fix: use same UNC path normalization logic with libuv (#306)

---------

Co-authored-by: LongYinan <[email protected]>

Author: sapphi-red

Cargo.lock

@@ -14,7 +14,7 @@ keywords     = ["node", "resolve", "cjs", "esm", "enhanced-resolve"]
 license      = "MIT"
 readme       = "README.md"
 repository   = "https://github.com/oxc-project/oxc-resolver"
-rust-version = "1.70"
+rust-version = "1.74"
 include      = ["/src", "/examples", "/benches"]
 
 [lib]
@@ -70,7 +70,6 @@ serde_json = { version = "1", features = [
   "preserve_order",
 ] } # preserve_order: package_json.exports requires order such as `["require", "import", "default"]`
 rustc-hash = { version = "2" }
-dunce = "1" # Normalize Windows paths to the most compatible format, avoiding UNC where possible
 once_cell = "1" # Use `std::sync::OnceLock::get_or_try_init` when it is stable.
 thiserror = "1"
 json-strip-comments = "1"

Cargo.toml

@@ -0,0 +1,3 @@
+const test = 'hello world'
+
+export default test

fixtures/pnpm/longfilename/index.js

@@ -0,0 +1,7 @@
+{
+  "name": "@oxc-resolver/test-longfilename-aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",
+  "private": true,
+  "version": "0.0.0",
+  "main": "index.js",
+  "type": "module"
+}

fixtures/pnpm/longfilename/package.json

@@ -3,6 +3,7 @@
   "version": "1.0.0",
   "private": true,
   "devDependencies": {
+    "@oxc-resolver/test-longfilename-aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa": "file:./longfilename",
     "axios": "1.6.2",
     "ipaddr.js": "2.2.0",
     "postcss": "8.4.33",

fixtures/pnpm/package.json

@@ -7,6 +7,11 @@ use cfg_if::cfg_if;
 #[cfg(feature = "yarn_pnp")]
 use pnp::fs::{LruZipCache, VPath, VPathInfo, ZipCache};
 
+#[cfg(windows)]
+const UNC_PATH_PREFIX: &[u8] = b"\\\\?\\UNC\\";
+#[cfg(windows)]
+const LONG_PATH_PREFIX: &[u8] = b"\\\\?\\";
+
 /// File System abstraction used for `ResolverGeneric`
 pub trait FileSystem: Send + Sync {
     /// See [std::fs::read_to_string]
@@ -164,55 +169,12 @@ impl FileSystem for FileSystemOs {
         cfg_if! {
             if #[cfg(feature = "yarn_pnp")] {
                 match VPath::from(path)? {
-                    VPath::Zip(info) => {
-                        dunce::canonicalize(info.physical_base_path().join(info.zip_path))
-                    }
-                    VPath::Virtual(info) => dunce::canonicalize(info.physical_base_path()),
-                    VPath::Native(path) => dunce::canonicalize(path),
+                    VPath::Zip(info) => fast_canonicalize(info.physical_base_path().join(info.zip_path)),
+                    VPath::Virtual(info) => fast_canonicalize(info.physical_base_path()),
+                    VPath::Native(path) => fast_canonicalize(path),
                 }
-            } else if #[cfg(windows)] {
-                dunce::canonicalize(path)
             } else {
-                use std::path::Component;
-                let mut path_buf = path.to_path_buf();
-                loop {
-                    let link = fs::read_link(&path_buf)?;
-                    path_buf.pop();
-                    if fs::symlink_metadata(&path_buf)?.is_symlink()
-                    {
-                      path_buf = self.canonicalize(path_buf.as_path())?;
-                    }
-                    for component in link.components() {
-                        match component {
-                            Component::ParentDir => {
-                                path_buf.pop();
-                            }
-                            Component::Normal(seg) => {
-                                #[cfg(target_family = "wasm")]
-                                {
-                                  // Need to trim the extra \0 introduces by https://github.com/nodejs/uvwasi/issues/262
-                                  path_buf.push(seg.to_string_lossy().trim_end_matches('\0'));
-                                }
-                                #[cfg(not(target_family = "wasm"))]
-                                {
-                                  path_buf.push(seg);
-                                }
-                            }
-                            Component::RootDir => {
-                                path_buf = PathBuf::from("/");
-                            }
-                            Component::CurDir | Component::Prefix(_) => {}
-                        }
-                        if fs::symlink_metadata(&path_buf)?.is_symlink()
-                        {
-                          path_buf = self.canonicalize(path_buf.as_path())?;
-                        }
-                    }
-                    if !fs::symlink_metadata(&path_buf)?.is_symlink() {
-                        break;
-                    }
-                }
-                Ok(path_buf)
+                fast_canonicalize(path)
             }
         }
     }
@@ -227,3 +189,77 @@ fn metadata() {
     );
     let _ = meta;
 }
+
+#[inline]
+fn fast_canonicalize<P: AsRef<Path>>(path: P) -> io::Result<PathBuf> {
+    #[cfg(windows)]
+    {
+        // fs::canonicalize was faster on Windows (https://github.com/oxc-project/oxc-resolver/pull/306)
+        Ok(node_compatible_raw_canonicalize(fs::canonicalize(path)?))
+    }
+    #[cfg(not(windows))]
+    {
+        fast_canonicalize_non_windows(path.as_ref().to_path_buf())
+    }
+}
+
+#[inline]
+#[cfg(not(windows))]
+// This is A faster fs::canonicalize implementation by reducing the number of syscalls
+fn fast_canonicalize_non_windows(path: PathBuf) -> io::Result<PathBuf> {
+    use std::path::Component;
+    let mut path_buf = path;
+
+    loop {
+        let link = fs::read_link(&path_buf)?;
+        path_buf.pop();
+        if fs::symlink_metadata(&path_buf)?.is_symlink() {
+            path_buf = fast_canonicalize(path_buf)?;
+        }
+        for component in link.components() {
+            match component {
+                Component::ParentDir => {
+                    path_buf.pop();
+                }
+                Component::Normal(seg) => {
+                    #[cfg(target_family = "wasm")]
+                    {
+                        // Need to trim the extra \0 introduces by https://github.com/nodejs/uvwasi/issues/262
+                        path_buf.push(seg.to_string_lossy().trim_end_matches('\0'));
+                    }
+                    #[cfg(not(target_family = "wasm"))]
+                    {
+                        path_buf.push(seg);
+                    }
+                }
+                Component::RootDir => {
+                    path_buf = PathBuf::from("/");
+                }
+                Component::CurDir | Component::Prefix(_) => {}
+            }
+
+            if fs::symlink_metadata(&path_buf)?.is_symlink() {
+                path_buf = fast_canonicalize(path_buf)?;
+            }
+        }
+        if !fs::symlink_metadata(&path_buf)?.is_symlink() {
+            break;
+        }
+    }
+    Ok(path_buf)
+}
+
+#[cfg(windows)]
+fn node_compatible_raw_canonicalize<P: AsRef<Path>>(path: P) -> PathBuf {
+    let path_bytes = path.as_ref().as_os_str().as_encoded_bytes();
+    path_bytes
+        .strip_prefix(UNC_PATH_PREFIX)
+        .or_else(|| path_bytes.strip_prefix(LONG_PATH_PREFIX))
+        .map_or_else(
+            || path.as_ref().to_path_buf(),
+            |p| {
+                // SAFETY: `as_encoded_bytes` ensures `p` is valid path bytes
+                unsafe { PathBuf::from(std::ffi::OsStr::from_encoded_bytes_unchecked(p)) }
+            },
+        )
+}

pnpm-lock.yaml

@@ -205,3 +205,13 @@ fn nested_symlinks() {
         Ok(dir.join("nm/index.js"))
     );
 }
+
+#[test]
+fn windows_symlinked_longfilename() {
+    let dir = dir();
+    let path = dir.join("fixtures/pnpm");
+    let module_path = dir.join("node_modules/.pnpm/@oxc-resolver+test-longfilename-aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa_m464apeldykmdsyzlfhtrggk24/node_modules/@oxc-resolver/test-longfilename-aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa/index.js");
+
+    let resolution = Resolver::new(ResolveOptions::default()).resolve(&path, "@oxc-resolver/test-longfilename-aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa").map(|r| r.full_path());
+    assert_eq!(resolution, Ok(module_path));
+}