add download endpoint for rustdoc archive

src/build_queue.rs

@@ -550,7 +550,8 @@ mod tests {
     fn test_invalidate_cdn_after_build_and_error() {
         crate::test::wrapper(|env| {
             env.override_config(|config| {
-                config.cloudfront_distribution_id_web = Some("distribution_id".into());
+                config.cloudfront_distribution_id_web = Some("distribution_id_web".into());
+                config.cloudfront_distribution_id_static = Some("distribution_id_static".into());
             });
 
             let queue = env.build_queue();
@@ -573,8 +574,16 @@ mod tests {
                 assert_eq!(
                     *ir,
                     [
-                        ("distribution_id".into(), "/will_succeed*".into()),
-                        ("distribution_id".into(), "/crate/will_succeed*".into()),
+                        ("distribution_id_web".into(), "/will_succeed*".into()),
+                        ("distribution_id_web".into(), "/crate/will_succeed*".into()),
+                        (
+                            "distribution_id_static".into(),
+                            "/rustdoc/will_succeed*".into()
+                        ),
+                        (
+                            "distribution_id_static".into(),
+                            "/sources/will_succeed*".into()
+                        ),
                     ]
                 );
             }
@@ -588,10 +597,26 @@ mod tests {
                 assert_eq!(
                     *ir,
                     [
-                        ("distribution_id".into(), "/will_succeed*".into()),
-                        ("distribution_id".into(), "/crate/will_succeed*".into()),
-                        ("distribution_id".into(), "/will_fail*".into()),
-                        ("distribution_id".into(), "/crate/will_fail*".into()),
+                        ("distribution_id_web".into(), "/will_succeed*".into()),
+                        ("distribution_id_web".into(), "/crate/will_succeed*".into()),
+                        (
+                            "distribution_id_static".into(),
+                            "/rustdoc/will_succeed*".into()
+                        ),
+                        (
+                            "distribution_id_static".into(),
+                            "/sources/will_succeed*".into()
+                        ),
+                        ("distribution_id_web".into(), "/will_fail*".into()),
+                        ("distribution_id_web".into(), "/crate/will_fail*".into()),
+                        (
+                            "distribution_id_static".into(),
+                            "/rustdoc/will_fail*".into()
+                        ),
+                        (
+                            "distribution_id_static".into(),
+                            "/sources/will_fail*".into()
+                        ),
                     ]
                 );
             }

src/cdn.rs

@@ -123,7 +123,17 @@ pub(crate) fn invalidate_crate(config: &Config, cdn: &CdnBackend, name: &str) ->
             distribution_id,
             &[&format!("/{}*", name), &format!("/crate/{}*", name)],
         )
-        .context("error creating CDN invalidation")?;
+        .context("error creating web CDN invalidation")?;
+    }
+    if let Some(distribution_id) = config.cloudfront_distribution_id_static.as_ref() {
+        cdn.create_invalidation(
+            distribution_id,
+            &[
+                &format!("/rustdoc/{}*", name),
+                &format!("/sources/{}*", name),
+            ],
+        )
+        .context("error creating static CDN invalidation")?;
     }
 
     Ok(())
@@ -168,6 +178,32 @@ mod tests {
         })
     }
 
+    #[test]
+    fn invalidate_a_crate() {
+        crate::test::wrapper(|env| {
+            env.override_config(|config| {
+                config.cloudfront_distribution_id_web = Some("distribution_id_web".into());
+                config.cloudfront_distribution_id_static = Some("distribution_id_static".into());
+            });
+            invalidate_crate(&*env.config(), &*env.cdn(), "krate")?;
+
+            assert!(matches!(*env.cdn(), CdnBackend::Dummy(_)));
+            if let CdnBackend::Dummy(ref invalidation_requests) = *env.cdn() {
+                let ir = invalidation_requests.lock().unwrap();
+                assert_eq!(
+                    *ir,
+                    [
+                        ("distribution_id_web".into(), "/krate*".into()),
+                        ("distribution_id_web".into(), "/crate/krate*".into()),
+                        ("distribution_id_static".into(), "/rustdoc/krate*".into()),
+                        ("distribution_id_static".into(), "/sources/krate*".into()),
+                    ]
+                );
+            }
+            Ok(())
+        });
+    }
+
     async fn get_mock_config() -> aws_sdk_cloudfront::Config {
         let cfg = aws_config::from_env()
             .region(Region::new("eu-central-1"))

src/config.rs

@@ -26,6 +26,10 @@ pub struct Config {
     #[cfg(test)]
     pub(crate) s3_bucket_is_temporary: bool,
 
+    // CloudFront domain which we can access
+    // public S3 files through
+    pub(crate) s3_static_domain: String,
+
     // Github authentication
     pub(crate) github_accesstoken: Option<String>,
     pub(crate) github_updater_min_rate_limit: u32,
@@ -67,6 +71,8 @@ pub struct Config {
     // CloudFront distribution ID for the web server.
     // Will be used for invalidation-requests.
     pub cloudfront_distribution_id_web: Option<String>,
+    /// same for the `static.docs.rs` distribution
+    pub cloudfront_distribution_id_static: Option<String>,
 
     // Build params
     pub(crate) build_attempts: u16,
@@ -125,6 +131,8 @@ impl Config {
             #[cfg(test)]
             s3_bucket_is_temporary: false,
 
+            s3_static_domain: env("S3_STATIC_DOMAIN", "https://static.docs.rs".to_string())?,
+
             github_accesstoken: maybe_env("DOCSRS_GITHUB_ACCESSTOKEN")?,
             github_updater_min_rate_limit: env("DOCSRS_GITHUB_UPDATER_MIN_RATE_LIMIT", 2500)?,
 
@@ -148,6 +156,7 @@ impl Config {
             cdn_backend: env("DOCSRS_CDN_BACKEND", CdnKind::Dummy)?,
 
             cloudfront_distribution_id_web: maybe_env("CLOUDFRONT_DISTRIBUTION_ID_WEB")?,
+            cloudfront_distribution_id_static: maybe_env("CLOUDFRONT_DISTRIBUTION_ID_STATIC")?,
 
             local_archive_cache_path: env(
                 "DOCSRS_ARCHIVE_INDEX_CACHE_PATH",

src/db/file.rs

@@ -38,8 +38,12 @@ pub fn add_path_into_remote_archive<P: AsRef<Path>>(
     storage: &Storage,
     archive_path: &str,
     path: P,
+    public_access: bool,
 ) -> Result<(Value, CompressionAlgorithm)> {
     let (file_list, algorithm) = storage.store_all_in_archive(archive_path, path.as_ref())?;
+    if public_access {
+        storage.set_public_access(archive_path, true)?;
+    }
     Ok((
         file_list_to_json(file_list.into_iter().collect()),
         algorithm,

src/db/migrate.rs

@@ -848,6 +848,11 @@ pub fn migrate(version: Option<Version>, conn: &mut Client) -> crate::error::Res
             "CREATE INDEX builds_release_id_idx ON builds (rid);",
             "DROP INDEX builds_release_id_idx;",
         ),
+        sql_migration!(
+            context, 35, "add public visibility to files table",
+            "ALTER TABLE files ADD COLUMN public BOOL NOT NULL DEFAULT FALSE;",
+            "ALTER TABLE files DROP COLUMN public;"
+        ),
 
     ];
 

src/docbuilder/rustwide_builder.rs

@@ -410,6 +410,7 @@ impl RustwideBuilder {
                             &self.storage,
                             &rustdoc_archive_path(name, version),
                             local_storage.path(),
+                            true,
                         )?;
                         algs.insert(new_alg);
                     };
@@ -421,6 +422,7 @@ impl RustwideBuilder {
                             &self.storage,
                             &source_archive_path(name, version),
                             build.host_source_dir(),
+                            false,
                         )?;
                         algs.insert(new_alg);
                         files_list

src/storage/database.rs

@@ -21,6 +21,32 @@ impl DatabaseBackend {
         Ok(conn.query(query, &[&path])?[0].get(0))
     }
 
+    pub(super) fn get_public_access(&self, path: &str) -> Result<bool> {
+        match self.pool.get()?.query_opt(
+            "SELECT public 
+             FROM files 
+             WHERE path = $1",
+            &[&path],
+        )? {
+            Some(row) => Ok(row.get(0)),
+            None => Err(super::PathNotFoundError.into()),
+        }
+    }
+
+    pub(super) fn set_public_access(&self, path: &str, public: bool) -> Result<()> {
+        if self.pool.get()?.execute(
+            "UPDATE files 
+             SET public = $2 
+             WHERE path = $1",
+            &[&path, &public],
+        )? == 1
+        {
+            Ok(())
+        } else {
+            Err(super::PathNotFoundError.into())
+        }
+    }
+
     pub(super) fn get(
         &self,
         path: &str,

src/storage/mod.rs

@@ -140,6 +140,20 @@ impl Storage {
         }
     }
 
+    pub(crate) fn get_public_access(&self, path: &str) -> Result<bool> {
+        match &self.backend {
+            StorageBackend::Database(db) => db.get_public_access(path),
+            StorageBackend::S3(s3) => s3.get_public_access(path),
+        }
+    }
+
+    pub(crate) fn set_public_access(&self, path: &str, public: bool) -> Result<()> {
+        match &self.backend {
+            StorageBackend::Database(db) => db.set_public_access(path, public),
+            StorageBackend::S3(s3) => s3.set_public_access(path, public),
+        }
+    }
+
     fn max_file_size_for(&self, path: &str) -> usize {
         if path.ends_with(".html") {
             self.config.max_file_size_html
@@ -620,9 +634,38 @@ mod backend_tests {
         Ok(())
     }
 
+    fn test_set_public(storage: &Storage) -> Result<()> {
+        let path: &str = "foo/bar.txt";
+
+        storage.store_blobs(vec![Blob {
+            path: path.into(),
+            mime: "text/plain".into(),
+            date_updated: Utc::now(),
+            compression: None,
+            content: b"test content\n".to_vec(),
+        }])?;
+
+        assert!(!storage.get_public_access(path)?);
+        storage.set_public_access(path, true)?;
+        assert!(storage.get_public_access(path)?);
+        storage.set_public_access(path, false)?;
+        assert!(!storage.get_public_access(path)?);
+
+        for path in &["bar.txt", "baz.txt", "foo/baz.txt"] {
+            assert!(storage
+                .set_public_access(path, true)
+                .unwrap_err()
+                .downcast_ref::<PathNotFoundError>()
+                .is_some());
+        }
+
+        Ok(())
+    }
+
     fn test_get_object(storage: &Storage) -> Result<()> {
+        let path: &str = "foo/bar.txt";
         let blob = Blob {
-            path: "foo/bar.txt".into(),
+            path: path.into(),
             mime: "text/plain".into(),
             date_updated: Utc::now(),
             compression: None,
@@ -631,16 +674,25 @@ mod backend_tests {
 
         storage.store_blobs(vec![blob.clone()])?;
 
-        let found = storage.get("foo/bar.txt", std::usize::MAX)?;
+        let found = storage.get(path, std::usize::MAX)?;
         assert_eq!(blob.mime, found.mime);
         assert_eq!(blob.content, found.content);
 
+        // default visibility is private
+        assert!(!storage.get_public_access(path)?);
+
         for path in &["bar.txt", "baz.txt", "foo/baz.txt"] {
             assert!(storage
                 .get(path, std::usize::MAX)
                 .unwrap_err()
                 .downcast_ref::<PathNotFoundError>()
                 .is_some());
+
+            assert!(storage
+                .get_public_access(path)
+                .unwrap_err()
+                .downcast_ref::<PathNotFoundError>()
+                .is_some());
         }
 
         Ok(())
@@ -1028,6 +1080,7 @@ mod backend_tests {
             test_delete_prefix_without_matches,
             test_delete_percent,
             test_exists_without_remote_archive,
+            test_set_public,
         }
 
         tests_with_metrics {