHUB_API_BASE_URI: &'static str = "https://api.github.com";
 const GITHUB_USER_CONTENT_BASE_URI: &'static str = "https://raw.githubusercontent.com";
 
-pub fn retrieve_file_at_path<S>(service: S, repo_path: &RepoPath, path: &RelativePathBuf) ->
+#[derive(Debug, Clone)]
-    impl Future<Item=String, Error=Error>
+pub struct RetrieveFileAtPath<S>(pub S);
-    where S: Service<Request=Request, Response=Response, Error=HyperError>
+
+impl<S> Service for RetrieveFileAtPath<S>
+    where S: Service<Request=Request, Response=Response, Error=HyperError> + Clone + 'static,
+          S::Future: 'static
 {
+    type Request = (RepoPath, RelativePathBuf);
+    type Response = String;
+    type Error = Error;
+    type Future = Box<Future<Item=Self::Response, Error=Self::Error>>;
+
+    fn call(&self, req: Self::Request) -> Self::Future {
+        let service = self.0.clone();
+
+        let (repo_path, path) = req;
         let path_str: &str = path.as_ref();
         let uri_future = format!("{}/{}/{}/HEAD/{}",
             GITHUB_USER_CONTENT_BASE_URI,
@@ -23,7 +35,7 @@ pub fn retrieve_file_at_path<S>(service: S, repo_path: &RepoPath, path: &Relativ
             path_str
         ).parse::<Uri>().into_future().from_err();
 
-    uri_future.and_then(move |uri| {
+        Box::new(uri_future.and_then(move |uri| {
             let request = Request::new(Method::Get, uri.clone());
 
             service.call(request).from_err().and_then(move |response| {
@@ -37,7 +49,8 @@ pub fn retrieve_file_at_path<S>(service: S, repo_path: &RepoPath, path: &Relativ
                     future::Either::B(decode_future)
                 }
             })
-    })
+        }))
+    }
 }
 
 #[derive(Deserialize)]

src/main.rs

@@ -9,6 +9,7 @@ extern crate base64;
 extern crate hyper;
 extern crate hyper_tls;
 #[macro_use] extern crate lazy_static;
+extern crate lru_cache;
 extern crate maud;
 extern crate ordermap;
 extern crate relative_path;

src/models/crates.rs

@@ -43,7 +43,7 @@ impl FromStr for CrateName {
     }
 }
 
-#[derive(Debug)]
+#[derive(Clone, Debug)]
 pub struct CrateRelease {
     pub name: CrateName,
     pub version: Version,

src/utils/cache.rs

@@ -0,0 +1,127 @@
+use std::fmt::{Debug, Display, Formatter, Result as FmtResult};
+use std::hash::Hash;
+use std::time::{Duration, Instant};
+use std::ops::Deref;
+use std::sync::Mutex;
+
+use failure::{Error, Fail};
+use futures::{Future, Poll};
+use futures::future::{Shared, SharedError, SharedItem};
+use lru_cache::LruCache;
+use tokio_service::Service;
+
+pub struct Cache<S>
+    where S: Service<Error=Error>,
+          S::Request: Hash + Eq
+{
+    inner: S,
+    duration: Duration,
+    cache: Mutex<LruCache<S::Request, (Instant, Shared<S::Future>)>>
+}
+
+impl<S> Debug for Cache<S>
+    where S: Service<Error=Error> + Debug,
+          S::Request: Hash + Eq
+{
+    fn fmt(&self, fmt: &mut Formatter) -> FmtResult {
+        fmt.debug_struct("Cache")
+            .field("inner", &self.inner)
+            .field("duration", &self.duration)
+            .finish()
+    }
+}
+
+impl<S> Cache<S> 
+    where S: Service<Error=Error>,
+          S::Request: Hash + Eq
+{
+    pub fn new(service: S, duration: Duration, capacity: usize) -> Cache<S> {
+        Cache {
+            inner: service,
+            duration: duration,
+            cache: Mutex::new(LruCache::new(capacity))
+        }
+    }
+}
+
+impl<S> Service for Cache<S>
+    where S: Service<Error=Error>,
+          S::Request: Clone + Hash + Eq
+{
+    type Request = S::Request;
+    type Response = CachedItem<S::Response>;
+    type Error = CachedError;
+    type Future = Cached<S::Future>;
+
+    fn call(&self, req: Self::Request) -> Self::Future {
+        let now = Instant::now();
+        let mut cache = self.cache.lock().expect("lock poisoned");
+        if let Some(&mut (valid_until, ref shared_future)) = cache.get_mut(&req) {
+            if valid_until > now {
+                if let Some(Ok(_)) = shared_future.peek() {
+                    return Cached(shared_future.clone());
+                }
+            }
+        }
+        let shared_future = self.inner.call(req.clone()).shared();
+        cache.insert(req, (now + self.duration, shared_future.clone()));
+        Cached(shared_future)
+    }
+}
+
+pub struct Cached<F: Future>(Shared<F>);
+
+impl<F> Debug for Cached<F>
+    where F: Future + Debug,
+          F::Item: Debug,
+          F::Error: Debug
+{
+    fn fmt(&self, fmt: &mut Formatter) -> FmtResult {
+        self.0.fmt(fmt)
+    }
+}
+
+impl<F: Future<Error=Error>> Future for Cached<F> {
+    type Item = CachedItem<F::Item>;
+    type Error = CachedError;
+
+    fn poll(&mut self) -> Poll<Self::Item, Self::Error> {
+        self.0.poll()
+            .map_err(CachedError)
+            .map(|async| async.map(CachedItem))
+    }
+}
+
+#[derive(Debug)]
+pub struct CachedItem<T>(SharedItem<T>);
+
+impl<T> Deref for CachedItem<T> {
+    type Target = T;
+
+    fn deref(&self) -> &T {
+        &self.0.deref()
+    }
+}
+
+#[derive(Debug)]
+pub struct CachedError(SharedError<Error>);
+
+impl Fail for CachedError {
+    fn cause(&self) -> Option<&Fail> {
+        Some(self.0.cause())
+    }
+
+    fn backtrace(&self) -> Option<&::failure::Backtrace> {
+        Some(self.0.backtrace())
+    }
+
+    fn causes(&self) -> ::failure::Causes {
+        self.0.causes()
+    }
+}
+
+impl Display for CachedError {
+    fn fmt(&self, f: &mut Formatter) -> FmtResult {
+        Display::fmt(&self.0, f)
+    }
+}

src/utils/mod.rs

@@ -1 +1 @@
-pub mod throttle;
+pub mod cache;

src/utils/throttle.rs

@@ -1,121 +0,0 @@
-use std::fmt::{Debug, Display, Formatter, Result as FmtResult};
-use std::time::{Duration, Instant};
-use std::ops::Deref;
-use std::sync::Mutex;
-
-use failure::{Error, Fail};
-use futures::{Future, Poll};
-use futures::future::{Shared, SharedError, SharedItem};
-use tokio_service::Service;
-
-pub struct Throttle<S>
-    where S: Service<Request=(), Error=Error>
-{
-    inner: S,
-    duration: Duration,
-    current: Mutex<Option<(Instant, Shared<S::Future>)>>
-}
-
-impl<S> Debug for Throttle<S>
-    where S: Service<Request=(), Error=Error> + Debug
-{
-    fn fmt(&self, fmt: &mut Formatter) -> FmtResult {
-        fmt.debug_struct("Throttle")
-            .field("inner", &self.inner)
-            .field("duration", &self.duration)
-            .finish()
-    }
-}
-
-impl<S> Throttle<S> 
-    where S: Service<Request=(), Error=Error>
-{
-    pub fn new(service: S, duration: Duration) -> Throttle<S> {
-        Throttle {
-            inner: service,
-            duration,
-            current: Mutex::new(None)
-        }
-    }
-}
-
-impl<S> Service for Throttle<S>
-    where S: Service<Request=(), Error=Error>
-{
-    type Request = ();
-    type Response = ThrottledItem<S::Response>;
-    type Error = ThrottledError;
-    type Future = Throttled<S::Future>;
-
-    fn call(&self, _: ()) -> Self::Future {
-        let now = Instant::now();
-        let mut current = self.current.lock().expect("lock poisoned");
-        if let Some((valid_until, ref shared_future)) = *current {
-            if valid_until > now {
-                if let Some(Ok(_)) = shared_future.peek() {
-                    return Throttled(shared_future.clone());
-                }
-            }
-        }
-        let shared_future = self.inner.call(()).shared();
-        *current = Some((now + self.duration, shared_future.clone()));
-        Throttled(shared_future)
-    }
-}
-
-pub struct Throttled<F: Future>(Shared<F>);
-
-impl<F> Debug for Throttled<F>
-    where F: Future + Debug,
-          F::Item: Debug,
-          F::Error: Debug
-{
-    fn fmt(&self, fmt: &mut Formatter) -> FmtResult {
-        self.0.fmt(fmt)
-    }
-}
-
-impl<F: Future<Error=Error>> Future for Throttled<F> {
-    type Item = ThrottledItem<F::Item>;
-    type Error = ThrottledError;
-
-    fn poll(&mut self) -> Poll<Self::Item, Self::Error> {
-        self.0.poll()
-            .map_err(ThrottledError)
-            .map(|async| async.map(ThrottledItem))
-    }
-}
-
-#[derive(Debug)]
-pub struct ThrottledItem<T>(SharedItem<T>);
-
-impl<T> Deref for ThrottledItem<T> {
-    type Target = T;
-
-    fn deref(&self) -> &T {
-        &self.0.deref()
-    }
-}
-
-#[derive(Debug)]
-pub struct ThrottledError(SharedError<Error>);
-
-impl Fail for ThrottledError {
-    fn cause(&self) -> Option<&Fail> {
-        Some(self.0.cause())
-    }
-
-    fn backtrace(&self) -> Option<&::failure::Backtrace> {
-        Some(self.0.backtrace())
-    }
-
-    fn causes(&self) -> ::failure::Causes {
-        self.0.causes()
-    }
-}
-
-impl Display for ThrottledError {
-    fn fmt(&self, f: &mut Formatter) -> FmtResult {
-        Display::fmt(&self.0, f)
-    }
-}