perf(compile): read embedded files as static references when UTF-8 an…

…d reading as strings (#27033)

cli/args/deno_json.rs

@@ -22,6 +22,8 @@ impl<'a> deno_config::fs::DenoConfigFs for DenoConfigFsAdapter<'a> {
     self
       .0
       .read_text_file_lossy_sync(path, None)
+      // todo(https://github.com/denoland/deno_config/pull/140): avoid clone
+      .map(|s| s.into_owned())
       .map_err(|err| err.into_io_error())
   }
 

cli/cache/mod.rs

@@ -116,6 +116,8 @@ impl<'a> deno_cache_dir::DenoCacheEnv for DenoCacheEnvFsAdapter<'a> {
     self
       .0
       .read_file_sync(path, None)
+      // todo(https://github.com/denoland/deno_cache_dir/pull/66): avoid clone
+      .map(|bytes| bytes.into_owned())
       .map_err(|err| err.into_io_error())
   }
 

cli/module_loader.rs

@@ -1060,7 +1060,10 @@ impl<TGraphContainer: ModuleGraphContainer> NodeRequireLoader
     self.npm_resolver.ensure_read_permission(permissions, path)
   }
 
-  fn load_text_file_lossy(&self, path: &Path) -> Result<String, AnyError> {
+  fn load_text_file_lossy(
+    &self,
+    path: &Path,
+  ) -> Result<Cow<'static, str>, AnyError> {
     // todo(dsherret): use the preloaded module from the graph if available?
     let media_type = MediaType::from_path(path);
     let text = self.fs.read_text_file_lossy_sync(path, None)?;
@@ -1075,15 +1078,18 @@ impl<TGraphContainer: ModuleGraphContainer> NodeRequireLoader
           .into(),
         );
       }
-      self.emitter.emit_parsed_source_sync(
-        &specifier,
-        media_type,
-        // this is probably not super accurate due to require esm, but probably ok.
-        // If we find this causes a lot of churn in the emit cache then we should
-        // investigate how we can make this better
-        ModuleKind::Cjs,
-        &text.into(),
-      )
+      self
+        .emitter
+        .emit_parsed_source_sync(
+          &specifier,
+          media_type,
+          // this is probably not super accurate due to require esm, but probably ok.
+          // If we find this causes a lot of churn in the emit cache then we should
+          // investigate how we can make this better
+          ModuleKind::Cjs,
+          &text.into(),
+        )
+        .map(Cow::Owned)
     } else {
       Ok(text)
     }

cli/node.rs

@@ -160,7 +160,7 @@ impl CjsCodeAnalyzer for CliCjsCodeAnalyzer {
           if let Ok(source_from_file) =
             self.fs.read_text_file_lossy_async(path, None).await
           {
-            Cow::Owned(source_from_file)
+            source_from_file
           } else {
             return Ok(ExtNodeCjsAnalysis::Cjs(CjsAnalysisExports {
               exports: vec![],

cli/resolver.rs

@@ -37,7 +37,7 @@ use crate::node::CliNodeCodeTranslator;
 use crate::npm::CliNpmResolver;
 use crate::npm::InnerCliNpmResolverRef;
 use crate::util::sync::AtomicFlag;
-use crate::util::text_encoding::from_utf8_lossy_owned;
+use crate::util::text_encoding::from_utf8_lossy_cow;
 
 pub type CjsTracker = deno_resolver::cjs::CjsTracker<DenoFsNodeResolverEnv>;
 pub type IsCjsResolver =
@@ -62,7 +62,10 @@ pub struct ModuleCodeStringSource {
 pub struct CliDenoResolverFs(pub Arc<dyn FileSystem>);
 
 impl deno_resolver::fs::DenoResolverFs for CliDenoResolverFs {
-  fn read_to_string_lossy(&self, path: &Path) -> std::io::Result<String> {
+  fn read_to_string_lossy(
+    &self,
+    path: &Path,
+  ) -> std::io::Result<Cow<'static, str>> {
     self
       .0
       .read_text_file_lossy_sync(path, None)
@@ -182,18 +185,21 @@ impl NpmModuleLoader {
 
     let code = if self.cjs_tracker.is_maybe_cjs(specifier, media_type)? {
       // translate cjs to esm if it's cjs and inject node globals
-      let code = from_utf8_lossy_owned(code);
+      let code = from_utf8_lossy_cow(code);
       ModuleSourceCode::String(
         self
           .node_code_translator
-          .translate_cjs_to_esm(specifier, Some(Cow::Owned(code)))
+          .translate_cjs_to_esm(specifier, Some(code))
           .await?
           .into_owned()
           .into(),
       )
     } else {
       // esm and json code is untouched
-      ModuleSourceCode::Bytes(code.into_boxed_slice().into())
+      ModuleSourceCode::Bytes(match code {
+        Cow::Owned(bytes) => bytes.into_boxed_slice().into(),
+        Cow::Borrowed(bytes) => bytes.into(),
+      })
     };
 
     Ok(ModuleCodeStringSource {

cli/standalone/binary.rs

@@ -282,14 +282,13 @@ impl StandaloneModules {
           .vfs
           .read_file_all(entry, VfsFileSubDataKind::ModuleGraph)?,
         Err(err) if err.kind() == ErrorKind::NotFound => {
-          let bytes = match RealFs.read_file_sync(&path, None) {
+          match RealFs.read_file_sync(&path, None) {
             Ok(bytes) => bytes,
             Err(FsError::Io(err)) if err.kind() == ErrorKind::NotFound => {
               return Ok(None)
             }
             Err(err) => return Err(err.into()),
-          };
-          Cow::Owned(bytes)
+          }
         }
         Err(err) => return Err(err.into()),
       };
@@ -694,7 +693,7 @@ impl<'a> DenoCompileBinaryWriter<'a> {
             &file_path,
             match maybe_source {
               Some(source) => source,
-              None => RealFs.read_file_sync(&file_path, None)?,
+              None => RealFs.read_file_sync(&file_path, None)?.into_owned(),
             },
             VfsFileSubDataKind::ModuleGraph,
           )

cli/standalone/mod.rs

@@ -91,6 +91,7 @@ use crate::resolver::CliNpmReqResolver;
 use crate::resolver::NpmModuleLoader;
 use crate::util::progress_bar::ProgressBar;
 use crate::util::progress_bar::ProgressBarStyle;
+use crate::util::text_encoding::from_utf8_lossy_cow;
 use crate::util::v8::construct_v8_flags;
 use crate::worker::CliCodeCache;
 use crate::worker::CliMainWorkerFactory;
@@ -516,13 +517,13 @@ impl NodeRequireLoader for EmbeddedModuleLoader {
   fn load_text_file_lossy(
     &self,
     path: &std::path::Path,
-  ) -> Result<String, AnyError> {
+  ) -> Result<Cow<'static, str>, AnyError> {
     let file_entry = self.shared.vfs.file_entry(path)?;
     let file_bytes = self
       .shared
       .vfs
       .read_file_all(file_entry, VfsFileSubDataKind::ModuleGraph)?;
-    Ok(String::from_utf8(file_bytes.into_owned())?)
+    Ok(from_utf8_lossy_cow(file_bytes))
   }
 
   fn is_maybe_cjs(

cli/standalone/virtual_fs.rs

@@ -743,15 +743,12 @@ impl deno_io::fs::File for FileBackedVfsFile {
     Err(FsError::NotSupported)
   }
 
-  fn read_all_sync(self: Rc<Self>) -> FsResult<Vec<u8>> {
-    self.read_to_end().map(|bytes| bytes.into_owned())
+  fn read_all_sync(self: Rc<Self>) -> FsResult<Cow<'static, [u8]>> {
+    self.read_to_end()
   }
-  async fn read_all_async(self: Rc<Self>) -> FsResult<Vec<u8>> {
+  async fn read_all_async(self: Rc<Self>) -> FsResult<Cow<'static, [u8]>> {
     let inner = (*self).clone();
-    tokio::task::spawn_blocking(move || {
-      inner.read_to_end().map(|bytes| bytes.into_owned())
-    })
-    .await?
+    tokio::task::spawn_blocking(move || inner.read_to_end()).await?
   }
 
   fn chmod_sync(self: Rc<Self>, _pathmode: u32) -> FsResult<()> {

cli/util/text_encoding.rs

@@ -11,6 +11,15 @@ use deno_core::ModuleSourceCode;
 static SOURCE_MAP_PREFIX: &[u8] =
   b"//# sourceMappingURL=data:application/json;base64,";
 
+#[inline(always)]
+pub fn from_utf8_lossy_cow(bytes: Cow<[u8]>) -> Cow<str> {
+  match bytes {
+    Cow::Borrowed(bytes) => String::from_utf8_lossy(bytes),
+    Cow::Owned(bytes) => Cow::Owned(from_utf8_lossy_owned(bytes)),
+  }
+}
+
+#[inline(always)]
 pub fn from_utf8_lossy_owned(bytes: Vec<u8>) -> String {
   match String::from_utf8_lossy(&bytes) {
     Cow::Owned(code) => code,

ext/fs/in_memory_fs.rs

@@ -3,6 +3,7 @@
 // Allow using Arc for this module.
 #![allow(clippy::disallowed_types)]
 
+use std::borrow::Cow;
 use std::collections::hash_map::Entry;
 use std::collections::HashMap;
 use std::io::Error;
@@ -457,11 +458,11 @@ impl FileSystem for InMemoryFs {
     &self,
     path: &Path,
     _access_check: Option<AccessCheckCb>,
-  ) -> FsResult<Vec<u8>> {
+  ) -> FsResult<Cow<'static, [u8]>> {
     let entry = self.get_entry(path);
     match entry {
       Some(entry) => match &*entry {
-        PathEntry::File(data) => Ok(data.clone()),
+        PathEntry::File(data) => Ok(Cow::Owned(data.clone())),
         PathEntry::Dir => Err(FsError::Io(Error::new(
           ErrorKind::InvalidInput,
           "Is a directory",
@@ -474,7 +475,7 @@ impl FileSystem for InMemoryFs {
     &'a self,
     path: PathBuf,
     access_check: Option<AccessCheckCb<'a>>,
-  ) -> FsResult<Vec<u8>> {
+  ) -> FsResult<Cow<'static, [u8]>> {
     self.read_file_sync(&path, access_check)
   }
 }

ext/fs/interface.rs

@@ -1,5 +1,6 @@
 // Copyright 2018-2024 the Deno authors. All rights reserved. MIT license.
 
+use core::str;
 use std::borrow::Cow;
 use std::path::Path;
 use std::path::PathBuf;
@@ -288,7 +289,7 @@ pub trait FileSystem: std::fmt::Debug + MaybeSend + MaybeSync {
     &self,
     path: &Path,
     access_check: Option<AccessCheckCb>,
-  ) -> FsResult<Vec<u8>> {
+  ) -> FsResult<Cow<'static, [u8]>> {
     let options = OpenOptions::read();
     let file = self.open_sync(path, options, access_check)?;
     let buf = file.read_all_sync()?;
@@ -298,7 +299,7 @@ pub trait FileSystem: std::fmt::Debug + MaybeSend + MaybeSync {
     &'a self,
     path: PathBuf,
     access_check: Option<AccessCheckCb<'a>>,
-  ) -> FsResult<Vec<u8>> {
+  ) -> FsResult<Cow<'static, [u8]>> {
     let options = OpenOptions::read();
     let file = self.open_async(path, options, access_check).await?;
     let buf = file.read_all_async().await?;
@@ -327,17 +328,25 @@ pub trait FileSystem: std::fmt::Debug + MaybeSend + MaybeSync {
     &self,
     path: &Path,
     access_check: Option<AccessCheckCb>,
-  ) -> FsResult<String> {
+  ) -> FsResult<Cow<'static, str>> {
     let buf = self.read_file_sync(path, access_check)?;
-    Ok(string_from_utf8_lossy(buf))
+    Ok(string_from_cow_utf8_lossy(buf))
   }
   async fn read_text_file_lossy_async<'a>(
     &'a self,
     path: PathBuf,
     access_check: Option<AccessCheckCb<'a>>,
-  ) -> FsResult<String> {
+  ) -> FsResult<Cow<'static, str>> {
     let buf = self.read_file_async(path, access_check).await?;
-    Ok(string_from_utf8_lossy(buf))
+    Ok(string_from_cow_utf8_lossy(buf))
+  }
+}
+
+#[inline(always)]
+fn string_from_cow_utf8_lossy(buf: Cow<'static, [u8]>) -> Cow<'static, str> {
+  match buf {
+    Cow::Owned(buf) => Cow::Owned(string_from_utf8_lossy(buf)),
+    Cow::Borrowed(buf) => String::from_utf8_lossy(buf),
   }
 }
 

ext/fs/lib.rs

@@ -17,6 +17,7 @@ pub use crate::interface::OpenOptions;
 pub use crate::ops::FsOpsError;
 pub use crate::ops::FsOpsErrorKind;
 pub use crate::ops::OperationError;
+pub use crate::ops::V8MaybeStaticStr;
 pub use crate::std_fs::RealFs;
 pub use crate::sync::MaybeSend;
 pub use crate::sync::MaybeSync;