feat(registry): add individual job endpoints for package data

Cargo.lock

@@ -3730,6 +3730,7 @@ dependencies = [
  "thiserror 2.0.7",
  "tokio",
  "tokio-tar",
+ "tokio-util",
  "toml",
  "tracing",
  "tracing-actix-web",

registry/CHANGELOG.md

@@ -10,6 +10,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Support deprecating and yanking packages by @daimond113
 - Add yanking & deprecating to registry by @daimond113
 - Log more information about configured auth & storage by @daimond113
+- Add individual endpoints for package data over using `Accept` header conditional returns by @daimond113
 
 ### Performance
 - Switch to using a `RwLock` over a `Mutex` to store repository data by @daimond113

registry/Cargo.toml

@@ -16,6 +16,7 @@ semver = "1.0.24"
 chrono = { version = "0.4.39", features = ["serde"] }
 futures = "0.3.31"
 tokio = "1.42.0"
+tokio-util = "0.7.13"
 tempfile = "3.14.0"
 fs-err = { version = "3.0.0", features = ["tokio"] }
 async-stream = "0.3.6"

registry/src/endpoints/mod.rs

@@ -1,4 +1,7 @@
 pub mod deprecate_version;
+pub mod package_archive;
+pub mod package_doc;
+pub mod package_readme;
 pub mod package_version;
 pub mod package_versions;
 pub mod publish_version;

registry/src/endpoints/package_archive.rs

@@ -0,0 +1,27 @@
+use actix_web::{web, HttpResponse};
+
+use crate::{
+	error::RegistryError,
+	package::read_package,
+	request_path::{resolve_version_and_target, AnyOrSpecificTarget, LatestOrSpecificVersion},
+	storage::StorageImpl,
+	AppState,
+};
+use pesde::names::PackageName;
+
+pub async fn get_package_archive(
+	app_state: web::Data<AppState>,
+	path: web::Path<(PackageName, LatestOrSpecificVersion, AnyOrSpecificTarget)>,
+) -> Result<HttpResponse, RegistryError> {
+	let (name, version, target) = path.into_inner();
+
+	let Some(file) = read_package(&app_state, &name, &*app_state.source.read().await).await? else {
+		return Ok(HttpResponse::NotFound().finish());
+	};
+
+	let Some(v_id) = resolve_version_and_target(&file, version, target) else {
+		return Ok(HttpResponse::NotFound().finish());
+	};
+
+	app_state.storage.get_package(&name, v_id).await
+}

registry/src/endpoints/package_doc.rs

@@ -0,0 +1,66 @@
+use crate::{
+	error::RegistryError,
+	package::read_package,
+	request_path::{resolve_version_and_target, AnyOrSpecificTarget, LatestOrSpecificVersion},
+	storage::StorageImpl,
+	AppState,
+};
+use actix_web::{web, HttpResponse};
+use pesde::{
+	names::PackageName,
+	source::{
+		ids::VersionId,
+		pesde::{DocEntryKind, IndexFile},
+	},
+};
+use serde::Deserialize;
+
+pub fn find_package_doc<'a>(
+	file: &'a IndexFile,
+	v_id: &VersionId,
+	doc_name: &str,
+) -> Option<&'a str> {
+	let mut queue = file.entries[v_id]
+		.docs
+		.iter()
+		.map(|doc| &doc.kind)
+		.collect::<Vec<_>>();
+	while let Some(doc) = queue.pop() {
+		match doc {
+			DocEntryKind::Page { name, hash } if name == doc_name => return Some(hash.as_str()),
+			DocEntryKind::Category { items, .. } => {
+				queue.extend(items.iter().map(|item| &item.kind))
+			}
+			_ => continue,
+		};
+	}
+
+	None
+}
+
+#[derive(Debug, Deserialize)]
+pub struct Query {
+	doc: String,
+}
+
+pub async fn get_package_doc(
+	app_state: web::Data<AppState>,
+	path: web::Path<(PackageName, LatestOrSpecificVersion, AnyOrSpecificTarget)>,
+	request_query: web::Query<Query>,
+) -> Result<HttpResponse, RegistryError> {
+	let (name, version, target) = path.into_inner();
+
+	let Some(file) = read_package(&app_state, &name, &*app_state.source.read().await).await? else {
+		return Ok(HttpResponse::NotFound().finish());
+	};
+
+	let Some(v_id) = resolve_version_and_target(&file, version, target) else {
+		return Ok(HttpResponse::NotFound().finish());
+	};
+
+	let Some(hash) = find_package_doc(&file, v_id, &request_query.doc) else {
+		return Ok(HttpResponse::NotFound().finish());
+	};
+
+	app_state.storage.get_doc(hash).await
+}

registry/src/endpoints/package_readme.rs

@@ -0,0 +1,27 @@
+use actix_web::{web, HttpResponse};
+
+use crate::{
+	error::RegistryError,
+	package::read_package,
+	request_path::{resolve_version_and_target, AnyOrSpecificTarget, LatestOrSpecificVersion},
+	storage::StorageImpl,
+	AppState,
+};
+use pesde::names::PackageName;
+
+pub async fn get_package_readme(
+	app_state: web::Data<AppState>,
+	path: web::Path<(PackageName, LatestOrSpecificVersion, AnyOrSpecificTarget)>,
+) -> Result<HttpResponse, RegistryError> {
+	let (name, version, target) = path.into_inner();
+
+	let Some(file) = read_package(&app_state, &name, &*app_state.source.read().await).await? else {
+		return Ok(HttpResponse::NotFound().finish());
+	};
+
+	let Some(v_id) = resolve_version_and_target(&file, version, target) else {
+		return Ok(HttpResponse::NotFound().finish());
+	};
+
+	app_state.storage.get_readme(&name, v_id).await
+}

registry/src/endpoints/package_version.rs

@@ -2,13 +2,14 @@ use actix_web::{http::header::ACCEPT, web, HttpRequest, HttpResponse};
 use serde::Deserialize;
 
 use crate::{
+	endpoints::package_doc::find_package_doc,
 	error::RegistryError,
 	package::{read_package, PackageResponse},
-	request_path::{AnyOrSpecificTarget, LatestOrSpecificVersion},
+	request_path::{resolve_version_and_target, AnyOrSpecificTarget, LatestOrSpecificVersion},
 	storage::StorageImpl,
 	AppState,
 };
-use pesde::{names::PackageName, source::pesde::DocEntryKind};
+use pesde::names::PackageName;
 
 #[derive(Debug, Deserialize)]
 pub struct Query {
@@ -27,50 +28,19 @@ pub async fn get_package_version(
 		return Ok(HttpResponse::NotFound().finish());
 	};
 
-	let Some((v_id, entry)) = ({
-		let version = match version {
-			LatestOrSpecificVersion::Latest => match file.entries.keys().map(|k| k.version()).max()
+	let Some(v_id) = resolve_version_and_target(&file, version, target) else {
+		return Ok(HttpResponse::NotFound().finish());
+	};
+
+	// TODO: this is deprecated, since the introduction of the specific endpoints for readme, doc and archive.
+	//	     remove this when we drop 0.5 support.
 	{
-				Some(latest) => latest.clone(),
-				None => return Ok(HttpResponse::NotFound().finish()),
-			},
-			LatestOrSpecificVersion::Specific(version) => version,
-		};
-
-		let mut versions = file
-			.entries
-			.iter()
-			.filter(|(v_id, _)| *v_id.version() == version);
-
-		match target {
-			AnyOrSpecificTarget::Any => versions.min_by_key(|(v_id, _)| *v_id.target()),
-			AnyOrSpecificTarget::Specific(kind) => {
-				versions.find(|(_, entry)| entry.target.kind() == kind)
-			}
-		}
-	}) else {
-		return Ok(HttpResponse::NotFound().finish());
-	};
-
 		if let Some(doc_name) = request_query.doc.as_deref() {
-		let hash = 'finder: {
-			let mut queue = entry.docs.iter().map(|doc| &doc.kind).collect::<Vec<_>>();
-			while let Some(doc) = queue.pop() {
-				match doc {
-					DocEntryKind::Page { name, hash } if name == doc_name => {
-						break 'finder hash.clone()
-					}
-					DocEntryKind::Category { items, .. } => {
-						queue.extend(items.iter().map(|item| &item.kind))
-					}
-					_ => continue,
-				};
-			}
-
+			let Some(hash) = find_package_doc(&file, v_id, doc_name) else {
 				return Ok(HttpResponse::NotFound().finish());
 			};
 
-		return app_state.storage.get_doc(&hash).await;
+			return app_state.storage.get_doc(hash).await;
 		}
 
 		let accept = request
@@ -90,6 +60,7 @@ pub async fn get_package_version(
 				app_state.storage.get_package(&name, v_id).await
 			};
 		}
+	}
 
 	Ok(HttpResponse::Ok().json(PackageResponse::new(&name, v_id, &file)))
 }

registry/src/main.rs

@@ -190,6 +190,24 @@ async fn run() -> std::io::Result<()> {
 							.to(endpoints::package_version::get_package_version)
 							.wrap(from_fn(auth::read_mw)),
 					)
+					.route(
+						"/packages/{name}/{version}/{target}/archive",
+						web::get()
+							.to(endpoints::package_archive::get_package_archive)
+							.wrap(from_fn(auth::read_mw)),
+					)
+					.route(
+						"/packages/{name}/{version}/{target}/doc",
+						web::get()
+							.to(endpoints::package_doc::get_package_doc)
+							.wrap(from_fn(auth::read_mw)),
+					)
+					.route(
+						"/packages/{name}/{version}/{target}/readme",
+						web::get()
+							.to(endpoints::package_readme::get_package_readme)
+							.wrap(from_fn(auth::read_mw)),
+					)
 					.service(
 						web::resource("/packages/{name}/{version}/{target}/yank")
 							.put(endpoints::yank_version::yank_package_version)

registry/src/request_path.rs

@@ -1,4 +1,7 @@
-use pesde::manifest::target::TargetKind;
+use pesde::{
+	manifest::target::TargetKind,
+	source::{ids::VersionId, pesde::IndexFile},
+};
 use semver::Version;
 use serde::{Deserialize, Deserializer};
 
@@ -46,6 +49,33 @@ impl<'de> Deserialize<'de> for AnyOrSpecificTarget {
 	}
 }
 
+pub fn resolve_version_and_target(
+	file: &IndexFile,
+	version: LatestOrSpecificVersion,
+	target: AnyOrSpecificTarget,
+) -> Option<&VersionId> {
+	let version = match version {
+		LatestOrSpecificVersion::Latest => match file.entries.keys().map(|k| k.version()).max() {
+			Some(latest) => latest.clone(),
+			None => return None,
+		},
+		LatestOrSpecificVersion::Specific(version) => version,
+	};
+
+	let mut versions = file
+		.entries
+		.iter()
+		.filter(|(v_id, _)| *v_id.version() == version);
+
+	match target {
+		AnyOrSpecificTarget::Any => versions.min_by_key(|(v_id, _)| *v_id.target()),
+		AnyOrSpecificTarget::Specific(kind) => {
+			versions.find(|(_, entry)| entry.target.kind() == kind)
+		}
+	}
+	.map(|(v_id, _)| v_id)
+}
+
 #[derive(Debug)]
 pub enum AllOrSpecificTarget {
 	All,

registry/src/search.rs

@@ -132,8 +132,8 @@ pub async fn make_search(
 			.add_document(doc!(
 				id_field => pkg_name.to_string(),
 				version => v_id.version().to_string(),
-				scope => pkg_name.as_str().0,
-				name => pkg_name.as_str().1,
+				scope => pkg_name.scope(),
+				name => pkg_name.name(),
 				description => latest_entry.description.clone().unwrap_or_default(),
 				published_at => DateTime::from_timestamp_secs(latest_entry.published_at.timestamp()),
 			))
@@ -165,8 +165,8 @@ pub fn update_search_version(
 	search_writer.add_document(doc!(
         id_field => name.to_string(),
 		schema.get_field("version").unwrap() => version.to_string(),
-        schema.get_field("scope").unwrap() => name.as_str().0,
-        schema.get_field("name").unwrap() => name.as_str().1,
+        schema.get_field("scope").unwrap() => name.scope(),
+        schema.get_field("name").unwrap() => name.name(),
         schema.get_field("description").unwrap() => entry.description.clone().unwrap_or_default(),
         schema.get_field("published_at").unwrap() => DateTime::from_timestamp_secs(entry.published_at.timestamp())
     )).unwrap();

registry/src/storage/fs.rs

@@ -9,6 +9,7 @@ use std::{
 	fmt::Display,
 	path::{Path, PathBuf},
 };
+use tokio_util::io::ReaderStream;
 
 #[derive(Debug)]
 pub struct FSStorage {
@@ -19,11 +20,11 @@ async fn read_file_to_response(
 	path: &Path,
 	content_type: &str,
 ) -> Result<HttpResponse, RegistryError> {
-	Ok(match fs::read(path).await {
-		Ok(contents) => HttpResponse::Ok()
+	Ok(match fs::File::open(path).await {
+		Ok(file) => HttpResponse::Ok()
 			.append_header((CONTENT_TYPE, content_type))
 			.append_header((CONTENT_ENCODING, "gzip"))
-			.body(contents),
+			.streaming(ReaderStream::new(file)),
 		Err(e) if e.kind() == std::io::ErrorKind::NotFound => HttpResponse::NotFound().finish(),
 		Err(e) => return Err(e.into()),
 	})