perf: improve data structures of bundle splitting (#11506)

Author: LingyuCoder

Cargo.lock

@@ -83,6 +83,10 @@ impl Identifier {
   pub fn to_string(&self) -> String {
     self.0.to_owned()
   }
+
+  pub fn precomputed_hash(&self) -> u64 {
+    self.0.precomputed_hash()
+  }
 }
 
 impl fmt::Display for Identifier {

crates/rspack_collections/src/identifier.rs

@@ -3,7 +3,7 @@ use std::collections::hash_map::Entry;
 use rspack_util::atom::Atom;
 
 use super::{ExportInfoData, ExportInfoTargetValue, Inlinable, UsageFilterFnTy, UsageState};
-use crate::{DependencyId, ExportsInfo, Nullable, RuntimeSpec};
+use crate::{DependencyId, Nullable, RuntimeSpec};
 
 impl ExportInfoData {
   pub fn reset_provide_info(&mut self) {
@@ -212,17 +212,12 @@ impl ExportInfoData {
     changed
   }
 
-  pub fn set_has_use_info(&mut self, nested_exports_info: &mut Vec<ExportsInfo>) {
+  pub fn set_has_use_info(&mut self) {
     if !self.has_use_in_runtime_info() {
       self.set_has_use_in_runtime_info(true);
     }
     if self.can_mangle_use().is_none() {
       self.set_can_mangle_use(Some(true));
     }
-    if self.exports_info_owned()
-      && let Some(exports_info) = self.exports_info()
-    {
-      nested_exports_info.push(exports_info);
-    }
   }
 }

crates/rspack_core/src/exports/export_info_setter.rs

@@ -163,30 +163,6 @@ impl ExportsInfo {
     new_info_id
   }
 
-  pub fn set_has_use_info(&self, mg: &mut ModuleGraph) {
-    let mut nested_exports_info = vec![];
-    let exports_info = self.as_data_mut(mg);
-    for export_info in exports_info.exports_mut().values_mut() {
-      export_info.set_has_use_info(&mut nested_exports_info);
-    }
-    exports_info
-      .side_effects_only_info_mut()
-      .set_has_use_info(&mut nested_exports_info);
-    if let Some(redirect) = exports_info.redirect_to() {
-      redirect.set_has_use_info(mg);
-    } else {
-      let other_exports_info = exports_info.other_exports_info_mut();
-      other_exports_info.set_has_use_info(&mut nested_exports_info);
-      if other_exports_info.can_mangle_use().is_none() {
-        other_exports_info.set_can_mangle_use(Some(true));
-      }
-    }
-
-    for nested_exports_info in nested_exports_info {
-      nested_exports_info.set_has_use_info(mg);
-    }
-  }
-
   pub fn set_used_without_info(&self, mg: &mut ModuleGraph, runtime: Option<&RuntimeSpec>) -> bool {
     let mut changed = false;
     let exports_info = self.as_data_mut(mg);

crates/rspack_core/src/exports/exports_info.rs

@@ -168,7 +168,7 @@ pub enum ExportProvided {
   Unknown,
 }
 
-#[derive(Debug, Hash, PartialEq, Eq, Default)]
+#[derive(Debug, Hash, PartialEq, Eq, Default, Clone)]
 pub struct UsageKey(pub Vec<Either<Box<UsageKey>, UsageState>>);
 
 impl UsageKey {

crates/rspack_core/src/exports/utils.rs

@@ -0,0 +1,98 @@
+use std::collections::hash_map::Entry;
+
+use rayon::prelude::*;
+
+use crate::{
+  ExportsInfo, ExportsInfoData, ExportsInfoGetter, ModuleGraph, ModuleIdentifier,
+  PrefetchExportsInfoMode, PrefetchedExportsInfoWrapper,
+};
+
+impl<'a> ModuleGraph<'a> {
+  pub fn get_exports_info(&self, module_identifier: &ModuleIdentifier) -> ExportsInfo {
+    self
+      .module_graph_module_by_identifier(module_identifier)
+      .expect("should have mgm")
+      .exports
+  }
+
+  pub fn get_prefetched_exports_info_optional<'b>(
+    &'b self,
+    module_identifier: &ModuleIdentifier,
+    mode: PrefetchExportsInfoMode<'b>,
+  ) -> Option<PrefetchedExportsInfoWrapper<'b>> {
+    self
+      .module_graph_module_by_identifier(module_identifier)
+      .map(move |mgm| ExportsInfoGetter::prefetch(&mgm.exports, self, mode))
+  }
+
+  pub fn get_prefetched_exports_info<'b>(
+    &'b self,
+    module_identifier: &ModuleIdentifier,
+    mode: PrefetchExportsInfoMode<'b>,
+  ) -> PrefetchedExportsInfoWrapper<'b> {
+    let exports_info = self.get_exports_info(module_identifier);
+    ExportsInfoGetter::prefetch(&exports_info, self, mode)
+  }
+
+  pub fn get_exports_info_by_id(&self, id: &ExportsInfo) -> &ExportsInfoData {
+    self
+      .try_get_exports_info_by_id(id)
+      .expect("should have exports info")
+  }
+
+  pub fn try_get_exports_info_by_id(&self, id: &ExportsInfo) -> Option<&ExportsInfoData> {
+    self.loop_partials(|p| p.exports_info_map.get(id))
+  }
+
+  pub fn get_exports_info_mut_by_id(&mut self, id: &ExportsInfo) -> &mut ExportsInfoData {
+    self
+      .loop_partials_mut(
+        |p| p.exports_info_map.contains_key(id),
+        |p, search_result| {
+          p.exports_info_map.insert(*id, search_result);
+        },
+        |p| p.exports_info_map.get(id).cloned(),
+        |p| p.exports_info_map.get_mut(id),
+      )
+      .expect("should have exports info")
+  }
+
+  pub fn set_exports_info(&mut self, id: ExportsInfo, info: ExportsInfoData) {
+    let Some(active_partial) = &mut self.active else {
+      panic!("should have active partial");
+    };
+    active_partial.exports_info_map.insert(id, info);
+  }
+
+  pub fn active_all_exports_info(&mut self) {
+    let active_partial = self.active.as_mut().expect("should have active partial");
+    for partial in self.partials.iter().rev().flatten() {
+      for (id, exports_info) in partial.exports_info_map.iter() {
+        match active_partial.exports_info_map.entry(*id) {
+          Entry::Occupied(_) => {}
+          Entry::Vacant(entry) => {
+            entry.insert(exports_info.clone());
+          }
+        }
+      }
+    }
+  }
+
+  pub fn reset_all_exports_info_used(&mut self) {
+    let exports_info_map = &mut self
+      .active
+      .as_mut()
+      .expect("should have active partial")
+      .exports_info_map;
+
+    exports_info_map
+      .par_iter_mut()
+      .for_each(|(_, exports_info)| {
+        for export_info in exports_info.exports_mut().values_mut() {
+          export_info.set_has_use_info();
+        }
+        exports_info.side_effects_only_info_mut().set_has_use_info();
+        exports_info.other_exports_info_mut().set_has_use_info();
+      });
+  }
+}

crates/rspack_core/src/module_graph/exports_info.rs

@@ -9,13 +9,13 @@ use swc_core::ecma::atoms::Atom;
 
 use crate::{
   AsyncDependenciesBlock, AsyncDependenciesBlockIdentifier, Compilation, DependenciesBlock,
-  Dependency, ExportInfo, ExportName, ExportProvided, ExportsInfoGetter, ModuleGraphCacheArtifact,
-  PrefetchExportsInfoMode, PrefetchedExportsInfoWrapper, RuntimeSpec,
+  Dependency, ExportInfo, ExportName, ModuleGraphCacheArtifact, RuntimeSpec,
 };
 mod module;
 pub use module::*;
 mod connection;
 pub use connection::*;
+mod exports_info;
 
 use crate::{
   BoxDependency, BoxModule, DependencyCondition, DependencyId, ExportsInfo, ExportsInfoData,
@@ -1026,62 +1026,6 @@ impl<'a> ModuleGraph<'a> {
     new_mgm.add_incoming_connection(*dep_id);
   }
 
-  pub fn get_exports_info(&self, module_identifier: &ModuleIdentifier) -> ExportsInfo {
-    self
-      .module_graph_module_by_identifier(module_identifier)
-      .expect("should have mgm")
-      .exports
-  }
-
-  pub fn get_prefetched_exports_info_optional<'b>(
-    &'b self,
-    module_identifier: &ModuleIdentifier,
-    mode: PrefetchExportsInfoMode<'b>,
-  ) -> Option<PrefetchedExportsInfoWrapper<'b>> {
-    self
-      .module_graph_module_by_identifier(module_identifier)
-      .map(move |mgm| ExportsInfoGetter::prefetch(&mgm.exports, self, mode))
-  }
-
-  pub fn get_prefetched_exports_info<'b>(
-    &'b self,
-    module_identifier: &ModuleIdentifier,
-    mode: PrefetchExportsInfoMode<'b>,
-  ) -> PrefetchedExportsInfoWrapper<'b> {
-    let exports_info = self.get_exports_info(module_identifier);
-    ExportsInfoGetter::prefetch(&exports_info, self, mode)
-  }
-
-  pub fn get_exports_info_by_id(&self, id: &ExportsInfo) -> &ExportsInfoData {
-    self
-      .try_get_exports_info_by_id(id)
-      .expect("should have exports info")
-  }
-
-  pub fn try_get_exports_info_by_id(&self, id: &ExportsInfo) -> Option<&ExportsInfoData> {
-    self.loop_partials(|p| p.exports_info_map.get(id))
-  }
-
-  pub fn get_exports_info_mut_by_id(&mut self, id: &ExportsInfo) -> &mut ExportsInfoData {
-    self
-      .loop_partials_mut(
-        |p| p.exports_info_map.contains_key(id),
-        |p, search_result| {
-          p.exports_info_map.insert(*id, search_result);
-        },
-        |p| p.exports_info_map.get(id).cloned(),
-        |p| p.exports_info_map.get_mut(id),
-      )
-      .expect("should have exports info")
-  }
-
-  pub fn set_exports_info(&mut self, id: ExportsInfo, info: ExportsInfoData) {
-    let Some(active_partial) = &mut self.active else {
-      panic!("should have active partial");
-    };
-    active_partial.exports_info_map.insert(id, info);
-  }
-
   pub fn get_optimization_bailout_mut(&mut self, id: &ModuleIdentifier) -> &mut Vec<String> {
     let mgm = self
       .module_graph_module_by_identifier_mut(id)
@@ -1108,22 +1052,6 @@ impl<'a> ModuleGraph<'a> {
     condition.get_connection_state(connection, runtime, self, module_graph_cache)
   }
 
-  // returns: Option<bool>
-  //   - None: it's unknown
-  //   - Some(true): provided
-  //   - Some(false): not provided
-  pub fn is_export_provided(
-    &self,
-    id: &ModuleIdentifier,
-    names: &[Atom],
-  ) -> Option<ExportProvided> {
-    self.module_graph_module_by_identifier(id).and_then(|mgm| {
-      let exports_info =
-        ExportsInfoGetter::prefetch(&mgm.exports, self, PrefetchExportsInfoMode::Nested(names));
-      exports_info.is_export_provided(names)
-    })
-  }
-
   // todo remove it after module_graph_partial remove all of dependency_id_to_*
   pub fn cache_recovery_connection(&mut self, connection: ModuleGraphConnection) {
     let condition = self

crates/rspack_core/src/module_graph/mod.rs

@@ -48,16 +48,16 @@ impl<'a> FlagDependencyUsagePluginProxy<'a> {
 
   fn apply(&mut self) {
     let mut module_graph = self.compilation.get_module_graph_mut();
+    module_graph.active_all_exports_info();
+    module_graph.reset_all_exports_info_used();
+
     for mgm in module_graph.module_graph_modules().values() {
       self
         .exports_info_module_map
         .insert(mgm.exports, mgm.module_identifier);
     }
     let mut q = Queue::new();
     let mg = &mut module_graph;
-    for exports_info in self.exports_info_module_map.keys() {
-      exports_info.set_has_use_info(mg);
-    }
 
     // SAFETY: we can make sure that entries will not be used other place at the same time,
     // this take is aiming to avoid use self ref and mut ref at the same time;

crates/rspack_plugin_javascript/src/plugin/flag_dependency_usage_plugin.rs

@@ -12,6 +12,7 @@ version.workspace = true
 dashmap            = { workspace = true }
 derive_more        = { workspace = true, features = ["debug"] }
 futures            = { workspace = true }
+itertools          = { workspace = true }
 rayon              = { workspace = true }
 regex              = { workspace = true }
 rspack_collections = { workspace = true }