Separate registry-api from the index, use the configured base url

Author: Nemo157

src/bin/cratesfyi.rs

@@ -16,7 +16,7 @@ use docs_rs::utils::{
 };
 use docs_rs::{
     start_background_metrics_webserver, start_web_server, AsyncStorage, BuildQueue, Config,
-    Context, Index, InstanceMetrics, PackageKind, RustwideBuilder, ServiceMetrics, Storage,
+    Context, Index, RegistryApi, InstanceMetrics, PackageKind, RustwideBuilder, ServiceMetrics, Storage,
 };
 use humantime::Duration;
 use once_cell::sync::OnceCell;
@@ -537,12 +537,10 @@ impl DatabaseSubcommand {
             }
 
             Self::UpdateCrateRegistryFields { name } => {
-                let index = ctx.index()?;
-
                 db::update_crate_data_in_database(
                     &mut *ctx.conn()?,
                     &name,
-                    &index.api().get_crate_data(&name)?,
+                    &ctx.registry_api()?.get_crate_data(&name)?,
                 )?;
             }
 
@@ -719,6 +717,7 @@ struct BinContext {
     service_metrics: OnceCell<Arc<ServiceMetrics>>,
     instance_metrics: OnceCell<Arc<InstanceMetrics>>,
     index: OnceCell<Arc<Index>>,
+    registry_api: OnceCell<Arc<RegistryApi>>,
     repository_stats_updater: OnceCell<Arc<RepositoryStatsUpdater>>,
     runtime: OnceCell<Arc<Runtime>>,
 }
@@ -734,6 +733,7 @@ impl BinContext {
             service_metrics: OnceCell::new(),
             instance_metrics: OnceCell::new(),
             index: OnceCell::new(),
+            registry_api: OnceCell::new(),
             repository_stats_updater: OnceCell::new(),
             runtime: OnceCell::new(),
         }
@@ -783,11 +783,15 @@ impl Context for BinContext {
             let config = self.config()?;
             let path = config.registry_index_path.clone();
             if let Some(registry_url) = config.registry_url.clone() {
-                Index::from_url(path, registry_url, config.crates_io_api_call_retries)
+                Index::from_url(path, registry_url)
             } else {
-                Index::new(path, config.crates_io_api_call_retries)
+                Index::new(path)
             }?
         };
+        fn registry_api(self) -> RegistryApi = {
+            let config = self.config()?;
+            RegistryApi::new(config.registry_api_host.clone(), config.crates_io_api_call_retries)?
+        };
         fn repository_stats_updater(self) -> RepositoryStatsUpdater = {
             let config = self.config()?;
             let pool = self.pool()?;

src/config.rs

@@ -2,13 +2,14 @@ use crate::{cdn::CdnKind, storage::StorageKind};
 use anyhow::{anyhow, bail, Context, Result};
 use std::{env::VarError, error::Error, path::PathBuf, str::FromStr, time::Duration};
 use tracing::trace;
+use url::Url;
 
 #[derive(Debug)]
 pub struct Config {
     pub prefix: PathBuf,
     pub registry_index_path: PathBuf,
     pub registry_url: Option<String>,
-    pub registry_api_host: String,
+    pub registry_api_host: Url,
 
     // Database connection params
     pub(crate) database_url: String,
@@ -140,7 +141,7 @@ impl Config {
 
             registry_index_path: env("REGISTRY_INDEX_PATH", prefix.join("crates.io-index"))?,
             registry_url: maybe_env("REGISTRY_URL")?,
-            registry_api_host: env("DOCSRS_REGISTRY_API_HOST", "https://crates.io".into())?,
+            registry_api_host: env("DOCSRS_REGISTRY_API_HOST", "https://crates.io".parse().unwrap())?,
             prefix: prefix.clone(),
 
             database_url: require_env("DOCSRS_DATABASE_URL")?,

src/context.rs

@@ -2,7 +2,7 @@ use crate::cdn::CdnBackend;
 use crate::db::Pool;
 use crate::error::Result;
 use crate::repositories::RepositoryStatsUpdater;
-use crate::{AsyncStorage, BuildQueue, Config, Index, InstanceMetrics, ServiceMetrics, Storage};
+use crate::{AsyncStorage, BuildQueue, Config, Index, InstanceMetrics, ServiceMetrics, Storage, RegistryApi};
 use std::sync::Arc;
 use tokio::runtime::Runtime;
 
@@ -16,6 +16,7 @@ pub trait Context {
     fn service_metrics(&self) -> Result<Arc<ServiceMetrics>>;
     fn instance_metrics(&self) -> Result<Arc<InstanceMetrics>>;
     fn index(&self) -> Result<Arc<Index>>;
+    fn registry_api(&self) -> Result<Arc<RegistryApi>>;
     fn repository_stats_updater(&self) -> Result<Arc<RepositoryStatsUpdater>>;
     fn runtime(&self) -> Result<Arc<Runtime>>;
 }

src/db/add_package.rs

@@ -2,7 +2,7 @@ use crate::{
     db::types::Feature,
     docbuilder::{BuildResult, DocCoverage},
     error::Result,
-    index::api::{CrateData, CrateOwner, ReleaseData},
+    registry_api::{CrateData, CrateOwner, ReleaseData},
     storage::CompressionAlgorithm,
     utils::MetadataPackage,
     web::crate_details::CrateDetails,

src/db/delete.rs

@@ -194,7 +194,7 @@ fn delete_crate_from_database(conn: &mut Client, name: &str, crate_id: i32) -> R
 #[cfg(test)]
 mod tests {
     use super::*;
-    use crate::index::api::CrateOwner;
+    use crate::registry_api::CrateOwner;
     use crate::test::{assert_success, wrapper};
     use postgres::Client;
     use test_case::test_case;

src/docbuilder/rustwide_builder.rs

@@ -13,7 +13,7 @@ use crate::utils::{
 };
 use crate::RUSTDOC_STATIC_STORAGE_PREFIX;
 use crate::{db::blacklist::is_blacklisted, utils::MetadataPackage};
-use crate::{Config, Context, Index, InstanceMetrics, Storage};
+use crate::{Config, Context, RegistryApi, InstanceMetrics, Storage};
 use anyhow::{anyhow, bail, Context as _, Error};
 use docsrs_metadata::{BuildTargets, Metadata, DEFAULT_TARGETS, HOST_TARGET};
 use failure::Error as FailureError;
@@ -87,7 +87,7 @@ pub struct RustwideBuilder {
     db: Pool,
     storage: Arc<Storage>,
     metrics: Arc<InstanceMetrics>,
-    index: Arc<Index>,
+    registry_api: Arc<RegistryApi>,
     rustc_version: String,
     repository_stats_updater: Arc<RepositoryStatsUpdater>,
     workspace_initialize_time: Instant,
@@ -105,7 +105,7 @@ impl RustwideBuilder {
             db: pool,
             storage: context.storage()?,
             metrics: context.instance_metrics()?,
-            index: context.index()?,
+            registry_api: context.registry_api()?,
             rustc_version: String::new(),
             repository_stats_updater: context.repository_stats_updater()?,
             workspace_initialize_time: Instant::now(),
@@ -519,8 +519,7 @@ impl RustwideBuilder {
 
                     let release_data = if !is_local {
                         match self
-                            .index
-                            .api()
+                            .registry_api
                             .get_release_data(name, version)
                             .with_context(|| {
                                 format!("could not fetch releases-data for {name}-{version}")
@@ -565,7 +564,7 @@ impl RustwideBuilder {
 
                     // Some crates.io crate data is mutable, so we proactively update it during a release
                     if !is_local {
-                        match self.index.api().get_crate_data(name) {
+                        match self.registry_api.get_crate_data(name) {
                             Ok(crate_data) => {
                                 update_crate_data_in_database(&mut conn, name, &crate_data)?
                             }

src/index/mod.rs renamed to src/index.rs

@@ -3,72 +3,40 @@ use std::{path::PathBuf, process::Command};
 
 use anyhow::Context;
 use crates_index_diff::gix;
-use url::Url;
 
-use self::api::Api;
 use crate::error::Result;
 use crate::utils::report_error;
 
-pub(crate) mod api;
-
 pub struct Index {
     path: PathBuf,
-    api: Api,
     repository_url: Option<String>,
 }
 
-#[derive(Debug, serde::Deserialize, Clone)]
-#[serde(rename_all = "kebab-case")]
-struct IndexConfig {
-    #[serde(default)]
-    api: Option<Url>,
-}
-
-/// Inspects the given repository to find the config as specified in [RFC 2141][], assumes that the
-/// repository has a remote called `origin` and that the branch `master` exists on it.
-///
-/// [RFC 2141]: https://rust-lang.github.io/rfcs/2141-alternative-registries.html
-fn load_config(repo: &gix::Repository) -> Result<IndexConfig> {
-    let file = repo
-        .rev_parse_single("refs/remotes/origin/master:config.json")
-        .with_context(|| anyhow::anyhow!("registry index missing ./config.json in root"))?
-        .object()?;
-
-    let config = serde_json::from_slice(&file.data)?;
-    Ok(config)
-}
-
 impl Index {
-    pub fn from_url(path: PathBuf, url: String, max_api_call_retries: u32) -> Result<Self> {
-        let diff = crates_index_diff::Index::from_path_or_cloned_with_options(
+    pub fn from_url(path: PathBuf, url: String) -> Result<Self> {
+        crates_index_diff::Index::from_path_or_cloned_with_options(
             &path,
             gix::progress::Discard,
             &AtomicBool::default(),
             crates_index_diff::index::CloneOptions { url: url.clone() },
         )
+        .map(|_| ())
         .context("initialising registry index repository")?;
 
-        let config = load_config(diff.repository()).context("loading registry config")?;
-        let api = Api::new(config.api, max_api_call_retries)
-            .context("initialising registry api client")?;
         Ok(Self {
             path,
-            api,
             repository_url: Some(url),
         })
     }
 
-    pub fn new(path: PathBuf, max_api_call_retries: u32) -> Result<Self> {
+    pub fn new(path: PathBuf) -> Result<Self> {
         // This initializes the repository, then closes it afterwards to avoid leaking file descriptors.
         // See https://github.com/rust-lang/docs.rs/pull/847
-        let diff = crates_index_diff::Index::from_path_or_cloned(&path)
+        crates_index_diff::Index::from_path_or_cloned(&path)
+            .map(|_| ())
             .context("initialising registry index repository")?;
-        let config = load_config(diff.repository()).context("loading registry config")?;
-        let api = Api::new(config.api, max_api_call_retries)
-            .context("initialising registry api client")?;
         Ok(Self {
             path,
-            api,
             repository_url: None,
         })
     }
@@ -102,10 +70,6 @@ impl Index {
         Ok(index)
     }
 
-    pub fn api(&self) -> &Api {
-        &self.api
-    }
-
     pub fn run_git_gc(&self) {
         let gc = Command::new("git")
             .arg("-C")

src/index/crates.rs

@@ -8,10 +8,12 @@ pub use self::context::Context;
 pub use self::docbuilder::PackageKind;
 pub use self::docbuilder::RustwideBuilder;
 pub use self::index::Index;
+pub use self::registry_api::RegistryApi;
 pub use self::metrics::{InstanceMetrics, ServiceMetrics};
 pub use self::storage::{AsyncStorage, Storage};
 pub use self::web::{start_background_metrics_webserver, start_web_server};
 
+mod registry_api;
 mod build_queue;
 pub mod cdn;
 mod config;

src/lib.rs

@@ -13,8 +13,8 @@ const APP_USER_AGENT: &str = concat!(
 );
 
 #[derive(Debug)]
-pub struct Api {
-    api_base: Option<Url>,
+pub struct RegistryApi {
+    api_base: Url,
     max_retries: u32,
     client: reqwest::blocking::Client,
 }
@@ -47,8 +47,8 @@ pub struct CrateOwner {
     pub(crate) login: String,
 }
 
-impl Api {
-    pub(super) fn new(api_base: Option<Url>, max_retries: u32) -> Result<Self> {
+impl RegistryApi {
+    pub fn new(api_base: Url, max_retries: u32) -> Result<Self> {
         let headers = vec![
             (USER_AGENT, HeaderValue::from_static(APP_USER_AGENT)),
             (ACCEPT, HeaderValue::from_static("application/json")),
@@ -67,12 +67,6 @@ impl Api {
         })
     }
 
-    fn api_base(&self) -> Result<Url> {
-        self.api_base
-            .clone()
-            .with_context(|| anyhow!("index is missing an api base url"))
-    }
-
     pub fn get_crate_data(&self, name: &str) -> Result<CrateData> {
         let owners = self
             .get_owners(name)
@@ -100,7 +94,7 @@ impl Api {
         version: &str,
     ) -> Result<(DateTime<Utc>, bool, i32)> {
         let url = {
-            let mut url = self.api_base()?;
+            let mut url = self.api_base.clone();
             url.path_segments_mut()
                 .map_err(|()| anyhow!("Invalid API url"))?
                 .extend(&["api", "v1", "crates", name, "versions"]);
@@ -142,7 +136,7 @@ impl Api {
     /// Fetch owners from the registry's API
     fn get_owners(&self, name: &str) -> Result<Vec<CrateOwner>> {
         let url = {
-            let mut url = self.api_base()?;
+            let mut url = self.api_base.clone();
             url.path_segments_mut()
                 .map_err(|()| anyhow!("Invalid API url"))?
                 .extend(&["api", "v1", "crates", name, "owners"]);