directly return rustdoc rustdoc json instead of redirect

Author: syphar

src/docbuilder/mod.rs

@@ -8,4 +8,6 @@ pub use self::rustwide_builder::{
 };
 
 #[cfg(test)]
-pub use self::rustwide_builder::RUSTDOC_JSON_COMPRESSION_ALGORITHMS;
+pub use self::rustwide_builder::{
+    RUSTDOC_JSON_COMPRESSION_ALGORITHMS, read_format_version_from_rustdoc_json,
+};

src/docbuilder/rustwide_builder.rs

@@ -54,7 +54,7 @@ pub const RUSTDOC_JSON_COMPRESSION_ALGORITHMS: &[CompressionAlgorithm] =
     &[CompressionAlgorithm::Zstd, CompressionAlgorithm::Gzip];
 
 /// read the format version from a rustdoc JSON file.
-fn read_format_version_from_rustdoc_json(
+pub fn read_format_version_from_rustdoc_json(
     reader: impl std::io::Read,
 ) -> Result<RustdocJsonFormatVersion> {
     let reader = BufReader::new(reader);

src/storage/mod.rs

@@ -419,14 +419,22 @@ impl AsyncStorage {
         self.get_stream(path).await?.materialize(max_size).await
     }
 
-    /// get a decompressing stream to an object in storage
+    /// get a raw stream to an object in storage
+    ///
+    /// We don't decompress ourselves, S3 only decompresses with a correct
+    /// `Content-Encoding` header set, which we don't.
     #[instrument]
-    pub(crate) async fn get_stream(&self, path: &str) -> Result<StreamingBlob> {
-        let blob = match &self.backend {
+    pub(crate) async fn get_raw_stream(&self, path: &str) -> Result<StreamingBlob> {
+        match &self.backend {
             StorageBackend::Database(db) => db.get_stream(path, None).await,
             StorageBackend::S3(s3) => s3.get_stream(path, None).await,
-        }?;
-        Ok(blob.decompress().await?)
+        }
+    }
+
+    /// get a decompressing stream to an object in storage.
+    #[instrument]
+    pub(crate) async fn get_stream(&self, path: &str) -> Result<StreamingBlob> {
+        Ok(self.get_raw_stream(path).await?.decompress().await?)
     }
 
     /// get, decompress and materialize part of an object from store

src/test/mod.rs

@@ -154,7 +154,7 @@ pub(crate) trait AxumRouterTestExt {
         path: &str,
         cache_policy: cache::CachePolicy,
         config: &Config,
-    ) -> Result<()>;
+    ) -> Result<AxumResponse>;
     async fn assert_success(&self, path: &str) -> Result<AxumResponse>;
     async fn get(&self, path: &str) -> Result<AxumResponse>;
     async fn post(&self, path: &str) -> Result<AxumResponse>;
@@ -269,7 +269,7 @@ impl AxumRouterTestExt for axum::Router {
         path: &str,
         cache_policy: cache::CachePolicy,
         config: &Config,
-    ) -> Result<()> {
+    ) -> Result<AxumResponse> {
         let response = self.get(path).await?;
         let status = response.status();
         assert!(
@@ -278,7 +278,7 @@ impl AxumRouterTestExt for axum::Router {
             response.redirect_target().unwrap_or_default()
         );
         response.assert_cache_control(cache_policy, config);
-        Ok(())
+        Ok(response)
     }
 
     async fn get(&self, path: &str) -> Result<AxumResponse> {

src/web/extractors/mod.rs

@@ -3,4 +3,4 @@ mod path;
 pub(crate) mod rustdoc;
 
 pub(crate) use context::DbConnection;
-pub(crate) use path::{Path, PathFileExtension};
+pub(crate) use path::{Path, WantedCompression};

src/web/extractors/path.rs

@@ -1,11 +1,15 @@
 //! custom axum extractors for path parameters
-use crate::web::error::AxumNope;
+use crate::{
+    storage::{CompressionAlgorithm, compression::compression_from_file_extension},
+    web::error::AxumNope,
+};
 use anyhow::anyhow;
 use axum::{
     RequestPartsExt,
     extract::{FromRequestParts, OptionalFromRequestParts},
     http::request::Parts,
 };
+use derive_more::Deref;
 
 /// custom axum `Path` extractor that uses our own AxumNope::BadRequest
 /// as error response instead of a plain text "bad request"
@@ -97,6 +101,58 @@ where
     }
 }
 
+/// get wanted compression from file extension in path.
+///
+/// TODO: we could also additionally read the accept-encoding header here. But especially
+/// in combination with priorities it's complex to parse correctly. So for now only
+/// file extensions in the URL.
+/// When using Accept-Encoding, we also have to return "Vary: Accept-Encoding" to ensure
+/// the cache behaves correctly.
+#[derive(Debug, Clone, Deref, Default, PartialEq)]
+pub(crate) struct WantedCompression(pub(crate) CompressionAlgorithm);
+
+impl<S> FromRequestParts<S> for WantedCompression
+where
+    S: Send + Sync,
+{
+    type Rejection = AxumNope;
+
+    async fn from_request_parts(parts: &mut Parts, _state: &S) -> Result<Self, Self::Rejection> {
+        parts
+            .extract::<Option<WantedCompression>>()
+            .await
+            .expect("can never fail")
+            .ok_or_else(|| AxumNope::BadRequest(anyhow!("compression extension not found in path")))
+    }
+}
+
+impl<S> OptionalFromRequestParts<S> for WantedCompression
+where
+    S: Send + Sync,
+{
+    type Rejection = AxumNope;
+
+    async fn from_request_parts(
+        parts: &mut Parts,
+        _state: &S,
+    ) -> Result<Option<Self>, Self::Rejection> {
+        if let Some(ext) = parts
+            .extract::<Option<PathFileExtension>>()
+            .await
+            .expect("can't fail")
+            .map(|ext| ext.0)
+        {
+            Ok(Some(WantedCompression(
+                compression_from_file_extension(&ext).ok_or_else(|| {
+                    AxumNope::BadRequest(anyhow!("unknown compression file extension: {}", ext))
+                })?,
+            )))
+        } else {
+            Ok(None)
+        }
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;

src/web/extractors/rustdoc.rs

@@ -2,6 +2,7 @@
 
 use crate::{
     db::BuildId,
+    storage::CompressionAlgorithm,
     web::{
         MatchedRelease, MetaData, ReqVersion, error::AxumNope, escaped_uri::EscapedURI,
         extractors::Path, url_decode,
@@ -573,6 +574,30 @@ impl RustdocParams {
         EscapedURI::from_path(path)
     }
 
+    pub(crate) fn json_download_url(
+        &self,
+        wanted_compression: Option<CompressionAlgorithm>,
+        format_version: Option<&str>,
+    ) -> EscapedURI {
+        let mut path = format!("/crate/{}/{}", self.name, self.req_version);
+
+        if let Some(doc_target) = self.doc_target() {
+            path.push_str(&format!("/{doc_target}"));
+        }
+
+        if let Some(format_version) = format_version {
+            path.push_str(&format!("/json/{format_version}"));
+        } else {
+            path.push_str("/json");
+        }
+
+        if let Some(wanted_compression) = wanted_compression {
+            path.push_str(&format!(".{}", wanted_compression.file_extension()));
+        }
+
+        EscapedURI::from_path(path)
+    }
+
     pub(crate) fn features_url(&self) -> EscapedURI {
         EscapedURI::from_path(format!(
             "/crate/{}/{}/features",
@@ -863,7 +888,7 @@ mod tests {
     use super::*;
     use crate::{
         db::types::version::Version,
-        test::{AxumResponseTestExt, AxumRouterTestExt},
+        test::{AxumResponseTestExt, AxumRouterTestExt, V1},
     };
     use axum::{Router, routing::get};
     use test_case::test_case;
@@ -1719,4 +1744,66 @@ mod tests {
             format!("/{KRATE}/0.14.0/{KRATE}/trait.Itertools.html")
         )
     }
+
+    #[test_case(None)]
+    #[test_case(Some(CompressionAlgorithm::Gzip))]
+    #[test_case(Some(CompressionAlgorithm::Zstd))]
+    fn test_plain_json_url(wanted_compression: Option<CompressionAlgorithm>) {
+        let mut params = RustdocParams::new(KRATE)
+            .with_page_kind(PageKind::Rustdoc)
+            .with_req_version(ReqVersion::Exact(V1));
+
+        assert_eq!(
+            params.json_download_url(wanted_compression, None),
+            format!(
+                "/crate/{KRATE}/{V1}/json{}",
+                wanted_compression
+                    .map(|c| format!(".{}", c.file_extension()))
+                    .unwrap_or_default()
+            )
+        );
+
+        params = params.with_doc_target("some-target");
+
+        assert_eq!(
+            params.json_download_url(wanted_compression, None),
+            format!(
+                "/crate/{KRATE}/{V1}/some-target/json{}",
+                wanted_compression
+                    .map(|c| format!(".{}", c.file_extension()))
+                    .unwrap_or_default()
+            )
+        );
+    }
+
+    #[test_case(None)]
+    #[test_case(Some(CompressionAlgorithm::Gzip))]
+    #[test_case(Some(CompressionAlgorithm::Zstd))]
+    fn test_plain_json_url_with_format(wanted_compression: Option<CompressionAlgorithm>) {
+        let mut params = RustdocParams::new(KRATE)
+            .with_page_kind(PageKind::Rustdoc)
+            .with_req_version(ReqVersion::Exact(V1));
+
+        assert_eq!(
+            params.json_download_url(wanted_compression, Some("42")),
+            format!(
+                "/crate/{KRATE}/{V1}/json/42{}",
+                wanted_compression
+                    .map(|c| format!(".{}", c.file_extension()))
+                    .unwrap_or_default()
+            )
+        );
+
+        params = params.with_doc_target("some-target");
+
+        assert_eq!(
+            params.json_download_url(wanted_compression, Some("42")),
+            format!(
+                "/crate/{KRATE}/{V1}/some-target/json/42{}",
+                wanted_compression
+                    .map(|c| format!(".{}", c.file_extension()))
+                    .unwrap_or_default()
+            )
+        );
+    }
 }