feat(incremental): named chunk ids (#8652)

Author: ahabhgk

crates/rspack_binding_values/src/chunk.rs

@@ -40,17 +40,22 @@ impl JsChunk {
 
   #[napi(getter)]
   pub fn id(&self) -> napi::Result<Either<&str, ()>> {
-    let (_, chunk) = self.as_ref()?;
-    Ok(match chunk.id() {
-      Some(id) => Either::A(id),
+    let (compilation, chunk) = self.as_ref()?;
+    Ok(match chunk.id(&compilation.chunk_ids) {
+      Some(id) => Either::A(id.as_str()),
       None => Either::B(()),
     })
   }
 
   #[napi(getter)]
   pub fn ids(&self) -> napi::Result<Vec<&str>> {
-    let (_, chunk) = self.as_ref()?;
-    Ok(chunk.id().map(|id| vec![id]).unwrap_or_default())
+    let (compilation, chunk) = self.as_ref()?;
+    Ok(
+      chunk
+        .id(&compilation.chunk_ids)
+        .map(|id| vec![id.as_str()])
+        .unwrap_or_default(),
+    )
   }
 
   #[napi(getter)]

crates/rspack_core/src/chunk.rs

@@ -9,6 +9,7 @@ use rspack_error::Diagnostic;
 use rspack_hash::{RspackHash, RspackHashDigest};
 use rustc_hash::{FxHashMap as HashMap, FxHashSet as HashSet, FxHasher};
 
+use crate::chunk_graph_chunk::ChunkId;
 use crate::{
   compare_chunk_group, merge_runtime, sort_group_by_index, ChunkGraph, ChunkGroupOrderKey,
   RenderManifestEntry,
@@ -58,7 +59,6 @@ pub struct Chunk {
   // - The name of chunks create by dynamic import is `None` unless users use
   // magic comment like `import(/* webpackChunkName: "someChunk" * / './someModule.js')` to specify it.
   name: Option<String>,
-  id: Option<String>,
   id_name_hints: HashSet<String>,
   filename_template: Option<Filename>,
   css_filename_template: Option<Filename>,
@@ -108,12 +108,23 @@ impl Chunk {
     self.css_filename_template = filename_template;
   }
 
-  pub fn id(&self) -> Option<&str> {
-    self.id.as_deref()
+  pub fn id<'a>(&self, chunk_ids: &'a UkeyMap<ChunkUkey, ChunkId>) -> Option<&'a ChunkId> {
+    ChunkGraph::get_chunk_id(chunk_ids, &self.ukey)
   }
 
-  pub fn set_id(&mut self, id: Option<String>) {
-    self.id = id;
+  pub fn expect_id<'a>(&self, chunk_ids: &'a UkeyMap<ChunkUkey, ChunkId>) -> &'a ChunkId {
+    self
+      .id(chunk_ids)
+      .expect("Should set id before calling expect_id")
+  }
+
+  pub fn set_id(
+    &self,
+    chunk_ids: &mut UkeyMap<ChunkUkey, ChunkId>,
+    id: impl Into<ChunkId>,
+  ) -> bool {
+    let id = id.into();
+    ChunkGraph::set_chunk_id(chunk_ids, self.ukey, id)
   }
 
   pub fn prevent_integration(&self) -> bool {
@@ -263,7 +274,6 @@ impl Chunk {
       filename_template: None,
       css_filename_template: None,
       ukey: ChunkUkey::new(),
-      id: None,
       id_name_hints: Default::default(),
       prevent_integration: false,
       files: Default::default(),
@@ -578,18 +588,14 @@ impl Chunk {
     chunks
   }
 
-  pub fn expect_id(&self) -> &str {
-    self
-      .id
-      .as_ref()
-      .expect("Should set id before calling expect_id")
-  }
-
-  pub fn name_for_filename_template(&self) -> Option<&str> {
+  pub fn name_for_filename_template<'a>(
+    &'a self,
+    chunk_ids: &'a UkeyMap<ChunkUkey, ChunkId>,
+  ) -> Option<&'a str> {
     if self.name.is_some() {
       self.name.as_deref()
     } else {
-      self.id.as_deref()
+      self.id(chunk_ids).map(|id| id.as_str())
     }
   }
 
@@ -602,7 +608,7 @@ impl Chunk {
   }
 
   pub fn update_hash(&self, hasher: &mut RspackHash, compilation: &Compilation) {
-    self.id.hash(hasher);
+    self.id(&compilation.chunk_ids).hash(hasher);
     for module in compilation
       .chunk_graph
       .get_ordered_chunk_modules(&self.ukey, &compilation.get_module_graph())
@@ -716,16 +722,20 @@ impl Chunk {
     &self,
     order: &ChunkGroupOrderKey,
     compilation: &Compilation,
-  ) -> Option<Vec<String>> {
+  ) -> Option<Vec<ChunkId>> {
     self
       .get_children_of_type_in_order(order, compilation, true)
       .map(|order_children| {
         order_children
           .iter()
           .flat_map(|(_, child_chunks)| {
-            child_chunks
-              .iter()
-              .filter_map(|chunk_ukey| compilation.chunk_by_ukey.expect_get(chunk_ukey).id.clone())
+            child_chunks.iter().filter_map(|chunk_ukey| {
+              compilation
+                .chunk_by_ukey
+                .expect_get(chunk_ukey)
+                .id(&compilation.chunk_ids)
+                .cloned()
+            })
           })
           .collect_vec()
       })
@@ -735,21 +745,22 @@ impl Chunk {
     &self,
     include_direct_children: bool,
     compilation: &Compilation,
-  ) -> HashMap<ChunkGroupOrderKey, IndexMap<String, Vec<String>, BuildHasherDefault<FxHasher>>> {
+  ) -> HashMap<ChunkGroupOrderKey, IndexMap<ChunkId, Vec<ChunkId>, BuildHasherDefault<FxHasher>>>
+  {
     let mut result = HashMap::default();
 
     fn add_child_ids_by_orders_to_map(
       chunk_ukey: &ChunkUkey,
       order: &ChunkGroupOrderKey,
       result: &mut HashMap<
         ChunkGroupOrderKey,
-        IndexMap<String, Vec<String>, BuildHasherDefault<FxHasher>>,
+        IndexMap<ChunkId, Vec<ChunkId>, BuildHasherDefault<FxHasher>>,
       >,
       compilation: &Compilation,
     ) {
       let chunk = compilation.chunk_by_ukey.expect_get(chunk_ukey);
       if let (Some(chunk_id), Some(child_chunk_ids)) = (
-        chunk.id.clone(),
+        chunk.id(&compilation.chunk_ids).cloned(),
         chunk.get_child_ids_by_order(order, compilation),
       ) {
         result

crates/rspack_core/src/chunk_graph/chunk_graph_chunk.rs

@@ -1,10 +1,18 @@
 //!  There are methods whose verb is `ChunkGraphChunk`
 
+use std::borrow::Borrow;
+use std::fmt;
+use std::sync::Arc;
+
 use hashlink::LinkedHashMap;
 use indexmap::IndexSet;
 use itertools::Itertools;
-use rspack_collections::{DatabaseItem, IdentifierLinkedMap, IdentifierMap, IdentifierSet};
+use rspack_cacheable::cacheable;
+use rspack_collections::{
+  DatabaseItem, IdentifierLinkedMap, IdentifierMap, IdentifierSet, UkeyMap,
+};
 use rustc_hash::{FxHashMap as HashMap, FxHashSet};
+use serde::{Serialize, Serializer};
 
 use crate::{
   find_graph_roots, merge_runtime, BoxModule, Chunk, ChunkByUkey, ChunkGraphModule,
@@ -21,6 +29,51 @@ pub struct ChunkSizeOptions {
   pub entry_chunk_multiplicator: Option<f64>,
 }
 
+#[cacheable]
+#[derive(Debug, Clone, PartialEq, Eq, Hash, PartialOrd, Ord)]
+pub struct ChunkId {
+  inner: Arc<str>,
+}
+
+impl From<String> for ChunkId {
+  fn from(s: String) -> Self {
+    Self { inner: s.into() }
+  }
+}
+
+impl From<&str> for ChunkId {
+  fn from(s: &str) -> Self {
+    Self { inner: s.into() }
+  }
+}
+
+impl fmt::Display for ChunkId {
+  fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+    write!(f, "{}", self.as_str())
+  }
+}
+
+impl Serialize for ChunkId {
+  fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
+  where
+    S: Serializer,
+  {
+    serializer.serialize_str(self.as_str())
+  }
+}
+
+impl Borrow<str> for ChunkId {
+  fn borrow(&self) -> &str {
+    self.as_str()
+  }
+}
+
+impl ChunkId {
+  pub fn as_str(&self) -> &str {
+    &self.inner
+  }
+}
+
 #[derive(Debug, Clone, Default)]
 pub struct ChunkGraphChunk {
   /// URI of modules => ChunkGroupUkey
@@ -550,7 +603,10 @@ impl ChunkGraph {
       .iter()
     {
       let chunk = compilation.chunk_by_ukey.expect_get(c);
-      map.insert(chunk.expect_id().to_string(), filter(c, compilation));
+      map.insert(
+        chunk.expect_id(&compilation.chunk_ids).to_string(),
+        filter(c, compilation),
+      );
     }
 
     map
@@ -912,4 +968,23 @@ impl ChunkGraph {
       })
       .unwrap_or(module.source_types().iter().copied().collect())
   }
+
+  pub fn get_chunk_id<'a>(
+    chunk_ids: &'a UkeyMap<ChunkUkey, ChunkId>,
+    chunk_ukey: &ChunkUkey,
+  ) -> Option<&'a ChunkId> {
+    chunk_ids.get(chunk_ukey)
+  }
+
+  pub fn set_chunk_id(
+    chunk_ids: &mut UkeyMap<ChunkUkey, ChunkId>,
+    chunk_ukey: ChunkUkey,
+    id: ChunkId,
+  ) -> bool {
+    if let Some(old_id) = chunk_ids.insert(chunk_ukey, id.clone()) {
+      old_id != id
+    } else {
+      true
+    }
+  }
 }

crates/rspack_core/src/chunk_group.rs

@@ -265,7 +265,7 @@ impl ChunkGroup {
         compilation
           .chunk_by_ukey
           .get(chunk)
-          .and_then(|item| item.id())
+          .and_then(|item| item.id(&compilation.chunk_ids))
       })
       .join("+")
   }

crates/rspack_core/src/compiler/compilation.rs

@@ -34,6 +34,7 @@ use crate::{
   cache::Cache,
   cgm_hash_results::CgmHashResults,
   cgm_runtime_requirement_results::CgmRuntimeRequirementsResults,
+  chunk_graph_chunk::ChunkId,
   chunk_graph_module::ModuleId,
   get_runtime_key,
   incremental::{Incremental, IncrementalPasses, Mutation},
@@ -181,6 +182,8 @@ pub struct Compilation {
   pub dependencies_diagnostics: IdentifierMap<Vec<Diagnostic>>,
   // artifact for module_ids
   pub module_ids: IdentifierMap<ModuleId>,
+  // artifact for chunk_ids
+  pub chunk_ids: UkeyMap<ChunkUkey, ChunkId>,
   // artifact for code_generation
   pub code_generation_results: CodeGenerationResults,
   // artifact for create_module_hashes
@@ -201,7 +204,6 @@ pub struct Compilation {
   pub incremental: Incremental,
 
   pub hash: Option<RspackHashDigest>,
-  pub used_chunk_ids: HashSet<String>,
 
   pub file_dependencies: IndexSet<ArcPath, BuildHasherDefault<FxHasher>>,
   pub context_dependencies: IndexSet<ArcPath, BuildHasherDefault<FxHasher>>,
@@ -289,6 +291,7 @@ impl Compilation {
       async_modules: Default::default(),
       dependencies_diagnostics: Default::default(),
       module_ids: Default::default(),
+      chunk_ids: Default::default(),
       code_generation_results: Default::default(),
       cgm_hash_results: Default::default(),
       cgm_runtime_requirements_results: Default::default(),
@@ -301,8 +304,8 @@ impl Compilation {
       old_cache,
       incremental,
       code_splitting_cache: Default::default(),
+
       hash: None,
-      used_chunk_ids: Default::default(),
 
       file_dependencies: Default::default(),
       context_dependencies: Default::default(),
@@ -1502,6 +1505,7 @@ impl Compilation {
       entrypoint_ukey: &ChunkGroupUkey,
       chunk_group_by_ukey: &ChunkGroupByUkey,
       chunk_by_ukey: &ChunkByUkey,
+      chunk_ids: &UkeyMap<ChunkUkey, ChunkId>,
       chunk_graph: &mut ChunkGraph,
     ) {
       let entrypoint = chunk_group_by_ukey.expect_get(entrypoint_ukey);
@@ -1517,14 +1521,15 @@ impl Compilation {
         runtime,
         chunk_by_ukey.get(&entrypoint.get_runtime_chunk(chunk_group_by_ukey)),
       ) {
-        chunk_graph.set_runtime_id(runtime, chunk.id().map(ToOwned::to_owned));
+        chunk_graph.set_runtime_id(runtime, chunk.id(chunk_ids).map(|id| id.to_string()));
       }
     }
     for i in self.entrypoints.iter() {
       process_entrypoint(
         i.1,
         &self.chunk_group_by_ukey,
         &self.chunk_by_ukey,
+        &self.chunk_ids,
         &mut self.chunk_graph,
       )
     }
@@ -1533,6 +1538,7 @@ impl Compilation {
         i,
         &self.chunk_group_by_ukey,
         &self.chunk_by_ukey,
+        &self.chunk_ids,
         &mut self.chunk_graph,
       )
     }
@@ -1862,11 +1868,16 @@ impl Compilation {
         .filter(|(_, (_, remaining))| *remaining != 0)
         .map(|(chunk_ukey, _)| self.chunk_by_ukey.expect_get(chunk_ukey))
         .collect();
-      circular.sort_unstable_by(|a, b| a.id().cmp(&b.id()));
+      circular.sort_unstable_by(|a, b| a.id(&self.chunk_ids).cmp(&b.id(&self.chunk_ids)));
       runtime_chunks.extend(circular.iter().map(|chunk| chunk.ukey()));
       let circular_names = circular
         .iter()
-        .map(|chunk| chunk.name().or(chunk.id()).unwrap_or("no id chunk"))
+        .map(|chunk| {
+          chunk
+            .name()
+            .or(chunk.id(&self.chunk_ids).map(|id| id.as_str()))
+            .unwrap_or("no id chunk")
+        })
         .join(", ");
       self.push_diagnostic(diagnostic!(severity = Severity::Warn, "Circular dependency between chunks with runtime ({})\nThis prevents using hashes of each other and should be avoided.", circular_names).boxed().into());
     }