refactor: use RwLock<Vec<Arc<PackageJson>> for package.json storage

Author: Boshen

Cargo.lock

@@ -82,6 +82,7 @@ indexmap = { version = "2", features = ["serde"] }
 json-strip-comments = "3"
 once_cell = "1" # Use `std::sync::OnceLock::get_or_try_init` when it is stable.
 papaya = "0.2"
+parking_lot = "0.12"
 rustc-hash = { version = "2" }
 serde = { version = "1", features = ["derive"] } # derive for Deserialize from package.json
 serde_json = { version = "1", features = ["preserve_order"] } # preserve_order: package_json.exports requires order such as `["require", "import", "default"]`

Cargo.toml

@@ -11,6 +11,7 @@ use cfg_if::cfg_if;
 #[cfg(feature = "yarn_pnp")]
 use once_cell::sync::OnceCell;
 use papaya::{HashMap, HashSet};
+use parking_lot::RwLock;
 use rustc_hash::FxHasher;
 
 use super::borrowed_path::BorrowedCachedPath;
@@ -21,11 +22,14 @@ use crate::{
     context::ResolveContext as Ctx, path::PathUtil,
 };
 
+pub type PackageJsonIndex = usize;
+
 /// Cache implementation used for caching filesystem access.
 #[derive(Default)]
 pub struct Cache<Fs> {
     pub(crate) fs: Fs,
     pub(crate) paths: HashSet<CachedPath, BuildHasherDefault<IdentityHasher>>,
+    pub(crate) package_jsons: RwLock<Vec<Arc<PackageJson>>>,
     pub(crate) tsconfigs: HashMap<PathBuf, Arc<TsConfig>, BuildHasherDefault<FxHasher>>,
     #[cfg(feature = "yarn_pnp")]
     pub(crate) yarn_pnp_manifest: OnceCell<pnp::Manifest>,
@@ -35,6 +39,7 @@ impl<Fs: FileSystem> Cache<Fs> {
     pub fn clear(&self) {
         self.paths.pin().clear();
         self.tsconfigs.pin().clear();
+        self.package_jsons.write().clear();
     }
 
     #[allow(clippy::cast_possible_truncation)]
@@ -110,38 +115,45 @@ impl<Fs: FileSystem> Cache<Fs> {
             .get_or_try_init(|| {
                 let package_json_path = path.path.join("package.json");
                 let Ok(package_json_bytes) = self.fs.read(&package_json_path) else {
+                    if let Some(deps) = &mut ctx.missing_dependencies {
+                        deps.push(package_json_path.clone());
+                    }
                     return Ok(None);
                 };
-
                 let real_path = if options.symlinks {
                     self.canonicalize(path)?.join("package.json")
                 } else {
                     package_json_path.clone()
                 };
-                PackageJson::parse(&self.fs, package_json_path, real_path, package_json_bytes)
-                    .map(|package_json| Some(Arc::new(package_json)))
-                    .map_err(ResolveError::Json)
+                PackageJson::parse(
+                    &self.fs,
+                    package_json_path.clone(),
+                    real_path,
+                    package_json_bytes,
+                )
+                .map(|package_json| {
+                    let arc = Arc::new(package_json);
+                    let mut arena = self.package_jsons.write();
+                    let index = arena.len();
+                    arena.push(arc);
+                    Some(index)
+                })
+                .map_err(ResolveError::Json)
+                // https://github.com/webpack/enhanced-resolve/blob/58464fc7cb56673c9aa849e68e6300239601e615/lib/DescriptionFileUtils.js#L68-L82
+                .inspect(|_| {
+                    ctx.add_file_dependency(&package_json_path);
+                })
+                .inspect_err(|_| {
+                    if let Some(deps) = &mut ctx.file_dependencies {
+                        deps.push(package_json_path.clone());
+                    }
+                })
             })
             .cloned();
 
-        // https://github.com/webpack/enhanced-resolve/blob/58464fc7cb56673c9aa849e68e6300239601e615/lib/DescriptionFileUtils.js#L68-L82
-        match &result {
-            Ok(Some(package_json)) => {
-                ctx.add_file_dependency(&package_json.path);
-            }
-            Ok(None) => {
-                // Avoid an allocation by making this lazy
-                if let Some(deps) = &mut ctx.missing_dependencies {
-                    deps.push(path.path.join("package.json"));
-                }
-            }
-            Err(_) => {
-                if let Some(deps) = &mut ctx.file_dependencies {
-                    deps.push(path.path.join("package.json"));
-                }
-            }
-        }
-        result
+        result.map(|option_index| {
+            option_index.and_then(|index| self.package_jsons.read().get(index).cloned())
+        })
     }
 
     pub(crate) fn get_tsconfig<F: FnOnce(&mut TsConfig) -> Result<(), ResolveError>>(
@@ -213,6 +225,7 @@ impl<Fs: FileSystem> Cache<Fs> {
                 .hasher(BuildHasherDefault::default())
                 .resize_mode(papaya::ResizeMode::Blocking)
                 .build(),
+            package_jsons: RwLock::new(Vec::with_capacity(512)),
             tsconfigs: HashMap::builder()
                 .hasher(BuildHasherDefault::default())
                 .resize_mode(papaya::ResizeMode::Blocking)

src/cache/cache_impl.rs

@@ -11,6 +11,7 @@ use cfg_if::cfg_if;
 use once_cell::sync::OnceCell as OnceLock;
 
 use super::cache_impl::Cache;
+use super::cache_impl::PackageJsonIndex;
 use super::thread_local::SCRATCH_PATH;
 use crate::{
     FileSystem, PackageJson, ResolveError, ResolveOptions, TsConfig, context::ResolveContext as Ctx,
@@ -28,7 +29,7 @@ pub struct CachedPathImpl {
     pub meta: OnceLock<Option<(/* is_file */ bool, /* is_dir */ bool)>>, // None means not found.
     pub canonicalized: OnceLock<Weak<CachedPathImpl>>,
     pub node_modules: OnceLock<Option<Weak<CachedPathImpl>>>,
-    pub package_json: OnceLock<Option<Arc<PackageJson>>>,
+    pub package_json: OnceLock<Option<PackageJsonIndex>>,
     pub tsconfig: OnceLock<Option<Arc<TsConfig>>>,
 }
 

src/cache/cached_path.rs

@@ -14,15 +14,20 @@ pub use serde::*;
 #[cfg(target_endian = "little")]
 pub use simd::*;
 
-use std::{fmt, path::PathBuf};
+use std::{fmt, path::Path};
 
 use crate::JSONError;
 
 /// Check if JSON content is empty or contains only whitespace
-fn check_if_empty(json_bytes: &[u8], path: PathBuf) -> Result<(), JSONError> {
+fn check_if_empty(json_bytes: &[u8], path: &Path) -> Result<(), JSONError> {
     // Check if content is empty or whitespace-only
     if json_bytes.iter().all(|&b| b.is_ascii_whitespace()) {
-        return Err(JSONError { path, message: "File is empty".to_string(), line: 0, column: 0 });
+        return Err(JSONError {
+            path: path.to_path_buf(),
+            message: "File is empty".to_string(),
+            line: 0,
+            column: 0,
+        });
     }
     Ok(())
 }

src/package_json/mod.rs

@@ -230,7 +230,7 @@ impl PackageJson {
         let json_bytes = if json.starts_with(b"\xEF\xBB\xBF") { &json[3..] } else { &json[..] };
 
         // Check if empty after BOM stripping
-        super::check_if_empty(json_bytes, path.clone())?;
+        super::check_if_empty(json_bytes, &path)?;
 
         // Parse JSON directly from bytes
         let value = serde_json::from_slice::<Value>(json_bytes).map_err(|error| JSONError {

src/package_json/serde.rs

@@ -269,7 +269,7 @@ impl PackageJson {
         }
 
         // Check if empty after BOM stripping
-        super::check_if_empty(&json_bytes, path.clone())?;
+        super::check_if_empty(&json_bytes, &path)?;
 
         // Create the self-cell with the JSON bytes and parsed BorrowedValue
         let cell = PackageJsonCell::try_new(MutBorrow::new(json_bytes), |bytes| {

src/package_json/simd.rs

@@ -1,7 +1,7 @@
 //! https://github.com/webpack/enhanced-resolve/blob/main/test/dependencies.test.js
 
 #[cfg(not(target_os = "windows"))] // MemoryFS's path separator is always `/` so the test will not pass in windows.
-mod windows {
+mod test {
     use std::path::PathBuf;
 
     use super::super::memory_fs::MemoryFS;
@@ -18,17 +18,6 @@ mod windows {
 
     #[test]
     fn test() {
-        let file_system = file_system();
-
-        let resolver = ResolverGeneric::new_with_file_system(
-            file_system,
-            ResolveOptions {
-                extensions: vec![".json".into(), ".js".into()],
-                modules: vec!["/modules".into(), "node_modules".into()],
-                ..ResolveOptions::default()
-            },
-        );
-
         let data = [
             (
                 "middle module request",
@@ -92,6 +81,15 @@ mod windows {
         ];
 
         for (name, context, request, result, file_dependencies, missing_dependencies) in data {
+            let file_system = file_system();
+            let resolver = ResolverGeneric::new_with_file_system(
+                file_system,
+                ResolveOptions {
+                    extensions: vec![".json".into(), ".js".into()],
+                    modules: vec!["/modules".into(), "node_modules".into()],
+                    ..ResolveOptions::default()
+                },
+            );
             let mut ctx = ResolveContext::default();
             let path = PathBuf::from(context);
             let resolved_path =