rust-lang · syphar · Aug 11, 2026
@@ -29,8 +29,7 @@ use docs_rs_rustdoc_json::{
     read_format_version_from_rustdoc_json,
 };
 use docs_rs_storage::{
-    AsyncStorage, Storage, compress, file_list_to_json, rustdoc_archive_path, rustdoc_json_path,
-    source_archive_path,
+    AsyncStorage, Storage, compress, rustdoc_archive_path, rustdoc_json_path, source_archive_path,
 };
 use docs_rs_types::{
     BuildId, BuildStatus, CompressionAlgorithm, CrateId, KrateName, ReleaseId, Version,
@@ -640,23 +639,22 @@ impl RustwideBuilder {
         let local_storage = tempfile::tempdir_in(&self.config.temp_dir)?;
 
         let mut algs = HashSet::new();
-        let (source_files_list, source_size) = {
+        let source_stats = {
             let _span = info_span!("adding sources into database").entered();
             debug!("adding sources into database");
             let temp_dir = tempfile::tempdir_in(&self.config.temp_dir)?;
 
             krate.copy_source_to(&self.workspace, temp_dir.path())?;
 
-            let (files_list, new_alg) = self.runtime.block_on(
+            let stats = self.runtime.block_on(
                 self.storage
                     .store_all_in_archive(&source_archive_path(name, version), &temp_dir),
             )?;
 
             fs::remove_dir_all(temp_dir.path())?;
 
-            algs.insert(new_alg);
-            let source_size: u64 = files_list.iter().map(|info| info.size).sum();
-            (files_list, source_size)
+            algs.insert(stats.alg);
+            stats
         };
 
         let successful = build_dir
@@ -740,16 +738,15 @@ impl RustwideBuilder {
                         target_build_logs.insert(target, (target_res.build_log, successful));
                     }
 
-                    let (file_list, new_alg) =
+                    let doc_stats  =
                         self.runtime.block_on(
                         self.storage.store_all_in_archive(
                             &rustdoc_archive_path(name, version),
                             local_storage.path(),
                         ))?;
-                    let documentation_size = file_list.iter().map(|info| info.size).sum::<u64>();
-                    self.builder_metrics.documentation_size.record(documentation_size, &[]);
-                    algs.insert(new_alg);
-                    Some(documentation_size)
+                    self.builder_metrics.documentation_size.record(doc_stats.original_size, &[]);
+                    algs.insert(doc_stats.alg);
+                    Some(doc_stats.original_size)
                 } else {
                     None
                 };
@@ -845,14 +842,13 @@ impl RustwideBuilder {
                     cargo_metadata,
                     &build.host_source_dir(),
                     &res.target,
-                    file_list_to_json(source_files_list),
                     successful_targets,
                     &release_data,
                     has_docs,
                     has_examples,
                     algs,
                     repository,
-                    source_size,
+                    source_stats.original_size,
                 ))?;
 
                 if let Some(repository_id) = repository {
@@ -1836,7 +1832,6 @@ mod tests {
                 },
                 Path::new("/unknown/"),
                 "x86_64-unknown-linux-gnu",
-                serde_json::Value::Array(vec![]),
                 vec![
                     "i686-pc-windows-msvc".into(),
                     "aarch64-unknown-linux-gnu".into(),

@@ -16,7 +16,7 @@ use docs_rs_rustdoc_json::{
     RUSTDOC_JSON_COMPRESSION_ALGORITHMS, RustdocJsonFormatVersion,
     read_format_version_from_rustdoc_json,
 };
-use docs_rs_storage::{AsyncStorage, file_list_to_json, rustdoc_archive_path, source_archive_path};
+use docs_rs_storage::{AsyncStorage, rustdoc_archive_path, source_archive_path};
 use docs_rs_storage::{compress, decompress, rustdoc_json_path};
 use docs_rs_types::{
     BuildId, BuildStatus, CrateId, KrateName, ReleaseId, ReqVersion, SimpleBuildError, Version,
@@ -113,15 +113,14 @@ async fn import_test_release_inner(
     .await?;
 
     let mut algs = HashSet::new();
-    let (source_files_list, source_size) = {
+    let source_stats = {
         info!("writing source files to storage...");
-        let (files_list, new_alg) = storage
+        let stats = storage
             .store_all_in_archive(&source_archive_path(name, version), &source_dir)
             .await?;
 
-        algs.insert(new_alg);
-        let source_size: u64 = files_list.iter().map(|info| info.size).sum();
-        (files_list, source_size)
+        algs.insert(stats.alg);
+        stats
     };
 
     let registry_data = registry_api
@@ -178,11 +177,10 @@ async fn import_test_release_inner(
     }
 
     info!("writing rustdoc files to storage...");
-    let (rustdoc_file_list, new_alg) = storage
+    let doc_stats = storage
         .store_all_in_archive(&rustdoc_archive_path(name, version), &rustdoc_dir)
         .await?;
-    let documentation_size: u64 = rustdoc_file_list.iter().map(|info| info.size).sum();
-    algs.insert(new_alg);
+    algs.insert(doc_stats.alg);
 
     info!("loading repository stats...");
     let repository_id = repository_stats
@@ -232,14 +230,13 @@ async fn import_test_release_inner(
         cargo_metadata.root(),
         &source_dir,
         default_target,
-        file_list_to_json(source_files_list),
         all_targets,
         &registry_data,
         true,
         false, // FIXME: real has_examples?
         algs,
         repository_id,
-        source_size,
+        source_stats.original_size,
     )
     .await?;
 
@@ -249,7 +246,7 @@ async fn import_test_release_inner(
         "rustc 1.95.0-nightly (873d4682c 2026-01-25)",
         BUILD_VERSION,
         BuildStatus::Success,
-        Some(documentation_size),
+        Some(doc_stats.original_size),
         None,
         None::<&SimpleBuildError>,
     )

@@ -425,38 +425,6 @@ mod tests {
         });
     }
 
-    #[test]
-    fn empty_file_list_dont_break_the_view() {
-        async_wrapper(|env| async move {
-            let release_id = env
-                .fake_release()
-                .await
-                .name("fake")
-                .version("0.1.0")
-                .source_file("README.md", b"hello")
-                .create()
-                .await?;
-
-            let path = "/crate/fake/0.1.0/source/README.md";
-            let web = env.web_app().await;
-            web.assert_success(path).await?;
-
-            let mut conn = env.async_conn().await?;
-            sqlx::query!(
-                "UPDATE releases
-                     SET files = NULL
-                     WHERE id = $1",
-                release_id.0,
-            )
-            .execute(&mut *conn)
-            .await?;
-
-            assert!(web.get(path).await?.status().is_success());
-
-            Ok(())
-        });
-    }
-
     #[test]
     fn latest_contains_links_to_latest() {
         async_wrapper(|env| async move {

@@ -0,0 +1,3 @@
+ALTER TABLE releases 
+ADD COLUMN archive_storage BOOL NOT NULL DEFAULT TRUE,
+ADD COLUMN files JSON ;
@@ -0,0 +1,3 @@
+ALTER TABLE releases
+DROP COLUMN files,
+DROP COLUMN archive_storage;
@@ -8,7 +8,6 @@ use docs_rs_types::{
 };
 use docs_rs_utils::rustc_version::parse_rustc_date;
 use futures_util::stream::TryStreamExt;
-use serde_json::Value;
 use slug::slugify;
 use std::{
     collections::{HashMap, HashSet},
@@ -21,19 +20,15 @@ use tracing::{debug, error, info, instrument};
 /// Adds a package into database.
 ///
 /// Package must be built first.
-///
-/// NOTE: `source_files` refers to the files originally in the crate,
-/// not the files generated by rustdoc.
 #[allow(clippy::too_many_arguments)]
-#[instrument(skip(conn, compression_algorithms, source_files))]
+#[instrument(skip(conn, compression_algorithms))]
 pub async fn finish_release(
     conn: &mut sqlx::PgConnection,
     crate_id: CrateId,
     release_id: ReleaseId,
     metadata_pkg: &MetadataPackage,
     source_dir: impl AsRef<Path> + fmt::Debug,
     default_target: &str,
-    source_files: Value,
     doc_targets: Vec<String>,
     registry_data: &ReleaseData,
     has_docs: bool,
@@ -72,14 +67,13 @@ pub async fn finish_release(
                keywords = $14,
                have_examples = $15,
                downloads = $16,
-               files = $17,
-               doc_targets = $18,
-               is_library = $19,
-               documentation_url = $20,
-               default_target = $21,
-               features = $22,
-               repository_id = $23,
-               source_size = $24
+               doc_targets = $17,
+               is_library = $18,
+               documentation_url = $19,
+               default_target = $20,
+               features = $21,
+               repository_id = $22,
+               source_size = $23
            WHERE id = $1"#,
         release_id.0,
         registry_data.release_time,
@@ -97,7 +91,6 @@ pub async fn finish_release(
         serde_json::to_value(&metadata_pkg.keywords)?,
         has_examples,
         registry_data.downloads,
-        source_files,
         serde_json::to_value(doc_targets)?,
         is_library,
         metadata_pkg.documentation,
@@ -337,8 +330,8 @@ pub async fn initialize_release(
     version: &Version,
 ) -> Result<ReleaseId> {
     let release_id = sqlx::query_scalar!(
-        r#"INSERT INTO releases (crate_id, version, archive_storage)
-         VALUES ($1, $2, TRUE)
+        r#"INSERT INTO releases (crate_id, version)
+         VALUES ($1, $2)
          ON CONFLICT (crate_id, version) DO UPDATE
          SET -- this `SET` is needed so the id is always returned.
             version = EXCLUDED.version
@@ -697,7 +690,6 @@ mod test {
             },
             tempdir.path(),
             DEFAULT_TARGET,
-            Value::Array(vec![]),
             vec![DEFAULT_TARGET.to_string()],
             &ReleaseData::default(),
             true,
@@ -1471,7 +1463,6 @@ mod test {
             },
             tempdir.path(),
             DEFAULT_TARGET,
-            Value::Array(vec![]),
             vec![DEFAULT_TARGET.to_string()],
             &ReleaseData::default(),
             true,

@@ -45,7 +45,6 @@ mime = { workspace = true }
 moka = { version = "0.12.14", features = ["future"] }
 opentelemetry = { workspace = true }
 rand = { workspace = true, optional = true }
-serde_json = { workspace = true }
 sqlx = { workspace = true } # for sqlite
 strum = { workspace = true }
 tempfile = { workspace = true }

@@ -1,6 +1,5 @@
 use docs_rs_mimes::detect_mime;
 use mime::Mime;
-use serde_json::Value;
 use std::path::PathBuf;
 
 /// represents a file path from our source or documentation builds.
@@ -17,21 +16,6 @@ impl FileEntry {
     }
 }
 
-pub fn file_list_to_json(files: impl IntoIterator<Item = FileEntry>) -> Value {
-    Value::Array(
-        files
-            .into_iter()
-            .map(|info| {
-                Value::Array(vec![
-                    Value::String(info.mime().as_ref().to_string()),
-                    Value::String(info.path.into_os_string().into_string().unwrap()),
-                    Value::Number(info.size.into()),
-                ])
-            })
-            .collect(),
-    )
-}
-
 #[derive(Debug, Clone, Eq)]
 pub enum FolderEntry {
     File(String, Mime),

@@ -6,6 +6,7 @@ mod config;
 pub(crate) mod errors;
 mod file;
 mod metrics;
+mod result;
 pub(crate) mod storage;
 #[cfg(any(test, feature = "testing"))]
 pub mod testing;
@@ -16,7 +17,8 @@ pub use blob::{Blob, BlobUpload, StreamingBlob};
 pub use compression::{compress, compress_async, decompress};
 pub use config::Config;
 pub use errors::{PathNotFoundError, SizeLimitReached};
-pub use file::{FileEntry, FolderEntry, file_list_to_json};
+pub use file::{FileEntry, FolderEntry};
+pub use result::ArchiveStatistics;
 pub use storage::blocking::Storage;
 pub use storage::non_blocking::AsyncStorage;
 pub use types::StorageKind;