Add benchmarks

htsget-http-actix/Cargo.toml

@@ -8,9 +8,18 @@ edition = "2018"
 [dependencies]
 
 actix-web = "3"
+actix-files = "0.5"
 envy = "0.4"
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
 futures-util = { version = "0.3.5", default-features = false }
 htsget-http-core = { path = "../htsget-http-core" }
-htsget-search = { path = "../htsget-search" }
+htsget-search = { path = "../htsget-search" }
+
+[dev-dependencies]
+criterion = "0.3"
+reqwest = { version = "0.11", features = ["blocking", "json"] }
+
+[[bench]]
+name = "request-benchmark"
+harness = false

htsget-http-actix/benches/request-benchmark.rs

@@ -0,0 +1,138 @@
+use criterion::measurement::WallTime;
+use criterion::{criterion_group, criterion_main, BenchmarkGroup, Criterion};
+use htsget_http_core::{JsonResponse, PostRequest, Region};
+use reqwest::{blocking::Client, Error as ActixError};
+use serde::Serialize;
+use std::collections::HashMap;
+use std::{convert::TryInto, time::Duration};
+#[derive(Serialize)]
+struct Empty {}
+
+const HTSGET_RS_URL: &str = "http://localhost:8080/reads/data/bam/htsnexus_test_NA12878";
+const HTSGET_REFSERVER_URL: &str = "http://localhost:8081/reads/htsnexus_test_NA12878";
+
+fn request(url: &str, json_content: &impl Serialize) -> Result<usize, ActixError> {
+  let client = Client::new();
+  let response: JsonResponse = client.get(url).json(json_content).send()?.json()?;
+  Ok(
+    response
+      .htsget
+      .urls
+      .iter()
+      .map(|json_url| {
+        Ok(
+          client
+            .get(&json_url.url)
+            .headers(
+              json_url
+                .headers
+                .as_ref()
+                .unwrap_or(&HashMap::new())
+                .try_into()
+                .unwrap(),
+            )
+            .send()?
+            .text()?
+            .len(),
+        )
+      })
+      .collect::<Result<Vec<_>, ActixError>>()?
+      .into_iter()
+      .sum(),
+  )
+}
+
+fn bench_request(
+  group: &mut BenchmarkGroup<WallTime>,
+  name: &str,
+  url: &str,
+  json_content: &impl Serialize,
+) {
+  println!(
+    "\n\nDownload size: {} bytes",
+    request(url, json_content).expect("Error during the request")
+  );
+  group.bench_function(name, |b| b.iter(|| request(url, json_content)));
+}
+
+fn criterion_benchmark(c: &mut Criterion) {
+  let mut group = c.benchmark_group("Requests");
+  group
+    .sample_size(150)
+    .measurement_time(Duration::from_secs(15));
+
+  bench_request(
+    &mut group,
+    "htsget-rs simple request",
+    HTSGET_RS_URL,
+    &Empty {},
+  );
+  bench_request(
+    &mut group,
+    "htsget-refserver simple request",
+    HTSGET_REFSERVER_URL,
+    &Empty {},
+  );
+
+  let json_content = PostRequest {
+    format: None,
+    class: None,
+    fields: None,
+    tags: None,
+    notags: None,
+    regions: Some(vec![Region {
+      reference_name: "20".to_string(),
+      start: None,
+      end: None,
+    }]),
+  };
+  bench_request(
+    &mut group,
+    "htsget-rs with region",
+    HTSGET_RS_URL,
+    &json_content,
+  );
+  bench_request(
+    &mut group,
+    "htsget-refserver with region",
+    HTSGET_REFSERVER_URL,
+    &json_content,
+  );
+
+  let json_content = PostRequest {
+    format: None,
+    class: None,
+    fields: None,
+    tags: None,
+    notags: None,
+    regions: Some(vec![
+      Region {
+        reference_name: "20".to_string(),
+        start: None,
+        end: None,
+      },
+      Region {
+        reference_name: "11".to_string(),
+        start: Some(4999977),
+        end: Some(5008321),
+      },
+    ]),
+  };
+  bench_request(
+    &mut group,
+    "htsget-rs with two regions",
+    HTSGET_RS_URL,
+    &json_content,
+  );
+  bench_request(
+    &mut group,
+    "htsget-refserver with two regions",
+    HTSGET_REFSERVER_URL,
+    &json_content,
+  );
+
+  group.finish();
+}
+
+criterion_group!(benches, criterion_benchmark);
+criterion_main!(benches);

htsget-http-actix/docker-htsget-refserver.sh

@@ -0,0 +1,2 @@
+docker image pull ga4gh/htsget-refserver:1.4.0
+docker container run -d -p 8081:3000 -v $(pwd)/../data:/data -v $(pwd):/config ga4gh/htsget-refserver:1.4.0 ./htsget-refserver -config /config/htsget-refserver-config.json

htsget-http-actix/htsget-refserver-config.json

@@ -0,0 +1,18 @@
+{
+    "htsgetConfig": {
+        "props": {
+            "port": 8081,
+            "host": "http://localhost:8081/"
+        },
+        "reads": {
+            "dataSourceRegistry": {
+                "sources": [
+                    {
+                        "pattern": "^(?P<id>.*)$",
+                        "path": "/data/bam/{id}.bam"
+                    }
+                ]
+            }
+        }
+    }
+}

htsget-http-actix/src/main.rs

@@ -1,3 +1,4 @@
+use actix_files::Files;
 use actix_web::{web, App, HttpServer};
 use htsget_search::{
   htsget::{from_storage::HtsGetFromStorage, HtsGet},
@@ -43,12 +44,13 @@ async fn main() -> std::io::Result<()> {
   }
   let config = envy::from_env::<Config>().expect("The environment variables weren't properly set!");
   let address = format!("{}:{}", config.htsget_ip, config.htsget_port);
+  let storage_base_address = format!("{}/data", address);
   let htsget_path = config.htsget_path.clone();
   HttpServer::new(move || {
     App::new()
       .data(AppState {
         htsget: HtsGetFromStorage::new(
-          LocalStorage::new(htsget_path.clone())
+          LocalStorage::new(&htsget_path, &storage_base_address)
             .expect("Couldn't create a Storage with the provided path"),
         ),
         config: config.clone(),
@@ -79,6 +81,7 @@ async fn main() -> std::io::Result<()> {
           .route("/{id:.+}", web::get().to(get::variants::<HtsGetStorage>))
           .route("/{id:.+}", web::post().to(post::variants::<HtsGetStorage>)),
       )
+      .service(Files::new("/data", htsget_path.clone()))
   })
   .bind(address)?
   .run()

htsget-http-core/src/json_response.rs

@@ -1,13 +1,13 @@
 use std::collections::HashMap;
 
 use htsget_search::htsget::{Class, Format, Response, Url};
-use serde::Serialize;
+use serde::{Deserialize, Serialize};
 
 /// A helper struct to convert [Responses](Response) to JSON. It implements [serde's Serialize trait](Serialize),
 /// so it's trivial to convert to JSON.
-#[derive(Debug, PartialEq, Serialize)]
+#[derive(Debug, PartialEq, Serialize, Deserialize)]
 pub struct JsonResponse {
-  htsget: HtsGetResponse,
+  pub htsget: HtsGetResponse,
 }
 
 impl JsonResponse {
@@ -20,10 +20,10 @@ impl JsonResponse {
 
 /// A helper struct to represent a JSON response. It shouldn't be used
 /// on its own, but with [JsonResponse]
-#[derive(Debug, PartialEq, Serialize)]
+#[derive(Debug, PartialEq, Serialize, Deserialize)]
 pub struct HtsGetResponse {
-  format: String,
-  urls: Vec<JsonUrl>,
+  pub format: String,
+  pub urls: Vec<JsonUrl>,
 }
 
 impl HtsGetResponse {
@@ -39,26 +39,25 @@ impl HtsGetResponse {
 
 /// A helper struct to convert [Urls](Url) to JSON. It shouldn't be used
 /// on its own, but with [JsonResponse]
-#[derive(Debug, PartialEq, Serialize)]
+#[derive(Debug, PartialEq, Serialize, Deserialize)]
 pub struct JsonUrl {
-  url: String,
-  headers: HashMap<String, String>,
-  class: String,
+  pub url: String,
+  pub headers: Option<HashMap<String, String>>,
+  pub class: Option<String>,
 }
 
 impl JsonUrl {
   fn new(url: Url) -> Self {
     JsonUrl {
       url: url.url,
-      headers: match url.headers {
-        Some(headers) => headers.get_inner(),
-        None => HashMap::new(),
-      },
-      class: match url.class {
-        Class::Body => "body",
-        Class::Header => "header",
-      }
-      .to_string(),
+      headers: url.headers.map(|headers| headers.get_inner()),
+      class: Some(
+        match url.class {
+          Class::Body => "body",
+          Class::Header => "header",
+        }
+        .to_string(),
+      ),
     }
   }
 }

htsget-http-core/src/lib.rs

@@ -118,7 +118,6 @@ mod tests {
     htsget::{from_storage::HtsGetFromStorage, Format, Headers, Url},
     storage::local::LocalStorage,
   };
-  use std::path::PathBuf;
   #[test]
   fn get_request() {
     let mut request = HashMap::new();
@@ -129,14 +128,10 @@ mod tests {
       get_response_for_get_request(&get_searcher(), request, Endpoint::Reads),
       Ok(JsonResponse::from_response(Response::new(
         Format::Bam,
-        vec![Url::new(format!(
-          "file://{}",
-          get_base_path()
-            .join("bam")
-            .join("htsnexus_test_NA12878.bam")
-            .to_string_lossy()
-        ))
-        .with_headers(Headers::new(headers))]
+        vec![
+          Url::new("http://localhost/data/bam/htsnexus_test_NA12878.bam")
+            .with_headers(Headers::new(headers))
+        ]
       )))
     )
   }
@@ -167,14 +162,10 @@ mod tests {
       get_response_for_get_request(&get_searcher(), request, Endpoint::Variants),
       Ok(JsonResponse::from_response(Response::new(
         Format::Vcf,
-        vec![Url::new(format!(
-          "file://{}",
-          get_base_path()
-            .join("vcf")
-            .join("sample1-bcbio-cancer.vcf.gz")
-            .to_string_lossy()
-        ))
-        .with_headers(Headers::new(headers))]
+        vec![
+          Url::new("http://localhost/data/vcf/sample1-bcbio-cancer.vcf.gz")
+            .with_headers(Headers::new(headers))
+        ]
       )))
     )
   }
@@ -200,14 +191,10 @@ mod tests {
       ),
       Ok(JsonResponse::from_response(Response::new(
         Format::Bam,
-        vec![Url::new(format!(
-          "file://{}",
-          get_base_path()
-            .join("bam")
-            .join("htsnexus_test_NA12878.bam")
-            .to_string_lossy()
-        ))
-        .with_headers(Headers::new(headers))]
+        vec![
+          Url::new("http://localhost/data/bam/htsnexus_test_NA12878.bam")
+            .with_headers(Headers::new(headers))
+        ]
       )))
     )
   }
@@ -260,27 +247,15 @@ mod tests {
       ),
       Ok(JsonResponse::from_response(Response::new(
         Format::Vcf,
-        vec![Url::new(format!(
-          "file://{}",
-          get_base_path()
-            .join("vcf")
-            .join("sample1-bcbio-cancer.vcf.gz")
-            .to_string_lossy()
-        ))
-        .with_headers(Headers::new(headers))]
+        vec![
+          Url::new("http://localhost/data/vcf/sample1-bcbio-cancer.vcf.gz")
+            .with_headers(Headers::new(headers))
+        ]
       )))
     )
   }
 
-  fn get_base_path() -> PathBuf {
-    std::env::current_dir()
-      .unwrap()
-      .parent()
-      .unwrap()
-      .join("data")
-  }
-
   fn get_searcher() -> impl HtsGet {
-    HtsGetFromStorage::new(LocalStorage::new("../data").unwrap())
+    HtsGetFromStorage::new(LocalStorage::new("../data", "localhost/data").unwrap())
   }
 }

htsget-search/Cargo.toml

@@ -11,3 +11,8 @@ noodles = { version = "0.5.0", features = ["bam", "bcf", "bgzf", "cram", "csi",
 
 [dev-dependencies]
 tempfile = "3.2.0"
+criterion = "0.3"
+
+[[bench]]
+name = "benchmark"
+harness = false