refactor(cache): align new cache snapshots with webpack (#15311)

Author: hardfist

crates/rspack_core/src/new_cache/file_cache_strategy.rs

@@ -8,7 +8,7 @@ use super::{
   CacheKey, Etag,
   cache_value::{CacheEntry, CacheValueDecoder, CacheValueEncoder, ErasedCacheValue},
   db::{Database, DatabaseFamily, DatabaseValue, DatabaseWrite},
-  snapshot::{BuildDeps, Snapshot},
+  snapshot::{BuildDeps, FileSystemInfo},
   validator::{CacheValidator, CacheValidatorResult},
 };
 use crate::cache::persistent::codec::CacheCodec;
@@ -53,7 +53,7 @@ impl FileCacheStrategy {
     rspack_pkg_version: String,
     cache_version: String,
     codec: Arc<CacheCodec>,
-    snapshot: Snapshot,
+    file_system_info: FileSystemInfo,
     build_deps: BuildDeps,
   ) -> Result<Self> {
     let (base_path, database_path) = database_paths;
@@ -63,7 +63,7 @@ impl FileCacheStrategy {
         rspack_pkg_version,
         cache_version,
         codec.clone(),
-        snapshot,
+        file_system_info,
         build_deps,
       ),
       codec,

crates/rspack_core/src/new_cache/mod.rs

@@ -22,7 +22,7 @@ pub use idle_file_cache::IdleFileCache;
 pub use memory_cache::{MemoryCache, MemoryCacheGetResult};
 use rspack_fs::ReadableFileSystem;
 
-use self::snapshot::{BuildDeps, Snapshot};
+use self::snapshot::{BuildDeps, FileSystemInfo};
 use crate::{
   CompilationLogger, CompilationLogging, CompilerOptions, cache::persistent::codec::CacheCodec,
 };
@@ -53,7 +53,11 @@ pub fn create_cache(
     None
   };
   let codec = Arc::new(CacheCodec::new(project_root));
-  let snapshot = Snapshot::new(options.snapshot.clone(), input_filesystem.clone());
+  let file_system_info = FileSystemInfo::new(
+    input_filesystem.clone(),
+    options.snapshot.clone(),
+    compiler_options.output.hash_function,
+  );
   let build_deps = BuildDeps::new(
     &options.build_dependencies,
     input_filesystem,
@@ -73,7 +77,7 @@ pub fn create_cache(
     rspack_workspace::rspack_pkg_version!().to_string(),
     options.version.clone(),
     codec,
-    snapshot,
+    file_system_info,
     build_deps,
   ) {
     Ok(strategy) => strategy,

crates/rspack_core/src/new_cache/snapshot/build_deps.rs

@@ -1,9 +1,10 @@
 use std::{collections::VecDeque, path::PathBuf, sync::Arc};
 
+use rspack_error::Result;
 use rspack_fs::ReadableFileSystem;
 use rspack_paths::{AssertUtf8, InternedPath, InternedPathSet};
 
-use super::{Snapshot, SnapshotEntry};
+use super::{FileSystemInfo, Snapshot, SnapshotValidationResult};
 use crate::{
   CompilationLogger,
   cache::persistent::build_dependencies::{Helper, is_node_package_path},
@@ -22,6 +23,14 @@ pub enum BuildDepsValidationResult {
   },
 }
 
+#[derive(Debug, Default)]
+pub struct ResolvedBuildDependencies {
+  pub(crate) dependencies: InternedPathSet,
+  pub(crate) files: InternedPathSet,
+  pub(crate) contexts: InternedPathSet,
+  pub(crate) missing: InternedPathSet,
+}
+
 /// Build dependencies manager.
 #[derive(Debug)]
 pub struct BuildDeps {
@@ -51,21 +60,35 @@ impl BuildDeps {
   ///
   /// For performance reasons, recursive searches stop at dependencies in
   /// `node_modules`.
+  ///
+  /// See webpack's build dependency resolution implementation:
+  /// https://github.com/webpack/webpack/blob/ce97d583e1cd8f3e47b70737de72e91b567a8497/lib/FileSystemInfo.js#L1873-L2523
   pub async fn resolve_dependencies(
     &mut self,
     current: &InternedPathSet,
     paths: impl Iterator<Item = InternedPath>,
-  ) -> InternedPathSet {
+  ) -> ResolvedBuildDependencies {
     let mut helper = Helper::new(self.fs.clone(), self.logger.clone());
-    let mut added = InternedPathSet::default();
+    let mut resolved = ResolvedBuildDependencies::default();
     let mut queue = VecDeque::new();
     queue.extend(self.pending.iter().cloned());
     queue.extend(paths);
 
     while let Some(dependency) = queue.pop_front() {
-      if current.contains(&dependency) || !added.insert(dependency.clone()) {
+      if current.contains(&dependency) || !resolved.dependencies.insert(dependency.clone()) {
         continue;
       }
+      match self.fs.metadata(dependency.assert_utf8()).await {
+        Ok(metadata) if metadata.is_directory => {
+          resolved.contexts.insert(dependency.clone());
+        }
+        Ok(_) => {
+          resolved.files.insert(dependency.clone());
+        }
+        Err(_) => {
+          resolved.missing.insert(dependency.clone());
+        }
+      }
       if is_node_package_path(&dependency) {
         continue;
       }
@@ -79,25 +102,44 @@ impl BuildDeps {
     }
 
     self.pending.clear();
-    added
+    resolved
   }
 
   /// Validate build dependencies.
   ///
   /// If any build dependency changed, this method returns an invalid result.
   pub async fn validate_snapshot(
     &self,
+    file_system_info: &FileSystemInfo,
     snapshot: &Snapshot,
-    entries: &[SnapshotEntry],
+    dependencies: &InternedPathSet,
     tracked_files: usize,
-  ) -> BuildDepsValidationResult {
-    let (modified_files, removed_files) = snapshot.calc_modified_paths(entries).await;
-    if !modified_files.is_empty() || !removed_files.is_empty() {
-      return BuildDepsValidationResult::Invalid {
-        modified_files,
+  ) -> Result<BuildDepsValidationResult> {
+    let validation = file_system_info.check_snapshot_valid(snapshot).await?;
+    let pending = self
+      .pending
+      .iter()
+      .filter(|path| !dependencies.contains(*path))
+      .cloned()
+      .collect::<InternedPathSet>();
+    match validation {
+      SnapshotValidationResult::Valid if pending.is_empty() => {
+        Ok(BuildDepsValidationResult::Valid { tracked_files })
+      }
+      SnapshotValidationResult::Valid => Ok(BuildDepsValidationResult::Invalid {
+        modified_files: pending,
+        removed_files: Default::default(),
+      }),
+      SnapshotValidationResult::Invalid {
+        mut modified_files,
         removed_files,
-      };
+      } => {
+        modified_files.extend(pending);
+        Ok(BuildDepsValidationResult::Invalid {
+          modified_files,
+          removed_files,
+        })
+      }
     }
-    BuildDepsValidationResult::Valid { tracked_files }
   }
 }