Cache refactor - add an in-memory cache, so redis is not needed

Cargo.toml

@@ -33,6 +33,7 @@ dhat = {version="0.3.2", optional = true}
 mimalloc = { version = "0.1.38", default-features = false }
 async-once-cell = {version="0.5.3"}
 actix-governor = {version="0.4.1"}
+mini-moka = "0.10"
 
 [dev-dependencies]
 rusty-hook = "^0.11.2"

src/bin/websurfx.rs

@@ -5,7 +5,9 @@
 
 use mimalloc::MiMalloc;
 use std::net::TcpListener;
-use websurfx::{config::parser::Config, run};
+use websurfx::{
+    cache::cacher::Cache, cache::redis_cacher::RedisCache, config::parser::Config, run,
+};
 
 /// A dhat heap memory profiler
 #[cfg(feature = "dhat-heap")]
@@ -30,6 +32,14 @@ async fn main() -> std::io::Result<()> {
 
     // Initialize the parsed config file.
     let config = Config::parse(false).unwrap();
+    let cache = match &config.redis_url {
+        Some(url) => Cache::new(
+            RedisCache::new(url, 5)
+                .await
+                .expect("Redis cache configured"),
+        ),
+        None => Cache::new_in_memory(),
+    };
 
     log::info!(
         "started server on port {} and IP {}",
@@ -44,5 +54,5 @@ async fn main() -> std::io::Result<()> {
 
     let listener = TcpListener::bind((config.binding_ip.clone(), config.port))?;
 
-    run(listener, config)?.await
+    run(listener, config, cache)?.await
 }

src/cache/cacher.rs

@@ -2,107 +2,53 @@
 //! from the upstream search engines in a json format.
 
 use error_stack::Report;
-use futures::future::try_join_all;
+use mini_moka::sync::Cache as MokaCache;
-use md5::compute;
+use std::time::Duration;
-use redis::{aio::ConnectionManager, AsyncCommands, Client, RedisError};
+use tokio::sync::Mutex;
 
-use super::error::PoolError;
+use super::{error::PoolError, redis_cacher::RedisCache};
 
-/// A named struct which stores the redis Connection url address to which the client will
+/// Different implementations for caching, currently it is possible to cache in-memory or in Redis.
-/// connect to.
 #[derive(Clone)]
-pub struct RedisCache {
+pub enum Cache {
-    /// It stores a pool of connections ready to be used.
+    /// Encapsulates the Redis based cache
-    connection_pool: Vec<ConnectionManager>,
+    Redis(RedisCache),
-    /// It stores the size of the connection pool (in other words the number of
+    /// Contains the in-memory cache.
-    /// connections that should be stored in the pool).
+    InMemory(MokaCache<String, String>),
-    pool_size: u8,
-    /// It stores the index of which connection is being used at the moment.
-    current_connection: u8,
 }
 
-impl RedisCache {
+impl Cache {
-    /// Constructs a new `SearchResult` with the given arguments needed for the struct.
+    /// Creates a new cache, which wraps the given RedisCache.
-    ///
+    pub fn new(redis_cache: RedisCache) -> Self {
-    /// # Arguments
+        Cache::Redis(redis_cache)
-    ///
-    /// * `redis_connection_url` - It takes the redis Connection url address.
-    /// * `pool_size` - It takes the size of the connection pool (in other words the number of
-    /// connections that should be stored in the pool).
-    pub async fn new(
-        redis_connection_url: &str,
-        pool_size: u8,
-    ) -> Result<Self, Box<dyn std::error::Error>> {
-        let client = Client::open(redis_connection_url)?;
-        let mut tasks: Vec<_> = Vec::new();
-
-        for _ in 0..pool_size {
-            tasks.push(client.get_tokio_connection_manager());
-        }
-
-        let redis_cache = RedisCache {
-            connection_pool: try_join_all(tasks).await?,
-            pool_size,
-            current_connection: Default::default(),
-        };
-        Ok(redis_cache)
     }
 
-    /// A helper function which computes the hash of the url and formats and returns it as string.
+    /// Creates an in-memory cache
-    ///
+    pub fn new_in_memory() -> Self {
-    /// # Arguments
+        let cache = MokaCache::builder()
-    ///
+            .max_capacity(1000)
-    /// * `url` - It takes an url as string.
+            .time_to_live(Duration::from_secs(60))
-    fn hash_url(&self, url: &str) -> String {
+            .build();
-        format!("{:?}", compute(url))
+        Cache::InMemory(cache)
     }
 
-    /// A function which fetches the cached json results as json string from the redis server.
+    /// A function which fetches the cached json results as json string.
     ///
     /// # Arguments
     ///
     /// * `url` - It takes an url as a string.
     pub async fn cached_json(&mut self, url: &str) -> Result<String, Report<PoolError>> {
-        self.current_connection = Default::default();
+        match self {
-        let hashed_url_string: &str = &self.hash_url(url);
+            Cache::Redis(redis_cache) => redis_cache.cached_json(url).await,
-
+            Cache::InMemory(in_memory) => match in_memory.get(&url.to_string()) {
-        let mut result: Result<String, RedisError> = self.connection_pool
+                Some(res) => Ok(res),
-            [self.current_connection as usize]
+                None => Err(Report::new(PoolError::MissingValue)),
-            .get(hashed_url_string)
+            },
-            .await;
-
-        // Code to check whether the current connection being used is dropped with connection error
-        // or not. if it drops with the connection error then the current connection is replaced
-        // with a new connection from the pool which is then used to run the redis command then
-        // that connection is also checked whether it is dropped or not if it is not then the
-        // result is passed as a `Result` or else the same process repeats again and if all of the
-        // connections in the pool result in connection drop error then a custom pool error is
-        // returned.
-        loop {
-            match result {
-                Err(error) => match error.is_connection_dropped() {
-                    true => {
-                        self.current_connection += 1;
-                        if self.current_connection == self.pool_size {
-                            return Err(Report::new(
-                                PoolError::PoolExhaustionWithConnectionDropError,
-                            ));
-                        }
-                        result = self.connection_pool[self.current_connection as usize]
-                            .get(hashed_url_string)
-                            .await;
-                        continue;
-                    }
-                    false => return Err(Report::new(PoolError::RedisError(error))),
-                },
-                Ok(res) => return Ok(res),
-            }
         }
     }
 
-    /// A function which caches the results by using the hashed `url` as the key and
+    /// A function which caches the results by using the `url` as the key and
-    /// `json results` as the value and stores it in redis server with ttl(time to live)
+    /// `json results` as the value and stores it in the cache
-    /// set to 60 seconds.
     ///
     /// # Arguments
     ///
@@ -110,43 +56,46 @@ impl RedisCache {
     /// * `url` - It takes the url as a String.
     pub async fn cache_results(
         &mut self,
-        json_results: &str,
+        json_results: String,
         url: &str,
     ) -> Result<(), Report<PoolError>> {
-        self.current_connection = Default::default();
+        match self {
-        let hashed_url_string: &str = &self.hash_url(url);
+            Cache::Redis(redis_cache) => redis_cache.cache_results(&json_results, url).await,
-
+            Cache::InMemory(cache) => {
-        let mut result: Result<(), RedisError> = self.connection_pool
+                cache.insert(url.to_string(), json_results);
-            [self.current_connection as usize]
+                Ok(())
-            .set_ex(hashed_url_string, json_results, 60)
-            .await;
-
-        // Code to check whether the current connection being used is dropped with connection error
-        // or not. if it drops with the connection error then the current connection is replaced
-        // with a new connection from the pool which is then used to run the redis command then
-        // that connection is also checked whether it is dropped or not if it is not then the
-        // result is passed as a `Result` or else the same process repeats again and if all of the
-        // connections in the pool result in connection drop error then a custom pool error is
-        // returned.
-        loop {
-            match result {
-                Err(error) => match error.is_connection_dropped() {
-                    true => {
-                        self.current_connection += 1;
-                        if self.current_connection == self.pool_size {
-                            return Err(Report::new(
-                                PoolError::PoolExhaustionWithConnectionDropError,
-                            ));
-                        }
-                        result = self.connection_pool[self.current_connection as usize]
-                            .set_ex(hashed_url_string, json_results, 60)
-                            .await;
-                        continue;
-                    }
-                    false => return Err(Report::new(PoolError::RedisError(error))),
-                },
-                Ok(_) => return Ok(()),
             }
         }
     }
 }
+
+/// A structure to efficiently share the cache between threads - as it is protected by a Mutex.
+pub struct SharedCache {
+    cache: Mutex<Cache>,
+}
+
+impl SharedCache {
+    /// Creates a new SharedCache from a Cache implementation
+    pub fn new(cache: Cache) -> Self {
+        Self {
+            cache: Mutex::new(cache),
+        }
+    }
+
+    /// A function which fetches the cached json results as json string.
+    pub async fn cached_json(&self, url: &str) -> Result<String, Report<PoolError>> {
+        let mut mut_cache = self.cache.lock().await;
+        mut_cache.cached_json(url).await
+    }
+
+    /// A function which caches the results by using the `url` as the key and
+    /// `json results` as the value and stores it in the cache
+    pub async fn cache_results(
+        &self,
+        json_results: String,
+        url: &str,
+    ) -> Result<(), Report<PoolError>> {
+        let mut mut_cache = self.cache.lock().await;
+        mut_cache.cache_results(json_results, url).await
+    }
+}

src/cache/error.rs

@@ -12,6 +12,7 @@ pub enum PoolError {
     /// This variant handles the errors which occurs when all the connections
     /// in the connection pool return a connection dropped redis error.
     PoolExhaustionWithConnectionDropError,
+    MissingValue,
 }
 
 impl fmt::Display for PoolError {
@@ -30,6 +31,9 @@ impl fmt::Display for PoolError {
                     "Error all connections from the pool dropped with connection error"
                 )
             }
+            PoolError::MissingValue => {
+                write!(f, "The value is missing from the cache")
+            }
         }
     }
 }

src/cache/mod.rs

@@ -3,3 +3,4 @@
 
 pub mod cacher;
 pub mod error;
+pub mod redis_cacher;

src/cache/redis_cacher.rs

@@ -0,0 +1,152 @@
+//! This module provides the functionality to cache the aggregated results fetched and aggregated
+//! from the upstream search engines in a json format.
+
+use error_stack::Report;
+use futures::future::try_join_all;
+use md5::compute;
+use redis::{aio::ConnectionManager, AsyncCommands, Client, RedisError};
+
+use super::error::PoolError;
+
+/// A named struct which stores the redis Connection url address to which the client will
+/// connect to.
+#[derive(Clone)]
+pub struct RedisCache {
+    /// It stores a pool of connections ready to be used.
+    connection_pool: Vec<ConnectionManager>,
+    /// It stores the size of the connection pool (in other words the number of
+    /// connections that should be stored in the pool).
+    pool_size: u8,
+    /// It stores the index of which connection is being used at the moment.
+    current_connection: u8,
+}
+
+impl RedisCache {
+    /// A function which fetches the cached json results as json string.
+    ///
+    /// # Arguments
+    ///
+    /// * `redis_connection_url` - It takes the redis Connection url address.
+    /// * `pool_size` - It takes the size of the connection pool (in other words the number of
+    /// connections that should be stored in the pool).
+    pub async fn new(
+        redis_connection_url: &str,
+        pool_size: u8,
+    ) -> Result<Self, Box<dyn std::error::Error>> {
+        let client = Client::open(redis_connection_url)?;
+        let mut tasks: Vec<_> = Vec::new();
+
+        for _ in 0..pool_size {
+            tasks.push(client.get_tokio_connection_manager());
+        }
+
+        let redis_cache = RedisCache {
+            connection_pool: try_join_all(tasks).await?,
+            pool_size,
+            current_connection: Default::default(),
+        };
+        Ok(redis_cache)
+    }
+
+    /// A helper function which computes the hash of the url and formats and returns it as string.
+    ///
+    /// # Arguments
+    ///
+    /// * `url` - It takes an url as string.
+    fn hash_url(&self, url: &str) -> String {
+        format!("{:?}", compute(url))
+    }
+
+    /// A function which fetches the cached json results as json string from the redis server.
+    ///
+    /// # Arguments
+    ///
+    /// * `url` - It takes an url as a string.
+    pub async fn cached_json(&mut self, url: &str) -> Result<String, Report<PoolError>> {
+        self.current_connection = Default::default();
+        let hashed_url_string: &str = &self.hash_url(url);
+
+        let mut result: Result<String, RedisError> = self.connection_pool
+            [self.current_connection as usize]
+            .get(hashed_url_string)
+            .await;
+
+        // Code to check whether the current connection being used is dropped with connection error
+        // or not. if it drops with the connection error then the current connection is replaced
+        // with a new connection from the pool which is then used to run the redis command then
+        // that connection is also checked whether it is dropped or not if it is not then the
+        // result is passed as a `Result` or else the same process repeats again and if all of the
+        // connections in the pool result in connection drop error then a custom pool error is
+        // returned.
+        loop {
+            match result {
+                Err(error) => match error.is_connection_dropped() {
+                    true => {
+                        self.current_connection += 1;
+                        if self.current_connection == self.pool_size {
+                            return Err(Report::new(
+                                PoolError::PoolExhaustionWithConnectionDropError,
+                            ));
+                        }
+                        result = self.connection_pool[self.current_connection as usize]
+                            .get(hashed_url_string)
+                            .await;
+                        continue;
+                    }
+                    false => return Err(Report::new(PoolError::RedisError(error))),
+                },
+                Ok(res) => return Ok(res),
+            }
+        }
+    }
+
+    /// A function which caches the results by using the hashed `url` as the key and
+    /// `json results` as the value and stores it in redis server with ttl(time to live)
+    /// set to 60 seconds.
+    ///
+    /// # Arguments
+    ///
+    /// * `json_results` - It takes the json results string as an argument.
+    /// * `url` - It takes the url as a String.
+    pub async fn cache_results(
+        &mut self,
+        json_results: &str,
+        url: &str,
+    ) -> Result<(), Report<PoolError>> {
+        self.current_connection = Default::default();
+        let hashed_url_string: &str = &self.hash_url(url);
+
+        let mut result: Result<(), RedisError> = self.connection_pool
+            [self.current_connection as usize]
+            .set_ex(hashed_url_string, json_results, 60)
+            .await;
+
+        // Code to check whether the current connection being used is dropped with connection error
+        // or not. if it drops with the connection error then the current connection is replaced
+        // with a new connection from the pool which is then used to run the redis command then
+        // that connection is also checked whether it is dropped or not if it is not then the
+        // result is passed as a `Result` or else the same process repeats again and if all of the
+        // connections in the pool result in connection drop error then a custom pool error is
+        // returned.
+        loop {
+            match result {
+                Err(error) => match error.is_connection_dropped() {
+                    true => {
+                        self.current_connection += 1;
+                        if self.current_connection == self.pool_size {
+                            return Err(Report::new(
+                                PoolError::PoolExhaustionWithConnectionDropError,
+                            ));
+                        }
+                        result = self.connection_pool[self.current_connection as usize]
+                            .set_ex(hashed_url_string, json_results, 60)
+                            .await;
+                        continue;
+                    }
+                    false => return Err(Report::new(PoolError::RedisError(error))),
+                },
+                Ok(_) => return Ok(()),
+            }
+        }
+    }
+}

src/config/parser.rs

@@ -19,7 +19,7 @@ pub struct Config {
     pub style: Style,
     /// It stores the redis connection url address on which the redis
     /// client should connect.
-    pub redis_url: String,
+    pub redis_url: Option<String>,
     /// It stores the option to whether enable or disable production use.
     pub aggregator: AggregatorConfig,
     /// It stores the option to whether enable or disable logs.
@@ -99,7 +99,7 @@ impl Config {
                 globals.get::<_, String>("theme")?,
                 globals.get::<_, String>("colorscheme")?,
             ),
-            redis_url: globals.get::<_, String>("redis_url")?,
+            redis_url: globals.get::<_, String>("redis_url").ok(),
             aggregator: AggregatorConfig {
                 random_delay: globals.get::<_, bool>("production_use")?,
             },

src/lib.rs

@@ -21,6 +21,7 @@ use actix_cors::Cors;
 use actix_files as fs;
 use actix_governor::{Governor, GovernorConfigBuilder};
 use actix_web::{dev::Server, http::header, middleware::Logger, web, App, HttpServer};
+use cache::cacher::{Cache, SharedCache};
 use config::parser::Config;
 use handlebars::Handlebars;
 use handler::paths::{file_path, FileType};
@@ -45,7 +46,7 @@ use handler::paths::{file_path, FileType};
 /// let listener = TcpListener::bind("127.0.0.1:8080").expect("Failed to bind address");
 /// let server = run(listener,config).expect("Failed to start server");
 /// ```
-pub fn run(listener: TcpListener, config: Config) -> std::io::Result<Server> {
+pub fn run(listener: TcpListener, config: Config, cache: Cache) -> std::io::Result<Server> {
     let mut handlebars: Handlebars<'_> = Handlebars::new();
 
     let public_folder_path: &str = file_path(FileType::Theme)?;
@@ -58,6 +59,8 @@ pub fn run(listener: TcpListener, config: Config) -> std::io::Result<Server> {
 
     let cloned_config_threads_opt: u8 = config.threads;
 
+    let cache = web::Data::new(SharedCache::new(cache));
+
     let server = HttpServer::new(move || {
         let cors: Cors = Cors::default()
             .allow_any_origin()
@@ -73,6 +76,7 @@ pub fn run(listener: TcpListener, config: Config) -> std::io::Result<Server> {
             .wrap(Logger::default()) // added logging middleware for logging.
             .app_data(handlebars_ref.clone())
             .app_data(web::Data::new(config.clone()))
+            .app_data(cache.clone())
             .wrap(cors)
             .wrap(Governor::new(
                 &GovernorConfigBuilder::default()

src/server/routes/search.rs

@@ -1,7 +1,7 @@
 //! This module handles the search route of the search engine website.
 
 use crate::{
-    cache::cacher::RedisCache,
+    cache::cacher::SharedCache,
     config::parser::Config,
     handler::paths::{file_path, FileType},
     models::{aggregation_models::SearchResults, engine_models::EngineHandler},
@@ -17,10 +17,6 @@ use std::{
 };
 use tokio::join;
 
-// ---- Constants ----
-/// Initialize redis cache connection once and store it on the heap.
-static REDIS_CACHE: async_once_cell::OnceCell<RedisCache> = async_once_cell::OnceCell::new();
-
 /// A named struct which deserializes all the user provided search parameters and stores them.
 #[derive(Deserialize)]
 pub struct SearchParams {
@@ -89,6 +85,7 @@ pub async fn search(
     hbs: web::Data<Handlebars<'_>>,
     req: HttpRequest,
     config: web::Data<Config>,
+    cache: web::Data<SharedCache>,
 ) -> Result<HttpResponse, Box<dyn std::error::Error>> {
     let params = web::Query::<SearchParams>::from_query(req.query_string())?;
     match &params.q {
@@ -125,6 +122,7 @@ pub async fn search(
                         safe_search
                     ),
                     &config,
+                    &cache,
                     query,
                     page - 1,
                     req.clone(),
@@ -136,6 +134,7 @@ pub async fn search(
                         config.binding_ip, config.port, query, page, safe_search
                     ),
                     &config,
+                    &cache,
                     query,
                     page,
                     req.clone(),
@@ -151,6 +150,7 @@ pub async fn search(
                         safe_search
                     ),
                     &config,
+                    &cache,
                     query,
                     page + 1,
                     req.clone(),
@@ -185,27 +185,19 @@ pub async fn search(
 async fn results(
     url: String,
     config: &Config,
+    cache: &web::Data<SharedCache>,
     query: &str,
     page: u32,
     req: HttpRequest,
     safe_search: u8,
 ) -> Result<SearchResults, Box<dyn std::error::Error>> {
-    // Initialize redis cache connection struct
-    let mut redis_cache: RedisCache = REDIS_CACHE
-        .get_or_init(async {
-            // Initialize redis cache connection pool only one and store it in the heap.
-            RedisCache::new(&config.redis_url, 5).await.unwrap()
-        })
-        .await
-        .clone();
     // fetch the cached results json.
-    let cached_results_json: Result<String, error_stack::Report<crate::cache::error::PoolError>> =
+    let cached_results_json = cache.cached_json(&url).await.ok();
-        redis_cache.clone().cached_json(&url).await;
     // check if fetched cache results was indeed fetched or it was an error and if so
     // handle the data accordingly.
     match cached_results_json {
-        Ok(results) => Ok(serde_json::from_str::<SearchResults>(&results)?),
+        Some(results) => Ok(serde_json::from_str::<SearchResults>(&results)?),
-        Err(_) => {
+        None => {
             if safe_search == 4 {
                 let mut results: SearchResults = SearchResults::default();
                 let mut _flag: bool =
@@ -216,8 +208,8 @@ async fn results(
                     results.set_disallowed();
                     results.add_style(&config.style);
                     results.set_page_query(query);
-                    redis_cache
+                    cache
-                        .cache_results(&serde_json::to_string(&results)?, &url)
+                        .cache_results(serde_json::to_string(&results)?, &url)
                         .await?;
                     return Ok(results);
                 }
@@ -266,9 +258,8 @@ async fn results(
                 results.set_filtered();
             }
             results.add_style(&config.style);
-            redis_cache
+            let json_results = serde_json::to_string(&results)?;
-                .cache_results(&serde_json::to_string(&results)?, &url)
+            cache.cache_results(json_results, &url).await?;
-                .await?;
             Ok(results)
         }
     }

tests/index.rs

@@ -9,7 +9,12 @@ fn spawn_app() -> String {
     let listener = TcpListener::bind("127.0.0.1:0").expect("Failed to bind random port");
     let port = listener.local_addr().unwrap().port();
     let config = Config::parse(false).unwrap();
-    let server = run(listener, config).expect("Failed to bind address");
+    let server = run(
+        listener,
+        config,
+        websurfx::cache::cacher::Cache::new_in_memory(),
+    )
+    .expect("Failed to bind address");
 
     tokio::spawn(server);
     format!("http://127.0.0.1:{}/", port)